EPA/600/R-94/191
November 1994
LAKE MICHIGAN URBAN AIR
TOXICS STUDY
by
Gerald J. Keeler, P.I.
The University of Michigan
School of Public Health
Ann Arbor, MI 48109-2029
T901758
Project Officer
Gary Evans
United States Environmental Protection Agency
Atmospheric Research and Exposure Assessment Laboratory,
AREAL
Research Triangle Park, N.C. 27711
This study was conducted in cooperation with the U.S. Environmental Protection Agency
ATMOSPHERIC RESEARCH AND EXPOSURE ASSESSMENT LABORATORY
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, N.C. 27711
-------
TECHNICAL REPORT DATA
1. REPORT NO.
EPA/600/R-94/191
2.
^ i iii iih iiiiii ii!
PB95-129102
4. TITLE AND SUBTITLE
Lake Michigan Urban Air Toxics Study
5 .REPORT DATE
May 1994
6.PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Gerald J. Keeler
8.PERFORMING ORGANIZATION REPORT
NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
The University of Michigan
School of Public Health
108 Observatory Rd., Ann Arbor, MI 48109-2029
10.PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Atmospheric Research & Exposure Assessment
Laboratory
Research Triangle Park, NC 27711
13 .TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
During the summer of 1991, an air toxics monitoring program was conducted in the
lower Lake Michigan area. This study was designed to take advantage of the
intensive meteorological and oxidant data base being generated concurrently by the
Lake Michigan Ozone Study (LMOS stations). Over 1,200 samples were collected and
analyzed to determine atmospheric levels of PCBs, pesticides, PAHs, VOCs, particle
mass, and trace elements. In addition, a research vessel and a small aircraft were
employed on selected days to measure micro-meteorological parameters and pollutant
concentrations at offshore locations near Chicago. The goals of this Great Waters
pilot study were to evaluate methods of sample collection and analysis, quantify
the atmospheric concentrations of toxic substances in the lower Lake Michigan area,
compare measurements made over land and over water, attempt to differentiate the
Chicago urban plume from regional background, identify categories of sources for
the target pollutants, and estimate deposition to the lake.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b. IDENTIFIERS/ OPEN ENDED
TERMS
c.COSATI
Atomospheric Deposition
Air Toxics, Atmospheric Monitorine
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (Ihis Report)
Unclassified
21.NO. OF PAGES
£/ "7
20. SECURITY CLASS (This Page)
Unclassified
22. PRICE
-------
Disclaimer
The information in this document has been funded wholly or in part by the United
States Environmental Protection Agency under contract number T901758 to Gerald J. Keeler,
The University of Michigan. It has been subjected to the Agency's peer and administrative
review, and it has been approved for publication as an EPA document. Mention of trade
names or commercial products does not constitute endorsement or recommendation for use.
ii
-------
Dedication
This report is gratefully dedicated to Mr. Terry Clark. Terry generously invested a
great deal of his time, over the last few months, constructively critiquing this report. His
insightful comments on the meteorological and modeling facets of the report were extremely
helpful in finalizing this document. Terry's dedication to his work which he embraced with
enthusiasm is highly respected by those of us who benefited from his generosity. Terry will be
deeply missed by those who have had the opportunity to work with him.
iii
-------
Acknowledgments
We would like to acknowledge the hard work and dedication of the many individuals
who made this study possible. This project was a true cooperative research study between the
U.S. EPA's AREAL and the University of Michigan. In particular, the support of the
USEPA-AREAL staff was crucial during the planning, site setup and field operations.
Initially, EPA's design and QA Plan for the study was prepared by Dale Pahl, Melissa
McMcullough, Pam Blakely, Gary Evans, Bob Stevens, Bob Lewis and Linda Porter. Gary
Evans and Alan Hoffman were instrumental in the coordination and implemention of the
research between the eight laboratories involved in the study in a very short amount of time.
Gary Evans, Alan Hoffman, Bob Stevens, Mack Wilkins, and Bobbie Edmonds all played vital
roles in the site selection and field portion of the study. Many hot hours were spent in the
field at South Haven and aboard the Laurentian for which we gratefully acknowledge. We
would also like to thank Jack Bowen, and his staff, of the U.S. EPA-AREAL for their helpful
field audit during the study. A special acknowledgment is given to Teri Conner, EPA-
AREAL, for her expertise, analysis, and writing of the CMB portions of this report. A special
thanks is given to the University of Michigan students who helped to implement various parts
of the field study. This group includes Frank Marsik together with several students in the
Atmospheric Science Department at the University of Michigan operated all meteorological
equipment while the ship was on-station.
The hard-working field teams for this study are gratefully acknowledged for their
stamina and resolve. The operators successfully collected the comprehensive suite of
pollutants using brand new equipment with great care and special devotion. A special thanks
is given to the individuals who collected samples aboard the R/V Laurentian: Carl Lamborg,
Steve Mischler and Carrie Monosmith, and Marion Hoyer. The University of Michigan crew
aboard the R/V Laurentian was also helpful and accommodating and deserve a special thanks.
The ships crew was comprised of Robert Nauta, Glen Tompkins. Christopher Whims and Jim
White.
iv
-------
Abstract
During the summer of 1991, a study of the dynamics of toxic pollutant transport and
deposition into southern Lake Michigan was performed. This initial effort was aimed at
establishing the feasibility of providing accurate and representative air samples and gradient
measurements of toxic contaminants and nutrients over Lake Michigan. This project was
designated the Lake Michigan Urban Air Toxics Study (LMUATS) and was designed to serve
as a pilot for the atmospheric measurements portion of the Lake Michigan Loading Study. The
sampling conducted allowed for the quantification of most of the 14 Critical Pollutants on the
International Joint Commission (UC) list. A comprehensive suite of atmospheric measurements
was performed for polychlorinated biphenyls (PCBs), polycyclic aromatic hydrocarbons (PAHs),
selected pesticides, volatile organic compounds (VOCs), particulate matter, trace elements
(including mercury), and fine carbon (organic and elemental). Three land-based sites were
operated during the study to provide one upwind sampling point, one urban site, and one
downwind site on the eastern shore of Lake Michigan. Two sites were also Lake Michigan
Ozone Study (LMOS) sites which provided a complete suite of meteorological and some
ancillary air quality data (ozone and nitrogen oxides). The University of Michigan Research
Vessel (R/VLaurentian) was utilized as a collection platform for three different sampling periods
during the one-month study.
The data collected during the LMUATS revealed large gradients and temporal changes
in the concentrations of most compounds measured. Concentrations for most pollutants,
especially the regionally distributed species, were elevated for a one-week period from 16-23
July which was associated with air mass transport from the southwest. This period enabled an
initial assessment of over-water pollutant transport. The concentrations of total PCBs and most
of the PAH compounds measured were greatest, by 1-2 orders of magnitude at the urban site
in Chicago. Ambient concentrations of trace metals were generally two to three times higher
in Chicago than at the rural sites. Atmospheric mercury concentrations were four to five times
higher in Chicago. Dry deposition fluxes of trace elements associated with fine (_< 2.5 nm) and
coarse (2.5-10 /j.m) particles over Lake Michigan were estimated using a combined dispersion-
deposition modeling approach. The effects of micrometeorological parameters, particle size
distributions, wave dynamics, and type and location of sampling sites on the dry deposition flux
estimates are discussed. This report was submitted in fulfillment of T901758 by the University
of Michigan under the sponsorship of the United States Environmental Protection Agency. The
report covers a period from June, 1991 to May, 1993 and work was completed as of May, 1994.
v
-------
Contents
1. Introduction 1-1
1.1. Objectives 1-4
1.2. Technical Approach 1-5
2. Conclusions 2-1
2.1. Concentration Gradients of Pollutants in Southern Lake Michigan Basin .... 2-1
2.2. Source Categories Responsible for Pollutants Measured 2-5
2.3. Urban Input of Toxic Contaminants 2-8
2.4. Comparison of Simultaneous Measurements Over-Land and Over-Water ... 2-12
3. Recommendations 3-1
4. Methods 4-1
4.1. Meteorological Measurements 4-1
4.2. Chemical Measurements 4-4
4.3. Sampling Schedule and Operations 4-5
4.4. Sample Analysis 4-9
4.5. Mercury Sampling and Analysis Methods 4-16
4.6. Quality Assurance 4-24
4.7. Data Processing 4-30
4.8. Mixed-Layer Trajectories 4-30
5. Composition of the Atmospheric Aerosol System 5-1
5.1. PM10 and Ionic Composition 5-1
5.2. Average Trace Element Concentrations 5-22
5.3. Atmospheric Mercury Levels 5-32
5.4. Average VOC Concentrations 5-43
5.5. Average PAH Concentrations 5-44
vi
-------
5.6. Average PCB Concentrations 5-54
5.7. Average Pesticide Concentrations 5-59
5.8. Over-Water versus Land-Based Measurements 5-76
6. Meteorological Analysis 6-1
6.1. Synoptic Overview for LMUATS 6-1
6.2. Micrometeorological measurements 6-13
7. Source Apportionment/Receptor Modeling 7-1
7.1. Introduction 7-1
7.2 Meteorological and Trajectory Analysis 7-4
7.3. Scanning Electron Microscopy (SEM) 7-17
7.4. Correlation Analysis of Observed HAPs 7-31
7.5. Principal Component Analysis 7-39
7.6. Chemical Mass Balance Modeling 7-51
8. Estimating Atmospheric Deposition 8-1
8.1. Introduction 8-1
8.2. Methods 8-2
8.3. Modeling Results 8-8
8.4. Discussion 8-27
9. References 9-1
vii
-------
Figures
Figure 1-1. The Location of the Surface Air Quality and Meteorological Sampling Sites
for the Lake Michigan Ozone Study 1-6
Figure 4-1. Bow Tower Assembly on the Research Vessel Laurentian 4-3
Figure 4-2. Sampling Locations for the Lake Michigan Urban Air Toxics Study 4-8
Figure 4-3. Collocated Elemental and Organic Carbon Samples at South Haven 4-28
Figure 4-4. LMUATS Data Aquisition Stream 4-31
Figure 4-5. LMUATS Data Processing Stream 4-32
Figure 5-1. Variations in Measured PM10 Concentrations at Four Measurement Sites... 5-2
Figure 5-2. Mixed Layer Trajectories Ending at the Three Land-Based Sites on 19
July, 1991 " 5-3
Figure 5-3. Variations in Measured Fine Fraction PM10 Concentrations 5-5
Figure 5-4. Variations in Measured Fine Particle S Concentrations 5-6
Figure 5-5. Variations in Measured Fine Particle S042" Concentrations 5-7
Figure 5-6. Mixed Layer Trajectories to the R/V Laurentian while On-Station
Off-Shore of Chicago 23-27 July and 5-8 August 1991 5-9
Figure 5-7. Variations in Measured Fine Particle H+ Concentrations 5-11
Figure 5-8. Variations in Measured Fine Particle NH4+ Concentrations 5-12
Figure 5-9. Variations in Measured Fine Particle NO3" Concentrations 5-13
Figure 5-10. Variations in Measured Gaseous NH3 Concentrations 5-14
Figure 5-11. Variations in Measured Gaseous S02 Concentrations 5-16
Figure 5-12. Variations in Measured Gaseous HNO3 Concentration 5-17
viii
-------
Figure 5-13. Variations in Measured Maximum Hourly O3 Concentrations 5-18
Figure 5-14. Variations in Measured Elemental Carbon Concentrations 5-20
Figure 5-15. Variations in Measured Organic Carbon Concentrations 5-21
Figure 5-16. Variations in Measured Fine Pb Particle Concentrations 5-26
Figure 5-17. Variations in Measured Coarse Particle Pb Concentrations 5-27
Figure 5-18. Variations in Measured Fine Particle Fe Concentrations 5-29
Figure 5-19. Variations in Measured Fine Particle Mn Concentrations 5-30
Figure 5-20. Variations in Measured Fine Particle Zn Concentrations 5-31
Figure 5-21. Variations in Measured Fine Particle Si Concentrations..: 5-33
Figure 5-22. Variations in Measured Fine Particle Se Concentrations 5-34
Figure 5-23. Variations in Measured Vapor-Phase Hg Concentrations 5-35
Figure 5-24. Variations in Measured Vapor-Phase Hg in Chicago 5-37
Figure 5-25. Variations in Measured Particle-Phase Hg Concentrations 5-39
Figure 5-26. Correlation of Hg(p) by INAA Measured at South Haven with the
Coarse-to-Fine-Mass Ratio 5-40
Figure 5-27. Results of Size Segregated Particulate Hg Sampling in Ann Arbor 5-42
Figure 5-28. Variations in Measured Retene Concentrations 5-45
Figure 5-29. Variations in Measured Coronene Concentrations 5-46
Figure 5-30. Variations in Measured Naphthalene Concentrations 5-51
Figure 5-31. Variations in Measured Benzo(a)pyrene Concentrations 5-52
Figure 5-32. Variations in Measured Total PCB Concentrations 5-57
Figure 5-33. Variations in Measured a- HCH Concentrations 5-62
Figure 5-34. Variations in Measured 7- HCH Concentrations 5-63
Figure 5-35. Variations in Measured DDT Concentrations 5-70
Figure 5-36. Variations in Measured DDE Concentrations 5-72
ix
-------
Figure 5-37. Variations in Measured DDD Concentrations 5-73
Figure 5-38. Variations in Measured Chlorpyrifos Concentrations 5-77
Figure 6-1. Plot of Surface Station Data for 11 July 1991 6-3
Figure 6-2. Plot of Surface Station Data for 24 July 1991 6-7
Figure 6-3. Plot of Surface Station Data for 2 August 1991 6-10
Figure 6-4. Plot of Surface Station Data for 6 August 1991 6-12
Figure 6-5. Schematic of the Micrometeorological Effects on Pollutant Deposition 6-16
Figure 6-6. Spectral Energy Plots for the Data Collected aboard the R/V Laurentian 24
July, 1991 6-24
Figure 6-7. Temporal Variation of the Momentum Flux Measured Aboard the R/V
Laurentian 24-25 July, 1991 6-30
Figure 7-1. Mixed-layer Backward Trajectories Associated with PM10 Concentrations
> 50 |ig/m^3 in South Haven, MI 7-6
Figure 7-2. Mixed-layer Backward Trajectories Associated with PM10 Concentrations
> 50 |J.g/m3 in Chicago, IL 7-7
Figure 7-3. Mixed-layer Backward Trajectories Associated with PM10 Concentrations
> 50 (ig/m3 in Kankakee, IL 7-8
Figure 7-4. Selected Point Sources in Southeast Chicago 7-11
Figure 7-5a. PM10 Concentrations in the Study Region on 17 July 91 7-13
Figure 7-5b. Total Particulate Concentrations in the Study Region on 17 July 91 7-13
Figure 7-6. Coal-Fired Power Utilities in the United States 7-16
Figure 7-7(a-n). Scanning Electron Micrographs 7-22
Figure 7-8. Profiles of PAH Ratios with Benzo(e)pyrene on 17 July 1991 7-49
Figure 7-9. Profiles of PAH Ratios with Benzo(e)pyrene on 6 August 1991 7-50
Figure 7-10. Ratios of Selected PAHs to BeP Measured on the R/V Laurentian 6 August 1991
7-67
Figure 8-1. Mixed Layer Backward Trajectories that Crossed Lake Michigan on 14
and 15 July, 1991 8-10
x
-------
Figure 8-2. Variation of the Deposition Velocity with MMD along Trajectories
Traversing Lake Michigan 8-11
Figure 8-3. Ratio of the Measured / Calculated Concentration of Trace Metals
Associated with Fine Particles Obtained from Backward Mixed-layer Trajectories for
South Haven on 19, 20 and 21 July 1991 8-14
Figure 8-4. Ratio of the Measured Calculated Concentration of Trace Metals
Associated with Coarse Particles obtained from Backward Mixed-layer Trajectories
for South Haven on 19, 20 and 21 July 1991 8-15
Figure 8-5. Backward Mixed-layer Trajectories for South Haven that traversed Lake
Michigan on the 19-21 July, 1991 8-16
Figure 8-6. Curves of Km and Kbs Coefficients vs. Particle Aerodynamic Diameters 8-18
Figure 8-7. Dry Deposition Load of Trace Metals Associated with Fine Particles
obtained from South Haven Measurements and Trajectories 8-20
Figure 8-8. Dry Deposition Load of Trace Metals Associated with Coarse Particles
obtained from South Haven Measurements and Trajectories 8-21
Figure 8-9. Night to Day Dry Deposition Flux Ratio Calculated from 12-hour Ambient
Measurements at South Haven 8-22
Figure 8-10. Dry Deposition Load of Trace Metals Associated with Fine Particles
obtained from Chicago Measurements and Trajectories 8-23
Figure 8-11. Dry Deposition Load of Trace Metals Associated with Coarse Particles
obtained from Chicago Measurements and Trajectories 8-24
Figure 8-12. Chicago to South Haven Dry Deposition Flux Ratio 8-25
Figure 8-13. Dry Deposition Flux for Selected Crustal and Anthropogenic Elements
for Backward Trajectories that Traversed Lake Michigan 8-29
xi
-------
Tables
Table 4-1. Pollutant Measurements Made during the LMUATS 4-5
Table 4-2. Samplers Deployed for The LMUATS 4-9
Table 4-3. Pollutant Species Measured during the LMUATS 4-11
Table 4-4. Vapor Phase Mercury Samples Collected during LMUATS 4-18
Table 4-5. Precision between Replicate Hg Standards 4-19
Table 4-6. Results of Field Blanks for Vapor Phase Mercury by Site 4-20
Table 4-7. XRF vs. INAA Comparison at IIT 4-26
Table 4-8. Comparison of PCBs and Pesticides in Collocated Samples at South Haven. ...4-27
Table 4-9. Comparison of Collocated XRF Fine Particle Data at South Haven 4-29
Table 4-10. Comparison of Collocated Coarse Particle XRF Data at South Haven 4-30
Table 5-1. Summary of ADS Concentration Data Collected from 8 July-9 August, 1991..5-8
Table 5-2. Average and Maximum Concentrations for Fine Fraction Carbon 5-19
Table 5-3. Average and Maximum Concentrations for Fine Fraction Trace Elements
Determined by XRF 5-23
Table 5-4. Average and Maximum Concentrations for Coarse Fraction Trace Elements
Determined by XRF ...5-24
Table 5-5. Average Compositional Ratios for Selected Elements 5-25
Table 5-6. Vapor-phase Hg Measurements in Chicago, on the R/V Laurentian and in
South Haven 5-36
Table 5-7. Results From TSP v. FM CVAFS Analysis of Hg 5-43
Table 5-8. Average and Maximum Concentrations of VOC 5-47
xii
-------
Table 5-9. Average and Maximum Concentrations of PAHs 5-48
Table 5-10. Comparison of Urban Polynuclear Aromatic Hydrocarbons Concentrations
Measured in Three Cities 5-49
Table 5-11. Comparison of PAH Concentrations Measured On or Near the Great Waters.5-53
Table 5-12. Ambient PCB Concentrations 5-55
Table 5-13. Polychlorinated Biphenyls Measured in the Great Lakes Basin 5-56
Table 5-14. Ambient Pesticides Concentrations 5-60
Table 5-15. Concentrations of Pesticides in Rural Locations in the Great Lakes Region. ..5-61
Table 5-16. Atmospheric Concentrations of a- and 7- Hexachlorocyclohexanes in the
Great Lakes Basin 5-64
Table 5-17. Atmospheric Concentrations of HCB in the Great Lakes Basin 5-66
Table 5-18. Atmospheric Concentrations of Chlordane in the Great Lakes Basin 5-67
Table 5-19. Atmospheric Concentrations of DDT in the Great Lakes Basin 5-71
Table 5-20. Atmospheric Concentrations of Dieldrin Measured in the Great Lakes Basin. 5-74
Table 5-21. 12-hour Average Concentrations for Fine Fraction Trace Elements
Determined by XRF When R/V Laurentian on Station Near Chicago 5-82
Table 5-22. Average Concentrations of PAHs when the R/V Laurentian was on Station
Near Chicago 5-83
Table 5-23. Ambient Pesticides Concentrations when the R/V Laurentian was on
Station Near Chicago 5-84
Table 7-1. Southeast Chicago Point Source Information 7-9
Table 7-2. East St. Louis and Granite City Point Source Information 7-10
Table 7-3. Selected Illinois Air Monitoring Sites and Surrounding Source Information 7-14
Table 7-4. Selected Illinois Special Purpose Air Monitoring Sites and Surrounding Source
Information 7-15
Table 7-5. Tabulation of Coarse Fraction Particulate Matter Collected on 21 July 91 Analyzed by
SEM 7-18
xiii
-------
Table 7-6. Tabulation of Fine Fraction Particulate Matter Collected on 21 July 1991 Analyzed by
SEM 7-19
Table 7-7. Individual Particle Analysis Summary 7-29
Table 7-8. Pearson Correlations Among Fine Fraction Elements (and Total Particulate Mercury)
Measured at IIT 7-32
Table 7-9. Atmospheric Tracer Elements and their Known Sources 7-40
Table 7-10. Factor Loadings from Varimax-Rotated PCA of IIT Fine Fraction Elemental Data.
7-41
Table 7-11. Factor Loadings from Varimax-Rotated PCA of Kankakee Fine Fraction Elemental
Data 7-42
Table 7-12. Factor Loadings from Varimax-Rotated PCA of South Haven Fine Fraction
Elemental Data 7-43
Table 7-13. Ratios of PAHs Observed at LMUATS Sites Compared to Other Sites 7-48
Table 7-14. Reported Ratios of PAHs which Represent Gasoline Exhaust 7-48
Table 7-15. Average Fine-Particle Concentrations (ng/m3) and Uncertainties of Elements Used in
CMB Source Apportionment 7-53
Table 7-16. Average Coarse-Particle Concentrations (ng/mJ) and Uncertainties of Elements used
in CMB Source Apportionment 7-54
Table 7-17. Ratios (± analytical uncertainty) of Soil-Related Elements to Si 7-55
Table 7-18. Chemical Mass Balance source profiles 7-59
Table 7-19. Summary of CMB Results for Coarse-particle Fraction in Percent of Measured Mass
with Uncertainty Estimates 7-60
Table 7-20. Summary of CMB Results for Fine-particle Fraction in Percent of Measured Mass
with Uncertainty Estimates 7-62
Table 8-1. Mass Median Diameter for Selected Trace Elements 8-9
Table 8-2. Model Parameters Obtained for Three MMDs along Backward Trajectory
Segments over Lake Michigan on 15 July 1991 8-13
Table 8-3. Comparison of Calculated Dry Deposition Flux with that Determined by
Holsen et cil. (1993) 8-17
xiv
-------
Table 8-4. Comparison of Calculated Dry Deposition Loads with that Estimated by
Holsen et al (1993) 8-26
Table 8-5. Average Dry Deposition Fluxes of Trace Metals to Lake Michigan 8-30
xv
-------
Chapter 1
Introduction
The presence of persistent toxic chemicals within the Great Lakes Basin has been a
matter of public and scientific interest in both the United States and Canada for many years.
Of particular concern are those contaminants which tend to bioaccumulate in the food chain.
This list includes several pesticides, polychlorinated biphenyls (PCBs), and heavy metals
(especially mercury). Advisories are frequently issued by local health authorities, warning
residents against over consumption of fish taken from the lakes. In recent years, much effort
has been directed toward reducing or eliminating direct discharges of contaminants to the
Great Lakes and their major tributaries. However, contaminated fish are often found in
remote lakes where direct water discharges can be ruled out as possible sources. These
findings confirm that the atmosphere represents a significant source or pathway to the Great
Lakes and the surrounding drainage basin for some of the contaminants of most concern.
The Clean Air Act Amendments (CAAA) of 1990 legislated that a wide range of
environmental research and regulatory activity be undertaken by the U.S. Environmental
Protection Agency (EPA) and other agencies to improve the quality of the nation's
environment. Hazardous air pollutants (HAPs) are addressed in Title III of the CAAA. In
regard to the Great Lakes, Section 112(m) states that the EPA Administrator, in conjunction
with NOAA, "shall c.iduct a program to identify and assess the extent of atmospheric
deposition of HAPs to the Great Lakes, the Chesapeake Bay, Lake Champlain, and coastal
waters. As part of such a program the Administrator shall... investigate the sources and rates
of atmospheric deposition of air pollutants."
In March 1991, three months after the signing of the CAAA, an international
workshop of environmental scientists was convened at the Gray Freshwater Biological
Institute in Navarre, Minnesota. Participants included representatives from EPA, the National
Oceanic and Atmospheric Administration (NOAA), Environment Canada, the Ontario
Ministry of the Environment, the Canadian Centre for Inland Waters, the Canadian
1-1
-------
Department of Fisheries and Oceans, and several states and universities. This group was
assembled to assess the state of science and identify critical research needs to answer
questions about the sources, concentrations, and fluxes of air toxics to the Great Lakes, as
well as the other waters listed in Section 112(m) of the CAAA. Among the highest priority
research needs identified during the workshop were: (1) measuring onshore and offshore
deposition rates simultaneously to determine whether or not onshore measurements represent
deposition to the bulk of the lake surface, and (2) characterizing the impact of local urban
plumes on the total toxic burden to the Great Lakes. Both of these research issues are critical
to the design of long-term monitoring networks to determine rates of atmospheric deposition
to the lakes.
In response to the 1990 CAAA and the 1987 Annex to the Great Lakes Water Quality
Agreement between the United States and Canada, a long-term monitoring program is being
jointly implemented by the two countries to assess the relative contribution from atmospheric
processes to water quality degradation in the Great Lakes. This program, known as the
Integrated Atmospheric Deposition Network (IADN), currently is measuring concentrations
of selected toxic substances in ambient air and precipitation at one shoreline location for each
of the five lakes. Eventually, five monitoring sites per lake are planned for the network. A
key objective of the IADN program is to detect trends over time in atmospheric loadings to
the lakes.
In addition to IADN, the U.S. Environmental Protection Agency (EPA) is planning to
conduct a shorter, intensive study of Lake Michigan over the next few years. The Lake
Michigan Loading Study will employ both monitoring and modeling techniques to provide
greater understanding of the sources, transport, and fate of toxic substances entering the lake.
For a period of one year, measurements will be made of target contaminant concentrations in
(and exchange between) lake water, tributaries, and the atmosphere. The resulting dat? will
be used to develop models for predicting the response of Lake Michigan and its fish to
proposed regulatory actions. This project will serve as a template for future studies and will
require information on the impact of local air emission sources, as well as the contribution
from regional air masses.
The maximum local source density near Lake Michigan occurs along its southwestern
shoreline which is dominated by the greater Chicago, Illinois and Gary, Indiana urban areas.
With a population of over eight million, this is the third largest metropolitan area in the
country. In addition to the usual urban air pollution sources, emissions occur from point
sources such as iron and steel manufacturing in East Chicago and Gary; petroleum refining in
1-2
-------
northwest Indiana and northeast Illinois; and other industrial and municipal activities within
the metropolitan area.
A persistent, regional air quality problem has long been experienced in the Lake
Michigan area with high summertime ozone levels. Multi-day ozone episodes frequently
develop in the region when the predominant wind direction is from the south to southwest,
temperatures are relatively high, and relative humidity is high. During such episodic periods,
the National Ambient Air Quality Standard (NAAQS) for ozone is often exceeded at routine
monitoring sites near the lake shore in all four states bordering Lake Michigan (i.e., Illinois,
Indiana, Michigan, and Wisconsin). Typically, ozone concentrations decrease with increasing
distance from the lake shore.
Following several years of unsuccessful attempts to address the summer ozone
problem around the lake through individual State Implementation Plans, the four states
involved decided to join forces to develop a regional response. A large-scale program was
undertaken, with assistance from EPA, to take intensive air quality and meteorological
measurements over and around Lake Michigan during the Summer of 1991. The resulting
database will provide the basis for a photochemical reactive grid model of the lower Lake
Michigan area. Once it has been fully validated and calibrated, the model will be used to
assess alternative regional ozone control strategies.
The field measurement portion of the program, known as the Lake Michigan Ozone
Study (LMOS), was conducted over the period from 17 June through 9 August 1991. Not
only were ground-based continuous measurements taken, but also selected measurements
were made of ozone, ozone precursors, and meteorological parameters aboard several vessels
operating on the lake and aircraft flying transects through the study domain. Upper-air
soundings were also collected with rawinsonde balloon systems on the intensive days and
radar profilers.
In April 1991, at the request of EPA's Region 5, a decision was made by EPA's
Atmospheric Research and Environmental Assessment Laboratory at Research Triangle Park,
North Carolina (AREAL/RTP) to take advantage of the extensive LMOS database by
conducting a concurrent air toxics monitoring study in the lower Lake Michigan area. This
project was designated the Lake Michigan Urban Air Toxics Study (LMUATS) and was
designed to serve as a pilot for the atmospheric measurements portion of the Lake Michigan
Loading Study, scheduled to begin in 1993. The LMUATS was performed by the University
of Michigan Air Quality Laboratory and the EPA AREAL/RTP. Other participants providing
1-3
-------
site operations, air craft measurements, or analytical services included the Illinois Institute of
Technology, Massachusetts Institute of Technology, NOAA's Atmospheric Turbulence and
Diffusion Division (ATDD), ManTech Environmental, Battelle-Columbus, Southwest
Research Institute, and Sunset Laboratories.
1.1. Objectives
The major goals established for the LMUATS were to quantify the concentrations of
selected air toxic species in the lower Lake Michigan area; to identify the source categories
responsible for these contaminants; to attempt to differentiate the contribution of the
Chicago/Gary urban plume from the regional air masses; to compare measurements made over
land with those made over water; to estimate the rates of dry deposition to the lower lake area
during the study period; and to evaluate methods for the sampling and analysis of toxic
substances in ambient air. This latter goal was of particular importance for measuring
mercury (in both the vapor and particulate phases), as the AREAL/RTP had very limited
experience in making such measurements.
It is unknown how much of the toxics in the air over the lakes originates from sources
near (within 20 km) the lake shoreline versus those sources farther upwind. This project was
the first step in the study of the dynamics of toxic pollutant transport over and deposition into
large lakes. This initial effort was aimed at establishing the feasibility of providing accurate
and representative air samples and gradient measurements of toxic contaminants and nutrients
over Lake Michigan. The primary tasks for the project were:
1) Measurements of the horizontal gradients in concentrations of selected toxic
contaminants across Lake Michigan;
2) Interpretation of the measured concentrations to ascertain the probable source areas
and pathways for the contaminants;
3) An initial assessment of the importance of over-water deposition monitoring for
specific toxic contaminants; and
4) An initial assessment of the relative importance of the urban input to the total loadings
of toxic contaminants to Lake Michigan.
1-4
-------
1.2. Technical Approach
The primary objective of this Cooperative Research Project was to assess the
atmospheric deposition of toxic contaminants to the Great Lakes using actual near-shore and
over-water measurements of the critical pollutants and relevant meteorological parameters for
a one-month period concurrent with the Lake Michigan Ozone Study. The States of
Michigan, Indiana, Illinois, and Wisconsin formulated a coordinated research program entitled
the Lake Michigan Ozone Study to investigate the formation and transport of pollutants from
the Chicago/Milwaukee metropolitan areas across Lake Michigan and to quantify the
atmospheric source-receptor relationships in the area surrounding Lake Michigan. A
comprehensive suite of surface and upper-air meteorological and air quality measurements
were collected. These measurements included airborne measurements with tethered balloons
around the lake and on the ships, and by several research aircraft. Figure 1-1 shows the
location and density of the surface sampling sites utilized for the LMOS. By performing
meteorological and standard air quality measurements aboard two research vessels
strategically positioned on Lake Michigan; the University of Michigan was an integral part of
the 1990 LMOS Pilot Study. The experience gained in taking the over-water measurements
during the summer of 1990 formed the basis for a proposal submitted to the Great Lakes
National Program Office in the Fall of 1990. The proposed concept was eventually adopted
by AREAL/RTP and ultimately became the basis for the LMUATS.
Observations made during the 1990 Pilot Study suggested that the study objectives
would best be served by positioning a vessel along the western shore of Lake Michigan at a
distance offshore from Chicago. Positioning the research vessel offshore of the major
urban/industrial centers allowed for the direct measurement of the transported urban toxic
contaminants. The influence of the various urban centers can be directly measured by tracking
the urban plumes out over the Lake, using meteorological and chemical tracer techniques, and
comparing the concentrations measured on the research vessels. Measurements of varying
sampling duration were performed from 12 to 24 hours in duration to allow the complex
meteorological and chemical transformations to be observed and investigated. The
measurements provide some essential data to investigate the "urban hot spot" concept that a
large fraction of the atmospheric deposition into Lake Michigan may occur at relatively short
distances from shore. In addition, the atmospheric contaminant and nutrient data coupled
with the meteorological and tracer measurements should provide the needed information for a
preliminary assessment of the relative importance of the urban centers in the total loading of
toxic contaminants to Lake Michigan.
1-5
-------
Figure 1-1. The Location of the Surface Air Quality and Meteorological
Sampling Sites for the Lake Michigan Ozone Study.
1-6
-------
These data can potentially be merged with climatological data to investigate deposition
trends or patterns to the lake. In addition, the study design may provide the necessary data to
allow a detailed source apportionment of the measured toxic contaminants. The local and
regional scale receptor modeling utilizes the VOC and trace element measurements taken
during the month-long investigation. This project was unique in its attempts to couple actual
over-water measurements of critical toxic species with intensive meteorological and tracer
measurements to begin to quantify the actual atmospheric deposition of these contaminants
while determining the potential sources of the measured species. A direct, but limited
comparison of the over-water and land-based measurements was carried out. This
comparison provided an initial assessment of the need for actual over-water measurements but
can not be extrapolated to other large lakes or to other time periods for Lake Michigan. A
hybrid receptor-deposition modeling framework was developed concurrently with this project
to provide a more comprehensive framework for assessment of atmospheric deposition to
large bodies of water. This framework can be utilized for the Lake Michigan Loading study
and for other investigations dealing with the Great Waters.
1-7
-------
Chapter 2
Summary and Conclusions
The Lake Michigan Urban Air Toxics Study was a pilot investigation aimed at
assessing the impact of the urban plume on the deposition of air toxic pollutants into Lake
Michigan. Several conclusions can be drawn from this study which can be used in
planning for future investigations on the other Great Lakes and for the Great Waters
including Lake Champlain, Chesapeake Bay, and coastal estuaries.
The major objectives of the study were: 1) to quantify the horizontal gradients in
concentrations of selected toxic contaminants across Lake Michigan; 2) to identify the
source categories and most probable pathways for these contaminants; 3) to provide an
initial assessment of the relative importance of the urban input to the total loadings of
toxic contaminants to Lake Michigan, including the differentiation of the contribution of
the Chicago/Gary urban plume from the regional air mass; and 4) to assess the importance
of over-water versus over-land deposition monitoring for specific toxic contaminants.
In order to accomplish these objectives, a comprehensive suite of atmospheric
measurements was performed during a month long study. The approach used in this study
was to quantify the levels of a large number of HAPs at three land-based locations
(upwind, central, and downwind of Chicago) often linked by air mass transport and at an
additional over-water platform along the path between the urban and the downwind site.
Kankakee, IL was identified as the site upwind of Chicago. IIT, approximately 1.6 km
from the shore of Lake Michigan, was chosen as the central Chicago site. The downwind
site was located across Lake Michigan in South Haven, MI. The University of Michigan
R/V Laurentian was deployed offshore of Chicago for over-water measurements of
HAPs.
2-1
-------
2.1 Concentration Gradients of Pollutants in the Southern Lake Michigan Basin
Generally, the monitoring site in urban Chicago (IIT) had the highest
concentrations of HAPs and trace elements (Si, K, Ca, Ti, V, Mn, Fe, Cu, Zn and Pb).
However, there were notable exceptions. The trace elements concentrations in
Kankakee were often comparable to Chicago. This is in contrast to the other two sites,
over-water and South Haven, where levels of most trace elements were at least a factor of
2 less than IIT. Specifically for Hg, the median concentration of vapor phase Hg was four
times higher at IIT site when compared to over-water and South Haven, while the
maximum concentration was more than 15 times. Particulate Hg levels were also 5 to 15
times higher in Chicago (IIT) than at South Haven or on the R/V Laurentian, suggesting
that local sources of Hg are likely to be responsible for the Hg observed at the Chicago
(IIT) site.
During the course of the 30-day investigation, two periods in which the prevailing
wind flow was from the south-southwest were encountered. This provided an
opportunity to observe the behavior of the many different classes of compounds as they
were advected from source regions in Illinois across the water to South Haven. The
concentrations of particulates and several vapor phase Hg compounds decreased as they
were advected from the urban source region. From the shift in the size distribution of the
particulate matter between Chicago (IIT) and South Haven, it was evident that the coarse
particulate matter was being depleted (by dry deposition) as the aging air mass was
advected over the lake to the downwind monitoring site in South Haven. As a result of
both dispersion and deposition, the concentration of some compounds, such as vapor and
particulate mercury, decreased rapidly to the regional 'background' levels within a short
distance from the urban/industrial source areas.
During periods of flow from the north and northwest (14, 15, 24-27 July 1991),
pollutant concentrations measured at Chicago, South Haven, and on the R/V Laurentian
were typically quite low. Levels of some of the fine fraction metals such as Pb, Fe, Mn
and Zn measured in Kankakee during these periods were elevated above the levels
observed at IIT. Since IIT in Chicago is only 1.6 km from the lake, the higher levels
observed at Kankakee suggest an impact due to source(s) between the two sites (e.g. in
southeast Chicago/Gary area).
2-2
-------
Concentrations of the more reactive compounds, such as many of the PAHs, were
uniformly higher concentration at IIT when compared to the other three monitoring sites.
Also, concentrations of carcinogenic compounds such as benzo(a)pyrene and
indeno(l,2,3-c,d)pyrene were approximately 10 times higher than the other sites.
Dispersion, deposition, and transformation of the compounds are responsible for these
lower concentrations. It is interesting to note that most of the measured PAHs were
found in higher concentrations over the water than at the South Haven site for several
days and on average. At Kankakee, there is an apparent impact due to combustion
processes, as evidenced by elevated levels of naphthalene and the peak concentration of
benzo(a)pyrene measured at Kankakee on 17 July 1991. Concentrations of several
compounds, such as naphthalene, acenaphthylene and retene exhibited somewhat similar
concentrations between sites, which indicates a more regional source distribution. Since
these compounds are predominantly found in the vapor phase in the atmosphere, their
lifetimes may be appreciably shorter than that of the larger molecular weight compounds
associated primarily with the particulate phase.
The ambient concentrations of most of the PCBs measured at IIT were at least a
factor of 3 greater than at the other measurement locations. Also, total PCB
concentrations at IIT were slightly higher than many of the reported PCB measurements in
the Great Lakes region during the last 5-years (Table 5-13). There were two exceptions
to this. First, the concentrations of low molecular weight PCBs (2-PCB and Total mono-
PCB) measured on the R/V Laurentian and in South Haven were 4 times greater than the
concentrations observed at IIT when flow was out of the east from Michigan. This is an
interesting finding, especially since quantification of these low molecular weight PCBs is
most likely an underestimate due to their high volatility and inherent difficulty to be
captured quantitatively. Second, the concentration of PCBs measured over the lake were
greater than measurements taken in Kankakee and South Haven.
A review of the total PCB measurements during the period for which the
prevailing flow was from the southwest indicates that total PCBs were not transported
from the urban Chicago area downwind to South Haven. This implies a strong
concentration gradient of PCBs around Chicago with rapid dispersion of the already
relatively low levels of these compounds. A similar trend was observed for mercury.
Dispersion alone results in concentrations of these compounds at levels too low to detect
with adequate precision within a short distance from the source(s). However, the low
levels of these compounds does not imply that they are not important with regard to the
2-3
-------
problem of deposition. On the contrary, the deposition of picogram and nanogram quantities
of pollutants such as mercury and PCBs which bioaccumulate in the food chain are currently
believed to be responsible for a substantial portion of these toxins entering the Great Lakes
(Strachan and Eisenreich, 1992).
The majority of the pesticides/herbicides quantified during the LMUATS were highest
on average at Kankakee. Notable exceptions were Aldrin and Simizine, which were found in
higher concentrations at IIT than at the other 3 sites; and DDT (and its derivatives) which
were greatly elevated primarily at the South Haven site. Chlorpyrifos had similar
concentrations at Kankakee and South Haven and Mirex had similar concetrations at IIT and
Kankakee. The elevated mean determined for several pesticides at Kankakee is largely due
to samples collected on 29 July, and 2, 5, and 6 August 1991 which were highly elevated
in a-HCH, k-HCH, hexachlorobenzene, Mirex, K-chlordane. and Dieldrin. Under certain
meteorological conditions (e.g. strong southwesterly flow), it is possible that elevated
concentrations of some pesticides in the Kankakee region are serving as an important source
of these compounds to the lakes. Average concentrations for many currently used
pesticides/herbicides (e.g. atrazine) are 2 to 3 times higher on average at Kankakee than at
IIT. Current pesticide use in areas along the Michigan coast, especially in the case of aircraft
application, may contribute substantially to the total burden of these compounds to the Great
Lakes. Significant concentrations of many of the pesticides were observed over-water both
in the northcentral portion of the lake (at the 100m station off Muskegon) as well as offshore
of Chicago. For most of the pesticides, concentrations varied dramatically by location and
magnitude on the basis of regional crop distribution, pesticide use patterns, and application
requirements. Emissions from these source areas will depend on the properties of the
pesticide and evaporation, which are largely controlled by precipitation and temperature.
Some pollutants had higher concentrations at the upwind (Kankakee) site and the
downwind (South Haven) site as a result of local emissions (present and historical) in these
locations. Kankakee experienced the highest mean concentration of fine mass as well as
several of the lighter elements such as Al, Si, S and CI in the fine fraction. In addition, the
mean fine Se concentration was highest at Kankakee, indicating that the regional coal
combustion influence has a dominant impact on this site. Observations of particulate matter
measured during the study indicate that locations such as Kankakee are highly susceptible to
frequently elevated pollutant levels as a result of transport from several large point sources
as well as strong regional source areas. The findings indicating the apparent importance of
2-4
-------
few large point sources was somewhat unexpected with the cement producing facilities in
Oglesby, IL 40 km west-northwest of Kankakee, and the many coal-fired power plants
distributed in the southern Illinois region. These sources make selection of a non-polluted
'upwind' site for Chicago measurements very problematic.
One of the most unexpected and important findings of the LMUATS was the
discovery of significantly elevated ambient concentrations of DDT, DDD and DDE in
South Haven, MI. The concentrations of the parent 4,4—DDT compound in samples
collected at South Haven are significantly elevated above other reported values. The
levels of the daughter compound, p,p'-DDE, are an order of magnitude greater in
concentration than previously reported total DDT levels. These high levels have been
confirmed by further studies at this site during 1992-1993 (Keeler, unpublished data). The
average DDT:DDE concentration ratio measured in South Haven is 0.3. Since the
concentration of the DDT breakdown product is three times that of the parent compound;
this provides sufficient evidence that levels observed are probably due to release of
previously applied DDT. The time for degradation of DDT to DDE is slow enough to
implicate re-emission of a much earlier DDT application as the source of this compound.
While the levels of DDT, DDD, and DDE, were interestingly high during the LMUATS
only at South Haven it should be noted that the levels continue to be elevated at this
southwestern Michigan site today. Samples analyzed for Semivolatile Organochlorine
(SOCs) compounds, including DDT. as part of a Michigan Great Lakes Protection Fund
sponsored project, confirm that the eievated levels during the 1-month study in 1991 were
not anomalous in nature or due to a one-time application of the banned pesticide (Keeler,
Monosmith. and Hermanson, unpublished data). The elevated levels of DDT seen in
South Haven are not observed at the other Michigan sites located across the state (Ann
Arbor, Pellston, or Deckerville).
The atmospheric chemistry of the pollutants as they are advected over the water
from source regions including the Chicago/Gary urban area is extremely complex.
Measurements made during LMUATS suggest that when transport is from the southwest,
over-water measurements of aerosol acidity (H~r), and the to SO42" ratios indicate that
the sulfate transported from upwind sources remained largely unneutralized (ratio close to
1 on the R/V Laurentian), and not totally neutralized in South Haven (average ratio of
0.4). The aerosol acidity has been hypothesized to have a large impact on the
composition, solubility, and biological availability of toxic species as they are deposited
from an air mass advected from a source region out over the lakes. Cadmium samples
2-5
-------
taken using Hi-volume samplers taken during the LMUATS were analyzed to determine
the water soluble and acids soluble fractions of this toxic metal (Hashim, J., 1993). The
water soluble fraction of the Cd was found to increase with increasing aerosol acidity
measured concurrently. The highly acidic air masses are also accompanied by highly
oxidative conditions caused by increased concentrations of ozone found in the air masses
as they impact the downwind shore. Air masses that arrived in South Haven after
transport from the southwest were often found to contain the chemical reactants involved
in complex oxidations involving both inorganic and organic compounds. Ozone levels in
South Haven exceeded the NAAQS on several days, reaching 158 ppb during the period
from 16-22 July which resulted from southwest transport.
2.2 Source Categories Responsible for Pollutants Measured
2.2.1. Sources of Particulate Mass and Trace Elements
Sources and source categories for a subset of the HAPs measured were assigned
based on specific source tracer species and emission profiles determined from previous
studies. This information was then used to interpret the results from principal component
analysis, chemical mass balance modeling, and other approaches such as ratios of specific
tracer compounds and elements, and individual particle analysis by scanning electron
microscopy.
Through careful review of extensive work performed by Sweet and Vermette
(1992, 1993) and through discussions with emissions and permit specialists at the State of
Illinois and other County Agencies, the major pollutant sources in Illinois, and the major
urban/industrial areas in the southern Lake Michigan Basin including Chicago, can be
inventoried. Among the major hazardous pollutant sources in the region are iron and steel
manufacturing, petroleum refining, municipal waster incineration, coal-fired power plants,
plating industry, and non-ferrous metal industry.
The influence of the iron-steel industry on the composition of the measured aerosol
during was clearly observed. The Fe concentrations in both fractions of the PM10, were
elevated, at all 3 sites and are above the contribution that would be expected from
windblown dust. Analysis of the IIT fine fraction trace element data using principal
component analysis yielded four dominant source categories: iron and steel manufacture,
soil derived components, metals industry, and a regional coal-combustion source
dominated by sulfur. A secondary A1 smelter located in the St. Louis area may be the
2-6
-------
source of elevated concentrations of several metals, in addition to Al, measured at the
land-based sites on several occasions.
At Kankakee the factor analysis results were less clear identifying a soil-derived
component, mixed metals source component, and the dominant regional sulfate
component. The factor analysis results at South Haven included a three factor model with
iron and steel emissions, resuspended soil and the regional sulfate as the dominant
contributors to the fine mass measured at South Haven.
The chemical mass balance approach applied to the IIT measurement data used
source categories based on the U.S. EPA data base SPECIATE in conjunction with source
profiles published by Vermette et al (1990). For the days during which the predominant
flow was from the southwest, this model was able to explain 41-69% of the coarse mass
with the largest contribution from crustally derived material, 12-19% was due to limedust,
3-5% was due to fugitive steelyard emissions and 1-2% was due to incinerator emissions.
However, due to the complexity of the airshed and the lack of appropriate source profiles
for many of the previously defined major source categories, the model could not
effectively explain the fine mass measured at IIT.
The episode of elevated PM10 observed from 16-22 July is also seen in the time
series plots of fine Fe, Mn, and Zn, which are marker elements for iron-steel combustion.
The levels of these elements are higher both at IIT and South Haven during the period 16-
20 July. This would suggest that the steel plants in the Gary/southeast Chicago area are
the most important contributor to the levels of these metals measured in South Haven. On
21 July the concentrations of Fine Fe are elevated at both Kankakee and IIT in Chicago
but not at South Haven. This is due to a stationary front sitting between South Haven and
Chicago which meteorologically cut South Haven off from the dominant southerly flow to
the sites south of the front. The levels observed at the 3 land-based sites on 22 July for
the iron-steel related elements are all elevated. The highest concentrations are first
observed in Kankakee. This is consistent with the mixed layer trajectories for this day
which meteorologically "connect" Kankakee with Chicago and South Haven. The
elevated concentrations of iron-steel plant marker elements were again seen on 5-8
August.
In order to determine the relative impact of combustion sources for some trace
elements measured at the sites versus the crustal components, the ratios of Al/Si, K/Fe,
and Fe/Si were calculated. The Kankakee site appears to be impacted by a combustion
2-7
-------
source emitting A1 in the PM10 range. The fine Al/Si concentration ratio for South Haven
is also elevated above the crustal average at 0.80. This average is strongly influenced by
the extremely elevated A1 concentration measured on 2 August which comprised more
than half of the mean. If this one sample were removed, the fine Al/Si ratio would drop to
0.4, which is closer to the IIT ratio and the crustal average. In addition, the ratios suggest
that a source or combination of sources of fine Fe is also contributing to the Fe observed
at all of the sites. Except for the coarse Al/Si ratio at Kankakee the remaining
concentration ratios are similar to the crustal average, suggesting that the coarse-fraction
aerosols are primarily of soil or crustal origin. This allows for the utilization of the crustal
ratios for these elements in investigating the origin of other elements which are of a mixed
soil and anthropogenic origin.
Sources of vapor phase Hg measured at IIT were described using a 4-component
model. The majority of vapor phase Hg was explained by the factor which indicated iron
and steel sources. A smaller amount of the variability in the data was accounted for by the
regional sulfate and soil components. Two important findings of this study are that coarse
particle Hg can be measured in both urban and rural locations; and that the form, and
perhaps the reactivity, of the particulate Hg seems to vary depending upon the source and
meteorological conditions. Since the study on Lake Michigan, data indicate that the
particulate Hg is not always submicron in aerodynamic size but can be found with super-
mircon particulate matter. As a result, some dry deposition estimates have probably
underestimated the dry deposition of this toxic compound.
In order to verify the statistical conclusions of source type, scanning electron
microscopy was utilized on a subset of samples collected at IIT to describe individual
particles in terms of the morphology and their elemental composition. This analysis,
although carried out for only 6 samples, lends strong evidence to support many of the
suspected source types including iron and steel, coal combustion, incineration, A1
production, diesel emissions, and resuspended soil or urban and dust. The use of SEM
points to areas for future study and has implications for attempts to model the deposition
and chemical characterization of the aerosols under study. The complexity of the
particulate matter is evident in the photomicrographs generated by SEM. The long chain
Fe spheres, proliferation of sulfate aerosols, soot conglomerates and intricately patterned
salts and other crustal elements illustrate the complex nature of the mixture that must be
sorted out in trying to understand the urban atmosphere.
2-8
-------
2.2.2. Sources of Organic Compounds
Recent studies of Vermette et al (1992) and Sweet and Vermette (1992) indicate
that sources for organic compounds in the region of Illinois encompassing the Kankakee
site include rubber reclaiming, petroleum refining, chemical manufacture and hazardous
waste incineration.
Sources of the specific PAHs quantified in this study, include a multitude of
stationary and mobile combustion sources. Most of the source apportionment techniques
used in the past to delineate sources of trace elements are difficult, or at best problematic,
to apply to PAHs. An attempt was made in this report to compare results of ratios of
several of the more stable particulate PAHs to those ratios observed by other investigators
in order to determine source signatures for gasoline exhaust and diesel exhaust. Although
some generalizations can be made, however, much more extensive work is required before
conclusive statements can be postulated concerning the origin of the PAH aerosols and
gases quantified in an urban air mass.
In general, the concentration ratios indicate that PAHs measured in Chicago at the
IIT site (benzo(g,h,i)perlyene, coronene, benzo(g,h,i)pereylene, indeno(l,2,3-c,d)pyrene,
benzofluoranthenes, crysene, benzo(e)pyrene, benzo(a)anthracene and benzo(a)pyrene)
have a substantial contribution from mobile sources including diesel and gasoline exhaust.
The ratios of PAHs calculated using the Kankakee, R/V Laurentian and South Haven
concentration data, were less similar to those found by Li and Kamen or Tong and
Karasek indicating that at those sites, sources for the higher molecular weight PAHs are
likely due to transport from distant sources or other local emissions, rather than
predominantly a result of local mobile emissions.
Sources for pescicides measured at Kankakee and South Haven are due to current
or historical application to local crops. Both Kankakee and South Haven are intensively
farmed for a variety of crops, mainly corn, beans and wheat in the Kankakee area and fruit
and table vegetables in the South Haven area. Historical use and subsequent re-emission
of DDT appears to continue to be a major source of this compound to the atmosphere at
South Haven.
In the extremely complex airshed encompassed in the LMUATS study, any one of
the techniques may not independently yield all of the necessary information regarding
2-9
-------
sources. Using additional hybrid modeling techniques such as Quantitative Transport Bias
Analysis should yield important interpretation to the data collected during LMUATS.
2.3 Urban Input of Hazardous Air Pollutants
The measurements of hazardous air pollutants documented in this study provide
evidence to support the hypothesis that urban areas are significant contributors to
atmospheric levels and, therefore, significant contributors to deposition of HAPs to
adjacent water bodies. The data also indicate that for some pollutants such as Hg and
PAHs, elevated concentrations and strong horizontal concentration gradient exists
downwind of a large urban/industrial area.
The near field input of potentially toxic pollutants to the Great Lakes from on-
shore urban/industrialized areas is likely elevated above levels found in the midlake regions
as well as those off-shore from rural, non-industrialized areas. The Clean Air Act
Amendments required that the sources and rates of atmospheric deposition of hazardous
air pollutants be investigated and determined. Critical to determining the rate of
deposition of atmospheric compounds is characterizing the impact of local urban plumes
on the total toxic burden to the Great Lakes.
2.3.1. Differentiating the Contribution of Toxic Pollutants To Lake Michigan From
Chicago/Gary versus the Contribution From Upwind Locations
During the LMUATS, it was discovered that the concentrations of many
hazardous trace compounds at the "upwind" site at Kankakee were elevated on occasion
above those concentrations measured in Chicago (IIT). Therefore, it was hypothesized
that pollutant sources upwind of Chicago/Gary must be contributing to the total loading of
atmospheric pollutants to Lake Michigan. In order to separate the source contribution of
the regional component from the Chicago/Gary urban input to the lake, tracer species (or
a source fingerprint of some type) would be needed by which the upwind contribution
could be distinguished, allowing it to be identified after mixing with the urban aerosol.
This is a difficult and complex task and has been accomplished only in part by the
techniques utilized in this study.
2-10
-------
There is a pronounced episode of elevated PM-10 concentrations from 16-22 July
where the levels exceeded 80 (ig/m3 at both Kankakee and Chicago (IIT), and exceeded
60 (j.g/m3 at South Haven. Mixed layer trajectories reveal that the stagnant air masses
were transported from the southwest during this period. The maximum PM-10
concentrations were observed on 19 July with air mass transport from the St. Louis area
to each site. The PM-10 concentrations at each site peaked again on 2 August when
winds again were light and from the southwest.
Perhaps more interesting then the elevated PM-10 levels during the study is the
change in the aerosol size distribution after transport from Kankakee to IIT to South
Haven as reflected by the fine fraction of the PM-10. One objective of the LMUATS was
to capture an episode where air mass transport "linked" the upwind site (Kankakee) with
the urban site (IIT) and the over-water transport site (R/V Laurentiari) and the downwind
site (South Haven). The episode from 16-23 July provided the perfect opportunity for
linkage between 3 of the 4 sites with the research vessel which was, unfortunately, not on
station during this period. However, at the beginning of the episode on 16 July, about
78% of the PM-10 observed at Kankakee was in the fine fraction which is typical of a
regionally transported aerosol. The PM-10 level and the percentage of coarse particulate
matter at Kankakee rose from 18-22 July as a result of transport from sources upwind of
Kankakee such as the St. Louis and East St. Louis urban/industrial area. As the air mass
moved into Chicago the concentration of the fine particulate mass dropped due to
dispersion. The addition of coarse mass occurred from the activities in the urban/industrial
area. The PM-10 levels observed at IIT were only slightly lower than those measured in
Kankakee but about 40% of the PM-10 was now in the fine fraction. As the air masses
continued across Lake Michigan, without additional sources en route to South Haven, the
fine fraction reasserted itself as the dominant portion of the PM-10. This is what one
typically observes at a site influence.! primarily by long range transport. The PM-10 levels
in South Haven during the episode were about 20 |ag/m3 less on the average than the
levels observed concurrently at Chicago (IIT). This decrease can be largely explained by
dispersion. The change in the particle size distribution, however, cannot simply be
explained by dispersion en route. It appears that larger particles observed in IIT do not
make it to South Haven and are probably lost by deposition. Therefore, the PM-10 is
primarily made up of fine particles in South Haven after over-water transport. This loss of
large particles probably would have been accentuated if the particles greater than 10 in
size had been sampled (Holsen et al., 1993).
2-11
-------
For some species, such as elemental carbon, differentiation of the urban plume
from upwind sources is simplified. During the transport period 16-22 July, the elemental
carbon levels observed in Kankakee do not indicate an impact from upwind sites and are
much lower than those levels at Chicago. This suggests that local combustion sources in
Chicago are responsible for the elevated levels measured there as well as at the downwind
site in South Haven.
More extensive source profiles for a variety of industrial types in both St. Louis
and Chicago would be needed to quantitatively differentiate the regional source
contributions from the urban aerosol. In lieu of this deficiency, routine monitoring data
from the States of Illinois, Indiana, Wisconsin, and Michigan were obtained to provide
better spatial coverage for measurements of particulate matter. The more comprehensive
set of PM10 and TSP measurements made in Illinois during the LMUATS period reveal
that there are more substantial sources of fine and coarse particulate matter in the vicinity
of the Kankakee site.
2.3.1. Relative Importance of the Urban Input of Toxic Compounds to the Total
Atmospheric Contaminant Loading to Lake Michigan
It is unknown how much of the toxics in the air over the lakes originated from
sources near (< 20 km) the lake shoreline versus those originating from sources further
upwind. Evidence suggests that during periods of southwest transport sources upwind of
Chicago were impacting this site, thus entering the air mass being advected over the lake.
An important goal of the LMUATS was to supply data that would provide an opportunity
to compare model results with measurements of deposition in order to ascertain the degree
of agreement between them. This study provides insufficient over-water HAPs
measurements, due to poor meteorological conditions, to adequately evaluate deposition
models. The data do provide critical information regarding whether the model results are
reasonable.
The relative contribution of the coarse and fine fractions to the overall particle
loadings from the urban sources is a very complex and difficult problem to address. The
extremely uniform PM10 and particulate sulfate concentrations measured from southern
Illinois across the lake into Michigan suggest that the fine fraction material is not the result
of the Chicago/Gary urban emissions. The course fraction of the PM10 and coarse
particles greater than 10 |am may be the dominant fraction to understand in relation to
urban deposition. One study of particulate mass distributions and dry deposition flux in
2-12
-------
Chicago and over-water during the LMUATS concluded that the majority of the particle
loading to the lake is the result of particles greater than 6.5 ^m (Holsen et al., 1993).
These conclusions are supported by measurements of the size distribution of the aerosol
system in Chicago.
The dry deposition fluxes to Lake Michigan for selected trace metals associated with
fine and coarse particles were estimated using a novel hybrid receptor-deposition modeling
approach. This model included parameters that take into account the variability in the
deposition flux over the lake as a function of meteorological parameters, particle size
distribution, parameters controlling the physics of water waves, and the type and location
of sampling sites. Significant temporal and spatial variations in the deposition velocities
were obtained for trajectories traversing lake Michigan from one day to the next. The
results of the dry deposition modeling indicate that one must not assume that the
deposition flux, calculated using dry deposition velocities, to be constant in time or space
for the total area of the lake. A large uncertainty in the deposition flux is associated with
assuming constant deposition velocities.
Holsen et al. (1993) estimated that fine particles are responsible for about 2% of
Pb and 0.6% of Ca in the total deposition flux at the Chicago (IIT) site. The Pb and Ca
deposition flux associated with coarse particles with an aerodynamic diameters in the
range of 2.5 to 10 ^m accounted for about 11.5% and 8% of the total fluxes of Pb and Ca,
respectively. Based on these estimates, the dry deposition flux to Lake Michigan during
the LMUATS by the fine and coarse fraction of the PM10 calculated by Holsen et al.
(1993) and by this study are shown in Table 7-3. The higher ratios for the coarse fraction
may be due to (a) particle growth in a humidity gradient (not considered in this study)
when the air parcel crosses the lake leaving submicron particles in the inertial deposition
range, yielding higher deposition fluxes; (b) position of the sampling station in the plume;
(c) statistical uncertainty related to deposition measurements; (d) small representativeness
of measured fluxes in one point for the whole lake area; (e) the obvious limitation of
annual estimates discussed by Holsen et al. (1993) in which deposition fluxes measured
during the summertime are assumed valid throughout 1991; and (f) different deposition
flux distributions for particle size obtained in Chicago (IIT) (used in this comparison)
might be obtained by sampling directly on Lake Michigan.
To estimate the urban contribution of the deposition loading to Lake Michigan and
to evaluate the choice of sampling site (i.e., rural, urban/industrial) and its location in the
basin on the deposition estimates, the ratio between the calculated Chicago (IIT) and
2-13
-------
South Haven dry deposition loads for trace metals associated with fine and coarse
particles was calculated. The ratios for most trace metals in the fine fraction are in the
range of 0.6 to 3 with the exception of total Hg which has a ratio of 16. The high ratio for
total Hg is probably due to strong local sources in the Chicago area since ambient Hg
concentrations measured in Chicago (IIT) are about 10 times higher than those measured
at South Haven.
