EPA 910/9-91-027
Puget Sound Estuary Program
Evaluation of the
Atmospheric Deposition of
Toxic Contaminants to Puget Sound
August 1991
-------
EPA 910/9-91-027
Puget Sound Estuary Program
Evaluation of the
Atmospheric Deposition of
Toxic Contaminants to Puget Sound
Prepared by: Puget Sound Water Quality Authority
Olympia, Washington
Prepared for: U.S. EPA Region 10
Seattle, Washington
August 1991
-------
PUGET SOUND WATER QUALITY AUTHORITY
PROJECT STAFF
Authority Chair
Christine Gregoire
Project Manager
Naydene Maykut
Editing
Marcia Lagerloef
John Dohrmann
Krag Unsoeld
Word Processing, Graphics, and Reproduction
Lorena Mrachek
111
-------
PROJECT SCIENTISTS
Battelle Marine Research Laboratory
Dr. Eric Crecelius
Project Manager
Aerosol and Deposition Study
AMTEST, Inc.
Chris Hansen
Nutrient Analysis
U.S. EPA, Region 10
Air Monitoring and Analysis Section
! Rob Wilson
Meteorological Data Collection
TRC Environmental Consultants
Kirk Winges
Project Manager
WYNDsoft*
Dr. Halstead Harrison
Diffusion/Transport Modeling
NBA, Inc.*
Dan Redline
Source Test/Receptor Modeling Field Study
Dr. David Kalman
PAH Analysis for Source Test
Joe Zukauskas*
Mobilization Coefficients
U.S. EPA Research Triangle Park
Atmospheric Research and Exposure Assessment Laboratory
Robert K. Stevens
Dr. Teri L. Conner
Receptor Modeling
Man Tech Environmental Technology, Inc.
Dr. William Ellenson
Receptor Modeling Laboratory Analysis
Battelle Memorial Research Institute
Deborah Smith
Al Pollack
VOC Analysis
PTI Environmental Services
Mary Louise Conte, Editor
*TRC Subconsultants
IV
-------
ACKNOWLEDGMENTS
This report is the product of U.S. Environmental Protection Agency Cooperative Agreement
No. CE-000412. The project was partially funded through the National Estuary Program under
the authorities of the Clean Water Act as amended. Funding was provided by the U.S. EPA,
Office of Marine and Estuarine Protection through the Region 10 office. The conclusions and
recommendations of this report do not necessarily reflect the views of the U.S. Environmental
Protection Agency, nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
U.S. EPA PROJECT MANAGER
Office of Puget Sound, Region 10
Dr. John Armstrong
Special technical support was provided by the Air Monitoring and Analysis Section of Region
10 EPA.
Jon Schweiss (Chief)
Rob Wilson
Marsha Lee
William Ryan
William Puckett
Guidance and oversight for the design of the studies was provided by the Atmospheric
Deposition Technical Work Team.
ATMOSPHERIC DEPOSITION TECHNICAL WORK TEAM
Dr. John Armstrong, Office of Puget Sound, Region 10 EPA
Leslie Carpenter, Washington Department of Ecology
Janet Chalupnik, American Lung Association
Dr. Eric Crecelius, Battelle Marine Research Laboratory
Art Dammkoehler, Air and Toxics, Region 10 EPA
Fred Fenske, Simpson Tacoma-Kraft Co.
Dave Galvin, METRO
Dr. Jack Hardy, Western Washington University
Dr. Halstead Harrison, University of Washington
Tom Hubbard, METRO
Dr. Cheryl Krone, NOAA
Dr. Tim Larson, University of Washington
Tony Paulson, NOAA
Jamie Randies, Northwest Air Pollution Control Agency
Ed Rashin, Washington Department of Ecology
Dave Renstrom, City of Bellevue
Jon Schweiss, Environmental Services, Region 10 EPA
Rob Wilson, Environmental Services, Region 10 EPA
John Yearsley, Environmental Services, Region 10 EPA
-------
Access to properties and the provision of power or other facilities necessary for sampling was
made possible by the following sampling site managers:
Alexander Avenue, Puget Sound Air Pollution Control Agency
Brown's Point Lighthouse, U.S. Coast Guard
Morse Industrial Supply, Tom Knackstad
Riverside School, Edward White
Sea-Land, Lynn Tarkenton
Tyee Marina, Margaret Wood
VI
-------
TABLE OF CONTENTS
Page
LIST OF TABLES xi
LIST OF FIGURES xiii
EXECUTIVE SUMMARY xvii
CHAPTER 1. INTRODUCTION AND BACKGROUND 1
Introduction 1
The Problem 1
The State of our Knowledge 2
Organization of this Report 3
Background 3
Toxic Contaminants in the Atmosphere 3
Atmospheric Deposition Process 4
Transport Processes 6
CHAPTER 2. STUDY PLAN 9
Study Objectives 9
Study Area Selection 9
Target Chemicals 9
Study Components 11
Sampling Network 11
Meteorological Data Collection 14
Six-Month Aerosol and Deposition Sampling 16
Field Study for Receptor Modeling 16
Emission Inventory 17
Source Test of Simpson Tacoma Kraft Pulp Mill 17
Receptor Modeling 17
Diffusion/Transport Modeling 18
CHAPTER 3. METEOROLOGICAL STUDY 19
Meteorological Data Collection Program 19
Monitoring Site 19
Sensors 19
Operation 20
Data Quality 20
Results 21
CHAPTER 4. SIX-MONTH AEROSOL AND DEPOSITION STUDY 25
Objectives 25
Sampling Network 25
Experimental Methods 25
Aerosol Sampling 25
Deposition Sampling 27
Chemical Analysis of Aerosol and Deposition Samples 27
Sample Selection 29
Results and Discussion 31
Aerosol Results 31
Vll
-------
Table of Contents cont'd Page
Deposition Results ................................ .... 43
Quality Control ......................................... 48
Quality Control Results for Field Sampling ..................... 48
Quality Control Results for Chemical Analyses ................... 48
Summary ............................ ................. 50
Aerosol (PM^,) ..................................... 50
Deposition ......................................... 51
CHAPTER 5. EMISSIONS ................................... 53
Emission Inventory ....................................... 53
Point Sources ....................................... 53
Area Sources ....................................... 57
Simpson Tacoma Kraft Source Test ............................. 59
Test Results ........................................ 59
Kaiser Particle-Size Source Test Results .......................... 60
CHAPTER 6. RECEPTOR MODELING STUDY ..................... 63
Objectives ............................................ 63
Ambient Monitoring ...................................... 63
Analytical Results of 18-Day Field Study and Discussion ............. 63
Motor Vehicle Tracers - Lead ............................. 78
Motor Vehicle Tracers - Volatile Organic Compounds .............. 78
Receptor Modeling Results .................. . ............... 80
Source Apportionment by Chemical Mass Balance ................. 80
Description of Sources ........ ........... . ............. 81
Results and Discussion ................................. 83
Summary ............................................. 91
Aerosol (PM10) ...................................... 91
Receptor Modeling of PMjs Emissions Sources ................... 91
CHAPTER 7. DIFFUSION/TRANSPORT MODELING ................. 93
Objectives ............................................ 93
WV3 Eulerian Grid Dispersion Model ........................... 93
Model Input ........................................... 94
Winds and Transport Coefficients .............................. 98
Results: Simulations ................................... ... 101
Simulations of Concentrations and Fluxes ....................... 101
Mobilization of Runoff to Commencement Bay ................... 109
CHAPTER 8. COMPARISON OF STUDIES ....... ................. 113
Paired Comparisons ...................................... 113
1. Aerosol/Deposition Sampling ............................ 113
2. Aerosol/Diffusion Modeling ............................ 120
3. Aerosol/Receptor Modeling ............................. 128
4. Deposition/Diffusion Modeling ........................... 131
5. Deposition/Receptor Model ..... . ....................... 134
6. Receptor Model/Diffusion Model .................... ..... 134
Summary ................................. . ........... 142
Vlll
-------
Table of Contents cont'd Page
CHAPTER 9. SYNTHESIS OF RESULTS 145
Context 145
Spatial 145
Temporal 145
Meteorology 146
Emissions 146
Synthesis 149
Components of the Aerosol 149
Applications to Commencement Bay and Puget Sound 152
Mass Loading to Commencement Bay 152
Input Timing and Location 154
Loading to Puget Sound 155
CHAPTER 10. CONCLUSIONS AND RECOMMENDATIONS 157
Conclusions 157
Relative Importance of Atmospheric Deposition 157
Effectiveness of the Tools 158
Recommendations 159
GLOSSARY 161
REFERENCES 163
IX
-------
LIST OF TABLES
Page
Chapter 2. Study Plan
Table 2-1. Height of Samplers Above Surface 14
Chapter 4. Six-Month Aerosol and Deposition Study
Table 4-1. Aerosol and Deposition Sampling Plan 25
Table 4-2. Extraction Procedures and Analytical Techniques for Aerosol Deposition
Samples 30
Table 4-3. Study PAHs 30
Table 4-4. Aerosol and Deposition Sample Analysis Summary 31
Table 4-5. Mean Paniculate Concentrations at Six Tacoma Sites (/ig/m3) 32
Table 4-6. Mean Elemental Aerosol Concentrations at Six Tacoma Sites (ng/m3) 34
Table 4-7. Mean Particulate PAH and Total Combustion PAH Concentrations at Six
Tacoma Sites (ng/m3) 40
Table 4-8. Mean Vapor PAH Concentrations at Six Tacoma Sites (ng/m3) 41
Table 4-9. Concentrations of Particulate and Vapor Aliphatic Hydrocarbons (C9 to
C36) and PCBs (21 Congeners) (ng/m3) 43
Table 4-10. Mean Concentrations of Metals in Atmospheric Deposition Samples
Collected in Tacoma from July to December 1989 44
Table 4-11. Mean Metals Deposition Rates 44
Table 4-12. Concentrations of Nutrients and Chloride in Atmospheric Deposition
Samples Collected in Tacoma from July to December 1989 45
Table 4-13. Mean PAH and Total Combustion PAH Atmospheric Deposition Rates
at Five Tacoma Sites (ng/m2/day) 46
Chapter 5. Emissions
Table 5-1. Point Source for PM10 .54
Table 5-2. Point Source for VOCs 56
Table 5-3. PM10 Area Source Inventory-1986 58
Table 5-4. Growth Factors 1986-1991 59
Table 5-5. Mean Percent of PM10 Emissions 60
Table 5-6. Kaiser Particle-Size Distribution Source Test Results - July, 1988 61
Chapter 6. Receptor Modeling Study
Table 6-1 (a). Average Fine-Particle Species Concentrations at the Alexander Avenue Site .... 64
Table 6-l(b). Average Fine-Particle Species Concentrations at the Morse Supply Site 65
Table 6-2(a). Average Coarse-Particle Species Concentrations at the Alexander Avenue Site . . 66
Table 6-2(b). Average Coarse-Particle Species Concentrations at the Morse Supply Site 67
Table 6-3(a). Average VOC Concentrations at the Alexander Avenue Site 69
Table 6-3(b). Average VOC Concentrations at the Morse Supply Site 70
Table 6-4. Average Concentrations of Soil-Corrected Potassium, K' 79
Table 6-5. Correlations of VOCs with Lead and K'(N=71) 80
Table 6-6. Percent Contribution of Major Sources to Calculated Fine-Particle Mass
for Daytime, Nighttime, and All Samples - Alexander Avenue Site ..89
Table 6-7. Percent Contribution of Major Sources to Calculated Fine-Particle Mass
for Daytime, Nighttime, and All Samples - Morse Supply Site 90
XI
-------
List of Tables cont'd
Page
Table 6-8. Percent Contribution of Major Sources to Fine-Particle Mass for
December and January Samples 90
Chapter 7. Diffusion/Transport Modeling
Table 7-1. PAH Emissions from the Kaiser Aluminum Smelter 96
Table 7-2. Total Emissions of PM10 for Diffusion/Transport Model 96
Table 7-3. Empirical Atmospheric Deposition Parameters 101
Table 7-4. Estimates by the WV3 Model of Fractions of Emitted Toxic Contaminants
Deposited Within the Modeling Domain (91.5 km2) and Mobilized Into
the "Prompt-Runoff Watershed of Commencement Bay Ill
Table 7-5. Projected Mass Loading of Airborne Toxic Contaminants to Puget Sound
Based on Emissions in 91.5 km2 WV3 Modeling Domain 112
Chapter 8. Comparison of Studies
Table 8-1. Mean Deposition Rate, Mean Aerosol Concentration, and Deposition
Velocity (Vd) at the Alexander Avenue Site for • Three Sampling
Periods 114
Table 8-2. Comparisons of Observed PM^,,, with Simulated PM10 by Various
Protocols 123
Table 8-3. Comparisons Between Observed PM^ and Simulations by WV3
Model 123
Table 8-4. Comparison Scores for Model Estimates of Three Metals (As, Pb, Zn)
vs Study Observations 125
Table 8-5. Comparison Scores for Model Estimates of Eight PAHs vs Study
Observations-Renormalized 125
Table 8-6. Summary of Scores Comparing PM^ Observations with WV3
Simulations, Stratified by Season 126
Table 8-7. Comparisons Between Observed PM10 and Simulations by WV3 Model 126
Table 8-8. Scores Comparing PSAPCA's PM10 Observations with Simulations by
WV3, Stratified by Season 126
Table 8-9. Scores Comparing 18-Day Receptor Modeling PM10 ("Coarse+Fine")
Observations with Simulations by WV3, All Points (Morse Supply and
Alexander Avenue) 126
Table 8-10. Deposition Velocities Inferred from PSWQA Measurements at All
Sites 133
Table 8-11. Receptor Modeling Results for Fine Particulates (PM25) Using Lead (Pb)
and a VOC (o-xylene) as Tracers of Vehicle Exhaust 135
Table 8-12. Fractional Source Attributions from WV3 December 1 through 31,
1989 135
Table 8-13. Receptor Modeling Results for (% Measured for PM10) Using Two
Alternative Methods of Calculation 136
Chapter 9. Synthesis of Results
Table 9-1. Contaminant Mass Loading to Commencement Bay 152
Table 9-2; Estimated Mass Loading to Commencement Bay Based on WV3 Model 153
Table 9-3. Estimated Mass Loading to Commencement Bay Based on Adjusting WV3
Model Results for Increased Deposition and Mobilization 153
Table 9-4. Predicted Concentrations of Specific Contaminants in the Sea-Surface
Microlayer 154
XII
-------
LIST OF FIGURES
Page
Chapter 1. Introduction and Background
Figure 1-1. Formation and Deposition of Fine and Coarse Particles 5
Chapter 2. Study Plan
Figure 2-1. Commencement Bay in Puget Sound, Washington 10
Figure 2-2. Commencement Bay Air and Deposition Sampling Sites 12-13
Figure 2-3. Tacoma Tideflats Stack Air Emissions, Toxic Release Inventory 15
Chapter 3. Meteorological Study
Figure 3-1 (a). Air Stagnation Episodes (November - December) 23
Figure 3-l(b). Air Stagnation Episodes (December - January) 23
Chapter 4. Six-Month Aerosol and Deposition Study
Figure 4-1. PS-1 Aerosol Sampler 26
Figure 4-2. Deposition Sampler 28
Figure 4-3. Suspended Paniculate (PM^) Concentrations, Alexander Avenue and
Tyee Marina Sites 33
Figure 4-4. Aerosol Iron (Fe) Concentrations, Alexander Avenue and Tyee
Marina Sites 36
Figure 4-5. Aerosol Potassium (K) Concentrations, Alexander Avenue and Tyee
Marina Sites 37
Figure 4-6. Aerosol Lead (Pb) Concentrations, Alexander Avenue and Tyee
Marina Sites 38
Figure 4-7. Aerosol Zinc (Zn) Concentrations, Alexander Avenue and Tyee
Marina Sites 39
Figure 4-8. Average Monthly Vapor PAH Concentrations, Alexander and Tyee
Marina Sites 42
Figure 4-9 Atmospheric Deposition of CPAH versus Time, Alexander Avenue
and Tyee Marina Sites 47
Chapter 5. Emissions
Figure 5-1. 12-km Diameter Circle Around Fire Station No. 12 55
Chapter 6. Receptor Modeling Study
Figure 6-1. Major Constituents of Fine-Particle Mass at the Alexander Avenue Site 71
Figure 6-2. Major Constituents of Fine-Particle Mass at the Morse Supply Site 71
Figure 6-3. Fine-Particle Mass Comparison of Alexander Avenue and Morse
Supply Sites versus Sampling Interval 73
Figure 6-4. Coarse-Particle Mass Comparison of Alexander Avenue and Morse
Supply Site versus Sampling Interval 74
Figure 6-5. Fine-Particle Elemental Carbon Comparison of Alexander Avenue
and Morse Supply Sites versus Sampling Interval 75
Xlll
-------
List of Figures cont'd Page
Figure 6-6. Fine-Particle Sulfate (by Ion Chromatography) Comparison of
Alexander and Morse Supply Sites versus Sampling Interval 76
Figure 6-7. Fine-Particle Organic Carbon Comparison of Alexander Avenue and
Morse Supply Sites versus Sampling Interval 77
Figure 6-8. Fine-Particle Receptor Modeling Results at the Alexander Avenue
Site Using Lead (Pb) as the Leaded-Fuel Vehicle Tracer, Total Motor
Vehicle Exhaust is Reported 85
Figure 6-9. Fine-Particle Receptor Modeling Results at the Morse Supply Site
Using Lead (Pb) as the Leaded-Fuel Vehicle Tracer, Total Motor
Vehicle Exhaust is Reported 85
Figure 6-10. Fine-Particle Receptor Modeling Results at the Alexander Avenue
Site Using o-xylene as the Motor Vehicle Tracer 87
Figure 6-11. Fine-Particle Receptor Modeling Results at the Morse Supply Site
Using o-xylene as the Motor Vehicle Tracer 87
Figure 6-12. Fine-Particle Receptor Modeling Results at the Alexander Avenue Site
Using 2,2,4-trimethylpentane as the Motor Vehicle Tracer 88
Figure 6-13. Fine-Particle Receptor Modeling Results at the Morse Supply Site
Using 2,2,4-trimethylpentane as the Motor Vehicle Tracer 88
Chapter 7. Diffusion/Transport Modeling
Figure 7-1. Modulation Function for Cars and Woodsmoke 95
Figure 7-2. Map of Sampling Sites Used for Modeling 95
Figure 7-3. Distribution of Surface Emissions of PM10 97
Figure 7-4. Distribution of Mid-Level Emissions of PM10 97
Figure 7-5. Wine-Rose Scattergram for 24-Hour Vector Winds 99
Figure 7-6. Wind Hodograph for 24-Hour Winds at 10m Height 100
Figure 7-7(a). Simulated Highest 24-Hour (midnight-to-midnight) Averages for
Arsenic (As) Aerosols 102
Figure 7-7(b). 186-Day Average for Arsenic (As) Aerosols 102
Figure 7-7(c). Sum of Wet and Dry Deposition for Arsenic (As) Aerosols 103
Figure 7-8(a). Simulated Highest 24-Hour (midnight-to-midnight) Averages for
Copper (Cu) Aerosols 103
Figure 7-8(b). 186-Day Average for Copper (Cu) Aerosols 104
Figure 7-8(c). Sum of Wet and Dry Depositions for Copper (Cu) Aerosols 104
Figure 7-9(a). Simulated Highest 24-Hour (midnight-to-midnight) Averages for Lead
(Pb) Aerosols 105
Figure 7-9(b). 186-Day Average for Lead (Pb) Aerosols 105
Figure 7-9(c). Sum of Wet and Dry Deposition for Lead (Pb) Aerosols 106
Figure 7-10(a). Simulated Highest 24-Hour (midnight-to-midnight) Averages for Zinc
(Zn) Aerosols 106
Figure 7-10(b). 186-Day Average for Zinc (Zn) Aerosols 107
Figure 7-10(c). Average Total Deposition Rate (Wet + Dry) for Zinc (Zn) Aerosols 107
Figure 7-11 (a). Simulated Highest 24-Hour (midnight-to-midnight) Averages for
Vapor + Paniculate PAH 108
Figure 7-1 l(b). 186-Day Average for Vapor + Particulate PAH . 108
Figure 7-1 l(c). Sum of Wet and Dry Deposition for Vapor + Particulate PAH 109
xiv
-------
List of Figures cont'd Page
Chapter 8. Comparison of Studies
Figure 8-1. Comparison of Mean Aerosol Concentration (ng/m3) and Mean
Deposition Rate (/*g/m2/day) at the Alexander Site for September 7 -
21, November 2 - 16, and November 30 - December 14 115
Figure 8-2(a). Aerosol Paniculate Metal Concentrations (ng/m3) at the Morse Supply
Site 116
Figure 8-2(b). Metal Deposition Rate (jig/m2/day) at the Morse Supply Site 116
Figure 8-3(a). Aerosol Paniculate Metal Concentrations (ng/m3) at the Alexander
Avenue Site 117
Figure 8-3(b). Metal Deposition Rate (/*g/m2/day) at the Alexander Avenue Site 117
Figure 8-4(a). Aerosol Paniculate PAH Concentration (ng/m3) at the Morse Supply
Site 118
Figure 8-4(b). PAH Deposition Rate (ng/m2/day) at the Morse Supply Site 118
Figure 8-5(a). Aerosol Paniculate PAH Concentrations (ng/m3) at the Alexander
Avenue Site 119
Figure 8-5(b). PAH Deposition Rate (ng/m2/day) at the Alexander Avenue Site 119
Figure 8-6. Scattergram of Observed PM^ vs PM10 Simulations of WV3 Model 122
Figure 8-7. Composited B as a Function of Time of Day 127
Figure 8-8. Composited PM10 Simulations as a Function of the Time of Day 127
Figure 8-9(a). Comparison of Aerosol Collected by Six-Month Study (PM^) and
the 18-Day Study (PM25, PM10) at the Alexander Avenue Site 130
Figure 8-9(b). Comparison of Aerosol Collected by Six-Month Study PM^) and
18-Day Study (PM^, PM10) at the Morse Supply Site 130
Figure 8-10. TSP and PM10 Measured during the Study Period at Fire Station No. 12 ... 131
Figure 8-11. Receptor Model Results of Low Wind Days (December 12 and 14)
at the Morse Supply Site 138
Figure 8-12. Receptor Model Results for High Wind Days (January 1 and 5) at the
Morse Supply Site 138
Figure 8-13. Receptor Model Results for Low Wind Days (December 12 and 14)
at the Alexander Avenue Site 139
Figure 8-14. Receptor Model Results for High Wind Days (January 4 and 5) at the
Alexander Avenue Site 139
Figure 8-15. Tracer-Rose Using Fire Station No. 12 PM10 Data 140
Figure 8-16. Tracer-Rose Using Alexander Avenue Site Data 141
Chapter 9. Synthesis of Results
Figure 9-1. Comparison of Monthly Mean Temperatures for the Study Period with
the Monthly Mean, Average Maximum, and Average Minimum for
1980-1988 147
Figure 9-2. Comparison of Monthly Mean Precipitation for the Study Period with
the Monthly Mean, Average Maximum, and Average Minimum for
1980-1988 147
Figure 9-3. Comparison of Monthly Mean PM10 at the Alexander Avenue Site for
the Study Period with the Monthly Mean, Mean High, and Mean Low
PM10 from 1986-1988 149
xv
-------
Executive Summary
INTRODUCTION
Evaluation of the Atmospheric Deposition of Toxic Contaminants was a study conducted to
more fully understand the contribution of airborne toxic contaminants to water quality problems
in Puget Sound. The study was managed by the Puget Sound Water Quality Authority, with
major funding provided by the National Estuary Program of the EPA through the Region 10
office. The Puget Sound Air Pollution Control Agency (PSAPCA) also contributed to the
study.
One of the main concerns that prompted the study was a growing recognition of the problem
of cross-media transfer of pollutants. The deposition of airborne particles and gases may be
responsible for contributing certain heavy metals, polycyclic aromatic hydrocarbons (PAHs),
and other organic compounds to Puget Sound. Deposition can occur directly as particles settle
onto the water surface, or indirectly as they settle on land and are subsequently washed or
blown off the land into the Sound. These toxic chemicals are then added to the chemicals in
the surface water layer (microlayer) of the Sound, the water column, and/or the sediments.
The resultant increase in toxicity from this deposition is a concern because of its potential effect
on the health and survival of aquatic life in the Sound, and ultimately, through the food chain,
on human health.
Atmospheric deposition of toxic contaminants has received only limited attention in the Puget
Sound region, but these studies pointed towards this pathway as a potentially important
contributor to the total loading of lead (Pb), arsenic (As), PAHs and other organic compounds
to the Sound. Atmospheric deposition has been found to be a major source of paniculate metals
and PAHs in the Southern California Bight; a source of metals, polychlorinated biphenyls
(PCBs), DDT, and other pesticides in the Great Lakes region; and a source of one-quarter of
the nitrogen compounds that cause serious eutrophication problems in Chesapeake Bay.
STUDY DESIGN
Commencement Bay, an embayment near Tacoma, Washington, was selected for a pilot-scale
study to address the following objectives:
* Develop a better understanding of the importance of atmospheric deposition relative to other
inputs of toxic contaminants to Commencement Bay.
* Develop efficient and cost-effective tools for assessing this question in other reaches and
embayments of Puget Sound, and other water bodies as well.
Commencement Bay was selected to represent a "worst case" test area, because it is a heavily
industrialized area with a complex mix of air pollution sources and documented high
concentrations of Pb, PAHs, and other chemicals in both the sediments and sea-surface
microlayer. The study design was focused on sampling and analysis of metals (e.g. lead, zinc,
arsenic, manganese) and combustion PAHs (three- to five-ring compounds). Low molecular
weight PAHs, PCBs, aliphatic hydrocarbons, and nutrients were also sampled and analyzed.
PAHs are of concern because some are persistent in the environment and are potent cancer-
causing and mutagenic agents. Metals are of concern because they are toxic to aquatic life at
certain elevated concentrations.
xvii
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
The "tools" employed for the study consisted of four field sampling components and two
modeling efforts. Some general capabilities and limitations of the tools were known in advance:
field measurements are resource-intensive and therefore cannot be used realistically to cover
a large or dense grid of sites, modeling can provide larger area coverage, but its accuracy is
dependent on the input data and the assumptions used to develop the model. In applying these
tools in concert in a single embayment, the aim was to compare and validate their effectiveness
and determine whether some more limited combination of tools would provide an adequate
assessment of the contribution of atmospheric contaminants to the water.
The field components consisted of:
» Six-month meteorological data acquisition - temperature, solar radiation, precipitation, and
wind speed and direction collected at one site.
* Six-month aerosol and deposition sampling — twice per week (three- or four-day) aerosol
samples (collecting particles up to 25-50 /*m in diameter) at six sites [broken into intensive
(five-site) and non-intensive (three-site) sampling periods] , and two-week deposition samples
at five sites (no definite upper particle size limit).
* 18-day sampling for receptor modeling ~ aerosol sampling for particles up to 10 /*m and
volatile organic compounds (VOCs) at two sites.
* Source testing of the Simpson Tacoma Kraft pulp mill — sampling and analysis of metals
and PAHs in the emissions.
The modeling efforts, which complemented the field studies and provided the tools for
interpreting and extrapolating from the field study results, included:
*• Receptor modeling ~ using a mathematical method to attempt to match patterns in
chemicals released from sources with patterns of chemicals collected at receptor sites in
order to apportion the ambient aerosol to its sources (focusing on particles less than
10
* Diffusion/transport modeling (including a mobilization coefficient model) — using emissions
data, emissions stack parameters, and meteorological data in a physical model to estimate
the aerosol and deposition concentrations over a 91.5 km2 region and then further estimating
the percentage of the deposited contaminants that are washed off the land (rather than held
in the soil) into Commencement Bay (focusing on the less than 10 /xm particles).
An emission inventory was a necessary data input for both models. The emission inventory was
prepared compiling previous direct measurements and estimates of both point source and area
source emissions in the Tacoma Tideflats.
A field sampling network was established, centered in the industrialized Tacoma Tideflats.
Six sites were located along two different transects through the study area: one transect crossed
the main industrial sector roughly southwest to northeast, the other stations were located along
the path of the prevailing wind through the area (southeast to northwest). The upwind and
downwind sites were selected to provide an indication of the background contaminant levels.
Stations were located either near or over the water.
XVlll
-------
Executive Summary
The field program was conducted from July 1989 through January 1990, encompassing six
months with a range of meteorological conditions and emissions scenarios.
Following completion of the individual studies, paired comparisons of the data were conducted
to bridge between the studies (particularly where different particle size fractions had been
investigated) to attempt to reconcile results and fill data gaps.
STUDY RESULTS
An important result of the investigation was a much more coherent picture, for the Tacoma
Tideflats/Commencement Bay region, of the contributing sources of atmospheric contaminants
and the temporal and spatial variations in the ambient aerosol and the amounts of contaminants
deposited. Some of the more significant findings include:
* The summer and winter high-concentration2 aerosols were qualitatively different. The
high-concentration summer aerosol appears to have consisted largely of resuspended larger
particles, such as fugitive dust. The highest concentrations of metals were found in the
industrial area, particularly at the Sea-Land site (Pb and Zn). The highest concentrations
of PAHs were measured at the Alexander Avenue site. The set of meteorological
conditions favoring transport of these high-concentration aerosols consisted of prolonged
periods of dry weather followed by moderate-to-high wind speeds.
* The 18-day study showed that on average the late fall/winter PM10 (particles less than
10 /xm) aerosol consisted largely of "fine" particles (particles less than 2.5 urn). A large
percentage of the total fine-particle mass consisted of organic compounds. The similar
aerosol concentrations throughout the sampling area suggest a substantial influence of
regional sources relative to local sources. High concentrations of the particles were most
often associated with air stagnation episodes and low wind speeds predominantly from the
southeast.
* During the summer/early fall sampling period, when larger particles appear to have been
an important component of the aerosol, earth-moving and construction activities (with
accompanying exhaust and road dust from heavy machinery) occurred in the vicinity of the
industrial area sampling sites. Also during the study period Kaiser had at their site
approximately eight acres of wet scrubber sludge containing PAHs, some of which had the
potential of becoming airborne dust during the dry season. Ore offloading operations
occurring near the Sea-Land site included offloading of "black ore" (ore from which lead,
zinc, silver, and gold are extracted by smelting) into open rail cars. These sources are
inferred based on the patterns of chemicals in the deposition samples and their proximity
to these industrial sites.
* The major source of the late fall and winter high-concentration fine-particle aerosol (less
than 2.5 /xm) appears to be woodsmoke that drains downslope and downwind from
residential areas into the industrial area. This unexpected result was supported by the
receptor modeling, a tracer rose analysis (which projects concentrations of contaminants
measured at a receptor as vectors pointing upwind) and comparisons between the diffusion
model simulations and data collected at Puget Sound Air Pollution Control Agency
(PSAPCA) monitors in the Tideflats.
2High-concentration aerosol refers to episodes during the study period when the highest
concentrations of aerosol were observed.
xix
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
* The sampling times and locations may not have been adequate to accurately reflect the
relative influence of the Simpson Tacoma Kraft mill, the largest emissions source for PM10
in the study area.
* The deposition rates for metals and PAHs were higher in the summer than in the fall.
High concentrations usually occurred after periods of dry weather and were associated
with moderate wind speeds. The substantial concentration gradients that were detected
between sampling sites reflected the proximity of the contaminant sources to the sampling
sites. These facts indicate that resuspended large particles were most likely a major part
of the deposition. Deposition rates for all the metals were greater at the industrial sites
than the background sites and were probably the result of nearby industrial practices. PAH
deposition rates were five to 10 times higher at the Alexander site than at Tyee Marina,
Morse, and Sea-Land. PAH deposition was probably dominated by large particles from
the Kaiser emissions or resuspended dust from the Kaiser site.
* The diffusion model simulations predicted that one to three percent of the PM10 emissions
would be deposited and reach Commencement Bay during normal rain event runoff. More
than 90 percent of the emissions would be advected beyond the boundaries of the 91.5 km2
modeling domain. Based on knowledge of the dynamics of small particles, and the
topography and meteorology of Puget Sound, another three to four percent of the emissions
are expected to be deposited farther downwind in Puget Sound.
Based on the study results some first order estimates of atmospheric loading to Commencement
Bay were computed and compared with other sources of contaminants to these waters. With
the data available only limited conclusions and conjectures can be drawn:
* Using deposition rates as measured at the Riverside School (low rate) and Sea-Land (high
rate) sites, and an assumed area of Commencement Bay of 10 km2, mass loading rates for
metals and combustion PAHs were calculated and compared with measured loadings from
municipal and industrial point sources and the Puyallup River. The metals loadings from
the atmosphere were considerably lower than the point source and the river loadings.
Similar data was not available for comparison of the PAH loadings. These deposition
estimates did not include runoff from land.
* Computing mass loading from the diffusion model simulations and the estimated
mobilization of contaminants into runoff also resulted in metals loadings that were minor
compared to the point sources and the river inputs. However, these estimates are probably
low because of the low assumed deposition velocities (0.1 cm/s) and mobilization
coefficients.
* Comparing estimates of the predicted microlayer concentrations of metals and PAHs, based
on deposition rates measured at Tyee Marina and Sea-Land, with measured microlayer
concentrations demonstrated that it was possible to account for the measured concentrations
with the estimated atmospheric inputs.
CONCLUSIONS AND RECOMMENDATIONS
The study was successful as a pilot-level effort in developing first order estimates of
atmospheric deposition and in providing a test and evaluation of a comprehensive set of field
sampling and modeling tools that can be used (with modifications) to study atmospheric
deposition elsewhere. The conclusions set atmospheric deposition in perspective relative to
other sources of contaminants in a heavily industrialized region of Puget Sound, where overall
xx
-------
Executive Summary
contaminant loading is high. It should not be assumed that the relative contribution of
atmospheric deposition is the same elsewhere in the Sound.
Relative Importance of Atmospheric Deposition
* Direct atmospheric deposition of metals appears to be a small contributor, in terms of mass
loading, relative to point source water discharges of metals to Commencement Bay.
>• The data available do not allow definitive conclusions on the relative importance of
atmospheric inputs of PAHs. The deposition measurements of PAHs suggest that Kaiser
may be an important local source of PAHs. [Note: Since the sampling conducted for the
study, Kaiser has covered PAH-contaminated scrubber sludges on their property that had
the potential to become part of the fugitive dust loading of PAHs to the Bay.]
* Woodsmoke, another source of PAHs, is an important contributor to the ambient aerosol
during winter periods of air stagnation. However, PAHs generally make up less than 1
percent of the woodsmoke emissions mass. In addition, woodsmoke emissions are
predominantly in the less than 2.5 pm particle size fraction, therefore they are not expected
to settle nearby and be an important component of atmospheric deposition to
Commencement Bay.
* The total input from atmospheric deposition cannot be assessed from the results of this
study because of uncertainty in the estimates of atmospheric deposition on land and its
subsequent runoff to the Bay.
* Atmospheric deposition may be significant relative to other inputs in particular zones,
especially close to large sources of fugitive dust, near discharge pipes for stormwater, and
at the sea surface microlayer (the top 50-100 pm of the water column). These focused
entry points into the water, coupled with pulsed inputs create periodic conditions where
atmospheric deposition may be the dominant source of toxic contaminants.
*• It is not possible from the data to make a definitive prediction of deposition in the far field
(Puget Sound). Emissions that are likely to be carried beyond Commencement Bay are
generally made up of vapors and particles less than 2.5 jtm in diameter. For this size
particle there is a rapidly decreasing probability of deposition with distance from the
source.
Effectiveness of the Tools
*• Direct aerosol and deposition measurements are central to estimating the atmospheric
contributions to the water. The experience in this study suggests that important
considerations for future application of these sampling approaches include: 1) consistency
in particle size fractions sampled; 2) length of the sampling period; 3) the number of
samples chemically analyzed; 4) sampling protocol and amount for PCBs, aliphatic
hydrocarbons, nutrients, and vapor PAHs; and 5) sampling locations relative to winds and
primary sources.
* Receptor modeling was effective in discerning important sources of the ambient aerosol,
particularly the woodsmoke advected into the study area. For future modeling applications,
important considerations are: 1) the temporal and spatial representativeness of the sampling;
2) accurate information on the chemical signatures of sources and their variability with
time; and 3) sample quantity for organic analysis.
xxi
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
* Diffusion modeling showed promise and performed best when compared with measurements
of PM10 mass. Model refinements should address: 1) accurate information on chemical and
particle size characteristics of emissions; 2) size of the model domain relative to location
of sources; 3) particle size fraction modeled; 4) deposition velocities relative to particle
sizes.
* Mobilization coefficient modeling was entirely theoretical for this study. For future
applications particular attention should be devoted to: 1) field validation of coefficients; 2)
mobilization during severe storm events; and 3) size of the model domain relative to runoff
sources.
Recommendations
Further basic research is needed (not necessarily in Puget Sound) to improve some critical
technical methods and the understanding of important physical processes, including:
* Aerosol particle distribution and chemistry;
* Fugitive dust transport and its controlling parameters;
>• Particle dynamics related to size and chemistry;
* Deposition sampling techniques;
> Chemical and physical dynamics of contaminant mobilization;
* Hybrid dispersion modeling.
If a more definitive analysis is desired for Commencement Bay and Puget Sound, the following
additional studies might be pursued either independently or as part of a larger program:
* More detailed characterization of ambient and source particle size distributions and
chemistry;
*• Sampling of the microlayer and nearshore zones to complement deposition sampling;
* Combined sewer overflow (CSO)/stormwater monitoring;
*• Refined emissions inventories, including temporal variations;
* Vertical and horizontal wind profiling;
*• Hybrid dispersion modeling;
*• More detailed monitoring of discharges of toxic substances from permitted air and water
discharges.
xxn
-------
Chapter 1. Introduction and Background
INTRODUCTION
This report presents the findings of a study that was conducted to more fully understand the
contribution of airborne toxic contaminants to water quality problems in Puget Sound.
Commencement Bay, an embayment near Tacoma, Washington, was used as a case study to:
1) determine the importance of atmospheric deposition relative to other sources of toxic
contaminants; and 2) develop and test tools for assessing the contribution of atmospheric
deposition in other reaches and embay men ts of the Sound. Both field sampling and modeling
were conducted, focusing on the six-month period from July 1989 to January 1990. The study
was managed by the Puget Sound Water Quality Authority, with major funding provided by the
National Estuary Program of the EPA through the Region 10 office. The Puget Sound Air
Pollution Control Agency (PSAPCA) also contributed to the study. Academic, agency, and
private sector scientists were involved in designing and conducting the study.
The Problem
One of the main concerns that prompted the study was a growing recognition of the potential
for cross-media transfer of pollutants. The atmosphere contains contaminants in the form of
particles and gases that originate from exhaust from cars, trucks, and other mobile sources;
woodstove smoke; smoke from slash burning; emissions from commercial and industrial
processes and materials storage; and dust from soil erosion. The deposition of airborne
particles and gases may be a key mechanism for contributing certain heavy metals, polycyclic
aromatic hydrocarbons (PAHs), and other organic compounds to Puget Sound. Deposition can
occur directly as airborne particles and gases settle onto the water surface, and indirectly as they
settle on land and are subsequently blown off the land or washed via surface-water runoff and
storm sewers into the Sound. These toxic chemicals are then added to the chemicals in the
surface-water layer (or microlayer) of the Sound, the water column, and/or the sediments. The
resultant increase in toxicity from this deposition is a concern because of its potential effect on
the health and survival of aquatic life in the Sound, and ultimately, through the food chain, on
human health.
Atmospheric deposition of toxic contaminants has received only limited attention in the Puget
Sound region, but these studies point towards this pathway as a potentially important contributor
to the total loading of lead (Pb), arsenic (As), PAHs and other organic compounds in the
Sound. During the early 1980s researchers measured the concentrations of toxic chemicals in
airborne particulates at several coastal locations (Prahl, Crecelius, and Carpenter, 1984).
During the mid 1980's other scientists related concentrations of toxic chemicals in marine
sediments to elevated rates of fish disease and altered populations of bottom-dwelling animals
like worms, clams, and shrimp, that serve as food for fish. They identified concentrations of
toxic chemicals above which they always saw biological harm (Barrick et al., 1988). In 1988,
the Washington Department of Ecology released draft sediment criteria for managing the quality
of Puget Sound sediments. A comparison of the annual average concentrations of five organic
compounds in airborne particles collected by Prahl et al. with the draft sediment criteria
revealed that the airborne particles from Seattle significantly exceeded the draft sediment
criteria. This comparison, while overly simplified, suggested that deposition of airborne
particles has the potential to create sediments that are toxic to marine life. The current study
therefore was designed to better assess the atmospheric inputs.
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
The State of our Knowledge
Prior to undertaking the study described in this report, an extensive literature review was
conducted of relevant research in Puget Sound and other coastal waters, including the Great
Lakes (see Appendix A). The intent was to identify how the design of this study could benefit
from prior experience and also to determine how this study could best contribute new
knowledge about the potential significance of atmospheric deposition as a source of toxic
contaminants to coastal waters. Most previous research performed in the Puget Sound area has
involved indirect measurements or estimates of atmospheric deposition. Left unanswered are
key questions about the rates of deposition and the amount of land-deposited material that
actually makes its way to the receiving water. These are difficult questions to answer
quantitatively.
A mass balance calculation done by Paulson et al. (1988b) estimated that municipal and
industrial water dischargers and atmospheric sources contributed roughly equivalent amounts
of Pb to the main basin of Puget Sound, for a total of 66% of the Pb. Romberg et al. (1984)
interpolated from TSP (total suspended particulate) maps of the region, assuming a constant
Pb/TSP relationship, and concluded that atmospheric deposition may account for one-quarter
of the total lead input to the Sound. Romberg et al. made similar calculations for PAHs and
concluded that atmospheric input of combustion-derived PAHs is significant.
Stormwater runoff is thought to be a primary mechanism for transporting toxic chemicals to
Puget Sound after they have been deposited on land. Galvin et al. (1984) found that street dust
and dirt on urban impervious surfaces were major contributors of toxics found in stormwater
[particularly phosphorus (P), lead (Pb), and zinc (Zn)]. However, they found no correlations
between metals and organics sampled in atmospheric suspended particles and those sampled in
street dust. Pitt and Bissonette (1984) suggested that airborne PAHs are combustion products
(from vehicles and other sources) while the street dirt PAHs are from petroleum product spills.
