£ United States
^& Environmental Protection Agency
LOVE CANAL
EMERGENCY
DECLARATION AREA
HABITABILITY STUDY
FINAL REPORT
VOLUME V
Peer Review Summary-
TRC Responses
TECHNICAL REVIEW COMMITTEE
U.S. Environmental Protection Agency Region
U.S. Department of Health and Human Servio
Centers for Disease Control
New York State Department of Health
New York State Department of Environment
-------
VOLUME V
Peer Review Summary-
TRC Responses
Prepared for
U.S. EPA REGION II
26 Federal Plaza
New York, New York 10278
Prepared by
CH2M HILL SOUTHEAST, Inc.
P.O. Box 4400
Reston, Virginia 22090
Under Contract No. 68-01-7251
and
ICAIR, LIFE SYSTEMS, Inc.
24755 Highpoint Road
Cleveland, Ohio 44122
Under Contract No. 68-01-7331
July 1988
-------
CONTENTS
Section Page
Acknowledgements xi
List of Abbreviations and Acronyms xii
1.0 Summary 1-1
2.0 Introduction 2-1
2.1 Background 2-1
3.0 Peer Review Meeting Summary 3-1
3.1 Air Assessment-Indicator Chemicals 3-1
3.1.1 Consensus Evaluation 3-1
3.1.2 Technical Issues 3-1
3.1.3 Recommendations 3-1
3.2 Soil Assessment-Indicator Chemicals 3-3
3.2.1 Consensus Evaluation 3-3
3.2.2 Technical Issues 3-3
3.2.3 Recommendations 3-3
3.3 Soil Assessment-2,3,7,8-TCDD 3-4
3.3.1 Consensus Evaluation 3-4
3.3.2 Technical Issues 3-4
3.3.3 Recommendations 3-5
3.4 Summary 3-6
4.0 Responses to Comments on the Air
Assessment for Indicator Chemicals 4-1
4.1 Design 4-1
-------
CONTENTS
(Continued)
Section Page
4.1.1 Sampling Strategy 4-1
4.1.2 Heating of Unoccupied Homes 4-2
4.1.3 Rainfall and Ground Water Levels 4-3
4.1.4 Design Modifications 4-4
4.2 Methods 4-4
4.2.1 TAGA Performance Criteria 4-4
4.2.2 TAGA Operation 4-6
4.2.3 Quality Assurance 4-7
4.3 Results 4-8
4.3.1 Sampling Activity 4-8
4.3.2 Analytic Results 4-10
4.3.3 Discussion of Detects 4-10
4.3.4 Variation of Potentially
Influencing Parameters 4-12
4.4 Other 4-20
4.4.1 Detection Limits and Selection
ofLCICs 4-20
4.4.2 Interpretation of the Results 4-21
4.4.3 Additional Information 4-23
5.0 Responses to Comments on the Soil
Assessment for Indicator Chemicals 5-1
5.1 Design 5-1
5.1.1 Statistical Sampling Design 5-1
5.1.2 Selection of Census Tracts 5-4
IV
-------
CONTENTS
(continued)
Section Page
5.2 Methods 5-5
5.2.1 Method Documentation (QAPPs) 5-5
5.2.2 Sample Collection and
Preparation Methods 5-6
5.2.3 Sample Analysis Methods 5-11
5.3 Results 5-12
5.3.1 Consistency of Data Qualifiers 5-12
5.3.2 Blank Sample Data 5-13
5.3.3 Data Completeness 5-14
5.3.4 Actual Versus Planned Sample
Sizes 5-15
5.3.5 Neighborhood Power Calculations 5-16
5.3.6 Magnitudes of the Differences
and Other Percentiles of LCIC
Concentrations 5-16
5.3.7 Analysis of Outlying Values 5-17
5.3.8 Correcting for Chance Outcomes
of the Univariate Tests 5-19
5.3.9 Spatial Analysis 5-20
5.3.10 Additional Information 5-21
6.0 Responses to Comments on the Soil
Assessment for 2,3,7,8-TCDD 6-1
6.1 Design 6-1
6.1.1 Representativeness of Samples
and Rationale for Selection of
Sampling Media 6-1
6.1.2 Assumed Hot Spot Size 6-4
6.1.3 Tilting of the Sampling Grid 6-6
-------
CONTENTS
(continued)
Section Page
6.1.4 Statistical Design 6-6
6.2 Methods 6-6
6.2.1 QAPP Documentation and
Implementation 6-6
6.2.2 Chemistry 6-9
6.2.3 Numbering of Samples 6-10
6.2.4 Data Review 6-10
6.2.5 Audits 6-11
6.3 Results 6-12
6.3.1 Moving of Sampling Points 6-12
6.3.2 Modifications During
Implementation 6-14
6.3.3 Statistical Interpretation 6-16
6.3.4 Interpretation of the Results 6-22
6.3.5 Additional Information/
Discussion Needed 6-24
6.4 Other 6-27
6.4.1 Issues Related to the
Habitability Decision 6-27
6.4.2 Documentation 6-31
REFERENCES R-l
APPENDDC A: Final Summary of Modifications
to TAGA Procedures, April 15,1988, Air
Assessments—Indicator Chemicals A-l
VI
-------
CONTENTS
(continued)
Page
APPENDIX B: Memorandum: Further Replies to
Peer Review Comments, June 13,1988 (Revised
July 7,1988), Air Assessment-Indicator
Chemicals B-l
APPENDIX C: Memorandum: Reply to Peer
Review Comments, May 19,1988, Air Assessment--
Indicator Chemicals C-l
APPENDIX D: Reasons for Denial of Permission
to Sample Love Canal Properties, Love Canal EDA
Habitability Study D-l
APPENDIX E: List of Errata, Air Assessment--
Indicator Chemicals E-l
APPENDIX F: Justification for Nonparametric
Statistical Comparisons, Soil Assessment--
Indicator Chemicals F-l
APPENDIX G: Empirical and Fitted
Distribution of LCICs by EDA Sampling Area
and Retrospective Power Analysis by EDA
Sampling Area and Neighborhood Using
Fitted Distributions, Soil Assessment-Indicator
Chemicals G-l
APPENDIX H: Detection Limits, Soil
Assessment—Indicator Chemicals H-1
APPENDIX I: Chemical Data Quality
Assessment, Soil Assessment--
Indicator Chemicals 1-1
Vll
-------
CONTENTS
(continued)
APPENDIX J: Sampling Area Comparisons
and Additional Sensitivity Analyses, Soil
Assessment-Indicator Chemicals J-l
APPENDIX K: Summary of Sensitivity Results
by EDA Sampling Area, Soil Assessment--
Indicator Chemicals K-1
APPENDIX L: Spatial and Correlation
Structures of LCICS in the EDA, Soil
Assessment-Indicator Chemicals L-l
APPENDDC M: Neighborhood Comparisons and
Redefined EDA Sampling Area Comparisons,
Soil Assessment-Indicator Chemicals M-1
APPENDIX N: List of Errata, Soil Assessment-
Indicator Chemicals N-1
APPENDIX O: Integrated Data Base, Soil
Assessment-Indicator Chemicals O-l
APPENDIX P: Summary of Health Study on
Dioxin Level of Concern, Soil Assessment-
2,3,7,8-TCDD P-l
APPENDIX Q: Summary of Assumptions for
1.0 ppb Level of Concern, Soil Assessment-
2,3,7,8-TCDD Q-l
APPENDIX R: Guide to Love Canal Dioxin
Soil Sampling Study Quality Assurance Project
Plan, Soil Assessment-2,3,7,8-TCDD R-l
vin
-------
CONTENTS
(continued)
Page
APPENDIX S: Method Validation Report,
2,3,7,8-TCDD in Soil and Sediment by Low
Resolution Mass Spectrometry S-l
APPENDIX T: Retrospective Probability of
Detecting the Target-Size Locally Contaminated
Area, Soil Assessment--2,3,7,8-TCDD T-l
APPENDIX U: Letter: Summary of Preliminary
Research on Sampling Point with Results Above
Level of Concern, March 31,1988, Soil Assessment--
2,3,7,8-TCDD U-l
APPENDIX V: Letter: Results of Followup Investi-
gation, July 6,1988, Soil Assessment--
2,3,7,8-TCDD V-l
APPENDIX W: Memorandum: Health Concentra-
tion Based on Results of the Soil Assessment
for 2,3,7,8-TCDD, June 17,1988 W-l
FIGURES
2-1 Habitability Study/EDA Sampling Areas and
Neighborhoods 2-3
2-2 Habitability Study/Location of EDA and Comparison
Areas in New York State 2-4
4-1 Air Assessment-Indicator Chemicals:
Distribution of Homes Sampled, Phases 1-4 4-9
4-2 Overburden Well Locations 4-15
IX
-------
CONTENTS
(continued)
Page
6-1 Soil Assessment--2,3,7,8-TCDD, Design
of Sampling Grid 6-19
6-2 Soil Assessment-2,3,7,8-TCDD, Probability of
Detecting Locally Contaminated Area One-Third
the Size of the Target Area 6-21
TABLES
3-1 Peer Review Panel on Love Canal Habitability Study
Final Reports 3-7
4-1 Water Level Elevations in Selected
Overburden Wells 4-14
4-2 Monthly Precipitation-Historical Record,
Niagara Falls International Airport 4-17
4-3 Mean Monthly Precipitation and Total
Monthly Precipitation 4-19
5-1 Length of Soil Column Obtained by Neighborhood
and Comparison Area, Soil Assessment--
Indicator Chemicals 5-7
5-2 Example Sample Preparation Schedule Provided Daily
to the Sample Preparation Teams, Soil Assessment--
Indicator Chemicals 5-10
6-1 Distance Actual Sampling Points Were Moved from
Grid Nodes, Soil Assessment~2,3,7,8-TCDD 6-13
6-2 Summary of Sample Results by Sampling
Month, Soil Assessment-2,3,7,8-TCDD 6-17
-------
CONTENTS
(continued)
6-3 Summary of Concentrations Measured from
0.5 ppb to 1.0 ppb and Their Associated Matrix
Spike Results, Soil Assessment--2,3,7,8-TCDD 6-25
XI
-------
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the efforts of all organizations and in-
dividuals who participated in the conceptualization, design, and implemen-
tation of this study. Special acknowledgement for their continuing guidance
goes to the U.S. Environmental Protection Agency (EPA) Region n and the
other agencies composing the Technical Review Committee (TRC): the U.S.
Department of Health and Human Services/ Centers for Disease Control
(DHHS/ CDC), the New York State Department of Health (NYSDOH), and
the New York State Department of Environmental Conservation (NYSDEC).
ICAIR, LIFE SYSTEMS, Inc., under separate EPA contract, organized the
peer review effort and facilitated the peer review meeting. ICAIR prepared
Sections 1.0 and 3.0 of this volume. Section 3.0 was concurred with by each
peer review panel member. CH2M HILL prepared responses to the peer
review panel's comments and managed the preparation of Volume V.
CH2M HILL was assisted by its subcontractor firms. The specific technical
contributions of these other organizations are noted in the Acknowledge-
ments sections of Volumes n through IV.
Finally, the participation and dedication of the six expert scientists who
served on the peer review panel is gratefully acknowledged.
Xll
-------
LIST OF ABBREVIATIONS AND
ACRONYMS
A-BHC
ATSDR
BHC
B-BHC
BQC
CB
CC
CLP
CNP
CT
DBB
D-BHC
DCB
DHHS/CDC
DQO
EDA
Alpha-hexachlorocyclohexane
Agency for Toxic Substances and Disease Registry
Hexachlorocyclohexane
Beta-hexachlorocyclohexane
Blind Quality Control
Chlorobenzene
Continuing Calibration
Superfund Contract Laboratory Program
2-chloronaphthalene
Chlorotoluene
1,4-dibromobenzene
Delta-hexachlorocyclohexane
1,2-dichlorobenzene
Department of Health and Human Services/Centers
for Disease Control
Data Quality Objective
Emergency Declaration Area
EMSL-LV/LEMSCO Environmental Monitoring Systems Laboratory-
Las Vegas/Lockhead Engineering and Management
Services Company
EPA
U.S. Environmental Protection Agency
Xlll
-------
FHB
FOIA
g
G-BHC
GC/MS
1C
IQR
IS
IS5
Kg
LCARA
LCIC
m3
MHB
MFC
MS
MS/MSD
NEIC
NYSDEC
NYSDOH
ORD
OTA
Field Handling Blanks
Freedom of Information Act
grams
Garnma-hexachlorocyclohexane
Gas Chromatograph/Mass Spectrometer
Initial Calibration
Interquartile Range
Internal Standard
Internal Standard No. 5 D-10 Pyrene
Kilogram
Love Canal Area Revitalization Agency
Love Canal Indicator Chemical
Cubic Meter
Method/Holding Blanks
Maximum Possible Concentration
Matrix Spike
Matrix Spike/Matrix Spike Duplicate
EPA National Enforcement Investigation Center
New York State Department of Environmental
Conservation
New York State Department of Health
Office of Research and Development
Congressional Office of Technology Assessment
XIV
-------
PC-AT Personal Computer--AT
PCI First Performance Check
PC2 Second Performance Check
PE Performance Evaluation
ppb parts per billion
QA Quality Assurance
QAPP Quality Assurance Project Plan
QC Quality Control
RPD Relative Percent Difference
RRF Relative Response Factor
RSD Relative Standard Deviation
RT Retention Time
SCS Soil Conservation Service
SICP Selected Ion Current Profile
SIM Selected Ion Monitoring
S/N Signal To Noise Ratio
SOP Standard Operating Procedure
SSB Shipping and Storage Blank
TAGA Trace Atmospheric Gas Analyzer
TBBP 2,4,6-tribromobiphenyl
TCB 1,2,4-trichlorobenzene
TeBB 1,2,4,5-tetrabromobenzene
XV
-------
TeCB 1,2,3,4-tetrachlorobenzene
TRC Technical Review Committee
2,3,7,8-TCDD 2,3,7,8-tetrachlorodibenzo-p-dioxin
%V Percent Valley
XVI
-------
Section 1.0
SUMMARY
The Love Canal Emergency Declaration Area; Proposed Habitability
Criteria (NYSDOH and DHHS/CDC, 1986) was developed to assist the
Love Canal TRC, a committee composed of senior-level officials from EPA,
CDC/DHHS, NYSDOH, and NYSDEC, in conducting a study of the
habitability of the Emergency Declaration Area (EDA). The criteria call for
the performance of three environmental studies: (1) an assessment of the
levels of Love Canal Indicator Chemicals (LCIC) in air in the EDA; (2) an
assessment of the levels of LCICs in the soil of the EDA and selected com-
parison areas; and (3) an assessment of the levels of 2,3,7,8-tetrachlorodiben-
zeno-p-dioxin (TCDD) in the soil of the EDA. These studies have been
completed, and the results are presented in Volumes n, IE, and IV, (CH2M
HILL, 1988a, b, and c) respectively, of this series.
Under the direction of the TRC, a peer review panel was convened to review
the implementation of the three environmental studies. This volume reports
on the findings of the peer review and provides the responses to that peer
review. The objectives of the peer review were to determine whether the
studies were properly designed and properly implemented, and whether the
resulting data are reliable and appropriately presented. The peer review panel
consisted of six expert scientists: an air chemist, a soil scientist, a civil en-
gineer, an epidemiologist, and two statisticians.
Each peer reviewer prepared preliminary written comments on Volumes n
through IV. The preliminary comments were supplied to the other peer
reviewers, to the TRC, and to the public. Responses to these comments were
submitted in the form of draft technical memoranda to the peer reviewers and
the public prior to the peer review meeting. (The contents of the technical
memoranda are included in this volume.)
The peer review meeting was held in Niagara Falls, New York, on June 20
and 21, 1988. The peer reviewers discussed their comments and concerns
with each other, the TRC, and with the scientists who performed the studies.
Members of the public were also given the opportunity to ask questions and
l-l
-------
make comments. Based on these discussions, the peer review panel reached
the following conclusions:
• Each of the three environmental studies was well planned and well
executed.
• There was a high degree of data quality control, and the data from
each study are appropriate for making a determination on
habitability.
• The statistical analyses of the data from the soil assessment for indi-
cator chemicals, as presented in Volume m, are appropriate for
making a determination on habitability.
The peer reviewers identified areas in which additional information or
clarification would be useful and recommended that these topics be ad-
dressed in this final report. The additional information and clarification is
contained in Sections 4.0 through 6.0 of this volume. In addition, the peer
reviewers and the TRC requested additional supplemental statistical analyses
(i. e., analyses that went beyond the study plan) that they felt would be use-
ful in the habitability decision. The bulk of the appendices to this volume
are dedicated to these supplemental analyses.
Upon acceptance of this volume, the TRC will transmit the five-volume
habitability report to the Commissioner of Health for the State of New York
for use in a determination of the habitability of the EDA.
1-2
-------
Section 2.0
INTRODUCTION
This is Volume V of a five-volume series. Volume I (ICAIR and
CH2M HILL, 1988) provides an introduction to the Love Canal EDA
Habitability Study and documents the decision-making and history during
its development; Volume n reports on the air assessment for indicator chemi-
cals; Volume m reports on the soil assessment for indicator chemicals; and
Volume IV reports on the soil assessment for 2,3,7,8-TCDD. Volume V
summarizes both the peer review comments of Volumes n, ffl, and IV and
the responses to those comments.
This section gives a brief background of the EDA and the habitability study.
Section 3.0 summarizes the discussion, conclusions, and recommendations
of the peer review panel, which were reviewed and approved by each panel
member. Sections 4.0,5.0, and 6.0 provide the responses to the peer review
comments concerning the air assessment for indicator chemicals, the soil as-
sessment for indicator chemicals, and the soil assessment for 2,3,7,8-TCDD,
respectively.
Several appendices contain more detailed technical information and analyses
to support the responses given in Sections 4.0,5.0, and 6.0.
2.1 BACKGROUND
The EDA is a residential area surrounding the Love Canal chemical waste
site in Niagara Falls, New York. The majority of the EDA residents vacated
the area because of concern over the potential health effects of chemicals
from the site. The site has undergone extensive remediation. (An extensive
history of the Love Canal site and the surrounding EDA is presented in
Volume I.)
The Love Canal TRC, composed of representatives from the EPA,
DHHS/CDC, NYSDOH, and NYSDEC, was established to coordinate and
oversee the remedial program and habitability study at the Love Canal. The
habitability study was designed to enable the Commissioner of Health for
2-1
-------
the State of New York to determine whether the EDA is now suitable for
human habitation.
The following studies and activities have taken place:
• Development of the habitability criteria
• Peer review of the habitability criteria
• Pilot study of the proposed soil and air assessments for indicator
chemicals
• Peer review of the pilot study
• Soil and air assessments for indicator chemicals and a soil assess-
ment for 2,3,7,8-TCDD
• Peer review of the assessments
This volume documents and responds to the peer review of the three environ-
mental assessments.
Figures 2-1 and 2-2 are included for the convenience of the reader.
Figure 2-1 shows the EDA sampling area and neighborhood boundaries, and
Figure 2-2 shows the relative location of the EDA and the comparison areas
used in the study.
2-2
-------
*&i-
_»*Wr7 A«
....n*ix_
aMOCTAerr
acHoo.a(rf
•»
1
3^
'5
*••
""%*
^rV-
'Si
*!
ij
^—^-^^.—^•W
^^ASaaQHfL araiiK-
1 - ; i
: t
i :W -
: ? 6 i
i | f
i '. ; i
% ViSTai
i :
oocvi^^g
^^ TT
4
SCALE: r-eeo1
LEGEND
"7
Figure 2-1 - -^
HABITABILITY STUDY
EDA SAMPUNG AREAS AND NEIGHBORHOODS
13 EMERGENCY DECLARATION AREA (EDA) NEIGHBORHOOD BOUNDARY
FENCE LINE AROUND LOVE CANAL REMEDIATION SITE
EDA SAMPLING AREA BOUNDARY
T02NO STREET
SOURCE: NEIGHBORHOOD BOUNDARIES ADAPTED FROM THE PROPOSED
-===•• HABITABILITY CRITERIA DOCUMENT (NYSDOH AND DHHS/CDC. 1986).
-------
\\\ NOBTH
lr\TONAWANDA
Scale: 1" a 3 Mil**
(Approx.)
msi
1 0 1
Figure 2-2
HABIT ABILITY STUDY
LOCATION OF EDA AND COMPARISON AREAS
IN NEW YORK STATE
-------
Section 3.0
PEER REVIEW MEETING SUMMARY
The objective of this chapter is to summarize the final conclusions and recom-
mendations of the peer review panel. This summary has been reviewed and
approved by each panel member. The panel's findings are presented for the
air assessment for indicator chemicals, the soil assessment for indicator
chemicals, and the soil assessment for 2,3,7,8-TCDD. An overall evaluation
of the suitability of the data for a decision on habitability is also presented.
3.1 AIR ASSESSMENT—INDICATOR CHEMICALS
3.1.1 Consensus Evaluation
The peer review panel unanimously agreed that the air assessment for LCICs
(CH2M HILL, 1988a) was adequately designed and implemented and that
the data obtained are reliable.
3.1.2 Technical Issues
Technical questions or concerns that were raised by the peer reviewers prior
to the peer review meeting were summarized and responded to by CH2M
HILL in technical memoranda provided to the peer reviewers. Comment
summaries and responses are presented in Section 4.0 of this volume.
3.1.3 Recommendations
The peer review panel identified several areas where additional information
or clarification would be useful and recommended that these be addressed.
These recommendations are summarized below.
• Provide a more thorough description of how the detection limit for
the Trace Atmospheric Gas Analyzer (TAGA) was calculated.
• Provide information on how the sensitivity of the TAGA for the
LCICs depended on humidity.
3-1
-------
• Provide more detailed information on rainfall levels during the sam-
pling period, with special reference to the levels of precipitation that
occurred during 1975 and 1976, when chemical migration from the
Love Canal was most obvious.
• Provide a more thorough description of the significance and the limi-
tations of the data. This would include discussion of the following
topics:
- Concentrations of contaminants in air are highly variable in time
and space, and even thorough studies such as this one cannot
provide complete assurance that all occasions of elevated con-
taminant levels have been detected.
- Heating a home creates a pressure gradient that tends to increase
infiltration of soil gas into the house. Since most of the EDA
residences were unoccupied and it was not practical to heat them,
this may have decreased the ability to detect LCICs in indoor air.
- The finding that one home out of 562 in the EDA had
measurable levels of air LCICs should be viewed in the perspec-
tive that, in the pilot study, 1 home out of 31 in the control area
also had measurable levels.
- It is not inconsistent to observe measurable levels of LCICs in
soil but not in air.
In addition, even though the air study identified only one home in the EDA
with measurable levels of indicator chemicals, if rehabitation occurs and if
individuals continue to have concerns, it may be desirable to make informa-
tion available on home ventilation devices that have been used elsewhere for
the control of radon gas. These devices might also be used for the control of
possible soil-gas accumulation in homes.
3-2
-------
3.2 SOIL ASSESSMENT—INDICATOR CHEMICALS
3.2.1 Consensus Evaluation
The peer review panel unanimously agreed that the assessment for soil in-
dicator chemicals (CH2M HILL, 19885) was particularly well planned and
well performed. In addition, the panel unanimously agreed that the statisti-
cal analyses of the data were appropriate and correct, and are suitable for
making a determination on habitability.
3.2.2 Technical Issues
Technical questions or concerns that were raised by the peer reviewers prior
to the peer review meeting were summarized and responded to by
CH2M HILL in technical memoranda provided to the peer reviewers. These
summaries and responses to comments are presented in Section 5.0 of this
volume.
3.2.3 Recommendations
The peer review panel identified several areas where additional information
or clarification would be useful, and recommended that these be addressed.
These recommendations are summarized below.
• Although the statistical analyses of the data reported in Volume ffl
are the most appropriate and are adequate for a habitability decision,
several additional statistical analyses or summaries may be useful.
These include:
- Analyses fitting parametric distributions (e.g., lognormal mix-
ture) and comparison of various percentile values (e.g., 50 per-
cent, 90 percent, 95 percent)
- Additional explanation regarding the purpose and meaning of the
various analyses of robustness (sensitivity)
- Summarization of the results of the existing statistical analyses,
organized by sampling area
3-3
-------
- Additional sensitivity or robustness analyses for 1,2,4-
trichlorobenzene (TCB), treating all values below 2.0 parts per
billion (ppb) as non-detects
• Provide a more detailed explanation of the procedure used to select
Census Tracts 221 and 225 as the comparison areas in Niagara
Falls.
• Provide additional description and explanation regarding the ab-
sence of blank correction in the data reported, with special attention
to the distinction between laboratory background and actual soil
sample data.
• Provide additional explanation as to why some samples were
"flagged," and the steps taken to ensure that this did not introduce a
bias.
• Review available hydrogeological and meteorological data to deter-
mine whether any likely transport mechanism from the Love Canal
can be identified to account for the elevated levels of soil LCICs ob-
served in Sampling Areas 1,2, and 3. Identification of such a mech-
anism would support the hypothesis that the observed differences
are due to the Love Canal.
3.3 SOIL ASSESSMENT—2,3,7,8-TCDD
3.3.1 Consensus Evaluation
The peer review panel unanimously agreed that the soil assessment for
2,3,7,8-TCDD (CH2M HILL, 1988c) was adequately designed and imple-
mented and that the data are reliable.
33.2 Technical Issues
Technical questions or concerns that were raised by the peer reviewers prior
to the peer review meeting were summarized and responded to by
CH2M HILL in technical memoranda provided to the peer reviewers. These
3-4
-------
comment summaries and responses are presented in Section 6.0 of this
volume.
3.3.3 Recommendations
The peer review panel identified several areas where additional information
or clarification would be useful, and recommended that these be addressed.
These recommendations are summarized below.
• Provide an overview of the data quality assurance activities as-
sociated with the soil dioxin study, with appropriate cross-refer-
ences to the relevant portions of the Quality Assurance Project Plan
(QAPP) (CH2M HILL, 1986 and 1987e).
• Provide the following additional information that might be useful in
interpreting the significance and the limitations imposed by having
sampled surface soil only.
- The possible occurrence of "hot spots" other than the one de-
tected, and the limitations imposed by having sampled surface
soil only
- The chance of exposure to dioxin-contaminated soil as a function
of the area and depth of the contaminated soil
- The chronic exposure conditions assumed by the CDC in the
derivation of the 1.0-ppb level of concern, as these relate to the
types of exposures that might be expected to occur in the EDA
- The relative magnitude of the potential health risks associated
with exposures to small and/or subsurface areas of dioxin con-
tamination, expressed in language readily understandable by the
average citizen
• Provide a summary of dioxin soil measurements as a function of the
time of year when samples were taken, to ensure that changes in the
sampling schedule did not introduce a significant source of varia-
tion.
3-5
-------
3.4 SUMMARY
A peer review panel of six expert scientists reviewed Volumes n, HI, and IV
of the Love Canal Habitability Study Final Reports, along with associated
supporting documentation. The panel members are listed in Table 3-1. The
panel unanimously agreed that each of the component parts of the habitability
study was well planned, well executed, and had a high level of data quality
assurance, and that the resulting data are of high quality and are appropriate
for making a determination on habitability. In addition, the statistical
analyses of indicator chemical levels in soil as reported in Volume ffl and
supporting documents are appropriate for the purposes of making a
habitability decision. The panel identified a number of areas where addition-
al information or clarification would be useful, and recommended that these
be addressed.
3-6
-------
Table 3-1
PEER REVIEW PANEL ON LOVE CANAL
HABIT ABILITY STUDY FINAL REPORTS
Dr. Janick F. Artiola
Assistant Research Scientist
Department of Soil and Water Science
University of Arizona
Tuscon, Arizona
Dr. Randall Charbeneau
Associate Professor of Civil Engineering
University of Texas
Austin, Texas
Dr. Joan Daisey
Staff Scientist in Air Chemistry
Lawrence Berkeley Laboratory
Berkeley, California
Dr. David Schoenfeld
Director of the Biostatistics Center
Massachusetts General Hospital and
Harvard University
Boston, Massachusetts
Dr. Carl Shy
Professor of Epidemiology
University of North Carolina
Chapel Hill, North Carolina
Dr. Michael R. Stoline
Professor of Statistics
Western Michigan University
Kalamazoo, Michigan
3-7
-------
Section 4.0
RESPONSES TO COMMENTS ON THE AIR
ASSESSMENT FOR INDICATOR CHEMICALS
This section briefly describes the peer review comments on Volume n, Air
Assessment-Indicator Chemicals (CH2M HILT., 1988a), and it gives the
responses to those comments. Both written preliminary peer review com-
ments and verbal peer review comments made during the peer review meet-
ing are addressed in this section. The peer review comments have been
organized under the categories of "Design," "Methods," "Results," and
"Other." Each comment is paraphrased, the peer reviewer making the com-
ment is referenced, and the response to the comment is given. Where lengthy
responses are required, the response is summarized in this section, while the
full response is provided in an appendix.
4.1 DESIGN
4.1.1 Sampling Strategy
COMMENT: Concentrations of contaminants in air are highly variable, and
even thorough studies such as this one cannot provide com-
plete assurance that all potential occasions of detectable con-
taminant levels have been identified. (Dr. Joan Daisey-Peer
Review Meeting Comments, June 20,1988.)
RESPONSE: The air assessment was designed to provide the most thorough
investigation of the presence of LCICs in ambient and indoor
air as was possible, given time and other practical constraints.
Sampling phases during a 5-month study period and sampling
times staggered within each phase were incorporated as part
of the design to best assess variation of diurnal and seasonal
parameters that could possibly affect the presence of LCICs
in homes.
It was not practical to sample under all conditions that may
potentially cause fluctuations in LCIC concentrations. Be-
cause concentrations of contaminants in air are variable, this
4-1
-------
study, as would any study, falls short of offering complete as-
surance that all potential occasions of detectable contaminant
levels have been identified. However, the study was specifi-
cally designed to include a wide variety of conditions that
could possibly influence the concentration of air LCICs in the
homes of the EDA.
COMMENT: Clarify how sampling was conducted during a 24-hour period
if the instrument could be operated only 15 hours per day.
(Dr. Joan Daisey-Written Preliminary Comments, p. 3.)
RESPONSE: Daily sampling was performed in shirts averaging ap-
proximately 12 hours. The starting times of these shifts were
varied so that sampling of unoccupied homes was performed
throughout all hours of the day and night during a given phase
to allow observation of any detectable diurnal variation of
LCICs. Sampling of occupied homes, however, was con-
ducted only between the hours of 7:00 a.m. and 9:00 p.m.
Table 5-1 on page 5-7 of the air assessment QAPP
(CH2M HILL, 1987a) depicts an example of this sampling
strategy.
4.1.2 Heating of Unoccupied Homes
COMMENTS: Document whether heat was on or off in unoccupied homes.
If the unoccupied homes were not heated, potential soil-gas
transport into the homes could be missed. (Dr. Joan Daisey-
- Written Preliminary Comments, p. 4.)
Heating a home creates a pressure gradient that tends to in-
crease infiltration of soil gas into the house. Since most of
the EDA residences were unoccupied and it was not practical
to heat them, this may have decreased the ability to detect
LCICs in indoor air. (Dr. Joan Daisey-Peer Review Meet-
ing Comments, June 20,1988.)
RESPONSE: Heating of unoccupied homes during the sampling effort was
not practical because of the sampling time frame and the ex-
tent of disrepair of the heating devices within many of the
4-2
-------
homes. However, no evidence of pressure-driven flow in-
fluencing LCIC concentrations was indicated during sam-
pling of heated, occupied homes. During Phase 4~the
coldest sampling period--23 occupied heated homes dis-
tributed throughout the EDA were sampled, and no LCICs
were detected.
4.1.3 Rainfall and Ground Water Levels
COMMENTS: Discuss how varying rainfall and ground water levels are in-
corporated into the study design. (Dr. Michael Stoline-Writ-
ten Preliminary Comments, p. 9.)
The design makes it difficult to assess the influence of rain-
fall and/or ground water levels on LCIC levels. Also, it is dif-
ficult to assess the influences of rainfall and ground water
levels outside the range encountered during sampling, within
the context of this design. (Dr. Michael Stoline-Written
Preliminary Comments, p. 2.)
No evidence is presented to suggest that the possible in-
fluence of ground water levels is properly accounted for or
tested. (Dr. Randall Charbeneau—Written Preliminary Com-
ments, p. 5.)
RESPONSE: Representative meteorological parameters, including hourly
wind and precipitation, are recorded at the nearby Niagara
Falls International Airport. Ground water levels near the
EDA are recorded by the NYSDEC. The influence of these
parameters on the presence of LCICs in the EDA air is not
known. Rainfall and ground water levels were not specifical-
ly assessed in Volume n because LCICs were virtually un-
detected throughout the 5-month assessment. Had there been
a higher frequency of LCICs detected during the assessment,
an evaluation of these and other parameters would have been
conducted. (In response to peer reviewer comments, ground
water and rainfall level variability during the study was com-
pared to historical variability—see Section 4.3.4 below.)
4-3
-------
4.1.4 Design Modifications
COMMENT: Discuss any other changes or modifications that may have
been implemented during the study. (Dr. Janick Artiola-
Written Preliminary Comments, p. 3.)
RESPONSE: Changes and modifications implemented during the study in-
clude modifications to TAGA analytical procedures, changes
in the phase-specific documentation, clarification of am-
biguities, correction of errors in original TAGA operating and
data reporting criteria, and additional documentation and
TAGA operational procedures. Refer to Appendix A con-
taining the memorandum from the EPA Environmental
Response Branch, dated April 15,1988, for additional details.
4.2 METHODS
4.2.1 TAGA Performance Criteria
COMMENTS: Origins of TAGA performance criteria (items 1 through 6 in
Table 4-1 of the QAPP, CH2M HILL, 1987a) are not proper-
ly referenced or justified. (Dr. Janick Artiola—Written
Preliminary Comments, p. 2.)
Provide a more thorough description of how the detection
limit for the TAGA was calculated. (Dr. David Schoenfeld-
-Peer Review Meeting Comments, June 20,1988.)
The QAPP (CH2M HILL, 1987a) lacks a calculation of the
actual detection limit as opposed to the nominal detection
limit. (Dr. David Schoenfeld—Written Preliminary Com-
ments, pp. 2-5.)
A perspective is needed on the selected detection limit objec-
tive of 4.0 ppb. Is this concentration lower than that expected
in ambient air if Love Canal were the source of the chemicals
(i.e., is the detection limit objective of 4.0 ppb appropriate
with respect to the overall study goal)? Were the levels
4-4
-------
detected prior to remediation significantly higher than the ob-
jective for detection limits (i.e., was the selection of the detec-
tion limit objective based on historical data)? Is there any
basis for human health concern if levels of LCICs below the
detection limits were to exist in the EDA (i.e., in retrospect,
could there be a health risk even though LCICs were not-for
the most part-detected)? (Dr. Carl Shy-Written Preliminary
Comments, p. 6.)
Provide information on how the sensitivity of the TAG A for
the LCICs depended on humidity. (Dr. Ed Horn
(NYSDOH)-Peer Review Meeting Comments, June 20,
1988.)
In Appendix A of the QAPP (CH2M HILL, 1987a) a factor
of two is missing from equation 2 in Section 4.3.3, p. 4-12
and EC is not defined on pp. vii-viii. (Dr. Joan Daisey-Writ-
ten Preliminary Comments, p. 4.)
RESPONSE: TAGA performance criteria were based on achievable instru-
ment limits and were derived from several months of method
development. The detection limit issue was addressed by two
different approaches: statistical models and review of detec-
tion limit verifications and instrument calibrations.
Two different statistical models were used to attempt to
predict actual detection limits with known confidence inter-
vals. Because of the TAGA's analytic complexities and the
need for these statistical models to make assumptions
simplifying the analytic procedure, the models were unable
to accurately predict reasonable detection limits.
A verification of detection limits was performed on a daily
basis during the assessment. A total of 44 verifications were
performed and in all of them, the LCICs were successfully
detected. The TAGA was calibrated using known LCIC con-
centrations of approximately 2.0 ppb and 4.0 ppb prior to
sampling each of the 562 different homes in the EDA. For
all but a few of those known concentrations very near the
nominal detection limit (<0.5 ppb), these concentrations met
all of the detection limit criteria.
4-5
-------
In controlled experiments performed at Edison, the maximum
decrease in TAG A sensitivity under maximum shifts of rela-
tive humidity (dry zero air to near-saturation) never exceeded
a factor of two. The TAGA's sensitivity was always highest
with zero air (no humidity) and suffered the largest relative
drop when the sample air was initially humidified. Therefore,
the effects of humidity changes on the detection limits for the
EDA houses should always be less than a factor of two be-
cause there must always be some humidity in the ambient and
indoor air.
For a detailed discussion of these issues and those raised by
Dr. Ed Horn of the NYSDOH, refer to Appendix B contain-
ing the memorandum from the EPA Environmental Response
Branch, dated June 13,1988 (Revised July 7,1988).
4.2.2 TAGA Operation
COMMENTS:Was the TAGA logbook audited? (Dr. Joan Daisey-Written
Preliminary Comments, p. 4.)
It is hard to discern the instrument startup and end-of-day
calibration procedures from those implemented prior to sam-
pling each house. (Dr. Janick Artiola—Written Preliminary
Comments, p. 2.)
RESPONSE: Auditing of the TAGA logbook was performed by personnel
from Northrop Services under contract with EPA's Office of
Research and Development (ORD). Instrument startup and
end-of-day procedures have been more clearly defined. Refer
to Appendix B, which contains the memorandum from the
EPA Environmental Response Branch, dated June 13,1988
(Revised July 7,1988).
COMMENTS: What air was used for calibration? If outdoor air was used
and LCICs were present, then aren't IQ and the lower limit of
detection higher than reported? (Dr. Joan Daisey-Written
Preliminary Comments, p. 4.)
4-6
-------
The procedure used for calibrating the TAGA assumes that
the ambient air is uncontaminated. If the ambient air was con-
taminated, another method was used. This procedure was to
be a revision to the QAPP (CH2M HILL, 1987a), but I did
not find it in the revisions. (Dr. David Schoenfeld-Written
Preliminary Comments, p. 5.)
RESPONSE: The TAGA is calibrated using ambient air. If ambient air were
contaminated, reported detection limits would be biased high
and corrective measures would be taken. Refer to Appen-
dices B and C, which contain the memoranda from the EPA
Environmental Response Branch, dated May 19 and June 13,
1988, for additional details.
COMMENT: Discuss that the sampling line was not heated during Phases 2,
3, and 4 of the sampling. (Dr. Joan Daisey—Written Prelimi-
nary Comments, p. 5.)
RESPONSE: Efficient transport of sample air through an unheated transfer
line was demonstrated during the continuing method
development that was conducted between Phases 1 and 2 of
the study. Therefore, the more flexible and lighter-weight un-
heated line was used in Phases 2, 3, and 4. Refer to Appen-
dix A, which contains the memorandum from the EPA
Environmental Response Branch, dated April 15, 1988, for
additional information.
4.2.3 Quality Assurance
COMMENT: Difficulty with the first blind canisters should be discussed,
specifically the time frame in which this occurred and whether
it compromised any of the actual sampling. (Dr. Joan Daisey-
-Written Preliminary Comments, p. 1.)
RESPONSE: Because non-field-proven state-of-the-art methods lead to in-
itial uncertainties in the concentration of LCICs in perfor-
mance evaluation, 6-liter-canister, standard gas cylinders
were substituted for blind canisters in the first phases. Exter-
nal checks of the TAGA accuracy were performed by the
4-7
-------
EPA's ORD contractor, Northrop Services, with one blind
16-liter canister audit sample per phase. During all four
phases, TAGA results were within the accuracy criteria. Data
quality was not compromised as a result of this substitution.
Refer to Appendix A, which contains the memorandum from
the EPA Environmental Response Branch, dated April 15,
1988, for additional information.
COMMENT: Clarify how it was determined that there were no interferences
for the selected ions. (Dr. Joan Daisey-Written Preliminary
Comments, p. 3.)
RESPONSE: Method development involved the search for interferent com-
pounds; none was found. In addition, interferences, if they
occurred, would bias sampling results high and have the
potential to produce false positives. Only one LCIC detect
was observed during the study, and this detect was confirmed
using a canister sample taken onsite. Refer to Appendix B,
which contains the memorandum from the EPA Environmen-
tal Response Branch, dated June 13,1988, for additional in-
formation.
4.3 RESULTS
4.3.1 Sampling Activity
COMMENT: An additional figure should be added showing the locations
of each of the homes that could not be sampled. (Dr. Janick
Artiola-Written Preliminary Comments, p. 3.)
RESPONSE: Figure 4-1 is a composite of Figures 6-1 through 6-4 of
Volume II; this figure identifies all homes sampled and not
sampled during the study.
COMMENT: What were the reasons for homeowner refusal? (Dr. Michael
Stoline-Written Preliminary Comments, pp. 1 and 6.)
4-8
-------
Figure 4-1
AIR ASSESSMENT-INDICATOR CHEMICALS
DISTRIBUTION OF HOMES SAMPLED
PHASES 1-4
SCALE: f.6501
LEGEND
—— EDA NEIGHBORHOOD BOUNDARIES (NUMBERED VO)
— • — FENCE LINE AROUND LOVE CANAL REMEDIATION SITE
• HOMES SAMPLED APPEAR SHADED ON FIGURE
SOURCE; NEIGHBORHOOD BOUNDARIES ADAPTED FROM THE PROPOSED
HABITABILITY CRITERIA DOCUMENT (NYSDOH AND DHHS/CDC. «6).
-------
RESPONSE: The reasons for homeowner refusal for each of the environ-
mental assessment programs have been compiled by the
NYSDOH and are contained in Appendix D.
4.3.2 Analytic Results
COMMENTS: I suggest that 95 percent confidence intervals be given on the
detection limits. (Dr. Joan Daisey-Written Preliminary
Comments, p. 6.)
The range of quantifiable detection limits is needed. (Dr.
Janick Artiola-Written Preliminary Comments, pp. 6-7.)
RESPONSE: Detection limits computed for each house analyzed have been
summarized, and a frequency distribution of those limits has
been developed for each sampling phase. For all phases,
95 percent of the detection limits were below 1.3 ppb for
chlorobenzene (CB) and 1.7 ppb for chlorotoluene (CT).
Refer to Appendices B and C, which contain the memoranda
from the EPA Environmental Response Branch, dated
May 19andJune 13,1988, for more information on the detec-
tion limits.
4.3.3 Discussion of Detects
COMMENT: Wind direction before, during, and after the sampling period
should be checked for the CT detect situation. The LCIC
might have entered the house by infiltration if the wind were
blowing from a source, and if the wind changed by the time
the sampling team arrived, the outdoor air would not show
detectable levels. However, because the air exchange rates
of houses are slow, typically 0.5 to 1.0 per hour, the house
would still contain the accumulated LCIC. (Dr. Joan Daisey-
-Written Preliminary Comments, p. 7.)
RESPONSE: Screening and investigative sampling for the home where CT
was detected took place over a 6-1/2-hour period, 7:30 a.m.
to 2:00 p.m. Based on wind data collected at the Niagara Falls
International Airport, a shift in wind direction (west-south-
4-10
-------
west to north-northeast) occurred at 9:30 a.m. Before this
house was sampled, winds were steady from the west-south-
west since 5:00 p.m. the previous day. In addition, ambient
air was sampled 1 hour and 1/2 hour before the detect at a
residence on the same street approximately 500 feet from the
home in which CT was detected Ambient air near the home
with the detect was sampled just before the sampling team
entered for indoor sampling. CT was detected on the main
floor of the home throughout this period. While ambient air
was sampled often during this time frame, LCICs were not
detected in any ambient air samples before or during the in-
vestigation.
CT was detected again at the same residence 7 days later,
while it had not been detected in any other home in the inter-
im. The probability of transient plume infiltration only in this
home during the assessment period is remote.
During the first sampling of this home in Phase 2, CT was
detected on the home's main floor only. During the second
sampling of this home, 7 days later in Phase 2, CT was
detected both on the main floor and in the basement; levels
of CT detected in the basement, however, were consistently
lower than those detected on the main floor. Detection limits
during this second sampling of Phase 2 were approximately
one-half those during the initial sampling.
COMMENTS: The source of CT could have been subsurface contamination,
contrary to what is implied in Volume n. A possible source
is contaminated Love Canal fill soil. The case made for an
outside air emission source is questionable based on the fact
that no CT was detected in ambient air; on the other hand, CT
emissions from industrial processes are not uncommon.
(Dr. Janick Artiola—Written Preliminary Comments, pp. 4-
5.)
Soil-gas transfer is a possibility. (Dr. Joan Daisey—Written
Preliminary Comments, p. 6.)
4-11
-------
RESPONSE: Fill dirt from the Love Canal is a possible mechanism of con-
tamination. However, if fill dirt was present at the residence
at which CT was detected, it had been there for many years.
CT was detected only during Phase 2 of this assessment, al-
though the same home was sampled again in Phases 3 and 4.
In addition, detected levels of CT were lower during the
second sampling of the home in Phase 2.
COMMENT: Is the detection of one LCIC in one home an isolated case of
contamination or a random event? (Dr.JanickArtiola-Writ-
ten Preliminary Comments, p. 6.)
RESPONSE: The fact that CT was detected twice in the same home 7 days
apart and that no LCICs were detected in any other home or
in ambient air during the assessment suggests that this detec-
tion is an isolated case rather than a random event.
4.3.4 Variation of Potentially Influencing Parameters
COMMENT: It might be appropriate to look at the ground water monitor-
ing data to see if a range of ground water levels was en-
countered during the sampling periods. (Dr. Randall
Charbeneau--Written Preliminary Comments, p. 5.)
RESPONSE: To evaluate the influence of local ground water levels on the
results of the air monitoring study, data from relatively shal-
low overburden wells located outside the capped portion of
the EDA were reviewed. Data from these wells best describe
the local water-table elevation in the area that is not directly
influenced by the hydraulic properties of the cap.
Only a limited amount of data was readily accessible. Data
on near-surface ground water before 1980 are particularly
limited, with less than 20 overburden wells investigated up to
that time. In contrast, several studies have been conducted
since 1980 that, combined, have involved approximately
1,000 samples of ground water, and a correspondingly large
number of water-level measurements.
4-12
-------
No data have been found that constitute a long-term histori-
cal record of local near-surface ground water at a single loca-
tion. Consequently, it is not possible to identify continuous
trends in water levels beginning from the time when the waste
problem was first identified. Over the last few years,
however, the NYSDEC has participated in water-level
monitoring programs for some of the wells within the EDA.
As a result, local water-level data are available for the period
in which the air monitoring study was conducted, as well as
for various times before the study.
Ground water-level data from selected overburden wells
(1151D, 1154D, and 1183D) are shown in Table 4-1. Wells
115 ID and 1154D are located at the southern end of the Canal
near the LaSalle Expressway, and well 1183D is situated at
the northern end, off Colvin Boulevard (Figure 4-2). The
data indicate that very little variation occurred in water-table
elevations during the time that the air study was conducted
(mid-July to mid-December 1987). In fact, ground water
levels in these wells have typically varied less than 1 foot
since the wells were installed in 1986.
Information on near-surface ground water before 1986 must
come from other overburden wells. Well 3151 is near wells
115 ID and 1154D, and well 5115 is close to well 1183D
(Figure 4-2). In 1980, the ground water elevation in
wells 3151 and 5115 was 566.75 feet and 563.00 feet above
mean sea level, respectively. These values are approximate-
ly 3 feet lower than the lowest water level reported for the
wells listed in Table 4-1.
COMMENTS: The study was conducted during a 5-month period and may
not be fully representative of the full range of meteorological
conditions in this area. Appropriate qualifying statements
should be included in the report. (Dr. Joan Daisey-Written
Preliminary Comments, p. 6.)
Uncertainties remain with respect to the effects of seasonal
variations (i.e., meteorological, temperature gradient indoor
to outdoor, others). (Dr. Joan Daisey-Written Preliminary
Comments, p. 8.)
4-13
-------
Table 4-1
WATER LEVEL ELEVATIONS IN SELECTED OVERBURDEN WELLS
(feet above mean sea level)
Well No.
Date
1986 4-29
5-13
6-4
7-30
8-26
10-1
10-30
12-5
12-29
1987 1-29
2-24
3-25
4-30
5-28
6-26
7-30
8-27
9-23
10-29
12-22
1151D
—
—
—
—
—
—
—
570.72
571.14
570.21
570.12
569.93
570.16
569.66
569.56
569.36
569.26
570.80
570.36
570.96
1154D
569.72
569.12
570.05
570.31
568.36
569.11
570.41
570.11
—
—
—
569.73
569.60
569.31
—
568.78
568.85
568.91
569.44
—
1183D
567.52
567.56
568.27
568.09
566.67
567.19
567.00
566.93
567.01
567.55
567.00
567.57
567.45
569.46
567.46
567.52
567.47
567.51
567.52
567.45
1988
3-15
570.54
569.52
567.45
-------
Figure 4-2
OVERBURDEN WELL LOCATIONS
SCALE: r-CKf
LEGEND
—— EDA NEIGHBORHOOD BOUNDARIES (NUMBERED
— - — FENCE LINE AROUND LOVE CANAL REMEDIATION SITE
• OVERBURDEN WELLS
SOURCE: NEIGHBORHOOD BOUNDARIES ADAPTED FROM THE PROPOSED
HABITABILITY CRITERIA DOCUMENT (NYSDOH AND DHHS/CDC. W86).
1
-------
What inferences can be made regarding the Canal chemicals
in EDA air if a period of prolonged heavy precipitation oc-
curs? (Dr. Michael Stoline—Written Preliminary Comments,
P- 12.)
Provide more detailed information on rainfall levels during
the sampling period, with special reference to the levels of
precipitation during 1975 and 1976, when chemical migra-
tion from the Love Canal was most obvious. (Dr. Michael
Stoline-Peer Review Meeting Comments, June 20,1988.)
RESPONSE: Sampling of indoor and ambient air was performed during
52 days between July 28 and December 13, 1987. The
meteorological conditions that occurred during the July-to-
December sampling period do not fully reflect the maximum
historical annual extremes, particularly for such parameters
as temperature and precipitation. However, these conditions
were typical of the annual means and extremes expected for
this period.
The maximum and minimum temperatures recorded during
sampling were 87°F and 20°F, respectively. Both heated and
unheated homes (i.e., occupied and unoccupied homes) were
sampled under conditions approaching these extremes. The
maximum and minimum temperatures occurring during this
5-month period were 91°F and 15°F, respectively.
Historical extreme temperatures were not expected during the
study period. However, a meaningful comparison of
recorded temperatures can be made with 30-year monthly
mean temperatures. The 30-year monthly mean maximum
and minimum temperatures range from 30eF to 81 °F and from
17°F to 62°F, respectively.
Mean monthly precipitation values, based on data collected
at the Niagara Falls International Airport, are shown in
Table 4-2. For comparison, the precipitation totals for each
month during the air monitoring study are included in
Table 4-3.
4-16
-------
Table 4-2
MONTHLY PRECIPITATION-HISTORICAL RECORD
NIAGARA FALLS INTERNATIONAL AIRPORT(a)
(inches precipitation)
Month/
Year
51
52
53
54
55
56
57
58
59
60
61
62
63
64
68
69
70
71
72
73
74
75
Jan.
4.13
2.27
2.20
1.93
2.66
4.62
3.12
3.88
2.37
2.45
2.09
.23
.60
3.04
.74
.47
4.01
1.11
1.50
2.15
1.46
1.85
2.92
4.28
4.01
4.32
4.59
4.31
4.62
Jul.
4.09
1.84
3.63
.55
2.23
3.59
3.49
2.03
.42
1.48
3.50
.97
3.23
2.64
3.16
1.61
1.78
1.99
1.32
Aug.
1.09
4.32
4.85
8.96
5.46
7.38
.79
2.87
4.48
6.00
1.06
4.41
3.90
5.23
1.50
2.53
4.98
Sep.
2.38
1.81
3.94
2.07
2.30
3.13
3.03
2.86
3.13
1.06
1.11
3.29
1.55
3.71
1.78
2.32
2.52
Oct.
1.65
.26
.49
6.77
7.18
.97
1.33
1.03
4.54
^*) 39
2'.18
.44
1.99
3.34
2.25
3.69
4.18
1.41
2.17
Nov.
6.06
1.86
2.53
3.57
2.07
2.13
2.33
2.56
2.30
2.44
5.03
*1.61
3.47
3.30
2.10
3.27
4.79
4.43
2.66
Dec.
3.41
2.86
2.07
3.81
2.74
2.86
4.04
1.09
3.73
2.77
2.42
1.73
3.08
3.10
3.17
3.57
3.46
4.20
2.44
3.51
All
Months
27.31
33.07
44.41
35.91
40.91
36.90
25.34
35.30
-------
Table 4-2
(continued)
Month/
Year
76
77
78
79
80
81
Period of Record
Mean
S.D.
Total OBS
Jan.
2.87
1.76
3.33
2.94
1.71
1.63
(1951-1964
2.46
.898
763
Feb. Mar.
2.26 4.96
.99 2.35
1.62 1.73
1.48 1.70
1.45 3.21
4.30 1.74
, 1968-1981)
2.40 2.88
1.181 1.202
679 744
Apr.
3.32
3.09
1.39
3.56
3.75
3.63
3.01
1.011
720
May
4.09
2.02
2.32
1.87
1.43
2.58
2.88
1.420
744
Jun.
4.86
2.12
2.38
1.58
3.74
4.30
2.72
1.538
738
Jul.
3.94
2.71
1.90
2.64
3.02
5.78
2.65
1.155
753
Aug.
1.01
4.09
3.40
3.49
2.28
2.14
3.75
2.105
713
Sep.
2.34
7.90
5.57
5.60
3.63
3.60
3.07
1.578
690
Oct.
2.42
2.10
3.72
4.15
4.11
3.19
2.76
1.867
747
NOV.
1.08
5.05
1.52
2.86
2.17
2.90
3.02
1.247
758
Dec.
2.29
4.71
2.05
2.91
2.94
3.97
3.04
.822
806
All
Months
35.44
38.89
30.93
34.78
33.44
39.76
35.01
4.749
8,855
(a)-Data based on daily observations and compiled by the Global Cllmatologlcal Branch, USAFETAC Air Weather Service.
(b)-Data based on less than full month's record.
-------
Month
August
September
October
November
December
Table 4-3
MEAN MONTHLY PRECIPITATION AND
TOTAL MONTHLY PRECIPITATION3
For Period July-December 1987
Mean Monthly
Precipitation (inches)b
3.7
3.1
2.8
3.0
3.0
Total
For Month
1987 (Inches)
4.3
4.7
2.1
4.1
0.7*
"Data recorded at Niagara Falls International Airport.
bMean monthly precipitation based on data recorded over
period 1951-1964,1968-1981.
°Value represents total precipitation during the period ot
sampling (December 1-13).
-------
Reports that chemicals were leaking from the landfill in-
creased during the mid- 1970s. During the 6-year period be-
tween 1972 and 1977, annual precipitation exceeded the
historical annual average each year except 1975. Highest
monthly precipitation occurred in September 1977; it was
more than twice the monthly average.
Although less representative of the EDA, the precipitation
recorded at Buffalo greatly exceeded the annual average over
this same period of record. The precipitation during 1977 was
1.5 times the annual average. The highest monthly precipita-
tion was 10.67 inches during July 1977, which is more than
three times the average for this month.
The data in Table 4-3 indicate that, with the exception of Oc-
tober 1987, the precipitation during the air sampling period
was greater than the historical mean monthly values.
A comparison of the data in both Tables 4-1 and 4-3 reveals
that, although the monthly precipitation during August and
September 1987 was significantly above the mean monthly
value, ground water levels in the wells changed very little.
4.4 OTHER
4.4.1 Detection Limits and Selection of LCICs
COMMENTS: Statistical certainties (about whether LCICs are present in the
EDA) can be obtained using only the upper range value of the
quantifiable detection limits. Habitability decisions may
have to consider LCIC exposure limits versus quantifiable
detection limits. (Dr. Janick Artiola—Written Preliminary
Comments, pp. 7-8.)
Assess the use of these LCICs. Can a probable determination
of the prevalence of other chemicals from Love Canal be
made? This assumption needs some resubstantiation in light
of the study outcome and other current information. (Dr.
Michael Stoline-Written Preliminary Comments, p. 12.)
4-20
-------
There is an underlying assumption that any chemical as-
sociated with the LCICs does not present a significant hazard
if the LCICs are present at levels below the limit of detection,
i.e., 1.0 ppb. (Dr. Joan Daisey-Written Preliminary Com-
ments, p. 8.)
Discuss the selection of indicator chemicals. (Dr. Carl Shy-
-Written Preliminary Comments, p. 6.)
RESPONSE: The evaluation approach and decision-making criteria used to
select LCICs have been previously peer reviewed. Based on
pilot study results, the air assessment was designed so that
any detected LCIC was considered significant and reflected
conditions different from areas outside the EDA. The LCICs
were selected to represent a large number of chemicals
originating from Love Canal. The selection of LCICs was
based on factors discussed in the habitability criteria docu-
ment (NYSDOH and DHHS/CDC, 1986), which did not in-
clude health risk. The habitability study was designed to
determine whether the EDA is as safe as similar inhabited
communities in Western New York State.
4.4.2 Interpretation of the Results
COMMENT: The finding that one home out of 562 in the EDA had
measurable levels of air LCICs should be viewed in the
perspective that, in the pilot study, 1 home out of 31 in the
control area also had measurable levels. (Dr. David Schoen-
feld-Peer Review Meeting Comments, June 20,1988.)
RESPONSE: The pilot study results contributed to the design of the air as-
sessment and are still an integral part of the overall study. The
Commissioner of Health for the State of New York will con-
sider both the pilot study and the air assessment results in
determining habitability.
COMMENT: Is it not inconsistent to observe measurable levels of LCICs
in soil but not in air? (Dr. Carl Shy—Peer Review Meeting
Comments, June 20,1988.)
4-21
-------
RESPONSE: It was noted by Dr. Shy in the peer review meeting that LCICs,
specifically chlorobenzenes, were detected in the EDA soil
but not in the air. Selection of LCICs was based upon specific
physical and chemical characteristics. The characteristics
best suited for an air LCIC were different from those best
suited for a soil LCIC and, therefore, the LCICs for the two
media were different: 1,2-dichlorobenzene (DCB), TCB, and
1,2,3,4-tetrachlorobenzene (TeCB) were chlorobenzenes
selected for soil while CB was selected for air.
However, even if LCICs were consistent between the two as-
sessments, it would still be likely that concentrations would
be detected in soil and not in the air influenced by this soil.
The rate of volatilization of a compound from soil to the air
is affected by many factors, including properties of the com-
pound, soil properties, and environmental conditions. It is
common to find concentrations of a compound in the soil at
several orders of magnitude above those in the air immediate-
ly influenced by that soil. Therefore, it would not be incon-
sistent to detect low level compounds in the soil and not detect
those same compounds in the air.
COMMENT: Concentrations of contaminants in air are highly variable, and
even thorough studies such as this one cannot provide com-
plete assurance that all potential occasions of detectable con-
taminant levels have been identified.
Even though the air study identified only one home in the
EDA with measurable levels of indicator chemicals, if
rehabitation occurs and if individuals continue to have con-
cerns, it may be desirable to make information available on
home ventilation devices that might be used to reduce soil-
gas accumulation in homes. (Dr. Joan Daisey—Peer Review
Meeting Comments, June 20,1988).
RESPONSE: As discussed in Section 4.1.1 above, this study, as would any
study, falls short of offering complete assurance that all
potential occasions of detectable LCIC levels have been iden-
tified. However, in its discussion of issues raised during the
peer review meeting, the TRC determined that the data ob-
4-22
-------
tained during the air assessment did not appear to indicate the
need for installation of home ventilation devices.
4.4.3 Additional Information
Also included in this volume as Appendix E is the list of errata for Volume n.
4-23
-------
Section 5.0
RESPONSES TO COMMENTS ON THE SOIL
ASSESSMENT FOR INDICATOR CHEMICALS
This section briefly describes the peer review comments on Volume ffl, Soil
Assessment-Indicator Chemicals (CH2M HILL, 1988b), and gives the
responses to those comments. Both written preliminary peer review com-
ments and verbal peer review comments made during the peer review meet-
ing are addressed in this section. The peer review comments have been
organized under the categories of "Design," "Methods," and "Results." Each
comment is paraphrased, the peer reviewer making the comment is
referenced, and the response to the comment is given. Where lengthy
responses are required, the response is summarized in this section, while the
full response is provided in an appendix.
5.1 DESIGN
5.1.1 Statistical Sampling Design
COMMENT: Since the soil LCIC assessment had a much lower percentage
of nonquantifiable results than anticipated, why was a non-
parametric analysis the only analysis conducted?
(Dr. Michael Stoline—Peer Review Meeting Comments,
June 20,1988.)
RESPONSE: The nonparametric analysis was adopted in part because of the
ability of this type of analysis to handle nonquantifiable (non-
detect) data. In addition, the nonparametric analysis requires
fewer assumptions about the distribution of concentration
values. The performance of a nonparametric analysis, as
measured by the power achieved for a given sample size, is
nearly as great as a parametric test when the assumptions of
the parametric test are met, and much greater when those as-
sumptions are not met.
In spite of the better than expected performance of the
laboratories, a parametric analysis still risks misinterpretation
5-1
-------
of the data if the statistical distribution is inappropriate. A
more complete discussion and review of the decision to use
nonparametric analyses is presented in Appendix F.
COMMENT: The pilot study peer review indicated that the sample sizes
would be independently reviewed. Was this done? (Dr.
Michael Stoline-Written Preliminary Comments, p. 14.)
RESPONSE: The sample size specified for the full study was reviewed by
the TRC and by statistical consultants. As the retrospective
power analysis presented in Volume m showed, the design
goal of 90 percent power to detect an order of magnitude dif-
ference was exceeded with the sample sizes used.
COMMENT: Explain the changes made for the simulation model since the
pilot study. (Dr. Michael Stoline-Written Preliminary Com-
ments, pp. 5-6.)
RESPONSE: The simulation model was originated prior to the pilot study
to aid in the decision of how to design the soil assessment
comparison study. The simulation model evolved as the
needs of the project evolved; therefore, the use and charac-
teristics of the model followed the phases of the project: pre-
pilot, pilot study design, post-pilot design, and soil
assessment study data analysis.
Before the pilot study began, the main purposes of the model
were to (1) aid in deciding which statistical test to use for
comparison and (2) estimate the number of samples needed
to achieve the power required by the habitability criteria. Be-
cause the distribution of the data was unknown, several poten-
tial data distributions were used to generate simulated data:
the generalized extreme value, the mixed exponential, the
Pearson type m, the lognormal, and the normal distribution.
Each of these distributions was regarded as a potential model
(with some censoring of below-detection-limit data) of the
comparison areas. One of several types of shifts (i.e., change
in location, scale, or both) was applied to each distribution to
simulate an EDA area.
5-2
-------
As reported in the pilot study report (CH2M HILL, 1987d),
a variety of potential test statistics were examined for use in
the soil assessment for indicator chemicals. These included
Fisher's Exact test, Student's t test, Wilcoxon rank sum test,
Moses scale test, Shorack's APF test, and Mood's test Of
these, the Wilcoxon rank sum test proved to be the most
robust to the distribution of the data.
The pilot study results demonstrated that the percentage of
nonquantifiable data would be lower than seen in previous
studies. It also showed that the interlaboratory variability was
a major component of the variability.
The simulation model was modified to estimate the sample
sizes required for the full LCIC soil assessment. The distribu-
tion that best fit the pilot study data was a lognormal mixture
distribution, which was used for further simulations to es-
timate the sample size required to achieve the power criteria.
The other candidate distributions were dropped.
During the sampling for the full LCIC soil assessment, the
model was further modified to include the analysis of data by
the Wilcoxon test, blocked across laboratories. The test that
was implemented is discussed in Appendix B of Volume HI.
After the analysis of the LCIC soil assessment data, a further
simulation was conducted to estimate the retrospective power
of the sampling and analysis as carried out. The results of
this simulation are given in Appendix G. This simulation
used a lognormal distribution fitted to the comparison area
data, as seen in the sampling program.
Note that the fitted parametric distributions were used only
for the purposes of estimating the power of statistical tests.
No attempt was made to perform any of the comparisons
using the parameters of the fitted distributions. The
parameters of the fitted distributions, while adequate for the
generation of artificial data, did not fit well enough to ade-
quately represent the actual data for purposes of comparison.
5-3
-------
COMMENT: Did flagging samples compromise "blind" analysis of samples
by laboratories? (Dr. David Schoenfeld-Written Prelimi-
nary Comments, p. 1, and Consensus Comment-Peer Revew
Meeting, June 20,1988.)
RESPONSE: The pilot study indicated that samples from Sampling Area 1
may contain concentrations of LCICs sufficiently elevated to
damage the calibration of the laboratory's Gas
Chromatograph/Mass Spectrometer (GC/MS). This would
cause delays in the analysis of further samples while the in-
struments were recalibrated. To avoid this, all samples from
Sampling Area 1 were flagged as being potentially elevated
so that the laboratories could perform an initial screening
analysis. An equal number of samples selected at random
from all other areas were given the same flag to be screened.
This preserved the "blind" aspect of analysis by preventing
the laboratories from knowing the area from which a sample
came.
5.1.2 Selection of Census Tracts
COMMENTS: Document the criteria used to select the two Niagara Falls
comparison areas, Census Tracts 221 and 225. (Dr. Michael
Stoline--Preliminary Written Comments, p. 14.)
Specifically, why was Census Tract 225 selected since it is in
such close proximity to the EDA? (Dr. Michael Stoline—Peer
Review Meeting Comments, June 20,1988.)
RESPONSE: This documentation is provided in Section 6.3, "Selection of
Comparison Areas" (pp. 6-13 through 6-20), in Volume I, In-
troduction and Decision-Making Documentation (1C AIR and
CH2M fflLL, 1988). Specifically, an area as close to the
EDA as possible was selected to provide a comparison area
with the same water supply and the same potential airborne
contaminant sources as the EDA. However, Census
Tract 225 complies with the criterion for selection by being
more than one-half mile from the Love Canal site proper and
other known chemical waste dump sites.
5-4
-------
5.2 METHODS
5.2.1 Method Documentation (QAPPs)
COMMENT: It is not clear how the QAPP and the final revised version of
the QAPP relate to each other. The revised version appears
to be a complete replacement of the original. However, the
revised version omits any mention of the matrix spike (MS)
samples or the blind quality control (QC) samples. (Dr. Joan
Daisey-Written Preliminary Comments, p. 7.)
RESPONSE: Two final QAPPs relate to the soil LCIC study addressed in
Volume HI. The first is entitled, Soil Sample Collection and
Preparation QAPP (Final Revised Version) (CH2M HILL,
1987b), and the second is entitled, Soil Sample Laboratory
Analysis QAPP (CH2M HILL, 1987c). These two QAPPs
addressed the field and laboratory phases of the study, respec-
tively. It was not the intent of the field QAPP to address the
MS or blind QC sample analyses, since these are laboratory
QC samples and not field QC samples.
COMMENT: Document any changes in the study methods made during im-
plementation of the QAPPs. (Dr. Janick Artiola—Written
Preliminary Comments, p. 3.)
RESPONSE: Changes in sampling or analysis procedures made during im-
plementation of the two respective QAPPs involved
responses to problems encountered in the field or in the
laboratories. These include problems with (1) coordinating
the schedules for sampling and analysis, resulting in a change
in the holding time requirements (discussed in Volume HI:
Section 5.2 and Appendix D); (2) extruding samples at the
sample preparation laboratory; this problem was solved by
taking replacement samples (see Volume m, pp. 6-1 and 6-
4); (3) intralaboratory split sample allocation (see comment
and response in Section 5.2.2 below); (4) blank interference
in the laboratories for which several corrective actions were
taken (see Volume m, p. 6-13); and (5) general analytical
procedures in one laboratory, which were responded to by
5-5
-------
collecting contingency samples (not utilized) and sending the
original samples to another laboratory for analysis (see
Volume III, pp. 6-4 and 6-14).
5.2.2 Sample Collection and Preparation Methods
COMMENT: What was the depth of the soil core relative to what the peer
reviewers had previously recommended? (Dr. Randall Char-
beneau-Written Preliminary Comments, p. 2.)
RESPONSE: During the peer review of the pilot study, the peer reviewers
recommended a soil core depth of 12 inches. The depth
criterion implemented was a 12-inch depth with a minimum
of 7 inches where 12 inches could not be achieved. The min-
imum of 7 inches was based on the minimum required
volume of soil for analysis.
As can be calculated from the numbers in Table 5-1,72 per-
cent of the soil samples collected for LCIC analysis had a
column length (core depth) of 12 inches, and 11 percent had
a column length of 11 inches. Samples with a column length
of less than 11 or 12 inches were collected from areas in
which the fill soil was not deep enough to provide a longer
soil column. These conditions were reported in the pilot
study report.
COMMENT: How much soil was collected from each location? Were the
soil materials (cores) segregated (sieved) in any manner?
How were the roots (plant materials) removed? How much
of the soil sample collected was used for analysis? Was there
any effort to determine the adequacy of the sample mixing
and splitting techniques used vis-a"-vis the representativeness
of the soil core collected? (Dr. Janick Artiola-Written
Preliminary Comments, p. 2.)
RESPONSE: The soil cores taken contained approximately 60 to 100 grams
of soil, depending on the length of the soil column that could
be obtained (see Table 5-1 for further information on actual
soil column lengths). The soil core materials were not
5-6
-------
Table 5-1
LENGTH OF SOIL COLUMN OBTAINED BY NEIGHBORHOOD AND COMPARISON AREA
SOIL ASSESSMENT-INDICATOR CHEMICALS
Length of Soil Column (Inches) Total
Area
EDA
Cheektowaga and
Tonawanda
Census Tract 221
Census Tract 225
TOTAL
Neighborhood
1
2
3
4
5
6
7
8
9
10
11
12
13
7
1
1
0
3
0
1
3
1
0
1
0
0
2
1
3
1
18
8
1
0
4
1
0
5
0
0
4
1
0
1
3
1
1
5
27
9
2
0
2
5
2
4
0
2
4
1
2
4
7
1
1
5
42
10
1
4
2
4
5
3
0
3
1
2
3
0
3
2
9
8
50
11
1
5
2
5
0
7
2
4
2
2
4
14
10
5
12
9
84
12
38
34
31
45
23
53
28
19
19
22
24
46
48
51
41
38
560
44
44
41
63
30
73
33
29
30
29
33
65
73
61
67
66
781
-------
segregated or sieved. Roots and other plant material were
removed by cutting and lifting them off with a clean knife.
While the complete collected soil sample was sent for
analysis, only 20 grams were needed for analysis (see Appen-
dix A of the soil sample analysis QAPP, CH2M HILL,
1987c).
The procedure for mixing the samples was developed as part
of the soil pilot study. Analytical results were used to deter-
mine the mixing time that obtained optimum soil
homogeneity while minimizing volatile and semivolatile
losses. Further information on the study conducted to deter-
mine the optimum sample preparation protocol is given on
page B-ll in Volume I of the pilot study report (CH2M
HILL, 1987d). A retrospective quantification of the
variability of the sample mixing and splitting techniques is
difficult because of the degree of variability introduced
through the sample extraction and analysis process.
COMMENT: Clarification and full, detailed documentation from the EPA
National Enforcement Investigation Center (NEIC)/Techlaw
are needed regarding what was observed during the field
audits and whether the sample collection teams followed the
sampling protocols in the sample collection and preparation
QAPP (CH2M HELL, 1987b). (Dr. Joan Daisey-Written
Preliminary Comments, pp. 1-2.)
RESPONSE: Several field and laboratory audits were conducted during the
soil LCIC study. Each audit comprised two parts: the eviden-
tiary documentation procedures audit that NEIC/Techlaw
was responsible for and the technical procedures audit that
the EPA Environmental Monitoring Systems Laboratory-Las
Vegas/Lockheed Engineering and Management Services
Company (EMSL-LV/LEMSCO) conducted. NEIC/
Techlaw's evidentiary audit summary report, contained in
Appendix G of Volume ffl, was based on reports of several
individual NEIC/Techlaw field and laboratory evidence
audits. Similarly, EMSL-LV/LEMSCO's technical proce-
dures audit summary report, contained in Appendix H of
Volume HI, was based on several individual field and
5-8
-------
laboratory technical audit reports. Including these addition-
al audit reports in Volume ni would have made the report un-
manageably large and less amenable to widespread public
distribution and use. Copies of the individual audit reports
may be obtained upon request from EPA Region n through
a Freedom of Information Act (FOIA) request
COMMENT: The previous peer review recommended that the soil material
clinging to the grass be shaken off and included. Was this
done? (Dr. Joan Daisey—Written Preliminary Comments,
p. 6.)
RESPONSE: The field sampling crews were not directed to shake the
removed sod layer, so the loose soil could be included in the
sample. However, the following procedure, reported on
page B-12 of Appendix B of the soil LCIC QAPP (CH2M
HILL, 1987b), was followed:
"1. Remove knife from packaging. Clear the area to be
sampled of any surface debris with the knife (twigs, rocks,
litter, snow, etc.). If the sampling area is covered with
grass or weeds, remove a small piece of sod with the knife
(3 to 4 inches square) and place it to one side, away from
the sampling area...."
COMMENT: While it is true that intralaboratory variability can be assessed
by the use of other QC samples, it is not clear how the over-
sight of the intralaboratory split samples occurred. Were the
Standard Operating Procedures (SOPs) not sufficiently clear
or was the split done so infrequently that it was forgotten?
(Dr. Joan Daisey--Written Preliminary Comments, p. 8.)
RESPONSE: The sample preparation team was given a daily schedule, like
the one shown in Table 5-2, that indicated which laboratories
were to receive samples and splits. As can be seen, the
schedule did not indicate which samples were to be split. This
was purposely done to allow the soil preparation laboratory
the freedom to select a full column (i.e., 12 inches) of soil as
was needed to provide a sufficient amount of soil for the
analysis of both halves; the actual length of soil column that
5-9
-------
Table 5-2
EXAMPLE SAMPLE PREPARATION SCHEDULE PROVIDED DAILY
TO THE SAMPLE PREPARATION TEAMS
SOIL ASSESSMENT-INDICATOR CHEMICALS
Station
Number
759
760
765
766
282
283
288
289
355
356
360
361
362
461
462
469
470
473
Split 3
Split 4
1019
1115
1222
Days
Preparation
Team
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
Analytical
Laboratory
CEC
AQI
CAA
CEC
AQI
CAA
AQI
MGM
AQI
MGM
VER
CEC
AQI
VER
CEC
CEC
AQI
NUS
EMS
CEC
CEC
NUS
AQI
-------
a particular location could yield could not be determined
before sample collection began. While the schedule desig-
nated that a split had to go to a certain laboratory, it did not
indicate intralaboratory, splits, i.e., that the original half and
the split half were to go to the same laboratory. As a
hypothetic example, assume that both Splits 3 and 4 in
Table 5-2 were to be intralaboratory splits. On Day 5, the
sample preparation crew selected samples from Station Num-
bers 282 and 283 to be split because these samples had full
columns. The original halves of these samples were destined
for laboratories AQI and CAA, respectively. Splits 3 and 4
were designated for laboratories EMS and CEC. Thus, the
intended intralaboratory splits were not sent to the same
laboratory as intended. In general, although the schedule was
carefully followed, most of the intended intralaboratory splits
were not sent to the same laboratory.
5.2.3 Sample Analysis Methods
COMMENT: Did modifications to the analytical method after the pilot study
actually result in lower detection limits? (Dr. Michael
Stoline-Written Preliminary Comments, pp. 11-12.)
RESPONSE: Yes, the detection limits were lowered by changes in the
analytical method after the pilot study. Qualitatively, this is
demonstrated by the detection of low-level concentrations in
the blind QC samples prepared by EMSL-LV. The quantita-
tive estimation of detection limits is discussed in Appendix H.
COMMENT: It appears from the box plots that 2-chloronaphthalene (CNP)
is ubiquitously present at levels below 0.2 ppb. Is this the
case? Did CNP routinely appear in the method/holding
blanks? This would indicate that the source may be
laboratory contamination rather than true concentrations in
field samples. Why weren't the concentration results for field
samples corrected for the blank concentration? (Dr. Janick
Artiola-Peer Review Meeting Comments, June 20,1988.)
5-11
-------
RESPONSE: The Method/Holding Blank (MHB) and field handling blank
(FHB) results are discussed in Appendix I, "Chemical Data
Quality Assessment." Considerable effort was expended to
keep the level of laboratory contamination as low as possible.
For example, all laboratories were provided with a common
supply of solvents and reagents that had been subjected to ex-
tensive analyses to demonstrate that the background levels of
LCICs and LCIC interferences were as low as possible. 2-
CNP was detected in most MHBs at a mean concentration
below 0.1 ppb, which is well below the 0.5-ppb blank accep-
tance level for the project. This indicates that CNP was a
ubiquitous laboratory or solvent/reagent contaminant present
at very low levels. (See Appendix F of Volume HI for more
details.)
Correcting sample concentration results using MHB con-
centrations is never routinely performed for this type of
chemical analysis. The procedures used in this study are
primarily based on those required for the EPA Superfund
Contract Laboratory Program (CLP). In the CLP semi-
volatile analysis procedure, results of the MHB analyses are
reported along with the field sample concentrations. This al-
lows the end user of the data to decide how best to handle the
MHB results. It is often observed that the concentration of a
compound in a field sample is lower than that observed in the
MHB. If these results were blank corrected, the reported con-
centration would be negative. In any event, the balanced
design of the statistical comparisons performed in this study
would tend to negate the effect of low-level laboratory con-
tamination. It was decided that correcting field sample results
for the MHB and the FHB concentrations was not advisable.
5.3 RESULTS
5.3.1 Consistency of Data Qualifiers
COMMENT: Some of the data qualifiers in the data base were not used con-
sistently by the six laboratories. Although this would not
5-12
-------
have an effect on the Good data, it would be appropriate to
make the qualifiers for the other data consistent in the final
data base. (Dr. Joan Daisey—Written Preliminary Comments,
p. 6.)
RESPONSE: Although it is desirable to have consistent qualifiers in the data
base, it is generally not good practice to change data qualifiers
assigned by the chemist in the laboratory at the time of
analysis. Problems could have been encountered that
(1) could not be indicated using the data qualifiers specified,
(2) are not otherwise documented, and (3) could have had an
effect on the quality of the data. The few qualifiers that were
used inconsistently were detailed qualifiers, which did not
have an effect on the assignment of the overall data usability
qualifiers of Good, Uncertain, and Bad.
5.3.2 Blank Sample Data
COMMENT: The report did not present data from the analysis of the FHBs
used to show that samples were not contaminated during the
sample collection phase of the study. A quantitative state-
ment about how much higher the FHBs were than the MHBs
should also be included. The statement that the MHBs all
showed very low levels of contamination should also be quan-
tified. Were they an order of magnitude lower than the limits
given, or 50 percent lower? (Dr. Joan Daisey—Written
Preliminary Comments, pp. 2-3.)
RESPONSE: Table I-11 of Appendix I summarizes the FHB results and al-
lows a comparison of the FHB results with the MHBs ex-
tracted in the same batch. Most LCICs show only a very
slight increase in concentration of the FHBs compared to the
MHBs. The largest effect is for DCB, which increased by an
average of 0.1 to 0.2 ppb in all laboratories. All other in-
creases were less than 0.1 ppb.
There are two possible causes for this marginal effect. The
first possibility is that there were low levels of LCICs in the
FHB matrix. The MHBs were based on a sand matrix care-
5-13
-------
fully prepared and provided by EPA to be used as an MHB
material. The FHBs were based on a soil matrix selected to
be similar in composition to soil from the Love Canal area.
While this soil was demonstrated by chemical analysis to be
acceptable as a blank matrix, there could be low levels of
LCICs present. Another possibility is that the slight eleva-
tion of FHB concentrations over the MHB concentrations are
caused by small amounts of contamination experienced
during the sample collection, soil preparation, transportation,
or laboratory sample extraction and cleanup processes. This
small degree of contamination did not trigger any corrective
action during the laboratory phase of the project and is not
expected to degrade the usefulness of the results since the con-
tamination levels observed are less than the precision of the
analysis method.
Table I-10 of Appendix I contains a summary of the MHB
results for each LCIC and each laboratory. This summary is
based solely on the computerized MHB equivalent concentra-
tions reported by the laboratories and does not include the
results of applying the MHB interpretation rules. All MHBs
reported in the study were accepted by the EMSL-LV when
the Selected Ion Current Profiles (SICP) were scrutinized.
The slightly elevated levels of DCB, especially notable in
Laboratory 8, were caused by minor laboratory
contamination. The high concentrations of beta-
hexachlorocyclohexane (B-BHC) and gamma-BHC (G-
BHC) were caused by the ubiquitous interferences arising
from the laboratory detergent used to clean glassware. All
laboratories experienced this interference to some extent and
had to be exceedingly careful to either minimize the con-
centrations of the interference or chromatographically
separate it from the BHCs.
5.3.3 Data Completeness
COMMENT: What was the data completeness for field and QC samples
relative to the 90 percent completeness goal in the QAPP
5-14
-------
(CH2M HILL, 19875 and c)? (Dr. Janick Artiola-Written
Preliminary Comments, p. 4.)
RESPONSE: For field samples, a total of 894 samples were planned and
781 were analyzed. This corresponds to a completeness of
87.4 percent. For field QC samples, 76 FHBs and 83 field
splits were planned for a total of 159 field QC samples. There
were 68 FHBs and 73 field splits analyzed for a total of 141
field QC samples analyzed. This corresponds to a complete-
ness of 88.7 percent. For laboratory QC samples, the com-
pleteness was 100 percent, in the sense that all laboratories
analyzed all laboratory QC samples at the planned frequen-
cy. The percentage of the laboratory samples meeting the QC
criteria is detailed in Table 1-5 of Appendix I.
5.3.4 Actual Versus Planned Sample Sizes
COMMENT: The actual sample size in many sampling areas varied from
the 75 samples discussed in the design. Did this reduce the
power of the statistical tests? (Dr. Michael Stoline-Written
Preliminary Comments, pp. 1-3.)
RESPONSE: The simulation study conducted after the pilot study suggested
that, with the distributions seen in the pilot study, ap-
proximately 50 samples would provide 90 percent power of
detecting an order-of-magnitude difference. Additional
samples were allocated to sampling areas to provide a contin-
gency for sample loss. Retrospective simulation of the power
indicates that, even with the loss of samples from
Laboratory 4, the power exceeded the goal of 90 percent for
detecting an order-of-magnitude difference, as shown in
Tables K-13 and K-14 of Volume m.
COMMENT: Reaffirm in Volume V that the originally planned statistical
analyses of the data remain the main or central analyses
despite the additional sensitivity analyses, etc., requested by
the peer reviewers. (Dr. Michael Stoline—Peer Review Meet-
ing Comments, June 20,1988).
5-15
-------
RESPONSE: Appendix J of this volume presents the additional sensitivity
analyses and their results, which are supplemental to those
presented in Section 6.0 of Volume in.
5.3.5 Neighborhood Power Calculations
COMMENT: Statistical tests should be performed using the original neigh-
borhoods. These analyses may require additional analyses.
For instance, if some of the original neighborhoods are not
contaminated (as determined by the results of the statistical
test), one would want to calculate retrospective powers for
these neighborhoods to show that they passed the 90 percent
power rule. (Dr. David Schoenfeld-Written Preliminary
Comments, p. 5.)
RESPONSE: Tables K-13 and K-14 in Appendix K of Volume HI present
the results of a retrospective power analysis for the univariate
and multivariate comparisons for the sampling areas. Appen-
dix G of this volume presents the retrospective power for the
univariate and multivariate comparisons based on a lognor-
mal mixture fit for the sampling areas and the neighborhoods.
The sampling area power calculations are performed with a
nominal sample size of 75 in both the comparison and EDA
areas, while the neighborhood power calculations are per-
formed with a nominal sample size of 35 in both areas. Since
actual sample sizes at both aggregation levels often exceeded
these numbers, the power estimates are biased low.
5.3.6 Magnitudes of the Differences and Other Percentiles of LCIC
Concentrations
COMMENTS: Highlight differences of magnitude of statistically significant
differences. (Dr. Carl Shy and Dr. Michael Stoline-Written
Preliminary Comments, pp. 5 and 17, respectively.)
Report other percentiles of LCIC concentrations such as the
90th or 95th. (Dr. Michael Stoline-Written Preliminary
Comments, p. 16.)
5-16
-------
Fit a lognormal or lognormal mixture distribution to the data
and report the upper percentiles for the fitted distribution.
(Dr. Michael Stoline—Written Preliminary Comments, p. 16.)
RESPONSE: Tables in Appendix G show both the empirical and the fitted
percentiles of the distributions for each neighborhood. These
tables can be used to an approximate magnitude for statisti-
cally significant differences. Calculation of the differences
is complicated by the fraction of non-detect values in each
distribution and is, therefore, not presented.
5.3.7 Analysis of Outlying Values
COMMENTS: There is an impression that, for most LCICs, there are more
outlying high values in the EDA areas than in the control
areas. A p-value for this hypothesis should be calculated.
(Dr. David Schoenfeld-Written Preliminary Comments,
p. 3.)
The sensitivity analysis at 1.0 ppb (shown in Table K-ll of
Volume III) satisfies my concerns about outliers, except for
possibly the LCIC TCB. Run a sensitivity analysis at 2.0 ppb
for TCB. (Dr. David Schoenfeld-Peer Review Meeting
Comments, June 21,1988.)
RESPONSE: The influence of high values is examined by the sensitivity
analysis in which all values less than 1.0 ppb are set
equivalent to non-detect values. The results of this analysis
are shown in Tables K-4 through K-12 of Volume in. As dis-
cussed during the peer review, 1.0 ppb is an appropriate cutoff
value for most LCICs. However, TCB appears to have
generally elevated concentrations so that 2.0 ppb may be
more appropriate for this compound. Table J-3 of Appen-
dix J gives the results for a 2.0-ppb cutoff value for all LCICs
including TCB.
The statistical test used is commonly referred to as compar-
ing the medians of two distributions. As discussed in
Volume in, this test more precisely tests whether a sample
5-17
-------
concentration from the EDA is equally likely to be greater
than or less than a sample concentration from the comparison
area. This test actually uses the whole probability distribu-
tion of the concentration values from both areas.
Two sensitivity analyses have been conducted to examine
whether the test is dominated by either the low values or the
high values. The first of these removed the 10 percent highest
values from the distribution in each comparison area. If fewer
comparisons are significant with this change in the data, then
high values tend to dominate the comparison, while if more
comparisons are significant, then low values tend to dominate
the comparisons.
The second sensitivity analysis to examine the influence of
outliers sets all low values, those less than 1.0 (or 2.0) ppb,
to be equivalent to a nonquantifiable value. Thus, the lowest
part of the distribution was forced to be the same. If fewer
comparisons are significant with this change in the data, then
low values tend to dominate the comparison, while if more
comparisons are significant, then high values tend to
dominate the comparisons.
Note that while complementary, the two sensitivity analyses
have different definitions of low and high. The first analysis
defines upper 10 percent. The second analysis defines low
and high variably: The range of low values can be from
50 percent to over 90 percent of the sample concentrations,
depending on the number of samples with concentrations
below the cutoff (1.0 or 2.0 ppb) for each sampling area and
each LCIC.