Using the hybrid modeling approach to estimate dry deposition fluxes, it was
determined that Si had the highest deposition flux during the LMUATS (360 tons) for
trace metals primarily of soil or crustal origin, followed by Ca (254 tons), Fe (160 tons)
and K (58.3 tons). The deposition flux of A1 (150 tons) was the highest loading followed
by S (55 tons), Zn (4 tons), Mn (2.35 tons), Cu (2.2 tons), Cd (1.07 tons), Pb (1.07 tons),
Se (0.93 tons), Ni (0.8 tons), Cr (0.53 tons), As (0.44 tons), Br (0.38 tons), and total Hg
(0.035 tons).
The dry deposition flux for specific compounds to Lake Michigan contributed by
sources in the urban area of Chicago/Gary is estimated to be 2 to 10 times greater than
that from regional sources. A factor of 10 increase in the deposition flux of many of the
SOCs attributed to the Chicago urban/industrial was also determined.
2.4 Comparison of Simultaneous Measurements Over-Land and Over-Water
The question of whether it is reasonable to utilize shoreline or inland monitoring
locations to represent over-lake deposition is critical to our ability to accurately assess the
atmospheric deposition component of the toxics loading to the Great Lakes. This
question can only be answered by a systematic study of the complex meteorological
conditions which are necessary to parameterize the deposition process. This project is
onlv the first step in the development of a program to study the dynamics of toxic
pollutant transport over and deposition into large lakes. Since historical data collected in
the Great Lakes utilized land-based monitoring locations, this study also provides an initial
comparison of the validity of land-based measurements with those made over-water. Due
to the expense of ship time, the number of days of concurrent over-water and over-land
measurements were limited. The ship was on-station approximately 20 miles west of
Muskegon, MI for an initial 'shake-down' cruise and the ship was on-station 5-10 miles
off-shore from Chicago/Gary for two periods later in the 30 day intensive. The mixed-
layer trajectories associated with these sampling days were given in Figure 5-6.
2-14
-------
2.4.1. Individual Cruise Descriptions
2.4.1.1. On-Station Off-Shore from Muskegon, MI 10-11 July 1991
During the sampling period off-shore from Muskegon, MI predominant easterly
flow advected pollutants from several sources in the Muskegon and Grand Rapids, MI
area to the lake. Elevated concentrations of Se, Zn and low molecular weight PCBs were
measured, in addition to one highly elevated sample with As. The concentrations of S042"
, NH3 and HNO3 were comparable on the R/V Laurentian and at South Haven. The
concentration of retene observed on the R/V Laurentian during this cruise was elevated
above the level observed simultaneously at South Haven, potentially indicating impact
from a combustion source utilizing wood or other vegetative material upwind. Levels of
coronene observed over-water during this period were similar to the observed
measurements in South Haven, indicating similar impact due to motor vehicle emissions or
other sources.
Average concentrations of several pesticides measured during this cruise were
similar on the R/V Laurentian and at South Haven (a-HCH, trans-nonachlor, Mirex,
Chlordane, AJdrin, Metolachlor, y-Chlordane and a-Chlordane), with a notable elevation
in Laurentian levels observed on 12 July. Concentrations of DDT and its breakdown
products as well as Dieldrin and Chlorpyrifos were highly elevated in South Haven.
Atrazine was the only pesticide measured which was elevated in concentration on the
Laurentian (11 July 93). Low molecular weight PCBs (total mono-PCB and 2-PCB)
were also measured in elevated levels on the Laurentian on 11 July as well as in South
Haven.
Analysis of samples collected for organic compounds were not analyzed from
Chicago (IIT) or Kankakee during the first cruise since funds available for this very
expensive analysis were limited. The prevailing transport pattern during the period
indicated that this data would be of less value than for different time periods during the
study.
2.4.1.2. On-Station Off-Shore from Chicago/Gary, 23-27 July 1991
For the first sampling period off-shore from Chicago/Gary, levels of the trace
elements measured aboard the R/V Laurentian were similar to or slightly lower than those
levels monitored at Chicago (IIT), largely due to the prevailing northerly flow that the
vessel received while on station. Since there are no large sources in the lake and the
2-15
-------
upwind fetch from the sampling point in southern Lake Michigan is 100s of kilometers of
open water, the levels of most pollutants would not be expected to be very high.
Exceptions to this were levels of vapor phase and particulate mercury as well as coronene
and naphthalene which were greatly elevated at Chicago (IIT) in comparison to those
concentrations observed on the R/V Laurentian. As a result of westerly winds from the
greater Chicago are, the daytime sample on 24 July aboard the R/V Laurentian revealed
elevated Pb levels, both fine and coarse, as well as elevated Zn .
Concentrations of acidic sulfate aerosols and were comparable at South Haven and
on the R/V Laurentian during this period. Gaseous NH3 concentrations were elevated on
the R/V Laurentian above the levels observed in South Haven especially on 24 July when
mixed layer flow was out of the west-southwest to the ship. Concentrations of NH3 and
other species associated with urban activities were higher during one twelve hour sampling
period.
2.4.1.3. On-Station Off-Shore from Chicago/Gary, 5-7 August 1991
During the second of the two cruises on which the R/V Laurentian was located
off-shore from Chicago/Gary, prevailing flow was from the east-southeast. These
conditions provided direct measurements of the transport of toxic compounds to the lake.
Increased levels of fine Fe and Mn, impacted by the plume from the Gary steel industry
were evident.
The PAH concentrations appear to mimic the behavior of atmospheric Hg, being
about three to ten times higher at Chicago (IIT) than over-water. The mean pesticide
concentrations measured at Chicago (IIT) and aboard the R/V Laurentian are within a
factor of 2 of each other with some being higher at Chicago (IIT) and others being greater
on the R/V Laurentian. Atrazine and Simazine were both found in greater concentrations
over the water than at IIT. The concentrations of pesticides observed in South Haven and
in Kankakee are generally much higher than those observed at either Chicago (IIT) or
over the lake. This is not surprising as both sites are in the middle of large agricultural and
fruit growing areas. Local application of pesticides strongly influences the levels seen at a
site and therefore, prevent a single land-based site from being used to estimate the over-
water levels with any certainty. However, on selected occasions, the concentration of
some pesticides measured on the R/V Laurentian were determined to be greater than the
concentration measured simultaneously at any of the land sites. This occurred on 5
August for the pesticide Atrazine and on 7 August for y-HCH. This finding may be the
2-16
-------
result of special meteorological conditions that allow for the advection of concentrated
pesticide-laden air masses over the water.
Also of interest is a second observation of elevated levels of low molecular weight
PCBs on board the R/V Laurentian, with high concentrations of 2-PCB and total mono-
PCB measured on the 5 and 7 August.
2.4.2. General Conclusions Regarding Over-Water Measurements
For the two periods when the R/V Laurentian was on-station, average fine mass
levels measured at the Chicago site and over-water were comparable. These
measurements suggests that estimates of deposition for many compounds would have been
fairly accurate using data collected at Chicago (IIT). The fairly close agreement in
measured values for most species may appear to indicate that the land-based site would
function as an adequate surrogate for over-water measurements of trace elements.
The major flaw with this assumption is that while the ship was on-station off
Chicago/Gary, two different and fairly uncommon summertime meteorological patterns
were observed: prevailing wind flow was strong and from the northwest during the first
cruise (providing a relatively non-polluted air mass to both the R/V Laurentian and
Chicago (IIT)) and transport to the ship was from the east southeast during the second
cruise (providing a relatively polluted air mass to the R/V Laurentian). The average of
these two events causes the over-water and Chicago based site to appear to experience
quite similar pollutant levels. In order to adequately assess the validity of utilizing land-
based monitors instead of over-water sites a representative sampling of several
meteorological conditions that reflect the complex dynamics of Lake-meteorology would
have to be obtained.
For example, in July 1990 during the preliminary study for the Lake Michigan
Ozone Study, a pollutant plume was observed as it traveled up the shoreline of Michigan
controlled in a tight, relatively non-dispersed plume. The role of these meteorological
phenomena (and others like it which provide for capture of a pollutant plume near the lake
surface as it is advected away from the source) in the deposition of toxic compounds to
the Lake may be highly significant. It is often the case that a few isolated instances with a
particular set of meteorological and source influence characteristics can be responsible for
major pollutant episodes.
2-17
-------
Important exceptions to the general finding of comparable levels of fine mass are
the striking differences in measured concentrations of Hg, several PAHs and pesticides
and noticeable differences in coarse mass which could greatly affect calculations of total
loading to the lake.
Generalizations regarding over-water measurements cannot be made from the
limited data collected in this initial study. The data indicate that over-water measurements
involve a very special set of circumstances that can, on occasion, play an important role in
the deposition of toxic compounds to the water. Further work in this area will help to
determine if taking into account seasonal differences in dominant meteorology and
atmospheric chemistry and transport to land based sites can be used as an adequate
surrogate for over-water deposition calculations on a yearly average basis. If the
uncertainty measured for such estimates turns out to be quite large, the air sampling
community may have to accept the burden of collecting this critical data in the most
accurate manner possible: taking representative samples over-water for limited portions
of each season, or potentially setting up unmanned monitoring stations for some of the
critical pollutants that would at least allow extrapolation to other compounds and would
provide accurate measure for a few of the species of interest.
2-18
-------
Chapter 3
Recommendations
This project was a first step in the study of the dynamics of toxic pollutant transport
over and deposition into large lakes. Our initial effort was aimed at establishing the
feasibility of providing accurate and representative air samples and gradient measurements of
toxic contaminants and nutrients over Lake Michigan. Among the highest priority research
needs identified are: (1) measuring onshore and offshore deposition rates simultaneously to
determine whether or not on-shore measurements represent deposition to the bulk of the lake
surface, and (2) characterizing the impact of local urban plumes on the total toxic burden to
the Great Lakes. Both of these research issues are critical to the design of long-term
monitoring networks to determine rates of atmospheric deposition to the lakes.
This investigation provided preliminary information to assess these questions. Because
of the complexity of the meteorology in the lake shore environment and the multitude of
sources coupled with the limited number of days of over-water measurements, the ability of
this study to adequately compare over-land versus over-water measurements was limited.
Some changes in the design and implementation of this study are proposed in attempt to
optimize similar research attempts made in the future. In addition, future research needed to
answer the original study goals is outlined. During the course of this investigation, additional
objectives were identified and are described below.
3.1. Proposed Changes in the Study Design
Changes in the present study which would enable a more complete description of the
airshed investigated include an improved location for an upwind location, improvements in
analysis, and most importantly, additional and flexible periods aboard the research vessel.
In light of the observation that Kankakee was often subjected to several, varied
sources in the Illinois area, an alternative 'upwind' control site should be selected in the region.
3-1
-------
Alternatively, if the source characterization for the Illinois/Indiana region is better described,
specific source influences on the upwind sites may be subtracted from the 'background' input
to the urban measurements.
In order to obtain accurate and reliable measurements of all species being measured,
sampler deployment, cleaning, analysis should be pre-tested and field blanks collected to
confirm the ability to obtain accurate quantification of the levels required. In addition, the
first samples collected at each site should be analyzed and results reported immediately to the
project officer and principal investigator in order to confirm that samples are being collected
and analyzed properly. Implementation of this procedure could have potentially avoided the
loss of the majority of the VOC data collected for this study.
Additional and more flexible time aboard the research vessel is imperative for
optimizing the sampling performed over-water during selected meteorological conditions.
This flexibility is essential to our ability to assess the question of over-water versus over-land
measurements.
Additional measurements such as analysis of both particulate and vapor phase SOCs
including pesticides, PCBs and PAHs, as well as other urban tracers, e.g. CO, at each of the
sampling sites would have helped in the modeling portion of this project. An additional site
on the west-side of Lake Michigan and additional measurement sites between IIT and Gary,
IN would allow for a better estimation of the total deposition to the southern portion of the
basin.
3.2. Future Research Needed to Meet the Original Objectives
The data also indicate that for some pollutants a high concentration and strong
gradient exist over the water downwL.d of an urban area, which points out the importance of
future investigations to include urban site monitoring in all proposed studies of the transport
and deposition of hazardous air pollutants to the lakes.
In order to accurately quantify the urban input of toxic contaminants to Lake Michigan
at least one additional monitoring location in Chicago/Gary region is necessary. This site is
critical to accurately estimating the loading to the lake during periods of south and southeast
flow. A measurement site closer to the industrial sources would allow us to better
characterize the size range of particulate matter emitted from the iron-steel coke ovens, oil
refining and other sources in the vicinity. If, as it is suspected, the large particle mass is an
3-2
-------
extremely important contribution to deposition in the southern portion of Lake Michigan, a
monitoring location in the heart of the industrial processes would greatly improve estimates of
loading to the lake.
To improve calculations of deposition and attempts at deposition modeling, both
theoretical and experimental research is needed to improve direct measurements to the water
as well as improvement of computational techniques to determine the impact of dry deposition
of atmospheric contaminants to water bodies. It is intrinsically difficult to compare the
estimated deposition fluxes from this work with those obtained by others based upon
measurements made with surrogate surfaces, or with estimates obtained using limited point
measurements at a few land-based sites, which may not represent the real physical situation at
the air-water interface during the transfer of particles from the atmosphere to the water
surface and vice versa.
3.3. Future Research Needed to Meet Additional Objectives
Additional Objectives
1. Developing an accurate emissions inventory for the compounds of highest priority.
2. Obtain a better source emissions library for modeling by performing stack testing
and source sampling of current sources including motor vehicles.
The urgent need for accurate information regarding source emission profiles for each
of the major source types in the Great Lakes Basin cannot be overemphasized. In the absence
of such information, models of transport, chemistry and deposition provide only crude
approximations of the total load and potential impact of the release of hazardous atmospheric
pollutants. In addition, the ability of legislators to formulate and implement appropriate
control regulations is highly dependent on identification of specific sources and the impact of
these emissions.
3. Perform research to improve parameterizations of the depositional processes that
can be utilized in both determinisitic models and hybrid receptor-deposition
models as described in this report.
This study was designed to serve as an intensive initial investigation of the importance
of atmospheric transport and deposition of hazardous air pollutants to Lake Michigan. The
data generated provides a unique resource for the design of future work in the Great Lakes.
3-3
-------
4. Characterize the large particle (> 10 fim) contribution to toxic deposition,
particulary in and near urban source areas.
5. Further evaluate the importance of over-water vs inland vs shoreline monitoring
locations.
6. Develop re-emission estimates for Hg, PCBs, and pesticides that have been banned
or discontinued in the U.S.
7. Need to better understand the gas-particle partitioning of semi-volatile compounds
between the source and the receptor site.
3-4
-------
Chapter 4
Methods
4.1. Meteorological Measurements
4.1.1. Tower Platform
Meteorological measurements were taken from a tower located 10 feet off the bow
of the ship. Figure 4.1 shows the design of the bow tower used in the study. The tower
assembly consisted of two main components: a retractable arm and a five-meter aluminum
tower. The retractable arm was designed to allow the tower to be pulled back onto the
ship's deck when instrument maintenance was necessary. The arm was attached to the
tower via a hinge, which allowed the tower to be tipped back before being slid back onto
the ship's deck.
The tower itself was triangular in shape, with each side approximately 6 inches in
length. When deployed, the tower base was two meters above the lake surface (during
relatively calm conditions). The tower was guyed to the ship using three steel cables.
Two cables were attached to the top of the tower, and each was attached to a side of the
ship. The third guy-wire ran from the bottom of the tower to the ship's bow. Once in
place, the guy-wires successfully kept the position of the tower fixed with respect to the
ship, so there was not any significant twisting of the tower about either the vertical or
horizontal axis. However, given that the tower was rigidly attached to the ship, significant
rolling of the tower was noted due to the movement of the ship in the lake's waves.
Motion sensors were to be used to assist in correcting for the ship's motion, however the
sensors were not available at the time of the study. As a result, other techniques were
used to remove the effect of the ship's motion from the data. These methods will be
discussed in the following section. The instruments that were placed on the tower are
described in detail below.
4-1
-------
4.1.2. Slow-Response Wind Measurements
Three R.M. Young wind vanes (Model 05701) were used in combination with a
sonic anemometer to measure the wind field over the lake surface. The wind vanes have a
specified working range of 0 to 40 meters/sec, a propeller threshold of 0.2 meter/sec and a
distance constant of 1.0 meter. The wind vane has a distance constant of 1.2 meters.
These instruments are widely used in many air quality measurement programs. As
indicated in Figure 4-1, the vanes were placed on the bottom half of the bow tower. The
vanes were spaced logarithmically (in combination with the sonic anemometer) so as to
test the validity of the "log wind profile" law in the atmospheric surface layer over the
lake. The lowest vane was placed such that its average height above the lake surface was
2.5 meters. The other wind vanes were then placed 3.0 and 4.0 meters above the lake
surface, respectively. Wind data from these instruments were sampled at 1 Hz.
4.1.3. Fast-Response Wind Measurements
Fast-response wind measurements were made using an Applied Technologies'
SWS-311/3K three-axis sonic anemometer. The ATI sonic has a horizontal measurement
range of + 20 meters/sec and a vertical measurement range of + 5 meters/sec. The
instrument has a specified accuracy of ± 0.05 m/sec for wind speed and ±0.1 degree for
wind direction. The sonic anemometer was placed on the tower so that it had an average
height of 5.5 meters above the lake surface. This instrument was sampled at a frequency
of 10 Hz and was recorded at this same frequency using LabTech Notebook software.
4.1.4. Temperature and Humidity Measurements
Temperature and humidity measurements were made using two Rotronic
Instrument Corporation MP-1 OOF Temperature/Relative Humidity probes. Each probe
was housed in a radiation shield provided by the same manufacturer. The instrument
ranges for temperature and relative humidity are -30 to + 70 °C and 0 to 100%,
respectively. The specified accuracies are + 0.2 °C and + 2.0 %, respectively. The time
constant for both sensors is listed as 10 seconds. One Temp./RH probe was placed at the
base of the tower ( 2 meters above the lake surface), while the second probe was placed at
the top of the tower (7 meters above the lake surface). The purpose of this placement was
to allow for a crude measurement of the atmospheric surface layer temperature gradient.
4-2
-------
Figure 4-1. Bow Tower Assembly on the Research Vessel Laurentian.
R/V Laurentian
Bow Tower Assembly
Temp. Probe
-f 3 meters from bov
Laurentian
7.0 m.a.s.
C02/H20v
i n
ATI Sonic
5.5 m.a.s.
¦—
3.5 m.a.s.
1—
2.5 m.a.s.
2.0 m.a.s.
l-l
Temp. Probe
4-3
-------
4.1.5. Carbon Dioxide and Water Vapor Measurements
The carbon dioxide and water vapor measurements were acquired using an
instrument borrowed from the National Oceanic and Atmospheric Administration's
Atmospheric Turbulence and Diffusion Division (NOAA-ATDD), Oak Ridge, TN. Since
wave-induced vertical motion rendered the vertical wind data from the ATI Sonic
unusable, we did not have such measurements to correlate with measured fluctuations in
the carbon dioxide and water vapor. For this reason, the results from the measurement of
these two variables are not presented.
4.1.6. Data Collection
4.1.6.1. Slow-Response Instruments
A Campbell Scientific 2IX Datalogger was used to collect the data from the "slow
response" instruments (temperature/relative humidity probes and wind vanes). Data were
sampled at a frequency of 1 Hz, then stored as one-minute averages. The information
stored includes.
- temperature and relative humidity at the 2.0 and 7.0 meter levels, and
- wind speed, wind direction and sigma-theta (standard deviation of the
wind direction) at the 2.0, 2.5 and 3.5 meter levels.
This information was later converted to one-hour averages, and can be found in the
Appendix.
4.1.6.2. Fast-Response Instruments
The data from the "fast response" instruments (ATI sonic anemon ~ter and the
NOAA-ATDD's CO2/H2O vapor sensor) were collected using a personal computer and a
commercial data acquisition software package (LabTech Notebook). Data were sampled
and at a frequency of 10 Hz. Data smoothing or filtering was performed during post-
processing.
4.2. Chemical Measurements
Table 4-1 summarizes the pollutants measured, sampling methods and number of
species of each pollutant quantified. Because of the relatively high costs for mass analysis
4-4
-------
of semi-volatile organic compounds (pesticides, PCBs and PAHs), it was decided in
advance to analyze only a subset of samples collected for these compounds. The PS-1
samples collected each day were shipped in cold packs to the appropriate laboratory.
Filters and traps were extracted together for each sample and were placed in cold storage.
A decision regarding which samples to analyze was made after an examination of the
particulate, trace element, and meteorological data.
Table 4-1. Pollutant Measurements Made during the LMUATS.
POLLUTANT CLASS
SAMPLER
NO. SPECIES
Pesticides
PS1/PUF
10
Total PCBs
PS1/PUF
20
PAHs
PS1/XAD
19
VOCs
Canister
44
Trace Elements
Dichot
18
Carbon
FPS
2
Gaseous Hg
Au Sand
1
Particulate Hg
GFF
1
Other Inorganics
ADS
8
4.3. Sampling Schedule and Operations
The EPA AREAL labs provided the majority of equipment utilized in this
investigation. In addition, support staff from EPA AREAL set up sampling equipment at
each of the sites, trained operators and performed field audits.
4.3.1. Schedule
The field-sampling portion of the study commenced on July 8 at all of the land-
based sites and continued until 9 August 1991. Integrated samplers were operated each
day from 8 AM CDT to 8 PM CDT at each site. Nighttime samples were also routinely
taken at the South Haven site from 8PM - 8AM throughout the study. Mercury sampling
and annular denuder samplers were operated more intensely during the 12 days that the
R/V Laurentian was on-station off-shore of Chicago. Two six-hour samples were taken
during the 12-hour daytime period from 8AM - 8PM CDT. A more intensive Hg sampling
4-5
-------
campaign was conducted at the IIT site while the annular denuders were only operated
aboard the R/V Laurentian and in South Haven.
4.3.2. Sampling Procedures
The sampling methods utilized in this study are detailed in the Sampling and
Quality Assurance Project Plan as well as in the Final Report submitted by Battelle entitled
"Analytical Support for Lake Michigan Urban Air Toxics Study". These documents
should be referred to for specific details of the sampling and analysis portions of this
study.
4.3.3. Sampling Locations
The sampling locations utilized in the study are shown in Figure 4-2. These sites
were selected after careful consideration of the data collected during the 1990 LMOS
Pilot Study in which the University of Michigan participated. The limited data from the
Pilot Study suggested that three sites strategically placed on-shore and at least one
sampling platform off-shore would allow an initial investigation of the importance of the
urban areas on deposition to Lake Michigan.
4.3.3. J. Over-water Measurement of A tmospheric Contaminants
The University of Michigan Research Vessel (R/V) Laurentian was operated for
the LMUATS for three different sampling periods during the one-month study. The ship
was positioned in two areas during the study. The sampling inlets were roughly 5 feet
above the deck but off the side of the bow of the ship (which stood an additional 20 feet
above the surface of the water). Sample collection took place only when the vessel was
anchored which kept the bow pointing into the prevailing wind at all times.
The first cruise was a test period to make sure that all of the sampling equipment
could be deployed safely from the vessel. This cruise was a two-day voyage that brought
the research vessel approximately 20 miles off-shore of Muskegon, MI. The second cruise
was a five-day sampling campaign that started in Grand Haven, MI with sampling at
locations from 4-6 km off-shore of the Chicago/Gary area. The cruise down to Chicago
took 9.5 hours on the R/V Laurentian and sampling occurred from 23-27 July. The last
cruise occurred from 5-8 August and the sampling occurred in the same general location
4-6
-------
Figure 4-2. Sampling Locations for the Lake Michigan Urban Air Toxics
Study.
4-7
-------
as during the previous cruise. Extremely violent weather on 8 August forced the research
vessel back into port.
4.3.3.2. Land-Based Sampling Locations
Three land-based sites were operated from 8 July through 9 August 1991. The
sites were chosen to provide one upwind sampling point, one urban site, and one site
downwind on the eastern shore of Lake Michigan. The upwind site chosen for the study
was located at the Kankakee Regional Airport in rural Illinois (Figure 4-2). This site was
chosen as the "upwind site" to provide the levels of atmospheric contaminants transported
into the Chicago/Gary Urban areas with southerly or southwesterly transport. The
Kankakee Regional Airport is in an agricultural area and during the period under study,
the site was primarily surrounded by corn crops. The urban site was located on the
campus of the Illinois Institute of Technology (IIT). This site had been utilized previously
for air pollution studies and provided a central Chicago location only 1.6 km from the lake
shore. The IIT site was atop a campus building which was the same height or higher than
buildings in the adjacent vicinity. The downwind site was located in South Haven, MI,
which was one of the LMOS Pilot Study Sites. The South Haven site was located in a
rural area approximately 5 miles from the lake. The South Haven area is largely
agricultural with fruit trees and vegetable produce dominating the local terrain. This site
was ideal for the intensive land-based operations that were necessary for the study. The
South Haven Site was the largest of the sites and duplicate sampling for all species was
performed there. The inlet to the samplers was approximately 7 feet above the ground in
an open field. Kankakee and South Haven were also LMOS sites which provided a
complete suite of meteorological and some auxiliary air quality data to be used. Auxiliary
data were also available at the IIT site. Continuous monitoring equipment were operated
at the sit": providing hourly ozone (03), NOx, and meteorological data.
4.3.4. Samplers
The air monitoring instrumentation employed for this project was well-tested
sampling equipment that provided the high-quality data necessary to meet the project
objectives. The EPA AREAL labs provided all the necessary instrumentation with the
exception of mercury sampling supplies which were provided by the University of
Michigan. The sampling equipment employed allowed for the quantification of most of
the 14 Critical Pollutants on the IJC list, including Hg. In addition, various airborne
nutrients, additional trace elements by Neutron Activation Analysis (INAA), a complete
4-8
-------
set of size-distribution particulate data for selected species over the lake, and size-
distribution measurements using scanning electron microscopy (SEM) on specially
collected substrates were also collected. A list of the samplers that were employed at each
location are given in Table 4-2. A duplicate sampler of each type was run at the Michigan
land site for the period of the study. The particle size information obtained from the
micro-orifice impactors and the SEM analysis, as well as the Hg determinations were
obtained for a subset of the sampling days on the R/V Laurentian and one shore line site.
Table 4-2. Samplers Deployed for the LMUATS.
Sampler
Samplers
Per Site
Sample
Duration
(hours)
Sampling Media
Analysis Method
PS1-PUF Sampler
2
12
PUF/Quartz
GC/MS
FPS
1
12
Filter
Carbon
Annular Denuder/FP
1
12
47 mm Quartz
IC
DICHOT
1
12
NaC03/Teflon
XRF/INAA/SEM
VOC Canister System
1
12
37 mm Teflon
GC/MS
Hg Systems
6-L Canisters
1. Au Sand/Filter
1
6/12
CVAFS
2. Filter w/INAA
1
12/24
Au /Quartz
INAA
3. Filter
1
12/24
47 mm Teflon
CVAFS
Micro-Orifice Impactor*
1
24
47 mm Quartz
37 mm Teflon
IC
* Micro-orifice impactors were of
y rim on the R V Laurentian and at the South
Haven site for a limited number of days.
4.4. Sample Analysis
Battelle Laboratory in Columbus, Ohio and Southwest Research Institute in San
Antonio, TX were responsible for the analyses of the organic compounds including VOCs,
PCBs, PAHs, and pesticides. Sunset Laboratory in Portland, OR performed all sample
preparation and analyses for organic and elemental carbon. EPA AREAL performed the
trace element determinations by X-Ray Florescence (XRF). Pollutant species measured
are given in Table 4-3.
4-9
-------
Table 4-3. Pollutant Species Measured during the LMUATS.
Inorganic
Species
PAH
Pesticides
and PCBs
VOCs
A1
As
Br
Ca
CI
Cr
Cu
Fe
Hg
K
La
Mg
Mn
Mo
Na
Ni
Pb
S
Sb
Se
Si
Sm
Ti
V
Zn
so42-
N03-
no2
NH^
H+
S02
HNO3
HONO
NH,
naphthalene
acenapthvlene
acenapthene
fluorene
phenanthrene
anthracene
fluorenone
retene
fluoranthene
pyrene
cyclopenta|c.d]pvrcne
benz( a]anthracene
chrvsene
benzofluoranthenes
benzo[e] pyrene
benzo [a] pyrene
indeno[ 1,2.3-c,d]pyrene
benzo[g,h, i]pervlene
coronene
monochlorobiphenvls
a-hexachlorocyclohexane
dichlorobiphenvls
hexachlorobenzene
atrazine
i-hexachlorocyclohexane
trichlorobiphenyls
tetrachlorobiphenvls
alachlor
mirex
aldrin
metolachlor
trans-Nonachlor
pentachlorobiphenyls
dieldrin
hexachlorobiphenyls
heptachlorobiphenyls
4,4'-DDT
octachlorobiphenvls
nonachlorobiphenvls
decachlorobiphenvls
trichlorofluoromethane (Freon-11)
dichlorodifluoromethane (Freon-12)
1,1,2-trichloro-1.2,2-tnfluorethane
(Freon-113)
1,2-dichloro-1,1,2,2-tetrafluorethane
(Freon-114)
methyl chloride
vinyl chloride
methyl bromide
ethyl chloride
1,1-dichloroethene
dichloromethane
3-chloropropene
1.1-dichloroethane
cis-1,2-dichloroethene
trichloromethane
1.2-dichloroethane
1,1,1 -trichloroethane
benzene
carbon tetrachloride
1,2-dichloropropane
trichloroethene
cis-1,3-dichloropropene
trans-1,3-dichloropropene
1,1,2-trichloroethane
toluene
1,2-dibromoethane
tetrachloroethene
chlorobenzene
ethvlbenzene
m&p-xvlene
stvrene
1,1,2,2-tetrachloroethane
o-xvlene
4-ethyl toluene
1,3,5 -trimethvlbenzene
1,2,4-trimethvlbenzene
benzyl chloride
m-dichlorobenzene
p-dichlorobenzene
o-dichlorobenzene
1,2,4-trichlorobenzene
hexachlorobutadiene
4-10
-------
The inorganic analyses and mercury analyses were primarily performed at the
University of Michigan with the assistance of Brooks Rand Ltd. in Seattle, WA. Ultra-
clean room facilities operated at the University of Michigan were utilized for preparation
and analyses of all mercury samples. The University of Michigan Ultra-Clean room facility
is a Class 100 laboratory designed to minimize any possible trace element contamination.
In addition, Neutron Activation Analysis was performed on a subset of the
dichotomous sampler filters. The INAA provided better sensitivity for many of the
important elements needed for receptor modeling. By providing more accurate and
sensitive determination for some critical elemental species, INAA was also performed on a
subset of filters for total particulate Hg determinations. The INAA was performed under
the direction of Dr. Ilhan Olmez at the MIT Nuclear Reactor Laboratory.
4.4.1. PAH Sampling and Analysis
The PS-1 samplers (General Metal Works, Cleves, OH) located at each sampling
site were used to collect both particle-bound and vapor-phase PAH onto a quartz fiber
filter (104 mm QAST, Pallflex, Putnam, CT) and XAD-2 (Supelco, Bellefonte, PA),
respectively. The clean filters and XAD-2 traps were prepared at Battelle and sent to each
sampling site. A standard operating procedure for loading, operation, and unloading of
PS-1 samplers was prepared for the field sampling teams by Battelle lab. In addition, at
the beginning of the field sampling campaign, an experienced Battelle technician went to
the South Haven site and demonstrated the proper sample handling procedure to minimize
any possible field contamination and to ensure the integrity of the collected samples.
PAHs were collected at a nominal flow rate of 4 cfm and the collected samples
were stored in the dark at 0°C before they were sent back to Battelle for analysis. Sample
tracking forms containing all sample collection information were filled out by the field
operators for each set of filter and XAD-2 samples and sent to Battelle with each sample.
The filter and corresponding XAD-2 trap were combined and extracted with
dichloromethane (DCM). The DCM extract was concentrated by Kuderna-Danish (K-D)
evaporation and analyzed by GC/MS in electron impact (EI) mode to determine target
PAHs. A Finnigan TSQ-45 GC/MS/MX operated in GC/MS mode was employed. Data
acquisition and processing were controlled by an INCOS 2300 data system. The MS was
operated in the selected ion monitoring (SIM) mode. Peaks monitored were the molecular
ions and characteristic fragment ions of the target analytes. The GC column was a DB5
4-11
-------
fused silica capillary column (30 m x 0.25 mm; 0.25 jim film thickness, Supelco). The GC
temperature was held at 70°C for 2 minutes, and then programmed to 290°C at 8°C/min.
Identification of target analytes was based on correct molecular ions, correct
fragmentation ions, and their GC retention times relative to that of the corresponding
internal standards (phenanthrene-d10 and/or 9-phenylanthracene). Quantification of target
analytes was based on the comparison of the respective integrated ion current responses of
target ions to that of the corresponding internal standard, with average response factors
generated from analyses of standard solutions. Target PAHs quantified are listed in Table
4-3.
4.4.2. Pesticide and PCB Sampling and Analysis
Polyurethane foam and quartz fiber filters for collection of gaseous and particulate
PCB's and pesticides were prepared by Southwest Research Institute and shipped to the
South Haven site for distribution to the other sampling locations. Combined PUF and
filter samples were received from the sampling sites and were analyzed by Southwest
Research Institute by GC/MS. Upon receipt, samples were Soxhlet extracted (PUF plug
with the corresponding quartz filter) with DCM. Analysis of the extract was performed
on a Fisons VG Autospec high-resolution mass spectrometer equipped with a Hewlett-
Packard 5890 gas chromatograph and CTC-A auto sampler. Since a mixture of heat-
sensitive pesticides were analyzed in conjunction with high-boiling PCBs, injector
temperature, inertness, and injection technique were optimized to result in the least
amount of thermal degradation of the target pesticides.
Target pesticides and PCBs quantified are listed in Table 4-3. Target compounds
were quantified using internal standards and a five point calibration curve. In order to
provide an indication of extraction efficiency, 3,3',4,4'-tetrachlorobiphenyl-13C12 was
introduced into each sample prior to extraction.
4.4.3. Elemental and Organic Carbon Sampling and Analysis
Quartz fiber filters, treated prior to sampling to remove all organic and elemental
carbon traces, were shipped in two batches to the South Haven site from which they were
distributed to the other sites. After sampling, filters were returned to aluminum foil-lined
petri dishes which were sealed with Teflon tape. At the conclusion of the study, all
samples and blanks were shipped to Sunset Labs for analysis. Fine particle samples were
analyzed using combustion flame ionization detection (FID) to measure total elemental
4-12
-------
and volatilizable carbon content. This information is useful for determining source
apportionment.
4.4.4. VOC Sampling and Analysis
VOC sampling canisters were cleaned by U.S. EPA and sent to the field for
sampling. A total of 129 canister samples were collected in the field and sent to Battelle
for analysis of VOC. Standards were analyzed for each of the target compounds of
interest in the LMUATS. The compounds analyzed are listed in Table 4-3.
Samples were analyzed using an automated GC system equipped with a Supelco
two-phase adsorbent trap which utilized a bed of Carbopack B and Carbosieve S-III
adsorbent. The system incorporates a Hewlett-Packard 5880 gas chromatograph and
parallel flame ionization and mass spectrometric detectors. Target compounds were
chromatographically resolved using a 50m-by-0.32mm internal diameter, OV-1 fused silica
column. The column exit flow was split to direct one third of the flow to the mass
spectrometric detector with the remaining flow passing through the flame ionization
detector. The mass spectrometer was operating in the selective ion monitoring (SIM)
mode. In this mode, the mass spectrometer monitored the characteristic ions of each
target VOC, rather than scanning all masses continuously between two mass limits. The
advantage of monitoring only pre-selected ions was increased sensitivity and improved
quantitative analysis. A detection limit of 0.10 ppbv was obtained in this mode.
4.4.5. Atmospheric Acidity Measurements
A primary goal of the study was to quantify the levels of toxic air pollutants in the
southern Lake Michigan Basin in order to determine how much of the airborne pollutants
are being deposited to aquatic and terrestrial ecosystems. Measurements of atmospheric
acidity, gaseous and aerosol strong acidity (H+) were performed to characterize the
chemical composition of the atmosphere and to investigate the behavior of the regional
and urban plumes advecting across Lake Michigan. The acidity of the aerosols was
hypothesized to be related to the form of the toxic trace elements present on aerosols that
are transported long distances.
Annular denuder/filter pack systems (ADS) were operated at two of the four sites
during the LMU ATS including South Haven, MI and aboard the University of Michigan's
research vessel (R/V) Laurentian.
4-13
-------
Acidic aerosol and gas measurements were taken 2 to 3 times a day at South
Haven throughout the duration of the study while the size fractionated samples were
collected on selected days. While the R/V Laurentian was docked, two 12-hour samples
were collected in South Haven from 8AM-8PM and 8PM-8AM CDT. While the ship was
on station, 3 samples were taken daily at South Haven and on the R/V Laurentian: 8AM-
2PM, 2PM-8PM, 8PM-8AM CDT. All size fractionated samples collected at South
Haven and on the R/V Laurentian were 24-hours in duration starting at 8AM CDT. ADS
samples were also collected in Ann Arbor as part of an ongoing study of atmospheric
acidity in Michigan.
The annular denuder sampling system was used for collection of acidic aerosols
and gases and has been described previously (Koutrakis, et al. 1988, Keeler et al., 1990a).
The ADS was utilized to quantify gaseous S02, HN03, HONO, NH3, and fine fraction
(<2.5 |im) particulate species S042", N03", NH4+, and aerosol strong acidity (H+).
Additionally, the system removes the gaseous ammonia and protects the collected
particulate matter from possible neutralization.
Size-fractionated samples were collected at South Haven and aboard the R/V
Laurentian using a six-stage micro-orifice type impactor (Keeler et al., 1990a). The six
stages have been characterized to separate atmospheric aerosols into the following size
ranges when operated at 30 LPM: #1: > 5 |im; #2: 5-2.5 fim #3: 2.5-l(im; #4:1-0.6 |im;
#5: 0.6-0.18 |am and; #6: <0.18 (im (Marple and Rubow, 1984). The system is designed
to operate with a minimal pressure drop so that vaporization of water and subsequent
alteration of the aerodynamic diameter of the particles being collected is avoided (Biswas
et al., 1987). The impactors were placed in a stand which forced incoming air to pass
through 8-citric acid-coated honeycomb-style aluminum denuders to remove ambient
ammonia (Koutrakis et al., 1988). The aerosol material was collected onto Teflon filters
(Teflo) and analyzed identically to the Teflon filters from the ADS.
Sulfate collected on ADS, MOI and on dichotomous filters (analyzed by XRF)
were compared to assess the precision of the three techniques. Regression analysis of
XRF S against ADS S042" shows quite good results with a slope of 0.33 ng/'m3 and a
correlation coefficient of 0.994. Likewise, the regression of ADS S042" versus MOI S042"
(fine fraction, stages 3, 4, 5 and 6) displays a slope of 0.93 and a correlation coefficient of
0.986. These results indicate that the collection and analytical techniques utilized were
comparable and precise.
4-14
-------
4.4.6. Scanning Electron Microscopy Analysis of Individual Particles
Six aerosol samples collected at the IIT were analyzed by Scanning Electron
Microscopy and Energy-Dispersive X-ray spectroscopy (SEM/EDX). The objective of
the SEM/EDX analyses was to characterize a representative number of individual aerosol
particles for each sample in order to provide additional information about particulate
sources impacting the IIT site. SEM/EDX analyses revealed significant differences in
aerosol composition among the six samples. These differences indicated that various
sources may have impacted the IIT site simultaneously. The observed differences are also
supported by results obtained using X-ray fluorescence (XRF) and receptor modeling
applied to the same sample set.
The six samples selected for SEM analysis were collected on 18 July, 19 July, 21
July, 2 August, 6 August, and 8 August 1991. The "bulk" elemental composition
(elements heavier than Mg) of these samples had previously been measured by XRF. All
samples were coarse fraction samples (2.5-10 (am aerodynamic dynameter) collected on
Teflon filters using a dichotomous sampler with an upper size cutoff of 10 (am. A feature
of this sampling device is that about 10% of all fine particles (< 2.5 ^m) also deposit on
the coarse filter, thus enabling both size fractions to be analyzed on the same filter. An
effort was made in the present study to characterize a minimum of 100 particles in each
size fraction for each sample. This is considered the minimum number required in order to
distinguish real differences in aerosol composition among the samples.
Each sample was analyzed using the following procedure: (1) after scanning the
sample at low magnification to confirm uniformity of loading and to avoid regions of
anomalous loadings, several fields of view were randomly selected for analysis; (2) for a
coarse or fine fraction particle analysis, all particles within a field with geometric diameter
>2.5 (im or < 1.5 |im, respectively are characterized; (3) for each particle the size,
morphology, elemental content and category are recorded in a data book. Elemental
composition is qualitatively determined by collecting an X-ray spectrum of the particle,
while the category is determined by the microscopist based on the particle's morphology
and composition; (4) micrographs are recorded for particles of special interest; (5) at the
conclusion of the analysis, particles of a given size fraction are sorted into a size/category
matrix as shown in the particle data tables included in Appendix A. It should be noted that
the EDX system is only capable of detecting major and minor elements in a particle, i.e.
elemental mass exceeding a few tenths of a percent of the total particle mass.
4-15
-------
For all of the samples analyzed, sulfate particles account for nearly all of the
particles in the fine fraction and are typically 0.3 (im to 0.4 (am in diameter. These
particles were not included in the 100 particles analyzed in the fine fraction. Instead, an
estimate of the sulfate concentration in each sample was made by obtaining an average
sulfate particle count from several micrographs of the sample at high magnification.
However, XRF analyses provided a better estimate of sulfate concentration.
4.5. Mercury Sampling and Analysis Methods
4.5.1. Vapor Phase Collection
A total of 147 vapor phase samples were taken at three sites during the period
from 8 July - 9 August 1991. Samples were only collected at the Barden Farm near South
Haven, Michigan (SHA), on the Research Vessel Laurentian (LAU) and at the Illinois
Institute of Technology (IIT) in Chicago, Illinois. Atmospheric vapor-phase mercury was
captured on gold-coated sand traps with pre-fired glass fiber pre-filters in Teflon filter
packs to exclude particles. Gold-coated sand traps supplied by Brooks Rand, Ltd. contain
approximately 0.6 cm3 of sand which has a coating of gold 1-2 atoms thick. The gold-
coated sand is supplied in a quartz tube 10 cm in length held in place by a quartz frit on
one end and quartz wool on the other end. Gold sand traps were connected to sampling
equipment by Teflon friction fit connectors. Clean techniques necessary for accurate
determination of low levels of mercury were employed in all procedures of sample
handling and custody.
Vacuum pumps at SHA and LAU utilized mass flow controlling units (Tylan Co.),
and at IIT flow was controlled by a needle valve and diaphragm pumping system set to
maintain flow below 0.3 LPM (0.27 at IIT, 0.29 at LAU and ^.28 at SHA). A minimum
of one blank was taken for every six samples. Thirty eight twelve-hour (8AM-8PM,
CDT) samples and nine blanks were collected at SHA (Table 4-4). For each day the
LAU was on-station, two daily six-hour samples (8AM-2PM, 2PM-8PM) and one
overnight 12 sample (8PM-8AM) were collected for a total of twenty five samples and six
blanks. At IIT, 58 samples and 10 blanks were collected on a more intensive schedule
than other sites. Two daily six-hour and one overnight 12-hour sample (8PM-8AM) were
taken when the LAU was on-station and one daily 12-hour sample was collected while the
LAU was in port.
4-16
-------
Table 4-4. Vapor Phase Mercury Samples Collected during the LMUATS.
Type of Sample IIT R/V Laurentian South Haven
Daytime Samples 46 18 38
Nighttime Samples 12 7 1
Blanks 10 6 9
4.5.2. Vapor Phase Analytical Procedure
After collection, samples were stored with end-plugs Teflon-taped inside individual
polyethylene tubes and then triple bagged in polyethylene zip-lock bags. Samples were
stored outdoors in waterproof containers and shipped at least every five days to Brooks
Rand, Ltd. for analysis. Elemental mercury levels were determined by thermal desorption
(at 400°C) using the dual amalgamation technique described by Bloom and Fitzgerald
(1988) followed by cold vapor atomic fluorescence spectrometry (CVAFS). Peak area
was used to quantify the amount of mercury detected. Standards were analyzed prior to
sample analysis and controls were analyzed in-between sample analyses. Results were
reported in nanograms Hg/trap and subsequently converted to ng Hg/m3. Values reported
are at ambient conditions and have not been converted to standard temperature and
pressure.
Analytical precision for replicate mercury standards analyzed before, during, and
after analysis of vapor phase mercury samples at Brooks Rand, Ltd. varied from 0.07-17%
with an a^rage precision of 6% (Table 4-5). A total of 50 standards were analyzed
during analysis of the 147 samples collected during LMUATS. These standards were
produced by purging a known amount of mercury from solution after reduction with
stannous chloride. Typically 1 and 2 ng standards were analyzed. However, on one
occasion standards of 0.5 to 10 ng Hg were analyzed.
The reproducibility of purged standards is strongly influenced by technician
handling and analytical set-up. Consequently, the 6% precision is likely to be more an
assessment of the ability to reproduce standards than a measure of analytical precision of
vapor phase samples. There are fewer places for error during analysis of a vapor phase
4-17
-------
sample than there are in the preparation of purged standards. The precision with which
the analyzer can reproduce response based on the same Hg input is much better than 6%.
Table 4-5. Precision between Replicate Hg Standards.
Number of Standards
50
Average % Precision for Standards
Median % Precision
6
7
Std Dev. of % Precision
7
Min. % Precision
0.07
Max. % Precision
17
The detection limit for CVAFS at an instrument gain setting of 90 is less than 1
picogram. During the course of this study the operational detection limit was 45.7 pg.
The operational detection limit was defined as 3 x the standard deviation of all field blanks
collected at the three sampling sites. Average field blank values by site are given in Table
4-6. The average field blank at IIT is strikingly lower than that at the other sites. This is
most likely due to operator error in following the protocol for field blanks. Field blanks at
SHA and on the LAU were prepared by removing the end plugs from the gold sand trap,
placing a pre-filter on the trap and putting the filter and gold trap in the sampling box for 2
minutes. At IIT it is likely that end plugs on the traps were not removed; consequently,
these would not be true field blanks. Because of this, vapor phase mercury measurements
reported here are not corrected for field blank values. In addition, the blank values are
below the level reported for ambient measurements (i.e. blanks are hundredths of a ng/m3
and ambient levels are reported to the tenth of a ng/m3).
4-18
-------
Table 4-6. Results of Field Blanks for Vapor Phase Mercury by Site.
Site
n
Average
Blank (pg)
Std Dev. of
Blanks (pg)
Detection Limit (3 a)
IIT
10
4.3
3.3
9.9 pg
LAU
6
17.7
19.0
57.0 pg
SHA
9
25.6
23.4
70.2 pg
Overall
25
15.9
15.2
45.6
4.5.4. Particulate-Phase Collection.
Particulate sample collection was carried out through filtration of moderately sized
air samples (approx. 20-40 m3) onto one of a variety of filter media. The media examined
during the sampling and development period included glass-fiber, quartz-fiber and Teflon.
Historically, these different media have been used for different applications because of a
variety of properties which they possess. These properties include inertness, pore size,
cost, Hg background, and applicability for other uses. The evaluation described below
focused on the last two factors, and sought to determine if filters which were to be used
for INAA (Teflon) might be usable for wet extraction and the amount of Hg background
each media might contribute.
4.5.4.1. Glass-Fiber
Millipore Corporation, Bedford, MA. 47mm, AP40 style. These filters are made
of borosilicate glass c\id formed into a mat of microfibers. They do not contain any
additional binders and may be heated to quite elevated temperatures to remove volatile
impurities, such as Hg. The listed pore size is 0.8 urn.
Gelman Sciences, Inc., Ann Arbor, MI. 47mm, A/E style. These filters are also
made of borosilicate glass. The listed "aerosol retention" is .3 (am at 32 LPM and
>99.98% efficiency.
4-19
-------
4.5.4.2. Quartz-Fiber
Pallflex Products Corp., Putnam, CN. 47 mm, 2500QAT-UP type. This material
is ultra-pure binderless quartz which is washed with soft water during production. Filter
efficiency is 99.99% at 0.3-0.05 jim. As with the glass, these filters may be baked to
reduce background Hg.
4.5.4.3. Teflon
Gelman Sciences Inc. 47 mm, Teflo with ring style. Teflon is an inherently clean
and inert medium, which is the industry standard for PM10 measurements using
Dichotomous samplers and non-destructive analytical techniques (e.g. INAA, XRF and
PIXE). The pore size is 2 nm. This media cannot be heated to lower the Hg background.
Investigations of the background concentrations of Hg on these filters indicate that
after firing at 500 °C for 3 hours, the glass and quartz-fiber materials are roughly
comparable in mercury content (40-200 pg Hg/filter). However, the glass appears to have
a slightly lower mercury content. The Teflon filters need not be treated before use and
have similar background levels (200 pg Hg/filter).
Filters were placed in acid-cleaned Teflon filter packs. Total suspended particulate
(TSP) samples were collected using an open-faced filter pack (Savillex Corp.,
Minnetonka, MN) while fine fraction (da<2.5 nm) particles were collected using a Teflon
filter pack connected to a Teflon-coated aluminum cyclone inlet system (University
Research Glassware, Carrboro, NC).
Air was pulled through these sampling systems using two different methods.
Diaphragm-type pumps (Gast Co., Benton Harbor, MI) were used for the moderate-to-
higher flow rate applications (10-30 LPM) often with a needle valve (Hoke) to control
flow. The flow rate during sampling was measured using a spirometer-calibrated
rotameter (Matheson), with the flowmeter on the inlet of a flow-test filter pack connected
to the sample pump. The use of a flow test filter pack was used to avoid contamination of
the true sample. The second approach employed was moderate-and high-flow-rate
sampling using mass-flow-controlled pump units. These units contained a variety of Tylan
mass flow control units and custom electronic panels which enabled careful flow rate
adjustment. These devices were also used for low flow (<1 LPM) applications related to
gas-phase Hg measurements. In most applications, total sample volume was measured
with spirometer-calibrated dry test meters (Schlumberger).
4-20
-------
All of the particulate mercury [Hg(p)] samples reported here were collected for a
nominal duration of 24 hours. While some samples were considerably shorter in duration,
most samples were found to have Hg levels significantly elevated above the detection
limit. As discussed below, adequately detectable samples may be expected from samples
as short as 6 hours under typical ambient conditions.
All sample handling was carried out using cleanroom (particle-free) vinyl gloves.
This included the loading and unloading of the filter packs as well as any handling of the
filter packs (i.e., during flow measurements). The loading and unloading of filters was
conducted out-of-doors to take advantage of the low Hg levels in ambient air.
Filters were loaded into the filter packs using acid-cleaned Teflon-coated forceps
just prior to sampling. Immediately after sampling, the filters were removed from the filter
packs and carefully placed into clean storage containers. Acid-cleaned polyethylene petri
dishes (Gelman Sciences, Inc.) or Teflon vials (Savillex) were the primary storage
containers used in all of the studies. After placing or folding the filters carefully into the
containers using the forceps, the container lids were sealed with Teflon tape to help
prevent diffusion of Hg vapor into the containers. Samples were kept in acid cleaned
Teflon jars which were taped and double-bagged in polyethylene "ziplock"-type bags and
stored at -40 °C in the dark until the time of analysis.
4.5.5. Particulate Phase Analytical Procedure
Two different analysis techniques were used to analyze the Hg(p) samples. They
are neutron activation analysis and dual-amalgamation cold vapor atomic fluorescence
spectrometry.
4.5.5.1. Neutron Activation Analysis
Neutron Activation Analysis is a non-destructive, radiation-based analytical
(Radiochemical) technique. Its usefulness stems from its high level of sensitivity combined
with the capability to simultaneously determine of many elements in a wide range of
sample media. The ability to quantitatively determine the elemental composition of
samples is particularly useful for source profiling and source/receptor modeling.
Briefly, INAA involves the activation of elemental constituents in a sample by
bombarding them with thermally excited neutrons which collide with the atomic nuclei and
convert some of the material present into radioactive isotopes of the original elements
4-21
-------
(radionuclides). The neutrons are frequently generated by the radioactive discharge of a
nuclear reactor. Following activation, the elements that are present and their abundance
are determined by gamma-ray spectrometry which identifies elements by the characteristic
energy of their gamma-ray discharge and their concentrations by the intensity of their
decay product emissions. Calibration of the results is achieved through comparison of the
detector response in the samples to the response from established Standard Reference
Materials (SRMs).
All of the INAA results which are presented here were obtained through the efforts
of Dr. Ilhan Olmez at the Massachusetts Institute of Technology Nuclear Reactor
Laboratory. This particular laboratory is unique in that the irradiation process takes place
near room temperature, whereas most reactors used for this technique expose samples to
elevated temperatures. For a highly volatile species such as mercury, this is critically
important in obtaining accurate and reliable data (Olmez, 1992, personal communication).
4.5.5.2. Cold Vapor A tomic Fluorescence Spectrometry
CVAFS is a member of the emission spectrometry family of analytical techniques.
In these methods, an aliquot of analyte material is temporarily suspended in the path of an
electromagnetic radiation source (usually ultraviolet or visible light) and excited by the
light. A photomultiplier tube (PMT) then detects transmitted light of the same frequency
or emitted light of the same or different frequency from the excited analyte. In atomic
fluorescence, the incident energy is long-wave UV (254 nm) supplied by a Hg vapor lamp.
This UV light excites Hg atoms suspended in a stream of carrier gas (He or Ar) and a
PMT situated perpendicular to the incident light beam detects the 254 nm photons
released during the fluorescent relaxation of the Hg atoms to their ground state energy.
Because of the geometry of the system and the inherently narrow bandwidth of the
fluorescent emissions, additional monochromators or filters are not necessary. Thus the
CVAFS method has a high energy throughput which results in excellent specificity and
sensitivity.
The "dual amalgamation" pre-concentration technique (Fitzgerald and Gill, 1979)
coupled with CVAFS (DACVAFS) aids in the analysis of minute amounts of mercury in
environmental samples from a variety of media. In this procedure, Hg° amalgamates to
gold-coated sand on a "sample" trap which contains approximately one gram of gold-
coated sand. While passing an inert gas (He or Ar) through the trap, the Hg is desorbed
by heating the trap to 400°C onto a second, "analytical" trap, from which it is desorbed
4-22
-------
into the spectrometer. This approach is effective in removing water, the primary positive
interference, from the system before quantification of Hg. This further increases the
sensitivity, specificity and applicability of the CVAFS method.
To use this analytical method for the determination of Hg(p), the Hg present in
various chemical forms in the particle phase (and for most other media as well) must be
captured as Hg° vapor on a gold-coated sand trap. This process involved the extraction
and volatilization of the Hg present on a collection filter. The Hg(p), presumably present
in a variety of chemical forms, was first separated from the particle mass matrix collected
on the filter material through an acid digestion process accompanied by sonication. The
sample filter was folded and placed into a 33-ml acid-cleaned Teflon vial. A 20-ml aliquot
of a 10% solution of 70% nitric, 30% sulfuric acid was then added as the primary
solubilizing agent in the extraction. The acids used for the extraction solution were
purged of Hg by bubbling with a Hg-free stream of pure nitrogen through the liquid for an
extended period of time (12-18 hours). The sample was then sonicated for 30 minutes to
break up the particle matrix without heating. Previous experiments indicated this to be a
more effective technique for extraction than heating the filter in the extraction solution.
This is possibly due to the lower temperature (estimated at 40-60 °C) of the technique and
less loss through volatilization. Following extraction, all of the solubilized forms present
were converted to Hg2+ through a 1-hour oxidation using 0.5 ml of BrCl solution. Also,
this step may aid in the extraction of Hg from the matrix. Next, the BrCl is reduced by the
addition of 0.1 ml of NH2OH solution to a 5-ml aliquot of the extract. This was necessary
to remove the halogens from solution which may damage the gilded surface of the gold-
sand traps. Finally, 0.5 ml of SnCl2 solution is added to the aliquot in a bubbler/impinger
setup which reduces the Hg2* to Hg°. The volatile Hg° was bubbled out of solution with
Hg-free N2, and the Hg vapor recaptured on the "sample" gold-sand trap. The analysis is
completed through detection as described above. Tv._»s, the DACVAFS approach yields a
value for Hg(p) which represents total "acid-extractable" Hg, and cannot distinguish the
chemical forms that were present at the time of sampling. The technique is precise (<15%
variability) and sensitive (detection limit -250 pg/filter or 5 pg/m3 for a 24-hour sample
collected at 30 LPM).
It was essential, especially when performing Hg(p) analysis using this technique, to
carry out the extraction/volatilization/quantification procedures under ultra-clean
conditions (Class 100) (Boutron, 1990). For most of the data presented here, a simple
cleanroom was constructed from a plastic frame draped with clear vinyl to isolate it from
4-23
-------
room air. A high efficiency particle filter (HEPA) and motor assembly supplied the
cleanroom with class 100 quality air, and particle-free gloves, suits and boots allowed
analysts to handle sample extraction and analysis in the cleanest possible conditions.
Ultra-clean conditions also demand extremely pure reagents, and many of the chemicals
which were used during analysis had to be further purified or purged of Hg before they
could be used. The simple cleanroom, in use for a part of this investigation, has been
replaced by a state-of-the-art ultra-trace element analysis facility at the University of
Michigan Air Quality Laboratory. The laboratory is constructed of all plastic and wood
with no exposed metal in the room. Vapor mercury levels are approximately 1 ng/m3.
4.6. Quality Assurance
Several steps were taken to assure the reliability and replicability of data collected
during the LMUATS. Site audits were performed at each of the four LMUATS sites
during the course of the study. Results of these audits determined that air samplers were
performed within the criteria set forth in the quality assurance project plan. The final
report of these performance and systems audits, prepared by Jack A. Bowen of the EPA
QATSD and Mike Pleasant of METI, are found in the Appendix.
Audit samples were prepared by EPA for analysis of VOCs, pesticides and PCBs
on PUF, and PAHs in solution. Results of these analyses are reported in the Appendix. In
addition, collocated samples were analyzed for elemental and organic carbon, PCBs,
pesticides, and a comparison between XRF and INAA analysis of 12 elements from
identical Teflon filters was conducted.
Collocated elemental and organic carbon analyses reveal excellent agreement
between samples (Figure 4-3). Mean elemental carbon values from the 26 collocated pairs
were 0.36 and 0.35 with standard deviations of 0.41 and 0.37, respectively; and 1^=0.92.
Mean organic carbon values (n=26) were 3.72 and 3.55 (standard deviation 2.62 and 2.74,
respectively), with an r2=0.94.
Fine fraction elements (Al, CI, K, Ti, V, Cr, Mn, Fe, Zn, As, Se and Br) were
analyzed on the same filter by both XRF and INAA for 25-31 samples from IIT (Table 4-
7). Results of this comparison indicate good agreement between the two analytical
methods for these elements.
4-24
-------
Figure 4-3. Collocated Elemental and Organic Carbon Samples at South
Haven.
Comparison of two samplers at South Haven
Elemental Carbon
Organic Carbon
Juiy August
Date
4-25
-------
Table 4-7. XRF vs. INAA Comparison at IIT.
Elements
n
XRF
Mean + Std Dev.
(ng/m3)
INAA
Mean + Std Dev.
(ng/m3)
r
r2
p value
Mn
32
6.18 ±10.72
5.81 ±8.55
0.99
0.98
0.0001
Zn
32
38.8 ±59.18
32.45 ±53.35
0.99
0.98
0.0001
Fe
32
124.78 ±122.82
154.38 ±159.21
0.89
0.79
0.0001
Se
32
0.99 ±1.03
1.05 ±0.79
0.88
0.77
0.0001
K
26
62.47 ±40.91
63.31 ±37.4
0.77
0.59
0.0001
A1
31
77.59 ±94.64
103.58 ±80.28
0.76
0.58
0.0001
Ti
25
4.88 ±7.67
9.3 ±6.08
0.75
0.56
0.0001
Cr
32
0.61 ±0.79
1.26 ±0.77
0.60
0.36
0.0003
CI
32
8.71 ±12.87
9.18 ±3.38
0.58
0.34
0.0004
As
32
0.21 ±0.93
0.51 ±0.43
0.56
0.31
0.0009
Br
32
2.81 ±1.02
0.83 ±0.94
0.47
0.22
0.0070
V
32
0.68 ±1.82
0.73 ±0.66
0.45
0.20
0.0100
PCB and pesticide replicate analyses are reported in Table 4-8. In general, there
was good agreement between several PCB isomers and congener groups and several
pesticides.
In addition, 46 collocated samples were collected at South Haven for fine and
coarse elements using the Dichot sampler. Results of XRF analysis of the paired filters are
presented in Tables 4.9 and 4.10. Fine elements and total fine mass were highly correlated
between the collocated samples for the majority of elements (Table 4-9).
Comparison of collocated filters for coarse particle mass (Table 4-10) indicates
that, in general, replicate samples could be reliably collected and analyzed for the majority
of coarse fraction elements.
4-26
-------
Table 4-8. Comparison of PCBs and Pesticides in Collocated Samples at
South Haven.