These studies clearly illustrate the complexity and difficulty of accurately determining the
amount and fate of deposited toxics.
Some of the more provocative studies that implicate atmospheric deposition and surface runoff
as sources of toxics to the Sound focus on the contamination measured in the sea surface
microlayer (approximately 50 to 100 /xm boundary layer between the atmosphere and the water
column). Scientists studying the microlayer have found pollutant enrichments at concentrations
100 to 10,000 times greater than in the water column. The concentrations of metals in the
microlayer were found to be two to 15 times greater in Elliott Bay (industrialized) than Sequim
Bay (rural) (Hardy et al., 1985a). Metals concentrations in the microlayer were found to be
six to 65 times greater than in water columns for both bays. The concentrations of the metals
measured in the study agreed well with predictions of atmospheric deposition [based on using
7Be as a tracer for the behavior of submicron particles (Crecelius, 1981)]. Hardy estimated that
up to half the total input of combustion PAHs and particulate metals entering coastal waters in
some areas of the U.S. originates from atmospheric deposition (Hardy et al., 1986b). Lead,
benzo(a)pyrene, and PCBs were reported in relatively high concentrations in the microlayer
(Hardy and Antrim, 1988a).
Studies in other areas of the country have shown that atmospheric deposition is a major source
of particulate metals and PAHs in the southern California Bight (Patterson and Settle, 1974);
a source of metals, polychlorinated biphenyls (PCBs), DDT, and other pesticides in the Great
Lakes region (Eisenreich, et al., 1979-1985); and a source of one-quarter of the nitrogen
compounds that cause serious eutrophication problems in Chesapeake Bay (Fisher, et al., 1988).
-------
Chapter 1. Introduction and Background
Organization of this Report
The remainder of this chapter provides background information on atmospheric deposition and
contaminant transport processes to assist the reader in understanding and interpreting the
findings of the study. Chapter 2 provides a discussion of the overall study design, including
the rationale for site selection, target chemicals, and the individual study components. Chapters
3-7 each address a specific study component and present the methods employed, the results,
and discussion. Chapter 8 compares and integrates the results of the separate studies. Chapter
9 synthesizes the results and discusses how these findings might apply to Commencement Bay
and Puget Sound. Chapter 10 presents conclusions and recommendations.
BACKGROUND
Toxic Contaminants in the Atmosphere
Toxic contaminants exist in the atmosphere in the form of aerosols (a suspension of colloidal
particles in a gas) and vapor. The principal contaminants investigated in this study are trace
metals ~ such as chromium, manganese, iron, nickel, lead, titanium, copper, and zinc — and
polycyclic aromatic hydrocarbons (PAHs). Due to coagulation, condensation, and gas
adsorption processes in the atmosphere, a single aerosol particle can contain many different
materials, including metals and PAHs.
PAHs are a class of complex organic compounds of natural and anthropogenic origin that are
widely-distributed around the world. They are of concern because some are persistent in the
environment and are potent cancer-causing and mutagenic agents. PAHs are produced largely
by the incomplete combustion of fossil fuels (gasoline, kerosene, coal, diesel fuel), waste
incineration, and burning of wood products (woodstoves, forest fires and slash burning).
Burning produces small (<3 /xm) airborne particulate matter (soot and fly ash) on which PAHs
are adsorbed. Lower molecular weight PAHs (2-3 fused aromatic rings) are acutely toxic,
causing illness or death to aquatic organisms. Within the group of higher molecular weight
PAHs, 20 to 30 are proven carcinogens (Neff, 1979). The fate and composition of deposited
PAHs are affected by the processes of sorption, biodegradation, photolysis, and volatilization
(Mills etal., 1985).
Trace metals are found naturally in the environment in crustal materials from which they are
released by erosion and other mechanical activities. Metals are also generated by human
activities and released as an emission from industrial processes and combustion of leaded fuel.
Although many metals are required in small quantities for biological processes, metals such as
mercury, lead, nickel, zinc, and cadmium are of environmental concern because they are toxic
to aquatic life at certain elevated concentrations. As chemical elements, metals are persistent
over time. Many trace metals are soluble in sea water and therefore could be available to
marine food webs (Hardy et al, 1985).
The size distribution of ambient aerosols is distinctly bi-model: fine (<2.5 /xm diameter) and
coarse (>2.5 /xm diameter). The two modes are relatively independent with respect to origin
and differ in their physical and chemical character. Coarse particles are generated mechanically
by weathering of crustal materials and biomass, by disturbance of soil or industrial process
material, or by droplet formation from seaspray. Fine particles are formed mainly from
condensation of hot gases or chemical conversion of gases in the atmosphere to particulate
species. They are typically associated with anthropogenic sources such as combustion or
industrial processes. Fine particles may grow in size to about 0.5 /xm by coagulation. Fine
particles are removed by rainout (processes taking place within clouds, such as the formation
of condensation nuclei), washout (removal of materials below cloud level by falling ice or
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
rain), or turbulent mixing to the surface. By contrast, coarse particles are removed mainly by
gravitational settling. The residence time of particles in the atmosphere is at a maximum for
particles in the 0.1 - 1.0 /xm range (see Figure 1-1).
Atmospheric contaminants are sampled by techniques that are optimized to collect a specific size
range of particles from ambient air. A PM10 sampler collects particles less than approximately
10 pm, which can in some samples be further split into coarse (2.5 - 10 /xm) and fine (0 -
2.5 /xm) fraction. A PM^.^ sampler collects particles up to approximately 50 /xm, depending
on the wind speed. A TSP, or Total Suspended Particulate, sampler is capable of collecting
particles up to about 75 /xm in diameter. Organic vapors are typically collected by capture in
an adsorbent medium, such as XAD-2 resin or polyurethane foam.
Atmospheric Deposition Processes
Deposition of contaminants from the atmosphere onto the surfaces of a watershed is known as
dryfall and wetfall. Within each of these two general categories, a number of specific processes
can be identified, depending on the particular contaminant.
Dry deposition consists of the settling of atmospheric particles and the condensation and
adsorption of trace gases onto the watershed surface. The settling velocity of a particle is a
function of many factors, including the particle mass, its surface area, and the ambient air
conditions. While gravity plays a key role in the deposition of larger particles, smaller particles
(<2.5 /xm) are deposited as a result of turbulent mixing bringing me particles into direct contact
with the surface on which they are being deposited.
Wet deposition or washout involves water vapor, rain, or solid phase water (snow, ice) as a
medium to transport atmospheric particles and condensed vapor to the land or water surface.
Some studies have found the deposition of PCBs to be dominated by wet deposition (Eisenreich
etal., 1981).
A number of studies have found that, overall, dry deposition dominates the input of
contaminants to the aquatic environment (McVeety and Kites, 1988; Webber, 1986). This
cannot be assumed to be the case during the wet winter months in the Pacific Northwest.
Accurately estimating deposition rates is exceedingly difficult because of shortfalls in our
understanding of the underlying physics, incomplete characterization of the particles and their
distribution, and insufficient empirical evidence (due to relatively primitive sampling
techniques).
Transport Processes
The question of how much airborne toxic material reaches a particular water body is difficult
to answer quantitatively. There are large uncertainties beyond the deposition rates, including
how much of the airborne toxic chemicals deposited on the land finds its way to the water.
Three of the processes affecting the fate of airborne contaminants deposited on land are
discussed below: surface water transport, groundwater transport, and fugitive dust.
Surface Water Transport:
Both bottom sediments and suspended matter in rivers contain significantly more trace metals
than are found in the dissolved phase (Horowitz, 1984). Equilibrium between metals in solution
and in the sorbed phase is attained on the order of seconds to minutes for most metals. Binding
of metals to the sediment is a consequence of sediment surface chemistry reactions. In general,
highest metals concentrations are associated with fine sediments having a large surface area to
-------
Chapter 1. Introduction and Background
Figure 1-1. Formation and Deposition of Fine and Coarse Particles (courtesy of R.
Stevens)
CHEMICAL CONVERSION
OF GASES TO
m
rr
i-
eo
a
iu
O
>
LOW VOLATILITY
VAPOR
HOMOGENEOUS
NUCLEATtON
L
CONDENSATION
GROWTH
OF NUCLEI
DROPLETS
J
COAGULATION
*
EROSION
0.01 0.10 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
100
I
WIND BLOWN DUST
&
EMISSIONS
RAINOUT
AND
WASHOUT
PARTICLE DIAMETER. MICROMETERS
FINE PARTICULATES »• : *— COARSE PARTICULATES •*
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
volume ratio. Metals in surface water runoff, like metals in rivers, are largely transported
while they are sorbed to suspended and bed sediments.
PAHs also generally tend to sorb to paniculate matter in the aquatic environment. Because of
their low aqueous solubilities, sorption of PAHs in natural waters occurs rapidly (Neff, 1979);
however sorptive equilibrium takes longer to attain for PAHs than for metals, and may not be
attained with overlying waters before transport occurs in rivers and streams.
The partitioning of metals and PAHs between the particulate and dissolved phase in surface
waters can have a large effect on their transport. In urban areas fine sediment particles can be
readily transported by the flow regimes normally encountered in drainage and sewer pipes. In
rural areas or developing areas, significant removal of sediment-bound contaminants can occur
through a combination of settling and filtration of particulates from overland flow and shallow,
concentrated flow, especially if the ground cover consists of fine-leaved vegetation.
Photolysis (chemical breakdown in the presence of light) can significantly reduce the levels of
PAHs present on dry watershed surfaces and at shallow depths in surface waters.
Environmental factors influence the rate of photolysis, and photolysis rates also vary among
PAHs. Biodegradation (breakdown by biological activity) could also significantly degrade PAHs
if the period between washoff events is on the order of several days. Many of the processes
that act to degrade PAHs can produce intermediate compounds and end products that may be
more harmful than the parent PAH compounds.
Groundwater Transport:
PAH concentrations in groundwater are about one-tenth the concentrations found in most rivers
in the United States (Neff, 1979), due to both the increased time of contact between PAHs and
sediments and biodegradation. Groundwater flow velocities are much lower than those
associated with either riverine or overland flow, allowing more time for PAHs to reach
equilibrium with the bulk water. In addition, in the case of groundwater, sediment sorption
sites are generally not limiting.
In situ biodegradation of toxic organic compounds by soil bacteria and fungi is well
documented. Under proper conditions of moisture, pH, temperature, and nutrients, significant
degradation of- a variety of organic compounds, including PAHs, has been demonstrated.
Except for biodegradation, all of the processes listed above, which act to reduce concentrations
of PAHs in groundwater, also act to reduce the concentrations of metals in groundwater relative
to surface water.
Fugitive Dust Transport:
Airborne particles that settle on land, become available for the fugitive dust load in the
watershed. Some of these particles may agglomerate, forming coarse particles which require
higher wind speeds for resuspension. Lead, zinc, chromium, copper, nickel, arsenic, cadmium,
and beryllium have been detected in street dust samples from both residential and commercial
areas (Cooper et al., 1985).
PAHs associated with regional (area-wide) aerosols differ in their sources and their dominant
chemistry from PAHs associated with street dust. PAHs associated with regional aerosols tend
to be associated with very small soot particles, less than or equal to 1 pm in diameter (Neff,
1979; Pederson et al., 1980, Thomas et al., 1968), and therefore can travel hundreds or even
thousands of miles before they are deposited. Some of the fugitive dust particles in which
PAHs are found are several orders of magnitude larger than the fine airborne particulate matter
-------
Chapter 1. Introduction and Background
and therefore tend to be transported to areas relatively close to the point where they last
resided. Regional aerosol PAHs are largely composed of combustion-derived PAHs, while
PAHs associated with street dust are largely composed of higher molecular weight compounds
thought to originate from uncombusted petroleum leaked from automobiles (Galvin and Moore,
1984).
Local wind gusts repeatedly suspend and redistribute fugitive dust until the dust particles reach
some short- or long-term resting place, such as incorporation into larger, relatively immobile
particles or deposition onto the surface of a nearby body of water. PAHs may be volatilized
or degraded during this local, wind-blown transport.
-------
Chapter 2. Study Plan
The following discussion presents the overall plan for the atmospheric deposition study and a
brief summary of the sampling and analytical scheme for each study component. Chapters 3
- 7 provide details on the methodology and results for the discrete study components.
STUDY OBJECTIVES
* Develop a better understanding of the importance of atmospheric deposition relative to
other inputs of toxic contaminants to Commencement Bay.
* Develop efficient and cost-effective tools for assessing this question in other reaches and
embayments of Puget Sound, and other water bodies as well.
Atmospheric deposition of concern in this study includes both direct and indirect (deposition on
land and subsequent runoff) inputs to Puget Sound. Both point sources (industrial emissions)
and area or regional sources (woodstoves, cars, etc.) are included.
STUDY AREA SELECTION
To make the multi-component study tractable, a single site was selected for intensive
investigation. The Technical Work Team evaluated a matrix of information pertaining to eight
embayments in Puget Sound and selected Commencement Bay near Tacoma, Washington
(Figure 2-1), as the most appropriate study area for the following reasons:
* Commencement Bay and its watershed is one of the most studied embayments in Puget
Sound. Therefore, databases exist for this embayment for ambient air quality, air
emissions, municipal/industrial discharges, wind speeds and directions, and marine sediment
quality.
* The matrix of air pollution sources is complex and diverse, including both large industrial
point sources and pervasive area sources, such as cars and woodstoves.
>• Both lead and combustion-derived PAHs have been measured at high concentrations in the
sediments and the sea-surface microlayer in this embayment.
*• Commencement Bay is a Superfund site with numerous environmental problems and
elevated chemical levels on land and in bottom sediment.
Commencement Bay therefore represents a "worst case" test area for determining if cross-
media (air to water) transfer of toxic contaminants is significant.
TARGET CHEMICALS
The study design was optimized for sampling and analysis of metals (e.g. lead, zinc, arsenic,
manganese), and combustion PAHs (three- to five-ring compounds). Low molecular weight
PAHs, PCBs, aliphatic hydrocarbons, and nutrients were also sampled and analyzed. Metals
and combustion PAHs were targeted because:
* Lead, some other metals, and, combustion-derived PAHs originate from both large
individual sources, such as industrial and commercial emissions, and pervasive sources,
such as cars and woodstoves.
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 2-1. Commencement Bay in Puget Sound, Washington
I
*
Victoria \
•4
Port Angeles
Port
Townsend
Area of
•• ' detail ,
CANADA
USA
Bellingham
Anadqrtes
Everett
Seattle
Commencement
Bay
Tacoma
Olympia
10
-------
Chapter!. Study Plan
* Based on previous studies, atmospheric sources have been estimated to contribute significant
amounts of these chemicals to Puget Sound and elsewhere.
* Metals and PAHs have been shown to be toxic to marine ecosystems and human health.
* Both lead and combustion-derived PAHs have been measured at high concentrations in the
sediments and sea-surface microlayer in Puget Sound.
STUDY COMPONENTS
The study consisted of four field components:
* Six-month meteorological data acquisition;
> Six-month aerosol and deposition sampling;
* 18-day field sampling for the receptor modeling; and
* Source testing of Simpson Tacoma Kraft pulp mill.
Two modeling efforts complemented the field studies and provided the tools for interpreting
and extrapolating from the field study results:
* Receptor modeling; and
* Diffusion/transport modeling (which also included a mobilization coefficient model).
In addition an emission inventory was prepared compiling previous direct measurements and
estimates for both point sources and area sources.
The field sampling network was established in the early summer of 1989 and data were
collected from July 1989 through early January 1990. The 18-day study was performed in
December 1989 and January 1990. The source test was also performed during January 1990.
The intent in selecting this six-month period was to encompass a range of meteorological
conditions (precipitation, temperature, winds) and emission scenarios such that results could be
used to characterize and extrapolate to other unsampled portions of the year.
Each study component focused on a specific particle size fraction or aspect of atmospheric
deposition. These distinctions will be explained below and will again be highlighted in Chapter
8, which compares and integrates the results from all the studies.
The scientists responsible for conducting each study component are listed at the front of this
report.
Sampling Network
Figure 2-2 presents the network of sampling sites that was set up in the Tacoma Tideflats. The
grid of stations included one transect across the industrial waterfront, roughly southwest to
northeast [Morse Industrial Supply (MS), Sea-Land Service, Inc. (SL), and Alexander Avenue
(AS)]; and another transect along the path of the prevailing wind through the area (southeast
to northwest), starting upwind of the industrial area in the Puyallup River and extending to a
site on the Puget Sound shoreline well downwind of the industrial area [Riverside School (RS),
Tyee Marina (TM), Brown's Point (BP)]. The far upwind and downwind sites (RS and BP)
were selected to provide an indication of background contaminant levels, while the industrialized
sites were chosen to represent contaminant levels within a heavily industrialized area.
11
-------
FIGURE 2-2
AEROSOL and DEPOSITION
SAMPLING SITES
Tacoma, Washington
SAMPLING SITES
® Brown's Point
® Tyee Marina
® Morse Industrial Supply
® Sea-Land Services, Inc.
® Alexander Site
® Riverside School
OTHER FEATURES
re] Fire Station No. 12
MAJOR INDUSTRIAL SOURCE AREAS
Kaiser Aluminum & Chemical Corp.
Simpson Tacoma Kraft Company
LINE FEATURES
Major Roads
Other Roads
A/1 Shorelines
A/ Streams
DATA SOURCES
Roads: VS. Geological Survey Digital Line Graph (DLG)
Transportation Data, original scale klOO.OW
Parcek Citv of Tacoma
Shorelines: National Oceanic and Atmospheric Administration,
National Ocean Service Nautical Charting Division
-------
Tft
ft-
a nrno
, d VH-H
1 4 i '-J ! i L_j
bv -
X ,- \
V
"\
A
-, v.—f
!<•"?
Li h
J
Commencement
!
i
p
IIL
-+-H-
j
f *
"S
/LJJJ II i.
\
\
Miles
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
To evaluate deposition to Commencement Bay, stations were located either near or over the
water. The sampling instruments at the Morse Industrial Supply, Sea-Land, Alexander Avenue,
and Riverside School were located on buildings. The Brown's Point instrument was located on
the ground floor of a lighthouse, and the Tyee Marina sampling site was located on top of a
grounded barge that is used by the marina for a breakwater. The height of the samplers off the
ground or water surface is presented in Table 2-1.
Table 2-1. Height of Samplers Above Surface
Sampling Site Height Above Surface
(feet)
Brown's Point 3
Tyee Marina 12
Morse Industrial Supply 20
Sea-Land 30
Alexander Avenue 10
Riverside School 12
To acquire representative samples of ambient air in the Tideflats area, stations were located at
least several blocks away from major point sources such as pulp mills, aluminum refineries, and
oil refineries. Figure 2-3 shows some of the industrial sites with air toxic emissions in the
Tacoma Tideflats. The map only includes those industrial sources required to report under
Section 313 of the Emergency Planning and Community Right-to-Know Act (EPCRA) of 1986.
(Reporting is required if the source manufactures, imports, processes, or uses certain toxic
chemicals in amounts greater than a specified threshold quantity. Fugitive emissions are not
included in the emissions indicated on the map.)
Two PSAPCA monitoring stations are located within the region of this sampling network: Fire
Station #12, which samples PM10, TSP, PM.,5, PM10 (nephelometer), and wind speed and
direction; and the Alexander Avenue site, where PM10 and wind speed and direction are
measured. Samples from these stations were used during the study for comparative data checks
and to provide a context for comparing conditions-encountered during the study period against
those occurring in other seasons and years.
Meteorological Data Collection
An integral part of the overall monitoring program was the collection of meteorological data
in the Tideflats during the entire six-month study period. Wind speed and direction,
precipitation, temperature, and solar radiation data were collected for input to the
diffusion/transport model and to provide information on the ambient environment for
interpretation of the other field study results. A meteorological tower was located atop the one-
story PSAPCA monitoring trailer at the Alexander Avenue site.
The meteorological measurements were recorded over two different averaging times. Five-
minute averages were recorded for wind speed, standard deviation of wind speed, wind
direction, standard deviation of wind direction, and precipitation. Hourly averages included
temperature at two meters above the ground, temperature difference (the temperature at ten
meters minus the temperature at two meters), and solar radiation.
14
-------
Figure 2-3.
TACOMA TIDEFLATS
Stack Air Emissions
Toxics Release Inventory
Ma submitted to EPA by facilities
regulated under the Emergency Planning
& Community Right-h-Kwo Act
Pounds Released in 1989
Q 1 - 1000
r\ 1,001 - 10,000
\w?
10,001 - 100,000
100.001 - 1,000.000
DRAFT:
facility locations approximate
Scale 1 : 51150
0 300 400 tOO 100 I DM
^EZS9
Meters
1
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Six-Month Aerosol and Deposition Sampling
The objective of this study was to conduct long-term (six-month) measurements of the ambient
aerosol and the deposition of atmospheric contaminants to sites adjacent to Commencement Bay.
Aerosol sampling was conducted at all six of the sampling network sites, although not all sites
were sampled during each time period. Deposition sampling was conducted at five of the sites
(Brown's Point, the background aerosol site, was not used for deposition).
Aerosol sampling collected vapor and particles up to a maximum diameter of approximately
50 /xm. Deposition sampling had no upper size limit.
Sampling intervals and frequencies were determined as a function of the following factors: the
number of chemical analyses that could be run within the project budget, the minimum required
temporal resolution of the data, and the chemical detection limits of the analyses. The sampling
schemes were as follows:
>• Deposition sampling—samples collected every two weeks at each of the five sites during the
entire six-month study period.
» Intensive aerosol sampling-samples collected twice per week (three- and four-day samples),
at five deposition sites (Brown's Point not included) during two sampling periods: July 20
to August 24, 1989 ~ primarily dry weather; and November 16 to December 21, 1989 -
primarily wet weather.
* Nonintensive aerosol sampling—samples collected twice per week, at two primary sites
(Alexander Avenue in the industrial area and Tyee Marina, downwind of the industrial area)
and Brown's Point during the remainder of the six-month study period.
The aerosol sampling was timed so that it ran in parallel with the deposition sampling.
All of the aerosol sample filters collected were weighed to determine total mass. Due to budget
constraints it was necessary to limit the number of samples that were chemically analyzed. The
samples selected were analyzed for metals and paniculate PAHs. In addition, ten of the aerosol
samples were analyzed for both vapor and paniculate PAHs, PCBs, and aliphatic compounds.
Selected deposition samples were analyzed for metals, paniculate PAHs, and nutrients. Ten
of the deposition samples were also analyzed for PCBs and aliphatic compounds.
Field Study for Receptor Modeling
This study was comprised of short-term aerosol sampling focused explicitly on the PM10 fraction
in the ambient air. These data were needed as input to the receptor modeling. Separate
samples were obtained of two size fractions of particles: fine (less than 2.5 pm) and coarse (2.5
- 10 /tm). To allow comparisons between the six-month aerosol and deposition monitoring and
the receptor modeling, two of the six-month sites were selected as the sites for acquiring
ambient receptor modeling data. One site was set up at Morse Industrial Supply; the second
receptor model sampling site was located at Alexander Avenue. Two dichotomous samplers
(capable of the size-fractionation noted above) were located at each site. In addition, a single
volatile organic compound (VOC) canister sampling unit was located at each site.
Concurrent ambient air sampling for PM10 particles and volatile organic compounds was
conducted for 18 days, split into two sampling periods: 7 a.m., December 5, 1989 to 7 a.m.,
December 16 1989; and 7 a.m., January 2, 1990 to 7 a.m., January 9, 1990. Twelve-hour
sampling was conducted for daytime (7 a.m. to 7 p.m.) and nighttime (7 p.m. to 7 a.m.)
16
-------
Chapter 2. Study Plan
periods. During the December sampling period, the Simpson Tacoma Kraft plant was not fully
operating due to routine maintenance as well as unscheduled downtime. Because of Simpson's
schedule, there was no ambient sampling for the receptor modeling study for the period from
December 16, 1989, to January 1, 1990.
All samples were analyzed for total mass, trace metals and other elements (silicon, sulfur,
bromine, chlorine, calcium, potassium). The fine fraction (<2.5 j*m) was selected for further
chemical analysis (organic carbon, elemental carbon) and was the focus of the receptor
modeling, because it constitutes those particles that are transported most broadly in the
atmosphere (thereby allowing for area-wide modeling) and because it is made up primarily of
combustion products from anthropogenic activities.
Emission Inventory
To provide input data for the diffusion/transport and receptor modeling, a detailed emission
inventory was compiled for the Tideflats area. Information was acquired from PSAPCA's
registration files for point sources, the source profile library, and relevant source tests.
Emission data for the metals and PAHs of interest for this study were acquired for point sources
(industrial emissions) and area sources (mobile sources such as cars, trucks, ships; woodstoves;
and road dust).
Source Test of Simpson Tacoma Kraft Pulp Mill
The Simpson Tacoma Kraft pulp mill is the major point source of particles less than 10 jtm in
diameter (PM10) in the Tideflats. Prior to this study, source-specific data on the emissions of
metals and PAHs were not available. In order to gather this data, a source test was performed
on the Simpson facility.
Paniculate samples were collected from the hogged fuel boiler, one of the lime kilns and the
No. 3 recovery furnace. Each source test included three separate runs. Testing was
accomplished between January 2, 1990, and January 6, 1990. Laboratory analyses were
conducted on the samples to determine the relative concentrations of metals and PAHs.
Receptor Modeling
Receptor or chemical mass balance (CMB) modeling was used to relate the ambient sample data
to the potential point and area emission sources. CMB modeling is a mathematical method that
attempts to match patterns of chemicals released from sources with ambient patterns collected
on filters or in canisters (presumably representing multiple sources) at the receptor site. The
mathematical model apportions the total amount of chemicals that are collected to the various
sources. Unless source-specific information (such as a unique tracer) is available, the CMB can
only distinguish source types, not individual culpable sources. This type of model does not
address spatial variations unless performed at a number of sites. One of the advantages of the
model, however, is that it does not require meteorological or stack parameter data (e.g. height,
diameter, and emission velocity and temperature).
In order to use this type of model, the sources that may be identified as contributors must be
known and chemically characterized (i.e., the chemical pattern or "fingerprint" of each source
must be known). Data from the Simpson Tacoma source test, from emission studies conducted
in other regions on area sources, and from the 18-day ambient study were analyzed. Because
carbon was not analyzed in the coarse particles collected by the 18-day study, the fine fraction
was the focus of the receptor modeling. The model was run using three alternative tracers for
vehicle exhaust, and the fine mass was apportioned to the source contributors.
17
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Diffusion/Transport Modeling
Diffusion/transport modeling was used in this study as a means of estimating the aerosol and
deposition concentrations in 91.5 km2 of the Tideflats area, based on the known emissions and
the ambient meteorological conditions. The modeling focused on the PM10 size fraction of the
atmospheric particles, because this was the only size fraction for which there was a detailed
emissions inventory. Existing modeling capability was enhanced in order to develop a useful
set of tools with which to estimate the contribution of atmospheric sources of contaminants to
Commencement Bay. The model developed for this study used the emission inventory, stack
parameters, and meteorological data gathered in the field study to produce maps of aerosol
concentrations.
Additional data were needed to estimate the deposition of contaminants from the aerosol. The
amount of deposition was based on estimates of the rates of transfer for each contaminant from
air to land and from air to water determined from the literature and other field studies. These
rates were added to the model in order to estimate the amount deposited.
It was also necessary to estimate the fraction of each air toxic deposited on the land which
would eventually find its way to the Sound. To do this, theoretical mobilization coefficients
were developed to represent the fraction of land-deposited pollutants that would be carried by
surface water runoff, by groundwater, or through a series of pipes into the waterway.
18
-------
Chapter 3. Meteorological Study
METEOROLOGICAL DATA COLLECTION PROGRAM
The purpose of this study was to obtain a high-quality, representative, and complete
meteorological data base to aid in interpreting and understanding the air quality and deposition
monitoring results. Meteorological data is required for operation of the diffusion/transport
model and is also critical to interpretation of the receptor modeling results.
Monitoring Site
The meteorological sensors were located at a single monitoring station set up at the Alexander
Avenue site atop the one-story PSAPCA monitoring trailer. The site is in a relatively open area
of the Tideflats, between the Blair and Hylebos waterways. The bluff along the northeastern
boundary of the Tideflats is a little more than one kilometer northeast of the site. The area in
the immediate vicinity is free from significant meteorological influences so that the data
collected are considered representative of the Tideflats area. The site is surrounded on three
sides (west, south, and east) by open fields, mostly mixed dirt and gravel, with minimal low
vegetation (grasses and weeds) cover. The Reichhold chemical plant is located to the north
of the site. The area between the site and the Reichhold structures, is an asphalt-covered,
truck-loading area.
Sensors
The wind measurements were made with a Wind Monitor-RE, Model 05701, manufactured by
the R. M. Young Company of Traverse City, Michigan. This sensor is a low-threshold, fast-
response, propeller-vane anemometer. Key specifications of this sensor include a speed
threshold less than 0.2 meters per second (m/s), direction threshold less than 0.4 m/s, a
damping ratio of 0.65, and a maximum delay constant of 1.0 meter. The wind sensor was
mounted on top of a 10-meter tower.
Precipitation was measured by a Texas Electronics tipping bucket rain gauge, Model 525.
Each bucket tip measures 0.01 inch of rainfall, with an accuracy of one percent at a rate of two
inches of rainfall per hour or less. The rain gauge was mounted on a one-meter pole, which
was mounted on top of the monitoring trailer. The gauge was located approximately five
meters from the tower to avoid any potential influence from the trailer and the tower. The
inlet height was 4.5 meters above the ground.
Solar radiation was measured by a Li-Cor silicon-photocell pyranometer, Model 200. Absolute
error is nominally ±3 percent. The sensor response is linear with a maximum deviation of one
percent over the full range. The pyranometer was mounted to the tower on a specially designed
mounting arm. The arm extended approximately one meter from the tower in a southerly
direction in order to avoid any shadow influences. The sensor was about 4.75 meters above
the ground.
Temperatures were measured at two and 10 meters above the ground with Campbell Scientific
Model 107 thermistor probes. The sensors' nominal accuracy is ±0.2°C from -30 to +45°C.
The probes were mounted in the radiation shields attached to the tower. The specific purpose
of the temperature measurements was to obtain measurements of temperature difference (AT)
19
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
in the lower atmosphere. However, it was discovered that this was not an optimum sensor
arrangement for accurate measurements of AT, because the sensors were not electronically
matched to a common temperature reference (as is typically done for AT measurements).
Therefore, the system was unable to achieve the required AT accuracy of 0.1 °C. Data quality
is considered further below.
All meteorological sensors were attached to a Campbell Scientific Model CR10 datalogger. The
datalogger was housed in a weatherproof shelter mounted to the tower. The datalogger
interrogated each sensor once per second, and converted the signals to appropriate units.
Further, the datalogger calculated averages over five minutes or one hour, as appropriate, and
archived the results in its internal memory. The data were summarized into hourly averages
and are listed in Appendix B-l.
Operation
New instruments were obtained for this meteorological data collection program. All sensors
were checked for proper operation. The temperature probes were calibrated together in an ice
bath for a one-point calibration. Installation was completed and monitoring began on July 6,
1989.
After startup of the meteorological instruments, the site was visited weekly to verify the good
condition and proper operation of the instrumentation and to download the data from the internal
datalogger memory to a portable computer. The data were then transported back to the office
where they were permanently archived.
Monitoring was continuous until approximately 10:00 AM PST on January 9, 1990, when the
meteorological instrumentation was dismantled. Data recovery rates for the program were
greater than 99 percent. A few hours of data were invalidated due to performance checks on
the instruments; the invalid data were replaced by a value of -999.
Data Quality
A number of instrument checks were performed during the monitoring program to assess the
adequacy of system performance. The variables that were assessed during the data quality
checks are discussed below.
Wind
Within 10 days of program startup, the wind direction orientation was checked and found to be
in error by 11°. At installation, the orientation of the wind direction sensor was aligned using
a compass. Subsequent checks of orientation with a compass confirmed that the compass was
subject to magnetic interference. Thus, the sensor had been misaligned, and the wind direction
data were subject to a systematic 11° error. The sensor was realigned to correct the error.
Two weeks later, the wind direction orientation was rechecked by an independent auditor using
a solar azimuth method. The orientation was found to be correct within 1° — well within the
system accuracy specification of +.5°. Based on the results of these checks, the wind direction
data for the first week of the program were corrected for the error by adding 11° to each value.
The wind direction orientation was checked again at the end of the program and found to be
within specification. Wind speed zero checks were also performed when the wind direction
checks were performed. Wind speed and direction thresholds were checked halfway into the
study, and again at the end of the program, and were found to be within manufacturer's
specifications.
20
-------
Chapter 3. Meteorological Study
Precipitation
The performance of the rain gauge was checked at the beginning and at the end of the program
by introducing the equivalent of 0.50 inches of rain into the gauge. The performance check
showed the gauge was low by four percent, but still within the system accuracy guideline of
+.10 percent.
Solar Radiation
The pyranometer was calibrated by the manufacturer just before the program began. Normally,
pyranometers only require annual calibration, so no further calibrations were performed. The
only checks on the performance of the pyranometer were zero checks, which were done
frequently throughout the program. In addition, the sensor surface was cleaned regularly during
the weekly site visits.
Temperature
As mentioned above, the instrumentation employed to measure AT was incapable of achieving
the required accuracy. This was determined at the completion of the project after a two-point
calibration was performed, and the results were discussed with the manufacturer. Based on the
last two-point calibration, a linear correction to the lower temperature sensor is possible (with
all temperatures in °C):
T (corrected) = T (unconnected) - (0.16/13.0) • T (uncorrected).
Even with this correction, however, the AT measurements are not considered accurate to within
+.0.1 °C because the thermistor probes were not matched to a common temperature reference,
and the accuracy of the individual sensors was ±0.2°C.
Another problem was encountered with the temperature measurements during the program. The
instantaneous readings from the upper sensor were observed to fluctuate within a range of about
0.0 - +0.5°C. The problem was never solved, even after considerable troubleshooting efforts,
including replacement of the top thermistor probe. The effect of this fluctuation is to bias the
upper (10 m) temperature measurements on the high side by approximately 0.1 °C. As a result
of the above problems, and even though the values appear reasonable, the AT measurements
must be considered outside the range of accuracy desired for this program.
Results
A listing of the hourly meteorological data is presented in Appendix B-l. A summary of daily
wind, solar insolation, and precipitation data is found in Appendix B-2.
In addition, Appendix E-5 contains 12-hour averages of meteorological data for the December
5 to 15 and January 2 to 8 sampling intervals (to coincide with the 18-day PM10 ambient
monitoring study). How conditions during the study period compare with previous years is
addressed in Chapter 8, Comparison of Studies.
Of particular interest for interpreting monitoring and modeling results, is a characterization of
the study period in terms of air stagnation episodes. To do this analysis both surface and
upper air meteorological data are required. Data collected at Quillayute, Washington, were
used, as this is the closest upper air station to the Puget Sound area which would be
representative of Puget Sound conditions. The methodology used to categorize air stagnation
21
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
episodes was based on a method developed by Holzworth (1972) and modified by Region 10
EPA to incorporate a method for computing mixing heights in marine climates.
Holzworth's method categorizes air stagnation episodes in terms of urban air pollution potential.
The more severe the episode, the more potential for urban air pollution. Holzworth's model
uses three meteorological parameters to predict the severity of stagnation events:
* Mixing height (depth of layer through which pollutants can mix);
* Average wind speed in mixing layer (sometimes called ventilation); and
> Precipitation (any precipitation event automatically terminates an episode).
Higher pollution potential exists when mixing depths are low and wind speeds are light.
Holzworth's method uses eight categories, with level one being the most severe. (Region 10's
modification added two additional levels, nine and ten.)
Appendix B-3 includes tables showing level 1,2, and 3 air stagnation episodes during the entire
study period. Appendix B-4 contains the meteorological data for the analysis, including actual
mixing depths, wind speeds, and precipitation events. Figures 3-1 (a) and (b) display the
results of the analysis for the November 1989 through January 1990 portion of the study period,
when there was the greatest potential for air stagnation episodes. Two points (a.m.: midnight-
noon; p.m.: noon-midnight) are plotted for each date. If the air stagnation level was constant
over that day, only one point was plotted.
The only episode of category 1 was of 48 hours duration and occurred from the afternoon of
December 25, 1989, through the morning of December 27, 1989. The most severe episode in
terms of duration was a category two of 180 hours length which occurred from the afternoon
of December 11, 1989 until the morning of December 19, 1989. At category three this episode
began 12 hours earlier, with a total length of 192 hours. The only episode in the summer was
a category two of 36 hours length, from the morning of July 26, 1989, through the morning
of July 27, 1989.
22
-------
Chapter 3. Meteorological Study
Figure 3-1 (a). Air Stagnation Episodes (November - December)
7
SOLID LINGS REPRESENT STAGNATION SEVERITY
1 CMOST SEVERE} THROUGH 10 CLEAST SEVERE}
FIGURE OKIE
Figure 3-l(b), Air Stagnation Episodes (December - January)
SOLID LINES RCPneSCMT STAONATION SEVER I TY
1 CMOST SBvene? THROUGH -10 c >- CAST sevene?
p i OURE TWO
23
-------
Chapter 4. Six-Month Aerosol and Deposition Study
OBJECTIVES
* Establish a network of stations with co-located samplers for monitoring ambient aerosol
and the deposition of atmospheric contaminants;
* Monitor continuously for six months (July 1989 to January 1990); and
> Analyze samples for total mass and specific chemical constituents (metals, PAHs, PCBs,
aliphatic hydrocarbons, and nutrients).
SAMPLING NETWORK
Chapter 2 contains a description of the six sites established for aerosol sampling-Morse
Industrial Supply (MS), Sea-Land Service, Inc. (SL), Alexander Avenue (AS), Riverside School
(RS), Tyee Marina (TM), and Brown's Point (BP)--and a map illustrating station placement
(Figure 2-2.). Aerosol sampling was split into two five-week intensive periods (July 20 to
August 24, 1989-primarily dry weather; and November 16 to December 21-primarily wet
weather) when all stations except Brown's Point were sampled; and a non-intensive sampling
scheme for the remainder of the six months when samples were only collected at Alexander
Avenue, Tyee Marina, and Brown's Point (background reference station). Aerosol samples
were collected twice per week (one three-day and one four-day sample). Deposition samples
were collected every two weeks at all of the above sites except Brown's Point (Table 4-1).
Table 4-1. Aerosol and Deposition Sampling Plan
Sampling
Sample Type Frequency Locations Length of Study
Aerosol 2 per week AS, MS, SL, 10 weeks
(intensive study) TM, RS
Aerosol 2 per week AS, TM, BP 16 weeks
(non-intensive)
Deposition 2 per month AS, MS, SL 6 1/2 months
TM, RS, + one
duplicate
EXPERIMENTAL METHODS
Aerosol Sampling
Aerosol samples were collected with General Metal Works Model PS-1 aerosol samplers (Figure
4-1). The upper size limit of particles captured by the PS-1 sampler is 25 to 50 fj.m, depending
on the wind speed. The samplers were equipped with 102 mm-diameter quartz (high purity)
filters and backed up with 6 cm-diameter, 8 cm-long polyurethane foam plugs (PUFs) for
collection of organic vapor. The quartz filters were used because they work well for
particulates and organics and are resistant to clogging.
25
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 4-1. PS-1 Aerosol Sampler^
Magnehelic
Gauge
0-100 in.
Exhaust
Ouct
(6 in. x 10 ft)
Voltage Variator
Elapsed Time Meter
26
-------
Chapter 4. Six-Month Aerosol and Deposition Study
Air was drawn through the sampler at about 0.25 m3 per minute and the sampling period was
three or four days (twice weekly). The sampling period and flow rate were chosen to provide
adequate sample for quantitative chemical analyses without overloading the air pump or the
filters. (During periods of air stagnation alerts, filters were changed daily to avoid plugging.)
This sampling frequency minimized the number of samples to be collected over the six-month
continuous sampling period. PS-1 samplers were operated following standard operating
procedures prescribed by EPA Region 10.
The PUFs and filters were cleaned before use following methods taken from the U.S. EPA
(1988) Compendium Method TO-13. All PUFs were cleaned with solvent (pesticide grade
acetone) in a Soxhlet extractor. Filters were cleaned in n^ethylene chloride, dried, and ashed.
An amplified discussion of these procedures is presented in Appendix C-l.
Filters were conditioned and weighed before field use following procedures in the EPA guidance
document for PM10 (U.S. EPA, 1987a). Clean filters were conditioned for 24 hours at 20°
during the summer and in a desiccator at 4°C in the fall, then weighed within 30 seconds after
removal from the desiccator. In the field, filters and PUFs were handled only with clean nylon
gloves and acid-cleaned, solvent-rinsed, Teflon-tipped forceps. Quartz filters and the PUFs
collected from the field were stored individually in a low-temperature (-70 °C) freezer until
analyzed.
Deposition Sampling
A large water-filled Pyrex pan with a surface area of 0.12 m2 was used as a deposition collector
(Figure 4-2). The pan contained a 1-cm layer of high purity water, simulating the surface of
Puget Sound. The water prevented dry deposited material from blowing out of the pan during
dry windy weather. A reservoir of water was used to maintain a constant level of water in the
pan in dry weather and an overflow drain and reservoir collected the excess water in wet
weather. The contents in the deposition pan were transferred biweekly to the deposition
reservoir container and taken to the laboratory for storage in a refrigerator. Deposition samples
were acidified to a pH of less than two with hydrochloric acid and held at 4°C in the dark until
analyzed. All deposition sampling equipment was made of glass or Teflon. See Appendix
C-l for cleaning and handling procedures.
Each pan was located on a 60 cm high table to reduce the input of resuspended dust and soil.
Vertical glass rods surrounded the pan to keep birds off and eliminate contamination from bird
droppings. The overflow reservoir was located under the table. The table was located near the
aerosol sampler, but protected from possible contamination from the aerosol sampling motor
by a plastic hose that directed the exhaust away from the deposition collector. Materials such
as feathers, leaves, and large bugs were removed with clean forceps.