COMMENT: A more detailed explanation of the purpose of each sensitivity
analysis should be provided. (Dr. David Schoenfeld-Peer
Review Meeting Comments, June 20,1988.)
RESPONSE: Explanations of each of the sensitivity analyses are given in
Appendix K. Sensitivity analyses were conducted on the fol-
lowing data sets:
5-18
-------
0 Data sets including data classified as Uncertain
« Data sets treating all concentrations below 1.0 ppb as
non-detects
• Data sets deleting the upper 10 percent of the concentra-
tions
« Data sets in which the ion identification criteria are
"relaxed"
• Data sets treating all concentrations below 2.0 ppb as
non-detects
COMMENT: Provide tables for each EDA sampling area that summarize the
results of the central analysis and all pertinent sensitivity
analyses reported in this volume and in Volume III. (Dr.
David Schoenfeld-Peer Review Meeting Comments,
June 20-21,1988.)
RESPONSE: Appendix K presents seven tables—one for each EDA sampling
area-that summarize the results of all of the sensitivity
analyses that were reported in this volume and in Volume HI.
The explanations of the issues addressed by each sensitivity
analysis are also contained in Appendix K.
5.3.8 Correcting for Chance Outcomes of the Univariate Tests
COMMENT: Construct a table similar to Table 6-9 in Volume III to display
the results of the univariate comparisons with a Bonferroni
correction for the number of LCICs. (Dr. Michael Stoline-
Written Preliminary Comments, p. 10.)
RESPONSE: Figures J-1 and 2 of Appendix J give a graphical presentation
of the Bonferroni corrected contrasts for eight LCICs. Other
corrections can be easily obtained by use of the table of p-
values (Volume III: Table K-l). For example, a 5 percent
level of significance is indicated for comparisons with a p-
value of 0.05 or less. With the Bonferroni correction for eight
LCICs, this level becomes 0.006 (0.05/8), so that any com-
5-19
-------
parison with this level or less is considered significant (EDA
greater than comparison area).
5.3.9 Spatial Analysis
COMMENT: Evaluate the effectiveness of the dioxin study in accounting
for hot spots, which were not addressed in the soil LCIC
study. (Dr. Randall Charbeneau—Written Preliminary Com-
ments, p. 5.)
RESPONSE: Although the soil assessment for indicator chemicals was not
designed to find locally contaminated areas, the random al-
location of samples to neighborhoods assumed that there were
no major spatial trends in indicator chemical concentration.
Appendix L discusses spatial analyses performed to check
this assumption. The LCIC soil assessment data were
analyzed by examining a Spearman correlation of concentra-
tion with distance from the Canal and by using the geostatis-
tical method of Kriging. Neither method demonstrated any
spatial patterns of sufficient magnitude to invalidate the as-
sumption of no major spatial trends.
COMMENT: Produce a graph of sample concentration by geographical
coordinate. (Dr. David Schoenfeld—Written Preliminary
Comments, pp. 4-5.)
RESPONSE: The results of analysis of each sample are given with the
sample location in Appendix I of Volume HI. One set of
maps will be placed on display for the public's review at the
NYSDEC's offices in Niagara Falls.
COMMENT: Examine the Spearman correlation of estimated concentration
with distance from the Canal. (Dr. David Schoenfeld-Writ-
ten Preliminary Comments, p. 5.)
RESPONSE: A spatial analysis of the concentration data was performed and
no spatial trends were identified. The spatial analyses are
reported in Appendix L.
5-20
-------
COMMENTS: Determine whether EDA Sampling Areas 2 and 3 are
downslope of the Love Canal or in or near a former swale
downstream from the Canal. Determine the transport
mechanism for detectable concentrations of soil LCICs.
(Dr. Randall Charbeneau-Peer Review Meeting Comments,
June 20,1988.)
Determine if EDA Sampling Areas 2 and 3 are downwind of
the Love Canal, (Public-Peer Review Meeting Comments,
June 20,1988.)
RESPONSE: These comments were discussed at the June 21,1988, TRC
meeting (see TRC Meeting Minutes). The TRC felt that such
analyses were beyond the scope of the study. The purpose of
the study was not to determine the mode of transport but
simply to determine whether concentrations of the selected
compounds in the EDA are different from concentrations in
areas not affected by a chemical landfill.
COMMENT: A statistical comparison should be performed using the
original neighborhoods rather than the sampling areas.
(Dr. David Schoenfeld-Written Preliminary Comments,
p. 5.)
RESPONSE: Results of the neighborhood comparisons are reported in Ap-
pendix M.
5.3.10 Additional Information
Also included in this volume as Appendix N is the list of errata from
Volume III.
Appendix O provides a description of the data base used in the soil assess-
ment for indicator chemicals.
5-21
-------
Section 6.0
RESPONSES TO COMMENTS ON THE SOIL
ASSESSMENT FOR 2,3,7,8-TCDD
This section briefly describes the peer review comments on Volume IV, Soil
Assessment-2,3,7,8-TCDD (CH2M HILL, 1988c), and gives the responses
to those comments. (2,3,7,8-TCDD is also referred to as dioxin throughout
this document.) Both written preliminary peer review comments and verbal
peer review comments made during the peer review meeting are addressed
in this section. The peer review comments have been organized under the
categories of "Design," "Methods," "Results," and "Other." Each comment
is paraphrased, the peer reviewer making the comment is referenced, and the
response to the comment is given. Where lengthy responses are required,
the response is summarized in this section while the full response may be
provided in an appendix.
6.1 DESIGN
6.1.1 Representativeness of Samples and Rationale for Selection of
Sampling Media
COMMENT: The splitting of samples was not done in an unbiased manner,
i.e., using a soil sample splitter. (Dr. Janick Artiola-Written
Preliminary Comments, p. 3.)
RESPONSE: A cone splitter was originally planned to be used at the
homogenization laboratory. When soil with a high-moisture
content was encountered in a large percentage of the samples,
the method of homogenizing the samples was changed; the
homogenization was done manually both in Las Vegas in
1986 and in the field in 1987. A cone splitter would not have
been effective with the high-moisture-content samples.
The results of the archive sample analysis versus the as-
sociated original sample analysis indicated that the com-
parability was good. In addition, the archive sample taken
from the sampling location that had dioxin at concentrations
6-1
-------
above the 1.0-ppb level was homogenized and split into five
aliquots. Each of these aliquots was analyzed, and the results
were 17.3, 19.9, 20.2, 20.9, and 21.2 ppb, with a mean of
19.9 ppb, indicating relatively close agreement.
COMMENTS: Appendix A of the QAPP indicates that, during field sam-
pling, pieces of soil larger than 0.3 centimeters would be ex-
cluded. How was this done? Was a sieve used to segregate
sizes? If so, was a record kept of the percentages (by weight
or volume) of soil excluded from each sample collected? This
is biased exclusion of soil particles. The unbiased but sys-
tematic exclusion of soil materials (other than live plant tis-
sue) should be made using sieves. (Dr. Janick
Artiola-Written Preliminary Comments, pp. 3-4.)
There is major uncertainty in the representativeness of the soil
samples collected and the premises and assumptions that were
used. Although much effort was placed in the actual logis-
tics and mechanics of the soil sample statistical design,
sample tracking, and QC procedures, less thought was used
in the definition of what would constitute a representative soil
sample. Less care was taken to provide a truly unbiased
sample collection and sample preparation prior to chemical
analysis. (Dr. Janick Artiola—Written Preliminary Com-
ments, p. 6.)
RESPONSE: The QAPP (CH2M HILL, 1986 and 1987e) called for the
removal of twigs as well as stones larger than 0.3 centimeters,
mostly for practical reasons. It was known from other dioxin
sampling efforts that these objects were not amenable to the
method of homogenization (blending) that was to be used. In
addition, these materials interfere with the analytical method
and are typically set aside in the laboratory. Also, since it is
highly unlikely that dioxin would be found within the stones,
any bias of the results caused by removing stones in the soil
samples would give a higher than actual concentration of
dioxin in the soil layer; i.e., the actual concentration would
perhaps be slightly lower than that reported.
6-2
-------
COMMENTS: Reasons for the arbitrary exclusions of the "sod" layer were
not properly explained. Use of the sod layer is justified based
on this statement in the report:
"Under the direction of the TRC, the top 2 inches of soil
were selected for sampling. This selection was based on
a consideration of the most likely exposure mechanisms.
Two common activities have been identified as probable
means by which human exposure of soil would occur (1)
children playing and (2) adults gardening, etc." (CH2M
HILL, 1988c)
These statements are based on a questionable premise-that
gardening and child-playing activities take place only in the
top 2 inches of soil.
The soil zone sampled was the most active (biologically and
chemically) and was, therefore, the least likely to show levels
of dioxin because of ongoing degradation. If you were real-
ly only interested in the top 2 inches of soil, perhaps a more
direct and cost-effective approach would have been to add
2 inches of clean topsoil to the entire EDA. A better approach
would have been to sample the top 12 inches of soil. This
would have provided a more realistic and easier-to-accept
buffer zone between any remaining buried dioxin-con-
taminated soil and the future inhabitants of the EDA.
(Dr. Janick Artiola-Written Preliminary Comments, pp. 6-7
and 9.)
Does the absence of samples deeper than 2 inches raise any
concern? (Dr. Carl Shy-Written Preliminary Comments,
p. 7.)
RESPONSE: Dioxin is not readily biodegradable, but it does undergo
photolysis in the presence of hydrogen donors and ultraviolet
light. It is strongly bound to the organic matter in soils; typi-
cally, the organic fraction of the soil would decrease with in-
creasing depth.
6-3
-------
The selection of the top 2 inches of soil was based on a
standard protocol used by EPA Region n for dioxin sampling
in residential areas to determine if there are levels of concern
in the accessible surface soils. This protocol has been
reviewed and approved by the CDC and the Agency for Toxic
Substances and Disease Registry (ATSDR). The level of
concern for 2,3,7,8-TCDD is based on exposure to "acces-
sible" residential surface soils containing concentrations of
dioxin at or above 1.0 ppb. The health effects associated with
dioxin and the relevant exposure patterns for soil, particular-
ly those for children, were considered in establishing the level
of concern.
The level of concern established by Kimbrough et al., sum-
marized in Appendix P, was specifically developed for use at
Missouri dioxin sites, although it applies generally to dioxin
sites in residential areas, since exposure patterns are typical
of residential areas and are not site-specific. Different levels
of concern are applied for accessible surface soils at commer-
cial and agricultural sites.
It is understood that adults gardening, etc., would come into
contact with soil at depths greater than 2 inches. Adult resi-
dents would also come into contact with soils at depths of
several feet if they were transplanting trees or doing construc-
tion-related activities. However, the top 2 inches of soil
presents the portion of the soil column that adults and children
most frequently come in contact with, and thus is a more ap-
propriate depth to reflect long-term exposure and chronic ef-
fects. Appendix Q summarizes the assumptions used to
calculate the level of concern.
6.1.2 Assumed Hot Spot Size
COMMENT: The assumed size of a hot spot seems large unless a great deal
of soil material was transported. (Dr. Randall Charbeneau--
Written Preliminary Comments, p. 5.)
6-4
-------
RESPONSE: The size of the target hot spot was discussed extensively by
the TRC. The mechanism of transport of dioxin into the EDA
is not known. In fact, at the time of sampling, it was not
known if any dioxin was present or had ever been present in
the soils of the EDA sampling areas. It was known that
(1) dioxin had been found in the EDA sewers and creeks and
(2) analysis of soil borings taken around the perimeter of
Ring n did not indicate the presence of dioxin. Based on
anecdotal information, it was theorized that material may
have been taken from the Canal and used to fill in low-lying
residential areas. If this were true, and the fill did contain
dioxin, then this may have been a mechanism of transport.
The selection of the hot spot size was based on the intent of
the study, the possible fill transport mechanism, and the
habitability study schedule. A much smaller target hot spot
size would have greatly increased the number of samples re-
quired. A primary consideration in the schedule was the
laboratory capacity; a significantly larger number of samples
could have delayed the study results and the overall
habitability study schedule. The number of sampling points
required is directly proportional to the area of the target hot
spot. To have the same probability of detecting a hot spot
one-fourth the area of the target size used in the study, four
times as many samples would have been required (ap-
proximately 10,000). The sampling was initiated in Novem-
ber 1986, and the final Quality Assurance (QA) of the
analytical results was not completed until March 1988.
COMMENTS: What is the justification for assuming that fill dirt would be
uniformly applied to all surface areas of a lot? (Dr. Michael
Stoline--Written Preliminary Comments, pp. 4,8.)
What is recommended if the assumption that fill was uniform-
ly applied over an entire lot cannot be made? (Dr. Michael
Stoline--Written Preliminary Comments, p. 4.)
RESPONSE: This was not specifically assumed. If a lot was contaminated,
all that was assumed was that the contamination would be
6-5
-------
above the 1.0-ppb level of concern and that these levels would
be reliably detected.
6.1.3 Tilting of the Sampling Grid
COMMENT: What was the reason for the seemingly arbitrary selection of
13 degrees for tilting of the sampling grid? (Dr. Janick Ar-
tiola-Written Preliminary Comments, p. 2.)
RESPONSE: An offset rotation of the sampling grid with respect to the
predominantly north-south, east-west layout of the streets
was chosen to avoid having entire rows or columns of the grid
overlaying a street or a set of houses. The offset was an ar-
bitrarily chosen prime number. A prime number was used
because it would be less likely to have a submultiple that
would interact with the street layout.
6.1.4 Statistical Design
COMMENT: How does the statistical design handle large numbers of non-
detect samples? (Dr. Janick Artiola-Written Preliminary
Comments, p. 8.)
RESPONSE: The sampling design for this study is based on the size of the
locally contaminated area it is designed to detect. The prob-
ability of detecting the contamination depends only on the
grid size (assuming that a sample taken from a locally con-
taminated area will contain detectable levels of contamina-
tion) and its relationship to the target size of the locally
contaminated area.
6.2 METHODS
6.2.1 QAPP Documentation and Implementation
COMMENT: How were the training and qualifications of all field and su-
pervisory personnel checked? Was there a specific review of
all subcontracting personnel made by the major contractor as
6-6
-------
part of the QAPP? (Dr. Janick Artiola--Written Preliminary
Comments, p.2.)
RESPONSE: The resumes of personnel from associate firms were checked,
particularly for staff with training in the required level of
protection and with training or experience in soil sampling.
COMMENT: The QAPP does not give guidelines for the definition of the
sod layer or mention the necessary presence of properly
trained personnel able to make such separation. Clearly, the
use of trained soil scientists during sample collection would
have been recommended. (Dr. Janick Artiola-Written
Preliminary Comments, p. 3.)
RESPONSE: We assumed that a precise separation of the sod layer was not
needed for the purposes of the dioxin study since the sam-
pling technique followed the standard protocol used by
Region II for dioxin sampling in residential surface soil. In
this sampling technique, the sod layer is removed by cutting
the sod layer vegetation (or live plant tissue) and lifting. If
there was no sod layer at the sample location, the surface soil
was sampled. This approach was taken after lengthy discus-
sion with the CDC and ATSDR.
COMMENT: The EPA acceptance windows for the Performance Evalua-
tion (PE) samples should have been included in the QAPP in
a more explicit format. At a concentration of 1.0 ppb, the PE
sample analysis results must fall within the range of 0.40 to
1.3 ppb (99 percent). According to Table 6-3 in Volume IV,
there are at least two values that appear to fall outside of this
range. It is not clear if the PE and other samples were
reanalyzed, which appears to be the requirement according to
Appendix E of the QAPP. The upper value of the range for
the 1.2-ppb target was 2.1, which is 64 percent higher than
the target value. The range for acceptable analysis of the PE
samples was difficult to find. (Dr. Joan Daisey—Written
Preliminary Comments, pp. 1,2, 5.)
6-7
-------
RESPONSE: The criteria for the PE sample results are advisory. Table
R-l of Appendix R summarizes the CLP QC criteria for
dioxin analysis.
COMMENTS: Some pages are missing from the beginning of Appendix C.
The revisions to the QAPP do not include any corrections.
(Dr. Joan Daisey-Written Preliminary Comments, p. 4.)
In general, the QAPP and revisions to the QAPP were not as
well written and organized as they might have been. The
criteria for the QA/QC plan should have been detailed ex-
plicitly in a table in an appropriate section. Some of the
figures in the original QAPP, e.g., A-5, were missing; others
were mislabeled. Certain things are still not clear even after
reading all three documents; e.g., was the form shown in
Figure A-11 of the original QAPP used for the Novem-
ber/December 1986 sampling? (Dr. Joan Daisey-Written
Preliminary Comments, p. 5.)
The documentation of this study is not as well organized or
as clear as it should be. It is essential that what was done be
clearly and completely documented. The QAPP documents,
in particular, should be revised and edited. (Dr. Joan Daisey-
-Written Preliminary Comments, p. 10.)
The QAPP and revised QAPP documents are difficult to un-
derstand and follow. The QAPP should be more "user-friend-
ly." Some guide to the use of the documents is recommended.
(Dr. Joan Daisey—Peer Review Meeting Comments, June 20,
1988.)
RESPONSE: It was determined that an incomplete copy of the QAPP had
been provided to Dr. Daisey, and a corrected copy was sup-
plied to her.
The QAPP for the soil assessment for 2,3,7,8-TCDD was dif-
ficult to follow for two primary reasons:
1. The document was revised during implementation to cor-
rectly reflect the changes in the sample homogenization
6-8
-------
procedure. Only the changed portions were included in
the revised version.
2. The field sampling effort was designed and implemented
by CH2M HILL and the laboratory analysis was con-
ducted through the EPA CLP. The appropriate portions
of the CLP User's Guide and SOP were included in the
QAPP as appendices and were not paraphrased and in-
tegrated into the main body of the QAPP. This may have
caused some confusion; however, the authors of the QAPP
felt it inappropriate to reword or paraphrase the CLP.docu-
ments.
To make the QAPP more user-friendly, a guide to the use and
organization of the QAPP has been prepared that will be in-
serted behind the title page of each future copy of the QAPP.
This guide is provided in Appendix R.
6.2.2 Chemistry
COMMENT: The statistical operating characteristics of the analytical pro-
cedure should be provided. The available information is in-
sufficient to determine whether the analytic procedure would
be able to reliably detect a concentration of 1.0 ppb and dis-
tinguish it from a lower concentration. (Dr. David Schoen-
feld-Written Preliminary Comments, p. 1.)
RESPONSE: A performance evaluation study was conducted by EMSL-
LV on the CLP analytical method for the analysis of 2,3,7,8-
TCDD in soils and sediments. The results of this study
indicated that the precision of the method can be characterized
by a percent relative standard deviation (RSD) of 7.4 percent
for fortified field blanks, and 12.6 to 23.1 percent for PE
samples. The study concluded that little, if any, bias exists.
Qualitatively, the method of performance is characterized by
a false positive rate of 1 to 4 percent and a false negative rate
of 2 to 3 percent at a concentration of 0.8 to 1.5 ppb. The
report on this performance evaluation is provided in Appen-
dix S.
6-9
-------
6.2.3 Numbering of Samples
COMMENT: It is not clear how the PE samples were numbered. Were these
samples given a random number that looked like the sample
numbers for the Love Canal samples? Could the analyst iden-
tify a PE sample from the number? (Dr. Joan Daisey-Writ-
ten Preliminary Comments, p. 2.)
RESPONSE: PE samples were numbered so that the analyst could not dis-
tinguish them from field samples.
COMMENT: Sample numbering is not clear from the QAPP documents or
Volume IV. How many numbers were associated with each
collected sample? How well were the numbers on the dif-
ferent forms and labels cross-checked? A table listing all
documents, labels, tags, etc., generated for each sample would
have been very helpful. (Dr. Joan Daisey-Written Prelimi-
nary Comments, p. 4.)
RESPONSE: Each sample was assigned one sample number. The forms
and labels were computer-generated and cross-checked as
part of the sample management program. The documents,
labels, and tags are described in the QAPP.
6.2.4 Data Review
COMMENT: Appendix C of Volume IV states that all analytical data were
evaluated by Region II analysts/reviewers trained in GC/MS
analysis and having specialized knowledge of dioxin
analysis. This document is ambiguous. A clearer statement
of exactly what was reviewed and an explicit statement that
the analyses met or did not meet the criteria should have been
included. (Dr. Joan Daisey—Written Preliminary Comments,
p. 4.)
RESPONSE: Analytical results for each sample and batch of samples were
reviewed using a checklist of criteria from the CLP SOP,
which is provided in Appendix C of Volume IV.
6-10
-------
6.2.5 Audits
COMMENT: It is not clear in Appendix C whether the CLP laboratories
were audited prior to any Love Canal sample analysis.
(Dr. Joan Daisey-Written Preliminary Comments, p. 3.)
RESPONSE: Laboratories analyzing dioxin under the CLP contract are
audited before samples are received and then annually there-
after. Eagle Picher, KCSI, and TMS were audited before they
received samples. ETC had participated in the CLP program
under a previous contract and, therefore, their performance
had been audited previously. ETC was audited again in
March 1987, approximately 2 months after it received
samples. ECAL analyzed dioxin samples under a special ser-
vices contract and had been audited by EPA Region n prior
to receipt of samples for other dioxin work. (See p. 2 of
Appendix C, Volume IV.)
COMMENT: The major substantive change in the QAPP was the sample
homogenization method. The change was reasonable;
however, since there was no field audit during the May 1987
field sampling, there is no information on whether the 2-
minute mixing protocol was followed. (Dr. Joan Daisey-
Written Preliminary Comments, p. 7.)
RESPONSE: The modified homogenization procedure (hand-mixing) that
was used and audited at the homogenization laboratory was
also used in the field. Page A-14 of Appendix A of the
revised dioxin QAPP states:
"5. Mix for 2 minutes with stainless steel spoon."
While no formal audit was conducted in May 1987, a form
outlining the 11-step procedure to be used for collection and
homogenization of each sample was signed and dated by the
sample collector after each sample was collected and
homogenized.
6-11
-------
6.3 RESULTS
6.3.1 Moving of Sampling Points
COMMENT: Why were some sampling points located beyond the 35-foot
limit rather than simply dropping the node, as suggested in
the QAPP documents? (Dr. Michael R. Stoline-Written
Preliminary Comments, pp. 6,8.)
RESPONSE: Less than 4 percent of the samples collected were located out-
side the 35-foot limit as a result of field activities that were
not detected until post- fieldwork QC review was completed.
The original grid nodes associated with these samples were
resampled where possible. The results of both analyses were
included in Volume IV.
In most cases, samples tied to grid nodes that had to be moved
more than 35 feet because of unsampleable surfaces
(i.e., pavement, house, etc.) should have instead been
dropped according to the plan. However, in some cases a
point was moved and sampled to satisfy the two-sample-per-
lot criterion. The alternative was to use an arbitrarily located
sampling point to satisfy the two-sample-per-lot criterion for
that lot. It was felt at the time that it was better to have a point
tied to a grid node than arbitrarily located. As shown in
Table 6-1,90 points were moved further than 35 feet from the
associated grid node, and 63 of these were satisfactorily
resampled at the grid node. For the remaining 27, it was not
possible to sample within 35 feet of the grid node for the fol-
lowing reasons:
1. Driveways, garages, or ponding areas on residential lots
were encountered at the designated grid node location.
The point was moved to the nearest soil surface which, in
some cases, was more than 35 feet from the grid node.
2. The designated grid node was located on a rooftop. Ac-
cording to the sample design, such nodes were to be
moved 5 feet from the structure. In some cases, this was
more than 35 feet from the designated grid node location.
6-12
-------
Table 6-1
DISTANCE ACTUAL SAMPLING POINTS WERE MOVED FROM GRID NODES
SOIL ASSESSMENT-2,3,7,8-TCDD
Distance Moved
Not Moved
0 to 5 feet
>5 to 15 feet
>1 5 to 25 feet
>25 to 35 feet
>35 feet
TOTAL SITES
Number of sites moved >35 feet that were resatn
Number of Sites
567
939
429
260
125
_2Q
2,410
pled 63
Number of sites moved >35 feet that were not resampled 27
-------
3. The grid node was located on paved areas such as road-
ways or parking lots. The point was moved to the nearest
soil surface which, in some cases, was more than 35 feet
from the grid node. In most of these cases the sampling
point was moved to the yard of a property where only one
grid node was located; this was done to provide the second
sampling point to fulfill the two-sample-per-lot criterion.
6.3.2 Modifications During Implementation
COMMENT: A short discussion on any changes or modifications that may
have been made in the course of the implementation of the
QAPP is suggested. (Dr. Janick Artiola-Written Prelimi-
nary Comments, p. 4.)
RESPONSE: All changes are addressed in the revised QAPP. The only
change during implementation was the change in the
homogenization procedure, which is discussed on page 5-1
of Volume IV.
COMMENTS: Effects of the sampling month, rainfall, and changes in the
sampling plan should have been evaluated. The original
schedule that called for all samples to be obtained in Novem-
ber/December 1986 was modified, resulting in samples being
collected in November/December 1986 and May and
July 1987. Reasons for this modification are not adequately
explained or considered in the analysis. Why was the month
of June skipped? Was there a marked difference in rainfall
over these months, such that dioxin might have been dis-
persed more in one of the months? More discussion should
be presented on the reasons for and circumstances surround-
ing this change in the sampling schedule, and some interpreta-
tion should be made of whether the change had any potential
impact on the results obtained. (Dr. Carl Shy—Written
Preliminary Comments, pp. 3-4,6,7.)
The results should be shown by month and by type of sample,
e.g., original sample, archive sample, QC sample. Because
of the differences in the mixing technique for samples col-
6-14
-------
lected in different time periods, the results of the analyses
should be displayed separately for the November/December
and the May and July samples. (Dr. Carl Shy-Written
Preliminary Comments, pp. 3,5,7.)
Was there any difference in the proportion of samples with
positive dioxin values for the two time periods? (Dr. Carl
Shy-Written Preliminary Comments, p. 3.)
A table summarizing the results by concentration by month
is preferred. (Dr. Carl Shy-Peer Review Meeting Com-
ments, June 20,1988.)
Are the trace levels at all associated with month of sampling,
homogenization method, or rainfall patterns? (Dr. Carl Shy-
-Written Preliminary Comments, p. 6.)
RESPONSE: Sampling was stopped from December to May because of in-
clement weather. As a result, approximately 600 samples
remained to be collected. In addition, a QA check of the
sample coordinates was made, and it was discovered that
several points had been moved more than 35 feet from the
original grid node. These grid nodes were resampled along
with the remaining 600 points during the sampling conducted
in May. After a QA of the sample coordinates taken in May,
a few points still had been moved more than the 35-foot limit.
This, coupled with the fact that more properties were now
available for sampling [Love Canal Area Revitalization
Agency (LCARA) had purchased more properties], led the
TRC to a decision to collect additional samples during July.
It was not expected that these changes would affect the
results, since the solubility of dioxin in water has been es-
timated to be approximately 32 parts per trillion (U.S. EPA,
1987) and dioxin binds tightly to soil organic matter. These
physical characteristics suggest that dioxin that has been in
the soil for several years should not be significantly affected
by these factors.
6-15
-------
Table 6-2 shows the results, by month collected, of the dioxin
samples. As can be seen, there is no apparent association of
higher concentrations with the time of year that the sample
was collected. Therefore, it can be inferred that the weather
or the different seasons did not affect the dioxin results.
6.3.3 Statistical Interpretation
COMMENT: An unanswered question remains: Was the 95 percent prob-
ability of detect lowered by moving some of the sample
points? (Dr. Janick Artiola~Written Preliminary Comments,
P-8.)
RESPONSE: The sampling program was designed to have a 95 percent
probability of detecting a locally contaminated elliptical area
approximately 126 feet long by 66 feet wide. This ellipse has
an area of approximately 6,500 square feet, which is the
median lot size in the EDA.
During the implementation of this sampling plan, several
sample locations had to be moved because of the infeasibility
of sampling the assigned location (pavement, structure, etc.}.
An estimate of the retrospective probability of detecting a lo-
cally contaminated area of 6,500 square feet was developed
by using Monte Carlo Simulation. The coordinates of the
EDA neighborhood boundaries and the points sampled were
used as a data base by the simulation. The computer program
estimated the probability of detecting a locally contaminated
area by generating a simulated locally contaminated area and
placing it at a random location and orientation on the map of
the EDA. If the simulated area was within the EDA and con-
tained a sampled location, it was counted as a hit. If the simu-
lated area was within the EDA but did not contain a sampled
location, it was counted as a miss.
The ratio of hits to misses is an empirical estimate of the prob-
ability of detecting a contaminated area. This estimate has a
slight high bias because 14 of the 2,274 points sampled had
no valid results. The estimate has a low bias because the EDA
6-16
-------
Table 6-2
SUMMARY OF SAMPLE RESULTS BY SAMPLING MONTH
SOIL ASSESSMENT-2,3,7,8-TCDD
Concentration
Range
(PPB)
Sampling Locations
With Highest Result
In Concentration Range
Total Results Using
All Reported Concentrations
for Each Location
ND
<0.2
*0.2and<0.4
i 0.4 and<0.6
^ 0.6 and<0.8
^0.8and<1.0
*1.0
Did Not Pass QC
Total Samples
Collected
Nov
645
16
2
0
0
1
0
664
8
672
Dec
934
20
1
0
0
0
0
955
2
957
May
594
4
2
0
2
0
1
603
4
607
Jul
38
0
0
0
0
0
0
38
0
38
Total
2,211
40
5
0
2
1
1
2,260
14
2,274
NOV
752
22
2
0
1
1
0
778
Dec
1,034
22
1
0
0
0
0
1,057
May
629
5
2
0
2
0
5
643
Jul
39
0
0
0
0
0
0
39
Total
2,454
49
5
0
3
1
5
2,517
-------
contained areas of 6,500 square feet or larger that did not have
accessible soil (i.e., the area was paved) but were not removed
from the simulated EDA. The estimated probability of
detecting the target-size locally contaminated area is 88 per-
cent. The details of the simulation can be found in Appen-
dix T.
COMMENTS: Those charged with making the habitability decisions should
at least be aware of how the probability of detecting a hot spot
changes when the size of the hot spot is only a fraction of the
size of a median EDA residential lot (Dr. Michael Stoline-
Preliminary Written Comments, p. 4.)
Emphasize that any dioxin-contaminated fill areas smaller
than the size of a median-size lot have a much less than 95 per-
cent chance of being detected with the statistical sampling
design used in this study. (Dr. Janick Artiola-Written
Preliminary Comments, p. 8.)
What is the probability that the sampling plan would detect
dioxin in an area one-third the size of a lot? (Dr. Michael R.
Stoline--Written Preliminary Comments, p. 4.) „
RESPONSE: Figure 6-1 shows the relationship between ellipse dimensions,
the grid size, and the probability of finding (or detecting) the
ellipse. The variables shown in the figure are the following:
L = Length of the semi-major axis of the ellipse
G = Grid size
S = Length of the minor axis divided by the length of the
major axis of the ellipse
B = 1 minus the probability of detecting the target-size
contaminated area
The paragraphs below describe the calculation used in the
sampling design and the subsequent calculation of the
retrospective probability of finding a target hot spot one-third
the size of the design target size.
6-18
-------
1.00
0.80
0.60
ft
0.40
0.20
0.00
0.00
Square sampling grid
0.10
0.20
0.30
0.40
0.50
L/G
Artwork reprinted with the permission of Chemical Engineering Magazine. Vol. 91, 1984.
0.60
0.70
0.80
0.90
1.00
Figure 6-1
SOIL ASSESSMENT-^, 3, 7. 8.-TCDD
DESIGN OF SAMPLING GRID
-------
Design: For the design, an ellipse with an area of ap-
proximately 6,500 square feet was used. (This ellipse was
selected to represent the median size of a residential lot in the
EDA.) A design criterion of 95 percent probability of detect-
ing the target-size ellipse was also used. The variables had
the following values:
L = 63 feet
G = ? (object of calculation)
S = 0.52 (=66/126)
B = 0.05 (= 1 - 95%)
As shown in Figure 6-1, with the above values assigned for
the variables, the L/G ratio must be about 0.91. That is, the
grid size (length) must be about 10 percent longer than the
length of the semi-major axis. The semi-major axis of the tar-
get ellipse is 63 feet; thus, a grid size of 69 feet was selected.
An area one-third that of a median-size lot: For an area one-
third that of the target size, the length of the ellipse is ap-
proximately 58 percent of the original ellipse (square root of
1/3 = 0.58). The variables used in the figure for this deter-
mination are as follows:
L = 36.5 feet (0.58 * 63)
G = 69 feet
L/G = 0.53
S = 0.52 (proportion of the ellipse remains constant)
B = ? (object of calculation)
As shown in Figure 6-2, the value of B for the above-assigned
values is about 0.55, which makes the probability of detect-
ing a target area this size approximately 45 percent (1 minus
0.55).
6-20
-------
1.00
0.80 -
0.60 -
ft
0.40 -
0.20 -
0.00
Square sampling grid
0.00 0.10
0.20 0.30
0.40
0.50
L/G
Artwork reprinted with the permission of Chemical Engineering Magazine. Vol. 91,1984.
Figure 6-2
SOIL ASSESSMENT -- 2,3,7,8-TCDD
PROBABILITY OF DETECTING LOCALLY CONTAMINATED AREA
ONE THIRD THE SIZE OF THE TARGET AREA (1 MINUS B = 45%)
-------
6.3.4 Interpretation of the Results
COMMENT: The reader should be given more details about the circum-
stances, if any are known, that might account for this elevated
dioxin concentration. Is there any evidence that fill dirt from
the Canal was spread on this site? Will deeper soil samples
be obtained from this site? (Dr. Carl Shy-Written Prelimi-
nary Comments, p. 6.)
RESPONSE: No visual evidence of a source was found. A summary of the
results of an effort by NYSDOH to research the site is at-
tached (Appendix U). Additional sampling, which has in-
cluded soil samples taken at depths varying from 2 inches to
12 inches, has been performed, and the sampling results are
summarized in Appendix V.
COMMENT: An underlying assumption of the study was that the recoveries
of the MS samples would reflect the recoveries of the EDA
samples. If the recoveries of the EDA samples were actual-
ly lower, then there could be more sites that should be clas-
sified as equal to or greater than 1.0 ppb. Neither the QAPP
nor Volume IV adequately addresses the issue of how the
recoveries and associated uncertainties apply to detect values
in the range of 0.5 to 0.9 ppb. The issue of .the recovery of
2,3,7,8-TCDD from the EDA samples is a remaining uncer-
tainty. (Dr. Joan Daisey-Written Preliminary Comments,
pp. 8,9.)
RESPONSE: Table 6-3 lists the MS recoveries associated with samples that
had a dioxin concentration between 0.5 ppb and 1.0 ppb.
Only four analytical results were in this range. Two of those
results had high MS recoveries (1.2 ppb versus 1.0 ppb) in-
dicating that estimated values for that batch, if biased, were
more likely to be biased high than low. The other two con-
centration values were associated with the same sample. For
the first analysis of this sample, the result was a0.63-ppb con-
centration with a 0.95-ppb MS result. A followup analysis
by high resolution mass spectrometry, which was requested,
indicated a sample concentration of 0.92 ppb; since this was
a special followup confirmatory analysis, an additional MS
was not analyzed.
6-22
-------
Table 6-3
SUMMARY OF CONCENTRATIONS MEASURED FROM 0.5 ppb to 1.0 ppb
AND THEIR ASSOCIATED MATRIX SPIKE RESULTS
SOIL ASSESSMENT~2,3,7,8-TCDD
Station No.
1932
1936
5234
DBNO.
030110
030108
016309
026413
Type of
Sample
Original
Original
Original
Archive
Initial Result
(PPb)
0.62
0.61
Rejected
0.63a
Initial
Matrix Spike
Result (ppb)
1.2
1.2
Rejected
0.95
Rerun Result
(PPb)
N/A
ND (MPC=0.04)
ND (MPC=0.36)
ND(MPC=0.19)
0.92"
Rerun
Matrix Spike
Result (ppb)
N/A
0.90
0.95
0.91
N/A
Analyzed by high resolution MS technique
-------
COMMENT: The sampling plan is based on the assumption that there was
no movement of dioxin from the Love Canal by water
transport, e.g., flooding and/or rainfall runoff. Since the land
is fairly flat, this seems reasonable. However, the clustering
of the detects in the contiguous sectors 6,7,8, and 9 suggests
the possibility of water transport Has this been considered?
How would this affect the validity of the statistically based
sampling plan? (Dr. Joan Daisey-Written Preliminary Com-
ments, p. 10.)
RESPONSE: The results have been mapped. Given the fact that only one
location sampled had dioxin at concentrations above the level
of concern and low levels of most of the dioxin detects [rela-
tive to the Maximum Possible Concentrations (MPCs) for the
non-detects and relative to health concerns], it does not ap-
pear that a sophisticated analysis of the detect data is merited.
6.3.5 Additional Information/Discussion Needed
COMMENT: No explanation is given for the statement that the concentra-
tion of dioxin in the original sample of the one soil specimen
that exceeded 1.0 ppb could not be quantified. It would have
been highly desirable to have comparable data for the original
and archive samples from this site. (Dr. Carl Shy—Written
Preliminary Comments, p. 6.)
RESPONSE: Two analyses of the original sample were rejected because of
low internal standard (IS) recoveries (4 percent and 5 per-
cent). The estimated values for these sample analyses were
17.5 and 13.4 ppb, respectively. The low IS recovery was ac-
counted for in calculating the sample concentration. The es-
timated concentrations of 13.4 and 17.5 ppb are in relatively
close agreement with the five archive sample results (17.3 to
21.2 ppb).
COMMENT: The report describes retrieval of 279 archive samples but fails
to state the basis for selecting the 170 archive samples and
30 associated QC samples that were sent to a new laboratory
for QC checks. Both the selection criteria and rationale for
6-24
-------
these analyses should be stated. (Dr. Carl Shy-Written
Preliminary Comments, p. 5.)
RESPONSE: A QA review of the first few batches of samples analyzed in
the program indicated that a laboratory was having trouble
successfully analyzing the samples. The EPA QA officer
asked the laboratory to rerun the samples. In addition, the
QA officer decided to perform an additional QC check on the
laboratories participating in the study.
It was determined that the 60 archive samples associated with
the sample batches that were rejected because of failed PEs
should be analyzed by a separate laboratory. Since, at that
time, most of the sample results were non-detects, an effort
was made to improve the chance for detects in this QC check
so that a QC comparison could be made. Therefore, 60 addi-
tional archive samples selected from Sampling Area 1 were
sent for analysis. (Pilot studies had shown that a higher fre-
quency of detectable levels of soil LCICs occurred in this
area, and it was therefore assumed that the probability of
detecting dioxin in samples from this area may be higher.) In
addition, 50 archive samples were selected randomly to
provide further comparison results. Overall, a total of 170 ar-
chive samples were equally distributed into 10 coolers and
were packaged and shipped in the same manner as regular
samples (i.e., each batch consisted of 16 field samples, 1 field
replicate, 1 Shipping and Storage Blank (SSB), 1 MS, and
1 PE).
COMMENT: Did the difference in homogenization procedures result in dif-
ferent sample rejection rates over the two time periods?
(Dr. Carl Shy—Written Preliminary Comments, p. 3.)
RESPONSE: Sample rejections were based on various factors that were not
related to the homogeneity of the sample but were instead re-
lated to laboratory performance. (Many of the sample
analyses were rejected because of failed QC sample analyses;
most of these samples were reanalyzed.)
6-25
-------
COMMENT: It would be useful to document the reasons for owner refusal
to participate in the sampling study. (Dr. Michael Stoline--
Written Preliminary Comments, p. 8.)
RESPONSE: Reasons for homeowner refusals for each of the environmen-
tal assessment programs are contained in Appendix D.
COMMENT: The data on the results (recoveries) of the MS analyses should
be presented in Volume IV. (Dr. Joan Daisey—Written
Preliminary Comments, p. 2.)
RESPONSE: The range and median of the results of the MS analyses are
given in Volume IV, p. 6-6.
COMMENT: There is insufficient information on the nature of the soil used
for the MS and PE samples. What was the source of the soil?
What were the criteria used to determine that it was com-
parable to the soil from the Love Canal area? Were soil par-
ticle size distributions and moisture contents similar to the
real samples? Were the appearances of the MS and PE
samples similar to those of the real samples or could the
analyst identify the PE sample by appearance? (Dr. Joan
Daisey--Written Preliminary Comments, p. 2.)
RESPONSE: PE samples were obtained from EMSL-LV and NUS through
EPA Region II. The PE samples obtained from EMSL-LV
were spiked sand or clay samples. The PE samples from NUS
were native soils from an EPA Region JJ dioxin site in the
Newark, New Jersey, area. For the MS samples, Soil Con-
servation Service (SCS) charts were used to locate soil similar
to EDA soil in Ipswich, Massachusetts, since the blanks were
being prepared there. The SCS groups soils as similar based
on distinctive soil relief and drainage. The soil physically
resembled EDA soil. Information on particle size distribu-
tion and moisture content of the EDA soils was not available
to further match the Ipswich soil. Analyses were conducted
on representative samples of the blank soil, and the results in-
dicated no dioxin.
6-26
-------
COMMENT: It is not clear why the discrepancies between the plotted sam-
pling point lot locations and the field-observed sampling
point lot locations occurred. This should be clarified.
(Dr. Joan Daisey-Written Preliminary Comments, p. 4.)
RESPONSE: The sampling points were surveyed to a permanent U.S.
Geological Survey benchmark, and the base map was not
The base map was created using aerial photographs, was
field-checked, but was not verified by comparison with
ground-truth. The property boundaries were based on tax
maps obtained from the City of Niagara Falls.
6.4 OTHER
6.4.1 Issues Related to the Habitability Decision
COMMENT: We must revise our previous assumptions to allow for the ex-
istence of several hot spots with the same amount of surface
area as the one that was discovered. (Dr. David Schoenfeld-
-Written Preliminary Comments, p. 2.)
RESPONSE: The contaminated area that was found during the dioxin soil
assessment is estimated to be less than 5 feet in radius; this is
based on subsequent sampling of the area. One hypothesis is
that this area was contaminated by an ascribable cause. If this
is the case, the type of design assumptions that were used in
the assessment sampling may be inappropriate for a con-
taminated area of this size.
However, if an ascribable cause is not found and the con-
taminated area is assumed to be representative of a random
occurrence in the EDA, an estimate of the maximum prob-
able frequency of randomly occurring 5-foot-or-less-radius
hot spots can be made.
If contaminated areas of 2 feet in radius are distributed ran-
domly throughout the EDA and it is assumed that the
2,260 samples collected (with valid analytical results) were
6-27
-------
located randomly (not unreasonable since the hypothesized
contaminated area is so small relative to the sample grid
design), then the point estimate of the probability of locating
such an area is 1 in 2,260 or 0.0004425.
The 95 percent upper confidence bound on this estimate is
obtained from:
L=((X+l)*F(0.05,dfl,df2))/(n-x+(x+l)*F(0.05^fl/i£2))
where df 1=2* (X+l)=4
df2=2* (N-X)=4518
X=1,N=2260,F(0.05,4,4518)=2.37
and L=0.002094. (Zar, 1984, p. 378)
N=0.002094*2260=4.7 (or about 5 hits for
2,260 samples)
This estimate indicates that if the contaminated areas are ran-
domly distributed over the EDA, another 2,260 samples
would have a 95 percent probability of finding fewer than five
such areas.
COMMENT: If the EDA is declared habitable, will the new dwellers be
relegated to keeping their yard activities within the top
2 inches? (Dr. Janick Artiola—Written Preliminary Com-
ments, p. 9.)
RESPONSE: The study design called for measurements of dioxin in the top
2 inches of soil to determine if dioxin was present above
1.0 ppb. For the purposes of this study, and consistent with
the CDC level of concern, it was assumed that activities
would not be restricted to the top 2 inches of soil if the levels
of dioxin were below 1.0 ppb.
COMMENT: The discussion section is inadequate and unsatisfactory. No
effort is made to interpret for the reader the meaning and sig-
nificance of the distribution and levels of positive dioxin soil
samples. Are these less-than-l.O-ppb samples a random find-
ing typical of an uncontaminated neighborhood? Do they
6-28
-------
represent traces of dioxin that were present at higher con-
centrations? Are the trace levels at all associated with sample
collection teams? (Dr. Carl Shy—Written Preliminary Com-
ments, pp. 6-8).
RESPONSE: The data are not available to determine whether the less-than-
1.0-ppb samples represent typical random findings. This
study was not designed to address this issue but rather to com-
pare the results to the CDC 1.0-ppb level of concern.
COMMENTS: A conclusion is needed: Do the authors believe that the
results suggest no problem with dioxin in soil? (Dr. Carl Shy-
-Written Preliminary Comments, pp. 6-8.)
How safe is the EDA for children playing? How safe is the
EDA for gardening activities? An approach to answering
these questions is to estimate the median-size play area for a
child in the EDA and to estimate the median-size garden area
for a typical EDA resident. Assume that fill dirt, if used,
would be uniformly spread over the typical play area or typi-
cal garden area, and determine the probability of detecting
hot spots of both of these sizes. How habitable is an area that
contains subareas, however small and isolated, that are known
to be contaminated or known to have been contaminated?
What does this say about the quality of life attainable?
(Dr. Michael Stoline-Written Preliminary Comments, pp. 4,
10,11.)
The results of this study should be explained in terms that can
be easily understood. (Dr. David Schoenfeld—Peer Review
Meeting Comments, June 20,1988).
RESPONSE: Appendices P and Q summarize the assumptions and ration-
ale for the 1.0-ppb level of concern for dioxin in residential
soils. The level of concern assumes uniform and average con-
tamination of soil at 1.0 ppb. The public health risk as-
sociated with exposure to soil is a complex combination of
the level of contamination, the size of a contaminated area,
the type of use of the area, the length of time persons are using
the area, and the age of persons using the area.
6-29
-------
The risk associated with exposure to a small area of elevated
contamination decreases as the area gets smaller. The con-
taminated area that was found in the EDA had dioxin up to
17.0 to 35.0 ppb in an area estimated to be less than 5 feet in
radius. Such an area (80 square feet) comprises about 1 per-
cent of the area of a median-size lot in the EDA. Thus, long-
term exposure to soil from a lot that might have such an area
of contamination poses a lower risk than if the entire lot were
contaminated at 1.0 ppb. If the contaminated area were the
focus of a child's play area or a home garden, the dioxin ex-
posure from soil may be higher than if the lot were uniform-
ly contaminated at 1.0 ppb. However, the chance that such a
contaminated area exists and would be the focus of such ac-
tivities is quite small.
At the request of EPA Region n, ATSDR reviewed the results
of dioxin sampling and prepared a health consultation. This
consultation is presented in Appendix W and concludes:
• The investigation for 2,3,7,8-TCDD contamination is
adequate to define the potential exposure of humans
posed by the chemical's presence in the area.
• There is no widespread 2,3,7,8-TCDD contamination in
the surface soil within the EDA that might present any
threat to human health for the residents.
COMMENT: Clearly discuss the chances of exposure to dioxin contamina-
tion as a function of target hot spot size and sample depth.
(Consensus Comment-Peer Review Meeting, June 20,
1988.)
RESPONSE: As discussed in Section 6.3.3, the likelihood of detecting a hot
spot decreases with the size of the hot spot. Figures 6-1 and
6-2 showed the relationship between hot spot size and the
probability of detection. Basically, the smaller the hot spot
the less likely it will be to find it with the same-size sampling
grid. However, it should be noted that the risk associated with
exposure to a hot spot also decreases with the size of the hot
spot. This is because the probability that a person will inter-
act with the hot spot decreases with the size of the hot spot.
6-30
-------
Only the top 2 inches of soil were sampled. The top 2 inches
were selected to sample the most accessible soil; this was
based on EPA Region n standard procedures for dioxin sam-
pling in residential soil. The followup depth sampling at the
location with a concentration above the level of concern in-
dicated that the concentration of dioxin decreased with depth;
this would support the decision to sample the top 2 inches if
it were assumed that the mode of transport at the hit location
was the only potential mode of transport. However, because
none of the potential modes of transport have been eliminated,
no conclusions can be drawn about the concentration of
dioxin at depths greater than 2 inches.
COMMENT: Discuss the chronic exposure conditions assumed by the CDC
in the derivation of the 1.0-ppb level of concern, as these re-
late to the types of exposures that might be expected to occur
in the EDA. (Consensus Comment-Peer Review Meeting,
June 20,1988.)
RESPONSE: The assumptions used by the CDC in deriving the 1.0-ppb
level of concern for dioxin in residential soils are summarized
in Appendix Q. The 1.0-ppb level of concern and assump-
tions used to arrive at this level of concern are considered to
be applicable to residential areas in general and, therefore, ap-
plicable to those types of exposure that might be expected to
occur in the EDA. Also, the CDC and the ATSDR were con-
sulted and concurred on the study design.
6.4.2 Documentation
COMMENT: A large body of data sheets, notebooks, analytical informa-
tion, etc., has been generated in this study. What is the ul-
timate fate of this documentation? The habitability criteria
states that all of the environmental data from these studies
will be made available to the general public and the scientific
community for analysis and interpretation. (Dr. Joan Daisey-
-Written Preliminary Comments, p. 10.)
RESPONSE: This information will be stored by EPA and will be available
to the public through an FOIA request to EPA Region JJ.
6-31
-------
REFERENCES
CH2M HILL. 1986. Love Canal Dioxin Soil Sampling Study Quality As-
surance Project Plan.
CH2M HILL. 1987a. Love Canal Full-Scale Air Sampling Study Quality
Assurance Project Plan.
CH2M HILL. 1987b. Love Canal Habitability Study- Soil Sample Collec-
tion and Preparation Quality Assurance Project Plan (Final Revised Ver-
sion).
CH2M HILL. 1987c. Love Canal Habitability Study-Soil Sample Labora-
tory Analysis Quality Assurance Project Plan.
CH2M HILL. 1987d. Pilot Study for Love Canal EDA Habitability Study.
(Vol. land II)
CH2M HILL. 1987e. Revisions to Love Canal Dioxin Soil Sampling Study
Quality Assurance Project Plan.
CH2M HILL. 1987f. Summary of Responses to The Peer Review of The
Pilot Study For The Love Canal EDA Habitability Study. (Vol. I and II)
CH2M HILL. 1988a. Love Canal Emergency Declaration Area
Habitability Study, Vol. II: Air Assessment-Indicator Chemicals.
CH2M HILL. 1988b. Love Canal Emergency Declaration Area
Habitability Study, Vol. Ill: Soil Assessment- Indicator Chemicals.
CH2M HILL. 1988c. Love Canal Emergency Declaration Area
Habitability Study. Vol. IV: SoilAssessment-2^,7,8-TCDD.
Gilbert, Richard 0.1987. Statistical Methods for Environmental Pollution
Monitoring. Van Nostrand Reinhold Company, New York.
ICAIR and CH2M HILL. 1988. Love Canal Emergency Declaration Area
Habitability Study, Vol. I: Introduction and Decision-Making Documenta-
tion.
R-l
-------
Kimbrough, Renate D.; Henry Falk; Paul Stehr, and George Fries. 1984.
"Health Implications of 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) Con-
tamination in Residential Soil." Journal of Toxicology and Environmental
Health, 14, pp. 47-93.
NYSDOH and DHHS/CDC. 1986. Love Canal Emergency Declaration
Area; Proposed Habitability Study.
U.S. EPA. 1987. The National Dioxin Study.
Zar, Jerrold H. 1984. Biostatistical Analysis. Prentice Hall, Englewood
Cliffs, New Jersey.
R-2
-------
APPENDIX A
Final Summary of Modifications to TAGA
Procedures, April 15, 1988
Air Assessment—Indicator Chemicals
Prepared by
U.S. EPA Environmental Response Branch
Environmental Response Team
Edison, New Jersey
WDC 63394.T1 (SA)
-------
Final Summary of Modifications to TAGA
Procedures Used During the Love Canal
Full-Scale Air Sampling Study
April 15, 1988
FROM: Thomas H. Pritchett
ERT QA/QC Coordinator (Love Canal)
Environmental Response Branch
U.S. Environmental Protection Agency
-------
INTRODUCTION
From the period from July 1987 to December 1987 the Environmental
Response Teams (ERT's) mobile mass spectrometer / mass spectro-
meter (the TAGA) conducted indoor air analyses at Love Canal as
part of the overall Love Canal Emergency Declaration Area (EDA)
Full Scale Habitability Study. The analyses were performed using
the procedures outlined in the Summary of the TAGA Standard
Operating and Reporting Procedures (SORP) for the Love Canal
Full-Scale Air Sampling Study which was Appendix A of the Love
Canal Full-Scale Air Sampling Study Quality Assurance Project
Plan (QAPP) prepared by CH2M Hill in July 1987 with the procedu-
ral modification outlined below.
The indoor air analyses were performed the ERT's EERU contractor
(Enviresponse, Inc.) in Phase 1 of the study and by the ERT's
REAC contractor (Roy F. Weston, Inc.) in Phases 2-4. Between
Phases 1 and 2 of the study the ERT changed its operations assis-
tance contract from the EERU contract to the REAC contract.
However, because Roy F. Weston hired for its REAC contract all
the old EERU personnel involved in the study, no key personnel were
lost during the contract transfer. All changes in key personnel
(primarly that of the Sampling Crew Chief) were made by on-site
training of the replacement by the replaced personnel. Thus the
data quality was never adversely affected by the change in ERT
contractors.
MODIFICATIONS TO TAGA ANALYTICAL PROCEDURES
The original TAGA analytical procedures were contained in the
Summary Of The TAGA Standard Operating Procedures For The Love
Canal Full-Scale Air Sampling Study (SORP), which was Appendix A
of the QAPP. At the start of the study this SORP was a set of
procedures and reporting formats which had never been tested
under a full set of field conditions. Consequently, based upon
the actual field experiences and data, several typographic errors
in criteria were discovered (e.g., the reversal in the Instrument
Startup Checklist of the maximum acceptable temperatures for
cryoshells #2 and #3), some procedures were modified and improved
after sufficient field data was gathered (e.g., after Phase 1 the
changing from a heated Teflon sampling line to unheated Teflon
tubing), and several data criteria were modified to reflect
lessons learned (e.g., the changing of the ID criteria to be
applied to data above the quantitation limit).
Such changes, modifications, and improvements are to be expected
whenever "state of the art" analytical methodologies and proce-
dures are first utilized in the field. Because of the time
constraints built into the sampling effort, a pilot field trial
of new analytical protocols was by-passed for this program and
draft procedures were finalized. Had sufficient time been
available for either one to two months of full trials at Edison
or a practice mobilization, many of these modifications would
have been made prior to their being finalized in QAPP.
-------
All such procedural and data criteria modifications were first
approved by the ERT QA/QC coordinator prior to their implementa-
tion. Also, all such changes were also documented by memorandum
to the Love Canal files written by the ERT QA/QC coordinator.
These memorandum are included as References in the final report
on the air analyses. Copies were also given to the Region II
Project Manager, the CH2M Hill Task Manager, and the ERT QA
Analytical Operations Director. In addition, the changes which
went into effect prior to the start of the Phase 2 analyses were
summarized in a memorandum dated September 14, 1987 written by
the ERT QA/QC Coordinator to the Region II Project Manager.
These modifications are summarized below in the approximate chro-
nological order of the first application (as documented in the
original memorandums to file). In a very few cases ambiguities
also occurred between the original memorandum and the more brief
description contained in the September 14 summary. The below
listing also clarifies these ambiguities.
1) Detection Limit Criteria
During the first mobilization a modification was made to the
decision tree used to determine whether an LCIC was detected.
This modification was initially made to take into account the
artificially high instrument noise levels which were encountered
during the first phase of monitoring. It resulted in higher
confidence in any reported levels of LCICs detected. The new
decision tree was as follows:
As an initial screen the individual ion pair signal were
first compared against the ion pair detection limits (de-
fined by the zero signal plus two times the standard devia-
tion). If any ion pair signal (in ion counts per second)
was below its detection limit, then the compound was con-
sidered not detected. (This initial screen is the test
specified in the QAPP.) However, if all three signals were
above their detection limits, then the corresponding concen-
tration values were then computed and the concentration
reported by the most sensitive ion pair was then compared
against the compound detection limit. The LCIC was not
considered to be detected unless the concentration computed
from the most sensitive ion pair was above the compound
detection limit.
2) The Use of a. Single Calibration for Successive OA/QC Analyses
During Phases 1-3 two QA/QC analyses, each requiring prior
calibrations, were performed successively at the end of alternate
days. These analyses were the precision canisters and the detec-
tion limit / quantitation limit verifications. The documentation
requirements for these analyses implied that separate analyses
were required for each. This implied procedure was changed such
that the calibration performed for the first analysis could be
used in reducing the data for the second analysis provided that
-------
the two analyses were performed with no delay between the two.
This modification resulted in more efficient utilization of
instrument time and had no affect on the data quality.
Starting in Phase 4, three quantitative QA/QC analyses were being
performed at the end of every other day. These analyses were a 6
Liter Summa performance evaluation canister analysis, a 16 Liter
Summa precision canister analysis, and the detection limit &
quantitation limit verification. Past data had shown that by the
end of the day the TAGA response factors were stable over a
ninety minute time period. Therefore, one calibration was
allowed for all three analyses provided that the analyses were
completed within ninety minutes. Since the 16 Liter performance
evaluation canister analysis and quantitation verification analy-
sis both involved accuracy checks against known spiked concentra-
tions, one of these analyses had to be the last analysis in order
to again verify the assumed stability of the instrument response.
This modification allowed the TAGA analysts to more efficiently
utilize the available instrument time.
3) Criteria Applied in Deciding To Restart TAGA Sampling After
Corrective Actions Had Been Taken As a. Result of Either an
Excessive Response Factor Decay or an Unacceptablv High LCIC
Detection Limit
After any corrective actions had been taken as a result
of either of the above two problems, a new calibration was per-
formed to determine whether the instrument now had the required
sensitivity. If it did not, then further corrective actions was
warranted. However, .if the instrument sensitivity was within
specifications, then the response factors from this calibration
were entered into the Daily Summary of Response Factors (Figure
4.3.7 of the TAGA SORP) but no % RF decay was computed. A second
calibration was then performed in order to ensure that the
instrument sensitivity was no longer significantly decaying and
that there was no danger of the instrument sensitivity decaying
to the point that the resulting detection limits would again be
unacceptable. The response factors from the second calibration
were also entered in the daily summary (% RF decay computed) and
were used as the calibratior for the next house or QA/QC analysis
provided that the starting of that analysis was pending the
completion of the corrective action.
This procedure was further modified, but not documented, by the
TAGA group during Phase 2. The second, post-corrective-action
calibration was dropped from the procedure provided a) that the
decay either had been reversed or had decreased to less than 15%
and b) that the new detection limits computed from the first
post-corrective-action calibration were less than 2 ppb. In
other words, once the appropriate corrective action was taken,
only one set of calibration data, not two sets, was used to
determine whether the TAGA should continue sampling.
-------
Although this field modification was made without the proper
documentation, it did not affect the overall data quality. In
all but two cases when this further modification was used the
first post-corrective-action calibration indicated an increase in
the instrument's sensitivity (i.e., the response factor decay was
reversed). In the other two cases, the decay reversed itself by
the next house and the instrument detection limits were less than
50% of their maximum allowable values.
4) Use of a Gas Standard Cylinder in Place of Unique £. Liter
Summa Canisters as Performance Evaluation Samples
This replacement was required during Phases 1 and 2 of sampling
due to uncertainties encountered with the known concentrations
of the initial two batches of Summa canisters from Northrop
services. It was not until Phase 3 that the team leader from
Northrop Services had sufficient confidence in the "known" con-
centrations for these canisters to be used as performance
evaluation samples. Because this was the first time thet RTP
had prepared performance evaluation samples for these
compounds, these initial problems were not unreasonable and are
the price of using non-field-proven "state of the art"
methodologies in such an intensive program.
Standard gas cylinders, which were analyzed by the TAGA using the
procedures developed for the 6 Liter Summa canister, became the
daily QC sample used to certify the daily accuracy of the TAGA.
In all performance evaluation analyses different gas standard
cylinders were used for the calibration and the actual perform-
ance evaluation analysis. These standard gas analyses could not
be considered to be an absolute check of the TAGA's accuracy; a
bias in the standard gas cylinder certifications could not have
been detected since the calibrations and cylinder analyses would
have had the same bias. Therefore, Northrop Services challenged
the TAGA group with a 16 Liter Summa canister audit sample during
each phase. Extra precautions were taken by Northrop Services to
ascertain the LCIC concentrations. During all four phases the
TAGA's results were within the accuracy criteria. Thus, the data
quality was not adversely affected during Phases 1 and 2 by the
necessary shift to the Scott standard gas cylinders as the
accuracy check samples.
5) Change in the Identification Criteria to Be Applied Whenever
an LCIC Concentration Was Greater Than the Quantitation Limit
Whenever an LCIC was detected above its quantitation limit, the
original QA/QC plan in the SORP specified that the ratios of the
various ion pair signals would be computed and that these ratios
would then be compared against the appropriate ion ratio ranges
calculated during the previous calibration. This criteria was
designed to test for the presence of potential interferences to a
given ion pair. This was the first time that ion ratios had been
applied to TAGA air monitoring data. Unfortunately, during the
quantitation limit verification, the performance evaluation, and
the precision canister analyses in Phase 1, this criteria repeat-
-------
edly failed to correctly identify the presence of pure compounds.
Therefore, the use of ion ratios was discarded as a means of
testing for interferences. A new and better set of identifica-
tion criteria was developed from the quantitation limit verifica-
tion data from the first eight days of Phase 1. This new cri-
teria never failed to correctly identify the pure, spiked LCIC
during all subsequent quantitation limit verifications. The new
criteria was as follows:
For any LCIC present above its quantitation limit, concen-
tration values were first computed from the signal for each
ion pair and tnen an average concentration was computed.
Next, the software computed the percent deviations of each
individual ion concentration from the overall average. The
compound was considered to be not interfered with if all of
the above percent deviations were less than or equal to 20%.
If any individual percent deviation exceeded 20% then the
compound was considered to be interfered with and the
reported average concentration would have been flagged
accordingly. No such interferences were detected during
this study.
The ID criteria was initially specified at 15% in the original
memo dated August 17. This criteria was widened to 20% in the
September 14 summary memo. The 20% criteria was used throughout
the sampling program.
Implicit in this modification of the ID criteria were changes in
the criteria to be applied during the data review and data vali-
dation and changes in several of the predefined data reporting
•spreadsheets. The data review / validation criteria lib of Table
7.1 of the SORP was changed to "for each quantitation limit
verification analysis check to insure that the reported data met
the new ID criteria". The following forms were modified as a
result of this ID criteria change:
Figure 1 - The LOTUS Summary of Calibration and Detection Limit
Data (Figure 4.3.6 of the SORP).
Figure 2 - LCIC Summary Data for Stationary Monitoring (Figure
4.6.5 of the SORP).
Figure 3 - LCIC Summary Data for Cylinder/Canister Analyses
(Figure 4.7.2 of the SORP).
Figure 4 & 5 - Verification of Detection & Quantitation Limits By
Serial Dilution (Figure 4.8.1 of the SORP). This form was,
in essence, three separate forms designed to report data
from three interrelated experiments: a) spikes below the
reported detection limits, b) spikes between the detection
and quantitation limits, and c) spikes above the detection
limits. When this form was modified between Phases 1 and 2,
it was broken into three separate but similar spreadsheets.
However, because of a change in air sampling pumps between
the two phases, the detection limits for the last three
phases were always too low for the TAGA group to be able to
reliably spike below the reported detection limits. There-
fore only the forms for parts b and c of the verification
-------
were generated during the last three phases of the study.
The new versions for parts b and c are found in figures 4
and 5, respectively. The two "chlorotoluene concentrations"
found in the bottom half of the form should read "chloro-
benzene concentrations". The error was seen but never
corrected in the LOTUS macro program since it did not affect
any of the computed values.
A new version of the "Calculation of Intermediate RFs, DLs, and
Resulting % Error Bars" (Figure 4.3.8 of the SORP) was also
prepared but has not been included since it was not included in
any of the final data packages. (Although the three calibrations
were performed during the initial survey and investigative survey
of the one house with a detectable level of an LCIC, the response
factors were so similar that no intermediate response factors
were used in reporting the data. The difference in the reported
concentrations were less than the least significant digit of the
reported value.)
6) Other Data Forms Which Were Modified During the Study
Two other forms were improved during the study. Figure 6 con-
tains the new version of the Data Flag Versus Sequence Number
Table (Figure 4.6.4 of the SORP) which is illustrated in figure
6. This offset allowed the operator to convert the sequence
number at which the air was sampled at the head of the hose to
the sequence number corresponding to the signal generated when
that sampled air actually reached the TAGA. In most cases, these
offset sequence numbers were used to reduce the stored data. One
of the only exceptions, during which an offset sequence was not
used in reducing the data, was the start sequence for the first
location sampled during a file when the hose had been stationary
for several minutes (e.g., the initial ambient air analysis for a
given house). This improved form was first used during Phase 1
of the study.
The Summary of Results From Precision Analyses (Figure 4.7.3 of
the SORP) was also modified. The new form, which is contained in
Figure 7, summarizes the results from both LCICs in one table
rather than two and includes the results from all cylinder and
canister analyses from that phase which could be used to evaluate
the precision. It also calculates the %error of the average
concentration versus the "known" concentration. The new form
does not include the dilution data since that data is contained
in the LCIC Summary Data for Cylinder/Canister Analyses (Figure
3) which are included in the daily data packages.
7) Modification to the Procedures Used in Analyzing the Precision
and Performance Evaluation Samples
The procedure for analyzing this samples given in section 4.7 of
the SORP stated that "Immediately after initiating the data
acquisition, the operator will open the cylinder/canister and
will start feeding the gas through the dilution system." This
procedure was changed to allow the operator to start feeding the
-------
sample gas to the dilution system prior to starting the data
acquisition. This change did not affect the data quality but did
allow the TAGA crew to save valuable time in performing the
analyses. The justification for this change is discussed in
detail in the above-mentioned September 14 memo.
8) Use of an Unheated Teflon Line in Sampling a. Given House
Method development performed between the first and second phases
determined that no loss of either LCIC occurred when an unheated
Teflon line was used at ambient air temperatures as low as 0°C.
Therefore, due to the various handling problems associated with
the heated transfer lines, unheated Teflon lines were used. The
use of an unheated transfer line greatly decreased the time spent
at each sampled residence. Additionally, because an unheated
lines was used during the % Transport Efficiency Analyses, the
outside ambient temperature was recorded on the TAGA Operator's
Logsheet for each test performed. The outside ambient tempera-
ture was also recorded at each house and these temperatures were
checked against the lowest temperature for which a set of %
Transport Efficiencies had been obtained.
9) Spike Concentration Used When the Detection Limits' of the Two
LCICs Differed bv More Than Ippb
Whenever the detection limits of the two LCICs differed by more
than Ippb, then the detection limit and quantitation limit deter-
minations were performed at the lowest possible dilution which
was within 1 and 2 ppb of the higher of the two detection and
quantitation limits, respectively. The original set of operating
procedures had not considered the possibility that either the
detection limits or the quantitation limits for the two LCICs
would differ by more than 1 ppb. This modification addressed
that situation.
10) Maximum Acceptable TAGA starting Vacuum Pressure
The original SORP specified that the TAGA must have pumped down
to at least 2 X 10A-7 torr before the TAGA crew could start
preparing the instrument for that day's analyses. This criteria
was based upon the manufacturer's recommendations and upon a two
hour shift in startup times for each day. According to the
manufacturer starting analyses before this pressure was reached
could both shorten the usable life of the instrument for that day
and could result in a severe decay in the instrument sensitivity
towards the end of the day. Starting in Phase 2, it was not
always possible to obtain this manufacturer's maximum recommended
starting pressure with the shift to fixed start times for three
continuous days at a time. On several days during the last three
phases the start time of the analyses would have had to be
delayed in order to meet this criteria. However, the existing
QA/QC plan was designed to quickly detect the very problems the
maximum acceptable starting pressure was designed to help prevent
(changes in instrument response). Therefore, the maximum accept-
able TAGA starting vacuum criteria was replaced by the judgment
-------
of the TAGA Senior Scientist. At no time during the study was an
analytical problem ever traced back to the TAGA group starting up
the instrument before the maximum acceptable pressure was
obtained.
11) Alternate Method for Computing Detection Limits When Ambient
Air Was Potentially Contaminated by LCIC
During the last two phases of the program, the background ambient
air occasionally gave rise to elevated signals for a given LCIC.
Once during the last phase the ambient signal was even greater
than the detection limit. In several occasions, an LCIC signal
actually dropped when the TAGA sampling line was moved into the
houses. The detection limit computations developed for this QAPP
were based upon the assumption that the LCICs were unique to Love
Canal and were not normally present in the ambient air in the
EDA. Elevated signals during the calibration zero point determi-
nations would bias the computed detection limits high and could
have potentially resulted in false negatives. A total of fifteen
calibrations were affected - 5 during Phase 3 and 10 during Phase
4. Therefore, an alternate procedure was developed for dealing
with these infrequent elevated signals.
The alternate procedure was as follows:
First, a clean calibration was Selected for the effected
LCIC from the calibration data for that day. The "clean"
calibration was defined as the closest calibration in which
the average zero concentration signal for the affected LCIC
was within one standard deviation of the average signal for
the unaffected LCIC. The alternate detection limit was then
computed by dividing the original detection limit by a
correction factor. This correction factor was computed by
averaging individual ion pair ratios which were calculated
by taking ion pair response factors from the "clean" cali-
bration and dividing them by the response factors from the
affected calibration. Thus, if the instrument sensitivity
had doubled from the "clean" calibration to the affected
calibration (i.e., the response factors had increased by a
factor of two), the alternate, corrected detection limit
would be half of the detection limit computed from the
"clean" calibration.
Because this was not the approved detection limit calculation,
the values reported in the final report were those computed using
the original calculations. However, these values are flagged and
the appropriate tables, which can be found in the report Refer-
ence section, are designated. In all cases the appropriate house
data were compared against both sets of detection limits. There
were no LCIC levels which would have changed to a "detect" using
the alternate detection limits from a "non-detect" using the
original limits. Therefore, this potential bias was considered
but no false positives were found among the fifteen houses with
the affected detection limits.
-------
12) Data Review Activity Loasheet
The SORP stated that this logsheet would be formalized. However,
it was found that the each data review crew worked better using
their own informal checklists. These checklists were too cryptic
for interpretation by non-TAGA users but were more than adequate
to insure that the data review was completed in a thorough
manner. This effectiveness was evidenced by the lack of signifi-
cant errors and omissions having to be caught by the data valida-
tion step. Therefore, no formal data review activity logsheet
was ever developed.
13) Data Validation Activity Logsheet
This logsheet was again specified as "Yet to be Completed" in the
SORP. This logsheet was formalized into a two part form (Figures
8 & 9). The logsheet served two purposes. The first part (Figure
8) summarized the QA/QC results for t?iat day's analyses and
compared the reported QC results against the appropriate Data
Quality Objectives. The second part (Figure 9) listed any data
discrepancies which had the potential of affecting the data
quality. Each item listed was then specifically reviewed by the
ERT QA/QC Coordinator in order to determine the necessary correc-
tive action. The actions included A) no action needed - data
quality not affected, B) Data Review action required in order to
resolve and document the noted discrepancy, C) explanatory memo
required from the ERT QA/QC coordinator documenting his profes-
sional judgment on the degree, if any, to which the data quality
was affected, and D) withdraw the affected data package from the
overall package; no corrective action possible except for
resampling.
14) No End-of-Phase SAF & MFC Calibration Checks for Phases 1 &. 3.
Table 6.3.1 of the SORP states that the SAF and MFC calibrations
were to be rechecked prior to the TAGA demobilizing from each
phase, it did not define the time period which corresponded to
"before demobilizing". Consequently, in Phases 1 & 3 the last
set of MFC and SAF calibration checks were not performed on the
last day of the phase. In Phase 1, the last calibrations checks
were performed on the twelfth day of a fourteen day sampling
effort and in Phase 3 they were performed between the sixth and
seventh days of a nine day effort. In both cases, the measured
calibration errors never exceeded 7% (Phase 3) and 8.6% (Phase
1) for the SAF and MFC calibration checks. Also, when these
calibrations were first checked in the subsequent mobilization
the maximum errors were 8.22% (Phase 2) and 3.61% (Phase 4) for
the MFC and SAF calibration checks. Although the measured
errors from these calibration checks did vary throughout the
program, they never exceeded the 10% criteria. Therefore, it
would have been extremely unlikely that either calibration had
drifted outside of the 10% criteria during those last two days
of each phase, especially when they were inside the criteria
during the first check of the subsequent phases. Consequently,
it is highly unlikely that these two deviations from the
10
-------
calibration check schedule adversely affected the data quality.
CHANGES IN THE PHASE-SPECIFIC DOCUMENTATION
Section A of Table 9-1 of the SORP describes the required, addi-
tional documentation for each two week sampling effort. These
documents were beyond those which would be included in the house
and daily data packages. All of these documents have been moved
to other locations within the total data package as described
below.
The documentation for the following items have all been placed in
the data package immediately prior to the data from the first day
of the first mobilization: the initial certification and the
recertifications of the standard cylinder concentrations by Scott
Specialty Gasses (Figure 3.4.1 of the SORP) and by Northrop
Services, the certification of the rotameter calibration (NBS-
traceable), and the certification of the Gilian PFS 500 Micro-
processor Flow Calibrator. The latter two certifications, which
were only documented once, were applicable to the complete
sampling effort. However, the certifications were indirectly
reconfirmed for each phase by the independent MFC and SAF flow
audits which were performed by Northrop Services. These audits
were always performed on the same days that the SAF and MFC
calibrations were checked by the rotameter and MFC, respectively.
The standard cylinder recertifications performed by Northrop
Services were only officially documented once during the program.
All other communications of the recertification results were
through verbal communications and informal memos. Because of the
time required for each analysis the actual recertifications of
each of the standard cylinders by Scott Specialty Gasses were
interspersed throughout the total program. In other words, only
selected cylinders were recertified by Scott prior to each phase.
All of these recertification results are summarized in Tables 37
& 38 of Volume 1 of the TAGA group's final report. This summary
also indicates when each recertification analysis was performed.
Consequently, the documentation for all of the individual Scott
recertification analyses are presented together with the documen-
tation of the Northrop recertifications. These documents are in
Appendix 1.
The documentation of the initial SAF and MFC calibration checks
were moved into the data package for the first day of that parti-
cular phase while the summary of the results from the precision
analyses were included as the last item for that particular
phase. This form (Figure 4.7.3 of the SORP) was modified and
these modifications are discussed above in section 6. An example
of the new form for reporting this data is contained in Figure 7
of this report.
Two sets of documents were dropped from the packages: the summary
TAGA Operating Logsheets (Figure 4.1.1 of the SORP) and the
Cylinder/Canister Chains of Custody (Figure 3.4.2 of the SORP).
11
-------
The summary logsheets were dropped since new, individual log-
sheets were filled out each day. These individual logsheets were
included in each daily data package. The chains of custody were
dropped because their use became sporadic after Phase 1. The
identity of each canister was clearly labeled and there never
arose an occasion in which a chain of custody was needed to
resolve which canister was being analyzed. Additionally, because
of the uniqueness of these canister samples, it would not have
been possible for any company or any other government agency to
have supplied us with Summa canisters containing the LCICs at the
10 - 25 ppm spiked concentrations during this study. Therefore,
the lack of proper chains of custody for all of these canisters
does not affect the data quality.
CLARIFICATION OF THE AMBIGUITIES AND
CORRECTION OF ERRORS IN THE ORIGINAL
CRITERIA SPECIFIED IN THE SORP
There were several ambiguities and errors in the SORP procedures
and criteria which became apparent during the sampling. These
were caused by the fact that the QAPP was finalized before many
of the procedures described in the SORP were fully field tested.
These errors and their corrections are summarized below:
1) The maximum allowable percent decay in the response factors
is reported as 15% in Table 4-1 of the main body of the QAPP but
as 30% in Table 6.3.1 and the text of Appendix A of the QAPP.
This discrepancy resulted from a change in the QAPP not being
incorporated into the SORP. The correct value was 15%.
2) The SORP dad not specify whether the decay criteria was to be
applied to the average decay or each individual ion pair response
factor. The criteria was applied to the average decay.
3) The 10% criteria listed for "Perform Periodic Calibrations"
in Table 6.3.1 of the SORP can not be applied to the actual
calibrations. This accuracy criteria was actually the criteria
of the SAF and mass flow controller calibrations. It was the
accuracy of those two sensor calibrations that, along with the
accuracy at which the standard gas concentrations were known,
determined the overall accuracy of the TAGA calibrations.
4) In the verification/criteria listed in Table 7.1 of the SORP
criteria 12 and 13 have figures misidentified. Specifically, in
criteria 12 Figure 4.7.2 should be Figure 4.8.1 and in criteria
13 Figure 4.8.1 should be Figure 4.9.2.
5) In Figure 4.1.4, the instrument startup checklist, the re-
quired temperatures for items four and five were reversed.
Specifically, the required temperature for cryoshell Temperature
#2 should have been <.75°K and the required temperature for Cryo-
shell Temperature #3 should been _<18.5°K.
12
-------
6) Throughout the SORP two different source pressure ranges were
used (0.9-1.1 torr versus .950-1.050 torr). The 0.9-1.1 torr
range was considered the proper criteria range.
7) The SORP did not specify which of the two concentration
values (Average versus Actual) reported in the results of the
Verification of Detection & Quantitation Limits analysis (Figure
4.8.1) should have been within 1 ppb and 2 ppb of the detection
limits and quantitation limits respectively. Since the "Actual
Concentration" was the actual diluted spiked concentration, it
was always the concentration compared against the quoted detec-
tion and quantitation limits.
8) The SORP failed to specify that the accuracy of the TAGA was
also to be evaluated during the quantitation limit verification.
This was an error of accidental omission which was caught by the
data validation crew during the first phase of sampling. The
percent error in the calculated "Average concentration" was com-
puted for each quantitation limit verification to ensure that the
accuracy of the TAGA at its quantitation limit was within the
specified 25%. The percent error was calculated as follows:
%error = [average - actual',/actual X 100.
9) While the SORP used integer values throughout all of its
criteria, the actual computed values for the various criteria
were computed as real numbers with fractional components. There-
fore, prior to a computed criteria being compared to the SORP
specified value, the computed criteria was always rounded to the
closest whole number using the standard rules for rounding.
10) Criteria 6 in Table 7.2 of the SORP specified that the
entries in the TAGA chronological logsheet should have been
consistent with the appropriate entries in the TAGA logsheet.
One of these entries was the source pressure reading. The gauge
which measured this reading was only precise to an hundredth of a
torr and had a typical measurement noise of 0.002 torr. The SORP
never specifically addressed this issue and the matter was raised
during the data validation. The criteria was further defined
such that the two readings were considered equal if they were
within 0.004 torr from each other.
11) The quadrupole warmup time criteria listed in the Instrument
Startup Checklist accidentally was given as <.30 minutes both on
the sheets and in the SORP. The correct value was >.30 minutes.
12) Data Quality Objective #2 in Table 6.1.1 of the SORP defined
the maximum allowable precision of the TAGA analyses to be 25% as
measured by "(% error in the reported concentration)". In addi-
tion, criteria 10 of Table 7.1, which outlines the data review
checks, also includes a check on the accuracy of the analysis.
Since this is the same wording as was used in defining the accu-
racy criteria in Data Quality Objective #1 of the same table, the
wording implies that for each precision analysis the percent
error of the reported concentration from the known value was to
13
-------
be compared against this precision criteria and that this compar-
ison would determine the overall precision of the TAGA. However,
Figure 4.7.3 in the SORP, which shows the format that was to be
used in reporting the precision results, indicate that a percent
relative standard deviation (RSD) was to be computed. This
ambiguity arose from the fact that the data was being used for
two related but different purposes: l) an internal check or
control analysis (accuracy evaluation) and 2) a repetitive sample
of constant concentration (precision evaluation). In the data
packages, the percent errors were calculated in the data summary
reports (Figure 3 of this report) for each analysis and a RSD was
computed for each set of precision analyses (Figure 7 of this
report).
The RSDs were used in both the TAGA group's final report and my
summary of the TAGA's performance to evaluate the TAGA's preci-
sion against this criteria. Each precision analysis was also
treated as an internal control sample and a percent j. -or was
computed. However, the comparisons of these percent error
results were included in the discussions of the TAGA's accuracy
versus the accuracy criteria of "less than a 25% error".
Because of the affect this ambiguity had on the review of the
TAGA's performance against the data quality objectives, I dis-
cussed it with Bill Coakley, the ERT's Quality Assurance Officer.
He concurred that only a standard deviation or a relative stan-
dard deviation were appropriate measures of precision. He also
concurred with the use of the percent error of these analyses as
another measure of the overall accuracy of the TAGA (Data Quality
Objective #1).
ADDITIONAL PROCEDURES ADDED
Several additional QC procedures and data validation steps were
added to the field analyses. These procedures were designed to
further improve the overall data quality even beyond what was
originally specified in the QAPP. These additional procedures
are as follow:
1) Check of Accuracy at the Ouantitation Limit
Although never specified in the QAPP or the Data Quality Objec-
tives, it was implicit in the design of the quantitation verifi-
cation that the quantitative accuracy of the TAGA should be
checked whenever a quantitation limit verification occurred.
Therefore, starting in Phase 2, the Verification of Quantitation
Limits spreadsheet (Figure 5) was modified to calculate the error
in each quantitation limit verification spike analysis. Both the
data review and data validation groups were then tasked to check
the reported error against the 25% accuracy criteria. In all
cases the reported values met this accuracy criteria.
14
-------
2) Additional Documentation of the Correct Sampling Air Flow and
Collision Gas Thickness Values
In Phase 2, the TAGA senior scientist noticed that the Sampling
Air Flow (SAF) and Collision Gas Thickness (CGT) readings occa-
sionally were affected when the TAGA initiated a data-stored-to-
file acquisition but not during the normal no-storage acquisi-
tions. He further determined that the settings themselves had
not changed but that the problem was apparently just a computer
gliche. Therefore, he initiated the inclusion of real-time, no-
storage, ion signal versus time trace (Figure 10) to each data
package. From then on, this sheet was used to get the correct
SAF, CGT, and discharge current (DI) readings.
15
-------
FIGURE 1
ciinuTioi in WICIIQI LIIIT im
Diti: 1-DIC-I7
Tin: Ol:10:)t
riliiui: UC400)
•••••••^••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••MM
CilorotoUtit
Cue. W.O/ fl.l
ppbr
0.0
2.0
1.1
1.1
If.)
21. 1
iepi
11
20f
lOf
OIS
IfOO
2f)2
idl
iepi
7
IS
IS
)f
4)
103
ir 101. ton
j-iitept •!
12
iittoiitj
ppb
Iitintitj of
Cblorotolam
•*«•*•*«•««»
1.317H5
O.ftSf
25
0.2
12f.O/ fS.O
iepi
11
S7
Of
103
)St
Sff
21.4111!
f.lffll)
O.fttl
Ditietiu Liiiti
2)
1.1
id2 121
iepi
f
IS
12
If
20
)t
tbi loit uniitifi ioi pair it eeipoud DL
qmicicitioi
(••••••••••s*l
liiit (ppbl*
(••••••••••••••••MMICMI
).2
•••••••••I
.O/ fl.O
iepi
7
70
1)1
21!
SIS
077
)).2I22
7.J2157!
O.fttO
13
O.I
lOt.l
>••••••••••••
id)
iepi
j
1)
20
22
)l
)7
•*••••••*
Cblerobusiii
Cue. 112
ppb f
0.0
2.0
4.1
0.1
If.)
2f.l
ir
y-iatept
12
iituiitr
PPb
.fl/ 77.0
iepi
11
112
345
71S
1)71
2211
is.jom
7.112152
o.ftH
2f
t.)
Idl
iepi
I
it
37
3t
fl
t4
112. O/ Sl.O
iepi
12
11)
270
sst
1151
17)f
15.1271
f.107127
l.tttl
Diuetioi lliiti
22
1.)
Id2 114. O/ 77.0
iepi
S
22
)f
jj
It
40
iepi
13
f2
12f
23t
1)7
7)0
27.4f77)
J.7577J2
O.ftt)
2S
l.t
Id)
iepi
f
0
13
If
30
4f
Iitiaiitj of tbi uit iinitifi iei pair it eoipeud 01 77.f
Cblerobiaiiii qiutititioi liiit Ippbl* 2.7
-------
FIGURE 2
LCZC SUMMARY DATA FOR STATIONARY MONITORING
CLE: LC4004
END SEQUENCE
LOCATION:
604
INTENSITY
ID
CTOL
CTOL
CTOL
PM/DM
126. O/ 91.
126. O/ 65.
128. O/ 91.
RF DL
101.91 0.25
21.42 1.07
33.24 0.39
AVG
15
14
13
SO
13
12
12
AVERAGE CHLOROTOLUENE CONCENTRATION
PPB %DEV FR
AVG AVG CONC
0.15 62.95%
0.65 64.53%
0.39 1.57%
0.4
AVERAGE % DEVIATION FROM CONCENTRATION AVG 43.02%
DETERMINATION
ID
CBEN
CBEN
CBEN
PM/DM
112. O/ 77.
112. O/ 51.
114. O/ 77.
RF
85.30
65.63
27.47
AVERAGE CHLOROBBNZBNB
INTENSITY
DL AVG
0.30 16
0.34 14
0.91 12
CONCENTRATION
SO
27
12
12
PPB
AVG
0.19
0.21
0.44
0.3
Qflj
%DEV FR
AVG CONC
32.83%
23.61%
56.44%
AVERAGE % DEVIATION FROM CONCENTRATION AVG
DETERMINATION
37.6%
-------
FIGURE 3
LCIC SUMMARY DATA FOR CYLINDERXCANISTER ANALYSES
FILE: LC4001
PRECISION CYLINDER AAL16555
START SEQUENCE: 220
END SEQUENCE:
CYLINDER ID:
332
AAL16555
SAF RDG:
MFC RDG:
INTENSITY
ID
CTOL
CTOL
CTOL
PM/DM
126/
126/
128/
91
65
91
RF
97.54
19.82
30.46
AVG
1234
275
406
SD
113
52
69
AVERAGE CHLOROTOLUENE DILUTED CONCENTRATION
ORIG.
CHLOROTOLUENE
CONCENTRATION
IN CYLINDER
1503 ml/sec
35 ml/Bin
PPB %DEV FR
AVG AVG
12
13
13
13
34
.7
.9
.3
.3 ppb
.2 ppm
CONC
4
4
0
.8%
.5%
.3%
AVG % DEVIATION FROM DILUTED CONC
%ERROR IN FINAL CONC
3.2%
INTENSITY
ID
CBEN
CBEN
CBEN
PM/DM RF
112/ 77 80.60
112/ 51 62.05
114/ 77 25.45
AVERAGE CHLOROBENZENE DILUTED
ORIG.