Sampler
Sampler
No. 933411
No. 933413
Compound
n
Mean
Std Dev.
Mean
Std Dev.
f*
r2
(pg/m3)
(pg/m3)
1
p,p'-DDD
5
6.6
9
7.6
6
0.99
0.98
Dieldrin
5
109.7
100.3
163.6
84
0.98
0.95
gamma-Chlordane
5
28.2
21.1
32.4
19.2
0.91
0.83
trans-Nonachlor
5
17.3
19.3
22.2
17.2
0.88
0.78
Mirex
5
6.2
4
9.8
4.4
0.88
0.78
2,2',4,4',5,6'-PCB
5
0.1
0.1
0.2
0.3
0.88
0.78
alpha-Chlordane
5
24.5
19.5
30.6
15.2
0.85
0.72
Atrazine
5
246.3
402.4
333.8
335.3
0.84
0.71
TOT.mono-PCB
5
191.7
101.9
240.6
118.2
0.82
0.67
TOT.hexa-PCB
5
46.4
35.4
52.9
17.5
0.81
0.66
2,2',3,4,5'-PCB
5
7.6
6.9
12.7
5.5
0.79
0.62
TOTAL PCBs
5
721.4
378.7
843.8
266.9
0.78
0.61
4,4'-DDT
5
224.5
175.2
252.9
74.5
0.78
0.61
2,3-PCB
5
20.6
10.7
24.5
5.5
0.77
0.60
TOT.pen-PCB
5
251.1
124.9
281.1
88.7
0.74
0.55
hexachlorobenzene
5
42.1
17.2
58.7
5.8
0.71
0.50
TOT.tri-PCB
5
54
33.8
68.4
28.6
0.70
0.48
TOT.tet-PCB
5
68.7
46.1
82
42
0.68
0.47
Metolachlor
5
87.9
130.2
44.9
61.9
0.49
0.24
p,p'-DDE
5
948.9
515.8
1271.6
334.1
0.40
0.16
Chlorpyrifos
5
146.5
116.4
222.8
185.3
0.37
0.14
TOT.di-PCB
5
104.4
61.8
113.6
28.9
0.27
0.07
alpha-HCH
5
84.9
42.6
125.9
22.9
0.27
0.07
TOT.hep-PCB
5
4.9
3.8
4.4
2.9
-0.16
0.03
Aldrin
5
0.3
0.4
0.4
0.4
j.15
0.02
TOT.octa-PCB
5
0.2
0.2
0.2
0.3
-0.10
0.01
gamma-HCH
5
49.9
29.8
75.7
21.9
0.11
0.01
TOT nona-PCB
5
0
0
0.1
0.2
-
-
Simazine
5
0
0.1
0
0
-
-
deca-PCB
5
0
0
0.5
1.1
-
-
Alachlor
5
0
0
33.4
74.7
-
-
2-PCB
5
0
0
3.4
5.1
-
-
2,4,5-PCB
5
0
0
0
0
-
-
2,2'3,4',5,6,6'-PCB
5
0
0
0.1
0.2
-
-
2,2',4,6-PCB
5
0.1
0.2
0
0
-
-
2,2',3,3',4,5',6,6'-PCB
5
0
0
0.1
0.2
-
-
4-27
-------
Table 4-9. Comparison of Collocated XRF Fine Particle Data at South
Haven.
Variable
n
Sampler 1
Mean + Std Dev.
(ng/m3)
Sampler 2
Mean + Std Dev.
(ng/m3)
r
r2
p value
S
46
2218 ±2649
2190 ±2591
0.98
0.96
0.0001
Mass
46
16710±14279
16072±13557
0.96
0.92
0.0001
Fe
46
90.37 ±93.98
68.5 ±65.02
0.96
0.92
0.0001
Pb
46
6.92 ±5.86
6.47 ±5.17
0.94
0.88
0.0001
K
46
54.08 ±37.01
50.23 ±31.13
0.92
0.85
0.0001
Zn
46
16.47 ± 14.2
15.98 ± 13.92
0.91
0.83
0.0001
Mn
46
3.45 ±3.39
3.25 ±2.65
0.86
0.74
0.0001
Ca
46
61.5 ±53.23
46.13 ±31.59
0.78
0.61
0.0001
Br
46
2.07± 1.05
1.99 ± 1.03
0.74
0.55
0.0001
Se
46
1.08 ±0.82
1.04 ±0.91
0.72
0.52
0.0001
Ti
46
4.37 ± 4.21
3.11 ±3.46
0.61
0.37
0.0001
Si
46
157.85 ± 137.53
122.5 ± 100.57
0.59
0.35
0.0001
Cu
46
3.22 ± 2.17
4.1 ±3.22
0.38
0.14
0.0090
CI
46
3.75 ± 4.71
7.14 ± 5.44
0.37
0.14
0.0100
Sc
46
0.22 ± 1.67
0.29 ± 1.65
0.36
0.13
0.0100
Te
46
1.29 ±6.78
-0.31 ±6.63
0.35
0.12
0.0200
Ni
46
0.01 ±0.7
0.05 ±0.74
0.34
0.12
0.0200
Ba
46
11.62 ± 13.76
12.29 ± 12.19
0.32
0.10
0.0300
4-28
-------
Table 4-10. Comparison of Collocated Coarse Particle XRF Data at
South Haven.
Variable
n
Sampler 1
Mean + Std Dev.
(ng/m3)
Sampler 2
Mean + Std Dev.
(ng/m3)
r
r2
p value
S
46
136.81 ± 119.73
125.52 ± 120.1
0.86
0.74
0.0001
Ca
46
307.34 ± 279.17
408.32 ±359.66
0.76
0.58
0.0001
Fe
46
186.07 ± 166.56
236.4 + 213.39
0.75
0.56
0.0001
Si
46
854-713
1121 ±834
0.74
0.55
0.0001
Mn
46
6.11 = 5.28
7.97 ±6.68
0.73
0.53
0.0001
A1
46
194.18 ± 186.14
258.41 ±207.62
0.69
0.48
0.0001
Pb
46
2.49 = 3.26
2.13 ± 3.13
0.69
0.48
0.0001
Ti
46
15.46 = 12.4
20.13 ± 15.06
0.67
0.45
0.0001
K
46
94.58 ± 56.27
108.99 ±59.83
0.60
0.36
0.0001
Ni
46
0.39 = 0.98
0.15 ± 1.45
0.60
0.36
0.0001
Mass
46
8071 - 4782
9288 ± 5003
0.47
0.22
0.0009
P
46
19.06 - 17.3
24.25 ± 17.03
0.45
0.20
0.0020
Sr
46
1 - 0.96
1.04 ± 1.09
0.40
0.16
0.0060
Cr
46
0.74 = 0.97
0.5 ± 1.02
0.34
0.12
0.0200
Y
46
0.77 = 4.02
0.92 ±4.9
0.32
0.10
0.0300
4-29
-------
4.7. Data Processing
All data processing and analysis was completed using strict quality control and quality
assurance measures. Finalized data sets received from laboratories at which the samples had
been analyzed were checked for consistency with master files of sampling dates and
identification numbers. Figure 4-4 describes the data acquisition stream for each measurement
made as part of the LMUATS and the analytical method and reporting organization
responsible for the data.
Figure 4-5 outlines the processing stream used to generate a complete, checked data
file for use in statistical analyses. Data from original files received from analysis
laboratories were used to prepare input files for SAS (Statistical Analysis Software, Cary,
NC) using a consistent format including site, date, time of sample collection, sample
identification number, pollutant concentration and note codes so that each row of data
described one unique sample. These files were saved as ASCII files with the filename
specifying site and group of compounds contained in the data set.
SAS programs were run to create SAS data sets for each group of compounds for
each site. SAS program and SAS data set names corresponded to site and group of
compounds, for example KAIPCB.SAS is a SAS program which reads in original PCB data
for Kankakee and generates a SAS data set named KAIPCB.SSD. Once SAS data sets were
created on personal computers, the data files were converted to XPO files and transferred to a
UNIX based system which could handle the manipulations of such a large data set with much
greater speed. Compound-specific data were merged to create four summary data sets by site.
A final data set was created by merging these four data sets into one main data set.
4.8. Mixed-Layer Trajectories
Surface and upper air meteorological observations were obtained from the National
Weather Service (NWS) with additional soundings from the Lake Michigan Ozone Study
measurements. On-site meteorological measurements of temperature, pressure, dew point and
wind speed and direction were also utilized to estimate the time of frontal passages. Mixed-
layer 72-h back trajectories were calculated for air masses arriving at the site at 0200, 0800,
1400, and 2000 EDT daily (Heffter, 1980). The backward trajectories were calculated
starting at monitoring sites moving backwards in time in 3-h time steps. The trajectories
represent the most probable path of an air parcel advected with the mixed-laver-averaged
winds. Trajectory calculations are not exact and have associated uncertainties which
generally increase with the time or distance upwind of the receptor location (Kahl and
Samson. 1986;Draxler, 1987).
4-30
-------
Figure 4-4. LMUATS Data Aquisition Stream.
4-31
-------
Figure 4-5. LMUATS Data Processing Stream.
4-32
-------
Chapter 5
Composition of the Atmospheric Aerosol System
This section of the report presents the results of the elemental and chemical
analysis of the samples collected during the study. The comprehensive characterization of
the atmospheric aerosol system (gases and particles) includes both inorganic and organic
chemical analysis, and in many cases in two or more size fractions. The extensive suite of
the measured chemical compounds provides critical information on the composition and
the sources of the potentially toxic air pollutants found in the lower Lake Michigan Basin.
The concentrations of several classes of compounds were determined at four
locations during the study period. The classes were PAHs, pesticides, PCBs, and metals
which included Hg, elemental and organic carbon, and inorganic species. The data
presented here are perhaps the most extensive attempt to completely characterize the
chemical composition of the atmosphere at multiple sites simultaneously. The data will be
discussed by compound class with attention focused on the inter-site differences for
specific compounds rather than day to day fluctuations in the concentrations of each
compound.
5.1. PM10 and Ionic Composition
Figure 5-1 shows the PM-10 concentrations measured between 8 July and 9
August, 1991. The fine fraction mass (<2.5 |im) is shown in black on the bottom of each
bar. There is a pronounced episode of elevated PM-10 concentrations from 16-22 July
where the levels exceeded 80 (ag/m3 at both Kankakee and IIT and exceeded 60 pig/m3 in
South Haven. Mixed layer trajectories reveal that the stagnant air masses were
transported from the southwest during this period. The maximum PM-10 concentrations
were observed on 19 July with air mass transport from the St. Louis area to each site
(Figure 5-2). The PM-10 concentrations also peaked on 2 August when winds were light
and from the southwest at each site.
5-1
-------
Figure 5-1. Variations in Measured PM10 Concentrations at Four
Measurement Sites.
Fine
Kanka kee
CZD
Coarse
100 -|
i r~i i i i i i i— i r
S 10 1 1 12 13 14 15 16 17 IB 19 20 21 22 23 24 25 26 27 2B 29 30 31 1 2
4 5 6 7 B 9
60 -
40
20 -
IIT
mm
lUi ntlil
9 10 1 1 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 2B 29 30 31 1234567B9
100 R/V Laurention
BO -
60 -
I-
i i r
1112
M4-
ii—i—i—r
23 24 25 26 27
ill
t-!—i—t~t—i—i
5 6 7
South Haven
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 i i i i i
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 123456789
July
Date
August
5-2
-------
Figure 5-2. Mixed Layer Trajectories Ending at the Three Land-Based Sites
on 19 July, 1991. (Squares indicate position of the air mass in 3-hour time
increments moving backward from the site.)
18Z 19 JUL 91 — 72hr Trajectories
5-3
-------
Perhaps more interesting than the elevated PM-10 levels during the study is the
change in the aerosol size distribution after transport from Kankakee to IIT to South
Haven as reflected by the fine fraction of the PM-10 shown in Figure 5-3. One objective
of the LMUATS was to capture an episode where air mass transport "linked" the upwind
site (Kankakee) with the urban site (IIT) and the over water transport site (R/V
Laurentiari) and the downwind site (South Haven). The episode from 16-23 July
provided the perfect opportunity for linkage between 3 of the 4 sites with the research
vessel. Unfortunately, the vessel was not on station during this period. However, Figure
5-3 shows that at the beginning of the episode on 16 July about 78% of the PM-10
observed at Kankakee was in the fine fraction which is typical of a regional transported
aerosol. From 18-22 July, the PM10 level at Kankakee rose as did the percentage of
coarse particulate matter due to transport from sources upwind of Kankakee such as the
St. Louis urban/industrial area. As the air mass moved into Chicago the concentration of
the fine particulate mass dropped due to dispersion. The addition of coarse mass occurred
from the activities in the urban/industrial area. The PM-10 levels observed at IIT were
only slightly lower than those measured in Kankakee but about 40% of the PM-10 was
now in the fine fraction. As the air masses continued across Lake Michigan, without
additional sources en route to South Haven, the fine fraction again reasserted itself as the
dominant portion of the PM-10. This again is what one typically observes at a site
influenced primarily by long range transport. The PM-10 levels in South Haven during the
episode were about 20 (j.g/m3 less on the average than the levels observed concurrently at
IIT. This decrease can be largely explained by dispersion. The change in the particle size
distribution, however, cannot simply be explained by en route dispersion. It appears that
larger particles observed in Chicago do not make it to South Haven and are probably lost
by deposition. Therefore, the PM-10 is primarily made up of fine particles in South Haven
after overwater transport. This loss of large particles probably would have been
accentuated if the particles greater than 10 jim in size had been sampled (Holsen et al.,
1993).
Figure 5-4 shows the fine S concentrations determined by XRF at the four
locations during the study. Sulfur, in the form of sulfate, is the largest contributor to the
fine mass at each site. The pattern in the fine S concentrations is very similar from site to
site with the maximum concentrations measured in Kankakee. Annular denuder systems
(ADS) were operated during the study at two of the four sites during LMUATS. There
were a total of 74 ADS samples collected in South Haven, and 22 on the R/V Laurentian
of 6 and 12 hours in duration. Figure 5-5 shows the time series for fine particulate S042"
5-4
-------
Figure 5-3. Variations in Measured Fine Fraction PM10 Concentrations.
Kankakee
IIT
South Haven
nil
i i i rrrr i i i i i i i i i —
8 9 10 11 12 13 1* 15 16 17 18 182021 22 23 24 25 26 27 2B 29 3031 1 2 3
B 9
July
Date
August
5-5
-------
Figure 5-4. Variations in Measured Fine Particle S Concentrations.
1 0000 -I
B000
'e 6000 -
cn
c 4000
2000 -
0
R/V Laurentian
¦'I
T'f*T H
~i—i—rnr—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—r
11 12 232* 25 26 27
-i—r
Jii
5 6 7
10000 -I
8000
South Haven
nE 6000 -
6 9 10 11 12 13 1* 15 16 17 18 19 20 21 2223 24 2526 27 28 29 3031 123456789
July Date August
5-6
-------
Figure 5-5. Variations in Measured Fine Particle SO42" Concentrations.
R/V Laurentian
300
270 -
240 -
210 -
n
[ 1B0-
o 150 -
E
c 120 -
90 -|
60
30
0
d
lllll 111
JfL
"I—I—i T T i—I—I—I—I—I—I—i—I—, T 7 T T 7 I—I—I—I—l—I—r
1 1 12 23 24 25 26 27 5 6 7
South Haven
July
August
Date
5-7
-------
measured during the study at the two locations. The 6-hour samples were averaged into
12-hour data for statistical analysis so that the data would be comparable to the 12-hour
integrated elemental and organic composition data. It should be noted that the XRF fine S
data agreed quite well with the ADS fine particulate S042" determined by ion
chromatography (r=0.99).
The average composition of the fine particles collected is given in Table 5-1. The
concentrations measured in Ann Arbor, MI during the same period are also shown in the
table for comparison. Levels measured on the R/V Laurentian were typically very similar
to those measured concurrently in South Haven. It should be noted that the levels
observed while the R/V Laurentian was on station are, in general, quite low as would be
Table 5-1. Summary of ADS Concentration Data Collected from
8 July - 9 August, 1991.
Laurentian
South Haven
Ann Arbor
n
Mean Std Dev.
n
Mean Std Dev.
n
Mean Std Dev.
NH3 (ppb)
14
1.7 ± 1.4
63
1.2 ±0.9
22
3.1 ±0.8
Maximum
5.2
5.8
4.4
HONO (ppb)
14
0.3 ±0.3
64
0.3 ±0.2
22
0.4 ±0.3
Maximum
1.0
1.1
1.5
HNO3 (ppb)
14
1.3+0.9
64
0.8 ±0.9
22
0.8 ±0.7
Maximum
3.4
3.9
2.6
S02 (ppb)
14
3.4+4.4
64
1.9 ±2.1
22
2.8 ±2.0
Maximum
15.8
8.9
8.7
SO42" (nmole/m3)
14
34 ±33
64
52 ±66
22
79 ± 102
Maximum
115
281
329
H+ (nmole/m^)
14
14 ±11
64
25 ±43
22
24 ±41
Maximum
34
241
130
NH4+ (nmole/m3)
14
53 ±64
64
82 ±98
22
131 ± 171
Maximum
204
398
544
5-8
-------
Figure 5-6. Mixed Layer Trajectories to the R/V Laurentian while On-Station
Off-Shore of Chicago 23-27 July and 5-8 August 1991.
5-9
-------
expected with the strong north-northwesterly or east-northeasterly flow to the vessel when
it was on-station near Chicago (Figure 5-6).
The concentrations measured in South Haven are typical of summertime values
previously measured in the midwestern U.S. (Pierson et al., 1989). The average levels
observed are also very similar to those measured during September 1989 in Green Bay,
Wisconsin (Mamane et al., 1993). Mean S042" concentrations in South Haven and
aboard the R/V Laurentian were 52 and 34 nmole/m3 (5.0 and 3.3 |_ig/m3), respectively.
The mean S042" concentration at Chicago over a 1-year period was 5.55 [ig/m3 (Lee et
al., 1993). The interesting difference in the sulfate observed during the LMUATS and
those previously observed in Chicago is the acidity of the sulfate. The mean and maximum
H+ concentrations measured in Chicago were 8 and 78 nmole/m3 compared to 14 and 34
on the R/V Laurentian and 25 and 241 nmole/m3 in South Haven. The fact that the
maximum H+ level observed in Chicago was one-third that measured in South Haven
suggests that SO2 to H2SO4 conversion is occurring over the lake en route to South
Haven. Figure 5-7 displays the H+ concentrations measured at the two sites during the
study. The figure clearly indicates that acidic S042" was observed in South Haven during
the transport episode 16-22 July. It is also clear that the S042" measured over the lake
was not neutralized to a great extent. However, as seen in Figure 5-8, NH4+ is the
dominant cation even during the transport episode. Considerable neutralization had
already occurred when the air mass reached South Haven during the 5-day period. Fine
particle nitrate concentrations shown in Figure 5-9 were typically very low during the
study and near detection limits during the transport episode previously discussed.
Changes can be observed in the chemical profile of fine fraction aerosols measured
during the LMUATS. The H+ to S042" ratios at different sites were 0.78, 0.44, and 0.16,
on the R/V Laurentian, South Haven, and Ann Arbor, respectively. This indicates that the
acidic S04 " aerosol measured over the lake is to a large extent unneutralized. However,
as the aerosol is transported inland (even a short distance as at South Haven) an additional
25% of the acidity is neutralized. This is most likely due to rapid fumigation of the air
mass after reaching the shoreline where relatively high levels of ammonia react rapidly
with the acidic S04 However, compared to the average inland values measured in Ann
Arbor during the study, both the over-lake and South Haven areas appear to be exposed
to relatively unneutralized sulfate.
Measurements of the gaseous compounds were limited to those species quantified
by the ADS or those measured as part of the LMOS. Figure 5-10 shows the gaseous NH3
5-10
-------
Figure 5-7. Variations in Measured Fine Particle H+ Concentrations.
R/V Laurention
ill
h—r—i—I—i—i—r
1112
1 I I T T T T T I—I T
23 24 25 26 27
5 6 7
South Haven
1 22 241213
July
August
Date
5-11
-------
Figure 5-8. Variations in Measured Fine Particle NH4+ Concentrations.
R/V Laurentian
400
360 -
320 -
280 -
240 -
200 -
160 -
120 -
60 -
40 -
0 -
41
i.i
iiiii
"i—i—r
\ \ i r
11 12
23 24 25 26 27
~ i T
5 6 7
South Haven
400
360 -
320 -
260 -
240 -
200 -
160 -
120 -
60 -
40 -
0 -
I ¦ ¦ ¦ i
iIii i "I i 1 I I I i I i I i i i I I I I i I i I
8 9 10 11 12 13 14 15 16 17 1B 19 20 21 22 23 24 25 26 27 28 29 30 31 123456789
July
August
Date
5-12
-------
Figure 5-9. Variations in Measured Fine Particle NO3" Concentrations.
R/V Laurentian
30 -1
25
10 -
"i i r~i i i r
1112
1—1—1—r
1—1—1—1—1—1—m—r~r
23 24 25 26 27
-Mr
5 6 7
South Haven
30
25 -
20 -
|
c
10 -I
8 9 10 11 12 13 1* 15 16 17 18 19 20 2122 23 24 25 26 27 28 29 30 31 1 23456789
July August
Date
5-13
-------
Figure 5-10. Variations in Measured Gaseous NH3 Concentrations.
R/V Laurentian
CD
CL
a
5.0
4.5
4.0 -
3.5 -
3.0 -
2.5 -
2.0 -
1.5 -
1.0 -
0.5 -
0.0
ll
1112
"l I r~T 1 1
23 24 25 26 27
"I i i i i i r
i r
5 6 7
South Haven
CD
CL
a.
5.0
4.5
4.0 H
3.5
3.0 H
2.5
2.0
1.5
1.0
0.5 H
0.0
5.5
i II I I I I I i
8 S 10 1112 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 123456788
July
August
Date
5-14
-------
concentrations measured at the two sites using the ADS. The NH3 concentrations on the
R/V Laurentian were elevated when the flow to the vessel was directly from shore. This
occurred both on the 11-12 July and on the 24 July when the vessel experienced a period
of transport from Chicago to the ship. SO2 concentrations measured at the two sites were
generally quite low with a few exceptions. The SO2 levels observed in South Haven
increased 4-5 fold during the period of southwesterly flow from 16-22 July. Southwest
flow also resulted in elevated concentrations on 2 August. The SO2 concentrations
measured on the R/V Laurentian 6-7 August reached 16 and 10 ppb respectively as seen
in Figure 5-11. The 12-hour average values were elevated due to direct plume impact
from the Fe-steel industry in Gary. The plume from the plant was in contact with the
vessel for about 2 hours during the 12-hours of sampling, thus, the concentrations in the
plume were substantially higher that the levels displayed.
The levels of photochemically derived pollutants HNO3 and O3 are shown in
Figures 5-12 and 5-13. Ozone levels exceeded the NAAQS on several days in South
Haven during the transport period from 16-22 July. HNO3 concentrations also were
elevated during this period at South Haven reaching 4.0 ppb on 18 July corresponding to
the highest O3 observed during the study of 153 ppb. In fact, the correlation between the
maximum hourly O3 and the 12-hour integrated HNO3 concentration was 0.92. This is
similar to the relationship observed in other areas with elevated photochemical pollutant
levels (Keeler et al., 1990a). The elevated oxidant concentrations measured in South
Haven after over-water transport is clearly seen. While the levels of all of the particulate
species are not elevated during this period, most of the combustion or secondary species,
such as sulfate, are elevated.
5.1.1. Carbon Measurements
The fine fraction carbon measurement data are shown in Table 5-2. The Cv refers
to organic carbon species in the fine particulate matter that could be volatilized. The Ce is
the elemental carbon or black carbon in the fine particulate matter often associated with
diesel exhaust. At the IIT site the mean elemental carbon concentration is only about 12%
of the organic carbon concentration. A ratio of Cg/Cv of 1/8 is similar to that reported in
the Green Bay study for the fine fraction carbon. However, the ratios of Cg/Cv at the
other sites are smaller with ratios of 1/12, 1/10, and 1/10 at the more rural Kankakee, R/V
Laurentian, and South Haven sampling locations.
5-15
-------
Figure 5-11. Variations in Measured Gaseous SO2 Concentrations.
R/V Laurentian
16 10
n—r
t—1—1—r
"i—1—r
iiUi
1—1—r
1—1—1—1 1— —r
5 6 7
1 1 r
1112
23 24 25 26 27
South Haven
10 11 12 13 14 15 1 6 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
July
August
Date
5-16
-------
Figure 5-12. Variations in Measured Gaseous HNO3 Concentration.
R/V Laurentian
5.0
4.5 -
4.0 -
3.5 -
3.0 -
2.5 -
2.0 -
1.5 -
1.0 -
0.5
0.0
1—i—r
1
w!
1—r iiii
23 24 25 26 27
1 I I I I I 1 T T T I I I
5 6 7
1112
South Haven
July
August
Date
5-17
-------
Figure 5-13. Variations in Measured Maximum Hourly 03 Concentrations.
200
175
150
125
100
75
50
25
0
l IT
|i|i llllikll 11 llllllililiiilillllllllli i UiWllilluii
i i i i i r r ~t "t i i i i i Ii i i i i i
B 9 10 1 1 12 13 14 15 1 6 17 1 8 19 20 21 22 23 24 25 26 27 2B 29 30 31 123456789
200 -|
175 -
150 -
125 -
100 -
75 -
50 -
25 -
0 -¦
R/V Laurentian
i—i—i—i—i—i—r
1—i—i—i—r
11
23 24 25 26
South Haven
July
Date
August
5-18
-------
Table 5-2. Average and Maximum Concentrations for Fine Fraction
Carbon (Units are ng/m^).
Kankakee
IIT
R/V Laurentian
South Haven
Mean Std Dev
Mean Std Dev
Mean Std Dev
Mean Std Dev
Cv
4.4 ± 1.8
6.7 ±3.5
3.2 ± 1.7
3.7 ±2.6
Max
7.9
26
6.1
10.1
Ce
0.3 ±0.3
1.21 ±1.0
0.3 ±0.4
0.35 ±0.4
Max
1.0
4.6
1.1
1.4
While the proportion of the carbon in the elemental and organic forms is similar to
that observed in Green Bay; the concentration levels observed in this study are on the
average, about two times greater with the mean at IIT being four times greater for both
elemental and organic carbon.
Figures 5-14 and 5-15 show the temporal variation observed in the carbon levels
measured during LMUATS. Both Cv (organic) and Ce (elemental) carbon levels are
elevated during the transport episode from 16-22 July at IIT and South Haven. The
elemental carbon levels observed in Kankakee are much lower during this period
suggesting that local combustion sources in Chicago are responsible for the elevated levels
there as well as at the downwind site in South Haven. Elemental carbon concentrations on
the lake were typically very low as would be expected from the previous discussion.
5.1.2. Micro Orifice Impactor Results
Results of the analysis of filters from the Micro Orifice Impactor (MOI) were used
to observe the size distribution of chemical species at the two sampling sites. S04
NH^ and H+ consistently appeared on stages 4, 5 and 6 (primarily 4 and 5). The
observation of these species in <1 [am diameter particles is typical of other similar
measurements (Pierson et al., 1989). The mass mean diameters (MMD) of H , S04 and
N03" were calculated from data from all 6 stages by determining the nominal diameter at
_|_
-------
Figure 5-14. Variations in Measured Elemental Carbon Concentrations.
3.0
2.5 -
2.0 -
1.5 -
1.0 -
0.5 -
0.0 --
R/V Laurentian
1—i—i—r
23 24 25 26 27
I T I
ii4L
"l l l l l I r
5 6 7
1112
3.0
2.5
„ 2.0 -h
E
\ 1.5 H
Ol
a
1.0 -
0.5 -
0.0
South Haven
"i—i—r
-f-T-
I I T T 1— T T T T T T I 1—r
15 16 17 1 8 19 20 21 22 23 24 25 26 27 28 29 30 31
I.I, i,»
123456789
July
Date
August
5-20
-------
Figure 5-15. Variations in Measured Organic Carbon Concentrations.
15
12
)
E 9
\
cn
= 6
3
0
26 16
15
3
R/V Laurentian
II
I I l I I I l
"i—i—l|H , , , ,
, , Jl* 1
11 12
23 24 25 26 27
5 6 7
15
12 -
5
E 9 -
\
oi
3 6
3 -
0
South Haven
1—i—i—i—r
u
¦ 1
T T i 1 I I I I I I I i I I i I I I I I i i I
15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 123456789
July
Date
August
5-21
-------
5.2. Average Trace Element Concentrations
Trace element determinations were made on Teflon filters collected using
dichotomous samplers as described earlier. X-Ray Fluorescence (XRF) analysis was
performed on each pair (fine and coarse) of samples collected at the 3 land-based sites and
on the R/V Laurentian. Tables 5-3 and 5-4 provide a statistical summary of the
concentrations of the 16 elements and the particulate mass in the fine and coarse fraction
of the PM10. The Kankakee site was found to have the highest mean concentrations for
fine mass as well as several of the lighter elements such as Al, Br, S, and CI in the fine
fraction. The fine fraction Se concentration was on average the highest, suggesting the
regional coal combustion influence is dominant at Kankakee. The highest concentrations
for the remainder of the elements, most of which are anthropogenic in origin, were
observed at the IIT site.
Sulfur, almost entirely in the form of sulfate, is the largest elemental contributor to
the fine particulate mass at each site. As discussed earlier, the measured sulfate is often
very acidic and is found in various states of neutralization. The next most abundant
elemental species was Si (with the exception of the Kankakee site, where Si composed an
average of about 10% of the total PM10). The soil-derived elements Al, K, Ca, and Fe
were the next most abundant in terms of mass, together contributing about 5% to the
PM10 measured.
The average Al concentration measured at the Kankakee site, both fine and coarse,
is elevated above the levels observed at the other sites. Table 5-5 gives a comparison of
the ratios of Al/Si, K/Si, Ca/Si, and Fe/Si at the sites to the average crustal abundance
given by Mason (1966) to investigate if other source impacts are found in addition to soil.
It is clear that the Kankakee site is being impacted by a combustion source emitting Al in
the PM10 range. The fine Al/Si concentration ratio for South Haven is 0.80 which is also
elevated above the crustal average. This average is strongly influenced by the extremely
elevated Al concentration measured on 2 August which comprised more than half of the
mean. If this one sample were removed, the fine Al/Si ratio would drop to 0.4, which is
closer to the IIT ratio and the crustal average. The table also suggests that a source or
combination of sources of fine Fe and Ca are also contributing to the Fe and Ca observed
at all of the sites. Except for the coarse Al/Si ratio at Kankakee, the remaining
concentration ratios are similar to the crustal average, suggesting that the coarse-fraction
aerosols are primarily of soil or crustal origin.
5-22
-------
Table 5-3. Average and Maximum Concentrations for Fine Fraction Trace
Elements Determined by XRF (ng/m3).
Kankakee
nT
R/V Laurentian*
South Haven
Mass*
Avg
20.1 ± 14.1
17.9 ± 11.9
9.8 ±6.8
13.6 ± 12.6
Max
54.7 (7/18)
45.4 (7/19)
28.6 (8/7)
50.1 (7/17)
A1
Avg
550.9 + 905.3
77.6 ±94.6
8.7 ±44.0
101.1 +356.6
Max
4529.9 (7/22)
429.9 (7/22)
79.9 (8/7)
2756.4 (8/2)
Si
Avg
154.3 +204.1
169.4 + 178.8
71.9 ±53.7
126.1 ± 138.4
Max
1044.8 (7/22)
858.2 (8/2)
162.3 (7/11)
911.7 (7/22)
S
Avg
2921.6 +2718.0
2437.3 +2153.7
1229.8 ± 1217.6
1811.3 ±2339.1
Max
10355.3 (7/17)
7566.3 (7/17)
4708.2 (8/7)
8994.3 (7/17)
CI
Avg
12.1 +8.9
8.7+12.9
8.7 ±7.0
6.3 ±5.3
Max
46.9 (8/2)
74.1 (8/8)
19.4 (7/12)
24.7 (7/11)
K
Avg
50.7 + 24.9
62.5 ±40.9
54.9 ±49.9
46.5 ±31.8
Max
124.0 (7/18)
175.4 (7/21)
169.2 (8/7)
145.0 (7/22)
Ca
Avg
69.0 + 52.9
78.6 + 57.0
44.7 ±24.2
44.2 ±33.1
Max
254.4 (7/25)
230.8 (7/22)
86.7 (7/24)
143.9 (7/22)
Ti
Avg
4.8+6.7
4.9 ±7.7
3.5 ±2.7
3.7 ±5.1
Max
32.0 (7/22)
33.9 (7/22)
9.6 (7/24)
34.7 (7/22)
V
Avg
0.5 + 1.2
0.7 ± 1.8
0.5 ± 1.4
0.4 ±1.7
Max
2.7 (7/20)
3.6 (7/22)
2.5 (8/5)
3.5 (7/22)
Mn
Avg
3.2 +
6.2 ± 10.7
4.1 ±4.9
2.9 ±2.5
Max
10.1 (8/1)
59.3 (7/22)
19.4 (8/6)
11.7 (11/17)
Fe
Avg
74.3 +67.6
124.8 ± 122.8
78.5 ± 102.6
63.7 ±67.5
Max
291.5 (7/22)
558.1 (8/8)
401.1 (8/6)
316.6 (11/22)
Ni
Avg
0.5 +0.9
0.2 ±0.7
0.1 ±0.7
0.0 ±0.7
Max
2.7 (7/29)
1.5 (7/14)
1.2 (8/5)
1.6 (8/2)
Cu
Avg
6.2+4.0
10.0 ±5.2
5.6 ±5.2
3.7 ±2.9
Max
20.2 (8/2)
22.3 (7/21)
23.2 (7/23)
15.4 (8/8)
Zn
Avg
22.3 + 15.5
38.8 ±59.2
21.2 ±20.1
13.8 ± 12.8
Max
78.8 (7/11)
294.1 (8/8)
63.9 (8/6)
63.9 (7/17)
Se
Avg
1.5 + 1.7
1.0 ± 1.0
1.1 ± 1.4
0.9 ±0.8
Max
9.4 (8/6)
4.1 (8/8)
4.5 (8/7)
3.0 (7/18)
Br
Avg
3.6 + 1.7
2.8 ± 1.0
2.4 ± 1.3
1.8 ± 1.0
Max
7.4 (8/6)
4.9 (8/8)
4.3 (7/12)
4.8 (7/17)
Pb
Avg
13.2+7.1
14.3 ± 14.9
8.5 ±9.8
5.7 ±4.9
Max
32.4 (7/12)
62.1 (8/8)
36.3 (8/6)
27.5 (7/24)
3
~ Hg/m (mass, only)
* Sampling Dates R/V on-station 7/11-7/12, 7/25-7/27, 8/5-8/8, reflect periods of predominantly northerly flow
5-23
-------
Table 5-4. Average and Maximum Concentrations for Coarse Fraction Trace
Elements Determined by XRF (ng/m^).
Kankakee
HT
R/V Laurentian
South Haven
Mass*
Avg
18511 + 13470
15135 +12386
6281 + 3636
8271 +4969
Max
54991 (8/2)
48282 (8/2)
14631 (7/24)
22697 (7/19)
A1
Avg
4736.2 ± 5251.1
615.6 + 582.7
165.9 + 188.2
232.1 ± 210.2
Max
19772.2 (8/2)
2472.9 (8/2)
533.5 (7/24)
757.6 (7/19)
Si
Avg
1066.5 + 722.1
2000.2 ± 1995.9
713.3 + 509.9
1010.9 + 785.1
Max
2859.4 (7/20)
8368.1 (8/2)
1862.5 (7/24)
3285.3 (7/19)
S
Avg
193.8+115.2
141.5 + 130.4
78.5 + 65.0
107.1 + 113.2
Max
591.5 (8/6)
510.3 (8/6)
257.9 (8/6)
448.9 (7/18)
CI
Avg
32.1 + 19.5
47.7 + 41.1
32.5 +47.3
24.9 ±27.1
Max
72.4 (7/26)
174.9 (8/6)
210.5 (8/6)
155.9 (7/19)
K
Avg
112.8 + 46.8
192.7 + 183.4
78.2 +45.7
98.5 + 58.5
Max
216.1 (8/2)
776.8 (8/2)
181.6 (7/24)
263.1 (7/19)
Ca
Avg
676.2 + 388.1
1150.7 + 1036.7
381.1 + 289.9
355.9 + 335.0
Max
1505.3 (7/17)
4082.5 (8/2)
1123.4 (7/24)
1276.0 (7/18)
Ti
Avg
23.3 + 15.4
44.8 + 41.4
16.6+11.1
18.0+ 14.8
Max
64.9 (7/22)
166.6 (8/2)
49.8 (7/24)
57.5 (7/19)
V
Avg
1.0 + 2.3
1.9 + 2.2
0.7 + 1.1
0.7 + 2.5
Max
3.7 (7/20)
7.7 (7/19)
2.5 (7/12)
3.6 (7/11)
Mn
Avg
7.9 + 5.2
15.6 + 14.6
7.6 + 7.3
6.9 + 6.3
Max
23.1 (7/26)
64.2 (8/6)
29.5 (8/6)
23.3 (7/19)
Fe
Avg
233.5 + 132.5
589.2 + 472.8
247.9 + 283.4
207.0 + 199.7
Max
609.8 (7/26)
1894.4 (8/2)
1153.2 (8/6)
742.5 (7/18)
Ni
Avg
1.3 + 1.5
0.5 + 1.2
0.1 +0.8
0.1 + 1.3
Max
4.4 (8/2)
5.4 (7/19)
1.6 (7/12)
7.7 (7/17)
Cu
Avg
13.4 + 12.0
9.9 + 15.8
3.1 + 3.1
3.7+6.1
Max
40.1 (7/19)
93.0 (7/19)
11.7 (8/6)
30.1 (8/5)
Zn
Avg
14.9+10.4
21.9 + 17.6
8.2 + 8.3
6.4 + 6.7
Max
42.2 (8/5)
67.4 (8/8)
31.2 (7/24)
37.1 (7/17)
Se
Avg
0.0 + 0.4
0.0 + 0.3
0.0+0.4
0.0 + 0.4
Max
1.1 (7/26)
0.8 (7/21)
0.6 (8/6)
1.0 (7/20)
Br
Avg
0.5+0.6
0.6+0.6
0.7 + 1.1
0.5+0.6
Max
1.7 (7/9)
1.8 (7/22)
3.9 (8/6)
2.0 (7/16)
Pb
Avg
3.3+2.2
10.6 + 14.9
2.9 + 2.9
1.8 + 2.8
Max
7.8 (7/19)
80.6 (7/9)
11.1 (8/6)
18.4 (7/17)
~ ug/m (mass, only)
* Sampling Dates R/V on-station 7/11-7/12,
7/25-7/27, 8/5-8/8, reflect periods of predominantly northerly flow.
5-24
-------
Table 5-5. Average Compositional Ratios for Selected Elements.
Site Al/Si K/Si Ca/Si Fe/Si
Fine Coarse Fine Coarse Fine Coarse Fine Coarse
Kankakee
IIT
R/V Laurentian
South Haven
Green Bay (Aerosol Study)
Crustal Average (Mason) 0.20 0.10 0.13 0.18
3.60
4.40
0.33
0.11
0.45
0.63
0.48
0.22
0.46
0.31
0.36
0.10
0.46
0.58
0.74
0.29
0.12
0.23
0.76
0.11
0.62
0.53
1.1
0.35
0.80
0.23
0.37
0.10
0.35
0.35
0.50
0.20
1.4
0.78
0.9
0.13
0.66
0.61
0.96
0.20
The levels of many elements measured during the LMUATS are similar from one
site to the other. While it was expected that Kankakee would have lower levels due to its
rural location with little local industrial activity, this was not the case. Kankakee observed
higher average levels of trace elements, in both size fractions, than the "downwind" rural
site in South Haven. For example, Figure 5-16 shows the fine Pb concentrations
measured at the four measurement sites. The range in Pb levels observed in Kankakee are
greater than what was observed in South Haven. Interestingly, the maximums observed
aboard the R/V Laurentian (nearly 40 ng/m3) are higher than those measured in
Kankakee. Both of the peaks observed on the R/V Laurentian were observed on days
when, for at least a portion of the sampling period, the vessel was downwind of the
Chicago or Gary urban/industrial areas.
The urban site in Chicago (IIT) observed comparable or slightly higher levels of
most trace elements found in the fine fraction. The concentrations of fine Mn, Fe, Cu, and
Zn were only two times higher than those measured in Kankakee and about three times
greater, on the average, than those measured in South Haven. The concentration of
coarse fraction elements was also about two to three times greater in Chicago than at the
other sites. Figure 5-17 displays the coarse Pb concentrations measured at the four sites.
The levels of coarse Pb are much higher at IIT than at any of the other sites.
5-25
-------
Figure 5-16. Variations in Measured Fine Pb Particle Concentrations.
c
40 -|
30 -
20
10 -
Kankakee
lllllllilmillili
l l l l l l l l l l l l i l l l I l I l T~t I T l ! I
9 10 1112 13 1*15 16 17 18 19 2021 222324 2526272829 3031 1 2 4 5 6 7 6 9
40
30
20 -
10 -
62
62
I IT
ilii
u-
u
<1 LI J
i i i i i i i 1 i I I I I I I I I T T T I
9 10 1 1 12 13 14 15 16 17 18 1S 20 21 22 23 24 25 26 27 28 29 30 31 1 23456769
U444+
40 -|
30 -
n
E
\ 20 H
cn
c
10 -J
R/V Laurentian
ll ll
IT i—i—i—i—;—i—r
Jll
~i—i
1112
23 24 25 26 27
5 6 7
40
30 H
E
\ 20
C
10
South Haven
hll l»i>iilll
I llliLi-T"! I'I'I'iMl'fh'TMlM'i
I [ I I I I I I " I 1 I i I i
8 9 10 11 12 13 14 15 16 17 18 19 2021 22 23 24 25 26 27 2B 29 30 31 123456789
July
Date
August
5-26
-------
Figure 5-17. Variations in Measured Coarse Particle Pb Concentrations.
Kankakee
1 ! H I I ! I U H I i T 1 I I 1 f I I T I f i T 1 i f !
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 4 5 6 7 B 9
81
I IT
32
+4
iHhlU
d
JiiLi
I I I I I T I I I I I I I I I ! i I I I I I I I
9 10 1 1 12 13 14 15 16 17 16 19 20 21 22 23 2425 26 2728 29 30 31 1 23456789
R/V Laurentian
iM'r !¦
"]—i—i—;—i i i r
23 24 25 26 27
Ll
11 12
5 6 7
South Haven
44-
f fc'H' T I'f I I
1 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 1
8 9 10 1 1 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 123456789
July
Date
August
5-27
-------
While Pb was typically sub-micron in size when leaded gasoline was still used prior to the
late 1980s, the industrial Pb emission is more bi-modal in size suggesting that fugitive
emissions may be comparable in importance to the current combustion emissions.
The levels of the trace elements measured aboard the R/V Laurentian were lower,
on the average, than those measured at the shoreline sites. This is largely because of the
prevailing northerly flow that the vessel received while on station. Since there are no large
sources in the lake and the upwind fetch from the sampling point in southern Lake
Michigan is hundreds of kilometers of open water, the levels of most pollutants would not
be expected to be very high. However, the first cruise on 11-12 July positioned the R/V
Laurentian 20 miles offshore from Muskegon and easterly flow resulted in elevated levels
of many elements including Se and Zn. There were also westerly winds from the Greater
Chicago area for the daytime sample on the 24 July which brought elevated Pb levels, both
fine and coarse, as well as elevated Zn. The last cruise was also successful in being
influenced by the industry in Gary, IN.
The influence of the iron-steel industry on the composition of the measured aerosol
is observed clearly in Figures 5-18 through 5-20. The fine Fe concentrations are seen in
Figure 5-18. Elevated Fe levels are seen at all three sites, and as suggested earlier, above
what would be expected from windblown dust. The peaks in the fine Fe are seen in
Chicago and aboard the R/V Laurentian during the last few days of the study. These days
were characterized by southeast winds bringing emissions from southeast Chicago and
Gary to both measurement sites. Concurrent peaks in the fine Mn concentration are seen
in Figure 5-19. The PM10 episode observed from 16-22 July is also obvious in these plots
of fine Fe, Mn, and Zn. The plots clearly show that the levels of these marker elements for
iron-steel combustion are higher both at IIT and South Haven during the period 16-20
July. This would suggest that the ...eel plants in the Gary/southeast Chicago area are the
most important contributor to the levels of these metals measured in South Haven. On
July 21, the concentrations of fine Fe are elevated at both Kankakee and IIT but not at
South Haven. This is because a stationary front sitting between South Haven and Chicago
cut this site off from the southerly flow hitting the sites south of the front. The levels
observed at the three land-based sites on 22 July are all elevated except that the highest
concentrations are observed first in Kankakee. This is consistent with the mixed layer
trajectories for this day which connect Kankakee with Chicago and South Haven. The
elevated concentrations of iron-steel plant marker elements is also seen on 5-8 August.
5-28
-------
Figure 5-18. Variations in Measured Fine Particle Fe Concentrations.
01
c
400 -i
300 -
200
100
Kankakee
xlUl
ii
h
i—i—i—i—i———i—i—i——i—i—i—i—i—i—i—I—i—i—i—i—i—i—i
9 10 11 12 13 14 15 16 17 1B 19 2021 22 23 24 23 26 27 28 29 30 31 1 2 4 5 8 7 8 9
1 I I T II i I f I f I
400
300 -
en 200
c
1 00
558
I IT
lltM I
itttttttttttttttttttttttttttTi
9 1011 1213141516171819 20 21 22 23 24 25 2627 28 29 30 31 1 2 3 4 5 6 7 8 9
4+
400 -i
300 -
cn
c
200 -
100 -
401
R/V Laurentian
fit
nil
't I't
1112
t—i—i—i—i—i—i—n—i —i—i—i
23 24 25 26 27 5 6 7
400
South Haven
8 9 10 11 12 13 14 15 16 17 18 19 2021 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8
July
Date
August
5-29
-------
Figure 5-19. Variations in Measured Fine Particle Mn Concentrations.
E
o>
25
20 .
15
c 10
5
0
25
20
n
E 15
CT>
C 10
5
0
Kankakee
M T III
+4-4+
+4
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 4 5 6 7 0 9
59
I IT
Iff
uu
fir?t 11 f ITI
"T i T~
I T 1—rTTTTTTTTTTTTTTTTTTTTTTT I TTTTT I
9 10 11 12 13 14 15 16 17 IB 192021 222324252627 2629 3031 1 2 3 4 5 6 7 6 9
25
20 -
'e 15 _
c io H
5 -
0
R/V Laurentian
i—i—r
iiL
i—i—i—r
T T T i i i
4
1111 111
5 6 7
1112
23 24 25 26 27
South Haven
July
Date
August
5-30
-------
Figure 5-20. Variations in Measured Fine Particle Zn Concentrations.
Kankakee
o>
c
211 294
c
100 -i
60
60 -
40 -
20 -
0
South Haven
HU
M1
¦In.I
i i i
8 9 10 1 1 12 13 14 15 16 17 18 19 2021 22 23 24 25 26 27 2B 29 30 31 123456789
July
Date
August
5-31
-------
This period was one in which the R/V Laurentian was on station off Chicago and was
impacted by the plume from the Gary steel industry. The winds were out of the southeast
during this period at both the IIT site and the R/V Laurentian.
The concentrations of fine Si are shown in Figure 5-21. Aerosol Si is typically a
good marker element for windblown soil but elevated concentrations of fine Si suggest
another source contributing to the levels observed. Peaks in the fine Si concentration are
observed at each site, but not always on the same day. It has been suggested that fine
particle Si is a product of the reduction of the original oxide in coal (SiC>2) during
combustion, causing volatilization of SiO followed by re-oxidation to form SiC^. While
this explanation is possible, it is hard to see a clear relationship between the fine Si and the
fine Se (marker for coal combustion) shown in Figure 5-22.
5.3. Atmospheric Mercury Levels
5.3.1. Measured Vapor-Phase Levels
The concentration of atmospheric Hg varied greatly from site to site and from day
to day during the study. At IIT the vapor-phase Hg concentrations measured from 10 July
- 9 August, 1991 ranged from 1.8 - 62.7 ng/m3, with an average of 8.7 ng/m3 (Table 5-6).
The peak concentration was measured from a 6 hour sample in the afternoon on 8 August.
The subset of samples collected at IIT and at South Haven for 6 hour periods were
averaged to generate a 12 hour mean for purposes of comparison among sites (Figure 5-
23). Twenty-five vapor-phase Hg samples were collected on the R/V Laurentian, during
the three separate cruises. The average vapor-phase Hg concentration measured on the
R/V Laurentian was 2.3 ng/m3. The mean vapor-phase Hg concentration observed in
South Haven was quite typical of previous measurements in the Great Lakes basin, 2.0
ng/m3. (Schroeder, 1982 Glass et al., 1990 and Fitzgerald et al., 1991). A total of four
duplicate samples were collected at South Haven. The average precision for the vapor-
phase sampling was better than 15%.
5-32
-------
Figure 5-21. Variations in Measured Fine Particle Si Concentrations.
800
700
600
500
400
300
200
100
0
Kankakee
1045
¦ ulnl.illi
i iff
9 10 11 12 13 14 13 16 17 1B IB 20 21 22 23 24 25 26 27 28 29 30 31 1 2
858
4 5 6 7 S 9
800
700
600
500
400
300
200
100
0
R/V Laurentiar
1—r
JL
-M*
i—i T"T*i—i—i—r
IM1^ i
1112
23 24 25 26 27
5 6 7
800
700 -
600 -
500
400 -
300
200 -
100 -
0
South Haven
912
't'Im M1?1!*
I I I i i i I I i I I i i I I 1 i I
S 9 10 11 12 13 14 15 1 6 17 1 8 19 20 21 22 23 24 25 26 27 2B 29 30 31 123456789
July
Date
August
5-33
-------
Figure 5-22. Variations in Measured fine Particle Se Concentrations.
Kankakee
9.4
Uul-lU*
1 i \ 1 T I
k
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 4 5 6 7 8 9
I IT
"i—i—r^T—r~T—i—i 1—i I—I T I T T T T I T I T~~T I I T T T T I
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 2B 29 30 31 1 234567B9
I LXl
u
R/V Laurentian
w
"I—I —i—i—I—I—i—I—I—I—I—I T7l T 7 I—!—I—I—I—I—I—I T T T I—T
1 1 12 23 24 25 26 27 5 6 7
South Hcven
B 9 10 11 12 13 14 15 16 17 18 19 2021 222324 2526272829 30 31 123456789
Ju|y Date August
5-34
-------
Figure 5-23. Variations in Measured Vapor-Phase Hg Concentrations.
59 18
i i i i i i i i i i i i i i i i i i r
1 1 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 2B 29 30 31 123456789
R/V Laurentian
JlU
1—I—I—I—I" T T 7 I—I—I
5 6 7
i i i i i r
1112
23 24 25 26 27
15 -i South Haven
July
August
Date
5-35
-------
5.3.1.1. Diurnal Variation of Vapor-Phase Mercury
Samples were collected more intensively at the IIT site with 18 samples collected
between 8AM-2PM (designated as AM), 17 samples collected between 2PM-8PM (PM),
11 daytime 12-hour samples collected between 8AM-8PM (DAY) and 12 nighttime
samples were collected from 8PM-8AM (NIGHT). This frequency of sampling allowed us
to investigate the diurnal behavior of vapor-phase Hg. The average concentration (ng/m3)
for the AM samples was 3 .3 times larger than that for the NIGHT samples. The average
concentration for the PM samples was 2.1 times larger than NIGHT samples (Figure 5-
24). The mean vapor-phase Hg concentration for AM and PM samples was 10.1 ng/m3,
while for the DAY samples it was 9.9 ng/m3.
Table 5-6. Vapor-phase Hg Measurements in Chicago (IIT), on the R/V
Laurentian (LAU) and in South Haven (SHA) (ng/m^).
SITE
n
MEDIAN
MEAN
STD DEV
MIN
MAX
HT
58
4.5
8.7
12.0
1.8
62.7
LAU*
25
2.2
2.3
0.7
1.3
4.9
SHA
38
1.8
2.0
0.6
1.8
4.3
*DATES: R/Von-station 7/11-7/12, 7/25-7/27, 8/5-8/8
5.3.1.2. Spatial Variation in Vapor-phase Hg
Vapor-phase Hg concentrations were significantly lower at sites distant from
Chicago. The concentrations of vapor-phase Hp measured in South Haven and aboard the
R/V Laurentian were on average half those concentrations measured at IIT. Two
episodes were evident during the study, marked by predominant flows from the southwest,
carrying pollutants such as sulfate and particulate matter from upwind sites to downwind
sites on the RV and in South Haven. During the first episode, 19-23 July, a slight increase
in Hg° at the IIT site was observed; a corresponding increase in Hg° at the South Haven
site was not detected.
5-36
-------
Figure 5-24. Variations in Measured Vapor-Phase Hg in Chicago.
N= 1 B
N=1 1
AM PM DAY NIGHT
8 om — 2 pm 2 pm — 8 pm 8 am - 6 pm 8 pm — 8 am
5-37
-------
5.3.2. Measured Particulate-Phase Levels
At two of the land sites, the Illinois Institute of Technology in Chicago (south side)
and South Haven, MI, and on the research vessel R/V Laurentian, vapor- and particle-
phase Hg samples were collected. The particle-phase samples [Hg(p)] were collected on
glass-fiber filters for 24 hours (8AM - 8AM CDT) at 30 LPM in open-faced filter packs
and at 10 LPM in cyclone-inlet filter packs. In some instances, additional open-faced filter
packs were taken with glass-fiber and Teflon filters. The glass-fiber TSP samples were
analyzed by CVAFS while the fine mass and TSP samples collected on Teflon were
analyzed by INAA at MIT. The results from these samples are summarized in Figure 5-
25.
The Hg(p) results from the different analytical techniques together with values for
vapor-phase Hg, aerosol mass, aerosol elemental content, and meteorological parameters
begin to define some potential trends in the atmospheric chemistry for Hg(p). The Hg(p)
detected using CVAFS was in almost all instances significantly lower than that measured
by INAA. Values for both techniques yielded higher concentrations for the Chicago site
(IIT) and less for the R/V Laurentian and in South Haven. However, the differences
between the radiochemical and wet extraction techniques were much smaller in Chicago
than in South Haven. Size-segregated samples (fine and total suspended mass) from
South Haven analyzed by INAA indicate significant amounts of Hg(p) in the coarse
fraction. The CVAFS and INAA results on fine fraction samples were very similar. This
apparent chemical- and/or size-dependent distinction strongly indicates a natural source
for Hg found in the coarse fraction at the South Haven site. In the coarse fraction,
particulate Hg which is strongly bound to particulate matter appears to be unavailable for
extraction using strong acids and is only quantified by radiochemical techniques. In
contrast, fine fraction particulate Hg appears to be equally recovered by acid digestion and
neutron activation analysis.
Other data collected during this study support this explanation for the difference
between results from the acid digestion versus neutron activation analysis of particulate
Hg. The Hg(p) value for TSP as determined by INAA correlates significantly (r =0.72
P<0001) with the coarse-to-fine-mass ratio (Figure 5-26). The values for fine Hg(p) by
INAA (no corroborative data for CVAFS) display peak values during the primary air
quality "episode" from 16-22 July
5-38
-------
Figure 5-25. Variations in Measured Particle-Phase Hg Concentrations.
720
910
500 —
300 —
200
100 —
0
300
200
1 00
R/V Laurentian
"i—i—r
-f-T-
> II
n—i—i—I—T—T—T—i—r
5 6 7
23 24 25 26 27
South Haven
l I I I I
JLjl
111
hi
l
r V r V V I V I i i ~i ~i T i i i i i i i i i i r
15 16 17 18 19 20 2122 23 24 25 26 27 28 29 30 31 123456789
July August
5-39
-------
Figure 5-26. Correlation of Hg(p) by INAA Measured at South Haven with the
Coarse-to-Fine-Mass Ratio.
200
1 50
cn
100 -
50 -
0
1 9
20
21
May
22
23
5-40
-------
During the 16-22 July episode, air flow was predominantly from the SW and was
accompanied by elevated ozone, sulfate, acidity and marker elements for various
metallurgical industries. Unfortunately, TSP mass concentrations at South Haven were
not measured. Information on typical soil concentrations of Hg given by Fergusson
(1991) suggests that soil Hg ranges from 0.01 - 0.5 ppm (pg Hg/^g soil) globally. In
order to observe the concentrations of Hg(p) found in South Haven by INAA (200
pg/m3), coarse fraction particle mass (in the size range 2.5 - 10 |im and >10^m) would
have to range between 4 - 200 ng/m3.
The observed coarse mass in the size range 2.5-10 (am at South Haven ranged
between 2-22 ng/m3, which is not enough to account for all of the Hg(p) measured.
This finding indicates that particles greater than 10|am are likely to be a contributing factor
to the particulate Hg measured, since particles of size 10 - 30 (am are collected as part of
particulate Hg sampling. The Hg content of the soil in South Haven was measured by the
University of Michigan Air Quality Laboratory to be 0.99 ppm, about twice the
unperturbed background level given above. Therefore, resuspended soil particles in the
size range 10-30 could be responsible for a significant portion of the Hg observed in
the coarse fraction. The local soil in South Haven may have been previously contaminated
through deposition of Hg or the application of mercury-containing compounds
(pesticides).
The behavior outlined above (between INAA and CVAFS) was not observed in
Chicago. At IIT, the INAA and CVAFS results more closely approximate each other in
most instances. The difference in behavior observed between South Haven and Chicago
can be explained by first assuming that anthropogenically emitted Hg(p) can exist in both
the fine and coarse fractions. This assumption has been initially confirmed from a brief
samplinp program carried out in Ann Arbor with the collection of TSP and fine mass
quartz-fiber niters side-by-side followed by analysis with CVAFS. Figure 5-27 and Table
5-7 summarize these results. It is clear from these samples that Hg(p) which is acid-
extractable by our technique may be found in both the coarse and fine fractions, but that
on some occasions, it resides primarily in the fine fraction. In Chicago, a similar profile
may exist, but during transport across Lake Michigan, the larger coarse fraction material is
lost from the urban plume more rapidly than the fine material by settling. Thus, the profile
of the urban plume will change during transport, becoming devoid of anthropogenic,
"acid-extractable" Hg in the coarse fraction when it reaches South Haven.
5-41
-------
Figure 5-27. Results of Size Segregated Particulate Hg Sampling in Ann Arbor.
5-42
-------
The findings just presented were further explored during a two-week sampling
campaign at the Ann Arbor NDDN site outside of Dexter. The same sampling and
analytical schemes were utilized in the more recent study. The findings from that longer
experiment corroborated the results presented here: Aerosol Hg(p) can be found in both
the fine and coarse fraction, and that the total particulate Hg is, on occasion, equal to the
fine. At this point we cannot state what controls the size distribution of the atmospheric
particulate Hg we have measured. The findings presented here may be the first to
substantiate the existence of a coarse particle Hg in the Great Lakes Basin. Ongoing
research at the UMAQL will continue to shed new light on this important topic.
Table 5-7. Results From TSP v. FM CVAFS Analysis of Hg (pg/m3).
Day
PM 2.5
TSP
% in Fine
1
182
187
97
2
56
112
50
3
43
72
60
4
10
27
37
5
85
158
54
5.4. Average VOC Concentrations
The concentrations of volatile organic compounds (VOC) were measured using
summa canisters as discussed previously in Section 3. A summary of the arithmetic mean,
standard deviation, and maximum concentrations determined by Battelle Columbus is
given in Table 5-8. Also shown is the date that the maximum occurred to help assist us in
determining the cause of the elevated concentrations measured at each site. This also
helps to identify concurrent maximums between sites, and to facilitate the cross correlation
of maximum days with other maximums found for the other compounds.
The VOC data were not utilized extensively for this report. The data were thought
to be somewhat suspect since the levels are significantly lower than typical ambient
concentrations. The poor data quality was explained by a contaminant problem in the
5-43
-------
majority of canister samples. Analytical contamination was ruled out by routine
processing of standard reference materials. A subset of samples collected at Kankakee
were collocated with 2 hour VOC collections as part of the LMOS study. For the three
collocated samples, only benzene concentrations agreed well. Concentrations of toluene,
ethyl benzene, and p+p-xylene were significantly lower in LMUATS measurements as
compared to LMOS study measurements. Potential reasons for the low values may
include poorly cleaned canisters, poor vacuum seal or preparation, inadequacies in sample
collection or handling or improper extraction procedures. This problem may have been
avoided if proper blanks had been collected and analyzed before shipment of canisters to
collection sites and immediate analysis of the first subset of samples from each of the sites
to ensure that valid data was being collected. The research teams should have been
notified immediately when a problem was discovered.
As a result of the inadequate data, the VOC data will not be useful in receptor
modeling as it has in other studies. This is a disappointment as the EPA-AREAL has
successfully utilized this type of data in the past in CMB-type analyses.
5.5. Average PAH Concentrations
The average and maximum of atmospheric concentrations of total (vapor +
particle) PAHs at four sites are summarized in Table 5-9. The concentrations at IIT were
the highest among four sites and the concentrations at South Haven were the lowest while
the concentrations on R/V Laurentian were slightly higher than those at Kankakee.