Chemical Analysis of Aerosol and Deposition Samples
The particulate load on filters was determined by accurately weighing the air filter before and
after use. The filters were conditioned in a desiccator before weighing to prevent weighing
errors due to humidity. During the fall and winter, filters were conditioned at 4°C instead of
20 °C to reduce the loss of volatile hydrocarbons.
Filters were subsampled for metals and organic compounds after being cut into pie-shaped
sections with a solvent-cleaned knife on a clean Teflon surface. These subsections were
weighed so that the percentage of the particulate load on the subsection could be calculated.
27
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 4-2. Deposition Sampler
Water Reservoir
Inside
Glass Rods
! i Collection
1 ' Bottle
Overt low Tube
28
-------
Chapter 4. Six-Month Aerosol and Deposition Study
Extraction procedures and analytical methods used are summarized in Table 4-2. Air filters
were analyzed for sulfur (S), chlorine (Cl), potassium (K), calcium (Ca), bromine (Br), titanium
(Ti), chromium (Cr), manganese (Mn), iron (Fe), vanadium (V), nickel (Ni), lead (Pb), arsenic
(As), selenium (Se), copper (Cu), and zinc (Zn). The minimum quantifiable limit was about
0.5 ng/m3, which was usually sensitive enough to quantify most of the elements on the air
filters. The minimum quantifiable limit for metals in water was about 1 /xg/L. The metals that
were quantified in the deposition samples included As, Cu, Pb, Zn, Ni, Mn, and Cr.
Quality control samples for metals included a field blank, a certified water sample, and a spike
blank sample containing metals concentrations comparable to the deposition samples. These
samples were also acidified and stored, and metal recoveries determined. Precision was
estimated from field duplicate samples.
Approximately 17 individual PAH compounds (Table 4-3) were quantified from filter, PUF, and
deposition samples. Before the filters, PUFs, and deposition samples were extracted, a known
amount of three deuterated PAH surrogates was added to each sample to determine method
recovery. Detection limits were about 0.1 ng/m3.
PAH compounds lower in molecular weight than phenanthrene exist primarily in the vapor
phase. While the PUF is intended to trap vapor phase compounds, studies have documented
significant sample breakthrough (escape through the PUF) as a function of vapor pressure,
temperature, and volume of air sampled (You and Bidleman, 1984; Chuang et al., 1987).
Therefore, ambient concentrations of low molecular weight compounds are likely to be
underestimated by the methods and sampling schemes used in this study. However, measured
concentrations of the 11 PAH compounds, anthracene through benzo(g,h,i)perylene, are
expected to be representative of the ambient concentrations.
Extracts were analyzed for selected PCB congeners (18, 33, 49, 52, 77, 97, 101, 105, 118,
126, 138, 149, 169, 180, 187, 194, and 195). These congeners were chosen because they span
a wide range of chlorination, molecular weight, solubility, and vapor pressure, and also because
they are prominent in aerosol samples and contribute significantly to the mass of commercially
sold PCBs (Aroclors). Surrogates and injection standards were used to verify analytical
recoveries.
Sample Selection
Due to budget constraints, it was necessary to limit the number of sample analyses. All of the
aerosol samples collected were weighed for determination of total mass. The Alexander Avenue
site was selected for the most comprehensive analysis, because it has a substantial record of
existing data from the PSAPCA monitoring co-located there, and because it is the site where
extensive meteorological measurements were made during this study. Aerosol samples from
the two highest, the two lowest, and the two weights closest to the mean were analyzed from
the Alexander Avenue site for each of the intensive monitoring periods (July 20 to August 24,
and November 16 to December 24) and also the non-intensive period. The selection of these
samples determined the set of dates from which aerosol samples were selected from the
remaining sites. Time periods corresponding to the intensive aerosol monitoring were chosen
for the deposition sample analysis. The sample sites and dates selected for analysis are
contained in tables in Appendix C-2.
29
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 4-2. Extraction Procedures and Analytical Techniques for Aerosol and Deposition
Samples
Sample
Chemical Extraction Method
Chemical
Analysis Technique
AEROSOL
Filter
DEPOSITION
Metals
PUF & Filter Organics
-PAHs
-PCBs
Methylene chloride (filter)
Hexane-Ether (PUF)
(Compendium Method TO-13,
EPA, 1988)
-Aliphatic
Hydrocarbons
Metals
Organics
Acid-digested for total
recoverable metals
(Method 3005, EPA, 1986c)
Methylene chloride
(Method 610)
Energy-dispersive x-ray
fluorescence (XRF)
(Nielson, 1977)
Gas chromatography/
mass spectrometry
(GC/MS)
Gas chromatograph/
electron capture
detection (GC/ECD)
(similar to Method 608)
Gas chromatograph/
flame ionization
detection (GC/FID)
Atomic absorption
flame spectropho-
tometry (EPA, 1986c)
(same as above for
aerosol samples)
Table 4-3. Study PAHs (listed in order of ascending molecular weight)
Naphthalene
Acenaphthylene
Acenaphthene
Fluorene
Dibenzothiophene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Benzo(b)fluoranthene
Benzo(k)fluoranthene
Benzo(a)pyrene
Indeno(l ,2,3-c,d)pyrene
Dibenzo(a,h)anthracene
Benzo(g, h, i)perylene
30
-------
Chapter 4. Six-Month Aerosol and Deposition Study
In addition the following samples were analyzed specifically to assist in comparison and
interpretation of the results of various studies:
* Samples collected from November 27 to 30 from the Alexander Avenue and Riverside
School sites: these samples were taken during the highest aerosol readings that occurred
during the study. The samples were analyzed to determine from the chemical patterns
which sources were the major contributors to the high readings.
> Samples collected from September 7 to 11 and September 14 to 18 from the Alexander
Avenue site: these samples were analyzed to composite the results with the September 11
to 14 and September 18 to 21 samples to compare the aerosol and deposition measurements
for the same time period during a two-week dry period.
* Samples from November 2 to 6 and November 13 to 16 from the Alexander Avenue site:
these samples were analyzed to composite the results with the November 6 to 9 and
November 9 to 13 samples to compare the aerosol and deposition measurements for the
same time period during a two-week period of heavy and continuous rain.
* November 30 to December 4, December 4 to 7, December 7 to 11, and December 12 to
13 samples were analyzed to composite their results with the December 11 to 12 and
December 13 to 14 samples. This was done to compare the results with the aerosol results
from the receptor modeling study.
Numbers of samples collected and analyzed are summarized in Table 4-4.
Table 4-4. Aerosol and Deposition Sample Analysis Summary
Samples Samples
Sample Type Locations Collected Analyzed
Aerosol AS, TM, SL, MS, RS 140 45
(intensive study)
Aerosol AS, TM, BP 93 16
(non-intensive)
Deposition AS, TM, SL, MS, RS 78 50
-I- one duplicate
RESULTS AND DISCUSSION
Aerosol Results
Particulates
The concentrations of suspended particulates were usually in the range of 20 to 100 /*g/m3. The
data for each air filter are presented in Appendix C-3, and the mean concentrations for each site
are presented in Table 4-5.
31
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 4-5. Mean Particulate Concentrations (itg/m3) at Six Tacoma Sites(a)
Site
Alexander
Avenue
(AS)
Tyee
Marina
(TM)
Brown's
Point
(BP)
Sea-Land
(SL)
Morse
Supply
(MS)
Riverside
School
(RS)
Entire Sampling Period
6-29-89 to 1-9-90 62 41 35°° 77(c) 64(c) 50(c)
Summer Intensive Period
7-20-89 to 8-24-89 66 22 No Data 50 42 42
Autumn Intensive Period
11-16-89 to 12-21-89 73 62 No Data 100 82 56
(a) The estimated error in particulate concentrations is approximately +.2.6 percent based on
duplicate samples collected at one site.
(b) Sampling period is 7-10-89 to 11-6-89, and does not include summer intensive period
(c) Summer and autumn intensive periods only
The suspended particle results for Alexander Avenue and Tyee Marina are plotted versus time
in Figure 4-3. The paniculate concentrations at Alexander Avenue were three times higher than
those at Tyee Marina during the summer. During the fall there are similar paniculate patterns
at these two sites. The highest paniculate concentrations occurred during periods of air
stagnation. During the fall intensive sampling period, the two highest paniculate concentrations
for the entire data set were measured at all of the five primary sites for filter periods that began
either on November 27 or December 11. On November 27 all of the filters were clogged due
to the high paniculate concentrations. (To prevent this from happening again, filters were
changed daily during air stagnation periods.) During the periods of highest paniculate
concentrations, similar paniculate concentrations (121 to 174 ^tg/m3) were observed at all five
primary sites.
The difference in paniculate concentrations between Alexander Avenue and Tyee Marina during
the summer may have been due to a massive ditch digging and earth moving construction
project that took place during July and August 1989 within 200 meters of the Alexander Avenue
site. Under dry summer conditions resuspension of dust is more likely to occur. The more
even gradient in paniculate concentrations among all the sites during the period of highest
concentrations in the fall suggests that the samples were more heavily influenced by a regional
source, most likely dominated by fine particles rather than a nearby source of coarse particles.
Elemental Concentrations
The concentrations of elements in air samples appear to be correlated with paniculate
concentrations. The concentrations of elements are generally lowest during wet, windy periods
and highest during air stagnation episodes. The concentrations of 15 elements in the air samples
are presented in Appendix C-4, and the mean concentrations for each site are presented in Table
4-6. High and variable concentrations of copper in air filter samples resulted from air
contamination by the air sampler motor armature, even though a plastic tube was used to divert
motor exhaust away from the sampler. Consequently, the copper data from the air filters were
not usable. Based on the limited number of filters from each site, the pattern is composed of
32
-------
Chapter 4. Six-Month Aerosol and Deposition Study
Figure 4-3. Suspended Particulate (PM^j,,) Concentrations, Alexander Avenue and Tyee
Marina Sites
a, - Alexander Avenue (AS)
t\A/
150
ri
<
E 100
a.
50
n
i-
-
>
T?
« )
> )
s v
S
IS ' ?
rs ' ?
j-y
ita' "
E:l :
?:>
i;ii::l
July
-
*
S
_l
<> > >0 n>
<> > > QM V
<>r * ? < I > > 5
'*'>(!!!)$
' August 5
>
>
>
>
X
>
9 >
• >
i >
> >
> >
y >
> >
> >
» >
>eptember
(C
c < «
c d<
; ' *
il
• Jp *
October
(
^
<
<
n <
I ;5
M
|; |33
November
s
|
>
' E
>§££
^ ^ K R
^ / C £ H
^Eh! n
^ K! ^s
ii ii
December
Filler Daie
so - Tyee Marina (TM)
August ' September October November ' December
Filler Daic
33
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 4-6. Mean Elemental Aerosol Concentrations at Six Tacoma Sites (ng/m3)(a)
Site (Number of Samples)
Element AS (26) TM (9) BP (3) SL (5) MS (6) RS (7)
s
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Ni
Zn
Sew
Pb
As
Br
2600
650
350
1600
100
10
5.6
45
1400
12
94
0.58
39
5.8
7.8
2200
800
200
740
54
9.6
3.3
31
860
9.9
78
0.75
27
2.6
6.2
1400
300
160
540
40
7.6
0.87
18
510
5.9
27
0.68
15
2.7
5.0
2000
380
330
980
100
9.1
6.8
43
1400
10
130
0.7
62
6.6
8.8
2000
470
320
970
84
6.4
8.6
45
1300
10
90
0.6
45
4
7.2
1400
300
310
360
64
8.4
6.8
37
770
9.9
72
0.47
52
3.6
9.3
(a) The estimated error in metal concentration is approximately ±10 percent based on chemical
analyses of duplicate samples collected at one site.
(b) Where selenium (Se) was undetected in samples, the detection limit of about 0.5 ng/m3 was
used to calculate an estimated mean.
34
-------
Chapter 4. Six-Month Aerosol and Deposition Study
higher concentrations of crustal elements (Ca, Fe, K, Mn, and Ti) at the industrial sites (AS,
MS, and SL) and the rural site (RS) than at the marine sites (TM and BP). The Sea-Land site
had the highest lead and zinc concentrations. Crustal metals such as iron and potassium were
generally highest during the summer at the Alexander Avenue site (Figures 4-4 and 4-5).
During the fall air stagnation episodes, the concentrations of lead and zinc at both Alexander
Avenue and Tyee Marina were much higher relative to crustal metals than in the summer
(Figures 4-6 and 4-7).
The probable source of lead and zinc at the Sea-Land site is a facility nearby at Terminal 7
where ore containing these metals is off-loaded from ships and loaded into open rail cars. The
peak in crustal metals concentrations at the Alexander Avenue site may reflect construction and
earth moving activities during dry summer weather.
Particulate PAHs
The concentrations of PAHs collected on air filters were approximately an order of magnitude
higher during air stagnation episodes, when total particulates were high, than during other
periods. The dominant five compounds are fluoranthene, pyrene, chrysene,
benzo(b)fluoranthene, and benzo(a)pyrene. The concentrations of these individual compounds
range from 0.1 to 10 ng/m3. These five compounds are often referred to as combustion PAHs
(CPAHs) because of their probable source. The PAH concentration data for 17 compounds are
presented in Appendix C-5 and the mean concentrations for each site are presented in Table 4-
7. The PAH compounds listed from naphthalene through anthracene are relatively volatile and
usually not detected on air filters and only partially recovered from PUF samples. Because of
the long sample collection period (three to four days) some paniculate PAHs may also have
partitioned off the filter and onto the PUF. The mean total CPAH paniculate concentrations
[sum of the ten compounds, fluoranthene through benzo(g,h,i)perylene] are relatively uniform
among the sites, except for Brown's Point, which has much lower concentrations.
Vapor PAH
The concentrations of volatile PAH compounds, such as naphthalene, acenaphthalene,
acenaphthene, and fluorene in air samples were severely underestimated by the sampling and
analytical methods used in this study. The two main causes are breakthrough of volatile PAHs
on PUF plugs and low recoveries of PAHs during the chemical analyses.
The vapor penetration of PAHs through PUF plugs has been studied by You and Bidleman
(1984). Their results indicate an air sample volume of about 1,000 m3 (as used in this study)
would result in approximately 50 percent breakthrough of phenanthrene in the PUF plugs.
Compounds more volatile than phenanthrene, such as fluorene or naphthalene, will break
through at an air volume of less than 200 m3. Therefore, the concentrations of fluorene and
more volatile PAH compounds will be underestimated for vapor PAHs in this study by at least
50 percent.
The analytical chemistry recovery results for the lower molecular weight PAH compounds were
estimated using two surrogates, naphthalene-d8 and acenaphthene-dlO. The recoveries of the
surrogate for PUF samples were in the range of 10 to 40 percent for this study, indicating that
the lower molecular weight PAH compounds, which are volatile, are partially lost during the
evaporation step of the analysis. Results of surrogate recoveries are in Appendix C-6.
35
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 4-4. Aerosol Iron (Fe) Concentrations, Alexander Avenue and Tyee Marina Sites
Iron - Alexander Avenue (AS)
E
'ci
a.
July August Septemoer Novemoer
Filler Date
Decemoer
Iron - Tyee Marina (TM)
ju'iy August September November
Filter Date
36
December
-------
Chapter 4. Six-Month Aerosol and Deposition Study
Figure 4-5. Aerosol Potassium (K) Concentrations, Alexander Avenue and Tyee Marina
Sites
1.1
l
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Potassium - Alexander Avenue (AS)
E>
July August September November
Filter Date
December
1.2
Potassium - Tyee Marina (TM)
0.8
"Sb
0.6
0.4
0.2
July August September November
Filter Date
37
December
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 4-6. Aerosol Lead (Pb) Concentrations, Alexander Avenue and Tyee Marina Sites
Lead - Alexander Avenue (AS)
0.12
0.1
0.08
E 0.06
"efc
a.
0.04
0.02
July August
September November
Filter Date
December
0.12
0.1
0.08
f>
'E o.06
"Sfc
0.04
0.02
Lead - Tyee Marina (TM)
July August September November
Filler Dale
December
38
-------
Chapter 4. Six-Month Aerosol and Deposition Study
Figure 4-7. Aerosol Zinc (Zn) Concentrations, Alexander Avenue and Tyee Marina Sites
Zinc - Alexander Avenue (AS)
E
a.
O.Jj
0.3
0.25
0.2
0.15
0.1
0.05
It
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
s
July August
September November
Filler Date
December
Zinc - Tyee Marina (TM)
July August
September November
Filler Date
December
39
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 4-7. Mean Particulate PAH and Total Combustion PAH(a) Concentrations at Six
Tacoma Sites fog/m3)""
Site (Number of Samples)
Compound AS (26) TM (5) BP (3) SL (5) MS (6) RS (7)
Naphthalene
Acenaphthalene
Acenaphthene
Fluorene
Dibenzothiophene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Benzo(b)fluoranthene
Benzo(k)fluoranthene
Benzo(a)pyrene
Indeno(l ,2,3-c,d)pyrene
Dibenzo(a, h)anthracene
Benzo(g , h , i)perylene
0.037
ND(C)
ND
0.001
0.005
1.2
0.034
4.4
3.3
3.1
7.3
4.3
2.8
2.7
2.0
0.42
2.5
0.16
ND
ND
ND
ND
0.79
0.082
4.7
4.5
4.2
10.2
5.9
2.1
1.9
1.2
0.46
1.5
0.093
ND
ND
ND
ND
0.11
ND
0.34 .
0.28
0.24
0.84
1.8
0.024
0.23
0.29
0.10
0.40
0.091
ND
ND.
ND
ND
1.2
0.084
4.0
3.8
2.9
5.6
3.9
2.6
2.7
1.9
0.40
2.4
0.22
ND
ND
ND
ND
0.81
ND
3.0
3.0
2.4
4.6
3.4
1.8
2.2
1.5
0.33
2.2
0.11
ND
ND
ND
ND
1.1
ND
3.2
3.5
4.3
6.9
5.1
3.4
4.4
2.7
0.60
3.1
Total CPAHs 32.9 36.8 4.55 30.1 24.4 37.3
(a) Combustion PAH concentrations are the sum of ten compounds (fluoranthene through
benzo (g,h,i) perylene)
(b) Samples usually three to four days, few one-day samples
Sample volume approximately 1,000 m3
(c) ND = Not detected
Given both the sample capture and analytical recovery errors discussed above, the vapor PAH
results presented below must be treated carefully and considered as only qualitative at best. The
biases in the sampling and analysis may mean that as little as 5 - 20% of the original vapor
PAH load was measured.
Despite the sampling and analytical losses the mean concentrations of vapor PAHs measured
are considerably higher than the mean concentrations of particulate PAHs. (As noted above,
the PUF may have sampled some of the particulate PAHs that broke through during the long
sampling period.) The mean total vapor PAH concentrations for all six sites except Brown's
Point are in the range of 101 to 237 ng/m3 (Table 4-8). The data for each PUF sample are
provided in Appendix C-6. Brown's Point concentrations are very low, but include only two
samples. The dominant four compounds are fluorene, phenanthrene, fluoranthene, and pyrene.
The latter two compounds are also present on the filters, however, only at levels one-fifth of
the concentration found in the vapor phase. Compounds that are higher in molecular weight
than pyrene are generally not detected or, if they are detected, they have lower concentrations
in the vapor phase than in the particulate phase.
40
-------
Chapter 4. Six-Month Aerosol and Deposition Study
Individual mean concentrations of the four dominant PAH compounds range between 2 and 124
ng/m3. A plot of monthly average vapor PAH concentrations at the Alexander Avenue and
Tyee Marina sites is shown in Figure 4-8. The highest concentrations at Tyee Marina occurred
during the December air stagnation episodes when the concentrations of particulates, metals,
and paniculate PAHs were highest. However, the highest concentration at the Alexander
Avenue site occurred in August with lower, but fairly similar concentrations in September,
November, and December.
Table 4-8. Mean Vapor PAH Concentrations at Six Tacoma Sites (ng/m3)(l
Site (Number of Samoles)
Compound
Naphthalene
Acenaphthalene
Acenaphthene
Fluorene
Dibenzothiophene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Behzo(b)fluoranthene
Benzo(k)fluoranthene
Benzo(a)pyrene
Indeno(l ,2,3-c,d)pyrene
Dibenzo(a,h)anthracene
Benzo(g,h,i)perylene
Total Vapor PAHs
AS (21)
2.4
2.2
4.3
12
9.2
124
14
40
25
0.84
2.4
0.07
0.01
ND
ND
ND
ND
237
TM(7)
0.54
1.3
3.3
9.2
8.4
101
7.1
29
21
0.62
1.7
0.18
ND00
ND
ND
ND
ND
183
BP(3)
0.63
0.25
0.80
2.2
1.0
27
1.1
13
8.4
0.63
3.1
0.10
ND
ND
ND
ND
ND
55
SL(5)
1.0
2.2
3.1
12
8.1
57
5.0
16
11
0.25
0.66
0.01
0.00
ND
ND
ND
ND
115
MS (6)
1.2
3.0
2.0
9.5
7.8
54
5.6
15
12
0.26
0.51
0.02
ND
ND
ND
ND
ND
110
RS(7)
2.4
5.2
1.7
7.1
3.8
49
5.9
14
11
0.37
0.80
0.10
ND
ND
ND
ND
ND
101
(a) Samples usually three to four days, few one-day samples
Sample volume approximately 1,000 m3
Data not corrected for breakthrough or recovery efficiency
(b) ND = Not detected
41
-------
N)
O
'P
ST
o
500
TOTAL AVERAGE MONTHLY VAPOR PAH, 1989
ALEXANDER AVENUE AND TYEE MARINA, TACOMA
100 H
,n c
ag
JUL
AUG
Alexander Avenue
SEP OCT
MONTH (1989)
NOV
|X\{ Tyee Marina
DEC
o
1
«5"
T3
O
n
o
I
a
I.
o
e
a
63
S
a,
re
re
85
8
a
g
65
ST
O
0
e
re
I
B-
o
re
•O
8.
s^
o*
B
o
a
65
83
I
(-»•
O
2s
JQ
-------
Chapter 4. Six-Month Aerosol and Deposition Study
Particulate and Vapor Aliphatic Hydrocarbons and PCBs
The mean concentrations of paniculate and vapor aliphatic hydrocarbons and PCBs are
summarized in Table 4-9, and the results for the individual filter and PUF plug samples are
included in Appendix C-7. Four air samples collected in July and five air samples collected
in December were analyzed for 27 aliphatic hydrocarbon compounds (AHC, C9 to C36) and
21 PCB congeners. The concentrations of the AHCs and PCBs were frequently below detection
limits for the sample sizes used in this study.
The general pattern is that the total AHCs (vapor and particulate) detected in December, when
air stagnation existed, were approximately four times higher than the total AHCs in July. There
usually were higher concentrations of vapor than particulate AHCs. This could reflect some
partitioning of particulates from the filter onto the PUF.
The total PCB concentration was approximately five times higher in December than in July.
The PCBs were present predominantly in the vapor phase. However, due to breakthrough, the
measured values may be low. Because of the limited number of samples analyzed and the
limited number of congeners that were detected, the data set is not of quantitative value,
however, the proportion of vapor to particulate is consistent with other studies (Chevreuil et al.,
1989; Duinker and Bouchertall, 1989).
Table 4-9. Concentrations of Particulate and Vapor Aliphatic Hydrocarbons (C9 to C36)
and PCBs (21 Congeners) (ng/m3)
AS
TM
July 27
SL
MS
Particulate Aliphatics 43 3 11 19
Vapor Aliphatics 27 130 30 15
Particulate PCBs 0.05 0.01 <0.02 0.04
Vapor PCBs 0.01 0.07 0.17 0.04
December 11
AS TM SL MS RS
Particulate Aliphatics 60 270 170 130 20
Vapor Aliphatics 100 350 600 99 120
Particulate PCBs 0.15 0.37 <0.02 <0.02 <0.02
Vapor PCBs 0.35 1.3 1.6 1.3. 0.38
Deposition Results
Metals
The concentrations of seven metals (As, Cr, Cu, Mn, Ni, Pb, and Zn) were determined in
atmospheric deposition samples collected at five sites for nine two-week intervals, and
deposition rates were calculated for each metal in each sample. These results are included in
Appendix C-8. The mean metal deposition concentrations for the sampling period are presented
in Table 4-10. The deposition concentrations for all of the metals are greater at the industrial
sites than at the marine sites. The mean metals deposition rates are presented in Table 4-11.
43
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 4-10. Mean Concentrations of Metals'** in Atmospheric Deposition Samples Collected
in Tacoma from July to December 1989
SITE As Cr Cu Mn Ni Pb Zn
Tyee Marina
Tyee Marina (dup)
Morse Supply
Sea-Land
Alexander Avenue
Riverside School
1.4
1.3
10.2
20.8
13.4
0.9
4.0
3.6
7.9
15.0
9.8
3.0
51
26
133
162
85
13
39
33
206
149
144
24
13
7.8
56
41
42
6.0
34
30
132
872
73
13
103
94
326
916
287
26
(a) Mean values for all times
Table 4-11. Mean Metals Deposition Rates
ue/m2/dav
SITE As Cr Cu Mn Ni Pb Zn_
Tyee Marina 1.8 4.1 58 43 17 35 107
Tyee Marina (dup) 1.8 4.0 30 42 8.4 42 125
Morse Supply 9.8 9.2 123 188 49 127 284
Sea-Land 18 17 149 142 41 653 872
Alexander Avenue 11 11 68 112 33 55 230
Riverside School 1.8 4.6 20 42 8.4 22 36
Nutrients
The concentrations of nutrients (nitrate plus nitrite, ammonia, total phosphate, orthophosphate,
and sulfate) were determined in deposition samples collected at five sites for six two-week
intervals. The concentrations measured (mg/L) are presented in Appendix C-9 along with the
sampling dates and the volumes collected. The concentrations of all nutrients appear lower
during the wet periods of November and December than during the drier periods of summer
and early fall due to dilution from rain. The data generally show poor agreement between field
duplicates collected in pans 10 feet apart. There could be numerous reasons for the variability
observed between duplicates and between stations. A primary concern is the long (two-week)
sampling period, which is far from optimal for nutrients subject to microbial uptake and
remineralization, as well as degassing from the water. Therefore, the qualitative observations
about the data should be considered as only preliminary. Mean deposition concentrations for
each nutrient at each site over the July to December 1989 period are summarized in Table 4-
12. There do not appear to be spatial trends in the mean data. Appendix C-9 also contains the
nutrient data expressed as deposition units of /tg/m2/day. Seasonal trends are not apparent.
44
-------
Chapter 4. Six-Month Aerosol and Deposition Study
Table 4-12. Concentrations of Nutrients and Chloride in Atmospheric Deposition Samples
Collected in Tacoma from July to December 1989 (Mean values)
; mg/L
Nitrate
Volume + Ammonia Total Ortho-
Site (L) Chloride Nitrite Nitrogen Phosphate Phosphate Sulfate
Tyee Marina 4.5 4.2 0.39 0.44 0.07 0.01 55
Tyee Marina (dup) 2.5 2.3 0.58 0.74 0.08 0.05 9.1
Alexander Avenue 4.8 2.9 0.45 1.6 0,24 0.01 33
Alexander Ave (dup) 7.5 1.3 0.36 0.13 0.04 0.00 12
Morse Supply 4.3 1.3 0.25 0.14 0.23 0.06 19
Sea-Land 4.3 2.2 0.29 0.32 0.13 0.00 27
Riverside School 6.4 0.67 0.25 0.32 0.05 0.00 4.7
Aliphatic Hydrocarbons and PCBs
The concentrations and deposition rates for AHCs and PCBs are presented in Appendices C-10
(a) and (b). Deposition samples collected in July (July 11 to 25) and December (November 30
to December 14) were analyzed, and the deposition rates for a total of 27 AHCs and 21 PCB
congeners are given in units of ng/m2/day. AHCs with molecular weight greater than C-18
were frequently present in deposition samples. Based on this limited data, there do not appear
to be regional or seasonal trends. The PCBs are rarely detected in the deposition samples and
no patterns are apparent. The sample sizes and detection limits were not satisfactory to provide
useful quantitative data.
Polycyclic Aromatic Hydrocarbons
The atmospheric deposition of PAHs at the five sites is dominated by four CPAHs:
fluoranthene, pyrene, chrysene, and benzo(b)fluoranthene. The mean PAH deposition rates by
site and the total PAH deposition rates are presented in Table 4-13. The PAH data for each
of the 48 deposition samples are in Appendix C-ll. These four compounds were also the four
most abundant compounds in the particulate air samples. This relationship is in agreement with
other PAH deposition studies, which reported that PAH deposition is due to the particulate, not
the vapor phase (McVeety and Kites, 1988). The individual mean deposition rates of the four
dominant compounds are in the range of 500 to 3,500 ng/m2/day at the Tyee Marina, Morse
Supply, Sea-Land, and Riverside School sites. The mean deposition rates at the Alexander
Avenue site are five to ten times higher than those of the Tyee Marina, Morse Supply, or Sea-
Land (see Table 4-13 and Figure 4-9). The maximum deposition occurred at the Alexander
Avenue site in September 1989. Deposition rates at Riverside School are about one-third those
at Tyee Marina, Morse Supply, and Sea-Land. The highest deposition rates at the Alexander
Avenue site are inconsistent with the particulate PAH aerosol data, which indicate similar
concentrations at all sites and show the highest concentration in December rather than
September.
The high deposition of PAHs at the Alexander Avenue site may be due to resuspension of
coarse soil particles which contain high concentrations of PAH. The local construction activity
near the Alexander Avenue site could have caused the resuspension of contaminated coarse soil
particles that were large enough so that they were not collected by the air sampler.
45
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 4-13. Mean PAH and Total Combustion PAH Atmospheric Deposition Rates at Five
Tacoma Sites (ng/mVday)
Site (Number of Samples)
COMPOUND
Naphthalene
Acenaphthalene
Acenaphthene
Fluorene
Dibenzothiophene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Benzo(b)fluoranthene
Benzo(k)fluoranthene
Benzo(a)pyrene
Indeno(l ,2,3-c,d)pyrene
Dibenzo(a,h)anthracene
Benzo(g , h ,i)perylene
AS (8)
140
21
66
150
150
3,200
210
12,000
7,800
3,200
11,000
7,900
2,100
2,600
2,000
640
1,800
TM(9)
120
ND(a)
3
21
19
520
15
2,000
1,200
420
2,900
1,800
39
210
180
75
190
SL(9)
120
16
47
69
46
730
90
2,900
2,100
710
3,400
2,700
600
690
830
270
920
MS (9)
170
75
89
150
60
920
190
1,700
1,600
630
1,900
1,100
540
580
470
230
520
RS (9)
160
10
ND
7
6
260
9
820
510
170
1,100
660
16
87
83
32
110
TOTAL CPAH 52,000 9,000 15,000 9,300 3,600
(a) ND = Not detected
46
-------
ATMOSPHERIC DEPOSITION OF CPAH VERSUS TIME
ALEXANDER AVENUE AND TYEE MARINA, TACOMA
150 -
140 -
130 -
120 -
0
5 110 -
I
o 100 -
O 90 -
{/)
3 80 -
3 70 ~
m 60 -
*~*
3 50 —
a w
gN 40 -
K>
^ 30 -
Q
iS 20 -
10 -
0 ~
-^-.
I
7/13
' /
7/27
/
/
/
f
8/10
^/
/
/
/
/
/
\|
~7
/
/
/
/
's
/
/
/
'j
/
/
jf
J
/
/
/
/
/
/
/
f
/
J S
/ /
/, /
/ /
/ /
/ /
/ /
/ / L
/ S \
-n /-^ x\
/
/
^r
;
/
/
/
/
/
/
i i i
Ss
\
\
\^
v
/
/
s
';
/
/
/
/
/
\
\
\
8/24 9/7 9/21 10/5 10/19 11/2 11/16 11/30 12/14
Y S\ Alexander Avenue
DATE
(\\! Tyee Marina
a?!?
i'l
g£
^^
p^>
B
1
re"
o*
S?
•O
1.
^^
S*
O
n
I
<
g
H
1
>
w
3
Dd
«
>.
fB
re
85
a
re
re
n
85
5T
JU
If
i
B
?
%
I
0
S3
a
^^
s1
i.
^^
S*
8
£*"
^
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
QUALITY CONTROL
Quality Control Results for Field Sampling
Quality control procedures for field sampling included collection of duplicate filters, PUFs, and
deposition samples, and field blanks.
Results of field blanks (filters, PUFs, and water) indicate the elements and compounds were
either not detected and/or were detected at concentrations that were insignificant (less than 10
percent) compared with the concentrations in field samples.
The results for duplicate samples are limited to only a few filters, PUFs, and deposition samples
because a limited number of the field samples were analyzed. All of the particulate duplicate
samples were analyzed and the results between samplers were usually very close; for example,
the mean difference between 11 particulate samples collected on two PS-1 samplers at the
Alexander Avenue site during the fall intensive period was 2.6 percent. There is also good
agreement between the particulate concentrations (TSP) measured daily at the PSAPCA Fire
Station No. 12 site and those measured for three- or four-day periods in this study.
The concentrations of metals in duplicate aerosol and duplicate deposition samples usually agree
within 10 to 20 percent. There is lesser agreement between nutrients in duplicate deposition
samples than for metals, which indicates that the nutrient data have higher variances. The
duplicates for PAHs in filters, PUFs, and deposition samples usually agree within 20 to 30
percent; however, approximately 10 percent of the PAH compounds in both vapor and
particulate samples differ by more than 50 percent between duplicates.
Quality Control Results for Chemical Analyses
Polycyclic Aromatic Hydrocarbons
Quality control samples for chemical analysis of PAH compounds in air filters included field
blanks, surrogate recoveries, and filter spikes, as well as analysis of National Institute of
Standards and Technology (NIST) certified reference urban dust (SRM 1649). The field blanks
did not contain detectable levels of PAH compounds except for naphthalene [Appendix C-12(a)].
The detection limits were about 0.1 ng/m3 [Appendix C-12(b)]. Surrogate recoveries of three
deuterated PAH compounds (naphthalene-d8, acenaphthene-dlO, and perylene-d!2) that were
added to each filter during the extraction were usually in the range of 10 to 40 percent for d8,
30 to 70 percent for dlO, and 60 to 120 percent for d!2. These recoveries are within
acceptable ranges. Blank filter spikes with approximately 1.5 ng/m3 of all PAH compounds had
recoveries in the range of 40 to 80 percent [Appendix C-12(c)]. The results for SRM 1649
averaged 11 percent lower than the certified concentrations [Appendix C-12(d)].
Analysis of PAH PUF plugs and deposition water samples included field blanks, surrogates, and
blank spikes. Field blank PUFs contained detectable PAH concentrations equivalent to
approximately 2 ng/m3 [Appendix C-12(a)]. Field blank deposition water occasionally contained
five to 10 ng/L of PAH compounds which would contribute about five ng/nf/day, or only a few
percent of the deposition rate for most field samples. Surrogate recoveries of d8 in PUFs and
deposition samples were usually in the range of 10 to 20 percent [Appendices C-12 (a) and (c)].
However, the dlO and d!2 recoveries were approximately 20 to 40 percent and 60 to 110
percent, respectively. Recoveries of spiked blanks were in the range of 60 to 112 percent for
PUF spikes at the level of 2.5 ng/m3 [Appendix C-12(c)]. Spike recoveries for water were in
the range of 38 to 130 percent with the lower recoveries for the most volatile PAH compounds.
48
-------
Chapter 4. Six-Month Aerosol and Deposition Study
PCBs and Aliphatics
Method detection limits for air and deposition samples were estimated by analyzing seven
replicates of spiked blanks, then multiplying the standard deviation of the mean by the Student
t-test values [Appendices C-12 (e) and (f)]. The concentrations of PCB congeners and AHCs
in field blanks were below the method detection limits. The spike recoveries for a blank filter
spiked with AHCs were in the range of 45 to 159 percent [Appendix C-12(g)].
Nutrients
Quality control procedures for nutrients included analysis of a field blank, laboratory duplicates,
matrix spikes, and matrix spike duplicates. Field blanks (water used to fill the deposition
sampler) contained 0.021 mg/L total phosphate, 0.014 mg/L nitrate + nitrite, and 0.013 mg/L
ammonia. The other nutrients were below detection limits. Laboratory duplicates usually
agreed within 10 percent and matrix spike recoveries were in the range of 83 to 112 percent.
Because of the poor agreement between nutrient deposition rates for field duplicate samples and
between different sites, the samples appear to have been subject to errors due to the sampling
procedure, and therefore the data are considered of lesser quality than the metals and PAH data.
Metals - XRF
The procedure for XRF analysis of aerosol-loaded filters was developed for low atomic weight
organic filter material such as Whatman 41, Teflon, cellulose acetate, International Paper
Corporation (IPC) filter material, polycarbonate, charcoal, and similar material on which the
aerosol loading would represent a significant percentage of the total mass observed by the XRF
analytical method. Quartz fiber filters were used because they are acceptable for the collection
and analysis of organic compounds. The problem encountered with the relatively high atomic
weight of the quartz filters is that the filter itself represents most of the mass defined as sample
by the scatter peaks, which results in an over-correction for absorption. The use of the quartz
filter also effectively forecloses retrieval of any usable concentration values for the analytes of
aluminum (Al), silicon (Si), and P. To deal with the problem of over-correction, a spectrum
of a blank filter was collected for one-half the acquisition time of the actual samples and the
blank spectrum was subtracted from the sample spectrum. To verify that the corrections were
of the proper magnitude, a fusion of a known area of three of the filters was accomplished and
analyzed for three analytes (Fe, Ca, and K) by an independent technique, atomic absorption
(AA) spectroscopy [Appendix C-12(h)].
Quality control for elements on filters included field blanks and XRF standards. Concentrations
of elements in filter blanks were usually below detection limits or at a low uniform level. For
several metals (As, Cr, Se, and V), the concentrations in some of the field samples were below
detection or, when the field samples were corrected for the filter blank, negative concentrations
resulted.
High and variable concentrations of copper in air filter samples resulted from air contamination
by the air sampler motor armature even though a plastic tube had been used to divert motor
exhaust away from the sampler. Consequently, the copper data for air filters were not usable.
The deposition samples, which were not collected as close to the air sampler motor, do not
appear to be affected by copper contamination.
Metals - Atomic Absorption
Quality control samples including two certified water samples (SLRS and 1643b), three field
blanks, certified urban dust (1648), and two matrix spikes were analyzed along with the
49
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
deposition samples [Appendix C-12(i)]. The results of quality control samples indicate the
analytical methods provided acceptable results. Except for Cr, Ni, and Zn recoveries from the
urban dust standard, the recoveries for metals in other standards and spikes were acceptable.
The low recoveries of chromium and zinc indicate the digestion method (total recoverable metal)
used was not able to completely dissolve these metals in urban dust. The high recovery of
nickel is unexplained. However, nickel recovery was acceptable in other standards and matrix
spikes. The concentrations of metals in field transport blanks were low compared with the
concentrations of metals in field samples.
SUMMARY
Aerosol
During the summer, the total paniculate mass concentrations were three times higher at
the industrial (Alexander Avenue) site than at the marine site (Tyee Marina).
Higher concentrations of crustal elements were observed at the industrial and rural sites
than at the two marine sites (Tyee Marina and Brown's Point).
The highest aerosol concentrations of Pb and Zn were measured at the Sea-Land industrial
site.
The mean total CPAH concentrations [fluoranthene through benzo(g,h,i,)perylene] were
relatively uniform among the sites, except for the background site at Brown's Point,
where CPAH concentrations were lower than at the other sites.
The highest total mass concentrations occurred during periods of air stagnation.
During periods of air stagnation, similar total mass concentrations were observed at all
sites.
The concentrations of the metals were highly correlated with the total mass concentrations
and were, therefore, generally lowest during wet and windy periods and highest during
air stagnation episodes.
Concentrations of crustal metals, such as Fe and K, were relatively higher in the summer
at the Alexander Avenue site.
During the air stagnation episodes in the fall, the concentrations of Pb and Zn were much
higher relative to the concentrations of crustal metals than they were in the summer.
Concentrations of particulate PAHs in the aerosol were approximately an order of
magnitude higher during the fall air stagnation episodes than when the total mass
concentrations were low.
The mean concentrations of vapor PAHs were considerably higher than the mean
concentrations of particulate PAHs. However, the concentrations of the more volatile
PAH compounds (naphthalene through fluorene) were underestimated by the sampling and
analytical methods used in this study, and some of the particulate PAHs may have broken
through during the sampling and been collected on the PUF.
Nutrient, aliphatic hydrocarbon, and PCB data were not adequate for reliable quantitative
results.
50
-------
Chapter 4. Six-Month Aerosol and Deposition Study
Deposition
The deposition rates for all of the metals were greater at the industrial sites. The highest
deposition rates for Pb and Zn were at the Sea-Land site.
The atmospheric deposition of PAHs at the five sites was dominated by the most abundant
compounds in the paniculate air samples.
The PAH deposition rates at the Alexander site were five to 10 times higher than those
at the Tyee, Morse, or Sea-Land sites. Deposition rates at the Riverside School site were
about one-third those at the Tyee, Morse, and Sea-Land sites.
The highest maximum PAH deposition occurred at the Alexander site in September, while
the highest paniculate PAH aerosol concentrations occurred throughout the monitoring
network in December.
Higher metals deposition rates occurred in the summer than in the fall.
51
-------
Chapter 5. Emissions
EMISSION INVENTORY
An emission inventory is a cataloguing of all emissions sources, with information regarding the
total mass and specific chemical constituents emitted by each source. It is important to have
as accurate and complete an emission inventory as possible to use as input for modeling studies.
The diffusion/transport model uses the emission inventory estimates, along with stack
parameters such as height and diameter of the stack, temperature and emission rate of the exit
gases, and meteorological data to estimate the ambient concentrations of chemicals via the
diffusion equations. The chemical mass balance (CMB) or receptor model compares patterns
of chemicals from the ambient data (at the receptor) with patterns of chemicals of the possible
contributing sources from the emission inventory.
The inventory was compiled for the PM10 fraction of emissions (particles with diameters of
10 p.m or less), because this is the focus in regulating emissions (and the size class most
responsible for human health effects). Most information is available for this size class. This
was the size class used in the diffusion/transport modeling and measured during the 18-day
receptor modeling study.