AVG
1004
787
338
CONCENTRATION
SD
106
88
60
CHLOROBENZENE CONCENTRATION IN CYLINDER
AVG % DEVIATION FROM
% ERROR IN
UPPER LOWER
FINAL CONC
PPB %DEV FR
AVG AVG
12.5
12.7
13.3
12.8 ppb
33.0 ppn
2.5%
10.7%
CONC
2.7%
1.0%
3.7%
-
-------
FIGURE 4
miriciTici or omcrioi t QDUTITITIOI
IIIITI IT lUIU DILUTION (cut)
nil: LC4020
DHICTIOI UD QOUTITIIIOI HUT VUiriClTIOl
OIICII1L CTIIIDII CILOIOTOLDIIB COICIITtiTIOl Ippil 21.33
ouciiiL CTIIIDII ciLoioiuiiii coicumnoi ippii 2S.54
miTSIQUUCI: US IIPIOC: 1547il/uc
IID SIQOHCI: 223 IK IK: S ll/iil
iiTiism PPI
ID PI/DI IP 1T6 1ft
CTOl 12f.O/ Jl.O 170.33 233 1.4
CTOL 121. fl/ (5.0 37.2) il l.i
CTOL 121. O/ ll.l 50.24 77 1.5
mtlH CILOIOTOLDIIE COICUTtlTIOI Ippbl
1CTB1L CILOIOTOLOUI COICIR11TIOI Ippbl
imi
DL
24
21
21
1.5
1.4
1
1
0
0
m
OL
OITICTI8?
iiTiism
ID
cm
cm
cm
PI/DI
112
112
114
.O/
.O/
.O/
77
51
77
.0
.0
.0
154.
113.
47.
IP
14
3f
25
PPI
1T6 116
201
155
76
1.
1.
1.
4
4
{
inn
DL
27
2)
24
PPI
DL
1.17
0.2f
1.51
mnct CILOIOIIIZUI COICUTUTIOI (ppb) 1.4
ICTOil CILOIOIIItlll COICUTUTIOI Ippb) 1.4
BITICTIO? hi
-------
FIGURE 5
Tuiricmoi or OHICTIOI t gourmuoi
lIIITf IT Illlll DILUTIOH, m? 3
mmiTioi or QCUTINTIOI mm
rill: LC4020
OETICTIOI no gouTiTiTioi IIIIT TUIFICITIOI
OIICIIIL CTIIIOU CIIOIOTOLUIH COICUTUTIOI (ppil 2i.33
OIIGII1L CTIIIOU CIIOIOIUZIII COICUItiTIOl Ippll 2S.S4
l?ll! SEQOIICI: 43? fir IOC: HI? ll/ue
UDSigUIICt: S4i WC IK: 11 ll/iil
•••••(••••••••••••^••••••••••••••••••••••••••••••••••••••••SMI
IRIISITT
ID ri/oi tr iic
CTOL 12i.l/ !l.fl 170.3) S23
CTOL 121. O/ iS.O 37.2! 125
CTOL 12l.fi/ tl.O 58.21 lil
1FC CILOIOTOIDEII COICIITilTIOI Ippb)
ACT CILOIOTOLDEII COICIITilTIOI Ippbl
11! 1 DIVI1TIOI riOI COHOTU ITS
UIIOI riOI 1CT JFIIID COICUTUTIOI
m
M
J.I
3.<
3.2
3.2
3.1
3.01
J.It
on ri
m
1.1
0.1
1.0
iTC
ion
fllTC
4.31
4.St
l.lt
3.01
Kcm
""
imiiiTT
10 N/U ir ATC
Clll 112.fi/ 77.0 1S4.{4 4S7
CIEI 112.fi/ 51.0 113,31 35f
Clll 114. O/ 77.0 47.2$ IS)
116 CILOtOTOLUEII COICUTUTIOI (ppbl
ACT CIlQIOTOtBUE COICUTUTIOI Ippb)
AT6 1 DirilTIOI flOI COffOTIO ATC
mioi rioi ACT JFIIID COICUTUTIOI
rn
ITC
3.0
3.1
3.4
3.2
3.0
4.SI
1.21
DIM!
ATC
0.2
0.0
0.2
ATC
•
•
«//*
-------
FIGURE 6
PartBtttra
Filt: LC4004
Titlt:
RESIDENCE TIME:
MINUTES/SEQUENCE:
OFFSET SEQUENCE I:
Acquired on:
Multiplt Ion
l-DEC-87 At: 08:18:42
57
0.01854
51
S.qu.nc. Oft»«t Stqutnc* Ti»e Fltg
0.07
1.05
1.39
2.39
2.54
3.54
3.67
4.69
4.82
5.84
6 00
7.00
7.15
8.15
8.31
10.85
10.90
A
B
C
D
E
F
C
H
I
J
K
L
M
N
0
P
0
-------
SUIinHRY OF THE PHRSE 2 PRECISION RESULTS
RESULTS FROM THE SCOTT CVLINDER PRECISION BNHLVSES
CC.
ra
cu
•
i CVL. ID.
IRRL 19177
iRRL 19 177
SRRL15177
:RRL 19177
:RHL 19177
iRRL 19 17?
•.RRL1917?
iRRL 19177
iRRL 19177
iRRL 19177
iRRL 19177
OHTE
9/19/87
: 9/16/67
19/17/87
i 9/18/87
! 9/19/87
: 9/2 1/87
i 9/22/87
i 9/23/87
! 9/24/87
! 9/29/87
! 9/26/87
CERTIFIED
PPH-CTOL
29.99
29.99
29.99
29.99
29.99
29.99
29.99
25.99
29.99
29.99
29.99
iRVERRGE RNRLVZEO CONCENTRRTION
i XERROR OF
RVG. CONC. FROH THEO
CONC.
PPH-CBEN
29.93
29.93
29.93
29.53
25.93
25.5 I
25.91
25.53
25.53
25.53
25.53
- PPtl
CONC.
isro DEV. OF RNRLVZEO CONC.-PPH
IK RELRTIVE
STO DEV. OF RNRLVZED
CONC.
THGH RNflLVSES
PPH-CTOL '
23.0
23.3
26.9
25. 1
27.3
27.7
27.2
26.4
25.3
28.0
26.6
26.1
0.32
1.7
6.32
RESULTS
PPH-CBEN
23.6
24.0
27.8
25.6
26.8
28.0
27.8
26.7
26.4
25.9
26.3
26.3
2.92
1.5
5.52
•A DEV. FROH HERN i
CTOL
-11.82
-10.62
3.22
-3.72
4.72
6.22
4.32
1.32
-3.02
7.42
2.02
CBEN i
-10.12!
-8.62!
9.82!
-2.52!
2.02!
6.62!
9.82!
1.72!
0.52!
-1.42!
0.12!
RESULTS FROn PRECISION 16-L SUflflR CRNISTERS
iCRNISTER
i ID.
iPEB-lR
iPEB-lR
iPEB-lR
iPEB-lR
iPEB-lR
i DRTE
i HNRLVZED
109/16/87
i 09/18/87
i 09/2 1/87
i 09/23/87
i 09/26/87
RTP THEORETICHL CONC
PPM-CTOL
36.6
36.6
36.6
36.6
36.6
iRVERRGE RNRLVZEO CONCENTRRTION
iKERROR OF
RVC. CONC. FROH THEO
PPH-CBEN
31.5
31.5
31.5
31.5
31.5
- PPH
CONC.
:STO DEV. OF RNRLVZED CONC.-PPH
•.X RELRTtVE
STO DEV. OF RNHLVZED
CONC.
TRGH RNRLVSES
RESULTS
PPO-CTOL PPH-CBEN
36.5
36.8
39.3
38.9
36.0
37.9
3.62
1.2
3 . 32
30.3
30.2
30.8
30.4
32.5
30.8
-2.12
1.0
3.12
2 DEV. FROM HERN i
CTOL
-3.72
-2.92
3.72
2.62
0.32
CBEN i
-1.82!
-2.12!
-0. 12!
-1.42!
5.42!
-------
DATA VALIDATION COMMENTS LOG
Date of sampling:
Date data package reel _
Date D.V. Logsheet completed:
Date data package finalized:
4*1 yi
ceivted:
i.
FIGURE 8
II
To minimize errors related to instrument response, check the following:
ACTION FREQUENCY CRITERIA COMMENTS
SAF
Calibration
l)upon arrival at LC
2)weekly thereafter
3)before demobilizing
<10% Error
Days since
last check
MFC
Calibration
l)upon arrival at LC
2) weekly thereafter
3) before demobilizing
<10% Error
Days since
last check
% Transport
Efficiency
1) start of day
2) end of day
> or = 85%
X\/
^ l
Verification
of Calculated
Quant itation
Limits
l)end of day
1) within Ippb
of calculated
Quant. Limits
2) <25% Error
Accuracy and
Precision
using Cylinder/
Canister of
known cone.
1) cylinder —
start of day
2) canister —
every other day,
at end of day
<25% Error
U
^l ~) -(f
Accuracy using
PE Cylinder/
Canister
l)one PE cylinder/
canister a day,
at start of day
<25% Error
-CTlOl- v-
Response
Factor
Decays
1)every calibration
all average
RF Decays
< or = 15%
-------
i-iliUKt y
KEY:
A - Comment does not affect data quality.
B - Comment requires Data Review action.
C - Comment requires explanatory action by ERT QA/QC Coordinator,
D « comment requires house data package withdrawal
(house to be scheduled for resampling).
COMMENTS:
B
dnJkL flns \K. SpmiyuiL -fabU^ <%>* Pits '. ^C>OD I
2)A -*/9 B
3) A B
4) A B.
5)A B_
6) A B_
-------
FIGURE 10
ICPS
i-nec-87
86! 14 1 55
6«r 1498.
DI
-a.
6001
CG
VL
126/ 91 126x 65 - - 11Z/ 77*--
294.
ON
77
e.0
2.0
4.6
6.0 e.e
TIfE CHIN.)
le.e
12.e
14.0
PE
TH
MIL
C
TO
VS
15.0
e.e
IB.
e.iee
i.
leee.
e.e
OR
CB
L3
L4
LS
L6
e.e
e.e
•30.0
•35.0
e.e
e.e
IN
Ml
FP
DrQ
mi
e.e
-3400.
20.0
0.210
0.190
RE3
RC1
Rl
R2
R3
158. SY
107. OP
-6.6 IS
-50.0 Gl
-7.0 62
G3
260.
vex
flPCI
0.000
0.000-
0.000
-------
APPENDIX B
Memorandum:
Further Replies to Peer Review Comments,
June 73, 7988 (Revised July 7, 1988)
Air Assessment-Indicator Chemicals
Prepared by
U.S. EPA Environmental Response Branch
Environmental Response Team
Edison, New Jersey
WDC 63394.T1 (SA)
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
EDISON, NEW JERSEY 08837
June 13, 1988
* (revised July 7, 1988)
MEMORANDUM
SUBJECT: Further Replies to Peer Review Comments on the Love
Canal Emergency Declaration Area Full Air Study
FROM: Thomas H. Pritchett, ERT QA/QC Coordinator (LoyjeCanal)
Environmental Response Branch
Bruce Peterson
CH2M Hill, inc., Seattle, WA
TO: Doug Garbarini
Love Canal EDA Habitability Study Project Manager
U.S. EPA Region II, ERRD-NYCRA
The peer reviewers raised several other questions concerning the
TAGA analyses and QAPP which I did not address in my first reply
memorandum (May 19, 1988). I am addressing those now.
Actual versus Nominal Detection Limits. Dr. Schoenfield pgs. 2-5:
Summary of Comment
Dr. Schoenfield addresses the issue of nominal detection limits
(those used by the TAGA group) versus actual detection limits
(those defined from a strictly statistical basis). He concluded
that the nominal detection limits used may have been lower than
the actual statistically derived detection limits. Because of
the high number of non-detects seen in the study, he wished to
see some discussion of the probabilities of false negatives.
From these probabilities confidence intervals could then be
reported. Unfortunately, it is not a trivial matter to develop a
statistical model which accurately predicts the TAGA's response
to a compound as will be shown below.
TAGA & Its signal Processing
In order to properly evaluate the applicability of various sta-
tistical models, one must first have a basic understanding of the
TAGA and its signal processing. The TAGA is a multi-channel
sequential signal counter. Each channel counts the ions which
first pass through one mass filter, are then converted to a
different mass, and finally pass through a second mass filter.
During the EDA habitability study three unique channels (referred
to as ion pairs) were used for each LCIC. The channels (or ion
pairs) are counted one at a time for brief time interval (usually
0.1 seconds) each and the signals are then converted to a rate
(ion counts per second or ICPS). One cycle through the channels
-------
is considered one unique measurement and occurs in just over a
second. Typically, at least thirty cycles were acquired for
every room in the EDA houses. Thus, when the TAGA obtained a
reading for a given LCIC in a room, that reading consisted of
averages from over thirty measurements on three separate
channels. More detailed descriptions of the TAGA and its data
acquisitions can be found in Appendix A of the Love Canal
Emergency Area Full Air study Quality Assurance Project Plan,
CH2M Hill, 1986.
These signals have several characteristics which must be taken
into account when one is developing a statistical model for
estimating detection limits. First, the signal response for each
channel of the TAGA is directly proportional to the concentration
of that compound. The proportionality constant for each ion pair
(channel) is called its response factor and is expressed in units
of ion-counts-per-second (signal)/part-per-billion (concen-
tration) . These constants can be drastically different for each
ion pair for that compound. For example, the addition of 5 ppb
of chlorotoluene will result in four times the signal increase in
the most sensitive ion pair than the signal increase in the least
sensitive ion pair. Second, the overall instrument sensitivity,
as measured by the ICPS/ppb(v) response factors, changes through-
out the day by as much as a factor of 2 and from day to day by an
half an order of magnitude. The relative sensitivity of each ion
pair to the other ion pairs for that compound do remain essent-
ially unchanged. Third, unlike a GC/MS the measured response is
for a direct sampling, realtime instrument. There are no target
compound signal peaks surrounded by baseline signals from the
clean carrier gas. Rather, sets of signal intensity averages
must be compared against each other to determine if there is a
significant difference - similar to the manner that signals from
a spectrophotometer are compared. Fourth, the measured signal
for each channel is composed of two separate components, a
constant background noise signal and the signal due to the com-
pound itself. However, the data reduction software does not
zero-correct the signals prior to computing concentrations.
Fifth, the standard deviations of the measured signals, converted
back to just ions counted, tend to agree with the theoretical
values predicted by the Poisson distribution.
Nominal Detection Limits
Because each ion pair channel has a different sensitivity, they
would all have different individual detection limits. The signal
for the most sensitive ion pair could be well above the back-
ground noise while the least sensitive ion pair signal was indis-
tinguishable from the noise. If one requires that all three
signals be statistically significant from the background noise,
as we did, then the overall compound detection limit will become
the detection limit of the least sensitive ion pair. Predicting
the probabilities for all three signals becomes complicated by
the facts that the different but linked sensitivities of the ion
pair signal channels to levels of true compound.
-------
To avoid false positives caused by shifts in the background noise
(an intermittent problem in phase 1), an additional detection
limit criteria was added. Even if all three signals were above
their individual detection limits, the compound was not con-
sidered to be detected unless the signal for the most sensitive
ion pair gave a concentration reading above the overall compound
detection limit. Because of the relative differences in the ion
pair sensitivities, the signal for the most sensitive ion pair
had to be 2X - 4X greater than its own detection limit in order
to generate the signal needed to meet this ID criteria. However,
if true compound was responsible for elevated signals, then the
elevated signals would have such a relative distribution of
intensities on the ion pair channels.
Quality Assurance Plan Coverage of
Detection Limit Concerns and Acquired Data
Because of the uncertainties in the true TAGA detection limits
(i.e., limit at which the false negative rate is less than 95%)
versus the limits used in this study, the quality assurance plan
had several features built into the analyses designed to gather
data on the validity of the TAGA detection limits. First, at the
end of each analysis day (w/ one exception caused by an instru-
ment malfunction) the TAGA group spiked a gas standard into the
ambient air at concentrations less than 1 ppb above the nominal
detection limit in order to specifically test the TAGA's ability
to detect trace levels of LCICs. Forty four detection limit
verifications were performed during the study and in all of them
the LCICs were successfully detected.
In addition, the TAGA calibrations used concentrations at approx-
imately 2 ppb and 4 ppb. Consequently, many of the calibration
curves contained concentration points below, just at, or slightly
above the nominal detection limits. Before preceding, it should
be noted that, because of the types of errors encountered when
using mass flow controllers, the low standard flow rates used to
prepare the 2 ppb air concentration were subject to the highest
relative errors. Therefore, the relative uncertainty of the true
concentration in the air was greatest with the "2 ppb" calibra-
tion point.
Forty six (46) chlorotoluene calibrations contained a concentra-
tion point less than 0.5 ppb above the nominal detection limit.
Of these points 10, or 22%, failed to meet one of the detection
limit criteria. The data for these points is summarized in Table
1A. In all ten calibrations the selected concentration point
failed to meet the ID criteria of having the signal of the most
sensitive ion pair exceed the equivalent signal for the compound
detection limit. In only three calibrations the least sensitive
ion pair signal failed to exceed its individual detection limit.
Eight of these ten false negatives were at concentrations just at
or 0.1 ppb above the detection limit. Since all but one of these
eight were at the lowest concentration in the curve which had the
highest relative error in the true concentration being spiked,
many of the observed "false positives" were probably spikes at a
-------
true concentration below the nominal detection limits. Thus, a
predicted 22% false positive rate would be overly conservative.
Chlorotoluene calibrations from phases one and four were also
selected based upon a calibration concentration between 0.5 ppb
and less than 1 ppb above the nominal detection limit. One
hundred and twelve (112) such calibrations were found. In all
cases the data at the target concentrations met all of the detec-
tion limit criteria. Thus, we can be certain that the difference
between the true detection limits and the used nominal detection
limits were less than 1 ppb.
Statistical Models of Detection Limits
Estimators for detection limits for the TAGA are not generally
available. However there are estimators of detection limits
which have been developed for other instruments which may be
applied to the TAGA. Two estimators which have been examined are
one based on the instrument response to background (Currie,
1968). A second method is based on the instrument calibration
curves (Hubaux and Vos 1967). Each method has its advantages and
drawbacks.
a) Minimum Detectable Difference Model
This method is based on the instrument response to the background
measurements made during calibrations with ambient air. The
basic concept of the method is to estimate the minimum detectable
difference between a blank measurement and a detectable concen-
tration. The key parameter required for this estimate is the
standard deviation of the instrument response to blanks.
One difficulty with the TAGA data currently available is that its
instrumentation produced averages of sets of 5 individual
measurements. These averages were themselves averaged to produce
estimates of overall standard deviations and means. Therefore
the instrument produced standard deviations may not be repre-
sentative of the individual sample (0.1 sec integration) standard
deviation.
The basic procedure for estimating the minimum detectable
difference (mdd) (ZAR, 1984) is to calculate
mdd = SQRT(2 Sp2/n') * (talpha/df + tbeta/df)
^
where Sp is the estimate of the pooled variance,
talpna df is the t statistic critical values for
controlling false positives,
tt)eta df is the t statistic critical value for
controlling false negatives,
and n' is the harmonic mean of the two sample sizes
n'= (2*n1*n2)/(n1+n2)
-------
Now for the TAGA the standard deviations provided by the
instrument may not represent the individual measurements and the
standard deviation of the instrument at the detection limit
concentration instrument response are not measured at all.
However, since the individual measurements are total counts of
ions detected, the distribution of these measurements may be
assumed to follow a poisson distribution. .This allows the
variance of the individual measurements to be estimated by the
mean of the individual measurements.
So the mdd can be estimated as follows:
let n be number of measurements of initial ion pair counts,
IQ (50 for Love Canal Air Assessment)
let m be the number of measurements of ion pair counts made
at the mdd. (30 for Love Canal Air Assessment)
and let t be the integration time for 1 measurement (0.1 for
Love Canal Air Assessment
Then the estimates of the standard deviations ( 1 is calibration,
2 is measurement at mdd+IQ)
S22 = mdd+I0/t
And the estimate of the pooled standard deviation is
Sp2= (I0*(n-l)+(mdd+I0)*(m-l) )/(t*(m+n-2) )
This can be plugged into the above equation to yield:
mdd2 - ((2*(m-l)*(talpna+tbeta)2)/((m+n-2)*t*n'))*mdd
-(2*I0Mtalpna+tbeta)2)/(t*n')) = 0
Since this is a quadratic equation it can be simplified if
terma = 1 ,
termb = ( (2* (m-l)*(talpna + tbeta)2)/( (m+n-2)*t*n» ) ) ,
and termc = ( 2*IQ* (talpha + tbeta) 2)/(t*n' ) ) .
Then mdd = (termb +/- SQRT(termb2+4*termc))/2
For 99% confidence and 99% power (1% false positive, 1%
false negative rates) talplla 35 = tbeta 3r = 2.438 and
for the Air LCIC assessment:'n=50, m=30'n'=37.5 df=35 and
t=0.l.
So termb= 4.7144 and termc= 12.6802*IQ.
-------
To then convert the minimal detectable difference (mdd) plus base
count back to concentrations, the instrument response factor is
divided into the mdd. Again, although each ion pair would have
similar values for the mdd as ion counts per second, the conver-
sion of mdd's to concentrations will result in vastly different
detection limits for each ion pair (3X -4X difference between
least sensitive and most sensitive). Therefore, the overall
compound detection limit would be the individual detection limit
of the least sensitive ion pair.
b) Calibration Model of Detection Limits
The second method is based on the instrument calibration curve.
This method uses the prediction limits of the regression relation
to estimate the method detection limit. One disadvantage of this
method is that it assumes that the concentrations of analytes
used for the are exactly known. In actuality the accuracy of the
metering mechanisms is less reliable at low levels. This can
increase the apparent non-linearity of the calibration curve. An
advantage of the calibration curve method of estimation is that
it accounts for not only the uncertainty in measurements at given
concentrations but also the uncertainty of the calibration curve.
The calibration curve method uses a regression relation developed
from the several levels of calibration concentrations used. This
model is:
RAi = a*C± + b + e^ i=l...n;
where n is the number of concentration levels used in the
calibration.
Let C' be the mean of the C^ ,
a' be the estimate of a,
and b' be the estimate of b.
Then
SSC = sumi(Ci-C»)2,
MSE = (sumi(RAi2) - b^sum^RAi) - a'*sumi(CiRAi) )/(n-2) ,
t = talpha.n-Z = tbeta.n-Z critical value of t statistic,
and T = t*SQRT(MSE*((n+l/n)+(C'2/SSC)))
Finally,
MDL = ((2*a'*T - (2*t2*C'*MSE/SSC))/(a'2 -(t2*MSE/SSC)).
The derivation of this equation is shown in the Pilot Study for
Love Canal EDA Habitability Study Volume 1 Appendix E (CH2M HILL
1987).
-------
c) Summary of Model Limitations
Both adaptations of standard method for estimating detection
limits have difficulties with aspects of the TAGA operation not
represented in the statistical model. The blank response method
assumes that the variability of the instrument in response to a
blank, which is primarily electronic noise, is similar to the
variability of the instrument response to a detectable level of
an analyte, which is due to a combination of both chemical and
electronic noise. The calibration curve method assumes that the
instrument variability is constant over the linear calibration
range and that the concentrations used during calibration are
exact. Both methods therefore give approximations to the detec-
tion limit but with unknown biases due to the assumptions that
must be used with either method.
In addition, both methods make their predictions based upon only
one ion pair signal. In their derivations both models assume
that the critical values for false negatives, tjpeta, and false
positives, talpha, are equal. This assumption is only true when
one is dealing with only a single signal. The TAGA was using
three interrelated signals for each compound. The true alpha and
beta values needed for 99% confidence will be somewhere between
the values used by the above methods and the alpha and beta
values that would be predicted by the probability of three
separate random events occurring simultaneously (false positive)
or individually (false negative). Because these signal averages
are interrelated but still subject to random errors of different
magnitude with different relative effects on the final determina-
tion, the true probabilities of type I and type II errors may not
agree with the statistically predicted probabilities. For
example, a false negative determination could result from slight
random ion counting errors in either the 112/51 or 114/77 ion
pair signals causing the reported averages to fall below the
individual ion detection limits or could result from a similar
but different magnitude ion counting error in the 112/77 signals
causing the reported average to fail to meet the ID criteria. At
a given threshold the probability of such a false negative occur-
ring will be drastically different than the false positive prob-
ability of random errors simultaneously causing all three
averages to exceed their individual detection limits and the most
sensitive ion average to exceed the ID criteria.
Comparison of Model Detection Limits
With Love Canal Calibration Data
Table IB contains a comparison of the predicted and nominal
detection limits for ten of the calibrations in which both com-
pounds were present at less than 1 ppb but greater than or equal
to 0.5 pbb above the nominal detection limits. Table 1C contains
a similar comparison of the various detection limits for chloro-
toluene calibrations for which all "detect" criteria were met at
a concentration less than 0.5 ppb above the nominal detection
limit. Table ID contains another set of comparisons for the
chlorotoluene calibrations for which a false negative occurred at
-------
a concentration above the nominal detection limits.
Some trends can be discerned from this data. First, the detec-
tion limits predicted by the minimum detectable difference model
were consistently lower than the used nominal detection limits.
Second, the calibration model detection limits were, with rare
exceptions, consistently higher than the nominal detection limits
and tended to be overly conservative. The excessive conservatism
is best illustrated in Table IB in which the calibration model
detection limits were on the average of 2X higher (which trans-
lated into usually > 2 ppb higher) than the nominal detection
limits) even though all of the empirical data indicates that the
nominal detection limits are less than 1 ppb lower than the true,
< 1% false positive detection limits. In the few instances in
which the calibration model predicted a lower detection limit
than the nominal value, the compound was detectable at less than
0.5 ppb above the nominal detection limit. Third, the minimum
detectable difference model may be adequate for predicting the
rate of false positive (a supposition which was not tested) but
fails to predict the probability of false negatives.
Conclusions
The nominal detection limits had an experimentally determined,
conservative false negative rate of approximately 22% when com-
pared against chlorotoluene calibration concentrations of less
than 0.5 ppb above the used detection limit criteria. This rate
was considered to be conservative because several of the observed
false negatives could have been caused by errors in the spiked
concentrations themselves and not in the observed signals. When
the criteria were applied to concentration data obtained at 0.5 -
< 1 ppb above the nominal detection limits, no false negatives
were observed. Consequently two different detection limit models
were evaluated for use with the data, the minimum detectable
difference model (Zar, 1984) and the calibration curve model
(Hubaux and Vos, 1968). Neither model compared favorably against
the calibration data. The calibration model tended to be overly
conservative and predicted detection limits at concentrations
much higher than the TAGA was able to repeatedly detect compounds
( > 0.5 ppb above the nominal detection limits). The minimum
detectable difference model repeatedly underestimated the detec-
tion limits at which an analysts could have reasonable certainty
that false negatives were not occurring.
References
Hubaux, Andre, and G. Vos; "Decision and Detection Limits for
Linear Calibration Curves", Analytical Chemistry, 42:8, 849-855,
(1968)
G.H. Zar; "Biostatistical Analysis, 2nd Edition", Prentice Hall,
Inc., Englewood, NJ, 1984.
CH2M Hill, "Pilot Study for Love Canal EDA Habitability Study,
Volume I", February 1987.
-------
L.A. Currie; "Limits for Qualitative Detection and Quantitative
Determination", Analytical Chemistry; 40:3, 586-593, (1968).
Origin of TAGA Performance Criteria, Artiola, pgs. 2, p. 6:
In his review of the QAPP, Dr. Artiola states that the criteria
for the TAGA data quality objectives (DQOs) were not properly
referenced or justified. In a similar manner Dr. Shy wanted
further information on the selection of 4 ppb as the maximum
acceptable detection limit. There were (and still are ) no
references upon which TAGA performance criteria could be based.
The selection of 4 ppb as the maximum acceptable detection limit
was based strictly upon the achievable detection limits of the
TAGA and was not driven by a health based criteria (except that
the DLs had to be as low as possible). All of the proposed DQO
criteria were set after several months of TAGA method development
during which the various instrument parameters were routinely
monitored when the TAGA was known to be operating properly.
These criteria were then reviewed and compared against the com-
bined 8 years of direct TAGA experience (as of Spring 87) of the
three senior TAGA scientists. During this review the only criti-
cism of these criteria was that they were probably too tight; the
TAGA could have difficulty consistently meeting these specifica-
tions in the field.
We have discussed these criteria with other TAGA groups since the
QAPP was written and the field work initiated. These groups /
operators included Raphe Pavlick - the Roy F. Weston TAGA
operator from the several years that they marketed TAGA applica-
tions; Tye Willingham - a former operator of the EPA's TAGA and a
current operator of the York Research TAGA; Sciex - the instru-
ment manufacturer; and Ministry of the Environment, Ontario - the
group operating the Canadian government's mobile TAGA. In all
cases they were pleasantly surprised that the TAGA was indeed
meeting such stringent criteria.
Interferences, Dr. Daisey p. 3:
Dr. Daisey asked how we determined that the ion pairs chosen were
not subject to interferences. This determination was based upon
three sets of data. First, during the combined 8 years of direct
TAGA experience none of the senior TAGA scientists had ever
encountered any compound except chlorobenzene and chlorotoluenes
forming all three of either set of ion pairs. Second, none of
the TAGA operators could hypothesize from their considerable
GC/MS experience any organic compound which, for either set of
ion pairs, could conceivably form both of the parent ions and
then have that those parent ions fragment to the selected daugh-
ter ions. Third, because of the constant raising of this ques-
tion (including at the last peer review committee / TRC meeting),
during our months of method development/refinement we specifical-
ly looked for interferences to any of the selected compounds. In
every case, whenever unexpected signals were seen for the
selected ion pairs, the signals were found to be caused by real
-------
levels of the target compounds. In fact, several leaks in the
TAGA inlet systems were discovered during this search for inter-
ferences. No true interferences were ever observed.
However, the interference issue is a moot point in regards to
this study. Interferences can only affect the data quality by
either biasing reported values high or causing false positives.
Only three sets of positive unknown data were ever acquired
during the study - two sets for the "detect" house and on set for
ambient air. Had any of these "hits" been caused by an interfer-
ent, based upon our previous experiences with interferences with
other compounds, the resulting individual ion pair concentrations
would have been skewed. This was not the case for these three
sets of data; the individual ion pair concentrations were consis-
tent. In addition, the house detect was confirmed by a Summa
canister grab sample followed by GC/MS analysis. Therefore, it
is safe to state that there was never a problem due to interfer-
ences to the selected ion pairs for chlorotoluene, the only
compound for which the issue of interferences would have been
applicable in this study.
Calibrations in Ambient Air, Dr. Daisey p. 4:
Dr. Daisey asked whether ambient air was used during the TAGA
calibrations and what the effect that contaminated ambient air
would have had on the detection limits. Ambient air was used.
If that ambient air was contaminated then the reported detection
limits would have been biased high. During phases 3 & 4 we did
encounter calibrations in which elevated IQ signals for chloro-
toluene were seen. Corrections to the detection limits were
applied. This elevated signal determination and the applied
corrections are covered in detail in my final summary of QAPP
modifications memo dated April 15, 1988 and in my initial reply
to the peer review comments dated May 19, 1988.
Typical Sequence of TAGA Events, Artiola p. 2:
Dr. Artiola stated that he found it difficult to discern in the
QAPP the start-up and end-of-day procedures from those imple-
mented prior to entering each house. This was partially due to
the fact that several of the procedures (e.g., LCIC calibrations)
were repetitively used throughout each sampling day. Another
confusing procedure was the "Analysis Procedure for Precision /
Performance Evaluation Cylinders/Canisters" (Section 4.7 of
Appendix A). This procedure was used during both the start-of-
day and end-of-day instrument checkouts - the only difference
being which sample was being analyzed using this procedure (i.e.,
6 Liter Summa canister, 16 Liter Summa canister, or Scott
Specialty standard cylinder).
To clarify this confusion I have enclosed Table 2, which lists
the typical TAGA analysis sequence for a given day. The table
references the Appendix A section applicable to each sequence
step. However, the table does not list the initial LCIC calibra-
tion which was required for the start-of-day Scott precision/P.E.
10
-------
cylinder analysis because the procedure itself states that a
prior calibration was required.
Auditing of TAGA Maintenance Logbook, Dr. Daisey p. 4:
Dr. Daisey asked whether the TAGA's maintenance logbook was ever
audited. The logbook was audited by Ken Caviston of Northrop
Services, the ORD Quality Assurance Branch operations contractor,
prior to the first phase of the study. It was not re-audited
afterwards.
Whenever TAGA maintenance or corrective actions were taken during
a sampling phase, these actions were documented on the TAGA
chronological logsheet for that day. In addition, the TAGA
senior scientist kept a personal logbook in which he recorded all
maintenance activities which occurred during each sampling
period. Therefore all TAGA maintenance during this study was
always recorded and was usually recorded in at least two separate
sets of documents.
Typographical Errors in Appendix A of the QAPP, Dr. Daisey pgs.
3-4, Items 5, 6, and 10:
Dr. Daisey noted several items which she identified as either
typographical or omission errors. These items were 1) "benzene"
in line 5 of paragraph 1 of page 3-4 should be "benzyl"; 2)
factor of 2 missing from equation 2, line 23 on page 4-12; and 3)
EC is not defined. On the first item she is correct - "benzene
chloride" should be "benzyl chloride" (alpha-chlorotoluene).
However, on the second item she is not correct. The noted equa-
tion is a direct transcription of equation A.3.2 of Appendix A of
the "Summary of the TAGA Standard Operating and Reporting Proce-
dures for the Love Canal Full-Scale Air Sampling Study" (Appendix
A of the QAPP). The issue of whether of factor of 2 should be
included in the calculations is an issue of the degree of confi-
dence one can place in the predicted errors bars. Since these
error bars are not statistically derived, confidence intervals
based upon standard deviations are not applicable. If factors
are needed to define the confidence intervals for these error
bars, these factors may ultimately be defined by both the time
interval and the relative differences between the bracketing
response factors. However, in this study these error bars would
have been applied to data with temporally close, bracketing
calibrations. Therefore, a factor of 2 would not have been
applicable. In fact, during the instances when intermediate
response factors would have been applicable (the "detect" house),
the bracketing calibrations had response factors which were so
similar that there was no significant difference between the
concentrations calculated with either set of bracketing response
factors and those calculated with intermediate response factors.
When Dr. Daisey made her comment on EC she correctly identified
EC as the uncertainty in the reported concentration. Although EC
was not defined in the text, it was mathematically defined in
11
-------
equation A.2.3. The confusion arose because of the use of the =
symbol in the equation. When this symbol is used in mathematical
derivations it is used to specify that the variable to the left
of the symbol is "defined as" the expression to the right of the
symbol. In writing this derivation I assumed that the readers
would be familiar with the meaning of the = symbol.
Effect of Changes in Humidity on TAGA Sensitivity
In controlled experiments performed at Edison the maximum de-
crease in TAGA sensitivity under maximum shifts of relative
humidity (dry zero air to near-saturation) never exceeded a
factor of 2. On two separate days the TAGA sensitivity was
evaluated in controlled humidity matrices ranging from dry zero
air to essentially saturated zero air. The results were repro-
ducible and the relative changes in instrument response were
consistent over a 5X range of concentrations. A typical set of
relative response versus absolute humidity data is illustrated in
Figure 1. (The high humidity reading is biased low by the inac-
curacy of the relative humidity probe at close to the saturation
point.)
The TAGA's sensitivity was always highest with zero air (no
humidity) and suffered the largest relative drop when the sample
air was initially humidified. Therefore the effects of humidity
changes on the detection limits for the EDA houses should always
be less than a factor of two because there must always be some
humidity in the ambient and indoor air.
cc: B. Coakley, U.S. EPA-ERT
G. Helms, CH2M Hill
D. Mickunas, Roy F. Weston, Inc. (REAC)
12
-------
FIGURE 1. RELATIVE SENSITIVITY OE TAGA
m
c
-------
TABLE 1A. RESULTS FROM CHLOROTOLUENE CALIBRATIONS IN WHICH A CONCENTRATION
LESS THAN 0.5 PPB FAILED ONE OR "DETECT" CRITERIA (FALSE NEGATIVE)
FILE
NAME
MLC046
MLC088
MLC150
MLC150
MLC158
MLC158
MLC003
MLC003
MLC002
MLC035
MLC2081
MLC3138
MLC114
MLC114
(I)
(S)
(I)
(S)
(I)
(S)
(I)
(S)
NOMINAL SPIKED
DETECTION CONG .
SIGNAL OBSERVED
DETECTION SIGNAL
LIMIT (PPB) (PPB) LIMIT (ICPS) (ICPS)
1.
1.
3.
3.
1.
1.
2.
2.
1.
1.
2.
2.
1.
1.
9
8
9
9
9
9
0
0
8
9
0
0
8
8
2
2
4
4
2
2
2
2
1
2
2
2
2
2
.0
.1
.0
.0
.0
.0
.0
.0
. 9
.0
.0
. 1
.0
.0
214
187
566
149
206
53
261
93
236
205
96
149
238
76
186
166
392
136
151
52
179
84
235
202
79
136
178
74
PPB TOTAL PPB
EQUIV. ADJUSTMENT
OF DIF. TO DL
0
0
1
0
0
0
0
0
0
0
0
0
0
0
.2
.2
.2
.3
.5
.0
.6
.2
.0
.0
.4
.2
.5
.0
0.
0.
1.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
3
5
3
4
6
1
6
2
1
1
4
3
7
2
NOTES :
Unless noted otherwise,
because the
signal
the Chlorotoluene was
of the most sensitive
the identification
suffices after the
missed because of
limit, then
the di
ion
criteria unless designated
file
an ion
name. When the
signal did not
not detected
did not meet
otherwise by
Chlorotoluene was
meet
its 2
fferent signals are noted as (I)
for ID criteria and
ion
signal criteria,
sigma
and (S)
respectively.
-------
TflBLE IB. COMPRRISON OF VRRIOUS DETECTION LIMITS WITH CRLIBRRTION DRTR IN WHICH BOTH LCICs WERE "DETECTED"
IN WHICH BOTH LCICs WERE "DETECTED" RT CONCENTRRTIONS 0.5 - < 1 PPB OF THE NOMINRL DETECTION LIMITS.
FILE
MLC4119
MLC042
MLC012
MLC4040
MLC4037
MLC4043
MLC178
MLC076
MLC4124
MLC010
RUERRGE
CHLOROTOLLIENE
SPIKED NOMINRL MOD
CONC. Det Lim DL (1)
2.1 1.6 1.1
2.0 1.5 0.7
2.0 1.4 0.7
2.0 1.4 0.9
2.0 1.4 1.0
2.0 1.3 1.5
2.1 1.3 0.6
2.0 1.3 0.7
2.1 1.3 0.9
1.9 1.2 0.6
7. DIF. PPB DIF. CHLOROBENZENE 7. DIF. PPB DIF.
FROM MRX. DL FROM FROM MRX. DL FROM
NOMINRL FROM NOMINRL SPIKED NOMINRL MOD NOMINRL FROM NOMINRL
Det Lim CRLIB (2) Det Lim CONC. Det Lim DL <1> Det Lim CRLIB <2> Det Lim
-31. -57. 5.2 3.6 2.1 1.4 0.8 -42. 17. 3.4 2.0
-53. 3X 3.2 1.7 2.0 1.3 0.6 -53. B7. 3.3 2.0
-50. OX 6.6 5.2 2.0 1.5 0.6 -60. OX 9.9 8.4
-35. 77. 4.1 2.7 2.0 1.2 0.8 -33. 3X 4.4 3.2
-28. 6X 3.0 1.6 2.0 1.2 0.7 -41. 77. 3.8 2.6
15. 4X 5.1 3.8 2.0 1.5 0.7 -53. 3X 4.2 2.7
-53. 8X 5.5 4.2 2.1 1.2 0.5 -58. 3X 3.8 2.6
-46. 2X 5.6 4.3 2.0 1.2 0.6 -50. OX 3.6 2.4
-30. 8X 1.6 0.3 2.1 1.1 0.7 -36. 4X 1.8 0.7
-50. OX 2.8 1.6 1.9 1.3 0.7 -46. 2X 2.1 0.8
-36. 4X - 2.9 - - - -47. 6X - 2.7
NOTES:
(1) - Detection limit based upon minimum detectable difference with the least sensitive ion pair.
(2) - Maximum of the ion pair detection limits calculated from the calibrations using method developed
by Hubaux and uos.
-------
TABLE 1C. COMPARISON OF VARIOUS DETECTION LIMITS WITH CALIBRATION DATA
IN WHICH CHLOROTOLUENE WAS "DETECTED" AT A CONCENTRATION
LESS THAN 0.5 PPB OF THE NOMINAL DETECTION LIMIT.
FILE
MLC008
MLC032
MLC3143
MLC4151
MLC3134
MLC4133
MLC2126
MLC2086
MLC3021
AVERAGE
MAX. DL
SPIKED NOMINAL MDD FROM
CONC. Det Lim DL (1) CALIB (2)
1.9 1.9 0.8 4.1
1.9 1.9 0.9 1.5
2.1 2.0 1.2 3.4
2.0 1.9 1.1 1.2
2.1 1.9 1.3 2.6
2.0 1.8 1.0 1.7
2.0 1.8 0.7 2.7
2.0 1.7 1.0 3.1
2.1 1.7 1.0 2.9
% SPIKED ABOVE THE
"DETECTION LIMIT"
MAX. DL
NOMINAL MDD FROM
Det Lim DL (1) CALIB (2)
0.0% 137.5% -53.7%
0.0% 111.1% 26.7%
5.0% 75.0% -38.2%
5.3% 81.8% 66.7%
10.5% 61.5% -19.2%
11.1% 100.0% 17.6%
11.1% 185.7% -25.9%
17.6% 100.0% -35.5%
23.5% 110.0% -27.6%
9.4% 107.0% -9.9%
NOTES :
(1) - Detection limit based upon minimum detectable difference
with the least sensitive ion pair.
(2) - Maximum of the ion pair detection limits calculated from
the calibrations using the method developed by Hubaux
and Vos.
-------
TABLE ID. COMPARISON OF VARIOUS DETECTION LIMITS WITH CALIBRATION DATA IN
WHICH CHLOROTOLUENE WAS "NOT DETECTED" AT A CONCENTRATION LESS
THAN 0.5 PPB OF THE NOMINAL DETECTION LIMIT (FALSE NEGATIVE).
FILE
MLC2081
MLC003
MLC150
MLC3138
MLC046
MLC035
MLC158
MLC002
MLC114
MLC088
AVERAGE
MAX. DL
SPIKED NOMINAL MDD FROM
CONC. Det Lim DL (1) CALIB (2)
2.0 2.0 1.3 4.2
2.0 2.0 0.6 6.3
4.0 3.9 1.0 9.5
2.1 2.0 1.2 4.4
2.0 1.9 0.9 6.8
2.0 1.9 0.8 3.6
2.0 1.9 0.9 5.5
1.9 1.8 0.7 4.8
2.0 1.8 0.7 5.4
2.1 1.8 0.8 6.7
% SPIKED ABOVE THE
"DETECTION LIMIT"
MAX. DL
NOMINAL MDD FROM
Det Lim DL (1) CALIB (2)
0.0% 53.8% -52.4%
0.0% 233.3% -68.3%
2.6% 300.0% -57.9%
5.0% 75.0% -52.3%
5.3% 122.2% -70.6%
5.3% 150.0% -44.4%
5.3% 122.2% -63.6%
5.6% 171.4% -60.4%
11.1% 185.7% -63.0%
16.7% 162.5% -68.7%
5.7% 157.6% -60.2%
NOTES :
(1) - Detection limit based upon minimum detectable difference
with the least sensitive ion pair.
(2) - Maximum of the ion pair detection limits calculated from
the calibrations using the method developed by Hubaux
and Vos .
-------
TABLE II. TAGA DAILY OPERATIONS SUMMARY
1. Completed TAGA operating log sheet (SORP figure 4.1.1)
*
2. Started tnd wanned up Instrument (SORP section 4.1) and completed
checklist (SORP figure 4.1.4)
3. Performed miss calibration of quidrupoles 1 and 3 (SORP section 4.1).
4. Performed Scott preclslon/p.e. cylinder analysis (SORP section 4.7)
5. Performed hose transportation efficiency verification (SORP section
4.2)
6. Performed compound calibration Immediately prior to sampling house
(SORP section 4.3) and checked response factor decay and detection limits.
7. Analyzed house (SORP section 4.6)
8. Repeated sections 6 and 7 until the end of the day.
9. Performed the end of day compound calibration (SORP section 4.3)
10. Performed the detection and quantization limit verification analysis
(SORP section 4.8)
11. Performed the 16-1 prec1s1on/p.e. summa canister analysis (SORP
section 4.7; the frequency of this analysis varied according to canister
availability and Q.C requirements)
12. Performed the 6-1 p.e. summa canister analysis (SORP section 4.7; the
frequency of this analysis varied according to canister availability and
Q.C* requirements)
13. Performed the end of day hose transportation efficiency verification
analysis (SORP section 4.2)
14. Shut down Instrument and recycled cryosystem
-------
APPENDIX C
Memorandum:
Reply to Peer Review Comments,
May 19, 1988
Air Assessment—Indicator Chemicals
Prepared by
U.S. EPA Environmental Response Branch
Environmental Response Team
Edison, New Jersey
WDC 63394T1 fSA)
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
EDISON, NEW JERSEY OMJ7
• May 19, 1988
MEMORANDUM
SUBJECT: Reply to Peer Review Comments on the Love Canal
Emergency Declaration Area Full A1r Study
FROM: Thomas H. Prltchett. ERT QA/QC Coordinator (Love Canal)
Environmental Response Branch
TO: Doug 6arbar1n1
Love Canal EDA Hab1tab111ty Study Project Manager
U.S. EPA Region II. ERRD-NYCRA
I have read the four sets of peer review comments that I received
and have verbally responded to questions from Dr. Schoenfleld over the
phone. In these comments and questions two major Issues directly related
to the TAGA arose which I should address.
The first Issue, which was raised by both Dr. Dalsey and Dr. Schoenfleld,
concerned our use of IQ to calculate the detection limits when the cali-
brations were performed 1n ambient air. Both noticed that this method assumes
that no LCICs or Interferences are present during the calibration. They both
questioned the validity of that assumption and correctly Identified the poten-
tial for an artificial positive bias 1n the detection limits 1f either an LCIC
or Interferent was present. Dr. Schoenfleld was specifically worried that
this bias could result 1n the occurrence of false negatives. As I explained
to Dr. Schoenfleld, we were aware of that potential and watched for elevated
I0s1gna1s during the calibrations. Elevated I0 signals for chlorotoluene
were observed five times during phase 3 and ten times 1n phase 4; none
were observed for chlorobenzene. Whenever an elevated I0 was observed,
an alternate set of detection limits were computed using the procedures
outlined on page 9 of 1n the "Final Summary of Modifications to TAGA Proce-
dures Used During the Love Canal Full-Scale Air Sampling Study", dated
April 15, 1988 and the occurrences were documented with ERT QA/QC memoranda
to the files. These Memoranda are contained within the TAGA group's
final report and are specifically referenced 1n report as Memo fs T, U,
and Z. In all cases-the Initial non-detect field determination remained
unchanged. Therefore, this potential biases towards false negatives was
checked for and, 1n the very few cases 1n which 1t could have occurred,
1t was never found.
The second issue concerned the variability 1n the TA&A's detection
limits and the effect that this variability would have on comparisons of
this data to past air data and to the hab1tab111ty criteria. The peer
-------
reviewers did not have the full set of data (specifically the detection
limits for each house) from which they could compute the statistical
summaries needed to answer their concerns. These I have enclosed.
Specifically Table 1 lists by phase the following detection limit data
for both compounds: the average, the standard deviation, the minimum
value used, and the maximum value used.. (When the original detection
limit was based upon an artificially high 10, I used the alternate detec-
tion limit 1n computing these statistics.) The attached figures graphically
Illustrate the frequency at which a given range of detection limits were
used for each phase and for the whole study. Because these distributions
are not Gaussian, I have also summarized the detection limit distribution
1n Table 2.
I have only addressed the TAGA-related concerns of which I am aware
of. As additional concerns are raised either by the remaining reviewers
or during a TRC meeting, I will be happy to make the appropriate reply.
Atta.chment
cc: B. Coakley, U.S. EPA-ERT
6. Helms, CH2M.H111 V
-------
TABLE 1. SUMMARY OF. DETECTION LIMITS USED FOR HOUSE
ANALYSES DURING THE LOVE CANAL EDA FULL AIR STUDY
PHASE 1
AVERAGE DETECTION LIMIT (ppb)
STANDARD DEVIATION (ppb)
MINIMUM DETECTION LIMIT
MAXIMUM DETECTION LIMIT
PHASE 2
AVERAGE DETECTION LIMIT (ppb)
STANDARD DEVIATION (ppb)
MINIMUM DETECTION LIMIT
MAXIMUM DETECTION LIMIT
PHASE 3
AVERAGE DETECTION LIMIT (ppb)
STANDARD DEVIATION (ppb)
MINIMUM DETECTION LIMIT
MAXIMUM DETECTION LIMIT
PHASE 4
AVERAGE DETECTION LIMIT (ppb)
STANDARD DEVIATION (ppb)
MINIMUM DETECTION LIMIT
MAXIMUM DETECTION LIMIT
SUMMARY
AVERAGE DETECTION LIMIT (ppb)
STANDARD DEVIATION (ppb)
MINIMUM DETECTION LIMIT
MAXIMUM DETECTION LIMIT
Chlorpbenzene
N *
1.04
0.46
0.50
3.70
N *
0.69
0.22
0.40
1.60
N -
0.88
0.24
0.50
1.80
N -
0.70
0.20
0.40
1.50
N *
0.82
0.32
0.40
1.50
Chlorotoluene
129
1.23
0.54
0.60
4.20
133
0.81
0.29
0.40
2.30
148
1.17
0.36
0.60
2.70
155
0.93
0.28
0.40
2.10
565
1.03
0.42
0.40
2.10
-------
TAOLE a.
FREQUENCv DISTRIBUTION FOR THE TAOA ocrccrioN Limrs USED
OUAINO THE LOVE CANAL EM FULL-SCALE lUR SAflPLIMO STUOV
OL
0
o
0
0
0
0
1
8
1
1
1
PHASE 1 1
CHLOROBCNZEME CNLOROTOLUENC
Fraa, Par. Fr«4j Par.
O O.OX 0 O.OX
1 O.OX 0 O.OX
• 7.0X * 2.3*
IT 20. •* • 7.O*
o ao. «* o 7. ox
aa 38.0X 14 t*.4X
ao 83. 81 14 30. a*
1« 48. a* ao 48.7*
14 7«. t* ai
88. 3*
88. 4*
•3. OX
•3.8X
•4.9X
94. «X
94.91
•4.**
•7.7*
•7.7*
•7.71
•7.7X
•7.7X
•7.7X
•7.7X
•7.7X
1 47. 4*
73.4*
T*. IX
02. ax
84. OX
87.4*
8«.*X
•3. OX
•3.8X
•3.8X
•0.3*
•4. IX
•4. IX
97.7X
' »7.7X
•8.4X
•8.4X
•7.7X 0 M.4X
•7.7X 0 .4*
•O.4X 0 .4*
••.ax o .4X
••.ax o .4*
••.ax o .4*
••.ax o .4x
•«.ax o .4*
100.0X 0 .4*
100.0X 0 .4X
too. ox i .ax
1OO. OX 0 .21
too.ox o .ax
too.ox i too.ox
«•*•*•
PHASE a
CHLOROBENZENE CHLOROTOLUENE
a I.BX i o.ox
20 20.3* 8 4.8*
44 ' 84. ** 27 27. IX
30 77.4* 40 B7.1X
8 77. 4X 8 87. t*
14 88. OX !• 71. 4X
8 •0.2X 13 80. B*
3 «a.BX 10 88.0*
4 98. 8X
a »7.ox
0 97. OX
a M.BX
a too .ox
o too. ox
o too.ox
o too. ox
o too. ox
o too. ox
o too. ox
o too. ox
.
100.0*
too. ox
too.ox
too.ox
too. ox
too.ox
too. ox
too.ox
too. ox
100.0*
too.ox
100. OX
too. ox
too. ox
too. ox
too.ox
too. ox
too.ox
too.ox
• 1.7X
•a.BX
•3. ax
•7.7X
•8.BX
M.BX
M.BX
M.BX
M.ax
••.ax
••.ax
100.0X
too. ox
too.ox
100.0*
too.ox
too.ox
100.0*
100.0*
1OO.O*
too.ox
100.0*
too.ox
100.0*
100.0*
100. OX
too.ox
too.ox
too. ox
too.ox
100.0*
PHASE 3 1
CHLOROBENZENE CHLOROTOLUENC 1
Fraq Par. Fraq Par.
0 O.OX 0 O.OX
• 4.1* O O.OX
10 10. 2* 0 3.4X
ao 31.01 0 8.8X
O St. OX O 8.8X
aa 44.4* 8 14.9X
27 44. •* at 2».l*
17 74.4* !• 41. •*
37 94. 4X 37 44.91
4 97. 3X 8 72.3*
1 M.OX It 78.7X
1 M.4X 11 87.2X
1 ••.3*
0 •9.3*
1 100.0*
o too.ox
O 100. OX
0 100. OX
o too. ox
0 1OO.OX
0 100. OX
o too. ox
o too. ox
0 100. OX
0 100. OX
o too. ox
o too. ox
o too.ox
o too.ox
o too. ox
o too.ox
o too.ox
o too.ox
•O.BX
•3. ax
•3.«X
•8.3*
•7.3X
M.3X
M.BX
M.3X
**.3X
•9.3X
••.3*
too.ox
too. ox
too. ox
100. OX
100. OX
too.ox
too. ox
100.0*
too. ox
too. ox
o too. ox o too. ox
o too.ox o too. ox
o too.ox o too. ox
o too.ox o too.ox
o too.ox o too.ox
o too. ox o too.ox
PHASE 4
CHLOROBENZENC CHLOROTOLUENC
13 7.7X t O.4X
28 a3.»X 8 B.8X
30 43. 2X 19 18. IX
•7 47. t* M It. OX
0 47. IX O 31. OX
at 80.41 18 41. 3X
17 »1.4* 'at B4.8X
•4.8X aa ft9.ox
•0.7X 37 84. BX
98.7X ta 94. ax
M.7X
too. ox
100. OX
too. ox
too.ox
too. ox
too. ox
too. ox
too.ox
too. ox
UM.OX
too.ox
too. ox
too.ox
too.ox
too.ox
too.ox
too.ox
too.ox
too.ox
too.ox
100. OX
too. ox
too. ox
too. ox
too. ox
too. ox
too. ox
UM.OX
88.11
97. 4X
98. IX
98.7X
99. 4X
9».4X
. 99.4X
too.ox
too.ox
too.ox
too.ox
too.ox
too.ox
too.ox
too. ox
1OO.OX
1OO.OX
too.ox
too.ox
too.ox
too.ox
too.ox
too.ox
too. ox
». too.ox
100.0X
too.ox
too.ox
too.oi
FOR ALL PHASES
CHLOROBENZEME CHLOROTOLUENE
14 a.BX a o.4x
oo tx. tx ift 3. ax
to* ai. ax 04 ia.7X
104 49.7X T4 38. OX
O 49.7* 0 ».•*
79 83. 7X 4O M.BX
8? T8.8X 88 48. BX
44 83.4* Tt 81. IX
01 92.4* 9V 78.2*
14 94. 9X 89 83. 4X
o 9B.8X aa 87. sx
It 97. 7X a* 91. 3*
.4X ta 93.BX
.IX
.3*
.3X
.3X
.BX
.BX
.BX
.BX
.8*
.8*
.BX
.BX
.BX
.BX
.4X
.8X
.81
.8X
.8S
.8*
too. ox
too. ox
too.ox
too. ox
UM.OX
94.7X
98. 4X
94. 4X
97. BX
M.2X
M.4X
M.9X
M.9X
M.3X
99. 3X
99. 4X
M.4X
M.4X
M.4X
M.4X
M.4X
M.4X
99.4X
M.4X
99.4X
M.4X
M.4X
M.BX
M.BX
M.BX
UM.OX t UM.OX
OL - Oat action Unit in p»b
Fraq. - Muxt
»ar a» haw«a« vith Mhich tha dat«cti«n linit *••• uaad.
Par. - Tha oarcant af hau*aa ttittt whieK • datactian Unit !••• than or aqu*l t* tha valtia at tha laft.
-------
l/J
u
(/)
D
O
I
u.
O
ct
u
tn
5
D
Z
190
180
170
160
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10
0
FREQUENCY DISTRIBUTION OF USED) DLs
SUMMARY FOR ALL PHASES
—
-
-
_
—
BJ_
~
^
—
—
;
k
r
;
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
1
s
\
\
s
X
X
X
X
s
;
\
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
i
„
X
X
X
X
X
s
X
s
X
X
X
X
X
X
X
X
X
7
/
/
/
/
/
/
/
/
/
/
/
/
s
-^
s
X
s
X
s
X
s
s
s
s
X
X
X
X
X
X
X
X
I
i
7
/
/
/
/
/
/
^
s
s
X
X
X
X
X
s
X
X
X
X
n
X
X pi
XX
rs s
/x y\ p-[\j jxj _p^ ,_,,_,,_, ,_ __
1 1 1 1 1 1 1 1 1 1 1 1 1—
0.4O
0.80 1.20
Chlorobenzene
1.60
2.00
2.40
2.80
3.20
3.60
4.00
DETECTION LIMIT RANGE (PPB)
IXX| Chlorotoluene
-------
z
I
fe
I
FREQUENCY DISTRIBUTION OF USED DLs
FOR PHASE 1
35 -
30 -
25 -
20 -
15 -
10 -
5 -
.
y
/
X
/
/
3
7
X
'
/
,
'
/
/
,/
/
/
/
X
/
/
'
/*
/
/
X
f
^
s
V
s
s
s
s
s
s
\
V
S
s
T
f
/
It
-
s
s
s
\
s
s
s
s
\
^
— \
/ V
/ s _ 3 xl
^H Nrfl n R n . R . . n n r
0.40
0.80
1.20 ' 1.60
2.00
2.40 2.80
3.20
3.60
4.00
17*71 cttloretanane
DETECTION LMT RAN
:PPB)
Chlorotoluene
*
K.
IU
f
FREQUENCY DISTRIBUTION OF USED DLs
0.40 030
1.20
FOR PHASE 2
BO -
70 -
60 -
50 -
40 -
30-
20-
10-
:u
„
x
'
X
/
/
"
l<
X
/
/
/
X
X
^
X
X
X
•
s
s
s
X
X
\
X
X
X
X
X
7
/
y
/
/
/
/
t>
/
/
/
/
/
s
\
s
s
s
X
\
V
V
s
s
\
s
s
s
V
s
X
•
•
.
R
\
X
X
X
F^nO n5
rSrTirnrn , — FI_ — D__ — , — , — . — . — ___
240 2.40
2.80
3.20 3.60
4.00
I7"71 Chlorob«nant
DETECTION UMT RANOEJPPB)
IXNl Ct)lQrptolu«n«
-------
1C
tf>
FREQUENCY DISTRIBUTION OF USED DLs
FOR PHASE 3
40 -
35 -
30 -
25 -
20 -
15-
10 -
5 -
/
/
X
X
X
X
X
X
X
X
X
X
X
X
X
^
3
r
X
X
X
X
X
x
X
X
x
X
x
x
x
x
X
X
x
x
x
x
X
x
/
•+
s
\
\
s
s
\
s
\
s
7
X
/
X
X
/
X
/
X
X
X
X
/
X
/
X
X
X
X
/
X
X
X
X
/
-.
s
s
V
s
\
s
s
s
s
s,
s
s
s
s
s
s
\
V
s
s
s
\
X
X
/
X
/
X
X
X
X
/
X
/
X
X
/
«;
V
s
X
s
s
s
s
\
s
X
s
s
s
s
\
s
V
X
s
\
•
,
p
T
S
S
s
s
\
s
s
s
s
\
p
V
s
s
s
V
s
s
s
•>
n n
J H n
1I*IIIIIIIII.
0.40
0.60
1.20
1.60
2.C.O
2.40 2.80 3.20 3.60
4.00
OETCCTION UI*T RANgEJPPB)
Chlorobenz«ne IXXI Chlerelelutne
FREQUENCY DISTRIBUTION OF USED DLs
FDR PHASE 4
40
8
50 -
40 -
30 -
20-
10 -
x1
X
x
x
x
n
X
f
/
^
X
X
X
x
x
x
X
X
X
X
X
X
X
x
X
X
X
_
s
s
>
s
s
•s
s
>
s
s
V
X
X
x
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X1
X
s
\
s
s
\
s
>
N
s
s
s
V
s
s,
s
X
X
X
/
/
/
X
X1
X
v
s
s
s
s
s
s
s
^s
s
s
s
\
s
s
s
s
s
•••
l_
r
y
p
s
s
•^
•s
s
\
S
s
s
s
\
N
S
S
s
s
NrR n , n , , , ,
0.40
0.80
1.20
1.60 2AO
2.40 2.80 3.20 3.60
4.00
DETECTION UMT H*N£EJPPB)
rvN] Chlerot0lu«nt
-------
APPENDIX D
Reasons for Denial of Permission to
Sample Love Canal Properties
Love Canal EDA Habitability Study
Prepared by
New York State Department of Health
Albany, New York
WDC 63394.T1 (SA)
-------
STATE OF NEW YORK - DEPARTMENT OF HEALTH
INTEROFFICE MEMORANDUM
TO: T1m Van Epp
CH2M Hill
FROM: Edward G. Horn, Ph.D., Environmental Scientist
Division of Environmental Health Assessment
SUBJECT: Reasons for Denial of Permission to Sample Love Canal Properties
DATE: June 14, 1988
The attached tables were prepared in response to a request by peer
reviewers of the Love Canal Emergency Declaration Area Habitability Study
for the reasons that homeowners and residents of the Emergency Declaration
Area denied permission to take samples from their property.
The two tables summarize what information could be found for the Air
Assessment of Indicator Chemicals (samples taken in homes) and for the Soil
Assessment of 2,3,7,8-TCDD. The reasons for denial of permission were
ascertained at the time through follow-up phone calls and, in some cases,
personal visits.
The Soil Assessment of Indicator Chemicals did not require the same
follow-up because the random selection of 50 or 75 samples per sampling area
did not necessitate maximum participation of homeowners. A summary of
permissions and denials could not be prepared in the time available.
However, the same individuals were involved as were in the Air Assessment,
and no significant number of additional denials were encountented in the Soil
Assessment of Indicator Chemicals. /• '
Attachment
cc: William Stasiuk
Doug Garbarini, EPA Region II
-------
DIOXIN SAMPLING
REASONS FOR REFUSING PERMISSION
Reason Number of times
Reason unknown/ 15
No response to mailing*
Responded "no" to participation and 1
asked not to be contacted again
* Property owners were contacted at least once by mail, and
follow-up was done by phone or mail as needed. Persons not
responding to the first mailing were either telephoned or
mailed a follow-up before the end of the sampling program.
One owner refused to participate but two samples were taken from his
lot in error.
One owner originally refused to participate due to concern over his lawn
being damaged but later allowed sampling.
Sixteen owners were opposed to the study, its purpose and its costs but
sold to LCARA before sampling ended. Samples were then collected on
these properties.
These 18 instances are not reflected in the above count.
NYSDOH
14 June 1988
-------
AIR SAMPLING
REASONS FOR REFUSING PERMISSION
Reason Number of horns*
Permission refused
Reasons unknown 9
(No Response to mall and phont Inquiries)*
Owner said no but didn't state a reason 8
Owner said yes but tenant refused access 3
Tenant opposed to study 1
Tenant reasons unknown 2
Husband said yes but wife refused 3
for housecleanlng reasons
Concerns about Integrity of sampling team 2
Owner said no, house was tested enough 2
Home not accessible
Structure physically Inaccessible . 4
Scheduling problems (resident couldn't.be 2
be there at stated times)
Unable to coordinate access for unoccupied 1
building
Owner died and no contact existed for estate 1
-*~People were contacted at least twice by ma11 In every -
Instance, and telephone follow-up was always attempted.
When telephone outreach was unsuccessful, the sampHnq
was asked to leave a note for the resident or try to talk
to them personally.
One person refused sampling due to concern that chemicals would be
detected; but she reconsidered and a sample was taken.
Some residents refused access because they were opposed to the study,
but they sold to LCARA In time for LCARA to allow air sampling.
These two Instances are not reflected 1n the above count.
NYSDOH
14 June 1988
-------
APPENDIX E
List of Errata
Air Assessment—Indicator Chemicals
WDC 63394.T1 (SA)
-------
Appendix E
LIST OF ERRATA
AIR ASSESSMENT—INDICATOR CHEMICALS
One erratum was identified that involves a change in content
to Volume II. This erratum is defined below.
ERRATUM
The air assessment report incorrectly stated the analytic
results of the sampling during Phase 2 of the house where CT
was detected. The following text should be changed:
1-2
Line
27
1-2
15-17
6-11
20
6-11
26
6-11
29
Erratum
"in the basement samples collected, suggesting"
should read "in the basement samples collected
during the initial sampling and investigation,
suggesting"
"during Phase 2 of the study, chlorotoluene was
detected on the main floor of the unoccupied,
single-story structure in Neighborhood 6. The
highest concentration detected was 3.4 ppb" should
read "during Phase 2 of the study chlorotoluene was
detected in an unoccupied, single-story structure in
Neighborhood 6. During the initial sampling and
investigation chlorotoluene was only detected on the
main floor of the home. Lower levels of
chlorotoluene were detected both on the main floor
and in the basement during the home's second sampling
in Phase 2. The highest concentration detected was
3.4 ppb on the main floor during the home's initial
sampling."
"Chlorotoluene was not detected in the basement."
should read "Chlorotoluene was not detected in the
basement during this initial sampling and
investigation."
"Levels in all six of the first floor rooms were
between the quahtitation and" should read "levels in
the main floor and the basement were between the
quantitation and"
"This investigation did not detect chlorotoluene in
the air in the basement of the home during either
sampling effort" should be deleted.
WDR358/052
E-l
-------
6-11 26 "Levels in all six of the first floor rooms were
between the quantitation and" should read "levels in
the main floor and the basement were between the
quantitation and"
6-11 29 "This investigation did not detect chlorotoluene in
the air in the basement of the home during either
sampling effort" should be deleted.
WDR358/052
E-2
WDR358/052/DRAFT/7-15-88
-------
APPENDIX F
Justification for Nonparametric
Statistical Comparisons
Soil Assessment-Indicator Chemicals
WDC 63394.T1 (SA)
-------
Appendix F
JUSTIFICATION FOR NONPARAMETRIC
STATISTICAL COMPARISONS
SOIL ASSESSMENT—INDICATOR CHEMICALS
In response to discussions with Dr. Stoline during the peer
review of Volume III, a summary of the reasons for using
nonparametric statistics for the analysis of the soil assess-
ment for LCIC data has been prepared. This summary recapit-
ulates discussions in the proposed habitability criteria
document (NYSDOH and DHHS/CDC, 1986), the pilot study for
soil LCIC (CH2M HILL, 1987) and the peer review of the pilot
study for soil LCIC (Life Systems, 1987).
APPROACH
Following the recommendations of the Love Canal Emergency
Declaration Area; Proposed Habitability Criteria (NYSDOH and
DHHS/CDC, 1986) and the Peer Review of the Love Canal
Full-Scale Sampling Plan (Life Systems, 1987), the goal of
the soil assessment for indicator chemicals was to design
and implement a sampling program that would achieve a
90 percent power of detecting an order-of-magnitude differ-
ence between an EDA sampling area and a comparison area with
95 percent confidence. The design of the soil assessment
was thus centered around a statistical comparison approach.
Statistical methods can be broadly divided into two groups:
parametric and nonparametric. Parametric methods require
knowledge (or an assumption) about the statistical distribu-
tion that generated the data; nonparametric methods do not.
The disadvantage of nonparametric methods is that they can
be less powerful than parametric methods in cases where the
assumptions of the parametric methods hold. The Wilcoxon
rank sum test (a nonparametric test), however, is practical-
ly as powerful as the t-test (a parametric test) for the
two-sample comparison problem under the normal distribution,
equal variance assumptions required by the t-test; the power
is 95 percent of the t-test (Lehmann, 1975, pp. 76-81).
If the assumptions required by a parametric test do not
hold, the true alpha-level of the parametric test (the prob-
ability of incorrectly declaring the EDA sampling area to be
different from the comparison area) can be grossly different
from the assumed alpha-level, rendering the statistical
results invalid. On the other hand, the alpha-level of a
nonparametric test does not depend on the underlying
distribution.
F-l
-------
Two-sample parametric tests are usually designed to detect
specific kinds of differences between two different distribu-
tions, such as location shifts, scale shifts, multiplicative
shifts, or upper tail shifts. Although nonparametric tests
are often used with a specific kind of shift in mind, they
can be effective at detecting other kinds of shifts as well.
The LCIC data from both the pilot study and the full-scale
study display two important characteristics that must be
accounted for by any statistical method used to perform the
comparisons:
o The distribution of the concentration estimates in
most sampling areas are skewed to the right. The
concentration estimates are bounded below by zero
and exhibit a few very large values.
o The number of non-detect concentrations for some
LCICs is quite large.
The major arguments for using a nonparametric method, as
opposed to a parametric method, for performing the
statistical comparisons can be summarized as follows:
1. The LCIC data do not conform to a normal distribution,
which is the usual distribution assumed for parametric
tests.
2. Although some transformation (e.g., natural logarithm)
of the LCIC data may induce a distribution that approxi-
mates a normal distribution, the question remains as to
how to deal with non-detects (since a non-detect cannot
necessarily be assumed to equal zero or any other arbi-
trarily chosen value). Several methods are suggested
in the literature for dealing with non-detects, which
includes Gilbert and Kinnison (1981), Gillion and
Helsel (1986), Gleit (1985), and Owen and DeRouen
(1980). No method, however, is universally accepted.
3. A Monte Carlo simulation performed during the design of
the sampling program and described in Appendix B of
Volume III showed that a nonparametric test—the
Wilcoxon rank sum test--gave the best performance in
terms of power and maintenance of the nominal
alpha-level over a wide range of distributions,
censoring percentages, and sample sizes.
WDR361/030
F-2
-------
REFERENCES
CH2M HILL, 1987. Pilot Study for Love Canal EDA
Habitability Study. Volumes I and II.
Gilbert, R.O. and Kinnison, R.R. 1981. "Statistical Methods
for Estimating the Mean and Variance From Radionuclide
Data Sets Containing Negative, Unreported, or Less-Than
Values," Health Physics, 40, 377-390.
Gilliom, R.J., and Helsel, D.R. 1986. "Estimation of
Distributional Parameters for Censored Trace Level
Water Quality Data, 1, Estimation Techniques," Water
Resources Research, 22, 135-146.
Gleit, A. 1985. "Estimation for Small Normal Data Sets With
Detection Limits," Environmental Science and
Technology, 19, 1201-1206.
Lehmann, E.L. 1975. Nonparametrics: Statistical Methods
Based on Ranks. Holden-Day, San Francisco.
Life Systems. 1987. Peer Review of the Love Canal
Full-Scale Sampling Plan.
NYSDOH and DHHS/CDC. 1986. Love Canal Emergency
Declaration Area; Proposed Habitability Criteria.
Owen, W.J., and DeRouen, T.A. 1980. "Estimation of the Mean
For Lognormal Data Containing Zeroes and Left-Censored
Values, With Application to the Measurement of Worker
Exposure to Air Contaminants," Biometrics, 36, 707-719.
WDR361/030
F-3
-------
APPENDIX G
Empirical and Fitted Distribution of LCICs by EDA
Sampling Area and Retrospective Power
Analysis by EDA Sampling Area and
Neighborhood Using Fitted Distributions
Soil Assessment-Indicator Chemicals
-------
Appendix G
EMPIRICAL AND FITTED DISTRIBUTION OF LCICs BY EDA
SAMPLING AREA AND RETROSPECTIVE POWER ANALYSIS BY EDA
SAMPLING AREA AND NEIGHBORHOOD USING FITTED DISTRIBUTIONS
Following the request of Dr. Stoline in the written prelimi-
nary comments, this appendix presents tables of the percen-
tiles of the empirical and lognormal mixture best-fit
distributions for each LCIC in each sampling area.
Table G-l (a through h) gives the actual percentile of the
empirical concentrations from the soil LCIC assessment,
i.e., the 90th percentile value is that concentration value
from the data set which is larger than 90 percent of the
other observed values. (Only Good data are presented in the
tables in this appendix.)
Table G-2 (a through h) gives the estimated percentile
values of the lognormal mixture fit, assuming that the fit-
ted distribution is the correct distribution. The 90th per-
centile in this table is the theoretical value which is
exactly greater than 90 percent of the potential observa-
tions from this distribution.
The fitted lognormal mixture distribution table has several
entries for which the maximum likelihood estimation proce-
dure failed to converge. This occurred when the lognormal
mixture model was a sufficiently inadequate model for a par-
ticular LCIC and sampling area that the percentile could not
be estimated.
Table G-2 also shows the estimated lognormal mixture parame-
ters when these are estimatable. These parameters (all in
log units) are the two mean values (uyl, uy2), the common
standard deviation (sy), and the mixing fraction (p). Note
that these parameters are all interdependent so that the
fitted values for one area cannot be compared singly to
those from another area, i.e., uyl of Sampling Area 3 cannot
be compared with uyl of CT225.
Figure G-l shows the cumulative frequency distribution for
both the empirical ('+' symbols) and the fitted (solid line)
lognormal mixture distributions for each LCIC and each area.
The parameters of the fitted lognormal mixture distributions
are used to calculate the retrospective power of the LCIC
soil assessment. Tables G-3 (univariate) and G-4 (multivar-
iate) show estimates of the retrospective power for the
sampling area comparisons with a sample size of 75. These
tables correspond to Tables B-5 and B-6 of Appendix B of
Volume III of the Habitability Study. Whereas the
Volume III tables used a fitted lognormal distribution,
G-l
-------
Tables G-3 and G-4 use the fitted lognormal mixture
distribution.
Although the estimates for retrospective power change, the
general result is unchanged: the goal of achieving 90 per-
cent power with 95 percent confidence for an order-of-
magnitude difference (delta = 10) is met.
Tables G-5 (univariate) and G-6 (multivariate) show the
estimates of retrospective power for neighborhood compari-
sons with a sample size of 35. The estimated power for
comparisons at the neighborhood level (with sample sizes
around 35) also meets the goal, except for beta-BHC, which
has an estimated power of 85 percent with sample size 35.
Note that these simulations assume that the sample size in
the comparison area is the same as the EDA. Since the com-
parison areas had more than 35 samples, these estimates of
power are likely to be biased low.
WDR359/033
G-2
-------
Table G-la
CONCENTRATION EMPIRICAL PERCENTILES
1,2-DICHLOROBENZENE
SOIL ASSESSMENT—INDICATOR CHEMICALS
Sampling
Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
Percentile
25th
0.24
0.32
0.29
0.76
0.28
0.30
0.28
0.30
0.26
0.32
50th
0.36
0.39
0.43
1.01
0.39
0.40
0.36
0.39
0.36
0.41
75th
0.58
0.53
0.65
1.51
0.63
0.63
0.50
0.58
0.56
0.57
90th
0.93
0.81
0.84
2.68
1.03
0.90
0.66
0.93
0.82
0.77
95th
1.01
1.06
0.99
4.22
2.67
1.24
0.91
1.20
1.02
0.86
99th
1.38
1.15
1.40
5.65
19.80
2.37
2.23
3.19
1.51
0.94
Table G-lb
CONCENTRATION EMPIRICAL PERCENTILES
1,2,4-TRICHLOROBENZENE
SOIL ASSESSMENT—INDICATOR CHEMICALS
Sampling
Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
Percentile
25th
0.11
0.46
0.33
5.63
0.51
0.55
0.29
0.31
0.27
0.36
50th
0.14
0.65
0.61
8.67
0.91
0.89
0.44
0.50
0.37
0.59
75th
0.19
0.98
1.18
13.76
1.79
1.50
0.96
1.13
0.55
0.95
90th
0.28
1.13
2.84
24.74
4.97
3.20
1.97
3.01
0.88
1.46
95th
0.31
1.85
3.90
41.24
8.93
4.91
4.05
4.07
1.15
2.97
99th
0.92
3.12
33.07
45.11
167.33
7.65
12.08
35.65
3.51
4.49
WDR361/011/1
-------
Table G-lc
CONCENTRATION EMPIRICAL PERCENTILES
1,2,3,4-TETRACHLOROBENZENE
SOIL ASSESSMENT—INDICATOR CHEMICALS
Sampling
Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
Percentile
25th
0.03
0.40
0.25
8.06
0.67
0.54
0.25
0.22
0.20
0.30
50th
0.05
0.53
0.61
11.48
1.17
1.07
0.39
0.43
0.31
0.54
75th
0.07
0.76
1.23
22.43
2.27
1.89
0.98
1.05
0.53
1.01
90th
0.13
0.97
2.81
45.42
7.94
2.78
2.35
2.61
0.90
1.61
95th
0.18
1.16
12.51
52.41
10.84
4.50
8.46
6.61
1.14
2.64
99th
0.84
1.33
64.25
67.20
66.66
9.55
182.41
168.64
2.90
3.91
Table G-ld
CONCENTRATION EMPIRICAL PERCENTILES
2-CHLORONAPHTHALENE
SOIL ASSESSMENT—INDICATOR CHEMICALS
Sampling
Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
Percentile
25th
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
50th
0.03
0.06
0.07
ND
0.04
0.04
0.04
0.06
0.06
0.06
75th
0.08
0.10
0.10
0.09
0.07
0.09
0.88
0.88
0.10
• 0.10
90th
0.10
0.13
0.12
0.16
0.10
0.10
0.11
0.11
0.12
0.13
95th
0.13
0.16
0.14
0.18
0.14
0.12
0.12
0.12
0.15
0.15
99th
0.21
0.22
0.32
0.21
0.24
0.15
0.13
0.16
0.19
0.18
WDR361/011/2
-------
Table G-le
CONCENTRATION EMPIRICAL PERCENTILES
ALPHA-BHC
SOIL ASSESSMENT—INDICATOR CHEMICALS
Sampling
Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
ED A 5
EDA6
EDA7
Percentile
25th
ND
0.06
ND
4.98
0.19
0.04
ND
ND
ND
0.05
50th
ND
0.18
0.11
8.25
0.40
0.22
0.13
0.13
0.07
0.14
75th
ND
0.29
0.59
18.06
0.85
0.58
0.76
0.35
0.22
0.26
90th
ND
0.42
2.16
26.07
2.14
1.27
3.87
1.47
0.45
0.51
95th
0.01
0.66
3.13
35.09
5.65
3.98
16.00
5.28
0.53
0.95
99th
0.17
1.19
33.97
69.70
100.31
16.05
152.53
83.51
1.99
2.42
Table G-lf
CONCENTRATION EMPIRICAL PERCENTILES
DELTA-BHC
SOIL ASSESSMENT—INDICATOR CHEMICALS
Sampling
Area
Buff
CT221
CT225
EDA1
EDA2
ED A 3
EDA4
EDA5
EDA6
EDA7
Percentile
25th
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
50th
ND
ND
ND
1.13
0.04
ND
ND
ND
ND
ND
75th
ND
ND
ND
1.93
0.25
0.12
ND
ND
ND
ND
90th
ND
ND
0.27
3.74
0.59
0.42
0.29
0.16
0.06
0.13
95th
ND
0.16
0.49
9.83
1.07
0.75
0.95
0.66
0.23
0.30
99th
0.00
0.94
5.40
38.83
2.33
79.99
3.04
9.96
3.44
1.25
WDR361/011/3
-------
Table G-lg
CONCENTRATION EMPIRICAL PERCENTILES
BETA-BHC
SOIL ASSESSMENT—INDICATOR CHEMICALS
Sampling
Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
Percentile
25th
ND
ND
ND
4.64
ND
ND
ND
ND
ND
ND
50th
ND
ND
ND
11.58
0.20
0.13
ND
ND
ND
ND
75th
ND
0.10
0.46
26.63
0.66
0.64
0.47
0.17
ND
0.31
90th
ND
0.44
1.41
47.88
1.80
2.02
18.26
1.21
0.31
0.80
95th
ND
0.80
15.61
72.16
3.98
15.69
538.82
13.84
0.59
1.37
99th
5.36
1.03
50.57
101.72
29.97
51.17
4,108.31
663.49
3.15
2.01
Table G-lh
CONCENTRATION EMPIRICAL PERCENTILES
GAMMA-BHC
SOIL ASSESSMENT—INDICATOR CHEMICALS
Sampling
Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
Percentile
25th
ND
ND
ND
0.79
ND
ND
ND
ND
ND
ND
50th
ND
ND
ND
1.73
0.09
ND
ND
ND
ND
ND
75th
ND
ND
0.10
3.23
0.24
0.20
0.12
0.04
ND
ND
90th
ND
0.26
0.52
9.17
1.03
0.46
0.81
0.33
0.14
0.13
95th
ND
0.49
1.76
12.16
1.83
1.92
16.33
4.43
0.37
0.35
99th
0.04
80.81
15.83
20.99
6.67
12.68
85.60
26.33
0.89
0.61
WDR361/011/4
-------
Table G-2a
CONCENTRATION LOGNORMAL MIXTURE PERCENTILES
1,2-DICHLOROBENZENE
SOIL ASSESSMENT—INDICATOR CHEMICALS
Percentile
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
ED A 7
25 50
.24 .35
.30 .39
.30 .42
.71 1.10
.26 .40
.30 .41
.25 .36
.29 .40
.26 .35
.31 .40
ESTIMATED
uyl
-1.304
-1.034
-1.159
.120
-.942
-.973
-.977
-1.006
-1.227
-1.075
75
.58
.52
.63
1.72
.64
.58
.51
.58
.56
.57
PARAMETERS
uy2
-.365
-.222
-.415
.094
1.908
.034
-5.891
.075
-.356
-.428
90
.87
.78
.83
2.57
1.03
.90
.69
.94
.81
.75
sy
.413
.306
.303
.661
.621
.428
.487
.421
.302
.270
95 99
1.07 1.52
.96 1.29
.95 1.22
3.27 5.14
1.78 10.80
1.20 1.89
.83 1.16
1.28 1.99
.95 1.23
.86 1.08
P
.661
.824
.559
.155
.956
.872
.950
.858
.689
.670
WDR359/032/1
-------
Table G-2b
CONCENTRATION LOGNORMAL MIXTURE PERCENTILES
1,2,4-TRICHLOROBENZENE
SOIL ASSESSMENT—INDICATOR CHEMICALS
Percentile
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
25
.10
.49
.34
4.96
.55
.49
.28
.32
.25
.37
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA 3
EDA4
EDA5
EDA6
EDA7
50
.14
.65
.63
8.97
1.01
.85
.49
.58
.39
.57
ESTIMATED
uyl
-2.041
-.456
-.507
-.352
-.023
-.283
-.802
-.648
-.952
-.621
75
.20
.86
1.18
15.08
1.90
1.52
.90
1.14
.60
.91
PARAMETERS
uy2
-1.642
1.056
3.262
2.296
4.418
1.178
1.635
1.699
-.952
.907
90 95
.27 .33
1.16 1.52
2.25 3.97
23.65 30.85
3.57 5.93
2.82 4.28
1.91 4.06
2.64 5.45
.90 1.15
1.54 2.35
sy
.485
.393
.884
.710
.891
.734
.791
.827
.661
.600
99
.47
3.75
42.54
50.56
119.02
9.04
12.57
15.68
1.80
4.98
P
.803
.961
.966
.103
.971
.882
.923
.905
.601
.919
WDR359/032/2
-------
Table G-2c
CONCENTRATION LOGNORMAL MIXTURE PERCENTILES
1,2,4,5-TETRACHLOROBENZENE
SOIL ASSESSMENT—INDICATOR CHEMICALS
Percentile
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
25
.04
.39
.28
6.29
.66
.53
.22
.25
.19
.31
50
.05
.54
.57
11.97
1.20
.99
.44
.49
.32
.53
75
.08
.73
1.20
22.35
2.35
1.84
.93
1.01
.52
.94
90
.11
.94
2.95
39.01
6.13
3.21
2.14
2.18
.82
1.57
95
.15
1.07
8.84
54.37
14.09
4.49
7.25
4.97
1.06
2.10
99
.64
1.38
44.23
102.78
39.15
8.39
57.61
90.24
1.75
3.54
ESTIMATED PARAMETERS
Sampling Area
uyl
_uy2
sy
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
-2.967
-.394
-.656
2.515
.062
.031
-.884
-.767
-1.121
-.931
-.585
-.999
2.765
-2.128
2.605
-.056
3.156
3.778
-6.087
.030
.562
.335
.982
.908
.816
.919
.992
.996
.724
.646
.975
.605
.933
.972
.898
.524
.945
.958
.970
.655
WDR359/032/3
-------
Table G-2d
CONCENTRATION LOGNORMAL MIXTURE PERCENTILES
2-CHLORONAPHTHALENE
SOIL ASSESSMENT—INDICATOR CHEMICALS
Percentile
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
25
.03
.04
.04
*
.03
.04
.04
.04
.04
.04
50
.05
.06
.07
a
.05
.05
.05
.06
.06
.06
75
.07
.10
.10
a
.07
.09
.08
.09
.10
.10
90
.10
.13
.13
a
.11
.10
.10
.11
.13
.13
95
.13
.15
.16
a
.13
.12
.11
.12
.14
.14
99
.20
.19
.21
a
.20
.13
.13
.16
.17
.17
ESTIMATED PARAMETERS
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
uyl
uy2
-3.005
-2.249
-2.513
-2.312
-2.930
.335
.160
.450
.288
-2.
-3.
-2.
-2.
-3.657
-3.132
-4.193
-6.293
-3.433
.157
.353
.320
,280
sy
-3.
•2.
-3.
-3,
-3.142
-2.217
.621
.277
.433
.449
.600
.174
.185
.282
.253
.231
.850
.451
.787
.475
.769
.333
.679
.521
.490
.584
Percentile estimation algorithm failed to converge.
WDR359/032/4
-------
Table G-2e
CONCENTRATION LOGNORMAL MIXTURE PERCENTILES
ALPHA-BHC
SOIL ASSESSMENT—INDICATOR CHEMICALS
Percentile
Sampling
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
Area 25
.10
.06
.01
5.16
.14
.03
.03
.02
.02
.05
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
50
.12
.17
.10
9.67
.36
.23
.12
.09
.08
.14
ESTIMATED
uyl
-2.090
-1.432
-6.469
2.386
-1.043
-.959
2.883
-2.571
-1.521
-1.672
75
.15
.30
.47
17.02
.98
.65
.57
.37
.23
.29
PARAMETERS
uy2
-2.090
-3.229
-1.757
.554
4.347
-7.000
-2.311
2.577
-3.751
-4.563
90
.19
.46
1.75
27.76
2.52
1.47
3.42
1.66
.42
.53
sy
.318
.611
2.000
.771
1.447
1.202
1.926
2.000
.759
.894
95
.21
.58
3.78
37.08
4.70
2.36
15.51
5.11
.59
.75
99
.26
.91
15.53
63.67
40.98
5.62
182.52
70.00
1.06
1.40
P
.806
.690
.191
.890
.985
.758
.087
.952
.515
.812
WDR359/032/5
-------
Table G-2f
CONCENTRATION LOGNORMAL MIXTURE PERCENTILES
DELTA-BHC
SOIL ASSESSMENT—INDICATOR CHEMICALS
Percentile
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
No feasible lognormal
25 50
a a
.01 .01
.00 .00
.97 1.83
.00 .05
.00 .00
.00 .00
.00 .01
.00 .00
.00 .01
ESTIMATED
uyl
a
-4.666
-7.000
.171
-1.392
-7.000
-7.000
.644
-7.000
-5.389
mixture fit.