Except retene, which showed a more regional distribution largely due to its natural
vegetative source (Figure 5-28), the concentrations of volatile PAHs in Chicago (IIT)
were about ten to fifty times greater than the concentrations observed at South Haven.
The concentrations of the higher molecular weight PAHs were about ten times higher in
Chicago. The steady concentration profile for coronene, a marker compound for motor
vehicles is shown in Figure 5-29. The elevated levels of coronene observed in Chicago are
expected due to the density of motor vehicle traffic in the large urban area. The 10-fold
decrease in the coronene concentrations observed both in South Haven and in Kankakee is
an indication of both the dearth of local traffic at those sites and dispersion that is taking
place between the source and the sites.
5-44
-------
Figure 5-28. Variations in Measured Retene Concentrations.
Kankakee
1.0
0.8
^ 0.6 -
C 0.4 -
0.2
0.0
1.0 -|
o.e
0.6
cn
c 0.4 -
0.2 -
0.0
I IT
~i—rr
u
~i—i—i i i
1 6 17 1B 19 20 21 22 23 24
1.3 1.0
"l I r~T I I I I T T T T I I
29 31 2 5 6 7 8
a
T T i—r
1112
-i—i—r—i—i—r
R/V Laurentian
I
23 24
"1 1 1 1 !
5 6 7
1.0 -|
0.8 -
E 0.6
en
c 0.4 -
0.2 -
0.0
South Haven
~rl
2
Uk
I I H
12
~ I I I
16 17 IB 192021 22 2324
29 31
July
Date
5 6 7 B 9
August
5-45
-------
Figure 5-29. Variations in Measured Coronene Concentrations.
10 -i
Kankakee
1 -
c
0.1 -
0.01
I I I I
i I I
16 17 18 19 20 21 22 23 24
"I I I I I I I I I I T T T T I I
29 31 2 5 6 7 6
10
I IT
0.01
t—r
"1 I I I I I T
1 6 17 18 19 20 21 22 23 2*
I I I I 1 I I
29 31 2 5 6 7 8
10 -i
R/V Laurentian
1 -
CT>
C
0.1
0.01
A
i—i T T i—i—i—i—i—i—i—i—i—r
1 1 12 23 24
~i—m—i—i—i—r
ii
I I T I I I
5 6 7
10 -i
South Haven
E
\
c
1 -
0.1 -
0.01
"I—I—I—I 7 I
12
I lllllll
16 17 18 19 2021 22 23 24
July
i r
29 31
I T T T T T I
5 6 7 8 9
Date
August
5-46
-------
Table 5-8. Average and Maximum Concentrations of VOC (ppbv).
Kankakee IIT Laurentian South Haven
1) dichlorodifluoromethane
Avg
0.52 + 0.21
0.68 ±0.20
0.54 ±0.20
0.48 ±0.08
Max
1.1 (7/17)
1.22 (7/24)
1.05 (7/26)
0.67 (7/17)
2) methyl chloride
Avg
0.49 ± 0.20
0.57 ±0.17
0.51 ±0.21
0.54
Max
0.72 (7/19)
0.89 (7/15)
0.75 (7/26)
0.54 (7/22)
8) trichlorofluoromethane
Avg
2.27 ±2.92
1.94 ± 1.99
0.78 ±0.21
0.73 ±0.13
Max
9.34 (7/21)
8.76 (7/21)
1.33 (7/27)
1.08 (7/21)
10) 1,1-dichloroethene
Avg
<0.1
0.38 ±0.13
<0.1
<0.1
Max
-
0.47 (7/19)
-
-
11) dichloromethane
Avg
0.29
0.63 ±0.38
0.31 ±0.03
0.28 + 0.07
Max
0.29 (7/30)
1.40 (7/31)
0.33 (7/24)
0.40 (8/6)
12) 3-chloropropene
Avg
<0.1
0.37 ±0.11
<0.1
0.97 ± 1.20
Max
-
0.45 (7/21)
-
1.82 (7/25)
13) 1,1,2-trichloro-1,2,2 -trifluoroethane
Avg
3.11+3.60
1.76 ±2.15
0.18 ±0.06
1.14 ±0.96
Max
15.31 (7/19)
6.32 (7/23)
0.35 (7/24)
3.47 (7/20)
17) 1,2-dichloroethane
Avg
0.16 + 0.06
<0.1
0.24
<0.1
Max
0.24 (7/29)
-
0.24 (8/6)
-
18) 1,1,1-trichloroethane
Avg
0.20 ±0.08
0.29 ±0.13
0.24 ±0.08
0.22 ±0.11
Max
0.46 (7/25)
0.76 (7/24)
0.41 (7/24)
0.52 (7/12)
19) benzene
Avg
0.41 ±0.15
0.85 ±0.68
0.33 ±0.14
0.34 ±0.45
Max
0.83 (7/19)
4.11 (8/2)
0.61 (8/7)
2.34 (7/21)
20) carbon tetrachloride
Avg
0.12 ±0.04
0.12 ±0.05
0.12 ±0.04
0.11 ±0.01
Max
0.22 (7/25)
0.25 (7/24)
0.23 (7/27)
0.11 (8/6)
22) trichloroethene
Avg
<0.1
0.14 ±0.06
0.23
0.20
Max
-
0.23 (7/24)
0.23 (7/24)
0.20 (7/12)
27) toluene
Avg
0.38 ±0.27
2.33 ±2.72
0.67 ±0.67
0.97 ± 1.86
Max
1.06 (7/19)
15.08 (8/2)
2.39 (7/11)
8.26 (7/21)
29) tetrachloroethene
Avg
0.69 ±0.67
0.33 ±0.36
0.23
0.13
Max
1.16 (7/19)
1.44 (7/9)
0.23 (7/24)
0.13 (7/18)
30) chlorobenzene
Avg
0.39 ±0.22
<0.1
<0.1
<0.1
Max
1.01 (7/22)
-
-
-
31) ethyl benzene
Avg
0.12 ±0.02
0.28 ±0.38
0.28 ± 0.12
0.44 ±0.48
Max
0.13 (7/25)
1.96 (8/2)
0.47 (7/11)
0.99 (7/21)
32) m+p-xylene
Avg
C./0 ±0.08
0.73 ± 1.19
0.72 ±0.62
0.69 ± 1.31
Max
0.28 (7/25)
6.87 (8/2)
2.10 (7/11)
3.90 (7/21)
33) styrene
Avg
<0.1
<0.1
0.18 ±0.06
0.59 ±0.81
Max
-
-
0.27 (7/11)
2.22 (7/12)
35) o-xylene
Avg
0.11
0.29 ±0.47
0.33 ±0.21
0.50 ±0.65
Max
0.11 (7/25)
2.52 (8/2)
0.72 (7/11)
1.46 (7/21)
36) 4-ethyl toluene
Avg
<0.1
0.24 ±0.25
<0.1
0.44
Max
-
0.79 (8/2)
-
0.44 (7/21)
37) 1,3,5-trimethylbenzene
Avg
<0.1
0.18 ± 0.16
<0.1
0.29
Max
-
0.55 (8/2)
-
0.29 (7/21)
38) 1,2,4-trimethylbenzene
Avg
<0.1
0.30 ±0.39
<0.1
0.63 ±0.56
Max
-
1.95 (8/2)
-
1.02 (7/21)
5-47
-------
Table 5-9. Average and Maximum Concentrations of PAHs (ng/m3).
Kankakee
HT
Laurentian
South Haven
Naphthalene
Avg
287.59 ± 342.48
507.25 + 179.84
119.38 + 125.36
66.32 + 66.06
Max
963.14 (7/16)
805.66 (7/16)
421.42 (8/7)
230.04 (7/17)
Acenaphthylene
Avg
2.61 +2.18
4.79 + 2.82
1.41 +0.98
0.49+0.24
Max
6.72 (7/29)
14.21 (8/8)
3.81 (8/7)
1.00 (8/6)
Acenaphthene
Avg
1.82 + 0.93
55.91 + 39.68
2.28 + 2.11
1.10+0.37
Max
3.42 (8/6)
133.38 (7/16)
8.10 (7/23)
1.78 (8/2)
Fluorene
Avg
3.69+1.81
53.67 + 33.55
7.17 + 4.50
3.45 + 1.90
Max
6.51 (8/6)
132.29 (7/16)
15.69 (8/7)
7.79 (7/17)
Phenanthrene
Avg
7.97 + 3.89
167.92 + 125.07
10.78 + 9.11
4.81 +2.05
Max
14.16 (8/6)
427.53 (7/22)
31.13 (7/23)
8.94 (7/18)
Anthracene
Avg
0.30 + 0.34
7.59 + 5.87
0.27+0.25
0.13 +0.06
Max
1.40 (7/17)
17.60 (7/16)
0.78 (7/23)
0.23 (8/5)
Fluorenone
Avg
1.02+0.43
12.08 + 7.30
1.11 +0.63
0.86+0.46
Max
2.00 (8/6)
23.42 (7/31)
2.39 (7/24)
1.99 (7/18)
Retene
Avg
0.27 + 0.15
0.58+0.25
0.57+0.31
0.45 ±0.22
Max
0.53 (7/16)
0.92 (7/18)
1.30 (7/11)
0.76 (7/18)
Fluoranthene
Avg
2.09+ 1.63
46.58 + 29.26
3.22+2.87
1.50 ±0.91
Max
7.13 (7/17)
109.70 (7/22)
8.80 (8/6)
3.72 (7/17)
Pyrene
Avg
1.11 + 1.17
23.58 + 14.37
1.60 + 1.52
0.77 ±0.42
Max
5.19 (7/17)
55.30 (7/22)
5.18 (8/6)
1.75 (7/17)
Benz(a)anthracene
Avg
0.25+0.66
3.02 + 2.78
0.26+0.39
0.14 ±0.15
Max
2.70 (7/17)
8.88 (7/22)
1.25 (8/6)
0.57 (7/17)
Chrysene
Avg
0.33+0.66
5.17 + 3.82
0.62+0.86
0.31 ±0.36
Max
2.77 (7/17)
12.96 (7/22)
2.50 (8/6)
1.32 (7/17)
Cyclopenta(c,d)pyrene
Avg
0.03 +0.03
0.22+0.16
0.09+0.15
0.05 ±0.05
Max
0.11 (8/6)
0.63 (8/2)
0.52 (8/6)
0.18 (7/17)
Benzofluoranthenes
Avg
0.58+ 1.21
10.16 + 9.55
0.91 + 1.24
0.57 ±0.71
Max
5.09 (7/17)
32.89 (7/31)
3.74 (8/6)
2.59 (7/17)
Benzo(e)pyrene
Avg
0.17 + 0.32
2.81 +2.61
0.25 +0.32
0.17 ±0.20
Max
1.35 (7/17)
9.09 (7/31)
1.03 (8/6)
0.74 (7/17)
Benzo(a)pvrene
Avg
0.26 + 0.60
3.04 + 3.88
0.25+0.31
0.14 ±0.16
Max
2.49 (7/17)
15.31 (7/31)
0.80 (8/7)
0.69 (7/17)
Indeno( 1,2,3-c,d)pyrene
Avg
0.30 + 0.53
3.90 + 3.17
0.41 +0.52
0.26 ±0.33
Max
2.26 (7/17)
10.21 (7/22)
1.63 (8/6)
1.32 (7/17)
Dibenzo(a.h)anthracene
Avg
0.19 + 0.16
1.39 + 0.89
0.21 +0.16
0.16 ±0.13
Max
0.76 (7/17)
3.24 (7/22)
0.58 (8/6)
0.60 (7/17)
Benzo(g, h, i)pery lene
Avg
0.23+0.39
3.32 ±2.35
0.33 +0.43
0.21 ±0.26
Max
1.68 (7/17)
7.99 (7/22)
1.39 (8/6)
1.02 (7/17)
Coronene
Avg
0.13+0.10
1.39 + 0.97
0.16+0.15
0.09 ±0.07
Max
0.34 (7/17)
3.90 (8/8)
0.43 (8/7)
0.31 (7/17)
5-48
-------
Naphthalene was the most abundant PAH at all four sites and
cyclopenta(c,d)pyrene was the least abundant. Results from studies of PAH
concentrations in Boston and Houston, reported by Kelly et al. (1992), were from
equivalent sampling and analytical protocols as those employed for LMUATS, allowing
direct comparison between these sites. The concentrations of 2- to 3- ring PAHs at IIT
were comparable to the levels measured in Boston during the Summer 1991 (Kelly et al.,
1992), while the concentration of the larger molecule PAHs was about five times higher at
IIT than at Boston (Table 5-10). In general, the concentration of PAHs in Chicago
appears to be significantly higher than those measured in Boston or Houston.
Table 5-10. Comparison of Urban Polynuclear Aromatic Hydrocarbons
Concentrations Measured in Three Cities (ng/m ).
IIT
Boston
Houston
Author
Time
This Study
Summer 91
Kelly et al.
Summer 91
Kelly et al.
Summer 91
Naphthalene
507
606
515
Acenaphthene
56
24.1
22.8
Fluorene
54
36.2
24.2
Phenanthrene
168
101
50.1
Anthracene
8
3
1.4
Fluorenone
12
10.2
5.4
Fluoranthene
47
16.7
10.2
Pyrene
24
8.4
7.9
Benz(a)anthracene
3
0.9
0.6
Chrysene
5
1
0.6
Cyclopenta(c,d)pyrene
0.22
0.3
0.2
Benzofluoranthenes
10
1.1
0.5
Benzo(e)pyrene
2.8
0.4
0.2
Benzo(a)pyrene
3.0
0.4
0.2
Indeno( 1,2,3 -c,d)pyrene
3.9
0.5
0.2
Dibenzo(a,h)anthracene
1.4
0.3
0.2
Benzo(g,h,i)perylene
3.3
0.6
0.3
Coronene
1.4
0.3
0.2
5-49
-------
Figure 5-30 shows the naphthalene concentrations measured at the four sites. The
maximum concentrations observed in Kankakee were associated with local stagnation and
with weak transport to the south of the site. Maximum concentrations were also reported
at IIT on the same days but they were lower in magnitude. The concentrations of
benzo(a)pyrene measured during the month are shown in Figure 5-31. The concentrations
are about an order of magnitude higher in Chicago than at any of the other sites.
The concentrations for five volatile PAHs measured on the R/V Laurentian were
comparable to the daytime average concentrations at Glendora, CA, approximately 20 km
northeast of Los Angeles and generally downwind of downtown Los Angeles. The
average concentrations observed on the R/V Laurentian were about five times higher than
the concentrations on Chesapeake Bay (Baker et al., 1992) except for benzofluoranthenes
and dibenzo(a,h)anthracene. Compared to the ambient concentrations measured on Lake
Superior as reported by Baker and Eisenreich (cited in Eisenreich and Strachan, 1992),
Table 5-11 shows the concentrations measured on Lake Michigan aboard the R/V
Laurentian were about five to fifty times higher, except for dibenzo(a,h)- anthracene,
which was about 500 times higher.
5-50
-------
Figure 5-30. Variations in Measured Naphthalene Concentrations.
Kankakee
963 825
~l—I—I—I—I TTTTTT TT I
16 17 IB IS 20 21 22 23 24
700
600
500
E 400
300
200
100
0
R/V Laurentian
"l—I—r
"i—i—i—i—i—i—i—r~T
In
1
1112
23 24
-—I—I—I—I—I—I—I—I T T I—I—I
5 6 7
o>
c
700 -i
600 -
500
400
300
200
1 00
0
South Haven
II
T-T-
4-r+
"i I—r~rn—r
16 17 18 162021 22 22 24 29 31
I T T 1 I
i r
12
5 6 7 B 9
July
Date
August
5-51
-------
Figure 5-31. Variations in Measured Benzo(a)pyrene Concentrations.
o>
c
10 -i
1 -
0.1 -
0.01
Kankakee
i i i—m—i i i
16 17 IB 19 20 21 22 23 24
I I I
1 T I T I I I ii I I I I
29 31 2 5 8 7 0
15
10
NT
cn
c
1 -
0.1 -
0.01
_l—,—!—!—r
"I TTTTTTTTT I I I I I T I I I T~T""
16 17 18 19 20 21 2223 24 29 31 2 5 6 7 B
10
R/V Laurentian
c
0.1 -
0.01
t—i—i n i
1112
Jl
Jl
t—i—i—i—i—i—i—i T T i—r
5 6 7
i i i i t r
23 24
July Date August
5-52
-------
Table 5-11. Comparison of PAH Concentrations Measured On or Near the Great
Waters (Units are ng/m^).
Lake
South
Lake
Lake
Chesapeake
Michigan
Haven
Superior
Superior
Bay
Author/Source
This Study
This Study
Baker, 1992
Hites, 1992
Baker, 1992
Period of Study
Summer '91
Summer '91
1986
1981
1990
Naphthalene
119
66
Acenaphthylene
1.4
0.49
Acenaphthene
2.3
1.1
Fluorene
7.2
3.5
0.5
0.7
Phenanthrene
10.8
4.8
2.7
2.1
2.1
Anthracene
0.27
0.13
0.004
0.06
Fluorenone
1.1
0.86
Retene
0.6
0.5
Fluoranthene
3.2
1.5
0.5
Pyrene
1.60
0.8
0.3
0.2
0.5
Benz(a)anthracene
0.26
0.14
0.03
0.008
0.05
Chrysene
0.62
0.31
0.06
0.2
0.1
Cyclopenta(c,d)pyrene
0.09
0.05
Benzofluoranthenes
0.91
0.57
0.05
0.2
Benzo(e)pyrene
0.25
0.17
0.006
0.009
0.08
Benzo(a)pyrene
0.25
0.14
0.004
0.003
0.04
Indeno( 1,2,3-c.d)pyrene
0.41
0.26
0.01
0.007
0.07
Dibenzo(a,h)anthracene
0.21
0.16
0.0004
0.01
Benzo(g,h,i)pervlene
0.33
0.21
0.02
0.005
0.08
Coronene
0.16
0.09
5-53
-------
5.6. Average PCB Concentrations
Atmospheric concentrations of total PCBs measured at IIT were 2.6-3 .3 times
higher than the total PCBs measured at the other 3 LMUATS sites (Table 5-12). In
addition, total PCB concentrations measured at IIT were slightly higher than many of the
other PCB measurements made in the Great Lakes region during the last five years (Table
5-13). Other investigations in the Great Lakes Basin in which elevated total PCBs were
measured include a study by Hermanson in northern Michigan (reported in Eisenreich and
Strachan, 1992) in which levels of total PCBs ranged from 31-1,860 pg/m3 and an
investigation by Hornbuckle et al. (1992) in Green Bay, WI where total PCBs measured
over the water ranged from 160-2200 pg/m3.
Average concentrations of total PCBs measured during LMUATS were slightly
greater over Lake Michigan on the R/V Laurentian than the two rural land-based sites
(Kankakee and South Haven). This same observation was reported by Hornbuckle et al.
(1992) for Green Bay, WI. However, in their study, even larger differences in over-water
vs. land-based concentrations were observed. The day-to-day variations in the total PCB
concentrations measured are seen in Figure 5-32. Total PCB levels observed at
Kankakee, on the R/V Laurentian and in South Haven ranged from 639 to 808 pg/m .
Other reported values for measurements made in Michigan and in other Great Lakes states
are similar to, or lower than, the concentrations observed in this study.
Analysis of atmospheric samples for PCBs has historically suffered from poor
source profile information, poor analytical and inter laboratory analytical comparisons and
marked differences in PCB profiles between source and receptor locations. Individual
PCB congeners exhibit widely varying physical and chemical properties, which determine
their preferential phase partitioning. As a result of this lack of information, source-
receptc. relationships for PCB congeners are very poorly understood. Further
complicating the ability to interpret source-receptor relationships, is the lack of
standardization of PCB congeners quantified and reported.
In the LMUATS PCB samples, 8 congeners were quantified: IUPAC Nos: 1, 5,
29, 50, 87, 154, 188 and 201 (Table 5-12). Since the numbers and the types of PCB
congeners measured are different between studies, few data are available for comparison.
In a study by Hoff et al. (1992) in Egbert, Ontario, PCB 87 levels measured were much
lower than those quantified at the 4 LMUATS sites. However, Hoff reports levels of PCB
201 of 0.92 pg/m from Egbert, which is almost identical to the concentration of this
congener measured at IIT.
5-54
-------
Table 5-12. Ambient PCB Concentrations (pg/nr*).
Kankakee
OT
Laurentian
South Haven
2-PCB
Avg
1± 2
6+21
25+ 66
13+52
Max
5 (8/7)
85 (7/19)
233 (8/7)
234 (8/7)
TOT. mono-PCB
Avg
60+ 49
77+ 68
164+113
101+ 105
Max
138 (7/16)
267 (7/16)
356 (7/11)
428 (7/10)
2,3-PCB
Avg
12+ 18
94+57
30+ 42
28+36
Max
51 (7/16)
189 (7/19)
160 (7/24)
171 (7/18)
TOT. di-PCB
Avg
83+ 106
186+ 124
107+ 73
82+60
Max
267 (7/19)
425 (7/18)
252 (7/24)
191 (7/19)
2,4,5-PCB
Avg
0.03+ 0.13
0.05+0.19
0.02+ 0.09
0.00+0.00
Max
0.51 (7/23)
0.74 (8/6)
0.32 (8/6)
0.00 (8/8)
TOT.tri-PCB
Avg
115+59
422+ 255
89+ 47
84+ 67
Max
254 (8/5)
779 (7/20)
199 (7/24)
338 (8/8)
2.2',4,6-PCB
Avg
0.7+2.2
0.2+0.4
0.4+0.6
0.9+2.6
Max
8.4 (8/2)
1.2 (7/24)
1.7 (8/7)
11.5 (8/8)
TOT.tet-PCB
Avg
92+ 56
463+ 262
101+36
86+57
Max
253 (7/18)
855 (7/23)
174 (7/24)
285 (8/8)
2,2',3,4,5'-PCB
Avg
11+24
47+37
16+7
9± 7
Max
95 (7/18)
134 (7/19)
30 (7/24)
25 (7/17)
TOT.pen-PCB
Avg
212+ 306
707+ 479
290+163
263+ 127
Max
1296 (7/18)
1744 (7/19)
606 (7/24)
519 (7/17)
2.2'.4,4',5,6'-PCB
Avg
0.2+0.3
0.3+0.4
0.1+0.2
0.1+0.2
Max
0.9 (7/23)
1.1 (7/16)
0.5 (7/24)
0.6 (7/15)
TOT.hexa-PCB
Avg
68+ 145
230+ 171
50+28
50+32
Max
589 (7/18)
660 (7/19)
101 (7/11)
137 (7/17)
2.2'3,4',5,6,6'-PCB
Avg
0.1+0.3
0.00+0.00
0.00+0.01
0.1+0.1
Max
1.3 (7/23)
0.00 (8/8)
0.04 (7/11)
0.5 (7/17)
TOT.hep-PCB
Avg
6.7+9.8
42+33
7± 5
5± 4
Max
41 (7/18)
108 (8/2)
16 (8/6)
13 (7/17)
2,2',3,3',4,5',6,6'-PCB
Avg
0.3+0.9
0.9+0.7
0.01+0.03
0.1+ 0.2
Max
3.4 (7/23)
2.2 (8/2)
0.1 (7/11)
0.8 (7/17)
TOT.octa-PCB
Avg
0.4+ 0.5
10+9
0.6+0.7
0.3+ 0.5
Max
1.6 (7/18)
27 (8/2)
2.2 (7/24)
2.2 (7/17)
TOT.nona-PCB
Avg
0.6+ 1.6
1.9+3.3
0.3+0.5
1.2+ 3.2
Max
6.2 (7/23)
11 (7/24)
1.6 (8/7)
11.9 (8/8)
deca-PCB
Avg
0.9+3.3
0.2+0.9
0.00+ 0.00
0.1+0.6
Max
12.6 (7/23)
3.8 (7/24)
0.00 (8/7)
2.5 (7/12)
Total PCBs
Avg
639+602
2139+1214
808+357
672+ 279
Max
2669 (7/18)
3999 (7/19)
1594 (7/24)
1216 (7/17)
5-55
-------
Table 5-13. Polychlorinated Biphenyls Measured in the Great Lakes Basin
(Units are pg/m"*).
Author, period of study
Location
EPCB
LMUATS, 1991
Kankakee, IL
639
Chicago, IL
2139
R/V Laurentian
808
South Haven, MI
672
Keelerefa/. (1993), 1993
Pellston, Michigan
24-860
(Unpublished data)
South Haven, Michigan
60 - 620
Deckerville, Michigan
91-399
Ann Arbor/Dexter, Michigan
83-640
Hermanson (1992), 1990-1991
Michigan
31 - 1,860
Holsenero/., 1991
IIT
940
Sweet (1992), 1989-1990
Michigan
120
Hornbuckle et al. (1992), 1991
N. Green Bay
100-300
Hornbuckle and Eisenreich (1992), 1991
Michigan
160 - 1,900
McConnell (1992), 1990
Huron
180
Hornbuckle and Eisenreich (1992), 1991
Huron
120 - 200
McConnell (1992), 1990
St. Clair, Erie
410, 520
Hornbuckle (1992), 1991
St. Clair, Erie
570,470
Sweet (1992), 1990-1991
Pt. Petre
310
Hornbuckle (1992), 1991
Ontario
440 - 600
Hoff, Muir, Grift (1992) 1988 - 1989
Great Lakes
380 - 470
Hoff, Bidleman (1992), 1988 - 1989
Great Lakes
78 - 290
McConnell (1992), 1988 - 1989
Great Lakes
80 - 170
Murphy (1992), 1985
Chicago (1980-1981 n=4)
5,500 +/-2000
Manchester-Neesvig and Andren (1989),
Northern Wisconsin
135 winter to
1989
1,820 summer
5-56
-------
Figure 5-32. Variations in Measured Total PCB Concentrations.
4000
3500 -
3000 -
£ 2500 -
c> 2000 -
1500 -
1000 -
500 -
0
Kankakee
I ! I I I T I I I f I I I I I ¦ I I f I
I I I 1 I—r—t—i—I 1 I l I I 1—I—I T I T I T I—I T T T T I—I
1617181920 222324 29 31 2 5 6 7 8
4000
3500 -
3000
E 2500 -
\
2000 -
a.
1500
1000 -
500 -
0
I IT
"I l I I—I I l I—TTTTTTTTT I I I I T I I I I I I I I T
16 17 16 19 20 21 22 23 24 29 31 2 5 6 7 8
111
3500
g
\ 2000
CJl
500
0
R/V Laurentian
1 ~! t i l ^ l—r—t—i—i i—r—i—i—i—Jl
1 Mini Jl'll i i
11 12
23 24
5 6 7
4000 -i
3500
3000 H
"e 2500
2000 -
a
1500 -
1000 -
500
0
South Haven
I I 7
ll «'l ¦ I' I
10 1112 13 14 15 16 17 18 19 2021 2324
July
Date
29 31 2 5 6 7 8
August
5-57
-------
During LMUATS, the concentration of congener 201 was three times lower in
Kankakee than IIT and was even lower at South Haven and not detected on the R/V
Laurentian. Schwackhamer et al. (1988) report levels of particulate 201 of 0.33 pg/m3,
similar to the total 201 measured at Kankakee. The particulate 201 measurements by
Schwackhamer et al. were made during the summer at Siskewit Lake on Isle Royal, a
remote island in Lake Superior. In the same investigation, the authors report summertime
levels of total PCBs that are almost 5 times higher than winter levels (2.9 vs. 0.62 ng/m3),
with the summer values exceeding those measured in Chicago during the LMUATS.
At IIT the di- to hexa-PCBs were the dominant forms present, however, high
concentrations of larger PCBs (hepta- and octa-PCB) were observed. Murphy et al.
(1985) measured predominantly tri- to hexa-PCB from incinerator stack (sewage sludge
and municipal incinerator) emissions in the Midwest.
Holsen et al. (1991) report data from samples collected at IIT in Chicago, peaks of
tri- and hexa-PCB homologues independent of season. In Summer, penta-PCBs also were
high in the Holsen study (and during Spring hepta-PCBs were high). At IIT in LMUATS,
PCB 5 and 87 were highly elevated. The most-elevated homologues quantified in samples
during the LMUATS were the di-, tetra- and penta-PCBs. Penta-PCB concentrations at
IIT were almost 2 times higher than other homologue groups.
In the Manchester-Neefig and Andren (1989) report, congeners 28+31 were
highest in concentration, followed by 52, 153+105+132, then 118. During LMUATS,
these congeners were not measured, however, the dominant fraction of congeners that
were measured were the mono-PCBs at South Haven and on the R/V Laurentian. The
mono-PCBs were in high concentration at IIT and Kankakee as well, but also with a
strong influence of a source of penta-PCB (87), which is observed at much lower levels in
South Haven and on the R/V Laurentian.
5-58
-------
5.7.
Average Pesticide Concentrations
Pesticide concentrations and isomer ratios measured during LMUATS may be
valuable tools in discerning urban vs. rural inputs in the transport of these and other toxic
compounds. Concentrations of the individual pesticides quantified in LMUATS samples
are listed by site in Table 5-14 with the maximum value for each sample recorded by site
and the date the maximum value occurred. Several studies have demonstrated the
temperature/seasonal dependence of atmospheric pesticide levels (Table 5-15). Since
LMUATS sampling occurred during the summer, the levels of the more volatile pesticides
as well as those still in use may be elevated above levels reported for yearly averages for
areas of the Great Lakes Basin.
Summertime a-HCH values measured at LMUATS sites were similar to or slightly
lower than the consensus average of 300 pg/m3 reported by Eisenreich and Strachan
(1992) for the Great Lakes Basin. The highest average value for this compound occurred
at Kankakee (268 pg/m^), the site at which a peak value of 1,243 pg/m3 was observed on
July 29, 1991 (Figure 5-33).
Average concentrations of y-HCH measured over the water and at South Haven
(103 and 90 pg/m , respectively) were very similar to the average summertime value of
100 pg/mJ reported for other rural Great Lakes locations. However, at IIT y-HCH
concentrations were almost one half the value at the other sites (58 pg/mJ) and at
Kankakee, the average level (180 pg/m3) measured was more than 1.5 times higher than
the average values of over-water and South Haven measurements. Figure 5-34 shows that
peak values of y-HCH on the R/V Laurentian were 3 times and 5 times the average
summertime concentrations for the Great Lakes region, while at IIT, the peak value of y-
HCH measured only slightly exceeded the 'consensus' average (Eisenreich and Strachan,
1992). At Kankakee, the peak value of y-HCH measured during LMUATS was 936
pg/m , strongly suggesting a large local source component.
5-59
-------
Table 5-14. Ambient Pesticides Concentrations (pg/m3).
Kankakee
IIT
Laurentian
South Haven
a-HCH
Avg
268+60
111.86+39
169.20+205
128.14+27
Max
1243 (7/29)
169 (7/23)
821 (8/5)
214 (7/24)
Hexachlorobenzene
Avg
139+ 204
68+40
104+154
54.90+23
Max
692 (7/29)
195 (7/19)
606 (8/5)
111 (7/29)
Atrazine
Avg
373+461
183+153
286+272
223.12+189
Max
1504 (7/22)
418 (7/31)
806 (8/5)
805 (7/15)
y-HCH(lindane)
Avg
180+ 251
58+36
103+100
90+104
Max
936 (8/6)
131 (7/21)
299 (7/12)
522 (7/29)
Alachlor
Avg
549+1186
76+208
57+205
26+65
Max
4713 (8/6)
610 (7/29)
739 (7/23)
190 (7/19)
Mircx
Avg
43+ 37
44+26
9+7
11+7
Max
151 (8/6)
114 (7/21)
24 (7/12)
33 (8/8)
Aldrin
Avg
1.0+1.3
6±5
0.3+0.4
0.4+0.6
Max
3.9 (8/8)
15 (8/8)
1 (7/11)
2 (8/7)
Metolachlor
Avg
184+182
8+27
10+29
42+65
Max
551 (7/29)
110 (7/29)
104 (7/12)
251 (7/23)
trans-nonachlor
Avg
65+40
55+29
12+7
31.2+21
Max
153 (8/6)
100 (7/18)
27 (7/12)
77 (7/17)
Dieldrin
Avg
371+302
159+76
46+25
169+90
Max
1176 (8/6)
271 (8/8)
111 (8/6)
402 (7/21)
Simazine
Avg
90+112
103+198
85+116
54+141
Max
319 (7/24)
637 (8/21
345 (7/24)
587 (8/2)
Chlorpynfos
Avg
83+75
44+26
12+10
88+122
Max
258 (7/19)
113 (7/21)
32 (8/6)
514 (7/13)
y-Chlordane
Avg
113+69
83+49
27+14
45+23
Max
288 (8/6)
182 (7/18)
49 (8/7)
90 (7/21)
a-Chlordane
Avg
88+54
76+39
22±9
42+22
Max
226 (8/6)
166 (7/8)
40 (7/23)
85 (7/20)
4,4'-DDT
Avg
10+16
183+133
33+34
339+167
Max
54 (8/8)
530 (8/2)
121 (8/6)
574 (7/20)
P.P'-DDE
Avg
36+26
119+99
57+63
1331+351
Max
95 (8/6)
395 (8/8)
244 (7/24)
1944 (8/2)
P,P'-DDD
Avg
3+6
4 ±6
3 ±5
12+7
Max
19 (8/6)
21 (7/31)
20 (8/7)
27 (8/7)
5-60
-------
Table 5-15. Concentrations of Pesticides in Rural Locations in the Great Lakes
Region (Consensus Values from Eisenreich and Strachan, 1992).
a-HCH, Summer 300
a-HCH, Fall & Spring 200
a-HCH, Winter 100
y-HCH, Summer 100
y-HCH, Fall & Spring 40
y-HCH, Winter 40
HCB, Winter 100
HCB, Other 150
L-Chlordane, Summer 80
S-Chlordane, Fall 30
Z-Chlordane, Winter 20
X-Chlordane, Spring 40
I-DDT, (mainly DDE), Summer - Superior. 50
Michigan, Huron
L-DDT. (mainly DDE), Summer - Erie, Ontario 100
Z-DDT, (mainly DDE), Other - all lakes 20
Dieldrin, Winter 20
Dieldrin, Other 80
Atrazine, May & June 5,000
Atrazine, Other 10
5-61
-------
Figure 5-33. Variations in Measured a- HCH Concentrations.
350 i
300
250
i
^ 200 -
o>
q. 150
100
50
0
Kankakee
1243
632
"i I tt T~"T i i i r
16 17 16 19 20 22 23 24
)T IT I T I I T T T T 1 I
29 31 2 5 6 7 0
350
300
, 250
\ 200 -
CP
150
1 00
50
0
I IT
1—I—I I I I I T
1 6 1 7 1 e 19 20 21 22 23 24
"1 I T T I 1
29 31 2 5 6 7 6
350
300
250
200
a. 150
100
50
0
R/V Laurentian
821
I
"i—rrri—i—i—i—i—i—i—i—i—r
1 1 12 23 24
i—i—i—i—i—I—r
H
5 6 7
350 -i
300
n 250
\ 200 -
a. 150 -
100 -
50 -
0
South Haven
T TTTTTTTTTTTT I T- I I I I T~T ; T I I T T T T I I
10 1 1 12 13 14 15 16 17 18 19 20 21 23 24 29 31 2 5 6 7 8
July
Date
August
5-62
-------
Q.
200
180
160
140
120
100
SO
50
40
20
0
Figure 5-34. Variations in Measured y- HCH Concentrations.
„ . . 405 293 454 936
Kankakee
~i—I—I—i—I I I I
16 17 16 19 20 222324
~i i i r~T i
29 31 2
I T l I
5 e 7 B
200
180
160
n 140
^ 120
a> 100
80
60
40
20
0
200
160
1 60
1 40
'e 120
\ 100
cn
cl 80
60
40
20
0
I IT
1 i i i 11 11
"I l 1 I I l T T T I— T I I I r~T I T—l T"
16 17 IB 19 20 21 22 23 24 29 31 2
299
T—i—i—I—r—r~I
5 6 7 B
255 258
-
R/V Laurentian
II
i i i I i I i I i I
i i i i i i i I l i I f
II
u.
1112
23 24
5 6 7
200
180
160
r, 140
E 120
"oi 100
a 80
60
40
20
0
522
South Haven
il
~i rn i i i i i i i i i i i i r
10 1112 13 14 15 16 17 18 19 20 21 23 24
1 I I I I I I I I I I I i
29 31 2 5 6 7 8
July
Date
August
5-63
-------
In general, previous investigations in the northern Hemisphere report higher
concentrations of a-HCH than y-HCH (Table 5-16). Typically, the a-/y-HCH ratios in the
northern Hemisphere are greater than 5 (Tatsukawa et al., 1990). Samples analyzed for
HCH during the LMUATS investigation revealed a-/y-HCH ratios of 8.3 at Kankakee,
8.8 at IIT, 8.6 on Lake Michigan, and 7.1 in South Haven. The lower ratio at South
Haven may reflect the use of Lindane in the vicinity (Lindane contains predominantly y-
HCH) versus the more commonly used technical HCB (which contains a higher
percentage of a-HCH than y-HCH). This theory is also supported by the ratio of peak a-/
y-HCH at South Haven of 0.4 (peak a-HCH occurred on 24 July and peak y-HCH
occurred on 29 July).
Table 5-16. Atmospheric Concentrations of a- and y- Hexachlorocyclohexanes in
3
the Great Lakes Basin (Units are pg/m ).
Author, period of study Location a-HCH y-HCH
Average Average
LMUATS, 1991
Kankakee, IL
268
180
Chicago, IL
112
58
R/V Laurentian
169
103
South Haven, MI
128
90
Reid (1992), 1989-1990
Lake Superior, Turkey Lake
0-20
0-50
Hornbuckle and Eisenreich
Lake Michigan, Green Bay
40-460
24-240
(1992), 1989
Reid (1992), 1989-1990
Lake Huron
0-270
0-130
Reid (1992), 1989-1990
Inland (Dorset)
0-500
0-500
Reid (1992), 1989-1990
Lake Erie
0-400
0-130
Reid (1992), 1989-1990
Ontario (Pt. Petre)
0-560
0-200
Lane, Hoff, Bidleman
Great Lakes, Summer
220-430
40-130
(1992), 1985-1990
Lane, Hoff, Bidleman
Great Lakes, Autumn
170-230
28-38
(1992), 1985-1990
Lane, Hoff, Bidleman
Great Lakes, Winter
70-190
13-40
(1992), 1985-1990
Lane, Hoff, Bidleman
Great Lakes, Spring
110-290
75-130
(1992), 1985-1990
5-64
-------
The average level of hexachlorobenzene (HCB) at Kankakee (139 pg/m3) is also
elevated above the average concentration measured at the other three sites. The value
reported here is also at the top end of the ranges of HCB reported from other
investigations in the Great Lakes Basin (Table 5-17) The maximum value of HCB
measured at Kankakee (642 pg/m3) during LMUATS occurred on 29 July and is one of
the highest concentrations of this compound reported for a rural location in the Great
Lakes Basin. The maximum value of 606 pg HCB/m was measured over Lake Michigan
on 5 August, 1991. The mixed-layer trajectories for this day revealed flow reaching the
vessel initially from the east and then to the north traversing southwestern Michigan.
Average HCB concentrations measured at IIT and SHA (68 and 55 pg/m3) were
significantly lower than levels at Kankakee and on the R/V Laurentian. However, on the
R/V Laurentian the mean is heavily weighted by the maximum HCB value.
Chlordane has been distributed widely in the U.S. as a potent pesticide, one of its
uses being primarily termite control (Bidleman et al., 1990). The concentrations of a-
chlordane measured during LMUATS are quite similar (by site) to the concentrations of y-
chlordane. Mean and maximum a- and y-chlordane concentrations at Kankakee are
slightly elevated above those concentrations measured at IIT. Concentrations of the two
isomers at IIT and Kankakee are elevated above total chlordane levels reported in the
Great Lakes Basin (Table 5-18). At IIT this may be due, in part, to the extensive use of
chlordane in the control of indoor insects, primarily termites.
Measurements of chlordane in South Haven and over Lake Michigan on the R/V
Laurentian are similar to other levels reported in the region. Since chlordane has been
banned in the U.S. for several years, the presence of this compound in the atmosphere
represents revolatilization of soil-bound chlordane, loss from stockpiled/stored sources, or
long distance transport.
The ratio of y (trans)- to a (cis)-chlordane yields information concerning the
nature and age of the air mass and source of the chlordane measured. A y/a ratio greater
than 1 indicates a regional source or local emission of chlordane, whereas a y/a ratio of
less than 1 indicates a transported and more-aged air mass. Ratios of the average
concentrations of each of these isomers measured during LMUATS were greater than one
at all four sites (1.3 at Kankakee, 1.1 at IIT, 1.2 on the R/V Laurentian and 1.1 in South
Haven).
5-65
-------
Table 5-17. Atmospheric Concentrations of HCB in the Great Lakes Basin
3
(Units are pg/m ).
Author, period of study Location
Average HCB
LMUATS, 1991 Kankakee, IL
139
Chicago, IL
68
R/V Laurentian
104
South Haven, MI
55
Kelly, et al. (1991) Lake Erie, Summer
50
Reid (1992), 1989-1990 Lake Superior, Turkey Lake
0-220
Hornbuckle and Eisenreich (1992), Lake Michigan. Green Bay
10-80
1989
Bidleman (1992), 1989 Michigan, Green Bay
160
Reid (1992), 1989-1990 Lake Huron
0-100
Reid (1992), 1989-1990 Inland (Dorset)
0-130
Reid (1992), 1989-1990 Lake Erie
0-180
Reid (1992), 1989-1990 Ontario (Pt. Petre)
0-230
Lane, Hoff, Bidleman (1992), 1985- Great Lakes, Summer
40-130
1990
Lane, Hoff, Bidleman (1992), 1985- Great Lakes, Autumn
28-38
1990
Lane, Hoff, Bidleman (1992), 1985- Great Lakes, Winter
13-40
1990
Lane, Hoff, Bidleman (1992), 1985- Great Lakes, Spring
75-130
1990
5-66
-------
Table 5-18. Atmospheric Concentrations of Total Chlordane in the Great Lakes Basin
3
(Units are pg/m ).
Author, period of study
Location
Average (pg/m3)
LMUATS, 1991
Kankakee, IL
201
Chicago, IL
159
R/V Laurentian
49
South Haven, MI
87
Reid (1992), 1989-1990
Lake Superior
0-42
Bidleman (1992), 1989
Green Bay, WI
9
Bidleman (1992), 1990
Green Bay, WI
42
Reid (1992), 1989-1990
Lake Huron
0-50
Reid (1992), 1989-1990
Dorset
0-105
Hoff (1992), 1988-1989
Egbert, Ontario, Summer
76
Hoff (1992), 1988-1989
Egbert, Ontario, Autumn
26
Hoff (1992), 1988-1989
Egbert, Ontario, Winter
15
Hoff (1992), 1988-1989
Egbert, Ontario, Spring
41
Reid (1992), 1989-1990
Lake Erie
0-182
Reid (1992), 1989-1990
Ontario, Pt. Petre
0-110
Bidleman (1992), 1990
Great Lakes, Summer
187
5-67
-------
Three chlordane-related and derivative pesticides were quantified in LMUATS
samples. These include trans-nonachlor, alachlor and metalachlor. Average trans-
nonachlor concentrations were lowest over Lake Michigan (12 pg/m3) and were similar at
Kankakee and IIT and were slightly lower in South Haven (65, 55 and 31 pg/m3). Peak
values for this compound were twice that observed in South Haven and 1.5 times the peak
value observed at IIT.
Alachlor and metalachlor have been listed as the two most heavily used pesticides
by weight in the Great Lakes Basin for 1986 (Baker and Richards, 1990). These
chlorinated compounds were used as herbicides in large agricultural areas. The levels of
Alachlor varied greatly at each of the LMUATS sites, by as much as an order of
magnitude. The highest concentrations of this herbicide were recorded from samples
collected at Kankakee where the average concentration was 549 pg/m3 and the peak value
for the study period was 4,713 pg/m3. The Alachlor concentration was lower at IIT by
almost an order of magnitude and even lower in over-water measurements. The lowest
values for alachlor were recorded at South Haven. While the concentration of this
compound decreased away from Kankakee at the measurements sites in this study, there
were occasions of highly elevated alachlor concentrations recorded at both IIT and over
Lake Michigan. These peak events occurred on 29 July and 23 July and represented
atmospheric concentrations of 610 and 739 pg/m3, respectively
Similar behavior, was observed for levels of Metalachlor, but the concentrations in
South Haven were also elevated. Average Metalachlor levels at Kankakee were higher
than those at the other sites by a factor of 2 to 25. Maximum concentrations of this
herbicide were recorded in Kankakee and IIT on the same sampling date (29 July)
However, the concentration observed at IIT on this day was 110 pg/mJ and the
concentration recorded at Kankakee was 551 pg/m . Maximum values recorded on Lake
Michigan and in South Haven (on 12 July and 23 July, respectively) suggest a local
source (or sources) in the South Haven area.
Figure 5-35 displays the time variation in the 4,4—DDT levels measured during the
month of study. DDT levels measured during the LMUATS study reveal alarming
information. Table 5-19 shows that the concentrations of the parent 4,4—DDT compound
in samples collected at South Haven are significantly elevated above other reported values.
The levels of the daughter compound. p,p'-DDE, shown in Figure 5-36, are an order of
magnitude greater in concentration than previously reported total DDT levels. These high
levels have been confirmed by further studies at this site during 1992-1993 (Keeler,
5-68
-------
unpublished data). The average DDT:DDE concentration ratio measured in South Haven
is 0.3, providing sufficient evidence that levels observed are due to release of previously
applied DDT since the concentration of the DDT breakdown product is three times that of
the parent compound. The time for degradation of DDT to DDE is slow enough to
implicate re-emission of a much earlier DDT application as the source of this compound.
At IIT, DDT and DDE levels measured were elevated (183 and 119 pg/m3,
respectively). Few reports document total DDT levels similar to those observed at IIT.
Concentrations of the DDT metabolite, DDD, shown in Figure 5-37 was significantly
lower than 4,4-DDT and p,p'-DDE at all the LMUATS sites, and represents a small
fraction of the ZDDT reported in other studies.
At IIT, the ratio of the average DDT:DDE concentration is 1.5, indicating that a
source of new DDT is present. The transport of new DDT is partially evident in the over-
water measurements since on the R/V Laurentian the average DDT:DDE concentration is
0.6. At Kankakee, the DDT:DDE ratio is 0.3 indicating a previously applied source of
this toxin. Current DDT:DDE ratios in air from the Great Lakes region are typically less
than 1. Nevertheless, there are some situations in which DDT:DDE is greater than 1 in
North America. McConnell et al. (1992) gives ratios of DDT DDE for samples collected
in 1989-1990 in the Great Lakes region. One sample taken in Green Bay during June,
1989 showed a DDT.DDE ratio of 1.8. This sample was also enriched in toxaphene and
chlordane, and was associated with airflow from the S-SW. Another air sample from
Lake St. Clair in August 1990 contained unusually high levels of DDT with a DDT:DDE
ratio of 2.4. Although preliminary, these observations suggest that pulses of "new" DDT
are being superimposed on a North American background of atmospheric residues that are
largely DDE.
In a separate study, Rapapon et al. (1985) hypothesized that since the DDT in
peat core samples from the Great Lakes region and eastern Canada was largely p,p'-DDT
and o,p'-DDT, it was freshly deposited and most likely the result of transport from current
use in Mexico and Central America. However, the recent finding of elevated DDT levels
in only two of the four sites in LMUATS (IIT and South Haven) indicates a local source.
Furthermore, the elevated DDT levels in Michigan seem to be isolated to the southwestern
corner of the state, based on this study and recent measurements made at four sites in
lower Michigan (Keeler, unpublished data). It is difficult to conceive of a Mexican or
Central American source that selectively impacts only one corner of the state.
5-69
-------
C7I
Q.
550
500
450
400
350 H
300
250 H
200
150 -
100 -
50 -
0
Figure 5-35. Variations in Measured DDT Concentrations.
Kankakee
l1 l i—[—i—I I l l I T T T I I I T I I I I I I T l l l l l l i i—rn
16 17 18 19 20 22 2324 29 31 2 5 6 7 B
4+
550
500
450
, 400
E 350
\ 300
2 250
200
150
100
50
0
NT
i - i i T T 1——r
29 31 2 5 6 7 8
i I i i i r
1 6 17 IB 19 20 21 22 23 24
550
500
450
400
> 350
^ 300
a- 250
200
1 50
100
50
0
R/V Laurentian
i !v i—i—i—i—i—i—r
1112
"i—i—i—r
4i
23 24
5 6 7
550
500
450
400
'g 350
\ 300
S 250
200
150
100
50
0
South Haven
574 552
t i i i r
10 11 12 13 14 15 16 17 18 19 2021 23 24
29 31
July
Date
I T T
5 6 7 8
August
5-70
-------
Table 5-19. Atmospheric Concentrations of DDT in the Great Lakes Basin.
Author, period of study
Location
Average
(particulate + vapor)
Z DDT (pg/m3 )
LMUATS, 1991
Kelly et al. (1991)
Kelly et al.( 1991)
Reid (1992), 1989-1990
Bidleman (1992), 1989
Hornbuckle and Eisenreich (1992),
1989
Reid (1992), 1989-1990
Reid (1992), 1989-1990
Hoff (1992), 1988-1989
Hoff (1992), 1988-1989
Hoff (1992), 1988-1989
Hoff (1992), 1988-1989
Reid (1992), 1989-1990
Reid (1992), 1989-1990
Bidleman (1992), 1990
Kankakee, IL
Chicago, IL
R/V Laurentian
South Haven, MI
Lake Erie, Summer
Lake Erie, Winter
Lake Superior, Turkey Lake
Lake Michigan, Green Bay
Lake Michigan, Green Bay
Lake Huron
Inland (Dorset)
Inland (Egbert), Summer
Inland (Egbert), Autumn
Inland (Egbert), Winter
Inland (Egbert), Spring
Erie
Ontario, Pt. Petre
Great Lakes, Summer
10
183
33
339
140
30
0-70
24
20-70
0-50
0-30
180
80
20
100
130-270
0-120
120
5-71
-------
Figure 5-36. Variations in Measured DDE Concentrations.
01
Q.
10000
1000 H
100
10 -
Kankakee
~l I I I I I I I 7 T T T T I I-
18 17 18 19 20 22 24
1 I I ! I I I I I ————I—I
29 31 2 5 6 7 B
10000
1000 -
cn 1 00 -
a.
10 -
I IT
1 i i—r
"i i i i
16 17 IB 19 20 21 22 23 24
~i i i i i r
29 31 2
T T I I
5 6 7 B
10000 -I
1000 -
tr>
a.
100
10 -
R/V Laurentian
"i—i' T"' i—r
1112
i—i—r
I
23 24
5 6 7
1 0000
1000 -
100 —
a.
10
South Haven
~l T"~T
10 11 12 13 14 15 16 17 18 18 20 21 2324
29 31
I I I I I
2 5 6 7 8
July
Date
August
5-72
-------
Figure 5-37. Variations in Measured DDD Concentrations.
0>
Q.
30
25
20
15
10
5
0
Kankakee
~\—I—I—r
~\—l T ! I l—I l l l I l—I ItI I ItI iTTTl I l
16171819 20 22 24 29 31 2 5 6 7 B
o>
a.
30
25 H
20
15
10
5
0
NT
i—i—i—i—r
4
—i ii i i r
1 6 1 7 1 8 1 9 20 21 22 23 24
I I I I I
29 31 2
T I T I I
5 6 7 B
30
R/V Lourentian
25 -
20 -
E
\ 15 -
tJ!
Q.
1 0
5 H
0
"i—i—r
n—i—i—i—i i T
1112
23 24
5 6 7
E
Q.
30
25 -
20 -
15
10
5 H
0
South Haven
4+
TTTTTTTTTTTT—I I i r
10 1 1 12 13 14 15 16 17 18 19 2021 2324
I I T
29 31
h
July
Date
5 6 7 8
August
5-73
-------
Dieldrin, a highly toxic pesticide, was quantified in LMUATS samples. Typical
concentrations for this pesticide measured in other samples collected in the Great Lakes
-> -i
Basin range from 20 pg/m in winter to 80 pg/m"3 in samples collected during other times
of the year (Table 5-20). In one set of samples collected from Egbert, Ontario, Hoff (cited
in Eisenreich and Strachan, 1992) reports Dieldrin levels from 23 - 620 pg/m3. This range
is closer to those values obtained during LMUATS sampling. The Dieldrin concentration
averaged 371 pg/m3 in Kankakee with a maximum of 1176 (on 6 August). Dieldrin
concentrations at IIT, over Lake Michigan and in South Haven were all lower than in
Kankakee, with levels at IIT and South Haven being roughly equivalent (159 and 169
pg/m , respectively). The concentration of Dieldrin in samples collected over Lake
Michigan averaged 46 pg/m with a peak value of 110 pg/m . These later levels closely
reflect rural concentrations reported by other investigators and indicate that Dieldrin is not
effectively transported from potential source locations.
Table 5-20. Atmospheric Concentrations of Dieldrin Measured in the Great
Lakes Basin.
Author, period of study
Location
Average
(pg/m3)
LMUATS. 1991
Kankakee. IL
371
Chicago, IL
159
R/V Laurentian
46
South Haven. MI
169
Kelly et al. (1991)
Lake Erie, Summer
10
Sweet (1992), 1989-1990
Green Bay. WI
20
Hornbuckle and Eisenreich (1992),
Green Bay, WI
3-130
1989
Hoff (1992), 1988-1989
Egbert, Ontario, Summer
70-100
Hoff (1992), 1988-1989
Egbert, Ontario, Autumn
24-70
Hoff (1992), 1988-1989
Egbert, Ontario. Winter
7-24
Hoff (1992), 1988-1989
Egbert, Ontario, Spring
23-620
5-74
-------
For several of the compounds measured in LMUATS, a consensus on the Great
Lakes regional atmospheric concentration does not exist. Among those compounds are
aldrin, mirex, simazine, chlorpyrifos, alcohol and metalachlor. Aldrin is used as an
insecticide and in modeling work completed by Stiver and Mackay (1990), it is predicted
that for steady-state soil emission, the atmospheric concentration of this compound would
be approximately 1 pg/m3. During the LMUATS this is, in fact, the concentration of
Aldrin found at Kankakee (0.95 pg/m3) and exceeds the concentration found over the
water (0.26 pg/m ) and in South Haven (0.35 pg/m ). However, Aldrin concentrations in
Chicago at IIT were six times higher than levels measured at any of the other sites,
indicating a local source for this compound. Interestingly, the peak value for Aldrin
measured in Kankakee and IIT was on 8 August and in South Haven on 7 August, days
with easterly flow to each of the sites.
Mirex is a highly chlorinated insecticide. The average and maximum
concentrations of Mirex in the atmosphere in both Kankakee and IIT were strikingly
similar as were the average levels and maximum values reported for R/V Laurentian and
South Haven measurements. One value reported by Hoff for Mirex collected in Egbert,
Ontario in the Great Lakes region was 22 pg/m (Eisenreich and Strachan, 1992). Levels
observed at LMUATS sites fall on both sides of this range and do not show wide
variability in concentrations, indicating that for this compound, the sources may be diffuse
and constant.
Additional compounds, to those mentioned above, which rank in the top twenty of
the pesticides for quantity used in the Great Lakes Basin in 1986, are Atrazine (ranked
3rd), Chlorpyrifos (ranked 16th) and Simazine (ranked 20th). Among the three, Atrazine
and Simazine are currently used heavily. Atrazine is a common herbicide applied to corn
crops as well as other vegetation. Ambient concentrations of Atrazine demonstrated
marked seasonal changes linked to the time of application. Eisenreich and Strachan
(1992) report consensus levels of Atrazine measured in the months of May and June,
averaging 5,000 pg/m , while levels averaged from samples collected during other months
of the year indicate atmospheric concentrations of 10 pg/mJ. In a springtime investigation
by Sweet (1992) in the Lake Michigan Watershed, Atrazine levels were reported to be
100,000 pg/m3. These reported values are significantly greater than average Atrazine
-> "1
levels measured during LMUATS which varied from 183 pg/m at IIT to 372 pg/m at
Kankakee. The maximum value observed was at Kankakee (1504 pg/m3 on 22 July).
This may have reflected the high percentage of corn crop land in the area and thus the use
of large amounts of Atrazine. The peak Atrazine concentrations observed over Lake
5-75
-------
Michigan and in South Haven were half the peak level measured at Kankakee. At IIT,
peak Atrazine was half the level measured over Lake Michigan and in South Haven.
Chlorpyrifos is a potent insecticide. In the LMUAT study, the average
atmospheric Chlorpyrifos concentration was similar in Kankakee and in South Haven (83
and 88 pg/m3) but underwent very wide fluctuation (maximum values at these sites were
560 and 514 pg/m with a standard deviation of 75 and 122 pg/m , respectively). The day
to day pattern for this insecticide is clearly very different than seen for the other pesticides
(Figure 5-38). Mean levels of this compound were significantly lower at IIT (44 pg/m )
and over Lake Michigan (12 pg/m").
Simazine is in widespread use as a herbicide and is applied heavily by Michigan
fruit growers. This application is most likely responsible for the peak in Simazine
concentration measured at South Haven on 2 August 1991 The peak concentration at
IIT was observed on the same date and is greater than the level observed in South Haven
(637 pg/m3). Average concentrations of Simazine measured during LMUATS do not
demonstrate a marked difference between sites. The average concentration of Simazine at
KAN was 90 pg/m , 103 pg/m" at IIT , 85 pg/m over Lake Michigan and 54 pg/m in
South Haven.
5.8. Over-Water versus Land-Based Measurements
The question of whether it is reasonable to utilize shoreline or inland monitoring
locations to represent over-lake deposition is critical to our ability to accurately assess the
atmospheric deposition component of the toxics loading to the Great Lakes. This
question can only be answered by a systematic study of the complex meteorological
conditions which are necessary to parameterize the deposition process. This project is the
first step in the development of a program to study the dynamics of toxic pollutant
transport over and deposition into large lakes. Since data collected historically in the
Great Lakes utilized land-based monitoring locations, this study also provides an initial
comparison of the validity of land-based measurements with those made over-water. Due
to the expense of ship time, the number of days of concurrent over-water and over-land
measurements was limited. The ship was on-station approximately 20 miles west of
Muskegon, MI for an initial pilot cruise and for two periods later in the 30-day intensive,
the ship was on-station 5-10 miles off-shore from Chicago/Gary. The mixed-layer
trajectories associated with these sampling days are given in Figure 5-6.
5-76
-------
Figure 5-38. Variations in Measured Chlorpyrifos Concentrations.
250
200
E 150 H
\
o>
100 -
258
50 -
Kanka kee
~ i r
i i i i i i r
16 17 18 19 20 22 24
11 i i J 111 f i i M f
T~l
29 31 2 5 6 7 8
250
200 -
E 150 -
C71
100 -
50 -
0 -
I IT
i—i—i—i—r
7 T
1 6 17 18 19 20 21 22 23 24
f f i 11—i r I r f
ILL
-i—i—r T r T i T i—r
29 31 2 5 6 7 8
250
200
'g 150
Ol
a. 100
50
0
250
200
E 150
cn
a 100
50
0
R/V Laurentian
T—i—I T1! i
i—i—r
i i
1112
514
680 269
23 24
If Hi!
5 6 7
South Hcven
T I TTTTTTTTTTTT I TT I I I I I T T
10 11 12 13 14 15 16 17 18 19 20 21 23 24 29 31 2
! I J T ? I
July
Date
5 6 7 B
August
5-77
-------
5.8.1.
Individual Cruise Descriptions
5.8.1.1. On-Station Off-Shore from Muskegon, MI J 0-11 July 1991
During the sampling period off-shore from Muskegon, MI predominant easterly
flow advected pollutants from several sources in the Muskegon and Grand Rapids, MI
area over the lake. Elevated concentrations of Se, Zn and low molecular weight PCBs
were measured, in addition to one sample with high As levels. The concentrations of
SO4 , NH3 and HNO3 were comparable on the R/V Laurentian and at the South Haven
site. The concentration of retene observed on the Laurentian during this cruise was
elevated above levels observed simultaneously at South Haven, potentially indicating
impact from a combustion source utilizing wood or other vegetative material upwind. The
elevated retene levels at the location farther to the north may also reflect the different
species and tree density north of Muskegon. Levels of coronene observed over-water
during this period were similar to the observed measurements in South Haven, indicating
similar impact due to motor vehicle emissions or other sources.
Average concentrations of several pesticides measured during this cruise were
similar on the Laurentian and in South Haven (a-HCH, trans-nonachlor, Mirex,
Chlordane, Aldrin, Metalachlor, y-Chlordane and a-Chlordane) with a notable elevation in
Laurentian levels observed on 12 July. Concentrations of DDT and its breakdown
products as well as Dieldrin and Chlorpyrifos were highly elevated in South Haven.
Atrazine was the only pesticide measured which was found in elevated concentration on
the Laurentian (11 July). Low molecular weight PCBs (total mono-PCB and 2-PCB)
were also measured in elevated levels on the Laurentian on 11 July as well as in South
Haven.
Analysis of samples collected for organic compounds were not analyzed at IIT or
Kankakee during the first cruise. The funds available for this expensive analysis were
limited and the prevailing transport pattern during the period indicated that this data would
be of less value than for different time periods during the study.