Emissions from point sources (industrial emissions) and area sources (mobile sources such as
cars, trucks, ships; woodstoves; and road dust) are included in the inventory. Point source
information was taken from the PSAPCA registration files, the source profile library, and
relevant source tests. Area source estimates originated from PSAPCA's PM10 emission
inventory for the Tacoma Tideflats. Despite the substantial database for compiling the
inventory, the data are limited in representing the spatial and temporal variations of source
emissions. In addition, specific source tests were not available for all point sources, so
estimates were derived from "generic" profiles in the source profile library. These limitations
affect the accuracy of the modeling and will be discussed further in Chapter 8, Comparison of
Studies.
The discussion that follows provides an overview of the emissions inventory. An amplified
discussion of the computations for the emissions inventory can be found in Appendix D.
Point Sources
PSAPCA Files
Point-source air emission data were taken from PSAPCA's registration files for industrial
sources. The information in these files is a combination of actual measured emissions and
estimates based on industrial process and facility specifications. Point sources within 12 km
of Fire Station No. 12 (located approximately in the center of the Tideflats industrial area) are
listed in descending order of emission quantity in Tables 5-1 and 5-2. In Table 5-1 the sources
are listed in descending order of PM10 emission quantity, because most of the metals and PAHs
exist in the environment in the small particles. Indeed, most of the major sources of concern
are listed in Table 5-1. However, a few additional sources, which contained no PM10
emissions, but might contain chemical emissions of interest, are listed in Table 5-2 in
descending order of VOCs.
From PSAPCA files, additional data were compiled on the metal and PAH emissions for the
major contributing sources. This information can be found in Appendix D-l. Limitations in
these data were addressed by using available source profiles from similar industrial point
sources located elsewhere.
53
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 5-1. Point Sources for PM,,
PM10
(T/YR)
NAME
FUNCTION
783.9 Simpson
330.2 Kaiser
52.4 Woodworth
44 Buffelen
43.7 Puget Sound Plywood
42.5 Lone Star Cement
42.4 West Coast Door
27.9 Continental Lime
27.9 U.S. Oil and Refining
25.4 Continental Grain
19.8 General Metals
17.8 Coastcraft
17.1 Domtar Gypsum
11.8 USG Interiors
8.9 Tacoma Port Facilities
6.9 Commencement Bay Mill
6.3 Canyon Sand and Gravel
4.5 Scofield George
3.8 Sound Refining
3.6 Canyon Concrete
1.4 Pacific International
1.3 Kaiser
1.3 Lianga
1.2 Occidental
1.0 Harmon
0.8 Simon Joseph and Sons
0.7 Pacific NW Terminals
0.6 Tacoma Public Works
0.3 Nalleys
0.3 Tacoma Boatbuilding
0.1 Sierra Sandblasting
0.1 Reichhold Chemicals
0.1 Monitor Inc.
0.1 Atlas Foundry
Paper mill
Aluminum production
Asphalt paving
Hardwood veneer and plywood
Softwood veneer and plywood
Minerals and earth
Millwork
Lime
Petroleum refining
Grain
Scrap and waste
Millwork
Gypsum production
Mineral wool
Marine cargo handling
Sawmill and planing mill
Construction
Ready-mix concrete
Petroleum refining
Ready-mix concrete
Concrete products
Aluminum production
Millwork
Alkalies and chlorine
Millwork
Secondary smelting and refining of
nonferrous metals
Marine cargo handling
Asphalt paving
Pickles, sauces and seasonings
Shipbuilding and repairing
Heavy construction
Industrial organic chemicals
Millwork
Steel Foundry
54
-------
Chapter 5. Emissions
Figure 5-1. 12-km Diameter Circle Around Fire Station No. 12
S >Ww'j{* r. V1' •i *2"'-"UlTWTm
IflfiliSSi
ip-T^c ^v-'c:f'ti:';Go'mmenc'e'ment '•;-• v,-^-
['•••"JJKX. I*-..1 CJ>•'•.•'••., •••. .'..•! '. :- '' ••'•.-,.Vv
55
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 5-2. Point Sources for VOCs
voc
(T/YR)
NAME
FUNCTION
247 Girard Custom Coaters Wood preserving
41.5 American Reinforced Plastic Plastic products
28.6 TAM Engineering Motor vehicle parts/accessory
23.5 Auto Warehousing Business association
21.1 Overall Laundry Industrial Launderers
6.0 B.P. Oil Petroleum bulk station/terminal
4.0 Buckeye Pipeline Refined petroleum pipeline
3.1 Superior Oil Petroleum bulk station/terminal
3.0 Unocal Petroleum bulk station/terminal
2.1 American (Yale) General industrial machinery
2.3 Pennwalt Inorganic Chemicals Industrial inorganic chemicals
0.7 Beloit Fabricated rubber products
0.5 Puget Sound By-Products Animal/marine fats and oils
0.2 Tacoma Public Works Utility Sewage systems
Source Profile Library
Source profiles are obtained through emissions testing at the source. A number of sources have
been tested in the Northwest and throughout the United States. The source profiles obtained
include information on the proportions of various chemicals in the emissions. Source profiles
used to estimate metal and PAH emissions in the Tideflats are listed in Appendix D-2. Some
of the profiles are based on source tests performed at Tacoma sources. However, some of the
profiles are taken from source tests performed at similar sources elsewhere in the country
(generic source profiles). Generic source profiles provide some information about metals and
PAHs which otherwise may be completely missing from the registration data, but often are not
truly representative of the source in question.
Source profiles are usually expressed as a percentage of the total mass contributed by each
chemical in the emission. To estimate the emissions of a particular plant, the specific chemical
percentages are multiplied by the total mass emissions of the plant to acquire chemical-specific
mass emission numbers.
Source Tests
According to the PSAPCA PMi0 emission inventory, 73 percent of the point source PM10
emissions in the Tacoma Tideflats come from the Simpson Tacoma Kraft pulp mill (51 percent)
and the Kaiser aluminum refinery (22 percent). Therefore, it seemed particularly important to
obtain the best emission information available for these two sources.
There are two sets of chemical source test information for Kaiser Aluminum (see Appendix
D-3). Both sets contain general emission information and information on quantities of specific
PAHs. Pollution control work has been done at Kaiser to reduce particulate emissions since
the 1985 source test. Therefore, the most recent set (1988) is being used to define the
emissions from Kaiser. A summary of the Kaiser particle size source test results is included
at the end of this chapter.
56
-------
Chapter 5. Emissions
The source test data on Simpson that were available at the beginning of this project did not have
specific information on the PAHs that were of interest for this study. Therefore, a source test
was performed on Simpson, with support from Ecology. A discussion of the Simpson source
test and the results comprise the second section of this chapter. A description of the sampling
procedures and the complete results of this source test are in Appendix D-4.
Area Sources
An area source emission inventory was compiled for the Tideflats industrial area using available
1986 emission estimates from PSAPCA and scaling them up for growth through 1991. To
inventory area sources, a grid system was set up for the Tideflats area. The Universal
Transverse Mercator (UTM) System was used to define grids with a spatial resolution of 1 km2.
A detailed discussion and calculations of area source emissions can be found in Appendix D-5.
Residential Heating
Emissions from residential heating are generally correlated with the population density.
Therefore, a detailed, block-by-block analysis was made for the Tacoma Tideflats area using
the latest census data (1980).
Distillate oil: Data on sales of distillate oil were obtained and adjusted for known usage by all
point sources. The remaining fuel was apportioned to each grid by population. PM10 emissions
were calculated using an emission factor from Compilation of Air Pollution Emission Factors
(AP-42, Vol. I, pg. 1.3-2 8/82) and assuming a PM10 fraction of 50 percent (AP-42, Vol. I, pg.
1.3-6, 10/86) and a population density of 1,000 people per square kilometer.
Residential wood combustion: To estimate PM10 emissions from residential wood combustion,
data from a study conducted by the Oregon State Department of Environmental Quality was
used to derive wood usage. The average usage for all homes was found to be 0.71 cords per
year. Emission factors for wood stoves/fireplace inserts and fireplaces were taken from AP-
42 (Vol. I, pg. 1.9-3, 5/83), and a PM10 fraction of 100 percent was assumed.
To adjust for the colder climate in Tacoma than Portland, the ratio of the climatological average
of heating degree days per year was also factored into the calculation.
Railroads
Data on distillate fuel (diesel) usage by switchyard locomotives were obtained from Burlington
Northern and Union Pacific. PM10 emissions were calculated using these data and an emission
factor from AP-42 (Vol. II, pg. II-2-1, 4/73), assuming a PM10 fraction of 80 percent.
Ships
Data on ship activity were obtained from the Port of Tacoma. An approximate emission factor
from AP-42 (Vol. II, pg. II-3-2, 1/75) and a PM10 fraction of 50 percent (AP-42, Vol. I, pg.
1.3-6, 10/86) were used and a 24-hour turnaround time was assumed.
Motor Vehicles
The calculation of exhaust emissions from motor vehicles includes a very slight dependence on
the vehicle speed and a strong dependence on the vehicle mix. The emission factors for light-
duty cars and trucks are similar. Heavy-duty gasoline vehicles emit roughly three times as
much as light-duty vehicles; however, the majority of the total emissions come from heavy-
i
57
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
duty diesel trucks. Even though they constitute only about 10 percent of the vehicles on the
road in these areas, they emit nearly 75 percent of the paniculate matter.
The method used to compute motor vehicle emissions per grid in the Tacoma Tideflats area
involved defining three speed zones (20, 35, and 55 mph), each with its own vehicle mix. The
20-mph zone was chosen to represent downtown and residential traffic (collector streets), and
the average vehicle mix that was used was provided by the Washington State Department of
Transportation (DOT). The 35-mph zone was chosen to represent the major arterial streets,
which, in these areas, have roughly four times as many heavy-duty diesel trucks and twice as
many heavy-duty gasoline trucks. The 55-mph zone was chosen to represent highways and
freeways, and contained the standard DOT mix.
The procedure used to calculate emission factors is contained in AP-42 (Vol. II, Appendix L,
9/85).
Traffic count data were acquired from the Traffic Engineering Division of the City of Tacoma,
and when multiplied by the appropriate length of roadway, these numbers give the vehicle miles
traveled (VMT.). The product of the VMT per grid and the appropriate emission factors gives
the total emissions per grid.
Resuspended Road Dust
The emission factor for road dust is a function of the vehicle speed. High-speed roadways are
swept clean by the turbulence induced by the flow of traffic.
Conclusions
Table 5-3 summarizes the results of the PM10 area source inventory for the base year 1986.
The totals in the table were computed by expanding the example calculations in Appendix D-5
to all area sources for the total Tideflats industrial area. The inventory estimates clearly show
the importance of resuspended road dust and motor vehicle exhaust. When calculated as a daily
emission rate, the contribution of woodstoves is also important. The emissions from the use
of natural gas and distillate oil for residential heating are negligible.
Table 5-3. PM10 Area Source Inventory-1986 '
Source Tideflats
Tons/year kg/day
Exhaust 58 143
Road Dust 357 887
Ships 9 22
Railroads 6 15
Airplanes — —(a)
Woodstoves IS 109
Total 448 1,176
(a) Insignificant
58
-------
Chapter 5. Emissions
For the 1991 baseline inventory the following growth factors were assumed:
Table 5-4. Growth Factors 1986-1991
Source Tideflats
Exhaust +3% per year
Road Dust +3% per year
Ships no change
Railroads no change
Airplanes N/A
Woodstoves no change
Growth factors for vehicular sources (i.e., exhaust and road dust) were determined by the
Tacoma Traffic Engineering Division based on the rate of growth in traffic volumes over the
years 1984-1986.
SIMPSON TACOMA KRAFT SOURCE TEST
Because the Simpson Tacoma Kraft plant is the largest PM10 point source in the Tideflats, a
source test was performed to update the site-specific metals and PAH emission data. Paniculate
samples were collected from three emission points at the Simpson plant: the hogged fuel boiler,
the No. 1 lime kiln, and the No. 3 recovery furnace. The source testing was started January
2 and completed January 6, 1990.
The methods used were similar to those used for the Pacific Northwest Source Profile Library
(PNSPL) study. Samples were analyzed to determine total mass and chemical analyses were
performed on the fine fraction using x-ray fluorescence (XRF), ion chrpmatography, and
combustion flame ionization (CFI) for organic and elemental carbon determination.
The PM10 size-segregating dilution sampler (SSDS) was used to collect samples from each
emission source. Two dichotomous samplers (total of 4 filters) with modified inlet tubes were
used to collect both fine and coarse samples from the dilution chamber. In general, the samples
were collected downstream of the emission controls on each of the stacks at sample ports
routinely used for compliance monitoring. Typical sampling duration was one hour. All source
tests included three replicate runs, so that a total of 12 filters were collected for each of the
source tests.
The sources were tested under normal operating conditions. The facility had recently gone
through a maintenance shutdown, and the sampling was delayed until plant operation returned
to acceptable operating conditions. An expanded description of the sampling equipment and
sample run conditions can be found in Appendix D-4(a).
Test Results
The laboratory results of the PAH, XRF, and ion chromatography analyses are presented in
Appendix D-4(c, d). Laboratory protocols are discussed in Appendix D-4(b).
59
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
PAH Analyses
The PAH analyses showed that there was relatively little organic mass in the source test
samples. This was indicated by both the organic carbon/total carbon fine mass fraction results
and by the mass spectrometer responses. The target quantity for the PAH analyses is 100 fj.g
organic carbon per assay, but there was only 10 /xg per assay or less available for the study
samples. As a result, most of the analytes were below the lower quantitation limits for the
samples, and a limited dynamic range was seen among those analytes that could be detected.
In particular, the five-ring PAH compounds that from previous chemical mass balance analyses
were most useful in differentiating among sources of polycyclic organic matter (POM) were
below detection limits. This characteristic of these samples is not consistent with previous
experience with source tests from Simpson's hogged fuel boilers. It may represent unusually
efficient combustion conditions during sampling or some aspect of sample collection or may be
the result of major modifications that the Simpson Tacoma Kraft company made to the boilers
in 1987 to reduce the emissions of unburned hydrocarbons.
Results from the PAH analyses [Appendix D-4(c)] show very little PAH in the lime kiln
samples; n-alkanes in the recovery boiler samples; and carbazole, fluoranthene, pyrene, and
retene in the samples from the hogged fuel boiler.
Metals. Ions, and Carbon Analyses
Analysis summaries for metals, ions, and carbon are presented in Appendix D-4(d). All three
sources sampled show significant percentages of sodium (26 - 32%), sulfur (9 - 19%), and
sulfate (26 - 58%). Of greater interest to this study are the relative emissions of Cu, Pb, Zn,
As, and organic carbon (OC).
As shown in Table 5-5, the lime kiln emissions are relatively enriched in lead while the hogged
fuel boiler emissions are relatively enriched in the other three metals. Both the hogged fuel
boiler and the recovery boiler emissions are relatively enriched in organic carbon.
Table 5-5. Mean Percent of PMin Emissions
Source Cu Pb Zn As OC
Hogged Fuel Boiler
Lime Kiln
Recovery Furnace
.049
.013
.006
.262
.485
.000
.607
.008
.037
.032
.001
.003
2.34
.30
1.84
KAISER PARTICLE-SIZE SOURCE TEST RESULTS
Particle-size distribution samples were collected from two roof monitors and units 36 and 37
dry scrubber baghouses at Kaiser Aluminum and Chemical Corporation's plant in Tacoma,
Washington, on July 21 through 27, 1988. Three to six particle-size distribution samples were
collected at each site. The average sampling period for each of the samples collected was
approximately 20 hours. The results of these tests are presented in Table 5-6.
60
-------
Chapter 5. Emissions
Table 5-6. Kaiser Particle-Size Distribution Source Test Results - July, 1988
Line I Monitor Line IV Monitor Dry Scrubber Baghouses
Run No. %PM10 Run No. %PM10 Unit-Stack-Run No. %PM,
1
2
3
1 -Special
2-Special
3-Special
13.3 1
11.5 2
18.8 3
18.9
18.0
19.4
32.7 36-2-1
35.7 36-2-1
43.9 36-2-3
36-2-4
37-2-1
37-2-2
37-2-3
37-2-4
36-1-1-Special
36-2-2-Special
36-2-3-Special
32.6
27.2
28.5
16.9
27.4
32.9
38.2
17.5
25.9
32.9
38.2
Average (%) 16.7 37.4 27.7
[Average for Monitors: 27.0%]
The results of the size distribution test show that the average percentage of paniculate matter
with an aerodynamic diameter of less than 10 /*m (PM10) was 27 percent. Therefore, the Kaiser
emissions consisted mainly (over 70 percent) of particles larger than 10 \un in diameter.
61
-------
Chapter 6. Receptor Modeling Study
OBJECTIVES
>• Conduct 18-day ambient monitoring of PM10 participates and volatile organic compounds.
*• Using source test information and known chemical "fingerprints" for emission sources, run
receptor or chemical mass balance (CMB) model to apportion sources of ambient aerosol.
Receptor or chemical mass balance (CMB) modeling uses mathematical tools and chemical
information on the ambient aerosol and its sources to determine the contributions of major
sources of pollutants to the total paniculate mass at a "receptor" site. Receptor modeling
generally focuses on the fine particles (<2.5 ^m) of the ambient aerosol, as this fraction is
transported most broadly in the atmosphere and is primarily a product of combustion or other
human activities.
AMBIENT MONITORING
The receptor modeling was supported by an 18-day ambient monitoring study conducted at the
Alexander Avenue and Morse Industrial Supply sites from December 5, 1989 to December 16,
1989, and from January 2, 1990 to January 8, 1990. Alexander Avenue and Morse Industrial
Supply are located within 3.5 km of each other (Figure 2-2).
Each site was equipped with two dichotomous PMW samplers and a VOC (volatile organic
compound) canister sampler. Two dichotomous samplers with two types of filter media (quartz
and Teflon) were needed to accommodate the variety of analytical techniques performed.
Twelve-hour day and night samples were obtained of fine- (<2.5 pm) and coarse- (2.5-10 pm)
particle metals, fine-particle ionic species, fine-particle organic and elemental carbon, and
VOCs. VOCs were monitored concurrently with fine particles to determine whether any of the
VOCs could serve as a tracer for mobile sources. (Although, lead has been used as a tracer
for mobile sources in the past, it is being phased out of the fuel supply, so an alternate tracer
is needed.) A detailed discussion of the sampling and analysis equipment that was used and
the types of analyses that were performed for this study is presented in Appendix E-l.
Analytical Results of 18-Day Field Study and Discussion
Average Composition
A detailed compilation of all results can be found in Appendix E. Average fine-particle
concentrations of organic carbon, elemental carbon and ionic species measured at the two
sampling sites are presented in Tables 6-1 (a) and (b). Averages are given for daytime (7 a.m.
to 7 p.m.), nighttime (7 p.m. to 7 a.m.), and all samples (day and night combined) for each
sampling site. Chemical species with average measurement uncertainty of 30 percent or less
for at least one of the sites are included in this summary. A compilation of all fine-particle data
can be found in Appendix E-2. .
Average coarse-particle species concentrations are presented in Tables 6-2 (a) and (b). Only
data on total mass and ionic species are available for the coarse particles. Averages are given
for daytime, nighttime, and all samples (day and night combined) for each sampling site.
Species with average measurement uncertainty of 40 percent or less for at least one of the sites
are included in this summary. All coarse-particulate data can be found in Appendix E-3.
63
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 6-l(a). Average Fine-Particle Species Concentrations at the Alexander Avenue Site
Parameter*
Mass
OC
EC
NO3-
SO4=
Al
S
Cl
K
Ca
Mn
Fe
Cu
Zn
Br
Pb
Method13
GRAY
THERM
THERM
1C
1C
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
Mean Concentration, ng/m3
Day
21900
8500
2300
1200
2600
143
937
173
105
49
24
181
31
51
7.0
32
Night
21800
8700
1200
920
1800
816
709
182
128
28
15
64
26
33
7.3
23
All
21800
8500
1700
1000
2100
480
823
177
116
39
19
123
28
42
7.2
27
Uncertainty
All
2%
7%
19%
17%
12%
20%
8%
10%
9%
14%
15%
13%
17%
14%
28%
19%
aMass = fine particle mass; OC = organic carbon as carbon; EC = elemental carbon.
bGRAV = gravimetry; THERM = thermal-optical; 1C = ion chromatography; XRF = X-ray
fluorescence.
64
-------
Chapter 6. Receptor Modeling Study
Table 6-1 (b). Average Fine-Particle Species Concentrations at the Morse Supply Site
Parameter*
Mass
OC
EC
NO3-
S04=
Al
S
Cl
K
Ca
Mn
Fe
Cu
Zn
Br
Pb
Method15
GRAY
THERM
THERM
1C
1C
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
Mean Concentration, ng/m3
Day
24700
10600
2400
1400
2300
136
879
226
146
55
41
289
10
51
8.7
45
Night
23000
10300
1500
1200
1800
148
708
251
175
28
18
89
5
30
7.5
27
All
23800
10200
1900
1200
1900
142
793
238
161
41
29
189
8
41
8.1
36
Uncertainty
All
2%
7%
19%
11%
22%
80%
8%
9%
8%
13%
12%
12%
42%
14%
23%
17%
aMass = fine particle mass; OC = organic carbon as carbon; EC = elemental carbon.
bGRAV = gravimetry; THERM = thermal-optical; 1C = ion chromatography; XRF = X-ray
fluorescence.
65
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 6-2(a). Average Coarse-Particle Species Concentrations at the Alexander Avenue
Site
Parameter*
Mass
Al
Si
S
Cl
K
Ca
Mn
Fe
Cu
Zn
Cr
Method15
GRAY
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
Mean Concentration, ng/m3
Day
8500
459
1092
194
358
64
313
10.4
450
18
27
3.2
Night
4600
203
428
137
302
41
125
6.6
156
12
17
1.0
All
6600
331
760
165
330
53
219
8.0
303
15
22
2.1
Uncertainty
All
4%
42%
39%
21%
19%
16%
9%
30%
11%
23%
27%
148%
"Mass = coarse particle mass.
bGRAV = gravimetry; XRF = X-ray fluorescence.
66
-------
Chapter 6. Receptor Modeling Study
Table 6-2 (b). Average Coarse-Particle Species Concentrations at the Morse Supply Site
Parameter*
Mass
Al
Si
S
Cl
K
Ca
Mn
Fe
Cu
Zn
Cr
Methodb
GRAY
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
XRF
Mean Concentration, ng/m3
Day
13000
1660
1373
154
394
82
336
15
636
13
30
9
Night
6900
1169
528
140
370
58
124
7.2
234
5.2
16
4
All
9600
1428
951
147
381
70
230
11
435
8.8
22
6
Uncertainty
All
4%
30%
29%
22%
17%
15%
9%
21%
10%
33%
17%
40%
"Mass = coarse particle mass.
bGRAV = gravimetry; XRF = X-ray fluorescence.
67
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
For some of those species with average measurement uncertainty greater than 30 percent, there
were individual samples with uncertainties of much less than 30 percent. For example, fine-
particle aluminum concentrations during three nighttime sampling intervals at the Alexander
Avenue site averaged greater than 1 /-tg/m3 with an associated uncertainty around 15 percent.
These samples were associated with light southeast winds or stagnant air conditions. The Kaiser
aluminum plant is 1 km southeast of the Alexander Avenue sampling site. Fine-particle
aluminum concentrations were not similarly high at the Morse Industrial Supply sampling site
on those or any sampling intervals. Conversely, coarse-particle aluminum concentrations on
many sampling days were much higher and measured with lower uncertainty at the Morse
Supply site than at the Alexander Avenue site. Because aluminum is the most abundant metal
found in the earth's crust in addition to being an aluminum smelter product, the high coarse-
particle aluminum concentrations (and high total coarse-particle mass) at the Morse Supply site
may indicate more resuspended soil dust at that site. Samplers were set up on a platform at this
site, and thus were closer to the ground than the Alexander Avenue site samplers, which were
located on the rooftop of a one-story building.
VOC data are summarized in Tables 6-3 (a) and (b) for those species where no more than 30
percent of the samples were below the detection limit. In cases where VOC data were below
the detection limit, the concentration was set equal to zero (plus or minus the detection limit)
for the averaging process. VOC data for all samples are tabulated in Appendix E-4.
Tables 6-1 to 6-3 show that the Morse Supply site tended to have somewhat higher
concentrations of most fine- and coarse-particle constituents and VOCs. One reason for this
is that there may be more vehicle traffic to resuspend dust near the Morse Supply site, because
it is closer to arterial grids that have higher traffic usage than the Alexander Avenue site.
Exceptions to the higher concentration at Morse Supply include fine-particle copper
concentrations, which were nearly four times higher at the Alexander Avenue site than at the
Morse Supply site. Coarse-particle copper was also higher at the Alexander Avenue site. The
elevated copper concentrations are most likely an artifact resulting from the operation of an
adjacent sampler with a type of motor containing copper wire brushes. These motors are
known to eject copper particles during their operation. A similar sampler was operated at the
Morse Supply site, but it was not immediately adjacent to the PM10 samplers.
Coarse-particle iron concentrations were twice as high at the Alexander Avenue site as at the
Morse Supply site, while fine-particle iron was about 50 percent higher at the Morse Supply
site. Coarse-particle chromium was about three times higher at the Morse Supply site than at
the Alexander Avenue site. Given the relatively close proximity of the two sampling sites,
these observations indicate that there were local activities contributing to some of the differences
seen in the specific composition of ambient air measured at each of the sites.
At both sites, the most outstanding feature of the data is the amount of fine-particulate organic
carbon compared with the total fine particles. Figures 6-1 and 6-2 show the relative
contributions of the major constituents of the fine-particle mass at the Alexander Avenue and
Morse Supply sampling sites. Organic carbon concentrations were multiplied by 1.4 to account
for the hydrogen and oxygen which, together with organic carbon, constitute the average
organic compounds in the ambient air (Countess et ah, 1980; Shah et al., 1984). The average
composition was about the same at each site, with organic compounds accounting for over one-
half of the fine-particle mass.
68
-------
Chapter 6. Receptor Modeling Study
Table 6-3(a). Average VOC Concentrations at the Alexander Avenue Site
Dichlorodifluoromethane
Trichlorofluoromethane
1,1, 1-Trichloroethane
Benzene
Carbon Tetrachloride
2,2,4-Trimethylpentane
Toluene
Ethylbenzene
m,p-Xylene
o-Xylene
4-Ethyltoluene
1 ,3,5-Trimethylbenzene
1 ,2,4-Trimethylbenzene
Mean Concentration, ppb
Day
0.72
0.74
0.62
2.24
0.10
0.28
4.57
0.76
2.91
1.02
0.25
0.23
0.77
Night
1.26
0.76
0.51
2.03
0.097
0.21
3.79
0.53
2.05
0.74
0.18
0.19
0.58
All
0.99
0.75
0.57
2.14
<0.1
0.24
4.18
0.65
2.48
0.88
0.22
0.21
0.67
Uncertainty
All
9%
15%
3%
8%
14%
7%
9%
6%
4%
5%
14%
11%
13%
69
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 6-3 (b). Average VOC Concentrations at the Morse Supply Site
Dichlorodifluoromethane
Trichlorofluoromethane
1,1,1 -Trichloroethane
Benzene
Carbon Tetrachloride
2,2,4-Trimethylpentane
Toluene
Ethylbenzene
m,p-Xylene
o-Xylene
4-Ethyltoluene
1 , 3 , 5-Trimethylbenzene
1 ,2,4-Trimethylbenzene
Mean Concentration, ppb
Day
2.03
0.74
0.75
3.30
0.11
0.41
6.94
1.04
3.86
1.36
0.34
0.33
1.04
Night
0.62
0.68
0.50
2.50
0.11
0.31
4.80
0.69
2.56
0.92
0.23
0.25
0.76
All
1.32
0.71
0.62
2.90
0.11
0.36
5.82
0.86
3.21
1.14
0.29
0.29
0.90
Uncertainty
All
15%
8%
2%
6%
9%
6%
6%
3%
2%
3%
14%
4%
6%
70
-------
Chapter 6. Receptor Modeling Study
Figure 6-1. Mtyor Constituents of Fine-Particle Mass at the Alexander Avenue Site
Other (17.2%)
Ammonium Sulfate (13.6%)
Ammonium Nitrate (6.1%)
Elemental Carbon (8.0%)
Organic Compounds
(55.1%)
Figure 6-2. Major Constituents of Fine-Particle Mass at the Morse Supply Site
Other (12.2%)
Ammonium Sulfate (1 1.7%)
Ammonium Nitrate (6.9%)
Elemental Carbon (8.1%)
Organic Compounds
(61.1%)
71
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Ion chromatographic data for nitrate and sulfate were corrected to reflect their possible
association with the ammonium ion. If sulfate existed as un-neutralized sulfuric acid, its
contribution to fine mass would be 25 percent less than the fully neutralized ammonium sulfate.
As no precautions were taken to preserve the original acidity of the sample, the maximum
neutralization was assumed for the calculations. Even so, sulfate expressed as ammonium
sulfate was a distant second in the contribution to fine-particle mass at only 13 percent on
average.
On average, 15 percent of the fine-particle mass falls in the "other" category. This category
includes metals plus other elements associated with them (such as oxides and anions). It may
also include water associated with species in the fine fraction.
Meteorological Conditions
As described in Chapter 3, meteorological data were collected throughout the study from a IO-
meter meteorological instrument tower at the Alexander Avenue site. Average conditions for
each 12-hour sampling period are presented in Appendix E-5. During the December 5 to 15
and January 2 to 8 sampling intervals, very different meteorological conditions prevailed. The
December sampling days were, with few exceptions, under fairly stagnant conditions with little
or no precipitation. Fine-particle concentrations averaged 30.3 +_ 0.6 /ig/m3 during this period.
By contrast, the January sampling days were associated with higher wind speeds, more
precipitation, and lower fine-particle loadings (10.2 +. 0.2 /tg/m3). Winds during the entire
study were most often from the southeast or the south. There were some important exceptions
which had a considerable influence on the mass loadings observed. A more thorough analysis
is given in the following sections.
Temporal Variations of Key Parameters and Correlations Between Sites
Figures 6-3 to 6-7 show the temporal variation of some of the chemical species measured in this
study. The total fine particle mass at each site tracked one another quite well. This is not
unreasonable because of the close proximity (3.5 km) of the two sampling sites. Organic
carbon and fine-particle mass data correlated quite well [correlation coefficient (r) = 0.96 in
a linear regression comparison]. This is a consequence of the fact that an average of 58 percent
of the fine-particle mass was composed of organic carbon compounds. The similarity of these
two parameters also serves a quality assurance role because the underlying analytical procedures
(gravimetric and thermal-optical) are fundamentally different. The similar temporal patterns of
all of the fine-particle species shown indicate the overriding influence of meteorology on fine-
particle concentrations. As discussed earlier, coarse particles exhibited more distinct differences
between the sites, which may be a reflection of both local emission sources and differences in
height of the samplers.
The highest concentrations of fine mass (42 jig/m3) and organic carbon (18 /tg/m3) found during
the 18-day study occurred between December 9 (p.m.) and 15 (a.m.). This approximately
corresponds to the time period (December 9 to 13) when the hogged-fuel boiler and most other
components of the Simpson Tacoma Kraft plant were not operating (see Figure 2-2 for location
of Simpson point source.) This is also a time period associated with the most stagnant air
conditions of the study. The winds were mostly from the southeast, following the flow through
the valley toward the bay. During this time period, the organic carbon concentration dropped
from its otherwise high levels twice (see Figure 6-7.) These lower concentrations (though not
as low as the average for the study) occurred on days when the winds had a northerly or
northeasterly component.
72
-------
Chapter 6. Receptor Modeling Study
Figure 6-3. Fine-Particle Mass Comparison of Alexander Avenue and Morse Supply Sites
versus Sampling Interval
12/05/89 12/09/89 12/13/89 1/02/90 1/06/90
12/07/89 12/11/89 12/15/89 1/04/90 1/08/90
Date
ALEXANDER SITE -»- MORSE SITE
73
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 6-4. Coarse-Particle Mass Comparison of Alexander Avenue and Morse Supply
Sites versus Sampling Interval _•
~" J I T I I I 1 I I I I I I I 1 I III I I I 1
12/05/89 12/09/89 12/13/89 1/02/90 1/06/90
12/07/89 12/11/89 12/15/89 1/04/90 1/08/90
Date
ALEXANDER SITE
MORSE SITE
74
-------
Chapter 6. Receptor Modeling Study
Figure 6-5. Fine-Particle Elemental Carbon Comparison of Alexander Avenue and Morse
Supply Sites versus Sampling Interval
12/05/89 12/09/89 12/13/89 1/02/90 1/06/90
12/07/89 12/11/89 12/15/89 1/04/90 1/08/90
Date
ALEXANDER SITE
MORSE SITE
75
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 6-6. Fine-Particle Sulfate (by Ion Chromatography) Comparison of Alexander
Avenue and Morse Supply Sites versus Sampling Interval
U I i i I T i ii i ii ii ii i i T i i i i
12/05/89 12/09/89 12/13/89 1/02/90 1/06/90
12/07/89 12/11/89 12/15/89 1/04/90 1/08/90
Date
ALEXANDER SITE
MORSE SITE
76
-------
Chapter 6. Receptor Modeling Study
Figure 6-7. Fine-Particle Organic Carbon Comparison of Alexander Avenue and Morse
Supply Sites versus Sampling Interval
LJ I I 1 I I I 1 I F T I F I i I I I f I 111
12/05/89 12/09/89 12/13/89 1/02/90 1/06/90
12/07/89 12/11/89 12/15/89 1/04/90 1/08/90
Date
ALEXANDER SITE
MORSE SITE
77
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
In December, a period of very low fine mass and organic carbon concentrations occurred from
December 7 (nighttime) to December 9 (daytime). This period was associated with precipitation
and with higher wind speeds with a southerly or southwesterly component. The January
sampling days were associated with precipitation, higher winds, and much lower filter loadings
of fine particles. Coarse-particle concentrations at the Morse Supply site were generally higher
than at the Alexander Avenue site, especially during the January sampling period.
Appendix E-6 contains figures with concentrations measured at the Morse Supply site plotted
versus concentrations measured at the Alexander Avenue site. All figures show very good
correlations between the two sites [correlation coefficient (r) = 0.92 or greater], with somewhat
higher concentrations measured at the Morse Supply site.
PAH Results
Total ion chromatograms and peak identifications for PAHs appear in Appendix E-7. All
identifications are based on automated searches of mass spectral libraries and should be
considered tentative. Tentative identifications were confirmed with selected ion monitoring
(SIM). The results of this analysis are also in Appendix E-7. Analyses were difficult because
PAH levels were near the detection limits of the instruments. The quantitative results should
be considered as not better than +50 percent. The qualitative identifications are somewhat
uncertain due to the constant loss of chromatographic performance caused by the high levels of
nonvolatile material in the extracts. PAHs may also have been missed due to interferences or
shifting retention times. Data with less uncertainty are required for receptor modeling
applications, therefore the PAH data acquired in this study were not used further.
Motor Vehicle Tracers - Lead
Lead and/or bromine have traditionally been used as tracers for motor vehicles. The correlation
between fine lead and bromine at the two sampling sites was generally good (r = 0.88), though
not as good as in past studies (Lewis et al., 1988a). As indicated in Tables 6-1 (a) and (b),
lead concentrations followed a weak diurnal pattern, with the higher concentrations mostly
occurring during the day when total motor vehicle use is at the highest. Elemental carbon
followed a similarly strong diurnal pattern, while total fine particles did not. Elemental carbon
could be a component of automobile emissions, especially for diesel powered vehicles. A
bromine-to-lead ratio in the range of 0.3 to 0.4 in urban areas (Lewis et al., 1986; Koutrakis
and Spengler, 1987) is consistent with concentrations of lead and bromine originating from
motor vehicles (this ratio is not expected to change as the total amount of lead is reduced in the
fuel supply). During this study, the ratio averaged 0.24 ±_ 0.07, which is somewhat low, but
still within range to be representative of motor vehicles. These findings, coupled with sufficient
loadings of lead, indicate that lead may be used as a tracer for motor vehicles in this study.
The low bromine-to-lead ratio indicates there may be a source of lead other than automobiles.
Motor Vehicle Tracers - Volatile Organic Compounds (VOCs)
Several VOCs were identified as candidates for tracers of fine-particle carbon and extractable
organic matter (EOM) from mobile sources as part of a wintertime air pollution study in Boise,
Idaho in 1986 and 1987 (Stevens et al., 1989; Zweidinger et al., 1990). Two of those
candidate tracers, which were also measured in this study, are o-xylene and 2,2,4-
trimethylpentane. In the Boise study, ambient concentrations of a candidate VOC were required
to have a high correlation with ambient concentrations of fine-particle lead. Furthermore, the
VOCs had to have a low correlation with soil-corrected fine-particle potassium, which has been
shown to be a tracer of woodsmoke (Lewis and Einfeld, 1985). The latter requirement is
necessary to exclude VOCs highly correlated with combustion sources other than motor
78
-------
Chapter 6. Receptor Modeling Study
vehicles. Assuming lead is still a reasonable tracer for motor vehicles, this same approach can
be used for the Tacoma Tideflats data.
Galloway et al. (1989) showed that water-soluble potassium represents the nonsoil derived
potassium in the fine fraction. It is likely that potassium is formed in water-soluble form by
all combustion sources that produce potassium. Usually the amount of potassium from
windblown soil is greater than potassium from other sources. Furthermore, windblown soil is
concentrated in the coarse-particle fraction, with some contamination in the fine fraction, so the
coarse-particle potassium concentration should be higher than the fine-particle potassium
concentration if soil is the dominant source of potassium. Tables 6-1 (a) and (b) and 6-2 (a)
and (b) show that the fine-particle potassium concentration was more than twice the coarse-
particle potassium concentration, suggesting that soil was not the dominant source of potassium.
Therefore, the correction for potassium originating from soil should be small.
The correction is made using the ambient coarse-particle potassium-to-iron (K:Fe) ratio as the
best estimate of the ratio in local soil. In this data set, that ratio was quite variable (±.120
percent), indicating that there may have been a nonsoil component in either the coarse potassium
or coarse iron. This is entirely possible in such an industrial area. If this nonsoil,
nonwoodsmoke, coarse-particle K:Fe ratio is the same as in the fine fraction, the correction is
still valid. There is no way to check that assumption with this data base.
Soil-corrected potassium, represented by K', is calculated using the method described by Lewis
and Einfeld (1985):
[K'] = [K]rme - [K:Fe]coane.[Fe]
ifinc
where [K] and [Fe] are the ambient fine or coarse-particle potassium and iron concentrations.
Because of the large coarse K:Fe variation, individual sample ratios were applied to
corresponding fine-particle data.
In the Tacoma Tideflats airshed, both residential and industrial woodsmoke, as well as other
nonmotor vehicle combustion sources, may contribute to the nonsoil potassium in the fine-
particle fraction. Motor vehicles have a negligible contribution to fine-particle potassium.
Unfortunately, there are no data that show the water-soluble potassium content of other major
sources in the airshed. Table 6-4 indicates that K' varied diurnally, with higher concentrations
occurring at night. This may indicate a significant residential woodsmoke source contribution.
Table 6-4. Average Concentrations of Soil-Corrected Potassium, K*
Alexander (AS) Morse (MS)
Avg. K', ng/m Avg. K', ng/m
Day
Night
All
78 + 11
108 + 12
93 + 11
109 + 16
148 + 17
128 + 16
The correlation between the VOCs and fine-particle lead and K' is presented in Table 6-5. The
VOCs measured in this study (with less than 30 percent of the measurements below detection)
are listed in descending order of their squared correlation coefficients with lead. Benzene tops
the list, but benzene also has the highest correlations with K' and is not expected to be a
reliable mobile source tracer. The Boise study indicates that woodsmoke may have significant
benzene and toluene components. O-xylene was ranked third in correlation with lead and had
a reasonably low correlation with K'. It is thus confirmed as a possible mobile source tracer.
79
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
The other promising mobile source tracer identified from the Boise data, 2,2,4-trimethylpentane,
was measured at low levels in this study and had a fairly high correlation with K'. Both o-
xylene and 2,2,4-trimethylpentane were mildly diurnal, with higher concentrations occurring
during the day.
Table 6-5. Correlations of VOCs with Lead and K>(N=71)
voc
Pb
(r2)
K'
(r2)
VOC
avg. ppb
Benzene 0.819 0.455 2.52
Toluene 0.799 0.386 5.00
o-Xylene 0.779 0.334 1.01
1,2,3-Trimethylbenzene 0.770 0.366 0.79
2,2,4-Trimethylpentane 0.758 0.437 0.30
m,p-Xylene 0.757 0.303 2.84
4-Ethyltoluene 0.752 0.349 0.25
Ethylbenzene 0.751 0.311 0.76
1,3,5-Trimethylbenzene 0.699 0.334 0.25
1,1,1-Trichloroethane 0.565 0.268 0.60
Trichlorofluoromethane 0.414 0.326 0.73
Dichlorodifluoromethane 0.016 0.009 1.16
Carbon tetrachloride 0.0017 0.0002 0.10
RECEPTOR MODELING RESULTS
Source Apportionment by Chemical Mass Balance
EPA's CMB model, version 7 (Watson et al., 1990), was used to apportion fine-particle
chemical species measured at the sampling sites to sources which may have been impacting
those sites. The CMB model consists of a least squares solution to a set of linear equations
which expresses each measured concentration of a chemical species i (Cj) as a linear sum of
products of the abundance of i as emitted by source j (A^) and the total mass concentration due
to source j (Mj):
Q = E A;jMj + e-,,
where &, is the residual that represents the difference between the measured and calculated
concentrations. This equation is solved for Mj by minimizing Chi2 in the expression:
Chi2 = E ei2/E,2,
where E, is the effective variance, which represents uncertainty in both A;j and Q. The set of
abundances of all species (i) as emitted by source (j) represents the "source profile" of source
(j).
In performing a CMB, it is assumed that: (1) the abundance of a species (i) used in the fitting
procedure is known for each source type; and (2) all major sources of each species (i) must be
included in the CMB. Usually, there are more species (i) fitting the above criteria man there
are sources.