75
a
.02
.01
3.49
.26
.08
.01
.04
.00
.01
PARAMETERS
uy2
a
-.736
-1.322
1.201
-5.522
-1.271
-1.540
-4.532
-1.764
-1.493
90 95
a a
.03 .06
.20 .55
6.15 8.54
.59 .90
.56 1.20
.27 .64
.14 .42
.04 .23
.13 .33
sy
a
.809
1.343
.765
.981
1.460
1.400
1.668
1.476
1.008
99
a
.81
2.19
15.48
1.90
4.33
2.35
6.76
1.29
.97
P
a
.959
.830
.568
.524
.687
.770
.045
.883
.859
WDR359/032/6
-------
Table G-2g
CONCENTRATION LOGNORMAL MIXTURE PERCENTILES
BETA-BHC
SOIL ASSESSMENT—INDICATOR CHEMICALS
Percentile
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
25
a
.00
.00
5.35
.00
.00
.01
.00
.00
.01
50
a
.00
.02
12.57
.22
.08
.06
.00
.00
.03
75
a
.10
.47
28.65
.84
.83
.30
.18
.00
.32
90
a
a
2.33
47.74
2.18
2.91
5.03
2.06
.33
.75
95
a
a
5.39
62.61
3.74
5.77
137.77
5.78
a
1.14
99
a
a
23.99
99.97
10.09
19.09
1868.44
32.62
a
2.34
ESTIMATED PARAMETERS
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
uyl
-1.021
-7.000
-.742
3.336
-.602
-7.000
-3.125
uy2
sy
000
000
-1
679
370
000
668
000
374
000
289
215
-.929
-4.472
.001
.956
.894
.619
.332
.586
.000
.000
.941
.897
.509
.701
.497
.509
.667
.450
.903
.674
.778
.421
Percentile estimation algorithm failed to converge.
WDR359/032/7
-------
Table G-2h
CONCENTRATION LOGNORMAL MIXTURE PERCENTILES
GAMMA-BHC
SOIL ASSESSMENT—INDICATOR CHEMICALS.
Percentile
Sampling Area
Buff
CT221
,CT225
EDA1 1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
Sampling Area
Buff
CT221
CT225
EDA1
EDA2
EDA3
EDA4
EDA5
EDA6
EDA7
No feasible lognormal
25 50
a a
.00 .01
.00 .00
.25 2.29
.00 .07
.00 .00
.00 .02
.00 .01
.01 .02
.01 .02
ESTIMATED
uyl
a
-5.040
-.963
.598
-1.856
-7.000
-4.332
-4.782
-4.315
-1.769
mixture fit.
75 90 95
a a a
.03 .10 .27
.05 .64 1.51
4.20 7.26 10.08
.27 .80 1.47
.20 .73 1.36
.09 .80 6.96
.05 .45 3.25
.03 .14 .31
.04 .15 .26
PARAMETERS
uy2 sy
a a
4.244 2.000
-7.000 1.514
1.044 .872
-7.000 1.528
-1.269 1.337
2.087 2.000
1.048 1.725
-1.357 .770
-4.165 .742
99
a
63.52
5.87
18.64
4.48
4.06
97.65
27.82
.75
.55
P
a
.981
.274
.484
.700
.579
.907
.894
.876
.175
WDR359/032/8
-------
Table G-3
POWER (AT ALPHA = 0.05) FOR UNIVARIATE TESTS BASED ON
LOGNORMAL MIXTURE PARAMETERS GIVEN IN TABLE G-2 FOR
GOOD DATA GIVEN IN TABLE B-4 OF FINAL REPORT VOLUME III,
WITH SAMPLE SIZE 75
LCICb
Delta3
.00 .046 .051 .052 .046 .048 .046 .047 .043
.05
.10
.20
.50
1.00
2.00
5.00
10.00
.173
.408
.847
1.000
1.000
1.000
1.000
1.000
.093
.187
.459
.961
1.000
1.000
1.000
1.000
.076
.119
.253
.678
.972
1.000
1.000
1.000
.066
.096
.200
.525
.878
.996
1.000
1.000
.062
.096
.136
.338
.676
.958
1.000
1.000
.066
.096
.200
.525
.878
.996
1.000
1.000
.059
.072
.100
.198
.360
.678
.943
.996
.059
.072
.105
.259
.520
.827
.991
1.000
LCIC concentrations in EDA sampling area are drawn from a statistical
population related to that of the comparison area by the factor
1+delta; delta=10.0 corresponds to an order of magnitude shift.
LCIC 1 = 1,2-Dichlorobenzene 2 = 1,2,4-Trichlorobenzene
3 = 1,2,3,4-Tetrachlorobenzene 4 = Chloronaphthalene
5 = alpha BHC 6 = delta BHC
7 = beta BHC 8 = gamma BHC
WDR359/030/2
-------
Table G-4
POWER FOR MULTIVARIATE TESTS FOR PARAMETERS GIVEN
IN TABLE G-2 FOR INTERVARIABLE CORRELATION
GIVEN IN TABLE B-7 OF FINAL REPORT VOLUME III,
FOR SAMPLE SIZE 75
ALPHA
Delta3 .0100 .0250 .0500 .1000 .2000
.00 .007 .026 .059 .095 .203
.05 .009 .022 .058 .113 .217
.10 .008 .030 .062 .128 .238
.20 .025 .057 .108 .175 .328
.50 .182 .292 .407 .539 .705
1.00 .686 .797 .873 .931 .965
2.00 .979 .988 .995 .998 .999
5.00 .999 .999 1.000 1.000 1.000
10.00 1.000 1.000 1.000 1.000 1.000
LCIC concentrations in EDA sampling area are drawn from a
statistical population related to that of the comparison
area by the factor 1+delta; delta=10.0 corresponds to an
order of magnitude shift.
WDR359/031/2
-------
Delta3
Table G-5
POWER (AT ALPHA = 0.05) FOR UNIVARIATE TESTS BASED ON
LOGNORMAL MIXTURE PARAMETERS GIVEN IN TABLE G-2 FOR
GOOD DATA GIVEN IN TABLE B-4 OF FINAL REPORT VOLUME III,
WITH SAMPLE SIZE 35&
LCICb
.00
.05
.10
.20
.50
1.00
2.00
5.00
.048
.118
.258
.580
.986
1.000
1.000
1.000
.060
.106
.162
.280
.728
.990
1.000
1.000
.046
.082
.116
.178
.424
.784
.986
1.000
.060
.148
.250
.602
.992
1.000
1.000
1.000
.048
.058
.075
.111
.231
.462
.780
.978
.046
.072
.100
.160
.328
.614
.914
.996
.062
.068
.078
.092
.146
.254
.416
.734
.046
.062
.072
.098
.164
.318
.562
.888
10.00 1.000 1.000 1.000 1.000 1.000 1.000 .882 .968
LCIC concentrations in EDA sampling area are drawn from a statistical
population related to that of the comparison area by the factor
1+delta; delta=10.0 corresponds to an order of magnitude shift.
LCIC 1 = 1,2-Dichlorobenzene 2 = 1,2,4-Trichlorobenzene
3 = 1,2,3,4-Tetrachlorobenzene 4 = Chloronaphthalene
5 = alpha BHC 6 = delta BHC
7 = beta BHC 8 = gamma BHC
WDR359/030/1
-------
Table G-6
POWER FOR MULTIVARIATE TESTS FOR PARAMETERS GIVEN
IN TABLE G-2 FOR INTERVARIABLE CORRELATION
GIVEN IN TABLE B-7 OF FINAL REPORT VOLUME III,
FOR SAMPLE SIZE 35
ALPHA
Delta5 .0100 .0250 .0500 .1000 .2000
.00 .006 .012 .038 .084 .202
.05 .006 .014 .042 .096 .212
.10 .006 .014 .048 .102 .222
.20 .010 .030 .056 .118 .242
.50 .032 .084 .164 .288 .454
1.00 .172 .320 .468 .604 .758
2.00 .556 .664 .752 .860 .936
5.00 .812 .924 .956 .978 .994
10.00 .914 .970 .982 .990 .994
LCIC concentrations in EDA sampling area are drawn from a
statistical population related to that of the comparison
area by the factor 1+delta; delta=10.0 corresponds to an
order of magnitude shift.
WDR359/031/1
-------
50.00
fw.oo
8
ESO.OO
220.00
-10.00
EDA I OBSERVATIONS
FITTOJ oisntiBuriON
OjnULATIVE PROBABILITY IP) EXCEEDANCE PROBA8IUTT ll-P)
44tt f M • •••••• I* I 1144441 41
PROBABILITY IPI EXCEEDANCE PROBABILITY II-PI
70.00
j-eo.ro
Iso.oo
140.00
gx.oo
u
•20.00
tn
N
-10.00
0
EDfi I OBStRVRTIONS
FITTED OISTRIBUTIIW
.20
£.13
1.10
i-08
CM
.05
.03
« «.l4Jlt • tt •»•*••• • I tt«4.l« M
CimULflTIVE PROB081L1TT IP) EXCEEDWO; PftOBflBlLITT 11-P)
mum EDA I OBSERVATIONS
FITTED DISTRIBUTION
CUnULBTIVE PROBABILITY IP) EXCEEDANCE PROBABILITY ll-P)
OMULflTIVE PROBflBlLITT IPI EXCEEOBNCE PW»flBILITT ll-f I
• I 144.1
CUnULRTIVE PftOBflBILlTT IP) EXCEEOBNCE PR080BII.ITT ll-P)
120.00
H (I H I! EDA 1 OBSERVATIONS
FITTED DISTRIBUTION
CUnULflTIVE PROBflSILITT IPI EXCEEOflNCE PROBflBlLITT ll-P)
CUnULBTIVE PROBBBILITY IP) EXCEEORNCE PROBflBILirr ll-P)
Figure G-1a SOIL ASSESSMENT •• INDICATOR CHEMICALS
CUMULATIVE FREQUENCY DISTRIBUTIONS FOR LCICs IN EDA SAMPLING AREA 1
-------
• «.1 4 J I ' I I M • »•»••• M t
CUnUlflTIVE PROBflBILITT (PI EXCEEOflNCE PROBRBILITY II-PI
ITS .00
SISO.OO
t
oilOO.OO
£ 75.00
E
A SO. 00
fg
zs.oo
0
EOA 2 OBSERVATIONS
FITTED DISTRIBUTION
t I r
OjnUimiVE PROBflBILITT (PI CXCEEOPNCE PRtBSBILITT II-PI
70.00
CUtlULflTIVE PROBS8ILITT IP) EXCEEOfiNCE PR06B8ILITT (1-PI
CUnULATIVE PROBflBILITT (PI EXCEEOflNCE PROBABILITY (1-PI
120.00
« 90.00
M
I 60.00
30.00
EDfi 2 OBSERVflTIONS
FITTED DISTRIBUTION
2.SO
"' i 'irti 'i i'i1 lJllttlllliai
-------
2.SO
|E?.oo
*
w
zl.SO
I i.oo
5
.so
COR 3 OaScdVBTIONS
FITTED DISTRIBUTION
8.00
m M a 4 J I ' I I • • »••••• w I VI44.I41 4>
CunULHTIVE PROBRBILITY IPI EXCEEOflNCE PTOBR8ILITY II-PI
4411 •
M • t
in EXCEEOPNCC PROBRBILITY II-PI
10.00
- s.oo
*
6.00
tC 2.00
EOR 3 OBSERVATIONS
FITTED DISTRIBUTION
CUTlULSTIVt PROBflBILlTY IP) EXCEEOBNCt PROBflBILITlr II-PI
.08
mum Eon 3
FITTED OlSTRieuTION
CUnitUBTIVE PHOBBBltlTT (PI EXCEEOBNCE PROBABILITY (I-PI
n.so
IS. 00
£12. so
M
710.00
= 1.50
S.OO
2.50
0
EOR 3 08SERVRTIONS
FITTED DISTRIBUTION
80.00
70.00
IEO.OO
gso.oo
30.00
20.00
10.00
EBB 3 OBSERVflTIONS
FITTED DISTRIBUTION
CunULRTIVE PROBRSaiTT IPI EXCEEORNCE fROSflSILITT II-PI
0 «' J^ 4 4 !' {' '["WffQUfV* • 1144 .141
OMJLRTIVE PROBRBILITY IPI EXCEEORNCE PROBRBILITY II-P)
60.00
SO. 00
-40.00
r. o.oo
20.00
'0.00
EOR 3 OBSEftVRTIONS
FITTED DISTRIBUTION
IS .00
_I2.00
o
fc
§9.00
6.00
3.00
EOfl 3 OBSERVRTIONS
FITTED DISTRIBUTION
CunULRTIVE PROBABILITY IPI CXCCCORNCE PROBHBILITY II-PI
CUrtULfltlVE PR08BBILITT IPI EXCCEOSNCE PROBflBILITr II-PI
Figure G-1C SOIL ASSESSMENT -- INDICATOR CHEMICALS
CUMULATIVE FREQUENCY DISTRIBUTIONS FOR LCICs IN EDA SAMPLING AREA 3
-------
12.SO
IID.OO
7.50
§ s.oo
- 2.50
|t|||||| EDfi 4 OBSERVRTIONS
FITTED DISTRIBUTION
pROBSBtLiTr IPI EXCEEDRNCE PROBRBILITY II-PI
CUnULRTIVE PWMRBILITY IPI EXCEEORNCE PROBaflRITY
-------
3.50
40.00
_35.00
-SO.OO
$25.00
§20.00
s.oo
0
COB S OBSERVATIONS
FITTED DISTRIBUTION +
OWULBTIVE PROBBBILITY IP) EXCEEOHNCE PROBABILITY (1-PI
• t i
PROBB81LITY IPI EXCEEOBNCE PROBABILITY (1-PI
1 75. 00
{.ISO. 00
SlJS.OO
i
. oo
§ 75.00
• SO.OO
ri
eg
- 25.00
0
EOS 5 OBSERVSTIONS
FITTED DISTRIBUTION
.18
CUHUIBTIVE PROBBBILtTf IPI EXCEEOPNCE PR080BILITT (1-P)
"
.08
.03
EDR 5 OBSERVATIONS
FITTED DISTRIBUTION
• .•.! 4 4 ri^V "iC'-'iC-W •••»• I* I I I 4 4 .141 41
CUHULATIVE PROBABILITY (PI EXCEEDANCE PROBABILITY 11-P]
90.00
80.00
-70.00
fc
~60.00
a50-00
^40.00
SO.OO
20.00
10.00
0
EOB 5 OBSERVBUONS
DISTRIBUTION
10.00
CunUUUIVE PROBBBILITY (PI EXCEEOBNCE PROBSBILITY (1-PI
CUnULBTIVE PROBflBlLITY IPI EXCEEDBNCE PR08HBILITY (1-PI
700.00
600.00
ISDO.OO
B 400.00
a
Ssoo.oo
200.00
100.00
0
IHHIII EOB S OBSERVRTIONS +
FITTED DISTRIBUTION
a ft
it
90.00
ZS.OO
-20.00
JEIS.OO
10.00
5.00
EDO S OBSERVRTIONS
FITTED DISTRIBUTION
CUnULBTIVE PROBRBILITY (PI EXCEEOBNCE PF06BB1LITY II-PI
CUHULBTIVt PROBBBILITY IPI EXCEEORNCE PROBABILITY 11-P)
Figure G-le SOIL ASSESSMENT -- INDICATOR CHEMICALS
CUMULATIVE FREQUENCY DISTRIBUTIONS FOR LCICs IN EDA SAMPLING AREA 5
-------
1.75
-I.SO
I
rji.zs
u
|l.OO
I-75
O
4.50
.K
0
EDR 6 OBSERVRTIONS
FITTED DISTRIBUTION
4.00
• A.I 4Jlt ll» «•••««• Ml • I 4 4 .!« 41
CimULBTIVE PROeflBILlTY IP) EXCEEOPNCE PROBB8ILITY (1-PI
CUTIULBTIVE PR08RBIUTT (PI EXCEEDRNCE PR06B8ILITY (I-PI
3.00
tz.so
i2-00
gl.50
SI.OQ
-------
1.00
o
g
-60
.40
.70
EON 1 OBSERVATIONS
FITTED DISTRIBUTION
CunULBIIVE PROBflBILITY IP) EXCEEOflNCE PROBflBILITY Il-Pl
• I I .1 Ji .1 M «
CUnULBTIVE PROBflBILITY IPI EXCEEOflNCE PROBABILITY II-P)
4.00
^.so
£3.00
SJ? .50
o
g
gj.oo
ul-SO
n 1.00
CM
" .50
0
EDB 1 08SERVSTIONS
FITTED DISTRIBUTION
.20
.18
j.08
s
".05
.03
COR 1 065ERVBTIONS
FITTED DISTRIBUTION
CUnULPTIVf PROBR6ILITT IPI EXCEEOBNCE PROBflBILIU Il-Pl
OJnULRIIVC PROBBBILITT IPI EXCEEOflNCE PROBflBILITif CI-P)
Z.SO
1-SO
-1-20
A
.60
.30
COR ^ OBSERVBTIONS
FITTED DISTRIBUTION
CUnULRTIVE PR08BBIL1TT IP) EXCEEOPNCE PROBflBILlTT II-P)
CUnULBTIVE PROBflBILlTT IPJ EXCEEOflNCE PR08BBILITT II-P)
CUnULRTIVE PR08BBILITT IP) EXCEEOflNCE PR08HBILITY (I-PI
CUHULflTIVE PROeflBILlTY IP) EXCEEOflNCE PR06BBILITY Il-Pl
Figure G-lg SOIL ASSESSMENT -- INDICATOR CHEMICALS
CUMULATIVE FREQUENCY DISTRIBUTIONS FOR LCICs IN EDA SAMPLING AREA 7
-------
1.20
1.30
.60
- .30
C221 OBSERVBTIONS
FIITEO DISTRIBUTION '
• • i ,\ f ' * * '—I
3.50
-3.00
f™
*4
|z.oo
*i so
U 1 «3U
oc
••71.00
rw
.50
cunuimivE rnoBBBitnr IPI ExCEtOflNCE PROBABILITY II-PI
C221 OBSMVRTIONS
FITTED DISTRIBUTION
CUnULBTIVE rROMBUITY IPI EXCKOIWt PROWBUItY d-PI
C221 OBSERVRTIONS
flTTEO DISTRIBUTION
it i i i .1
CUMW.PT1VE PROBBBILITT IPI EXCEEOONCE PROBB8KITT II-PI
C221 OBSCRVRTIONS
FITTED DISTRIBUTION
CUnULBTIVE PROBflBlLITY IPI EXCEEDflNCE PROBBBtLITY (I-PI
1.20
C221 08SERVBTIONS
FITTED DISTRIBUTION
1.00
'& i i i i i i . . ,
pRoeneiLiir IPI CXCEEOPNCE PWWWILITT i I-PI .
C221 OBSERVATIONS
FITTED DISTRIBUTION
CunuiflTIVE PROBB8JLITY IPI EXCEEOflNCE PROBB8H.ITT II-P)
1.20
C221 OBSERVBTIONS
FITTED DISTRIBUTION
CUnUlBTIVC PROBABILITY IPI EXCEEOflNCE PROBfl81llTT II-PI
BO. 00
-70.00
-60.00
T50.00
§40.00
30.00
20.00
10-00
ft
mum Q221 oesERVBTlONS
I • f • u...y.|H-w..M...J|r .JjJn^T ^ f t | J J .
CunuLnTIVE PR08H8ILITY IPI EXCEEOflNCE PROBB8ILITY II-PI
Figure G-1h SOIL ASSESSMENT - INDICATOR CHEMICALS
CUMULATIVE FREQUENCY DISTRIBUTIONS FOR LCICs IN THE CENSUS
TRACT 221 COMPARISON AREA
-------
I.SO
C225 OBSERVATIONS
FITTED DISTRIBUTION
.90
K.OO
C225 OBSCTVRTIONS
FITTEO DISTRIBUTION
cunuLfliivt nweneiuTT IPI EXCCEDMCC PROBRBILITY II-PI
cunuuiTtvE MUBMiLiTY IPI EXCCEDRNCE PROBABILITY II-PI
C22S OBSEKVATIONS
FITTED DISTRIBUTION
cunuumvE pROBflaain IPI EXCCCORNCE PHOBABILITT II-PI
CimULRTIVE PROBRBILITY IPI EXCEEORNCE PROBftBILITY ll-p]
C225 OeSERVBTIONS
FITTED DISTRIBUTION
6.001—
5.00
"«.00
D3.00
2.00
1.00
C225 OBSERVBTIONS
FITTED DISTRIBUTION
CUtULATIVE PROBABILin IPI CXCtEORNCE PMBR81LITY II-PI
CUnULBTIVE PROBBBILIIT IPI EXCEEOBNCE PROBflBILITY ll-P)
C225 OBSERVATIONS
FITTED DISTRIBUTION
60.00
SO. 00
-40.00
?.30.00
20.00
10.00
C225 OBSERVOT10NS
FITTED DISTRIBUTION
CUnULRTIVt PROBB8ILITT IPI EXCEEOONCE PROBftBILITT II-PI
• 41.1 4 4 I 1 f U WWHT'lT'llI •• Ml 1144.1 41 41
CUnULflWE PROBftBILITT IPI EXCEEDBNCE PR08S61LITT II-PI
Figure G-11 SOIL ASSESSMENT -- INDICATOR CHEMICALS
CUMULATIVE FREQUENCY DISTRIBUTIONS FOR LCICs IN THE CENSUS
TRACT 225 COMPARISON AREA
-------
I.SO
I"
g
9 .60
.30
OBSERVflTIONS
FITTED DISTRIBUTION
0 41 «.! 44lt IN •»•••»• •• 1144 .\m . 41
OW.BTIVE PROBBBILITT IPI EXCEEOflNCE PROBRBILIH II-PI
1.00
.80
.60
- .20
C&T OBSERVRTIONS
FITTED DISTRIBUTION
-HI i 1.1 .i.j j . i .
• «.! 4411 It* • »•••»•
• I I 4 4 .1M M
CUU*t«rt PttOBABILITT (PI EXCEEOBNCE PROBBBILITT II-PI
C4T OBSERVfltlOHS
FITTED OISTRIBUTION
CunULBllVE PROBBBILIIr IPI EXCEEOflNCE PROBBBILITT II-PI
CUnULflTIVE PROBBBILITT IPI EXCEEDBNCE PROBRBILITY (I-PI
.IB
.15
C&T OBSERVflTIONS^
FITTED OISIRIBUTIO
CUnuLftMVE PROBR6ILITT IPI EXCEEOflNCE PROBBBILITT II-PI
6fn
•UU
5.00
9-
-4.00
fi
-3 00
§ '
2.00
1.00
0
C&T OBSERVflTIONS
FiTirn nicT0miTinN +
_+_____ —
.05
.04
-.04
ffl
s
-.03
£
T.03
§.02
.02
.01
.01
o
C&T OBSETWUIONS
firtrn nixra in 11,1x1 *•
I < .1
CUnULflTIVE PROeneiLlTT IPI EXCCCOfXC PROeflBKMY (I-P)
CUnULBTIVE PROBBBILITT IPI EXCEEOflNCE PROBBBILITT II-PI
Figure G-1J SOIL ASSESSMENT - INDICATOR CHEMICALS
CUMULATIVE FREQUENCY DISTRIBUTIONS FOR LCICs IN THE
CHEEKTOWAGA AND TONAWANDA COMPARISON AREA
-------
i APPENDIX H
\ Detection Limits
i
i
! Soil Assessment-Indicator Chemicals
WDC 63394.T1 (S«l
-------
Appendix H
DETECTION LIMITS
SOIL ASSESSMENT—INDICATOR CHEMICALS
Dr. Stoline, in his preliminary written comments inquired as
to whether or not the detection limits had been improved
from the pilot study. An estimate of the operational detec-
tion limit was provided in Volume III of this series. This
appendix discusses the variety of definitions of detection
limits in the current literature and provides other estima-
tors of the detection limit. The following sections provide
background information, discuss physical characteristics of
the laboratory analysis, and present a model for estimating
the method.
BACKGROUND
The 1980 Love Canal study found that more than 95 percent of
the samples analyzed had concentration levels that were too
low to quantify. This was an unexpected result of the
study, and no systematic effort had been made during the
laboratory analysis to determine the level of detection for
the analytical methods. A retrospective analysis indicated
that the level of detection was no greater than 1.0 ppm. A
review by Congressional Office of Technology Assessment
(OTA) of the study concluded that the number of samples with
non-detect concentrations made the statistical comparison
difficult. The OTA reported that, based on the available
information, it was not possible to conclude whether or not
unsafe levels of contamination existed in the EDA (OTA,
1983) .
The current Love Canal indicator chemical soil assessment
responded to these concerns on three levels. First, the
analytical techniques were improved to allow detection of
concentrations that are two to three orders of magnitude
smaller than those believed to have been detected in the
1980 study. Second, statistical techniques were used to
perform the sampling area to comparison area comparisons,
which were robust to the presence of nonquantifiable values.
Third, specific analyses were conducted to enable an esti-
mate of the detection limit. These analytical methods and
statistical techniques are discussed in the sample analysis
QAPP (CH2M HILL, 1987) and Volume II, respectively. This
appendix will discuss the methods used to estimate the detec-
tion limits of the analytical method.
The term "detection limit" is not as well defined as one
might think. The term has been used in the literature with
a number of different definitions and associated estimators
(Currie, 1988). Currie lists four types of detection limits
H-l
-------
in current use: lower limit of detection, instrumental
detection limit, method detection limit, and limit of detec-
tion. Each of these types has one or more estimators that
can be used to quantify that particular detection limit.
The detection limit of interest for this study is the method
detection limit, which is defined here as that concentration
of an analyte in soil that can be differentiated from a soil
matrix containing zero concentration with 95 percent confi-
dence and 95 percent power. This corresponds to the statis-
tical definition of the minimum detectable difference
between a sample with a concentration of an analyte at the
method detection limit and a sample with a zero concentra-
tion of an analyte.
Three different estimators of the method detection limit
have been proposed. One method quantifies the uncertainty
(variance) of instrument response when measuring blank soil
samples (Currie, 1968) . This variance is then assumed to be
similar to the variance of the instrument response when a
low level of an analyte is present. Standard statistical
hypothesis testing techniques are then used to estimate the
minimum detectable difference from the blank as the method
detection limit.
A second technique, similar to the first, uses low-level
spiked (known) concentrations, rather than blanks, to esti-
mate the variability of the instrument response (Glaser, et
al., 1981). The variance of blanks is then assumed to be
similar to that of low-level spikes, and standard hypothesis
testing techniques are again used to estimate the minimum
detectable difference.
The third method uses the linear calibration curve to
estimate the detection limit (Sharaf, et al., 1986). This
method uses the instrument sensitivity to pure solutions of
the target compounds as well as the uncertainty in response
at several concentrations to estimate the detection limit.
An estimator for the method detection limit that is similar
to that based on the calibration curve but based on estimat-
ed concentrations is the most appropriate for the Love Canal
indicator chemical soil assessment. This method has three
advantages:
o It includes the variance of instrument response at
several concentrations; the methods of Currie and
Glaser only use a single concentration.
o It includes the variance of the calibration curve
that the other methods do not.
H-2
-------
o It is based on estimated concentrations of measure-
ments on soil samples and, therefore, includes
some of the variability resulting from chemical
interference.
There are, however, limitations to this estimator that are
not resolved in the current literature. An understanding of
the analytical process is necessary to discuss the advan-
tages and limitations of this estimator.
PHYSICAL CHARACTERISTICS OF THE LABORATORY ANALYSIS
The analytical method used in the Love Canal soil assessment
for indicator chemicals is discussed in detail in the sample
analysis QAPP (CH2M HILL, 1987). The method can be sum-
marized briefly by discussing the theory of operation of the
GC/MS, the method of calibration, and the analysis of
unknown samples.
A GC/MS analysis begins by the injection of a small quantity
of a solution containing the target analytes (chemicals of
interest, here LCICs) into the GC part of the instrument.
The solution is vaporized into a gas and fed into a long
capillary column. The interaction of the molecules of the
vaporized solution with the interior of the capillary tube
causes different molecular species to pass through the col-
umn at different speeds. The time that each species takes
to pass through the column (elute) is a characteristic that
can be used to help identify the molecular species.
The time required for a molecular species to elute from the
chromatographic column is not, however, a unique identifier
of an analyte. Identification is accomplished by feeding
the gas eluting from the GC into the MS.
In the MS, the molecular species are bombarded with an
electron beam that breaks the molecules into a number of
charged ions. Each ion has a mass and is usually singly
charged. The ions are screened by an "electronic window"
that only allows ions of a given mass-charge ratio to pass
at a given time. The ions that pass this window are then
counted by a detector.
Since the "electronic window" can be switched very rapidly
to allow passage of ions of different masses, an estimate
can be made of the numbers of ions in several mass cat-
egories for each molecular species. The count of ions must
take place during the time that the particular molecular
species is eluting from the GC. Therefore, the more ion
masses that are counted, the less time there is for counting
ions at each mass. This decreases the resolution of the
instrument.
H-3
-------
One of the goals of the soil assessment for indicator
chemicals was to decrease the detection limit. One of the
methods used to accomplish this was to increase the instru-
ment resolution by decreasing the number of ion masses
examined for each LCIC. This mode of operation for a GC/MS
is known as Selected Ion Monitoring (SIM). The SIM method
used for LCIC detection used three predefined ion masses for
each LCIC. The total counts of ions at these three masses
then formed the basis for positive identification and quanti-
tation of each LCIC.
Identification of an LCIC is accomplished in the MS after
three criteria are met. The candidate molecules for an LCIC
must elute from the GC within an appropriate time frame.
The primary and secondary (first and second most common) ion
counts must be present in a ratio that is close to the theo-
retical ratio. Finally, all three ions must be present.
Once an LCIC has been identified by these criteria, the
concentration is estimated using the quantity of primary
ions present (as estimated by the extrapolated total ion
count), the total count of the internal standard primary
ion, the calibration information from the continuing cali-
bration done at the beginning of the shift, and information
from the initial calibration. The general procedure
involves comparing the instrument response to the unknown
quantity of LCIC with the instrument response to the known
LCIC concentrations used during the calibration process.
The GC/MS is calibrated as part of setting the instrument
for analyzing LCICs. The calibration is checked regularly,
and the instrument is recalibrated whenever it fails the
calibration criteria. The initial calibration consists of
analyzing five solutions containing three classes of chemi-
cals: the LCICs, the Internal Standards (IS), and the surro-
gate standards. The LCICs are at five different concentration
levels that span the range of interest and to which the
instrument response is thought to be linear. The data from
these five analyses are used to set up the calibration
curve.
A constant quantity of IS chemicals is added to solutions
before injection into the GC and used for quantitating the
LCICs in unknown solutions. The internal standards are cho-
sen to elute near groups of LCICs and are used to compensate
for small variations in injected volume as well as run-to-
run variation in instrument responsiveness.
The surrogate standard chemicals are added in known
quantities to each soil sample before extraction. The
amount recovered in the form of estimated concentrations
from the analysis is used to provide a rough measure of the
efficiency of the extraction process.
H-4
-------
The IS and surrogate standards are always analyzed at a
single, specific concentration.
The relation between the LCICs and the IS is established
during the initial calibration. The relation is updated
with a continuing calibration analysis of the lowest LCIC
calibration level at the beginning of each shift.
The analysis of unknown samples starts with the addition of
the surrogate chemical solution to the soil sample. This
soil then has solvents added to extract the LCICs and surro-
gates. After several steps in the extraction process, the
extract is concentrated and stored. When the extract is to
be analyzed, the ISs are added to the solution, and a por-
tion is removed for injection into the GC/MS.
This process has several sources of variability that
contribute to the uncertainty associated with concentration
estimates:
o The efficiency of the extraction process in
removing all the LCICs from the soil is unknown
and cannot be estimated. The extraction effi-
ciency of the surrogates is similar to, but not
sufficiently highly correlated with, the extrac-
tion efficiency of LCICs to allow a recovery cor-
rection to concentration estimates.
o The extraction process also obtains other
chemicals from the soil matrix. Many of these
chemicals will elute from the GC at similar times
as the LCICs and even have an ion in common with
an LCIC in the MS. This produces interferences
and chemical noise in the analysis. If the matrix
chemicals that interfere are present in sufficient
quantity, they may prevent identification of an
LCIC by throwing off the ratio of LCIC ions or by
obscuring the presence of an ion. Even low levels
of matrix chemicals can contribute additional uncer-
tainty in the quantitation of the LCICs.
o Finally, the GC/MS instrument has some inherent
variability in the concentration estimates. This
is estimated from the difference of the actual
instrument response from that predicted by the
calibration curve.
The bulk of the uncertainty associated with estimating LCIC
concentration is associated with the matrix. The matrix
effects are the major determinates of the detection limits.
The major matrix interferences that prevent the identifica-
tion of the LCICs are not included in this model. The
occurrence of such interferences is uncorrelated with the
H-5
-------
concentration of LCICs. The effect of the process, whereby
matrix interferences are sufficient to prevent identifica-
tion, is to make the detection limit model conditional on
the absence of interferences of this type or magnitude.
METHOD DETECTION LIMIT MODEL
An estimator for the method detection limit can be developed
based on the calibration curve method, but using recoveries
of known amounts of analytes spiked into soil rather than
pure analytes in solvent (Hubaux and Vos, 1967) . The method
consists of a regression of the estimated concentration on
the actual spiked concentration. This model has the follow-
ing limitations:
o The model is conditional on the absence of
interferences sufficient to prevent identifica-
tion. A separate statistical model is required to
estimate the probability of a field sample having
nonquantifiable concentrations due to matrix
interferences.
o The model assumes that the variability of the
estimates is independent of the magnitude of the
estimates.
o The model assumes that the estimated concentra-
tions are a linear function of the true
concentrations.
o The model assumes that a concentration can be
estimated validly for an arbitrarily low spiked
concentration. Any interfering compounds present
will be sufficient to prevent the proper identi-
fication of sufficiently small spiked
concentrations.
o The model is sensitive to the concentration levels
of the spiking solutions. This is because of the
uncertainty in regression relation resulting in
prediction limits that become wider the further
removed the detection limit estimate is from the
mean spiked concentration.
The parameters of the linear regression reflect the various
processes involved in the chemical analysis. The uncertain-
ty associated with an estimate at a given concentration is
due to the matrix effects and instrumental variance. The
slope of the regression reflects the sensitivity of the
instrument response to the LCIC concentration. The inter-
cept of the regression reflects the presence of background
H-6
-------
concentrations of LCICs in the soil or in the laboratory
solvents.
Figure H-l illustrates the various aspects of the model and
shows several concentrations estimated for spiked concen-
trations. A regression relation (solid line) is estimated
from these points, and the 95th percentile of predicted
estimated concentrations is also estimated (dashed line). A
pair of line segments that form a right angle graphically
illustrates the detection limit process.
The method detection limit is estimated from the regression
by examining the prediction limits of the relation. These
limits are the 5th and 95th percentile bounds for predicting
what the estimated concentration would be for a given known
concentration in the sample. The upper prediction limit for
a blank corresponds to the highest concentration that would
be estimated 95 percent of the time when analyzing a blank.
(Note that this is not quite appropriate for a GC/MS, as
blank samples would be expected to fail identification
criteria.)
The true concentration whose expected estimated concentra-
tion corresponds to the 95th percentile of estimated concen-
trations when a sample contains no LCICs (a blank) is then
the criterion level concentration. This concentration is
greater than that estimated for 95 percent of all blanks.
The true concentration, whose 5th percentile bound on the
predicted concentration estimate corresponds to the 95th per-
centile bound on predicted concentrations for blanks, is
then the method detection limit. This is the concentration
that will be differentiated from a blank 95 percent of the
time.
This estimate is shown in Figures H-2 (a through h), H-3
(a through h), and H-4 (a through h) for LCICs during the
soil assessment study. These figures show the Hubaux-Vos
estimates of detection limits for six laboratories and eight
LCICs for three different data sets. Each graph shows the
regression line between spiked and estimated concentrations,
the 5th and 95th percentile prediction bounds, and the
detection limits. The detection limits are shown by tracing
a line from the 95th percentile bound for blanks to the 5th
percentile bound for the method detection limit. The line
then drops from the 5th percentile bound down to the actual
concentration axis.
Occasionally, the line illustrating the detection limit
estimation process will not exactly meet the lower predic-
tion interval curve. This is caused by a round-off error in
the computer, and it does not interfere with the
H-7
-------
illustration of the range and magnitude of the Hubaux and
Vos detection limits seen in the study.
Figures H-2 (a through h) show the detection limits
estimated for each laboratory for each LCIC. These esti-
mates are based on the PE data obtained when the labora-
tories were qualifying for participation in the study. The
samples analyzed were spiked at several low levels by
EMSL-LV and analyzed in duplicate by all the laboratories.
Figures H-3 (a through h) are a similar series of plots.
This is the same data set with the spiked concentrations
greater than 1.0 ppb removed. This series of plots indi-
cates the sensitivity of the Hubaux and Vos estimators to
the range of concentrations used. Note that the detection
limits estimated with this data set are lower than those
using the full data set.
Figures H-4 (a through h) are the same series of plots using
data from the EMSL-LV BQC samples. These samples were
analyzed during the course of the study at a rate of one BQC
sample for every 20 field samples. The BQC samples covered
a wider range of spikes than the PE sample spikes.
After reviewing these figures, one can note that the
estimates of the method detection limit are quite variable
from laboratory to laboratory, from LCIC to LCIC, and from
data set to data set. What these estimates provide is an
upper bound on the concentration levels likely to be present
and still to be found nonquantifiable (non-detect). The
estimates include the variability of matrix effects and
recovery efficiencies, and thus, they represent a "worst
case" of what can be detected. An estimate of a detection
limit from this method does not mean that lower concentra-
tions cannot be detected; rather, this is an indication of
the concentration that probably would be detected if it were
present.
WDR357/031
H-8
-------
REFERENCES
CH2M HILL. 1987. Love Canal Habitability Study. Soil
Sample Laboratory Analysis Quality Assurance Project
Plan.
Currie, Lloyd. 1988. Detection in Analytical Chemistry
Importance, Theory and Practice. American Chemical
Society. Washington D.C. Pp 1-64
Currie, Lloyd. 1968. Limits for Qualitative Detection and
Quantitative Determination. Application to Radio-
chemistry. Analytical Chemistry. Volume 40, No. 3.
Pp 586-593.
Hubaux, Andre and Gilbert Vos. 1970. Decision and Detection
Limits for Linear Calibration Curves. Analytical Chem-
istry. Volume 42, No. 8. Pp 849-855.
Glaser, John; Denis Foerst, Gerald McKee, Stephan Quave,
William Budde. 1981. Trace Analyses for Wastewaters.
Environmental Science and Technology. Volume 15,
No. 12. Pp 1426-1435.
OTA. 1983. Habitability of the Love Canal Area, An Analy-
sis of the Technical Basis for the Decision on the
Habitability of the Emergency Declaration Area.
Sharaf, Muhammod, Deborah Illman, Bruce Kowalski.
Chemometrics. John Wiley and Sons. New York.
P 122ff.
WDR357/031
H-9
-------
2.0 -
.a
Q.
S 1.5
c
o
TS
o
u
O 1.0
T3
0)
S
J3
O
95th percentile of
predicted observations
when true concentration
Is zero.
0.5 -,
0.0 -
Hubaux-Vos
estimated
detection limit
2.0
True Concentration (ppb)
Figure H-1. KEY TO FIGURES H-2 THROUGH H-4
-------
Hubaux-Vos Method Detection Limits by Lab
2.O
-1.5
1 .O-
0.5
0.0 •<
0.0
LABlD-l
0.5 I ,O 1.5
True Concentration (ppb)
2.0
2.0
1 0 5
o.o •
o.o
LAB I 0-6
0.5 1.0 1.5
Tru* Concentration (ppb)
2.0
1
2.0-
1.5-
1 .O •
0.5
0.0
0.0
LAB I 0-2
O.S 1.0 1.5
Tru« Concent rot ion (ppb)
2.0
1
2.0 i
1 .5
1 .0
O.5
O.O
0.0
LAB I 0-7
0.5
Tr>
1 .0
1 .5
(ppb)
2.0
LABIO-3
LABIO-S
2 .O
;1.S-
1.0-
0.5 -
0.0 -
O
.0
1
2.0-|
1 .0
0.5
0.0 •
O.S 1.O 15 2.0 0.0 0.5 1.0
Figure H-2a. ESTIMATE OF DETECTION LIMITS USING PE DATA. (1, 2-DICHLOROBENZENE)
1 .5
2.0
-------
Hubaux-Vos Method Detection Limits by Lab
LAB I 0-6
2.O
'1.5:
1.0
0.5 •
o.o
o.o
O.S I .O 1.5
True Conc*ntrotion (ppb)
2.0
2.0
'1.5.
1 .0
O.S
o.o
O.O
0.5 1.0 1.5.
Tru* Concentration (ppb)
2.0
2.0
.s '
1 .O-
0.5 -
O.O
O.O
LABID-2
O.5 1.O 1.5
Tru* Concentration (ppb)
2.0
"5 0.5
00
0.0
LABIO-7
0.5 1 .O 1.5
Tru* Concentration (ppb)
2 .0
O.O
LABID-J
O.5
1 .O
1 .5
2.0
2 .0 -I
= 15
1.0
0.5
0.0
0.0
LABIO-B
1.5
2.0
Figure H-2b. ESTIMATE OF DETECTION LIMITS USING PE DATA. (1, 2, 4-TRICHLOROBENZENE)
-------
Hubaux-Vos Method Detection Limits by Lab
2.0 -
1.5-
.5
B
1 1.0-
S
"8 0.5 •
o.o
0.0
LAB I 0-1
0.5 1.0
Tru» Cone»nIrotion (ppb)
2.0
2.0
0.0
0.0
LAB I 0-6
0.5 1.0 1.5
Tru» Concontrotion (ppb)
2.0
LAB 10-2
LAB 10-7
2.O
'1.5
1.0
0.5
0.0
0.0
O.5 1.O 1.5
Truft Conc*ntrotion (ppb)
2.0
2.0
1.5
1.0-
OS
0.0
0.0
05 1.0 I .5
Tru* Conccnlrolion (ppb)
2.0
2 0-j
1
""
1 .O-
0.0
LA8ID-3
0.5
I .0
1 .5
2.0
2.0
: 1 .5
1.0
0.5
0.0 •
O.O
LAB I 0-8
0.5
1 .0
1 .5
2.0
Figure H-2C. ESTIMATE OF DETECTION LIMITS USING PE DATA. (1,2.3.4-TETRACHLOROBENZENE)
-------
Hubaux-Vos Method Detection Limits by Lab
LAB I 0-1
LAB 10-6
2.0
'1.5-
1.0
o.s
0.0
0.0
0.5 1.0 1.5
True Concentration (ppb)
2.O
2.O •
't.S
1.0
0.5
O.O
O.O
0.5 1.0 1.5
Tru* Concentration (ppb)
2.0
LAB I 0-2
LABID-7
2.O-
;1.5
1.0
0.5
O.O
O.O
O.5 1.0 1.5
True Concentration (ppb)'
2.O
2.0
: 1.5
I .O
0.5
O.O
0.0
0.5 1.O 1.5
True Concentration (ppb)
2.0
LAB I 0-3
O.S
1 .0
1 .5
2.0
2.0 H
&
10
0.0
0.0
LAB I 0-8
1 . 5
2.0
Figure H-2d. ESTIMATE OF DETECTION LIMITS USING PE DATA. (2-CHLORONAPHTHALENE)
-------
Hubaux-Vos Method Detection Limits by Lab
LABID-6
[3.0
'2.5-1
2.0
1 .5
1 .O •
05
O.O
0.0
0.5 1.0 1.5
Tru* Cene*nlrot ion (pf>t>>
2.O
.1
i.O •
2.5 •
2.0
J1.5
1.0
H o.s
JK 0.0
O.O
0.5 1.0 I .5
Tru* Conc*ntrotion (ppb)
2.O
LABlD-2
LAB I 0-7
: j.o-i
'2.5
2.O-
1.5-
1 O
O.S-
0.0 ^
00
O.S 1.0 1.5
Tru* Cone»ntrolion (ppb)
2.0
" J.O
= 2.5
2.0
1.5-
1 .0
0.3J
0.0 -I
O.O
0.5 1.O 1.5
Tru* Conc*ntrolion (ppb)
2.0
LABlO-3
LABIO-B
[3.O -
!2.5-
2.O
1.5-
10
0.5
O.O
O.S
2.0-
15-
' 0
0.5 -
O.O
1.0 1.5 2.0 "~* O. O 0.5 1.0
Figure H-2e. EXTIMATE OF DETECTION LIMITS USING PE DATA. (ALPHA-BHC)
1 .5
2.0
-------
Hubaux-Vos Method Detection Limits by Lab
LAB I 0-1
LAB I 0-6
True Concentration (ppt>>
2.5
i.O -
'2.5
2.0 •
1.5
1.0
0.5
0.0
0
.0
0.5 1.0 1.5
Tru* Concentration (ppb)
2.0
2. 5
LAB 10-2
LAB 10-7
2.0
1.5
1.0
0.5
0.0
O.9
l.O 1.5 2.0
True Concentration (ppb)
2.5
0.5
1.0 1.9
True Concentration (ppb)
2.O
2.5
LABID-3
LABID-B
.i.O-j
•*.»\
2.O
1.5
1 .O
O.S
O.O
O.S
1 .0
1 .5
2.5
.3.0 -
= 2.5
2.O
1 .5
1O
OS
O.O
O.O
0. 5
1 .0
1 .5
2.0
2.5
Figure H-2f. ESTIMATE OF DETECTION LIMITS USING PE DATA. (DELTA-BHC)
-------
Hubaux-Vos Method Detection Limits by Lab
LAB I 0=1
LABID-6
.3.0-
'25
2.0
1.3
1.0
0.5
0.0
O
0 0.5 1.0 1.5 2.0 2.5 3.0
True Concentration (ppb)
3.5 4.O
<
0.5
o.o
0.0 0.3 1.0 1.5 2.0 2.5 3.0
True Concentration (ppb)
3.5 4.0
LA8ID-2
LAB ID-7
.3.OH
:2.5
2.0-
1 .5 -
1.O -
O.5 -
0.0
O
0 0.5
1.0 I.5 2.0 2.» 3.0
Tru* Concentration (ppb)
3.5 4.0
.3.0-
:2.5 •
2.0 •
1 •* •
1.0-
0.5 -
O.O
0.0 0.5
1.O 15 2.0 25 3.O
Tru* Concentration (ppb)
3.5 4.O
LAB 10-3
LABIO-B
-JO
•2 5
2.0 -
1 .5
1 .O
0.5 •
O O
O O 0.5 1.O 1.5 2.O 2.5 JO 3.5 4.O
2 . O •
10-
O. 5 -
0.0 •
O.O O.5 1.O 1.5 2.O 2.5 3.O 3.5 4.0
Figure H-2g. ESTIMATE OF DETECTION LIMITS USING PE DATA. (BETA-BHC)
-------
Hubaux-Vos Method Detection Limits by Lab
LAB ID-I
LAB IO-6
.3.0-1
'25
2.0
1 •»•
I.O
0.5
• o.o
o.o o.s i.o
1.5 2.0 2.5
Tru» Conc«ntrotion (ppb)
3.O 3.5 4.O
.3.0-1
'2.5
2.0
1.5-
1 .0
0.5
0.0
0.0 0.5 I.O
1.5 2.0 2.5
Tru* Concentration (ppb)
3.0 3.5 4.0
LABIO-2
LABID-7
.3.0
'2.5
2.0
1 .5
1 .O
0.5
0.0
O.O O.3 1.0
I .5 2.O 2.5 3.O
uft Concentrotion (ppb)
3.5 4.0
.3.0 -\
2 *
2.0
> .5
1 .0 -I
0.5
O.O •
O.O O.5 I.O
1.5 2.O 2.3
Tru* Concftntrot& on (ppb)
3.0 3.5 4.0
LABIO-3
LABID-8
.30
!2.3-|
2 .O
1.3-
1.0-
O.S
0.0-
.3.0
!2.5
2.0
1 .5
1 .0
0.5
0.0
0.0 0.5 I.O 1.5 2.O 2.5 3.O 3.5 4.O
O.O 0.5 I.O 1.5 2.O 2.3 3.0 3.5 4.O
Figure H-2h. ESTIMATE OF DETECTION LIMITS USING PE DATA. (GAMMA-BHC)
-------
Hubaux-Vos Method Detection Limits by Lab
LAB ID-I
LAB I 0-6
2.0-j
!-
r
1 O.5 •
0.0
0.0
O.5 1.0 1 .5
Tru* Concentrotion (ppb)
2.0
2-0-1
1 .5
10
0.5
O.O •
0.0
0.5 1.0 1.5
Tru* Concentrotion (ppb)
2.0
LABID-2
LAB 10-7
Tru* Conctntrotion (ppb)
2.0
.i
s
2.0
" 1 .5 •
1.0-
0.5
O.O
0.0
05 1.0 1.5
Tru* Concentration (ppb)
2.O
LABID-3
LAB 10-8
2.0
1.5
1.0 •
0.5 •
O.O -T--"
O 0
0.5
1 .5
2.0
2.O •
'l.S-
1 .0
0.5
O.O •
O.O
0.5
1.0
1 .5
Figure H-3a. ESTIMATE OF DETECTION LIMITS USING PE DATA SET WITH CONCENTRATIONS >1 PPB REMOVED.
(1. 2-DICHLOROBENZENE)
2.O
-------
Hubaux-Vos Method Detection Limits by Lab
2.0-
'1.5
1 .0
0.5
0.0
0.0
LAB ID-I
0.5 1.0 1.5
Tru» Concent rot ion (ppb)
2.0
I
2.0 -
'1.5-
1 .0
0.5
0.0
0.0
LAB 10-6
0.5 1.0 1.5
Tru» Concentration (ppb)
2.0
LAB 10-2
LAB ID-7
2.0-
1.5-
1.0-
05
O.O-
O.O
O.5 I O 1 .5
True Concentration (ppb)
2.0
2 0 H
: 1.5
to
0.5
O.O
0.0
0.5 1.O 1.9
Truo ConconIrotion (ppb)
2.0
2.0 •
= 1.5
1 .O
0.5
O.O
0.0
LABIO-3
0.5
1 .5
2.0
2-0-1
.1
S
1 .5
O.O •
O.O
LABIO-B
0 5
1.0
1 .5
Figure H-3b. ESTIMATE OF DETECTION LIMITS USING PE DATA SET WITH CONCENTRATIONS >1 PPB REMOVED.
(1,2.3-TRICHLOROBENZENE)
2.O
-------
Hubaux-Vos Method Detection Limits by Lab
2.O
1.5-
1.0 •
0.5
O.O
0.0
LABID-1
0.5 1.0 1.5
True Concentration (ppb)
2.0
2. 0
1.5-
1.0 •
0.5
0.0
O.O
LAB I 0*6
0.5 1.0 1.5
Tru* Concentration (ppb)
2.0
2.0
' 1 .5
1 O
0.5 •
0.0
0.0
LABID-2
0.5 1.0 1.5
True Concentration (ppb)
2.O
2.0
,J
s
1 . 5
0.5
00
0
LAB I 0-7
0.5
Tri
1.0 1.5
ntrotion (ppb)
2.0
2.0-
'1.5-
1 .O
0.5 •
00
0
LA8IO-3
O.S
1 .O
1 .5
2.0
2.0 i
1
I
S
g
Ǥ
O.S
0.0 •
O.O
LAB 10-8
0.5
1.0
1 .5
Figure H-3C. ESTIMATE OF DETECTION LIMITS USING PE DATA SET WITH CONCENTRATIONS >1 PPB REMOVED.
(1,2, 3, 4-TETRACHLOROBENZENE)
2.0
-------
Hubaux-Vos Method Detection Limits by Lab
LABlD-1
LAB I 0-6
20
'1.5
1.O •
0.5
O.O
0.0
0.5 1.0 1.5
True Concentration (ppb)
2.0
2.0
'l.S
1.0
0.5
0.0
0.0
0.5 1.0 1.5
True Concentrolion (ppb)
2.0
LABI 0-2
LABID-7
True Concentrotion (ppb)
2.0
2.0-1
True Concentration (ppb)
2.0
0.5
LAB 10-3
1 O
1 .5
2.0
2 .O
= 1.5.
1.0
0.5
O 0 -I
O.O
LABio-e
0.5
1 .0
I .5
Figure H-3d. ESTIMATE OF DETECTION LIMITS USING PE DATA SET WITH CONCENTRATIONS > 1 PPB REMOVED
(2-CHLORONAPHTHALENE)
2.0
-------
Hubaux-Vos Method Detection Limits by Lab
LABIO-l
LAB I 0-6
* 3.0
f»-»
"g 2.0
J1.5
1.0-
I 0.5
§ 0.0
** 0
0.3 1.0 1.5
True Concentration (ppb)
2.0
2'5'
g 2.0 -
J~~ 1.5-
i.o
H 0.5
Jo.o-
S
0.0
0.5 1.0 (.5
Tru* Concentration (ppb)
2.0
LABID-2
LAB 10-7
.3.0-1
!2.5
2.0-
> .S
1 .0
OS
o.o-i
o'o
0.9 1.0 1.5
True Concentration (ppb)
2.0
Tru* Concentration (ppb)
2.0
LABlD-3
LABIO-B
.30-1
!2.5-
2.0
1.5-
1.O
0.5
O.O
0.0
O.S 1.O 1.5
Tru* Concentration (ppb)
2.O
" 2.5
2.0-1
I .5
1.0
0.5 .
0.0
O.O
0.5 1.O 1.5
True Concentration (ppb)
2.0
Figure H-3c. ESTIMATE OF DETECTION LIMITS USING PE DATA SET WITH CONCENTRATIONS >1 PPB REMOVED. (ALPHA-BHC)
-------
Hubaux-Vos Method Detection Limits by Lab
LABID-1
LAB I 0-6
0.5
1.0 1.5
Tru» Concentration (ppb)
2.0
2.5
0.5
1.0 1.5
Tru* Conc»ntration (ppb)
2.0
2.5
LAB I 0-2
LAB 10-7
2 *
2.0
* *
» o
O. 5
OO
O.O
O.5
1 O I . S
True Concentration (ppb)
2.0
2.5
-a.3.0
f 2'51
'•§ 2. O •
J~* 15
1.0
"$ 0.5
BOO
0.0
0.5
1.0 1.9
Tru* Conc»ntrotion (ppb)
2.0
2.5
LABlD-3
LAB10-8
(ppb)
2.5
0. 5
1 .O
1 .5
2.0
True Concentration (ppb)
2.5
Figure H-3f. ESTIMATE OF DETECTION LIMITS USING PE DATA SET WITH CONCENTRATIONS >1 PPB REMOVED. (DELTA-BHC)
-------
Hubaux-Vos Method Detection Limits by Lab
LAB I 0-1
LABIO-6
.->
'•»•
2.0-
I.S
1.0
0.5-
0.0 O.S
1.0 1.5 2.O 2.5 3.0
True ConcentroIion (ppb)
35 4.0
.3.0
'2.5-
2.0
1 .5
1.0-
O.S
0.0
0.0 O.S 1.0 1.S 2.0 2.S 3.0
Tru* Concentration (ppb)
3.5 4.0
LABID-2
LAB 10-7
4.O
|2.0
J 1 .0
1[ 0.5
§ 0.0 •
o.o o.s
1.0 1.S 2.O 2.5 JO
Tru* Concontrotion (ppb)
3.5 4.0
LAB I 0-3
LABIO-B
•J
'2.5
2.0
1 .5
1.0-
O.S •
O.O •
O.O O.S 1.0 1.S 2.O 2.5 3.0
True Concentration (ppb>
3.5 4.0
Figure H-3g. ESTIMATE OF DETECTION LIMITS USING PE DATA SET WITH CONCENTRATIONS >1 PPB REMOVED. (BETA-BHC)
-------
Hubaux-Vos Method Detection Limits by Lab
LAB I 0-1
LABID-6
.3-0-j
'2.3
2.O •
1.5
1 O
0.5
o.o •
0
.0
0.5
I.O 1.5 2.0 2.5 3 O
Tru* Conc«nlrotion (ppb)
3.5 4.0
,i
3.0 -\
2. 5
20
1.5-
1.0
O. 5
0. O
0
0.5 1.0 1.5 2.0 2.5 3.0
Tru* Concentration (ppb)
3.5 4.0
LAB 10-2
LAB 10-7
3.5 4.O
.i
.30
"2.5
2.0-
1.5-
1 .O
0.5 •
00
0
0.5
1.0 1.5 2.0 2.5 3.0
Tru* ConctnIrotI on (ppb)
3.5 4.0
LABID-3
LABIO-B
T2 5
-§ 2.O
'] 1.5-1
J 10
"Eos
800
O.O O.S 1.0 1.5 2.O 2 5 3.O
True Concent rotion (ppb)
3.5
4.O
3.0
= 25
20
1.5-
I.O
O.S
O.O •
0
0.5
1 .0
1 .5
• Cc
2.0
introl i<
2.5
i (ppb)
3.0
3.5
Figure H-3H. ESTIMATE OF DETECTION LIMITS USING PE DATA SET WITH CONCENTRATIONS >1 PPB REMOVED. (GAMMA-BHC)
-------
Hubaux-Vos Method Detection Limits by Lab
LAB I 0-1
LABID-6
2.0
1.0
0.5
0.0
0.0
0.5 1 .O 1.5
Tru» Concentration (ppb)
2.0
2.0-
'•••
'•<>
0.5
...,
0.0
0.5 ^ 1.0 1.5
Tru* Concentration (ppb)
2.0
LABID-2
LAB 10-7
2.0-
!l.»-
1 .O-
0.5
O.O-
O.O
O.S 1.0 1.5
Truft Conc*ntration (ppb)
2.0
2.0
1 1 .5 •
1.0
O.5
00
0.0
0.5 1.0 1.5
Tru* Cortcftn t r a t i on (ppb)
2.0
20
: 1 .5-
1 .O
0.5-
00
0.0
LABID-3
O.S
2.0
1.5-
0.5
0.0
0.0
LAB I 0-8
0.5
1.0
1 .5
Figure H-4a. ESTIMATE OF DETECTION LIMITS USING BQC DATA. (1, 2-DICHLOROBENZENE)
2.0
-------
Hubaux-Vos Method Detection Limits by Lab
LABIO-1
LABIO»6
1
2.0 -
1.3-
I O
05
0.0
O.O
2.0 •
1 .O-
05
0.0-
0
2.0
'1.3
i .o •
OS
0.0
0.0
0.3 1.0 1.5
True Concentration (ppt>)
LABID-2
O.3 1.O 1 .5
True Concentration (ppb>
LABID-J
O'. 3 1 .0 15
True Concentration (ppb)
2.0
2.0
20-|
'l.S
1 .0
0.5
0.0
0.0
0.5 1.0 1.5
Tru* Concentration (ppb)
LAB ID-7
2.0
2.0 •
= 1.5
1.0
0.3
00
0.0
O.S 1.0 1.3
true Concentration (ppb)
LA8IO-8
2.0
20
'1.5-
1.0
0.5 •
0.0 -
0.0
0.3
1 .O 1.3
tfotion {ppb)
2.0
BQC sample data
Figure H-4b. ESTIMATE OF DETECTION LIMITS USING BQC DATA. (1,2, 4,-TRICHLOROBENZENE)
-------
Hubaux-Vos Method Detection Limits by Lab
LAB ID-I
LAB 10-6
1
2.0
1.3
1 .0 -
0.5
0.0
0.0
O.5 1.0 1.5
True Concentration (ppb)
2.O
I
20
1.5-
1.0
O. 5
0.0
o.o
0.5 1.0 1.5
True Concentration (ppb)
2.0
20
05
0.0
LAB 10-2
LABID-7
O.5 1.0 1.5
Tru* Concentration (ppb)
2.0
S
2.0-1
' 1.5
I .O
0.5
00
O.O
O.5 1.0 1.5
True Concentration (ppb)
2.O
LABlD-3
LAB 10-8
2.O-J
I
J
B
I ' O-
1 ° *
0.0 •
O
O.S 1.0 1.5
2.0
'l.5-
1 O
0.5
0.0 •
0.0
BQC sample data
0.5 1.0 1.5
True Concentration (ppb)
20
Figure H-4C. ESTIMATE OF DETECTION LIMITS USING BQC DATA. (1,2, 3, 4-TETRACHLOROBENZENE)
-------
Hubaux-Vos Method Detection Limits by Lab
LABIO-l
LAB I 0-6
2.0 ^
' \.i
I.O
0.5-
0.0
0.0
0.5 I.O 1.5
Tru* Concentration (ppb)
2.0
o.o
0.5 1.0 15
Tru* Concentration (ppb)
2.0
LABID-2
LAB 10-7
2.O -j
1.5
1.0
0.5
O.O
O.O
O.S I.O 1.5
Tru* Concentration (ppb)
2.0
f:
2.0-j
1.5
I .0 •
05
O.O
O.O
0.5 1.0 1.5
Tru* Conc*ntration (ppb)
2.0
LABIO-3
LABID-B
2.O
1.5
1 .O
0.5
O.O
O.O
2.0 -
"' 1 .5 •
1.0
O.S
0.0
O.S 1.0 1.5 2.O O.O O.S 1.0 I.5
Figure H-4d. ESTIMATE OF DETECTION LIMITS USING BQC DATA. (2-CHLORONAPHTHALENE)
2.0
-------
Hubaux-Vos Method Detection Limits by Lab
LAB I 0-1
LAB I 0-6
3
'2.*
2.0 •
1.5-
1.0
0.5
0.0
0.0
0.5 1.0 1.5
True Concentration (ppb)
2.0
"S 2.0 -
I"
51.0-
1 0.5
Jj 0. 0
0
O
0.5 1.0 1.5
True Concentration (ppb)
2.0
LAB I 0-2
LAB 10-7
2.0-
1.5-
0.5
0.0
O.S 1.0 1.5
Tru* Concentration (ppb)
2.O
2.5
2.0
1 5
1.0
0.5
0.0
o.o
0.5 1.0 1.5
True Concentration (ppb)
2.O
LAB 10-3
LAB I 0-6
True Concentration (ppb)
2.0
True Concentrotion (ppb)
2.0
Figure H-4e. ESTIMATE OF DETECTION LIMITS USING BQC DATA. (ALPHA-BHC)
-------
Hubaux-Vos Method Detection Limits by Lab
LABID=1
LAB I 0-6
.1.0
'2.5
2.O
1.5-
I .0
0.5
o.o-l
0.0
O.5 1 .O 1.3 2.0
True Concentration (ppb)
2.5
True Concentration (ppb)
2.5
LAS I 0-2
LAB ID-7
2.0
1 .O
O.5 •
0.0 -
O.O
O.5 1.O 1.3 2.O
True Concentration (ppb)
2.5
2.0
1 . 5
1.O
0.5
O.O
O.O
0.5
1.0 1.5
True Concentration (ppb)
2.O
2.5
LAB 10-3
LAB I 0-8
2.0-
I.S
1.0
0.5
O.O
0.5
1.O 1.5
True Concentration (ppb)
2.0
True Concentration (ppb)
2.5
Figure H-4f. ESTIMATE OF DETECTION LIMITS USING BQC DATA. (DELTA-BHC)
-------
Hubaux-Vos Method Detection Limits by .Lab
LABlO-1
LAB I 0-6
O.5
1.0 1.5 2.0 .2.5 JO
True Concentration (ppb)
3.5 4.0
1.O 1.5 2.0 2.5 i.O
Tru* Concentration (opb)
3.5 4.0
LABID-2
LABID-7
aj •
"S" 2.5-
'•f 2.0
I »•»•
3 i.o-
"8 0.5
J O.O-
O
0 O.S
1.0 1.5
True Co
2.O
iO t f O
2.5 3.O
> (ppb)
3.5 4.O
.3.0 -
!2.5
2.0 •
i .5 •
1.0-
0.5 -
00
00 0.5
1.0 1.5 2.O 2.3 3.0
tru« Concentration (ppb)
3.5 4.0
LABID-3
LAB I 0-8
L3.0
S2 5
2.O-
1 .5
IO
05
O.O •
O.O 0.5
1.0 1.5 2.O 2.5 3.O
True Concentrotion (ppb)
3.5 4.0
-S
~
20
1.5-
1.0
O 5
O.O -
0
.O O. 5
2.0
2.5 3.0
i (ppb>
Figure H-4g. ESTIMATE OF DETECTION LIMITS USING BQC DATA. (BETA-BHC)
3.5 4.0
-------
Hubaux-Vos Method Detection Limits by lab
LAB ID-I
LAB I 0-6
0.0 0.5
1.0 1.5 2.0 2.5 3.0
True Concentration (ppb)
3.5 4.0
.3.0-1
'2.5
2.0
1.5-
1 .0
O.5
O.O
0
0 0.5 t.O 1.5 2.0 25 3.0
True Concentration (ppb)
3.5 4.0
LAB 10-2
LAB 10-7
.3.0
'2.5
2.O •
».5
1.O
0.9
0.0
0.0 O.5 1.0
1 .9
True Co
2.O
2.5
' (ppb)
3.0 3.5 4.O
_3.<
20
1.5-
1 .O •
0.5-
O O
0
.0 0.5 1.0 1.5 2.O 2.5 3.0
True Concentrotion (ppb)
3.5 4.0
LABIO-3
LABIO-B
.3.0
!2 5-
2.O •
1.5
1 .O -
0.5
0.0
O.O O.S 1 .O
1 .5
True Co
2.0
It i
2.5
n (ppb)
3.0 3.5 4.O
3. OH
2'5
2.0
1 0
0.0 O.S 1.O 1.5 2.O 2.5 3.O
True Concentration (ppb)
3.5 4.0
Figure H-4h. ESTIMATE OF DETECTION LIMITS USING BQC DATA. (GAMMA-BHC)
-------
APPENDIX I
Chemical Data Quality Assessment
Soil Assessment-Indicator Chemicals
-------
Appendix I
CHEMICAL DATA QUALITY ASSESSMENT
SOIL ASSESSMENT—INDICATOR CHEMICALS
Since the chemical data quality and the interlaboratory
variability were significant issues during the pilot study
(CH2M HILL, 1986) , the authors feel it is appropriate to
expand on the limited presentation of the evaluation of data
quality contained in Section 6.2 and Appendix J of
Volume III. The soil assessment for indicator chemicals was
designed to provide chemical data of known and defensible
quality. As this evaluation is important for all future
uses of the data, this appendix expands the discussion.
Also contained in this discussion are the results of the MHB
and FHB analyses requested by Dr. Joan Daisey and the data
completeness information requested by Dr. Janick Artiola.
BACKGROUND
Two general approaches are used to assess data quality:
classical and chemometric. The classical approach is based
on descriptive statistics and the comparison of results
against control limits that were determined before the labo-
ratory analyses were initiated. This evaluation is based on
the Data Quality Objectives (DQO) of representativeness,
completeness, accuracy, precision, and comparability, which
were specified before the project was conducted. The
chemometric approach (Delaney, 1984; Massart et al., 1988)
uses aggregate assessments of data quality. It is conducted
as an exploratory data analysis in which large amounts of
data are examined using informative graphical presentations
to expediently discover biases, differences in variability,
or time trends that might not have been adequately con-
trolled in the chemical analysis and that might adversely
affect some uses of the chemical results.
The quantitation accuracy was 60 to 80 percent for all LCICs
except 2-CNP, for which the accuracy was 50 percent. The
within-laboratory precision, as measured using Matrix Spike/
Matrix Spike Duplicate (MS/MSD) Relative Percent Differences
(RPD), was typically below 20 percent except for 2-CNP, for
which the RPD averaged 20 to 40 percent in five of the six
laboratories performing chemical analysis. The between-
laboratory precision, as measured using field-split samples,
was below 50 percent for the non-BHC LCICs and was in the
range of 40 to 70 percent for the BHC LCICs.
Quantitation of 2-CNP appears to be less stable and biased
lower than the other LCICs. While this difficulty is not
expected to seriously affect the statistical comparisons for
1-1
-------
habitability, alternative uses of the 2-CNP results should
be made with caution.
Laboratory 6 had the roost consistent performance, as evi-
denced by high accuracy and precision. Laboratory 1 had
high precision but was biased slightly low relative to the
other laboratories. Laboratory 3 displayed consistent evi-
dence of increased variability relative to the other
laboratories.
Both the classical and the chemometric assessment support
the conclusion that there were no major defects in chemical
data quality. Within the constraints of the preset control
limits for each QC measure, there were subtle differences in
accuracy, precision, and comparability from one laboratory
to another. With the minor exception of the advisory recov-
ery limits for matrix spikes, all laboratories were able to
meet the DQOs.
APPROACH
The chemical data from the soil assessment for indicator
chemicals are examined in terms ot the five DQOs: represen-
tativeness, completeness, accuracy, precision, and compar-
ability. This examination includes both qualitative
descriptions and quantitative reporting of results. (The
DQO of sensitivity is discussed in Appendix H, which dis-
cusses detection limits.)
Each of the five DQOs is presented separately in the follow-
ing section. A familiarity with the chemical analysis
method used in the soil assessment for indicator chemicals
and with the concepts of analytical chemistry and statistics
is assumed.
Chemometric and statistical tools were used for visual
examination of the data. Box plots and trend plots were
employed to readily facilitate the examination of a large
amount of data. The laboratories are indicated on these
plots by numbers 1 through 8. (Laboratory 5 is not shown
and ultimately did not participate in the study; samples
prepared by Laboratory 4 were analyzed by Laboratory 1.)
Where appropriate, control limits are indicated by horizon-
tal lines. On the box plots, points that are beyond the
range of the plot are included by numerical annotations.
(See page 6-18 of Volume III for more details on how to read
box plots.) On the trend plots, a least squares regression
line has been fit to the data. The vertical lines indicate
the dates of initial calibrations.
This appendix includes summary tables of the laboratory per-
formance and examples of the box plots and trend plots used
to examine the chemical data. At each reference to an
1-2
-------
example figure in the text, a parenthetical note is given
describing the specific example provided. These examples
are included to show the reader the form of data presenta-
tion. The entire set of figures was used to develop the
observations and comparisons that are discussed in the text
of this appendix.
DISCUSSION
As stated previously, the soil assessment for indicator
chemicals was designed to provide data of known and
defensible quality. To achieve this goal, several data
QA/QC objectives were selected; these objectives relate to
the data's representativeness, completeness, accuracy, pre-
cision, and comparability.
Representativeness is discussed in terms of the number of
samples collected and analyzed in the study.
Completeness of the field sample data has already been dis-
cussed in Volume III. Statistically useful results were
obtained for 85.7 percent of the samples analyzed in the
study. Completeness of the QA/QC data is assessed in terms
of the number of QA/QC analyses that meet the specified QC
criteria. Tabulations of these results are presented.
Both accuracy and precision are examined in three ways.
First, a summary table of descriptive statistics is presen-
ted for each of the QC measures that relate to accuracy or
precision. Second, examples are given to show the qualita-
tive assessment of accuracy and precision performance dis-
played by the participating laboratories. Finally, a
quantitative, statistically based statement is made regard-
ing the accuracy or precision of the data from this study.
Comparability of the data produced by each of the partic-
ipating laboratories has been assessed by studying box plots
of a wide variety of instrument and method performance mea-
sures. Time-series trend plots have been used to determine
the extent to which various chemical analysis parameters,
such as surrogate recoveries and relative response factors,
remained stable for the duration of the laboratory analysis
phase of the project.
The following paragraphs assess whether these QA/QC objec-
tives have been achieved.
REPRESENTATIVENESS
To obtain representative samples, the sampling design spec-
ified that sample collection be randomly allocated among
collection teams and neighborhoods. A detailed discussion
1-3
-------
of the sampling design to achieve the sample representative-
ness can be found in the sample collection and preparation
QAPP (CH2M HILL, 1987a). As indicated by Table 1-1, a total
of 879 field samples were collected out of the 894 planned.
These soil samples were collected from the seven sampling
areas of the EDA and the three comparison areas (Cheektowaga
and Tonawanda combined, Census Tract 221, and Census Tract
225). The agreement between the actual and planned sample
allocations confirms that the original randomization was
maintained with respect to the sampling crew and sample
analysis laboratory. The representativeness criterion, as
described in the design of the sampling program, was
achieved.
COMPLETENESS
Of the 879 field samples collected and sent to seven analyt-
ical laboratories for analysis of the LCICs (Table 1-1), 781
samples were actually analyzed; 97 samples were not analyzed
because of problems at Laboratory 4, and one sample was not
analyzed at Laboratory 3. In addition to the regular field
samples, two types of field QC samples were collected and
analyzed by the laboratories. Table 1-2 shows the number of
field samples and field QC samples that were analyzed by
each laboratory.
After the sample analysis, the electronic and hardcopy data
provided by the laboratories were subjected to a detailed
data assessment by EMSL-LV. As a result, an assignment of
data qualifiers were assigned to each sample and each LCIC
analyte within a sample. Each LCIC result was defined as
Good, Uncertain, or Bad. Table 1-3 gives the percentage of
the Good, Uncertain, and Bad summary qualifiers for each
laboratory. As shown in Table 1-3, the total Bad data of
only 6.6 percent indicate the excellent performance of the
laboratories using a method of this complexity.
Table 1-4 summarizes the laboratory QC samples required by
the QA/QC program and analyzed by the laboratories to
improve the reliability of the data. In addition to these
QC samples, other QA/QC measures, e.g., surrogates, internal
standards, etc., as specified in the analysis QAPP (CH2M
HILL, 1987b) were also performed by the laboratories. QC
control limits were set for the various measures of these QC
samples. The QAPP-specified goal was that the laboratories
should meet the QC control limits for either 90 percent or
100 percent of the analyses, depending on the type of QA/QC
measures involved. Table 1-5 summarizes the results of all
QC sample data. The results from all field, field QC, and
laboratory QC samples are included and summarized in the
table.
1-4
-------
As shown in Table 1-5, the QAPP-specifled goal was met for
all types of QA/QC measures, with the minor exception of
MS/MSD analyses, which had advisory control limits. The
results of these QA/QC measures indicate that the analytical
program was successfully implemented within a constraining
schedule and heavy sample analysis load. The laboratories,
in general, followed the QA/QC requirements, as specified in
the analysis QAPP. This increases the confidence of the
results reported by the laboratories.
ACCURACY
For this study, accuracy has been assessed using three types
of measures: surrogate recovery, matrix spike recovery, and
BQC sample recovery. The MHB and FHB results have also been
examined for evidence of bias. Three surrogate standard
compounds, 1,4-dibromobenzene (DBB), 2,4,6-tribromobiphenyl
(TBBP), and 1,2,4,5-tetrabromobenzene (TeBB), were spiked
into each field sample and QC sample at the beginning of the
sample preparation process. While they are not LCICs, the
surrogate compounds were chosen to mimic the behavior of the
LCICs during sample preparation and analysis. 1,4-DBB and
2,4,6-TBBP were spiked at the 10.0-ppb level, and
1,2,4,5-TeBB was spiked at 1.0 ppb as a low-level surrogate
to check the sensitivity of the method. Since surrogates
were spiked into each field and QC sample, a large amount of
surrogate recovery data is available.
The MS/MSD QC samples were created for 1 out of every
20 field samples by spiking 5.0 ppb of LCICs into a portion
of a field sample. The MS recovery measures the accuracy of
the chemical analysis method.
The BQC samples, ranging in LCIC concentrations from 0.2 ppb
to 3.0 ppb, were prepared in a soil or sand matrix by
EMSL-LV. The participating laboratories extracted one BQC
sample with each extraction batch of up to 10 field samples.
The BQC samples were single-blind, since the laboratories
knew which samples were BQCs but did not know the concen-
tration of LCICs in the BQC samples. The recovery of LCICs
from the BQC samples is a measure of method accuracy.
Surrogate Recoveries
Box plots were prepared to illustrate the recoveries
achieved by each laboratory for each of the three surro-
gates. A separate box plot is shown for each surrogate for
the field samples and for the QC samples analyzed by each
laboratory. An example is shown in Figure 1-1 (1,4-DBB sur-
rogate recoveries). The box plots for surrogate recoveries
from all laboratories had whiskers that came close to, or
went beyond, one or both control limits. Thus, meeting the
surrogate recovery criteria was reasonably difficult, and
1-5
-------
reinjections and re-extractions were needed for surrogate
recoveries outside the control limits. When the surrogate
recovery was outside the control limit in the same direction
for both the original sample and the re-extraction, the lab-
oratory submitted both sets of data to show that there was a
consistent problem beyond its control. Such problems tended
to happen occasionally when a matrix interference caused a
high surrogate recovery.
A summary of surrogate recovery results is shown in
Table 1-6 for field samples and in Table 1-7 for laboratory
QC samples. Table 1-6 includes surrogate results for 18
duplicate sample analyses, as well as for any sample that
required dilution because of high LCIC concentration.
As shown in the tables and box plots, the surrogate recovery
tends to be lower than 100 percent because of unavoidable
losses from the sample extraction and cleanup. For 1,4-DBB,
the mean recoveries ranged from 66 to 87 percent and the
median recoveries ranged from 64 to 80 percent. Labora-
tories 1 and 3 tended to be on the low side, and Labora-
tories 2 and 8 tended to be on the high side. However, no
significant volatility discrimination was observed among the
laboratories. That result is contrary to the findings of
the pilot study; it indicates that method modifications and
additional training in the extraction protocol were success-
ful in producing more uniform extractions among the
laboratories.
For 2,4,6-TBBP, the mean and median recoveries ranged from
75 to 104 percent and 70 to 100 percent, respectively. Lab-
oratories 1 and 6 were on the low side, while Laboratories 3
and 7 were on the high side. Laboratories 3 and 7 tended to
experience more outliers outside the advisory upper control
limit than the other laboratories.
For 1,2,4,5-TeBB, the low-level surrogate, the mean and
median recoveries ranged from 71 to 214 percent and 65 to
95 percent, respectively. Laboratory 1 experienced
unusually high recoveries of 1,2,4,5-TeBB for one field sam-
ple and its re-extraction/reinjection analysis. These high
surrogate recoveries, in the range of 2,000 to 10,000 per-
cent, were probably caused by a matrix interference and con-
tributed to a high mean percent recovery. Laboratory 3
again showed high recoveries and tended to be outside the
upper limit.
It is interesting to note that all three surrogates tended
to have slightly higher recoveries in the QC samples than in
the field samples. The effect is most notable for the first
surrogate, 1,4-DBB, suggesting that the soil matrix or inter-
ferences may cause recoveries to decrease.
1-6
-------
MS Recoveries
The MS and MSD recoveries were examined using box plots for
each laboratory by LCIC. An example is shown in Figure 1-2
(A-BHC matrix spike recoveries). The advisory limits for MS
recovery are also shown on the plot. A summary of MS and
MSD recoveries is shown in Table 1-8. The table contains a
count of MS/MSD samples, the mean recovery, the relative
standard deviation, the maximum recovery, and the minimum
recovery.
The MS recovery bias was comparable from laboratory to labo-
ratory, with medians typically in the range of 60 to 80 per-
cent. There are, however, several notable deviations:
(1) Laboratory 1 had low MS recovery for all LCICs
except 2-CNP.
(2) Laboratory 4, for which only one MS/MSD pair was
analyzed (because of the sample preparation prob-
lems noted in Volume III), also showed this
behavior.
(3) All laboratories experienced low recoveries for
2-CNP, with medians in the range of 28 to 60 per-
cent and means in the range of 11 to 56 percent.
This tendency toward low recovery for 2-CNP was
noted during the method validation, but a complete
explanation was never found.
There was good agreement in median and mean MS recovery
between the laboratories for all LCICs. Such an agreement
indicates that samples were being prepared and analyzed
using comparable procedures in all laboratories. The excep-
tion is Laboratory 1, which had a slight negative bias rela-
tive to the other laboratories for all LCICs except 2-CNP.
Laboratory 4 also appeared to have difficulty (low recov-
eries) , but there is not enough information to fully assess
this problem.
BQC Sample Recoveries
Box plots and summary tables of percent recovery for EPA BQC
samples were examined for each LCIC by laboratory. An exam-
ple box plot is shown in Figure 1-3 (G-BHC BQC recoveries).
Table 1-9 summarizes the percent recovery of the BQC sample
analysis. The mean percent recovery was used to assess
bias. The mean percent recoveries ranged from 60 to 89 per-
cent for all LCICs with the exception of 2-CNP, which had
mean percent recoveries of 50 percent. The relative stan-
dard deviation of BQC recovery ranged from 15 to 72 percent.
The percent recovery for each laboratory and LCIC tended to
be lower than 100 percent because of the unavoidable losses
1-7
-------
experienced during sample extraction and cleanup. There
were no major differences in bias. Laboratory 2's percent
recoveries were biased slightly low for six of eight LCICs
and, to a lesser extent, Laboratory 1's recoveries were low
for five of eight LCICs. Laboratory 8 showed a slightly
high bias for 1,2-dichlorobenzene (1,2-DCB), caused by a
small amount of laboratory contamination, as indicated from
the method/holding blank results.
Table 1-9 also shows that the percent recoveries of the LCIC
from sand samples were generally lower than from the soil
samples, except for Laboratory 7. This notable effect could
be a consequence of differences in the way the sand and soil
BQC samples were prepared.
MS/BQC Comparison
An interesting comparison is seen when the MS and BQC recov-
ery results are examined together. An example is shown in
Figure 1-4 (B-BHC MS and BQC recoveries). No acceptance
limits are shown for BQC recovery because these acceptance
limits varied with the concentration of the LCICs in the BQC
samples. The MS samples were spiked at the 5.0-ppb level,
while the BQC concentrations ranged from 0.2 ppb to 3.0 ppb.
However, there is a nearly consistent subtle bias between
the MS and BQC recoveries. For 1,2-DCB and 1,2,4-trichloro-
benzene (TCB), the BQC recoveries tend to be higher than the
MS recoveries. For the rest of the LCICs, with the excep-
tion of 1,2,4,5-TeCB, the MS recoveries tend to be higher
than the BQC recoveries. This minor effect might be
explained by the relative volatility of the LCICs and the
manner in which these two types of QC samples were prepared.
The MS samples were prepared just before extraction, while
the BQC samples were prepared in advance and stored before
extraction.
Method/Holding Blanks
An MHB sample, consisting of soil that was demonstrated to
be uncontaminated with LCICs or LCIC interferences, was
extracted with every batch of field samples. The MHB was
acceptable if the equivalent concentration of LCIC or LCIC
interferences was below 0.5 ppb (0.6 ppb for 1,2-DCB).
Because of the inclusion of a blank requirement on LCIC
interferences, the acceptance rules for MHBs were involved
so that unnecessary re-extractions were minimized. The MHB
interpretation rules were slightly subjective, because a
visual examination of the selected ion current profile was
needed to determine if a potential LCIC interference peak
was close enough to the LCIC retention time to affect
quantitation.
1-8
-------
Table 1-10 provides a summary of the MHB results for each
LCIC and each laboratory. This summary is based solely on
the computerized MHB equivalent concentrations reported by
the laboratories and does not include the results of apply-
ing the MHB interpretation rules. All MHBs reported in the
study were accepted by the EMSL-LV when the SICPs were
scrutinized.
The slightly elevated levels of 1,2-DCB, especially notable
in Laboratory 8, were caused by minor laboratory contamina-
tion. The high concentrations of B-BHC and G-BHC were
caused by the ubiquitous interferences arising from the lab-
oratory detergent used to clean glassware. All laboratories
experienced this interference to some extent and had to be
exceedingly careful to either minimize the concentration of
the interference or to chromatograhically separate it from
the BHCs.
Field Handling Blanks
A number of double-blind FHBs were analyzed by each labo-
ratory during the study to demonstrate that field samples
were not being contaminated with LCICs or LCIC interferences
during the sample collection, soil preparation, or chemical
analysis processes. Table 1-11 summarizes the FHB results
and allows a comparison of the FHB results with the MHBs
extracted in the same batch. Most LCICs show only a very
slight increase in concentration of the FHBs compared to the
MHBs. The largest effect is for 1,2-DCB, which increased by
an average of 0.1 to 0.2 ppb in all laboratories. All other
increases were less than 0.1 ppb.
There are two possible causes for this marginal effect. The
first possibility is that there were low levels of LCICs in
the FHB matrix. The MHBs were based on a sand matrix care-
fully prepared and provided by EPA to be used as a method
blank material. The FHBs were based on a soil matrix
selected to be similar in composition to soil from the Love
Canal area. While chemical analysis showed this soil to be
acceptable as a blank matrix, there could be low levels of
LCICs present. Another possibility is that the slight ele-
vation of FHB concentrations over the MHB concentrations is
caused by small amounts of contamination experienced during
sample collection, soil preparation, or transportation.
This small degree of contamination did not trigger any cor-
rective action during the laboratory phase of the project
and is not expected to degrade the usefulness of the
results, since the contamination levels observed are less
than the precision of the analysis method.
1-9
-------
PRECISION
Precision can be assessed using the same three QC measures
used to assess accuracy: surrogate recovery, MS recovery,
and BQC recovery. Variations of surrogate recovery, MS
recovery, and BQC recovery are each measures of precision.
This assessment would include sources of variability over
the course of the laboratory phase of the project. The
intrinsic, within-laboratory precision of the analysis
method can be assessed by comparing the MS and MSD recov-
eries, since these pairs of samples are generally extracted
and analyzed at the same time. The MS/MSD precision is
usually expressed as an RPD of the paired recoveries. In
addition, field split samples have been used to assess
interlaboratory precision.
Surrogate Recoveries
The box plots and summary tables of surrogate recoveries
described above under Accuracy (Figure 1-1 and Tables 1-6
and 1-7) were also used to qualitatively assess the preci-
sion of sample preparation. For 1,4-DBB, the precision of
surrogate recoveries was comparable among the laboratories,
with Laboratory 1 being slightly less variable and Labora-
tory 2 slightly more variable. For 2,4,6-TBBP, Laboratory 3
was slightly more variable and Laboratory 6 was slightly
less variable than the other laboratories. For 1,2,4,5-
TeBB, Laboratory 3 was noticeably more variable than the
other laboratories, while Laboratories 2 and 6 were slightly
less variable.
MS Recoveries
Using the MS/MSD percent recovery box plots described above
under Accuracy (Figure 1-2, A-BHC surrogate recoveries), it
is possible to assess the sample preparation and analysis
precision achieved over the course of the project. There
were some notable differences among laboratories in preci-
sion. Laboratories 1 and 6 showed better precision than the
other laboratories for four to five LCICs. Laboratories 3
and 8 showed poorer precision than the other laboratories
for five to seven LCICS.
Within-laboratory precision was assessed using data from the
MS/MSD pairs. Table 1-12 summarizes the precision results
for MS/MSD analyses. The Relative Percent Difference (RPD)
between MS and MSD percent recoveries were examined for each
LCIC by laboratory, using Table 1-12 and box plots. An
example of the box plots is shown in Figure 1-5 (5a: 2-CNP,
5b: A-BHC MS/MSD RPDs).