5.8.1.2. On-Station Off-Shore from Chicago/Gary, 23-27 July 1991.
For the first sampling period off-shore from Chicago/Gary, levels of the trace
elements measured aboard the R/V Laurentian were similar to or slightly lower than those
levels monitored at IIT, largely due to the prevailing northerly flow that the vessel
5-78
-------
received while on station. Since there are no large sources in the lake and the upwind
fetch from the sampling point in southern Lake Michigan is hundreds of kilometers of
open water, the levels of most pollutants would not be expected to be very high. The
exceptions to this observations were levels of vapor-phase and particulate mercury as well
as coronene and naphthalene which were greatly elevated at IIT in comparison to those
concentrations observed on the Laurentian. Also, the daytime sample on 24 July aboard
the Laurentian revealed elevated Pb levels, both fine and coarse, as well as elevated Zn as
a result of westerly winds from the greater Chicago area.
Concentrations of most acidic aerosols were comparable at the South Haven and
Laurentian sites during this period, with the exception of NH3 which was elevated above
the levels observed in South Haven.
5.8.1.3. On-Station Off-Shore from Chicago Gary, 5-7 August 1991
During the second of the two cruises for which the Laurentian was located off-
shore from Chicago/Gary, prevailing flow was from the east-southeast. These conditions
provided direct measurements of the transport of toxic compounds to the lake. Levels of
fine Fe and Mn, impacted by the plume from the Gary steel industry were evident.
The PAH concentrations appear to mimic the behavior of atmospheric Hg, about a
factor of three to ten times higher at IIT than was measured over-water. The mean
pesticide concentrations measured at IIT and aboard the R/V Laurentian are within a
factor of 2 of each other with some being higher at IIT and others being greater on the
Laurentian Atrazine and Simazine were both found in greater concentrations over the
water than at IIT. The concentrations of pesticides observed in South Haven and in
Kankakee are generally much higher than those observed at either IIT or over the lake.
This is not surprising as both sites are in the middle of large agricultural and fruit growing
areas. Local application of pesticides strongly influences the levels seen at one site and,
therefore, prevent a single land-based site from being used to estimate the over-water
levels with any certainty. However, on selected occasions the concentration of some
pesticides measured on the Laurentian were determined to be greater than the
concentration measured simultaneously at any of the land sites. This occurred on 5
August for the pesticide Atrazine and on 7 August for y-HCH. This finding may be the
result of special meteorological conditions that allow for the advection of concentrated
pesticide-laden air masses over the water.
5-79
-------
Also of interest is a second observation of elevated levels of low molecular weight
PCBs on board the Laurentian, with high concentrations of 2-PCB and total mono-PCB
measured during 5 and 7 August.
5.8.2. General Conclusions Regarding Over-Water Measurements
For the two periods when the R/V Laurentian was on-station, average fine mass
levels measured at the Chicago site and over the water were comparable. These
measurements suggest that estimates of deposition for many compounds would have been
fairly accurate using data collected at the Chicago site. The fairly close agreement in
measured values for most species appear to indicate that the land-based site would
function as an adequate surrogate for over-water measurements of trace elements.
The major flaw with this assumption is that while the ship was on-station off
Chicago/Gary, two different and fairly uncommon summertime meteorological patterns
were observed. For instance, the prevailing wind flow was strong and from the northwest
during the first cruise. This provided a relatively non-polluted air mass to both the
Laurentian and the Chicago site. During the second cruise, transport to the ship was from
the east-southeast which provided a relatively polluted air mass to the Laurentian coming
from the Chicago/Gary area. Both of these events produced similar levels of pollutants
measured over water and at the Chicago land based site. In order to adequately assess the
validity of utilizing land-based monitors instead of over-water sites, a representative
sampling of several meteorological conditions that reflect the complex dynamics of lake
meteorology would have to be obtained.
For example, during the preliminary study for the Lake Michigan Ozone Study in
July 1990 a tight, relatively non-dispersed pollutant plume was observed as it traveled
North along the shoreline of Michigan. The role of these meteorological phenomena in
the deposition of toxic compounds to the Lake may be highly significant because they may
provide capture of a pollutant plume near the lake surface as it is advected away from the
source. It is often the case that a few isolated instances with a particular set of
meteorological and source influence characteristics can be responsible for major pollutant
episodes.
Important exceptions to the general finding of comparable levels of fine mass are
the striking differences in measured concentrations of mercury, several PAHs and
5-80
-------
pesticides and noticeable differences in coarse mass which could greatly affect calculations
of total loading to the lake.
Generalizations regarding over-water measurements cannot be made from the
limited data collected in this initial study. The data indicate that over-water measurements
involve a very special set of circumstances that can, on occasion, play an important role in
the deposition of toxic compounds to the water. Further work in this area will help to
determine if considering seasonal differences in dominant meteorology, atmospheric
chemistry, and transport to land based sites can be used as an adequate surrogate for over-
water deposition calculations on a yearly average basis. If the uncertainty measured for
such estimates turns out to be quite large, then the air sampling community may have to
accept the burden of collecting this critical data in the most accurate manner possible. For
example, representative samples taken over-water for limited portions of each season or
using unmanned monitoring stations for some of the critical pollutants may be necessary to
extrapolate to other compounds. This could provide an accurate measurement for a few
of the species of interest.
The values given in Table 5-21 suggest that for the brief LMUATS period when
the R/V Laurentian was off-shore of Chicago, the levels observed at IIT are quite similar
to those measured over the water. In fact, the averages are indistinguishable from each
other. The exception to this is Hg which was much higher at IIT, in both the vapor and
particulate phases, than that measured aboard the Laurentian. The levels of fine mass and
most trace metals, PAHs and several of the pesticides observed at Kankakee and in South
Haven are quite different and would not agree well with the Laurentian values. The levels
of fine fraction elements measured at South Haven are approximately one-half of those
measured aboard the Laurentian for this period. Again, the Hg levels observed at South
Haven were more similar to those measured concurrently aboard the Laurentian than
those measured at IIT.
Table 5-22 shows that the average PAH concentrations measured at the four
locations during the period when the R/V Laurentian was on station near Chicago. The
PAH concentrations appear to mimic the behavior of atmospheric Hg and are about a
factor of three to ten times higher, on average, at IIT than measured over-water. For
example, naphthalene concentrations averaged 397 ng/m3 at IIT and were 151 ng/m3 on
the Laurentian Anthracene was found to be 3 ng/m3 on average at IIT and only 0 3
ng/m3 6 miles offshore.
5-81
-------
Table 5-23 lists the average concentrations of pesticides measured at the 4
LMUATS monitoring sites. The mean pesticide concentrations measured at IIT and
aboard the R/V Laurentian are within a factor of 2 of each other with some being higher
at IIT and others being greater on the Laurentian. Atrazine was 5 times greater over the
water than at IIT (258 vs. 51 pg/m3). Simazine was also much higher on the Laurentian
than at IIT with an average concentration of 103 pg/m3 compared to 6 pg/m3. The
concentrations of pesticides observed in South Haven and in Kankakee are generally much
higher than those observed at either IIT or over the lake. This is not surprising as both
sites are in the middle of large agricultural and fruit growing areas. Local application of
pesticides strongly influences the levels seen at one site and therefore, prevent a single
land-based site from being used to estimate the over-water levels with any certainty.
However, on selected occasions, the concentration of some pesticides measured on the
Laurentian were determined to be greater than the concentration measured simultaneously
at any of the land sites. This occurred on 5 August for the pesticide Atrazine and on 7
August for y-HCH. This finding may be the result of special meteorological conditions
that allow for the advection of concentrated pesticide-laden air masses over the water.
Table 5-21. 12-hour Average Concentrations for Fine Fraction Trace Elements
Determined by XRF When R/V Laurentian on Station Near Chicago (ng/m3).
Kankakee
IIT
R/V Laurentian
South Haven
Mass
1410+ 1030
1070 ^810
920 ± 870
580 ±360
A1
293 ± 306
74 ± 109
26 ±41
24 ±59
Si
76 + 32
63 ± 32
59 ±54
67 ± 138
S
1700 ±1718
1183 + 1442
1171 ±1293
553 ±395
CI
13 ± 7
6 ± 6
7 ± 7
5 ± 5
K
51+24
46 ±42
54 ±50
27 ±32
Ca
77 ±76
43 ±28
42 ±24
31 ±33
iMn
4 ± 3
4 ± 4
4 ± 5
2 ± 3
Fe
69 ±54
98 ± 132
83 ± 103
32 ±68
Cu
4 ± 1
9 ± 7
6 ± 5
3.4 ± 3
Zn
25 ± 15
16 ± 17
20 ±20
8.4 ± 13
Se
2 ± 3
0 85 ± 1.1
1 ± 1
1 ± 1
Br
4 ± 2
2.6 ± 1.0
2 ± 1
2 ± 1
Pb
15 ± 7
9 ± 10
9 ± 10
3 ± 5
5-82
-------
Table 5-22. Average Concentrations of PAHs when the R/V Laurentian was on
Station Near Chicago (ng/m3).
Naphthalene
Acenaphthvlene
Acenaphthene
Fluorene
Phenanthrene
Anthracene
Fluorenone
Retene
Fluoranthene
Pyrenc
Benz(a)anthraccne
Chrysene
Cyclopcnta(c.d)pyrenc
Bcnzofluoranthcnes
Benzo(e)pyrene
Benzo(a)pyrene
Indeno(1.2.3-c.d)pyrene
Dibenzo(a.h)anthracene
Benzo(g.hi)perylene
Coronene
Kankakee
Avg
162.32 +270.66
1.45 +0.95
1.96 +1.36
3.99 +2.76
8.52 +4.71
0.24 +0.15
1.32 +0.52
0.21 +0.17
1.72 +1.05
0.79 +0.48
0.10 +0.08
0.20 ±0.12
0.04 +0.04
0.32 +0.15
0.11 +0.05
0.14 +0.05
0.19 +0.09
0.17 +0.07
0.15 +0.07
0.12 +0.06
IIT
Avg
396.64 +156.38
3.82 +1.56
20.04 +12.50
23.42 +10.58
51.56 +22.53
3.07 +1.89
4.69 +2.55
0.38 +0.22
18.8 +9.10
9.18 +4.00
1.14 +0.78
2.09 +1.28
0.20 +0.19
3.88 +2.70
1 04 +0.69
1.64 +1.26
1.84 +1.22
0.76 +0.41
1.65 +0.89
0.74 +0.27
R/V
Laurentian
Avg
151.28 +130.47
1.60 +1.06
2.57 +2.38
8.00 +4.97
12.94 +9.51
0.31 +0.27
1.23 +0.60
0.52 +0.23
4.05 +2.86
2.01 +1.55
0.34 +0.42
0.80 +0.93
0.12 +0.17
1.17 +1.35
0.32 +0.35
0.31 +0.34
0.52 +0.57
0.23 ±0.17
0.42 ±0.47
0.18 +0.16
South Haven
Avg
25.85 +23.84
0.52 +0.33
0.85 +0.36
2.06 +0.73
3.90 +0.73
0.16 +0.07
0.60 +0.30
0.28 +0.14
1.08 +0.39
0.53 ±0.22
0.05 ±0.02
0.11 ±0.03
0.03 ±0.03
0.12 ±0.04
0.05 ±0.01
0.05 ±0.03
0.09 ±0.05
0.11 ±0.04
0.07 ±0.04
0.06 +0.03
5-83
-------
Table 5-23. Ambient Pesticides Concentrations when the R/V Laurentian was on
Station Near Chicago (pg/m3).
Kankakee IIT Laurentian South Haven
Avg Avg Avg Avg
alpha-HCH
204+124
135+40
192+232
152±37
Hexachlorobenzene
172+277
71 + 16
121±174
69±13
Atrazine
254+123
51 ±70
258+254
220±123
gamma-HCH(lindane)
299+399
43±25
90±93
59+12
Alachlor
1096+2045
0+0
74+234
0±0
Mircx
48+59
26±21
8+6
8+6
Aldrin
1 ±2
3 ±2
0.2+0.4
0.9+1.1
Metaiachlor
97+96
0+0
3 ±8
66+109
trans-nonachlor
55+55
29+18
11+6
17+7
Dieldrin
492±423
83+44
48+28
95+70
Simazine
199±147
6+13
103+121
18+40
Chlorpyrifos
38±25
29+20
11 + 10
18± 15
gamma-Chlordanc
128±110
45+23
29±14
31±6
alpha-Chlordane
98+88
38+20
23+9
27+13
4,4'-DDT
12+16
104+72
38±37
267+145
P.P'-DDE
60+26
71+32
63+71
1127+371
P.P'-DDD
8+8
3+2
4+6
13+9
5-84
-------
Chapter 6
Meteorological Analysis
6.1. Synoptic Overview for LMUATS
The overall weather pattern during the first half of the Lake Michigan Urban Air
Toxics Study period (8 July - 9 August 1991) was rather non-dynamic. Throughout the
first two weeks of the study, the jet stream set up in a zonal (west to east) flow pattern,
lying mainly across the U.S.-Canadian border. The result of this pattern was a series of
alternating high and low pressure systems that passed directly over the Great Lakes
region. With the jet stream to the north of the Great Lakes, the various weather systems
tended to move north of the study area. Under these conditions, the southern Lake
Michigan Basin experienced prolonged periods of northerly winds. As areas of high
pressure moved across the Great Lakes, the winds were initially northwesterly (i.e., from
the northwest), gradually becoming northeasterly and then easterly. As areas of low
pressure moved across the Great Lakes, winds became southerly for a short period, before
returning to the northwest with the storm's passage.
The final week of July proved to be the most meteorologically dynamic of the
LMUATS period. Early in the week, an upper-level low developed over the Hudson Bay
and essentially remained in place during the entire week. The upper-level low produced a
series of fronts that moved across the study area during this week. This system eventually
weakened and moved eastward during the weekend of 27-28 July 91. As a result, no
sustained flow patterns were able to form over the region during this period.
The final week of the study saw the return to a zonal upper-level flow pattern
across the northern United States. A split in the jet stream left the Great Lakes under the
influence of the weak, southern branch of the jet. As a result, a slowly moving, broad area
of high pressure dominated the weather during the final week.
6-1
-------
6.1.1. Week of 8-13 July 91
Monday, 8 July 91, was the first day of the LMUATS period. On this day, a cold
front moved through the study area during the early morning hours. The passage of the
front brought scattered showers and a switch from warm, southerly winds to much cooler,
northwesterly winds. Behind the front, an area of high pressure was located across the
northern Great Plains. As the high moved toward the Great Lakes, it helped to maintain
the north to northwest winds for the remainder of the day. Despite clearing skies,
temperatures were considerably cooler behind the front. Temperatures across the
southern Great Lakes were in the 90s during the previous day, but only warmed into the
mid 70s Monday. The high continued to move toward the Great Lakes during the night,
with winds becoming more northerly across the study area.
On Tuesday morning, the high pressure area that had been in the Great Plains the
previous day was now located near Houghton Lake, MI. As a result, the winds across the
southern Great Lakes were from the east at 5-10 mph. Skies across the study area were
partly sunny, while temperatures remained cool. Afternoon temperatures only warmed
into the upper 70s. Without much upper-level support, the high pressure area weakened
over the Great Lakes Tuesday night. As a result, the overnight winds across the study
area were light and variable.
The high pressure area stalled across the state on Wednesday, maintaining a rather
diffuse wind field across the entire Great Lakes region. The predominant wind direction
across the study area was from the east during this period. Skies were sunny across the
region on Wednesday, with afternoon temperatures rebounding into the low to mid 80s.
The exception was along the Wisconsin/Illinois shorelines, where temperatures were
cooler due to weak onshore flow.
An upper-level ridge strengthened over the Great Lakes on Thursday, causing a
strong high to develop. The wind field across the study area therefore became more
organized, with east to northeast winds dominating during the morning hours. As can be
seen in Figure 6-1, the difference in the wind field on opposite sides of the lake appears to
have been caused by a very weak trough of low pressure. However, this feature did not
persist through the remainder of the day, and winds across the entire study area gradually
became easterly.
6-2
-------
Figure 6-1. Plot of Surface Station Data for 11 July 1991.
-------
The high began to drift eastward on Friday, allowing winds across the study area
to become southeasterly ahead of an approaching low pressure center. The low pressure
center was located over Madison, WI around sunrise. As the low moved through the
southern Great Lakes, clouds increased over the region. Scattered light precipitation was
reported in the study area, with rainfall totals generally less than 0.1 inch. With the
passage of the system across southern Lake Michigan Friday night, winds were variable
across the southern Lake Michigan Basin.
On Saturday morning, most of the precipitation ended as the low pressure system
worked its way eastward. Skies were variably cloudy during the remainder of the day,
with a few isolated afternoon showers developing. Behind the departing low, the winds
became northerly, with speeds ranging from 10-15 mph. As the low pressure area
continued to pull away from the Great Lakes, wind speeds across the study area
diminished to 8-10 mph. Despite the northerly winds that had developed during the day,
afternoon highs warmed into the mid to upper 70s.
6.1.2. Week of 14-20 July 91
The second week of the study was much like the first, with a slow-moving area of
high pressure dominating the weather in the Great Lakes for several days. On Sunday, a
large area of high pressure was located over the northwestern Great Lakes. Cool,
northerly winds were reported statewide with weak land breeze characteristics evident in
the light wind pattern during the morning. In the afternoon, onshore winds kept
temperatures cool near the southern lake shore, while sunny skies helped to warm the
inland areas into the lower 80s. Late in the day, the center of the high pressure area
moved over western Michigan. As a result, the overnight winds became light and variable.
Monday saw the high pressure center pass across the central Lower Peninsula of
Michigan. Winds remained light and variable through the day. The high pressure kept
skies sunny across the entire Great Lakes area. Temperatures warmed into the low 80s in
most areas, but were cooler along the southern Lake Michigan shoreline. As the high
continued to drift eastward overnight, some light southerly winds developed across the
eastern Great Lakes. However, the winds across the study area remained light and
variable through the night.
By Tuesday morning, the center of the high pressure area had shifted eastward into
Pennsylvania. The morning winds around the Chicago area were offshore, while all other
6-4
-------
shoreline locations reported calm winds. As the day progressed the winds became more
organized. South to southwesterly winds were reported across the study area by late
afternoon, with temperatures rising to near 90 across the entire southern Lake Michigan
Basin. Winds remained southwesterly through the night.
A weak trough worked into the northern Great Lakes Wednesday morning, but
had little effect on conditions in the study area. High pressure across the eastern U.S.
helped to produce a weak, but very warm south to southwesterly flow across the region.
Afternoon highs again climbed to 90°F. With the East Coast high stalling just off the
Carolina coast, conditions remained constant from Wednesday through Saturday. Skies
remained generally sunny, though haze was reported across the southern Lake Michigan
Basin each day. Temperatures soared during the period, with readings in the 90s reported
through Saturday. Temperatures in Chicago, IL reached the 100-degree mark on Friday
and Saturday. On Saturday, a cold front approached the Great Lakes from the northwest
and stalled across central Michigan during the early part of the day. The front remained
stationary for the remainder of the weekend, producing a strong south to southwesterly
flow across the entire study area.
6.1.3. Week of 21-27 July 91
As noted in the above section, a stationary front continued to hold across the
central Great Lakes on Sunday. The front stretched from just north of Milwaukee, WI to
near Port Huron, MI. Scattered showers and thundershowers persisted near the front
during the day Sunday, with generally light amounts reported across the study area (0.1 to
0.3 inches). The flow across the study area remained from the south-southwest, with
speeds reported around 10 mph. With the front to the north of the study area, afternoon
tenr- matures warmed into the mid 80s on Sunday.
On Monday, a strong upper-level low started to move out of the Canadian
Rockies. A surface cold front associated with this upper-level feature began to move out
of the Central Plains states. In response to the approaching cold front, winds across the
study area increased from the southwest to around 10-15 mph. The strengthening
southwesterly winds helped to push the stationary front northward, allowing very warm
air to move into the study area. Temperatures climbed into the upper 90s to near 100 on
Monday, with Chicago, IL reporting a high of 102°F.
6-5
-------
On Tuesday morning, a cold front from the Central Plains moved across the study
area. The winds behind the front were from the northwest, with speeds ranging from 10-
20 mph. Scattered precipitation accompanied the front, with all precipitation ending by
12Z (8AM EDT). As the day progressed, the winds remained strong from the northwest.
This northwest flow pattern resulted in significantly cooler temperatures on Tuesday, with
afternoon highs remaining in the upper 70s. On this day the R/V Laurentian made its first
trip to the southern Lake Michigan sampling station.
A weak trough of low pressure moved through the study area early Wednesday
morning. As a result, the winds across the southern part of Lake Michigan were from the
southwest (Figure 6-2). However, the winds quickly returned to the west-northwest with
the passage of the trough. Despite the northwesterly winds reported on the ship,
emissions from the Gary, Indiana industrial area were observed moving to the northeast
(i.e., over the lake) for several hours. Figure 6-2 shows that Gary did have winds from the
southwest during this period. Further analysis determined these southwesterly flows were
likely caused by the deceleration of the weak trough. It is likely that the wind-shift line
remained between the R/V Laurentian and Gary for several hours before moving to the
southeast. A second weak trough moved through the study area that night.
The trough that moved into the study area Wednesday night was located just to
the south of Lake Michigan on Thursday morning. When the Canadian upper-level low
started to pull to the northeast, an area of surface high pressure was able to move in
behind the trough. Winds across the study area were generally from the north. Along the
Wisconsin shoreline, the wind appeared to be most affected by the surface high and the
resulting flow was from the northwest. Along the Michigan shoreline, the winds were
affected by the departing trough and the flow was from the northeast throughout the day.
The R/V Laurentian similarly reported northeasterly winds on Thursday.
Early Friday morning, the area of high pressure was located over northern
Wisconsin. The high gradually drifted eastward through the day and consequently, the
winds remained from the northeast across the study area. Skies remained sunny across the
Great Lakes, with afternoon highs only warming into the lower 70s. Little change was
noted on Saturday, as the high continued to slowly make its way toward the East Coast.
6-6
-------
Figure 6-2. Plot of Surface Station Data for 24 July 1991.
20 20
Plot of Surface Station data for 12Z 24 JUL 91
\% It
-------
6.1.4. Week of 28 July-3 August 91
The high pressure system that dominated the weather on 26 July and 27 July
slowly moved to the northeast on Sunday. A weak low-pressure system moved out of
eastern Nebraska and introduced clouds into the Great Lakes. The winds were generally
from the east on Sunday due to the continued influence of the Canadian high pressure
area. The combination of cloudy skies and easterly winds resulted in relatively cool
temperatures with afternoon highs in the 70s. As the night progressed, scattered showers
associated with the approaching low moved into the Great Lakes.
Monday began with scattered showers across the southern Lake Michigan Basin.
The surface low producing the showers was located just east of Rockford, IL at sunrise.
As a result, early morning winds were from the southeast across the study area. As the
storm center passed over the south-central part of Lake Michigan, winds across the area
became variable for a short time. Once the storm had passed, rain showers tapered off and
winds across the study area remained light and northwesterly. Generally light rainfall
totals were reported with most areas receiving less than 0.1 inch of rain. However, just to
the north of the study area heavier rains were reported, with Muskegon, MI reporting 1.70
inches.
By Tuesday morning, the low pressure system had moved into southern Ontario.
A weak area of high pressure then moved into the Great Lakes that afternoon. Winds
were light and generally from the north during the day, though the wind field was light
enough to be altered by land breeze mechanisms during the morning hours. As a result,
the morning winds across the Lake Michigan shoreline contained a lakeward component.
As the day progressed, the weak surface high continued to drift eastward. In response,
winds across the southern Great Lakes became light and southerly late in the day. Cool
temperatures prevailed over the area on Tuesday, with afternoon highs remaining in the
70s under sunny skies.
On Wednesday, the weak high pressure system moved into the Ohio Valley.
Consequently, the winds across the study area were mild and southerly. With skies
remaining sunny through the day, temperatures warmed well into the 80s across the
region. Afternoon highs rose to near 90°F in both Milwaukee, WI and Chicago, IL. As
the day progressed, winds increased in speed as a cold front began moving into the
Midwest from the Northern Plains. Winds remained from the south for most of the night.
6-8
-------
The approaching cold front moved through the Chicago area early Thursday
morning. The southerly winds across the study area became west to northwest with the
front's passage. The cold front was relatively weak with some clouds developing but no
precipitation reported. A weak high pressure area quickly moved in behind the front,
producing generally sunny skies. The temperature remained virtually unchanged from the
day before, with the exception of the southern shores of Lake Michigan. The Thursday
afternoon temperatures there were only in the 70s due to the onshore flow ahead of the
high. The surface high pressure center weakened during the day on Thursday. As a result,
the weak cold front that had passed through the study area Thursday morning was not
pushed far to the south. A developing low pressure system in Nebraska produced a
southerly flow to the south of the front, which caused it to move northward Thursday
night, eventually stalling across the study area.
As can been seen in Figure 6-3, the front stretched from the Wisconsin/Illinois
border to Flint, MI Friday morning. Winds to the north of the front were easterly, while
south of the front the winds were from the south. The arrival of the front brought variably
cloudy skies to the southern Great Lakes during the morning hours Friday. As the day
progressed, two low pressure areas developed along the front and headed to the east,
toward the Great Lakes. The first low pressure system moved across the study area
Saturday morning. Rainfall totals from the storm's passage ranged from .02 inches in
Chicago, IL to .57 inches in Muskegon, MI. As the first storm center moved east of the
study area, winds across the southern Lake Michigan Basin became northerly. The front,
as well as the second storm system, moved southeastward Saturday afternoon. A large
area of high pressure then moved in from Manitoba, Canada, bringing clearing skies and
brisk northerly winds. Cooler air associated with this system moved in, resulting in
Saturday afternoon temperatures in the low to mid 70s.
6.1.5. Week of 4 August - 10 August 1991
On Sunday, 4 August, high pressure continued to push toward the Great Lakes.
While the center of the high pressure was located along the North Dakota-Manitoba,
Canada border, it was still able to influence the winds across the Great Lakes. The winds
across the region were from a northerly direction. Early in the day, a weak trough was
responsible for variable wind directions across the study area. At 12Z (8AM EDT), the
trough ran from just south of Lansing, MI to near the South Haven, MI sampling site.
The winds to the north of the trough were due north at the time of observation, while
6-9
-------
Figure 6-3. Plot of Surface Station Data for 2 August 1991.
On
i
O
-------
south of the trough, the winds were northwesterly across southern Michigan. At that
time, the winds across northern Illinois were all from the northeast. There is some
evidence of a second weak trough in this area, which would explain the variation in the
wind directions across the lake. The wind speed generally increased throughout the day.
The strong winds (10-20 mph) were blowing straight down the north/south axis of the
lake for approximately 24 hours. As a result, the waves on Lake Michigan swelled to 3-6
feet. Sunday was to be the first day of the second sampling intensive aboard the R/V
Laurentian. However, due to the rough lake conditions, the ship was docked in Benton
Harbor for the night. During the night, the high pressure area continued its push toward
the Great Lakes. As the high moved closer, the winds diminished, allowing the R/V
Laurentian to reach its sampling position off the shore of Chicago.
By Monday morning, the center of the high was located over the western Upper
Peninsula of Michigan. The high was still close enough to produce sunny skies across the
southern part of Lake Michigan, though a cloud line was located near Chicago. From the
R/V Laurentian, skies south of Chicago appeared quite dark and capable of producing
rain. Indeed, records show that rain did fall just to the south of the lake, while skies over
the ship's location remained partly sunny throughout the day. With the approach of the
area of high pressure, the winds across the study area become northeasterly and eventually
easterly during the day.
By sunrise Tuesday, the winds had become slightly south of east. The center of
the high was now located near Saulte Ste. Marie, MI. As can be seen in Figure 6-4, a
weak trough of low pressure extended from eastern Iowa to southern Lake Michigan. As
a result, the flow over the study area was from a southeasterly direction. Conversely, the
remainder of the Great Lakes experienced easterly winds. The wind speeds over the lake
ranged from 5-10 mph. That evening, ti.e trough began to shift to a more west-east
orientation, and the winds observed from the ship's location shifted to a northeasterly
direction (SE -> NE in 30 minutes). It was decided that movement of the ship farther to
the south would place it in a more favorable wind flow pattern for sampling. For this
reason, the ship was moved southward that evening. In the ship's new location, the winds
were from the southeast once again, and remained that way throughout the night. It is
believed that the difference in the wind directions at the two locations was a result of the
trough moving slightly to the south of the ship's earlier location.
On Wednesday, the high pressure area was still the dominant feature across the
Great Lakes. The storm system that was developing in the Central Plains continued to
6-11
-------
Figure 6-4. Plot of Surface Station Data for 6 August 1991.
\V 20
-------
move slowly toward the Great Lakes. These two synoptic features helped to maintain
southeasterly winds across the study area for much of the day, with winds becoming
easterly during the early evening. The winds remained from the east for several hours.
During the early morning hours, the winds shifted to the southeast once again. The
change in wind direction was in response to the approaching low pressure system from the
Central Plains. The storm system arrived in the study area shortly after midnight, with
light rain beginning at the R/V Laurentian at 1:15 AM.
By 7AM Thursday morning, the storm center was located in eastern Iowa.
Showers and thunderstorms continued across the area throughout the day. The winds
over the lake increased significantly from the southeast as the storm system approached.
At that time, the winds over the lake shore areas were from the east. Two factors are
believed to be responsible for the observed divergent wind conditions. First, at 12Z a
weak trough ran from Milwaukee, WI to Muskegon, MI. The orientation of the pressure
field was conducive to a localized southeasterly flow over the southern half of the lake. In
addition, for the given pressure pattern, less friction across the lake surface (as compared
to the land surface) would also be conducive to a flow more parallel to the isobars (lines
of constant pressure). Winds reached 20-25 mph from the southeast, and large waves
developed on the lake. Sampling activities were suspended and the ship returned to shore
to avoid unsafe conditions.
As the day progressed elsewhere across the region, the winds across the southern
Great Lakes continued from an easterly direction. The storm center passed through the
southern Great Lakes during the afternoon and evening hours of Thursday, causing the
winds to become northerly behind the surface low. By morning, the storm center was
located over western Pennsylvania. Thursday precipitation totals across the area ranged
from 0.5-2.0 inches. With the storm now off to the eas1" nigh pressure began moving in
from the Northern Plains on Friday morning. The skies across the area began to clear
somewhat, with variably cloudy skies reported through the day. Despite the sunshine,
temperatures remained cool due to the northerly winds that had developed ahead of the
advancing high. Afternoon temperatures only warmed into the mid 70s across the study
area.
6.2. Micrometeorological Measurements
One of the goals of the Lake Michigan Urban Air Toxics Study (LMUATS) was to
investigate the possible relationship between the concentration of toxics in the ambient air
6-13
-------
over Lake Michigan and their corresponding concentrations in the lake surface water
below. If such a correlation were found to exist, further information would be needed to
understand the potential exchange processes occurring at the air-lake interface. For this
purpose, a series of meteorological measurements were taken aboard the R/V Laurentian
during two ship borne measurement periods: 23-27 July 91 and 5-8 August 91. These
measurements were used to characterize the micrometeorological environment that existed
in the atmospheric surface layer above Lake Michigan during these intensives. These data
were also used to compute surface layer fluxes of heat, momentum and moisture. These
fluxes were then used to gain insight into the aforementioned exchange processes
occurring at the air-lake surface interface.
6.2.1. Analysis of Micrometeorological Data
As noted earlier in this report, one of the goals of the ship borne intensives during
the Lake Michigan Urban Air Toxics Study was to investigate the possible relationship
between the concentration of toxic pollutants in the air over Lake Michigan and their
corresponding concentrations in the lake surface water below. In order to accomplish this
task, it was necessary to characterize the state of the atmospheric surface layer over the
lake during the measurement periods. The state of the surface layer will greatly influence
the various processes by which pollutants are transported through the atmosphere and to
the surface. To make this characterization, a set of micrometeorological measurements
was conducted off the bow of the ship. Once collected, the data were disseminated in two
unique ways to determine the likelihood of the deposition of the airborne toxics. One
method of analysis involves simply looking at the stability of the atmosphere. The second
more qualitative method used meteorological data to determine the actual deposition
velocity of the pollutants. In the sections that follow, the methods by which pollutants can
be deposited to a surface are discussed as well as the aforementioned methods of
determining if there was a likelihood of deposition during this study period.
6.2.2. Deposition Processes
Once emitted into the atmosphere, pollutants are transported by the prevailing
winds. Heavier particles will usually fall out of the atmosphere quickly, mainly due to the
effects of gravity. Smaller, lighter particles and gases are less likely to fall out of the
atmosphere due to this gravitational settling. As a result, the lighter particles are more
likely to be transported over longer distances. Since the lighter particles will not settle out
6-14
-------
of the atmosphere on their own, they must rely on turbulent motions in the atmosphere to
assist in their deposition to the Earth's surface (in the absence of precipitation). These
turbulent motions can be generated via mechanical and/or via thermal effects (Figure 6-5).
Mechanical turbulence is caused by wind shear (a change in the wind speed/direction
with height). The particles or gases become caught up in the turbulent motions or eddies,
which in turn help to transport them toward the surface. In much the same way, particles
can be transported toward the surface via thermally induced turbulence. In this case, if the
air above a surface is cooler than the surface itself, the layer of air immediately above the
surface is warmed. This warmer air will rise, causing the development of buoyant,
turbulent eddies. Once again, the eddies will aid in the transport of the lighter particles or
gases toward the surface.
To a large degree, the intensity of turbulence in the atmosphere depends on the
stability of the atmosphere. Some of the micrometeorological measurements made aboard
the R/V Laurentian were done specifically to determine the stability of the atmosphere
over the lake. In the next section, the role that stability plays in the control of turbulence
intensity will be addressed. In addition, measurement results, conclusions regarding the
stability of the atmospheric surface layer during the intensives, and how this may have
affected pollutant deposition will be discussed.
6.2.3. Stability of the Atmospheric Surface Layer
Regardless of how turbulence is produced, the stability of a layer in the
atmosphere will determine if this turbulence is damped or allowed to grow in that layer. A
stable atmosphere will act to suppress turbulence, while an unstable atmosphere will allow
for the generation and/or growth of turbulence. During the ship borne intensive
measurement periods, continuous measurements were made of the wind and temperature
fields in the lowest 7 meters of the atmospheric surface layer above the lake. In addition,,
temperature measurements were taken from the lake surface layer. These parameters
were used to determine the stability of the atmospheric surface layer.
A layer is considered statically stable if the potential temperature in that layer
increases with increasing height. The potential temperature of a layer is the temperature
that the layer of air would have if it were adiabatically lowered to the surface. Since the
potential temperature of any parcel (or layer) of dry air does not change when moved
6-15
-------
Figure 6-5. Schematic of the Micrometeorologicai Effects on Pollutant Deposition.
Mechanically Induced Turbulence
u
u
0)
a
o
o
Turbulent eddies carry
particles tovnrd lake's surface.
Thermally Induced Turbulence
6-16
-------
vertically, vertical movement in a stable layer will result in the parcel being colder and
denser than its surroundings. Therefore the parcel will want to sink back to its original
position. In a statically unstable atmosphere, the same parcel would be warmer than its
surroundings when lifted, and the parcel would want to continue to rise. In these ways,
turbulence is either damped or enhanced by the thermal structure (and thus stability) of a
particular layer.
For this study, it was the stability of the atmospheric surface layer above Lake
Michigan that was of interest. The hypothesis stated that if the layer was stable,
turbulence would be suppressed. In the absence of turbulence, the lighter particulates and
gases would not have a significant deposition mechanism. If the layer was unstable,
enhanced turbulence would be expected, resulting in enhanced deposition of pollutants.
While numerical results will not be presented at this time, the following can be said
regarding the stability of the surface layer during the intensive periods.
6.2.3.1. 24 July 1991
On 24 July 1991, weak high pressure initially prevailed across the southern Lake
Michigan Basin, producing sunny skies and warm temperatures. Afternoon high
temperatures along the Lake Michigan shoreline ranged from 80-85° F. As a cold front
approached from the northwest during the afternoon, the winds became southeasterly with
speeds ranging from 8-12 mph. Under these conditions, mild air was carried across the
southern portion of Lake Michigan and across the area in which the R/V Laurentian was
anchored. During the afternoon, measurements aboard the R/V Laurentian indicated that
the temperature was constant through the atmospheric surface layer, with an average
temperature of approximately 77°F. During this same period, measurements were taken
of the temperature of the lake surface water. These readings indicated a surface
temperature of approximately 75°F. A cold front moved across the area that evening
(9PM-12AM). With the passage of the front, winds became northwesterly at 12-16 mph,
and significantly cooler air moved in behind the front.
The above information suggests the following about the atmospheric surface layer
stability for that day. The surface temperature of the lake was significantly cooler than the
temperature of the airmass as it left the Lake Michigan shoreline. The afternoon
temperature readings taken aboard the R/V Laurentian indicate that the airmass cooled as
it traveled to the ship's position. This cooling of the airmass' lower layer would have
6-17
-------
produced a thermally neutral-to-stable atmospheric surface layer during the afternoon and
evening periods. The net effect would be to stabilize the surface layer.
A Reynolds number was also calculated for this period. The Reynolds number is a
measure of atmospheric stability which considers both thermal and mechanical effects.
During the afternoon and evening hours, the Reynolds number indicated neutral to slightly
unstable conditions. Values of the Reynolds number ranged from 0.0 to -0.1.
When the two sets of information are both considered, the following can be
deduced about the stability conditions over Lake Michigan that day. Wind data show
adequate surface wind shear to produce some mechanical turbulence. The apparent
stability of the surface layer suggests that in all likelihood this turbulence would have been
damped. As a result, long-range transport of pollutants from the Lake Michigan urban
region across the lake was probable. The highest probability for pollutant deposition to
the lake's surface would have been during and after the passage of the frontal systems.
6.2.3.2. 25 July 1991
As noted in the discussion of 24 July, the passage of a cold front on Tuesday
evening caused a shift in surface wind directions. While the winds initially became
northwesterly after the front's passage, the direction became north to northeast during the
early morning of 25 July. Wind speeds ranged from 12-16 mph during the early morning
hours, before diminishing somewhat to 10-14 mph for the remainder of the day. The
passage of the front brought a new, cooler airmass across the southern Lake Michigan
Basin. The intrusion of the cooler airmass was somewhat masked by the typical diurnal
fall in temperatures. However, its effects became apparent during the daylight hours of 25
July. While the afternoon highs on 24 July were in the range of 80-85°F, the daytime
highs -cross the region on 25 July ranged from 70-73°F. Skies remained sunny as a new
area of high pressure moved into the Great Lakes region.
As previously observed, the surface water temperature was approximately 74-75°F
during this period. Unlike conditions of 24 July, the airmass that flowed off the land and
over the water was now colder than the lake surface. Rather than cooling the lower part
of the atmospheric surface layer, the lake now served to warm the lower portion of the
surface layer. The temperature at the lake surface was as much as 0.5°F warmer than the
temperature 7 meters above the lake surface. This trend began to develop during the early
morning hours, when the air temperature dropped below the temperature of the lake
6-18
-------
surface. This situation is termed statically unstable, as the potential temperature was
decreasing with height. These conditions persisted throughout the day. Now, rather than
suppressing turbulence, the thermal structure was such that it should have enhanced any
existing turbulence.
As before, Reynolds number values were estimated. The calculated values ranged
from -0.1 to -0.3, and were indicative of unstable conditions. This information, together
with the thermal evidence discussed above, strongly suggests that the conditions present
throughout 25 July were increasingly conducive to turbulence and enhanced pollution
deposition to the lake surface.
The prediction of increased turbulence on 25 July is further supported in the next
section of this report. In this section, a second method of analysis for investigating
pollutant deposition will be discussed. The concept of deposition velocity will be
reviewed. Although it is not practical to measure the deposition velocity directly, it can be
calculated by measuring the degree of turbulence in the surface layer. This turbulence can
be inferred through the measurement of a quantity called momentum flux.
6.2.4. Deposition Velocity
A second method of investigating the deposition of pollutants to any surface is
through consideration of a quantity called deposition velocity. This parameter describes
the average velocity at which a particular pollutant is deposited under a given set of
circumstances. Calculations of the deposition velocity consider stability, the chemical
nature of the pollutant and the characteristics of the surface to which deposition is
occurring. There are two types of deposition velocities which are usually considered: wet
deposition and dry deposition. Wet deposition involves the removal of pollutants from the
atmosphere via various precipitation processes. For the days being considered (24 and 25
July), no precipitation occurred. Therefore, only dry deposition will be discussed in this
section.
When considering the dry deposition of pollutants to any surface, it is often useful
to think in terms of an electrical resistance analogy. In a series circuit, the amount of
current flowing through the circuit is inversely proportional to the amount of resistance in
that circuit. Similarly, the velocity at which a pollutant is deposited to a surface is
inversely proportional to the resistance to deposition offered by a variety of factors. The
expression for the deposition velocity is as follows:
6-19
-------
Vd=l/[Ra + Rs + Rt]
(1)
where Vj is the deposition velocity, Ra is the aerodynamic resistance, Rs is the surface
layer resistance and Rt is the transfer resistance (Seinfeld, 1986).
The aerodynamic resistance accounts for the turbulent diffusion of material from
the free atmosphere to the surface laminar sub layer (immediately next to the surface).
This aerodynamic resistance is a function of typical meteorological parameters, such as
wind speed, surface roughness and atmospheric stability. The surface layer resistance
depends on parameters characterizing diffusion across the laminar sub layer (directly next
to the surface in question). This resistance is dependent on molecular, rather than
turbulent, properties. Finally, the transfer resistance depends on the physical-chemical
interaction between the pollutant and the surface.
For most species where the deposition under consideration is to a water surface,
the limiting (or dominant) term in equation (1) is the aerodynamic term. For this reason,
the other two terms will not be discussed at this time. Further information on these terms
can be found in the previously mentioned reference.
6.2.5. Aerodynamic Resistance
As noted above, the aerodynamic resistance is a function of wind speed,
atmospheric stability and surface roughness. It is independent of the species being
deposited. Seinfeld states that this resistance can be viewed as being characteristic of the
resistance of momentum transfer to the surface. He offers the following relationship:
VM(zi,z0) = U,2/U(z1) (2)
where Vm(z1,z0) is the momentum transfer velocity, z\ is the height above the surface,
z0 is the roughness length, U^2 is the momentum flux and U(zi) is the average wind
speed at height, z\. Seinfeld goes on to state that if the aerodynamic resistance is the only
(or dominant) resistance to dry deposition, then Vd = Vm. Simply, what equation (2)
tells us is that the velocity with which a pollutant is deposited to a surface is proportional
to the momentum flux above that surface. While it is not currently possible to measure the
deposition velocity directly, it is possible to measure the momentum flux. Therefore, our
goal was to measure the flux of momentum to Lake Michigan, then infer the deposition
velocity of pollutants over the lake from these measurements.
6-20
-------
6.2.6. Momentum Flux Calculations
The flux of any quantity can be described as the transfer of that quantity through a
unit area per unit time. There are four standard techniques that are generally used for
calculating fluxes of various parameters. In order of preference, these techniques are:
1. eddy correlation (or direct correlation),
2. inertial subrange energy dissipation,
3. flux/gradient (or profile), and
4. bulk aerodynamic methods.
During the LMUATS, the fluxes of momentum, heat, moisture, CO2 and ozone
were of primary interest. While planning the measurement strategy before the intensives,
it was decided that the eddy correlation technique to estimate the fluxes of interest would
be used. This is the most direct procedure for measuring turbulent fluxes and is thus the
most accepted method. Unfortunately, considerable problems were encountered during
the intensives. Significant wave action was encountered on Lake Michigan during both
ship borne measurement periods. This caused many problems when trying to compute
the momentum flux from the measurements. As a result, several methods were used to
compute the momentum flux, in an attempt to remove wave motion from the data.
Methods 1, 2 and 4 were chosen for the analysis. Of the 3 techniques, only method 4
proved successful.
The following is a discussion of the three methods utilized in the analysis. The
discussions of methods 1 and 2 include explanations as to why these methods were
deemed unsuccessful. The discussion of method 4, the bulk aerodynamic method,
includes interpretations of the final results. Finally, this section concludes with
suggestions on how to carry out such measurements in the future.
6.2.7. Eddy correlation
The eddy correlation technique is the most direct method available for determining
the atmospheric flux of various quantities caused by the turbulent (or eddy) motions in the
atmosphere. In essence, the method tries to correlate fluctuations in the wind with
fluctuations in the quantity of concern. The ultimate goal was to estimate the relative
magnitude of vertical transport of pollutants to the lake surface. Therefore, the main
6-21
-------
concern was with making a correlation between fluctuations in the vertical component of
the wind and fluctuations in the species (or meteorological variable) in question. To
calculate the momentum flux described above, it was necessary to correlate fluctuations in
the vertical component of the wind with fluctuations in the horizontal component of the
wind.
At any instant in time, the vertical velocity may be broken into two parts: the
average velocity and some perturbation about that average. This is mathematically stated
as:
W = W + W' (3)
where W is the average value of W and W' is the perturbation about the mean.
Likewise, for the horizontal component of the wind:
U = U + U' (4)
In lieu of a lengthy, mathematical discussion, it can simply be stated that the time
averaged value of the correlation between the U- and W-components of the wind can be
written as:
uw = UW + U'W' (5)
where t/W is the product of the average U- and average W-wind components during
some time interval and U' W' is the average of the product of the U- and W-perturbations
during that same time interval. Over any appreciable length of time W - 0, so the second
term drops out (that is, air is "l/viously escaping into space). Therefore, equation (5)
reduces to:
UW = U'W' (6)
The quantity U'W' is referred to as the eddy momentum flux. That is, U'W' is the flux
of momentum that is caused by turbulent eddies in the atmosphere. When U'W' < 0, the
flux of momentum is to the surface (indicating the opportunity for pollutant deposition).
The larger the value of U' W', the better the chances for deposition. When U' W' >0, the
flux of momentum is away from the surface.
6-22
-------
Turbulent eddies come in a large variety of sizes, ranging from centimeters in
diameter to hundreds of meters. In order to properly measure the fluctuations caused by
the smallest of these eddies, measurements must be taken of the wind direction and speed
over very small time intervals. For this purpose, a set of "fast-response" instruments were
utilized. These instruments were able to sample various fields at a rate of 10 Hz (that is,
10 times each second). The instruments used are explained in detail in the Methods
section (Chapter 4). Calculations of U' and W' were made using a data analysis package
written for this study. Ten-minute averages of the momentum flux were then calculated
using the eddy correlation method described above.
The results of this analysis were unrealistic and alternated randomly from positive
to negative values. It should be noted that, while it is possible to have instantaneous
positive values of the momentum flux, the time-averaged values should always be negative
when calculations are computed for the surface layer. The questionable results were not
completely unexpected however, since the bow tower was constantly in motion due to the
rocking of the R/V Laurentian in the waves. As a result, the measured "W" component of
the wind was never truly just the "W" component. The data were contaminated by this
motion in two ways. First, the tower was often tilting, such that it was never really
vertical. For this reason, a true W-component was not measured. Rather, a combination
of the W-component and a horizontal component was measured. Pond et al. (1971)
found that without correction for an axis tilt of as little as 5°, such data were worthless.
The tilt of the bow tower exceeded this limit. As noted earlier, equipment was ordered
that would have allowed for corrections of these motions. However, the delivery of this
equipment was delayed and it was unavailable for use during the data analysis.
The W-component data were also contaminated by the vertical bouncing of the
boat and tower caused by the interaction of the ship with the waves on the lake surface.
As waves passed under the ship, the ship (and the bow tower) would move up and down
in response. This vertical motion was superimposed upon the actual W-component. It
was originally felt that this effect could be eliminated by using a mathematical technique
called spectral analysis. This technique is described below
The turbulent motions of the atmosphere can be thought of as a series of
fluctuations that repeat themselves with a certain frequency (e.g. once every second (1
Hz) or 5 times every second (5 Hz)). When the strength of the different fluctuations is
plotted versus their frequency of occurrence on a log-log plot, the result is a plot similar to
Figure 6-6. When these turbulent wind measurements are taken from a stable platform in
6-23
-------
Figure 6-6. Spectral Energy Plots for the Data Collected aboard the R/V Laurentian
24 July 1991 [W-component (top) U-component (bottom)].
W-coaponent, 24 JUL 1991
190010-190334 E.D.T.
.01 .1 1.0
Frequency (Hz = see-!)
-5/3
10. Of
Total U-component, 24 JUL 1991
190010-190334 E.D.T.
"TH
m • .i it: ;i
1;
ll"i ijlAnli (l 1!! ¦!!'
Ui
.01 .1 1.0
Frequency (Hz = sec"l)
¦5/3
6-24
-------
a neutral boundary layer, these spectral energy values fall off with a slope of -5/3 Any
variations to this -5/3 fall-off are said to be caused by external forcing. In this case,
external forcing was due to the bouncing of the ship in the waves. Note that the external
forcing in the data plot occurs at 0.3 Hz. This would represent an occurrence of once
every three seconds. This matches the observed frequency of the waves on the lake during
this particular measurement period.
The above analysis can be performed mathematically using a technique called
Fourier transform. This type of analysis takes a time series plot of the wind data and
mathematically transforms the plot into a sum of sine and cosine terms. These terms are
based on the different frequencies of the various fluctuations. Mathematically, this is
expressed as:
W{f) = £Af sin(//) + £Bf sin (ft) (7)
j f
where f is the "frequency" of the fluctuation and "A" and "B" describe the magnitudes of
the fluctuations. Since the frequency of the external forcing is known, the terms that
include this frequency or range of frequencies can be removed or filtered. A reverse
transform can then be computed, resulting in a new time series that is theoretically the
same as the original time series, minus the frequency caused by the external forcing.
A Fourier analysis was performed on the data for 24 and 25 July 91. The new time
series was then used to compute the momentum flux using the eddy correlation technique
described above. Again, the results were unrealistic. The values that were calculated
continued to show an oscillatory trend by alternating randomly between positive and
negative values. The magnitudes of the numbers were also quite small, indicating that the
filtering technique had removed too much energy and did not generate realistic results.
Since both attempts at using the eddy correlation method did not result in
reasonable values for the momentum flux, a decision was made to try several indirect
methods for calculating the momentum flux. These methods are discussed below.
6-25
-------
6.2.8. Energy Dissipation
At the beginning of this section, the generation of turbulence was attributed to
mechanical and thermal processes. The intensity of turbulence is often expressed as the
turbulent kinetic energy or (TKE). The turbulence that is often felt by airplane passengers
is obviously quite strong if it can cause a plane to shake. This would be an example of
turbulence with high TKE. Weaker turbulence would be classified as having lower TKE.
Once formed, turbulent eddies will gradually break into smaller eddies due to their
interactions with each other. Eventually, the eddies are small enough that molecular
viscosity or friction will cause them to dissipate, transferring any remaining energy into
heat. A range of middle-sized eddies exisits such that the change in the TKE of the eddies
is not attributed to either the production processes mentioned earlier, or the effects of
viscosity forces. Rather, the eddies exchange energy inertially from the larger eddies as
they break up. This range of eddy sizes (and their corresponding frequencies of
occurrence) is called the inertial subrange. The inertial subrange can been seen on the
spectral power (spectral energy / sample size) plots in Figure 6-6 as the part of the plots
that show a -5/3 fall-off. The "inertial subrange" plays a prominent role in the second
analysis technique, energy dissipation.
The energy dissipation technique is based on the assumption that in near neutral
conditions, the rate of production of turbulent energy at a particular height is equal to the
rate of dissipation of turbulent energy at that same height. Mathematically, this is stated
as:
-U' W —— (production)= 8 (dissipation) (8)
dz
where 8 is the turbulent energy dissipation rate. Pond et al. (1971) showed that the
dissipation rate can also be related to the momentum flux using the following relationship:
-U'W' =(K8Z)2/3 (9)
where K is the von Karman constant (= 0.40 ), 8 is the dissipation rate, and z is the height
at which the measurements were taken. Therefore, if the dissipation rate is known, the
momentum flux can be inferred indirectly.
6-26
-------
The dissipation rate can be determined by returning to the spectral energy plots
discussed previously (Figure 6-6). The energy dissipation rate for a period in time is
related to the spectral power plot for that same period by the following relationship:
S(k) = ak2/3 S2/3k"5/3 (10)
where S(k) the spectral energy at a particular wave number, k is the wave number ( = 271
n/U, with n = frequency ) and is the Kolmogorov constant ( =0.55). To determine the
dissipation rate for any given period, the following steps were undertaken:
1. a Fourier transform was performed on the data,
2. the inertial subrange was located,
3. any frequency within this range was picked and
the corresponding value of S(k) were recorded, and
4. equation (10) was solved for values of £.
Ten-minute average values of 8 were computed and then substituted into equation (9) to
acquire the ten minute average momentum flux values. A similar analysis can be
performed to acquire dissipation rates for other scalar variables, such as moisture and
heat. These dissipation rates can then be used to compute moisture and heat fluxes,
respectively.
To compute the momentum flux, a spectral analysis on the horizontal component
of the wind must be performed. Figure 6-6 shows such a plot (bottom). In the frequency
range of 0.1-0.3 Hz, there is a slight deviation from the -5/3 slope. This is another
signature of the ship motion. This signature is not as prominent as seen in the W- wind
component plot (top-Figure 6-6). The advantage of the dissipation technique is that only
the spectral energy for one frequency needs to be considered. Therefore, the analysis
could be completed using a frequency outside the 0.1-0.3 Hz range (i.e. a frequency range
that was much higher than the frequency of the water waves). Thus, the information used
for this analysis was in a range unaffected by the bouncing of the ship. A frequency of 0.7
Hz was chosen for this analysis.
The results of the calculations were not consistent with previous research. While
not shown here, the magnitudes of the momentum flux values calculated were much
higher than the results published by other investigators for measurement programs similar
to those during the LMUATS. One possible source of error in the data is that the bow
6-27
-------
tower was located only 3 meters ahead of the R/V Laurentian. As a result, the
interaction of the horizontal wind with the wake of the ship likely distorted the horizontal
wind flow. If this occurred, the horizontal turbulence field measured would not be
representative of the actual turbulence field over the lake (per conversation with Tim
Crawford, NOAA-ATDD, Oak Ridge National Lab, 1992).
As noted above, similar analysis was possible for scalar variables such as moisture.
Since fast-response measurements of water vapor were made during the period 24-25 July
91, analysis was performed on this data using the energy dissipation method. The values
of the moisture flux calculated from this analysis were much higher than other data
published by other researchers. Upon further analysis of the moisture data, additional
evidence indicated that the horizontal wind data were tainted. Computation of the
moisture flux via the energy dissipation method requires use of both the energy dissipation
rates for moisture and horizontal wind. When values of the energy dissipation rates for
moisture were compared with those published by other researchers, the values compared
rather favorably (e.g. Pond et al. (1971) during BOMEX). As a result, the difference
between the moisture flux values and those from other studies can be traced back to
erroneous values of energy dissipation rates for the horizontal wind.
6.2.9. Bulk Aerodynamic
The final method used to compute the momentum flux was the bulk aerodynamic
method. This method computes the momentum flux from a bulk, aerodynamic quantity
such as the five-minute average wind speed at the height in interest. While this method is
a very indirect method of computing the momentum flux, the influence of ship motion is
not a consideration. Over a long period of time (e.g. five minutes), these motions will be
averaged out. The resulting data will not require that correction', be used. The
momentum flux is calculated using the following relationship:
-U*W'= CDU2 (11)
where Cd is the momentum drag coefficient and U is the mean horizontal wind (Arya,
1988).
One disadvantage of this simplistic approach is the value of Cd varies considerably
as the characteristics of the underlying surface change. Therefore, the drag coefficient is
6-28
-------
really a function of the roughness length (a relative measure of surface roughness). Values
for the roughness length vary from 10*4 meters for a calm, open sea surface to 3 meters
for urban areas. Arya (1988) uses the following relationship to calculate values of the
drag coefficient:
CD = K2 / [ In ( Z/ Z0) ]2 (12)
where K is the von Karman constant, Z is the height of the measured wind speed in
equation (11) and Zo is the roughness length. For our calculations, a value of 2 x 10"4
was used for Zo, yielding a value of 0.0015 for Cd.
The momentum flux results obtained with this method are given in the Appendix.
Unlike the results using the previous two methods, the values obtained using the bulk
aerodynamic method do compare favorably with values obtained by others who have
carried out similar experiments. The results are shown in a graphical form in Figure 6-7
on the following page.
6.2.10. Interpretation of Bulk Aerodynamic Results
From Figure 6-7, it can be seen that the momentum flux values were quite small
during the afternoon hours of 24 July, with values remaining less than 0.010 m2/sec2 until
approximately 6PM. Recalling previous discussions of deposition velocities, the
deposition velocities of pollutants during afternoon hours can be inferred to be low.
Therefore, long-range transport would have been likely throughout the afternoon. This
hypothesis is supported by the chemical measurements made for the period of 9AM-9PM
EDT, showing high levels of many pollutants at both the R/V Lanrentian and South
Haven sites.
That evening, a cold front approached the southern Lake Michigan Basin. A
gradual increase in the momentum flux values was observed during the front's approach.
Surface weather observations indicate that the front passed by the R/V Laurentiaris
position between 8PM and 8:30PM EDT. Shortly after the front's passage, the values of
the momentum flux increased dramatically. The flux values peaked near 0.110 m2/sec2
around 1AM EDT on 25 July, before diminishing somewhat for the remainder of the
measurement period (until 8PM EDT 25 July ). During this period, deposition velocities
were likely quite high.
6-29
-------
Figure 6-7. Temporal Variation of the Momentum Flux Measured Aboard
the R/V Laurentian 24-25 July, 1991.
ON
I
o
0.12
CM
t> 0.1
8
w 0.08
E,
a 0.06
£ .
2 0.04
0.02
Frontal
Passage
H h
H 1 1 1 1 H
H t-
-t H
H H
o
o
o
o
o
o
o
o
o O
O
o
O
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
O
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
in
to
1^
00
o>
o
CM
CO
T-
CM
CO
in
o
*—
CM
CO
in
<£>
r-
oo
CT>
o
T—
T—
T-
T—
T—
CM
CM
CM
CM
T-
*—
T-
V-
T-
T—
T-
*-
CM
Timq of Day (E.D.T.)
24-Jul
25-Jul
-------
It should be noted that the information inferred from the momentum flux measurements
parallels the conclusions derived from stability data. However, it is difficult to draw any
concrete conclusions without the support of results from other flux measurements of
moisture, heat and CO2. Had attempts at using the eddy correlation method been more
successful, these additional fluxes could have been calculated. While calculation of the
momentum flux via the bulk method was possible, the measurement techniques used for
the other variables prohibited the use of gradient or bulk methods to accurately calculate
their respective fluxes.
6.2.11. Suggestions For Future Flux Measurements
Despite the difficulties encountered during the measurement campaign, the
measurement of meteorological and chemical fluxes to and from the Great Lakes is a very
attainable goal. However, special consideration must be given to several issues.
When the objective of a field program is to measure the small-scale structure of an
atmospheric turbulence field, any motion experienced by the sensing instrument may well
be of the same magnitude as the turbulence being measured. This problem is of special
importance when trying to make direct, in situ measurements of a marine surface layer (as
was the case for this project). The ideal situation would involve no instrument motion, but
this is often unrealistic especially during ship borne measurements. The degree to which
the data are contaminated by the motion of the sensing instrument depends on which flux
measurement technique is being employed and to what type of platform the instrument is
attached.
6.2.11.1. Ship Borne Platforms
If the flux measurements are being made from a ship borne platform, such as in this
study, several precautions must be taken. It is inevitable that the sensing instrument will
experience some form of motion. Even during nearly calm wind conditions some degree
of wave action will be present, manifesting itself in the motion of the ship and affecting the
sensing instruments.
If the bulk aerodynamic method is to be used, the effects of the ship motion will be
averaged out by the time averaging of the data. This should be true whether the
instrumentation used is either the fast or slow response variety.
6-31
-------
If the gradient method is used, the flux calculations of the variable of interest will
require the accurate measurement of the vertical gradient (or change with height) of that
variable. This task will be very difficult. As the instruments move vertically due to wave
motions influencing the ship, the levels at which the measurements will actually be taken
will be changing constantly. A great degree of instrument separation would be required to
alleviate this effect which is usually not practical. As a result, this method is not suggested
for use on ship borne platforms.
The energy dissipation method of flux calculation has been shown to be successful
in previous field programs. As discussed above, this analysis technique allows the
investigator to retrieve the flux from the fluctuations of a particular frequency outside the
frequency range of the ship/sensor motions. The only requirement is that the chosen
frequency resides within the inertial subrange. While zero ship/instrument movement
would be preferred, the energy dissipation method allows for the measurement of
meteorological and scalar fluxes under low to moderate wave conditions. The energy
dissipation method was suggested as the most practical method for ship borne flux
measurements by several investigators, including Dr. Ken Davidson (Naval Post-Graduate
School) and Dr. Tim Crawford (Oak Ridge National Laboratory - ATDD). After
completing analysis of the data collected during the LMUATS, it appears that this method
would be the preferable method of analysis.
While the eddy correlation method is the most direct method of flux calculation,
significant post-processing of the data will be necessary to remove the effects of the ship
motion from the data. For this to be possible, it will be necessary to account for the
position of the instrument sensor at all times. Pond et al. (1971) were able to use the eddy
correlation method during BOMEX, though the floating platform that they used was said
to be very stable compared to a surface ship. Post-processing procedures included use of
a complicated mathematical scheme called coordinate rotation. According to Pond, this
method was based on some strong assumptions, but seemed to produce values which
compared favorably with calculations using the bulk and energy dissipation methods. The
results from similar calculations on ship borne data would be suspect, at best.