80
-------
Chapter 6. Receptor Modeling Study
Based on 1988 PM10 emissions data for the Tacoma Tideflats, the Simpson Tacoma Kraft
Company and the Kaiser Aluminum and Chemical Corporation are by far the largest point
sources in the area. It is presumed that a large fraction of the PM10 emissions, which are from
combustion sources, are in the fine-particle (< 2.5 ^im) mode. Simpson operates a hogged fuel
boiler which is responsible for much of the Simpson PM10 emissions. Kaiser emits particles
through the roof monitors and baghouses. However, less than 30 percent of the Kaiser
emissions are less than 10 /*m in diameter (see Chapter 5). Other major point sources include
two veneer and plywood manufacturers and a petroleum refinery.
Major area sources are expected to be residential woodsmoke and motor vehicle exhaust.
Motor vehicle exhaust is a mix of emissions from leaded-fuel vehicles, unleaded-fuel vehicles,
and diesel-operated equipment.
Most of the major area and point sources in the Tacoma Tideflats airshed emit nonnegligible
amounts of fine-particle organic carbon (e.g., motor vehicles, woodstoves, residual oil burning,
aluminum production, and veneer dryers), and there are multiple sources of potassium (hogged
fuel boilers and woodstoves) and elemental carbon (diesel vehicles, woodstoves, and residual
oil). Species such as chlorine and bromine should only be used as fitting species if absolutely
necessary, as they may not be stable on particles. Thus, resolving individual source types in
the Tideflats airshed presented a considerable challenge. In terms of the chemical mass balance
equation, the problem was underdetermined, so the fine-particle mass apportionment could not
be accomplished by CMB alone. Other indirect methods of apportioning some of the sources
were implemented. Motor vehicle sources were apportioned by alternate methods in addition
to CMB because the source profile of the average vehicle mix is not well known.
Description of Sources
The following components were included in the CMB calculations for fine particles: residential
woodstoves, hogged fuel boilers, aluminum production (one site only), residual oil boilers, scrap
metal fugitives, and motor vehicle exhaust. For the Alexander Avenue site, all source types
listed were included in the CMB calculation. The species used in the Chi2 minimization were
either Al, Cl, K, Zn, and Pb or VOCs, OC, and SO4". For the Morse Supply site samples,
the aluminum production source was not included because it produced a poor fit, and because
fine aluminum was near detection limits and thus had high uncertainties. The Morse Supply
site is farther from the Kaiser aluminum plant than the Alexander Avenue site and was not
downwind as often. In this case, the species used in the Chi2 minimization were Cl, K, Zn,
and Pb or VOCs, OC, and SO4=. Attempts were made to include an urban dust (or crustal
dust) source category, but it contributed insignificantly to fine mass. Its major chemical
element, silicon, was measured at levels too low to yield consistently good data. Similarly,
attempts were made to include veneer dryers but this resulted in collinearities with other sources
and an insignificant contribution to fine mass in the CMB.
Two alternative CMB/source apportionment calculations were performed for each sampling site.
The first (Vehicle Exhaust I) used a leaded-gas vehicle signature to calculate the contribution
from leaded-gas vehicles. This number was scaled up using emissions and fuel use data to get
the total vehicle exhaust contribution. The second CMB/source apportionment (Vehicle Exhaust
II) used a VOC as the tracer for total vehicle exhaust. This apportionment method was
attempted with o-xylene and 2,2,4-trimethylpentane as VOC vehicle exhaust tracers.
Residential Woodsmoke
In this study, over one-half of the fine-particle mass was comprised of organic compounds. A
potentially significant source of organic carbon is woodsmoke. In previous studies in western
81
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
states (Lewis et al., 1988a, 19885) residential woodsmoke was shown to be a significant source
of fine particles. Other residential burning may include the burning of grass, trees, brush, yard
and garden waste, and household waste. A survey conducted by the PSAPCA indicates the use
of woodstoves dominates the residential burning activities in Pierce County, which incorporates
the Tideflats and surrounding area. The highest fine-particle mass and organic carbon
concentration measured during the 18-day study occurred when the Simpson Tacoma Kraft
plant, a major industrial wood fuel source, was not operating. In order to draw definitive
conclusions about the dominant sources for all stagnation conditions, it would be necessary to
compare CMB results for a similar set of meteorological conditions when the Simpson plant was
operating. In this CMB application, the source signature for woodsmoke used was Profile
Number 42106 for residential woodstove emissions from the Pacific Northwest Source Profile
Library (PNSPL). This profile represents laboratory data using a typical Portland/Seattle wood
mix. The abundances of organic carbon and potassium in this profile were 47 percent and .25
percent, respectively.
Vehicle Exhaust I
A fraction of the total vehicle fleet in the Tideflats area is comprised of vehicles that burn
leaded gasoline. While the number of leaded fuel vehicles on the road is continually
decreasing, leaded fuel combustion is still expected to make a significant contribution to fine-
particle lead. A source profile for leaded fuel exhaust was used to estimate the contribution of
leaded fuel exhaust to the total fine-particle mass. The leaded vehicle source signature used was
Profile Number 31103 from the PNSPL. Including a diesel vehicle source in the CMB was also
attempted (using Profile Number 32203 from the PNSPL and elemental carbon as the fitting
species), but this resulted in collinearity between diesels and woodsmoke, forcing this approach
to be abandoned.
Vehicle Exhaust II
A second apportionment of the motor vehicle exhaust was performed using VOCs as the tracer
for the total vehicle fleet. Zweidinger et al. (1990) identified several VOCs as contaminants
of fine-particulate carbon and extractable organic matter (EOM) from mobile sources in Boise.
Two of those candidate contaminants which were measured in this study are p-xylene and 2,2,4-
trimethylpentane. These were confirmed as possible mobile source contaminants in this study,
although 2,2,4-trimethylpentane is the weaker candidate and was measured at low levels.
Concentrations of o-xylene and 2,2,4-trimethylpentane were comparable to concentrations of
these two compounds measured in the Boise study. EOM was regressed against each VOC (in
separate regressions) and a woodsmoke tracer K' in the Boise study. The inverse of the EOM
coefficients for motor vehicles represents the abundance of each VOC with respect to EOM.
This abundance was multiplied by 78 percent to get the abundance with respect to total fine
particles. The 78 percent was derived from Lewis et al. (1986) using their calculated motor
vehicle source profile. The calculated volatile carbon abundance (65 percent) was multiplied
by 1.2 to estimate the total EOM abundance in the fine-particle fraction for motor vehicles.
Aluminum Production
A major point source in the Tideflats area is the Kaiser Aluminum and Chemical Corporation.
This source is located just 1 km southeast of the Alexander Avenue sampling site, and therefore
was upwind of the site for most of the study. The source profile for an aluminum reduction
potline (Profile Number 29102 from the U.S. EPA Receptor Model Source Composition
Library, EPA-450/4-85-002) was used to represent the Kaiser aluminum plant in the CMB. The
organic carbon abundance is 28.2 percent in this profile; the aluminum abundance is 15.2
percent.
82
-------
Chapter 6. Receptor Modeling Study
Hogged Fuel Boiler
Another major point source in the Tideflats area is the Simpson Tacoma Kraft Company.
Among its emitting sources is a hogged fuel boiler, which is suspected to be a major contributor
to the fine-particle mass in the Tideflats airshed. This source is located north of the Morse
Supply sampling site and is closer to this site than to the Alexander Avenue site, which is east
of Simpson Tacoma Kraft and the Morse Supply site (Figure 2-2). Several other point sources
in the area operate hogged fuel boilers, though they are much smaller than the Simpson boiler.
These other sources include Buffelen Woodworking, Puget Sound Plywood, and West Coast
Door. Data collected during the source test at the Simpson hogged fuel boiler (Chapter 5) were
used in the CMB source profile for hogged fuel boilers. The abundances of organic carbon and
potassium in this profile are 2.3 and 12.7 percent, respectively (compared with 47 percent
organic carbon and 0.25 percent potassium for woodstove emissions). While both residential
woodstoves and hogged fuel boilers burn wood or wood products, the chemical profiles of those
two sources are sufficiently different that the two sources can be resolved in the CMB.
Residual Oil Boiler
U.S. Oil and Refining operates a residual oil-fired boiler in the center of the Tideflats. Oil-
fired boilers could be a major source of fine-particle sulfur in the Tideflats. Sulfates comprised
about 13 percent of the fine-particle mass as calculated if they existed as ammonium sulfate.
Profile Number 13502 from the PNSPL was used for the residual oil boiler signature in the
CMB. This profile was produced from several samples taken at the U.S. Oil and Refining
residual oil boiler in Tacoma, Washington, in 1988.
Scrap Metal Fugitives
Lead was used as a fitting species (for leaded fuel exhaust) in one version of the CMB for this
study. CMB requires that all significant sources of a fitting species be included in the
calculation. General Metals of Tacoma engages in scrap-metal handling activities that can
contribute to the toxic metal loading, including lead, in the Tideflats. Most of the
aforementioned source types do not contribute significant quantities of zinc, and there are likely
no other significant sources of zinc other than scrap metals. Therefore, zinc could be used as
a fitting species for metals. The PACS Profile Number 3193 was used in the CMB to represent
many sources of metals, such as scrap-metal handling, car shredding, and boat building. Ore
off-loading is another industrial activity in the Tideflats area, however there was no source
profile to represent this activity. It was assumed that the ore off-loading contribution is
primarily in the coarse particle fraction, so it was not used in the CMB.
Results and Discussion
Each CMB calculation was applied separately to data for the average of all samples from the
Alexander Avenue site, for the average of all data from the Morse Supply site, and for the
average daytime and nighttime samples for each site.
Two CMB/source apportionment calculations were performed for each sampling site. The first
used a leaded-gas vehicle signature to calculate the contribution to fine mass from leaded gas
vehicles, which turned out to be less than 1 percent. Fuel consumption data for 1988 from
Chevron, a major distributor in the area, indicates that leaded fuels comprised 25 percent of the
total gasoline sales, while unleaded fuels comprised the remaining 75 percent in the Seattle-
Tacoma area (Fade, 1990). The numbers for 1989 are not expected to be significantly different.
The contribution of leaded fuel exhaust to the fine-mass was multiplied by four to calculate the
total gasoline (leaded plus unleaded) exhaust contribution. If leaded vehicles emit more fine-
83
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
paniculate matter per gallon than unleaded vehicles, which is likely due to the age of leaded
fuel vehicles, then this calculation is an upper estimate for gasoline-powered vehicles.
According to a survey conducted by the PSAPCA (Fade, 1990), all nondiesel vehicles account
for 16 percent of the total grams per vehicle mile of motor vehicle emissions. Therefore,
dividing the mass estimate for all gasoline-powered vehicles by 0.16 yields the total contribution
to fine mass for all motor vehicles, diesel and nondiesel. This estimate is highly uncertain
because of the high uncertainty in the CMB estimate for leaded fuel exhaust (> ±50 percent)
and uncertainties in the fuel use data and vehicle emission inventory.
Results of the CMB/source apportionment using leaded fuel as the tracer for motor vehicles are
depicted in Figures 6-8 and 6-9 for the Alexander Avenue and Morse Supply sampling sites,
respectively. Motor vehicle exhaust contributed only seven to 10 percent of the fine-particle
mass, according to the estimate scaling up the contribution of leaded fuel exhaust. Fine-particle
mass was over-predicted by 16 to 24 percent. Contribution of a source type to fine-particle
mass is normalized by expressing it as the percentage of calculated mass rather than measured
mass, thus simplifying the graphical presentation of the data. Uncertainties in the source
estimates range from 10 to 30 percent, except for the motor vehicle uncertainty, which was at
least 50 percent.
At both sampling sites, the largest fine-particle mass component was residential woodsmoke,
accounting for 62 to 72 percent of the fine-particle mass. By contrast, the industrial
woodsmoke component (hogged fuel emissions) accounted for only two to three percent of the
fine-particle mass. These results are consistent with the prevailing meteorology during the study
in which winds were infrequently from the direction of the Simpson Tacoma Kraft plant.
Furthermore, Simpson shut down its operations during part of the December sampling period.
However, levels of hogged fuel emissions identified by the CMB were similar during operation
and shut-down. These results indicate a small impact of other hogged fuel boilers on the
sampling sites during the study. The Kaiser aluminum production appears to have made a
significant impact on the Alexander Avenue site, which was about 1 km distant and downwind
for most of the study. This source did not make a significant impact on the Morse Supply site,
which was more distant and not frequently downwind. There were several Alexander Avenue
site samples with aluminum loadings from about 1 to 8 /*g/m3, all collected at night.
The U.S. Oil and Refining residual oil boiler was apportioned assuming that the sulfate was
primary [i.e., no sulfur dioxide (SOj) from the boiler (or any other source) was converted to
sulfate before arriving at the sampling sites]. This assumption yields an upper estimate of the
contribution of the residual oil boiler to fine mass.
Including veneer dryers in the CMB resulted in many collinearities, because the most abundant
species in the veneer dryer profile by far is organic carbon, which is emitted in significant
amounts by woodsmoke and other sources. An upper-limit estimate of veneer dryers'
contribution to fine-particle mass was made by comparing annual PM10 emissions of sources
operating veneer dryers with emissions from Kaiser Aluminum, whose fine-particle mass
contribution was estimated successfully. At the Alexander Avenue site, Kaiser contributed
about 12 percent of the fine-particle mass, and, based on meteorology and proximity, probably
had the largest impact on that site of any of the major point sources. Kaiser Aluminum plant's
annual PM10 emissions are about 330 tons/year. Puget Sound Plywood and Buffelen
Woodworking each had annual PM10 emissions of about 44 tons/year. Both of these sources
operate veneer dryers. If their impact on the Alexander Avenue site was the same as Kaiser's,
then they would each contribute 44/330 • 12 percent =1.6 percent of the total fine-particle
mass. Buffelen Woodworking is located to the northwest of the Alexander Avenue sampling
site. Puget Sound Plywood is located in Tacoma, but not in the Tideflats area. It is unlikely
that these point sources had the same potential to affect the Alexander Avenue site as did
84
-------
Chapter 6. Receptor Modeling Study
Figure 6-8. Fine-Particle Receptor Modeling Results at the Alexander Avenue Site Using
Lead (Pb) as the Leaded-Fuel Vehicle Tracer, Total Motor Vehicle Exhaust is Reported*
Aluminum Production (12.0%)
Residual Oil (10.0%)
Scrap Metal (7.0%)
Hogged Fuel (2.0%)
/-Vehicle Exhaust (7.0%)
Residential Woodsmoke (62.0%)
* Fine mass was overpredicted by 16%; values shown are normalized so that the sum is 100%.
Figure 6-9. Fine-Particle Receptor Modeling Results at the Morse Supply Site Using Lead
(Pb) as the Leaded-Fuel Vehicle Tracer, Total Motor Vehicle Exhaust is Reported*
Residual Oil (10.0%)~\
Scrap Metal (5.0%)
Hogged Fuel (3.0%)
/-Vehicle Exhaust (10.0%)
Residential Woodsmoke (72.0%)
* Fine mass was overpredicted by 16%; values shown are normalized so that the sum is 100%.
85
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Kaiser, given the location of those sources and the prevailing meteorology (mostly south to
southeast winds). Furthermore, only a portion of the total PM10 output of those sources would
be due to veneer dryers. It is therefore concluded that the veneer dryers can be regarded as
insignificant in the CMB and the source apportionment.
The second CMB/source apportionment used the VOCs o-xylene and 2,2,4-trimethylpentane
as tracers for total vehicle exhaust. The abundance of each of these VOCs with respect to fine-
particle mass was used as the motor vehicle source profile in separate CMB calculations to
estimate the motor vehicle contribution to fine particles. Results using o-xylene as a motor
vehicle tracer indicate that motor vehicles contribute 18 to 20 percent to the calculated fine
particles (Figures 6-10 and 6-11). Results using 2,2,4-trimethylpentane as a motor vehicle
tracer indicate that motor vehicles contribute three to four percent of the calculated fine-particle
mass (Figures 6-12 and 6-13). These results compare well with the seven to 10 percent
estimate from the other motor vehicle apportionment described above, considering the
uncertainty of the estimates. There is little difference between the fine mass contribution of the
sources other than vehicle exhaust calculated by the two CMB/source apportionment methods.
Collinearity Assessment in the CMB Calculation
The CMB program contains diagnostic procedures to identify collinearity among the source
profiles or high uncertainties in the individual profiles. For the most part, no collinearities were
detected by the CMB program's diagnostic procedure for the samples presented in this report,
with the exception of average nighttime samples collected at the Alexander Avenue site (Table
6-6).
The CMB values indicate that woodstoves were by far the dominant source of fine-particle mass
during the study. The magnitude of the woodstove contribution was verified by performing the
CMB calculations without the woodstove source profile. Organic carbon and potassium have
significant contributions from residential woodsmoke, but they are not unique tracers of
residential woodsmoke in this airshed. Leaving these species in the CMB without the
woodstove source profile violates one of the fundamental assumptions of the CMB calculations.
Therefore, CMBs without woodstoves were calculated first excluding and then including organic
carbon and potassium as fitting species. Leaded-fuel vehicle contributions were scaled up to
estimate total motor vehicle exhaust, as detailed above (Vehicle Exhaust I).
The sum of individual source contributions was 44 +. 7 percent of the measured fine particles
for the Alexander Avenue site and 36 ±. 7 percent for the Morse Supply site for the source
apportionment without the woodstove source profile and without organic carbon and potassium
as fitting species. In comparison, source contributions summed to 57 i 7 percent (Alexander
Avenue) and 39 +_ 7 percent (Morse Supply) when organic carbon and elemental carbon are
included as fitting species while still excluding woodstoves. The source apportionment
including the woodstove source profile (and organic carbon and potassium) estimated that the
sum of non-woodsmoke sources accounted for 44 +. 6 percent of the measured fine-particle
mass at the Alexander Avenue site and 34 +. 6 percent at the Morse Supply site.
There is little difference between the CMBs including and excluding the woodstove profile. The
small increase in the non-woodstove estimate when organic carbon and potassium are included
as fitting species in the absence of a woodstove profile indicates the CMB is fairly insensitive
to collinearities between woodstove and non-woodstove signatures.
86
-------
Chapter 6. Receptor Modeling Study
Figure 6-10. Fine-Particle Receptor Modeling Results at the Alexander Avenue Site Using
o-xylene as the Motor Vehicle Tracer*
Aluminum Production (10.7%)
Residual Oil (8.9%)
Scrap Metal (6.3%)
Hogged Fuel (1.6%)
Vehicle Exhaust (17.6%)
Residential Woodsmoke (54.9%)
* Fine mass was overpredlcted by 31%; values shown are normalized so that the sum is 100%.
Figure 6-11. Fine-Particle Receptor Modeling Results at the Morse Supply Site Using
o-xylene as the Motor Vehicle Tracer*
Residual Oil (9.0%)-^
Scrap Metal (5.0%)
Hogged Fuel (2.4%)
Vehicle Exhaust (19.6%)
Residential Woodsmoke (64.0%)
* T7i
Fine mass was overpredicted by 39%; values shown are normalized so that the sum is 100%.
87
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 6-12. Fine-Particle Receptor Modeling Results at the Alexander Avenue Site Using
2,2,4-trimethylpentane as the Motor Vehicle Tracer*
Aluminum Production (12.6%)
Residual Oil
Scrap Metal (7.4%)
Hogged Fuel (1.9%)
Vehicle Exhaust (3.4%)
Residential Woodsmoke (64.3%)
* Fine mass was overpredicted by 12%; values shown are normalized so that the sum is 100%.
Figure 6-13. Fine-Particle Receptor Modeling Results at the Morse Supply Site Using
2,2,4-trimethylpentane as the Motor Vehicle Tracer*
Residual Oil (10.7%)^
Scrap Metal (5.9%)
Hogged Fuel (2.8%)
r-Vehicle Exhaust (4.5%)
Residential Woodsmoke (76.1%)
* Fine mass was overpredicted by 17%; values shown are normalized so that the sum is 100%.
88
-------
Chapter 6. Receptor Modeling Study
Table 6-6. Percent Contribution of Major Sources to Calculated Fine-Particle Mass for
Daytime, Nighttime, and All Samples - Alexander Avenue Site
Day Night All
Source
Residential woodsmoke
Aluminum production
Residual oil boiler
Motor vehicles
Scrap metals
Hogged fuel
% of
calculated
mass
63 + 7
3 + 2
16 + 3
8 + 5
8 + 1
2 + 0.3
}
% of
calculated
mass
87 ±8
5 + 3
6 + 1
2 + 0.3
% of
calculated
mass
62 + 7
12 + 3
10 + 3
7 + 4
7 + 1
2 + 0.3
Day versus Night
CMB calculations using leaded vehicles to apportion mobile sources were performed for daytime
and nighttime averaged samples. Results are presented in Tables 6-6 and 6-7. The errors
represent the standard error estimated by the CMB. For the Morse Supply site, there was little
difference between the day and night sample source apportionments, considering the limits of
the standard errors of the source contributions estimated by the CMB program. Automobile
exhaust emission estimates varied little from day to night within the large estimated
uncertainties. The scrap metal source was found to have a greater impact during the day than
at night, indicating that it is associated with mostly daytime activities. For the Alexander
Avenue site, the nighttime average source contributions could not be completely resolved.
Woodsmoke, residual oil boilers, and aluminum production were grouped together as one source
contribution. However, daytime average source contributions for the Alexander Avenue site
were completely resolved, and they indicate that the aluminum production source contributed
significantly less to the fine-particle mass for daytime samples (3 +. 2 percent) compared with
daytime plus nighttime averaged samples (12 +. 3 percent). It can therefore be inferred that
aluminum production activities were, on average, significantly higher at night than during the
day. This is consistent with the ambient data in which nighttime average aluminum
concentrations were 816 ng/m3, compared with an average of 143 ng/m3 aluminum for the
daytime samples. The three samples with aluminum concentrations in the 1 to 8 ng/m3 range
were all nighttime samples. The diurnal pattern of aluminum cannot be explained by nighttime
stagnation, as other elements measured in the fine-particle fraction showed no such diurnal
pattern, (i.e., very high nighttime concentrations; see Tables 6-1 (a) and (b). Auto exhaust
and scrap metals at the Alexander Avenue site showed the same diurnal behavior as at the
Morse Supply site. The scrap metal handling source type contributed from 20 percent (Morse
Supply, nighttime) to 34 percent (Alexander Avenue, daytime) of the fine-particle lead. This
makes scrap metal activities competitive with mobile sources as a source of fine-particle lead.
89
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 6-7. Percent Contribution of Major Sources to Calculated Fine-Particle Mass for
Daytime, Nighttime, and All Samples - Morse Supply Site
Day Night All
% of % of % of
calculated calculated calculated
mass mass mass
Residential woodsmoke 68 ± 7 76 ± 7 12 ±1
Residual oil boiler 12 ± 3 10 ± 3 10 ± 3
Motor vehicles 12 ±6 8±4 10 ±5
Scrap metals 7±1 4±1 5±1
Hogged fuel 2 ± 0.3 3 ± 0.3 3 ± 0.4
December versus January
Source apportionments were performed separately for the December and January samples
because of the differences in meteorology between December (dry, lower wind speeds) and
January (wet, higher wind speeds) and the differences in total fine-particle mass (about 30 ng/m3
in December versus about 10 /ig/m3 in January). Results are presented in Table 6-8.
Differences between the December and January source apportionments at either site were small.
The hogged fuel boiler source had a greater impact on the Morse Supply site in January (5 ±
0.8 percent) than in December (2 ± 0.2 percent). This is consistent with the shutdown of the
Simpson Tacoma Kraft hogged fuel boiler during most of the December sampling period. The
aluminum production impact on the Alexander Avenue site in January (17 ± 4 percent) was
higher than in December (11 ± 3 percent).
Table 6-8. Percent Contribution of Major Sources to Total Fine-Particle Mass for
December and January Samples
Alexander Site Morse Site
December January December January
Residential woodsmoke
Aluminum production
Residual oil boiler
Motor vehicles
Scrap metal
Hogged fuel
Measured mass, ng/m3
Calculated mass, ng/m3
Percent overestimated
63 + 7
11 + 3
10 + 3
7 + 4
7 + 1
2 ±0.3
29,500
33,535
14
57 + 7
17 + 4
12 + 4
5±3
6 + 1
2 ±0.3
8,910
12,345
38
72 + 7
0
11 + 3
10 + 5
6 + 1
2 ±0.2
31,200
40,839
31
70 + 7
0
9 + 3
11 + 6
5 + 1
5 ±0.8
11,400
12,730
12
90
-------
Chapter 6. Receptor Modeling Study
SUMMARY
Aerosol (PM10)
> The outstanding feature of the data at both sites (Morse and Alexander Avenue) was the
high relative amount of fine-particle organic compounds compared with the total mass.
* High fine-particle aluminum concentrations measured at the Alexander site were associated
with light southeast winds or stagnant conditions.
> Coarse-particle aluminum concentrations were much higher and measured with lower
uncertainty at the Morse site than at the Alexander site.
* The Morse site tended to have somewhat higher concentrations of most fine- and coarse-
particle constituents and VOCs. An exception was the copper concentration at the
Alexander site, which may be an artifact of sampler operation.
* Coarse particle iron concentrations were twice as high at the Alexander site, while fine-
particle iron concentrations were 50 percent higher at the Morse site. Coarse-particle
chromium concentrations were three times higher at the Morse site.
* Organic carbon and fine-particle mass data correlated quite well between the two sites.
There were similar temporal patterns of all of the fine-particle species.
* The highest concentrations of fine-particle total mass and organic carbon were associated
with the most stagnant conditions of the study. The winds were mostly from the southeast.
The Simpson Tacoma Kraft hogged fuel boiler was not operating during this period.
* Insufficient loadings of PAHs in the 18-day ambient sampling study precluded their use
in the receptor modeling study.
Receptor Modeling of PM2-S Emissions Sources
[It should be kept in mind when evaluating the study results that the measurements were made
over a limited time and spatial scale. Meteorology and season can strongly influence the
contribution of the various sources of contaminants to paniculate mass. This study covers only
a subset of conditions which may exist in the area in any one year and from year to year.]
* Residential woodsmoke was found to be by far the dominant source of fine particles,
accounting for 62 to 72 percent of the fine-particle mass by one estimation and 55 to 64
percent by the other.
* Residential woodsmoke accounted for a similar percentage of the fine-particle organic
carbon.
> Motor vehicle exhaust, residual oil boilers, and aluminum production at the Alexander
Avenue site were a distant second, each accounting for around 10 percent of the fine-
particle mass. (This estimation used lead as a tracer for leaded fuel vehicles.)
* Using selected VOCs as tracers for motor vehicle exhaust, estimates of this source
contribution ranged from three to 20 percent.
* Diesel powered vehicles accounted for most of the motor vehicle emissions.
91
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
* Scrap metal handling and similar activities contributed five to seven percent of the fine-
particle mass and more than 20 percent of the fine-particle lead, making these activities
competitive with leaded fuel vehicles as a source of fine-particle lead in the Tideflats
airshed.
» Industrial woodsmoke from hogged fuel boilers was estimated to contribute three percent
or less of the fine-particle mass. (Note: Winds were infrequently from the direction of
the Simpson Tacoma Kraft plant during the study. Also, the plant shut down operations
during part of the December sampling period.)
* No significant differences were observed between the averaged daytime and nighttime
samples, with the exception of aluminum production, which contributed significantly less
fine-particle mass during the day, and by inference, significantly more at night.
* Any differences between daytime and nighttime for motor vehicles were overshadowed by
the large uncertainty in their estimates.
*• The percentage that woodsmoke contributed to total fine-particle mass did not vary
diurnally.
92
-------
Chapter 1. Diffusion/Transport Modeling
OBJECTIVES
>• Develop a customized, state-of-the-art diffusion model to predict the aerosol and deposition
concentrations of contaminants in the Tacoma Tideflats area.
* Develop a model to estimate the fractions of contaminants entering the Commencement
Bay stormwater runoff following deposition on the watershed.
* Run these models for the entire field sampling period, using the meteorological and
emissions data to drive the daily simulations for transport and deposition.
The analysis of ambient aerosol and deposition samples is a labor- and resource-intensive
method to assess atmospheric contributions of contaminants. As a consequence the number of
possible samples and sampling sites is limited. A numerical model provides an alternative
approach to the problem: it can be an efficient tool for processing emissions source and wind
data into predictions of contaminant concentrations and fluxes as functions of time and space.
WV3 EULERIAN GRID DISPERSION MODEL
A mathematical model was developed to simulate the emission, dispersion, and deposition of
toxic contaminants, through the atmosphere and into the watershed of Commencement Bay.
WYNDvalley, version 3.01 (hereafter WV3), is a time-dependent Eulerian grid simulation that
integrates dispersion and deposition equations in three dimensions with arbitrarily varying
emissions, both in time and space. The model permits flexible boundary conditions and the
separation of the deposition velocity, Vs, into steady and time-varying components, such as
rain-modulated removal of soluble contaminants. Appendix F-l contains a discussion of the
model equations.
WV3 (and earlier versions of the model) have been compared with observations and with
predictions from RAM, a standard Gaussian Plume model, for PM10 at six sites in the Pacific
Northwest, including data from 1985 and 1986 at Fire Station No. 12 in the Tacoma Tideflats.
The earlier version of WV3 proved superior to RAM in five out of six scoring criteria, which
depend on matching observations with simulations, both in time and space. For the sixth,
neither model was superior to the other (Harrison et al., 1990).
The modeling domain for the present study is an irregular area comprising 91.5 km2 in the
Tacoma Tideflats. WV3 is implemented in three vertical layers of 356 cells, each one being
500 x 500 meters wide and 67 meters high. Time-varying source intensities (in this case, PM10
emissions) are specified for each cell in the surface layer and for cells in the middle layer that
are associated with stacks emitting contaminants at heights between 67 and 133 meters. The
PM10 size fraction was used in the diffusion modeling. The aerosol sampler used for the aerosol
and deposition studies collected aerosol particles up to a cutpoint between 25 and 50 /xm,
therefore the modeled concentrations are expected to be underestimates of the measured values.
93
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Emissions are modulated as one of four daily schedules (industrial, woodstoves, automotive or
dayshift), and by workday/weekend. Figure 7-1 shows the modulation functions used to
simulate automotive and woodsmoke emissions over a 24-hour period. Measured wind
directions are rounded into one of eight primary octants (n, ne, e, etc.) and time steps are
computed as functions of wind velocity to minimize numerical diffusion. Sub-scale eddy
diffusion (Kx, Ky, Kz) is added from measurements. Depositions are separated into wet and
dry components, the former proportional to rain rates through a washout coefficient, Wr, and
the latter as a constant dry deposition velocity, Vd.
Model output is copious, with records of 24-hour averages of the contaminant concentrations
and depositions in every surface cell, and hourly averages at seven receptor sites that were
placed on the grid at locations representing Brown's Point, Tyee Marina, Morse Industrial
Supply, Fire Station No. 12, Sea-Land, Alexander Avenue, and Riverside School (Figures 2-2
and 7-2). Hourly concentrations simulated at the seven receptor sites were averaged over
three- to four-day intervals, as appropriate, to coincide with those of the field aerosol
measurements.
MODEL INPUT
The model is used to simulate the distribution of the concentrations of atmospheric contaminants
and their fluxes to the surface. Required inputs to the model include the spatial and temporal
distributions of the sources of emissions, the dispersing winds, and the wet and dry processes
that remove the contaminants from the modeled domain.
Point source PM10 emission information was taken from the PSAPCA registration files, the
source profile library, and relevant source tests as described in Chapter 5. The actual input data
used by the model appear in Appendix F-2. Emission estimates were included for all sources
exceeding 1 kg/day of PM10 emissions, and located within 10 km of the Alexander Avenue
site. The source profiles estimate the weight fractions of the various tracers compiled from
source tests. Appendix F-2 (c) contains profiles for Cu, Zn, As, and Pb, and Appendix F-2
(d) contains profiles for several PAHs. Emissions from the Kaiser aluminum smelter (Table
7-1) are a special case. These were measured directly (AMTEST, 1988). It should be noted,
however, that the uncertainties in these numbers may exceed 40 percent (based on a comparison
with previous test results).
The total estimated PM10 emitted in the modeled area was 5,692 kg/day. There are 59 point
sources, with emissions totaling 4286 kg/day, of which 1014 kg are vented below 67 meters
and the remainder above. Emissions from area sources (vehicle exhaust, road dust, railroads,
ships, woodstoves) account for the remainder. Table 7-2 shows the total estimated emissions
used in the diffusion/transport modeling. Estimates are listed for PM10, Cu, Zn, As, Pb, the
sum of the four metals, several PAHs, and the sum of the PAHs. Note that 98 percent of the
modeled PAH emissions were from Kaiser.
Figures 7-3 and 7-4 are block contour maps showing estimated PM10 emissions at the surface
(level 1) and from elevated stacks (level 2, or mid-level).
94
-------
Chapter 7. Diffusion/Transport Modeling
Figure 7-1. Modulation Function for Cars and Woodsmoke
2.8
l.B
e.e
.uoodsnoke
i a o aID
cars
a o a a
8
- tine of day [PSN —>.
24
Figure 7-2. Map of Sampling Sites Used for Modeling
1 - Brown's Point (BP)
2 - Tyee Marina (TM)
3 - Fire Station No. 12
4 - Sea-Land (SL)
5 - Alexander Avenue (AS)
6 - Morse Industrial Supply (MS)
7 - Riverside School (RS)
Dir= 51 deg
Vel= 0.2 n/s
Tine =
Frane-
3.30 hours
19
Level = 1
Not raining
MENU <— M
95
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 7-1. PAH Emissions from the Kaiser Aluminum Smelter
kg/day
204.0
2.4
46.6
30.1
190.0
13.1
31.1
20.4
1.27
3.43
1.39
0.72
0.34
0.24
0.12
0.24
0.72
Code
NAPH
ACNY
ACNE
FLOR
PHEN
ANTH
FLUO
PYRE
BAA
CHRY
BBF
BKF
BAP
INDP
DBA
BZP
BEP
Comment
naphthalene
acenaphthylene
acenaphthene
fluorene
phenanthrene
anthracene
fluoranthene
pyrene
benzo(a)anthracene
chrysene
benzo(b)fluoranthene
benzo(k)fluoranthene
benzo(a)pyrene
indeno(l ,2,3-)pyrene
dibenzo(a,h)anthracene
benzo(g,h,i)perylene
benzo(e)pyrene
Total 546.2
Table 7-2. Total Estimated Emissions of PM10 for Diffusion/Transport Model
kg/day
PM10 (excluding PAH) 5692
Copper 3.85
Zinc 9.88
Arsenic 0.74
Lead 7.07
Cu+Zn+As+Fb 21.54
Fluorene 30
Phenanthrene 190
Anthracene 13
Fluoranthracene 41
Pyrene 20
Chrysene 4
Benzo(a)pyrene 0.3 .
Retene , 0.1
All other Kaiser PAHs 261
AllPAHs 560
All Kaiser PAHs 546
Non-Kaiser PAHs 14
NOTE: 1. 98% of all PAHs are from Kaiser.
2. PAHs comprise 10% of PM10.
96
-------
Chapter 7. Diffusion/Transport Modeling
Figure 7-3. Distribution of Surface Emissions of PM
10
0ir=294 deg Tine = 0.61 hours Level = 1 No* raining
Vel= 1.9 n/s Fr»ne= 10 MENU <— M
Figure 7-4. Distribution of Mid-Level Emissions of PM
10
3880
'18
m< '88
< 18
D!r=105 deg
Vel= 1.3 n/s
Tine =
Frane=
6.25 hours
34
Level = 2
Not r«In i ng
MEMU <~ M
97
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
WINDS AND TRANSPORT COEFFICIENTS
The meteorological data gathered at the Alexander Avenue site were processed into daily
summaries beginning at noon on July 7, 1989, and ending at noon on January 7, 1990. The
velocities and directions were combined as vector averages, the insolation as scalar averages,
and the precipitation as simple sums. For each 24-hour period percentages were computed
when the winds were within 22.5° of north, north-east, east, etc., and the percentage of
five-minute winds less than one meter per second. This summary is presented in Appendix
B-2, with dates labeling the first day of each 24-hour period, starting at noon.
Figure 7-5 shows a polar (wind-rose) scattergram of the 24-hour vector winds for all the data,
and Figure 7-6 shows a cumulative trajectory for winds at 10 meter height. A similar
scattergram to Figure 7-5, which was selected for winds when it is raining, reveals that rainy
winds were mostly southerly. Figure 7-6 reveals a dramatic "seasonal corner" on October 5th,
when there is a shift from summer winds with a set westerly flow to winter winds with a set
southerly flow.
The quality and volume of the meteorological data collected for the present study greatly exceed
that which is normally available for atmospheric dispersion studies. In particular, the horizontal
variances in wind directions and speeds, combined through Monin-Obhukov similarity theory
with the vertical temperature gradients, permit explicit determination of the horizontal and
vertical eddy diffusivities.
The vertical eddy diffusivities (Kz) are summarized in Appendix F-3(a) as functions of wind
velocity and the vertical temperature gradient between two and 10 meters above the surface.
The AT measurement errors affect calculations of Kz, when atmospheric stabilities are near
neutral. A +0.1° measurement bias of AT (between thermistors separated by eight meters) at
a wind speed of 1 m/s would reduce the perceived Kz from 1.66 mY1 to 0.67 mV1 [see
Appendix F-3(a)]. The sensitivity of contaminant concentrations computed by WV3 to estimates
of Kz varies markedly with wind speed and direction, but is everywhere else less than 0.3.
Thus, a +0.1 ° bias in AT would include positive errors in computed tracer concentrations that
should be everywhere less than 45 percent.
Appendix F-3(b) contains two scattergrams, one demonstrates approximate linearity of the
horizontal eddy diffusivities with the wind velocities, and the second shows that long-term
averages of the transverse and longitudinal horizontal eddies are equal (i.e. that the horizontal
turbulence is not dependent on the horizontal direction.
WV3 also requires deposition velocities and washout coefficients to estimate the efficiencies
with which suspended particles are deposited on the surface by dry- and rain-modulated
processes. For the contaminant tracers of this study, median values for identical or chemically
similar tracers, as reported by McMahon et al. (1979), were assumed (see Table 7-3).
98
-------
Chapter 7. Diffusion/Transport Modeling
Figure 7-5. Wind-Rose Scattergram for 24-Hour Vector Winds
Npts= 186 (Number of Points)
Wind-rose Scattergram for Tacoma Winds. 1 circle = 1 m/s
99
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 7-6. Wind Hodograph for 24-Hour Vector Winds at 10m Height
Npts = 186 (Number of Points)
< •
i
i
/
i
.'•
J
f
/
/
/ /
1 /
I 1
i
\\
\ \
\ \
\
"•"-,
"-<
..•-""
'"' ^
m
jf
__i—
' /"'
/ ,
! (
!
\v
\ •
x- N
\
V
\ ^
~-^-._^
-""
^
--•"'"
/ ,--
f /
(
\
"""
V.._
-,^_
^
*" ~ — "-—
^"""-— -..
^^x
"v
i
--N \
-A '
I ^_- -.
. /' 1
'^
.-'
^^
__^
"•— .
"--._
^ ^
^« !
-, '\
X
\
••_ \
-. 1
\ \
'•. %
\ \
11
I/ /
/
/ /
.»'
K X
^
\
\
\ \
\\ \
V\ \
/ \ \
f, \ ]
'}
\
-4 I j
/ /
/ /
/
/
/
'
Cumulative Trajectory for Tacoma Winds July 7, 1989 to January 7, 1990
100
-------
Chapter 7. Diffusion/Transport Modeling
Table 7-3. Empirical Atmospheric Deposition Parameters
Tracer Vd (cm/sec)(a) Wr M
Arsenic
Copper
Lead
Zinc
PAH
0.1
0.1
0.1
0.1
0.1
1000
1000
1000
1000
300
(a) Vd is defined as the dry deposition velocity.
(b) Wr is defined as (concentration of a tracer in rain)/(concentration of the tracer in air)
RESULTS: SIMULATIONS
The WV3 model was run separately for the inventory emissions of As, Cu, Pb, Zn, and for a
generic PAH compound that was assumed from the Kaiser plant only, at a nominal emission
rate of 1 kg/day. [Because greater than 90 percent of the PAH compounds that are emitted in
the Tacoma Tideflats derive from a single source (the Kaiser aluminum smelter), a useful
approximation is to simulate these as if from a single tracer emitted at a standard rate (1
kg/day).] The hourly concentrations from these simulations were averaged over the same three-
or four-day intervals as the aerosol sampling at each site (see Appendix F-4).
Simulations of Concentrations and Fluxes
A principal reason for attempting to model the concentrations and fluxes of concentrations
through the air and into the watershed of Commencement Bay is to develop both interpolative
and extrapolatiye tools that may be used at other sites within the modeled domain and at
different domains entirely. For this purpose, estimates of the spatial distributions of the
concentrations and fluxes are crucial.
Estimates using the WV3 model are displayed in Figures 7-7 through 7-11 (each with subscripts
a-c). Concentrations and fluxes are estimated for As (Figure 7-7), Cu (Figure 7-8), Pb (Figure
7-9), Zn (Figure 7-10), and a generic PAH tracer (Figure 7-11) that is assumed to have been
emitted from Kaiser with a flux of 1 kg/day. (Concentrations and flux values for individual
PAHs can be acquired by multiplying by the Kaiser emissions for that PAH; the patterns will
remain identical.)
Subscripts in Figures 7-7 through 7-11 represent the following for the pollutant of concern:
(a) The highest 24-hour average concentration;
(b) The 186-day average aerosol concentration; and
(c) The sum of the dry and wet depositions.
Model output simulations for the highest brief episode (five minute), the second-highest 24-
hour aerosol concentrations, dry deposition fluxes, and wet deposition rates for each of the five
chemical constituents appear in Appendix F-5. Comparisons of modeled concentrations with
measured concentrations and an evaluation of model performance are addressed in some detail
in Chapter 8, Comparison of Studies.