The advisory limit for MS/MSD RPD was 30 percent (35 percent
for D-BHC). This advisory limit was met for
1-10
-------
nearly all LCICs by all laboratories except for 2-CNP at
roost laboratories and B-BHC at Laboratory 8. As shown in
Figure 1-5 and Table 1-12, Laboratories 1 and 6 achieved
noticeably lower RPDs than roost other laboratories for five
or six LCICs. Laboratories 3 and 8 had RPDs that were
higher than the other laboratories for five LCICs. Of note
were the RPDs for 2-CNP, which were higher in all labora-
tories except Laboratory 1 than for all other LCICs. As
noted above in the Accuracy section, this variability was
also accompanied by markedly low recovery. This effect was
noted during the method development and validation, but no
acceptable explanation has been found.
BQC Sample Recoveries
The variability of percent recovery for BQC samples was
examined for each LCIC by laboratory using the box plots
(Figure 1-3, G-BHC BQC recoveries) and summary table
(Table 1-9) described above under Accuracy. The Interquar-
tile Range (IQR), which is the height of the box, was used
to assess precision. The IQRs varied from a low of about
10 percent to a high of about 50 percent.
Precision, as seen from the IQR, varied somewhat from LCIC
to LCIC. Laboratories 2, 6, and 7 showed slightly lower
precision for one or two of the eight LCICs. Laboratories 1
and 6 showed high precision for five of the eight LCICs.
As demonstrated in Table 1-9, which shows the sand samples
separately from the soil samples, the percent Relative Stan-
dard Deviation (RSD) for the sand samples was generally
lower than for the soil samples. This difference is
probably because the sand samples were more homogeneous and
had less interferences than the soil samples. The matrix
interference in some soil samples also prevented the reli-
able identification of LCICs, especially for the BHCs. This
may be the reason for the zero percent recovery of the BHC
compounds.
Field Split Samples
A number of field samples were split in half in the soil
preparation laboratory and were sent to two different labo-
ratories. A comparison of the results from these field
splits was made to assess the analytical method precision.
These results are expressed as RPD between the pairs of
results and are presented as box plots in Figure 1-6. (Only
two field splits were sent to the same laboratory, as dis-
cussed in Volume III. These within-laboratory splits are
denoted "w" in Figure 1-6, while the between-laboratory
splits are denoted "b.") Only the results with a detected
concentration have been included in Figure 1-6. There were
actually 62 pairs of between-laboratory split samples, but
1-11
-------
many of these yielded a non-detect result for one or both
halves of the split for the BHCs and 2-CNP. For the non-BHC
LCICs, the median RPDs were below 50 percent. For the BHCs,
the median RPDs ranged from 40 percent for A-BHC up to
70 percent for D-BHC.
COMPARABILITY
Comparability is only qualitatively defined and assessed in
the classical QA/QC approach, but it can be examined more
rigorously using chemometrics. These assessments include
both intralaboratory and interlaboratory comparability. A
lack of comparability means that using these data for pur-
poses that depart from the balanced design of the soil
assessment for indicator chemicals could result in erroneous
trends or results.
Initial Calibration (1C)
Box plots of the RRFs for each of the five calibration stan-
dards from each of the ICs for each LCIC and surrogate stan-
dard by laboratory were examined. Figure 1-7 (2,4,6-TBBP 1C
RRFs for Laboratory 3) shows an example. The number of ICs
ranged from 3 for Laboratory 8 to 10 for Laboratory 3.
Since Laboratory 3 performed a large number of ICs, it
appears that there was a problem with maintaining stable
instrument performance in that laboratory. There tended not
to be a consistent pattern in the RRF values from one 1C to
the next within the laboratories; this was especially
noticeable in Laboratory 3. This indicates that either the
instrument sensitivity was not gradually degrading during
the course of the study as might be expected, or the instru-
ment operators were compensating for any degradation by
changing instrument operating parameters.
The RRFs generally agreed from laboratory to laboratory for
each LCIC and surrogate, except for 1,2-DCB at Laboratory 6,
which had a noticeably higher Relative Response Factor (RRF)
than the other laboratories. In general, Laboratory 6 tend-
ed to have tight, precise ICs, with RRFs that varied in mag-
nitude from one 1C to another, while Laboratory 8 tended to
have a broad range of 1C RRFs that did not change much from
one 1C to another. This suggests that the GC/MS in Labo-
ratory 6 was more linear over the concentration range of the
calibration standards while the GC/MS in Laboratory 8, while
not quite as linear, was more stable. This conclusion is
supported by the observation that Laboratory 8 only needed
to perform three ICs over the course of the project.
Performance Check Ion Ratios
The identification of the presence of LCICs in samples is
confirmed using the ratio of the secondary to the primary
1-12
-------
ion area. For a confirmed identification this ratio must be
within ± 20 percent of the theoretical ratio. If a labor-
atory typically gets ion ratios that are different from the-
ory, it may get a higher fraction of false negatives because
of failed ion ratios. Non-detects resulting from failed ion
ratios can also be caused by chemical interferences. The
ion ratios obtained for each performance check chromatogram
were examined using box plots. An example is shown in
Figure 1-8 (2-CNP ion ratios).
All laboratories had good agreement between the first (PCI)
and second (PC2) performance check ion ratios. For the non-
BHC LCICs, all laboratories were well within the ion ratio
control limits, and the ion ratios were comparable from
laboratory to laboratory. For the BHCs, ion ratio variabil-
ity often approached the control limits. There were also
notable biases in the ion ratios from laboratory to
laboratory.
For 1,2-DCB, all laboratories had comparable ratios except
Laboratory 6, which had high outliers. For 1,2,4-TCB, Lab-
oratory 6 was more variable than the other laboratories.
For 1,2,3,4-TeCB, Laboratory 3 had notably tight ion ratios.
For 2-CNP, Laboratories 2 and 3 were biased slightly high
and Laboratory 6 was biased slightly low. For all four
BHCs, Laboratories 2 and 8 had more variable ion ratios that
approached the upper control limit. Laboratory 3 tended to
have less variable ion ratios for the BHCs. For the inter-
nal standard DIO-pyrene (IS5), the ion ratios were precise
and comparable except for Laboratories 2 and 3, which had
ratios that were variable and biased high.
Performance Check Sensitivity
During each performance check chromatogram, the sensitivity
of the GC/MS is evaluated by measuring the Signal to Noise
Ratio (S/N) for the LCIC B-BHC and the surrogate standard
1,2,4,5-TeBB at the 0.5-ppb level. The control limit is
that the S/N be above 2.5. These S/Ns were examined using
box plots. An example is shown in Figure 1-9 (B-BHC S/N).
A low S/N indicates that the instrument is not performing
sensitively. In this case, the laboratory may experience
false negatives because of inadequate sensitivity. A tight
dispersion of S/N indicates good precision and instrument
stability.
For B-BHC, the S/N tended to be in the range of 5 to 25.
Laboratory 2 tended to have a low but precise S/N of about
5. Laboratory 3 had a high but highly variable S/N from
below 2.5 to 90. Typically, the PC2 S/N was slightly lower
than that for PCI, indicating that the instrument tended to
lose some sensitivity during the analysis shift. For TeBB,
1-13
-------
the S/Ns were comparable from laboratory to laboratory at
about 5 to 10, with two exceptions: Laboratory 8 had a
higher S/N of about 10 to 20, and Laboratory 3 had a highly
variable S/N from below 2.5 to about 65.
Performance Check Resolution
Chromatographic efficiency had to be demonstrated at the
beginning and the end of each analysis shift. This was
determined using the resolution between 1-chloronaphthalene
(CNP) and the LCIC 2-CNP for both the PCI and PC2 and
between B-BHC and G-BHC in PC2. The resolution is expressed
as the fractional height of the valley between the two peaks
relative to the average height of the two peaks. This is
referred to as the percent valley (%V). Chromatographic
resolution was examined using box plots. An example is
shown in Figure 1-10 (2-CNP PC1%V).
Laboratories 3 and 7 had small, precise %Vs, indicating
excellent resolution and good Chromatographic stability.
The %V for Laboratories 1, 6, and 8 was larger and more
variable, indicating less ideal Chromatographic performance.
None of the laboratories had difficulty meeting the %V con-
trol limit. For BHC resolution, Laboratories 3 and 7 con-
tinued to have excellent and precise resolution.
Laboratories 1 and 8 had some outliers that approached the
%V control limit.
Continuing Calibration (CC) Relative Response Factors
In the CC chromatogram at the beginning of each analysis
shift, the RRF for each LCIC and surrogate is determined and
compared to the mean RRF obtained from the most recent ini-
tial calibration. Calibration stability and comparability
were examined using box plots of the percent difference
between the CC and 1C RRFs for each laboratory by LCIC and
surrogate. An example is shown in Figure 1-11 (B-BHC: CC
RRF) .
While there were some notable differences in precision,
there was not a consistent pattern from one LCIC to another.
It seems that some laboratories were able to hold a stable
calibration for one LCIC and other laboratories for another
LCIC. The precision for the non-BHC LCICs appears to be
more stable than for the BHCs and the surrogates. This
might mean that compounds eluting early in the chromatogram
are quantitated more reliably than those eluting later.
IS Area Variation
Box plots of the percent difference between the IS area for
each field and QC sample and the corresponding CC standard
were examined for each laboratory. An example is shown in
1-14
-------
Figure 1-12 (IS4 area percent difference). The horizonal
lines indicate the control limits for this parameter.
Generally, the performance for all five ISs was comparable
among all laboratories. In many cases the box plot whiskers
came close to, or extended beyond, the control limits. This
indicates that the laboratories experienced some difficulty
in staying within the limits. There were only slight biases
between field and QC samples within a laboratory. These
biases were not consistent from laboratory to laboratory.
In general, Laboratories 1 and 6 tended to be slightly more
precise, while Laboratories 2 and 7 tended to be slightly
more variable.
IS Raw Area Trends
Trend plots of raw IS areas versus the date of analysis were
examined for each of the five ISs to assess the long-term
stability of the operation of the GC/MS instrument in each
laboratory. An example is shown in Figure 1-13 (IS area
trend for Laboratory 1). For this and the remaining trend
plots, vertical lines indicate the dates of initial
calibration.
The typical behavior for the IS areas is to decrease
gradually over time as the instrument is used, suggesting
that the sensitivity of the analysis is gradually decreas-
ing. Such a downward trend was seen in Laboratories 2 and
8. Other laboratories tended to show this trend over
shorter periods of time, with some discontinuities, after
which noticeably higher IS areas were seen. These breaks in
the trend are probably caused by cleaning the MS source or
quadrupoles, by replacing the GC column or injection port
liner, or by increasing the MS accelerating voltage or
detector gain. Abrupt changes in IS area are generally syn-
chronized with new ICs. An upward trend could be caused by
an increase in the concentration of the IS spiking solution
due to evaporation of the solvent.
IS Retention Time (RT) Trends
Long-term stability of the GC/MS was also examined using
trend plots of IS RT versus the analysis date. An example
is shown in Figure 1-14 (ISS RT trend for Laboratory 1).
Typical behavior is a gradual decrease in RT as the GC col-
umn stationary phase is degraded or compromised by the field
sample extract matrix or as short pieces of the front end of
the fused-silica GC column are cut off to maintain accept-
able GC performance. Such a trend is seen in Laboratory 8.
Discontinuities in IS RT can be caused, for example, by
replacing the GC column, changing the GC temperature pro-
gram, or by changing the carrier gas flow rate. Such behav-
ior of stability, punctuated by breaks in the IS RT, were
1-15
-------
seen at Laboratories 1, 2, and 7. Laboratory 6 displayed a
less obvious trend, with IS RTs changing over a smaller
range than changes at other laboratories. Laboratory 3 had
a gradual upward trend with one upward discontinuity,
indicating that perhaps the carrier gas flow rate was
decreasing.
CC RRF Trends
Ideally, the RRFs should display no discernible trends over
the course of the study. Any change in the response of the
instrument to an LCIC should be reflected by a corresponding
change in the response to the associated IS, keeping the RRF
constant. Continuing calibration RRF stability was examined
by preparing time series plots for each LCIC and each labo-
ratory. An example is shown in Figure 1-15 (CC RRF trend
for Laboratory 6). The trends themselves were well-behaved,
either gradually increasing, decreasing, or staying essen-
tially flat. The pattern of trends for each LCIC across
laboratories or for each laboratory across LCICs was, how-
ever, complex. Each laboratory displayed a combination of
trends that differed from every other laboratory. This sug-
gests that the relative sensitivity of the GC/MS instruments
to LCICs and ISs was changing independently in each labo-
ratory. Presumably, the CC performed at the beginning of
each analysis shift correctly determined these relative sen-
sitivity changes, resulting in accurate quantitation. Since
the laboratories passed the criteria for every EPA check
standard for calibration accuracy, it appears that there was
accurate quantitation, thus demonstrating the importance of
CC.
Surrogate Recovery Trends
Surrogate standards are spiked into each sample prior to
extraction. Surrogate recoveries are used as a QA/QC mea-
sure to demonstrate that sample extraction and cleanup have
been efficiently performed. Low surrogate recovery is taken
as an indication that the laboratory has failed to success-
fully prepare samples. Excessively high surrogate recovery
can be caused by matrix interferences. Trend plots of sur-
rogate recovery versus date of analysis were examined to
assess the long-term stability of the sample preparation
process. An example is shown in Figure 1-16 (1,2,4,5-TeBB
surrogate recovery for Laboratory 8).
For most laboratories, only random variation was seen. Only
very slight upward or downward trends were discernible,
indicating that the sample preparation process was being
employed consistently over the course of the project. Lab-
oratories 3 and 4 exhibited a higher degree of variability
or more frequent outliers, indicating a lower level of dili-
gence at those laboratories. Laboratory 7 had a higher
1-16
-------
degree of variability at the beginning of the study, showing
some difficulties in initiating the project. Laboratory 8
showed the opposite trend, indicating that sample prepara-
tion diligence degraded near the end of the project.
SUMMARY
A summary assessment of chemical data quality for this proj-
ect is difficult because of the wide range of performance
indicators examined. Certainly, no major defects in data
quality were found, and the notable observations presented
in this appendix are predominantly subtle differences in
laboratory performance.
Laboratory 6 had the most consistent performance, as evi-
denced by high accuracy and precision. Laboratory 1 had
high precision but was biased slightly low relative to the
other laboratories. Laboratory 3 displayed consistent evi-
dence of increased variability relative to the other
laboratories.
Quantitation of 2-CNP appears to be less stable and biased
lower than the other LCICs. While this difficulty is not
expected to seriously affect the statistical comparisons for
habitability, alternative uses of the 2-CNP results should
be made with caution.
WDR363/022
1-17
-------
REFERENCES
CH2M HILL 1986. Pilot Study for Love Canal EDA Habitability
Study. (Vol. I and II.)
CH2M HILL 1987a. Love Canal Habitability Study—Soil Sample
Collection and Preparation Quality Assurance Project
Plan.
CH2M HILL. 1987b. Love Canal Habitability Study—Soil
Sample Analysis Quality Assurance Project Plan.
Delaney, M.F. 1984. "Chemometrics." Analytical Chemistry,
56-261R-277R.
Massart, D.L., Vandeginste, B.G.M., Deming, S. N., Michotte,
Y., and Kaufman, L. 1988. Chemometrics; A Textbook.
Elsvier Science Publishing Company, Inc. Amsterdam.
WDR363/022
1-18
-------
Table I-l
NUMBER OF FIELD SAMPLES RECEIVED AND ANALYZED
BY LABORATORY
Analytical.
Laboratory'
1
2
3
4
6
7
8
TOTAL
No. Of
Samples
Planned
122
125
128
135
137
110
137
894
No. of
Samples
Received
119
121
125
135
133
109
137
879
No. of
Samples
Analyzed
119 + 38b
121
124*:
Od
133
109
137
781
Laboratory 5 was not selected for study participation.
Thirty-eight samples extracted by Laboratory 4 were sent to
Laboratory 1 for analysis.
•*
'One sample was not analyzed.
Laboratory 4 did not successfully analyze any samples.
WDR355/010
WDR355/010
-------
Type of
Samples
Field Samples
Field Handling
Blank
Field Split
Total
Table 1-2
SUMMARY OF FIELD SAMPLES ANALYZED
BY LABORATORY AND SAMPLE TYPE
Laboratory
119 + 38
14 + 3
22 + 1
197
2
121
12
17
150
3
124
11
14
149
6 7
133 109
10 6
9
152
4
119
8 Total
137 781
11 67
6
154
73'
921
Laboratory 4 did not participate in the sample analysis.
Thirty-eight samples extracted by Laboratory 4 were sent to
Laboratory 1 for analysis.
-i
'Three of Laboratory 4's FHB were analyzed by Laboratory 1.
One field split from Laboratory 4 was analyzed by Laboratory 1.
^
"Valid results were obtained for both halves in 64 field split pairs.
WDR355/020
-------
Table 1-3
SUMMARY OF PERCENTAGE OF GOOD, UNCERTAIN, AND BAD FLAGS*
Analytical
Laboratory
% of Results
Designated
As Good
!b
2
3
6
7
8
TOTAL
84.6
83.7
84.3
94.4
94.3
72.4
85.7
% of Results
Designated
As Uncertain
2.4
4.1
12.6
4.7
3.4
18.3
7.6
% of Results
Designated
As Bad
12.9
12.2
3.1
0.1
2.3
9.2
6.6
Table represents counts of flags associated with the 781 field
samples used in the statistical comparisons.
"'include samples originally sent to and extracted by
Laboratory 4 but analyzed by Laboratory 1.
Type of QC
Samples
Table 1-4
SUMMARY OF LABORATORY QC SAMPLES ANALYZED
BY LABORATORY AND SAMPLE TYPE
a
Laboratory
1 2 3 6 7
Method/Holding 47b 19 34 28 20
Blanks
Blind QC Samples 43° 46 33 29
MS/MSD 20d 23 23 24
EPA Check 7 8 J/7 8
TOTAL 117 96 107 89
8
28
21
16
5
62
27
16
8
79
199
e
122
53
550
Laboratory 4 did not participate in the sample analysis.
Eight of Laboratory 4 ' s MHBs were analyzed by Laboratory 1 .
Six of Laboratory 4's BQCs were analyzed by Laboratory 1.
Two of Laboratory 4 ' s MS/MSDs were analyzed by Laboratory 1 .
£
Duplicate analyses were included where reported by the laboratory.
WDR355/021
-------
Table 1-5
QC DATA SUMMARY
Data Meeting QC Criteria
QA Measure
Method Holding
Blank
Surrogate Spike
Recovery
Matrix Spike
Recovery (Advisory
only)
Matrix Spike Duplicate
Precision (Advisory
only)
Initial Calibration
Standard
EPA Check Standard
Recovery
Performance Check
Standard
Continuing Calibration
Standard
Internal Standard
Area Variation
Compounds
All LCICs
1,4-DBB
2,4,6-TBBP and
1,2,3,4-TeBB
All LCICs
All LCICs
All LCICs plus
1,4-DBB, 2,4,6-
TBBP, & 1,2,4,5-
TeBB
All LCICs plus
1,4-DBB, 2,4,6-
TBBP, & 1,2,4,5-
TeBB
All LCICs plus
~Pyrene» 1-CNP
1,2,4,5-TeBB
All LCICs plus
1,4-DBB, 2,4,6-
TBBP, 1,2,4,5-
TeBB
d.-l,4-DCB
dg-Naphthalene
d-Tg-acenaphthene
d,Q-phenanthrene
d1Q-pyrene
Laboratory
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
Goal
90
90
90
90
100
100
100
100
90
Achieved
99.7
99.1
99.1
98.7
100
92.4
96.1
98.7
97.4
98.9
95.5
97.3
65.6
84.7
80.6
88.0
80.4
83.9
96.3
94.3
83.3
95.8
89.3
87.5
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
99.5
100
100
100
99.7
99.9
98.4
99.3
99.0
94.8
98.5
-------
Table 1-5
(continued)
Data Meeting QC Criteria
Goal Achieved
QA Measure Compounds Laboratory (%) (%)
BOC Recovery All LCICs 1 90 93.0
2 90.8
3 92.1
6 94.8
7 95.8
8 90.3
Note: In this table all results for both field and QC samples of all data usabilities
(G, U, and B) have been included.
WDR355/023
-------
Table 1-6
SUMMARY STATISTICS OF SURROGATE ANALYSIS RESULTS
(HS SAMPLES ONLY)
Analyte
1,4-DBB
Lab
1
2
3
4
6
7
8
Count
121
121
128
40
137
113
139
Mean
% Recovery
67.7
80.8
65.8
66.2
70.9
76.3
81.8
Std. Dev.
% Recovery
7.3
10.6
9.5
13.1
9.2
10.9
11.9
% RSD
10.7
13.1
14.4
19.8
12.9
14.3
14.5
Minimum
% Recovery
49.8
51.5
50.2
33.5
42.2
47.4
36.0
Maximum
% Recovery
83.1
104
111
85.0
90.4
102
103
TOTAL
799*
73.5
Median 73.0
11.9
16.2
2,4,6-TBBP
TOTAL
799*
89.1
Median 88.1
17.4
19.5
1245-TeBB
TOTAL 799* 102
Median 79.6
384
376
*Results from 18 duplicate samples' analyses were included.
33.5
32.4
0.0
111
1
2
3
4
6
7
8
121
121
128
40
137
113
139
74.5
94.4
94.8
73.0
85.8
102
89.8
10.9
15.8
18.4
15.1
11.5
16.5
15.6
14.6
16.7
19.4
20.6
13.4
16.2
17.4
56.1
59.3
57.5
32.4
58.5
66.4
38.8
134
148
175
97.6
129
168
154
175
1
2
3
4
6
7
8
121
121
128
40
137
113
139
214
86.9
88.3
66.9
74.9
82.4
83.3
982
11.6
22.3
19.6
14.3
17.5
16.7
458
13.3
25.2
29.3
19.1
21.3
20.1
28.1
53.2
0.0
16.0
0.0
0.0
44.7
10025
114
211
127
126
113
144
10025
WDR355/024
-------
Table 1-7
SUMMARY STATISTICS OF SURROGATE AKALYSIS RESULTS
(Laboratory QC Samples Only)
Analyte Lab
1,4-DBB 1
2
3
6
7
8
Count
110
88
90
81
57
71
Mean
% Recovery
75.8
87.1
74.1
78.2
83.3
85.4
Std. Dev.
% Recovery
8.6
9.1
10.3
7.5
15.2
12.6
% RSD
11.3
10.4
13.9
9.7
18.2
14.7
Minimum
% Recovery
34.9
68.7
51.0
57.5
46.1
58.8
Maximum
% Recovery
90.1
111.0
110.0
96.3
108.0
122.0
TOTAL
497
80.2
11.5
14.4
34.9
122.0
2,4,6-TBBP
1
2
3
6
7
8
110
88
90
81
57
71
77.4
96.8
95.1
87.3
104.0
91.9
10.9
15.0
14.1
12.2
18.3
15.0
14.2
15.5
14.8
14.0
17.5
16.4
35.5
66.2
61.3
59.9
57.8
58.8
101.0
139.0
151.0
121.0
138.0
121.0
TOTAL
497
90.8
16.4
18.0
35.5
151.0
1,2,4,5-TeBB
1
2
3
6
7
8
110
88
90
81
57
71
70.8
86.3
94.4
81.2
90.0
87.0
10.1
12.4
31.6
15.4
17.2
13.6
14.3
14.3
33.5
19.0
19.1
15.6
28.1
57.9
66.0
58.2
57.6
56.1
94.4
114.0
325.0
188.0
113.0
111.0
TOTAL
497
84.1
19.8
23.5
28.1
325.0
Table includes surrogate results for Method Blanks, Matrix Spike, and Matric Spike
Duplicate analyses and EPA Blind QC analyses.
WDR355/025
-------
Table 1-8
SUMMARY STATISTICS OF MATRIX SPIKE AND MATRIX SPIKE DUPLICATE RECOVERY RESULTS
Mean
Compound
1,2-DCB
1,2,4-TCB
1,2,3,4-TeCB
2-CNP
A-BHC
D-BHC
B-BHC
G-BHC
Laba
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
I-
Count0
10
11
8
12
7
7
10
11
8
12
7
7
10
11
8
12
7
7
10
11
8
12
7
7
10
11
8
12
7
7
10
11
8
12
7
7
10
11
8
12
7
7
10
11
8
12
7
7
(% Rec)
MS
53.7
79.2
67.2
69.4
73.1
68.4
59.1
80.1
77.0
73.6
70.8
74.2
64.0
82.2
83.7
77.1
75.8
77.4
55.2
55.6
50.0
37.4
37.4
47.4
63.1
81.5
77.8
79.9
76.9
89.0
59.4
67.3
57.2
81.8
72.9
70.7
57.6
67.3
63.2
77.3
78.3
81.1
63.4
83.2
94.2
85.8
85.0
90.1
MSD
53.2
79.9
64.5
67.1
70.9
62.3
61.6
79.9
69.1
71.3
73.7
69.8
67.5
82.3
77.0
74.4
79.8
75.0
53.9
54.3
44.1
40.4
33.0
47.2
63.7
79.1
74.9
79.1
74.8
82.6
62.1
69.8
59.8
82.3
70.3
64.3
61.1
67.3
62.8
79.3
75.4
77.7
64.8
80.3
92.7
85.8
87.3
82.9
% RSD
(% Rec)
MS
11.8
11.6
25.5
13.8
5.9
19.3
12.4
13.3
31.6
12.8
16.9
13.4
15.3
11.4
16.1
11.0
13.0
15.2
16.2
39.6
34.4
53.3
60.9
32.3
17.5
15.9
52.0
14.1
12.3
19.5
18.4
24.0
90.5
17.7
15.1
28.6
16.4
24.8
94.7
26.0
15.3
28.6
17.7
15.6
19.6
18.6
12.8
19.8
MSD
11.7
12.7
33.2
14.5
14.7
20.2
11.6
10.0
39.5
14.1
19.5
14.2
11.8
11.8
36.5
11.0
14.0
19.7
15.5
24.0
40.4
47.2
66.3
28.0
14.8
14.6
60.5
13.7
14.9
18.5
19.7
30.7
93.7
15.5
17.9
28.9
15.4
31.8
107.0
25.2
15.1
23.1
15.7
18.7
31.9
16.5
20.5
16.6
Maximum
(% Rec)
MS
63.9
99.6
96.6
88.5
79.5
80.0
73.1
99.6
103.5
88.4
90.1
85.9
82.9
101.8
105.8
87.6
89.3
91.1
66.8
93.1
67.8
62.4
80.3
60.1
83.3
107.7
108.8
100.0
89.5
108.2
74.2
88.4
94.0
115.7
84.0
100.4
74.5
91.5
104.8
106.6
91.3
111.3
87.0
106.6
112.6
123.9
99.3
115.2
MSD
62.5
94.0
96.5
79.1
82.1
84.3
73.2
95.8
92.8
88.5
93.1
86.1
83.1
104.2
99.3
87.8
89.8
98.7
69.0
69.3
65.2
65.3
74.0
72.5
81.7
95.9
113.1
97.8
87.6
103.3
82.4
105.1
107.0
104.8
83.7
90.9
78.4
99.2
111.4
99.2
87.8
104.7
87.3
102.7
119.7
112.3
114.7
101.1
Minimum
(% Rec)
MS
44.4
65.2
44.9
49.6
64.9
46.4
52.4
61.6
30.2
52.7
57.2
55.7
49.4
70.5
61.7
58.6
63.4
56.2
44.1
8.5
12.1
2.0
7.6
22.3
45.9
62.0
-14.3
56.5
64.6
58.7
36.4
39.3
-61.1
66.7
57.9
4U.1
42.9
33.4
-76.7
31.1
60.1
40.6
49.2
65.8
71.7
65.5
72.4
64.4
MSD
44.2
64.1
23.2
47.5
51.2
50.8
51.7
69.5
15.4
53.4
50.8
56.3
55.4
73.4
11.5
61.9
59.0
59.7
43.8
26.5
12.0
1.5
11.0
33.4
52.0
61.3
-15.6
58.5
61.7
65.1
39.2
36.4
-61.4
64.7
47.0
44.2
47.7
26.7
-87.6
35.1
56.3
60.2
54.7
59.4
36.6
68.6
65.5
64.3
aLaboratory 1 performed one MS/MSD set of analyses on one field sample originally sent to
Laboratory 4.
bCount of MS/MSD pairs for which all these sample results (native sample, MS, and
MSD) were reported.
WDR355/026
-------
Table 1-9
SUMMARY OF EPA BLIND QC SAMPLE RESULTS
1,2-DCB
1,2,4-TCB
1,2,3,4-TeCB
2-CNP
A-BHC
D-BHC
B-BHC
G-BHC
Mean
Count
Lab
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
Sand
7
8
10
7
3
6
7
8
10
7
3
6
7
8
10
7
3
6
7
8
10
7
3
6
7
8
10
7
3
6
7
8
10
7
3
6
7
8
10
7
3
6
7
8
10
7
3
6
Soil
36
38
23
22
18
21
36
38
23
22
18
21
36
38
23
22
18
21
36
38
23
22
18
21
36
38
23
22
18
21
36
38
23
22
IB
21
36
38
23
22
18
21
36
38
23
22
18
21
(% Rec)
Sand
61.4
62.2
68.4
78.8
93.9
107
47.8
46.0
69.2
75.8
82.9
76.7
44.0
36.1
61.9
72.0
87.0
71.1
54.1
34.1
55.8
71.9
80.4
70.2
52.1
34.7
63.7
69.5
85.6
66.6
49.1
29.2
56.4
70.1
74.5
51.3
56.4
41.2
71.7
72.9
96.4
73.5
51.8
41.0
61.9
71.5
96.5
69.5
Soil
88.3
85.3
82.1
87.6
83.8
119
72.5
76.5
101.3
89.0
76.3
81.4
61.0
64.8
90.5
82.3
69.4
67.2
53.6
49.7
46.4
53.2
51.0
43.5
60.2
61.2
80.9
77.9
60.4
68.9
47.4
49.9
76.0
76.3
53.6
66.8
53.4
57.8
73.4
52.2
54.2
48.6
57.0
60.7
78.5
80.0
65.6
71.3
% RSD
(% Rec)
Sand
39.6
16.8
23.3
24.3
4.4
32.4
32.0
19.5
24.0
19.8
3.1
23.2
29.6
25.8
23.3
17.3
4.1
26.4
33.2
12.7
21.5
22.4
7.6
31.6
32.1
22.3
25.3
17.6
13.7
23.6
37.0
12.8
25.1
18.4
3.7
46.7
32.5
30.7
26.1
16.7
3.6
29.5
30.5
20.0
46.1
13.8
3.1
25.6
Soil
47.0
38.7
44.8
37.6
34.2
58.0
21.6
17.8
54.6
19.0
20.6
23.6
14.9
14.9
58.8
12.4
15.2
17.6
28.6
18.0
40.1
26.4
26.2
41.9
14.1
16.5
41.1
14.0
18.4
32.6
26.7
23.6
53.9
28.7
31.9
23.2
24.1
23.5
41.5
65.8
57.8
82.6
17.6
18.8
56.9
27.4
16.6
32.2
Maximum
(% Rec)
Sand
103
78.3
100
110
98.3
160
68.1
59.3
84.4
102
84.4
107
68.0
50.7
76
88
90
102
82.5
42.5
72.5
97.5
87.5
110
76.2
46.4
83.6
87.1
97.1
85.7
77.8
34.4
83.3
86.7
77.2
96.7
84.9
56
94.7
88
100
111
75.4
52.3
97.7
81.5
99.2
95.4
Soil
230
230
218
187
155
380
120
110
330
127
115
145
85
92
329
105
90
92
95
70
110
87.5
76
75
77.1
86
225
113
88.6
100
68.3
81.7
199
122
81.7
103
72.0
94
181
100
93.3
97.8
83.1
98
245
119
83.8
102
Minimum
(% Rec)
Sand
24.2
50.0
44.4
46.7
90
64.2
26.7
35.6
38.5
53.3
80
83.3
26.0
24.0
39.3
50.5
83
53.5
31.7
30.0
35.8
46.9
76.3
46.9
37.1
22.9
41.4
48.6
73.6
48.9
32.8
25
36.7
47.5
71.7
34.4
39.6
25.3
44.9
51.7
93
49.7
35.4
29.2
0.0
53.8
93.3
49.2
Soil
31.5
55.8
25.5
52.8
49
56.7
48.5
52.2
54.4
62.2
50
53.3
43.5
46
54
64.7
53.5
46
37.5
30
0
17.5
32.5
11.7
42.0
40.7
58.5
60
48.9
0.0
0.0
30
0.0
0.0
0.0
38
0.0
34
0.0
0.0
0.0
0.0
37.5
36.2
0.0
0.0
49.2
0.0
WDR357/003
-------
Table 1-10
SUMMARY OF LABORATORY METHOD HOLDING BLANK RESULTS3
Compound
1,2-DCB
1,2, 4-TCB
1,2,3,4-TeCB
2-CNP
A-BHC
D-BHC
B-BHC
Lab
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
b
Count
47
19
34
28
20
28
47
19
34
28
20
28
47
19
5
28
3
28
47
19
32
28
20
28
47
19
32
10
20
28
47
19
30
4
8
28
47
19
32
25
19
28
No. of Count
Exceeded
Control Limits
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
2
1
3
0
6
Mean
Cone.
(ppb)
0.26
0.22
0.15
0.20
0.17
0.49
0.048
0.074
0.041
0.053
0.052
0.065
0.005
0.010
0.004
0.001
0.086
0.005
0.067
0.053
0.051
0.056
0.063
0.070
0.019
0.020
0.011
0.055
0.029
0.038
0.010
0.034
0.021
0.023
0.011
0.040
0.13
0.036
0.073
0.39
0.25
0.43
Maximum
Cone.
(ppb)
0.48
0.38
0.26
0.38
0.45
0.73
0.19
0.21
0.086
0.092
0.14
0.097
0.025
0.022
0.005
0.005
0.15
0.011
0.14
0.076
0.47
0.076
0.12
0.17
0.10
0.066
0.096
0.080
0.10
0.015
0.093
0.070
0.11
0.040
0.030
0.14
4.96
0.28
0.51
2.57
0.45
3.54
Minimum
Cone.
(ppb)
0.05
0.11
0.08
0.084
0.085
0.30
0.019
0.040
0.021
0.024
0.019
0.038
0.001
0.002
0.001
0.001
0.011
0.001
0.023
0.027
0.017
0.044
0.020
0.022
0.001
0.006
0.001
0.019
0.003
0.005
0.001
0.005
0.001
0.016
0.003
0.007
0.001
0.007
0.001
0.085
0.114
0.014
-------
Compound
G-BHC
Lab
1
2
3
6
7
8
b
Count
47
19
32
0
11
28
Table 1-10
(continued)
No. of Count
Exceeded
Control Limits
0
0
2
0
0
7
Mean
Cone.
(ppb)
0.022
0.021
0.071
0.033
0.62
Maximum
Cone.
(ppb)
0.16
0.039
0.65
0.16
2.61
Minimum
Cone.
(ppb)
0.001
0.008
0.001
0.004
0.005
Table includes results for primary ion concentrations only.
Only detects are included in the counts and statistics.
WDR357/072
-------
Table I-11
SUMMARY RESULTS OF FIELD HANDLING BLANK AND
LABORATORY METHOD HOLDING BLANK RESULTS FROM
THE SAME EXTRACTON BATCH
Compound
1,2-DCB
1,2, 4-TCB
1,2,3,4-TeCB
2-CNP
a-BHC
b-BHC
g-BHC
Number
Detected
Lab
1
2
3
4
6
7
8
1
2
3
4
6
7
8
1
2
3
6
7
8
1
2
3
4
6
7
8
1
2
7
1
3
2
7
Count
14
12
11
3
10
7
11
14
12
11
3
10
7
11
14
12
11
10
7
11
14
12
11
3
10
7
11
17
12
7
17
11
12
7
MHB
14
12
10
3
10
7
8
14
10
10
3
10
7
8
0
1
0
0
1
0
14
9
4
3
10
5
5
0
0
1
0
1
1
1
FHB
14
12
11
3
10
7
11
14
11
10
3
9
6
11
4
8
3
1
2
2
14
6
3
3
9
6
6
2
2
0
1
0
0
0
Mean
Cone (ppb)
MHB
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
_
0.
-
-
0.
-
0.
0.
0.
0.
0.
0.
0.
_
-
0.
_
0.
0.
0.
30
23
13
10
20
16
52
05
07
04
03
05
04
06
-
01
-
-
15
-
08
05
03
03
05
06
09
_
-
10
_
36
03
16
FHB
0.41
0.39
0.25
0.28
0.26
0.28
0.76
0.10
0.11
0.12
0.09
0.09
0.13
0.15
0.09
0.03
0.02
0.01
0.04
0.01
0.09
0.08
0.05
0.09
0.04
0.11
0.10
0.06
0.01
—
2.16
—
—
Maximum
Cone (ppb)
MHB
0.46
0.38
0.20
0.15
0.37
0.25
0.73
0.05
0.20
0.09
0.04
0.06
0.09
0.08
—
0.01
—
—
0.15
—
0.10
0.07
0.04
0.05
0.08
0.10
0.10
__
—
0.10
__
0.36
0.03
0.16
FHB
0.55
0.62
0.30
0.29
0.42
0.43
1.12
0.39
0.18
0.19
0.11
0.16
0.16
0.31
0.31
0.12
0.02
0.01
0.05
0.01
0.14
0.09
0.08
0.11
0.06
0.15
0.14
0.12
0.02
—
2.16
—
—
Minimum
Cone (ppb)
MHB
0.21
0.11
0.10
0.05
0.13
0.09
0.30
0.04
0.04
0.02
0.02
0.04
0.02
0.04
—
0.01
—
—
0.15
—
0.05
0.03
0.02
0.02
0.04
0.04
0.07
__
—
0.10
__
0.36
0.03
0.16
FHB
0.26
0.22
0.21
0.27
0.17
0.13
0.59
0.07
0.08
0.06
0.08
0.06
0.07
0.09
0.01
0.01
0.01
0.01
0.04
0.01
0.06
0.07
0.02
0.07
0.03
0.03
0.07
0.01
0.01
—
2.16
—
—
Delta-BHC had no detects.
WDR357/055
-------
Table 1-12
SUMMARY OF MATRIX SPIKE AND MATRIX SPIKE DUPLICATE
PRECISION RESULTS
Compound
1,2-DCB
1,2,4-TCB
1,2,3,4-TeCB
2-CNP
A-BHC
D-BHC
Laba
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
1
2
3
6
7
8
Count
10
11
8
12
7
7
10
11
8
12
7
7
10
11
8
12
7
7
10
11
8
12
7
7
10
11
8
12
7
7
10
11
8
12
7
7
Mean
% RPD
4.57
7.91
14.8
5.63
12.1
11.8
7.65
7.48
22.9
6.96
11.5
6.99
8.62
8.01
22.1
8.46
8.36
8.47
6.03
28.5
49.6
23.4
24.4
24.2
4.79
8.74
18.4
3.99
4.82
11.0
7.76
15.7
23.1
7.35
11.8
19.2
Std.
Dev.
% RPD
4.22
5.65
23.3
7.74
12.2
9.88
9.58
6.49
37.8
6.59
15.9
8.08
12.9
6.42
49.3
9.27
11.1
6.73
6.09
39.0
47.8
26.5
29.2
18.8
3.52
5.61
25.7
2.40
4.68
7.76
7.33
15.5
20.8
4.76
9.69
11.7
Max.
% RPD
12.5
17.6
70.6
27.6
36.8
29.9
32.5
18.4
114
24.7
46.2
21.8
37.2
21.7
144
32.8
32.0
23.4
18.2
130
122
80.8
81.9
56.9
12.4
23.2
80.6
7.88
12.6
24.8
25.4
57.8
64.2
17.8
29.2
39.7
Min.
% RPD
0.12
1.34
0.16
0.58
1.78
0.36
1.03
0.04
1.59
1.94
0.47
0.24
0.18
0.66
1.42
0.86
1.42
3.40
1.10
0.57
4.81
0.63
3.23
3.85
1.44
2.31
3.87
0.23
0.23
1.15
0.32
1.23
0.48
0.73
0.17
5.41
-------
Compound
B-BHC
G-BHC
Lab
1
2
3
6
7
8
1
2
3
6
7
8
Coui
10
8
11
12
7
7
10
11
8
12
7
7
Table 1-12
(continued)
Mean
% RPD
4.03
9.22
19.4
6.03
6.75
13.8
9.72
17.8
18.5
7.89
6.98
26.9
10.5
14.8
20.4
3.95
5.16
15.3
36.1
55.1
64.8
14.2
16.6
46.8
4.35
4.99
19.5
5.68
6.58
9.30
12.8
17.3
64.8
' 18.4
19.3
26.1
0.57
1.
2,
.66
.16
0.85
0.95
10.6
0.23
1.
5.
,34
.74
0.70
0.16
2.17
Laboratory 1 performed one MS/MSD set of analyses on one field sample
originally scheduled for Laboratory 4.
Count of MS/MSD pairs for which all three sample results (native sample,
MS, and MSD) were reported.
WDR357/008
-------
uo
120
110
% 100
R 90
0 80
V
E 70
R
Y 60
50
40
30
t
1 FS 1 OC 2 FS 2 OC 3 FS 3 OC 4 FS 4 OC 6 FS 6 OC 7 FS 7 OC 8 FS 8 OC
LABORATORY
Note: Analytical Results for Samples With Dilution Factor > 1 and for Instrument Blanks are not Included
Figure I -1
SOIL ASSESSMENT--
INDICATOR CHEMICALS
SURROGATE RECOVER-
IES FOR FIELD AND QC
SAMPLES
SURROGATE = DBB
140
130
120
no
R 10°
E
0 90
V
R 80
Y
70
60
50
40
3 4
LABORATORY
Figure I - 2
SOIL ASSESSMENT--
INDICATOR CHEMICALS.
MATRIX SPOKE RECOV-
ERY SUMMARY
LCIC = ALPHA - BHC
-------
120-
110
100 i
90
80-
<
R 70-
E
0 60
V
t M.
Y 40-
30-
20-
10-
0
LABORATORY
Figure I - 3
SOIL ASSESSMENT--
INDICATOR CHEMICALS
EPA BLIND QUALITY
CONTROL SAMPLE SUM-
MARY
LCIC = GAMMA - BHC
190-
180-
170-
160-
150-
140-
« 130-
R '20-
E 110-
C 100-
0 90-
V 80-
R 70"
Y 60-
50-
40
30-
20-
10-
0-
r
$&
1
1 MS 1 BOC 2 MS 2 80C 3 MS 3 80C 4 MS 4 BOC 6 MS
'LABORATORY AND MS OR BOC
6 BOC
7 MS
7 BOC
8 MS
8 BOC
Figure I - 4
SOIL ASSESSMENT--
INDICATOR CHEMICALS
COMPARISON OF MATRIX
SPIKE AND BQC RECOV-
ERIES
LCIC = BETA - BHC
-------
140
130
120
110
100
90
R 80
P 70
0 60
50
40
30
20
10
0
1
4
LABORATORY
Figure I - 58
SOIL ASSESSMENT--
INDICATOR CHEMICALS
MATRIX SPIKE RELATIVE
PERCENT DIFFERENCE
SUMMARY
LCIC = CNP
90
80
70
60
50
40
30
20
10
0
LABORATORY
Figure I - 5b
SOIL ASSESSMENT-
INDICATOR CHEMICALS
MATRIX SPIKE RELATIVE
PERCENT DIFFERENCE
SUMMARY
LCIC = ALPHA - BHC
-------
200
160
160
140
R 120
P 100
D 60
60
40
20
0
JE
w
D
c
B
r
[
^
h
H
f
b
0
B
r
w
T
c
B
•
"H
i
T1
b
1
C
B
~E
w
T
e
C
B
«
«
r
I
n
J
f
b
T
e
B
w
c
N
P
4
[
h
1
r
b
C
N
P
~t
w
A
B
H
C
d
D
1
T
b
A
B
H
C
,
i — i
—
T
w
D
B
H
C
b
D
B
H
C
SS
w
B
B
H
C
*
1
T
b
B
B
H
C
--e-
w
C
B
H
C
ICIC
Note: Pairs with a Non-Defect Result for Either Half Are
0.11
R 0.10
E
I
1 0.09
V
E
R 0.08
N 0.07
S
r
A °'06
0
R 0.05
0.04
1
1
1
•
J
l
1
i
2
Not Included
J
4
..
5
CALIBRATION
ft
6
,
1
7
V
|
B
9
NUMBER
•L
Figure 1 - 6
0/-MI AOC»l-00»«|-M-r
b INDICATOR CHEMICALS
c RELATIVE PERCENT
B DIFFERENCE IN FIELD
c SPLIT RESULTS WITHIN
AND BETWEEN LABORA-
TORIES
L
Figure 1 - 7
SOIL ASSESSMENT-
INDICATOR CHEMICALS
INITIAL CALIBRATION
10 ' SUMMARY FOR LABORA-
TORY 3
SURROGATE = TBBP
-------
0.00
0.00
0.39
0.38
0.37
0.36
0.35
0 0.34
N 0.33
R 0.32
* 0.31
1 0.30
0.29
0.28
0.27-1
0.26
0.25
1 PCI t PC2 2 PC1 2 PC2 3 PCI 3 PC2 4 PCI 4 PC2 6 PCI 6 PC2 7 PCI 7 PC2 8 PCI 8 PC2
LABORATORY AND PCI OR PC2
Figure I - 8
SOIL ASSESSMENT--
INDICATOR CHEMICALS
PERFORMANCE CHECK
ION RATIOS
LCIC = CNP
too-i
? 90
G
N 80
* 70^
0 60-1
N 50
0
^ 40-
* 30
R
A 20-|
T
0 '°
0
^T
I
I
I PCI I PC2 2 PCI 2 PC2 3 PCI 3 PC2 4 PCI 4 PC2 6 PCI 6 PC2 7 PC1 7 PCI 8 PCI 8 PC2
LABORATORY AND PCI OR PC2
Figure I - 9
SOIL ASSESSMENT--
INDICATOR CHEMICALS
PERFORMANCE CHECK
SENSITIVITY
ON-BETA-BHC 217 ION
-------
301
20
10
LABORATORY
Figure I • 10
SOIL ASSESSMENT--
INDICATOR CHEMICALS
PERFORMANCE CHECK
RESOLUTION
LCIC-CNP-PC1
30-
20-
10-
N -10
P
[-20-
N
1
-30-
-40
LABORAIORY
Figure I-11
SOIL ASSESSMENT--
INDICATOR CHEMICALS
CONTINUING CALIBRA-
TION RF SUMMARY
LCIC = BETA - BHC
-------
120-
110
90
D 60
1 70
F 60
E 50
t 40
N JO
C 20.
10
1 0
N
-10
« -20
-30
-40
-50
-60
1
\{
n
«
I
f
i
•*
CSt
\
4
r
*\
*
i* 1
! i
•0
S
T
i
5
1
•
m
1
~~ M
n
1 FS I OC 2 FS 2 OC 3 FS 3 OC 4 FS 4 OC 0 FS 6 OC
LAOORA10RY
7 FS 7 OC 8 FS 8 OC
Figure I -12
SOIL ASSESSMENT--
INDICATOR CHEMICALS
INTERNAL STANDARD
AREA PERCENT VARI-
ATION
IS4 = D10-PHENAN-
THRENE
-------
1900000
1800000
1700000
1600000
1500000
1400000
1300000
1200000
0 1100000
t>
* 1000000
o
OL
900000
800000
700000
600000
500000
400000
300000
200000
t
____»-• —
_— —
i
1
p
a
p a
_«-— — ~~
D
P
'
a
nn
B
p
I1
a
B
p
p
i
p
i
i
" " i
p
n
a a a
plBn
p
B||Q
PR"
p
B
p
p
p
p a
a
H o
°B • P
H B P P
u n H
0 PH
y D D D H B
00 Ha p p
H B
B
if"f rr"""
— FTii
k— -^^B
P D
P
p
P
B
p
p
P P
H P
a p
P D
n
p
p
p
p
p .
_____
— — rr a
^^^ 1 1 •••
P
P
_^_— —
^——-~~~~~~~^ P
a P
P P
p
« B
P
p H "
p
a n
B
10/26/87 11/05/87 11/15787 " 11/25/87 12/05/87 12/15/87 12/25/87 01/04/88
Dole of Anolysis
Note: Plot Includes all Data: Good, Uncertain, and Bad.
Figure H3 SOIL ASSESSMENT -- INDICATOR CHEMICALS
TREND ANALYSIS FOR LABORATORY 1
RAW PEAK AREA FOR D4 - DCB
-------
28
27
26
t>
E
o
**
c
ec.
23
22-1
I
• B
a • B
I
10/26/87 11/05/87 11/15/87
Note: Plot Includes all Data: Good, Uncertain, and Bad.
11/25/87 12/05/87
Dote of Anolysis
12/15/87
12/25/87
01/04/88
Figure 1-14 SOIL ASSESSMENT -- INDICATOR CHEMICALS
TREND ANALYSIS FOR LABORATORY 1
RAW RETENTION TIME D10 - PYRENE
-------
u
o
in
c
o
CC.
2.5
2.4
2.3
2.2
2.1
2.0
1.8
i o
a P
1.4
1.3
1.2
1.1
1.0
10/16/87 10/26/87 11/05/87 11/15/87 11/25/87
Note: Plot Includes all Data: Good, Uncertain, and Bad.
12/05/87 12/15/87 12/25/87 01/04/88 01/14/88 01/24/88 02/03/88
Dote o( Anolysis
Figure MS SOIL ASSESSMENT -- INDICATOR CHEMICALS
TREND ANALYSIS FOR LABORATORY 6
-------
150
140
130
120
110
100
90
I 80
t>
Q.
o
60
50
40
30
20
10
p
a
10/16/87 10/26/87 11/05/87 11/15/87 11/25/87 12/05/87 12/15/87 12/25/87 01/04/88
Note: Plot Includes all Data: Good, Uncertain, and Bad. Dote ol Ano lysis
Figure 1-16 SOIL ASSESSMENT -- INDICATOR CHEMICALS
TREND ANALYSIS FOR LABORATORY 8
SURROGATE NO. 3
-------
APPENDIX J
Sampling Area Comparisons and Additional
Sensitivity Analyses
Soil Assessment—Indicator Chemicals
-------
Appendix J
SAMPLING AREA COMPARISONS AND
ADDITIONAL SENSITIVITY ANALYSES
SOIL ASSESSMENT—INDICATOR CHEMICALS
In response to requests by Dr. Schoenfeld and Dr. Stoline
during peer review (June 20 and 21, 1988) and by the TRC,
additional comparisons were performed to provide context for
the statistical comparisons of the soil LCIC assessment and
to examine the sensitivity of the statistical comparisons to
the particular data used.
APPROACH
The tables that are presented in this appendix compare all
"EDA sampling areas and comparison areas. (These tables
incorporate Tables 6-8 and 6-9 in Volume III and show com-
parisons among EDA sampling areas.) Two original sensi-
tivity analyses were reported in Volume III. In addition,
three other sensitivity analyses are included in this
appendix, and an alternative approach to the comparison of
all LCICs is also included here.
The comparisons are presented in a format identical to that
in Volume III. The following analyses are presented:
o All EDA sampling areas and comparison areas
compared among themselves for each LCIC (uni-
variate test), based only on the observations
classified as Good (Table J-la through c)
o All EDA sampling areas and comparison areas
compared among themselves for all LCICs considered
together (multivariate test), based only on the
observations classified as Good (Table J-2a and b)
o A sensitivity analysis of the results presented in
Table J-la, based on classifying all observations
less than or equal to 2.0 ppb as non-detect
(Table J-3)
o A sensitivity analysis of the results presented in
Table J-la, based on trimming the upper 10 percent
of the data set (Table J-4a through c)
o Percent increase in number of detects, by LCIC, in
the Good data due to reclassifying non-detects
with an ion ratio between 20 and 40 percent as
detects (Table J-5)
J-l
-------
o A sensitivity analysis of the results presented in
Tables J-la through c, based on a data set
modified by relaxing the ion ratio criterion for
identification (detection) (Table J-6a through c)
o A sensitivity analysis of the results presented in
Tables J-2a and b, based on the data set modified
by relaxing the ion ratio criterion for
identification (detection) (Table J-7a and b)
o Comparisons between the EDA sampling areas and the
comparison areas for each LCIC (univariate test),
with the Bonferroni correction for number of LCICs
(Figure J-l)
o Comparisons between each comparison area and the
other two comparison areas for LCIC (univariate
test), with the Bonferroni correction for number
of LCICs (Figure J-2)
DISCUSSION OF RESULTS
COMPARISONS AMONG EDA SAMPLING AREAS
Tables J-l (a through c) and 2 (a and b) supplement
Tables 6-8 and 6-9 of Volume 111 by including comparisons
among EDA sampling areas. Table 6-8 in Volume III lists the
results of comparing each of the seven EDA sampling areas
and the three comparison areas with the three comparison
areas for each LCIC (univariate test), based only on the
observations classified as Good. Table 6-9 of Volume III
shows the analogous results based on the multivariate test.
Neither table shows comparisons between EDA sampling areas.
Following the request of the TRC, Tables J-l and 2 provide
the results of comparisons of EDA sampling area to EDA
sampling area.
TRIMMING DATA BELOW 2.0 PPB
In addition to the two sensitivity analyses presented in
Volume III, a new sensitivity analysis is also presented
here. At the request of a peer reviewer (Dr. David
Schoenfeld—Peer Review Meeting Comment, June 21, 1988), the
additional sensitivity analysis was performed in which all
data below 2.0 ppb were classified as non-detects.
Table J-3 shows the results of this analysis.
The purpose of this sensitivity analysis is similar to that
of the sensitivity analysis reported in Volume III in which
all concentrations less than 1.0 ppb were treated as non-
detects. The purpose, as stated in Volume III (p. 5-10), is
to assess the sensitivity of the comparisons to the
J-2
-------
laboratory detection limits. "If the results had changed
dramatically with the [artificially] increased detection
limits, then it could be concluded that the detection limits
were exerting a strong influence on the results" (Volume
III, p. 5-10).
Since most observations are less than 1.0 ppb for all of the
LCICs except tri- and tetrachlorobenzene, the 1.0-ppb cutoff
was used in the original sensitivity analysis. Since for
the LCICs TCB and TeCB, most observations are less than
2.0 ppb, Dr. Schoenfeld requested that the analysis treating
data below 2.0 ppb as non-detects be performed.
These sensitivity analyses which treat near-detection limit
values as non-detect address another issue as well. Because
most observations are less than 1.0 or 2.0 ppb for most
LCICs, the effect of treating all observations less than or
equal to 1.0 or 2.0 ppb as non-detect is to emphasize the
characteristic of the Wilcoxon rank sum test, which tests
for differences in the upper tails of the distributions.
That is, this sensitivity analysis can be used to test
whether there are more outlying high values in one EDA
sampling or comparison area than in another.
TRIMMING THE UPPER 10 PERCENT OF THE DATA
The "upper 10 percent trimmed" data set used to produce
Table J-4 (a through c) consists of the original data set
modified by deleting the largest 10 percent of the
observations in each comparison and EDA sampling area.
The purpose of deleting the upper 10 percent of the data in
each area is to explore the effects of treating the largest
10 percent of the observations in each EDA sampling area and
comparison area as outliers. The question this sensitivity
analysis proposes to answer is whether a few large values in
any particular EDA sampling or comparison area will have a
great influence on the outcome of the statistical compari-
sons. It should be noted that it is not possible to perform
a meaningful multivariate analysis based on an upper
10 percent trim of the multivariate observations, because an
upper 10 percent trim is not well-defined for a multivariate
observation. For any particular EDA sampling or comparison
area, the samples with the largest 10 percent of the obser-
vations for one LCIC may not correspond to the largest
10 percent of the observations for another LCIC.
RELAXING ION RATIO CRITERION
The data set used to produce Tables J-5, 6 (a through c),
and 7 (a and b) include the original data set, and it has
been modified in the following way: non-detect observations
classified as non-detect solely because the observed ratio
J-3
-------
of primary to secondary ions deviated from the theoretical
ratio by more than 20 percent but less than 40 percent were
reclassified as detect, and the primary ion concentration
has been used to estimate the concentration of the LCIC.
(Note that three other criteria were used to establish
whether an observation was recorded as detect or non-detect.
These were (1) all three ions present, (2) the peaks for the
three ions all within two scans of each other, and (3) the
peaks for the three ions all within the + 0.005 relative
retention time window. If an observation failed any of
these three criteria, it remained classified as a non-detect
in the modified data set.) Table J-5 shows the percent
increase in number of detects, by LCIC, for the Good data
based on the reclassification of non-detects using the
criteria described above. The purpose of generating
Tables J-6 (a through c) and 7 (a and b) is to explore the
effects of LCIC interferences on the statistical
comparisons.
BONFERRONI CORRECTION
At the request of another peer reviewer (Dr. Michael
Stoline—Written Preliminary Comments, p. 10), an alterna-
tive version of Figure 6-9 of Volume III is presented in
Figure J-l, in which the results of the statistical com-
parisons are modified using a Bonferroni correction (Miller,
1966) over the eight LCICs. The Bonferroni method is an
alternative to the multivariate approach (Figure 6-10 of
Volume III) to the multiple comparisons problem (see
Appendix A of Volume III).
For the original comparisons, shown in Figure 6-9, an EDA
sampling area has been determined to be statistically
significantly different from a comparison area (indicated by
an up or down arrow) if the two-sided p-value of the
statistic associated with that comparison (p-values are
given in Table K-l) was less than or equal to 0.05. The
direction of the arrow is determined by the sign of the
statistic. The Bonferroni approach guarantees that the
overall alpha-level for the eight tests associated with
comparing a particular EDA sampling area to a particular
comparison area (one test for each LCIC) is no more than
0.05. That is, the probability of at least one false
positive (EDA concentration incorrectly assumed to be larger
than comparison area concentration) or false negative (EDA
concentration incorrectly assumed to be less than comparison
area concentration) among the eight tests is no more than
5 percent.
For the comparisons shown in Figure J-l of this appendix, an
EDA sampling area is determined to be statistically signifi-
cantly different from a comparison area if the two-sided
J-4
-------
p-value of the statistic associated with that comparison is
less than or equal to (0.05/8) = 0.00625.
CENTRAL ANALYSIS
Although the additional comparisons have been made, it
should be emphasized that the peer review committee has
recommended that the results of the original comparisons
(given in Tables 6-8, 6-9, 6-10, K-l, and K-2, and
Figures 6-9, 6-10, and 6-11 of Volume III) should be the
primary basis for the habitability decision (Dr. David
Schoenfeld—Peer Review Meeting Comment, June 21, 1988).
WDR356/008
J-5
-------
REFERENCE
Miller, Rupert G. 1966. Simultaneous Statistical
Inference. McGraw-Hill Book Co., New York.
WDR356/008
J-6
-------
LCIC
DCB
TCB
1
2
3
4
5
6
7
221
225
C&T
Table J-la
RESULTS FOR NONPARAMETRIC UNIVARIATE SAMPLING/ h „ A
COMPARISON AREA TO SAMPLING/COMPARISON AREA COMPARISONS3'D/c'Q'e
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
Median
Cone.
(ppb) 1
1.01
1.01
0.39 -n-
0.40 ++
0.36 ++
0.39 ++
0.36 ++
0.41 ++
0.39 ++
0.43 ++
0.36 ++
2
0.39
—
o
++
o
o
o
o
0
++
EDA
3
0.40
—
o
++
o
•f
o
o
o
++
Sampling
4
0.36
—
—
—
-
o
—
—
-
o
Areas
5
0.39
—
o
o
+
+
o
o
o
o
Comparison Areas
6
0.36
—
o
-
o
-
-
—
o
o
7
0.41
—
o
o
++
o
+
o
o
+
221
0.39
—
o
o
++
o
•M-
O
o
++
225
0.43
—
o
o
+
o
o
0
o
0
C&T
0.36
—
—
—
o
o
o
-
—
o
8.67
0.91
0.89
0.43
0.52
0.37
0.58
0.65
0.60
0.13
8.67
0.91
0.89
o
0.43
o
o
o
0.52
o
o
o
0.37
0.58
o
o
o
o
•M-
0.65
o
o
o
0.60
o
o
o
o
0.13
11.48 1.17
TeCB
1
2
3
4
5
6
7
221
225
C&T
11.48
1.17
1.06
0.38
0.43
0.31
0.54
0.53
0.61
0.05
1.06
o
0.38
o
o
o
0.43
0.31
o
o
o
0.54
o
o
o
o
0.53
o
•n-
o
0.61
0.05
o
o
o
o
WDR357/d.l
-------
a-BHC
d-BHC
b-BHC
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
Median
Cone.
(ppb)
ND
0.04
0.04
0.04
0.06
0.05
0.06
0.06
0.07
0.03
8.25
0.40
0.22
0.12
0.13
0.07
0.14
0.18
0.11
ND
1.13
0.04
ND
ND
ND
ND
ND
ND
ND
ND
11.58
0.20
0.13
ND
ND
ND
ND
ND
ND
ND
Table J-la
(continued)
EDA Sampling
1
ND
o
0
0
o
0
0
o
-
o
8.25
4-4-
4-4-
4-4-
4-4-
4-4-
4-+
4-4-
4-4-
4-4-
1.13
4-4-
+4-
4-4-
++
4-4-
+4-
4-4-
4-4-
4-4-
11.58
4-4-
4-4-
4-4-
++
4-4-
4-4-
4-4-
4-4-
4-4-
2
0.04
o
0
0
o
o
0
o
-
o
0.40
4-
4-4-
4-+
++
4-4-
4- +
4-4-
4-4-
0.04
+
4-4-
4-4-
4-4-
4-4-
4-4-
4-4-
4-4-
0.20
O
O
4-4-
4-4-
4-4-
4-4-
O
4-4-
3
0.04
0
o
o
o
o
-
o
—
o
0.22
-
o
4-
4-4-
4>
0
O
4-4-
ND
_
o
o
4-4-
+4-
4-+
0
4-4-
0.13
O
O
4-
4-4-
O
4-4-
O
4-4-
4
0.04
O
O
O
o
o
o
o
o
o
0.12
—
o
o
o
o
0
o
4-4-
ND
-_
o
0
o
o
4-
0
4-4-
ND
o
o
o
4-4-
o
4-
o
+ 4-
Areas
5
0.06
o
o
0
0
o
o
0
—
o
0.13
—
-
o
o
o
o
o
4-4-
ND
—
o
o
4-
0
4-
O
4-4-
ND
—
-
o
o
o
o
0
4-4-
6
0.05
o
o
o
o
o
o
o
o
o
0.07
—
—
o
o
-
-
o
4-4-
ND
—
—
0
-
o
o
o
+4-
ND
—
—
—
o
—
o
—
4-4-
7
0.06
o
o
+
o
o
o
o
o
4-
0.14
—
-
o
o
4-
o
0
++
ND
—
—
0
o
o
0
o
4-4-
ND
—
o
o
o
4-4-
O
O
4.4-
Comparison Areas
221
0.06
o
o
o
o
o
o
o
o
o
0.18
o
o
o
4-
O
o
4-4-
225
ND
O
o
o
o
ND
o
o
o
0.11
o
o
o
o
o
o
4-4-
ND
o
o
o
o
o
o
4-4-
ND
o
o
o
o
4-4-
o
4-
4-4-
C&T
0.03
o
o
o
o
o
o
ND
ND
ND
WDR357/d.l
-------
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
CST
Median
Cone.
(ppb)
1.73
0.09
ND
ND
ND
ND
ND
ND
ND
ND
TOTALS
All Symbols
Table J-la
(continued)
EDA Sampling^
1 2
1.73 0.09
-n-
++ o
++ +
++ ++
++ ++
++ ++
++ ++
::
63 40
0 4
8 20
1 1
0 7
3
ND
o
o
+
++
++
+-f
o
26
8
27
3
8
4
ND
-
o
0
+
o
+
0
7
4
39
5
17
Areas
5
ND
—
-
o
o
o
o
o
8
4
41
5
14
6
ND
—
—
-
o
o
o
**
6
0
27
8
31
7
ND
—
—
o
o
o
o
0
10
5
39
1
17
Comparison Areas
221
ND
»
-
o
o
o
o
10
3
36
6
17
225
ND
o
o
o
+
o
o
-
10
7
41
2
12
C&T
ND
V V
--
—
--
—
--
0
0
12
3
57
72
72
72
72
72
72
72
72
72
72
a.++ = Column sampling/comparison area > Row sampling/comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling/comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling/comparison area at 0.01 significance level.
b. All test results reported are based on two-sided p-values.
c. The first entry in each column is median concentration; ND indicates non-detect.
d. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
e. Based on observations classified as Good.
WDR357/d.l
-------
LCIC
DCB
TCB
TeCB
CMP
a-BHC
Table J-lb
TWO-SIDED p-VALUES FOR NONPARAMETRIC
DNIVARIATE SAMPLING/COMPARISON AREA h
TO SAMPLING/COMPARISON AREA COMPARISONSa'D
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
EDA Sampling Areas
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.964
0.815
0.275
0.519
0.175
0.181
0.110
0.047
0.693
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2
0.000
1.000
0.989
0.004
0.413
0.052
0.564
0.805
0.539
0.008
0.000
1.000
0.245
0.000
0.001
0.000
0.000
0.001
0.010
0.000
0.000
1.000
0.069
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.964
1.000
0.964
0.657
0.403
0.128
0.107
0.055
0.013
0.914
0.000
1.000
0.011
0.003
0.000
0.000
0.000
0.000
0.001
0.000
3
0.000
0.989
1.000
0.005
0.682
0.041
0.740
0.980
0.583
0.002
0.000
0.245
1.000
0.000
0.041
0.000
0.002
0.014
0.032
0.000
0.000
0.069
1.000
0.000
0.000
0.000
0.000
0.000
0.005
0.000
0.815
0.964
1.000
0.155
0.593
0.207
0.022
0.059
0.005
0.814
0.000
0.011
1.000
0.260
0.045
0.000
0.041
0.069
0.207
0.000
4
0.000
0.004
0.005
1.000
0.029
0.349
0.003
0.003
0.025
0.915
0.000
0.000
0.000
1.000
0.140
0.219
0.089
0.011
0.077
0.000
0.000
0.000
0.000
1.000
0.404
0.127
0.055
0.040
0.054
0.000
0.275
0.657
0.155
1.000
0.599
0.789
0.246
0.597
0.107
0.103
0.000
0.003
0.260
1.000
0.623
0.069
0.890
0.630
0.410
0.000
5
0.000
0.413
0.682
0.029
1.000
0.044
0.639
0.617
0.779
0.058
0.000
0.001
0.041
0.140
1.000
0.002
0.964
0.268
0.828
0.000
0.000
0.000
0.000
0.404
1.000
0.006
0.228
0.446
0.250
0.000
0.519
0.403
0.593
0.599
1.000
0.305
0.166
0.311
0.018
0.605
0.000
0.000
0.045
0.617
1.000
0.164
0.516
0.270
0.942
0.000
6
0.000
0.052
0.041
0.349
0.044
1.000
0.035
0.009
0.312
0.146
0.000
0.000
0.000
0.219
0.002
1.000
0.000
0.000
0.005
0.000
0.000
0.000
0.000
0.127
0.006
1.000
0.000
0.000
0.000
0.000
0.175
0.128
0.207
0.789
0.305
1.000
0.368
0.383
0.322
0.193
0.000
0.000
0.000
0.069
0.164
1.000
0.029
0.028
0.192
0.000
7
0.000
0.564
0.740
0.003
0.639
0.035
1.000
0.992
0.556
0.022
0.000
0.000
0.002
0.089
0.964
0.000
1.000
0.123
0.808
0.000
0.000
0.000
0.000
0.055
0.228
0.000
1.000
0.446
0.982
0.000
0.181
0.107
0.022
0.246
0.166
0.368
1.000
0.965
0.762
0.018
0.000
0.000
0.041
0.890
0.516
0.029
1.000
0.509
0.577
0.000
Comparison
221
0.000
0.805
0.980
0.003
0.617
0.009
0.992
1.000
0.757
0.007
0.000
0.001
0.014
0.011
0.268
0.000
0.123
1.000
0.368
0.000
0.000
0.000
0.000
0.040
0.446
0.000
0.446
1.000
0.346
0.000
0.110
0.055
0.059
0.597
0.311
0.383
0.965
1.000
0.824
0.091
0.000
0.000
0.069
0.630
0.270
0.028
0.509
1.000
0.346
0.000
225
0.000
0.539
0.583
0.025
0.779
0.312
0.556
0.757
1.000
0.194
0.000
0.010
0.032
0.077
0.828
0.005
0.808
0.368
1.000
0.000
0.000
0.000
0.005
0.054
0.250
0.000
0.982
0.346
1.000
0.000
0.047
0.013
0.005
0.107
0.018
0.322
0.762
0.824
1.000
0.010
0.000
0.001
0.207
0.410
0.942
0.192
0.577
0.346
1.000
0.000
Areas
C&T
0.000
0.008
0.002
0.915
0.058
0.146
0.022
0.007
0.194
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.693
0.914
0.814
0.103
0.605
0.193
0.018
0.091
0.010
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
l.O'OO
WDR357/d.3/l
-------
Sampling/
Comparison
LCIC Area
d-BHC
b-BHC
g-BHC
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
CST
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2
0.000
1.000
0.018
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.348
0.188
0.001
0.000
0.001
0.000
0.105
0.000
0.000
1.000
0.338
0.030
0.001
0.000
0.000
0.000
0.014
0.000
Table J-lb
(continued)
EDA Sampling Areas
3
0.000
0.018
1.000
0.063
0.220
0.001
0.008
0.000
0.150
0.000
0.000
0.348
1.000
0.335
0.012
0.000
0.079
0.000
0.460
0.000
0.000
0.338
1.000
0.156
0.038
0.000
0.002
0.001
0.179
0.000
4
0.000
0.000
0.063
1.000
0.829
0.108
0.351
0.032
0.573
0.000
0.000
0.188
0.335
1.000
0.051
0.000
0.411
0.016
0.755
0.000
0.000
0.030
0.156
1.000
0.404
0.011
0.098
0.020
0.711
0.000
5
0.000
0.000
0.220
0.829
1.000
0.034
0.169
0.015
0.560
0.000
0.000
0.001
0.012
0.051
1.000
0.099
0.293
0.453
0.118
0.000
0.000
0.001
0.038
0.404
1.000
0.158
0.466
0.278
0.461
0.000
6
0.000
0.000
0.001
0.108
0.034
1.000
0.474
0.563
0.372
0.008
0.000
0.000
0.000
0.000
0.099
1.000
0.002
0.247
0.001
0.004
0.000
0.000
0.000
0.011
0.158
1.000
0.434
0.921
0.035
0.002
7
0.000
0.000
0.008
0.351
0.169
0.474
1.000
0.206
0.667
0.002
0.000
0.001
0.079
0.411
0.293
0.002
1.000
.0.072
0.632
0.000
0.000
0.000
0.002
0.098
0.466
0.434
1.000
0.457
0.140
0.000
Comparison
221
0.000
0.000
0.000
0.032
0.015
0.563
0.206
1.000
0.075
0.070
0.000
0.000
0.000
0.016
0.453
0.247
0.072
1.000
0.033
0.000
0.000
0.000
0.001
0.020
0.278
0.921
0.457
1.000
0.095
0.007
225
0.000
0.000
0.150
0.573
0.560
0.372
0.667
0.075
1.000
0.001
0.000
0.105
0.460
0.755
0.118
0.001
0.632
0.033
1.000
0.000
0.000
0.014
0.179
0.711
0.461
0.035
0.140
0.095
1.000
0.000
Areas
CST
0.000
0.000
0.000
0.000
0.000
0.008
0.002
0.070
0.001
1.000
0.000
0.000
0.000
0.000
0.000
0.004
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.002
0.000
0.007
0.000
1.000
a. 221 = Census Tract 221.
225 = Census Tract 225.
CST = Cheektowaga and Tonawanda.
b. Based on observations classified as Good.
WDR357/d.3/2
-------
Table J-lc
SUMMARY OF NONPARAMETRIC UNIVARIATE
SAMPLING/COMPARISON TO SAMPLING/COMPARISON AREA COMPARISONS
a,b,c,d
Sampling/
Comparison
Area
Totals Over All LCICs
++
++
++
+ +
++
++
++
EDA Sampling Areas
1
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
2
0
0
1
0
7
0
2
6
0
0
5
1
2
0
0
6
0
2
0
0
6
0
2
0
0
6
0
2
0
0
3
0
0
1
0
7
0
0
6
2
0
3
0
5
0
0
1
4
3
0
0
6
1
1
0
0
4
1
2
1
0
4
0
0
1
0
7
0
0
2
1
5
0
0
0
5
3
0
0
8
0
0
1
1
6
0
0
0
0
7
0
1
5
0
0
1
0
7
0
0
2
0
6
0
0
3
4
1
0
1
7
0
0
2
2
4
0
0
0
0
8
0
0
6
0
0
1
0
7
0
0
2
0
6
0
0
1
1
6
0
0
6
1
1
0
0
4
2
2
0
0
3
2
3
7
0
0
1
0
7
0
0
2
0
6
0
1
2
1
4
1
0
7
0
0
0
0
8
0
0
3
2
3
0
0
Comparison Areas
221
0
0
1
0
7
0
0
2
0
6
0
0
3
1
4
1
2
2
3
0
0
0
6
2
0
2
1
4-
0
0
0
0
8
0
0
225
0
1
0
0
7
0
1
2
1
4
1
0
5
1
1
0
1
7
0
0
0
1
7
0
0
3
1
4
0
0
0
0
8
0
0
C&T
0
0
1
0
7
0
0
1
0
7
0
0
1
0
7
0
0
2
0
0
0
0
2
0
6
0
0
2
0
6
0
0
0
2
6
WDR357/3.4/1
-------
Table J-lc
(continued)
Sampling/
Comparison
Area
8
Totals Over All LCICs
++
EDA Sampling Areas
1
7
0
1
0
0
2
6
0
2
0
0
3
4
1
3
0
0
4
0
3
2
2
1
5
0
0
8
0
0
6
0
0
4
1
3
7
0
0
8
0
0
Comparison Areas
221 225
1
0
7
0
0
C&T
0
0
2
0
6
9
10
++ 7
+ 0
o 0
1
0
++ 7
+ 0
0 1
0
0
4
1
2
1
0
7
0
1
0
0
7
0
0
1
0
4
1
2
1
0
1
1
5
0
1
0
0
7
1
0
0
0
7
1
0
0
0
4
1
3
0
0
8
0
0
0
0
7
1
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
6
0
2
0
0
6
0
2
0
0
6
0
2
0
0
6
2
0
0
0
6
0
2
0
0
0
0
1
1
6
7
0
1
0
0
++ = Column sampling/comparison area > Row samp1ing/comparisen area
at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area
at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling/comparison area
at 0.05 significance level.
-- = Column sampling/comparison area < Row sampling/comparison area
at 0.01 significance level.
All test results reported are based on two-sided p-values.
221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
Based on observations classified as Good.
WDR357/d.4/2
-------
Table J-2a
RESULTS FOR NOKPARAMETPIC
MULTIVARIATE SAMPLING/COMPARISON . .
AREA TO SAMPLING/COMPARISON AREA COMPARISONSa'DfC'a
Sampling/
Comparison
Area 1
1
2 ++
3 ++
4 ++
5 + +
6 ++
7 ++
221 ++
225 ++
C&T ++
EDA Sampling Areas
2345
__
0
o
++ ++ o
+•»• -H- 0
+4- ++ 0 O
++ ++
++ ++ +
++ -f O
"
6 7
__
—
__
o ++
o +
++
—
++
o
++
Comparison Areas
221
225
C&T
a. ++ = Column sampling/c»niparison area > Row sampling/comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling comparison area at 0.01 significance level.
b. The direction of the difference between the EDA Neighborhoods and Comparison Areas is based on the sign of the sum of the
elements of the 8x1 vector of rank-sums (Volume III, Appendix B).
c. 221 = Census Tract 221.
225 = Census Tract 225.
CST = Cheektowaga and Tonawanda.
d. Based on observations classified as Good.
WDR357/d.5
-------
Table J-2b
TWO-SIDED p-VALUES FOR NONPARAMETRIC
MULTIVARIATE SAMPLING/COMPARISON AREA .
TO SAMPLING/COMPARISON AREA COMPARISONS3'13
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
EDA Sampling Areas
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2
0.000
1.000
0.613
0.001
0.000
0.000
0.000
0.000
0.002
0.000
3
0.000
0.613
1.000
0.004
0.000
0.000
0.000
0.000
0.022
0.000
4
0.000
0.001
0.004
1.000
0.084
0.086
0.009
0.000
0.095
0.000
5
0.000
0.000
0.000
0.084
1.000
0.220
0.028
0.037
0.023
0.000
6
0.000
0.000
0.000
0.086
0.220
1.000
0.008
0.001
0.019
0.000
7
0.000
0.000
0.000
0.009
0.028
0.008
1.000
0.004
0.778
0.000
a. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
b. Based on observations classified as Good.
Comparison Areas
221
0.000
0.000
0.000
0.000
0.037
0.001
0.004
1.000
0.017
0.000
225
0.000
0.000
0.022
0.095
0.023
0.019
0.778
0.017
1.000
0.000
C&T
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
WDR357/d.6
-------
Table J-3
RESULTS FOR NONPARAMETRIC UNIVARIATE SAMPLING/COMPARISON
AREA TO SAMPLING/COMPARISON AREA COMPARISONS WITH OBSERVATIONS
LESS THAN OR EQUAL TO 2.0 ppb TREATED AS NONDETECT3' '°' f6'
Sampling/
Comparison
LCIC Area
DCB 1
2
3
4
5
6
7
221
225
C&T
Median
Cone. EDA
(ppb) 123
ND ND ND
ND -(--)
ND +(++) o
ND -M- o
ND ++ o(++) o (+•»•)
ND ++ o o
ND ++ o o(+)
ND ++ o o
ND ++ o o
ND ++ o o
ND ++ o(++) o(-n-)
Sampling Areas
4
ND
~
o(~)
o(— )
o(-)
o
o(-)
o(— )
o(-)
o
5
ND
—
o
o
o(+)
o(-t-)
0
o
o
o
6
ND
~
o
o(-)
o
o(-)
o(-)
o(~)
o
o
7
ND
~
o
o
o(++)
o
o(+)
o
o
o(+)
Comparison Areas
221
ND
~
o
o
o(++)
o
O(-M-)
O
o
O(-M-)
225
ND
~
o
o
o(+)
o
o
o
o
o
C&T
ND
~
O ' — — J
o ( •••• j
o
o
o
o(-)
o(— )
o
8.67
TCB
1
2
3
4
5
6
7
221
225
C&T
8.67
ND
ND
ND
ND
ND
ND
ND
ND
ND
++
ND
o
O(-M-)
ND
ND
ND
ND
ND
ND
ND
ND
o(— )
0
0(0)
o
o(-)
o
+ (++)
o(-)
o
++
+ (o)
+ (o)
o
•n-
o(— )
-(o) o
-(o)
o(++)
o(-)
o( — ) o
-(o)
o(++) o(++)
o(-)
o(+)
-(o)
O(-M-)
o
-(o)
o(++)
o(-)
o
0
++
•Ho)
+ (o)
-n-
-------
LCIC
TeCB
CNP
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
Median
Cone.
(ppb)
11.48
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Table J-3
(continued)
EDA Sampling Areas
1
11.48
++
++
++
++
++
++
•n-
•n-
*+
ND
o
o
o
o
0
o
o
o(-)
0
2
ND
o
++
+ (++)
++
++
++
O(-H-)
++
ND
o
o
o
o
o
o
o
o(-)
o
3
ND
o
O(-M-)
O(-M-)
•n-
++
++
O(-M-)
++
ND
o
o
o
o
o
o(-)
o
o(~)
o
4
ND
—
o(~)
o
++(o)
o
+ (-)
o
++
ND
o
o
o
o
o
o
o
o
o
5
ND
-(--)
o(— )
o
++
o
•n-(o)
o
++
ND
o
o
o
o
o
o
o
o(-)
o
6
ND
—
—
— (o)
—
O V"™ r
O ' — ™l
—
o(++)
ND
0
o
o
o
o
o
o
o
0
7
ND
—
—
o
o
O(-M-)
O
-(o)
o(-n-)
ND
o
o
o(+)
o
o
o
o
o
o(+)
Comparison Areas
221
ND
--(o)
O(-M-)
o(-n-)
ND
o
o
o
o
o
o
o
o
o
225
ND
o
o
ND
o(-i-)
o
o
o
C&T
ND
ND
o
o
o
o
o
o
o(-)
o
o(-)
WDR357/030/2
-------
IOC
a-BHC
d-BHC
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
CST
Median
Cone.
(ppb)
8.25
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Table J-3
(continued)
EDA Sampling Areas
123
8.25 ND ND
•M- o(-)
++ o(+)
++ o(++) o
++ o(++) o(+)
•M- ++ +(++)
++ + (++) o(+)
++ + (++) 0
+ + O(-M-) 0
++ +(«•) +<++)
ND ND ND
++ o(-)
++ o(+)
++ o(++) o
++ o(++) o
++ o(++) o (++)
+•*• o(++) o(-n-)
+ + O(-M-) 0 (•«•*)
++ o(++) o
++ o(++) o (++)
4
ND
~o(-)
o
o
-H-(O)
•M-(O)
+ (o)
O
++
ND
0(")
o
o
o
o
o(+)
o
o(++)
567
ND ND ND
'"(_-) " "(-)
o(-) -(--) o(-)
o --(o) — (o)
--(o) -(o)
++(o) o(+)
+ (o) o(-)
+ (o) o(-) o
o — (o) -(o)
•*•(++) o(++) o(++)
ND ND ND
"0(..} "0{-, "o(-)
o o( — ) o( — )
o o o
o(-) o
o(+) o
o o
o(+) o o
o o o
o(++) o(++) o (+•«•)
Comparison Areas
221
ND
"-(--)
o
-(o)
-(o)
o(+)
o
-(o)
O(-M-)
ND
'o(-)
o( — )
o(-)
o(-)
o
o
o
0
225
ND
~o(-)
o
o
o
•M-(O)
+ (o)
+ (o)
+ (-^)
ND
'o(-)
0
o
o
o
o
o
o(++)
CST
ND
"(--)
_(__)
«
_(__)
o(-)
o(--)
o(— )
_(_-)
ND
~o(-J
o(--)
O \ ~ ""/
o(--)
O '"""")
o ( ^^ )
o
o(«)
-------
LCIC
b-BHC
g-BHC
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
Median
Cone.
(ppb) 1
11.58
11.58
ND ++
ND + +
ND ++
ND + +
ND ++
ND + +
ND -M-
ND + +
ND ++
ND
ND
ND + +
ND + +
ND + +
ND ++
ND + +
ND ++
ND -n-
ND + +
ND ++
Table J-3
(continued)
EDA Sampling Areas
2 34567
ND ND ND ND ND ND
—
o o o( — ) -( — ) -( — )
o o o(-) -( — ) -(o)
00 O — — (o)
o (+•*•) o(+) o -(o) o
+ (++) +(++) ++ +(0) O(-M-)
+ (•*-+) -Mo) ++(o) o o(~)
+ (++) + (++) +-»•(+) +(o) o o
o o o o o( — ) o
o(++) + (++) •*•(++) o(++) o(++) o (•«•+)
ND ND ND ND ND ND
__
o o(-) o(--) o(~) o(~)
o o o(-) -( — ) o( — )
o(+) o o --(-) — (o)
O(-H-) o(+) 0 -(o) -(o)
o(++) + (++) ++(+) +(o) o
O(++) O(-H-) -H-(o) +(o) O
o (++) o (•*•+) o(+) o o o
o(+) o o o o(-) o
0(++) O(-H-) 0(++) +(++) O(-M-) 0(++)
Comparison Areas
221
ND
—
-(--)
-(--)
--(-)
-(o)
o
o
-
0(++)
ND
—
O \ ™~)
o ( ^^ /
o(-)
o
o
o
o
O(-M-)
225
ND
—
o
o
o
o
O(-H-)
0
+
o(+0
ND
—
o(-)
o
o
o
+ (+)
0
o
o(++)
C&T
ND
—
o(— )
-(--)
_(__)
o( — )
o(~)
o( — )
o( — )
o(--)
ND
~
o(--)
o(-)
o( — )
_(__)
o(— )
o(— )
o( — )
o(— )
WDR357/030/4
-------
LCIC
Sampling/
Comparison
Area
Median
Cone.
(ppb)
TOTALS
Table J-3
(continued)
1
62
1
9
0
0
2
8
8
51
1
4
EDA
3
6
6
54
0
6
Sampling
4
8
5
52
0
7
Areas
5
6
7
52
1
6
6
0
0
47
5
20
7
0
0
53
9
10
Comparison Areas
All Symbols
72
72
72
72
72
72
72
221
0
0
49
12
JJL
72
225
6
4
56
0
_6
72
CST
0
0
49
7
16
72
a. ++ = Column sampling/comparison area > Row sampling/comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling/comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling/comparison area at 0.01 significance level.
b. All test results .reported are based on two-sided p-values.
c. The first entry in each column is median concentration. ND indicates nondetect.
d. C&T denotes Cheektowaga and Tonawanda.
e. Based on observations classified as Good.
f. Any change in results between this table and the original comparisons reported in Table STM2-la
is indicated by two sets of symbols; the second set of symbols denotes the original results.
-------
TCB
TeCB
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
0.94
0.35
0.39
0.33
0.36
0.33
0.38
0.38
0.39
0.35
7.80
0.83
0.79
0.41
0.44
0.36
0.54
0.63
0.56
0.13
11.34
1.09
0.94
0.35
0.38
0.28
0.49
0.50
0.54
0.05
Table J-4a
RESULTS FOR NONPARAMETRIC UNIVARIATE SAMPLING/COMPARISON
AREA TO SAMPLING/COMPARISON AREA COMPARISONS WITH
UPPER 10 PERCENT TRIMMED IN EACH AREA ' ' ' ' '
EDA Sampling Areas
•M-
11.34
1.09
Comparison Areas
1
0.94
•*•+
++
•n-
++
++
++
•n-
7.80
++
2
0.35
o
o
0
o
o
o
++
0.83
3
0.39
o
0
o
o
o
0
•n-
0.79
o
4
0.33
—
—
o(-)
o
—
—
-
o
0.41
—
5
0.36
o
o
o(-i-)
o
o
o
o
0.44
—
6
0.33
o
o
o
o(-)
-
—
0
o
0.36
__
7
0.38
o
o
o
o
0
++(+)
0.54
—
221
0.38
o
o
o
o
o
++
0.63
—
225
0.39
0
o
o
0
o
o
o
0.56
—
C&T
0.35
—
—
o
o
o
— (-)
—
o
0.13
—
0.94
-(o)
o
o
-(o)
0.35
o
o
-(o)
-(o)
o
o
o
0.38
o
o
o
0.28
•(o)
o
.+
o
o
•+
49
++(+)
o
o
o
++
0.50
0
o
o
o
+-T
0.54
•Mo)
o
o
o
o
M-
O
O
O
O
0.05
WDR357/d.7/l
-------
LCIC
CNP
a-BHC
d-BHC
b-BHC
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
Median
Cone.