In each of the above methods, significant consideration would have to be given to
the physical placement of the instrumentation with respect to the ship and/or instrument
platform. Any physical object will cause a distortion of wind flow. Care must be taken to
place the sensing instrument in an area unaffected by platform or ship waking effects.
6-32
-------
6.2.11.2. Stationary Platforms
The use of stationary platforms for the measurement of fluxes to and from the
Great Lakes would greatly simplify the post measurement analysis of data. Depending on
the stability of the platform, instrument motion effects could be virtually eliminated.
However, several new problems would result. Few stationary platforms, if any, exist
which would be suited for such an application. During this study, it was noted that several
large, cement structures were located offshore of Chicago. These structures house the
intake ducts for the city's water supply. Such structures could serve as sites for instrument
towers for flux measurement instruments. However, these structures are not easily
accessible and would likely cause considerable distortion of the wind flow. The tower
would need to be constructed so as to place the instrumentation out of the area affected by
the distortion.
A second idea would be to construct a tower on a sandbar just offshore. Such a
structure would need to be strong enough to withstand the considerable force produced by
the motion of the lake waters. If this could be accomplished, the platform would also
allow for the measurement of fluxes using any of the previously described methods. One
question that would need to be answered, however, is how representative the flux
measurements would be of fluxes occurring over the center of the lake. Also, if such a
lake shore platform were built, data collection might be limited to onshore flow conditions
only. Depending on shoreline topography, flow from land to the measurement platform
could contain motions induced by flow over the lake shore topography. Such flow would
not be representative of conditions at mid-lake, though these results would be useful for
studying near shore effects on fluxes and deposition.
6-33
-------
Chapter 7
Source Apportionment/Receptor Modeling
7.1. Introduction
An understanding of the transport and fate of contaminants emitted to the
atmosphere is crucial for determining potential impacts on human health and the
environment. Source characterization is a powerful tool for the assessment of emissions
from local sources. There are a number of methods available which identify sources or
source regions contributing to the pollution load measured at a given receptor. These can
be categorized into statistical methods, chemical and isotopic trace methods, individual
particle analysis, meteorological methods including trajectory techniques, dispersion
modeling and various combinations of these approaches. Some of these methods are
briefly described by Zweidinger et al. (1990) and Sweet and Vermette (1992 and 1993) in
their reports on VOCs and trace elements in Illinois and St. Louis. These are further
expanded upon in their references as well as in Gordon (1988), Henry et al. (1991) and
Hopke (1991).
7.1.1. Source Characterization Techniques
Traditionally, dispersion models have been the method of choice for calculating
source-receptor relationships for air pollutants. These models require detailed emissions
inventories for the sources of the pollutants of interest (e.g., S02, HCs, NOx, etc.).
However, even if the dispersion modeling could be done accurately, it is very unlikely that
the source emissions inventories for the pollutants of interest would be adequate. As
discussed earlier in this document, emissions inventories for the criteria pollutants have
many short-comings. These inadequacies are even more severe for hazardous pollutants
or pollutants which have large contributions from fugitive emissions, natural sources and
dusts.
7-1
-------
The limitations of the dispersion-oriented methods led to the development of
receptor-oriented models. These have been used effectively to identify the natural and
anthropogenic sources of atmospheric particles in both urban and rural areas (Gordon,
1980, 1988). Receptor models assess contributions from various sources of atmospheric
pollutants based on measurements at sampling or receptor sites. The measurement used
most often and most successfully is the elemental composition of atmospheric particulate
matter (Dzubay et al., 1988; Keeler et al., 1990). Pollutant properties, such as particle
morphology, particle size distribution, vapor to particulate partitioning, and elemental
composition, can then be used to identify sources or source types. Meteorological
information can also be used in conjunction with receptor models for assessing the
contributions of distant sources. These are often referred to as hybrid models (Keeler,
1987).
Another method of source characterization is the chemical element balance, or
chemical mass balance (CMB) method. It is based upon the premise that the emissions
characteristics of various source types are different enough in chemical and elemental
composition, as well as physical size and morphology so that one can identify their
contributions by measuring these characteristics in samples collected at a receptor site.
Thus, identifying the number and types of important sources of air pollutants is an
important first step in the application of CMB models to apportion the sources of air
pollutants measured in a specific urban area. The CMB models assume that the
composition of all contributing source types are known. However, this is often not the
case due to the fact that the sources are not easily sampled and/or the source classes have
widely varying compositions (Henry, 1991). Thus, the lack of specific source profile
information is an important limitation of the CMB method and this ultimately limits the
application of this approach. Particularly, emissions data have been sparse in the past,
particularly for SOCs, but the situation is improving.
While the CMB method has been applied primarily to urban-scale data, Rahn and
Lowenthal (1984, 1985) applied this technique to the regional signatures they identified to
apportion sulfate and trace metals observed on particulate matter. The application of
receptor models to regional and global scale problems has been controversial and has yet
to be fully developed to the level necessary for it to be definitive in nature. However, an
independent verification of the appropriateness of the trace element ratio approach was
performed, and it was indicated that this technique can be quite powerful (Keeler, 1987,
Keeler and Samson, 1987).
7-2
-------
Statistical methods have been developed that use information on the chemical
composition of aerosols to study the contribution of sources or even source regions to the
contamination at a given receptor. The APCA method, originally applied to total
suspended particulate concentrations (particles measured in both the fine and coarse
fractions), determines the composition of major source components, such as coal
combustion, crustal, or sea-salt, which may have contributed to the measured
concentrations at the receptors. Further improvement of this modeling method was
realized when APCA was applied to aerosol elemental concentration measurements in
separate particle size fractions (Li and Winchester, 1990). The results of this APCA
application provided the basis for the interpretation of coupled chemical reactions and
physical processes in remote locations. Also, information concerning atmospheric aging
processes and, therefore, the history of the aerosols was obtained. For example, it was
noticed that some crustal components were found in all particle size fractions.
Carbonaceous fuel combustion pollutants were identified by the presence of S, Si and CI
and the absence of A1 in all size fractions. This was the result of the release of volatile SiO
from the reduction of silicon dioxide by carbon during the combustion of coal with high
ash content. Since A1 is non-volatile during coal combustion, its absence when Si and
other metals were present indicates that the aerosol was generated from burning high-ash-
content coal.
Single or Individual Particle Analysis (IPA) has been applied in aerosol research to
investigate the sources and morphology of the collected atmospheric particulate matter
(Dzubay and Mamane, 1989; Mamane and Dzubay, 1987; Mamane, et al., 1986; Sheridan,
1989). Single Particle Analysis provides critical size distribution information that can be
used directly to calculate the deposition of pollutants as a function of size. In a review by
Sheridan (1989), it was observed that particles emitted by anthropogenic sources, such as
carbon soot and coal combustion sphei^s. occurred simultaneously with the highest
concentrations of H2SO4 droplets. Thus. IPA can be used to estimate the source
apportionment, as well as physical and chemical processes occurring during long-range
transport of air pollutants. One analytical method for IPA is the use of scanning electron
microscopy (SEM). Mamane (1990) utilized this technique to estimate the contribution of
refuse incinerators to the pollutant loadings of Philadelphia. SEM was also used in the
analysis of atmospheric particles collected during the Green Bay Aerosol Study (Mamane
et al., 1993). The particle size distributions and chemical compositions provided
important information that was utilized in receptor modeling as well as deposition
calculations.
7-3
-------
The utility of back-trajectory analysis for identifying probable source regions
contributing to elevated levels of atmospheric contaminants was illustrated over a decade
ago (Samson, 1980). In that work, trajectories corresponding to high and low resultant
sulfate concentrations measured at Allegheny Mountain were plotted separately to give a
picture of the meteorological flow patterns associated with each of the two sulfate
categories. Keeler (1987) also utilized this technique 6-years later to investigate the
meteorological conditions associated with high and low sulfate concentrations measured at
the same sites during the Allegheny-Laurel-Deep Creek Lake Experiment.
In order to assess atmospheric deposition to a receptor such as Lake Michigan, it
is necessary to know the relative contribution of various sources to the total mass loading.
Emissions to the atmosphere of the compounds discussed in this report are multitudinous.
Most of the compound classes result from human activities, but in some cases, natural
sources may make a contribution as well. For example, PAHs are released both by
anthropogenic combustion and forest fires, while Hg is emitted during coal combustion
and naturally from mines and soils. Techniques for deducing sources included in this
section are trajectory and meteorological analyses, individual particle analysis, emissions
inventories, multi-variate statistical methods, and chemical mass balance (CMB) models.
7.2 Meteorological and Trajectory Analysis
On-site meteorological parameters, such as wind speed and temperature
(maximum and average over the 12-hour sampling period), were also used to investigate
the simple chemical and elemental relationships in the measured data. The maximum 1-
hour and the 12-hour average temperatures for each sampling period were calculated, and
were found to be positively correlated at South Haven with the gas phase species ozone,
HNO3 (r > 0.75), SO2 (r = 0.66), particulate SO42" (r = 0.62), and aerosol acidity (r =
0.61). This is not surprising since the highest concentrations of these pollutants, mostly
secondary in nature, were found when there was a strong flow from the southwest and
higher temperatures. The average wind speed measured over the sampling period was
also correlated with the pollutant measurements. Weaker correlations were observed for
the same species (r=0.5), suggesting that wind speed was not as strong an influence on the
observed concentrations during the study.
Attempts to understand the sources and transport of hazardous air pollutants are
only fairly recent. To begin our analysis of the meteorological transport of the hazardous
7-4
-------
pollutants, it is instructive to investigate the meteorological conditions associated with
both high and low levels of compounds or measurements for which we have some past
understanding (i.e., PM10 or aerosol sulfate). Figures 7-1 through 7-3 show the mixed-
layer trajectories associated with elevated PM10 concentrations (> 50 (ig/m3) measured at
the three land-based measurement sites during the LMUATS.
Several interesting observations can be made from these figures. First, the highest
concentrations of PM10 at each site are associated with transport from the southwest.
Also, on all five days when PM10 > 50 fig/m3 in South Haven, the PM10 at IIT and
Kankakee also exceeded this amount.
In addition, both IIT and Kankakee recorded PM10 >50 on the same days
except for one, 16 July. The fraction of the PM10 measured in the fine mode (<2.5 (im)
decreased on the days when PM10 >50 ng/m3 at both Kankakee and IIT. This is due to
the fact that the fine-fraction mass is comprised primarily of sulfate aerosol in various
states of neutralization. On average, about 60% of the fine mass measured at all of the
sites was sulfate, associated ions (e.g. NH4"1", H+) and water.
The trajectory analysis of the PM10 measurements at the 3 land-based sites forced
a consideration of two large urban/industrial source regions that may actually contribute to
substantial deposition of HAPs. The Chicago/Gary area was the initial focus of the study
and is directly adjacent to Lake Michigan. The second important urban/industrial area is
the East St. Louis-Granite Citv-St. Louis area which has similar insudtrial sources and
magnitudes as the Chicago/Gary area. Both of these source regions contain a wide variety
of industrial point sources including many known to emit HAPs. Emissions information
for the important sources in the two regions was obtained to help us in interpreting the
chemical data as well as for identifying source profiles for the subsequent CMB analysis.
The southeast Chicago source region is dominated by iron and steel and related
industries (Table 7-1). These industries are concentrated along either side of the Calumet
River. In addition to direct industrial point source emission, fugitive emissions from
storage areas and stockpile operations have been determined to be of measurable
significance (Figure 7-4). Large tracts of land are utilized for storage of coal, slag, scrap
steel, limestone and other raw materials. Additional significant point sources include grain
handling, landfills, chemical production and other miscellaneous manufacturing. The east
St. Louis/Granite City source region is dominated by iron/steel related industries, metal
smelting and organic chemical production (Table 7-2)
7-5
-------
Figure 7-1. Mixed-layer Backward Trajectories Associated with PM10
Concentrations > 50|ig/m3 in South Haven, MI.
7-6
-------
Figure 7-2. Mixed-layer Backward Trajectories Associated with PM10
Concentrations > 50pig/m in Chicago, IL.
7-7
-------
Figure 7-3. Mixed-layer Backward Trajectories Associated with PM10
3
Concentrations > 50ng/m in Kankakee, IL.
7-8
-------
Table 7-1. Southeast Chicago Point Source Information.
Source
Source Type
1991
Particulate
Emissions
Cargill, Inc.
Marketing Grain
54
Chicago Blast Furnace
Steel Manufacturing
324
Chicago Coke Plant
Steel Manufacturing
152
CID Landfill
Landfill Site
4
Cinders
Slag Processing
143
Con-Ed Peaking Units
Electric Utility
7
Continental-Elv B
Marketing Grain
121
Domtar Industries
Refining Sodium Chloride
13
Ford Motor Company
Auto Manufacturing
9
General Mills, Inc.
Milling Grain
154
Great Lakes Carbon
Petro & Coal Production
7
Heckett En. Harsco
Slag Processing
111
Heckett Eng.
Slag Processing
77
Heckett-Plant 27
Slag Processing
37
Inland Metals
Refining Nonferrous
1
Interlake-Riverdale
Steel Manufacturing
374
International Materials
Marine Cargo Handling
17
Jay's Foods
Food Preparation
12
LTV Steel
Steel Manufacturing
515
Marblehead Lime
Lime Manufacturing
130
Mississippi Line
Marine Cargo Handling
12
PVS Chem;cals
Inorganic Chemicals
130
Rail-to-water
Marine Cargo Handling
12
Riverdale Plating
Plating
19
SCA Chemical
Refuse Disposal
16
Sherwin-Williams
Paint Manufacturing
2
Stauffer Chemical
Fertilizer Manufacturing
21
Stolt Terminals
Warehouse & Storage
7
U.S. Steel-Southworks
Steel Manufacturing
110
Adapted From Sweet et al., 1990
7-9
-------
Table 7-2. East St. Louis and Granite City Point Source Information.
1991
Particulate
Emissions
Tons/Year
Amax (Big River) Zinc
Zinc Smelter
120
Archer Daniels Midland
Soybean Processing
107
Bulk Service
Marine Cargo Handling
104
Central Soya
45
Cerro Copper Products
Secondary Copper Smelter
49
Corn Sweeteners, Inc.
Grain Processing
2
Ethyl Petroleum Products
Organic Chemicals
3
Granite City Steel
Steel Manufacturer
2,282
International Mill Service
Ground Minerals
49
Jennison Wright Corporation
Wood Preserving
13
Kerr-McGee
Wood Preserving
8
Midwest Rubber Reclaiming
Reclaimed Rubber
376
Monsanto
Organic Chemicals
168
Nestle Company
Food Manufacturer
36
Pfizer Pigments
Inorganic Pigments
157
Phillips Pipeline Company
Petroleum Terminal
1
Pillsbury
Grain Handling
132
SCI
Secondary Aluminum Smelter
52
St Louis Lead Recyclers
Lead Recycling
5
St. Louis Slag Products
Slag Processing
242
Tara Corporation
Secondary Lead Smelter
74
Trade Waste Incinerator
Hazardous Waste Incineration
71
U.S. Army
Construction
10
Union Electric
Electric Utility
8
Adapted from Sweet et al.. 1990
7-10
Source Name Source Type
-------
Figure 7-4. Selected Point Sources in Southeast Chicago.
(From Sweet et al., 1990)
7-11
-------
As previously discussed herein, the highest measured pollutant concentrations at
all of the land-based sites (Kankakee, IIT, and South Haven) were associated with air flow
from the southwest. On 17 July 1991, air mass transport was from the southwest (Figure
7-1). The backward mixed-layer trajectories suggest that the regional air mass passed
over both major source regions before passing over Lake Michigan. PM10 concentrations
measured at the 3 land-based sites were above 60 |ag/m3 at both Kankakee and South
Haven and above 80 ng/m3 at IIT. Almost 50% of the PM10 on 17 July was particulate
sulfate with 30 (ig/m3 measured at Kankakee.
Several Illinois State National Air Monitoring Stations (NAMS) are located
between the East St. Louis and southeast Chicago source regions. On 17 July 1991,
samples were taken at these non-continuous sites (a one-in-six day sampling day). PM10
and total suspended particulate results are shown for these sites in Figures 7-5a and 7-5b,
respectively. The results indicate that total suspended particulate is predominantly
influenced by strong local sources whereas PM10 and sulfate concentrations appear to be
driven by the strong regional coal combustion sources. The spatial distribution of the
coal-fired utilities is shown in Figure 7-6 and the point sources are listed in Section 404 of
the 1990 CAAA. Five sites of importance to this analysis are listed in Table 7-3.
The Nilwood site, located in Macoupin County, has no local sources and is used as
a background reference by the Illinois Environmental Protection Agency. The Peoria site
in Peoria County and the Decatur site in Macon County have similar industrial point
sources and correspondingly similar measured ambient concentrations. Oglesby, however,
has two large cement manufacturing facilities that are major sources of particulate
emissions. The total suspended particulate and PM10 concentrations observed at Oglesby
were 328 (ig/m3 and 167 ng/m3, respectively. The Joliet site in Will County had typical
total suspended particulate values, but PM10 concentrations were elevated reaching 77
Hg/m3. This appears to be the result of transport from the Oglesby area.
Information from several Illinois State Special Purpose Monitoring Sites (SPMS)
was examined to explain several observed concentration deviations. These SPMS are
listed in Table 7-4. The Randolph County Sites are sited in a rural area to monitor a coal
extraction and processing operation. The measured PM10 concentrations for these sites
were 126 |J.g/m3 and 91 |ig/m3 respectively. The remaining SPMS sites examined were
the Chemetco sites in Madison County and the Chicago-Horsehead sites in Cook County.
These sites appeared to contribute to the observed elevated Pb concentrations.
7-12
-------
Figure 7-5a. PM10 Concentrations in the Study Region on 17 July 91.
Figure 7-5b. Total Particulate Concentrations in the Study Region on 17 July 91.
' " J
V I
4 ii
7-13
-------
Table 7-3. Selected Illinois Air Monitoring Sites and Surrounding Source
Information.
Monitoring
Site
Source
Source Type
1991
Particulate
Emissions*
Tons/Year
Decatur
Archer Daniels Midland
Corn Milling
5,000
Firestone
Tire Manufacturer
179
A.E.Staleys
Manufacturing
Corn Milling
6,800
Wagner Casting
Iron Foundry
458
Muller Company
Gray Iron Foundry
74
Joliet
Commonwealth Edison
Electric Utility
3,400
Crosfield Chemicals, Inc.
Industrial Inorganic Chem.
12
Desoto Inc.
Soap & Detergent
21
Olin Corporation
Industrial Inorganic Chem.
1,220
Oglesby
Illinois Cement Company
Cement Manufacturer
42,621
Lone Star Cement Co.
Cement Manufacturer
564
Peoria
Archer Daniels Midland
Corn Sweetener Plant
300
Caterpillar
Assembly Plant
300
Keystone Steel & Wire
Steel Wire & Related Prod.
200
Nilwood
None
* Based on Illinois Environmental Protection Agency, Emissions Inventory
7-14
-------
Table 7-4. Selected Illinois Special Purpose Air Monitoring Sites and
Surrounding Source Information.
Monitoring
Site
Source
Source Type
Particulate
Emissions*
Tons/Year
Lead
Emissions*
Tons/Year
Chemetco
Chemetco
Secondary Smelter
567
78
Clark Oil & Refining
Oil Refinery
167
Shell Oil
Oil Refinery
16,174
Chicago-
Horsehead
Horsehead Resource
Development Co.
Hazardous Waste
Recycling
1,161
250
Randolph
County
Illinois Power-Baldwin
Electric Utility
30,118
Peabody Coal
Company
Open Pit Mine &
Coal Prep. Plant
66,111
*Based on 1992 & 1993 Illinois State Emission Inventory
7-15
-------
Figure 7-6. Coal-Fired Power Utilities in the United States.
7-16
-------
7.3. Scanning Electron Microscopy (SEM)
Six aerosol samples collected at the IIT site during the LMUATS were analyzed
by Scanning Electron Microscopy and Energy-Dispersive X-ray spectroscopy
(SEM/EDX). The objective of the SEM/EDX analyses was to characterize a
representative number of individual aerosol particles for each sample in order to provide
additional information about particulate sources impacting the IIT site. SEM/EDX
analyses revealed significant differences in aerosol composition among the six samples
which in turn reflected differing mixes of sources. The observed differences generally
support results obtained by X-ray fluorescence (XRF) and by receptor modeling applied to
the same sample set. The results are described in the following section.
7.3.1. SEM Results
Results of the SEM analyses for both coarse and fine particles are tabulated in the
particle data tables in the Appendix. Examples of the tablulation are shown here for the
sample collected on 21 July 1991 (Tables 7-5 and 7-6). Particles are sorted into three
major classes: industrial, mineral, and organic with subclasses within each major class.
These tables report the actual number of particles characterized for the given filter. The
information provided in the header for each table includes the area of the filter scanned by
SEM as well as the effective sample volume for the size fraction analyzed. Note however
that the data have not been normalized to these parameters so that direct comparison of
different samples on a number basis is not valid. It is however meaningful to compare the
composition of different samples expressed as percentages.
Photomicrographs of selected industrial particles are shown in Figure 7-7(a-n).
Fibrous strands in the background are Teflon filter material. The high density of sulfate
particles is clearly illustrated in Figure 7-7a which also includes an iron sphere of 1 (j.m.
Iron spheres, less than 0.5 |im in diameter are evident in aggregated long chains in Figure
7-7b. Figures 7-7c, 7-7d, and 7-7e show an iron aggragate, an aluminum particle, and
aluminum oxide. Mineral particles are shown in Figures 7-7f through 7-7j including
ammonium sulfate, magnesium chloride, aluminosilicate flyash, silica spheres, and calcium
aluminum silicate particles. Biological and other organic particles are shown in Figures 7-
7k through 7-7n. These large particles include soot, spores and organic material.
Aluminum oxide particles were found in abundance on the sample collected on 8 August
1991.
7-17
-------
Table 7-5. Tabulation of Coarse Fraction Particulate Matter Collected on 21 July 91
Analyzed by SEM.
LAKH MICHIGAN URBAN AIR TOXICS STUDY
SAMPLE ID: 1027 COARSE FRACTION
SITE: IIT
DATE OF SAMPLING: 7/21/91
VOLUME (m3): 11.96 SEM AREA (urn2): 19,200
MINERALS
PARTICLE SIZE (urn)
1.5-2.1
<3.1
<5 1
<7.1
<10
>10
Total
%
Silicate
19
12
7
3
41
31.8
Quartz
5
1
6
4.7
Ca-rich
4
4
3
11
8.5
Ca & S
13
2
5
20
15.5
Dolomite
0
0.0
Dol. Limestone
2
4
6
4.7
Other
2l)
3Z)
5
3.9
ORGANICS
MINERALS
89
69.1
Spores
1
2
3
2.3
Plant Debris
0
0.0
Soot
2
1
3
2.3
INDUSTRIAL
ORGANICS
6
46
Flv Ash
3
3
2.3
Al-rich
0
0.0
Fe-rich
9
7.0
Zn-rich
0
0.0
S-rich
7
7
5.4
Other
84>
73)
15
11.6
INDUSTRIAL
34
26.?
TOTALS I 72
tt
21
3 | 0 | 0 [ 129
100 0
% | S6
26
16
2 | 0 | 0 1 100
1) Includes (1) Si-S-Ca-Ti-Fe-K-Mg smooth particle surrounded by film.
2) (1) Iron-rich; (1) Ca-S-Cl-Mg-K; (1) Aluminum-rich.
3) Mostly spheres.
4) Mg-Cl-Fe-S-Mn; Fe & Zn; NaSC>4, S-P-K-Ca (possibly fertilizer); NaS04 & CaCO^,
Mn-Fe-Zn spheres (possibly smelter or blast furnace); possibly K phosphate
5) Mn & Fe; Zn-Cl-Fe-Ca-Si; Pb-Cl-K-Zn (incinerator); Al/Si-Mn-Cl-S-Ca-Ba-Fe-Zn-Pb;
(NH4)2S04 & KS04; Na-Si-S-Ca-Ti-Fe-K-Mg.
7-18
-------
Table 7-6. Tabulation of Fine Fraction Particulate Matter Collected on 21 July 1991
Analyzed by SEM.
LAMM MlCBiGAN URBAN AO* TOXICS
SAMPLE ID: 1027 FINE FRACTION >ITE: IIT
DATE OF SAMPLING: 7/21/91 VOLUME (m3): 1.26 SEM AREA (nm2): 12,000
MINERALS
PA1
*TICLE SIZE (Jim)
<0*5
<0.7
<0,9
1
1.2
Other
9^)
j 10)
211)
91
3U)
19
22.4
INDUSTRIAL
58
6 8.3
TOTALS
36
7 | 9
10 | 10 I 13
&5 ! JQO.O j
%
42
8 j 11
12 j 12 J 15
m
1) Majority of Fe-rich particles are combustion spheres. 8) All contain Fe as well as Mn and/or Zn.
2) Fe & Mn; Fe & Cr (sphere); Fe-Ca-K-Ti-S.
3) Includes (4) Zn & S. (1) Zn & Pb, (1) Zn & Na.
4) Zn-Na-S-Cl.
5) Zn-S.
6) Avg. diameter ~ 0.3-0.4 urn.
7) S & trace Zn.
9) Pb & trace of Fe and Zn; Fe-Zn.
10) K-S-Na.
11) Possible NH4CI sphere.
12) (1) CI particle; (1) Na & C particle.
13) Pb-Zn-Cl; Si-Fe-S-Zn.
7-19
-------
Table 7-7 below summarizes the results of the SEM analyses and highlights
similarities or dissimilarities among the samples. Silicates and Ca-rich minerals (limestone,
gypsum, dolomite) from soils, road dust, etc. dominate the coarse fraction of all samples
except for the one collected on 8 August. Elements associated with these minerals include
Mg, Al, Si, K, Ca, Ti, and Fe. The elements Cr, Mn, Cu, Zn, and Pb are associated almost
exclusively with industrial-derived fine particles. Iron concentrations are generally split
between silicate minerals and industrial sources - primarily mineral sources in the coarse
fraction and industrial sources in the fine fraction.
Samples collected on 21 July and 8 August were strongly impacted by industrial
sources and have the least dolomite and dolomitic limestone content. The particulate
sample from 2 August reflected the least impact from industrial sources as it was rich in
quartz and Ca-containing minerals. Samples from 21 July, 6 August, and 8 August
showed the greatest influence from industrial sources including many Fe combustion
spheres, with the sample collected on 6 August particularly rich in Fe spheres. The relative
concentration of these spheres changed dramatically over the 3-week sampling period
revealing the changing impact of the iron and steel industry at the IIT site. In addition,
samples collected on 21 July and 8 August were strongly impacted by a source or sources
of Mn-Fe-Zn (apparently associated together, sometimes in combustion spheres). The
sample from 21 July also had a number of Cl-rich and Pb-rich particles. The sample from
8 August, and to a lesser extent the sample from 19 July, reflected a combustion source of
aluminum oxide particles which were observed only in the coarse fraction (Table 7-7).
Aluminum silicate fly ash spheres from coal-fired power plants appeared much less
frequently than Fe spheres, but were observed in the fine fraction of all samples. The
sample collected on 19 July showed the highest concentration. Some Cu-rich particles
were found in the fine fraction of samples collected on 6 August and 19 July. Samples
collated on 6 and 8 August had relatively high concentrations of organic particles (soot,
spores and pollens, and/or plant debris) in the coarse fraction. The sample collected on 6
August also had a relatively high percentage of fine fraction soot, possibly associated with
diesel combustion sources. Samples collected on 21 July and 6 August showed a number
of Na-rich particles of unknown origin in both size fractions. The sodium was observed to
be sometimes associated with Zn, sometimes sometimes with S, and sometimes with no
other metals (sodium carbonate or sodium hydroxide).
7-20
-------
7,3.2. SEM Summary
SEM analyses of six air samples from the Lake Michigan Urban Air Toxics Study
revealed significant differences between samples collected at the same site on different
days. The observed differences reflect changes in the mixture of sources impacting the
sampling site. Particles found in the samples could be associated with the following
source categories: soil and/or road dust, coal-fired power plants, the iron and steel
industry (combustion source), the alumina industry (combustion source), diesel
combustion, and botanical (plant debris, spores, pollen). In addition, SEM provided
evidence of an industrial source or sources of Mn-Fe-Zn that produced both combustion
spheres and non-spherical particles. Other particles were identified that may represent an
incineration source (particles rich in Zn, CI, Zn-Cl) and a smelter (particles rich in Pb or
Cu). The results obtained by SEM are generally consistent with the bulk XRF analyses.
(SEM results on the 8 August sample did not however reflect the high Pb and CI values
measured by XRF in the fine fraction of this sample).
Some of the particulate samples chosen for SEM analysis were selected because of
high particulate mercury concentrations seen during analysis by INAA. Among this
subset, particulate mercury varied from 135 pg/m to 290 pg/m for samples collected on
18 and 19 July and 6 and 8 August. On 21 July, one of the highest particulate mercury
concentrations measured at IIT, 720 pg/m3, occurred. Upon inspection by SEM, the
particulate matter collected on 21 July contained a large percentage of Fe spheres, and
other industrial/combustion source particulates in the coarse fraction (mainly in the 1.5-2.1
f.im size range) and predominantly industrial particles in the fine fraction (Table 7-7).
SEM inspection of the sample collected on 8 August revealed many similarities in the
coarse and fine fractions to the 21 July sample, with a couple of distinct differences: 1)
The coarse fraction of the 8 August sample contained many A1203 spheres and 2) The size
range of the fine fraction industrial components in the 8 August sample contained a
significant fraction of particles in the 0.5 to 0.9 (im range compared to the predominance
of industrial particles < 0.5 |im in the 21 July sample. Absolute principal component
scores for each of these days (detailed in Section 7.5.) indicate the iron and steel industry
as a source for both 21 July and 9 August. However, individual particle examination
revealed a iarge number of industrial particles in the 21 July sample, suggesting not only
smelting and steel industry, but also incineration. The presence of significant numbers of
Al-,SO, spheres in the 8 August sample also indicates that the alumina industry responsible
for this emission may not be associated with high particulate mercury levels.
7-21
-------
Figure 7-7. Industrial Particles.
,y
cGI
1 0 0 0 0
1 IJ 0 0 5 1 1
a) Iron Sphere and Sulfate Particles
£*
V
J*
0.
-
>
*
S V \
A V
\ ^ -
20KV X10008 Til 030 1 0/c?93'
b) Aggregate of Iron Spheres
7-ZZ
-------
Figure 7 -7. Industrial Particles.
d) Aluminum
7-23
-------
Figure 7-7. Mineral Particles.
e) Magnesium Chloride
f) Aluminosilicate Flyash
7-24
-------
Figure 7-7. Mineral Particles.
£0KV X5000 10U 02? 10/27.-'93?
g) Silica sphere
7-25
-------
Figure 7-7. Mineral Particle.
*
£ 0 K V X 1 0 0 0 0 Tu 005 1 1/10 / 3 3 =
i) Aluminum Sulfate
j) Aluminum Oxide
7-26
-------
Figure 7-7. Organic Particles.
k) Soot
/) Spores
7-27
-------
Figure 7-7. Organic Particles.
m) Organic Material
n) Spore
7-28
-------
Table 7-7. Individual Particle Analysis Summary.
Sample
Date
Coarse Fraction
Fine Fraction
18 July 91 86% Mineral, 4% Industrial, 10%
Organic (by number)
Highest Organic content (many
spores/pollens) and lowest
Industrial content.
No Fe spheres, 1 Fly ash sphere.
Some large soot scattered over
filter.
71% Mineral, 26% Industrial, 3%
Organic, (excluding Sulfate).
High % of Mineral particles.
High Sulfate content.
Some Al/Si Fly ash spheres, 1 Fe
sphere, and several non-spherical
Fe-rich particles. Several Zn & CI
particles.
19 July 91 85% Mineral, 9% Industrial, 6%
Organic (by number).
Relatively high Quartz &
Dolomitic Limestone content.
Relatively high Soot component.
Industrial particles include large
Aluminum Oxide particles, some
Al/Si Fly ash, some non-spherical
Fe-rich particles.
68% Mineral, 24% Industrial, 8%
Organic, (excluding Sulfate).
High % of Mineral particles.
Relatively high Soot content and
Sulfate content.
Highest concentration of Fly ash
spheres. Some Cu particles.
Lowest Concentration of Zn-rich
particles.
21 July 91 69% Mineral, 26% Industrial, 5%
Organic (by number).
Intermediate % (by number) as
Minerals: High Gypsum & Calcite,
but no Dolomite or Dolomitic
Limestone. High number % of
industrial particles including many
Fe combustion spheres, some Al/Si
Fly ash, and a large number of
industrial "other" particles
suggestive of smelting, steel
industry, and/or incineration (Mn,
Zn, Fe, CI, Pb).
No AI2O3 particles.
Similar coarse particle size
distribution to 8 August sample.
28% Mineral, 68% Industrial, 4%
Organic (excluding Sulfate).
No Dolomite or Dolomitic
Limestone.
High % of Industrial particles
(excluding S). Many Fe spheres,
Zn-rich, and industrial "other"
particles related to smelters, steel
industry, and/or incineration.
No Al/Si Fly ash however.
Moderate concentration of Zn-rich
particles.
Some Pb & CI particles.
7-29
-------
Table 7-7. Individual Particle Analysis Summary (continued).
Sample
Date
Coarse Fraction
Fine Fraction
2 August 91 91% Mineral, 4% Industrial, 5%
Organic (by number).
Highest Mineral content, highest
quartz. Sample has high Ca
content in Calcite, Dolomite, and
Dolomitic Limestone, but no
Gypsum. Fair number of spores.
Lowest % of Industrial particles,
includes only 1 Al/Si Fly ash
sphere, no Fe spheres, and some
industrial AI2O3. Similar size
distribution to coarse fraction of 18
July.
6 August 91 62% Mineral, 23% Industrial,
15% Organic (by number).
Intermediate % of Mineral
particles. Highest % of Ca-rich
class.
High % of plant debris and
spores/pollens.
Moderately high % of Industrial
particles. Highest concentration of
Fe combustion spheres of all
samples.
No AI2O3 particles.
8 August 91 37% Mineral, 40% Industrial,
23%) Organic (by number).
Lowest % of Mineral particles.
Relatively little dolomite or
Dolomitic Limestone.
Relatively high % of spores, plant
debris, and soot. Highest % of
Industrial particles. Unusually high
concentration of industrial AI2O3
particles. Many Fe combustion
spheres. Some Fe-Zn particles.
Most Coarse particles are in the
smallest size bin (1.5-2.1fj.m)
(similar to 21 July coarse fraction).
79% Mineral, 17% Industrial, 3%
Organic, (excluding Sulfate).
Highest % of Mineral particles.
High Silicate content.
Lowest % of Industrial particles,
very few Fe spheres or Zn-rich
particles.
Some Aluminum Oxide particles
and Al/Si Fly ash.
26% Mineral, 56% Industrial,
18% Organic (excluding Sulfate).
Relatively high % of Soot.
High concentration of Fe spheres
but almost no Zn-rich particles.
Some Cu particles.
Several Na-rich particles.
Sulfate particles are smaller than
the typical sizes.
19% Mineral, 79% Industrial, 2%
Organic (excluding Sulfate).
Highest % of Industrial particles
and lowest % of Mineral class.
Relatively little Dolomite,
Dolomite Limestone, Quartz;
relatively high Ca & S.
Industrial particles include some
Al/Si Fly ash and the highest
concentration of Fe spheres and
Zn-rich particles. Only 1 AJ2O3
particle.
Many Fe-Mn-Zn particles.
Note: The Fine fraction in all samples is dominated bv number by Sulfate particles (typical diameter -0.3 to 0.4 |im).
7-30
-------
Since only about 100 particles could be examined in each size fraction per sample,
one must be careful in assuming that the data are representative of the sample as a whole.
In future studies, a larger number of particles should be quantified (at least 300 particles).
This objective should be much more readily achieved with the use of a computer-
controlled SEM/EDX (CCSEM) system in 1994 to replace the present manually-operated
SEM.
The present study underscores the need for good source samples. Appropriate
source signatures based on XRF or INAA are essential to chemical mass balance
modeling. The same samples, if properly loaded, could also provide the microscopist with
a particle library containing characteristic morphologies and compositions of source-
specific particles. This would greatly enhance the source-apportionment capabilities of
SEM/EDX.
7.4. Correlation Analysis of Observed HAPs
As a first approach for investigating potential relationships between the chemical
compounds (both gaseous and particulate) and elements measured during the LMIJATS,
Pearson Correlation coefficients were calculated using the statistical software package,
SAS (Cary, N.C.). Since the number of compounds measured during the study is quite
large, the intra-site correlations within compound classes or analysis technique are shown
first, followed by the correlations between classes (fine metals and rare earth elements <
2.5(j.m with PAHs, PCBs and pesticides). Correlations with p-values less than 0.05 were
considered significant and are highlighted in Table 7-8, as well as in the tables in the
Appendix. Only variables of importance to the present discussion will be presented.
7.4.1. Relationships at the Illinois Institute of Technology (IIT)
Three methods of analysis were used to determine the elemental composition of
particulate matter in samples collected at IIT. XRF analysis was used to obtain data for
fine mass, Al, Si, S, CI, K, Ca, Ti, V, Mn, Fe, Ni, Cu, Zn, Se, Br, and Pb. INAA was used
to obtain values for Na, Mg, Cr, As. Mo, Sb, La, and Sm while CVAFS was used to
analyze particulate mercury. All elements used in these correlations are from fine-fraction
concentrations, except for particulate Hg which is total particulate matter.
Correlations of fine-fraction elements at IIT utilized a total of 32 data points. The
highest elemental correlation at IIT was between fine mass and sulfur (r=0.98). A subset
7-31
-------
Table 7-8. Pearson Correlations Among Fine Fraction Elements (and Total Particulate Mercury) Measured at IIT.
MASS Al Si
1.00 0.27 0.48
CI
K
Ca
Ti
V
Mn
Fe
Ni
Cu
Zn
Se
Br
Pb
Na
Mg
Cr
As
Mo
Sb
La
Sm
HgV
0.30
0.66
0.71
0.49
0.22
0.38
0.54
0.25
0.51
0.44
0.67
0.65
0.52
0.78
0 33
0.47
0.47
0.36
0.56
0.74
0 18
0.33
MASS
0 01
0.36
0.46
0.54
0.30
003
0.24
0.31
0.02
0.06
0.10
0.36
0.20
0.23
0.06
-0.09
-0.02
-0.18
0.03
0 12
-0.05
0.14
Al
-0.05
0.48
0.79
0.71
0.26
0.09
0.38
0.48
0.29
0.14
0.33
0.41
0.37
0.35
0.67
0.14
0.09
-0.05
0.15
0.29
0.33
0.07
Si
0.24
0.58
0.65
0.39
0.17
029
0.44
0.17
0.46
0.36
0.61
0.60
0.43
0.76
026
0.43
0.42
0.33
0.57
0.75
0.19
0.26
S
1.00
0.40
-0.03
0.02
0.12
0.87
0.63
0.22
0.18
0.76
0.58
0.47
0.55
0.14
0.20
0.66
0.80
0.72
0.24
0.30
0.01
0.76
CI
1.00
0.56
0.54
0.27
0.67
0.85
0 12
0.69
0.72
0.81
0.70
0.85
0.64
025
0.52
0.60
0.46
0.45
0.56
0.01
0.61
K
1.00
0.68
0 25
0 12
0.46
0.40
0.39
0.17
0.34
0.43
0.38
0.56
0.46
0.27
0.18
0 02
0.41
0.64
0.13
-005
Ca
1 00
0.71
0 15
0.33
0.50
0.14
0.21
0.34
0.44
0.38
0.42
0 32
0 12
0 15
0.00
0.07
0 34
0 14
0 12
Ti
1.00
0 12
0 12
0.47
-0.11
0 16
0.24
0 30
023
0 26
0.16
0.18
0.17
-0 01
0.05
0 18
021
0.14
V
1.00
0.86
0 11
0.49
0.96
0.77
0.55
0.85
0.38
0.29
0.72
0.83
0.77
0.42
0.44
0.09
0.87
Mn
1.00
0.11
0.69
0.83
0.81
0.64
0.90
0.50
0.35
0.63
0.77
0.64
0.61
0.52
0.01
0.68
Fc
1.00
-0 09
0.07
0.10
0 16
0.10
0 13
0.43
0.28
0.16
0.00
0 06
0.22
0.24
-0.05
Ni
1.00
0.54
0.68
0.50
0.67
0.59
0.31
0.35
0.41
0.38
0.50
0.40
006
0.39
Cu
1.00
0.78
0.55
0.92
0.54
0.31
0.69
0.75
0.70
0.40
0.48
0.18
0.91
Zn
1.00
0.71
0.82
0.64
0.40
0.61
0.74
0.69
0.53
0.45
0.20
0.70
Sc
1.00
0.60
0.61
0.26
0.34
0.53
0.42
0.45
0.41
0.00
0.53
Br
1.00
0.65
0.43
0.65
0.71
0.59
0.53
0.54
0.22
0.78
Pb
1.00
0.36
0.40
0.32
0.29
0.51
0.61
0.22
0.42
Na
1.00
0.51
0.32
0.24
0.25
0.22
0.57
0.27
Mg
1.00
0.83
0.66
0.60
0.60
0 33
0.55
Cr
1.00
0.81
0.59
0.50
0.17
0.65
As
1.00
0.46
0.36
0.26
0.58
Mo
I 00
0.66
0.18
0.19
Sb
1.00
0.15
0.27
La
1.00
0.15
Sm
1.00 HgV
7-32
-------
of highly correlated fine elements at IIT included Fe, K, Mn, Zn. Se and Pb. These
elements are good markers for iron-steel manufacturing and coal combustion. This group
of elements also correlated with the sample-average temperature and with the sample-
average ozone concentration. Ca showed the highest correlation with temperature and
ozone at IIT with a correlation coefficient of 0.83 and 0.74, respectively. This is largely
due to the coincident wind direction, southwest, that resulted in both elevated ozone and
higher temperatures. Particulate Hg also showed a positive correlation with temperature,
but the sample size was limited.
Since leaded gasoline is generally not used in motor vehicles, Pb is no longer a
marker for emissions of that type. Other elements showing high correlations included Mn
with CI, and Hg with Mn, Zn, CI, Fe, Se, and Pb. Elemental carbon may indicate a source
from fuel combustion, including diesel truck emission and wood burning. Both elemental
and organic carbon were significantly correlated with total fine mass, S, and Ca,
supporting the suspected combustion source. Among the fine fraction elements analyzed
by INAA, Na, Cr, As, Mo, Sb, and La were all correlated at least weakly, but in many
cases with r>0.6, indicating mixed combustion sources. These potentially include
incinerator emissions, steel industry, smelting as well as a regional background
component. Each of these metals was also correlated with Ca, indicating a potential
natural source. Both elemental carbon and organic carbon were significantly correlated
with fine mass, S, and Ca at IIT.
Several PAHs were highly correlated (r>0.90, n=16). One group of PAHs with
r=0.9 or greater included chrysene, benz(a)anthracene, benzo(e)pyrene.
benzofluoranthenes, indeno(l,2,3-c.d)pyrene. dibenzo(a.h)anthracene. and
benzo(g,h,i)perylene. These PAHs were also highly correlated with some PCBs and
pesticides. A second highly correlated ^-oup of PAHs included fluoranthene,
phenanthrene, and pyrene. Similar correlations were also found for this group with the
PCBs and pesticides, but pyrene had lower correlation coefficients. The two compounds
fluorene and acenaphthene were also highly correlated and had similar correlations with
PCBs and pesticides. Acenaphthene, fluorene, phenanthrene, and fluorenone are all
moderately to highly correlated with the maximum temperature and ozone at IIT.
Acenaphthene showed the highest correlation with ozone with an r=0.86. Two PAHs,
acenaphthylene and cyclopenta(c.d)pyrene were not significantly correlated with any other
PAHs at IIT.
7-33
-------
Ratios of selected PAHs with respect to benzo(e)pyrene were calculated in order
to further characterize PAH sources. Benzo(e)pyrene, found in particulate form, generally
is a more stable PAH. When compared to the other more volatile and reactive PAH
species, ratios compared between sites indicate source composition. Compounds chosen
for comparison to benzo(e)pyrene were pyrene, benz(a)anthracene, chrysene,
cyclopenta(c,d)pyrene, benzofluoranthenes, benzo(e)pyrene, benzo(a)pyrene,
indeno(l,2,3-c,d)pyrene, dibenzo(a,h)anthracene, benzo(g,h,i)perylene, and coronene. On
days with similar prevailing wind directions, the ratio of these compounds to
benzo(e)pyrene showed similar patterns.
PAH profiles for various source categories are not well characterized. In addition,
PAH correlations and ratios are difficult to use for predicting sources because of differing
emissions of PAHs from one source type and the reactivity of PAHs. Thus, the task of
attributing specific sources based solely on PAH data is difficult. Another method of
investigating sources of PAHs is to examine their relationship to specific tracer elements.
Correlations between fine-fraction elements and PAHs at IIT revealed that PAHs which
had very similar inter-correlation patterns also had similar correlations with fine-fraction
elements. For example, chrysene, acenaphthene and fluoranthene were all correlated with
Ca and Ni, while chrysene and fluoranthene were correlated with A1 and Si as well. Three
PAHs had highly significant correlations (r>0.80) with some metals. These include
chrysene with Ca, acenaphthylene with CI, Mn and Zn, and fluorenone with Ca. Chrysene
also was significantly correlated with Al, Si, and Ni, while acenaphthylene was also
correlated with Fe, Se, Pb, and Hg. These relationships may indicate coal combustion
sources as well as smelting and the local iron and steel industry. Naphthalene and
benzo(a)pyrene, which are indicative of some combustion types, did not show significant
correlations with any of the elements at IIT.
Three subsets of PCB compounds were highly correlated (r>0.90) for the 16
samples analyzed. One group, consisting of total tri-, total tetra-, and total PCBs, had
similar correlations with the PAHs and pesticides. The second group including total
penta-, hexa-, hepta- and total PCBs, was also highly correlated with ozone. Total penta-
PCBs were correlated with temperature (r=0.82) and ozone (r=0.85). The total hexa- and
total hepta-PCBs also had similar correlations with the PAHs and pesticides. The third
highly correlated group was the octa-PCBs which had similar correlations with the PAHs
and pesticides. Air of these groups of PCBs were significantly correlated with each other
as well (r>0.72).
7-34
-------
Two PCB compounds had no correlations with other PCBs. These were the tri-
PCB and the hexa-PCBs, both of which had concentrations within a very small range (0 -
0.74 pg/m3 for the tri- and 0 - 1.11 pg/m3 for the hexa-PCB). Other PCBs with a small
range of values had only one or two significant correlations including the tetra-PCBs, the
octa-PCBs, and the total nano- and deca-PCBs.
Analysis of PCB correlations with fine-fraction elements indicated no high
correlations (r>0.9) between members of these two categories. The mono-PCBs were not
significantly correlated with any of the elements, with the exception of a negative
correlation with V (r=-0.63). The rest of the PCBs were significantly correlated with the
same group of elements including fine mass, S, Si, and Ca. Two PCBs, total hepta- and
total octa-PCBs, also had significant correlations with the INAA data for Na, Mg, Cr, Sb,
La, and Sm.
Three pesticides, trans-Nonachlor, Dieldrin, and alpha-Chlordane, were highly
correlated (r>0.80) for the 16 samples at IIT. Trans-Nonachlor and alpha-Chlordane had
similar correlations with the PAHs and PCBs, but Dieldrin had different correlations.
Three compounds were not correlated with other pesticides, HCB, Simazine, and p,p'-
DDD. Of these, p,p'-DDD was also not significantly correlated with any PAHs or PCBs,
and HCB was only significantly correlated with one PCB. Three other compounds,
Atrazine, Alachlor and Metalachlor, had only one significant correlation with another
pesticide. Atrazine and Metalachlor were also not significantly correlated with any PAHs
or PCBs.
Interesting groupings were seen when the pesticides measured at IIT were
correlated with several different fine-fraction elements. However, these correlations were
never greater than 0.80. The group of highly correlated pesticides mentioned above
produced similar correlations with many elements, but each pesticide had additional
elements with which it correlated significantly. Alpha and gamma-HCH did not correlate
with any elements in common. .Alpha HCH demonstrated only negative correlations with
fine-fraction metals including Si, Ca, and Ni. Gamma-HCH was positively correlated with
V. HCB demonstrated the only other significant negative correlation (-0.76, with V).
Both DDT and Simazine were correlated with the same elements at IIT: Na, Mg, Cr. As,
Mo, Sb, La, and Sm. Mirex, Dieldrin, Chlorpyrifos and p, p'-DDE were all correlated
with Zn, Se and Pb among other elements. Three pesticides (p,p'-DDD, Alachlor and
Metolachlor) had no significant correlations with any elements. Two pesticides had only
negative correlations, alpha-HCH with Si, Ca, and Ni. and HCB with V.
7-35
-------
7.4.2. Relationships at South Haven, MI
At South Haven, sulfate, nitrate, nitric acid, strong aerosol acidity (H+), sulfur
dioxide and ammonia were quantified in addition to fine-fraction elements. The strongest
correlation was between fine mass and sulfur (r=0.98), which, as expected, was similar to
the correlation between fine mass and sulfate. The dominant grouping of elements that
correlated well at South Haven were S, Si, K, Ca, Mn, Fe, Zn, Se, Br, Pb, elemental
carbon, organic carbon, S04"2, N03", HN03, and S02. A regional/transported
combustion source is indicated by this combination of elements, as well as a natural
component.
At South Haven, the majority of PAHs measured correlated with one another.
Two slightly distinct groupings of highly correlated PAHs were: 1) fluoranthene, pyrene
benz(a)anthracene, chrysene, cvclopenta(c,d)pyrene, benzofluoranthenes, benzo(e)pyrene,
benzo(a)pyrene, indeno(l,2,3-c,d)pyrene, dibenzo(a,h)anthracene, benzo(g,h,i)perylene,
and coronene; and 2) fluorene and phenanthrene were correlated with fluoranthene,
pyrene, benz(a)anthracene, chrysene, cyclopenta(c,d)pyrene, benzofluoranthenes,
benzo(e)pyrene, benzo(a)pyrene, indeno(l,2,3-c,d)pyrene, dibenzo(a,h)anthracene,
benzo(g,h,i)perylene and coronene. This may represent an aged airmass that has
undergone substantial reaction with other atmospheric components.
All PAHs except acenaphthylene and anthracene correlated (r>0.5) with fine mass,
S, K, Ca (except for naphthalene), Mn, Fe, Zn (except for acenaphthene and retene), Se,
Br (except for retene), Pb, S04 , HN03, and S02 (except for coronene), indicating a
regional sulfate mixture and other potential combustion sources. In addition, H"1" was
correlated with acenaphthene, retene. fluoranthene and benz(a)anthracene.
Few PCBs at South Haven were correlated with one another. 2,2',3,4,5'-PCB was
correlated with penta-PCBs, 2,2',3,4,5,6,6-PCB, Hexa-PCBs, 2,2',3,4,5,6,6'-PCB, hepta-
PCBs, 2,2',3,3',4,5,6-PCB, and octa-PCBs. Mono-PCBs had a weak negative correlation
with di-PCBs, 2,2',3,4,5'-PCB, penta-PCBs, and hepta-PCBs. In general, the individual
isomers quantified were correlated with the congener groups of which they are a member
(for example, 2,2',4,6-PCB was correlated strongly, r=0.83, with tetra-PCBs). There was
no surprise in that finding.
Several PCBs at South Haven were correlated with the same set of fine-fraction
elements with which a large set of PAH compounds were correlated. Hexa-PCBs, hepta-
7-36
-------
PCBs, octa-PCBs and penta-PCBs were correlated with fine mass, Si, S, K, Ca, Mn, Fe,
Zn, Se. Br, Pb, S04"2, HNO^, and S02- Several PCBs were also correlated with vapor
phase mercury including nona-PCBs, di-PCBs, 2,2',3,4,5-PCB, penta-PCBs, hexa-PCBs,
octa-PCBs, and total PCBs.
Few pesticides at South Haven were correlated, and those compounds that did
demonstrate a correlation, were generally weak (r~0.5). The few compounds that were
highly correlated were trans-Nonachlor, alpha and gamma Chlordane and trans-Nonachlor
with Dieldrin. The weaker correlations include hexachlorobenzene with alpha HCH,
Alachlor with trans-Nonachlor, alpha and gamma Chlordane, DDT with DDE, and alpha
and gamma Chlordane, and negative correlations between Aldrin and Dieldrin and DDT.
The same subset of fine-fraction metals that correlated with groupings of PAHs
and PCBs measured at South Haven also correlated with a few of the pesticides.
Alachlor, trans-Nonachlor and alpha and gamma-Chlordane were all correlated with fine
mass, Si, S, K, Ca, Ti, Mn, Fe, Zn, Se, Br, Pb, strong acidity, S04"2, HN03, and S02. In
addition, Dieldrin was correlated with all of these elements except for Ca, Zn, and Br.
DDT was correlated with fine mass, Si, S, Ca, Ti, Fe, strong acidity, S04 , HN03, and
S02. Simazine was correlated with a different set of elements, Al, Si, Ca, Ni, and Cu,
while alpha HCH demonstrated a different correlation pattern as well with negative
correlations with Si, CI, K, Ni, and Cu.
Ozone correlated with many elements and compounds at South Haven. The
highest correlation (r=0.91) was with HNO,, followed by Cv or organic carbon (r=0.90),
Ce or elemental carbon (r=0.89), and fine and coarse Fe (r=0.87 and 0.86, respectively).
The relatively high correlation of ozone with coarse Fe is an interesting one. The
correlation of coarse mass with the maximum hourly ozone measured during the sampling
period (r=0.82) was slightly higher than the correlation observed with fine mas., (r=0.81).
Coarse Ca was also correlated with the maximum hourly ozone (r=0.82). This interesting
relationship between coarse particulate species and the maximum hourly ozone is largely
due to the location of the iron and steel industry in the urban areas upwind of South
Haven in the same direction. The correlation also reflects the temperature
interdependence with mixed layer flow from the southwest to South Haven.
7-37
-------
7.4.3.
Relationships at Kankakee, IL
The correlations found between fine elements measured at Kankakee are generally
weak, except for the correlation (r=0.96) between fine mass and S. Other correlations
include S and Se, and moderate correlations between Al, Si, CI, K, Ca, Ti, Fe, and Cu.
This combination indicates a regional background with indication of an influence that may
be attributed to combustion. In addition, Fe correlated with Al, Si, K, Ca, Ti, and Mn,
while Cu correlated with fine mass, Al, Si, CI, and slightly with S, K, Ca, and Ti.
Several PAHs quantified in samples collected at Kankakee were highly correlated.
These include anthracene, fluoranthene, pyrene, benz(a)anthracene, chrysene,
benzofluoranthenes, benzo(e)pyrene, benzo(a)pyrene, indeno(l,2,3-c,d)pyrene and
dibenzo(a,h)anthracene. Coronene was also significantly correlated with all of the same
PAHs (r~0.6). In addition, acenaphthene, fluorene and phenanthrene were correlated with
cyclopenta(c,d)pyrene, retene and fluorenone.
Very few PAHs quantified at Kankakee were correlated with fine-fraction elements
measured at that site. This may be due to a lack of coincident sources or may indicate an
aged airmass that has undergone extensive reaction and conversion resulting in
relationships that cannot be attributed to specific source types. However, pyrene,
benz(a)anthracene and benzofluoranthenes were correlated with S, phenanthrene was
correlated with CI, acenaphthene with K, acenaphthene and cyclopenta(c,d)pyrene with
Mn, acenaphthene with Fe and acenaphthene, and fluorene and phenanthrene with Zn.
Several PCB isomers were highly correlated with quantified PCB congener groups.
2,2',3,4,5'-PCB and penta-PCBs were highly correlated with hexa-PCBs, hepta-PCBs,
octa-PCBs and total PCB. 2,2',3,4,5,6,6'-PCB and 2,2',3,3',4,5,6-PCB were highly
correlated with nona-PCBs and deca-PCBs, while nona-PCBs were highly correlated with
deca-PCBs. Correlations between PCBs and fine-fraction elements were very sparse, as
was the case for PAHs. Total mono-PCBs, total hexa-PCBs and total hepta-PCBs were
correlated weakly with fine mass, total hepta-PCBs were correlated with S, 2,2',4,6-PCB
was highly correlated with CI, total mono-PCB and 2,2',4,6-PCB were correlated with Cu,
and negative correlations were observed for total mono-PCBs and 2,3-PCB with Ni, and
tri-PCBs with Ti.
Pesticides measured in Kankakee samples revealed a correlation pattern in which
Dieldrin, alpha and gamma Chlordane, DDE, DDD, gamma HCH, Alachlor and Mirex
7-38
-------
were all correlated. In addition, trans-Nonachlor was correlated with alpha and gamma
Chlordane, Chlorpyrifos, Dieldrin, Alachlor and Mirex. Simazine was correlated with
DDD, gamma HCH and Alachlor. DDT and Aldrin were only correlated with each other.
Correlations between fine-fraction elements and pesticides were stronger, in
general, than those correlations at the other sites. Several pesticides were correlated with
Mn, Zn, Se and Br (including Alachlor, Mirex, trans-Nonachlor, Dieldrin, gamma-
Chlordane, alpha-Chlordane and DDE). In addition, certain pesticides such as
Chlorpyrifos demonstrated a correlation pattern with a subset of fine-fraction elements,
fine mass, Si, K, Ca, Ti, Fe and slightly with Al. Atrazine was correlated with Al, Si, K,
Ca, Ti, Fe and slightly with Mn and Ni. DDD was correlated with Zn, Se, Br and Pb.
DDT did not correlate positively with any of the fine-fraction metals. However, weak
negative correlations were observed with fine mass, Al, CI, K, Ca and Cu.
7.5. Principal Component Analysis
In order to investigate the sources and transport patterns for the hazardous air
pollutants measured in the southern Lake Michigan Basin a receptor modeling technique,
principal component analysis (PCA), was applied to selected measurements collected
during LMUATS. Table 7-9 lists ten source categories and their known elemental
'fingerprint' or tracer species. The number of samples collected on the R/V Lanrentian
was too small to allow use of PCA. Consequently, this receptor model was used only to
apportion sources in Kankakee. Chicago and in South Haven. Elements included in the
statistical analysis were chosen on the basis of 1) quality of analytical data, and 2) use in
apportioning sources (marker compound or element).
7.5.1. Absolute Principal Component Analysis Applied to Trace Elements
APCA was performed on the fine-fraction trace elemental and carbon data
collected in Chicago (Table 7-10). The four factor model accounted for 91% of the
variability in the data. A majority of the variance in the data was explained with model.
Factor 1 explained the largest amount of variability in the data set (49%) and was
dominated by elements generally associated with the iron and steel industry, and other
metal production or manufacturing. Interestingly, when Hg vapor was included in the
APCA, it was highly loaded on this factor as well, explaining 81% of the variability in the
mercury concentrations measured. Factor 2 had moderate loadings of S, K, and elemental
carbon, and high loadings for fine Si and Ca indicating a mixed component with soil
7-39
-------
Table 7-9. Atmospheric Tracer Elements and their Known Sources (Keeler, 1987).
SOURCE
KEY ELEMENTS
Soil
Mg, Al, Si, K, Ca, Sc, Ti, Mn, Fe, Ga, Sr,
La, Ce, Sm
Coal Combustion
Soil Elements & Fine B, Ge, As, Se, Sb,
Ba, W, Hg, U
Oil Combustion
Fine Particle V, Ni, Mo
Petroleum Refinery
La, Ce, Nd
Motor Vehicles
Elemental Carbon
Smelters
Ni, Cu, As, Cd, In, Sn, Sb, Pb
Marine Aerosol
Na, CI
Vegetative Burning
Organic and Elemental Carbon, CI, K, Zn
Iron-Steel
Fine Particle Mg, Cr, Mn, Fe, Co, Ni
Incineration
CI, Zn, Ag, In, Sb, Hg
7-40
-------
Table 7-10. Factor Loadings from Varimax-Rotated PCA of IIT Fine Fraction
Elemental Data.
Factor 1
Factor 2
Factor 3
Factor 4
Fe-Steel
Soil +
Metal Industry
Regional S
s
0.15
0.41
0.37
0.78
Si
0.14
0.95
-0.05
0.10
K
0.76
0.51
0.05
0.25
Ca
0.06
0.90
0.33
0.17
Mn
0.95
-0.03
0.20
0.01
Fe
0.87
0.34
0.28
-0.02
Zn
0.95
0.01
0.12
0.10
Se
0.81
0.15
0.15
0.48
Pb
0.91
0.27
0.16
0.08
EC
0.12
0.58
0.61
0.16
As
0.79
-0.05
0.46
0.15
Sb
0.32
0.17
0.83
0.15
Cr
0.66
0.01
0.61
0.16
Eigenvalue
5.84
2.73
2.1
1.2
% variance
49
23
18
10
influence as well as a hint of cement industry. Factor 3 appears to represent a smelting or
metals processing source with high factor loadings from Sb, Cr, and EC, and moderate
loadings for As and S. Factor 4 appears to be the regional S component with high
loadings of S and moderate loadings of Se and elemental carbon. S was found to be
smeared onto all of the components at IIT.