101
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 7-7(a). Simulated Highest 24-Hour (midnight-to-midnight) Averages for Arsenic
(As) Aerosols
Highest 24-hour average, lower level
Figure 7-7(b). 186-Day Average for Arsenic (As) Aerosols
388 jH»|j«HHp»!J|jHspH[«l<^!i>ipji™j<>8«™
Average Concentration over 186 days
102
-------
Chapter 7. Diffusion/Transport Modeling
Figure 7-7(c). Sum of Wet and Dry Depositions for Arsenic (As) Aerosols
388
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 7-8 (b). 186-Day Average for Copper (Cu) Aerosols '
Average Concentration over 186 days
Figure 7-8(c). Sum of Wet and Dry Depositions for Copper (Cu) Aerosols
XXXXKXXXXXXXXKXKXXXM
EXKXXXXXXjtXXXKXXXX
m&xxxxkxxKxxxVxKx- - -
Average Total Deposition Rate: Milligrams/hectare/day
104
-------
Chapter 7. Diffusion/Transport Modeling
Figure 7-9(a). Simulated Highest 24-Hour (midnight-to-midnight) Averages for Lead (Pb)
Aerosols
476
418
361
246 <=< 303
188
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Rgure 7-9 (c). Sum of Wet and Dry Depositions for Lead (Pb) Aerosols
Average Total Deposition Rate: MilligraMS/hectare/day
Figure 7-10(a). Simulated Highest 24-Hour (midnight-to-midnight) Averages for Zinc (Zn)
Aerosols
616
531
446
361
276
itt
21 < '•'<
Highest 24-hour average, lower level
106
-------
Chapter 7. Diffusion/Transport Modeling
Figure 7-10(b). 186-Day Average for Zinc (Zn) Aerosols
XXXXXX XXXXXXV XXX\XX
XXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXX
Average Concentration over 186 days
Figure 7-10(c). Average Total Deposition Rate (Wet + Dry) for Zinc (Zn) Aerosols
300
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 7-ll(a). Simulated Highest 24-Hour (midnight-to-midnight) Averages for Vapor +
Particulate PAH
.......
Highest 24-hour average, lower level
Figure 7-ll(b). 186-Day Average for Vapor + Particulate PAH
Average Concentration over 186 days
108
-------
Chapter 7. Diffusion/Transport Modeling
Figure 7-ll(c). Sum of Wet and Dry Depositions for Vapor + Participate PAH
XXXKXXXXXKX
XXXXXXXXXXW
XX XXVx KJ&tXXXVw X
XXXKXXXXXKXXXXXXXXXX
XMXKXXXXXKXXXXXXXXXXXXXXX
XM X &XX&XXXXX XXVXXXVXM!
KXXXXXXXXKXXXMXXXXXXXXXXXXXXX
XXXKXXXXXXXXXXXXXXXXXJWXX
"° "'•"""""'•'""•""*'""""'"•j^jf
XXXXXXXMXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXX
XXKXXKXXXXXXXXXHX
XXXXX
KXXKXXX
xxxxxxxx
ixxi&xxx*
XXXXKXXK
XKXXKXX,
ixxxSfxxx i
xxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
X HH X
HMHWKXM.^
XMXKXXXXXXXXXyXXXX
XXXXKXXXXXXMXXXXXXXX
'*
KXXXXXXXXKXXKXXKXXXMXXXXXXXXXXX
XXXXXXKXXXXXXHXXHMKHX•-
Average Total Deposition Rate:
Mobilization into Runoff to Commencement Bay
The "bottom line" for the simulations, and indeed for the study itself, is estimates of the
fractions of each contaminant that are deposited onto the watershed and actually reach
Commencement Bay. For these estimates, not only are the spatial distributions of the fluxes
of each contaminant to the surface needed (Figures 7-7 to 7-11), but also the spatially-dependent
efficiencies with which each contaminant in each grid cell is mobilized into the runoff to
Commencement Bay (mobilization coefficients). These coefficients have not been
experimentally determined for the study area. Instead, annual average mobilization coefficients
were estimated with a linear washoff model computed by cell for the area of the
Commencement Bay watershed modeled by the diffusion model. Insufficient data were
available to allow the calculation of mobilization coefficients for individual metals and PAHs.
Therefore, for modeling purposes, both metals and PAHs were assumed to be completely
associated with fine sediments, and were considered as a single sediment-bound contaminant.
The development of the Mobilization Coefficient Model is summarized here and the results for
the Tacoma Tideflats area are contained in Appendix F-6. A stormwater transport model was
developed to serve as the basis of the contaminant transport model. Surface runoff doesn't
travel far in the Tacoma Tideflats area before it either enters a drainage pipe or ditch, or
infiltrates into the soil. For flow in drainage pipes, bare impervious ditches, and bare pervious
ditches that contain water throughout most of the year 100% of the contaminants are assumed
to be transported. For ditches that are dry throughout most of the year, but have pervious
linings and for ditches that contain substantial vegetation, all metals and PAHs are assumed
to be bound to sediment particles and completely removed by filtration and settling. The
flatness of the area provides little driving force for overland flow, thereby enhancing stormwater
109
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
infiltration and settling. In the developed industrial area north of Interstate 5 pipe flow is more
important than in the rural area south of the Interstate. To model contaminant transport within
each cell, watershed data including land use, SCS hydrological soil group, surface cover type,
and drainage pathways was obtained from aerial photos and site inspections.
The phase in which a contaminant exists largely determines its environmental fate. This
information was not known for the contaminants considered here. Degradation by bacteria,
sunlight, and hydrplytic actions could change the chemical form what was originally deposited.
In addition dramatic pH changes can occur from bock to block in industrial areas. To simplify
the development and implementation of the Mobilization Coefficient Model, it was assumed that
100% of all metals and PAHs which accumulate during a typical dry period are sorbed to fine-
particulate matter by the beginning of the next rainfall.
The mobilization coefficient is a number between 0 and 1 that represents the fraction of the
land-deposited contaminants that will be carried by surface water runoff, by groundwater, or
through a series of pipes into the waterway. Where the cell is over Commencement Bay or the
Puyallup River the coefficient is unity, and where it is partially over the Bay, a lower limit to
the mobilization coefficient is that fraction of the cell that is subtended by water. For cells over
land areas, the linear model predicts that less than 10 percent (mobilization coefficient less than
or equal to 0.1) of the deposited contaminants will reach Commencement Bay through "prompt
runoff (runoff occurring during the period of this study). Factors contributing to this relatively
low value include the existence of significant areas of vegetated open space, unlined vegetation
drainage ditches, and the flatness of the land. Another factor in the low mobilization is the
predominance of small storms in the annual rainfall pattern of the Pacific Northwest. For
exceptional storms, where large and sudden flows mechanically scour the drainage ditches, it
is plausible that deposits that have accumulated over fairly long periods may be flash-mobilized.
These deposits will likely have a reduced contaminant load, as compared with the original
deposited material, due to photolysis, biodegradation, biological uptake or vertical migration
through soils.)
Since mobilization coefficients were never experimentally determined, there are no values for
comparison with the computed model values, and therefore the errors associated with the model
values cannot be precisely determined. Potential sources of error include the "typical year"
estimate of rainfall, the assumption that contaminants are bound to particles, the washoff model,
the deposition estimates, and the assumption that contaminants not washed off by one storm will
be unavailable for washoff by a later storm event. Without verification and error analysis the
model can only provide rough, semi-quantitative estimates of pollutant mobilization coefficients.
Because of the above-noted uncertainties in the mobilization coefficients, the total fluxes to
Commencement Bay have been computed twice: once with those coefficients derived by the
Mobilization Model (Appendix F-6), and once with the same coefficients modified so as to
never be less than 0.1 in any cell. The results of these computations are presented in Table
7-4.
110
-------
Chapter 7. Diffusion/Transport Modeling
Table 7-4. Estimates by the WV3 Model of Fractions of Emitted Toxic Contaminants
Deposited Within the Modeled Domain (91.5 km1) and Mobilized Into the "Prompt-Runoff"
Watershed of Commencement Bay;
I. With washoff coefficients estimated in Appendix F-6:
Source Fractions of Emissions
Emissions Deposition/Emission Transported
Modeled Domain Commencement Bay
Q (kg/day) Wet/Q Dry/Q Sum/Q W/Q D/Q Tot/Q
Arsenic
Copper
Lead
Zinc
PAHs
0
3
7
9
1
.74
.85
.07
.88
.00(*)
4
5
3
3
1
.74
.53
.65
.36
.33
2.46
2.47
2.42
2.40
0.85
7.
8.
6.
5.
2.
20
00
07
76
18
1.
2.
1.
1.
0.
97
12
53
48
48
1.42
1.09
1.47
1.55
0.33
3.39
3.20
2.99
3.03
0.80
II. With washoff coefficients estimated so that they are never less than 0.1 in any cell:
Source Fractions of Emissions
Emissions Deposition/Emission Transported
Modeled Domain Commencement Bay
Q (kg/day) Wet/Q Dry/Q Sum/Q W/Q D/Q Tot/Q
Arsenic
Copper
Lead
Zinc
PAHs
0.74
3.85
7.07
9.88
1.00(*)
4.74
5.53
3.65
3.36
1.33
2.46
2.47
2.42
2.40
0.85
7.20
8.00
6.07
5.76
2.18
2.36
2.62
1.83
1.76
0.60
1.57
1.29
1.60
1.68
0.41
3.93
3.91
3.43
3.43
1.01
NOTES:
(*) PAH emissions are nominal, and should be renormalized for each separate PAH
compound (see Table 7-1).
Wet/Q values are the ratios of all of the rainborne deposits to all the emissions,
within the modeled domain, expressed as percents.
Dry/Q values are the ratios of all of the dry deposits to all the emissions, within
the modeled domain.
Sum/Q = Wet/Q + Dry/Q.
W/Q are the ratios of wet deposits times washoff coefficients, to all of the
emissions. These are the fractions of emitted contaminants that are promptly
deposited or washed into Commencement Bay by rain.
Ill
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
D/Q are the ratios of dry deposits times washoff coefficients, to all the emissions.
These are the fractions of emitted contaminants that are either dry-deposited into
Commencement Bay, or dry-deposited onto the nearshore watershed and washed
into Commencement Bay by subsequent runoff.
Tot/Q = W/Q + D/Q Fractions of all emissions transported into Commencement
Bay.
1. Assumed dry deposition velocities : 0.1 cm/s
2. Assumed washout ratios (metals) : 1000
(PAH) : 300
(A washout ratio is the concentration of a tracer in rain/concentration of the tracer
in air.)
From Sum/Q it can be noted that greater than 90 percent of the emitted tracers (PMW size
fraction) are advected beyond the boundaries of the modeled domain. Perhaps one to three
percent of emitted contaminants reaches Commencement Bay directly within the
"prompt-runoff watershed. Based on the particle deposition processes as modeling in WV3
and the mesoscale meteorology and topography of Puget Sound, it is likely that an additional
three to four percent of the airborne tracers (As, Cu, Pb, Zn, and PAHs) emitted in the Tacoma
Tideflats will reach Puget Sound beyond the 91.5 km2 modeling domain. This projected loading
is summarized in Table 7-5.
Table 7-5. Projected Mass Loading of Airborne Toxic Contaminants to Puget Sound
Based on Emissions in 91.5 km1 WV3 Modeling Domain
Estimated Mass
Loading kg/year
Contaminant (3-7% of emissions)
As
Cu
Pb
Zn
PAH"
8
42
77
108
11
- 19
- 98
-181
-252
- 19
"Based on nominal emission rate of 1 kg/day. Total Should be multiplied by total PAH
emissions in Table 7-2.
112
-------
Chapter 8. Comparison of Studies
This study examined atmospheric contaminants from a number of different perspectives. Each
sampling program or model effort focused on a somewhat different array of particle sizes and
time and space scales, in effect capturing a piece of the total picture. This chapter provides
data comparisons to bridge between and reconcile the individual study components. This
additional analysis is critical to fully test and evaluate the tools used and to address gaps in
sampling. Both qualitative and quantitative comparisons are attempted within the bounds of the
study assumptions and limitations.
PAIRED COMPARISONS
1. Aerosol/Deposition Sampling
Objective of Comparison: How predictive is one sample of the other? Can we forego one type
of sampling and still have a reasonably good idea of what will be in the other sample? What
do differences in the samples tell us about the sources and the transport/deposition processes
involved?
Limitations in Performing the Comparison:
Aerosol
* Upper size range of particles sampled was limited to 25-50 /xm (depending on wind speed);
larger particles in the ambient aerosol may have been missed.
*• Sampler performance is variable, affected by precipitation and wind speed; collection
efficiency of smaller particles (< 1.0 /tin) unknown.
> Long time period for collection of the sample (three or four days) increased possibility of
vapor breakthrough from the PUF and changes in vapor-particle partitioning. (Direct
measurement of breakthrough not made for the field samples.)
»> Limited number of samples analyzed.
Deposition
* Particle size spectrum in sample not measured.
* Long collection period (two weeks) introduced opportunities for revplatilization of samples
or transformation of chemicals (e.g. photolysis, biodegradation), particularly during summer
conditions.
* Performance of samplers for collection of large particles blown horizontally across the
sampling surface not known.
> Samplers at all sites not at same height, therefore different potential to capture large
particles.
113
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Approach: 1) Compare mean aerosol concentration and mean deposition rate for samples that
match up in time and space. Rely on metals data rather than PAH, to reduce potential sampling
and analytical error in comparison. 2) Compare aerosol and deposition data for all sample dates
analyzed for a particular station to examine patterns.
Results: Only three deposition samples had all aerosol samples during the corresponding sample
period analyzed: the two week samples beginning on September 7, November 2, and November
30 at Alexander Avenue. The data for these samples are presented in Table 8-1. Figure 8-1
is a plot of the mean aerosol concentration and mean deposition rate for all three periods for
As, Cr, Mn, Ni, Pb, and Zn. Figures 8-2 through 8-5 show all analyzed aerosol and deposition
samples for the Alexander Avenue and Morse sampling sites plotted first for metals and then
for PAHs. (Similar plots for the Sea-Land, Riverside, and Tyee Marina are included in
Appendix G-l.)
Table 8-1. Mean Deposition Rate, Mean Aerosol Concentration, and Deposition Velocity
(Vd) at the Alexander Avenue Site for Three Sampling Periods
Mean Deposition Mean Aerosol Deposition Velocity
/xg/mvday ng/m3 cm/sec
September 7-21
As 14.1 19.4 0.84
Cr 1.8 5.7 0.37
Mn 187 66 3.3
Ni 57 16.2 4.0
Pb 91 45.4 2.3
Zn 360 110 3.8
November 2-16
As 7.4 2.5 3.4
Cr 12.2 4.5 3.1
Mn 70.1 19.5 4.2
Ni 5.8 5.0 1.3
Pb 38 18.3 2.4
Zn 249 39.5 7.3
November 30 - December 14
As 5.4 5.0 1.3
Cr 7.4 6.8 1.3
Mn 61 61 1.2
Ni 44 19.8 2.6
Pb 21 41 0.59
Zn 147 161 1.1
114
-------
Chapter 8. Comparison of Studies
Figure 8-1. Comparison of Mean Aerosol Concentration (ng/m3) and Mean Deposition
Rate (/ig/mVday) at the Alexander Site for September 7-21, November 2 - 16, and
November 30 - December 14. All Data for As, Cr, Mn, Ni, Pb, and Zn are plotted
c
0
•p
•H
0
u
Q
C
t)
350
300
250
200
150
100
50
n
Sept (+)
* Nov (»)
Nov/Dec (o)
• • —
•*
'•*•
o
*
•*• °
•t- °
0 20 40 60 80 100
Mean Aerosol
120 140 160 180
115
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 8-2(a). Aerosol Particulate Metal Concentrations (ng/m3) at the Morse Supply Site
300
250
200
150
100
50
.1.2/1.1/89.
/I
/ I
/ i
8/14/89
Arsenic
Q
Chromium
Manganese
... A ...
Lead
Zinc
Dates
Figure 8-2(b). Metal Deposition Rate Qtg/mVday) at the Morse Supply Site.
700
600
500
400
300
200
100
0 L-
Arsenic
Chromium
_ — D— —
Manganese
... A ...
Nickel
Lead
Zinc
Dates
116
-------
Chapter 8. Comparison of Studies
Figure 8-3 (a). Aerosol Particulate Metal Concentrations (ng/m3) at the Alexander Avenue
Site
350
300
250
200
150
100
50
0
Fi^u
500
400
300
200
100
n
_ 1271 1/89
i1
9/11/89 .1
A * 1
I |(
I' 11/27/89 | I
l\ fi ' i
_ (.1 /.V...I
• 1 / \ II
7/24/89 J ' * . ! '
1 - L Ki
~ i fir; ,r£..Tj^..
\ A "f »A / / \ '• 1
,££S^IS& &&£$£
Arsenic
Chromium
Manganese
Nickel
Lead
Zinc
A
Dates
re 8-3 (b). Metal Deposition Rate (pg/mVday) at the Alexander Avenue Site
7/27/89
/ ' 9/07/89
/ t ,
y\ / 11/02/89
. \ / \
At/ i
' 7"":: V ^ ' V
*
A'' 'a •' \ .^*.
Arsenic
Chromium
Manganese
Nickel
Lead
Zinc
Dates
117
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 8-4(a). Aerosol Particulate PAH Concentration (ng/m3) at the Morse Supply Site
25
20
15
10
12/11/89
.../!
/ j
n
i
Naphthalene
— Q
Phenanthrene
— _D- _
Fluoranthene
... A ...
Pyrene
Chrysene
Dates
Figure 8-4(b). PAH Deposition Rate (ng/mVday) at the Morse Supply Site
6,000
5,000
4,000
3,000
1.000
7/27/89
Naphthalene
Q
Phenanthrene
Fluoranthene
Pyrene
Chrysene
Dates
118
-------
Chapter 8. Comparison of Studies
Figure 8-5(a). Aerosol Participate PAH Concentrations (ng/m3) at the Alexander Avenue
Site
40
30
20
10
12/11/89
11/27/89
ft
!'• il
•9/117&9-
•
'/V
-"
Naphthalene
^^^^^^^^^^^^m
Phenanthrene
_ _D— —
Fluoranthene
... A ...
Pyrene
Chrysene
Dates
Figure 8-5(b). PAH Deposition Rate (ng/m2/day) at the Alexander Avenue Site
35,000
30,000
20,000
15,000
10.000
5,000
.9/.07/89.
11/30/89
..A.
A- '
/
:/ *
I j, I I I I I I
Dates
Naphthalene
Phenanthrene
D
Fluoranthene
Pyrene
Chrysene
119
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Analysis: For the reasons listed above under Limitations, direct comparison of the aerosol and
deposition data is complicated by numerous potential unquantified biases in the sampling and
analyses. It is not possible with the limited data to sort out the separate effects of meteorology,
different particle sizes sampled, and different emissions mixes. The data in Table 8-1 reveal
different deposition velocities for each month and each metal, with no clear pattern between the
different metals. The September sampling period was dry, while November and December
were wet (90 and 108 mm rain collected). November 2 corresponded with the beginning of a
48-hour Type 2 air stagnation episode for Tacoma. The data in Figure 8-1 are too limited to
make definitive judgements, however they show a pattern of higher deposition versus aerosol
concentrations in November and the opposite in September. These data cannot, however, be
treated as independent, given the likelihood that certain metals co-occur on the same particle.
Deposition velocities are quite high, far exceeding deposition velocities assumed for PM10 (0.1-
0.3 cm/sec). The deposition velocities are within the range of measurements elsewhere for TSP
(total suspended particulars) and are probably associated with very large particles.
Observations of the deposition pans noted that particles in the sand/grit size range were
collected.
Figures 8-2 through 8-5 must be viewed cautiously because the aerosol and deposition samples
dp not coincide precisely in time. However, these plots show an interesting pattern of generally
higher metals deposition rates in the summer and early fall and higher aerosol metals
concentrations in the December sampling period. These differences might be attributed to the
potential for the two sampling devices to capture different size fractions of particles. PAH
aerosol concentrations and deposition rate do not exhibit this same relationship. At the Morse
site the pattern in deposition parallels the aerosol concentrations. At the Alexander Avenue site
the deposition rate in early September is very high, not following the same pattern as Morse
relative to the aerosol concentration. The September sampling period was characterized by
construction activities at a site near Alexander Avenue.
Summary: Uncertainties in the data, based on the number of potential sampling biases, and the
minimal number of samples analyzed preclude making any definitive comparisons. It would
have been desirable to make comparisons chemical by chemical (taking into account co-
occurring metals) and regime by regime (wet vs. dry, high temperature vs. low temperature,
high wind vs. low wind, woodstoves on vs. off, etc.). From the data it cannot be concluded
that one type of measurement is an adequate predictor of the other. Differences observed
seasonally in the aerosol and deposition and the lack of comparison with each other might be
attributable to variability seen in particle distribution by season, the chemistry in these
distributions, and/or the particle size sampling efficiency of the methods.
2. Aerosol/Diffusion Modeling
Objective of Comparison: Can diffusion modeling be used with confidence in place of aerosol
sampling or to extend it in space and time? Do the observations and simulations compare well
quantitatively, qualitatively? Are there particular conditions under which there is better, or
worse agreement? What do differences between the observations and simulations tell us about
the sources or size fractions?
Limitations in Performing the Comparison:
Aerosol
> Upper size range of particles sampled was limited to 25-50 pirn (depending on wind speed);
larger particles in the ambient aerosol may have been missed.
120
-------
Chapter 8. Comparison of Studies
Sampler performance is variable, affected by precipitation and wind speed; collection
efficiency of smaller particles (< 1.0 /*m) unknown.
Long time period for collection of the sample (three or four days) increased possibility of
vapor breakthrough from the PUF and changes in vapor-particle partitioning. (Direct
measurement of breakthrough not made for the field samples.)
Limited number of samples analyzed.
Range in the upper particle size captured by the aerosol sampler precludes using a single
proportionality factor to relate aerosol (PM^s-a) and model (PM10) numbers.
* Three- to four-day sampling times preclude finer temporal resolution of the data.
* Limited temporal and spatial scale of 18-day aerosol PM10 sampling (two stations separated
by 3.5 km).
Diffusion Model
> Only the PM10 size fraction modeled.
* Spatial resolution of the model limited to 0.5 x 0.5 km grids; may not be descriptive of
the smaller-scale, near field source-receptor relationships.
* Emission inventory accuracy limited by lack of information on the chemistry and mass
emissions specific for Tacoma facilities and their temporal fluctuations.
* Emissions were limited to the inventory for the modeling domain; did not account for
advection into the domain.
* Wind field data for the model limited to a single meteorological site at Alexander Avenue.
*• Modeling did not account for chemical transformations or vapor emissions.
* Deposition velocities assumed by the model were very low compared to the range of
measurements made elsewhere.
Approach: 1) Compare mass estimates of model with observations from PM^ and PM10
ambient aerosol sampling. 2) Compare model simulations with PM^ observations for metals
and PAHs. 3) Further break down the data spatially and temporally and compare it for
particular emissions and meteorological regimes.
Results: Observations (PM^) and simulations (PM10) of aerosol mass for all six sites sampled
are plotted as a scattergram in Figure 8-6. Hourly output from the model for each receptor site
was averaged over the sampling time of the corresponding aerosol sample. Table 8-2 contains
a comparison of the observations and simulations based on 199 pairs of data. From the center
"Mean" column it can be seen that the mean of all PM10 simulations underpredicts the PM^^,
measurements by 48 percent, 61.8 percent of the simulations agree with the observations within
a factor of two, and the weighted mean relative error (wMRE) is 64.8 percent. This scoring
factor (wMRE) expresses the root-mean-square fractional differences between the observations
and simulations, weighted to give more emphasis to the larger measurements. (This is done
because larger measurements are of more concern in a regulatory context.) The lower the
wMRE, the better the model performance.
121
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 8-3 contains site-by-site comparisons and summary scores that aggregate the data so that
scores for all sites are compared with the scores for just the central Tideflats industrial area
sites (Tyee Marina, Morse Industrial, Alexander Avenue, Sea-Land), omitting the two sites at
the northwestern and southeastern boundaries of the Tideflats, Brown's Point and Riverside
School. At Brown's Point and Riverside School (wMRE 106.4 and 155.8 percent) the model
underpredicts the observations by factors of three and five (11.8 /xg/m3 versus 29.0 /*g/m3 and
9.0 /ig/m3 versus 49.6 /*g/m3). For the other sites combined the wMRE is 52.9 percent. For
the four sites the average ratio of simulated PM10 to observed PM^ is 68 percent.
Figure 8-6. Scattergram of Observed
vs PM10 Simulations by WV3 Model
188
a
£
e
all sites _
• ;....-.:. '• •••
* » i * **
• . ..*'••••.:.••' •*•
• i * •***
* * *4 * **
***** * + *#
_L
B
— TSP
(observed) — >
188
122
-------
Chapter 8. Comparison of Studies
Table 8-2. Comparisons of Observed PM^,, with Simulated PM10 by Various Protocols
NUMBER OF CASES = 199
Mean of Predictions
Mean of Observations
2*[P-O]/[P+O]
S.D. of Predictions
S.D. of Observations
2*[P-O]/[P+0]
FOR PAIRED DATA:
Agree within a factor of two
Slope: d[P]/d[O]
Correlation coefficient [r] 0.78
wMRE= 2*rms/rms
Robust MRE
90%
35.5
57.7
-0.41
24.6
37.7
-0.24
67.3%
0.68
0.71
72.7%
74.1%
75%
34.3
55.7
-0.45
23.9
35.9
-0.31
65.3%
0.66
0.66
69.4%
70.8%
Mean
32.9
53.9
-0.48
22.5
33.4 i
-0.39
61.8%
0.63
0.62
64.8%
66.1%
25%
31.6
51.8
-0.52
21.0
30.9
-0.47
58.4%
0.60
0.56
60.1%
63.7%
10%
29.9
49.7
-0.55
20.0
28.5
-0.52
55.4%
0.58
56.4%
60.4%
NOTE: S.D. = Standard deviation
Table 8-3. Comparisons Between Observed PM^,, and Simulations by WV3 Model
ouc c
Name
Browns Point
Tyee Marina
Morse Supply
Alexander Avenue
Sea-Land
Riverside
All Data
T+M+A+S
i ampie
Size
24
54
22
54
22
23
199
152
nvi^s-x -"Acaii
29.0
42.3
63.5
65.0
77.1
49.6
53.9
58.5
YY V J IVltXIJl
Mg/m3
11.8
27.5
51.4
44.7
46.8
9.0
32.9
39.9
75%
110.8
53.9
45.3
55.6
66.1
158.4
69.4
55.1
W1VJ.IVJ-,,
50%
106.4
51.2
44.1
53.1
60.4
155.8
64.8
52.9
25%
101.2
48.7
40.4
49.7
54.2
. 148.1
60.1
50.7
123
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Simulations and observations are compared on a chemical basis in Tables 8-4 and 8-5. Table
8-4 is based on comparison of 104 pairs of data for the metals As, Pb, and Zn. The robust
mean relative error (rMRE) is 59.7 percent. Table 8-5 is based on 229 comparisons for eight
PAH compounds. The rMRE is 96. 1 percent and observations and simulations agree within a
factor of two for only 35.6 percent of the comparisons. For the comparison in Table 8-5, the
PAH data were renormalized to attempt to correct for biases in the data.
Appendix G-2 contains data plots along a northwest (Brown's Point) to southeast (Riverside)
axis for each sampling period. Measured PM^ and simulated PM10 are shown. This spatial
segregation shows that samples beginning on 7/24, 11/27, 11/30, and 12/11 (periods
characterized by generally southeasterly winds) show low simulation values at the southeast
boundary of the modeling domain (Riverside School), and high observed values. The sample
beginning on 12/4 shows an anomalously high aerosol observation at the Sea-Land site (156
Another way of segregating the data is to make comparisons by month and by season. (Within
the study period, "summer" is defined as all data between July 7, 1989 and October 5, 1989,
when the winds shifted abruptly from net westerly to net southerly.) Table 8-6 shows the
wMRE scores by month, by season, and by season for non-stagnation periods only (here defined
as PM10 _<.100 /xg/m3). The winter season shows the best model performance (wMRE = 61.2
percent for all points and 46.7 percent for non-stagnant conditions).
Because of the differences in the size of particles measured and simulated, another more
appropriate comparison to test the model's predictive ability is to use the PMi0 data from the
PSAPCA monitors at Fire Station No. 12 and Alexander Avenue. Table 8-7 contains these
comparisons, showing a 50th percentile wMRE score of 41.7 percent. When these data are
segregated by season and stagnation/non-stagnation the wMRE scores, are 39.8 percent in the
summer, 42.2 percent for winter for all data, and 44.5 percent for winter without the stagnation
period data (Table 8-8).
A similar comparison was made between the model simulations and the ambient PM10 measured
during the 18-day receptor modeling study. Those ambient measurements never exceeded 100
Hg/m , so there is no segregation into stagnation vs. non-stagnation cases (despite the actual
stagnation event that occurred during December). Table 8-9 shows the wMRE scores. These
scores are not as good as those achieved using the PSAPCA data.
Another possible temporal comparison of the simulations is with nephelometer bv measurements
(particles < 1.0 pm) taken by PSAPCA at Alexander Avenue. Figure 8-7 shows the PSAPCA
data composited by time of day for June and January. When compared with the modulation
function used for traffic and woodsmoke (Figure 7-1), the June curve in Figure 8-7 shows a
weak diurnal pattern consistent with the traffic pattern, while the January curve shows the
highest b at nighttime and a marked diurnal pattern with a minimum in the afternoon. The
January data are consistent with some combination of increased nighttime sources and low
nighttime air flow, and relatively better air flow in the daytime afternoons. Figure 8-8 shows
a composited plot of PM10 simulations. Comparison with the bv observations shows that the
winter nighttime maximum is missing and the traffic pulses are exaggerated for both summer
and winter.
124
-------
Chapter 8. Comparison of Studies
Table 8-4. Comparison Scores for Model Estimates of Three Metals (As, Pb, Zn) vs
Study Observations
NUMBER OF CASES = 104
Mean of Predictions
Mean of Observations
2*[P-0]/[P+0]
S.D. of Predictions
S.D. of Observations
, 2*[P-0]/[P+0]
FOR PAIRED DATA:
Agree within a factor of two
Slope: d[P]/d[0]
Correlation coefficient [r]
Robust MRE
Actual
45.1
47.1
-0.04
36.2
60.5
-0.50
59.6%
0.75
0.76
59.6%
90%
49.5
54.6
0.07
38.1
68.6
-0.35
66.3%
0.87
0.81
64.4%
Median
45.3
46.5
-0.02
36.1
58.3
-0.48
60.4%
0.78
0.76
59.7%
10%
39.9
38.8
-0.13
33.0
50.9
-0.59
53.5%
0.70
0.71
54.6%
S.D. = Standard deviation
Table 8-5. Comparison Scores for Model Estimates of Eight PAHs vs Study Observations-
Renormalized
NUMBER OF CASES - 229
Mean of Predictions
Mean of Observations
2*[P-0]/[P+0]
S.D. of Predictions
S.D. of Observations
2*[P-0]/[P+0]
FOR PAIRED DATA:
Agree within a factor of two
Slope: d[P]/d[0]
Correlation coefficient [r]
Robust MRE
Actual
3.5
3.5
-0.00
6.2
7.0
-0.11
35.4%
0.92
0.61
96.0%
90%
4.2
4.6
0.23
7.8
9.1
0.30
42.6%
1.28
0.78
102.0%
Median
3.4
3.4
-0.01
5.8
6.0
-0.08
35.6%
0.94
0.61
96.1%
10%
2.7
2.5
-0.18
3.7
3.9
-0.48
29.7%
0.68
0.40
88.9%
NOTE: S.D. = Standard deviation
125
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 8-6. Summary of Scores Comparing PM^ Observations with WV3 Simulations,
Stratified by Season _ '• _ ___
wMRE% 75% 50% 25%
Jul
Aug
Sep
Oct
Nov
Dec
Summer
Winter
Summer
Winter
84.4
69.8
85.9
56.5
82.2
61.9
77.8
65.6
73.0
49.3
75.1
65.6
81.3
53.7
71.7
56.8
74.6
61.2
69.9
46.7
67.6
78.0
78.0
50.8
61.0
51.8
71.7
57.0
67.5
44.2
*- all points
*- all points
- TSP < 100 ng/m3
*- TSP j< 100 jig/m3
Table 8-7. Comparisons Between Observed PM10 and Simulations by WV3 Model
one o
Name
Alexander Avenue
Fire Station No. 12
mmpit jrr
Size
73
174
rjL10 ivicaii TV
Mg/m3
39.1
38.8
V J 1VAC4U1
fjLg/m3
44.7
41.6
75%
38.8
45.8
WIVilVd
50%
37.0
43.6
25%
35.2
41.6
All Data 247 38.9 42.5 43.3 41.7 39.9
Table 8-8. Scores Comparing PSAPCAs PM10 Observations with Simulations by WV3,
Stratified by Season
wMRE% 75% 50% 25%
Summer
Winter
Summer
Winter
41.9
49.3
41.9
46.3
39.8
42.2
39.8
44.5
37.5
40.2
37.5
42.0
*- all points
*- all points
*- PM10 < 100 /Kg/m3
*- PM10 <_100 /ig/m3
Table 8-9. Scores Comparing 18-Day Receptor Modeling PM,0 ("Coarse+Fine")
Observations with Simulations by WV3, All Points (Morse Supply and Alexander Avenue)
wMRE% 75% 50% 25%
PM10 72.6 70.0 67.1
126
-------
Chapter 8. Comparison of Studies
Figure 8-7. Composited B,n as a Function of Time of Day
1.2
I
a*
8.8
e
— tine of day [PSI1 —>
24
Figure 8-8. Composited PM10 Simulations as a Function of the Time of Day
180
58
§?
CD
e
winter
- tine of day [PSI1 —>
127
24
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Analysis: The model underpredicted the PM^ observations, as might be expected, since only
PM10 was modeled. However, when the sites at the northwestern and southeastern model
boundaries were removed from the analysis, the model's predictive capability improved. One
possible explanation is that the model neglects a background source which is relatively more
important at the two sites removed from the industrial emissions core of the model domain.
Based on the receptor modeling analyses in Chapter 6, one possible explanation is woodsmoke
imported into the modeling domain from outside.
The model performed considerably better with the PSAPCA PM10 data than with the PM^ or
18-day receptor modeling aerosol data. The model overpredicted the PSAPCA data by only 9
percent, but overpredicted values from the 18-day study by 70 percent. One reason may be the
difference in the sampling periods. The 18-day study sampled in 12-hour periods from 7 a.m.
to 7 p.m. and from 7 p.m. to 7 a.m. PSAPCA samples are collected at midnight and noon.
The chemical-specific simulations are predictably less accurate than those for total mass due to
a compounding of the inaccuracies in the emissions estimates used as input to the model and
the potential sampling and analysis biases that could affect the aerosol data.
The patterns of discrepancies pointed out from the spatial plots of mass in Appendix G-2 are
consistent with the theory of woodsmoke being advected into the modeling domain from
upvalley. A likely source is the town of Puyallup, 11 km southeast of Fire Station No. 12.
The b.p comparison with the model for June and July also supports this theory. The winter b
maximum could be attributed to woodsmoke. A separate analysis done by Harrison (in press)
calculates that the town of Puyallup, with a population of 24,000 people, could be responsible
for woodsmoke emissions that are advected under certain conditions down valley to contribute
up to 50 percent of the winter PM10 observed at Fire Station No. 12.
The high December 4 aerosol sample recorded on the contour map at the Sea-J-^nd site could
be attributed to a local source, such as the ore offloading facility nearby.
Comparison of the model simulations with the b9 observations strongly suggests that the model
overestimates traffic emissions, perhaps by a factor of two or more and underestimates
woodsmoke, perhaps by a factor of three.
Appendix G-3 references several reports that served as a basis for much of the modeling and
analysis discussion above.
Summary: The performance scores for the model using mass data were good, particularly when
PSAPCA PM10 data are the basis of comparison. The temporal comparison and analysis pointed
put some emissions (woodsmoke, traffic) that clearly need to be more accurately estimated to
improve the model's predictive capability. Many of the observations and their "discrepancies"
point towards an explanation that relies on a large woodsmoke emission component being
advected into the model domain during winter light wind conditions. This could be corrected
by using a larger modeling domain.
3. Aerosol/Receptor Modeling
Objective of Comparison: How well does the PM10 aerosol collected during the 18-day field
study for receptor modeling agree with the PM^ aerosol collected at three- to four-day
intervals in the six-month study? Are there certain conditions under which PM10 is a better
estimator of PM^.^? What do the differences between the two types of samples indicate about
the likely sources?
128
-------
Chapter 8. Comparison of Studies
Limitations in Performing the Comparison:
Aerosol
>• Upper size range of particles sampled was limited to 25-50 /xm (depending on wind speed);
larger particles in the ambient aerosol may have been missed.
> Sampler performance is variable, affected by precipitation and wind speed; collection
efficiency of smaller particles (< 1.0 /*m) unknown.
* Long time period for collection of the sample (three or four days) increased possibility of
vapor breakthrough from the PUF and changes in vapor-particle partitioning. (Direct
measurement of breakthrough not made for the field samples.)
* Limited number of samples analyzed.
Receptor Model
*• Focus of all sampling was particles in < 10 jtm size fraction.
*• Limited number (2) and short distance between the sites (3.5 km) used for PM10 aerosol
sampling.
* Short temporal coverage of 18-day study; no seasonal comparison.
*• Inadequate amount of sample for PAH analysis.
Approach: 1) Compare six-month and 18-day aerosol samples at the two sites in common
(Alexander Avenue and Morse Industrial Supply) for fine mass (<2.5 /tin), PM10 (fine plus
coarse mass), and PM^; 2) Compare PSAPCA TSP and PM10 data for entire study period
to look at seasonal patterns.
Results: Figures 8-9 (a) and (b) show plots of total mass for PM25, PM,0 and PM^jo aerosol
samples for Alexander Avenue and Morse Industrial Supply. The figures reveal similar patterns
at both sites. Fine mass accounts for most of the PM10 fraction. PM10 tracks the changes in
PM25-» quite wgU> with the possible exception of the December 12 sampling date at the Morse
site. The PM10 mass is a larger fraction of the PM^ mass when PM^ is low (December
15).
Figure 8-10 shows a plot of the PSAPCA TSP versus PM10 data collected at Fire Station No.
12 during the entire study period. The two curves track fairly well. Qualitatively, the
comparison appears to show a pattern of larger differences between the TSP and PM10 numbers
during the summer period (July through September). During the fall/winter period there
appears to be a pattern similar to that observed in Figures 8-9 (a) and (b), with PM10 mass
constituting a larger fraction of the TSP mass during periods low when TSP levels are low.
Analysis: The relative proportions between the coarse and fine mass are as expected from other
studies. Most combustion product emissions are in the <2.5 pm size range. The December
1 1 sampling date was the start of a significant air stagnation episode. The relative increase in
PM^jo versus PMW during this period is not consistent with the receptor modeling analysis that
attributes much of the emissions during this period to woodsmoke. The larger difference
between TSP and PM10 in the summer is consistent with the theory that resuspended dust is a
more significant factor in the drier months.
129
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 8-9(a). Comparison of Aerosol Collected by Six-Month Study (PM^J and the 18-
Day Study (PM^, PMto) at the Alexander Avenue Site
150
160
H
C
«
h
n 50
o
b
0
«-.» PH25-58
o..o PH10
+-•-* PM2.5
/\
\>-°"'V\
V.
6 8 10 12 14 16 18 20
December
Figure 8-9(b). Comparison of Aerosol Collected by Six-Month Study (PM^) and 18-Day
Study (PMu, PM,0) at the Morse Supply Site
o
•f
o
6 150
A
3
0
i; 100
e
«
u
n 50
o
u
0
«-.• PH25-50
o..o PN10
+—+ PM2.5
°
8 IB 12
Decenber
14 16 18 28
130
-------
Chapter 8. Comparison of Studies
Figure 8-10. TSP and PM,n Measured during the Study Period at Fire Station 12
250
280
I)
c
3
0
\
M
| 100
a
o
jj 50
O N
Month
D
Summary: The comparison between PM^ and PM10 shows that the two size fractions reflect
similar patterns, despite the high day-to-day variability in the ambient aerosol mass. The
seasonal comparison using the PSAPCA data indicates that the ratio between the two size
fractions may be different for summer and winter/fall conditions, so that a single constant
proportionality factor shouldn't be assumed to attempt to convert one number to another. The
seasonal difference is also consistent with the probability of a larger resuspended dust
component during the dry months. The study data are for two sites relatively close together
(3.5 km) and the PSAPCA data are for only one site, therefore the observations and conclusions
should not be assumed to necessarily hold true for all sites.
4. Deposition/Diffusion Modeling
Objective of Comparison: How well does the model perform qualitatively and quantitatively
in predicting the deposition at a receptor site? How do deposition velocities computed for the
deposition samples compare with the deposition velocities assumed in the model; what does this
comparison tell us?
Limitations in Performing the Comparison:
Deposition
* Particle size spectrum in sample not measured.
131
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
* Long collection period (two weeks) introduced opportunities for revolatilization of samples
or transformation of chemicals (e.g. photolysis, biodegradation), particularly during
summer conditions.
* Performance of samplers for collection of large particles blown horizontally across the
sampling surface not known.
* Samplers at all sites not at same height, therefore different potential to capture large
particles.
Diffusion Model
* Only PMu modeled, not total potential deposition.
* Deposition velocities used from literature may or may not be appropriate for ambient
aerosol.
> Deposition calculations dependent on accurate aerosol modeling, which is subject to
accuracy in emissions inventory, limited by scale of model, and affected by unknown
temporal variation in emissions.
Approach: 1) Compare deposition velocities calculated from deposition samples with velocities
assumed in the diffusion model. [Note: It would have been desirable to compare the deposition
of particular metals on a station by station basis for discrete meteorological or emissions
regimes. The model output for deposition was averaged over the 186-day study period, not
allowing for these comparisons.]
Results: Table 8-10 presents net deposition velocities calculated from the deposition samples
collected during the six-month study. The values in the table are the average deposition velocity
(wet + dry) computed as:
Vd = (Flux of tracers to the surface)/(Concentrations of the tracers in the air)
The standard deviations are for determinations of the means. The ranges of the distributions
are about six-fold greater than the standard deviations.