(ppb)
ND
0.04
ND
0.04
0.05
0.05
0.05
0.05
0.07
0.02
7.51
0.35
0.18
0.08
0.06
0.06
0.13
0.18
0.07
ND
0.79
ND
ND
ND
ND
ND
ND
ND
ND
ND
10.73
0.17
ND
ND
ND
ND
ND
ND
ND
ND
Table J-4a
(continued)
EDA Sampling
1 2
ND 0.04
o
o
o o
o o
O 0
-(o) -(o)
o -(o)
-(o) -(o)
— (-)
o o
7.51 0.35
•M-
+ + + +(•*•)
-n- ++
+ + ++
++ ++
++ ++
++ ++
t+ t+
0.79 ND
H *:r
•n- ++
++ ++
++ ++
+t tt
10.73 0.17
++
•f + O
•n- o
•f-f ++
3
ND
o
o
o
o
o
-(o)
—
o
0.18
--(-)
o
+
++
+
o
o
ND
o
++
++
++
+ (+)
ND
o
o
+•»•(+]
4
0.04
o
o
0
0
o
o
o
o
0
0.08
—
o
0
o
o
o
o
ND
-(o)
o
+ (o)
o
+
o
ND
o
o
1 + (o)
Areas
5
0.05
o
o
o
o
o
0
o
-
o
0.06
—
-
o
o
o
o
o
ND
o
o
++(+)
+ (o)
++(+)
o
ND
—
-(o)
6
0.05
+ \O i
^* (o)
0
o
o
o
o
o
o
0.06
—
--
o
o
-
— (-)
o
ND
-Co)
o
o
o
O(-M-)
ND
—
__
o
7
0.05
o
+ (o)
+
o
o
o
o
o
+
0.13
—
-
o
o
+
o
o
ND
o
-(o)
o
o
o
ND
—
o
o
o
•n-
o
o
o
o
o
o
o
Comparison Areas
221
0.05
+ (o)
+ (o)
+ (o)
o
0
o
o
o
0
0.18
0
o
o
++ (+)
o
o
ND
~~
-
— (-)
o
o
o
o
ND
~_
—
o
o
-(o)
225
0.07
+
++(+)
++
o
+
0
0
o
-n-(-i-)
0.07
o
o
o
o
o
o
++
ND
-(o)
o
o
o
o
o
+ <++)
ND
o
o
o
•Ko)
-n-
0
C&T
0.02
o
o
o
0
o
o
-
o
— (-)
ND
"•"
—
--
—
--
—
"""*
ND
__
--
—
o( —
0
- ( —
ND
—
--
~
-(o)
WDR357/d.7/2
-------
Table J-4a
(continued)
LCIC
g-BHC
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
CST
Median
Cone.
(ppb)
1.48
0.03
ND
ND
ND
ND
ND
ND
ND
ND
EDA Sampling Areas
123
1.48 0.03 ND
+ + o
++ o
•f-t- ++(+) O
++ +4- ++(+)
+ + ++ ++
++ ++ ++
•n- ++ ++
++ ++(+) o
++ ++ ++
4
ND
--(-)
o
o
++ (+)
o
+
o
++
5
ND
—
— (-)
0
o
o
0
o
+-f
6
ND
—
—
--(_)
o
o
o
— (-)
+ (++)
7
ND
—
—
o
o
o
o
o
++
Comparison Areas
221
ND
o
o
o
225
ND
o
o
p
o
o
C&T
ND
TOTALS
All Symbols
63
0
6
3
_0
72
44
1
16
3
_8
72
30
5
24
3
72
8
5
34
5
20
72
10
1
40
4
17
72
4
3
27
3
35
72
10
8
35
2
17
72
12
4
32
6
18
72
12
7
38
1
14
72
0
0
14
4
54
72
a. ++ = Column sampling/comparison area > Row sampling/comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling/comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling/comparison area at 0.01 significance level.
b. All test results reported are based on two-sided p-values.
The first entry in each column is median concentration; ND indicates non-detect.
C&T = Cheektowaga and Tonawanda.
Based on observations classified as Good.
c.
d.
e.
f.
Any change in results between Table STMS2-la and this table is indicated by two sets of symbols; the second set of
symbols denotes the original results reported in Table STM2-la.
WDR357/d.7/3
-------
Table J-4b
TWO-SIDED p-VALUES FOR NONPARAMETRIC UNIVABIATE
SAMPLING/COMPARISON AREA TO SAMPLING/COMPARISON AREA .
COMPARISONS WITH UPPER 10 PERCENT TRIMMED IN EACH AREA3'
LCIC
DCB
TCB
TeCB
CNP
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
EDA Sampling Areas
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.698
0.516
0.171
0.218
0.046
0.055
0.035
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
2
.000
.000
.854
.008
.427
.092
.901
.861
.711
.005
.000
.000
.233
.000
.000
.000
.000
.003
.003
.000
.000
.000
.046
.000
.000
.000
.000
.000
.000
.000
.698
.000
.792
.614
.343
.038
.032
.018
3
0.000
0.854
1.000
0.005
0.558
0.063
0.949
0.918
0.560
0.002
0.000
0.233
1.000
0.000
0.009
0.000
0.000
0.016
0.006
0.000
0.000
0.046
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.516
0.792
1.000
0.312
0.654
0.160
0.022
0.049
4 .
0.000
0.008
0.005
1.000
0.060
0.292
0.001
0.003
0.018
0.894
0.000
0.000
0.000
1.000
0.134
0.218
0.035
0.002
0.059
0.000
0.000
0.000
0.000
1.000
0.468
0.117
0.020
0.007
0.030
0.000
0.171
0.614
0.312
1.000
0.771
0.800
0.279
0.572
5
0.000
0.427
0.558
0.060
1.000
0.077
0.373
0.405
0.914
0.073
0.000
0.000
0.009
0.134
1.000
0.003
0.677
0.072
0.771
0.000
0.000
0.000
0.000
0.468
1.000
0.007
0.090
0.142
0.164
0.000
0.218
0.343
0.654
0.771
1.000
0.244
0.141
0.299
6
0.000
0.092
0.063
0.292
0.077
1.000
0.030
0.009
0.328
0.158
0.000
0.000
0.000
0.218
0.003
1.000
0.000
0.000
0.006
0.000
0.000
0.000
0.000
0.117
0.007
1.000
0.000
0.000
0.000
0.000
0.046
0.038
0.160
0.800
0.244
1.000
0.379
0.382
7
0.000
0.901
0.949
0.001
0.373
0.030
1.000
0.932
0.554
0.008
0.000
0.000
0.000
0.035
0.677
0.000
1.000
0.052
0.637
0.000
0.000
0.000
0.000
0.020
0.090
0.000
1.000
0.765
0.790
0.000
0.055
0.032
0.022
0.279
0.141
0.379
1.000
0.955
Comparison Areas
221
0.000
0.861
0.918
0.003
0.405
0.009
0.932
1.000
0.817
0.002
0.000
0.003
0.016
0.002
0.072
0.000
0.052
1.000
0.180
0.000
0.000
0.000
0.000
0.007
0.142
0.000
0.765
1.000
0.568
0.000
0.035
0.018
0.049
0.572
0.299
0.382
0.955
1.000
225
0.000
0.711
0.560
0.018
0.914
0.328
0.554
0.817
1.000
0.111
0.000
0.003
0.006
0.059
0.771
0.006
0.637
0.180
1.000
0.000
0.000
0.000
0.000
0.030
0.164
0.000
0.790
0.568
1.000
0.000
0.016
0.003
0.007
0.069
0.010
0.238
0.532
0.810
C&T
0.000
0.005
0.002
0.894
0.073
0.158
0.008
0.002
0.111
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.440
0.935
0.704
0.099
0.551
0.139
0.017
0.064
a-BHC
225 0.016 0.003 0.007 0.069 0.010 0.238 0.532 0.810 1.000 0.004
C&T 0.440 0.935 0.704 0.099 0.551 0.139 0.017 0.064 0.004 1.000
1
2
3
4
5
6
7
221
225
C&T
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0
1
0
0
0
0
0
0
0
0
.000
.000
.003
.000
.000
.000
.000
.000
.000
.000
0.000
0.003
1.000
0.239
0.014
0.000
0.026
0.122
0.098
0.000
0.000
0.000
0.239
1.000
0.585
0.113
0.724
0.931
0.352
0.000
0.000
0.000
0.014
0.585
1.000
0.189
0.337
0.087
0.895
0.000
0.000
0.000
0.000
0.113
0.189
1.000
0.011
0.006
0.276
0.000
0.000
0.000
0.026
0.724
0.337
0.011
1.000
0.321
0.394
0.000
0.000
0.000
0.122
0.931
0.087
0.006
0.321
1.000
0.116
0.000
0.000
0.000
0.098
0.352
0.895
0.276
0.394
0.116
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
WDR357/d.8/l
-------
LCIC
d-BHC
b-BHC
g-BHC
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
CST
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
Table J-4b
(continued)
EDA Sampling Areas
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2
0.000
1.000
0.002
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.249
0.074
0.000
0.000
0.001
0.000
0.051
0.000
0.000
1.000
0.244
0.007
0.000
0.000
0.000
0.000
0.003
0.000
3
0.000
0.002
1.000
0.047
0.118
0.000
0.001
0.001
0.026
0.000
0.000
0.249
1.000
0.358
0.002
0.000
0.082
0.001
0.417
0.000
0.000
0.244
1.000
0.148
0.005
0.000
0.000
0.000
0.082
0.000
4
0.000
0.000
0.047
1.000
0.714
0.021
0.111
0.024
0.329
0.009
0.000
0.074
0.358
1.000
0.018
0.000
0.693
0.027
0.776
0.000
0.000
0.007
0.148
1.000
0.304
0.004
0.068
0.016
0.705
0.000
5
0.000
0.000
0.118
0.714
1.000
0.002
0.040
0.004
0.355
0.003
0.000
0.000
0.002
0.018
1.000
0.055
0.088
0.673
0.025
0.000
0.000
0.000
0.005
0.304
1.000
0.061
0.471
0.148
0.306
0.001
6
0.000
0.000
0.000
0.021
0.002
1.000
0.284
0.371
0.105
0.371
0.000
0.000
0.000
0.000
0.055
1.000
0.000
0.139
0.000
0.008
0.000
0.000
0.000
0.004
0.061
1.000
0.195
0.837
0.008
0.036
7
0.000
0.000
0.001
0.111
0.040
0.284
1.000
0.148
0.572
0.090
0.000
0.001
0.082
0.693
0.088
0.000
1.000
0.026
0.601
0.000
0.000
0.000
0.000
0.068
0.471
0.195
1.000
0.307
0.105
0.002
Comparison
221
0.000
0.000
0.001
0.024
0.004
0.371
0.148
1.000
0.069
1.000
0.000
0.000
0.001
0.027
0.673
0.139
0.026
1.000
0.035
0.000
0.000
0.000
0.000
0.016
0.148
0.837
0.307
1.000
0.088
0.030
225
0.000
0.000
0.026
0.329
0.355
0.105
0.572
0.069
1.000
0.045
0.000
0.051
0.417
0.776
0.025
0.000
0.601
0.035
1.000
0.000
0.000
0.003
0.082
0.705
0.306
0.008
0.105
0.088
1.000
0.000
Areas
CST
0.000
0.000
0.000
0.009
0.003
0.371
0.090
1.000
0.045
1.000
0.000
0.000
0.000
0.000
0.000
0.008
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.001
0.036
0.002
0.030
0.000
1.000
a. 221 = Census Tract 221.
225 = Census Tract 225.
CST = Cheektowaga and Tonawanda.
b. Based on observations classified as Good.
WDR357/d.8/2
-------
Table J-4c
SUMMARY OF NONPARAMETRIC UNIVARIATE SAMPLING/COMPARISON
AREA TO SAMPLING/COMPARISON AREA COMPARISONS WITH UPPER
10 PERCENT TRIMMED IN EACH AREA3' /C/
Sampling/
Comparison
Area
1 ++
•f
o
-
—
2 ++
+
o
-
—
3 ++
+
o
-
—
4 ++
+
0
-
—
5 ++
+
o
-
—
6 ++
+
o
-
Totals Over all LCICs
EDA Sampling
1
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
0
1
0
2
0
0
1
0
7
2
I
5
0
0
5
0
3
0
0
6
0
2
0
0
6
0
1
1
0
3
0
0
1
0
7
0
0
5
1
2
3
1
4
0
0
4
1
3
0
0
6
0
2
0
0
4
0
0
1
0
7
0
0
2
0
6
0
0
4
1
3
0
1
7
0
0
2
1
5
0
0
Areas
5
0
0
1
0
7
0
0
2
0
6
0
0
3
1
4
0
0
7
1
0
3
0
5
0
0
£
0
1
0
0
7
0
1
1
0
6
0
0
2
0
6
0
0
5
1
2
0
0
5
0
3
7
0
0
1
0
7
0
1
1
0
6
0
1
2
1
4
1
2
5
0
0
0
0
7
1
0
3
2
3
0
0
Comparison Areas
221
0
1
0
0
7
0
1
1
0
6
0
1
2
1
4
3
0
2
3
0
0
0
7
0
1
4
0
4
0
0
225
0
1
0
0
7
1
0
2
0
5
1
0
4
1
2
0
2
6
0
0
0
2
6
0
0
4
0
4
0
0
C&T
0
0
1
0
7
0
0
1
0
7
0
0
1
0
7
0
0
2
0
6
0
0
2
0
6
0
0
3
1
4
WDR357/d.9/l
-------
Sampling/
Comparison
Area 1
7 ++7
+ 0
0 1
0
0
221 ++ 7
+ 0
o 0
1
0
225 ++ 7
+ 0
o 0
1
0
C&T ++ 7
+ 0
0 1
0
0
Table J-4c
(Continued)
Totals Over All LCICs
EDA Sampliing Areas
2
6
0
1
1
0
6
0
1
1
0
5
0
2
0
1
7
0
1
0
0
3
4
1
2
1
0
4
1
2
1
0
2
1
4
0
1
7
0
1
0
0
4
0
0
5
2
1
0
3
1
0
4
0
0
6
2
0
6
0
2
0
0
5
0
1
7
0
0
1
0
7
0
0
0
0
6
2
0
6
0
2
0
0
6
0
0
3
2
3
0
0
5
0
3
0
0
4
0
4
4
1
3
0
0
7
0
1
7
0
0
0
0
8
0
0
6
1
1
0
0
Comparison Areas
221
0
0
7
1
0
0
0
7
1
0
5
1
2
0
0
225
0
0
8
0
0
0
1
7
0
0
6
1
1
0
0
C&T
0
0
1
1
6
0
0
2
1
5
0
0
1
1
6
a. ++ = Colummn sampling/comparison area < Row sampling/comparison area
at 0.01 significance level.
+ = Column sampling/comparison area < Row sampling/comparison area
at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area > Row sampling/comparison area
at 0.05 significance level.
— = Column sampling/comparison area > Row sampling/comparison area
at 0.01 significance level.
b. All tests results reported are based on two-sided p-values.
c. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
d. Based on observations classified as Good.
WDR357/d.9/2
-------
Table J-5
COUNTS OF SAMPLES VERSUS FAILED LCIC IDENTIFICATIONS CAUSED BY
ION RATIOS IN THE RANGE OF +/- 20 TO 40 PERCENT OF THE
THEORETICAL RATIO*
SOIL ASSESSMENT—INDICATOR CHEMICALS
LCIC
Category
b
Detect
c
20% to 40%
d
>40%
Other NDs
f
Flagged as U/B
Total
Percent Increase
in Detects Using
20% to 40% Ion Ratios
DCB
649
0
2
4
126
781
0.0
TCB
684
0
0
1
96
781
0.0
TeCB
658
2
2
14
105
781
0.3
CNP
399
55
23
163
141
781
13.8
A-BHC
464
63
49
109
96
781
13.6
D-BHC
152
32
64
425
108
781
21.0
B-BHC
268
53
75
238
147
781
19.8
G-BHC
215
31
52
383
100
781
14.4
Only Good data included in this table
b
Passed all ID criteria
Failed only the ion ratio criteria in the +/- 20% to 40% theoretical range
Failed only the ion ratio criteria in >40% theoretical range
Failed ID criteria other than only the ion ratio
Failed the EMSL/LV data usability review as Uncertain or Bad data
d.
WDR363/010
-------
Table J-6a
RESULTS FOR NONPARAMETRIC UNIVARIATE SAMPLING/COMPARISON
AREA TO SAMPLING/COMPARISON AREA COMPARISONS.
BASED ON ION RATIO-MODIFIED ^0'0'0'
LCIC
DCB
TCB
TeCB
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
Median
Cone.
(ppb)
1.01
0.39
0.40
0.36
0.39
0.36
0.41
0.39
0.43
0.36
8.67
0.91
0.89
0.44
0.50
0.37
0.58
0.65
0.60
0.13
11.48
1.17
1.06
0.39
0.43
0.31
0.54
0.53
0.61
0.05
EDA S amp lintL Ares
Comparison Areas
1 2
1.01 0.39
++
++ o
•M- O
++ 0
++ 0
++ 0
++ O
++ ++
3
0.40
o
0
o
o
0
++
4
0.36
--
—
o
—
—
-
o
5
0.39
o
o
o(-t-)
o
0
o
o
6
0.36
0
-
o
o(-)
•
—
o
0
7
0.41
o
o
o
•f
o
o
+
221
0.39
o
o
o
o
o
•M-
225
0.43
o
o
o
o
o
o
o
C&T
0.36
—
—
0
o
o
-
—
o
8.67
•M-
•H-
11.48
•n-
0.91
++
•M-
1.17
0.89
o
1.06
o
0.44
o
o
o
o
•f-t-
0.39
o
o
o
o
o
0.50
•n-
o
o
o
0.43
o
o
o
0.37
0.31
0.58
o
o
o
o
0.54
o
o
o
o
0.65
o
o
o
•M-
0.53
o
o
o
o
0.60
o
o
o
o
0.61
o
o
o
o
0.13
0.05
•n-
WDR357AJ.10/1
-------
LCIC
CNP
a-BHC
d-EHC
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
Median
Cone.
(ppb)
0.06
0.05
0.05
0.04
0.06
0.07
0.06
0.07
0.07
0.04
8.25
0.40
0.24
0.13
0.14
0.10
0.17
0.18
0.15
ND
1.39
0.10
ND
ND
ND
ND
ND
ND
ND
ND
Table J-6a
(continued)
EDA Sampling Areas
1
0.06
0
o
o
o
o
o
0
o(-)
0
8.25
£
+4-
4+
44
£
1.39
H
44
44
4+
44
44
2 3
0.05 0.05
o o
o
o
o o
o o
0 O
-(o) -(o)
o(-)
0 O
0.40 0.24
44 o
44 4
+4- 44
44 o(+)
44 o
0.10 ND
44 o
44 O
44 44
44 4 +
44 44
44 O
4
0.04
o
o
o
o
o
-(o)
o
-(o)
o
0.13
o
o
4(o)
o
o
o
ND
o
o
+ (o)
o
+
o
5
0.06
o
o
o
o
o
o
o
o
0.14
o
0
o
o
o
.ND
0
o
+
o
+
o
6
0.07
o
o
o
o
o
o
o
o
o
0.10
-(o)
o
— (-)
-(o)
ND
-(o)
-
o
o
o
7
0.06
0
o
++(+
+ (o)
0
o
o
o
++(+
0.17
o(-)
o
0
++(+
o
o
ND
o
o
0
0
o
Comparison Areas
221
0.07
o
O
o
o
o
0.18
-(o)
o
o
o
o
+4
ND
o
o
225
0.07
o
o
o
0.15
o
o
o
•t-(o)
o
o
ND
o
o
o
o
o
o
C&T
0.04
o
o
o
o
o
o
-(o)
ND
ND
-(o)
WDR357/d.lO/2
-------
LCIC
b-BHC
g-BHC
1
2
3
4
5
6
7
221
225
C&T
Table J-6a
(continued)
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
Median
Cone.
(ppb) 1
11.58
11.58
0.29 ++
0.15 ++
0.07 ++
ND ++
ND ++
0.10 -M-
ND ++
0.12 ++
ND ++
EDA Sampling Areas
2
0.29
—
o
+ (o)
+ 4
++
++
++
O
+H-
3
0.15
—
o
o
o(+)
++
o
++
o
++
4
0.07
—
-(o)
o
o
++
o
+
o
++
5
ND
—
—
o(-)
o
+ (o)
o
0
o
++
6 7
ND 0.10
—
—
o
o
-(o) o
++
—
o •*•(+)
o
++ ++
Comparison Areas
221
ND
—
—
—
-
o
o
-(o)
-
++
225
0.12
--
o
o
o
o
+ +
o
•f
++
C&T
ND
—
—
—
--
—
—
—
-_
—
1.48
0.09
ND
ND
ND
ND
ND
ND
ND
ND
1.48
•n-
09
0
+
•+
-+
•+
•+
•+(+)
ND
o
o
+
++
++
++
o
ND
-
o
o
++(+)
o
o(+)
o
ND
—
-
0
0
0
o
o
ND
—
--
— (-)
o
o
o
o(-)
ND
—
--
o
o
o
o
o
ND
—
--
o(-)
o
o
o
o
ND
-
0
o
o
o
o
o
-(-
(+)
ND
WDR357/d.lO/3
-------
Table J-6a
(continued)
TOTALS ++ 63 42 26 8 8 6 13 11 10 0
+ 0374404560
o 9 19 27 36 42 25 38 32 42 10
0 1 27470714
_0 ,_7 10 17 14 34 17 17 13 58
All Symbols 72 72 72 72 72 72 72 72 72 72
a. -w- = Column sampling/comparison area > Row sampling/comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling/comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling/comparison area at 0.01 significance level.
b. All test results reported are based on two-sided p-values.
c. The first entry in each column is median concentration, ND indicates nondetect.
d. 221=Census Tract 221.
225=Census Tract 225.
C&T=Cheektowaga and Tonawanda.
e. Based on observations classified as Good.
f. Any change in results between Table STMS2-la and this table is indicated by two sets of symbols; the second set of
symbols denotes the original results, reported in Table STM2-la.
HDR357/d.lO/4
-------
Table J-6b
TWO-SIDED p-VALUES FOR NONPARAMETRIC UNIVARIATE
SAMPLING/COMPARISON AREA TO SAMPLING/COMPARISON AREA .
COMPARISONS BASED ON ION RATIO-MODIFIED CONCENTRATIONS3'D
LCIC
DCB
TCB
TeCB
CNP
a-BHC
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
CST
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
2
3
4
5
6
7
221
225
C&T
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0,000
0.000
0.000
0.000
0.000
0.000
1.000
0.445
0.693
0.843
0.910
0.557
0.361
0.320
0.202
0.946
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
2
.000
.000
.989
.004
.397
.052
.564
.805
.539
.008
.000
.000
.245
.000
.001
.000
.000
.001
.010
.000
.000
.000
.069
.000
.000
.000
.000
.000
.000
.000
.445
.000
.885
.517
.428
.254
.074
.019
.057
.573
.000
.000
.010
.001
.000
.000
.000
.000
.000
.000
EDA Sampling^ Areas
3
0.000
0.989
1.000
0.006
0.644
0.041
0.740
0.980
0.583
0.002
0.000
0.245
1.000
0.000
0.029
0.000
0.002
0.014
0.032
0.000
0.000
0.069
1.000
0.000
0.000
0.000
0.000
0.000
0.005
0.000
0.693
0.885
1.000
0.638
0.566
0.346
0.008
0.024
0.009
0.559
0.000
0.010
1.000
0.263
0.019
0.000
0.051
0.045
0.219
0.000
4
0.000
0.004
0.006
1.000
0.030
0.361
0.003
0.004
0.026
0.896
0.000
0.000
0.000
1.000
0.221
0.185
0.112
0.015
0.092
0.000
0.000
0.000
0.000
1.000
0.562
0.102
0.071
0.054
0.064
0.000
0.843
0.517
0.638
1.000
0.524
0.365
0.012
0.112
0.045
0.398
0.000
0.001
0.263
1.000
0.312
0.028
0.736
0.785
0.868
0.000
5
0.000
0.397
0.644
0.030
1.000
0.050
0.588
0.534
0.817
0.057
0.000
0.001
0.029
0.221
1.000
0.002
0.855
0.208
0.750
0.000
0.000
0.000
0.000
0.562
1.000
0.009
0.184
0.360
0.214
0.000
0.910
0.428
0.566
0.524
1.000
0.405
0.083
0.186
0.036
0.329
0.000
0.000
0.019
0.312
1.000
0.219
0.146
0.190
0.425
0.000
6
0.000
0.052
0.041
0.361
0.050
1.000
0.035
0.009
0.312
0.146
0.000
0.000
0.000
0.185
0.002
1.000
0.000
0.000
0.005
0.000
0.000
0.000
0.000
0.102
0.009
1.000
0.000
0.000
0.000
0.000
0.557
0.254
0.346
0.365
0.405
1.000
0.093
0.120
0.332
0.230
0.000
0.000
0.000
0.028
0.219
1.000
0.001
0.009
0.035
0.000
7
0.000
0.564
0.740
0.003
0.588
0.035
1.000
0.992
0.556
0.022
0.000
0.000
0.002
0.112
0.855
0.000
1.000
0.123
0.808
0.000
0.000
0.000
0.000
0.071
0.184
0.000
1.000
0.446
0.982
0.000
0.361
0.074
0.008
0.012
0.083
0.093
1.000
0.995
0.643
0.002
0.000
0.000
0.051
0.736
0.146
0.001
1.000
0.808
0.392
0.000
Comparison
221
0.000
0.805
0.980
0.004
0.534
0.009
0.992
1.000
0.757
0.007
0.000
0.001
0.014
0.015
0.208
0.000
0.123
1.000
0.368
0.000
0.000
0.000
0.000
0.054
0.360
0.000
0.446
1.000
0.346
0.000
0.320
0.019
0.024
0.112
0.186
0.120
0.995
1.000
0.726
0.026
0.000
0.000
0.045
0.785
0.190
0.009
0.808
1.000
0.698
0.000
225
0.000
0.539
0.583
0.026
0.817
0.312
0.556
0.757
1.000
0.194
0.000
0.010
0.032
0.092
0.750
0.005
0.808
0.368
1.000
0.000
0.000
0.000
0.005
0.064
0.214
0.000
0.982
0.346
1.000
0.000
0.202
0.057
0.009
0.045
0.036
0.332
0.643
0.726
1.000
0.018
0.000
0.000
0.219
0.868
0.425
0.035
0.392
0.698
1.000
0.000
Areas
C&T
0.000
0.008
0.002
0.896
0.057
0.146
0.022
0.007
0.194
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.946
0.573
0.559
0.398
0.329
0.230
0.002
0.026
0.018
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
WDR357/d.ll/l
-------
Sampling/
Comparison
LCIC Area
d-BHC 1
2
3
4
5
6
7
221
225
C&T
b-BHC 1
2
3
4
5
6
7
221
225
C&T
g-BHC 1
2
3
4
5
6
7
221
225
C&T
Table J-6b
(continued)
EDA Sampling Areas
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
2
0.000
1.000
0.029
0.004
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.185
0.050
0.001
0.000
0.001
0.000
0.117
0.000
0.000
1.000
0.377
0.043
0.000
0.000
0.000
0.001
0.008
0.000
3
0.000
0.029
1.000
0.151
0.202
0.000
0.001
0.000
0.117
0.000
0.000
0.185
1.000
0.224
0.100
0.000
0.239
0.000
0.589
0.000
0.000
0.377
1.000
0.137
0.018
0.000
0.004
0.010
0.081
0.000
4
0.000
0.004
0.151
1.000
0.725
0.022
0.118
0.018
0.216
0.000
0.000
0.050
0.224
1.000
0.488
0.001
0.969
0.048
0.710
0.000
0.000
0.043
0.137
1.000
0.336
0.009
0.247
0.162
0.517
0.000
5
0.000
0.000
0.202
0.725
1.000
0.032
0.124
0.019
0.490
0.000
0.000
0.001
0.100
0.488
1.000
0.011
0.544
0.134
0.314
0.000
0.000
0.000
0.018
0.336
1.000
0.142
0.936
0.984
0.590
0.000
6
0.000
0.000
0.000
0.022
0.032
1.000
0.522
0.629
0.368
0.002
0.000
0.000
0.000
0.001
0.011
1.000
0.000
0.187
0.000
0.000
0.000
0.000
0.000
0.009
0.142
1.000
0.121
0.192
0.057
0.002
7
0.000
0.000
0.001
0.118
0.124
0.522
1.000
0.264
0.667
0.000
0.000
0.001
0.239
0.969
0.544
0.000
1.000
0.022
0.850
0.000
0.000
0.000
0.004
0.247
0.936
0.121
1.000
0.871
0.539
0.000
Comparison Areas
221
0.000
0.000
0.000
0.018
0.019
0.629
0.264
1.000
0.111
0.023
0.000
0.000
0.000
0.048
0.134
0.187
0.022
1.000
0.024
0.000
0.000
0.001
0.010
0.162
0.984
0.192
0.871
1.000
0.542
0.000
225
0.000
0.000
0.117
0.216
0.490
0.368
0.667
0.111
1.000
0.000
0.000
0.117
0.589
0.710
0.314
0.000
0.850
0.024
1.000
0.000
0.000
0.008
0.081
0.517
0.590
0.057
0.539
0.542
1.000
0.000
C&T
0.000
0.000
0.000
0.000
0.000
0.002
0.000
0.023
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.002
0.000
0.000
0.000
1.000
a. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
b. Based on observations classified as Good.
WDR357/d.ll/2
-------
Table J-6c
SUMMARY OF NONPARAMETRIC UNIVARIATE SAMPLING/COMPARISON
AREA TO SAMPLING/COMPARISON AREA COMPARISONS BASED ON
ION RATIO MODIFIED CONCENTRATIONS
,a,t>,c,d
Sampling/
Comparison
Area
Totals Over All LCLCs
++
++
+ +
++
++
++
EDA Sampliing Areas
I
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
2_
0
0
1
0
7
1
1
6
0
0
5
2
1
0
0
6
0
2
0
0
6
0
2
0
0
_3
0
0
1
0
7
0
0
6
1
r
3
0
5
0
0
1
3
4
0
0
6
1
1
0
0
£
0
0
1
0
7
0
0
1
2
5
0
0
5
0
3
0
0
7
1
0
2
2
4
0
0
5_
0
0
1
0
7
0
0
2
0
6
0
0
4
3
1
0
1
7
0
0
2
2
4
0
0
£
0
0
1
0
7
0
0
2
0
6
0
0
1
1
6
0
1
3
2
2
0
0
4
2
2
_7
0
6
1
0
7
0
0
2
0
6
1
0
3
0
4
0
1
7
0
0
0
0
8
0
0
4
1
3
0
0
Comparison Area
221
0
0
1
0
7
0
1
1
0
6
0
1
1
1
5
1
1
4
2
0
0
0
7
1
0
4
0
4
0
0
225
0
0
1
0
7
0
0
3
0
5
1
0
2
1
4
0
2
6
0
0
0
1
7
0
0
3
1
4
0
0
CST
0
0
1
0
7
0
0
1
0
7
0
0
1
0
7
0
0
2
0
6
0
0
2
0
6
0
0
2
0
6
WDR357/d.l2/l
-------
Sampling/
Comparison
Area
221
225
C&T
Table J-6c
(continued)
Totals over all LCLCs
EDA Sampliing Areas
1
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
2
6
0
2
0
0
6
0
1
1
0
5
0
3
0
0
7
0
1
0
0
3
4
0
3
0
1
4
2
1
1
0
1
1
5
0
1
7
0
1
0
0
4
0
0
6
1
1
0
2
4
1
1
0
0
6
2
0
6
0
2
0
0
5
0
0
8
0
0
0
1
7
0
0
0
0
7
1
0
6
0
2
0
0
6
0
0
3
1
4
0
0
4
0
4
0
0
4
1
3
6
0
2
0
0
7
0
1
7
0
0
0
0
8
0
0
7
1
0
0
0
Comparison Area
221
0
0
7
1
0
0
0
7
1
0
6
2
0
0
0
225
0
0
8
0
0
0
1
7
0
0
6
1
1
0
0
C&T
0
0
0
1
7
0
0
0
2
6
0
0
1
1
6
a. ++ = Column sampling/comparison area > Row sampling/comparison area
at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area
at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling/comparison area
at 0.05 significance level.
— = Column sampling/comparison area < Row sampling/comparison area
at 0.01 significance level.
b. All test results reported are based on two-sided p-values.
c. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
d. Based on observations classified as Good.
WDR357/d.l2/2
-------
Table J-7a
RESULTS FOR NONPARAMETRIC MULTIVARIATE SAMPLING/COMPARISON AREA K -
TO SAMPLING/COMPARISON AREA COMPARISONS BASED ON ION RATIO-MODIFIED CONCENTRATIONS3'D'c'a
Sampling/
Comparison
Area 1 2
1
2 •*•+
3 ++ o
4 ++ ++
5 ++ ++
6 ++ ++
7 ++ ++
221 ++ ++
225 ++ ++
C&T ++ + +
EDA Sampling Area
345
__ H_ M_
o
—
++ O
++ O
+ + O 0
•n-
+ + — o
+ 00
6 7
— — ••_>
—
—
o ++
o +
+
-
+
o
++ ++
Comparison Areas
221
o
•n-
225
o
o
o
o
C&T
a. ++ = Column sampling/c»mparison area > Row sampling/comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling comparison area at 0.01 significance level.
b. The direction of the difference between the EDA and Comparison Areas is based on the sign of the sum of the elements of
the 8x1 vector of rank-sums (Volume III, Appendix B).
c. 221 = Census Tract 221.
225 = Census Tract 225.
CST = Cheektowaga and Tonawanda.
d. Based on observations classified as Good.
WDR357/d.l3/l
-------
Table J-7b
TWO-SIDED p-VALUES FOR NONPARAMETRIC
MOLTIVARIATE SAMPLING/COMPARISON AREA
TO SAMPLING/COMPARISON AREA COMPARISONS
BASED ON ION RATIO-MODIFIED CONCENTRATIONS3
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
EDA Sampling Areas
Comparison Areas
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.591
0.000
0.000
0.000
0.000
0.000
0.000
0.000
3
0.000
0.591
1.000
0.004
0.000
0.000
0.000
0.000
0.032
0.000
0.000
0.000
0.004
1.000
0.094
0.056
0.006
0.003
0.070
0.000
0.000
0.000
0.000
0.094
1.000
0.283
0.040
0.115
0.062
0.000
0.000
0.000
0.000
0.056
0.283
1.000
0.010
0.003
0.023
0.000
0.000
0.000
0.000
0.006
0.040
0.010
1.000
0.015
0.924
0.000
221
0.000
0.000
0.000
0.003
0.115
0.003
0.015
1.000
0.083
0.000
225
0.000
0.000
0.032
0.070
0.062
0.023
0.924
0.083
1.000
0.000
CST
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
a. 221 = Census Tract 221.
225 = Census Tract 225.
CST = Cheektowaga and Tonawanda.
b. Based on observations classified as Good.
WDR357/d.l4/l
-------
00000000
00000000
LEGEND:
fLCIC concentration in this EDA "~~N-...
sampling area is greater than that in the
comparison area at the Bonferroni corrected
0.05 level of significance (two-sided)
O No significant difference
I LCIC concentration in this EDA
I sampling area is less than that in the
* comparison area at the Bonferroni corrected
0.05 level of significance (two-sided)
Top Line = Census Tract 221
Middle Line = Census Tract 225
Bottom Line = Cheektowaga and Tonawanda
LCICs are in the following order on the line:
— TT "
3
0, ,f\ .1 r\ f\ T T T
tttttttt
5 1
2
s
1
Ott0tttt
Qtt0tttt
^^i
v. ^N:
»• m
Scale: 1" = 780'
ttt0tttt
tttptttt
ttt-tttt
SOURCE: Neighborhood boundaries adopted from
the Proposed Habitability Criteria document
(NYSDOH and DHHS/CDC, 1986)
Figure J-1
SOIL ASSESSMENT — INDICATOR CHEMICALS
SUMMARY OF UNIVARIATE COMPARISONS:
EDA SAMPLING AREA TO COMPARISON AREA
WITH BONFERRONI CORRECTION FOR NUMBER
OF LCICs
-------
CENSUS TRACT 225
CT221O O O O O O O O
t t t t t t
j !
CENSUS TRACT 221
j
CT225Q O O O O O O O
C&TO t t
t o t
CT221
CT225
4
4
4
4
@_
* ° *
4- 4
\ 4
4
LEGEND:
LCIC concentration in this
comparison area is greater at the
Bonferroni corrected 0.05 level of
significance (two-sided) than that
in the comparison area indicated
on the line.
No significant difference
LCIC concentration in this
comparison area is less at the
Bonferroni corrected 0.05 level of
significance (two-sided) than that
in the comparison area indicated
on the line.
LCICs are in the following order on the line:
Figure J-2
SOIL ASSESSMENT - INDICATOR CHEMICALS
SUMMARY OF UNIVARIATE COMPARISONS:
COMPARISON AREA TO COMPARISON AREA
WITH BONFERRONI CORRECTION FOR NUMBER
OF LCICs
-------
APPENDIX K
Summary of Sensitivity Results
by EDA Sampling Area
Soil Assessment—Indicator Chemicals
-------
Appendix K
SUMMARY OF SENSITIVITY RESULTS BY EDA SAMPLING AREA
SOIL ASSESSMENT—INDICATOR CHEMICALS
In response to a comment from Dr. David Schoenfeld at the
peer review meeting, this appendix presents the results of
the statistical comparisons between the EDA sampling areas
and the comparisons areas by sampling area. The results are
presented in Tables K-l through 7 (one table for each EDA
sampling area). Most of these results were presented in
Volume III; however, the results in Volume III were grouped
by LCIC rather than by sampling area.
It should be emphasized that the peer review panel has
recommended that the results of the original comparisons
(presented in Tables 6-8 through 6-10, K-l and K-2, and
Figures 6-9 through 6-11 of Volume III) should be the
primary basis for the habitability decision.
The results of the sensitivity analyses conducted on the
comparisons are also presented in these tables. The issues
addressed by each sensitivity analysis are explained below.
Tables K-4 through K-8 of Volume III present the results of
a sensitivity analysis that explores the effect of relaxing
the data quality control to include Uncertain data, as well
as the Good data. A description of the meaning of the data
qualifiers "Good" and "Uncertain" is provided in Appendix H
(page H-23) of Volume III.
The sensitivity analysis that includes data with the usabil-
ity flag of Uncertain explores the effect of expanding the
data base to include data that has some QA/QC problems. The
extent to which the comparison results change when the
expanded data set is used yields information on how
sensitive the comparison analysis is to the particular set
of data used.
Tables K-9 through K-ll of Volume III present the results of
a sensitivity analysis that explores the effect of classify-
ing all data below 1.0 ppb as non-detects. The purpose of
this sensitivity analysis, as stated in Volume III, is to
assess the sensitivity of the comparisons to the laboratory
detection limits. "If the results had changed dramatically
with the (artificially) increased detection limits, then it
could be concluded that the detection limits were exerting a
strong influence on the results" (Volume III, page 5-10).
This sensitivity analysis also addresses another issue,
determining the effects of outliers on the statistical
comparisons. Such a sensitivity analysis was requested by a
peer reviewer in the written preliminary comments (Dr. David
K-l
-------
Schoenfeld, page 3). Because most observations are less
than 1.0 ppb for most LCICs (except for TCB and TeCB; see
below) in the EDA sampling areas and comparison areas
(except for EDA Sampling Area 1), the effect of treating all
observations less than or equal to 1.0 ppb as non-detect is
to emphasize the characteristic of the Wilcoxon rank sum
test that evaluates for differences in the upper tails of
the distributions. That is, this modification can be used
to test whether there are more outlying high values in one
EDA sampling area or comparison area than in another.
For the LCICs TCB and TeCB, most observations are less than
2.0 ppb; thus, a peer reviewer (Dr. David Schoenfeld—Peer
Review Meeting Comment, June 21, 1988) requested that a
similar analysis be performed for these LCICs with the
cutoff at 2.0 ppb rather than 1.0 ppb.
At the request of the TRC, a sensitivity analysis was
performed that explores the effects of deleting the largest
10 percent of the observations for each LCIC in each EDA
sampling area and comparison area. The question this
sensitivity analysis proposes to answer is whether a few
large values in any particular EDA sampling area or
comparison area have a large influence on the outcome of the
statistical comparisons.
It was not possible to perform a meaningful multivariate
analysis based on an upper 10 percent trim of the multi-
variate observations, because an upper 10 percent trim is
not well-defined for a multivariate observation: for any
particular EDA sampling area or comparison area, the samples
with the largest 10 percent of the observations for one LCIC
may not correspond to the largest 10 percent of the observa-
tions for another LCIC.
At the request of the TRC, a sensitivity analysis was
performed that explores the effects of reclassifying
non-detects as detects if the only reason for the non-detect
classification was that the observed primary-to-secondary
ion ratio deviated from the theoretical ratio by more than
20 percent but not more than 40 percent. The purpose of
this sensitivity analysis is to explore the effects of LCIC
interferences on the statistical comparisons.
WDR363/001
K-2
-------
Table K-l
SUMMARY OF SENSITIVITY RESULTS FOR EDA SAMPLING AREA
Good
Comparison
LCIC Area
DCB
TCB
TeCB
CNP
a-BHC
d-BHC
b-BHC
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
Original And Uncertain
Median
Cone.
(ppb)
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
Median
Test Cone.
Result (ppb)
1.01
++ 0.40
++ 0.42
++ 0.36
8.67
++ 0.64
++ 0.60
++ 0.14
11.48
++ 0.55
++ 0.64
++ 0.05
ND
0 0.06
0.07
o 0.03
8.25
++ 0.18
++ 0.11
++ ND
1.13
++ ND
++ ND
++ ND
11.58
++ ND
++ ND
++ ND
Test
Result
1.01
++
++
++
8.67
++
++
++
12.93
++
++
++
ND
O
-
o
9.47
++
++
++
1.20
++
++
++
11.68
++
++
++
£1 ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
1.01
S2 ppb
Non-detect
Median
Cone. Test
Result
8.67
11.48
8.25
1.13
11.58
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
8.67
11.48
8.25
ND
11.58
Upper 10%
Trim
Median
Cone. Test
(ppb) Result
0.38
0.39
0.35
0.63
0.56
0.13
0.50
0.54
0.05
ND
ND
ND
ND
O
O
o
ND
ND
ND
ND
o
o
0
0.05
0.07
0.02
0.18
0.07
ND
ND
ND
ND
ND
ND
ND
0.94
7.80
11.34
ND
o
7.51
0.79
10.73
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
0.07
0.07
0.04
0.18
0.15
ND
ND
ND
ND
ND
0.12
ND
1.01
8.67
11.48
0.06
o
o
o
8.25
1.39
11.58
WDR363/002/7-13-88/1
-------
Table K-l
(Continued)
LCIC
g-BHC
ALL
Comparison
Area
221
225
C&T
221
225
C&T
Original
Median
Cone. Test
(ppb) Result
1.73
ND ++
ND ++
ND ++
Good
And Uncertain
Median
Cone. Test
(ppb) Result
1.73
ND ++
ND ++
ND ++
£1 ppb
Non-detect
Median
Cone. Test
(ppb) Result
1.73
ND ++
ND ++
ND ++
£2 ppb
Non-detect
Median
Cone. Test
(ppb) Result
ND
ND ++
ND ++
ND ++
Upper 10%
Trim
Median
Cone. Test
(ppb) Result
1.48
ND ++
ND ++
ND ++
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
1.48
ND ++
ND ++
ND ++
Could
Not be
Determined
Could
Not be
Determined
Could
Not be
Determined
++ = EDA sampling area > comparison area at 0.01 significance level.
+ = EDA sampling area > comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
= EDA sampling area < comparison area at 0.05 significance level.
— = EDA sampling area < comparison area at 0.01 significance level.
All test results reported are based on two-sided p-values.
°The first entry in each column is median concentration; ND indicates non-detect.
221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
eFor multivariate results (All), the direction of the difference between the EDA and the comparison
areas is based on the sine of the sum of the elements of the 8x1 vector of rank sums (Vol. Ill, Appendix B).
WDR363/002/7-13-88/2
-------
Table K-2
SUMMARY OF SENSITIVITY RESULTS FOR EDA SAMPLING AREA 2
a,b,c,d,e
Good
LCIC
DCB
TCB
TeCB
CNP
a-BHC
d-BHC
b-BHC
Original
Comparison
Area
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
Median
Cone.
(ppb)
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
Test
Result
0.39
o
o
++
0.91
++
++
++
1.17
++
++
++
0.04
0
-
o
0.40
++
++
++
0.04
++
++
++
0.20
++
o
++
And Uncertain
Median
Cone.
(ppb)
0.40
0.42
0.36
0.64
0.60
0.14
0.55
0.64
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
Test
Result
0.41
o
o
+
0.88
++
+
++
1.12
++
++
++
0.04
o
-
o
0.36
++
++
++
ND
++
++
++
0.19
++
o
++
SI ppb
Non-detect
S2 ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
ND
o
o
0
ND
++
o
++
1.17
++
++
++
ND
o
o
o
ND
++
0
++
ND
o
0
o
ND
++
o
++
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
ND
0
o
o
ND
+
o
++
ND
++
o
++
ND
o
0
o
ND
+
o
+
ND
0
O
o
ND
+
o
0
Upper 10%
Trim
Median
Cone. Test
(ppb) Result
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
0.38
0.39
0.35
0.63
0.56
0.13
0.50
0.54
0.05
0.35
o
o
0.83
1.09
ND
ND
ND
ND
ND
ND
ND
0.17
O
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
ND
ND
ND
ND
0.12
ND
0.39
o
o
0.91
1.17
ND
ND
ND
ND
ND
ND
ND
o
o
o
ND
++
0
++
ND
ND
ND
ND
ND
ND
ND
o
0
o
ND
+
O
+
0.05
0.07
0.02
0.18
0.07
ND
0.04
-
—
0
0.35
++
++
++
0.07
0.07
0.04
0.18
0.15
ND
0.05
-
o
o
0.40
++
++
++
0.10
0.29
o
WDR363/003/7-13-88/1
-------
LCIC
g-BHC
ALL
Comparison
Area
221
225
C&T
221
225
C&T
Original
Median
Cone.
(ppb)
ND
ND
ND
Test
Result
0.09
++
+
++
++
++
Good
And Uncertain
Median
Cone. Test
(ppb) Result
ND
ND
ND
Table K-2
(Continued)
Si ppb
Non-detect
Median
Cone.
(ppb)
0.05
++
++
++
++
ND
ND
ND
Test
Result
ND
o
o
No
Results
Here
S2 ppb
Non-detect
Upper 10%
Trim
Median
Cone.
(ppb)
ND
ND
ND
Median
Test Cone. Test
Result (ppb) Result
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
ND
o
o
o
0.03
ND
ND
ND
No
Results
Here
•H-
No
Results
Here
ND
ND
ND
0.09
++
++
++
++
++
++
++ = EDA sampling area > comparison area at 0.01 significance level.
+ = EDA sampling area > comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
= EDA sampling area < comparison area at 0.05 significance level.
— = EDA sampling area < comparison area at 0.01 significance level.
All test results reported are based on two-sided p-values.
°The first entry in each column is median concentration; ND indicates non-detect.
d221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda. 221 = Census Tract 221. 225 = Census Tract 225.
eFor multivariate results (ALL), the direction of the difference between the EDA sampling area and the comparison
area is based on the sine of the sum of the elements of the 8x1 vector of rank sums (Vol. Ill, Appendix B).
WDR363/003/7-13-88/2
-------
Table K-3
SUMMARY OF SENSITIVITY RESULTS FOR EDA SAMPLING AREA 3
a ,t> ,c, d ,e
Original
LCIC
DCB
TCB
Comparison
Area
221
225
C&T
221
225
C&T
Median
Cone.
(ppb)
0.39
0.43
0.36
0.65
0.60
0.13
Test
Result
0.40
0
o
++
0.89
+
+
++
Good
And Uncertain
Median
Cone.
(ppb)
0.40
0.42
0.36
0.64
0.60
0.14
Test
Result
0.42
o
o
++
0.89
++
+
++
Si ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
Test
Result
ND
o
o
o
ND
++
o
++
S2 ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
Test
Result
ND
o
o
o
ND
o
o
++
Upper 10%
Trim
Median
Cone.
(ppb)
0.38
0.39
0.35
0.63
0.56
0.13
Test
Result
0.39
o
o
++
0.79
+
++
++
Ion
Ratio Modified
Median
Cone.
(ppb)
0.39
0.43
0.36
0.65
0.60
0.13
Test
Result
0.40
o
o
++
0.89
+
+
++
TeCB
CNP
a-BHC
d-BHC
b-BHC
221
225
C&T
221
225
C&T
221
225
C&T
1.06
1.07
1.06
ND
0.94
0.53
0.61
0.05
0.55
0.64
0.05
ND
ND
ND
ND
ND
ND
0.50
0.54
0.05
ND
ND
ND
ND
++ ND
o ND
++ ND
ND
++ ND
0 ND
++ ND
ND
O
o
o
ND
ND
ND
ND
O
O
o
ND
ND
ND
ND
ND
ND
0.13
++
o
++
ND
ND
ND
0.07
++
o
++
ND
ND
ND
ND
ND
ND
ND
ND
O
ND
ND
ND
ND
ND
O
0.53
0.61
0.05
ND
ND
ND
ND
0.12
ND
1.06
221
225
C&T
221
225
C&T
0.06
0.07
0.03
0.18
0.11
ND
0.04
o
-
o
0.22
o
0
++
0.06
0.07
0.03
0.18
0.11
ND
ND
O
-
o
0.22
+
o
++
ND
ND
ND
ND
ND
ND
ND
o
o
0
ND
++
o
++
ND
ND
ND
ND
ND
ND
ND
0
O
o
ND
0
O
+
0.05
0.07
0.02
0.18
0.07
ND
ND
-
—
0
0.18
o
0
++
0.07
0.07
0.04
0.18
0.15
ND
0.05
-
—
o
0.24
+
o
++
ND
O
0.15
o
WDR363/004/7-13-88/1
-------
LCIC
g-BHC
ALL
Comparison
Area
221
225
C&T
221
225
CS.T
Original
Median
Cone. Test
(ppb) Result
ND
ND
ND
ND
Good
And Uncertain
Test
Result
ND
ND
ND
ND
Table K-3
(Continued)
Si ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
Test
Result
ND
o
o
ND
No
Results
Here
S2 ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
Test
Result
ND
0
0
o
Upper 10%
Trim
Median
Cone.
(ppb)
ND
ND
ND
Test
Result
ND
++
o
++
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
ND
ND
ND
ND
No
Results
Here
No
Results
Here
++ = EDA sampling area > comparison area at 0.01 significance level.
+ = EDA sampling area > comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = EDA sampling area < comparison area at 0.05 significance level.
— = EDA sampling area < comparison area at 0.01 significance level.
All test results reported are based on two-sided p-values.
°The first entry in each column is median concentration; ND indicates non-detect.
d221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
eFor multivariate results (ALL), the direction of the difference between the EDA sampling area and the
comparison area is based on the sine of the sum of the elements of the 8x1 vector of rank sums (Vol. Ill,
Appendix B).
WDR363/004/7-13-88/2
-------
Table K-4
SUMMARY OF SENSITIVITY RESULTS FOR EDA SAMPLING AREA 4
a,b,c,d,e
Good
LCIC
DCB
TCB
TeCB
CNP
a-BHC
d-BHC
b-BHC
Comparison
Area
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
Original
Median
Cone.
(ppb)
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
Test
Result
0.36
—
-
o
0.43
-
o
++
0.38
-
o
++
0.04
0
-
o
0.12
o
o
++
ND
+
o
++
ND
+
O
++
And Uncertain
Median
Cone.
(ppb)
0.40
0.42
0.36
0.64
0.60
0.14
0.55
0.64
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
Test
Result
0.38
—
o
o
0.45
-
o
++
0.42
o
o
++
0.04
o
-
o
0.12
o
o
++
ND
++
0
++
ND
+
o
++
Si ppb
Non-detect
S2 ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
ND
o
o
0
ND
o
0
++
ND
+
o
++
ND
o
o
o
ND
++
o
++
ND
o
o
0
ND
++
o
++
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
ND
o
o
o
ND
o
0
+
ND
+
o
++
ND
O
o
o
ND
+
o
++
ND
O
o
0
ND
++
0
+
Upper 10%
Trim
Median
Cone. Test
(ppb) Result
0.38
0.39
0.35
0.63
0.56
0.13
0.50
0.54
0.05
0.05
0.07
0.02
0.18
0.07
ND
ND
ND
ND
ND
ND
ND
0.33
o
0.41
o
0.35
0.04
o
o
o
0.08
o
o
ND
O
ND
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
0.07
0.07
0.04
0.18
0.15
ND
ND
ND
ND
ND
0.12
ND
0.36
o
0.44
o
0.39
o
o
0.04
o
0.13
o
o
ND
o
0.07
o
WDR363/005/7-13-88/1
-------
LCIC
g-BHC
ALL
Comparison
Area
221
225
CS.T
221
225
C6T
Original
Median
Cone. Test
(ppb) Result
ND
ND +
ND o
ND ++
Good
And Uncertain
Median
Cone. Test
(ppb) Result
ND
ND +
ND o
ND ++
Table K-4
(Continued)
£1 ppb
Non-detect
Median
Cone. Test
(ppb) Result
ND
ND o
ND o
ND o
S2 ppb
Non-detect
Median
Cone. Test
(ppb) Result
ND
ND o
ND 0
ND o
Upper 10%
Trim
Median
Cone. Test
(ppb) Result
ND
ND +
ND o
ND ++
Ratio
Ion
Modified
Median
Cone. Test
(ppb) Result
ND
ND
ND
ND
o
o
++
o
++
No
Results
Here
No
Results
Here
No
Results
Here
o
++
++ = EDA sampling area > comparison area at 0.01 significance level.
+ = EDA sampling area > comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = EDA sampling area < comparison area at 0.05 significance level.
— = EDA sampling area < comparison area at 0.01 significance level.
All test results reported are based on two-sided p-values.
°The first entry in each column is median concentration; ND indicates non-detect.
d221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
eFor multivariate results (ALL), the direction of the difference between the EDA sampling area and the
comparison area is based on the sine of the sum of the elements of the 8x1 vector of rank sums (Vol. Ill,
Appendix B).
WDR363/005/7-13-88/2
-------
Table K-5 . ,
SUMMARY OF SENSITIVITY RESULTS FOR EDA SAMPLING AREA 5a'D'c'a'e
Good
LCIC
DCB
TCB
TeCB
CNP
a-BHC
d-BHC
b-BHC
Original
Comparison
Area
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
Median
Cone.
(ppb)
0.39
0.43 "
0.36
0.65
0.60
0.13
0.53
0.61
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
Test
Result
0.39
0
o
o
0.52
0
o
++
0.43
o
o
++
0.06
o
-
o
0.13
o
o
++
ND
+
O
++
ND
O
o
++
And Uncertain
Median
Cone.
(ppb)
0.40
0.42
0.36
0.64
0.60
0.14
0.55
0.64
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
Test
Result
0.39
o
o
0
0.57
o
o
++
0.46
o
o
++
0.06
o
-
o
0.14
o
o
++
ND
+
o
++
ND
o
o
++
sl ppb
Non-detect
S2 ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
ND
o
o
0
ND
O
o
++
ND
++
O
++
ND
o
o
0
ND
+
o
++
ND
0
0
o
ND
+
0
+
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
ND
0
O
o
ND
+
0
++
ND
++
O
++
ND
O
o
o
ND
+
0
+
ND
0
o
o
ND
+
o
o
Upper 10%
Trim
Median
Cone. Test
Db) Result
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
0.38
0.39
0.35
0.63
0.56
0.13
0.50
0.54
0.05
0.36
o
o
o
0.44
o
o
0.38
o
o
ND
ND
ND
ND
ND
ND
ND
O
ND
O
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
ND
ND
ND
ND
0.12
ND
0.39
o
o
o
0.50
o
o
0.43
o
o
ND
ND
ND
ND
o
o
0
ND
ND
ND
ND
O
o
o
0.05
0.07
0.02
0.05
o
-
o
0.07
0.07
0.04
0.06
o
-
0
0.18
0.07
ND
0.06
o
0
++
0.18
0.15
ND
0.14
o
0
++
ND
O
ND
o
o
WDR363/006/7-13-88/1
-------
Table K-5
(Continued)
LCIC
g-BHC
ALL
Comparison
Area
221
225
C&T
221
225
C&T
Original
Median
Cone.
(ppb)
Test
Result
Good
And Uncertain
Median
Cone.
(ppb)
Test
Result
Si ppb
Non-detect
Median
Cone.
(ppb)
Test
Result
S2 ppb
Non-detect
Median
Cone.
(ppb)
Test
Result
Upper 10%
Trim
Median
Cone.
(ppb)
Test
Result
Ion
Ratio Modified
Median
Cone.
(ppb)
Test
Result
ND
ND
ND
ND
o
o
ND
ND
ND
ND
o
o
ND
ND
ND
ND
o
o
No
Results
Here
ND
ND
ND
ND
o
o
ND
ND
ND
ND
o
o
ND
ND
ND
No
Results
Here
No
Results
Here
ND
O
o
o
o
++ = EDA sampling area > comparison area at 0.01 significance level.
+ = EDA sampling area > comparison area at 0.05 significance .level.-
o = No significant difference at 0.05 significance level.
- = EDA sampling area < comparison area at 0.05 significance level.
— = EDA sampling area < comparison area at 0.01 significance level.
All test results reported are based on two-sided p-values.
cThe first entry in each column is median concentration; ND indicates non-detect.
d221 = Census Tract 221.
225 = Census Tract 225.
C6T = Cheektowaga and Tonawanda.
eFor multivariate results (ALL), the direction of the difference between the EDA sampling area and the comparison
area is based on the sine of the sum of the elements of the 8x1 vector of rank sums (Vol. Ill, Appendix B).
WDR363/006/7-13-88/2
-------
Table K-6
SUMMARY OF SENSITIVITY RESULTS FOR EDA SAMPLING AREA
Good
ppb
LCIC
DCB
TCB
TeCB
CNP
a-BHC
d-BHC
b-BHC
Original
Comparison
Area
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
Median
Cone.
(ppb)
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
Test
Result
0.36
—
o
0
0.37
—
—
++
0.31
—
—
++
0.05
0
-
o
0.07
-
o
++
ND
o
o
++
ND
o
—
++
And Uncertain
Median
Cone.
(ppb)
0.40
0.42
0.36
0.64
0.60
0.14
0.55
0.64
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
Test
Result
0.36
-
o
0
0.38
—
—
++
0.32
--
—
++
0.05
o
-
o
0.07
-
o
++
ND
0
0
+
ND
o
—
++
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
ND
o
o
o
ND
—
—
0
ND
o
—
o
ND
o
o
0
ND
0
-
O
ND
O
o
o
ND
o
o
o
S2 ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
ND
o
o
o
ND
O
—
0
ND
o
—
0
ND
0
o
o
ND
O
—
o
ND
O
o
0
ND
O
0
o
Upper 10%
Trim
Median
Cone. Test
(ppb) Result
0.38
0.39
0.35
0.63
0.56
0.13
0.50
0.54
0.05
0.05
0.07
0.02
0.18
0.07
ND
ND
ND
ND
ND
ND
ND
0.33
o
o
0.36
0.28
0.05
o
o
o
0.06
o
ND
O
o
o
ND
O
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
0.07
0.07
0.04
0.18
0.15
ND
ND
ND
ND
ND
0.12
ND
0.36
o
o
0.37
0.31
0.07
o
o
o
0.10
ND
o
o
ND
o
WDR363/007/7-13-88/1
-------
LCIC
g-BHC
ALL
Comparison
Area
221
225
C&T
221
225
CS.T
Original
Median
Cone. Test
(ppb) Result
ND
ND
ND
ND
o
Good
And Uncertain
Median
Cone.
(ppb)
ND
ND
ND
Test
Result
ND
o
Table K-6
(Continued)
Si ppb
Non-detect
Median
Cone. Test
(ppb) Result
<2 ppb
Non-detect
Upper 10%
Trim
ND
ND
ND
ND
o
No
Results
Here
Median
Cone.
(ppb)
ND
ND
ND
Median
Test Cone. Test
Result (ppb) Result
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
ND
o
o
o
ND
ND
ND
ND
O
ND
ND
ND
ND
o
o
No
Results
Here
No
Results
Here
++ = EDA sampling area > comparison area at 0.01 significance level.
+ = EDA sampling area > comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
= EDA sampling area < comparison area at 0.05 significance level.
— = EDA sampling area < comparison area at 0.01 significance level.
All test results reported are based on two-sided p-values.
°The first entry in each column is median concentration; ND indicates non-detect.
d221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
eFor multivariate results (ALL), the direction of the difference between the EDA sampling area and the comparison
area is based on the sine of the sum of the elements of the 8x1 vector of rank sums (Vol. Ill, Appendix B) .
WDRJbJ/UOV/ /-l_S-«B/<2
-------
Table K-7
SUMMARY OF SENSITIVITY RESULTS FOR EDA SAMPLING AREA 1 ' ' ' '
LCIC
DCS
TCB
TeCB
CNP
a-BHC
d-BHC
b-BHC
Original
Comparison
Area
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
Median
Cone.
(ppb)
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
Test
Result
0.41
0
o
+
0.58
o
o
++
0.54
o
o
++
0.06
o
-
o
0.14
0
o
++
ND
o
o
++
ND
0
0
++
Good
And Uncertain
Median
Cone. Test
(ppb) Result
£1 ppb
Non-detect
S2 ppb
Non-detect
0.40
0.42
0.36
0.64
0.60
0.14
0.55
0.64
0.05
0.41
o
o
0.59
o
o
0.55
o
o
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
ND
-
o
o
ND
o
o
++
ND
++
O
++
ND
O
o
o
ND
o
—
o
ND
• o
o
o
ND
o
o
o
221
225
C&T
221
225
C&T
0.06
0.07
0.03
0.18
0.11
ND
0.06
o
-
o
0.14
0
o
++
0.06
0.07
0.03
0.18
0.11
ND
0.05
o
-
o
0.15
o
o
++
ND
ND
ND
ND
ND
ND
ND
O
O
o
ND
o
—
o
ND
ND
ND
ND
ND
ND
ND
o
0
0
ND
0
-
O
0.05
0.07
0.02
0.18
0.07
ND
0.05
o
0
+
0.13
0
o
++
0.07
0.07
0.04
0.18
0.15
ND
0.06
o
o
++
0.17
o
0
++
ND
ND
ND
ND
ND
ND
ND
o
o
++
ND
0
0
++
ND
ND
ND
ND
ND
ND
ND
o
o
++
ND
o
o
++
ND
ND
ND
ND
ND
ND
ND
• o
o
o
ND
o
o
o
ND
ND
ND
ND
ND
ND
ND
o
o
o
ND
O
o
o
ND
ND
ND
ND
ND
ND
ND
0
O
O
ND
+
O
++
ND
ND
ND
ND
0.12
ND
ND
0
O
++
0.10
+
0
++
Median
Cone.
(ppb)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Test
Result
ND
o
0
o
ND
o
-
o
ND
O
-
0
ND
o
0
0
ND
0
-
o
ND
o
o
o
ND
O
o
o
Upper 10%
Trim
Median
Cone. Test
(ppb) Result
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
0.38
0.39
0.35
0.63
0.56
0.13
0.38
o
o
0.54
o
o
0.39
0.43
0.36
0.65
0.60
0.13
0.41
o
o
0.58
o
o
0.50
0.54
0.05
0.49
o
0
++
0.53
0.61
0.05
0.54
o
o
++
WDR363/008/7-13-88/1
-------
Table K-7
(Continued)
LCIC
g-BHC
ALL
Comparison
Area
221
225
C&T
221
225
C6.T
Original
Median
Cone.
(ppb)
ND
ND
ND
Test
Result
ND
o
o
++
Good
And Uncertain
Median
Cone.
!£pb)_
ND
ND
ND
Test
Result
ND
o
o
++
£1 ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
Test
Result
ND
o
-
o
S2 ppb
Non-detect
Median
Cone.
(ppb)
ND
ND
ND
Test
Result
ND
o
o
o
Upper 10%
Trim
Median
Cone. Test
Result
++
o
++
o
++
No
Results
Here
ND
ND
ND
No
Results
Here
Ion
Ratio Modified
Median
Cone. Test
(ppb) Result
ND
o
o
ND
ND
ND
ND
O
O
No
Results
Here
++ = EDA sampling area > comparison area at 0.01 significance level.
+ = EDA sampling area > comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
= EDA sampling area < comparison area at 0.05 significance level.
— = EDA sampling area < comparison area at 0.01 significance level.
All test results reported are based on two-sided p-values.
The first entry in each column is median concentration; ND indicates non-detect.
d221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
eFor multivariate results (ALL), the direction of the difference between the EDA sampling area and the comparison
area is based on the sine of the sum of the elements of the 8x1 vector of rank sums (Vol. Ill, Appendix B).
WDR363/008/7-13-88/2
-------
APPENDIX L
Spatial and Correlation Structures
of LCICs in the EDA
Soil Assessment—Indicator Chemicals
WDO R3334.T1 !SV
-------
Appendix L
SPATIAL AND CORRELATION STRUCTURES OF LCICs IN THE EDA
SOIL ASSESSMENT—INDICATOR CHEMICALS
Although the Love Canal soil assessment for indicator
chemicals was not designed to analyze spatial and correla-
tion structures (i.e., the assessment was concerned with
comparing averages rather than with identifying hot spots),
questions about such structures or patterns were raised
during the peer review. While this appendix presents spa-
tial and correlation structure analyses, the results of
these analyses should be regarded as qualitative.
The correlation structure of the LCICs was examined to indi-
cate the degree of association among the LCICs. Since both
univariate and multivariate comparison analyses were con-
ducted, knowledge of the correlations among the LCICs indi-
cates how much weight should be given to multiple signifi-
cant univariate differences.
The spatial structure of the LCICs was examined for
indications that the comparison design assumptions were not
violated. The sampling strategy of the comparison study was
based on the assumption (on the basis of pilot study
results) that there was no localized contamination by LCICs
in the EDA (i.e., contamination, if present, would occur
uniformly over the EDA). This meant that samples could be
randomly allocated to each neighborhood and that the results
of these random samples would be representative of the
neighborhood.
One method of examining the spatial structure is to
correlate LCICs with distance from the canal. Since one
hypothesis for LCIC concentrations is that the LCICs
migrated from the Canal, significant negative correlations
would indicate that such a hypothesis might be valid.
A second method of examining the spatial structure is to
generate semivariograms for each sampling area. This tech-
nique, adapted from Kriging, indicates whether sampled
points correlated more with nearby points than those more
distant. A consistent pattern in the semivariograms would
indicate spatial structure at a resolution smaller than the
sampling area.
Correlations among LCICs and analyses using the two methods
of examining spatial structure are presented in separate
sections below.
L-l
-------
CORRELATIONS AMONG LCICs
Table L-l presents Spearman rank (r) correlations among LCIC
concentrations for all samples taken in both the EDA and the
comparison areas. The concentrations of members of the same
chemical family exhibit a relatively high degree of correla-
tion. For example, the correlations among the BHC compounds
(A, D, G, B) all exceed 0.5, with the correlation between A-
and B-BHCs the strongest (r=0.71). The chlorobenzene
compounds are also correlated, with the correlation
strongest between TCB and TeCB (r=0.94), and weakest between
DCB and TeCB (r=0.38). The most interesting aspect of the
correlation matrix, however, is the observed relationship
between members of different chemical families. For ex-
ample, TCB and TeCB are highly correlated (r>0.75) with
A-BHC. The correlations between these chlorobenzenes and
the other BHC compounds are somewhat weaker, but not
inconsequential (r>0.5). Plots demonstrating some of these
relationships are given in Figures L-l through 3. Note
that, prior to plotting, the concentrations were transformed
to the natural logarithm scale.to reduce the influence of a
single, large observation on the scale of the plot. In
addition, 1 was added to each concentration prior to the
logarithmic scaling. Thus, zero in the original scale
transforms to zero in log scale.
The first question that might be asked is whether these cor-
relations are related to the measurement process itself. In
particular, one might ask whether the correlations are re-
lated to laboratory differences. This seems highly unlikely
considering the magnitude of the correlations. However, to
examine this possibility, laboratory-specific correlation
matrices are presented in Tables L-2 through 8. From these
matrices, we see that the correlation pattern is remarkably
stable among laboratories. The relationship among the
chlorobenzenes is similar, with the extraordinarily high
(>0.9) correlation between TCB and TeCB common to all the
laboratories. In addition, the pattern among the BHCs is
similar in all the laboratories, and the relationship
between the chlorobenzenes and BHCs is seen in each
laboratory.
In the correlation matrices examined thus far, all samples
were included irrespective of whether they came from the EDA
or from the comparison areas. Because significant differ-
ences were found between the EDA sampling areas and the
comparison areas, correlation matrices for the EDA and each
of the comparison areas are given in Tables L-9 through 12.
From these matrices, it is shown that the combined
comparison area of Cheektowaga and Tonawanda exhibits strong
correlations among the chlorobenzenes, although it appears
that the correlations are weaker than those seen in the EDA.
L-2
-------
Because of the absence of the BHCs at any appreciable level
in the Cheektowaga and Tonawanda samples, the correlations
among these compounds are very low. For the same reason,
the correlations between the BHCs and the chlorobenzenes are
also smaller than those seen in the EDA.
Like the combined sampling area of Cheektowaga and
Tonawanda, Census Tract 221 exhibits a correlation pattern
among the chlorobenzenes that is quite similar to that found
in the EDA. It also shows correlations between
chlorobenzenes and the BHC compounds similar to those seen
in EDA samples. Although the correlations are not as large
as those from the EDA, there is evidence that a systematic
relationship exists between the two families of chemicals in
samples from Census Tract 221. Census Tract 225's correla-
tion matrix is remarkably similar to that of the EDA. Al-
though some of the correlations between the BHCs and the
chlorobenzenes are not as strong as those seen in the EDA,
the overall pattern is the same, namely, strong correlations
between chlorobenzenes and BHC compounds.
In addition to the correlation analyses for laboratories and
sampling areas, correlation matrices were obtained for com-
binations of laboratories and sampling areas. These
analyses showed no apparent differences among the labora-
tories in any of the three comparison areas or the EDA. The
correlation analyses provide evidence that there are fairly
strong correlations among the chlorobenzenes used as LCICs.
The relationship among these compounds exists and is of
similar strength in samples taken from both the EDA and all
the comparison areas. There is also evidence of somewhat
weaker correlations among the BHC compounds in both the EDA
and two of the comparison areas. Furthermore, one of the
comparison areas exhibited correlations among the compounds
that were nearly identical to those observed in EDA samples.
Since the comparison areas exhibit similar relationships
among the LCICs, this reduces the validity of using con-
sistency of results across the LCICs as evidence of migra-
tion. The results of statistical analyses of individual
LCICs should be interpreted with caution. If, for example,
a comparison of TCB levels in an EDA sampling area and in
Census Tract 225 yields a statistically significant result,
we should not expect a similar result for TeCB to add much
to the weight of evidence concerning contamination of the
area. Thus, the multivariate test results, which take the
correlation structure in the LCICs into account, are the
most appropriate tools for assessing differences between the
EDA and the comparison areas.
L-3
-------
CORRELATION OF LCICs WITH DISTANCE FROM THE CANAL
To ascertain whether the levels of LCICs can be related to
proximity to the Canal, an analysis using simple correlation
techniques was conducted. To implement this analysis,
individual data were mapped to obtain a general sense of the
spatial orientation of the data. The concentrations for
local groups of values of each LCIC were averaged on a grid
encompassing the EDA and Census Tract 225. The comparison
area was included to assess evidence of large-scale spatial
trends across the EDA and the closest comparison area. Maps
of the EDA and the closest comparison area (Census
Tract 225) are shown in Figures L-4 through 11. In general,
each plotted value represents an average of four or five
sampling locations.
While there is little visual evidence that the LCIC concen-
trations decrease with distance from the Canal, a simple
analysis to verify this initial conclusion was performed to
assess the strength of spatial gradients in LCIC concentra-
tions. Correlations of LCIC concentrations at the sampling
locations with their distances from the center of the Canal
and to the nearest boundary of the Canal were computed. The
results are given below.
CORRELATION OF LCIC CONCENTRATION WITH
DISTANCE FROM CENTER OF THE LOVE CANAL
Spearman
Che-nical Correlation
DCB 0.100 0.028
TCB 0.092 0.042
TeCB 0.058 0.202
CNP -0.002 0.954
A-BHC 0.138 0.002
D-BHC 0.044 0.331
B-BHC 0.139 0.002
G-BHC 0.116 0.011
L-4
-------
CORRELATION OF LCIC CONCENTRATION WITH
DISTANCE FROM NEAREST THE CANAL BOUNDARY
Spearman
Ch«dc.l correlation
DCB 0.008 0.859
TCB -0.056 0.208
TeCB -0.075 0.092
CNP 0.079 0.086
A-BHC -0.057 0.199
D-BHC -0.141 0.002
B-BHC -0.069 0.135
G-BHC -0.074 0.094
This analysis gives very little evidence that LCIC concen-
trations decrease with increasing distance from the Love
Canal. The analysis relating concentration level to
distance from the center of the Canal gave no evidence of
this relationship, with correlations that were either
positive or indistinguishable from zero. The analysis based
on distance from the boundary of the Canal gives some weak
evidence of a negative relationship. However, even the one
correlation that is significantly different from zero
(p<0.01) is quite small. The high degree of statistical
significance associated with the rejection of the zero
correlation hypothesis for such small correlations is the
result of the large sample size. As an example, a plot of
G-BHC concentration versus distance from the Canal
(Figure L-12) does not provide a convincing impression of a
relationship between distance and concentration.
KRIGING ANALYSIS OF LCIC CONCENTRATIONS
One of the assumptions made in the statistical analysis of
the soil chemistry data was that the data were independently
distributed under the null hypothesis (no difference in soil
chemistry between sampling areas and comparison areas).
Essentially, this is equivalent to saying that the data have
no spatial structure, for instance, the soil LCIC concentra-
tions do not vary within a sampling or comparison area in
such a way that knowledge of the concentration and location
of a sample would allow prediction of the concentration of a
nearby sample. Inspection of the sampling results and the
site locator map given in Appendix I of Volume III of the
final report suggests that this is a reasonable assumption.
However, given the interest in possible spatial patterns in
the soil chemistry results, a more detailed analysis was
undertaken.
L-5
-------
One common approach to the analysis of spatial data is
random field theory, a special case of which is Kriging.
Random field theory is discussed in a number of textbooks.
The general idea is that the structure of a set of spatial
data (that is, observations with which a spatial location,
such as the sample site coordinates, can be associated) can
be described in terms of a function known as the variogram,
•y (h)=E [Z (x,) -Z (x_) ] , where h= | |x. -x_ | | . A best linear un-
biased estimator of the value of the random field at any
unmeasured location can be developed as a function of the
known observations and y(h). In this form, the random field
Z(x) is assumed to be homogeneous and isotropic, that is, y
depends only on h and not on the orientation or specific
location of the samples. More general formulations' of the
variogram are possible. For this exploratory analysis, how-
ever, an isotropic condition was assumed. In the special
case of a stationary random field where the correlation be-
tween observations with separation distance h is B(h), it
can easily be shown that y(h)=a [l-p(h)], where a is the
variance of the random field. If the data have no spatial
structure, that is, the observations are independent of the
separation distance, the variogram is constant.
In addition to a stationary field, the variogram is applica-
ble to a more general case where there is an underlying lin-
ear "drift" or trend surface. Because the variogram is
based on a difference between observations and not their
products, its estimator is not confounded by a first-order
trend. A more general formulation, which uses a generalized
covariance rather than the variogram, is applicable to cases
in which the trend surface can be described by higher-order
multinomials. One might question the implicit assumption of
a linear trend surface, given the complexity of most spatial
processes. In practice, though, it is only the behavior of
the random function in a neighborhood containing the few
nearest observations that is important. For the Love Canal
soil sampling data, this is typically a few hundred feet.
Variograms were estimated for each of the LCICs within each
sampling area (using the original sampling area designations
described in Volume III of the Final Report). Variogram
estimation is usually accomplished by separating data pairs
into distance classes in such a way that there are enough
pairs within each distance class to avoid excessive sampling
variability. For n observations, there are no (n-l)/2
unique pairs, so even for a sample size of 50 there are over
1,000 pairs. For the Love Canal soil chemistry data,
variogram estimation was complicated by the differences
among laboratories, as discussed in the Volume III of the
final report. Therefore, the following approach was taken.
First, within a given area, the distances corresponding to
each data pair (regardless of the laboratory that performed
the analysis) were computed. Next, the 10th, 20th, ... 90th
L-6
-------
percentiles of the separation distances were computed.
Then, for all samples collected by a given laboratory, the
appropriate distance class was identified. Non-detect
observations were replaced with a uniformly distributed
random number in the range between zero and an assumed
effective detection limit of 0.5 ppb (the use of a random
distribution in place of the non-detectable values properly
reflects lack of knowledge of the true value; use of any
fixed value (for instance, zero or the detection limit) ,
would bias the estimate of the variogram) . In addition, the
variogram estimation was performed on the logarithms of the
observations, rather than the raw observations, to avoid
giving disproportionate weight to the large observations.
Finally, the sums of squares of differences in observations
within each laboratory were computed and averaged to provide
the variogram estimator,
2
where S. is the squared difference associated with observa-
tion pair i. This variogram estimator was consistent with
the earlier analysis in that observations collected by dif-
ferent laboratories were not compared directly. The number
of observation pairs in each distance class was substantial-
ly reduced over what would have been available if across-
laboratory comparisons could be made; however, it was still
adequate for exploratory analysis. For most sampling areas,
the number of observation pairs in each distance class
ranged from 15 to 40. In cases where the number of observa-
tions in a distance class was less than 10, distance classes
were aggregated until this minimum was achieved.
The results are shown in Figures L-13 through 20. Given the
number of observation pairs in each distance class, virtu-
ally all the sample variograms suggest an absence of spatial
structure (constant variogram) . It should be noted that
there is not a formal test to determine, for instance,
whether the sample variogram is nonconstant. Therefore, the
results must be viewed entirely as exploratory. Note that
y(0)=0 by definition, so no significance should be attached
to the joining of -y(O) and y (h.. ) .
Inspection of Figures L-13 through 20 indicates the
following LCICs and areas for which the variogram is
apparently not flat and for which there is some possible
spatial structure:
o DCB: EDA5, EDA7 , 221
o TCB: EDA2, EDA4 , 221
o TeCB: EDA3, EDA4 , 221
o 2-CNP: EDA7
L-7
-------
o
o
o
o
A-BHC :
B-BHC:
D-BHC:
G-BHC:
EDA2,
EDA2
221
EDA1,
221
221
Overall, only 16 of the 80 possible area-LCIC combinations
showed any apparent spatial structure. It is of interest
that of the 16, Census Tract 221, a comparison area,
appeared six times. This suggests that there may be some
smooth variation in the concentrations of many of the LCICs
within Census Tract 221, but that for most of the EDA
sampling areas the concentrations were more or less random,
and the assumption of statistical independence is warranted.
Two comments should be made. First, all the observations
are subject to measurement error. Because of the randomiza-
tion in the experimental design, the measurement error
should be statistically independent. Its effect is to
dilute any spatial structure that might be present in the
data. Second, the variogram analysis does not reflect the
magnitude of the concentrations within a given area, nor
were variations analyzed among sampling areas. It is clear,
for instance, that Sampling Area (Neighborhood) 1 had
substantially higher concentrations of most LCICs than did
the other areas. The effect of the analysis is only to
confirm that within Sampling Area 1, the observations were
essentially random. Also, because the analysis was
conducted within areas, the maximum separation distance
class was limited to about 1,500 feet. If the analysis were
conducted for all sampling areas simultaneously, some of the
between-area differences detected using the Wilcoxon rank
sum test might be reflected in the variogram shape at
distances on the order of the typical distance among areas.