The results of a three-factor model of the fine-fraction trace elemental and carbon
data collected in Kankakee (Table 7-11) accounted for 86% of the variability in the data
used in the analysis. Factor 1 explained the largest amount of variability in the data set
(35%) and was highly loaded with elements generally of soil origin (Si, Ca, K. Fe). As
discussed in Section 5 many of these elements were very enriched over their crustal
abundance suggesting that iron-steel emissions may also be contributing to this factor.
7-41
-------
Factor 2 appears to represent iron-steel emissions as well as Pb and Zn smelting or metal
processing emissions with high factor loadings from Mn, Zn, and Pb. This factor is
associated with northerly air mass transport from Chicago/Gary to Kankakee. Factor 3
represents the regional S component, which is dominated by coal combustion sources,
with high loadings of S and Se. The mass contributions were calculated using APCA for
Kankakee. The largest source of fine mass at this site was the regional S source,
contributing on average, 14 fig/m3 of the 20 |j.g/m3.
Table 7-11. Factor Loadings from Varimax-Rotated PCA of Kankakee Fine
Fraction Elemental Data.
Factor 1 Factor 2 Factor 3
S
0.11
-0.22
0.87
Si
0.91
-0.17
0.00
Ca
0.71
-0.17
0.02
K
0.87
0.39
0.12
Mn
0.44
0.74
-0.17
Fe
0.90
0.27
-0.15
Zn
-0.03
0.81
-0.07
Se
-0.11
0.21
0.85
Pb
-0.09
0.79
0.15
Eigenvalue
3.12
2.19
1.58
% variance
24
18
In South Haven, the three-factor model for fine-fraction elements explained 93%
of the variability in the data with 42% explained by the first factor (Table 7-12). This
varimax-rotated model accounts for 92% of the variability in the fine mass as well. Factor
1 was found to have high loadings from several elements including Mn, Zn, Pb, elemental
carbon, as well as moderate loadings for K, Fe, Ca, and Se. These elements represent
emissions from the iron and steel industry and perhaps diesels. Factor 2 is highly loaded
with fine Si and moderately loaded with K, Ca, Mn, Fe, and elemental carbon. This factor
looks like a soil factor but the addition of elemental carbon suggests that soil is mixed with
7-42
-------
other source influences. The highest mass contributions for this component were on 22
July and 2 August when air mass trajectories were from the southwest over the lake
through Chicago and towards South Haven. Factor 3 indicates the regional S component
with high loadings for S and Se, and moderate loadings for Pb, elemental carbon, Zn and
K. Factor 3 was the dominant component during the pollution episode observed in South
Haven and the other two sites from 16-22 July. This component was responsible for over
half of the fine mass, on average, contributing 8.3 |ig/m3 out of 13.6 ng/m3. The first and
second factors contributed 2.7 and 3 .4 |ig/m3 to the fine mass, respectively.
Table 7-12. Factor Loadings from Varimax-Rotated PCA of South Haven Fine
Fraction Elemental Data.
Factor 1
Factor 2
Factor 3
Iron-Steel
Soil +
Regional S
s
0.39
0.25
0.85
Si
0.14
0.97
0.11
K
0.55
0.65
0.47
Ca
0.49
0.75
0.33
Mn
0.76
0.42
0.43
Fe
0.59
0.69
0.39
Zn
0.72
0.19
0.61
Se
0.44
0.22
0.83
Pb
0.76
0.32
0.52
EC
0.77
0.41
0.41
Eigenvalue
3.4
3.1
2.9
% variance
36
33
31
The blurring of several source types onto one component that is a large distance
from the sources is not uncommon (Keeler et al, 1991). As the distance from the source
increases relative to the distance between sources, this blurring seems to increase. The
second reason for several sources being loaded on the same factor has to do with the
location of the sources and the magnitude of the source influence. The regional S
component was the dominant fine mass contributor at all three land-based sites. This
7-43
-------
component represents coal combustion sources that are upwind of the Kankakee site and
when transport connects the three sites, such as on 17-22 July, this regional S component
dominates the source apportionment. Emissions are added to the regional plume as it is
transported through the urban/industrial areas of Chicago/Gary and carried out over Lake
Michigan. The combined emissions are then mixed and transported to South Haven
together, building in a correlation due to wind direction. Since transport from the
southwest also brings hot humid weather to the Lake Michigan Basin, elevated ozone and
photochemical activity is observed during the same period.
7.5.2. Principal Component Analysis Applied to PCB Data
Principal component analysis was utilized to investigate the variability, potential
source(s) and source impacts observed in the PCB congeners or homologue categories
measured at the IIT site. The tri-, tetra-, penta-, hexa-, hepta- and octa-PCB homologues
and Total PCBs are loaded on the same factor making source characterization difficult.
The second factor is dominated by congener 50, total nona-PCBs and deca-PCBs, while
the third factor is dominated by mono- and di-PCBs and congener 5. Only one PCB,
congener 1, loaded strongly on factor 4, with a small contribution from total tetra-PCB.
However, this may be an artifact of the data, since congener 1 was underestimated during
sample collection due to its high volatility.
At the conclusion of their report on PCB congeners and potential source
contributions, Manchester-Neevig and Andren (1989) stated that overall, the congener
distributions were surprisingly similar between seasons despite different air trajectories.
Consequently, source apportionment is a difficult task with these compounds.
7.5.3. Principal Component Analysis Applied to Pesticide P'_ta
In order to discern urban versus rural contributions to levels of pesticides
measured over the lake or at any receptor site, chemical mass balance, pesticide isomeric
ratio correlations (with tracer elements), and principal component analysis can be utilized.
However, interpreting source signatures for pesticides measured in ambient samples is a
difficult endeavor. Distinguishing revolatilization of previously applied or transported
pesticides from recent applications or transport of recently applied pesticides poses a
major challenge in attempting to discern sources. Use of APCA for pesticides measured at
Kankakee did not prove fruitful since the four factor model used did not separate
7-44
-------
contributions from several of the pesticides measured. Factor 1 was described by high
loadings from 7-HCH, Alachlor, Mirex, Trans-Nonachlor, Dieldrin, a- and 7-Chlordane,
DDE, and DDD.
At IIT, the four factor model separated sources for the a- and 7-HCH isomers.
Factor 1 describes the majority of the variability in the data for 7-HCH, Mirex, Trans-
Nonachlor, Dieldrin, Chlorpyrifos, a- and 7-Chlordane, and DDE. This separation of
pesticide contributions is similar to the pattern observed at Kankakee. However, Factor 2
at IIT also has high loadings of a-Chlordane, DDT, Trans-Nonachlor, Aldrin, Atrazine
and factor 2 is negatively correlated with a-HCH.
At South Haven, pesticides that loaded strongly on Factor 1 include Alachlor,
Trans-Nonachlor, Dieldrin. a- and 7- Chlordane and. to a lesser degree. 4,4'-DDT. Factor
2 describes a common source for DDT and DDE as well as 7-HCH. Aldrin was negatively
correlated with Factor 2 at South Haven. Factor 3 had large contributions from a-HCH,
HCB, Aldrin and Factor 4 in the model is also described by high loading from HCB, but
also with Metolachlor and a negative correlation with Simazine.
7.5.4. Principal Component Analysis Applied to PAH Data
Source apportionment of PAHs is plagued by some of the same attributes which
make pesticides difficult to use as tracer species. Because of their reactivity and exchange
between different phases in the atmosphere (gas-to-particle phase distribution), organic
compounds behave less conservatively than elemental tracers. Thus, changes in the
relative abundance of individual compounds occur in transit from source to receptor due
to differential reactivity and rates of atmospheric deposition. The limitations imposed by
alteration of profiles are not well understood (Keeler et al., 1993). However, recent work
has elucidated potential PAH tracer species.
In a recent paper by Li and Kamens (1993), ratios of specific tracer PAHs were
calculated in order to determine source signatures for gasoline exhaust, diesel exhaust and
wood combustion. Ratios of benzo(g,h,i)perylene/coronene,
benzo(g,h,i)perylene/indeno( 1.2.3-c,d )pyrene, benzofluoranthenes/indeno( 1,2,3-c,d)
pyrene. chrysene/benzo(e)pyrene, benz(a)anthracene/benzo(a)pyrene and
benzofluoranthenes/coronene were calculated for the LMUATS and are shown in Table 7-
13. The source contribution for the pattern observed by Li and Kamens and Tong and
7-45
-------
Karasek (1984) attributed to diesel exhaust is also given with the LMUATS ratios. The
ratios for gasoline exhaust were identified and are given in Table 7-14.
In order to further characterize PAH source signatures, ratios of selected PAHs
were calculated with respect to benzo(e)pyrene (BeP). Benzo(e)pyrene, generally found
in particulate form, is a more stable PAH. When contrasted with other more volatile and
reactive PAH species, ratios to BeP compared between sites may indicate source
composition and transport. Compounds chosen for comparison to BeP were pyrene,
benz(a)anthracene, chrysene, cyclopenta(c,d)pyrene, benzofluoranthenes, benzo(e)pyrene,
benzo(a)pyrene, indeno(l,2,3-c,d)pyrene, dibenzo(a,h)anthracene, benzo(g,h,i)perylene,
and coronene. On days when there was similar prevailing wind direction, the ratio of
these compounds to BeP showed similar patterns. For example, on 17 July 91 when the
wind directly connected IIT and South Haven, the PAH ratio profiles are similar (Figure
7-8). However, the ratios for some PAHs at South Haven were higher than at IIT such as
the ratios of retene and dibenzo(a,h)anthracene which might be due to the influence of
local sources. The profiles on the R/V Laurentian were always very similar to the profiles
at IIT, for example the profiles on 6 August 91 shown in Figure 7-9. This may be a result
of the R/V Laurentian's position close to Chicago and to the influence of coke ovens
southeast of the site in Chicago. Figure 7-10 reveals the striking similarities between the
PAH ratios found in coke oven emissions reported by Daisey et al. (1986) compared to
those measured on the Laurentian on the night of 6 August 1991. All of the ratios seen in
Figure 7-10 are within a factor of two which is quite remarkable considering the potential
modifications of the ratios caused by transformations en route to the sampling station.
Only on 5 August 91 does the PAH profile on the Laurentian demonstrate substantial
difference from the PAH profiles at IIT.
7.5.5. Source Attribution Using Combined Data Sets and Ratios
Elemental carbon is also a good indicator of diesel exhaust. Correlations were
calculated between the PAH ratios above and elemental and organic carbon by site in
order to determine the potential strength of these relationships as an additional indicator of
source contribution. At Kankakee, significant correlations were observed for the
benzo(g.h,i)perylene/coronene ratio with organic carbon, mass, Cu, ozone, and S.
Benzo(g.h,i)perylene/indeno(l,2.3-c,d)pyrene was negatively correlated with Al, Ca, Cu.
and mean-temperature. Benzofluoranthenes/indeno(l,2.3-c,d)pyrene was correlated only
with organic carbon, ozone, and S. Chrysene/benzo(e)pyrene was correlated with Ni and
7-46
-------
Zn. Benz(a)anthracene/benzo(a)pyrene was correlated with Ti, Mn, Fe, Se. K, and Ca,
and benzofluoranthenes/coronene was correlated with organic carbon, mass, S, and ozone.
At IIT, two of the PAH ratios (chrysene/benzo(e)pyrene and
benz(a)anthracene/benzo(a)pyrene) were not correlated significantly with any elements
measured, and benzofluoranthenes/indeno(l,2,3-c,d)pyrene was correlated only with
ozone. Benzo(g,h,i)perylene/coronene was correlated with ozone, Ca, and negatively
correlated with vapor phase mercury. Benzo(g,h,i)perylene/indeno( 1,2,3-c,d)pyrene was
negatively correlated with ozone, Ca, and positively correlated with CI, Mn, Zn, and vapor
phase Hg. Benzofluoranthenes/coronene was correlated only with ozone and Ca. None
of the PAH ratios at IIT were correlated with organic or elemental carbon.
On the R/V Laurentian, significant correlations with
benzo(g,h,i)perylene/coronene were observed with K, Ca, Mn, Fe, Pb, and nitrate (similar
to the correlations observed at Kankakee for this ratio). The
benzo(g,h,i)perylene/indeno(l,2.3-c,d)pyrene ratio was correlated only with V and Ni.
Benzofluoranthenes/indeno(l,2,3-c,d)pyrene was correlated with S, K, Mn, Fe, Pb, and
elemental carbon. Chrysene/benzo(e)pyrene was not correlated with any elements
measured on the Laurentian. Benz(a)anthracene/benzo(a)pyrene and
benzofluoranthenes/coronene were both correlated with K, Mn, and Fe.
Benz(a)anthracene/benzo(a)pyrene was also correlated with nitrate, and
benzofluoranthene/coronene was also correlated with Zn and Pb.
At South Haven, three of the PAH ratios were correlated with organic carbon
(benzo(g,h,i)perylene/coronene. benzofluoranthenes/indeno( l,2,3-c,d)pyrene and
benzofluoranthenes/coronene) and both benzo(g,h,i)perylene/coronene and
benzofluoranthenes/coronene were correlated with elemental carbon, mass. S. K, Ca, Mn,
Fe, Se, Pb, S04, H+. HN03, SO?, ozone, and temperature.
Benzofluoranthenes/indeno(l,2,3-c.d)pyrene was also correlated with mass, S, Fe, H+,
S04 HN03 ozone, and temperature. The three PAH ratios
Benzo(g,h,i)perylene/Indeno( l,2.3-c,d)pyrene, Chrvsene/Benzo(e)pyrene and
Benz(a)anthracene/benzo(a)pyrene were not correlated with any of the elements measured
at South Haven.
7-47
-------
Table 7-13. Ratios of PAHs Observed at LMUATS Sites Compared to Other Sites.
PAH Ratios Kankakee IIT R/V South Li and Tongand
Laurentian Haven Kamens Karasek
(1993) (1984)
Benzo(g,h,i)perylene/ 1.7 2.4 1.6 2.1 2.5 1.9
Coronene
Benzo(g,h,i)pereylene/ 0.8 0.9 0.8 0.8 1.1 1.1
Indeno( 1,2,3-c,d)pyrene
Benzofluoranthenes/ 1.7 2.5 1.9 2.0 0.5 0.3
Indeno( 1,2,3-c,d)pyrene
Chrysene/ 1.8 2.1 2.0 1.9 1.6 1.6
Benzo(e)pyrene
Benz(a)anthracene/ 0.9 1.6 0.8 1.1 1.0 1.7
Benzo(a)pyrene
Benzofluoranthenes/ 4.2 7.4 4.1 5.9 2.9 1.8
Coronene
Table 7-14. Reported Ratios of PAHs which Represent Gasoline Exhaust.
PAH Ratios Li and Kamens Grimmer
(1993) (1977)
Benzo(g,h,i)perylene/ j ^ ^
Coronene
Benzo(g,h,i)pereylene/ ^ 5
Indeno(l,2,3-c,d)pyrene
Benzofluoranthenes/
Indeno( 1,2,3 -c, d)pyrene
Benz(a)anthracene/ q ^
Benzo(a)pyrene
Benzofluoranthenes/ Coronene
7-48
3.8
0.4 0.2
Chrysene/ Benzo(e)pyrene 2.5 2.2
0.7
0.3 0.2
-------
Figure 7-8. Profiles of PAH Ratios with Benzo(e)pyrene on 17 July 1991.
H
:a er
v/.vs,i
E?v
7-49
-------
Figure 7-9. Profiles of PAH Ratios with Benzo(e)pyrene on 6 August 1991.
7-50
-------
7.6.
Chemical Mass Balance Modeling
A subset of the samples collected at the IIT site were selected for a detailed source
apportionment of both the fine and coarse fractions. Results of individual particle analyses
by Scanning Electron Microscopy and Energy Dispersive X-ray Spectroscopy
(SEM/EDS) were used to supplement the CMB results. Particle morphology and
elemental analysis has been shown to be useful in resolving source types which cannot be
resolved by conventional means (Dzubay and Mamane, 1989).
The CMB model yields quantitative source contribution estimates for the major
sources contributing to the mass for each sample. In practice, the model is limited by the
availability and variability of the chemical composition of the potential sources. The
source chemical compositions, i.e. source profiles, are used in the model in conjunction
with ambient data and uncertainties in each to estimate the mass concentration, with
uncertainties, contributed by each source at the sampling site during each sampling period.
The uncertainties of profiles may not realistically represent the actual variability of the
source categories they represent. This should be considered in the interpretation of the
CMB results.
Fortunately, the PM-10 particulate pollutant source apportionment for Chicago is
well characterized (Vermette et al., 1991; Sweet et al., 1993). Recently, Sweet et al.
(1993) reported using locally obtained fugitive dust source profiles to supplement
literature data to apportion the sources of particulate pollutants in southeast Chicago and
East St. Louis. This work builds upon work reported previously by the same group
(Vermette et al., 1991). Among their most significant findings are that: 1) surface dust
emissions are usually more important contributors to both fine- and coarse-particle mass
than a. ~ stack emissions, and 2) some sources may contribute significantly to specific toxic
metals without affecting overall PM-10 levels. The findings from these studies are based
on samples collected near sources of many of the fugitive dusts sources. Stack emissions
became important in the fine fraction only when winds were blowing directly from the
stack. The IIT site is about 17 km NNW of the source profile collection sites.
Nevertheless, these findings may still be applicable to the downtown Chicago samples
being studied here and will be considered in this analysis.
Tables 7-15 and 7-16 list the samples selected for CMB and SEM/EDS analysis
and the measured composition of their coarse and fine fractions, respectively. In the fine
7-51
-------
fraction, some elemental concentrations were available from both XRF and INAA. INAA
is clearly the better method for some elements, such as As. For other elements, the 2
methods are comparable. For CMB analysis, it is best to have source profiles and ambient
data produced by the same analytical procedures. The researchers who produced the
potentially important fugitive dust profiles (see discussion below) chose to report the
INAA results for V, As, and La; therefore, the ambient data used in this CMB analysis will
include the INAA results for those elements also.
Carbon data were available for the fine fraction only. While carbon may not be
useful in the CMB model because of the lack of carbon data in many of the source
profiles, it is important to include to illustrate the fraction of the mass accounted for by
carbon. The fractions of elemental and organic carbon of the total fine mass are listed in
Table 7-15. Organic carbon concentrations were multiplied by 1.4 to account for the
hydrogen and oxygen which, together with carbon, comprise the average particulate
organic compounds in the ambient air (Countess et al., 1980). Organic and elemental
carbon combined account for 36% to 58% of the fine mass. Since most of the source
profiles available do not report carbon, the CMB analysis will not be able to yield
quantitative estimates for all sources of carbon.
While carbon is certainly a substantial contributor to the fine mass, the largest
constituent is sulfur in the form of sulfate, including sulfuric acid and ammonium sulfate
(Dzubav et al1982; Dzubay et al., 1987; Stevens et al., 1980). In the absence of sulfur
speciation for these samples, sulfur is represented by sulfuric acid in Table 7-15. This
illustrates the minimum contribution of sulfur to fine mass, which ranges from 37% to
55%. Ammonium sulfate estimates would be about 25% higher than the sulfuric acid
estimates. In addition, any sulfates present would have some amount of water associated
with them which could substantially increase the mass. The amount of the increase
depends on the form of the sulfate and the relative humidity history (Dzubay et al., 1987;
Vossler and Macias, 1986). Sulfate particles are the result of conversion of gaseous SO2
emissions, largely from coal combustion, to the particulate sulfate and as such are regional
and generally associated with long-range transport.
7-52
-------
Table 7-15. Average Fine-Particle Concentrations (ng/m3) and Uncertainties of
Elements Used in CMB Source Apportionment.
18 July
19 July
20 July
21 July
2 August
6 August
8 August
MASS
36,748±l 175
45,347±1354
30,291 ±1118
36,261±1191
30,616± 1024
15.327±1008
29,620±1122
Na
97 ±6
60 ±4
36 ±2
120 ± 8
54 ±4
41 ± 3
50 ±4
Mg
320 ± 70
100 ±20
39 ± 17
65 ±40
82 ±20
47 ±25
190 ± 10
A1
69 ± 44
156 ±54
94 ±43
182 ± 51
173 ±47
87 ±41
27 ±40
Si
387 ±57
434 ± 64
233 ±37
259 ±41
293 ± 44
85 ± 19
131 ±25
P
10± 17
36 ±20
18 ± 17
-13 ± 16
48 ± 16
19 ± 12
2 ± 15
S
5820 ± 396
7526 ±512
5402 ± 367
5411 ± 369
4812± 326
1854 ± 127
3794 ± 259
CI
5 ± 4
16 ± 5
9 ± 4
10 ±4
12 ± 4
7 ± 4
74 ±7
K
77 ±6
79 ±6
56 ± 5
175 ± 12
80 ±6
98 ±7
146 ± 10
Ca
177 ± 12
171 ± 12
83 ±6
95 ±7
198 ± 14
89 ± 7
63 ± 5
Ti
13 ± 5
2 ± 5
7 ± 4
11 ±4
10 ± 4
-8 ±5
6 ± 4
V
0.98 ± 0.18
1.30 ±0.3
1.30 ± 80
1.20 ±0.20
0.75 ± .07
0.70 ±.06
0.94 ± .11
Cr
1.6 ±0.7
1.4 ±0.7
-0 6 ± 0.6
0.9 ± 0.6
0.4 ±0.6
0.9 ± 0 6
2.6 ± .7
Mn
5.1 ±0.8
5.4 ±0.8
1.6 ± 0.7
23.9 ± 1.9
4.6 ±0.7
13.0 ± 1.2
59.3 ±4.1
Fe
158 ± 14
154 ± 14
88 ± 8
336 ± 30
170 ± 15
393 ± 35
558 ±50
Ni
0.6 ±0.7
0.7 ±0.7
0.0 ±0.6
-0.6 ± 0.6
0.4 ± 0.6
-0.8 ± 0.6
0.7 ±0.7
Cu
14.6 ± 1.6
14.3 ± 1.6
6.4 ± 1.0
22.3 ±2.3
7.9 ± 1.0
21.0 ± 2.1
15.6 ± 1.7
Zn
40 ±4
27 ±3
16 ± 2
212 ± 19
42 ±4
51 ± 5
294 ± 26
As
0.52 ±0.05
0.46 ±0.06
0 46 ±0.05
0.66 ±0.07
0.67 ±0.07
0.93 ±0.10
2.30 ±0.20
Se
2.1 ±0.5
1.4 ±0.4
1.6 ±0.4
3.2 ± 0.5
0.3 ±0.3
1.7 ±0.4
4.1 ± 0.6
Br
4.0 ±0.6
4.0 ±0.7
2.7 ±0.6
4.4 ± 0.7
3.4 ±0.6
3.1 ±0.6
4.9 ±0.7
Sr
0.8 ±0.6
1.2 ±0.6
0.4 ± 0.5
0.4 ± 0.5
-0.4 ±0.5
-0.4 ± 0.5
0.0 ±0.6
Zr
-0.7 ± 3.0
2.1 ± 3.3
-8.1 ±2.7
3.6 ± 3.1
-3.6 ±2.6
-0.6 ± 3.0
3.6 ± 3.1
Mo
1.8± 3.0
5.9 ±3.3
"\9 ± 3.1
-3.1 ±2.8
-1.6 ± 2.6
4.4 ± 3.1
1.9 ± 3.0
Pd
-3.5 ±2.1
-1.1 ±2.3
-1.5 ±2.1
2.6 ±2.2
2.5 ±2.0
1.6 ±2.3
-2.0 ±2.1
Cd
2.1 ±2.8
0.2 ±2.9
5.0 ±2.9
2.2 ±2.8
1.2 ± 2.5
-2.0 ±2.8
-1.6 ± 2.8
Sn
20.1 ± 5.4
2.2 ± 5.1
3.0 ±4.8
7.0 ±4.8
-6.1 ±4.3
6.6 ± 5.0
11.7 ± 5.1
Sb
1.60 ±0.10
1.40 ±0.10
0.72 ±0.06
1.10 ± 0.10
0.98 ±0.07
3.00 ±0.20
1.90 ±0.10
Ba
21 ± 10
39 ± 11
11 ±9
9 ± 9
9 ± 8
33 ± 10
18 ± 10
La
0.140 ± .01
0.240 ± 0.020
0.098±.008
0.170 ±0.01
0.240±0.02
0.093±.007
0.230±.020
Pb
23.3 ±2.6
10.3 ± 1.7
8 7 ± 1.5
61.7 ± 5.8
11.9± 1.6
29.9 ± 3.1 -
62.1 ± 5.8
S as
17,823
23,049
16,543
16,570
14,736
5651
11,620
h2so4
h2so4/
0.49
0.51
0.55
0.46
0.48
0.37
0.39
MASS
EC
2420 ± 242
2210 ±221
1010 ± 101
1270± 127
2050 ± 205
1170 ± 117
1870 ± 187
OC
10,700 ±1070
16,200±1620
6870 ± 687
10,020 ± 100
8190 ± 819
5550 ± 555
6230 ± 623
EC/
0.07
0.05
0.03
0.04
0.07
0.08
0.06
MASS
OC*l 4
0.40
0.50
0.32
0.38
0.37
0.50
0.30
MASS
7-53
-------
Table 7-16. Average Coarse-Particle Concentrations (ng/m3) and Uncertainties of
Elements used in CMB Source Apportionment.
18 July
19 July
20 July
21 Julv
2 August
6 August
8 August
MASS
31,785±1095
37,753±1248
22.845±995
10.997±898
48,282±1270
19,567±966
10,017±900
Na
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Mg
N/A
N/A
N/A
N/A
N/A
N/A
N/A
A1
1097 ± 328
1865 ± 546
1330 ± 392
250 ± 109
2493 ± 716
413 ± 148
734 ± 223
Si
3971 ±991
5552± 1380
4786± 1183
1182 ±303
8368±2062
1556 ± 387
430 ± 115
P
53 ±27
34 ±25
25 ±20
0.7± 15.6
72 ±34
5.8 ± 18
15 ± 16
S
259 ± 176
270 ± 222
208 ± 162
251± 165
304 ± 154
510 ± 98
290 ± 126
CI
67 ±9
42 ± 8
33 ±6
126 ± 15
68 ± 10
175 ± 19
130 ± 16
K
396 ± 40
479 ± 48
409 ±41
109 ± 14
777 ± 77
154 ± 17
55 ±8
Ca
2888 ± 217
2458 ± 185
1514 ± 114
815 ±62
4083 ± 306
1980 ± 149
537 ±41
Ti
84 ± 14
123 ± 17
100 ± 14
32 ±6
167 ± 25
47 ±7
19 ± 5
V
2.6 ± 1.9
7.7 ± 2.1
2.0 ± 1.7
1.0 ± 1.5
3.2 ± 2.1
4.4 ± 1.7
3.1 ± 1.6
Cr
3.8 ± 1.0
7.9 ± 1.4
3.2 ±0.9
0.5 ± 0.7
5.8 ± 1.1
4.6 ± 1.0
3.7 ±0.9
Mn
24.8 ±2.4
25.5 ±2.6
23.7 ±2.3
17.0 ±2.0
49.3 ±4.3
64.2 ± 5.6
23.2 ± 2.9
Fe
954 ± 96
1133 ± U3
854 ± 85
905 ± 93
1894 ± 188
1518 ± 153
884 ± 93
Ni
1.2 ± 0.7
5.4 ± 1.1
-0.6 ±0.7
0.5 ±0.7
0.7 ±0.7
0.5 ±0.7
1.1 ±0.7
Cu
13.1 ± 1.8
93 ± 10
7.9 ± 1.2
9.6 ± 1.6
18.7 ± 2.3
8.5 ± 1.5
9.2 ± 1.5
Zn
29.5 ±3.7
49.8 ±5.7
21.2 ±2.6
44.4 ± 7.6
55.1 ± 6.4
31.8 ± 4.1
67 ± 11
As
1.5 ±0.9
0.0 ±0.9
0.7 ±0.8
-1.1 ±0.8
2.0 ± 1.0
0.8 ±0.8
0.0 ±0.8
Se
-0.3 ±0.4
-0.9 ±0.4
0.0 ±0.4
0.8 ±0.4
0.0 ±0.3
0.3 ±0.4
-0.5 ±0.3
Br
1.1 ±0.5
0.4 ±0.5
0.6 ±0.5
1.1 ±0.4
0.8 ±0.5
0.6 ±0.5
0.8 ±0.5
Sr
5.1 ±0.7
5.1 ±0.8
3.1 ±0.6
1.5 ± 0.5
6.5 ±0.8
2.7 ±0.6
1.3 ±0.5
Zr
0.2 ± 2.8
9.4 ± 3.5
3.9 ± 3.0
-3.9 ±2.6
1.5 ± 2.5
-0.9 ± 2.7
2.9 ±3.1
Mo
-2.6 ± 2.6
8.3 ± 3.3
0.9 ± 2.8
0.8 ±2.6
-0.8 ± 2.4
1.4 ± 2.8
0.8 ± 2.8
Pd
1.3 ± 2.0
-1.3 ± 2.1
-0.7 ± 2.0
0.0 ± 2.0
O
If
00
0.3 ±2.1
-1.9 ± 2.0
Cd
2.8 ±2.7
0.9 ± 2.7
-0.5 ±2.6
3.5 ± 2.6
1.1 ± 2.3
-0.5 ± 2.6
0.2 ± 2.6
Sn
5.2 ±4.6
-4 .1 ± 4 .8
-4.6 ±4.5
0.5 ± 4.3
6.1 ±4.1
8.4 ± 4.7
1.9 ± 4.6
Sb
2.5 ± 3.9
-7.2 ±4.1
0 0 ±3.9
4.6 ± 3.8
-0.4 ± 3.4
11.3 ±4.3
2.1 ± 4.0
Ba
31.3 ± 9.2
24.8 ±9.7
20.9 ±9.0
14.6 ± 8.4
41.7 ± 8.9
9.7 ± 8.8
16.3 ± 9.1
La
28 ±21
6 ±22
54 ±22
0 .7± 21.1
17 ± 19
33 ±22 .
-5 ±22
Pb
17.4 ±2.2
21.7 ± 2.5
12.8 ± 1.8
11.4 ± 2.1
32.0 ± 3.3
9.0 ± 1.7
8.8 ±2.0
A relationship can be observed between the meteorology described for these
samples in Chapter 6 and the fraction of fine mass consisting of sulfates. Recall that in
Chapter 6, the meteorology during the second week of the study (July 14 - 20) was
described by regional flow from the S-SW. This likely explains the higher fraction of
sulfate in the samples collected on 18-20 July 91 listed in Table 7-15. The sulfate fraction
drops to below 40% for samples collected on 6 and 8 August 91 which were associated
with local transport from the S.E. Chicago area. These samples plus the sample collected
7-54
-------
on 21 July 91 had the largest fine-particle concentrations of K, Mn, Cr, Zn, Se and Pb. In
the coarse fraction, these samples had the largest CI concentration and the smallest soil-
related element concentrations.
Comparison of ratios of soil-related elements (Al, K, Ca, Fe) as measured in the
ambient air with ratios representative of crustal material can reveal whether these elements
have non-crustal sources. It was assumed that Si in both the fine- and coarse-particle
fraction is dominated by crustal sources. In Table 7-17, these ratios are compared with
crustal ratios based on crustal shale and crustal limestone profiles obtained from U.S.
EPA's Volatile Organic Compound (VOC)/Particulate Matter (PM) Speciation Data
System (U.S. EPA, 1991) (henceforth referred to as SPECIATE). Again, the influence of
local transport of pollutants is evident for samples 6 and 8 August 91 from the ratios of
fine fraction Fe to Si which are much greater than crustal values.
Table 7-17. Ratios (± analytical uncertainty) of Soil-Related Elements to Si.
Sample
Date
Size
Fraction
Al/Si
K/Si
Ca/Si
Fe/Si
18 July 91
19 July 91
20 July 91
21 July 91
2 August 91
6 August 91
8 August 91
Coarse
Coarse
Coarse
Coarse
Coarse
Coarse
Coarse
0.27 ±0.11
0.36 ±0.13
0.28 ±0.11
0.21 ±0.11
0.30 ±0.11
0.27± 0.12
1.71 ±0.69
0.100 ±0.027
0.086 ±0.023
0.085 ±0.023
0.092 ±0.026
0.093 ±0.025
0.099 ±0.027
0.127 ±0.039
0.73 ±0.19
0.44 ±0.12
0.32 ±0.08
0.69 ±0.18
0.49 ±0.12
1.27 ± 0.33
1.25 ±0.35
0.24 ±0.06
0.20 ±0.05
0.18 ±0.05
0.76 ± 0.21
0.23 ± 0.06
0.98 ±0.26
2.06 ±0.59
18 July 91
19 July 91
20 July 91
21 July 91
2 August 91
6 August 91
8 August 91
Fine
Fine
Fine
Fine
Fine
Fine
Fine
0.18 ± 0.12
0.36 ±0.13
0.40 ±0.20
0.70 ±0.23
0.59 ±0.18
1.02 ±0.54
0.21 ±0.31
0.20 ±0.03
0.18 ±0.03
0.24 ± 0.04
0.68 ±0.12
0.27 ±0.05
1.16 ± 0.28
1.12 ± 0.23
0.46 ±0.07
0.39 ±0.06
0.36 ±0.06
0.36 ±0.06
0.68 ±0.11
1.05 ±0.25
0.48 ± 0.10
0.41 ±0.07
0.36 ±0.06
0.38 ±0.07
1.30 ±0.24
0.58 ± 0.10
4.64 ± 1.13
4.26 ± 0.91
Crustal Shale
Crustal
Limestone
0.293
0.175
0.097
0.112
0.081
12.58
0.173
0.158
7-55
-------
In general, all of the crustal elements in the fine fraction have some non-crustal
component and, as stated above, this component appears to be quite substantial for some
elements and samples. In the coarse fraction, K/Si is consistent with crustal estimates for
every sample. Coarse fraction Al/Si is similar to crustal values with the exception of the
sample collected on 8 August 91 which deviates substantially from the crustal value. The
coarse Ca/Si is consistently higher than crustal shale values, indicating influence of some
Ca-rich source, most likely limestone which is used in the steel-making process or in
making cement and is abundantly present in exposed piles of raw materials. The influence
of this Ca-rich source is especially evident for samples collected on 6 August 91 and 8
August 91. Coarse Fe/Si likewise indicates the influence of some non-crustal Fe source,
especially for samples collected on 6 and 8 August 91. These observations shall be
considered in the CMB analysis and interpretation.
7.6.1. The Chemical Mass Balance Model
The U.S. EPA/DRI Chemical Mass Balance Model, version 7 (Watson et al.,
1990) was used to quantitatively apportion the major sources of fine and coarse-particle
chemical species measured on the selected sampling days at IIT to their major sources.
The CMB model consists of an effective variance least squares solution to a set of linear
equations which expresses each measured chemical species concentrations as a linear sum
of the contributions of each source to the chemical species. The effective variance
solution gives the most weight to the source or ambient measurements with the lowest
uncertainty estimates. Source contributions are expressed as the product of the abundance
of the species as emitted by the source and the total mass concentration contributed by the
source. The set of abundances of all species as emitted by a particular source represents
the "source profile". In practice, it is not possible to apportion the total mass to every
individual contributing source in an airshed. Individual sources may be too similar to one
another, too numerous, or may not contribute significantly to the total mass loading or to
individual species. Similar sources are generally considered as a single "source category"
or "source type". Because there are many incinerators in the Chicago airshed, they were
considered as one incineration source type and represented by a single profile.
In performing chemical mass balance calculation, it is assumed that: 1) the
abundance of each species used in the fitting procedure is known for each source type, and
2) all major sources of each species used in the fitting procedure must be included in the
CMB. Other assumptions made for the CMB model are listed in Watson et al., (1990).
7-56
-------
One of the performance goals of the CMB model was to account for the total mass
of each element, within the uncertainties of the measurements involved. A complete
apportionment of the PM10 mass measured in this study was made more difficult because
few source profiles included carbon abundances, and total carbon concentrations were not
determined for the coarse fraction ambient data. It should be noted that biological
material (pollens, spores, plant debris) may comprise a significant portion of the coarse-
particle mass. Also, organic and elemental carbon are important components of motor
vehicle particulate emissions, especially for diesel vehicles. Apportionment of vehicle
particulate emissions is more difficult since tetraethyl lead has been removed from the fuel
supply. Historically, Pb has been used as a tracer for motor vehicle particulate emissions.
It was postulated that volatile organic species measured during the LMUATS could be
used as tracers for the motor vehicle particulate emissions (Zweidinger et al., 1990).
However, the VOC data collected has limited applicability to CMB modeling because it
was found to be unreliable.
7.6.2. Source Profiles
A combination of profiles available in the literature and in the U.S. EPA Speciation
Data System or "SPECIATE" (U.S. EPA, 1991) were used to predict ambient species
concentrations. Sources considered in the CMB analysis were chosen based on the results
of previous work and knowledge of emissions in the Chicago area, as well as on the
examination of crustal element ratios and meteorological conditions. Much of the fine
fraction was assumed to originate from regional sources and consist primarily of sulfate.
A single constituent source profile for sulfur represented by sulfuric acid was included to
account for the minimum contribution of sulfate to the fine-particle mass. A profile
represented sulfur as ammonium sulfate would yield an estimate about 25% higher than
the sulfuric acid estimates. While this assumption accounts for the secondary sulfate, and
thus a large portion of the fine mass, it does not yield any information on the specific
source types that contributed to the secondary sulfate.
Sweet et al. (1993) reported that steel-mill stack emissions are large contributors
to the ambient loadings of certain toxic elements. Furthermore, the PCA and SEM
analysis, discussed earlier in this chapter, confirmed that iron-steel emissions are important
sources in the southern Lake Michigan Basin. Unfortunately, the profiles available which
represent steel-related stack emissions are poor quality. However, surface dust profiles
associated with steel making (Vermette et al., 1990) were be the only steel-related profiles
7-57
-------
included in this analysis. A locally obtained lime dust profile (Vermette et al., 1990) is
included to account for the excess crustal calcium values. An urban road dust profile
developed from S.E. Chicago samples (Vermette et al., 1988) was also considered.
Incinerator emissions in urban areas can account for most of the Zn and Pb
concentrations, although they do not contribute substantially to the overall particulate
concentration (Sweet et al., 1993; Vermette et al., 1991; Conner et al., 1993). The best
available incineration profile in terms of quality of sample collection and analysis is profile
#17105 found in the U.S. EPA's SPECIATE data system. This profile was determined
from samples collected at a municipal incinerator in Philadelphia (Olmez et al., 1988). A
profile developed from samples collected in E. Chicago, IN was available (#17107), but
the methods of collection and analysis were not consistent with the procedures employed
during LMUATS (i.e. XRF analysis was not performed). In addition, fine and coarse
fraction particle samples were not collected as they were for the Philadelphia incinerator.
The Philadelphia incinerator profile demonstrates that the fine and coarse fractions differ
substantially in their elemental composition. Despite the differences in the two incinerator
profiles, the Philadelphia incinerator profile is similar to the E. Chicago incinerator profile
for some important elements: 11.5% versus 11.4% for Zn, 8.1% versus 6.9% for Pb.
From these results it was determined that the Philadelphia incinerator source profile would
be used for the current application.
The SPECIATE profile for crustal shale (#43305) was used to account for crustal
element sources. Some local industries burn oil to produce power and refineries burn
residual oil. Both an oil power plant (#11501) and a residual oil profile (#13501) from
SPECIATE are considered in the analysis to account for some of the Ni and V. The same
sources were considered for the coarse fraction as for the fine fraction, with the exception
of tK single-constituent sulfate profile. Source profiles are listed in Table 7-18 along with
literature references and abbreviations used in this analysis.
7-58
-------
Table 7-18. Chemical Mass Balance source profiles.
Profile
ID
Profile Description
Abbreviation
Reference
N/A
Sulfuric acid
H2S04
N/A
05P014
Steel yard road dust
STLYARD
Vermette et al., 1990
06P001
Coke yard road dust
COKEYARD
Vermette et al., 1990
06U004
Coal yard road dust
COALYARD
Vermette et al., 1990
07S301
Limestone dust
LIMEDUST
Vermette et al., 1990
N/A
Urban Road Dust
URBDUST
Vermette et al., 1988
43305
Crustal shale
SHALE
SPECIATE (U.S.
1991)
EPA,
17105
Incinerator emissions
INCIN
SPECIATE (U.S.
1991)
EPA,
11501
Oil-fired power plant
OILPP
SPECIATE (U.S.
1991)
EPA,
13501
Residual oil combustion
RE SOIL
SPECIATE (U.S.
1991)
EPA,
7.6.3. CMB Results
Results of the CMB analysis are discussed below. These results represent a
preliminary approximation of an obviously complex airshed. It should be noted that all
source types could not be included in the source estimates, but every effort was made to
provide a complete apportionment for as many of the toxic p'ements as possible with the
limited amount of information available. It should be noted that all source types could
not be included in the source estimates. However, meteorology and knowledge of
emissions are considered in the interpretation of results to establish a physical basis for
source estimates. Results of the SEM/EDS analyses are discussed later in light of the
CMB results to support or supplement those results. SEM/EDS may offer clues as to
which types of sources need to be included in future CMB efforts in this airshed.
7-59
-------
7.6.3.1. Coarse Fraction
The estimates of source contribution to the coarse fraction particulate samples are
summarized in Table 7-13. The source contributions for samples collected on 18 - 20 July
and 2 August include profiles for INCIN, SHALE, STLYARD and LIMEDUST. In
addition, a RESOIL profile was included for the sample collected on 19 July to
supplement the information for Ni and V. Also, an URBDUST profile was used to
provide additional information for most of the elements for the sample collected on 6
August. A satisfactory apportionment was not obtained for the sample collected on 8
August using the available source profiles. The appropriate species that provided an
optimum fit to the model were selected to maximize the degrees of freedom (number of
species minus the number of sources) and minimize Chi2 and maximize r^. These
indicators were within the desired values established by Watson et al. (1991).
Table 7-19. Summary of CMB Results for Coarse-particle Fraction in Percent of
Measured Mass with Uncertainty Estimates.
SOURCE CATEGORY
Sample
Date
RESOIL
URBDUST
INCIN
SHALE
STLYARD
LIMEDUST
UN-
EXPLAINED
18 July 91
0
0
1.1±0.3
41.0±4.1
4.1±1.6
18.9±3.0
39.4 ±5.3
19 July 91
0.2±.07
0
2.0±0.4
44.7±4.4
2.7±1.6
13.3±2.2
37.1 ±5.2
20 July 91
0
0
1.2±0.4
64.1±7.6
5.2±2.4
12.1±2.3
17.4 ±8.3
21 July 91
0
0
2.6±0.9
30.6±4.4
26.8±3.4
8.0±2.6
32.0 ±6.2
2 Aug 91
b
0
1.2±0.4
51.3±6.2
7.4±1.5
16.6±3.0
23.5 ± 7.1
6 Aug 91
0
13.6±3.6
0
21.3±3.4
31.2±4.7
10.0±3.4
23.9 ±7.6
8 Aug 91
-
0
-
-
-
-
-
The CMB results for 18-20 July 91 and 2 August 91 were similar. Also, the
meteorological conditions were similar for these days. For each of those sampling days
the PM10 concentration was > 50 (ig/m3 and the total coarse particle concentration was
7-60
-------
> 20 (ig/m3. Mixed-layer backward trajectories for these days (see Figure 7-2) indicated
long-range transport from the East St. Louis area. For each of these samples crustal
material, represented by SHALE, comprises the largest fraction of the coarse mass (41%-
69%), followed by LIMEDUST (12%-19%), STLYARD (3%-5%) and INCIN (l%-2%).
The latter two profiles accounted for significant amounts of some elements but
contributed very little to the total coarse mass concentration. The RESOIL applied to the
sample collected on 19 July 91 accounted for a very small (0.2%) portion of the coarse
particle mass but was necessary to account for the higher mass of V and Ni. These
elements are characteristic of oil combustion sources. While most of the coarse mass was
apportioned, it should be noted that a significant amount (17-39%) could not be
apportioned using the CMB model.
The major source contribution for the sample collected on 21 July was STLYRD
followed by SHALE and LIMEDUST. The winds were strong on this day coming from
the SSW. The mixed-layer backward trajectory indicated that the air mass passed
northward away from the East St. Louis area, and passed instead across the common
border of Iowa, Missouri and Illinois. The coarse particle mass concentration for this day
was 11 |ig/m3, significantly lower than for the other samples measured. However, the
local wind characteristics may provide some additional information.
The meteorological conditions were significantly different on 6 August 91 with
winds generally from the ESE to SE. This difference was reflected in the change in the
overall apportionment. Winds from the ESE to SE would result in greater influence of the
industrial and fugitive sources associated with S.E. Chicago. Indeed, STLYARD is
estimated to be the dominant source (31%) of the coarse mass for this sampling day,
followed by SHALE (21%), URBDUST (14%) and LIMEDUST (10%). The URBDUST
profile was needed to account for nearly half of the Cr and Mn. Howe1'',., the individual
species apportionment indicates that some species, most notably Pb, were significantly
overpredicted by STLYARD. However, the only stack-related steel emissions profiles
available were of poor quality and thus were not included in any final apportionments,
although unsuccessful attempts were made to include them in preliminary calculations.
7.6.3.2. Fine Fraction
The estimates for source contribution to the fine fraction particulate samples are
summarized in Table 7-20. The steel-related fugitive dusts (STLYARD, COKEYARD,
COALYARD) could be used interchangeably in the CMB calculations with equivalent
7-61
-------
mathematical fits to the data. There was not sufficient meteorological and emissions
information available to determine which dust profile best represented the true source
contributions, so the range of results is reported for each sample. These profiles cannot be
included in the CMB simultaneously because they are collinear. The appropriate species
that provided an optimum fit to the model were selected to maximize the degrees of
freedom (number of species minus the number of sources) and minimize Chi2 and
maximize r^.
Table 7-20. Summary of CMB Results for Fine-particle Fraction in Percent of
Measured Mass with Uncertainty Estimates.
SOURCE CATEGORY
h2so4
RESOIL
INCIN
SHALE
URBDUST
(21 July 91)
STLYARD/
COKEYARD/
COALYARD
MMEDLST
UN-
EXPLAINED
18 July 91
47.8±5.8-
0
0.8±0.1-
0.7±1.0 -
1.6 ±0.3 -
0.5±0.3 -
28.2±7.6 -
48.4±0.9
0.9±0.1
3.1±0.6
22.0 ±4.8
1.3±0.2
45.2±5.9
19 July 91
50.3±6.1-
0
0.35±.04-
1.2±0.9 -
1.2 ±0.3 -
0.4±0.2 -
30.5±7.3 -
50.7±6.1
0.45±0.05
3.2±0.6
17.3 ±3.9
1.0±0.2
43.8±6.1
20 July 91
54.3±6.6-
0
0.36± 04-
1 6±0.8
0.7 ±0.3 -
0.3±0.2 -
32.6±7.6 -
54.6±6.6
0.42±0.05
2.8±0.5
10.8 ± 3.7
0.7±0.1
41.0±6.6
21 July 91
44.7±5.5-
0
2.3±0.2-
2.1±0.3 -
6.1 ±0.9 -
0
22.4±6.9 -
45.3±5.5
3.2±0.3
4.0±0.6
26.4 ±4.2
44.2±5.6
2 Aug 91
47.3±5.7-
0
0.6±0.1-
0.2±0.9 -
1.9 ± 0.4 -
0.8±0.3 -
22.8±7.9 -
48.0±5.8
1.0±0.1
3.2±0.6
28.3 ±5.4
1.8±0.3
44.9±5.9
CMB source apportionment of the fine fraction proved more problematic than the
coarse fraction. This may be -.ttributed to the limited useful source profiles for non-
fugitive steel mill emissions and to the overwhelming dominance of sulfate in the fine
fraction. Also, the absence of a representative motor vehicle profile combined with the
lack of carbon abundancies in some of the existing profiles further limit the fine fraction
source apportionment.
The CMB results for samples collected on 18-20 July 91 and 2 August 91 were
similar, which is consistent with the results obtained for the coarse fraction. As discussed
previsously, the meteorological conditions were similar for these days. For each of these
sampling days the total PM10 concentration was >50 |ig/m3 and the total fine particle
7-62
-------
concentration was >30 ng/m3. Source contribution estimates were determined using
profiles for H2SO4, RESOIL, INCIN, SHALE, LIMEDUST and steel-related fugitive
dusts (STLYARD, COKEYARD, or COALYARD). Sulfur, represented by sulfuric acid
in the CMB, contributed the largest fraction of the fine mass. The other sources
accounted for most of the measured elements, but only a small portion of the total fine
mass. The steel-related fugitive dust contribution was largest for COKEYARD and
smallest for STLYARD. The SHALE estimate varied inversely to these estimates. Since
each type of fugitive dust may be transported from the same general source region, more
detailed profiles and meteorological information are needed to determine which, if any, is
contributing to the fine mass. The individual species apportionments revealed that each of
the fugitive dust profiles accounted for much of the Fe and Mn. It should be noted that no
stack-type steel-making profile was available to be included in the apportionment. Such a
profile could significantly change the apportionment of the steel-related elements.
However, this would not significantly affect the source contribution to the total fine mass.
The sample collected on 21 July 91 could not be described well based on the
source profiles utilized. This sample differed from the samples discussed above in several
ways: 1) it had the smallest sulfate contributions and the largest INCIN contributions, 2) a
locally-obtained urban dust profile (URBDUST) was required in place of SHALE to
account for much of the soil-related elements, 3) Ca was overpredicted by the fugitive
dust profiles and 4) diagnostics indicating that the apportionment may not be reliable as
evidenced by the larger Chi2 and the smaller r2.
As previously discussed for the coarse fraction, the meteorological conditions
during sample collection on 6 and 8 August 91 was significantly different from the other
sampling periods. These samples indicated an increased impact of local pollutant sources
contributing to the fine fraction samples. Because the profiles of the local sources may
have been adequate, these fine fraction samples were determined to be "unapportionable".
This indicates a need for an improvement in the profiles of the local sources.
A characteristic of all of the fine-fraction CMB apportionments is the substantial
overprediction of CI. A loss of CI after sample collection is seen frequently, and is
presumed to be due to on-going reactions with atmospheric pollutants or volatilization. In
addition, Cu could not be accounted for by the sources considered in the apportionments.
Its underprediction could be caused by incorrect or inadequate local profiles. However,
long-range transport of fine Cu from smelting operations in the East St. Louis area could
account for the underprediction.
7-63
-------
There was a relatively large fraction (22% - 45%) of the total fine mass which
could not be accounted for by the sources included in the calculations. This may be due to
the inadequate profiles for current motor vehicle emissions. However, this may also be
due to water associated with sulfate particles.
7.6.4. Comparison of Individual Particle Analysis and CMB Results
Results of the SEM/EDS analyses are discussed to support or supplement the
CMB results. SEM/EDS analysis was useful in providing clues as to which types of
sources need to be included in future CMB efforts for this airshed. A detailed discussion
of the SEM/EDS results are presented in section 7.3 of this chapter. It should be noted
that the SEM/EDS results are reported as percent by number and not by mass, as are the
CMB results. Thus, because of differences in particle size and density, the two results are
not equivalent. Because of the time required for manual SEM/EDS analysis, only about
100 particles per size-fraction per sample could be examined, limiting the
representativeness of each analysis.
7.6.4.1. Coarse Fraction
The SEM/EDS analysis reports that all coarse fraction samples were dominated by
minerals, which includes both Si-rich and Ca-rich components. The mineral fraction was
largest for samples collected on 18 and 19 July 91 and on 2 August 91. Results for these
samples are discussed together because they are similar for both SEM/EDS and CMB
results. SEM/EDS was not performed on the sample collected on 20 August 91. Silicates
and quartz comprise 56-59% by number of the coarse fraction of these 3 samples, a little
greater than the 41-51% source contribution estimate for SHALE. Some of the Si-rich
particles may be associated with other fugitive dusts such as STLYARD. Particles
containing Ca comprised 24-31% compared with 13-19% estimated for LIMEDUST by
CMB. This disparity can be explained by the fact that some of the Ca-rich particles may
occur in the SHALE, STLYARD, or other fugitive dusts. The carbonaceous particles
account for 5-10% by number and could explain some of the unapportioned mass. These
particles represent spores and plant debris (natural sources) and soot (combustion). Plant
debris was generally the largest type of particle found (5-10 ^m) and could thus represent
more mass than other types of particles. Industrial particles represented only a small
fraction (4-9%) of the coarse particles, consistent with CMB results. However, individual
particle analysis revealed a complexity which was not possible to show with CMB. For
instance, some flv ash spheres were found, but this type of source could not be
7-64
-------
apportioned by CMB because they were similar in elemental composition to other sources.
It should also be noted that the absence of some particles from the analysis does not
indicate a lack of influence of the responsible source. However, this may indicate that an
insufficient number of particles were examined.
The SEM/EDS results for 21 July 91 differed considerably from the 3 samples
discussed above, as did the CMB results. Both individual particle analysis and CMB show
that industrial emission were more important for this sample than the previous 3 samples.
Particles identified as originating from industrial sources comprised 26% of the coarse
particles; 7% were Fe spheres emitted from combustion sources and the remainder was a
mixture of elements from industrial sources. The CMB yielded a remarkably similar result
(27%) by representing Fe-rich sources with the STLYARD fugitive dust profile. This
agreement may be fortuitous, as a dust profile implies that particles are not the result of
combustion. On the other hand, some of the combustion products from the nearby stacks
may have settled to the ground in the area where the STLYARD samples were collected.
There is a disparity between the SEM/EDS estimate for Ca-rich particles (29%) and the
CMB estimate for LIMEDUST (8%) which cannot be explained by the available evidence.
The sample collected on 6 August 91 had a significant amount of industrial
particles (23%), more than half of which were Fe spheres from combustion sources. This
is consistent with the winds blowing from the heavily industrial S.E. Chicago area on that
day. In the CMB analysis, most of the Fe (71%) was attributed to STLYARD, a fugitive
dust profile. The lack of information on the particle morphology of the source samples
limited interpretation of these results. Twenty-five percent of the coarse particles were Si-
rich, compared with a SHALE source contribution estimate of 21%. SHALE is the
source profile richest in Si. Again, the contribution of Ca-rich particles (29%) is higher
than what would be expected from the CMB source contribution estimate for LIMEDUST
(10%) which suggests that an additional source of the Ca impacted the sample. This
sample contained the largest number of plant debris and spores/pollen (14%) which could
explain much of the mass unaccounted for by the CMB calculation.
7.6.4.2. Fine Fraction
The SEM/EDS analysis reported that all fine fraction samples were dominated by
sulfate. Because more than 99% of the fine particle identified in the SEM/EDS analysis
were sulfate particles, a quantitative comparison with CMB results is impractical. For this
reason, only a qualitative discussion is presented here. Any percentages reported refer to
7-65
-------
the non-sulfate population of fine particles. The sulfate particles identified were all in the
smallest size category (< 0.5 um) and appeared as round droplets. The droplet
appearance supports the hypothesis that water is associated with the sulfate particles and
could account for much of the mass not accounted for by the CMB calculations.
Samples identified as having large numbers of Fe-rich combustion spheres were
those samples which were not apportionable (6 and 8 August 91) or which had
problematic apportionments (21 July 91). Industrial particles accounted for 56 - 79% of
the non-sulfate particles in the 3 samples which were difficult to apportion. Clearly, a
proper apportionment of the fine fraction requires good source profile information for
local industrial sources, which can be responsible for many of the toxic elements.
SEM/EDS results for samples collected on 18 and 19 July 91 and 2 August 91
were similar, as they were for the CMB results. These samples had the smallest
percentage of industrial fine particles (17% -26% of the non-sulfate particles).
Nevertheless, good industrial profiles are needed for these samples as well.
For all samples collected, the soot carbon comprised 2% - 8% of the non-sulfate
fine particles. The size of the soot particles was variable. Since soot may comprise a
substantial portion of motor vehicle emissions, the number and size of soot particles could
potentially be used to estimate an upper limit for motor vehicle emissions.
Some particles were identified as originating from incineration, supporting the
inclusion of the FNCIN profile in the CMB calculations. Fly ash particles were also
identified in some of the samples, but the profile for a coal combustion source did not
provide an adequate apportionment in the CMB calculations.
The SEM/EDS results seem to support the conclusion that the source
apportionment of the fine fraction was often problematic ana that more source profile
information is needed. Source profiles for the major sources in the southern Lake
Michigan Basin should include measurements of carbon and mercury to better characterize
the source impacts in this area.
7-66
-------
Figure 7-10. Ratios of Selected PAHs to BeP Measured on the R/V Laurentian 6 August 1991.
BAA
CPP
CHRY
BFRT
BEP
BAP BGHIP IND COR
7-67
-------
Chapter 8
Estimating Atmospheric Deposition
8.1. Introduction
The transfer of trace metals from the atmosphere to a receptor surface via dry
deposition of particles or gaseous contaminants has been the subject of many studies over
the last twenty years (Slinn etai, 1978; Sehmel, 1980; Nicholson, 1988; Noll et al., 1988;
Wu et al., 1992). However, an in-depth evaluation of the dry deposition flux of trace
metals to large water bodies that considers the dynamics of micro-meteorological
parameters has not been performed. Such an evaluation would have to account for
influences on dry deposition such as particle size distribution, phoretic forces, electrostatic
gradients, and/or the dynamics of water waves controlling the flux of particles at the air-
water interface. The theoretical and experimental knowledge of particle dry deposition
over a water surface is limited due to the inherent difficulties in making direct flux
measurements over water (Wesely and Williams, 1981; Sievering, 1982). Past
measurements of direct dry deposition fluxes made over Lake Michigan (Wesely and
Williams, 1981) and the Atlantic Ocean (Sievering, 1982) were likely to have been
affected by the large uncertainty caused by particle growth in the humidity gradient
present over water. Furthermore, results using the gradient technique have not been
reported in the literature due to the near impossibility in detecting small concentration
gradients (usually 1%) at the air-water interface for particle diameters <1 (j.m .
In order to estimate dry deposition fluxes of contaminants to land and water
surfaces, ambient concentration measurements are typically combined with deposition
velocities obtained from the literature. Another method that has been developed over the
past 10 years is the use of surrogate surfaces (Noll et al1985; Holsen et al., 1993).
However, there are only a few data points using this technique in the Lake Michigan
Basin, and most of these are on land (Strachan and Eisenreich, 1988, 1992; Holsen et al.,
1993). Moreover, there are statistical uncertainties associated with these surrogate
8-1
-------
surface estimates that may be significant due to the fact that (a) the measurements are not
made directly on the water (the information may not be applicable to the air-water
interface); and (b) the dry deposition sampler is operated at a distance above the water
surface (at least 5-15 meters if it is located on board a ship) where the effect of critical
parameters such as phoretic and electrostatic forces, particle growth in the humidity
gradient at the air-water interface, and dynamics of water waves are not accounted for in
the measured fluxes.
It is also difficult to determine the effect of changes in the meteorological
conditions (wind speed, atmospheric stability, surface roughness) on the total dry
deposition of trace metals from the atmosphere to the water surface and on the ambient
concentrations as air masses cross the lake. A lagrangian process can not be easily
implemented because of the small number of direct measurements made at a few fixed
points. Furthermore, a simple extrapolation of the point estimates to the total surface area
of Lake Michigan (about 57,800 km2) may provide highly uncertain results.
This section discusses a new hybrid receptor-deposition modeling approach used
to evaluate the trace metal dry deposition over Lake Michigan during LMUATS. The
dynamic model accounts for both temporal and spatial variations in the dry deposition
fluxes over Lake Michigan for air masses traversing the lake. The variations in the dry
deposition flux are the result of horizontal and vertical dispersion, variations in wind
speed, the action of water waves, and the temperature gradient at the air-water interface.
The results calculated for this study are compared to direct measurements of dry
deposition fluxes and ambient concentrations of trace metals made at land-based sampling
sites as well as over-water measurements on Lake Michigan. Finally, the average dry
deposition fluxes to Lake Michigan are presented for measured trace metals along with a
discussion of the band of uncertainty related to the above mentioned parameters.
8.2. Methods
Dry deposition processes are affected by a wide range of meteorological
parameters (i.e., wind speed, atmospheric stability, relative humidity), pollutant
characteristics (i.e., chemical speciation, particle size and shape) and receptor surface
characteristics (i.e., surface roughness, type of surface). Early attempts to evaluate the
dry deposition flux to large water-bodies assumed that deposition velocities and ambient
concentrations were constant in both time and space and did not explicitly take into
8-2
-------
account the effect of processes that influence the dynamics of particle transfer from the
atmosphere to the water surface and vice versa. These processes include atmospheric
dispersion, particle growth within the humidity gradient, diffusophoresis, thermophoresis,
and wave breaking. Finally, atmospheric processes occur on many different time scales
which also must be considered when evaluating the deposition flux to large water surfaces.
A combined dispersion-deposition modeling approach was developed to estimate
the deposition flux of contaminants over Lake Michigan. The dry deposition flux, F, of
trace metals is related to the ambient concentrations X, the dry deposition velocity of
particles, V^, and the area, S, of the receptor surface through the following expression
(Chamberlain, 1983).
F(0=X(t)Vd(t)S (1)
The flux, F(/), is a function of X and V(j which are not constant in time or space, but
change during air mass transport over Lake Michigan. Mixed layer 72-hour backward and
forward trajectories (Heffier, 1980) were calculated for each LMUATS monitoring site.