With the exception of the lighter PAHs in the middle block of the table (which are more likely
to be affected by sampling and analysis errors), the remaining net (wet + dry) deposition
velocities exceed 2.2 cm/sec, and all but one exceed 3.4 cm/sec.
\
By comparison the diffusion model used an assumed value for dry deposition velocity of 0.1
cm/sec.
Analysis: The numbers in Table 8-10 exceed by an order of magnitude reported deposition
velocities for gaseous tracers or particles with diameters < 100 /tm (McMahon and Denison,
1979). They also exceed estimates of deposition velocity calculated using the Monin-Obhukov
similarity theory (Vd = 0.4 cm/sec). (These estimates are for small particles whose deposition
is a function of turbulent mixing.) One possible explanation for the high deposition velocities
is that large particles were advected into the deposition sampling pans. This is supported by
the observations of sand and grit in the pans. The suspected source of these large particles is
resuspended soil dust, which would not behave as the modeled eddy-driven deposition of
particles <10 ^m. Biases in the ambient aerosol sampling may also have resulted in
underestimates of the ambient aerosol. This would affect the calculation of Vd to yield a higher
132
-------
Chapter 8. Comparison of Studies
deposition velocity. As noted under comparison #1 above, there is a suspected lack of
correspondence between what was measured as ambient PM^jo aerosol and what may have been
deposited, particularly in the larger particle size range.
Resuspended dust may be transported over short distances (tens to hundreds of meters) most of
the time. Strong gradients in the vertical distributions of resuspended dust are likely; therefore
differences in the height of the deposition pans above the surface would be expected to be a
factor in comparing between sites. A quantitative analysis of the objective effect of this is not
possible.
Examination of the deposition data (Table 4-10) shows that the industrial sites (SL, AS, MS)
have higher deposition concentrations than the sites more removed from the center of the
Tideflats industrial area (TM, RS). These differences in concentration are not nearly as
pronounced for the aerosol data (Table 4-6). The deposition pattern cannot be completely
accounted for by the aerosol samples. The gradients in the deposition samples imply that local
(nearby) sources are affecting deposition, which is consistent with resuspension of dust as one
of the sources.
Table 8-10. Deposition Velocities Inferred from PSWQA Measurements at All Sites
cm/s
Arsenic 3.69 ±1.02
Chromium 2.22 ± 0.34
Manganese 3.38 ± 0.47
Nickel 4.27 ± 0.70
Lead 3.83 ± 0.86
Zinc 3.83 ± 0.73
Naphthalene 0.69 ±0.13
Acenaphthene 0.06 ± 0.02
Acenaphthylene 0.08 ± 0.04
Fluorene 0.04 ± 0.02
Phenanthrene 0.10 ± 0.05
Anthracene 0.11 ± 0.05
Fluoranthene 0.99 ± 0.53
Pyrene 0.95 ± 0.50
Benzo(a)anthracene 3.01 ± 1.58
Chrysene 4.58 ± 1.58
Benzo(a)fluoranthene 7.27 ± 3.82
Benzo(k)fluoranthene 3.77 ± 1.98
Benzo(a)pyrene 3.56 ±1.82
Indeno(l,2,3-c,d)pyrene 4.06 ± 1.72
Dibenzo(a,h)anthracene 8.11 ± 3.91
Benzo(g,h,i)perylene 3.45 ± 1.38
Summary: The comparisons between the computed deposition velocities and those used in the
model point strongly towards collection of larger size particles as part of the explanation for the
much higher deposition velocities for the field data. The data would not be expected to
compare precisely because the modeling only addressed particles < 10 /*m. Model deposition
133
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
simulations were not available for station-by-station, regime-by-regime comparisons to attempt
to further refine the analysis.
Attempting to model resuspension would be difficult due to uncertainties of lift coefficients as
functions of particle size and land-surface type, and size-dependent transport distances. Most
transport is concentrated into a few dry-windy episodes; thus, errors from statistical sampling
would require simulations of many years.
5. Deposition/Receptor Model
Objectives of the Comparison: How well do measurements of ambient PM10 predict the
qualitative and quantitative patterns in the deposition? Are the patterns in the deposition
consistent with the receptor modeling analysis of emissions sources?
Limitations in Performing the Comparison:
Deposition
* Deposition samples collected over two week period, therefore limited number of samples
(1) to compare with the 18-day PM10 aerosol.
* Lack of temporal match in initiation of the two-week sample with start of 18-day study.
Receptor Model
* Limited spatial and temporal scale for comparison.
Approach: As noted under Limitations there is insufficient data for a reasonable comparison
of the receptor modeling aerosol and the deposition data. Analysis of comparisons #1 and 4
above suggests that the deposition samples may contain large particles not collected by the
aerosol samplers. If so, the PM10 aerosol samples would be expected to have more
discrepancies with the deposition 'data than was noted for the PM^ aerosol.
6. Receptor Model/Diffusion Model
Objective of the Comparison: Do the two modeling approaches confirm each other as to the
sources of the ambient aerosol?
Limitations in Performing the Comparison:
Receptor Model
* Limited number (2) and short distance between the sites (3.5 km) used for PM10 aerosol
sampling.
> Short temporal coverage of 18-day study; no seasonal comparison.
> Inadequate amount of sample for PAH analysis.
Diffusion Model
*• Spatial resolution of the model limited to 0.5 x 0.5 km grids; may not be descriptive of
the smaller-scale, near field source-receptor relationships.
134
-------
Chapter 8. Comparison of Studies
Emission inventory accuracy limited by lack of information on the chemistry and mass
emissions specific for Tacoma facilities and their temporal fluctuations.
Emissions were limited to the inventory for the modeling domain; did not account for
advection into the domain.
Wind field data for the model limited to a single meteorological site at Alexander Avenue.
Modeling did not account for chemical transformations or vapor emissions.
Approach: 1) Compare receptor model (PMjs) and diffusion model source apportionments.
2) Scale up receptor model to PM10 and compare with diffusion model source apportionment.
3) Compare model results with other approaches for identifying sources.
Results: Table 8-11 summarizes the information from the receptor model pie charts (using Pb
and o-xylene as tracers of vehicle exhaust) found in Chapter 6, Figures 6-9 through 6-12.
These source apportionments are for the fine particle fraction (PMjj), based on all samples
collected during the 18-day study. At both sites residential woodsmoke accounted for the
largest component of fine-particle mass, followed by vehicle exhaust.
Table 8-12 contains fractional source apportionments for the diffusion model, grouped into
categories that were used to modulate emissions in the model over a 24-hour period. Vehicle
exhaust and industrial sources operating around the clock are the primary contributors.
Table 8-11. Receptor Modeling Results (Normalized Percent Contribution) for Fine
Particulates (PM2 5) Using Lead (Pb) and a VOC (o-xylene) as Tracers of Vehicle Exhaust
_ Alexander Avenue _ Morse _
_ Pb _ o-xylene _ Pb _ o-xylene
Vehicle Exhaust 7 ± 4 18 ± 5 10.0 ±5 20 ± 6
Woodstoves 62 ± 7 55 ± 6 72.0 ±7 64 + 6
Hogged Fuel 2.0 ± 0.3 1.6 ± 0.3 3.0 ± 0.4 2.4 ± 0.3
Al Production 12.0 ± 3 10.7 ± 2.5
Scrap Metal 7.0 ±1 6.3 ± 0.8 5.0 ±1 5.0 ± 0.9
Residual Oil 10.0 ±3 8.9 ± 2.5 10.0 ±3 9.0 ± 2.3
Table 8-12. Fractional Source Attributions from WV3 December 1 Through 31, 1989
Morse Supply _ Alexander Avenue
3-shift factories
1-shift factories
woodstoves
transportation
0.037 ± 0.053
0.170 ± 0.057
0.079 ± 0.066
0.446 + 0.051
0.360 ± 0.051
0.157 ± 0.043
0.100 ± 0.067
0.379 + 0.057
1.002 0.996
135
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Table 8-13 contains the results of the receptor model scaled up to PM10. The Fine CMB
(Chemical Mass Balance) was calculated using fine particle data only and fine particle source
profiles. Values in the table under this column were calculated as follows:
calculated fine mass concentration from source x 100
total measured fine + coarse mass concentration
Fine particles were overpredicted in the Fine CMBs by 16-24%. The columns under "PM10
CMB" were calculated using PMJO data and source profiles. Woodstoves could not be included
directly in these calculations because organic and elemental carbon were not measured for the
coarse particle size fraction. The unexplained mass ("other") could be considered an upper
limit for woodstoves.
To examine the influence of meteorology, source apportionments were performed separately for
high-wind and low-wind days, again using the fine particle data scaled up to PM10. From the
18-day study the days identified as very windy were December 8, January 3, 4, 5, 6, and 7.
Of these six days the only days when there were both a.m. and p.m. samples for both receptor
sites were January 4 and 5. Winds were mostly from the south (SW - SE) on those two days.
The days identified as having very little wind were December 5, 12, and 14. Of these three
days, the only days when there were both a.m. and p.m. samples for both sites were December
12 and 15. Winds were mostly from the NE - E - SE sectors during those days. Results of
this CMB modeling are presented in Figures 8-11 through 8-14. At the Morse site on windy
days there was a much larger "other" contribution than on non-windy days. This is a result of
the higher concentrations of coarse particles at that site on windy days. At the Alexander site
the fine-particle mass was overpredicted for most of the windy-day samples. The result of
overpredicting the fine particles is to reduce the "other" category.
Table 8-13. Receptor Modeling Results (% Measured PM10) Using Two Alternative
Methods of Calculation
Alexander Avenue Morse Supply
FineCMB(a) PM,n CMB00 Fine CMB PM,n CMB
Crustal Dust
Autos
Woodstove
Hogged Fuel
Al Potlines
Al Fugitives
Metals
Residual Oil Boilers
Other
...
6.0 ± 3.0
55.2 ± 6.1
1.6 ± 0.3
10.8 ± 2.5
—
6.3 ± 0.8
9.0 ± 2.5
11.1 ± 7.7
14.1 ± 2.8(c)
5.9 ± 3.4
—
6.0 ± 1.0
9.2 ± 5.0(c)
—
9.0 ± 1.5
23.7 ± 6.9(c)
32.1 ± 9.8
...
9.0 ± 4.2
63.4 ± 6.3
2.3 ± 0.3
—
—
4.9 ± 0.9
8.9 ± 2.3
11.5 ± 8.0
9.4 ± 1.9
9.4 ± 4.0
—
6.5 ± 1.1
—
12.1 ± 4.9
6.7 ± 1.1
21.3 ± 5.8
34.6 ± 8.9
(a) Fine CMB indicates that CMB was performed on fine particle data using fine-particle
profiles. Both fine CMBs overpredicted total fine mass (by 16% for Alexander Avenue
and 24% for Morse Supply). Calculated source contributions were divided by measured
fine plus coarse (PM10).
(b) PMW CMB indicates that CMB was performed on PM10 data (fine + coarse) using PM10
profiles, where available. If PM10 profiles were not available (as for Simpson hogged
136
-------
Chapter 8. Comparison of Studies
fuel), fine profiles were used. (Note: for other hogged fuel profiles, fine + PM10 were
nearly the same because most of the emissions are fine.) Woodstoves could not be
included in the PM10 CMB. They may be at least partly attributed to "other". Calculated
source contributions were divided by measured PM10.
(c) Crustal dust, aluminum potlines, and residual oil boiler (and possibly metal shredding) are
listed as a similarity cluster in the CMB.
Two other methods were attempted to gain information about the source apportionment and the
behavior of the emissions: covariance analysis and construction of polar fluxgrams or PM10
tracer roses. A covariance analysis is a mathematical technique used to isolate sets of chemical
species that behave similarly in time and space. Vectors or sets of chemical species are
identified that minimize the residual variance. These sets of chemicals are then used to identify
sources or source categories that significantly impact receptor sites. The covariance analysis
is described further in Appendix G-4. Two significant vectors were identified from the metals
data:
Vector 1: Pb + Zn + Mn + Ni + Br + (S) - Ca
Vector 2: Ti + Fe + Mn + Cr + K - Cl - (S)
The relative contributions of vector 1, associated with automotive transport, were higher at
Brown's Point and Riverside School and lowest at Morse Supply and Alexander Avenue. The
relative contributions of vector 2, associated with soil minerals, were lowest at Brown's Point
and Riverside School and highest at Morse and Alexander Avenue. From the PAH data two
vectors were identified, one representing all 17 PAH compounds, and associated with the Kaiser
emissions, and the other representing naphthalene, benzo(k)fluoranthene and
benzo(g,h,i)perylene, with an unidentified source.
PM,o tracer roses were constructed from the PSAPCA data from Alexander Avenue and Fire
Station No. 12 during the period of the study (Figures 8-15 and 8-16). Vector amplitudes are
proportional to the tracer concentrations measured at the receptor, and the vector direction
points upwind. Figure 8-15 displays the results of 175 daily-averaged PM,0 measurements from
Fire Station No. 12, Figure 8-16 the results of 73 daily-averaged measurements between July
7, 1989 and January 7, 1990 for Alexander Avenue. Each small circle represents one of the
daily measurements. The lobes "point fingers" at conspicuous upwind sources. An interior thin
circle in each curve depicts a normalized isotropic distribution. Inner and outer thin irregular
curves bracket this isotropic circle with expected sampling errors of ± one standard deviation.
Both figures show a common southeasterly lobe which runs parallel to the Puyallup valley axis.
At Alexander Avenue there is a similar lobe to the northwest. Further discussion can be found
in Appendix G-5.
Analysis: The source apportionments for residential woodsmoke and vehicle exhaust differ quite
dramatically among the several methods employed for calculations. Very high values are
computed for woodsmoke using Pb as a tracer of vehicle exhaust. These numbers remain high
in the "Fine CMB" because the fine mass dominated the measured PM10. The PM10 CMB
cannot be reconciled with the apportionment from the diffusion model. However, the
comparison between the diffusion model simulations composited for January and the bv
observations (see comparison #2) points strongly towards the diffusion model using too large
a vehicle exhaust emission input and too small a woodsmoke input. The high-wind versus low-
wind comparisons of the receptor model were based on very limited data, but showed a higher
contribution of woodsmoke on low-wind days. The difference in the amount of coarse particles
measured at the Morse site is likely to be a reflection of the sampler being located closer to the
ground than the Alexander Avenue sampler.
137
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 8-11. Receptor Model Results for Low Wind Days (December 12 and 14) at the
Morse Supply Site
Hogged Fuel (1.4%)-( ,-AII Autos (6.9%)
-Scrap Metals (4.7%)
Residual Oil (10.8%)
Woodstoves (65.7%)
Other (10.4%)
Figure 8-12. Receptor Model Results for High Wind Days (January 1 and 5) at the Morse
Supply Site
Hogged Fuel (2.0%)
Woodstoves (43.7%)
All Autos (6.3%)
Scrap Metals (5.9%)
Residual Oil (5.2%)
Other (36.9%)
-------
Chapter 8. Comparison of Studies
Figure 8-13. Receptor Model Results for Low Wind Days (December 12 and 14) at the
Alexander Avenue Site
Other (16.0%)
All Autos (5.8%)
Residual Oil (10.9%)
Scrap Metals (5.9%)
Al Production (4.5%)
Hogged Fuel (1.1%)
Woodstoves (55.8%)
Figure 8-14. Receptor Model Results for High Wind Days (January 4 and 5) at the
Alexander Avenue Site
Other (2-6%Vl_rAII Autos (9.9%)
Residual Oil (19.7%) -**«*«——^
Scrap Metals (3.4%)
Al Production (9.5%)
Hogged Fuel (0.1%)
Woodstoves (54.8%)
139
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Figure 8-15. Tracer-Rose Using Fire Station No. 12 PM1B Data
U(a) = 27.86 V.
N - 175 days
Bias =.0001"
west
Fire Station 812
July 7, 1989 - Jan 7, 1990
Natural Heights
10-03-1998
13:30:55
east
sou
140
-------
Chapter 8. Comparison of Studies
Figure 8-16. Tracer-Rose Using Alexander Avenue Site PM10 Data
U = 15.84
N = 73
18-03-1990
10:11:51
west
Alexander Street
July 7, '8? - Jan 7, '9
Natural Heights
/east
sou
141
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Results of the covariance analysis support the theory that resuspended dust plays an important
role in the input of metals measured at the sampling sites in the industrial area.
The southeasterly lobe in the PM10 tracer roses at both sites may be attributable to emissions
from a source at Kaiser, which is two miles east-southeast of Fire Station No. 12 and one mile
southeast of Alexander Avenue. This lobe and the one pointing northwest at Alexander Avenue
are also aligned with the valley axis. This would be consistent with the theory that woodsmoke
is advected down the valley, commingling with industrial emissions from the northwest at the
Alexander Avenue site.
Summary: While the current diffusion model run and the receptor modeling don't agree on the
relative importance of woodsmoke and vehicle exhaust (during the 18-day receptor model
sampling period) subsequent analyses argue that the woodsmoke component is larger than that
anticipated in the diffusion model.
The tracer rose developed using the PSAPCA data shows the potential for a strong woodsmoke
influence on the Tideflat monitoring stations from a source in a southeasterly direction.
SUMMARY
The following conclusions are drawn from the study comparisons. Individual paired
comparisons were limited by the data available and the inherent sampling biases and
assumptions used in sampling and modeling. Collectively, the comparisons reinforce the
individual study results and help develop a more coherent picture of the processes and
parameters affecting the atmospheric contaminants in the Tideflats.
> Aerosol and deposition samples, as collected in this study, are not sufficiently similar that
one can serve as a predictor of the other. The differences are confounded by the different
particle capturing abilities of the aerosol and deposition samplers, which are a function of
sampler design and location. The differences suggest that large particles (>50 /xm) not
collected by the aerosol samplers, are a significant component of the deposition samples
at certain times (e.g. summer dry weather and moderate to high wind speeds).
* Diffusion modeling, focused on the PM10 size fraction, performs best when compared with
measurements of the PM10 aerosol mass. The model's ability decreases as a larger size
spectrum of particles or specific chemical constituents are examined. The differences
suggest that the model's predictive ability would be improved by enlarging the modeling
domain and thereby increasing the emissions inputs for woodsmoke. Additional refinement
in all the emissions inputs to the model (e.g. their temporal variability, chemistry, size
fraction of particles) is needed to improve the performance in simulating the ambient
aerosol.
* From samples at a limited number of stations, it appears that the patterns in PM10 aerosol
and PM^jo track each other quite well, however the proportion of the total mass that is
PM10 changes seasonally, being higher in the winter. This is consistent with resuspended
dust (generally larger particles) comprising a larger component in the summer, and
woodsmoke (primarily <2.5 /tm) comprising a larger component of the aerosol in the
winter.
» Differences between the model assumptions of deposition velocity and the deposition
velocities computed from the deposition samples suggest that the samples collected particles
considerably larger than PM10. The model's predictive capability would be improved by
assuming a range of deposition velocities, including values appropriate for TSP.
142
-------
Chapter 8. Comparison of Studies
Comparison of the diffusion modeling-and receptor modeling source attributions suggests
that during winter stagnant air conditions, woodsmoke is an important component of the
ambient fine (<2.5 jtm) aerosol. Tracer rose analysis also supports the concept that this
woodsmoke is imported into the modeling domain from the southeast.
143
-------
Chapter 9. Synthesis of Results
Results of individual studies have been summarized at the end of the chapters. This chapter
assembles all of the information into a synthesis of what has been learned. First, study
conditions are compared with non-study periods to set a larger context for interpreting and
applying study results. Results are presented and discussed pertinent to the components, of the
ambient aerosol, the sources of the aerosol, deposition, and the transport of atmospheric
contaminants to Commencement Bay. The chapter concludes with an assessment of the
significance of these inputs to Commencement Bay relative to other sources of toxic
contaminants to the water.
CONTEXT
A reasonable question in assessing the study results is to inquire whether the conditions during
the study represented a worst case scenario. The spatial, temporal, meteorological, and
emissions characteristics of the study are examined below to attempt to make this determination.
Historical data and projections of future conditions are used for comparison where available.
Spatial
Sampling stations for ambient aerosol and deposition were centered within the Tacoma Tideflats
industrial area, with more distant stations both "upwind" at Riverside School, and "downwind"
at Brown's Point. The upwind and downwind orientations are geared to the dominant local
wind patterns. Sampling did not include equivalent downwind stations for other wind
directions, nor was it able to include stations further out over Commencement Bay. Brown's
Point, while generally in a downwind direction, is physically "around the corner" and therefore
removed from the direct downwind path. Tall stacks from some of the major industrial
emissions sources, such as Simpson Tacoma Kraft, may also move the emissions beyond the
relatively close-in sampling network of the study.
The diffusion modeling domain of 91.5 km2 included the Tideflats and the communities of
Milton, Fife, and SE Tacoma. As discussed in Chapter 8, this is probably not the only source
of emissions affecting the ambient air in the Tideflats, and it only includes a small portion of
the total watershed that drains into the Puyallup River and Commencement Bay. For example,
the modeling domain was bounded by the bluffs on either side of Commencement Bay and
therefore did not include all of Tacoma. As a consequence, the projected loading from
atmospheric deposition on land and runoff to Commencement Bay is probably underestimated.
Because of paved surfaces and steep slopes in downtown Tacoma, mobilization (the percentage
of contaminants available for runoff into Commencement Bay) will be relatively high from this
source compared with rural, unpaved regions of the watershed, where sediment and vegetation
filter out most contaminants prior to washoff. The sampling and modeling domain probably
captured a worst case for examining the effects of woodstove emissions on ambient air.
Temporal
The overall study period extended from July 1989 to January 1990. Ambient aerosol (PM^)
and deposition samples were collected during the entire six-month study; stations in use varied
according to the "intensive" and non-intensive sampling regimes described in Chapters 2 and 4.
145
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
A separate short-term aerosol study to support the receptor modeling was conducted for eighteen
days in December 1989 and January 1990. The measurements therefore are a series of
"snapshots" over the equivalent of half of a year and encompass a range of meteorological and
emission conditions.
Meteorology
State climatological summaries issued by the National Weather Service were used to plot
monthly means and minimum and maximum readings of temperature and precipitation for the
period from 1980 through 1988. Figures 9-1 and 9-2 illustrate the conditions during the study
period relative to the nine-year historical data. For the most part the study period is typical of
the normal pattern. Figure 9-2 shows, however, that the study period included two apparent
variances in precipitation from the historical conditions: extremely low precipitation in
September 1989 (0.18"), and an extraordinarily wet January 1990 (10.90"). However, the
highest rainfall in January occurred on January 9 (3.66"), just after completion of the sampling
studies.
Winds during the study period are summarized in Chapter 7 with a wind rose and a plot of the
cumulative trajectory of winds during the study period. The latter figure shows the expected
seasonal shift from net westerly summer winds to south or southeasterly winds during late fall
and winter. Air stagnation, which occurs during periods of light winds and increased
stratification, is more likely in the fall and winter. Air stagnation in the Tacoma region is often
accompanied by a light drainage wind that flows from the Puyallup area, down the river valley,
and over Commencement Bay. During the summer and fall there is a daily reversal in the
winds due to land-water temperature differences; this does not occur or rarely occurs during the
late fall when the land is cold. January and December are the two months of the year that
usually have the strongest temperature inversions and therefore the greatest potential for air
stagnation and elevated air contaminant levels.
During the study period Ecology issued two air stagnation forecasts covering the Tacoma
region: one on November 30, which lasted not quite 24 hours; and one on December 11, which
lasted until December 17. This compares with two forecasts issued in 1987 and one in 1988.
As described in Chapter 3, an analysis of meteorological conditions after-the-fact confirms the
air stagnation that was forecast in mid-December and notes a lesser episode in late December.
These episodes were of shorter duration (and severity) than some earlier episodes of prolonged
air inversion (e.g. approximately two weeks in late December 1985).
Besides the light wind/stagnant air conditions found at times in the winter, the study period also
included winter periods with strong winds, and summer and fall periods with prolonged dry
weather and moderate-to-high wind speeds. While this did not include all potential
meteorological conditions during a year, it did reflect conditions under which individual
emissions sources varied in their relative influence.
Emissions
An accurate accounting of all emissions sources, including their chemistry, particle size
fractions, and temporal variations is extremely difficult to construct. PSAPCA is continually
receiving updated information on specific Tacoma sources as well as revisions of the emission
factors used to estimate the emissions of particular source categories. The air emissions
inventory for the Tacoma Tideflats is not continuously updated to incorporate these changes.
To attempt to set emissions levels during the study into a longer-term context, one can look at
the PM10 measured by PSAPCA at their Alexander Avenue site during the study period and
compare this with the preceding three years of data from the same site during the same portion
146
-------
Chapter 9. Synthesis of Results
Figure 9-1. Comparison of Monthly Mean Temperatures for the Study Period with the
Monthly Mean, Average Maximum, and Average Minimum for 1980-1988
H
U
u
h
0)
u
a
80
70
60
50
40
30
Monthly Temperature at Tacoma
. Study year only
Excluding study year
+ Monthly nin and nax
O N
Month
D
Figure 9-2. Comparison of Monthly Mean Precipitation for the Study Period with the
Monthly Mean, Average Maximum, and Average Minimum for 1980-1988
V)
0
£
U
e
12
10
8
Monthly Precipitation at Tacoma
. Study year only
_ Excluding study year
* Monthly nin and nax
O
Month
147
N
D
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
of the year. Figure 9-3 illustrates this comparison. PM1(, during the study was higher than the
average maximum during September 1989, when precipitation was very low, and lower than
the average minimum level during January, when precipitation exceeded the norm.
Woodstoves are a clear example of a source with emissions modulated over a seasonal cycle.
The study period captured portions of the two extremes in this cycle: high and minimal
emissions.
Various earth-moving and construction projects, which probably contributed increased fugitive
dust, were underway in the Tideflats industrial area during the summer and fall months of the
study. Anecdotal information on these activities has been incorporated into some of the
discussion of study results.
The Simpson Tacoma Kraft pulp mill, which is the major source of particles less than PM10 in
the Tideflats, was not operating during part of December 1989, when the field sampling was
conducted for the receptor modeling study. In addition, the two monitoring sites for that study
were not downwind of Simpson, so this source was not fully reflected in the receptor modeling
analysis.
Since the period of the study several actions have or are taking place to further control
emissions: in the summer of 1990 Kaiser consolidated and covered eight acres of PAH-
contaminated scrubber sludges that were exposed and had the potential of becoming airborne;
in July 1991 the off-loading of "black ore" (ore from which lead, zinc, silver and gold are
extracted by smelting) at the Port of Tacoma's Terminal 7 was scheduled to cease; PSAPCA
is continuing efforts to improve the effectiveness of its communication and enforcement of burn
bans, thereby aiming to decrease this emissions source during the most sensitive meteorological
conditions; more stringent regulations on wood stove performance for new stove installations
are also now in effect. The likely overall effect of these actions will be to reduce air emissions,
particularly some local sources of fugitive dust. It therefore seems reasonable to conclude that
portions of the study period (summer/fall dry conditions and moderate to high winds, winter
stagnant air/heavy wood stove use) represent "worst case" emissions relative to current and
anticipated conditions.
148
-------
Chapter 9. Synthesis of Results
Figure 9-3. Comparison of Monthly Mean PM10 at the Alexander Avenue Site for the
Study Period with the Monthly Mean, Mean High, and Mean Low PM10 from 1986-1988
(Data is from PSAPCA monitoring ) .
4»
I)
e
o
•H
A
U
\
0)
o
h
0
•H
e
o
H
E
80
70
60
50
40
30
20 -
10 -
0
Study year only
Excluding study year
Mean high ft low
• ' -4- '
J A SO N D
Month
SYNTHESIS
Components of the Aerosol
>• The summer and winter high-concentration2 aerosols were qualitatively different. The
composition, concentration gradients between stations, and the differences measured
between co-located aerosol and deposition samplers suggest that the high-concentration
summer aerosol may have consisted largely of resuspended larger particles, such as fugitive
dust. These larger particles could be composed of, or serve as transport media for, metals
and PAHs. Larger particles tend to settle nearer their source than fine particles, and may
not be collected efficiently, or at all, by the aerosol sampler used in this study. The
highest concentrations of metals were found in the industrial area, particularly at the Sea-
Land site (Pb and Zn). The highest concentrations of PAHs were measured at the
Alexander Avenue site. The set of meteorological conditions favoring transport of the
summer aerosols from their sources to their receptors consisted of prolonged periods of dry
weather followed by moderate-to-high wind speeds. Concentrations of crustal metals, such
as Fe and K, were relatively higher in the summer at Alexander Avenue. A covariance
analysis of the data identified several soil minerals that contributed most strongly in the
industrial area at the Morse and Alexander Avenue sites.
2High-concentration aerosol refers to episodes during the study period when the highest
concentrations of aerosol were observed.
149
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
> The 18-day study showed that on average the late fall/winter PM10 aerosol consisted largely
of fine particles. A large percentage of the total fine-particle mass consisted of organic
compounds. High fine-particle aluminum concentrations were measured at the Alexander
Avenue site when there were light southeast winds or stagnant air. The similar PM^so
aerosol concentrations throughout the five-site sampling area during air stagnation periods
suggest a substantial influence of regional sources relative to local sources. High
concentrations of the particles were most often associated with air stagnation episodes and
low wind speeds predominantly from the southeast.
Sources of the aerosol
> All of the industrial sources, mobile sources, and residential and commercial combustion
sources in the Tideflats area contributed to some degree to the aerosol. The primary
parameters affecting the estimated contribution of each source to the aerosol at each
sampling point included the emissions levels, the particle sizes emitted, the location of
the emission source relative to the sampling point, and the meteorological conditions.
* During the summer/early fall sampling period, when larger particles appear to have been
an important component of the aerosol, earth-moving and construction activities (with
accompanying exhaust and road dust from heavy machinery) occurred in the vicinity of
the industrial area sampling sites. Fence posts were moved on the Reichhold property
adjacent to the Alexander Avenue site. An unpaved field 80 meters west of the Alexander
site was used for parking and short-term storage of large diesel trucks. Also during the
study period Kaiser had at their site approximately eight acres of wet scrubber sludge
containing PAHs, some of which had the potential of becoming airborne dust during the
dry season (Schmeil, personal communication). Two ore offloading operations of note
occur near the Sea-Land site. Kaiser offloads alumina, which is removed from the hold
of the ships by open crane and moved by conveyor belts to a storage dome. The Port
offloads lead and zinc at an adjacent berth and loads the ore into open rail cars. These
sources can only be inferred at this point based on the patterns of chemicals in the
deposition samples and their proximity to these industrial sites.
* The major source of the late fall and winter high-concentration fine-particle aerosol
(<2.5 /*m) appears to be woodsmoke that drains downslope and downwind from residential
areas into the industrial area. This was one of the more surprising results of the study,
however a number of different analyses bear out this conclusion. Receptor modeling
indicated woodsmoke was the largest contributor to fine-particle aerosol. The tracer rose
analysis of PM10 data (which projects concentrations of contaminants measured at a receptor
as vectors pointing upwind) indicated that a southeastern emissions source was dominant.
Followup studies comparing diffusion model simulations to PSAPCA data revealed that the
model input for woodsmoke emissions was an underestimate by a factor of three.
Calculations based on the population of the town of Puyallup (to the southeast) and the
assumed woodstove usage per capita were sufficient to account for the woodsmoke needed
to reconcile the model results with the PSAPCA measurements (Harrison, 1990).
> The sampling times and locations may not have been adequate to accurately reflect the
relative influence of the Simpson emissions. The Simpson mill is the largest emissions
source for PM10 in the study area. During most of the receptor modeling field study the
Simpson hogged fuel boilers were shut down for maintenance. The tracer rose analysis
also raises this question; there is only a modest western "finger point" despite Simpson's
location one mile east of the Fire Station No. 12 PSAPCA sampler. One reason for this
may be the effectiveness of the tall emissions stack in mitigating the ground level impacts
in the study area. Wind analysis shows that with the seasonal shift from summer to winter
150
-------
Chapter 9. Synthesis of Results
the percentage of winds from the SE increases. Therefore, samplers may not have been
in the prevailing downwind direction for Simpson.
* A number of the sampling locations were a short distance from major roadways. The
study results did not reach a clear conclusion on the relative importance of automotive
sources. They ranged from three to 20 percent of the source apportionment in the CMB
analyses, but the error in that analysis was estimated at ±50 percent. Diesel powered
vehicles accounted for most of the emissions. The coyariance analysis found auto
emissions (vector 2) to be most important at Brown's Point and Riverside School, the
sites furthest removed from the industrial emissions sources.
Deposition
>• Deposition rates for all the metals were greater at the industrial sites than at the Brown's
Point and Riverside School sites. PAH deposition rates were five to ten times higher at
the Alexander site than at Tyee Marina, Morse, and Sea-Land. Deposition rates for both
metals and PAHs were higher in the summer. PAH deposition was dominated by the most
abundant compounds in the paniculate air samples.
»• Kaiser appears to be the major source of the high-concentration deposition PAHs. Kaiser
is by far the largest PAH emissions source in the Tideflats area (546 kg/day of PM10 PAHs
versus an estimated 14 kg/day from all other area and point sources) and is in close
proximity to the sampler with the highest PAH deposition rates. As described above, the
measured PAH deposition could be either from stack emissions or from resuspended dust
contaminated by scrubber sludge held on the Kaiser property. It is also worth noting that
>70 percent of Kaiser's emissions are larger than PM10, and therefore more likely than
fine emissions to be deposited nearby.
* Deposition velocities calculated from the deposition samples were very high (most were
>2.2 cm/sec) compared to the deposition velocities assumed in the diffusion model (0.1
cm/sec). The high numbers may be an artifact of biases in aerosol or deposition sampling;
they could also be consistent with deposition of larger particles (compared with the PM10
fraction used for the diffusion model). This cannot be determined without data on the
specific size fractions in the deposition samples.
Transport to Commencement Bay
> The diffusion model attempted to simulate deposition and transport within the 91.5 km2
modeling domain. A separate Mobilization Coefficient Model was used to predict transport
to the receiving water. The models did not include all of the Puyallup River drainage
basin. The models calculated that one to three percent of the PM10 emissions would be
deposited and reach Commencement Bay in "prompt runoff." The diffusion model further
estimated that more than 90 percent of the PM10 emissions would be advected beyond the
boundaries of the model domain. As fine particles moving away from the source are
vertically transported into air above the surface boundary layer, their dry deposition fluxes
(which are proportional to concentration at the surface) drop rapidly with increasing
downwind range. The best guess is that another three to four percent of exported
emissions may be deposited farther afield in Puget Sound. It is important to note that these
results are based on modeling of the PM10 size fraction. Other researchers have estimated
deposition velocities closer to 0.3 cm/sec (Sehmel, 1984), which would increase computed
net deposition by a factor of three. Higher deposition rates, as suggested by the deposition
sampling results, are more likely with larger aerosol particles, however this deposition will
be limited to the area close to the emissions sources.
151
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
>• The transport calculations used mobilization coefficients that are not based on direct
measurements. These coefficients are quite low and assume that soils and vegetation will
absorb most of the deposited contaminants. The coefficients also assume typical year rain
events. If complete washoff is assumed for the modeled deposition during a severe storm
event, the model still predicts that less than 10 percent of the emissions will be transported
to Commencement Bay.
* The modeled transport numbers cannot be verified without direct measurements of runoff.
APPLICATIONS TO COMMENCEMENT BAY AND PUGET SOUND
Taking what was learned from the measurements of deposition and the model simulations of
deposition and runoff, some first order estimates of atmospheric loading to Commencement Bay
and Puget Sound can be computed and compared with other sources of contaminants to these
waters. Only limited conjectures and conclusions can be drawn from these comparisons with
the data presently available.
Mass Loading to Commencement Bay
One approach is to calculate mass loading by atmospheric deposition, using as upper and lower
bounds the mean deposition rates measured at Riverside School (a sampling site with low rates
of deposition) and the rates measured at Sea-Land (the sampling site with the highest metals
deposition rates). These rates were multiplied by an assumed area of Commencement Bay (10
km2). Table 9-1 demonstrates that these estimates are significantly lower than mass loading
estimates from point source water discharges given by Tetra Tech (1988) and Paulson et al.
(1989). Though the Tetra Tech and Paulson data differ by up to a factor of four, they are
both significantly greater than the upper bound of the atmospheric contributions. (Note: The
Paulson et al. estimate includes the Ruston shoreline and therefore covers a larger source area
than that considered by Tetra Tech.) The Puyallup River contributes more Cu, Pb, and Zn than
the point sources. One would conclude from these data that atmospheric deposition of metals
to Commencement Bay is of minor importance relative to other sources.
Table 9-1. Contaminant Mass Loading to Commencement Bay
Sources (mt/y)
Contaminant
Atmospheric
Deposition
RSfa) SL(b)
Point Sources
Tetra Tech (1988)
Point Sources
Paulson et al. (1989)(c)
Puyallup River
Curl et al. (1982)
As 0.007 0.07 5.2
Cu 0.07 0.5 3.4 17 19.2
Pb 0.08 2.4 3.1 15 7.4
Zn . 0.1 2.4 8.4 21 24
CPAH 0.013 0.055
(a) Mean deposition rate based on Riverside School deposition samples.
(b) Mean deposition rate based on Sea-Land deposition samples.
(c) Paulson et al. numbers have been adjusted to subtract the ASARCO discharges and the
nonpoint contributions.
152
-------
Chapter 9. Synthesis of Results
This comparison is still not definitive because of how the atmospheric deposition was computed.
The Sea-Land deposition rate is probably atypically high due to site-specific activities nearby
that generate fugitive dust. Therefore, extrapolation of this rate to the entire bay is probably
an overestimate of the direct deposition over the water surface. Conversely, because these
estimates for atmospheric deposition are based solely on direct deposition to the water, there
is an unqualified contribution of runoff from the watershed that is missing. Including the
runoff would yield a higher estimate for the overall contribution of atmospheric deposition.
To try and incorporate runoff, another approach for a mass loading budget is to rely on the
diffusion model and the runoff calculated using the mobilization coefficient model. A range of
one to three percent of the emitted contaminants is calculated to reach the modeled region of
Commencement Bay, either by direct deposition or "prompt" runoff (see Table 7-4). These
estimated atmospheric loadings are presented in Table 9-2.
Table 9-2. Estimated Mass Loading to Commencement Bay Based on WV3 Model
(modeling domain of 91.5 km2) •
Contaminant 4 - 7% of Emissions
As
Cu
Pb
Zn
PAH
CPAH
.0027-
.0141-
.0258 -
.0361 -
2.044 -
.269 -
.0081 mt/y
.0422 mt/y
.0774 mt/y
. 1082 mt/y
6.132 mt/y
0.087 mt/y
These rates are almost certainly an underestimate: they are based on a very low assumed
particle deposition velocity (0.1 cm/sec), only the PM10 fraction of emissions is modeled
(thereby missing the large particles), some emissions sources outside the modeling domain may
be important and are not included, and the mobilization coefficients used are very low
(assuming that most contaminants will be filtered out before stormwater flows into the bay).
In addition, the modeling domain did not include downtown Tacoma, and runoff from this
source into Commencement Bay. If deposition velocities are increased by a factor of three,
which is not unreasonable, this would increase the runoff by a factor of three. Another
adjustment that might be made in these calculations is to assume a worst case scenario of all
the deposited contaminants being available to be mobilized into runoff to the bay during severe
storms. This would further double the estimates. Using the upper range estimates from Table
9-2, the recalculated loadings from the model would be:
Table 9-3. Estimated Mass Loading to Commencement Bay Based on Adjusting WV3
Model Results for Increased Deposition Velocity and Mobilization
Contaminant Estimated Mass Loading
As .0486 mt/y
Cu .2532 mt/y
Pb .4644 mt/y
Zn .6492 mt/y
PAH 36.792 mt/y
CPAH 4.842 mt/y
153
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
These loadings are still significantly less than the point source loadings for metals included in
Table 9-1. It is not reasonable to project further with these data without better measurements
of runoff. While many studies have been undertaken in recent years to examine stormwater,
particularly in the Tideflats area, most are not able to separate the measured stormwater
contaminants into their respective sources to arrive at a reliable estimate for the airborne
contribution.
Input Timing and Location
An alternative approach to examine the potential significance for Commencement Bay and Puget
Sound of atmospheric deposition is to look more closely at how the atmospheric inputs are
occurring and the specific zones in the receiving water that are most likely to be affected.
Direct deposition to the bay and some intertidal stormwater enter the receiving water at the sea
surface in a 50 - 100 micron-thick zone that has been described as the sea surface microlayer.
The residence time of metals in the sea surface microlayer was determined by Hardy and Apts
(1983) to be on the order of three to 20 hours for urban air particulate matter [using a
microlayer thickness of 50 microns and a wind speed of 3.6 m/sec (average wind speed in Puget
Sound)]. The concentrations of metals in the sea-surface microlayer of Commencement Bay
can be estimated using the mean deposition rates for Sea-Land and Tyee Marina sites from
Table 4-11 and using the assumptions of a thickness of 50 microns and a residence time of three
or twenty hours.
The estimated bounds of chemical concentrations in the sea-surface microlayer are presented
in Table 9-4 for (1) the Tyee Marina site (representing the low end of deposition rates) at a
residence time of three hours, and (2) the Sea-Land site (representing the high end of the
deposition rates) at a residence time of 20 hours. The last column of the table contains
literature values for actual chemical analyses of sea-surface microlayer from Puget Sound urban
bays (Hardy et al., 1987).
Table 9-4. Predicted Concentrations of Specific Contaminants in the Sea-Surface
Microlayer(a)
ug/L
Contaminant
TM
3-hour
SL
20-hour
Literature Value
Hardy et al. (1987)
As 4.5 300
Cu 144 2,470 51 - 3,200
Pb 87 10,800 38 - 650
Zn 267 14,500 135 - 1,420
CPAH 11 225 8 - 148
(a) Concentrations assume a 50 micron-thick microlayer, three-hour or 20-hour residence
times, and mean deposition rates at Tyee Marina or Sea-Land.
The estimates of contaminant concentrations from atmospheric deposition (Table 9-4, columns
2 and 3) indicate that it is possible to account for the high measured concentrations (column
4) with the cross-media transfer of these toxic contaminants from the air, even at the lower
input rate.