WDR363/009
L-8
-------
Table L-1 LOVE CANAL CORRELATION ANALYSIS
VARIABLE
MEAN
STD DEV
MEDIAN
MINIMUM
SPEARMAN CORRELATION COEFFICIENTS / PROD > |R| UNDER HO:RHO=0 / NUMBER OF OBSERVATIONS
CONC1 CONC2 CONC3 CONC4 CONC5 CONC6 CONC7 CONC6
CONC1
DCB Cone
CONC2
TCB Cone
CONC3
OCB Cone
CONC4
CN Cone
CONC5
A-BHC Cone
CONC6
D-BHC Cone
CONC7
B-BHC Cone
CONCe
6-8HC Cone
1.00000
0.0000
654
0.48077
0.0001
651
0.38408
0.0001
643
0.22939
0.0001
608
0.35283
0.0001
653
0.24113
0.0001
646
0.34725
0.0001
608
0.25078
0.0001
648
0.48077
0.0001
651
1.00000
0.0000
683
0.94244
0.0001
672
-0.04135
0.2977
636
0.75998
0.0001
681
0.50081
0.0001
669
0.62668
0.0001
630
0.54945
0.0001
677
0.38408
0.0001
643
0.94244
0.0001
672
1.00000
0.0000
675
-0.06273
0.1149
633
0.76969
0.0001
673
0.52533
0.0001
661
0.64084
0.0001
622
0.56983
0.0001
669
0.22939
0.0001
608
-0.04135
0.2977
636
-0.06273
0.1149
633
1.00000
0.0000
639
0.02362
0.5515
638
0.07598
0.0574
626
-0.01597
0.6997
586
0.02768
0.4870
633
0.35283
0.0001
653
0.75998
0.0001
681
0.76969
0.0001
673
0.02362
0.5515
638
1.00000
0.0000
684
0.53080
0.0001
671
0.71231
0.0001
631
0.62400
0.0001
678
0.24113
0.0001
646
0.50081
0.0001
669
0.52533
0.0001
661
0.07598
0.0574
626
0.53080
0.0001
671
1.00000
0.0000
672
0.55897
0.0001
625
0.55018
0.0001
666
0.34725
0.0001
608
0.62668
0.0001
630
0.64084
0.0001
622
-0.01597
0.6997
586
0.71231
0.0001
631
0.55897
0.0001
625
1.00000
0.0000
633
0.59500
0.0001
631
0.25078
0.0001
648
0.54945
0.0001
677
0.56983
0.0001
669
0.02766
0.4670
631
0.62400
0.0001
678
0.55016
0.0001
666
0.59500
0.0001
631
1.00000
0.0000
680
MAXIMUM
CONC1
CONC2
CONC3
1 CONC4
CONC5
CONC6
CONC7
COMC8
654
683
675
639
664
672
633
680
0.56125382
1.98970717
2.73677037
0.05275430
1.90288012
0.36209821
12.80699842
0.76722059
0.90064293
7.83548659
11.69210578
0.05134786
10.34717751
3.50556707
179.29544162
5.24521318
0.39000000
0.57000000
0.52000000
0.05000000
0.15000000
0.00000000
0.00000000
0.00000000
0
0
0
0
0
0
0
0
19.60000000
167.33000000
182.41000000
0.32000000
152.53000000
79.99000000
4108.31000000
85.60000000
-------
Table L-2 LOVE CANAL CORRELATION ANALYSIS
LAB1
VARIABLE
N
MEAN
STD DEV
MEDIAN
MINIMUM
MAXIMUM
CONC1
CONC2
CONC3
CONC4
CONC5
CONC6
CONC7
CONC6
96
117
117
117
117
117
109
116
0.49333333
1.05384615
1.35282051
0.08555556
1.89393162
0.12358974
39.02220183
1.79758621
0.24556987
1.52279113
2.80981281
0.03806629
12.18321959
0.33559895
393.42723777
10.95619393
0.42000000
0.52000000
0.47000000
0.09000000
0.15000000
0.00000000
0.07000000
0.00000000
0
0
0
0
0
0
0
0
1.73000000
8.86000000
18.08000000
0.17000000
129.90000000
2.08000000
4108.31000000
85.60000000
SPEARMAN CORRELATION COEFFICIENTS / PROB > iRl UNDER HO:RHO=0 / NUMBER OF OBSERVATIONS
CONC1 CONC2 CONC3 CONC4 CONC5 CONC6 CONC7 CONC8
CONC1
DCS Cone
CONC2
TCB Cone
CONC3
QCB Cone
CCNC4
CN Cone
CONC5
A-BHC Cone
CONC6
D-BHC Cone
CONC7
B-BHC Cone
CONC8
6-BHC Cone
1.00000
0.0000
96
0.72155
0.0001
96
0.64582
0.0001
96
0.41858
0.0001
96
0.59263
0.0001
96
0.53331
0.0001
96
0.49668
0.0001
95
0.40628
0.0001
95
0.72155
0.0001
96
1.00000
0.0000
117
0.91819
0.0001
117
0.25281
0.0060
117
0.63011
0.0001
117
0.62018
0.0001
117
0.61181
0.0001
109
0.65172
0.0001
116
0.64582
0.0001
96
0.91819
0.0001
117
1.00000
0.0000
117
0.20240
0.0286
117
0.87650
0.0001
117
0.63946
0.0001
117
0.65465
0.0001
109
0.72835
0.0001
116
0.41858
0.0001
96
0.25261
0.0060
117
0.20240
0.0286
117
1.00000
0.0000
117
0.27219
0.0030
117
0.34835
0.0001
117
0.20593
0.0317
109
0.30209
0.0010
116
0.59263
0.0001
96
0.63011
0.0001
117
0.87650
0.0001
117
0.27219
0.0030
117
1.00000
0.0000
117
0.63827
0.0001
117
0.73307
0.0001
109
0.72480
0.0001
116
0.53331
0.0001
96
0.62018
0.0001
117
0.63946
0.0001
117
0.34835
0.0001
117
0.63827
0.0001
117
1.00000
0.0000
117
0.62393
0.0001
109
0.61863
0.0001
116
0.49666
0.0001
95
0.61181
0.0001
109
0.65465
0.0001
109
0.20593
0.0317
109
0.73307
0.0001
109
0.62393
0.0001
109
1.00000
0.0000
109
0.56275
0.0001
108
0.40628
0.0001
95
0.65172
0.0001
116
0.72635
0.0001
116
0.30209
0.0010
116
0.72480
0.0001
116
0.61663
0.0001
116
0.56275
0.0001
108
1.00000
0.0000
116
-------
nIABLE
Table L-3 LOVE CANAL CORRELATION ANALYSIS
LABS
MEAN
STD DEV
MEDIAN
MINIMUM
SPEARMAN CORRELATION COEFFICIENTS / PROB > |R| UNDER HO:RHO=0 / NUMBER OF OBSERVATIONS
CONC1 CONC2 CONC3 CONC4 CONC5 CONC6 CONC7 CONC6
MAXIMUM
:iCl
:iC2
.:c3
;iC4
:iC5
:i-6
:iC7
.ICO
101
102
101
65
100
101
102
102
0.82089109
1.4206B627
1.70663366
0.07152941
2.17540000
0.96930693
19.21294110
1.05960784
0.32880724
2.63586507
4.00550378
0.06142176
15.28452994
7.V8427671
171.32464748
5.37379614
0.75000000
0.47500000
0.39000000
0.08000000
0.07000000
0.00000000
0.00000000
0.00000000
0.00000000
0.11000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
2.46000000
13.84000000
23.0900COOO
0.21000000
152.53000000
79.99000000
1729.10000000
47.05000000
CONCl
DCB Cone
CONC2
TCB Cone
CONC3
QCB Cone
CONC4
CN Cone
CONC5
A-BHC Cone
CONC6
D-BHC Cone
CONC7
B-BHC Cone
CONC8
6-BHC Cone
1.00000
0.0000
101
0.41565
0.0001
101
0.32644
0.0009
100
-0.11316
0.3024
85
0.22286
0.0258
100
0.19067
0.0561
101
0.17319
0.0833
101
0.33633
0.0006
101
0.41565
0.0001
101
1.00000
0.0000
102
0.94181
0.0001
101
-0.29922
0.0054
85
0.75993
0.0001
100
0.31703
0.0012
101
0.61528
0.0001
102
0.55087
0.0001
102
0.32644
0.0009
100
0.94181
0.0001
101
1.00000
0.0000
101
-0.41319
0.0001
85
0.73664
0.0001
99
0.36383
0.0002
100
0.65301
0.0001
101
0.54399
0.0001
101
-0.11318
0.3024
85
-0.29922
0.0054
85
-0.41319
0.0001
85
1.00000
0.0000
85
-0,40779
0.0001
84
-0.28300
0.0087
85
-0.46155
0.0001
85
-0.3436B
0.0013
65
0.22286
0.0258
100
0.75993
0.0001
100
0.73664
0.0001
99
-0.40779
0.0001
84
1.00000
0.0000
100
0.32961
0.0008
100
0.61439
0.0001
100
0.46767
0.0001
100
0.19067
0.0561
101
0.31703
0.0012
101
0.36383
0.0002
100
-0.28300
0.0087
85
0.32961
0.0008
100
1.00000
0.0000
101
0.29997
0.0023
101
0.35123
0.0003
101
0.17319
0.0633
101
0.61528
0.0001
102
0.65301
0.0001
101
-0.46155
0.0001
85
0.61439
0.0001
100
0.29997
0.0023
101
1.00000
0.0000
102
0.37746
0.0001
102
0.33633
0.0006
101
0.55087
0.0001
102
0.54399
0.0001
101
-0.34368
0.0013
85
0.46767
0.0001
100
0.35123
0.0003
101
0.37746
0.0001
102
1.00000
0.0000
102
-------
Table L-4 LOVE CANAL CORRELATION ANALYSIS
LAB 2
VARIABLE
N
MEAN
STD DEV
MEDIAN
MINIMUM
MAXIMUM
CONC1
CONCZ
COHC3
CONC4
CONC5
CONC6
CONC7
CONC8
94
104
103
100
10*
99
9*
10*
0.43861702
1.21461538
1.74728155
0.03550000
1.05125000
0.09383838
0.96957447
0.23221154
0.29342106
2.87207341
5.38747480
0.04207677
3.45365732
0.35518523
3.49958244
0.77694067
0.36000000
0.47000000
0.44000000
0.00000000
0.15000000
0.00000000
0.00000000
0.00000000
0.17000000
o.oaoooooo
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
2.68000000
24.82000000
44.23000000
0.14000000
24.65000000
1.93000000
24.39000000
4.80000000
SPEARMAN CORRELATION COEFFICIENTS / PROB > iRl UNDER HO:RHO-0 / NUMBER OF OBSERVATIONS
CONC1 CONCZ CONC3 CONC4 CONC5 CONC6 CONC7 COMC6
CONC1
OCB Cone
CONCZ
TCB Cone
CONC3
QCB Cone
CONC4
CN Cone
CONC5
A-BHC Cone
CONC6
0-BHC Cone
CONC7
B-BHC Cone
CONC8
G-BHC Cone
1.00000
0.0000
94
0.51689
0.0001
94
0.44526
0.0001
93
0.28385
0.0067
90
0.43708
0.0001
94
0.38955
0.0001
94
0.30573
0.0047
84
0.49548
0.0001
94
0.51689
0.0001
94
1.00000
0.0000
104
0.94997
0.0001
103
0.05758
0.5694
100
0.74530
0.0001
104
0.48783
0.0001
99
0.52582
0.0001
94
0.63436
0.0001
104
0.44528
0.0001
93
0.94997
0.0001
103
1.00000
0.0000
103
0.01751
0.8635
99
0.73863
0.0001
103
0.48395
0.0001
98
0.51952
0.0001
93
0.63355
0.0001
103
0.28385
0.0067
90
0.05758
0.5694
100
0.01751
0.8635
99
1.00000
0.0000
100
0.16149
0.1085
100
0.05686
0.5710
95
0.05451
0.6099
90
0.06487
0.5214
100
0.43708
0.0001
94
0.74530
0.0001
104
0.73863
0.0001
103
0.16149
0.1085
100
1.00000
0.0000
104
0.49768
0.0001
99
0.68093
0.0001
94
0.67242
0.0001
104
0.38955
0.0001
94
0.48783
0.0001
99
0.48395
0.0001
98
0.05886
O.S710
95
0.49768
0.0001
99
1.00000
0.0000
99
0.59309
0.0001
89
0.62871
0.0001
99
0.30573
0.0047
84
0.52582
0.0001
94
0.51952
0.0001
93
0.05451
0.6099
90
0.66093
0.0001
94
0.59309
0.0001
89
1.00000
0.0000
94
0.66069
0.0001
94
0.49546
0.0001
94
0.63436
0.0001
104
0.63355
0.0001
103
0.06487
0.5214
100
0.67242
0.0001
104
0.62871
0.0001
99
0.66069
0.0001
94
1.00000
0.0000
104
-------
Table L-5 LOVE CANAL CORRELATION ANALYSIS
LAB 6
MEAN
STD DEV
MEDIAN
MINIMUM
MAXIMUM
:l
.2
:3
A
:s
:6
.7
•a
12ft
127
128
128
128
128
109
127
0.52367167
2.48574803
4.40953125
0.02375000
2.15093750
0.30421875
8.60247706
0.59622047
0.66252187
6.68724307
17.53058151
0.03236115
9.30755822
1.03499219
64.17857659
2.73889792
0.35000000
0.78000000
0.81500000
0.00000000
0.20500000
0.00000000
0.11000000
0.00000000
0.00000000
0.08000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
4.84000000
52.60000000
168.64000000
0.24000000
83.51000000
9.96000000
663.49000000
26.33000000
SPEARMAN CORRELATION COEFFICIENTS / PROB > |R| UNDER HO:RHO=0 / NUMBER OF OBSERVATIONS
CONC1 CONC2 CONC3 COHC4 CONC5 CONC6 CONC7 CONC8
CONC1
DCB Cone
CONC2
TCB Cone
CONC3
QCB Cone
CONC4
CN Cone
CONC5
A-BHC Cone
CONC6
D-BHC Cone
CONC7
B-BMC Cone
CONC8
G-BHC Cone
1.00000
0.0000
128
0.65876
0.0001
127
0.63760
0.0001
128
0.16333
0.0655
128
0.51815
0.0001
128
0.40710
0.0001
128
0.62048
0.0001
109
0.22467
0.0111
127
0.65876
0.0001
127
1.00000
0.0000
127
0.96686
0.0001
127
0.03890
0.6641
127
0.82953
0.0001
127
0.63725
0.0001
127
0.78617
0.0001
108
0.53425
0.0001
126
0.63760
0.0001
128
0.96686
0.0001
127
1.00000
0.0000
128
0.03559
0.6901
128
0.79648
0.0001
128
0.60352
0.0001
128
0.76112
0.0001
109
0.51273
0.0001
127
0.16333
0.0655
128
0.03890
0.6641
127
0.03559
0.6901
128
1.00000
0.0000
126
0.08028
0.3677
128
0.04737
0.5954
128
0.02986
0.7579
109
-0.09959
0.2653
127
0.51815
0.0001
128
0.62953
0.0001
127
0.79646
0.0001
128
0.08026
0.3677
128
1.00000
0.0000
128
0.69902
0.0001
128
0.82641
0.0001
109
0.61106
0.0001
127
0.40710
0.0001
128
0.63725
0.0001
127
0.60352
0.0001
126
0.04737
0.5954
128
0.69902
0.0001
128
1.00000
0.0000
128
0.73957
0.0001
109
0.57162
0.0001
127
0.62046
0.0001
109
0.78617
0.0001
108
0.76112
0.0001
109
0.02986
0.7579
109
0.62641
0.0001
109
0.73957
0.0001
109
1.00000
0.0000
109
0.61304
0.0001
108
0.22467
0.0111
127
0.53425
0.0001
126
0.51273
0.0001
127
-0.09959
0.2653
127
0.61108
0.0001
127
0.57162
0.0001
127
0.61304
0.0001
108
1.00000
0.0000
127
-------
VARIABLE
N
Table L-6 LOVE CANAL CORRELATION ANALYSIS
LAB 7
KAN
STD DEV
MEDIAN
MINIMUM
SPEARMAN CORRELATION COEFFICIENTS / PROB > iRl UNDER HO:RHO=0 / NUMBER OF OBSERVATIONS
CONC1 CONC2 CONC3 CONC4 CONC5 CONC6 CONC7 CONC8
CONC1
DCS Cone
CONCZ
TCB Cone
CONC3
QCB Cone
CONC4
CN Cone
CONC5
A-BHC Cone
CONC6
D-BHC Cone
CONC7
B-BHC Cone
CONC8
G-BHC Cone
1.00000
0.0000
107
0.54321
0.0001
106
0.49249
0.0001
105
0.54622
0.0001
93
0.50844
0.0001
107
0.36157
0.0002
101
0.44974
0.0001
96
0.33235
0.0005
107
0.54321
0.0001
106
1.00000
0.0000
106
0.92997
0.0001
104
0.21236
0.0421
92
0.78678
0.0001
106
0.62983
0.0001
100
0.65087
0.0001
95
0.49332
0.0001
106
0.49249
0.0001
105
0.92997
0.0001
104
1.00000
0.0000
105
0.19920
0.0556
93
0.77368
0.0001
105
0.65294
0.0001
99
0.65440
0.0001
94
0.50774
0.0001
105
0.54622
0.0001
93
0.21236
0.0421
92
0.19920
0.0556
93
1.00000
0.0000
93
0.22144
0.0329
93
0.13508
0.2122
87
0.17892
0.1078
82
0.06112
0.5606
93
0.50844
0.0001
107
0.76678
0.0001
106
0.77368
0.0001
105
0.22144
0.0329
93
1.00000
0.0000
107
0.61391
0.0001
101
0.75618
0.0001
96
0.63180
0.0001
107
0.36157
0.0002
101
0.62983
0.0001
100
0.65294
0.0001
99
0.13508
0.2122
87
0.61391
0.0001
101
1.00000
0.0000
101
0.61862
0.0001
96
0.59230
0.0001
101
0.44974
0.0001
96
0.65087
0.0001
95
0.65440
0.0001
94
0.17892
0.1078
82
0.75618
0.0001
96
0.61862
0.0001
96
1.00000
0.0000
96
0.64504
0.0001
96
0.33235
0.0005
107
0.49332
0.0001
106
0.50774
0.0001
105
0.06112
0.5606
93
0.63180
0.0001
107
0.59230
0.0001
101
0.64504
0.0001
96
1.00000
0.0000
107
MAXIMUM
CONC1
CONCZ
C011C3
CONC4
COUC5
COHC6
CONC7
CONC8
107
106
105
93
107
101
96
107
0.42392523
2.01358491
3.16266667
0.07645161
1.69915888
0.58188119
3.33312500
0.45504673
0.47226024
5.74417306
9.47963934
0.05625341
5.90383568
3.91645394
11.04027691
2.00440052
0.31000000
0.46000000
0.47000000
0.07000000
0.13000000
0.00000000
0.00000000
0.00000000
0.00000000
0.07000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
4.22000000
41.74000000
64.25000000
0.32000000
35.09000000
38.83000000
55.51000000
15.83000000
-------
ABLE
Table L-7 LOVE CANAL CORRELATION ANALYSIS
LAB 4
MEAN
STO OEV
MEDIAN
MINIMUM
SPEARMAN CORRELATION COEFFICIENTS / PROB > |R| UNDER HO:RHO=0 / NUMBER OF OBSERVATIONS
CONCl CONC2 CONC3 CONC4 CONC5 CONC6 CONC7 CONC8
CONCl
DCB Cone
CONCZ
TCB Cone
CONC3
QCB Cone
CONC4
CN Cone
CONC5
A-BHC Cone
CONC6
D-BHC Cono
CONC7
B-BHC Cone
CONC8
G-BHC Cone
1.00000
0.0000
20
0.78864
0.0001
20
0.83806
0.0001
17
0.52005
0.0168
20
0.58592
0.0066
20
0.47773
0.0331
20
0.60191
0.0050
20
0.52210
0.0182
20
0.78864
0.0001
20
1.00000
0.0000
20
0.99202
0.0001
17
0.26612
0.2568
20
0.83055
0.0001
20
0.72602
0.0003
20
0.89753
0.0001
20
0.85092
0.0001
20
0.83806
0.0001
17
0.99202
0.0001
17
1.00000
0.0000
17
0.23828
0.3571
17
0.65986
0.0001
17
0.72844
0.0009
17
0.85587
0.0001
17
0.85986
0.0001
17
0.52005
0.0188
20
0.26612
0.2566
20
0.23828
0.3571
17
1.00000
0.0000
20
0.08538
0.7204
20
0.27813
0.2351
20
0.17797
0.4528
20
0.06716
0.7785
20
0.58592
0.0066
20
0.83055
0.0001
20
0.85986
0.0001
17
0.06538
0.7204
20
1.00000
0.0000
20
0.66335
0.0001
20
0.88969
0.0001
20
0.97611
0.0001
20
0.47773
0.0331
20
0.72602
0.0003
20
0.72844
0.0009
17
0.27813
0.2351
20
0.66335
0.0001
20
1.00000
0.0000
20
0.80587
0.0001
20
0.84273
0.0001
20
0.60191
0.0050
20
0.69753
0.0001
20
0.85587
0.0001
17
0.17797
0.4526
20
0.68969
0.0001
20
0.80567
o.oool
20
1.00000
0.0000
20
0.91933
0.0001
20
0.52210
0.0182
20
0.65092
0.0001
20
0.65966
0.0001
17
0.06716
0.7785
20
0.97611
0.0001
20
0.64273
0.0001
20
0.91933
0.0001
20
1.00000
0.0000
20
MAXIMUn
1
z
1
4
5
6
r
•»
20
20
17
20
20
20
20
20
0.43750000
2.62150000
5.12352941
0.07800000
5.54400000
0.82100000
17.02000000
1.16100000
0.52776465
6.45622121
12.79056681
0.03819617
15.85198457
2.25276181
55.23798016
2.95461441
0.27500000
0.38000000
0.43000000
0.08000000
0.00000000
0.00000000
0.18500000
0.01500000
0.16000000
0.13000000
.0.04000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
2.47000000
24.02000000
50.20000000
0.17000000
69.70000000
9.83000000
241.25000000
12.20000000
-------
VARIABLE
Table L-8 LOVE CANAL CORRELATION ANALYSIS
LAB 3
MEAN
STD OEV
MEDIAN
MINIMUM
SPEARMAN CORRELATION COEFFICIENTS / PROS > (Rl UNDER HO:RHO=0 / NUMBER OF OBSERVATIONS
CONC1 CONCZ CONC3 CONC4 CONC5 CONC6 CONC7 CONC0
MAXIMUM
CONC1
CONC2
CONC3
CONC4
CONC5
COHC6
COHC7
coNca
IDS
107
104
96
108
106
103
104
0.68907407
3.54093458
3.39519231
O.OZ458333
2.23055556
0.07122642
1.96543689
0.32046077
1.95157714
16.72734105
18.35840164
0.03777821
10.46227427
0.30532950
6.79504127
1.25041770
0.38000000
0.87000000
0.68000000
0.00000000
0.13500000
0.00000000
0.00000000
0.00000000
0.00000000
0.04000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
19.80000000
167.33000000
182.41000000
0.17000000
100.31000000
2.04000000
44.89000000
9.53000000
CONC1
DCS Cone
CONCZ
TCB Cone
CONC3
QCB Cone
CONC4
CN Cone
CONC5
A-BHC Cone
CONC6
D-BHC Cone
CONC7
B-BHC Cone
CONC8
G-BHC Cone
1.00000
0.0000
108
0.75902
0.0001
107
0.59222
0.0001
104
-0.00318 •
0.9755
96
0.45183
0.0001
108
0.14995
0.1250
106
0.43503
0.0001
103
0.30731
0.0015
104
0.75902
0.0001
107
1.00000
0.0000
107
0.91393
0.0001
103
-0.02426
0.8155
95
0.64143
0.0001
107
0.30281
0.0017
105
0.57702
0.0001
102
0.51064
0.0001
103
0.59222
0.0001
104
0.91393
0.0001
103
1.00000
0.0000
104
-0.01790
0.8640
94
0.64791
0.0001
104
0.31411
0.0013
102
0.56465
0.0001
99
0.48510
0.0001
100
-0.00318
0.9755
96
-0.02426
0.8155
95
-0.01790
0.8640
94
1.00000
0.0000
96
0.07518
0.4666
96
-0.05288
0.6127
94
-0.09122
0.3898
91
-0.00750
0.9434
92
0.45183
0.0001
108
0.64143
0.0001
107
0.64791
0.0001
104
0.07518
0.4666
96
1.00000
0.0000
108
0.21661
0.0257
106
0.64382
0.0001
103
0.57433
0.0001
104
0.14995
0.1250
106
0.30281
0.0017
105
0.31411
0.0013
102
-0.05288
0.6127
94
0.21661
0.0257
106
1.00000
0.0000
106
0.29954
0.0023
101
0.32670
0.0008
102
0.43503
0.0001
103
0.57702
0.0001
102
0.56465
0.0001
99
-0.09122
0.3898
91
0.64382
0.0001
103
0.29954
0.0023
101
1.00000
0.0000
103
0.61180
0.0001
103
0.30731
0.0015
104
0.51064
0.0001
103
0.48510
0.0001
100
-0.00750
0.9434
92
0.57433
0.0001
104
0.32670
0.0008
102
0.61180
0.0001
103
1.00000
0.0000
104
-------
Table L-9
VARIABLE
LOVE CANAL CORRELATION ANALYSIS
EDA ONLY
MEAN
STD DEV
MEDIAN
MINIMUM
SPEARMAN CORRELATION COEFFICIENTS / PROB > iRl UNDER HO:RHO=0 / NUMBER OF OBSERVATIONS
CONC1 CONC2 CONC3 CONC4 CONC5 CONC6 CONC7 CONC8
CONC1
OCB Cone
CONC2
TCB Cone
CONC3
QCB Cone
CONC4
CN Cone
CONCS
A-BHC Cone
CONC6
0-BHC Cone
CONC7
B-BHC Cone
CONC8
6-BHC Cone
1.00000
0.0000
488
0.54411
0.0001
487
0.44546
0.0001
479
0.19218
0.0001
451
0.42286
0.0001
487
0.28125
0.0001
482
0.42286
0.0001
452
0.29783
0.0001
484
0.54411
0.0001
487
1.00000
0.0000
510
0.93407
0.0001
501
-0.07298
0.1133
472
0.77341
0.0001
508
0.52413
0.0001
498
0.65782
0.0001
469
0.57425
0.0001
506
0.44546
0.0001
479
0.93407
0.0001
501
1.00000
0.0000
502
-0.10021
0.0302
468
0.77586
0.0001
500
0.53976
0.0001
490
0.66173
0.0001
461
0.58580
0.0001
498
0.19218
0.0001
451
-0.07298
0.1133
472
-0.10021
0.0302
468
1.00000
0.0000
473
-0.01011
0.8267
472
0.06600
0.1567
462
-0.02941
0.5421
432
-0.01048
0.8209
469
0.42286
0.0001
487
0.77341
0.0001
508
0.77586
0.0001
500
-0.01011
0.8267
472
1.00000
0.0000
509
0.54250
0.0001
498
0.74655
0.0001
468
0.64676
0.0001
505
0.28125
0.0001
482
0.52413
0.0001
498
0.53976
0.0001
490
0.06600
0.1567
462
0.54250
0.0001
498
1.00000
0.0000
499
0.56232
0.0001
463
0.53715
0.0001
495
0.42286
0.0001
452
0.65782
0.0001
469
0.66173
0.0001
461
-0.02941
0.5421
432
0.74655
0.0001
468
0.56232
0.0001
463
1.00000
0.0000
470
0.61751
0.0001
469
0.29783
0.0001
484
0.57425
0.0001
506
0.58580
0.0001
498
-0.01048
0.8209
469
0.64676
0.0001
505
0.53715
0.0001
495
0.61751
0.0001
469
1.00000
0.0000
507
MAXIMUM
CONC1
CONC2
CONC3
CONC4
CONCS
CONC6
CONC7
CONCS
488
510
502
473
509
499
470
507
0.59290984
2.34939216
3.30209163
0.05124736
2.51029470
0.46935672
16.97814094
0.81497041
1.03001990
8.85880155
13.16388954
0.05015155
11.84650271
4.05610042
207.94620076
4.85880694
0.39500000
0.64000000
0.61000000
0.05000000
0.18000000
0.00000000
0.00000000
0.00000000
0
0
0
0
0
0
0
0
19.80000000
167.33000000
182.41000000
0.24000000
152.53000000
79.99000000
4108.31000000
85.60000000
-------
Table L-10 LOVE CANAL CORRELATION ANALYSIS
BUFFALO COMPARISON AREA
VARIABLE
CONC1
CONC2
CONC3
CONC4
COHC5
CONC6
CQNC7
CONC8
N
57
60
61
54
61
61
59
59
MEAN
0.44607018
0.16950000
0.06766805
0.04222222
0.00442623
0.00000000
0.09694915
0.00067797
STO DEV
0.28395446
0.13029855
0.11187962
0.04772827
0.02446253
0.00000000
0.69857800
0.00520756
MEDIAN
0.36000000
0.13500000
0.05000000
0.03000000
0.00000000
0.00000000
0.00000000
0.00000000
MINIMUM
0.11000000
0.07000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
MAXIMUM
1.38000000
0.92000000
0.84000000
0.21000000
0.17000000
0.00000000
5.36000000
0.04000000
SPEARMAN CORRELATION COEFFICIENTS / PROB > iRl UNDER HO:RHO=0 / NUMBER OF OBSERVATIONS
CONC1 CONC2 CONC3 CONC4 CONC5 CONC6 CONC7 CONC8
CONC1
OC8 Cone
CONC2
TCB Cone
CONC3
QCB Cone
CONC4
CN Cone
CONC5
A-BHC Cone
CONC6
D-BHC Cone
CONC7
B-BHC Cone
CONC8
6-BHC Cone
1.00000
0.0000
57
0.61722
0.0001
56
0.25871
0.0520
57
0.32103
0.0230
50
0.12136
0.3685
57
•
57
0.04896
0.7226
55
0.05574
0.6860
55
0.61722
0.0001
56
1.00000
0.0000
60
0.63094
0.0001
60
-0.08586
0.5410
53
0.10591
0.4206
60
60
0.06063
0.6512
58
-0.01190
0.9294
58
0.25871
0.0520
57
0.63094
0.0001
60
1.00000
0.0000
61
0.01290
0.9263
54
0.04745
0.7165
61
•
61
0.06221
0.6397
59
0.00389
0.9767
59
0.32103
0.0230
50
-0.08586
0.5410
53
0.01290
0.9263
54
1.00000
0.0000
54
0.02291
0.8694
54
54
0.00400
0.9775
52
0.16996
0.2284
52
0.12136
0.3685
57
0.10591
0.4206
60
0.04745
0.7165
61
0.02291
0.8694
54
1.00000
0.0000
61
*
61
0.38974
0.0023
59
-0.03038
0.8193
59
•
57
*
60
61
54
61
1.00000
0.0000
61
»
59
59
0.04896
0.7226
55
0.06063
0.6512
58
0.06221
0.6397
59
0.00400
0.9775
52
0.38974
0.0023
59
59
1.00000
0.0000
59
-0.02503
0.8521
58
0.05574
0.6860
55
-0.01190
0.9294
58
0.00389
0.9767
59
0.16996
0.2284
52
-0.03038
0.8193
59
•
59
-0.02503
0.8521
58
1.00000
0.0000
59
-------
'ABLE
Table L-11 LOVE CANAL CORRELATION ANALYSIS
CENSUS TRACT 221 COMPARISON AREA
MEAN
STD OEV
MEDIAN
MINIMUM
SPEARMAN CORRELATION COEFFICIENTS / PROB > |R| UNDER HO:BHO=0 / NUMBER OF OBSERVATIONS
CONC1 CONC2 CONC3 CONC4 CONC5 CONC6 COHC7 CONC8
CONC1
DCB Cone
CONC2
TCB Cone
CONC3
QCB Cone
CONC4
CN Cone
CONC5
A-BHC Cone
CONC6
D-BHC Cone
CONC7
B-BHC Cone
CONC8
G-BHC Cone
1.00000
0.0000
50
0.45846
0.0008
50
0.29713
0.0381
49
0.39256
0.0058
46
0.19688
0.1706
50
0.00968
0.9474
49
0.04220
0.7807
46
0.01222
0.9329
50
0.45846
0.0008
50
1.00000
0.0000
53
0.89487
0.0001
52
-0.28646
0.0416
51
0.53882
0.0001
53
0.04723
0.7395
52
0.24249
0.0932
49
0.06631
0.6371
53
0.29713
0.0381
49
0.69487
0.0001
52
1.00000
0.0000
52
-0.28957
0.0393
51
0.55730
0.0001
52
0.20600
0.1470
51
0.28393
0.0505
48
0.07371
0.6036
52
0.39256
0.0058
48
-0.28646
0.0416
51
-0.28957
0.0393
51
1.00000
0.0000
51
-0.16195
0.2562
51
0.04294
0.7671
50
-0.20043
0.1767
47
0.25338
0.0728
51
0.19688
0.1706
50
0.53662
0.0001
53
0.55730
0.0001
52
-0.16195
0.2562
51
1.00000
0.0000
53
0.16077
0.2549
52
0.36954
0.0090
49
0.22553
0.1044
53
0.00968
0.9474
49
0.04723
0.7395
52
0.20600
0.1470
51
0.04294
0.7671
50
0.16077
0.2549
52
1.00000
0.0000
52
0.49012
0.0004
49
0.10012
0.4801
52
0.04220
0.7607
46
0.24249
0.0932
49
0.28393
0.0505
48
-0.20043
0.1767
47
0.36954
0.0090
49
0.49012
0.0004
49
1.00000
0.0000
49
0.21454
0.1388
49
0.01222
0.9329
50
0.06631
0.6371
53
0.07371
0.6036
52
0.25338
0.0728
51
0.22553
0.1044
53
0.10012
0.4801
52
0.21454
0.1388
49
1.00000
0.0000
53
MAXIMUM
-.1
.2
:3
:4
.5
:6
:7
:8
50
53
52
51
53
52
49
53
0.45320000
0.77377358
0.56557692
0.06156663
0.20667925
0.02615365
0.11183673
1.56415094
0.22515700
0.52141902
0.26411631
0.05465682
0.21730905
0.13619700
0.24001453
11.09518268
0.39000000
0.65000000
0.53000000
0.06000000
0.18000000
0.00000000
0.00000000
0.00000000
0.19000000
0.25000000
0.22000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
1.15000000
3.12000000
1.33000000
0.22000000
1.19000000
0.94000000
1.03000000
80.81000000
-------
Table L-12 LOVE CANAL CORRELATION ANALYSIS
CENSUS TRACT 225 COMPARISON AREA
VARIABLE
MEAN
STD DEV
MEDIAN
MINIMUM
MAXIMUM
CONCl
CONC2
CONC3
cone*
CONC5
CONC6
COHC7
COHC8
59
60
60
61
61
60
55
61
0.50033098
1.82666667
2.58466667
0.06639344
1.10196721
0.12933333
2.10737273
0.41934426
0.26216582
5.15182366
8.71472890
0.05793769
4.53728914
0.70161525
9.18768912
2.06940963
0.43000000
0.60500000
0.62500000
0.07000000
0.11000000
0.00000000
0.00000000
0.00000000
0.18000000
0.12000000
0.06000000
0.00000000
0.00000000
0.00000000
0.00000000
0.00000000
1.40000000
33.07000000
64.25000000
0.32000000
33.97000000
5.40000000
50.57000000
15.83000000
SPEARMAN CORRELATION COEFFICIENTS / PROB > |R| UNDER HO:RIIO=0 / NUMBER OF OBSERVATIONS
CONC1 CONC2 CONC3 CONC4 CONC5 CONC6 CONC7 CONC8
CONC1
OCB Cone
CONC2
TCB Cone
CONC3
QCB Cone
CONC4
CN Cone
CONC5
A-BHC Cone
CONC6
0-BHC Cone
CONC7
B-BHC Cone
CONC8
6-BHC Cone
1.00000
0.0000
59
0.36374
0.0050
58
0.34417
0.0082
58
0.23062
0.0789
59
0.16323
0.2167
59
0.14308
0.2840
58
0.13839
0.3136
55
0.14490
0.2735
59
0.36374
0.0050
58
1.00000
0.0000
60
0.93863
0.0001
59
0.01355
0.9182
60
0.52374
0.0001
60
0.42648
0.0008
59
0.45435
0.0006
54
0.50298
0.0001
60
0.34417
0.0082
58
0.93863
0.0001
59
1.00000
0.0000
60
-0.03954
0.7642
60
0.55685
0.0001
60
0.45365
0.0003
59
0.49776
0.0001
54
0.55258
0.0001
60
0.23062
0.0789
59
0.01355
0.9182
60
-0.03954
0.7642
60
1.00000
0.0000
61
0.20727
0.1090
61
0.23532
0.0703
60
0.06134
0.6564
55
0.11294
0.3862
61
0.16323
0.2167
59
0.52374
0.0001
60
0.55685
0.0001
60
0.20727
0.1090
61
1.00000
0.0000
61
0.51500
0.0001
60
0.63398
0.0001
55
0.59011
0.0001
61
0.14308
0.2840
58
0.42648
0.0008
59
0.45365
0.0003
59
0.23532
0.0703
60
0.51500
0.0001
60
1.00000
0.0000
60
0.46609
0.0004
54
0.72293
0.0001
60
0 . 1 3839
0.3136
55
0.45435
0.0006
54
0.49776
0.0001
54
0.06134
0.6564
55
0.63398
0.0001
55
0.46609
0.0004
54
1.00000
0.0000
55
0.53048
0.0001
55
0.14490
0.2735
59
0.50298
0.0001
60
0.55258
0.0001
60
0.11294
0.3862
61
0.59011
0.0001
61
0.72293
0.0001
60
0.53048
0.0001
55
1.00000
0.0000
61
-------
5.5
5.0
4.5
4.0
3.5
T
C
B
3.0
C
o
n
c Z.S
2.0
1.5
1.0
0.5
0.0
A
A
A A
A A
A
A A
A
AAA A
AA A C
A A AA A
A B
A A AA BAA
A A A BA
BAAA BA A A A
A AA A AA A
B D AA A
A AA A AAABA A A
C BAA BCBACBB A
A BAABBABCCAAAC A
BE EHEFGAEBCAA A
A BBOHGKDJHBB A A AA
BEIPNLHBECBOBAA
A NPZSLIBBB A A
BJTZXTCF C
IRZZIDA
*A A
-*—
0.0
0.5
1.0
1.5
2.0
2.5
3.0
9C8 Cone
3.5
4.0
4.5
5.0
5.5
6.0
NOTES:
• A » 1 observation, 8 = 2 observations, etc.
• For 64 of the samples, one of the two chemicals plotted had
no valid results, and the results for those samples are not shown
• 15 sample results are not shown because more than 26
samples (2 = 26) had the same coordinate concentrations.
Figure L-1 LOG CONCENTRATION OF 1,2,4,-TRICHLOROBENZENE
VS. LOG CONCENTRATION OF 1,2,3,4-TETRACHLOROBENZENE
-------
5.5 »
5.0
4.5
4.0
A 3.5
B
H
C 3.0
C
o
n 2.5
e
2.0 *
I
IA
I
1.5 »
IA
IA
I
1.0 »
A A
A B
A A
B AA
A AA
A A
A A A
B
A B AA
A AA A A
A A AAB
A A
A B A
1C A B AACA
I AA CA A B B A
IB A A ADBCCBA AA
0.5 *D ACCEACB B A A
IN ADEJBA A
IZBrtHKDCA AA
IZINCE BB
0.0 +ZJHBBAA A
-4. 4 4
0.0 0.6 1.2
A
—4_-
1.8
—4—
2.4
—4.-
3.0
3.6
4.2 4.8
B-BHC Cone
5.4 6.0 6.6 7.2 7.B 8.4 9.0
NOTES:
• A = 1 observation, B = 2 observations, etc.
• For 105 of the samples, one of the two chemicals plotted had
no valid results, and the results for those samples are not shown.
• 264 sample results are not shown because more
than 26 samples (Z = 26) had the same coordinate concentrations.
Figure L-2 LOG CONCENTRATION OF ALPHA-BHC
VS. LOG CONCENTRATION OF BETA-BHC
-------
5.5
5.0 *
4.5
4.0 •»
3.5
T
C
B
3.0
C
o
n
c 2.5
2.0
1.5
1.0
0.5
0.0
A
A A
A B A
A A A
A A A
C
A A B B B
A A A A A A
AA AA A AA A A A
AA A A A A
A BA AAAA A
1C A CAA A A B
+0 BA BCCAAAAB AA
IB ABCCACB ABA A A AA
IIAAAOIGEABAAEAAA AA A
IG CLHJCCHBBAB A A A
+QACWMIHA CCB B AA A
IZKZTHEC A
IZUSJEA A A
IZGF A
+A
_> -_____+-____—_+—_-__._+ ______*_ .
0.0 0.5 1.0 1.5 2.0
A A
A A A A
A A
A
A A
A A
A
A
—4. 4 «. «. * »—
2.5 3.0 3.5 4.0 4.5 5.0
A-BHC Cone
5.5 6.0
NOTES:
• A = 1 observation, B = 2 observations, etc.
• For 55 of the samples, one of the two chemicals plotted had
no valid results, and the results for those samples are not shown.
• 101 sample results are not shown because more than 26
samples (Z - 26) had the same coordinate concentrations.
Figure L-3 LOG CONCENTRATION OF 1,2,4-TRICHLOROBENZENE
VS. LOG CONCENTRATION OF ALPHA-BHC
-------
0.51
0.27
03,0.48
0.58
0.38
0.55 0.42 0.M
0.48
0.54
. .„
°-40 o.ee
0X6 O.T?'77
0.38
0.52
0..7 -
0.43 0.50
0.49 °'37
0.43 0.48
1.03
0.29
0/45
0.53
0.53
0.34
0.40
0.87 """
0.38
0.36 0.41
0.54
0.38 0X1
0X9
0.47
0.51
0.40
0X9
0.41
0.37
NOTE:
Concentrations shown represent an average
of 4 or 5 locations.
Love
Canal
Site
1.83
0X3 0.19
0.53
0.39
0X4 0X4 0.34
0.40
0.54 OXOo.71
0.31 0.53 0.51 0X9
O.W°-« 0X6 0.32
0X7 0X8 0.35
0.43
1.34
0X0
0.99 4.20 0.39
3.10
0.96
0.81
0.96
0.760X1
Figure L-4 SOIL ASSESSMENT - INDICATOR CHEMICALS
MAP OF 1,2,-DICHLOROBENZENE CONCENTRATIONS
-------
0.49
0.86
0.95 0.49
0,40
9.04
0.23
1.13
1.65
1.63
0.80 0.160.56
0.52
0.83
0.33 '
0.90 0.59
1.24 0.71
0.68 0.93
1.53
0.31
7.85
0.31
0.59
0.96 0.41
1.83
0.22
0.53
0.51
1.08
0.36 0.47
0.52
0.32
1.01 °-89 032
°-81 0.68 0.47 Q44
1.25
Love
Canal
Site
0.96
0.29
0.64
1.92 1.063.54
0.75 1-*7 1.05 1.»*
2-41 1.23 0.70 0.29
2.17 0.73 0.80
5-65 10.56
26.84
2.63
12.11
7.78 ••
0.52
135 24.63 LOO
1 0.951.25
NOTE:
Concentrations shown represent an average
of 4 or 5 locations.
Figure L-5 SOIL ASSESSMENT -- INDICATOR CHEMICALS
MAP OF 1,2,4-TRICHLOROBENZENE CONCENTRATIONS
-------
°'38 nj*
0.64 °*3
0.90 0.57
0.15
0.83
5.01
4.76
16.93
1.31
0.130.54
0X8
0.77
0.82
0.31
0.18
0.81 0.51
1.02 0.67
0.65 0.83
0.33
30.77 O-37
1.91
0.24
0.70 0.32
2.59
0.31 0.44
0.16
0.60 1-15 0.15
0.56 0.43
0.68
OA4
0.74
0.47 0.38
0.24
1.24
0.31
Love
Canal
Site
8.32
17.93
033
1.33 5.35
0.73
1.59 1.363.21
0.82 1-35 1-07 2.01
4-94 1.63
0.80 oj»8
3.27 0.02
3.46
18.55
0.98
1.86
0.54
40.93
11.61
I2.7g6.55
7.19 1-25
1.09 1.94
NOTE:
Concentrations shown represent an average
of 4 or 5 locations.
Figure L-6 SOIL ASSESSMENT -- INDICATOR CHEMICALS
MAP OF 1,2,3,4-TETRACHLOROBENZENE CONCENTRATIONS
-------
0.00
O.OB
0.09
0.06 0.11
0.03 0.07
0.07 o.08
0.04
0.05
0.08
0.05
NOTE:
Concentrations shown represent an average
of 4 or 5 locations.
0.11
0.02
0.04
0.09 0.07 o.13
0.08
0.00
0.06
0.07
0>07 0.100.03
0-°0 0.06
0.02
0.11
0.05
0.09
0.05
0.05 0.06
0.12
0.04 0.04
0.04 O-09 0.10
0.05 0.06 0.06 0.04
.0.04
0.04 0.04
0.06
0.03'
0.06 0-05
0.06
0.06
Love
Canal
Site
0.04 0.06J.Q3
0.02 0>04 °-09 °*03
0.00 0.02 0.07 0.00
0.01
0.05
0.10
0.07
0.03
0.09
0.03
0.00
OM 0.06 0.04
0.06
0.00
0.07
0.04
0.06
0.08
0.07 0.07
Figure L-7 SOIL ASSESSMENT -- INDICATOR CHEMICALS
MAP OF 2-CHLORONAPHTHALENE CONCENTRATIONS
-------
0.16
0.20
o.oo °-°'
0.39 0.22
0.25 0.30
0.18 0.25
0.03
0.14
0.13
0.28
NOTE:
Concentrations shown represent an average
of 4 or 5 locations.
0.00
O.eiO.34
0.37
0.08
0.83
0.32 0.40
0.13 O.OtO.01
0.12 0>09-oo
1.03
0.07
19.19
0.00
°-OB
0.11
0.16
0.09
0.28
0.18
0.08 °-62 0.03
°-42 0.39 0.22 o.19
0.68 50.45 191
1^*25
0.94
Love
Canal
Site
0.92 0.593.59
0.14 0^0 1.10 0^1
t-4* 0.72 0.18 o.OO
0.09
0.45
6.10
00
0.64
2.07
0.19
1.08
8.17
22.58
7'91
14.903.33
0.00
14.54 0.67
0.801.34
Figure L-8 SOIL ASSESSMENT -- INDICATOR CHEMICALS
MAP OF ALPHA-BHC CONCENTRATIONS
-------
0.00
0.00
o 140.00
0.04
0.07
0.00 0.03
0.00
1.48
015
0.00
0.04 0.0tf>-00
•00 0.00
0.00
0.00
0.00 -
0.09 0.00
0.00 °-°°
0.02 o.02
0.00
0.01
0.00
0.21
0.22
0.00
1.83
0.03
0.00
0.02
0.03 0.02
O.T7
0.06
0.03 0.43
0.01
0.00
0.01 °-°° 0.00
0.06 0>02 o.02 002
0.05 °-00
0.21
0.00
0.24
Love
Canal
Site
0.07
2.14
0.63
0.03
0.21 o.oq.oe
0.04 0.05 0.00 0.00
0.06 o.OO
26.70 o.32
0.45 0.12 0.07
0.17
0.15
0.56
0.93
4.55 0.28
0.00
0.04 0.16
0.080.28
NOTE:
Concentrations shown represent an average
of 4 or 5 locations.
Figure L-9 SOIL ASSESSMENT -- INDICATOR CHEMICALS
MAP OF DELTA-BHC CONCENTRATIONS
-------
0.00
0.27 0.34
11.88
0.06
1.06
0.50 0.35
0.00
12.89
"° °'" o.,o
0.00 o.OO
0.00
0.00
0.19
0.00 °-00
0.33 0.07
0.39
O.H
0.07
0.37
0.02
0.07
0.03
0.52
0.03
0.09 0.09
1.11
0.08 0.15
0.02
0.08
0.05
0.00 0.08 2-75 0.03
109.78 °-00 °'17 0.24 0.16 008
6.24 °-16
0.73 1013.48 •*
0.12 2Jn 0.6226^4
131 0.00
Love
Canal
Site
0.08
0.49
6-12 6.18
42.09
0.12 1'01
9J7 t.14
0^6 O.T7
0.33
175
0.33 o.OO
O.U
0.00
°-3e 5.16 0.54
19.153-05
1621.68
NOTE:
Concentrations shown represent an average
of 4 or 5 locations.
Figure L-10 SOIL ASSESSMENT -- INDICATOR CHEMICALS
MAP OF BETA-BHC CONCENTRATIONS
-------
0.00
0.03
0.00 °-00
0.06 0.04
0.00 0.06
0.03 0.04
0.00
0.02
0.00
0.08
NOTE:
Concentrations shown represent an average
of 4 or 5 locations.
0.00
0.13
0.07 0.08
0.04
4.09
0.00
0.07
0.88
0.59
0.0tf>-00
0.02
0.98
0.00
6.71
0.00
0.01
0.01
0.03 0.02
0.32
0.05
0.01 O-25 0.00
°-18 0.05 0.15
t 4O 0.00
0.08 23.10
0.04
O.U 0.05
0.04
0.02
Love
Canal
Site
0.
1.41 0.2Q5.87
0.16 0.09 0.00
0-13 0.15 0.03 o.oo
0.32 0.51 0.08
0.00
0.14
0.92
0.00
0.35
0.48
1.76
7.09
0.24
0.00
0.97 0.19
0.13 0.38
Figure L-11 SOIL ASSESSMENT -- INDICATOR CHEMICALS
MAP OF GAMMA-BHC CONCENTRATIONS
-------
PLOT OF CONC6WIST LEGEND: A = 1 DBS, B = 2 083, ETC.
4.0
3.5
3.0
B 2.5
H
C
C
o 2.0
n
e
1.5
1.0
0.5
0.0 *
A A A B
E A A A
B A
A A A
A A A
A A A A A A A
BAA A
A A A C A B A
A A A A B A A
ABAACACAAB AA AAAA AA A AA
ABBABAAAAABACAOA A A BBA AAA BBA
MEAEBGCBGAIFAC FHAP DIANAADCQACOCHAEFBNACGOEAAQAJDFEDICBFCOCCCBG EO BC EA BAOACAA BABD C
200
400
600
800
1000
1200
OIST
1400
1600
1800
2000
2200
2400
NOTES:
• A *= 1 observation. B = 2 observations, etc.
• For 51 of the samples, one of the two chemicals plotted had
no valid results, and the results for those samples are not shown.
Figure L-12 SOIL ASSESSMENT -- INDICATOR CHEMICALS
LOG CONCENTRATION OF DELTA-BHC
VS. DISTANCE FROM LOVE CANAL BOUNDARY
-------
Sat/MOGRAM FOR AREA EM2 IOC U4CMOROEBCBE
E •«
SX
SCPflBSTION DISTANCE
&MHOGRAM FOR AREA EW3 LOC1 J-OOUROBENZENE
>oo «o loo no
KPOUITIIM DISTPMCC
SEWWOGRAM FOR WEA EW5 LOG U-DCHLOROBENZENE
|
MO 7SO
SfPWtRIIOK OISIDNCC
S8IVWOGRAM FOR WEA EDA7 LOG U-DCHLOROBENZENE
no soo iso 1000 ino isoo 1750
stPSRnnox DISIWCE
SEMVAHOGWM FOR WEA C221 LOG 1i-OICHLOROBENZENE
O .13
1.,
8-
BO 100 TSO
1000
' DISTPNCC
ino IKE ino
SEUVMOGRUI FOR MEA C225 LOC U-OCHLOROBENZENE
no no TSO 1000 ia> noo ino
scrawiiON DISTANCE
•noo IOOOD IBOO IBOD ITSOO xoooo moo
KnWDTION DISTWCt
Notes: Separation distances are in feet; nondetectable values were assigned a uniform
random number between 0 and an assumed effective detection limit of 0.5 ppb.
Figure L-13 SAMPLE VARIOGRAMS FOR LOGARITHMS OF 1,2-DICHLOROBENZENE
-------
X 1.15
Bl.OO
'"
100 ZQO 100
JOO IK
KKKATION OISIWCE
no soo no tooo ino
StPORBTION DISTBNCE
i.w
1.10
-.1.15
StT(»P7ION DISTSNCt
too nc tno 1900
BO SOO ISO
StMRflTION DISTflNCC
1000 ino
SBMHOGRAM FOR AREA C221lflCU4-mCHLOflOBENZE«
no too no 1000 tno
KPflRmiON OISTHNCC
? no wo lao tooo ino isoo two
DIStiMtc
no too wo uoo tsoo
DISTftttt
noo 11000 inoo iiaoo
KTORHTION OISIDNCC
Notes: Separation distances are in feet; nondetectable values were assigned a uniform
random number between 0 and an assumed effective detection limit of 0.5 ppb.
Figure L-14 SAMPLE VARIOGRAMS FOR LOGARITHMS OF 1,2.4-TRICHLOROBENZENE
-------
i.n
1.10
I-
no wo MO
l.l»
I.SO
A
•
nc wo no tooo iao
* OlSTflNCE
"°
no wo
•no 1000
DISTANCE
HO <00
SfPSWITION OISTSNCC
I
JI.OO
JOO 60C WC 1KB 1500
$00
SEPARATION
IK tooo ino
ij
no HO 150 IODO IHO ISOO tlSO
! DISTflNCC
ISO 100 ISC 1000 IISO ISCC IISO
StPBRBTIOK DISTBHCE
r
BO no iso 1000 tiso ino iiso
KPPRBTION 01 STANCE
woo 10000 iaoo mo IIBO nooo moo
DISTPHCE
Notes: Separation distances are in feet; nondetectable values were assigned a uniform
random number between 0 and an assumed effective detection limit of 0.5 ppb.
Figure L-15 SAMPLE VARIOGRAMS FOR LOGARITHMS OF 1,2,3,4-TETRACHLOROBENZENE
-------
SBMHOGRMI FOR tf& EMI LflC 2-CHOfiONtfHMM
IK » «0
KPflRBTK* DISTfHCE
•o no
straw ION OISTIMCE
in ino
ISO MO ISO 1000 IZSO
SEWSOTION DISTANCE
*100 BO 100 «00 too 100 '
5CPARBTION DISTANCE
r.o
I-'
100 no iioo is
senxanoN DISTANCE
SBMROGRAUFORAfft^
I.
..
i» too no 1000
SEFMtmiDN DISTANCE
ao too -no IODO IBO isoo ino
no KB -no m tno in>
CEPMtRTlON OI8TIMCC
OISTDNCC
Notes: Separation distances are in feet; nondetectable values were assigned a uniform
random number between 0 and an assumed effective detection limit of 0.5 ppb.
Figure L-16 SAMPLE VARIOGRAMS FOR LOGARITHMS OF 2-CHLORONAPTHALENE
-------
SEUMflOGRAU FOR AREA EW1 LOG ALWA-BHC
I"1
in no loo «o wo
SEPOWTION OISKHCC
SBIVAROGRAU FOR ABEA BDti LOC ALPHA-BHC
SaiV/WOGRW FOR WEA EW5 LflC AlPW-BHC
SBilVIWOGRHI FOR MEA EW LOG AlPHA-BHC
scrawl ON DISTPMCC
SaiVWOGRWI FOR AREA EW7 LOG ALPHA-BHC
no no im ino IKO
KPWHTION OtSTONtt
SBIVAHOGR/IM FOR AREA C221 LOG ALPHA-BHC
no HO IM 1000 ino im IT
01STDNCE
SBi/ABOGRAM FOR AREA C225 IOC AIPHA«C
SBi/AWGRAU FOR AREA C«T LtJC ALPHA-BHC
SCnUUITION OISTDNCC
Notes: Separation distances are in feet; nondetectable values were assigned a uniform
random number between 0 and an assumed effective detection limit of 0.5 ppb.
Figure L-17 SAMPLE VARIOGRAMS FOR LOGARITHMS OF ALPHA-BHC
-------
SaMRDGfiAU FOR AREA EDA1 ICC BETA-BHC
i.a
fjo
i.it
Si .00
» «o too no mo too
SCPHKIIIION DISTANCE
WURPTIO* OISIMCC
tooc ino
SEUVMOGRAM FOR AREA BWIOCBETA-BHC
10cn030040C900«01QC«]CtaO
SEPORfiTiOK DISTW4CE
4.90
4.00
j.»
Sl.OO
gl.SOj
11.00
SBIWKGRAM FOR WEAHW IOC BETA-BHC
ioo m no ITOC isoo
SCPPRATION DISTANCE
SBAfADOGRMI FOR AflEA HUE LOG ETA-BHC
m
SCPOUJTION OISTRNCE
I
E «
SBAfADOGRAM FOR AREA EDA7 LOC BETA-BHC
•no laao ino isoo 1150
I"'
SBIVAnOSRAM FOR AREA (221 LOG BETA-BHC
f» 100 150 1000 IBO 1100
SEHWBIIW DISTANCE
SBWAWGRAM FOR AREA C22510C BETA-BHC
Notes: Separation distances are in feet; nondetectable values were assigned a uniform
random number between 0 and an assumed effective detection limit of 0.5 ppb.
Figure L-18 SAMPLE VARIOGRAMS FOR LOGARITHMS OF DELTA-BHC
-------
SBIWWGRAM FOR AREA BA1 tOCDELTA-BHC
SBIVMKGRAUFORMEADULQCDELTA-BHC
KPWBIION DISlflNCE
DISTPMCE
I.IS
I.SO
11.is
g,.,
SatVWWGWM FOR WEA HlWiaDELW-BHC
1.00
1-19
11.to
SI-IS
SBMUKGRMIFORME/iBlMUXDELTA-eHC
too too 300 400 no wo 100 no
SEPfKBTION DISTANCE
I-
SBIVAWGRWI FOfl AREA EW5 LOC DELW-BHC
SBifADOGRMI FOR WEA HW LOG OELTA-BHC
S£P«01I»- DISTDNCE
ISO SOO T50 1000 1ZSO
SEPARATION DISTANCE
I.H
i-n
Wimm FOR MEA EDAI LOG DELTA-BHC
no 1000 tno ist
SEPARATION OlStANCE
SBMftOGMM FOR AREA C221 IOC DELTA-BHC
no too iso 1000 izso isoo iiso
SCPRTUTION DISTANCE
SBi/AK)GRAM FOR AREA C225 LOCOELTA-BHC
f» sx 750 looo \no ttoc fno
OISTIMCE
SBIVAIIOGRAMFORAI)EAC«TLaCDaTA-BHC
noo BOO noo IOOOD inoo itooo tftoo nooo moo
KttRRTlON 01STWCE
Notes: Separation distances are in feet; nondetectable values were assigned a uniform
random number between 0 and an assumed effective detection limit of 0.5 ppb.
Figure L-19 SAMPLE VARIOGRAMS FOR LOGARITHMS OF BETA-BHC
-------
SBKMOGRUIFORMEAEM1LOCGMMA-BHC
iooiooia)4ooiao«Biooioo
StrflRSIKX OlSTRNCt
I.to
i.n
I"'
C .15
K
SBMHOGRAM FOR WEA BM2 LOC GMItt-BHC
no n no
KMKHTIIM OISISNCE
S.OO
it.so
I'"
!-«>
StPflRflTIW DI5TKNCE
>D 400 100 MB KB 100
SfPWWTION DISTANCE
9.SO
1.00
fil.90
SWADOGRAM FOR AREA EW5 LOG GAMUA-BHC
JDO IDC no two
I DISTANCE
JJ3D
r-so
I"
SBIVMOGMU FOR AREA E)W LOC GAMMA-BHC
StPORBTIW OISIfKCE
SEMWABOGRAM FOR AREA HWIOCGWIMA-BHC
SaWAf»GfW(FOflW£AC221LOC««MA-BHC
BO BO ISO 1000 1190
OISTMNCE
wo ino mo
tja
I.T9
t.K
I-
!•*
SEfDRRTlON DISIfKE
BOO MOD WOO
Notes: Separation distances are in feet; nondetectable values were assigned a uniform
random number between 0 and an assumed effective detection limit of 0.5 ppb.
Figure L-20 SAMPLE VARIOGRAMS FOR LOGARITHMS OF GAMMA-BHC
-------
APPENDIX M
Neighborhood Comparisons and Redefined
EDA Sampling Area Comparisons
Soil Assessment-Indicator Chemicals
-------
Appendix M
NEIGHBORHOOD COMPARISONS AND
REDEFINED EDA SAMPLING AREA COMPARISONS
SOIL ASSESSMENT—INDICATOR CHEMICALS
At the request of the TRC and in response to a request by
Dr. Schoenfeld (peer review meeting June 22, 1988) addi-
tional comparisons were performed. The proposed habit-
ability criteria specified that comparisons were to be done
between EDA sampling areas and each of the three comparison
areas. The additional comparisons were requested to provide
a context in which the non-statistical significance of the
statistical comparisons could be judged.
The comparisons requested are presented in a format
identical with that of the comparisons presented in
Volume III. The following analyses are presented:
o EDA neighborhoods and comparison areas compared
with other EDA neighborhoods and comparison areas
for each LCIC (univariate test), based only on the
observations classified as Good (Table M-la and b
and Figure M-l).
o EDA neighborhoods and comparison areas compared
with other EDA neighborhoods and comparison areas
for all LCICs considered together (multivariate
test), based only on the observations classified
as Good (Table M-2a and b and Figure M-2).
o Redefined EDA Sampling Areas 2" ("two prime")
and 3' compared with all other EDA sampling areas
and comparison areas for each LCIC (univariate
test), based only on the observations classified
as Good (Table M-3a through c and Figure M-3).
o Redefined EDA Sampling Areas 2' and 3' compared
with all other EDA sampling areas for all LCICs
considered together (multivariate test), based
only on the observations classified as Good
(Table M-4a and b).
o EDA sampling areas and one aggregate comparison
area (all three comparison areas combined)
compared to other EDA sampling areas and the one
aggregate comparison area for each LCIC (univar-
iate test), based only on the observations
classified as Good (Table M-5a and b).
o EDA sampling areas and one aggregate comparison
area (all three comparison areas combined)
compared to other EDA sampling areas for all LCICs
M-l
-------
considered together (multivariate test), based
only on the observations classified as Good
(Table M-6a and b).
o One aggregate EDA sampling area (all seven EDA
sampling areas combined) compared to the three
comparison areas for each LCIC (univariate test),
based only on the observations classified as Good
(Table M-7a and b).
o One aggregate EDA sampling area (all seven EDA
sampling areas combined) compared to the three
comparison areas for all LCICs considered together
(multivariate test), based only on the
observations classified as Good (Table M-8a
and b).
DISCUSSION
NEIGHBORHOOD/COMPARISON AREA COMPARISONS
Tables M-l (a and b) and M-2 (a and b)and Figures M-l
and M-2 present the results of comparing the EDA
neighborhood/comparison areas. Although the statistical
comparisons reported in Volume III were performed based on
the EDA sampling areas (following the recommendation of the
peer review), during design of the field study, the
randomization of sampling sites (along with other effects
considered important) was performed at the neighborhood
level in the EDA. It is therefore statistically valid to
compare EDA neighborhoods with the other EDA neighborhoods
and the comparison areas. There are, however, two drawbacks
in performing the comparisons at the neighborhood level:
1. The number of comparisons increases from 45 (45 unique
pairs among the 7 EDA sampling areas and 3 comparison
areas) per LCIC to 120 per LCIC for the univariate test
and from 45 to 120 for the multivariate test. This
yields an increase in total comparisons from 405 to
1,080 and aggravates the multiple comparisons problem
(discussed in Appendix A of Volume III).
2. Because the number of observations per neighborhood is
less than or equal to the number of observations per
sampling area, the power to detect a difference of a
specific size (e.g., order of magnitude) is smaller at
the neighborhood level than at the sampling-area level.
REAGGREGATION OF SAMPLING AREAS 2 AND 3
As a background to the second sensitivity analysis described
herein, the original aggregation of EDA neighborhoods
defined EDA Sampling Area 2 as a combination of the two more
M-2
-------
southerly neighborhoods (2 and 3). The original EDA
Sampling Area 3 was defined as a combination of the two more
northerly neighborhoods (4 and 5). (See Figure M-l.)
Following the request of the TRC, new tables and a figure
are presented in this appendix snowing the results of
comparisons were conducted using recombined EDA Neigh-
borhoods 2, 3, 4, and 5. The purpose of these comparisons
is to explore the effect of the proximity of the Canal on
the results of these particular EDA sampling area compar-
isons. The new aggregation couples the two more westerly
neighborhoods closest to the Canal (Neighborhoods 2 and 4)
into one sampling area, denoted EDA Sampling Area 2", and
the two more easterly neighborhoods farther away from the
Canal (Neighborhoods 3 and 5) into another sampling area,
denoted 3'.
Tables M-3 (a through c) and M-4 (a and b) and Figure M-3
show the results of comparing different aggregations of EDA
Neighborhoods 2 through 5 with the other EDA sampling areas
and comparison areas.
COMPARISON AREAS COMBINED AND ALL EDA SAMPLING AREAS
COMBINED
Tables M-5 (a and b) and M-6 (a and b) present the results
of comparing the original seven EDA sampling areas with one
aggregate comparison area (all three comparison areas
combined). Tables M-7 (a and b) and M-8 (a and b) presented
the results of comparing one aggregate EDA sampling area
(all the original seven EDA sampling areas combined) with
the three separate comparison areas. The purpose of
generating Tables M-5 (a and b) through M-8 (a and b) is to
explore the pattern of differences between the EDA region
and the comparison areas on a grosser scale than that of the
original neighborhoods or sampling areas.
WDR356/009
M-3
-------
Table M-la
RESULTS FOR NONPARAMETRIC UNIVARIATE
NEIGHBORHOOD/COMPARISON AREA TO NEIGHBORHOOD/COMPARISON AREA COMPARISONS
LCIC
DCS
TCB
Neigh-
borhood/
Comparisoi
Area
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
Median
i Cone.
(ppb) 1
1.01
1.01
0.42 -*•+
0.33 ++
0.42 -M-
0.39 -H-
0.36 -H-
0.41 ++
0.39 ++•
0.36 +•+•
0.42 -H-
0.31 ++•
0.38 ++
0.37 ++
0.39 ++
0.36 +•+
0.43 ++
8.67
8.67
1.24 ++
0.84 ++•
0.91 ++
0.79 -M-
0.43 ++
0.85 ++
0.59 ++
0.40 ++
0.59 ++
0.32 ++
0.39 -n-
0.56 ++
0.65 ++
0.13 -M-
0.60 ++
EDA Neighborhoods
2
0.42
—
o
o
o
•f+
o
o
•f-f
o
•f-f
o
o
o
o
•f-f
1.24
—
o
o
o
++
o
+
++
•f
•f-f
•f-f
•f+
++
•f
•f-f
3
0.33
—
o
o
o
o
o
0
o
o
o
o
o
0
o
o
0.84
—
o
o
o
•f-f
o
o
++
•f
•f-f
++
•f-f
•f
•f
4"f
4
0.42
—
o
0
o
+
o
o
o
o
o
o
o
o
o
•*•+
0.91
—
o
o
o
•f-f
o
o
++
+
++
++
•f
•f
+
•f-f
5
0.39
—
o
0
0
+
o
0
o
o
o
o
0
o
o
+
0.79
—
o
o
0
++•
o
o
o
0
•f-f
•f-f
•f
0
o
++
6
0.36
—
—
0
-
-
—
-
o
—
o
o
-
—
-
o
0.43
—
—
—
—
—
0
o
o
0
o
o
o
-
o
•n-
7
0.41
—
o
0
o
o
++
o
4-
O
+
O
o
o
o
o
0.85
—
0
0
o
o
0
o
+
0
•f
-f
o
o
o
•H-
8
0.39
—
o
o
o
o
+
o
o
o
+
•f
o
o
o
+
0.59
—
-
o
o
0
o
o
o
0
++
+
o
0
o
+•+
9
0.36
—
—
o
o
o
o
-
o
o
o
o
o
0
o
o
0.40
—
—
—
—
o
o
-
o
-
o
0
o
—
0
•M-
10
0.42
—
o
0
0
o
++
o
o
o
++
o
0
o
o
+
0.59
—
-
-
-
0
o
o
o
+
++
++
o
o
o
++
11
0.31
—
—
o
o
o
o
-
-
o
—
o
o
—
o
o
0.32
—
—
—
—
—
o
-
—
o
—
o
—
—
0
++
12
0.38
—
o
o
o
o
o
o
-
o
o
o
o
o
o
o
0.39
—
—
—
—
—
o
-
-
o
—
o
—
—
—
++
13
0.37
—
o
o
0
o
+ -
o
o
o
o
o
o
o
o
o
0.56
—
—
—
-
-
o
0
o
o
o
++
•f-f
o
o
+4-
Comparison Areas
221
0.39
—
o
o
o
o
++
0
o
o
o
+-f
o
o
o
-1"*-
0.65
—
—
-
-
o
•f
o
o
-f+
o
•f+
•H-
O
O
-M-
225
0.43
—
o
o
o
o
+
0
0
o
o
o
o
0
o
o
0.60
—
-
-
-
o
o
o
o
o
o
o
++•
o
o
++
C&T
0.36
—
—
0
—
-
0
o
-
o
-
o
o
o
—
o
0.13
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
WDR356/054/1
-------
Neigh-
borhood/ Median
Comparison Cone.
LCIC Area (ppb)
TeCB
CNP
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
11.48
1.39
1.11
0.99
1.08
0.38
0.82
0.48
0.32
0.64
0.30
0.31
0.53
0.53
0.05
0.61
ND
0.04
0.04
0.02
0.04
0.04
0.07
0.04
0.07
ND
0.04
0.07
0.06
0.06
0.03
0.07
Table M-la
(continued)
EDA Neighborhoods
1
11.48
++
++
++
++
++
++
•M-
•M-
++
•H-
++
++
++
++
++
ND
o
0
o
o
o
o
o
o
o
o
o
-
o
-
o
2
1.39
o
+
o
++
+
++
•M-
++
•M-
•M-
•M-
++
++
++
0.04
0
o
o
o
o
o
o
o
0
o
o
o
o
0
o
3
1.11
o
o
o
++
+•
+
++
+
++
•n-
++
++
++
++
0.04
0
0
o
0
o
0
o
o
o
o
o
-
o
-
o
4
0.99
-
o
0
++
o
0
++
+
++
++
++
++
+
++
0.02
0
o
o
o
o
o
o
o
o
0
-
—
o
—
o
5
1.08
0
0
o
++
o
+
++
0
++
++
++
++
o
++
0.04
0
o
0
o
o
o
0
o
o
o
o
-
o
o
o
6
0.38
—
—
—
—
-
o
0
o
0
o
o
-
0
++
0.04
o
0
o
0
o
o
0
o
o
o
o
-
o
o
o
7
0.82
-
-
o
o
+
o
0
o
+
++
0
0
o
++
0.07
o
o
0
o
o
o
o
o
o
0
o
o
o
o
0
8
0.48
—
-
o
-
0
o
o
o
0
o
o
o
0
++
0.04
o
o
o
o
o
o
0
o
o
o
0
-
o
o
0
9
0.32
—
—
—
—
0
o
0
-
o
o
0
-
o
++
0.07
o
o
o
o
o
0
o
o
0
0
o
o
0
0
o
10
0.64
—
-
-
o
0
o
o
+
++
++
o
0
o
•n-
ND
0
o
o
o
o
o
0
o
o
0
o
—
o
-
0
11
0.30
—
—
—
—
o
-
o
o
—
o
—
—
-
++
0.04
o
o
o
o
o
o
o
o
o
o
-
-
-
-
o
12
0.31
—
—
—
—
o
—
o
0
—
0
—
—
—
++
0.07
0
o
0
+
o
o
o
o
o
o
+
o
0
o
•f
13
0.53
—
—
—
—
o
0
o
o
o
•»•+
++
o
0
•M-
0.06
+
0
+
++
+
+
o
+
o
++
+
0
o
o
++
Comparison Areas
221
0.53
—
—
—
—
+
o
o
+
0
++
++
o
o
++
0.06
0
o
o
o
o
o
o
o
o
o
+
o
o
o
o
225
0.61
—
—
-
o
o
0
o
0
o
+
++
o
0
++
0.07
+
0
•f
++
o
o
o
0
o
+
+
0
o
o
•f
C&T
0.05
—
—
—
—
—
—
—
—
—
—
—
—
—
—
0.03
0
o
o
o
o
o
0
o
o
o
o
-
—
o
-
WDR356/054/2
-------
Neigh-
borhood/ Median
Comparison Cone.
LCIC Area (ppb)
a-BHC
d-BHC
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
8.25
0.33
0.50
0.20
0.23
0.12
0.04
0.13
0.16
0.15
0.06
0.08
0.14
0.18
ND
0.11
1.13
0.10
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Table M-la
(continued)
EDA Neighborhoods
1234
8.25 0.33 0.50 0.20
++ o -
++ o
++ + +
++ O 0 0
++ + + O
++ O + O
•M- + + 0
++ ++ ++ o
++ ++ ++ +
-n- ++ ++ ++
++ +•»• ++ ++
++ ++ ++ O
++ ++ ++ 0
++ + ++ o
1.13 0.10 ND ND
++ o o
++ O 0
++ o o
++ ++ O 0
++ ++ ++• +
++ ++ 0 O
•M- -f O O
++ ++ + +
++ ++ -M- ++
++ ++ ++ ++
•M- ++ ++ -n-
++ ++ ++ +
++ ++ ++ ++
++ ++ ++ o
++ ++ -H- ++
5
0.23
o
o
o
o
0
o
0
o
+
o
o
o
o
ND
—
0
o
o
o
o
o
+
0
o
o
+
o
++
6
0.12
-
-
o
o
0
o
o
o
o
0
o
o
o
ND
—
—
-
o
o
0
o
0
o
o
o
+
o
•n-
7
0.04
o
-
0
0
0
o
0
o
o
o
o
o
o
ND
—
o
o
o
o
0
o
+
o
o
o
+
o
•M-
8
0.13
-
-
0
0
o
o
o
0
o
o
o
o
o
ND
-
o
o
o
o
o
0
+
o
o
o
•«••«-
o
++
9
0.16
—
—
o
o
o
0
o
o
0
o
o
0
o
ND
—
-
-
o
o
o
o
o
o
o
o
o
o
++
10
0.15
—
—
-
0
o
0
o
0
0
o
o
o
o
ND
—
—
—
-
o
-
-
o
o
o
o
0
0
o
11
0.06
—
—
—
-
o
o
o
o
o
o
o •
-
o
ND
—
—
—
o
o
o
o
o
o
o
o
0
o
++
12
0.08
—
—
—
o
o
o
0
o
o
o
o
o
o
ND
—
—
—
o
o
o
o
o
o
o
o
o
o
+
13
0.14
—
—
o
o
o
o
o
o
o
o
o
o
o
ND
—
—
-
0
o
o
o
o
o
o
o
o
o
++
Comparison Areas
221
0.18
—
—
0
0
0
o
0
o
o
+
o
o
o
ND
—
—
—
-
-
-
—
o
o
o
o
o
o
o
225
0.11
-
—
0
0
0
o
0
o
0
o
o
o
0
++
ND
—
—
o
o
o
o
o
o
o
o
o
o
0
++
C&T
ND
—
—
—
—
--
—
—
—
—
—
—
—
--
™"
ND
—
—
—
—
—
—
—
—
o
—
-
—
o
—
WDR356/054/3
-------
Neigh-
borhood/ Median
Comparison Cone.
LCIC Area (ppb)
b-BHC
g-BHC
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
11.58
0.16
0.39
0.13
0.08
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
1.73
0.05
0.09
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Table M-la
(continued)
EDA Neighborhoods
1 2
11.58 0.16
++
++ o
++• o
++ o
++ o
++ 0
++ o
++ ++
++ o
+•+ +
++• ++
++ o
+•+• ++
++ o
+•*• •*•+
1.73 0.05
++
++ o
++ o
++ o
++ 0
++ o
++ 0
++ ++
++ ++
+•+ ++
++ -M-
+•+ +•+
++• ++
++ 4-
3
0.39
o
o
o
o
+•
0
++•
+
++•
++
++
++•
+•
++
0.09
o
0
o
+•
+
o
+•+•
+•+•
++
++
++•
++•
o
4
0.13
o
o
0
0
o
o
•f-f
0
+
++
0
++
o
++
ND
o
o
o
0
0
o
++•
++•
++
+•+
+
+•+•
o
5
0.08
0
o
o
0
0
o
o
o
+
++
o
+
0
++
ND
o
0
o
0
o
0
0
+
o
+
o
+
o
6
ND
o
0
o
o
0
0
+
o
0
++
o
+
o
•n-
ND
0
-
o
o
o
0
0
+
o
+
o
+
o
7
ND
o
-
0
o
o
0
o
o
0
0
o
o
o
++
ND
o
-
o
o
o
o
o
o
o
0
o
o
o
8
ND
o
0
0
o
o
o
o
o
o
+
o
o
o
++
ND
o
o
o
o
o
o
o
•f
o
+
o
+
o
9
ND
—
—
—
o
-
o
o
0
o
o
o
o
-
++
ND
—
—
—
o
0
0
o
o
0
o
o
o
o
10
ND
0
-
o
o
o
0
0
o
0
+
o
o
o
++
ND
—
—
—
-
-
o
-
o
o
0
o
o
-
11
ND
-
—
-
-
o
o
0
o
o
o
o
o
o
++
ND
—
—
—
o
o
o
o
o
o
o
o
o
o
12
ND
—
—
—
—
—
o
-
o
-
o
—
o
—
+
ND
—
—
—
-
-
o
-
o
o
o
o
o
o
13
ND
o
— -
o
0
o
o
o
o
o
o
++
o
o
++
ND
—
—
-
o
o
o
o
o
o
o
o
o
o
Comparison Areas
221
ND
—
—
—
-
-
0
0
o
o
0
0
o
-
++
ND
—
—
—
-
-
o
-
0
o
o
o
o
o
225
ND
o
-
0
0
o
o
0
+
o
0
++
o
+
++
ND
-
0
o
o
o
o
o
0
+
o
o
o
o
C&T
ND
—
—
—
—
—
—
—
—
—
—
-
—
—
—
ND
—
—
—
—
—
—
—
-
-
-
—
—
—
—
WDR356/054/4
-------
Median
Comparison Cone.
LCIC Area (ppb)
TOTALS ++
+
o
-
—
All Symbols
Table M-la
(continued)
EDA Neighborhoods
1
105
0
12
2
1
120
2
50
26
37
0
7
120
3
43
15
53
2
7
120
4
31
13
63
4
9
120
5
16
12
83
1
8
120
6
7
6
73
13
21
120
7
8
9
90
5
8
120
8
8
10
87
7
8
120
Q
5
1
80
9
25
120
10
10
5
73
15
17
120
11
5
1
65
14
35
120
12
4
6
63
8
40
120
13
14
7
75
4
20
120
Comparison Areas
221
13
5
67
11
24
120
225
10
10
81
7
12
120
C&T
0
0
22
10
88
120
a.
++ = Column neighborhood/comparison area > Row neighborhood/comparison area at 0.01 significance level.
+ = Column neighborhood/comparison area > Row neighborhood/comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column neighborhood/comparison area < Row neighborhood/comparison area at 0.05 significance level.
— = Column neighborhood/comparison area < Row neighborhood/comparison area at 0.01 significance level.
b. All test results reported are based on two-sided p-values.
c. The First entry in each column is median concentration; ND indicates non-detect.
d. 221 = Census Tract 221.
225 = Census Tract 225.
C&T Cheektowaga and Tonawanda.
e. Based on observations classified as Good.
WDR356/054/5
-------
Table M-lb
TWO-SIDED p-VALUES FOR NONPARAMETRIC
UNIVARIATE NEIGHBORHOOD/COMPARISON AREA .
TO NEIGHBORHOOD/COMPARISON AREA COMPARISONS '
LCIC
DCB
TCB
Neigh-
borhood/
Comparison
Area
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
EDA Neighborhoods
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2
0.000
1.000
0.276
0.236
0.407
0.003
0.486
0.730
0.009
0.225
0.009
0.100
0.167
0.275
0.241
0.004
0.000
1.000
0.246
0.131
0.092
0.000
0.231
0.049
0.001
0.016
0.000
0.000
0.000
0.002
0.042
0.000
3
0.000
0.276
1.000
0.470
0.382
0.136
0.471
0.308
0.536
0.365
0.486
0.664
0.589
0.445
0.694
0.120
0.000
0.246
1.000
0.814
0.743
0.000
0.283
0.222
0.001
0.050
0.000
0.000
0.002
0.039
0.035
0.000
4
0.000
0.236
0.470
1.000
0.679
0.021
0.391
0.667
0.098
0.907
0.112
0.170
0.563
0.814
0.529
0.006
0.000
0.131
0.814
1.000
0.986
0.000
0.991
0.286
0.004
0.043
0.000
0.000
0.011
0.039
0.046
0.000
5
0.000
0.407
0.382
0.679
1.000
0.038
0.952
0.921
0.363
0.559
0.091
0.269
0.959
0.582
0.981
0.018
0.000
0.092
0.743
0.986
1.000
0.005
0.834
0.305
0.102
0.205
0.000
0.001
0.034
0.070
0.275
0.000
6
0.000
0.003
0.136
0.021
0.038
1.000
0.009
0.043
0.824
0.004
0.796
0.258
0.020
0.003
0.025
0.915
0.000
0.000
0.000
0.000
0.005
1.000
0.081
0.134
0.925
0.303
0.132
0.557
0.090
0.011
0.077
0.000
7
0.000
0.486
0.471
0.391
0.952
0.009
1.000
0.659
0.045
0.753
0.029
0.240
0.300
0.708
0.482
0.055
0.000
0.231
0.283
0.991
0.834
0.081
1.000
0.448
0.027
0.525
0.017
0.013
0.137
0.393
0.271
0.000
8
0.000
0.730
0.308
0.667
0.921
0.043
0.659
1.000
0.053
0.967
0.013
0.028
0.618
0.828
0.359
0.021
0.000
0.049
0.222
0.286
0.305
0.134
0.448
1.000
0.180
0.703
0.005
0.022
0.453
0.504
0.846
0.000
9
0.000
0.009
0.536
0.098
0.363
0.824
0.045
0.053
1.000
0.079
0.685
0.957
0.136
0.060
0.356
0.976
0.000
0.001
0.001
0.004
0.102
0.925
0.027
0.180
1.000
0.049
0.499
0.451
0.059
0.006
0.090
0.000
10
0.000
0.225
0.365
0.907
0.559
0.004
0.753
0.967
0.079
1.000
0.004
0.147
0.429
0.728
0.742
0.023
0.000
0.016
0.050
0.043
0.205
0.303
0.525
0.703
0.049
1.000
0.007
0.001
0.962
0.224
0.714
0.000
11
0.000
0.009
0.486
0.112
0.091
0.796
0.029
0.013
0.685
0.004
1.000
0.645
0.075
0.002
0.168
0.640
0.000
0.000
0.000
0.000
0.000
0.132
0.017
0.005
0.499
0.007
1.000
0.879
0.005
0.000
0.073
0.000
12
0.000
0.100
0.664
0.170
0.269
0.258
0.240
0.028
0.957
0.147
0.645
1.000
0.389
0.109
0.529
0.091
0.000
0.000
0.000
0.000
0.001
0.557
0.013
0.022
0.451
0.001
0.879
1.000
0.002
0.000
0.009
0.000
13
0.000
0.167
0.589
0.563
0.959
0.020
0.300
0.618
0.136
0.429
0.075
0.389
1.000
0.780
0.611
0.078
0.000
0.000
0.002
0.011
0.034
0.090
0.137
0.453
0.059
0.962
0.005
0.002
1.000
0.167
0.906
0.000
Comparison
221
0.000
0.275
0.445
0.814
0.582
0.003
0.708
0.828
0.060
0.728
0.002
0.109
0.780
1.000
0.757
0.007
0.000
0.002
0.039
0.039
0.070
0.011
0.393
0.504
0.006
0.224
0.000
0.000
0.167
1.000
0.368
0.000
225
0.000
0.241
0.694
0.529
0.981
0.025
0.482
0.359
0.356
0.742
0.168
0.529
0.611
0.757
1.000
•0.194
0.000
0.042
0.035
0.046
0.275
0.077
0.271
0.846
0.090
0.714
0.073
0.009
0.906
0.368
1.000
0.000
Areas
CST
0.000
0.004
0.120
0.006
0.018
0.915
0.055
0.021
0.976
0.023
0.640
0.091
0.078
0.007
0.194
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
WDR356/056/1
-------
LCIC
TeCB
CNP
Neigh-
borhood/
Comparison
Area
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
CST
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.948
0.901
0.775
0.958
0.275
0.506
0.948
0.497
0.619
0.806
0.097
0.036
0.110
0.047
0.693
2
0.000
1.000
0.274
0.039
0.120
0.000
0.018
0.001
0.000
0.001
0.000
0.000
0.000
0.000
0.002
0.000
0.948
1.000
0.779
0.800
0.444
0.803
0.636
0.868
0.589
0.354
0.841
0.222
0.133
0.232
0.078
0.645
3
0.000
0.274
1.000
0.383
0.343
0.000
0.027
0.019
0.000
0.011
0.000
0.000
0.000
0.000
0.001
0.000
0.901
0.779
1.000
0.771
0.617
0.761
0.637
0.435
0.136
0.678
0.752
0.119
0.018
0.120
0.041
0.989
4
0.000
0.039
0.383
1.000
0.987
0.000
0.206
0.053
0.000
0.030
0.000
0.000
0.000
0.000
0.012
0.000
0.775
0.800
0.771
1.000
0.962
0.149
0.725
0.775
0.497
0.714
0.549
0.037
0.002
0.086
0.010
0.835
5
0.000
0.120
0.343
0.987
1.000
0.000
0.235
0.034
0.001
0.055
0.000
0.000
0.001
0.001
0.069
0.000
0.958
0.444
0.617
0.962
1.000
0.575
0.939
0.991
0.682
0.494
0.570
0.456
0.048
0.214
0.093
0.770
Table M-lb
(continued)
EDA Neighborhoods
6
0.000
0.000
0.000
0.000
0.000
1.000
0.046
0.610
0.499
0.168
0.150
0.302
0.084
0.040
0.054
0.000
0.275
0.803
0.761
0.149
0.575
1.000
0.598
0.818
0.636
0.194
0.166
0.303
0.024
0.597
0.107
0.103
7
0.000
0.018
0.027
0.206
0.235
0.046
1.000
0.459
0.113
0.911
0.012
0.009
0.372
0.246
0.636
0.000
0.506
0.636
0.637
0.725
0.939
0.598
1.000
0.630
0.678
0.435
0.223
0.070
0.073
0.371
0.060
0.596
8
0.000
0.001
0.019
0.053
0.034
0.610
0.459
1.000
0.418
0.785
0.062
0.077
0.583
0.546
0.316
0.000
0.948
0.868
0.435
0.775
0.991
0.818
0.630
1.000
0.053
0.422
0.569
0.244
0.022
0.346
0.139
0.772
9
0.000
0.000
0.000
0.000
0.001
0.499
0.113
0.418
1.000
0.013
0.555
0.634
0.055
0.029
0.061
0.000
0.497
0.589
0.136
0.497
0.682
0.636
0.678
0.053
1.000
0.124
0.279
0.400
0.131
0.649
0.078
0.454
10
0.000
0.001
0.011
0.030
0.055
0.168
0.911
0.785
0.013
1.000
0.001
0.001
0.509
0.342
0.769
0.000
0.619
0.354
0.678
0.714
0.494
0.194
0.435
0.422
0.124
1.000
0.701
0.053
0.003
0.089
0.018
0.478
11
0.000
0.000
0.000
0.000
0.000
0.150
0.012
0.062
0.555
0.001
1.000
0.789
0.008
0.001
0.038
0.000
0.806
0.841
0.752
0.549
0.570
0.166
0.223
0.569
0.279
0.701
1.000
0.017
0.013
0.017
0.025
0.401
12
0.000
0.000
0.000
0.000
0.000
0.302
0.009
0.077
0.634
0.001
0.789
1.000
0.001
0.001
0.001
0.000
0.097
0.222
0.119
0.037
0.456
0.303
0.070
0.244
0.400
0.053
0.017
1.000
0.159
0.866
0.875
0.034
13
0.000
0.000
0.000
0.000
0.001
0.084
0.372
0.583
0.055
0.509
0.008
0.001
1.000
0.676
0.836
0.000
0.036
0.133
0.018
0.002
0.048
0.024
0.073
0.022
0.131
0.003
0.013
0.159
1.000
0.316
0.345
0.001
Comparison
221
0.000
0.000
0.000
0.000
0.001
0.040
0.246
0.546
0.029
0.342
0.001
0.001
0.676
1.000
0.346
0.000
0.110
0.232
0.120
0.086
0.214
0.597
0.371
0.346
0.649
0.089
0.017
0.866
0.316
1.000
0.824
0.091
225
0.000
0.002
0.001
0.012
0.069
0.054
0.636
0.316
0.061
0.769
0.038
0.001
0.836
0.346
1.000
0.000
0.047
0.078
0.041
0.010
0.093
0.107
0.060
0.139
0.078
0.018
0.025
0.875
0.345
0.824
1.000
0.010
Areas
C&T
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.693
0.645
0.989
0.835
0.770
0.103
0.596
0.772
0.454
0.478
0.401
0.034
0.001
0.091
0.010
1.000
WDR356/056/2
-------
Neigh-
borhood/
Comparison
LCIC Area
a-BHC 1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
d-BHC 1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
CST
Table M-lb
(continued)
EDA Neighborhoods
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2
0.000
1.000
0.568
0.036
0.141
0.018
0.054
0.010
0.005
0.003
0.000
0.000
0.000
0.000
0.017
0.000
0.000
1.000
0.126
0.053
0.002
0.001
0.003
0.016
0.001
0.000
0.000
0.000
0.000
0.000
0.001
0.000
3
0.000
0.568
1.000
0.028
0.372
0.010
0.010
0.010
0.002
0.002
0.000
0.000
0.000
0.000
0.003
0.000
0.000
0.126
1.000
0.785
0.060
0.003
0.104
0.189
0.016
0.000
0.001
0.000
0.001
0.000
0.006
0.000
4
0.000
0.036
0.028
1.000
0.950
0.242
0.133
0.323
0.122
0.047
0.003
0.005
0.174
0.134
0.213
0.000
0.000
0.053
0.785
1.000
0.133
0.030
0.275
0.365
0.045
0.001
0.010
0.002
0.049
0.000
0.067
0.000
5
0.000
0.141
0.372
0.950
1.000
0.625
0.082
0.566
0.438
0.143
0.026
0.108
0.341
0.143
0.610
0.000
0.000
0.002
0.060
0.133
1.000
0.592
0.700
0.661
0.828
0.031
0.167
0.198
0.777
0.043
0.906
0.000
6
0.000
0.018
0.010
0.242
0.625
1.000
0.332
0.984
0.915
0.917
0.111
0.172
0.822
0.630
0.410
0.000
0.000
0.001
0.003
0.030
0.592
1.000
0.734
0.431
0.481
0.062
0.250
0.198
0.839
0.032
0.573
0.000
7
0.000
0.054
0.010
0.133
0.082
0.332
1.000
0.322
0.948
0.305
0.795
0.969
0.408
0.354
0.391
0.000
0.000
0.003
0.104
0.275
0.700
0.734
1.000
0.549
0.548
0.039
0.138
0.161
0.404
0.049
0.472
0.000
8
0.000
0.010
0.010
0.323
0.566
0.984
0.322
1.000
0.908
0.956
0.219
0.523
0.607
0.381
0.538
0.000
0.000
0.016
0.189
0.365
0.661
0.431
0.549
1.000
0.577
0.024
0.128
0.065
0.544
0.006
0.405
0.000
9
0.000
0.005
0.002
0.122
0.438
0.915
0.948
0.908
1.000
0.794
0.343
0.084
0.870
0.517
0.804
0.000
0.000
0.001
0.016
0.045
0.828
0.481
0.548
0.577
1.000
0.103
0.588
0.444
0.641
0.216
0.701
0.001
10
0.000
0.003
0.002
0.047
0.143
0.917
0.305
0.956
0.794
1.000
0.315
0.124
0.671
0.743
0.748
0.000
0.000
0.000
0.000
0.001
0.031
0.062
0.039
0.024
0.103
1.000
0.280
0.337
0.063
0.670
0.098
0.107
11
0.000
0.000
0.000
0.003
0.026
0.111
0.795
0.219
0.343
0.315
1.000
0.968
0.114
0.045
0.555
0.000
0.000
0.000
0.001
0.010
0.167
0.250
0.138
0.128
0.588
0.280
1.000
0.827
0.425
0.905
0.541
0.005
12
0.000
0.000
0.000
0.005
0.108
0.172
0.969
0.523
0.084
0.124
0.968
1.000
0.111
0.093
0.167
0.000
0.000
0.000
0.000
0.002
0.198
0.198
0.161
0.065
0.444
0.337
0.827
1.000
0.187
0.319
0.413
0.011
13
0.000
0.000
0.000
0.174
0.341
0.822
0.408
0.607
0.870
0.671
0.114
0.111
1.000
0.468
0.616
0.000
0.000
0.000
0.001
0.049
0.777
0.839
0.404
0.544
0.641
0.063
0.425
0.187
1.000
0.071
0.772
0.000
Comparison
221
0.000
0.000
0.000
0.134
0.143
0.630
0.354
0.381
0.517
0.743
0.045
0.093
0.468
1.000
0.346
0.000
0.000
0.000
0.000
0.000
0.043
0.032
0.049
0.006
0.216
0.670
0.905
0.319
0.071
1.000
0.075
0.070
225
0.000
0.017
0.003
0.213
0.610
0.410
0.391
0.538
0.804
0.748
0.555
0.167
0.616
0.346
1.000
0.000
0.000
0.001
0.006
0.067
0.906
0.573
0.472
0.405
0.701
0.098
0.541
0.413
0.772
0.075
1.000
0.001
Areas
CST
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.001
0.107
0.005
0.011
0.000
0.070
0.001
1.000
WDR356/056/3
-------
Neigh-
borhood/
Comparison
LCIC
b-BHC
g-BHC
Area
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2
0.000
1.000
0.106
0.901
0.862
0.586
0.385
0.243
0.005
0.145
0.015
0.000
0.084
0.005
0.570
0.000
0.000
1.000
0.804
0.443
0.300
0.120
0.123
0.150
0.002
0.000
0.001
0.000
0.002
0.001
0.032
0.000
3
0.000
0.106
1.000
0.179
0.169
0.084
0.015
0.115
0.001
0.021
0.001
0.000
0.003
0.000
0.032
0.000
0.000
0.804
1.000
0.642
0.374
0.027
0.045
0.194
0.004
0.000
0.004
0.000
0.002
0.000
0.060
0.000
4
0.000
0.901
0.179
1.000
0.700
0.296
0.299
0.220
0.005
0.087
0.014
0.000
0.199
0.000
0.409
0.000
0.000
0.443
0.642
1.000
0.477
0.120
0.183
0.689
0.006
0.001
0.009
0.001
0.036
0.002
0.137
0.000
5
0.000
0.862
0.169
0.700
1.000
0.594
0.180
0.475
0.062
0.095
0.016
0.002
0.366
0.012
0.822
0.000
0.000
0.300
0.374
0.477
1.000
0.609
0.595
0.763
0.159
0.016
0.090
0.027
0.330
0.031
0.746
0.000
Table M-lb
(continued)
EDA Neighborhoods
6
0.000
0.586
0.084
0.296
0.594
1.000
0.272
0.549
0.013
0.381
0.078
0.000
0.574
0.016
0.755
0.000
0.000
0.120
0.027
0.120
0.609
1.000
0.558
0.677
0.099
0.029
0.054
0.046
0.367
0.020
0.711
0.000
7
0.000
0.385
0.015
0.299
0.180
0.272
1.000
0.698
0.126
0.607
0.364
0.073
0.896
0.192
0.611
0.000
0.000
0.123
0.045
0.183
0.595
0.558
1.000
0.384
0.206
0.143
0.397
0.510
0.709
0.596
0.560
0.001
8
0.000
0.243
0.115
0.220
0.475
0.549
0.698
1.000
0.192
0.505
0.396
0.047
0.673
0.323
0.667
0.000
0.000
0.150
0.194
0.689
0.763
0.677
0.384
1.000
0.058
0.012
0.118
0.035
0.483
0.026
0.519
0.000
9
0.000
0.005
0.001
0.005
0.062
0.013
0.126
0.192
1.000
0.379
0.446
0.540
0.062
0.649
0.013
0.005
0.000
0.002
0.004
0.006
0.159
0.099
0.206
0.058
1.000
0.646
0.913
0.630
0.218
0.688
0.080
0.014
10
0.000
0.145
0.021
0.087
0.095
0.381
0.607
0.505
0.379
1.000
0.413
0.014
0.528
0.137
0.681
0.000
0.000
0.000
0.000
0.001
0.016
0.029
0.143
0.012
0.646
1.000
0.428
0.279
0.062
0.476
0.029
0.044
11
0.000
0.015
0.001
0.014
0.016
0.078
0.364
0.396
0.446
0.413
1.000
0.179
0.463
0.927
0.082
0.000
0.000
0.001
0.004
0.009
0.090
0.054
0.397
0.118
0.913
0.428
1.000
0.891
0.279
0.688
0.153
0.010
12
0.000
0.000
0.000
0.000
0.002
0.000
0.073
0.047
0.540
0.014
0.179
1.000
0.001
0.197
0.001
0.037
0.000
0.000
0.000
0.001
0.027
0.046
0.510
0.035
0.630
0.279
0.891
1.000
0.212
0.764
0.077
0.002
13
0.000
0.084
0.003
0.199
0.366
0.574
0.896
0.673
0.062
0.528
0.463
0.001
1.000
0.100
0.695
0.000
0.000
0.002
0.002
0.036
0.330
0.367
0.709
0.483
0.218
0.062
0.279
0.212
1.000
0.201
0.484
0.000
Comparison
221
0.000
0.005
0.000
0.000
0.012
0.016
0.192
0.323
0.649
0.137
0.927
0.197
0.100
1.000
0.033
0.000
0.000
0.001
0.000
0.002
0.031
0.020
0.596
0.026
0.688
0.476
0.688
0.764
0.201
1.000
0.095
0.007
225
0.000
0.570
0.032
0.409
0.822
0.755
0.611
0.667
0.013
0.681
0.082
0.001
0.695
0.033
1.000
0.000
0.000
0.032
0.060
0.137
0.746
0.711
0.560
0.519
0.080
0.029
0.153
0.077
0.484
0.095
1.000
0.000
Areas
C&T
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.005
0.000
0.000
0.037
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.001
0.000
0.014
0.044
0.010
0.002
0.000
0.007
0.000
1.000
a. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
b. Based on observations classified as Good.