The path of the air parcel and the transport layer depth were calculated using both routine
upper-air observations from the National Weather Service (NWS) and supplementary data
taken from selected periods of the Lake Michigan Ozone Study (LMOS) for air parcels
crossing the lake to and from the monitoring sites.
Atmospheric concentrations of contaminants along a trajectory are affected by the
dispersion and deposition of pollutants emitted from sources located in the basin. Since a
complete emissions inventory was not available for all of the sources of hazardous air
pollutants in the Lake Michigan Basin, a virtual source (VS) was constructed from
measured concentrations at the sampling sites in order to estimate the variation of ambient
concentrations due to atmospheric dispersion along the over water trajectory. The VS is
defined as an area source, equivalent to all sources operating in the basin which are along
the path of the calculated air mass trajectory in either the forward or backward direction
from Lake Michigan. The distance of the VS from the lake shore is an important
parameter which can significantly affect the calculated concentrations of contaminants in
the air mass as it traverses the lake. Therefore, the distance of the VS from the coast was
determined through an optimization process that minimized the ratio of the predicted to
the observed concentrations at the monitoring stations. The sampling station where the
8-3
-------
ratios are evaluated is used as a source (or receptor) with respect to the sampling station
where the backward (or forward) trajectories end (or start).
Once the optimal distance of the VS from the lake was defined, the dispersion
parameters were evaluated using the monitoring site as the end point for each backward
trajectory (or the starting point for forward trajectories) crossing the lake and sampling
site. The change in atmospheric concentration of a contaminant along a trajectory,
assuming that the contaminant is uniformly distributed in the volume, dV=dx dy dz, is
governed by the following differential equation:
dM(t) / dt = [X(t) dx(t) dy(y) dz(t) / dt] - Vd X(t) dx(t) dy(t) (2)
where dM(t) is the mass of contaminant in the volume dV = dx dy dz. Assuming no loss
of contaminant through the upper level of the transport layer, the variation of the ambient
concentration along the trajectory is obtained from Eq. 1 by the following mass balances:
Xij (xy yy Zij) - Xy V^y (At xy yy) = Xy.i (xy.j yy.j Zy_i) (3)
^ij-l (xij-l Yij-1 zij-l)~ ^ij-l Vdy.i (At Xij_i yy-i) — Xjj (xy yy zy) (3 )
Eqs.3 and 3' are for trajectories in the backward and forward mode, respectively. The
index i represents the trajectory while j is the trajectory segment index which increases
along the trajectory (for both forward and backward mode) beginning at the monitoring
site. Xy is the concentration of the contaminant in the volume xy yy zy, V^y is the
deposition velocity of the particulate contaminant, and At is the time step which is
constant (At = 3 hr) for each segment.
The first term of these mass balance equations represents the mass of contaminant in
the volume xy yy zy and the second term is the mass of contaminant deposited over the
water surface area, xy yy. The term on the right side is the mass of contaminant advected
along the subsequent trajectory segment after accounting for the lateral spread of the
plume and the depletion of the ambient mass by dry deposition during the previous time
step. The time dependent pollutant concentrations along the backward or forward
trajectories are determined by Eqs. 3 and 3' which leads to the following expressions:
8-4
-------
^¦i,n ^i,l (xi,l Yi,l zi,l ! xi,n Yi,n zi,n) 1—^=2,n D" Vdij At / Zjj]
(4)
^¦i,n ^i,l (xi,l Yi, 1 zi,l / xi,n Yi,n zi,n) n-j=l,n-l Vdy At ! zij] (4)
where Xi,i is the ambient concentration of the pollutant measured at the sampling site
(j=l), Xij is the distance traveled by the plume, z;j is the transport layer depth, yy is the
crosswind plume width, and n is the number of 3-hour trajectory segments. Equations 4
and 4' consider the sampling site as the receptor in backward mode and as the source in
forward mode.
One of the earliest experiments that measured lateral dispersion over long
distances was completed by Richardson and Proctor (1925) with a re-assessment by
Sutton (1932). The data consisted mainly of the distance of balloons from release points
as a function of time and distance in Brighton and Reagent Parks in London. Sutton's
(1932) analysis concluded that the horizontal spread of the balloons varied with a power
law: xq (q = 0.875) for downwind distances greater than 500 km. This finding agrees
fairly well with results obtained later by Braham et al. (1952), Crozier and Seely (1955)
and Pasquill (1956). Data collected from the Mt. Isa smelter plumes in central Australia
(Bigg et al., 1978) with a wider range of downwind distances (15 to 1000 km) provided
an exponential law very close to that estimated by Sutton (1932) for the Richardson and
Proctor (1925) data set. The analysis of data collected over the sea by Crabtree (1982)
for a plume emitted by the Eggborough Power Station followed a power law with a
coefficient (q = 0.5) which is lower than that estimated for plumes transported over land.
The significant difference between sea and land data may be explained by the large
reduction, if not complete absence, of thermally induced turbulence over the water.
However, the observation of the constant power law behavior of plume width with
distance is partially due to the wind directional shear across the boundary layer (Pasquill
and Smith, 1983). Therefore, based on these previous results in the literature, a power
law with coefficient q = 0.5 was chosen to estimate the lateral spread of the plume
crossing the lake. The dispersion parameters for backward trajectories are given by:
yij = (Uij At/2)q j = n (5)
yy = [ Zk=j+i,n-i U,,k At + Uij At / 2 ]q 1 < j < n (6)
8-5
-------
The dispersion parameters for forward trajectories are given by:
y1J = (U1JAt/2)q j=l (5')
yij = [Zk=J+i)n-i Ui,k At + Uy At/2 ]q l 0 (11)
8-6
-------
The Monin-Oboukhov length L was estimated using California Institute of Technology's
linear approximation to Golder's plot (Golder, 1972).
The deposition velocity for fine and coarse particles along each trajectory segment
was calculated using a deposition model formulated for water surfaces. The approach is
based upon Williams' (1982) model which separates the atmosphere into two layers; the
upper layer where particle transport is controlled by turbulent transfer and the lower layer
where particle transport is controlled by gravitational settling. The transfer of particles is
controlled by two parallel paths, one is from air to a smooth water surface while the other
path is from air to a broken water surface with formation of spray and bubbles. The two-
layer formulation then provides the following expression for the deposition velocity of
particles over a water surface (similar to the electrical circuit analysis of Williams, 1982).
Vdlj = (A/B) [(1-ot) (Kss + Vgw) + Km a (Kbs + Vgw) / (Km + a ( Kab + Kbs + Vgw))] (12)
where:
A = Km [(1 - a) Kas + ct Kab + Vgd] + (1 - a) (Kas+ Vgd) a (Kab + Kbs + Vgw)
B = Km [(1 - a) (Kas + Kss) + a (Kab + Kbs) + Vgw] + (1 - a) (Kas + Kss + Vgw) a (Kab+Kbs+Vgw)
and a is the fraction of water surface that becomes broken due to the action of wind speed
(Wu, 1979). The settling velocity is calculated for dry Vgd and wet Vg^- particle
diameters. Further, A and B are related to turbulent transfer coefficients from the
atmosphere to the smooth (Kas) and broken (Kab) water surfaces formulated by Hess and
Hicks (1975), and to the smooth surface transfer coefficient (Kss) found in Slinn and Slinn,
1980. Kp- is the lateral transfer coefficient and Kbs is the broken surface transfer
coefficient thai must be estimated empirically since a mathematical formulation is not
widely accepted. The coefficients (Kas) and (Kab) were used by Williams (1982) as
constants with a range between 0 to oc and Kss = 1000 cm/s. Later in this report, the
values of Kbs and Km are empirically defined using the ambient concentrations and dry
deposition fluxes for fine (<2.5 |im) and coarse (2.5-10 |im) particles measured over Lake
Michigan.
The model of Sehmel and Hodgson (1978) which has been used extensively for
evaluating deposition velocities was used here. While this model was not specifically
formulated for water surfaces, eddy diffusivity, gravitational settling and particle inertia,
8-7
-------
the main mechanisms influencing the dry deposition velocity of particles, were considered.
Finally, the model combines these terms with the terminal settling velocity and Brownian
diffusion to predict deposition velocities.
8.3. Modeling Results
Aerosol samples were collected on 37-mm Teflon filters using dichotomous
samplers at each monitoring site. Trace element concentrations in the fine and coarse
fractions were determined using X-Ray fluorescence (XRF) analysis (Dzubay et al., 1982).
Tables 5-3 and 5-4 list the mean concentrations of selected trace metals in the fine (<2.5
(im) and coarse (2.5-10 (am) fractions of the PM10 measured at Chicago, EL and South
Haven, MI. Concentrations of fine fraction elements were generally lower at South Haven
than at Chicago with the exception of Al which was more abundant in South Haven on the
average. In the coarse fraction, crustal elements were found to be 2 to 3 times lower in
concentration at South Haven. The concentrations of primarily anthropogenically derived
elements such as Cd, Cr and Se were found at similar level at both sites, while Al, Mn, Zn,
and Pb were found to be 2 to 5 times lower at South Haven than those measured at
Chicago.
The deposition velocities of trace metals over Lake Michigan were calculated
using the particle size distributions for specific elements found in Holsen et al. (1993).
These were measured at Chicago (IIT) during the LMUATS. The size distributions were
used to calculate mass median diameters (MMD) in the fine (<2.5 (am) and coarse (2.5-10
(am) fractions. MMDs for elements not measured by Holsen, such as Hg, S, K, Ni, As,
Se, and Br, were taken from the literature (Milford and Davidson, 1985; Sadisivan, 1978).
Table 8-1 shows the MMDs used in this study for trace metals in both size fractions.
8-8
-------
Table 8-1. Mass Median Diameter (|im) for Selected Trace Elements.
Fine
(< 2.5 jam)
Coarse
(2.5-10 |im)
Reference
Al
0.75
4.0
1
Si
1.30
5.0
1
S
0.50
4.0
2
K
1.30
5.0
2
Ca
1.30
5.0
1
Cr
0.75
4.0
1
Mn
0.75
4.0
1
Fe
1.30
5.0
1
Ni
0.75
4.0
2
Cu
0.75
4.0
1
Zn
0.75
4.0
1
As
0.75
4.0
2
Se
0.50
4.0
2
Br
0.50
4.0
2
Cd
0.75
4.0
1
Hg (*)
0.61
—
2
Pb
0.50
4.0
1
(1) Holsen et al. (1993)
(2) Milford and Davidson (1985); Sadisivan (1978), Dodd et al. (1991)
Figure 8-1 shows the mixed layer trajectories from Chicago for 14 and 15 July
1991, moving backwards in time in 3-hour time-steps. The temporal trends in the
calculated deposition velocities for fine particles along the trajectories that crossed Lake
Michigan from north * j south are shown in Figure 8-2. The variations in the calculated
deposition velocities over water are quite large for micron and submicron particles.
8-9
-------
Figure 8-1. Mixed Layer Backward Trajectories that Crossed Lake
Michigan on 14 and 15 July, 1991.
(IIT = Chicago. Small squares indicate the position of the air mass as it
moves backward from the site in 3-hour increments)
8-10
-------
Figure 8-2. Variation of the Deposition Velocity with MMD along Trajectories
Traversing Lake Michigan, (a) 0.50 (im (b) 0.75 |im (c) 1.30 |im.
14-Jul
10
10
0.120 -(c)
0.090
14-Jul
0.060 -I
I
-I
0.030 1
15-Jul
0.000
4 5 6 7
Trajectory Time Step
10
8-11
-------
The dry deposition fluxes of trace metals associated with fine particles are affected
by large variations in the deposition velocity. The variations are about a factor of 3 for a
0.5 jam particle diameter (Figure 8-2a), 10 for a 0.75 (am particle diameter (Figure 8-2b)
and 15 for a 1.3 |j.m particle diameter (Figure 8-2c). Variations of 10 to 20% in
deposition velocity occur for a 4 fim particle diameter and 3 to 10% for a 6 ;a.m particle
diameter along trajectories such as those calculated for the 14 and 15 July.
Table 8-2 gives the model parameters utilized for three different MMDs along the
trajectory calculated for 15 July (shown in Figure 8-1). The coefficient j denotes the
trajectory segment starting at the receptor site (IIT) and moving backwards in time away
from the starting point. The estimated deposition velocities agree fairly well with those of
Williams (1982) and Slinn (1983), whose predicted deposition velocities were 10- to 100-
fold greater for particle diameters >0.8 (am.
Figures 8-3 and 8-4 show the measured/calculated ratios for both fine and coarse
particles obtained for trajectories that traversed Chicago, Lake Michigan and South Haven
on the 19, 20 and 21 of July 1991 (Figure 8-5). The measured/calculated ratios were
obtained assuming a VS operating at 30 km from Lake Michigan with a power law
coefficient of q = 0 .5 to account for the lateral spread of the plume crossing the lake. The
measured/calculated (m/c) concentration ratios for fine particles are generally in the range
of 0.75 to 2. The exceptions to this were A1 and total Hg, which had m/c ratios of 4.6
and 8, respectively on 21 July. On 19 July, Si had a m/c ratio of 6.8 while Pb had a m/c
equal of 3.6 for the trajectory on July 19. Measured/calculated ratios for coarse particles
range from 0.5 to 3 .6 with the exception of Pb, Zn and Mn which have ratios of 5, 8 and
6.5 respectively, for the trajectories on July 19, 20 and 21.
The high ratios obtained for some elements in the fine and coarse fraction may be
explained by (a) an increase in the horizontal spread of the plume (the power-law
coefficient was greater than the value used) caused by unstable atmospheric conditions
yielding lower concentrations in the air parcel crossing the lake; and (b) significant
fumigation leading to higher depletion rates of contaminants from the plume.
Next, the calculated dry deposition fluxes for trace metals are compared with
annual estimates made by Holsen et al. (1993) using surrogate surfaces mounted on the
R/V Laurentian. These surfaces were approximately 5 meters above the water surface,
the same level that the other aerosol measurements were performed.
8-12
-------
Table 8-2. Model Parameters Obtained for Three MMDs along Backward
Trajectory Segments over Lake Michigan on 15 July 1991.
j
ws
U*
10"3a
Km
Kbs
Kas
Kab
Kss
Vgd
Vd
(m/s)
(cm/s)
(cm/s)
(cm/s)
(cm/s)
(cm/s)
(cm/s)
(cm/s)
(cm/s)
Dp =
0.5 mm
1
1.0
0.84
-36.82
0.002
1.0
7
0.0060
0.0071
0.00001
0.0011
0.0012
2
0.9
0.65
-44.41
0.001
1.0
7
0.0041
0.0048
0.00001
0.0011
0.0012
3
1.7
2.61
-14.06
0.010
1.0
7
0.0294
0.0387
0.00020
0.0011
0.0014
4
1.6
2.32
-15.72
0.010
1.0
7
0.0251
0.0326
0.00020
0.0011
0.0014
5
2.6
5.51
-6.32
0.060
1.0
7
0.0759
0.1063
0.00060
0.0011
0.0020
6
2.5
5.17
-6.81
0.050
1.0
7
0.0703
0.0980
0.00050
0.0011
0.0019
7
3.3
7.93
-3.96
0.150
1.0
7
0.1172
0.1672
0.00100
0.0011
0.0025
8
3.3
7.93
-3.96
0.150
1.0
7
0.1172
0.1672
0.00100
0.0011
0.0025
9
4.6
12.74
-2.00
0.520
1.0
7
0.2008
0.2895
0.00180
0.0011
0.0037
10
4.5
12.36
-2.09
0.480
1.0
7
0.1941
0.2797
0.00170
0.0011
0.0036
Dp =
0.75 mm
1
1.0
0.84
-36.82
0.002
1.1
50
0.0060
0.0073
0.00001
0.0026
0.0026
2
0.9
0.65
-44.41
0.001
1.1
50
0.0042
0.0049
0.00001
0.0026
0.0026
3
1.7
2.61
-14.06
0.010
1.1
50
0.0298
0.0413
0.00020
0.0026
0.0033
4
1.6
2.32
-15.72
0.010
1.1
50
0.0254
0.0347
0.00010
0.0026
0.0031
5
2.6
5.51
-6.32
0.060
1.1
50
0.0767
0.1159
0.00050
0.0026
0.0058
6
2.5
5.17
-6.81
0.050
1.1
50
0.0710
0.1067
0.00040
0.0026
0.0054
7
3.3
7.93
-3.96
0.150
1.1
50
0.1181
0.1839
0.00080
0.0026
0.0101
8
3.3
7.93
-3.96
0.150
1.1
50
0.1181
0.1839
0.00080
0.0026
0.0101
9
4.6
12.74
-2.00
0.520
1.1
50
0.2012
0.3208
0.00140
0.0026
0.0259
10
4.5
12.36
-2.09
0.480
1.1
50
0.1946
0.3099
0.00140
0.0026
0.0243
Dp =
1.3 mm
1
1.0
0.84
-36.82
0.002
6.0
800
0.0061
0.0077
0.00001
0.0077
0.0082
2
0.9
0.65
-44.41
0.001
6.0
800
0.0042
0.0052
0.00001
0.0077
0.0080
3
1.7
2.61
-14.06
0.010
6.0
800
0.0304
0.0457
0.00010
0.0077
0.0140
4
1.6
2.32
-15.72
0.010
6.0
800
0.0259
0.0382
0.00010
0.0077
0.0127
5
2.6
5.51
-6.32
0.060
6.0
800
0.0785
0.1331
0.00030
0.0077
0.0361
6
2.5
5.17
-6.81
0.050
6.0
800
0.0727
0.1222
0.00030
0.0077
0.0327
7
3.3
7.93
-3.96
0.150
6.0
800
0.1207
0.2145
0.00060
0.0077
0.0655
8
3.3
7.93
-3.96
0.150
6.0
800
0.1207
0.2145
0.00060
0.0077
0.0655
9
4.6
12.74
-2.00
0.520
6.0
800
0.2050
0.3795
0.00100
0.0077
0.1404
10
4.5
12.36
-2.09
0.480
6.0
800
0.1983
0.33664
0.00100
0.0077
0.1314
8-13
-------
Figure 8-3. Ratio of the Measured / Calculated Concentration of Trace Metals
Associated with Fine Particles Obtained from Backward Mixed-layer
Trajectories for South Haven on 19, 20 and 21 July 1991.
A1 Si S K Ca Cr Mn Fe Ni Cu Zn As Se Br Hg Pb
Trace Elements
8-14
-------
Figure 8-4. Ratio of the Measured Calculated Concentration of Trace Metals
Associated with Coarse Particles obtained from Backward Mixed-layer
Trajectories for South Haven on 19, 20 and 21 July 1991.
100*
^ »;-*p ¦**- »::c"
A1 Si S K Ca Cr Mn Fe Cu Zn As Se Br Cd Hg Pb
Trace Elements
8-15
-------
Figure 8-5. Backward Mixed-layer Trajectories for South Haven that
Traversed Lake Michigan on the 19- 21 July, 1991.
(IIT = Chicago, SHA = South Haven. Small squares indicate the position of the air
mass as it moves backward from the site in 3-hour increments)
8-16
-------
Holsen et al. (1993) estimated that fine particles are responsible for about 2% and
0.6% of the total dry deposition flux of Pb and Ca, respectively, at the Chicago site. The
Pb and Ca dry deposition flux associated with coarse particles with aerodynamic diameters
in the range of 2.5 to 10 (im accounted for about 11.5 and 8% of total fluxes of Pb and
Ca, respectively. Based on these estimates, the dry flux deposited to Lake Michigan by
the fine and coarse fractions of the PM10 were calculated for this study (Table 8-3) and
also by Holsen et al. (1993). The higher ratios for the coarse fraction may be due to (a)
particle growth in a humidity gradient (not considered in this study) yielding submicron
particles in the inertial deposition range (and therefore higher dry deposition fluxes); (b)
position of the sampling station in the plume; (c) statistical uncertainty related to dry
deposition measurements; (d) small representativeness of measured fluxes at one point for
the whole lake area; (e) a different dry deposition flux distribution for particle sizes
measured in Chicago (used in this comparison) versus those found directly on Lake
Michigan.
Table 8-3. Comparison of Calculated Dry Deposition Flux (|j.g/m2-h) with that
Determined by Holsen et al. (1993).
PM10 Fraction PM10
Fine
Fine
Coarse
Coarse
Pb
Ca
Pb
Ca
(1) Holsen et al. (10)
0.0019
0.294
0.010
4.6
(2) This Study
0.0020
0.216
0.022
5.5
Ratio (2) / (1)
1.00
0.74
2.20
1.2
The previous results obtained for ambient concentrations and dry deposition fluxes
were used to define the optimal values for Km and Kbs. Both of these cr efficients in the
work by Williams (1982) work ranged from 0 to oc cm/s for Km, and 0 to 1000 cm/s for
K^s- Figure 8-6 displays the Km and curves obtained by cubic spline interpolations of
estimated values for aerodynamic particle diameters in the range of 0.5 to 6 ^m.
Obviously, the previous results are not thought to represent the real physical situation that
occurs at the air-water interface, since they were validated with measurements made using
surrogate surfaces not directly on the water surface. However, the high uncertainty in the
direct measurements of dry deposition fluxes to water surface using gradient and eddy flux
techniques (Wesely and Williams, 1981; Sievering, 1982) would not provide results any
more reliably.
8-17
-------
Figure 8-6. Curves of Km and K^s Coefficients vs. Particle Aerodynamic
Diameters.
8-18
-------
Dry deposition fluxes over Lake Michigan are estimated for trajectories crossing
Lake Michigan, Chicago, and South Haven for trace metals associated with the fine and
coarse PM10 particles. Figure 8-7 shows the average dry deposition loadings for trace
metals in the fine fraction obtained from trajectories crossing South Haven and Lake
Michigan. The dry deposition loads for trace metals primarily of crustal origin range from
26 tons for Si to 9.5 tons for K. Trace metals primarily of anthropogenic origin have a dry
deposition load in the range of 18.86 tons for S to 0.04 tons for Br. In the coarse fraction
(Figure 8-8), the dry deposition load for crustal or soil elements ranges from 333 tons for
Si to 48.8 tons for K, while elements of anthropogenic origin are in the range of 146.5
tons for A1 to 0.31 tons for Br.
To evaluate the variability of the dry deposition fluxes over a 24-hour period, the
deposition fluxes obtained from the daytime (18Z) backward trajectories were compared
to nighttime (beginning at 6Z) backward trajectories. Ambient concentrations measured
for the 12-hour nighttime sample and for the 12-hour daytime period were used to
evaluate the dry deposition fluxes of trace elements. Figure 8-9 shows the ratio of the
trajectory dry deposition fluxes calculated for the night/day (6Z/18Z) for trace metals
associated with fine and coarse particles for the South Haven site. The variability of the
dry deposition fluxes associated with fine particles ranged from 0.6 to 1.0 for most
elements with the exception of Se 1.4 and A1 1.65. The coarse fraction showed a dry
deposition flux ratio for most trace metals in the range of 0.6 to 1.4 with the exception of
A1 (2.2), Br (0.3) and Pb (0.3).
Figures 8-10 and 8-11 show the dry deposition loading of trace metals associated
with fine and coarse particles over Lake Michigan obtained using the backward
trajectories starting in Chicago. Crustal elements are slightly more abundant than those
estimated from South Haven trajectories with the exception of K which gave lower v .iues
using the Chicago trajectories. Anthropogenic elements are again more abundant than that
estimated from South Haven trajectories with the exception of A1 (2.04 tons), Cd (0.08
tons), As (0.06 tons), Se (0.01 tons) and Cr (0.1 tons). Deposition loads associated with
coarse particles (Figure 8-12) are two to three times higher than that obtained from the
South Haven trajectories for both crustal and anthropogenic elements, with the exception
of A1 and S which were similar at both sites.
8-19
-------
Figure 8-7. Dry Deposition Load of Trace Metals Associated with Fine Particles
obtained from South Haven Measurements and Trajectories.
8-20
-------
Figure 8-8. Dry Deposition Load of Trace Metals Associated with Coarse Particles
obtained from South Haven Measurements and Trajectories.
1000.00^
A1 Si S K Ca Cr Mn Fe Ni Cu Zn As Se Br Cd Pb
Trace Elements
8-21
-------
Figure 8-9. Night to Day Dry Deposition Flux Ratio Calculated from 12-hour
Ambient Measurements at South Haven. Nighttime Samples Collected
8PM - 8PM and Daytime Samples Collected 8AM-8PM.
Trace Elements
8-22
-------
Figure 8-10. Dry Deposition Load of Trace Metals Associated with Fine Particles
obtained from Chicago Measurements and Trajectories.
100.00
10.00
1.00^
0.10=
0.01
0.00
A1 Si S K Ca Cr Mn Fe Ni Cu Zn As Se Br Cd Hg Pb
Trace Elements
8-23
-------
Figure 8-11. Dry Deposition Load of Trace Metals Associated with Coarse Particles
obtained from Chicago Measurements and Trajectories.
Trace Elements
8-24
-------
Figure 8-12. Chicago To South Haven Dry Deposition Flux Ratio.
*
3
e
OS
=
3
e
c/o
e
on
ee
u
100.00
10.00
1.00
0.10:
0.01
A1 Si S K Ca Cr Mn Fe Ni Cu Zn As Se Br Cd Hg Pb
Trace Elements
8-25
-------
In order to evaluate the dependence of the type of sampling site (i.e., rural,
industrial) and location in the basin on the dry deposition estimates, the ratios between
Chicago and South Haven dry deposition loads for trace metals associated with fine and
coarse particles were calculated. The ratios for most of the fine fraction trace metals
(Figure 8-12) were in the range of 0.6 to 3 with the exception of Cr 0.1 and total Hg 16.
The high ratio for total Hg was probably due to strong local sources in the Chicago area,
as ambient Hg concentrations measured in Chicago were about 5-15 times higher than
those measured for vapor and particulate Hg in South Haven.
Table 8-4 gives the dry deposition load estimated to Lake Michigan for the
LMUATS period. The estimated dry deposition flux to Lake Michigan was computed by
Holsen et al. (1993) using the average elemental flux measured over the water and
assuming that this measured flux was typical for the southern one-half of the lake. The
average flux over the northern one-half of the lake was 25% of this amount. The
difference in the dry flux between the northern and southern portions of Lake Michigan is
roughly equivalent to the difference in total fluxes measured by Eisenreich (1980) with
bulk deposition collectors from September 1975 to December 1976. The value reported
for this study is the average of the dry deposition flux calculated from the data at South
Haven and Chicago (IIT). The spatial averaging is implicitly included in the hybrid-
receptor modeling approach described in this chapter. Surprisingly, the results from the
two approaches still agree within a factor of 2-4 for the two elements, Pb and Ca, given
below. The estimates based on the hybrid-receptor modeling approach can be applied to
any pollutant species and may be improved using additional sampling sites on both sides of
the lake.
Table 8-4. Comparison of Calculated Dry Deposition Loads (kg) with that
Estimated by Holsen et al (1993).
PM10 Fraction PM10
Fine
Pb
Fine
Ca
Coarse
Pb
Coarse
Ca
Total
Pb
Total
Ca
(1) Holsen et al. (1993)
49.6
7673
261
120,060
311
112,073
(2) This Study
83.5
9020
919
229,680
1002
238,700
Ratio (2)/(l)
1.7
1.2
3.5
1.9
3.2
2.1
8-26
-------
8.4.
Discussion
Modeling the dry deposition flux may best be performed by invoking a strictly
lagrangian framework since the concentration profile, aerosol size distribution,
atmospheric chemistry, and meteorological conditions characterizing an air mass change as
a function of transport time. Dry deposition fluxes over Lake Michigan may vary by
several orders of magnitude due to variations in the aerodynamic particle diameter and
meteorological conditions en route. Deposition velocities for submicron particles may
vary by a factor of 3 to 10, while particles slightly larger than 1 |j.m are found to have
variations in deposition velocities of a factor- of 15. Particles with an aerodynamic
diameter greater than 4|j.m can have variations of 3 to 20% with respect to the mean. The
large range in deposition velocities suggests that dry deposition fluxes measured at one
point near the lake will have a large uncertainty.
Fluxes measured at one sampling point do not account for the effect of variations
of meteorological parameters, particle size and ambient concentrations which occur when
the air mass crosses the lake. Further, the deposition velocity of particles over a water
surface is affected by the dynamics of water wave breaking and spray formation which are
included in estimates reported here through the coefficients Km and K^. Based on the
comparison of calculated with measured dry deposition fluxes and ambient concentrations,
optimal values of Km and Kbs are derived for particles in the size range of 0.5 to 6 |i.m.
One must be careful, however, not to over-interpret the ratio of the calculated/measured
dry deposition flux in the light of the assumptions made in calculating the dry deposition
flux in this study. Furthermore, this caution should also be applied when considering that
the measured fluxes are from one point on the lake and can not provide statistically
meaningful results for the large surface area of Lake Michigan.
The results presented in this work were also compared to those obtained using the
Sehmel and Hodgson (1978) model. The total flux ratio (S and H model / this study)
obtained for most of the trace metals was in the range of 0.85 to 1.72 with a mean ratio of
1.35. The good agreement suggested by these ratios is surprising considering that the
Sehmel and Hodgson (1978) formulation was not originally intended for a water surface,
as it does not take into account the effect of wave breaking or spray formation on the
transfer of particles from the atmosphere to the water surface and vice versa. On the other
hand, in optimizing the Km and Kbs coefficients, the role played by diffusophoretic and
8-27
-------
thermophoretic forces on the downward and upward motion of particles was not
considered, since direct measurement data are not available to validate the result.
The range in the dry deposition velocity caused by particle size and meteorological
conditions and, hence, its influence on the dry deposition flux becomes smaller for
particles in the inertial deposition range. Therefore, to account for the effect of all of the
previously defined parameters, average values of the dry deposition fluxes calculated along
each trajectory crossing Lake Michigan are presented. Figure 8-13 shows the distribution
of dry deposition fluxes for several trace metals in the fine and coarse fraction for
trajectories crossing Lake Michigan during LMUATS. The dry deposition flux is plotted
on a log scale to allow the large variations in flux from element to element to be displayed
on the same figure. The average dry deposition fluxes are calculated for each backward
and forward trajectory crossing the lake. Table 8-5 gives the average dry deposition flux
to Lake Michigan calculated for both fine and coarse particles during the LMUATS
period. The dry deposition flux for trace metals typically of crustal or soil origin in the
fine fraction ranges from 0.214 ng/m2-h for K to 0.58 |j.g/m2-h for Si, while in the coarse
fraction the dry deposition flux ranges from 1.1 |ig/m2-h for K to 8.5 |ig/m2-h for Si.
Trace elements primarily of anthropogenic origin associated with fine particles have a dry
deposition flux in the range of 0.001 |ag/m2-h for Br and Se to 0.425 n,g/m2-h for S; in the
coarse fraction the dry deposition fluxes range from 0.007 |ig/m2~h for Br to 3.3 |ig/m2-h
for Al.
8-28
-------
Figure 8-13. Dry Deposition Flux for Selected Crustal and Anthropogenic Elements
for Backward Trajectories that Traversed Lake Michigan.
l
o.i
0.01
0.001
9 10 11 12 13 14 IS 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9
O
¦B
N
E
a
9 10 11 12 13 14 15 16 17 18 192021 22 23 24 2'5 26 27 28 29 30 31 123456789
M
S
E
a
o
JZ
o
a.
u
Q
t
C
1000
100-
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9
0.01
0.001
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9
July August
8-29
-------
Table 8-5. Average Dry Deposition Fluxes (ng/m2-h) of Trace Metals to Lake
Michigan.
Element
Fine
(< 2.5 |im)
Coarse
(2.5-10 fim)
A1
0.086
3.3
Si
0.581
7.5
S
0.425
0.82
K
0.214
1.1
Ca
0.216
5.5
Cr
0.002
0.01
MN
0.003
0.05
Fe
0.288
3.3
Ni
0.002
0.018
Cu
0.004
0.045
Zen
0.013
0.075
As
0.002
0.010
Se
0.001
0.020
Br
0.001
0.007
Cd
0.003
0.020
Total Hg
0.001
Pb
0.002
0.022
8-30
-------
References
Arey, J., Atkinson, R., Zielinska, B., McElroy, P.A. Diurnal Concentrations of Volatile
Polycyclic Aromatic Hydrocarbons and Nitroarenes During a Photochemical Air
Pollution Episode in Glendora, California. Environ. Sci. Technol. 23: 321-327,
1989.
Arya, S.P. Introduction to Micrometeorology, Academic Press, San Diego, California,
1988. pp. 141-156.
Atkinson, R. Gas Phase Tropospheric Chemistry of Organic Compounds: A review.
Atmos. Environ. 24A: 1-41, 1990.
Baker, J.E., referenced in Eisenreich, S.J. and Strachan, W.M.J. Estimating Atmospheric
Deposition of Toxic Substances to the Great Lakes, In: Proceedings of Workshop
held at the Canada Centre for Inland Waters, Burlington, Ontario, January 31 -
February 2, 1992. p.48.
Baker, J.E., Church. T.M., Ondov. J.M., Scudlark, J.R., Conko, K.M., Leiseter, D.L. and
Wu, Z.Y. Chesapeake Bay Atmospheric Deposition Study, Phase I: July 1990-
June 1991. Report to the Department of Natural Resources, Annapolis, Maryland,
1992.
Baker. J.E. and Eisenreich, S.J. Concentrations and Fluxes of Polycyclic Aromatic
Hydrocarbons and Polvchlorinated Biphenyls Across the Air-Water Interface of
Lake Superior. Environ. Sci. Technol. 24:342-352, 1990.
Baker, D.B. and Richards, R.P. Transport of Soluble Pesticides Through Drainage
Networks in Large Agricultural River Basins. In: Kurtz D.A.(ed.), Long Range
Transport of Pesticides Lewis Publishers, Chelsea, Michigan, 1990. p. 241.
Ballschmiter, K. Global Distribution of Organic Compounds. Environ. Carcino. Ecotox.
Rev. C9:1-46, 1991.
Barrie, L.A., Gregor, D.J., Hargrave, B.T., Lake, R., Muir, D.C.G., Shearer, R.. Tracey,
B., Bidleman. T.F. Arctic Contaminants: Sources, Occurrence, and Pathways.
Science of the Total Environment. 122:1-74, 1992.
9-1
-------
Bidleman, T.F. Eisenreich, S.J. and Strachan, W.M.J. Estimating Atmospheric Deposition
of Toxic Substances to the Great Lakes, Workshop held at the Canada Centre for
Inland Waters, Burlington, Ontario, January 31 - February 2. 1992. p. 37.
Bidleman, T.F., Patton, G.W., Hinckley, D.A., Walla, M.D., Cotham, W.E. and Hargrave,
B.T. Chlorinated Pesticides and Polychlorinated Biphenyls in the Atmosphere of
the Canadian Artie. In: Kurtz D.A. (ed.), Long Range Transport of Pesticides.
Lewis Publishers, Chelsea, MI., 1990. p. 347.
Bigg, E.K., Ayers, G.P. and Turvey, D.E. Measurement of the Dispersion of a Smoke
Plume at Large Distances From the Source. Atmos. Environ. 12: 1815, 1978.
Biswas, P., Jones, C.L., and Flagan, R.C. Distortion of Size Distributions by Condensation
and Evaporation in Aerosol Sampling Instruments. J. Aerosol Sci. Tech. 7:231-
247, 1987.
Bloom, N. and Fitzgerald, W.F. Determination of Volatile Mercury Species at the
Picogram Level by Low-Temperature Gas Chromatography With Cold-Vapour
Atomic Fluorescence Detection. Analytica Chimica Acta. 208:151-161, 1988.
Boutron, C.F. A Clean Laboratory for Ultralow Concentration Heavy Metal Analysis.
Fresenius J. Anal. Chem. 337: 482-491, 1990.
Braham, R.R., Seely, B.K. and Crozier, W.D. A Technique for Tagging and Tracing Air
Parcels. Trans. Amer. Geophys. Union. 33:825, 1952.
Bunce. N.J., Landers, J.P., Langshaw, J-A., Nakai, J.S. An Assessment of the Importance
of Direct Solar Degradation of Some Simple Chlorinated Benzenes and Biphenyls
in the Vapor Phase. Environ. Sci. Technol. 23:213-218. 1989.
Bunce. N.J., Nakai, J.S. Atmospheric Chemistry of Chlorinated Phenols. ./. Air Pollut.
Com. Assoc. 39:820-823, 1989.
Businger, J.A., Wyngaard. J.C., Izumi. Y. and Bradley, E.F. Flux-Pofile Relationships in
the Amospheric Surface Layer. J. Atmos. Sci. 28:181-189, 1971.
Chamberlain, A.C. Deposition and Resuspension, In: Pruppacher, H.R., Semonin, R.G.
and Slinn, G.N.(eds.). Precipitation Scavenging, Dry Deposition, and
Resuspension. Elsevier Science Publishing Co.. Inc., New York, New York,
1983. p.731.
Clean Air Act (CAA). 1990. 42 U.S.C. 7401-7642 and the rules and amendments
promulgated thereunder.
Cleckner. L.B. Personal Communication, January 1993.
9-2
-------
Conner, T.L., Miller. J.L. and Willis, R.D. Source Apportionment of Fine and Coarse
Particles in Southern Ontario, Canada. In: Proceedings of the 86t'1 Annual
Meeting of the A&WMA, paper 93-TP-58.05, Denver, Colorado, 1993.
Countess, R.J., Wolff, G.T. and Cadle, SH. The Denver Winter Aerosol: A
Comprehensive Chemical Characterization. JAPCA. 30:1194, 1980.
Crabtree, J. Studies of Plume Transport and Dispersion over Distances of Travel up to
Several Hundred Kilometers, In: Proceedings 13th NATO/CCMS Conference, on
Air Pollution Modelling and its Applications, Plenum Press, New York, New York,
1982.
Crawford, T.L. Personal Communication, March 1992 and December 1992.
Crozier, W.D. and Seely, B.K. Concentration Distribution in Aerosol Plumes Three to
Twenty-Two Miles from a Point Source. Trans. Amer. Geophys. Union. 36:42,
1955.
Daisey, J.M. Cheney, J.L. and Lioy, P.J. Profiles of Organic Particulate Emissions from
Air Pollution Sources: Status and Needs for Receptor Source Apportionment
Modeling. J. Air Pollut. Control Assn. 36:17-33
Dodd, J.A., Ondov, J.M., Tuncel, G., Dzubay, T.G., Stevens, R.K. Multimodal Size
Spectra of Submicrometer Particles Bearing Various Elements in Rural Air.
Environ. Sci. Technol. 25:890-903, 1991.
Draxler. R.R. Sensitivity of a Trajectory Model to the Spatial and Temporal Resolution of
the Meteorological Data During CAPTEX. J. Clim. Appl. Met. 26:1557-1588.,
1987.
Dzubay. T.G. and Mamane. Y. Use of Electron Microscopy Data in Receptor Models for
PM-10. Atmos Environ. 23:467-476. 1989.
Dzubay, T.G., Stevens. R.K. and Gordon, G.E. A Composite Receptor Method Applied
to Philadelphia Aerosol. Environ. Sci. Technol. 22(l):46-52, 1987.
Dzubay, T.G. and Mamane. Y. Use of Electron Microscopy Data in Receptor Models for
PM-10. Atmos. Environ. 23:467-476, 1989.
Dzubay, T.G., Stevens, R.K., Lewis. C.W. et al. Visibility and Aerosol Composition in
Houston, Texas. Environ. Sci. Technol. 16:514-524, 1982.
Eisenreich, S.J. and Strachan. W.M.J. Estimating Atmospheric Deposition of Toxic
Substances to the Great Lakes. In: Procedings of Workshop held at the Canada
Centre for Inland Waters. Burlington. Ontario. January 31 - February 2, 1992.
9-3
-------
Fairfall. C.W. and Larsen, S.E. Inertial Dissipation Methods and Turbulent Fluxes at the
Air-Ocean Interface. Boundary Layer Meteorology. 34:287-302, 1986.
Fergusson, J.E. The Heavy Elements: Chemistry, Environmental Impact and Health
Effects. Pergamon Press, New York, 1990.
Fitzgerald, W. F., Vandal, G. M. and Mason, R. P. Atmospheric Cycling and Air-Water
Exchange of Mercury over Mid-Continental Lacustrine Regions. WASP. 56:745-
768, 1991.
Fitzgerald, W.F., Gill, G.A. and Kim, J.P. An Equatorial Pacific Ocean Source of
Atmospheric Mercury. Science. 224:597-599, 1984.
Glass, G.E., Sorensen, J.A., Schmidt, K.W. and Rapp, G.R. New Source Identification of
Mercury Contamination in the Great Lakes. Environ. Sci. Technol. 24:1059-1069,
1990.
Golder, D. Relations Among Stability Parameters in the Surface Layer. Boundary Layer
Meteorology. 3:47-58, 1972.
Gordon, G.E. Receptor Models. Environ. Sci. Technol. 14:792-800.1980.
Gordon, G.E. Receptor Models. Environ. Sci. Technol. 22:1132-1142, 1988.
Grimmer. G. Analysis of Automobile Exhaust Condensates. IARC scientific publication
no. 16, 1977. pp. 29-39.
Harrison. P.R.. Rahn. K.A. Dams. R., Robbins, J.A., Winchester. J.W.. Brar, S.S., Nelson,
D.M. Areawide Trace Metal Concentrations Measured by Multielement Neutron
Activation Analysis. JAPCA. 21:563-570, 1971.
Heffter. J.L. Atmospheric Transport and Dispersion Model, Air Resources Laboratories,
NOAA Technical Memorandum ERL ARL-81, Silver Spring, Maryland, 1980.
Henry, R.C. Mult'" ariate Receptor Models. In: Hopke. P.K. (ed.) Receptor Modeling
for Air Quality Management. New York, New York. 1991. pp. 117-147.
Hermanson, M., referenced in Eisenreich, S.J. and Strachan. W.M.J. Estimating
Atmospheric Deposition of Toxic Substances to the Great Lakes. In: Proceedings
of Workshop held at the Canada Centre for Inland Waters, Burlington, Ontario,
January 31 - February 2. 1992.
Hess, G.D. and Hicks, B.B. The Influence of Surface Effects on Pollutant Deposition
Rates over the Great Lakes. In: Proceedings on the Second Federal Conference on
the Great Lakes, 1975. pp.238-247.
9-4
-------
Hites, J.E., referenced in Eisenreich, S.J. and Strachan, W.M.J. Estimating Atmospheric
Deposition of Toxic Substances to the Great Lakes. In: Proceedings of Workshop
held at the Canada Centre for Inland Waters, Burlington. Ontario, January 31 -
February 2, 1992. p.48.
Hoff, R.M., referenced in Eisenreich, S.J. and Strachan, W.M.J. Estimating Atmospheric
Deposition of Toxic Substances to the Great Lakes, Workshop held at the Canada
Centre for Inland Waters, Burlington, Ontario, January 31 - February 2, 1992.
p.37.
Hoff, R.M., Muir, D.C.G. and Grift, N.P. Annual Cycle of Polychlorinated Biphenyls and
Organohalogen Pesticides in Air in Southern Ontario. I. Air Concentration Data.
Environ. Sci. Technol. 26:266-275, 1992.
Holsen, T.M., Noll, K.E., Liu, S. and Lee, W. Dry Deposition of Polychlorinated
Biphenyls in Urban Areas. Environ. Sci. Technol. 25:1075-1081, 1991.
Holsen, T.M., Noll, K.E., Fang, G.C., Lee, W.J., Lin, J.M. and Keeler, G.J. Dry
Deposition and Particle Size Distributions Measured During the Lake Michigan
Urban Air Toxics Study. Environ. Sci. Technol.21:\321-\333, 1993.
Hopke, P.K. An Introduction to Receptor Modeling. In: Hopke. P.K. (Ed.), Receptor
Modeling for Air Quality• Management., Elsevier, New York., 1991. pp. 1-10.
Hornbuckle, K.C. Eisenreich, S.J. and Strachan, W.M.J. (1992) Estimating Atmospheric
Deposition of Toxic Substances to the Great Lakes, Workshop held at the Canada
Centre for Inland Waters, Burlington, Ontario, January 31 - February 2, 1992.
p.37.
Hornbuckle, K.C. and Eisenreich. S.J. referenced in Eisenreich. S.J. and Strachan. W.M.J.
Estimating Atmospheric Deposition of Toxic Substances to the Great Lakes,
Workshop held at the Canada Centre for Inland Waters. Burlington, Ontario,
January 31 - February 2. 1992. p.40.
Hornbuckle, K.C., Achman, D.R. and Eisenreich, S.J. Over-Water and Over-Land
Concentrations of Polychlorinated Biphenyls in Green Bay, Lake Michigan.
Environ. Sci Technol. 27:87-98, 1992.
Kahl, J.D. and Samson, P.J. Uncertainty in Trajectory Calculations Due to Low-resolution
Meteorological Data. J. Clim. Appl. Met. 25:1816-1831, 1986.
Keeler, G.J. et al. Unpublished data from the Michigan Protection Fund: Michigan
Toxics Network, 1993.
Keeler, G.J., Pacyna, J.M., Bidleman, T.F. and Nriagu. J.O. Identification of Sources
Contributing to the Contamination of the Great Waters by Toxic Compounds. A
Report to the U.S. EPA Office of Air Quality Planning and Standards. 1993.
9-5
-------
Keeler, G.J., Spengler, J.D. and Castillo, R. Acid Aerosol Measurement at a Suburban
Connecticut site. Atmos. Environ. 24A:2915-2923, 1991.
Keeler, G.J., Spengler, J.D., Koutrakis, P., Allen, G.A., Raizenne, M. and Stern, B.
Transported Acid Aerosols Measured in Southern Ontario. Atmos. Environ.
24A:2935-2950, 1990a.
Keeler, G.J. Japar, S.M., Brachaczek, W.W., Gorse, R.A., Jr. and Pierson, W.R. The
Sources of Aerosol Elemental Carbon at Allegheny Mountain. Atmos. Environ.
24A:2795-2805, 1990b.
Keeler, G.J. and Samson P.J. On the Spatial Representativeness of Trace Element Ratios.
Environ. Sci. Technol. 23:1358-1364, 1987.
Keeler, G.J. A Hybrid Approach for Source Apportionment of Atmospheric Pollutants in
the Northeastern United States. PhD Dissertation, Univ. of Michigan, 1987.
Kelly, T.J., Chuang, J.C., Callahan, P.J., Lewis, R.G. and Pleil, J. Field Measurements of
Atmospheric Polynuclear Aromatic Hydrocarbon Concentrations and Phase
Distribution at TAMS Sites. In: Measurement of Toxic and Related Air
Pollutants, Proceeding of the 1992 U.S. EPA/A&WMA International Symposium,
1992. 31-36 pp.
Kelly, T.J., Czuczwa, J.M., Sticksel, P.R. and Sverdrup, G.M. (1991) Atmospheric and
Tributary Inputs of Toxic Substances to Lake Erie. J. Great Lakes Res. 17:504-
516.
Koutrakis, P., Wolfson, J.M., Slater, J.L., Brauer, M., Spengler, J.D.. Stevens, R.K.,
Stone, C.L. Evaluation of an Annular Denuder/Filter Pack System to Collect
Acidic Aerosols and Gases. ES&T 22:1463-1468, 1988.
Lane, D.A., Hoff, R.M. and Bidleman, T.F. referenced in Eisenreich, S.J. and Strachan,
W.M.J. Estimating Atmospheric Deposition of Toxic Substances to the Great
Lakes. Workshop held at the Canada Centre for Inland Waters. Burlington,
Ontario, January 31 - FeKuary 2, 1992. p.40.
Lee, H.S., Wadden, R.A. and Scheff, P.A. Measurement and Evaluation of Acid Air
Pollutants in Chicago Using an Annular Denuder System. Atmos. Environ.
27A:543-553, 1993.
Li. C. K. and Kamens, R.M. The Use of Polycyclic Aromatic Hydrocarbons as Source
Signatures in Receptor Modeling. Atmos. Environ. 27A:523-532, 1993.
Li, S.-M. and Winchester, J.W. Haze and Other Aerosol Components in Later Winter
Arctic Alaska. 1986'. J. Geophys. Res. 95(D2):1797 - 810, 1990.
9-6
-------
Mamane,Y., Stevens, R.K., and Dzubay, T.G. Levels and Sources of Fine Mass and
Carbon in the Great Lakes Basin. Atmos. Environ. In Press, 1993.
Mamane, Y. Estimate of Municipal Refuse Incinerator Contribution to Philadelphia
Aerosol Using Single Particle Analysis - II. Ambient Measurements. Atmos.
Environ. 24B: 127-135, 1990.
Mamane, Y. and Dzubay, T. Characterstics of Individual Particles at a Rual Site in the
Eastern United States. JAPCA 36:906-911, 1987.
Mamane, Y., Miller, J.L. and Dzubay, T.G. Characterization of Individual Fly Ash
Particles Emitted from Coal- and Oil- Fired Power Plants. Atmos. Environ.
21:2125-2135, 1986.
Manchester-Neesvig, J.B. and Andren, A.W. Seasonal Variation in the Atmophseric
Concentration of Polychlorinated Biphenyl Congeners. Environ. Sci. Technol.
23:1138-1148, 1989.
Marple. V.A. and Rubow, K.L. Development of a Micro-orifice Uniform Deposit
Cascade Impactor, Final Report, DOE Contract DE-FG22-83PC61255,
Pittsburgh Energy Tech. Cent., Pittsburgh, PA, 1984.
Mason, B. Principles of Geochemistry, 3rd Edition, John Wiley and Sons, Inc., New
York, New York, 1966.
McBean, G.A., Ed. The Planetary Boundary Layer. Tech. Note No. 165, WMO 530,
1979, pp 201.
McConnell, L.L. Eisenreich, S.J. and Strachan, W.M.J. Estimating Atmospheric
Deposition of Toxic Substances to the Great Lakes, Workshop held at the Canada
Centre for Inland Waters, Burlington, Ontario. January 31 - February 2, 1992.
p.37.
McConnell, L.L., Cotham, W.E. and Bidleman. T.F. Gas Exchange of
Hexachlorocyclohexanes in the Great Lakes. Environ. Sci. Technol., 27:1304-
1311,1993.
Mildford, J.B. and Davidson C.I. The Sizes of Particulate Trace Elements in the
Atmosphere-A Review. JAPCA 35:1249-1260, 1985.
Miller, S.S., Friedlander, S.K. and Hidy, G.M. A Chemical Element Balance for the
Pasadena Aerosol. J. Colloid Interface Sci. 39:165-178, 1972.
Murphy, T.J. referenced in Eisenreich. S.J. and Strachan. W.M.J. Estimating Atmospheric
Deposition of Toxic Substances to the Great Lakes. Workshop held at the Canada
Centre for Inland Waters, Burlington, Ontario. January 31 - February 2, 1992.
p.37.
9-7
-------
Murphy, T.J., Formanski, L.J., Brownawell, B. and Meyer. J.A. Polychlorinated Biphenyl
Emissions to the Atmosphere in the Great Lakes Region. Municipal Landfills and
Incinerators. Environ. Sci. Technol. 19:942-946, 1985.
Nicholson, K.W. The Dry Deposition of Small Particles: A Review of Experimental
Measurements. Atmos. Environ. 22:2653-2666, 1988.
Noll, K.E., Fang, Y.P. and Watkins, L.A. Characterization of the Deposition of Particles
from the Atmosphere to a Flat Plate. Atmos. Environ. 22:1461-1468, 1988.
Noll, K.E., Pontius, A., Frey, R. and Gould, M. Comparison of Atmospheric Coarse
Particles at an Urban and Non-Urban Site. Atmos. Environ.19:1931-1943, 1985.
Olmez, I. Personal Communication, 1992.
Olmez, I., Sheffield, A.E. and Gordon, G.E. Compositions of Particles from Selected
Sources in Philadelphia for Receptor Modeling Applications. JAPCA 38:1392-
1402, 1988.
Pasquill, F. Meteorological Research at Porton. Nature 177:1148, 1956.
Pasquill, F. and Smith, F.B. Atmospheric Diffusion, John Wiley & Sons, New York,
1983.
Phelps, G.T. and Pond, S. Spectra of the Temperature and Humidity Fluctuations and the
Fluxes of Moisture and Sensible Heat in the Marine Boundary Layer. J. Atmos.
Sci. 28:918-929, 1971.
Pierson. W.R.. Brachaczek. W.W., Gorse, R.A., Japar, S.M. and Norbeck. J.M.
Atmospheric Acidity Measurements on Allegheny Mountain and the Origins of
Ambient Acidity in the Northeastern United States. Atmos. Environ. 23:431-459,
1989.
Pond, S., Phelps, G.T. and Paquin, J.E. Measurements of the Turbulent Fluxes of
Momentum, Moisture and Sensible F^at Over the Ocean. J. Atmos. Sc/.29:901-
917, 1971.
Rahn, K.A. and Lowenthal, D.H. Pollution Aerosol in the Northeast: Northeastern-
Midwestern Contributions. Science 228:275-284, 1985.
Rahn, K.A. and Lowenthal, D.H. Elemental Tracers of Distant Regional Pollution
Aerosols. Science. 23:132-139, 1985.
Rapaport, R.A., Urban, N.R., Capel. P.D., Baker, J.E.. Looney, B.B.. Eisenreich. S.J. and
Gorham. E. "New" DDT Inputs to North America-Atmospheric Deposition.
Chemosphere 14:1167-1173, 1985.
9-8
-------
Reid, N. Ontario Ministry of the Environment, Great Lakes Toxics Deposition Network,
unpublished data, referenced in Eisenreich, S.J. and Strachan, W.M.J. Estimating
Atmospheric Deposition of Toxic Substances to the Great Lakes, Workshop held
at the Canada Centre for Inland Waters, Burlington, Ontario, January 31 -
February 2, 1992. p.40.
Richardson, L.F. and Proctor, D. Diffusion Over Distances Ranging from 3km to 86 km.
Memoirs of the R. Met. Soc. 1(1), 1925.
Roscoe, B.A., Hopke, P.K., Dattner, S.L. and Jenks, J.M. The Use of Principal
Component Factor Analysis to Interpret Particulate Compositional Data Sets.
JAPCA 32:637-642, 1982.
Sadisivan, S. Trace Elements in Size Separated Aerosols Over Sea. Atmos. Environ.
12:1677, 1978.
Samson, P.J. Trajectory Analysis of Summertime Sulfate Concentrations in the
Northeastern United States. J. Appl. Met. 19:1382-1394, 1980.
Scheff, P.A. and Wadden, R.A. Receptor Modeling for Volatile Organic Compounds. In:
Receptor Modeling for Air Quality Management. Ed: Hopke, P.K. Elsevier, New
York. pp. 213-253, 1991.
Schroeder, W.H. Sampling and Analysis of Mercury and its Compounds in the
Atmosphere. Environ. Sci. Technol. 16: 394-400, 1982.
Sehmel, G.A. Particle and Gas Dry Deposition: A Review. Atmos. Environ. 14:983-
1011, 1980.
Sehmel, G.A. and Hodgson, W.A. Model for Predicting Dry Deposition of Particles and
Gases to Environmental Surfaces. PNL-SA-6721, Battelle Pacific Northwest
Laboratories, Richland, Washington, 1978, pp. 44.
Seinfeld, J.H. Atmospheric Chemistry and Physics of Air Pollution. John Wiley and Sons,
New York, 1986.
Sheridan, P.J. Characterization of Size Segregated Particles Collected over Alaska and
the Canadian High Arctic. AGASP-II flights 204-206. Atmos. Environ. 23:237-86,
1989.
Sievering, H. Profile Measurements of Particle Mass Transfer at the Air-Water Interface.
Atmos. Environ. 15:123-129, 1981.
Sievering, H. Profile Measurements of Particle Dry Deposition Velocity at an Air-Land
Interface. Atmos. Environ. 16:301-306. 1982.
9-9
-------
Slinn, W.G.N., Hasse, L., Hicks, B.B.. Hogan, A.W., Lai, D., Liss, P.S., Munnich, K.O.,
Sehmel, G.H. and Vittori, O. Some Aspects of the Transfer of Atmospheric Trace
Constituents Past the Air-Sea Interface. Atmos. Environ. 12:2055-2087, 1978.
Slinn, S.A. and Slinn, W.G.N. Prediction for Particle Deposition on Natural Waters.
Atmos. Environ. 14:1013-1016, 1980.
Slinn, W.G.N. A Potpourri of Deposition and Resuspension Question, in Precipitation
Scavenging, Dry Deposition, and Resuspension. Eds: Pruppacher, H.R.,
Semonin, R.G. and Slinn, G.N., Elsevier Science Publishing Co., Inc., New York,
1983.
Stevens, R.K., Dzubay, T.G. and Shaw G.R.W. Characterization of the Aerosol in the
Great Smoky Mountains. Environ. Sci. Technol. 14:1491-1498, 1980.
Stiver, W. and Mackay, D. Limitations of the Compartmental Approach to Modeling Air-
Soil Exchange of Pesticides. In: Long Range Transport of Pesticides Ed: Kurtz
D.A. Lewis Publishers, Chelsea, MI., 1990, p. 303.
Strachan, W.M.J, and Eisenreich. S.J. Estimating Atmospheric Deposition of Toxic
Substances to the Great Lakes - An Update. In: Proceedings of a Workshop held
at the Canada Centre for Inland Waters. January 31-February 2, 1992, Burlington,
Ontario.
Strachan, W.M.J, and Eisenreich, S.J. Mass Balancing of Toxic Chemicals in the Great
Lakes: The Role of Atmospheric Deposition. In: Proceedings of Workshop on the
Estimation of Atmospheric Loadings of Toxic Chemicals to the Great Lakes Basin,
October 29-31, 1986, Windsor, Ontario, 1988.
Sutton, O.G. A Theory of Eddy Diffusion in the Atmosphere. Proc.Roy.Soc. A, 146:701.
1932.
Swackhamer, D.L., McVetty, B.D. and Hites, R.A. Deposition and Evaporation of
Polychlorobiphenyl Congeners to and from Siskiwit Lake, Isle Royale, Lake
Superior Environ. Sci. Technol. 22:664-672, 1988.
Sweet, C.W., Vermette, S.J. and Landsberger, S. Sources of Toxic Trace Elements in
Urban Air in Illinois. Environ. Sci. Technol. 27:2502-2510, 1993.
Sweet, C.W. and Vermette, S.J. Toxic Volatile Organic Compounds in Urban Air in
Illinois. Environ. Sci. Technol. 26:165-173, 1992.
Sweet, C.W. referenced in Eisenreich. S.J. and Strachan, W.M.J. Estimating Atmospheric
Deposition of Toxic Substances to the Great Lakes. Workshop held at the Canada
Centre for Inland Waters, Burlington, Ontario, January 31 - February 2. 1992.
p.37.
9-10
-------
Tatsukawa, R., Yamaguchi, Y., Kawano, M., Kannan, N. and Tanabe, S. Global
Monitoring of Organochlorine Insecticides- An 11-Year Case Study (1975-1985)
of HCHs and DDTs in the Open Ocean Atmosphere and Hydrosphere. In: Long
Range Transport of Pesticides Ed: Kurtz D.A. Lewis Publishers, Chelsea, MI.,
1990, p. 127.
Tong, H.Y. and Karasek, F.W. Quantification of PAHs in Diesel Exhaust Particulate
Matter by HPLC Fractionation and High Resolution GC. Anal. Chem. 56:2129-
2134.1984.
U.S. Department of Commerce, National Oceanic and Atmospheric Administration.
Agricultural Pesticide Use in Coastal Areas: A National Summary. Pait, A.S.,
De Souza, A.E. and Farrow, D.R.G. (eds.), 1992.
U.S. EPA Volatile Organic Compound (VOC)/Particulate Matter (PM) Speciation Data
System, Version 1.4. EPA-450/4-91-027. U.S.EPA Research Triangle Park 1991.
U.S. EPA Federal Register. National Emissions Standards for Hazardous Air Pollutants,
Fed. Reg. 38, 8820-8845. US Government Printing Office, Washington D.C.,
1973.
van Ulden, A.P. and Holtslag, A.A.M. Estimation of Atmospheric Boundary Layer
Parameters for Diffusion Applications, J. of Climate and Appl. Meteor. 24:1196-
1207.1985.
Vermette, S.J. and Landsberger. S. Airborne Fine Particulate Matter (PMiq) in Southeast
Chicago. Illinois State Water Survey Contract Report 525, Champaign, Illinois,
1991.
Vermette, S.J. Williams, A.L. and Landsberger, S. Surface Dust Elemental Profiles -
Southeast Chicago. Illinois State Water Survey Contract Report 488, Champaign,
Illinois, 1990.
Vermette, S.J., Sweet. C.W. and Landsberger, S. Airborne Fine Particulate Matter (PM-
10) in Southeast Chicago: Preliminary Report II. Illinois Staie Water Survey
Contract Report 481, Champaign, IL. 1988.
Voldner. E.C. and Smith, L. Production. Usage and Atmospheric Emissions of 14 Priority
Toxic Chemicals. Appendix 2 to the Proceedings of the Workshop on Great
Lakes Atmospheric Deposition. International Air Quality Advisory Board of the
International Joint Commission. ISBN 1-895085-20-9, 1991.
Vossler. T.L. and Macias, E.S. Contribution of Fine Particle Sulfates to Light Scattering
in St. Louis Summer Aerosol. Environ. Sci. Techno!. 20:1235-1243, 1986.
9-11
------- |