154
-------
Chapter 9. Synthesis of Results
The calculated runoff loadings also do not take into account that this input is usually delivered
in a pulsed manner, as a result of rainfall, and is likely to be channeled into the Bay via an
intertidal or subtidal pipe, generally nearshore. As with the microlayer, it is conceivable that
despite low overall loadings, the pulsed timing and the input location may make atmospheric
runoff a critical factor in determining the water quality in a circumscribed zone near the
stormwater pipe. The study results also suggest that large particles, which settle near their
source, are an important component of the deposition resulting from the Tideflats emissions.
If so, this would act in concert with the stormwater inputs to subject the nearshore zone to the
highest contaminant levels from atmospheric inputs.
Loading to Puget Sound
If large particles, such as fugitive dust, are likely to deposit close to their source, a remaining
question is the fate of the fine (<2.5 /*m) particles in the ambient aerosol. The study identified
woodsmoke as a significant component of the fine-particle mass during winter air stagnation
periods. As described above, the diffusion model predicted that only one to three percent of
the particles < 10 /*m would be deposited and transported to Commencement Bay in "prompt"
runoff. Combining the diffusion model results with some general information about dominant
wind patterns and the topography of Puget Sound results in an estimate that an additional three
to four percent of the Tideflats PM10 emissions would reach Puget Sound beyond the immediate
modeling domain of Commencement Bay (see Chapter 7). As the emissions move downwind
their probability of being deposited decreases rapidly with distance from the source. The
prevailing regional wind direction is towards the northeast, therefore the emissions from the
Tideflats will not be carried over the bulk of Puget Sound. Data from the sampling done for
this study are inadequate to allow further speculation on the fate of these fine particle emissions.
155
-------
Chapter 10. Conclusions and Recommendations
This study addressed two related management concerns:
* Developing a better understanding of the importance of atmospheric deposition relative to
other inputs of toxic contaminants to Commencement Bay; and
»> Developing efficient and cost-effective tools for assessing this question in other reaches
and embayments of Puget Sound, and other water bodies as well.
The study was scaled as a pilot-level effort to develop a first-order approximation of the relative
significance of atmospheric deposition. The results provide: 1) new insight into the contributing
sources and the composition of the ambient aerosol in the Tacoma Tideflats, 2) first order
estimates of atmospheric deposition in regions adjacent to Commencement Bay, and 3)
recommendations on how the tools used for this study need to be refined for future application.
This was a relatively short-term study of a complex air emissions system; not surprisingly, some
important questions remain unanswered, particularly how the study results apply beyond
Commencement Bay to other regions of the Sound.
CONCLUSIONS
Relative Importance of Atmospheric Deposition
[Note: These conclusions set atmospheric deposition in perspective relative to other sources of
contaminants in a heavily industrialized region of Puget Sound, where the overall contaminant
loading is high. It should not be assumed that the relative contribution of atmospheric
deposition is the same elsewhere in the Sound.]
> Direct atmospheric deposition of metals (i.e. atmospheric contaminants falling directly on
the surface of Commencement Bay) appears to be a small contributor, in terms of mass
loading, relative to point source water discharges of metals to Commencement Bay.
> The data available do not allow definitive conclusions on the relative importance of
atmospheric inputs of PAHs. The deposition measurements of PAHs showed a marked
gradient between sites, with Alexander Avenue, the closest site to the Kaiser Aluminum
facility, exhibiting the highest PAH deposition by more than a factor of three. Kaiser is
the largest PAH emissions source in the Tideflats area. Because greater than 70 percent
of Kaiser's emissions are particles larger than 10 microns, these emissions are likely to
settle nearby, and therefore may be an important local source of PAHs. [Note: After the
sampling was conducted for the study, Kaiser covered PAH-contaminated scrubber sludges
on their property that had the potential to become part of the fugitive dust loading of PAHs
to the Bay.]
* Wopdsmoke, another source of PAHs, is an important contributor to the ambient aerosol
during winter periods of air stagnation. However, PAHs generally make up less than 1
percent of the woodsmoke emissions mass. In addition, woodsmoke emissions are
predominantly in the less than 2.5 micron particle size fraction, therefore they are not
expected to settle nearby and be an important component of atmospheric deposition to
Commencement Bay.
157
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
* The total input from atmospheric deposition cannot be assessed from the results of this
study because of uncertainty in the estimates of atmospheric deposition on land and its
subsequent runoff to the Bay.
* Atmospheric deposition may be significant relative to other inputs in particular zones,
especially close to large sources of fugitive dust, near discharge pipes for stormwater, and
at the sea-surface microlayer (top 50-100 pm of the water column). These focused entry
points into the water, coupled with pulsed inputs (such as stormwater) create periodic
conditions where atmospheric deposition may be the dominant source of toxic contaminants,
resulting in a steep gradient in contaminant concentrations in the receiving water.
* The samples obtained for nutrients, PCBs, and aliphatic hydrocarbons were inadequate to
serve as a basis for assessing the relative importance of these atmospheric contributions.
»• It is not possible from the data to make a definitive prediction of deposition in the far field
(Puget Sound beyond Commencement Bay). Emissions that are likely to be carried beyond
Commencement Bay are generally made up of vapors and particles less than 10 microns
in diameter. For this size particle there is a rapidly decreasing probability of deposition
with distance from the source.
Effectiveness of the Tools
Aerosol and Deposition Study
These measurements were central to the study. There were a number of potential unqualified
biases in the sampling, however the data reveal patterns that were fundamental to developing
the overall understanding of the components and behavior of the ambient aerosol. These data
were verified and supported by the other studies and analyses. In designing future sampling
the following factors deserve particular attention:
* Consistency in the particle size fraction sampled for each study component;
> Length of sample collection period;
* Number of samples chemically analyzed;
* Sampling protocol/amount for PCBs, aliphatic hydrocarbons, nutrients, and vapor PAHs;
> Sampling locations relative to wind field and primary sources.
Receptor Modeling
This modeling effort was based on a very short (18 days) and spatially limited (two sites 3.5 km
apart) monitoring study. Nevertheless, the chemical mass balance (CMB) or receptor modeling
discerned some important features of the ambient "fine" aerosol (particles less than 2.5
microns), which were later borne out in the paired comparisons of studies. Specifically, the
model highlighted the importance of woodsmpke in the fine aerosol during winter air stagnation.
The design of future receptor modeling applications should pay particular attention to:
* Temporal and spatial representativeness of sampling;
* Information on the chemical signatures specific to the local sources and their variability
over time;
158
-------
Chapter 10. Conclusions and Recommendations
* Sample quantity for organic chemical speciation.
Diffusion Modeling
The diffusion model showed promise when compared with field measurements of the aerosol
mass, particularly for particles in the less than 10 micron size fraction. There was less
agreement between model simulations and field measurements for specific chemicals, but this
is not surprising given that the representativeness of sampling sites is uncertain when there are
significant local sources and variability in emissions with time. CMB modeling and additional
model comparisons with PSAPCA data (Harrison, 1990) helped to diagnose some potential
emissions input errors in the model. In addition, the intercomparison with the field deposition
measurements suggests that deposition velocities need to be varied to more realistically represent
the behavior of different chemicals and different particle sizes. For future applications, model
refinements should address:
*• Site-specific chemical and particle distribution information;
* Extension of the modeling domain;
* Particle size fractions modeled;
* Deposition velocities specific to a range of particle sizes.
Mobilization Coefficient Model
Mobilization coefficients were a critical component of computing total transport to
Commencement Bay. The coefficients used were theoretically derived and were not field
verified. The following should receive particular attention in future development of these
coefficients:
*• Field validation of the mobilization coefficients;
> Incorporation of severe storm events;
>• Size of model domain.
RECOMMENDATIONS
Analysis of the study results and performance of the tools suggests several areas where further
basic research is needed on certain critical technical methods and physical processes, including:
* Aerosol particle distribution and associated chemistry;
* Fugitive dust transport relative to wind speed, soil moisture, and vehicle activities;
> Particle deposition dynamics for different particle sizes and chemical constituents;
>• Deposition sampling techniques;
* Chemical and physical dynamics of contaminant mobilization;
* Hybrid dispersion modeling (a combination of models).
159
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
The current study produced first order estimates of some important processes. Significant
refinements in the methods are recommended if a more definitive picture of the relative
importance of atmospheric deposition is desired. To augment what has already been done,
additional studies that might be pursued either independently or as part of a larger program
include:
>• Extensive characterization of ambient and source particle size distributions and chemistry;
* Sampling of the microlayer and nearshore zones to complement the deposition sampling;
* Combined sewer overflow (CSO)/stormwater monitoring;
> Refinement of emissions inventories, including temporal variations;
> Vertical and horizontal wind profiling;
* Hybrid dispersion modeling.
160
-------
GLOSSARY
ADSORPTION
Adhesion of the molecules of a gas, liquid, or dissolved substance to a surface.
AEROSOL
A suspension of colloidal particles in a gas.
AEROSOL WASHOUT
Below-cloud scavenging of particles in the air by falling water (in liquid or solid form).
CHEMICAL MASS BALANCE MODELING
Apportioning the chemicals measured at a site to the potential sources.
COLLOIDAL
Very small particles larger than molecules, but small enough they remain suspended in a fluid
medium.
COMBUSTION-DERIVED PAHs (CPAHs)
Polycyclic aromatic hydrocarbons resulting from burning, generally considered to be
fluoranthene through benzo(g,h,i)perylene.
CONDENSATION
The process of changing from a vapor to a liquid.
DRY DEPpSITION
The deposition of particles during dry weather conditions. This can occur either by
gravitational settling or as a results of turbulent mixing causing the particles to impact the
deposition surface.
FUGITIVE DUST -
Paniculate matter or any visible air contaminant, other than uncombined water, that is not
collected by a capture system and emitted by a stack, but is released to the atmosphere at the
point of generation.
MASS BALANCE MODEL
Balance of incoming, outgoing, and resident chemicals.
MICROLASER
The approximate 50/i-thick boundary layer at the sea-surface.
PRECIPITATION SCAVENGING
Collection of particles in the air by rain, snow, sleet, etc.
RAIN-OUT
Scavenging of aerosol components by processes taking place within clouds, such as the
formation of condensation nuclei.
STRATIFICATION
The formation of layers with respect to buoyancy.
161
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
TRANSFER COEFFICIENTS
Factors which describe the rate at which transfers occur.
VAPOR EXCHANGE
Exchange of vapor principally between the air and water surfaces.
VOC
Volatile organic compound.
VOLATIZATION
Evaporation.
WET REMOVAL MECHANISM
Removal of particles from the atmosphere by both in-cloud and below-cloud processes.
XRF
Energy-dispersive x-ray fluorescence. A technique used for metals analysis.
162
-------
REFERENCES
Allen, A.G., M. Radojevic, and R.M. Harrison. 1988. Atmospheric speciation and wet
deposition of alkyllead compounds. Environmental Science and Technology 22:517-522.
Anderson, M.A. and A.J. Rubin. 1981. Adsorption of inorganics at solid-liquid interfaces.
Ann Arbor Science Publ., Ann Arbor, MI.
Arimoto, R. 1987. Atmospheric deposition of chemical contaminants to the Great Lakes.
Center for Atmospheric Chemistry Studies. University of Rhode Island, Narragansett, RI.
59pp.
Barrick, R.C. 1982. Flux of aliphatic and polycyclic aromatic hydrocarbons to Central Puget
Sound from Seattle (West Point) primary sewage effluent. Environmental Science and
Technology 16:682-692.
Barrick, R.C. et al. 1988. Sediment quality values refinement: 1988 update and evaluation of
Puget Sound AET. Volume 1. Final Report. Prepared for Tetra Tech, Inc. and the U.S.
Environmental Protection Agency Region 10, Office of Puget Sound, Seattle, WA. PTI
Environmental Services, Bellevue, WA. 74 pp. + appendices.
Bates, T.S. et al. 1987. Hydrocarbon distributions and transport in an urban estuary.
Environmental Science and Technology 21:193-198.
Bierman, V.J. and W.R. Swain. 1982. Mass balance modeling of DDT dynamics in Lakes
Michigan and Superior. Environmental Science and Technology 16:572-579.
Boehm, P.D. et al. 1985. Contaminant residence times and ecosystem recovery rates for
shelf and estuarine ecosystems. Final Report. Submitted to National Oceanic and Atmospheric
Administration. Battelle, Washington, D.C. 37 pp.
Bopp, R.F. 1983. Revised parameters for modeling the transport of PCB components across
an air-water interface. Journal of Geophysical Research 88:2521-2529.
Broman, D. et al. 1988. A multi-sediment-trap study on the temporal and spatial variability
of polycyclic aromatic hydrocarbons and lead in an anthropogenic influenced archipelago.
Environmental Science and Technology 22:1219-1228.
Buat-Menard, P. and R. Chesselet. 1979. Variable influence of the atmospheric flux on the
trace metal chemistry of oceanic suspended matter. Earth and Planetary Science Letters 42:399-
411.
Buffo J. 1979. Water pollution control early warning system. Section 1: Non-Point Source
Loading Estimates. A report to Seattle Metro., WA. 47 pp.
Galloway, C.P. et al. 1989. A refinement of the potassium tracer method for residential
woodsmoke. Atmospheric Environment 23:67-69.
Cambray, R.S., D.F. Jefferies, and G. Topping. 1979. The atmospheric input of trace
elements to the North Sea. Marine Science Communications 5:175-194.
163
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Canadian Embassy Newsletter. 1988. Environmental perspectives: air toxic contamination of
the Great Lakes. Vol. II, Number I.
Cannon, G.A. and M.W. Grigsby. 1982. Observations of currents and water properties in
Commencement Bay. NO A A Technical Memorandum OMPA-22. National Oceanic and
Atmospheric Administration, Boulder, CO.
Chevreuil, M. et al. 1989. Atmospheric pollution and fallout by PCBs and organochlorine
pesticides (Ile-de-France). Water, Air, and Soil Pollution 43:73-83.
Chuang, J.C., S.W. Hannan, and N.K. Wilson. 1987. Field comparison of polyurethane foam
and XAD-2 resin for air sampling for polynuclear aromatic hydrocarbons. Environmental
Science and Technology 21(5):804-810.
Cooper, J.A., C.A. Frazier and I.E. Houck. 1985. Seattle-Tacoma aerosol characterization
study (STAGS) Vol. 1: Quantitative source apportionment of paniculate matter on days of
excessive paniculate concentrations using chemical mass balance methods. Final report
prepared for the Puget Sound Air Pollution Control Agency, Seattle, WA.
Core, J. 1989. Pacific Northwest source profile library. Report to the U.S. Environmental
Protection Agency, Washington, D.C.
Countess R.J., G.T. Wolff and S.H. Cadle. 1980. The Denver winter aerosol: a
comprehensive chemical characterization. J. Air Pollut. Control Assoc. 30:1194-1200.
Cox, W.M. 1988. Protocol for determining the best performing model. Personal
communications. U.S. EPA, Office of Air Quality and Planning Standards, Technical Support
Division, Source Receptor Analysis Branch, RTP, Washington, D.C.
Crecelius, E.A. 1981. Prediction of marine atmospheric deposition rates using total Be7
deposition velocities. Atmospheric Environment 15:579-582.
Crecelius, E.A. 1980. The solubility of coal fly ash and marine aerosols in seawater. Marine
Chemistry 8:245-250.
Crecelius, E.A. et al. 1980. Background air paniculate'chemistry near Colstrip, Montana.
Environmental Science and Technology 14(4): 422-428.
Cross, J.N. et al. 1988. Contaminant concentrations and toxicity of sea-surface microlayer
near Los Angeles, California. Marine Environmental Research 23:307-323.
Curl, H.C., Jr. (ed.). 1982. Estuarine and coastal pollutant transport and transformation:
the role of particulates. The NOAA/OMPA Section 202 Research Program, Pacific Marine
Environmental Laboratory, Seattle, Washington.
Currie, L.A., G.A. Klouda, and J.A. Cooper. 1980. Radiocarbon 22:349-362
Doskey, P.V. and A.W. Anders. 1981. Modeling the flux of atmospheric polychlorinated
biphenyls across the air/water interface. Environmental Science and Technology 15:705-711.
Duce, R.A. et al. 1972. Enrichment of heavy metals and organic compounds in the surface
microlayer of Narragansett Bay, Rhode Island. Science 176:161-163.
164
-------
References
Duinker, J. C. and F. Bouchertall. 1988. On the distribution of atmospheric polychlorinated
biphenyl congeners between vapor phase, aerosols, and rain. Environmental Science and
Technology 23:57-62.
Dzubay T.G. et al. 1982. Visibility and aerosol composition in Houston, Texas.
Environmental Science and Technology 16, 514-525.
Ebbert, J.C., J.E. Poole, and K.L. Payne. 1985. Data collected by the U.S. Geological
Survey during a study of urban runoff in Bellevue, Washington, 1979-1982. File Report 84-
064. U.S. Geological Survey, Tacoma, WA.
Edgington, D.N. and J.A. Robbins. 1975. Records of lead deposition in Lake Michigan
sediments since 1800. Environmental Science and Technology 10:266-274.
Eisenreich, S.J. 1980. Atmospheric input of trace metals to Lake Michigan. Water, Air,
and Soil Pollution 13:287-301.
Eisenreich, S.J. (ed.) 1981. Atmospheric pollutants in natural waters. Ann Arbor Science
Publishers, Ann Arbor, MI. 512 pp.
Eisenreich, S.J. 1985. Atmospheric deposition workshop on organic contaminant deposition
to the Great Lakes Basin. Report to the U.S. Environmental Protection Agency, Great Lakes
National Program Office, Chicago, IL. 30 pp.
Eisenreich, S.J. and G.J. Hollod. 1979. Accumulation of polychlorinated biphenyls (PCBs)
in surficial Lake Superior sediments: atmospheric deposition. Environmental Science and
Technology 13:569-573.
Eisenreich, S.J., B.B. Looney, and J.D. Thornton. 1981. Airborne organic contaminants in
the Great Lakes ecosystem. Environmental Science and Technology 15:30-38.
Elder, J. et al. 1987. Toxic air pollution in the Great Lakes Basin: a call for action. A
preliminary report. Sierra Club Midwest. 28 pp.
Envir. Law Handbook. 1989. Tenth edition. Government Institutes, Inc., Rockville, MD.
p. 263.
Fisher, D. et al. 1988. Polluted coastal waters: the role of acid rain. Environmental Defense
Fund, N.Y. 102 pp.
Forstner, U. and G. Wittman. 1979. Metal pollution in the aquatic environment. Springer-
Verlag, NY. 486 pp.
Galloway, J.N. et al. 1982. Trace metals in atmospheric deposition: a review and assessment.
Atmospheric Environment 16:1677-1700.
Galvin, D.V. 1987. Toxicants in urban runoff, pp. 176-210 In: R.W. Seabloom and G.
Plews (eds.). Northwest nonpoint source pollution conference, proceedings, Seattle, March 24-
25, 1987. Washington State Department of Social and Health Services, Olympia, WA.
Galvin, D. V. and R. K. Moore. 1984. Toxicants in urban runoff. Metro Toxicant Program
Report No. 2. Report to the Toxicant Control Planning Section, Water Quality Division,
Municipality of Metropolitan Seattle, WA.
165
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Gatz, D.F. 1975. Pollutant aerosol deposition into southern Lake Michigan. Water, Air,
and Soil Pollution 5:239-251.
Gatz, D.F. et al. 1988. Great Lakes atmospheric deposition (GLAD) network, 1982 and
1983: data analysis and interpretation. GLNPO Report No. 2. Great Lakes National Program
Office, Chicago, IL. 69 pp.
Gibbs, R. 1967. Amazon River: environmental factors that control its dissolved and suspended
load. Science (156)1734:1737.
Gschwend, P.M. and R.A. Kites. 1981. Fluxes of polycyclic aromatic hydrocarbons to marine
and lacustrine sediments in the northeastern United States. Geochimica et Cosmochimica Acta
45:2359-2367.
Gucinski, H., J.T. Hardy, and H.R. Preston. 1988. Implications of toxic materials
accumulating in the surface microlayer in Chesapeake Bay. In: Understanding the estuary:
advances in Chesapeake Bay Research, proceedings. Baltimore, Maryland, April, 1988. U.S.
EPA, Chesapeake Bay Program Liaison Office, Annapolis, MD.
Hardy, J.T. and L. Antrim. 1988. Distribution and biological effects of sea-surface
contamination in Puget Sound. In: Puget Sound Research Conference, proceedings, Seattle,
March 18-19, 1988. Puget Sound Water Quality Authority, Olympia, WA. 400-408 pp.
Hardy, J.T. and C.W. Apts. 1989. Photosynthesis carbon reduction: high rates in the sea-
surface microlayer. Marine Biology 101(3):411-417.
Hardy, J.T. et al. 1988. A hydrophobic large-volume sampler for collecting aquatic surface
micrplayers: characterization and comparison with the glass plate method. Canadian Journal
of Fisheries and Aquatic Science 45:822-826.
Hardy, J.T. 1987. Contamination and toxicity of the sea-surface microlayer in Puget Sound.
In: Grey, R.H. et al., eds. Proceedings of the 24th Hanford Life Sciences Symposium—Health
and Environmental Research on Complex Organic Mixtures pp. 643-655. U.S. Department
of Energy and Battelle Memorial Institute, Richland, WA.
Hardy, J.T. 1987. Guest editorial: anthropogenic alteration of the sea surface. Marine
Environmental Research 23:223-225.
Hardy, J.T. et al. 1990. Aquatic surface microlayer contamination in Chesapeake Bay.
Marine Chemistry. 28(4):333-351.
Hardy, J.T. et al. 1987. The sea-surface microlayer of Puget Sound: Part I. Toxic effects
on fish eggs and larvae. Marine Environmental Research 23:227-249.
Hardy, J.T. et al. 1987. The sea-surface microlayer of Puget Sound: Part II. Concentrations
of contaminants and relation to toxicity. Marine Environmental Research 23:251-271.
Hardy, J.T. and C.E. Cowan. 1986. Model and assessment of the contribution of dredged
material disposal to sea-surface contamination in Puget Sound. PNL-5804. Prepared for the
U.S. Army Corps of Engineers. Battelle, Pacific Northwest Laboratories, Richland, WA. 8
pp. + appendices.
166
-------
References
Hardy, J.T., E.A. Crecelius, and R. Kocan. 1986. Concentration and toxicity of sea-surface
contaminants in Puget Sound. PNL-5834. Prepared for National Oceanic and Atmospheric
Administration. Battelle, Pacific Northwest Laboratories, Richland, WA. 46 pp.
Hardy, J.T. and J.B. States. 1986. Workshop on the sea-surface microlayer in relation to
ocean disposal, proceedings, Airlie, Virginia, December 18-19, 1985. BN-SA-2367. Battelle,
Pacific Northwest Laboratories, Richland, WA.
Hardy, J.T. 1985. Draft letter report: technical support for the ocean incineration regulation
model-review and recommendations for model revisions. Attachment 1: The sea surface
microlayer. Prepared for U.S. Environmental Protection Agency. Battelle, Washington
Environmental Program Office, Washington, D.C. 25 pp.
Hardy, J.T. et al. 1985. Sea-surface microlayer metals enrichments in an urban and rural
bay. Estuarine, Coastal and Shelf Science 20:299-312.
Hardy, J.T. et al. 1985. The sea-surface microlayer: fate and residence times of atmospheric
metals. Limnology and Oceanography 30:93-101.
Hardy, J.T. and C.W. Apts. 1984. The sea-surface microlayer: phytoneuston productivity
and effects of atmospheric paniculate matter. Marine Biology 82:293-300.
Hardy, J.T. and C.W. Apts. 1983. Fate and effects of heavy metals in the sea surface
microlayer. In: Proceedings of the International Conference on Heavy Metals in the
Environment. Commission of European Communities, Ltd., Edinburgh, United Kingdom.
Hardy, J.T. 1982. The sea surface microlayer: biology, chemistry and anthropogenic
enrichment. Progressive Oceanography 11:307-328.
Hardy, J.T. and E.A. Crecelius. 1981. Is atmospheric particulate matter inhibiting marine
primary productivity? Environmental Science and Technology 15:1103-1105.
Harrison, H.C. et al. 1990. Air quality during stagnations: a comparison of WYNDvalley with
RAM at five sites. J. Assoc. of Air and Waste Management 40:47-52.
Harrison, H. 1988. WYNDvalley: an eulerian-grid air quality dispersion model with versatile
boundaries, sources, and winds. Presented at APCA-EPA Specialty Conference, Receptor
Models in Air Resource Management, San Francisco, February 25-26, 1988. Air Pollution
Control Association, Pittsburgh, PA. 9pp.
Harrison, H. (in press). Where does it come from? Polar "fluxgrams" for air-quality
management. Journal of the Air and Waste Management Association.
Harrison, R.M., M. Radojevic, and S.J. Wilson. 1986. The chemical composition of highway
drainage waters, IV. Alkyllead compounds in runoff waters. The Science of the Total
Environment 50:129-137.
Hawthorne, S.B. et al. 1988. Identification of methoxylated phenols as candidate tracers for
atmospheric wood smoke pollution. Environmental Science and Technology 22:1191-1196.
Helz, G.R. 1976. Trace element inventory for the northern Chesapeake Bay with emphasis
on the influence of man. Geochimica et Cosmochica Acta 40:573-5£
167
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Hoffman, EJ. et al. 1985. Stormwater runoff from highways. Water, Air, and Soil Pollution
25:349-364.
Holzworth, G. 1972. Mixing heights, wind speeds, and potential for urban air pollution
throughout the contiguous United States. Publications AP-101. U.S. Environmental Protection
Agency, Office of Air Programs, Washington, D.C.
Hopke, P.K. 1985. Receptor modeling in environmental chemistry. Wiley Interscience, N.Y.
319 pp.
Horner, R.R. et al. 1986. Design of monitoring programs for determination of ecological
change resulting from nonpoint source water pollution in Washington State. A report to the
Washington State Department of Ecology. Environmental Engineering and Science Program,
Department of Civil Engineering, University of Washington, Seattle, WA. 195 pp.
Horowitz, A. J. 1984. A Primer on Trace Metal-Sediment Chemistry. U. S. Geological
Survey. Open-file Report No. 84-709. U.S. Dept of Interior, Geological Division.
Hubbard, T. 1988. Urban stormwater management in the Seattle-King County region~An
issue paper. Draft Report. Municipality of Metropolitan Seattle, WA. 30 pp.
Hunsaker, C.T., M.V. Huq, and S.M Adams. 1987. A regional screening method for
classifying pollutant status of coastal waters. In: Coastal Zone '87. Volume 3. The Fifth
Symposium on coastal and ocean management, proceedings, Seattle, Washington, May 26-29,
1987. American Society of Civil Engineers, N.Y. pp. 2712-2725.
Hunter, K.A. 1980. Processes affecting paniculate trace metals in the sea surface microlayer.
Marine Chemistry 9:49-70.
Johnson R.L., J.J. Shah, R.A. Gary and J.J. Huntzicker. 1981. An automated thermal-optical
method for the analysis of carbonaceous aerosol. In: Atmospheric aerosol source/air quality
relationships. Macias, E.S., and P.K. Hopke eds., ACS Symposium Series 167:223-233.
Kalman, D. and T.V. Larson. 1987. Puget Sound receptor modeling feasibility study.
Prepared for the Puget Sound Air Pollution Control Agency, Seattle, WA.
Karickhoff, S.W. 1980. Sorption kinetics of hydrophobic pollutants in natural sediments.
pp. 193-206. In: Baker, R.A., ed., Contaminants and sediments. Vol. 2: Analysis, chemistry,
biology. Ann Arbor Science Publ., Ann Arbor, MI.
Koutrakis, P. and J.D. Spengler. 1987. Source apportionment of ambient particles in
Steubenville, OH using specific rotation factor analysis. Atmospheric Environment 21: 1511-
1519.
Lewis, C.W. et al. 1988. Contribution of woodsmoke and motor vehicle emissions to ambient
aerosol mutagenicity. Environmental Science and Technology 22:968-971.
Lewis, C.W. et al. 1986. Receptor modeling study of Denver winter haze. Environmental
Science and Technology 20:1126-1136.
Lewis, C.W. et al. 1988. Sources of fine particle organic matter in Boise. In: 1988
EPA/APCA International Symposium: Measurement of Toxic and Related Air Pollutants,
proceedings, Raleigh, NC, May 1988. Air Pollution Control Association, Pittsburgh, PA.
168
-------
References
Lewis, C.W. and W. Einfeld. 1985. Origins of carbonaceous aerosol in Denver and
Albuquerque during winter. Environment International 11:243-247.
Long, E.R. 1982. An assessment of marine pollution in Puget Sound. Marine Pollution
Bulletin 13:380-383.
Ludwick, J.D., T.D. Fox, and S.R. Garcia. 1977. Elemental concentrations of northern
hemispheric air at Quillayute, Washington. Atmospheric Environment 11:1083-1087.
Mamane, Y. 1990. Estimate of municipal refuse incinerator contribution to Philadelphia
aerosol using single-particle analysis - II ambient measurement. Atmospheric Environment
24B: 127-135.
McMahon, T.A. and PJ. Denison. 1979. Empirical atmospheric deposition parameters: a
survey. Atmospheric Environment 13:571-585.
McVeety, B.D. and R.A. Kites. 1988. Atmospheric deposition of polycyclic aromatic
hydrocarbons to water surfaces: a mass balance approach. Atmospheric Environment 22:511-
536.
Mills, W.B. et al. 1985. Water quality assessment: a screening procedure for toxic and
conventional pollutants in surface and ground water. EPA/600/6-85/002. U.S. Environmental
Protection Agency, Athens, GA.
Medine, A.J., and S.C. McCutcheon. 1987. Fate and transport of sediment-associated
contaminants. EPA/600/D-87/356. U.S. Environmental Protection Agency, Washington, D.C.
Metcalf and Eddy. 1971. Storm water management model, Vol. I. Final Report. EPA Report
No. 11024 DOC 07/71. U.S. Environmental Protection Agency, Washington, D.C. 352 pp.
Moller, U. and G. Schumann. 1970. Mechanisms of transport from the atmosphere to the
earth's surface. Journal of Geophysical Research 75:3013-3019.
Morel, F.M. 1983. Principles of aquatic chemistry. Wiley, N.Y., 466 pp.
Murphy, P.P. et al. 1988. The transport and fate of paniculate hydrocarbons in an urban
fjord-like estuary. Estuarine, Coastal and Shelf Science 26:1-22.
Murphy, T.J. 1985. Background planning document for the design of a Great Lakes
atmospheric inputs and sources (GLAIS) network. Prepared for U.S. Environmental Protection
Agency, Great Lakes National Program Office. Chicago, IL. 29 pp.
Murphy, T.J. and C.P. Rzeszutko. 1977. Precipitation inputs of PCBs to Lake Michigan.
Journal of the Great Lakes Research 3:305-312.
0
National Academy of Sciences. 1972. Paniculate polycyclic aromatic organic matter. NTIS:
PB-212-940. Washington, D.C. 375 pp.
Neff, J.M. 1979. Polycyclic aromatic hydrocarbons in the aquatic environment: sources,
fates and biological effects. Applied Science Publ., N.Y. 262pp.
169
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Nielson, K. K. 1977. Matrix corrections for energy dispersive x-ray fluorescence analysis
of environmental samples with coherent/incoherent scattered x-rays. Analytical Chemistry
48(4):645-648.
Fade, J. 1990. Personal Communication. Puget Sound Air Pollution Control Agency, Seattle,
WA.
Pattenden, N.J., R.S. Cambray, and K. Playford. 1981. Trace and major elements in the
sea-surface microlayer. Geochimica et Cosmochimica Acta 45:93-100.
Patterson, C. and D. Settle. 1974. Contribution of lead via aerosol deposition to the Southern
California Bight. Journal de Recherches Atmospheriques 8:957-960.
Paulson, A.J. et al. 1989. Separate dissolved and particulate trace metal budgets for an
estuarine system: an aid for management decisions. Environmental Pollution 57:317-319.
Paulson, A.J. et al. 1988. Sources and sinks of Pb, Cu, Zn and Mn in the main basin of
Puget Sound. NOAA Technical Memorandum ERL PMEL-77. Pacific Marine Environmental
Laboratory, Seattle, WA. 26 pp.
Paulson, A.J. and R.A. Feely. 1985. Dissolved trace metals in the surface waters of Puget
Sound. Marine Pollution Bulletin 16:286-291.
Pederson, P. et al. 1930. Effects of fuel, lubricant, and engine operating parameters on the
emissions of polycyclic aromatic hydrocarbons. Environmental Science Technology
Peirson, D.H., P. A. Cawse, and R.S. Cambray. 1974. Chemical uniformity of airborne
particulate material, and a maritime effect. Nature 251:675-679.
Pitt, R. and P. Bissonnette. 1984. Bellevue urban runoff program. Summary Report, pp.
57-63. City of Bellevue Storm and Surface Water Utility, Bellevue, WA.
Pollack, A.J. and D.L. Smith. 1990. Final data report on Tacoma source apportionment
VOC sampling study, Work Assignment Contract No. 8Y865NALX.
Prahl, F.G., E.A. Crecelius, and R. Carpenter. 1984. Polycyclic aromatic hydrocarbons in
Washington coastal sediment: an evaluation of atmospheric and riverine routes of introduction.
Environmental Science and Technology 18:687-693.
Prych, E.A. and J.C. Ebbert. 1986. Quantity and quality of storm runoff from three urban
catchments in Bellevue, Washington. Water-Resources Investigation Report 86-4000. U.S.
Geological Survey, Tacoma, WA. 85 pp.
Puget Sound Air Pollution Control Agency. 1986. 1986 Air quality data summary: counties
of King, Kitsap, Pierce and Snohomish. Seattle, WA. 54 pp.
Richards, R.P. et al. 1987. Pesticides in rainwater in the northeastern United States. Nature
327:129-131.
170
-------
References
Riggin, R.M., W.T. Winberry and N.V. Tilley. 1986. Supplement to EPA/600/4-84/041:
compendium of methods for the determination of toxic organic compounds in ambient air.
EPA/600/4-87-006. U.S. EPA Environmental Monitoring Systems Laboratory, Research
Triangle Park, NC. 3 pp.
Romberg, G.P. et al. 1984. Toxicant pretreatment planning study technical report Cl:
presence, distribution and fate of toxicants in Puget Sound and Lake Washington. Prepared for
Municipality of Metropolitan Seattle, Seattle, Washington. 231 pp. + appendices.
Schmeil, P. 1991. Personal communication. Kaiser Aluminum and Chemical Corporation.
Sehmel, G.A. 1984. Deposition and resuspension. pp. 533-583 In: Darryl Randerson (ed.).
Atmospheric Science and Power Production. DOE/TIC 27601. Prepared for Technical
Information Center, U.S. Department of Energy, Washington, D.C.
Sehmel, G.A. and S.L. Sutler. 1974. Particle deposition rates on a water surface as a function
of particle diameter and air velocity. Journal de Recherches Atmospheriques 8:911-920.
Settle, D.M. and C.C. Patterson. 1982. Magnitudes and sources of precipitation and dry
deposition fluxes of industrial and natural leads to the North Pacific at Enewetak. Journal of
Geophysical Research 87:8857-8869.
Shah J.J., J.G. Watson, J.A. Cooper and J.J. Huntzicker. 1984. Aerosol chemical
composition and light scattering in Portland, Oregon: the role of carbon. Atmospheric
Environment 18:235-240.
Sievering, H. 1975. Dry deposition loading of Lake Michigan by airborne particulate matter.
Water, Air, and Soil Pollution 5:309-318.
Slinn, W.G.N. et al. 1978. Some aspects of the transfer of atmospheric trace constituents
past the air-sea interface. Atmospheric Environment 12:2055-2087.
Soil Conservation Service. 1986. Urban hydrology for small watersheds. Technical Release
55. GPO no. 1983-420-939/1580. U.S. Govt. Printing Office, Washington, D.C. 160 pp.
Stevens, R.K. et al. 1989. Volatile hydrocarbons as mobile source tracer species for receptor
modeling. In: Man and his ecosystem. Eighth World Clean Air Congress 1989, proceedings,
The Hague, The Netherlands, 11-15 September 1989. Volume 5. Elsevier, N.Y.
Strachan, W.M.J. and S.J. Eisenreich. 1988. Mass balancing of toxic chemicals in the Great
Lakes: the role of atmospheric deposition. In: Appendix I from the Workshop on the
Estimation of Atmospheric Loadings of Toxic Chemicals to the Great Lakes Basin, proceedings,
Scarborough, Ontario, October 29-31, 1986. International Joint Commission, Great Lakes
Regional Office, Windsor, Ontario.
Strachan, W.M.J. and H. Huneault. 1979. Polychlorinated biphenyls and organochlorine
pesticides in Great Lakes precipitation. Journal of the Great Lakes Research 5:61-68.
Strayer, D.E. and S.P. Paylou. 1988. Sources of contamination in Puget Sound. NOAA
Estuary-of-the-Month Seminar Series No. 8. Puget Sound: issues, resources, status, and
management. Envirospere Company, Bellevue, WA. 28 pp.
Stumm, W. and J.J. Morgan. 1981. Aquatic Chemistry. John Wiley, N.Y. 780 pp.
171
-------
Evaluation of the Atmospheric Deposition of Toxic Contaminants to Puget Sound
Swackhamer, D.L. and D.E. Armstrong. 1986. Estimation of the atmospheric and
nonatmospheric contributions and losses of polychlorinated biphenyls for Lake Michigan on the
basis of sediment records of remote lakes. Environmental Science and Technology 20:879-
883.
Swackhamer, D.L., B.D. McVeety, and R.A. Kites. 1988. Deposition and evaporation of
polychlorobiphenyl congeners to and from Siskiwit Lake, Isle Royale, Lake Superior.
Environmental Science and Technology 22:664-672.
Tetra Tech. 1988. Commencement Bay Nearshore/Tideflats feasibility study, Volume I. TC-
3218 Public Review Draft. Tetra Tech, Inc., Bellevue, WA.
Thomas, J.F., M. Mukai, and B. Tebbens. 1968. Fate of airborne benzo(a)pyrene.
Environmental Science Technology 2:33-39.
Thrane, K. E. and A. Mikalsen. 1981. High-volume sampling of airborne polycyclic aromatic
hydrocarbons using glass fiber filters and polyurethane foam. Atmospheric Environment
15(6):909-918.
Tuominen, J. et al. 1988. Polynuclear aromatic compounds and genotoxicity in paniculate
and vapor phases of ambient air: effect of traffic, season, and meteorological conditions.
Environmental Science and Technology 22:1228-1234.
Turro, NJ. 1978. Modern molecular photochemistry. Benjamin/Cummings Publ., Menlo
Park, CA. 628 pp.
U.S. Environmental Protection Agency. 1988. Compendium method TO-13. Determination
of benzo(a)pyrene [B(a)P] and other polynuclear aromatic hydrocarbons (PAHs) in ambient
air using gas chromatographic (GC) and high performance liquid chromatographic (HPLC)
analysis. U.S. Environmental Protection Agency, Quality Assurance Division, Environmental
.Monitoring Systems Laboratory, Research Triangle Park, NC.
U.S. Environmental Protection Agency. 1987. EPA guidance on the use of the reference
method for PM10 when using a dichotomous sampler. Section 2.10.1. Washington, D.C.
U.S. Environmental Protection Agency. 1987. Nonpoint source guidance. Office of Water
Regulations and Standards, Washington, D.C. 33 pp.
U.S. Environmental Protection Agency. 1987. Proposed modification of the Great Lakes
Atmospheric Deposition (GLAD) network to include toxic organics. Great Lakes National
Program Office. Chicago, IL. 26 pp.
U.S. Environmental Protection Agency. 1985. Compilation of air pollution emission factors,
4th edition. AP-42. Office of Air Quality Planning and Standards, Research Triangle Park,
NC.
U.S. Environmental Protection Agency. 1986. Test methods for evaluating solid waste:
physical/chemical methods. 3rd ed. SW-846. U.S. Environmental Protection Agency,
Washington, DC.
U.S. Environmental Protection Agency. 1986. Quality criteria for water 1986. EPA
440/5-86-001. U.S. Environmental Protection Agency, Washington, DC.
172
-------
References
U.S. Environmental Protection Agency. 1983. Results of the nationwide urban runoff
program, Vol. I. Final report. PB84-185552. Water Planning Division, Washington, D.C.
Wakeham, S.G. 1977. Hydrocarbon budgets for Lake Washington. Limnology and
Oceanography 22:952-957.
Wakeham, S.G. 1977. Synchronous fluorescence spectroscopy and its application to
indigenous and petroleum-derived hydrocarbons in lacustrine sediments. Environmental Science
and Technology 11:272-276.
Wakeham, S.G. 1977. A characterization of the sources of petroleum hydrocarbons in Lake
Washington. Journal WPCF (Water Pollution Control Federation) 7:1680-1687.
Wallace, G.T., G.L. Hoffman and R.A. Duce. 1977. The influence of organic matter and
atmospheric deposition on the paniculate trace metal concentration of northwest Atlantic surface
seawater. Marine Chemistry 5:143-170.
Washington Department of Ecology. 1988. Ecology acid deposition program: precipitation
monitoring. Water Quality Program, Olympia, WA. 6 pp.
Watson, J.G. et al. 1990. The USEPA/DRI chemical mass balance receptor model, CMS 7.0.
Environmental Software 5(l):38-49.
Webber, D.B. 1986. Dryfall: an important constituent of atmospheric hydrocarbon deposition.
Organic Geochemistry 9:57-62.
Winchester, J.W. 1972. A chemical model for Lake Michigan pollution: considerations on
atmospheric and surface water trace metal inputs, pp. 317-332 In: H.E. Allen and J.R. Kramer
(eds.). Nutrients in Natural Waters. Wiley, N.Y.
Winchester, J.W. and G.D. Nifong. 1971. Water pollution in Lake Michigan by trace
elements from pollution aerosol fallout. Water, Air, and Soil Pollution 1:50-64.
You, F. and T. F. Bidleman. 1984. Influence of volatility on the collection of polycyclic
aromatic hydrocarbon vapors with polyurethane foam. Environmental Science & Technology
18(5):330-333.
Zweidinger, R.B. et al. 1990. Identification of volatile hydrocarbons as mobile source tracers
for fine-particulate organics. Environmental Science & Technology 24:538-542.
173
------- |