WDR356/056/4
-------
Table M-2a
RESULTS FOR NONPARAMETRIC MULTIVARIATE NEIGHBORHOOD/COMPARISON AREA
TO NEIGHBORHOOD/COMPARISON AREA COMPARISONS ' ' 'Q
Neigh-
borhood/
Comparison
Area
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
12345
«» __ __ — —
-M- 0 0 O
4-4- 0 0 O
4-4- O O O
44- o o o
4-4- 4-4- 4- 4-4- o
+4- 0 0 0 O
4- 4-4- 0 0 O
4-4- +4 4-4- 4-4- +
4-4- 4 0 4- O
4-4 4-+ 4-4- 4-4- +4-
4-4- 4-4- 4-4- 44 4-4-
+4 4-4 ++ 44- 4-4-
4-4- 4-4 4-4- ++• ++
++ + ++ + 0
+4- +-f ++ +4- 4-4-
EDA
6
^ _
—
-
—
o
-
o
o
0
0
o
-
—
o
4-4-
Ne ighborhoods
7
__
0
o
o
0
4-
o
o
0
o
o
o
0
o
8
_
—
0
o
o
0
0
4-
0
o
o
4-
O
• o
"
9
__
—
—
—
-
0
o
-
o
o
o
o
-
o
10
.._
-
o
-
0
o
o
o
0
o
4-4-
-
-
o
11
__
—
—
—
—
o
o
0
0
o
o
o
-
o
12
__
—
—
—
—
o
o
o
o
—
o
-
—
o
13
_ —
—
—
—
—
4-
o
-
o
4-
o
4-
4-
o
4-4-
Comparison Areas
221
225
C&T
++
o
o
4-
4-
4-
O
O
o
o
o
o
o
o
o
++
a.
b.
c. 221
d.
= Column neighborhood/comparison area > Row neighborhood/comparison area at 0.01 significance level.
= Column neighborhood/comparison area > Row neighborhood/comparison area at 0.05 significance level.
= No significant difference at 0.05 significance level.
= Column neighborhood/comparison area < Row neighborhood/comparison area at 0.05 significance level.
= Column neighborhood/comparison area < Row neighborhood/comparison area at 0.01 significance level.
The direction of the difference between the EDA Neighborhoods and Comparison Areas is based on the sign of the sum of the elements of
the 8x1 vector of rank-sums (Volume III, Appendix B).
= Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
Based on observations classified as Good.
WDR356/057
-------
Table M-2b
TWO-SIDED p-VALUES FOR NONPARAMETRIC
MULTIVARIATE NEIGHBORHOOD/COMPARISON AREA .
TO NEIGHBORHOOD/COMPARISON AREA COMPARISONS '
Neighborhood/
Comparison
Area
1
2
3
4
5
6
7
8
9
10
11
12
13
221
225
C&T
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.033
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2
0.000
1.000
0.056
0.582
0.076
0.003
0.239
0.003
0.001
0.038
0.003
0.000
0.000
0.000
0.050
0.000
3
0.000
0.056
1.000
0.251
0.867
0.037
0.125
0.105
0.008
0.081
0.000
0.000
0.000
0.000
0.000
0.000
4
0.000
0.582
0.251
1.000
0.378
0.006
0.135
0.101
0.002
0.018
0.001
0.000
0.000
0.000
0.017
0.000
5
0.000
0.076
0.867
0.378
1.000
0.128
0.168
0.496
0.038
0.139
0.001
0.001
0.008
0.002
0.198
0.000
EDA Neighborhoods
6
0.000
0.003
0.037
0.006
0.128
1.000
0.034
0.675
0.432
0.054
0.298
0.092
0.015
0.000
0.095
0.000
7
0.000
0.239
0.125
0.135
0.168
0.034
1.000
0.732
0.305
0.153
0.145
0.054
0.055
0.220
0.223
0.000
8
0.033
0.003
0.105
0.101
0.496
0.675
0.732
1.000
0.043
0.419
0.134
0.180
0.035
0.054
0.129
0.000
9
0.000
0.001
0.008
0.002
0.038
0.432
0.305
0.043
1.000
0.072
0.851
0.081
0.192
0.038
0.140
0.000
10
0.000
0.038
0.081
0.018
0.139
0.054
0.153
0.419
0.072
1.000
0.207
0.003
0.024
0.022
0.113
0.000
11
0.000
0.003
0.000
0.001
0.001
0.298
0.145
0.134
0.851
0.207
1.000
0.405
0.282
0.024
0.177
0.000
12
0.000
0.000
0.000
0.000
0.001
0.092
0.054
0.180
0.081
0.003
0.405
1.000
0.032
0.003
0.056
0.000
13
0.000
0.001
0.000
0.000
0.008
0.015
0.055
0.035
0.192
0.024
0.282
0.032
1.000
0.022
0.829
0.000
Comparison Areas
221
0.000
0.000
0.000
0.000
0.002
0.000
0.220
0.054
0.038
0.022
0.024
0.003
0.022
1.000
0.017
0.000
225
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
C&T
0.000
0.050
0.000
0.017
0.198
0.095
0.223
0.129
0.140
0.113
0.177
0.056
0.829
0.017
1.000
0.000
d221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
Based on observations classified as Good.
WDR356/058
-------
Table M-3a
RESULTS FOR NONPARAMETRIC UNIVARIATE
LCIC
DCB
TCB
TeCB
Sampling/
Comparison
Area
1
2'
3'
4
5
6
7
221
225
C&T
1
2'
3'
4
5
6
7
221
225
C&T
1
2'
3'
4
5
6
7
221
225
C&T
SAMPLING/COMPARISON AREA TO
COMPARISONS WITH SAMPLING AREAS
Median
Cone.
(ppb) 1
1.01
1.01
0.42 -M-
0.38 ++
0.36 ++
0.39 ++
0.36 ++
0.41 ++
0.39 ++
0.43 ++
0.36 ++
8.67
8.67
0.98 ++
0.83 ++
0.43 ++
0.52 ++
0.37 ++
0.58 ++
0.65 ++
0.60 ++
0.13 ++
11.48
11.48
1.11 ++
1.09 -M-
0.38 ++
0.43 -M-
0.31 ++
0.54 -M-
0.53 ++
0.61 ++
0.05 ++
EDA
2' 3'
0.42 0.38
—
o
o
++ +
0 O
++ o
o o
0 O
O 0
++ -f
0.98 0.83
__ __
o
o
++ ++
++ +
++ ++
++ ++
•M- +
+ +
++ ++
1.11 1.09
—
o
0
++ ++
++ ++
++ ++
++ ++
++ •(•+
++ ++
++ -n-
SAMPLING/COMPARISQN. AREA ,
2 and 3 REDEFINED3 'D'TJ7U'e't
Sampling
4
0.36
—
—
-
-
0
—
—
-
0
0.43
_-
—
—
o
0
o
-
o
++
0.38
—
—
—
o
0
o
-
o
++
Areas
5
0.39
—
o
0
+
+
o
0
o
o
0.52
_-
—
-
o
++
o
0
0
•t"f
0.43
—
—
—
o
++
o
0
o
•f+
Comparison Areas
6
0.36
—
—
o
o
-
-
—
0
o
0.37
--
—
—
o
—
—
—
—
**
0.31
—
—
—
o
—
—
—
—
++
7
0.41
—
o
o
++
o
+
o
o
•f
0.58
--
—
—
0
o
++
o
o
*+
0.54
—
—
—
o
o
•M-
O
o
++
221
0.39
—
o
o
++
o
•f+
0
o
-1"*-
0.65
--
—
-
+
o
++
o
o
•n-
0.53
—
—
—
+
o
++
o
o
++
225
0.43
—
o
o
+
0
o
o
0
o
0.60
--
-
-
0
o
++
0
o
++
0.61
—
—
—
0
o
++
o
o
++
C&T
0.36
—
—
-
o
0
o
-
—
0
0.13
--
—
—
—
—
—
—
—
—
0.05
—
—
—
—
—
—
—
—
—
WDR356/059/1
-------
a-BHC
d-BHC
Sampling/
Comparison
Area
1
2'
3'
4
5
6
7
221
225
C&T
1
21
3'
4
5
6
7
221
225
C&T
1
2'
3'
4
5
6
7
221
225
C&T
Median
Cone.
(ppb)
ND
0.04
0.04
0.04
0.06
0.05
0.06
0.06
0.07
0.03
8.25
0.27
0.35
0.12
0.13
0.07
0.14
0.18
0.11
ND
1.13
ND
ND
ND
ND
ND
ND
ND
ND
ND
Table M-3a
(Continued)
EDA Sampling
1 2'
ND 0.04
o
o
o o
o o
o o
o o
o
o o
-
o o
8.25 0.27
++
44 O
44 O
44 44
44 44
44 44
44 44
44 4
44 44
1.13 ND
++
44 O
44 44
44 44
44 44
44 44
44 44
44 44
44 44
3'
0.04
o
o
o
o
. o
o
o
-
o
0.35
o
o
44
44
44
44
4
44
ND
o
4
O
44
44
44
4
44
4
0.04
o
o
o
o
o
o
o
o
o
0.12
o
o
o
o
o
o
o
++
ND
—
-
0
o
o
4
O
44
Areas
5
0.06
o
o
o
o
o
o
o
-
o
0.13
—
—
o
o
o
o
0
++
ND
—
0
o
4
O
4
O
+4
6
0.05
o
o
o
o
o
0
o
o
o
0.07
__
~
o
o
-
-
o
44
ND
—
—
o
-
o
o
o
44
7
0.06
o
4
o
o
o
0
o
0
+
0.14
__
—
0
o
4
o
o
44
ND
—
—
o
o
o
o
o
44
Comparison Areas
221
0.06
o
o
o
o
o
o
o
o
o
0.18
o
o
4-
o
ND
o
o
o
o
225
0.07
o
o
o
0.11
o
o
o
o
o
ND
o
o
o
o
o
C&T
0.03
o
o
o
o
o
o
ND
ND
WDR356/059/2
-------
Table M-3a
(Continued)
LCIC
b-BHC
g-BHC
TOTALS
Sampling/
Comparison
Area
1
2'
3'
4
5
6
7
221
225
C&T
1
2'
3'
4
5
6
7
221
225
C&T
++
+
0
-
—
Median
(ppb) 1
11.58
11.58
0.13 ++
0.19 -M-
ND ++
ND ++
ND -t-t-
ND ++
ND ++
ND ++
ND ++
1.73
1.73
0.01 -M-
ND ++
ND ++
ND ++
ND ++
ND ++
ND ++
ND ++
ND ++
63
0
8
1
0
EDA Sampling Areas
21
0.13
—
o
o
++
++
+
++
o
++
0.01
—
o
o
++
4-+
4-+
4-4
+
++
37
4
22
1
8
3'
0.19
—
o
o
+4
44
++
+4-
o
44
ND
—
o
o
+
44
44
++
o
++
29
9
26
1
7
4
ND
—
o
o
o
•n-
o
+
o
++
ND
. —
o
o
o
+
o
+
o
++
7
4
40
6
15
5
ND
—
—
—
o
0
o
o
0
++
ND
—
--
-
o
0
o
o
0
++
8
4
41
3
16
6
ND
—
—
—
—
o
—
o
—
•f+
ND
—
—
—
-
o
o
o
-
•n-
6
0
27
7
32
7
ND
—
-
—
o
o
++
o
0
*+
ND
—
--
—
o
o
o
o
o
+4-
10
5
38
1
18
Comparison Areas
221
ND
—
—
—
-
o
o
o
-
++
ND
—
--
—
-
o
o
o
o
++
10
3
35
6
18
225
ND
—
o
o
o
o
++•
o
+
++
ND
—
-
o
o
0
+•
o
o
++
10
7
39
6
10
C&T
ND
—
—
—
—
—
—
—
—
—
ND
—
--
—
—
—
—
—
—
—
0
0
12
4
56
All Symbols
72
72
72
72
72
72
72
a++ = Column sampling/comparison area > Row sampling/comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling/comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling/comparison area at 0.01 significance level.
All test results reported are based on two-sided p-values.
°The First entry in each column is median concentration; ND indicates non-detect.
d221 = Census Tract 221.
225 = Census Tract 225.
C&T Cheektowaga and Tonawanda.
eBased on observations classified as Good.
fEDA Sampling Area 2' = EDA Neighborhoods 2 and 4; EDA Sampling Area 3' = EDA Neighborhoods 3 and 5.
WDR356/059/3
72
72
72
-------
Sampling/
Comparison
LCIC Area
DCB
TCB
TeCB
CKP
a-BHC
1
2'
3'
4
5
6
7
221
225
C&T
1
21
3'
4
5
6
7
221
225
C&T
1
2'
3'
4
5
6
7
221
225
C&T
1
21
3'
4
5
6
7
221
225
C&T
1
2'
3'
4
5
6
7
221
225
C&T
Table M-3b
TWO-SIDED p-VALUES FOR NONPARAMETRIC
UNIVARIATE SAMPLING/COMPARISON AREA
TO SAMPLING/COMPARISON AREA COMPARISONS
WITH SAMPLING AREAS 2 AND 3 REDEFINEDa/D'c
EDA Sampling Areas
Comparison Areas
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.855
0.996
0.275
0.519
0.175
0.181
0.110
0.047
0.693
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2'
0.000
1.000
0.384
0.001
0.231
0.008
0.174
0.391
0.214
0.001
0.000
1.000
0.371
0.000
0.003
0.000
0.000
0.002
0.011
0.000
0.000
1.000
0.595
0.000
0.000
0.000
0.000
0.000
0.001
0.000
0.855
1.000
0.715
0.279
0.454
0.103
0.050
0.052
0.006
0.778
0.000
1.000
0.501
0.051
0.002
0.000
0.000
0.005
0.031
0.000
3'
0.000
0.384
1.000
0.020
0.982
0.197
0.545
0.453
0.877
0.019
0.000
0.371
1.000
0.000
0.017
0.000
0.001
0.018
0.035
0.000
0.000
0.595
1.000
0.000
0.000
0.000
0.000
0.000
0.001
0.000
0.996
0.715
1.000
0.474
0.559
0.273
0.053
0.065
0.015
0.986
0.000
0.501
1.000
0.059
0.002
0.000
0.000
0.001
0.024
0.000
4
0.000
0.001
0.020
1.000
0.029
0.349
0.003
0.003
0.025
0.915
0.000
0.000
0.000
1.000
0.140
0.219
0.089
0.011
0.077
0.000
0.000
0.000
0.000
1.000
0.404
0.127
0.055
0.040
0.054
0.000
0.275
0.279
0.474
1.000
0.599
0.789
0.246
0.597
0.107
0.103
0.000
0.051
0.059
1.000
0.623
0.069
0.890
0.630
0.410
0.000
5
0.000
0.231
0.982
0.029
1.000
0.044
0.639
0.617
0.779
0.058
0.000
0.003
0.017
0.140
1.000
0.002
0.964
0.268
0.828
0.000
0.000
0.000
0.000
0.404
1.000
0.006
0.228
0.446
0.250
0.000
0.519
0.454
0.559
0.599
1.000
0.305
0.166
0.311
0.018
0.605
0.000
0.002
0.002
0.617
1.000
0.164
0.516
0.270
0.942
0.000
6
0.000
0.008
0.197
0.349
0.044
1.000
0.035
0.009
0.312
0.146
0.000
0.000
0.000
0.219
0.002
1.000
0.000
0.000
0.005
0.000
0.000
0.000
0.000
0.127
0.006
1.000
0.000
0.000
0.000
0.000
0.175
0.103
0.273
0.789
0.305
1.000
0.368
0.383
0.322
0.193
0.000
0.000
0.000
0.069
0.164
1.000
0.029
0.028
0.192
C.OOO
7
0.000
0.174
0.545
0.003
0.639
0.035
1.000
0.992
0.556
0.022
0.000
0.000
0.001
0.089
0.964
0.000
1.000
0.123
0.808
0.000
0.000
0.000
0.000
0.055
0.228
0.000
1.000
0.446
0.982
0.000
0.181
0.050
0.053
0.246
0.166
0.368
1.000
0.965
0.762
0.018
0.000
0.000
0.000
0.890
0.516
0.029
1.000
0.509
0.577
0.000
221
0.000
0.391
0.453
0.003
0.617
0.009
0.992
1.000
0.757
0.007
0.000
0.002
0.018
0.011
0.268
0.000
0.123
1.000
0.368
0.000
0.000
0.000
0.000
0.040
0.446
0.000
0.446
1.000
0.346
0.000
0.110
0.052
0.065
0.597
0.311
0.383
0.965
1.000
0.824
0.091
0.000
0.005
0.001
0.630.
0.270
0.028
0.509
1.000
0.346
0.000
225
0.000
0.214
0.877
0.025
0.779
0.312
0.556
0.757
1.000
0.194
0.000
0.011
0.035
0.077
0.828
0.005
0.808
0.368
1.000
0.000
0.000
0.001
0.001
0.054
0.250
0.000
0.982
0.346
1.000
0.000
0.047
0.006
0.015
0.107
0.018
0.322
0.762
0.824
1.000
0.010
0.000
0.031
0.024
0.410
0.942
0.192
0.577
0.346
1.000
0.000
C&T
0.000
0.001
0.019
0.915
0.058
0.146
0.022
0.007
0.194
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.693
0.778
0.986
0.103
0.605
0.193
0.018
0.091
0.010
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
WDR356/060/1
-------
Sampling/
Comparison
LCIC Area
d-BHC
b-BHC
g-BHC
1
2'
3'
4
5
6
7
221
225
C&T
1
2'
3'
4
5
6
7
221
225
C&T
1
21
3'
4
5
6
7
221
225
C&T
Table M-3b
(continued)
EDA Sampling Areas
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2'
0.000
1.000
0.082
0.003
0.001
0.000
0.000
0.000
0.003
0.000
0.000
1.000
0.399
0.353
0.004
0.000
0.020
0.000
0.366
0.000
0.000
1.000
0.752
0.079
0.003
0.000
0.000
0.000
0.030
0.000
3'
0.000
0.082
1.000
0.016
0.113
0.000
0.000
0.000
0.039
0.000
0.000
0.399
1.000
0.165
0.002
0.000
0.006
0.000
0.118
0.000
0.000
0.752
1.000
0.081
0.016
0.000
0.000
0.001
0.141
0.000
4
0.000
0.003
0.016
1.000
0.829
0.108
0.351
0.032
0.573
0.000
0.000
0.353
0.165
1.000
0.051
0.000
0.411
0.016
0.755
0.000
0.000
0.079
0.081
1.000
0.404
0.011
0.098
0.020
0.711
0.000
5
0.000
0.001
0.113
0.829
1.000
0.034
0.169
0.015
0.560
0.000
0.000
0.004
0.002
0.051
1.000
0.099
0.293
0.453
0.118
0.004
0.000
0.003
0.016
0.404
1.000
0.158
0.466
0.278
0.461
0.000
6
0.000
0.000
0.000
0.108
0.034
1.000
0.474
0.563
0.372
0.008
0.000
0.000
0.000
0.000
0.099
1.000
0.002
0.247
0.001
0.000
0.000
0.000
0.000
0.011
0.158
1.000
0.434
0.921
0.035
0.002
7
0.000
0.000
0.000
0.351
0.169
0.474
1.000
0.206
0.667
0.002
0.000
0.020
0.006
0.411
0.293
0.002
1.000
0.072
0.632
0.000
0.000
0.000
0.000
0.098
0.466
0.434
1.000
0.457
0.140
0.000
Comparison Areas
221
0.000
0.000
0.000
0.032
0.015
0.563
0.206
1.000
0.075
0.070
0.000
0.000
0.000
0.016
0.453
0.247
0.072
1.000
0.033
0.000
0.000
0.000
0.001
0.020
0.278
0.921
0.457
1.000
0.095
0.007
225
0.000
0.003
0.039
0.573
0.560
0.372
0.667
0.075
1.000
0.001
0.000
0.366
0.118
0.755
0.118
0.001
0.632
0.033
1.000
0.000
0.000
0.030
0.141
0.711
0.461
0.835
0.140
0.095
1.000
0.000
C&T
0.000
0.000
0.000
0.000
0.000
0.008
0.002
0.070
0.001
1.000
0.000
0.000
0.000
0.000
0.000
0.004
0.000
0.000
0.000
1.000
0.000
0.000
0.000
0.000
0.000
0.002
0.000
0.007
0.000
1.000
a. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
b. Based on observations classified as Good.
c. EDA Sampling area 2' = EDA Neighborhoods 2 and 4;
EDA Sampling area 3" = EDA Neighborhoods 3 and 5.
WDR356/060/2
-------
Table M-3c
SUMMARY OF NONPARAMETRIC UNIVARIATE
SAMPLING/COMPARISON AREA TO SAMPLING/COMPARISON AREA
a.b.c.d.e
COMPARISONS WITH SAMPLING AREAS 2 and 3 REDEFINED ' ' ' '
Totals Over All LCICs
Sampling/
Comparison
Area
1 ++•
+
o
-
— •
2' ++
+
o
-
— •
31 ++
+
O
-
--
4 ++
+
0
-
--
5 ++
+
o
-
—
6 ++
+
o
-
—
7 ++
+
0
-
—
221 ++
+
0
-
— •
225 ++
+
0
-
—
EDA Sampling Areas
1
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
1
0
0
7
0
0
1
0
11
0
0
1
0
7
0
0
8
0
0
4
0
4
0
0
5
0
2
1
0
7
0
1
0
0
5
1
1
0
0
6
0
2
0
0
2
3
2
0
1
3'
0
0
1
0
7
0
0
8
0
0
2
2
4
0
0
3
2
3
0
0
5
0
2
1
0
6
0
2
0
0
5
1
2
0
0
1
3
3
1
0
f
0
0
1
0
7
0
0
4
0
4
0
0
4
2
2
0
0
7
1
0
1
1
6
0
0
0
0
7
1
0
0
3
2
2
1
0
0
7
1
0
5
0
0
1
0
7
0
0
2
0
6
0
0
3
2
3
0
0
7
1
0
2
2
4
0
0
0
0
8
0
0
0
1
7
0
0
0
0
7
1
0
6_
0
0
1
0
7
0
0
1
0
7
0
0
2
1
5
0
0
6
1
1
0
0
4
2
2
0
0
3
2
3
0
0
4
1
3
0
0
4
1
3
2
0
0
i
0
7
0
1
1
1
5
0
0
2
0
6
1
0
7
0
0
0
1
7
0
0
3
0
3
2
0
0
0
8
0
0
0
0
8
0
0
Comparison Areas
221
0
0
1
0
7
0
0
2
0
6
0
2
2
0
4
1
0
2
5
0
0
1
7
0
0
3
0
3
2
0
0
0
8
0
0
0
0
7
1
0
225
0
1
0
0
7
1
0
2
4
2
0
1
3
3
1
0
1
6
1
0
0
0
8
0
0
3
0
4
1
0
0
0
8
0
0
0
0
7
1
0
C&T
0
0
1
0
7
0
1
0
0
7
0
0
1
1
6
0
0
2
0
6
0
0
2
0
6
0
0
2
0
6
0
0
0
2
6
0
0
2
0
6
0
0
1
1
6
WDR356/061/1
-------
Table M-3c
(continued)
Totals Over All LCICs
Sampling/
Comparison EDA Sampling Areas Comparison Areas
Area T ]F J^ T | 57 221 125
C&T ++776666666
+ 00100000 0
o 11122202 1
00000020 1
00000000 0
a. ++:Column sampling/comparison area > Row sampling/comparison area at
0.01 significance level
+: Column sampling/comparison area > Row sampling/comparison area at
0.05 significance level
o: No significant difference at 0.05 significance level
-: Column sampling/comparison area < Row sampling/comparison area at
0.05 significance level
—: Column sampling/comparison area < Row sampling/comparison area at
0.01 significance level
b. All test results reported are based on two-sided p-values.
c. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
d. Based on observations classified as Good.
e. EDA Sampling area 2' = EDA Neighborhoods 2 and 4;
EDA Sampling area 3' = EDA Neighborhoods 3 and 5.
WDR356/061/2
-------
Table M-4a
RESULTS FOR NONPARAMETRIC MULTIVARIATE SAMPLING/COMPARISON AREA
TO SAMPLING/COMPARISON AREA COMPARISONS WITH
SAMPLING AREAS 2 AND 3 REDEFINED8' '''
Sampling/
Comparison
Area
1
2
3
4
5
6
7
221
225
C&T
1
4-4-
4-4-
4-4-
4-4-
4-4-
4-4-
+4-
4-4-
4-4-
2'
__
+
4-4-
4-4-
4-4-
4-4-
4-4-
4-4-
4-4-
3'
— _
-
4-4-
4-4-
4-4-
+4-
4-4-
4-4-
4-4-
EDA Sampling Areas
4 5
— _ _ _
—
—
o
o
o o
—
4-
0
++
6 7
__ B_
—
—
0 ++
0 +
4-4-
—
4-4-
O
4-4- 4-4-
Comparison Areas
221
C&T
225
o
4-
4-
O
4-
a.
++ = Column sampling/comparison area > Row sampling/comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling/comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling/comparison area at 0.01 significance level.
b. The direction of the difference between the EDA Neighborhoods and Comparison Areas is based on the sign of the sum of the
elements of the 8x1 vector of rank-sums (Volume III, Appendix B).
c. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
d. Based on observations classified as Good.
e. EDA Sampling area 2' = EDA Neighborhoods 2 and 4;
EDA Sampling area 3' = EDA Neighborhoods 3 and 5.
WDR356/062
-------
Table M-4b
TWO-SIDED p-VALUES FOR NONPARAMETRIC
MULTIVARIATE SAMPLING/COMPARISON AREA
TO SAMPLING/COMPARISON AREA COMPARISONS
WITH SAMPLING AREAS 2 AND 3 REDEFINED 'D'c
Sampling/
Comparison
Area
I
2
3
4
5
6
7
221
225
C&T
EDA Sampling Areas
1
1.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2'
0.000
1.000
0.027
0.000
0.000
0.000
0.000
0.000
0.002
0.000
3'
0.000
0.027
1.000
0.006
0.001
0.000
0.000
0.000
0.001
0.000
4
0.000
0.000
0.006
1.000
0.084
0.086
0.009
0.000
0.095
0.000
5
0.000
0.000
0.001
0.084
1.000
0.220
0.028
0.037
0.023
0.000
6
0.000
0.000
0.000
0.086
0.220
1.000
0.008
0.001
0.019
0.000
7
0.000
0.000
0.000
0.009
0.028
0.008
1.000
0.004
0.778
0.000
a. 221 = Census Tract 221.
225 = Census Tract 225.
CST = Cheektowaga and Tonawanda.
b. Based on observations classified as Good.
c. EDA Sampling area 2' = EDA Neighborhoods 2 and 4.
EDA Sampling area 3' = EDA Neighborhoods 3 and 5.
Comparison Areas
221
0.000
0.000
0.000
0.000
0.037
0.001
0.004
1.000
0.017
0.000
225
0.000
0.002
0.001
0.095
0.023
0.019
0.778
0.017
1.000
0.000
CST
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.000
WDR356/063
-------
Table M-5a
RESULTS FOR NONPARAMETRIC UNIVARIATE _ h _ fl _
EDA SAMPLING AREA TO COMBINED COMPARISON AREAS COMPARISONSa/D'c'Q'e
LCIC
DCB
TCB
TeCB
Comparison
Area
Combined
Combined
Combined
Median
Cone.
(ppb)
0.38
EDA Sampling Areas
1.01 0.39 0.40 0.36 0.39 0.36 0.41
++00-000
8.67 0.91 0.89 0.43 0.52 0.37 0.58
0.39 ++ ++ ++ o++ o ++
11.48 1.17 1.06 0.38 0.43 0.31 0.54
0.33 ++ ++ ++ + ++ o -M-
CNP Combined
a-BHC Combined
d-BHC Combined
b-BHC Combined
g-BHC Combined
Totals ++
+
o
-
—
All Symbols
ND
0.06 o
8.25
. ND ++
1.13
ND ++
11.58
ND ++
1.73
ND -n-
7
0
1
0
0
8
0.04
o
0.40
++
0.04
•n-
0.20
++
0.09
++
6
0
2
0
0
8
0.04
o
0.22
++
ND
++
0.13
++
ND
++
6
0
2
0
0
8
0.04
o
0.12
++
ND
++
ND
++
ND
++
4
1
2
1
0
8
0.06
0
0.13
*
ND
++
ND
o
ND
o
3
1
4
0
0
8
0.05
0
0.07
o
ND
o
ND
o
ND
o
0
0
8
0
0
8
0.06
o
0.14
++
ND
0
ND
+*
ND
0
4
0
4
0
0
8
a.
++ = Column sampling/comparison area > Row sampling/comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling/comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling/comparison area at 0.01 significance level.
b. All test results reported are based on two-sided p-values.
c. The first entry in each column is median concentration; ND indicates non-detect.
d. Combined comparison areas consist of Census Tract 221, Census Tract 225, and CheeKtowaga
and Tonawanda.
e. Based on observations classified as Good.
WDR356/064
-------
Table M-5b
TWO-SIDED p-VALUES FOR NONPARAMETRIC UNIVARIATE
EDA SAMPLING AREA TO COMBINED COMPARISON AREAS COMPARISONS^
LCIC
DCB
TCB
TeCB
CNP
a-BHC
d-BHC
b-BHC
g-BHC
Comparison
Area
Combined
Combined
Combined
Combined
Combined
Combined
Combined
Combined
EDA Sampling Areas
1
0.000
0.000
0.000
0.118
0.000
0.000
0.000
0.000
2
0.144
0.000
0.000
0.067
0.000
0.000
0.000
0.000
3
0.108
0.000
0.000
0.058
0.000
0.000
0.000
0.000
4
0.033
0.110
0.032
0.759
0.001
0.006
0.001
0.003
5
0.495
0.001
0.001
0.146
0.014
0.002
0.303
0.127
6
0.385
0.618
0.404
0.769
0.166
0.448
0.302
0.948
7
0.203
0.000
0.000
0.434
0.000
0.114
0.007
0.339
a. Combined comparison areas consist of Census Tract 221, Census Tract 225, and Cheektowaga
and Tonawanda.
b. Based on observations classified as Good.
WDR356/065
-------
Table M-6a
RESULTS FOR NONPARAMETRIC MULTIVARIATE EDA SAMPLING AREA
TO COMBINED COMPARISON AREAS COMPARISON3'
Comparison EDA Sampling Areas
Area I 23 4
Combined •*•+ ++ ++
a.
++ = Column sampling/comparison area > Row sampling/comparison area at 0.01
significance level
+ = Column sampling/comparison area > Row sampling/comparison area at 0.05
significance level
o = No significant difference at 0.05 significance level
- = Column sampling/comparison area < Row sampling/comparison area at 0.05
significance level
— = Column sampling/comparison area < Row sampling/comparison area at 0.01
significance level
b. The direction of the difference between the EDA Sampling Areas and combined
Comparison Areas is based on the sign of the sum of the elements of the 8x1
vector of rank-sums (Volume III, Appendix B).
c. Combined comparison areas consist of Census Tract 221, Census Tract 225, and
Cheektowaga and Tonawanda.
d. Based on observations classified as Good.
WDR356/066
-------
Table M-6b
TWO-SIDED p-VALUES FOR NONPARAMETRIC MULTIVARIATE EDA SAMPLING AREA
TO COMBINED COMPARISON AREAS COMPARISONS3'b
Comparison EDA Sampling Areas
Area I23
-------
Table M-7a
RESULTS FOR NONPARAMETRIC UNIVARIATE COMBINED
EDA SAMPLING AREAS TO COMPARISON AREA COMPARISONS3' 'C''6
LCIC
DCB
TCB
TeCB
CNP
a-BHC
d-BHC
b-BHC
g-BHC
Totals
Comparison Area
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
221
225
C&T
++
+
o
-
—
Median
Cone.
(ppb)
0.39
0.43
0.36
0.65
0.60
0.13
0.53
0.61
0.05
0.06
0.07
0.03
0.18
0.11
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Combined EDA
Sampling Area
0.40
o
o
++
0.64
o
0
++
0.61
o
o
++
0.05
o
—
o
0.18
o
o
++
ND
++
o
++
ND
++
O
++
ND
++
O
++
10
0
13
1
0
All Symbols
24
WDR356/068/1
-------
Table M-7a
(Continued)
a. ++ = Column sampling/comparison area > Row sampling/
comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/
comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level,
- = Column sampling/comparison area < Row sampling/
comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling/
comparison area at 0.01 significance level.
b. All test results reported are based on two-sided
p-values.
c. The First entry in each column is median concentration;
ND indicates non-detect.
d. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
e. Based on observations classified as Good.
WDR356/068
-------
Table M-7b
TWO-SIDED p-VALUES FOR NONPARAMETRIC UNIVARIATE
COMBINED EDA SAMPLING AREAS TO COMPARISON AREAS
COMPARISONS3'
Comparison Combined EDA
LCIC Area Sampling Area
DCB 221 0.757
225 0.573
C&T 0.003
TCB 221 0.923
225 0.372
C&T 0.000
TeCB 221 0.108
225 0.440
C&T 0.000
CNP 221 0.164
225 0.021
C&T 0.244
a-BHC 221 0.186
225 0.088
C&T 0.000
d-BHC 221 0.001
225 0.075
C&T 0.000
b-BHC 221 0.002
225 0.615
C&T 0.000
g-BHC 221 0.005
225 0.384
C&T 0.000
a. Combined comparison areas consist of Census Tract 221,
Census Tract 225, and Cheektowaga and Tonawanda.
b. Based on observations classified as Good.
WDR356/069
-------
Table M-8a
RESULTS FOR NONPARAMETRIC MULTIVARIATE COMBINED,
EDA SAMPLING AREAS TO COMPARISON AREAS COMPARISONS3' /C/
Combined EDA
Comparison Areas Sampling Area
221 ++
225 o
C&T ++
a.
++ = Column sampling/comparison area > Row sampling/
comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/
comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling
comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling
comparison area at 0.01 significance level.
b. The direction of the difference between the EDA Combined
Sampling Areas and Comparison Areas is based on the sine
of the sum of the elements of the 8x1 vector of rank-
sums (Volume III, Appendix B).
c. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
d. Based on observations classified as Good.
WDR356/070
-------
Table M-8a
RESULTS FOR NONPARAMETRIC MULTIVARIATE COMBINED
EDA SAMPLING AREAS TO COMPARISON AREAS COMPARISONS*' /C'
Combined EDA
Comparison Areas Sampling Area
221 ++
225 o
C&T ++
a.
++ = Column sampling/comparison area > Row sampling/
comparison area at 0.01 significance level.
+ = Column sampling/comparison area > Row sampling/
comparison area at 0.05 significance level.
o = No significant difference at 0.05 significance level.
- = Column sampling/comparison area < Row sampling
comparison area at 0.05 significance level.
— = Column sampling/comparison area < Row sampling
comparison area at 0.01 significance level.
b. The direction of the difference between the EDA Combined
Sampling Areas and Comparison Areas is based on the sign
of the sum of the elements of the 8x1 vector of rank-
sums (Volume III, Appendix B).
c. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
d. Based on observations classified as Good.
WDR356/070
-------
Table M-8b
TWO SIDED p-VALUES FOR NONPARAMETRIC MULTIVARIATE
COMBINED EDA SAMPLING AREAS TO COMPARISON
AREAS COMPARISONS3'
Combined
Comparison Area Sampling Area
221 0.000
225 0.180
C&T 0.000
a. 221 = Census Tract 221.
225 = Census Tract 225.
C&T = Cheektowaga and Tonawanda.
b. Based on observations classified as Good.
WDR356/071
-------
00000
o o o o o o
t t
t t t
-e-e
il
0 0 t * *
t * * ° °
•tt-tttt
44444°
0 * * 0 f *
11
v\
\\
•*•
r~
t t
^.^ SAMPUNG
AREA
BOUNDARY
11
VL
t t t
t
t
t
t
LEGEND:
LCIC concentration in this EDA
^. neighborhood is greater than that
' in the comparison area at the 0.05
level of significance (two-sided)
O No significant difference
-1 LCIC concentration in this EDA
i neighborhood is less than that in
the comparison area at the 0.05
level of significance (two-sided)
Top Line = Census Tract 221
Middle Line = Census Tract 225
Bottom Line = Cheektowaga and Tonawanda
LCICs are in the following order on the line:
r
LOVE CANAL
o t T o t T T T
o t t o T T o T
TttgTTtt
.'/
5
s. —
1 1 IS
_r
_r *
_r 2 v
\
\
^ l»W,
0 t Q „ t t t
0 t T 0 0 t
0 t t 0 0
t t t I T t
Scale:1" = 780'
1
\
1 0 * * 0 * *
! o t t ft
1 n f f o T f
Q
SOURCE: Neighborhood boundaries adapted from
the Proposed Habitatoility Criteria document
(NYSDOH and DHHS/CDC, 1986)
Figure M-1
SOIL ASSESSMENT — INDICATOR CHEMICAL4
SUMMARY OF UNIVARIATE COMPARISONS:
EDA NEIGHBORHOOD TO COMPARISON AREAfl
-------
LEGEND:
t
Scale: 1" = 780'
LCIC concentrations in this EDA
neighborhood are generally greater
than those in the comparison area
at the 0.05 level of significance
(two-sided)
Q No significant difference
I LCIC concentrations in this EDA
I neighborhood are generally less
i than those in the comparison area
Y at the 0.05 level of significance
(two-sided)
Comparison area symbols are in the
following order
CT221 CT225 CK & TON
NOTE: The direction of the difference between
the EDA. neighborhoods and comparison areas is
based on the sign of the sum of the elements of
the 8 x 1 vector of rank-sums (Appendix B). "•
SOURCE: Neighborhood boundaries adapted from
the Proposed Habitability Criteria document
(NYSDOH and DHHS/CDC, 1986)
Figure M-2
SOIL ASSESSMENT — INDICATOR CHEMICALS
SUMMARY OF MULTIVARIATE COMPARISONS
(LCICsASAGROUP):
EDA NEIGHBORHOOD TO COMPARISON AREA
-------
n 4
t t t
o t t t
o o o o
T
t
t t t t t
o-J^t.
o-J-J-
t t t
t
t
T
t
t
t
t
MM
i
0
f
t
t
f
f
t
f
t
1
f
f
t
iilMir
t
t
t
LCIC concentration in this EDA
sampling area is greater than that in the
T comparison area at the 0.05 level of
significance (two-sided)
O No significant difference
LCIC concentration in this EDA
4 sampling area is less than that in the
comparison area at the 0.05 level of
significance (two-sided)
Top Line = Census Tract 221
Middle Line = Census Tract 225
Bottom Line = Cheektowaga and Tonawanda
LCICs are in the following order on the line:
fill i i i T
t t t t
t t t t
t
t
t
2'
t
t
t
t
t
t
t
MM
4
©_
•
t
t
t
t
t
t
t
t
•
t
t
HUM
t
t
t
NOTE: The direction of the difference between
the EDA and comparison areas is based on the
•ign of the sum of (he elements of the
6x^ vector of rank-sums (Volume III. Appendix B).
Figure M-3
SOIL ASSESSMENT-INDICATOR CHEMICALS
SUMMARY OF STATISTICAL (UNIVARIATE
AND MULTIVARIATE) COMPARISONS:
EDA SAMPLING AREAS 2 AND 3 TO COMPARISON
AREAS AND EDA SAMPLING AREAS 2' AND 3' TO
COMPARISON AREAS
-------
APPENDIX N
List of Errata
Soil Assessment—Indicator Chemicals
-------
Appendix N
LIST OF ERRATA
SOIL ASSESSMENT—INDICATOR CHEMICALS
Below is a list of errata for the subject report. Some of
these are straightforward typographical errors, but others
involve a change in content (Note—negative line numbers
indicate number of lines from the bottom of the page.)
SECTIONS 1-6
Page Line Erratum
1-2 2 "The EDA comparison areas" should read
"The EDA sampling areas"
1-3 1 "A total of 5320" should read "A total of
5329."
4-5 3 The random selection of sampling sites was
done at the neighborhood level in the EDA,
not at the sampling area level, as implied.
6-2 See attached Table 6-5, which shows the
corrections.
6-45 3 The superscripts "a,b,c" should read
"a,b,c,d"
6-48 3 The superscripts "a,b" should read "a,b,c"
APPENDIX A
There is a statement on page A-6, line 9, that ties the
habitability decision to the "minimum unacceptable
difference." It should be emphasized that the "delta" of
one order of magnitude that was used in the design of the
study (performed by CH2M HILL) is in no way related to any
suggested rule for a habitability decision (which will be
made by the Commissioner of Health for the State of New
York.
Other Appendix A errata are as follows:
N-l
-------
Paqe Line Erratum
A-3 "mean" should read "median"
A-5 "mean" should read "median"
A-8 Delete this page; it is a repeat of page A-6,
A-9 -19 "sampling area-" should read "EDA sampling
area-"
-16 "neighborhood" should read "sampling area"
-13 "neighborhood" should read "sampling area"
-12 "neighborhood" should read "sampling area"
A-10 10 "neighborhood" should read "sampling area"
APPENDIX B
Paqe Line Erratum
B-2 -14 The text after "analytical methods?" should
start a new paragraph that is not part of
question number 3.
B-ll 24 "(Anderson and McLean)" should read
"(Anderson and McLean, 1974)"
B-13 8 "extension to Eq. B.3" should read "extension
to Eq. B.2"
B-14 -1 In the lower right-hand box, on the last
line, the subscript "k" (lower case) should
be "K" (upper case)
B-15 13 "sampling areas" should read "(sampling
areas)"
B-16 5 "standard normal variable" should read
"standard normal random variable"
B-20 Table B-4—LCICs are mislabeled. Labels
should be:
1,2 Dichlorobenzene (1)
1,2,4 Trichlorobenzene (2)
1,2,3,4 Tetrachlorobenzene (3)
Chloronaphthalene (4)
alpha-BHC (5)
N-2
-------
delta-BHC (6)
beta-BHC (7)
gamma-BBC (8)
APPENDIX F
In Figure F-l, "Blank QC" should be "Blind QC," and in
Figure F-3, abbreviations "EMDC" should be "EMPC," and
"Performance Handling Blank" should be "Preparation Handling
Blank."
APPENDIX K
Tables K-l, K-2, and K-4 to K-14 are missing some or all of
the superscripts in the title that indicate the relevant
footnotes.
Table K-12 is a duplicate of Table K-ll in Volume III.
WDR277/G49
N-3
-------
ERRATA
Table 6-5
NUMBER OF FIELD QC SAMPLES RECEIVED AND ANALYZED BY LABORATORY
Analytical
Laboratory
1
2
3
4
6
7
8
a
FHB
14
12
11
12
11
10
6
Field QC Samples
Received
Splith
22
17
14
11
9
4
6
PHB
8
7
10
9
6
7
9
SSB
19
18
19
21
18
17
18
Field QC Samples
Analyzed
FHB
14+3d
12
11
0
11
10
6
Split PHB SSB
22+19
17
14
0
9
4
6
Totals 76 83 56° 130 676 731
FHB = Field handling blank
PHB = Preparation handling blank
SSB = Shipping and storage blank
Laboratory 5 was not selected for study participation.
b
One split was not sent to Laboratory 4.
£
A total of 57 pHBs were used; however, one was unextrudable.
Three of Laboratory 4's FHBs were analyzed by Laboratory 1.
Nine FHBs were not analyzed by Laboratory 4.
PHBs and SSBs were stored but not analyzed.
gA field spLit extracted by Laboratory 4 was analysed by Laboratory 1.
A total of 83 pairs of splits were sent for analysis. One-half of each
split was treated as a field sample and used in the statistical analysis,
The other half is represented in this table.
Valid results were obtained for both halves of 64 pairs of splits.
WDR277/049
N-4
-------
APPENDIX O
Integrated Data Base
! Soil Assessment—Indicator Chemicals
-------
Appendix 0
INTEGRATED DATA BASE
SOIL ASSESSMENT—INDICATOR CHEMICALS
An integrated data base was created using the data collected
from each phase of the soil assessment for indicator
chemicals. This data base is available by FOIA request. A
discussion of the data and information systems used to
create the data base is given in Appendix E of Volume III.
This appendix provides a brief description of the data base
and a data base element list so that potential users can be
aware of the data available.
DESCRIPTION
The integrated data base contains 1.8 million "cells" of
data and has undergone QA/QC review to identify and correct
errors. The integrated data base is composed of six master
files and four ancillary files:
o Field sample master file and ancillary file
o Initial calibration master file and ancillary file
o Continuing calibration master file and ancillary
file
o Quality control sample master file and ancillary
file
o Form I master file
o Blind quality control spike master file
The ancillary files each correspond to a master file and
contain the original values for GC/MS data corrected by the
analyst via the LabData Corrections subsystem.
A full user's guide will be available with the data base
through FOIA request to EPA Region II.
0-1
-------
KEY TO THE LOVE CANAL INTEGRATED DATA BASE
DATA ELEMENT LIST
Data Element ID—Identification of the form AA.Bnn, where,
AA is an abbreviation of the file:
FS = Field Sample Master File
FA = Field Sample Ancillary File
1C = Initial Calibration Master File
IA = Initial Calibration Ancillary File
CC = Continuing Calibration Master File
CA = Continuing Calibration Ancillary File
QC = Quality Control Master File
QA = Quality Control Ancillary File
Fl = Form 1 Master File
BQ = Blind Quality Control Master File
B is an alphabetical character assigned to a data
element group
nn is sequentially assigned within a data element group
Name—SAS data element name
Type—type of data element:
CHAR for character, or
NUM for numeric
Length—number of bytes used to store the data element
Dimensions (for an array)—shows the number of dimensions
for the data element array and the number of elements
in each dimension [e.g., (8,3) is a two-dimensional
array with eight rows and three columns)]
Description—text description of the data element, including
"by" phrase(s) to describe any array dimensions (e.g.,
by 8 LCICs)
Source System—identifies the system that provided the data
element values, either captured/reported or computed
(marked with a "*" prefix)
WDR361/031
0-2
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
******
FS.A01
FS.A02
*******
FS.B01
FS.B02
FS.B03
FS.B04
FS.B05
FS.B06
FS.B07
FS.B08
FS.C01
FS.C02
FS.C03
FS.C04
FS.C05
FS.C06
FS.C07
FS.D01
FS.D02
FS.003
FS.D04
FS.D05
FS.D06
FS.D07
FS.D08
FS.D09
FS.D10
FS.D11
FS.D12
FS.D13
*******
FS.E01
FS.E02
FS.E03
FS.E04
FS.E05
FS.E06
FS.E07
NAME
ARRAY TYPE
LENGTH
DESCRIPTION
SOURCE
Keys and Status **************************************************************************************
ALANALID
FSSTATUS
Inter-File
I CANAL ID
CCANALID
MSANALID
MO ANAL ID
MBANALID
RBANALID
QCANALID
CLEANHS
Intra-Fi le
HSID
FDUPID
FHBID
SSBIDF
SSBIDP
PHBID
PLSID
Chronology
FSFCDTE
FSFCTME
FSFSDTE
FSPLDTE
FSPMDTE
FSPMTME
FSPSDTE
FSALDTES
FSALDTEL
FSAEDTE
FSAADTE
FSAATME
FSDBDTE
CHAR
CHAR
Links/Secondary
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
Links/Secondary
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
23
1
Analysis Lab Sample Id (Lab Id + Lab Analysis Id)
Status Flag
LabData
IntDBBld
Keys *********************************************************************
23
23
23
23
23
23
23
Initial Calibration Id (Lab Id + Lab Analysis Id
Continuing Calibration/Performance Check Id
(Lab Id + Lab Analysis Id of CC/PC1)
Matrix Spike Id (Lab Id + Lab Analysis Id)
Matrix Spike Duplicate Id (Lab Id + Lab Analysis
Lab Method Blank Id (Lab Id + Lab Analysis Id)
Reagent Blank Id (Lab Id + Lab Analysis Id)
of IC1) LabOata
LabData
LabData
Id) LabData
LabData
LabData
Quality Control Id (Lab Id + Lab Analysis Id) not currently used
8 Original Project Id (HS Number) IntDBBld
6 Project Field Sample Id (HS #) SampTrac
6
6
6
6
6
6
8
8
8
8
8
8
8
8
8
8
8
8
8
Revised Site Id (of the form nnnn or nnnnR);
if no R suffix, resampling not done
Field Handling Blank Id (HS #)
Field Group Ship/Store Blank Id (HS #)
Prep Group Ship/Store Blank Id (HS #)
Prep Lab Handling Blank Id (HS #)
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
Prep Lab Split Id (HS #) SampTrac
Field Collection Date SampTrac
Field Collection Time
Field Ship Date
Prep Lab Login Date
Prep Lab Mix Date
Prep Lab Mix Time
Prep Lab Ship Date
Analysis Lab Login Date (Sample Tracking)
Analysis Lab Login Date (LabOata)
Analysis Lab Extract Date
Analysis Lab Analysis Date
Analysis Lab Analysis Time
Data Validation Date
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
LabOata
LabData
LabData
LabData
DataVal
Data Transfer Tracki ng ************************************** ";"'**************************************
FSGENDTE
FSGENTME
FSADDDTE
FSADDTME
FSUPDDTE
FSUPDTME
FSUPDCNT
NUM
NUM
NUM
NUM
NUM
NUM
NUM
8
8
8
8
8
8
8
LabData Gen Date
LabData Gen Time
DB Add Date
08 Add Time
08 Most Recent Update Date
DB Most Recent Update Time
DB Update Count
LabData
LabData
IntDBBld
IntDBBld
IntDBBld
IntDBBld
IntDBBld
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
*******
FS.F01
FS.F02
FS.F03
FS.F04
FS.F05
FS.F06
FS.F07
FS.F08
FS.F09
FS.F10
FS.F11
FS.F12
FS.F13
FS.F14
FS.F15
FS.F16
FS.F17
FS.F18
FS.F19
FS.F20
FS.F21
FS.F22
FS.F23
FS.F24
FS.F25
*******
FS.G01
FS.G02
FS.G03
FS.G04
FS.G05
FS.G06
FS.G07
FS.G08
FS.G09
FS.G10
FS.G11
FS.G12
FS.G13
FS.G14
FS.G15
FS.G16
FS.G17
NAME ARRAY
TYPE
LENGTH
DESCRIPTION
SOURCE
Field Sampling Data *********************************************************************************
COL FORM
COLFORMR
SAMPTYPE
SITENBR
STREET
MEDIA
XCOORD
YCOORD
XACTUAL
YACTUAL
LOCDIFF
AREA ID
NEIGHID
TEAMNBR
COLLECTR
SOILCOMP
TUBENBR
COLMLEN
DUPLICAT
FLDCOM (3)
FCOCFORM
FTRFFORM
FLFEDEX
FLSRMKS
FLPRVBLK
Preparation Lab Data
PREPLAB
PREPCOND
PREPCOMM
PLRECBY
SPFORM
MIXRNAME
MIXTEAM
PREPCOM (2)
TRAKCOM
SPLIT
PCOCFORM
PTRFFORM
MSFLAG
PLJARNBR
PLSPACE
PLPOSHI
PLSHMETH
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
NUM
NUM
NUM
NUM
NUM
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
NUM
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
5
5
6
4
70
1
8
8
8
8
8
1
2
2
20
80
4
8
1
80
4
4
10
80
6
Sample Collection Form #
Sample Collection Replacement Form #
Sample Type
Site Number
Street Address
Media (a constant of 'S')
Site X Coordinate
Site Y Coordinate
Site X Coordinate
Site Y Coordinate
Difference in feet between X/YCOORD & X/YACTUAL
Area Id
Neighborhood Id
Sampling Team Number
Sample Collector Name (text)
Soil Composition
Collection Tube Number
Length of Soil Column
Duplicate Flag currently
Field Comments (text 3X80)
Field Chain of Custody Form # (of the form Fnnn)
Traffic Report for Tubes Form # (of the form nnnn)
Fed Ex Airbill Number
Field Shipping Special Instructions
Field QC Samples Only Data
Previous Field Blank Id (HS #) currently
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
IntDBBld
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
not used
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
not used
********************************************************************************
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
NUM
NUM
CHAR
CHAR
3
30
12
20
5
20
1
80
19
1
5
5
1
8
8
1
7
Prep Lab Id (a constant of 'CAA')
Condition on Receipt by Prep Lab (text)
Sampling Team Comment
Person Receiving in Prep Lab
Sample Preparation Form # (of the form Cnnnn)
Mixer Name (text)
Mixing Team
Prep Comments (text 2X80)
Tracking Comment
Designated Split
Prep Lab Chain of Custody Form # (of the form 'Ennn ')
Traffic Report for Prepared Samples Form #
(of the form 'Dnnn ')
Matrix Spike Flag currently
Preparation Lab Jar Number
Head Space
Possible High Analysis Values
Preparation Lab Shipping Method (a constant of 'FED EX')
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
SampTrac
not used
SampTrac
SampTrac
SampTrac
SampTrac
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
FS
FS
FS
FS
FS
FS
.G18
.G19
.G20
.G21
.G22
.G23
*******
FS
FS
FS
FS
FS
FS
.H01
.H02
.H03
.H04
.H05
.H06
FS.H07
FS.I01
FS
FS
FS
FS
.102
.103
.104
.105
*******
FS
FS
FS
FS
.J01
.J02
.J03
.J04
*******
FS
FS
FS
FS
FS
FS
FS
FS
FS
FS
FS
FS
FS
FS
FS
.K01
.K02
.K03
.K04
.K05
.K06
.K07
.K08
.K09
.K10
.K11
.K12
.K13
.K14
.K15
NAME
PLFEDEX
PLSRMKS
PLPRVBLK
EXCPFLAG
FTTEMP
TRTEMP
Analysis Lab
FSSMPL
FSALID
FSALCOND
ALRECBY
FSRERUN
FSRRSTAT
FSORIG
Analysis Lab
FSTWTEXT
FSWTEXT
FSMOIST
FSCONCDF
FSEXTANL
Analysis Lab
FSINSTID
FSANLST
FSFILE
FSINJVOL
Analysis Lab
FSQFILE
FSQMTH
FSQION
FSLCPH
FSISPH
FSPYRPH
FSSSPH
FSHLCS
FSHLCR
FSHISS
FSHISR
FSHPYRS
FSHPYRR
FSHSSS
FSHSSR
ARRAY TYPE
CHAR
CHAR
CHAR
NUM
NUM
NUM
LENGTH
10
80
6
8
8
8
DESCRIPTION
Fed Ex Airbill Number
Prep Lab Special Shipping Instructions
Prep Lab QC Samples Only
Previous Prep Lab Blank Id (HS #) currently
Exception Flag
Temperature inside cooler on arrival at preparation
lab, in degrees Celsuis
Temperature inside cooler on arrival at analytic lab,
in degrees Celsuis
SOURCE
SampTrac
SampTrac
not used
SampTrac
SampTrac
SampTrac
Sample Login Data **********************************************************************
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
10
3
30
20
2
1
CHAR 1
Sample Extraction Data
NUM 8
NUM
NUM
NUM
CHAR
Injection Data
CHAR
CHAR
CHAR
NUM
Interpretation
CHAR
(8,3) CHAR
(8) CHAR
(8,3) NUM
(5) NUM
NUM
(3) NUM
(8,3) NUM
(8,3) NUM
(5) NUM
(5) NUM
NUM
NUM
(3) NUM
(3) NUM
8
8
8
8
Project Sample Id
Analytic Laboratory Id
Condition on Receipt by Analysis Lab (text)
Person Receiving in Analysis Lab
Re-run Flag currently
Re- run Type currently
LabData
SampTrac
SampTrac
SampTrac
not used
not used
Orig. Found Flag currently not used
Target Weight to Extract (gm) LabData
Weight Extracted (gm)
Percent Moisture
Concentration Dilution Factor
Extraction Analyst
LabData
LabData
LabData
LabData
*************************************************************************
15
8
15
8
GC/MS Instrument Id
GC/MS Analyst
GC/MS Datafile Id
Injection Volume (ul)
LabData
LabData
LabData
LabData
Data ********************************************************************
12
1
1
8
8
8
8
8
8
8
8
8
8
8
8
GC/MS Shift Results File Name
Quant i tat ion Method Flag, by 8 LCI Cs, by 3 Ions
LCIC Quantisation Ion Selection, by 8 LCICs
Peak Height, by 8 LCICs, by 3 Ions
Peak Height, by 5 Int. Stds. (Primary Ion)
Peak Height Pyrene-D10 (Secondary Ion)
Peak Height, by 3 Surrogates (Primary Ion)
Scan for Peak Height, by 8 LCICs, by 3 Ions
Retention Time for Peak Height, by 8 LCICs, by 3 Ions
Scan for Peak Height, by 5 Int. Stds.
Retention Time for Peak Height, by 5 Int. Stds.
Scan for Peak Height for Pyrene-D10
Retention Time for Peak Height for Pyrene-D10
Scan for Peak Height, by 3 Surrogates
Retention Time for Peak Height, by 3 Surrogates
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
it'triritit'ttlt
FS.L01
FS.L02
FS.L03
FS.L04
FS.L05
FS.L06
FS.L07
FS.L08
FS.L09
FS.L10
FS.L11
FS.L12
FS.L13
FS.LK
FS.L15
*******
FS.M01
FS.M02
FS.M03
FS.M04
FS.M05
FS.M06
FS.M07
FS.M08
FS.M09
FS.M10
FS.M11
FS.M12
*******
FS.N01
FS.N02
FS.N03
FS.N04
FS.NOS
FS.N06
FS.N07
FS.NOS
FS.N09
FS.N10
FS.N11
FS.N12
FS.N13
FS.NH
NAME
ARRAY
FSCDATE
FSCTIME
FSCANAL
FSCLCA
FSCLCS
FSCLCR
FSCISA
FSCISS
FSCISR
FSCPYRA
FSCPYRS
FSCPYRR
FSCSSA
FSCSSS
FSCSSR
Analysis Raw
FSLCA
FSLCS
FSLCR
FSISA
FSPYRA
FSISS
FSPYRS
FSISR
FSPYRR
FSSSA
FSSSS
FSSSR
(8,3)
(8,3)
(8.3)
(5)
(5)
(5)
(3)
(3)
(3)
TYPE
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
Data (After
(8,3)
(8,3)
(8,3)
(5)
(5)
(5)
(3)
(3)
(3)
LabOata Computations
FSRR
FSRRC
FSRRF
FSCONCB
FSCONCA
FSIRAT
FSIONF
FSMINS
FSMAXS
FSRANG
FSRANGF
FSIDEVT
FSIDEVF
FSPRESCR
(8)
(8)
(8)
(8)
(8)
(8)
(8)
(8)
(8)
(8)
(8)
(8)
(8)
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
LENGTH
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
DESCRIPTION
Analysis Lab Analysis Date
Analysis Lab Analysis Time
Analyst Entering the Correction Data
Area, by 8 LCICs, by 3 Ions
Scan, by 8 LCICs, by 3 Ions
Retention Time, by 8 LCICs, by 3 Ions
Area, by 5 Int. Stds. (Primary Ion)
Scan, by 5 Int. Stds. (Primary Ion)
Retention Time, by 5 Int. Stds. (Primary Ion)
Area Pyrene-D10 (Secondary Ion)
Scan Pyrene-D10 (Secondary Ion)
Retention Time Pyrene-D10 (Secondary Ion)
Area, by 3 Surrogates (Primary Ion)
Scan, by 3 Surrogates (Primary Ion)
Retention Time, by 3 Surrogates (Primary Ion)
SOURCE
LabData
LabOata
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
Corrections if any) ******************************************************
8
8
8
8
a
8
8
8
8
8
8
8
for Field and
NUM
NUM
CHAR
NUM
NUM
NUM
CHAR
NUM
NUM
NUM
CHAR
NUM
CHAR
CHAR
8
8
1
8
8
8
1
8
8
8
1
8
1
3
Area, by 8 LCICs, by 3 Ions
Scan, by 8 LCICs, by 3 Ions
Retention Time, by 8 LCICs, by 3 Ions
Area, by 5 Int. Stds. (Primary Ion)
Area Pyrene-D10 (Secondary Ion)
Scan, by 5 Int. Stds. (Primary Ion)
Scan Pyrene-D10 (Secondary Ion)
Retention Time, by 5 Int. Stds. (Primary Ion)
Retention Time Pyrene-D10 (Secondary Ion)
Area, by 3 Surrogates (Primary Ion)
Scan, by 3 Surrogates (Primary Ion)
Retention Time, by 3 Surrogates (Primary Ion)
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabOata
LabData
LabData
QC Samples *******************************************************
Relative Retention Time, by 8 LCICs (Quanti tation Ion)
Relative Retention Time Criteria, by 8 LCICs
Flag for Rel. Ret. Time Out of Criteria, by 8 LCICs
PrelD-criteria Concentration, by 8 LCICs
(i.e. primary ion equivalent concentration)
Id Criteria applied Concentration, by 8 LCICs
Ion Ratio, by 8 LCICs
Flag for All Ions Being Present, by 8 LCICs
Minimum Scan Number of 3 Ions, by 8 LCICs
Maximum Scan Number of 3 Ions, by 8 LCICs
Scan Range (Max - Min), by 8 LCICs
Flag for Scan range >1, by 8 LCICs
Ion Ratio % Dev. from Theoretical Values, by 8 LCICs
Flag for Ion ratio % Dev. Out of Criteria, by 8 LCICs
Flag for Analysis Pre-screen
*LabData
LabData
*LabOata
*LabData
*LabOata
*LabOata
•LabData
LabData
LabData
•LabOata
•LabData
•LabData
•LabData
LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
FS.N15
FS.001
FS.002
FS.003
FS.004
FS.005
FS.006
FS.P01
FS.P02
FS.P03
FS.P04
FS.P05
FS.P06
FS.P07
FS.P08
FS.Q01
FS.Q02
FS.Q03
FS.Q04
FS.Q05
FS.Q06
FS.Q07
FS.Q08
FS.Q09
FS.Q10
FS.Q11
FS.Q12
FS.Q13
FS.Q14
FS.Q15
FS.Q16
FS.Q17
FS.Q18
FS.Q19
FS.Q20
FS.Q21
FS.Q22
FS.Q23
FS.Q24
NAME
FSSULFUR
ARRAY
LabOata Computations
FSPR (3)
FSPRF
FSSSADD
FSPRCL
FSPRCU
FSPROUT
(3)
(3)
(3)
(3)
LabData Computations
FSISADD
FSRD
FSAD
FSRF
FSAF
FSRDC
FSADCL
FSADCU
(5)
(5)
(5)
(5)
(5)
(5)
(5)
Data Validation nags
FSSURRA
FSSURRB
FSSURRC
FSSURCM
FSINTA
FSINTB
FSINTC
FS1NTCM
FSIDA
FSIDB
FSIDC
FSIDD
FSIDE
FSIDCOM
FSQTA
FSQTB
FSQTC
FSQTCOM
FSGENA
FSGENB
FSGENC
FSGEND
FSGENCM
FSDQ
(3)
(3)
(3)
(3)
(3)
(8,5)
TYPE
LENGTH
CHAR 1
for Surrogate
NUM 8
CHAR
NUM
NUM
NUM
1
8
8
8
NUM 8
DESCRIPTION
SOURCE
Flag for Sulphur Cleanup Performed LabData
% Recovery by 3 Surrogates *LabData
Flag for % Rec. Out of Criteria, by 3 Surrogates
Amount of added(ng), by 3 Surrogates
% Recovery Criteria lower limit, by 3 Surrogates
% Recovery Criteria upper limit, by 3 Surrogates
Number of % Rec. Out of Criteria
NUM 8 Internal Standard Quantity Added (in nanograms)
NUM
NUM
CHAR
CHAR
NUM
NUM
NUM
and
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
NUM
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
8
8
1
1
8
8
8
Comments
1
1
1
60
1
1
2
60
1
1
1
2
2
60
1
1
2
60
2
2
2
2
60
2
Ret. Time Difference From CC Val, by 5 Int Stds
Area Difference % From CC Val, by 5 Int Stds
Flag for Ret. Time Out of Criteria, by 5 Int Stds
Flag for Area Out of Criteria, by 5 Int Stds
Retention Time Difference Criteria, by 5 Int Stds
Area % Diff. Crit. lower limit, by 5 Int Stds
Area % Diff. Crit. upper limit, by 5 Int Stds
Check surrogates spiked in all samples
Surrogates recoveries within criteria
Surrogates Miscellaneous
Surrogates Comments
Internal Standards RT Criteria
Internal Standards Area Criteria
Internal Standards Miscellaneous
Internal Standards Comments
Id All ions maximize simultaneously
Id Appropriate flags used.
Id All peaks reported that meet id criteria
Id Low level peaks examined
Id Miscellaneous
Id Comments
Quantisation appropriate RRF's used if nonstandard
Check integration parameters if manual quant, used
Quantisation Miscellaneous
Quantisation Comments
Review case narrative and address all problems
Examine SICPS
Examine quant i tat ion reports
General Miscellaneous
General Comments
Data Qualifier flags samples by analyte
•LabData
LabOata
LabData
LabData
LabData
LabData
•LabData
•LabData
*LabData
*LabData
LabData
LabData
LabData
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
******* integrated DB Audit System Flags ********************************************************************
FS.R01 FSTRACK CHAR 1 Flag for Sample NOT Found in Sample Tracking IntDBBld
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
FS.R02
FS.R03
FS.R04
FS.R05
FS.R06
FS.R07
FS.R08
FS.R09
FS.R10
FS.R11
FS.R12
FS.R13
FS.R14
FS.R15
FS.R16
NAME ARRAY TYPE LENGTH DESCRIPTION SOURCE
FSF1FLAG
FSICREF
FSCCREF
FSMSREF
FSMDREF
FSRBREF
FSMBREF
FSDUP
FSFHB
FSSSBF
FSSSBP
FSPHB
FSPLS
FSRECALC
FSDIFLAB
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
1 Flag for Sample Match to FORM1 Master File IntDBBld
1 Flag for 1C not found or date/time inconsistent IntDBBld
1 Flag for CC not found or date/time inconsistent IntDBBld
1 Flag for MS not found IntDBBld
1 Flag for MSD not found IntDBBld
1 Flag for RB not found IntDBBld
1 Flag for MB not found IntDBBld
1 Flag for Field Duplicate not found currently not used
1 Flag for Field Handling Blank not found currently not used
CHAR 1 Flag for Shipping & Storage Blank not found currently not used
CHAR 1 Flag for Shipping & Storage Blank not found currently not used
CHAR 1 Flag for Prep Lab Handling Blank not found currently not used
CHAR 1 Flag for Prep Lab Split not found currently not used
CHAR 1 Flag for recalculation descrepancy on at least IntDBBld
one data element computed by LabData system
(within .2% tolerance)
CHAR 1 Flag for diff. lab between LabData & Samptrac currently not used
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
****** |
FA.A01
*******
FA.B01
FA.B02
FA.B03
FA. 804
FA. BOS
FA.B06
FA.B07
*******
FA.M01
FA.M02
FA.M03
FA.M04
FA. COS
FA.C06
FA.C07
FA. COS
FA.C09
FA.C10
FA.C11
FA.C12
NAME ARRAY TYPE
(eys **********
ALAN AL ID
Data Transfer
FSGENDTE
FSGENTME
FSADDDTE
FSADDTME
FSUPDDTE
FSUPDTME
FSUPDCNT
Data Replaced
FSLCA
FSLCS
FSLCR
FSISA
FSPYRA
FSISS
FSPYRS
FSISR
FSPYRR
FSSSA
FSSSS
FSSSR
»***
***********
CHAR
LENGTH
*******
23
DESCRIPTION
******************************************************
Analysis Lab Sample Id (Lab Id + Lab Analysis Id)
SOURCE
*************
LabData
Tracking******************************************************************************
by
<8,
(8,
(8,
(5)
(5)
(5)
(3)
(3)
(3)
NUM
NUM
NUM
NUM
NUM
NUM
NUM
Corrections
3) NUM
3) NUM
3) NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
8
8
8
8
8
8
8
LabData Gen Date
LabData Gen Time
DB Add Date
DB Add Time
DB Most Recent Update Date
DB Most Recent Update Time
DB Update Count
LabData
LabData
IntDBBld
IntDBBld
IntDBBld
IntDBBld
IntDBBld
************************************************************************
8
8
8
8
8
8
8
8
8
8
8
8
Area, by 8 LCICs, by 3 Ions
Scan, by 8 LCICs, by 3 Ions
Retention Time, by 8 LCICs, by 3 Ions
Area, by 5 Int. Stds. (Primary Ion)
Area Pyrene-D10 (Secondary Ion)
Scan, by 5 Int. Stds. (Primary Ion)
Scan Pyrene-DIO (Secondary Ion)
Retention Time, by 5 Int. Stds. (Primary Ion)
Retention Time Pyrene-D10 (Secondary Ion)
Area, by 3 Surrogates (Primary Ion)
Scan, by 3 Surrogates (Primary Ion)
Retention Time, by 3 Surrogates (Primary Ion)
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID NAME ARRAY TYPE LENGTH DESCRIPTION SOURCE
***** Keys and Status ***************************************************************************************
IC.A01 1C ANAL ID CHAR 23 Initial Calibration Id LabOata
(Lab Id + Lab Analysis Id of the IC1 analysis)
IC.A02 ICANAL (4) CHAR 23 Initial Calibration Id LabData
(Lab Id + Lab Analysis Id of the IC2-IC5 analyses)
IC.A03 ICSTATUS CHAR 1 Status Flag IntDBBld
******* £h rono I ogy ******************************************************************************************
IC.B01 ICAADTE (6) NUM 8 Analysis Lab Analysis Date, LabOata
by 5 Calibration Concentrations and EMSL PC
IC.B02 ICAATME (6) NUM 8 Analysis Lab Analysis Time LabData
by 5 Calibration Concentrations and EMSL PC
******* Qgta Transfer Tracking ******************************************************************************
IC.C01 ICGENDTE NUM 8 LabData Gen Date LabData
IC.C02 ICGENTME NUM 8 LabData Gen Time LabData
IC.C03 ICADDDTE NUM 8 DB Add Date IntDBBld
IC.C04 ICADDTME NUM 8 DB Add Time IntDBBld
1C. COS ICUPDDTE NUM 8 DB Most Recent Update Date IntDBBld
IC.C06 ICUPDTME NUM 8 DB Most Recent Update Time IntDBBld
IC.C07 ICUPDCNT NUM 8 DB Update Count IntDBBld
******* Analysis Lab Sample Login Data **********************************************************************
IC.D01 ICSMPL (6) CHAR 10 Project Sample Id for IC1 thru IC5 and EMSL PC LabData
1C. 002 ICALID CHAR 3 Analytic Laboratory Id LabData
******* Analysis Lab Injection Data *************************************************************************
IC.E01 ICINSTID CHAR 15 GC/MS Instrument Id LabData
IC.E02 ICINSTMF CHAR 20 GC/MS Instrument Manufacturer LabData
IC.E03 ICINSTMN CHAR 20 GC/MS Instrument Model Number LabData
IC.E04 ICISCODE CHAR 1 Internal Standard Solution Code LabData
1C. EOS ICSOLCD (5) CHAR 1 Initial Calibration Solution Codes (5) LabOata
IC.E06 ICCCCODE CHAR 1 Continuing Calibration Solution Code LabData
IC.E07 ICCOLMMF CHAR 20 Column Manufacturer LabData
1C. EOS ICCOLMSN CHAR 20 Column Serial Number LabData
IC.E09 ICANLST (6) CHAR 8 Analyst, LabData
by 5 Calibration Concentrations and EMSL PC
IC.E10 ICFILE (6) CHAR 15 Datafile Id, LabData
by 5 Calibration Concentrations and EMSL PC
IC.E11 ICINJVL (6) NUM 8 Injection Volume (ul), LabData
by 5 Calibration Concentrations and EMSL PC
******* Analysis Lab Interpretation Data ********************************************************************
IC.F01 ICQFILE CHAR 12 GC/MS Shift Results File Name LabData
For 1C2 ONLY:
IC.F02 ICQMTH (8,3) CHAR 1 Quantitation Method Flag, by 8 LCICs, LabData
by 3 Ions
IC.F03 ICQION (8) CHAR 1 LCIC Quantitation Ion Selection, by 8 LCICs LabOata
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
IC.F04
IC.F05
IC.F06
IC.F07
IC.F08
IC.F09
IC.F10
IC.F11
IC.F12
IC.F13
IC.FH
IC.F15
*******
IC.G01
IC.G02
IC.G03
IC.G04
IC.G05
IC.G06
IC.G07
IC.G08
IC.G09
IC.G10
IC.G11
IC.G12
IC.G13
IC.GH
IC.G15
*******
IC.H01
IC.H02
IC.H03
IC.H04
IC.H05
IC.H06
IC.H07
IC.H08
IC.H09
IC.H10
NAME
ICLCPH
ICISPH
ICPYRPH
ICSSPH
ICHLCS
ICHLCR
ICHISS
ICHISR
ICHPYRS
ICHPYRR
ICHSSS
ICHSSR
Correction
ICCDATE
ICCTIME
ICCANAL
ICCLCA
ICCLCS
ICCLCR
ICCISA
ICCISS
ICCISR
ICCPYRA
ICCPYRS
ICCPYRR
ICCSSA
ICCSSS
ICCSSR
ARRAY
(8,3)
(5)
(3)
(8,3)
(8,3)
(5)
(5)
(3)
(3)
Flags (For
(8,3)
(8,3)
(8,3)
(5)
(5)
(5)
(3)
(3)
(3)
Analysis Raw Data, By
(After
ICLCA
ICLCS
ICLCR
ICISA
ICPYRA
ICISS
ICPYRS
ICISR
ICPYRR
ICSSA
Corrections
(6,8,3)
(6,8,3)
(6,8,3)
(6,5)
(6)
(6,5)
(6)
(6,5)
(6)
(6,3)
TYPE
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
IC2
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
LENGTH
8
8
8
8
8
8
8
8
8
8
8
8
DESCRIPTION
Peak Height, by 8 LCICs, by 3 Ions
Peak Height, by 5 Int. Stds. (Primary Ion)
Peak Height Pyrene-D10 (Secondary Ion)
Peak Height, by 3 Surrogates (Primary Ion)
Scan for Peak Height, by 8 LCICs, by 3 Ions
Retention Time for Peak Height, by 8 LCICs,
by 3 Ions
Scan for Peak Height, by 5 Int. Stds.
Retention Time for Peak Height, by 5 Int.
Stds.
Scan for Peak Height for Pyrene-DIO
Retention Time for Peak Height for
Pyrene-D10
Scan for Peak Height, by 3 Surrogates
Retention Time for Peak Height, by 3
Surrogates
SOURCE
LabData
LabOata
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
ON L Y ) *********************************************************************
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
5 Calibration
. if
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
any)
8
8
8
8
8
8
8
8
8
8
Analysis Lab Analysis Date
Analysis Lab Analysis Time
Analyst
Area, by 8 LCICs, by 3 Ions
Scan, by 8 LCICs, by 3 Ions
Retention Time, by 8 LCICs, by 3 Ions
Area, by 5 Int. Stds. (Primary Ion)
Scan, by 5 Int. Stds. (Primary Ion)
Retention Time,
by 5 Int. Stds. (Primary Ion)
Area Pyrene-DIO (Secondary Ion)
Scan Pyrene-DIO (Secondary Ion)
Retention Time Pyrene-010 (Secondary Ion)
Area, by 3 Surrogates (Primary Ion)
Scan, by 3 Surrogates (Primary Ion)
Retention Time,
by 3 Surrogates (Primary Ion)
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
Concentrations and EMSL Performance check************************
Area, by 8 LCICs, by 3 Ions
Scan, by 8 LCICs, by 3 Ions
Retention Time, by 8 LCICs, by 3 Ions
Area, by 5 Int. Stds. (Primary Ion)
Area Pyrene-010 (Secondary Ion)
Scan, by 5 Int. Stds. (Primary Ion)
Scan Pyrene-D10 (Secondary Ion)
Retention Time, by 5 Int. Stds. (Primary Ion)
Retention Time Pyrene-D10 (Secondary Ion)
Area, by 3 Surrogates (Primary Ion)
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
IC.H11
IC.H12
*******
1C. 101
1C. 102
1C. 103
1C. 104
1C. 105
1C. 106
1C. 107
*******
IC.J01
IC.J02
IC.J03
IC.J04
1C. JOS
IC.J06
1C. JOT
1C. JOS
IC.J09
*******
IC.K01
IC.K02
IC.K03
IC.K04
IC.K05
NAME
ICSSS
ICSSR
ARRAY
(6,3)
(6,3)
LabOata Computations
ICMRF
ICRSD
ICRSDC
ICRSDF
ICRSDOUT
ICRF
ICCONC
(11)
(11)
(11)
(11)
(11,5)
(5)
LabData Computations
PCTCON
PCPRLC
PCPRUC
PCPROUT
PCCONC
PCPR
PCPRF
PCISADD
PCVOLHCS
(11)
(11)
(11)
(11)
(11)
(11)
Data Validation Flags
ICCALA
ICCALB
ICCALC
ICCALD
I CDVCOM
(3)
TYPE
NUM
NUM
LENGTH
8
8
DESCRIPTION
Scan, by 3 Surrogates (Primary Ion)
Retention Time, by 3 Surrogates (Primary Ion)
SOURCE
LabData
LabData
for Initial Ca I i brat i on*********************************************************"
NUM
NUM
NUM
CHAR
NUM
NUM
NUM
8
8
8
1
8
8
8
Mean Response Factors, by 8 LCICs and 3 Surrogates
% Relative Std. Dev., by 8 LCICs and 3 Surrogates
% Relative Std. Dev. criteria, by 8 LCICs
and 3 Surrogates
Flag for % Relative Std. Dev. out of criteria,
by 8 LCICs and 3 Surrogates
Number of % Relative Std. Dev. out of criteria
Response Factors, by 8 LCICs and 3 surrogates,
by 5 Calibration Concentrations
Concentrations, by 5 Calibration Concentrations
*LabOata
•LabData
LabData
*LabData
LabData
•LabData
LabData
for EMSL Performance Check *****************************************************
NUM
NUM
NUM
NUM
NUM
NUM
CHAR
NUM
NUM
and
CHAR
CHAR
CHAR
CHAR
CHAR
8
8
8
8
8
8
1
8
8
Comments
1
2
2
2
60
Theoretical Concentrations, by 8 LCICs
and 3 Surrogates
% Recovery Lower Criteria, by 8 LCICs
and 3 Surrogates
% Recovery Upper Criteria, by 8 LCICs
and 3 Surrogates
Number of % Recovery Out of Criteria
Concentrations, by 8 LCICs and 3 Surrogates
% Recovery, by 8 LCICs and 3 Surrogates
Flag for % Recovery Within Criteria,
by 8 LCICs and 3 Surrogates
Amount of Internal Standard Added
Volume of Check Standard Solution
LabData
LabData
LabData
LabData
•LabData
•LabOata
•LabData
LabData
LabData
******************************************************************
1C % RSD exceptions checked
1C check quantitation reports for evidence of editing
1C examine chromatograms
1C Miscellaneous
1C Comments
DataVal
DataVal
DataVal
DataVal
DataVal
******* integrated DB Audit System Results ******************************************************************
IC.L01
ICRECALC
CHAR
1 Flag for calculation descrepancies
IntDBBLd
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
NAME ARRAY TYPE LENGTH
DESCRIPTION
SOURCE
***** Keys **************************************************************************************************
IA.A01
*******
IA.B01
IA.B02
IA.B03
IA.B04
IA.B05
IA.B06
IA.807
*******
1A.C01
IA.C02
IA.C03
IA.C04
IA.C05
IA.C06
IA.C07
IA.C08
IA.C09
IA.C10
IA.C11
IA.C12
I CANAL ID
Data Transfer
ICGENDTE
ICGENTME
ICADDDTE
ICADDTME
ICUPDDTE
ICUPDTME
ICUPDCNT
Data Replaced
ICLCA
ICLCS
ICLCR
ICISA
ICPYRA
ICISS
ICPYRS
ICISR
ICPYRR
ICSSA
ICSSS
ICSSR
CHAR
23
Initial Calibration Id
(Lab Id + Lab Analysis Id of the IC1 analysis)
LabOata
T rack i ng ******************************************************************************
NUM
NUM
NUM
NUM
NUM
NUM
NUM
by Corrections
(6,8,3) NUM
(6,8,3) NUM
(6,8,3) NUM
(6,5) NUM
(6) NUM
(6,5) NUM
(6) NUM
(6,5) NUM
(6) NUM
(6,3) NUM
(6,3) NUM
(6,3) NUM
8
8
8
8
8
8
8
LabData Gen Date
LabData Gen Time
DB Add Date
DB Add Time
DB Most Recent Update Date
DB Most Recent Update Time
DB Update Count
LabData
LabOata
IntDBBld
IntDBBld
IntDBBld
IntDBBld
IntDBBld
************************************************************************
8
8
8
8
8
8
8
8
8
8
8
8
Area, by 8 LCICs, by 3 Ions
Scan, by 8 LCICs, by 3 Ions
Retention Time, by 8 LCICs, by 3 Ions
Area, by 5 Int. Stds. (Primary Ion)
Area Pyrene-D10 (Secondary Ion)
Scan, by 5 Int. Stds. (Primary Ion)
Scan Pyrene-D10 (Secondary Ion)
Retention Time, by 5 Int. Stds. (Primary Ion)
Retention Time Pyrene-D10 (Secondary Ion)
Area, by 3 Surrogates (Primary Ion)
Scan, by 3 Surrogates (Primary Ion)
Retention Time, by 3 Surrogates (Primary Ion)
LabData
LabData
LabData
LabOata
LabOata
LabData
LabData
LabData
LabData
LabData
LabData
LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
*******
CC.
cc.
CC.
A01
A02
A03
*******
cc.
cc.
cc.
cc.
B01
B02
B03
B04
*******
cc.
cc.
cc.
cc.
cc.
C01
C02
C03
C04
COS
CC.C06
cc.
C07
*******
cc.
cc.
D01
D02
*******
cc.
cc.
cc.
cc.
EOT
E02
E03
E04
*******
cc.
cc.
cc.
cc.
cc.
cc.
cc.
cc.
cc.
cc.
cc.
F01
F02
F03
F04
F05
F06
F07
F08
F09
F10
F11
NAME
ARRAY
TYPE
LENGTH DESCRIPTION
SOURCE
Keys and Status *************************************************************************************
CCANALID
CCPC2ID
CCSTATUS
CHAR
CHAR
CHAR
23
23
1
Continuing Calibration/Performance Check 1 Id
(Lab Id + Lab Analysis Id of CC/PC1)
Performance Check 2 Id
(Lab Id + Lab Analysis Id of PC2)
Status Flag
LabOata
LabOata
LabOata
Chronology ******************************************************************************************
CCAADTE
CCAATME
PC2AADTE
PC2AATME
Data Transfer
CCGENDTE
CCGENTME
CCADDDTE
CCADDTME
CCUPDDTE
CCUPDTME
CCUPDCNT
Analysis Lab
CCSMPL
CCALID
Analysis Lab
CCINSTID
CCANLST
CCFILE
CCINJVL
Analysis Lab
CCOFILE
CCQMTH
CCO I ON
CCLCPH
CCISPH
CCPYRPH
CCSSPH
CCHLCS
CCHLCR
CCHISS
CCHISR
NUM
NUM
NUM
NUM
8
8
8
8
CC/PC1 Analysis Lab Analysis Date
CC/PC1 Analysis Lab Analysis Time
PC2 Analysis Lab Analysis Date
PC2 Analysis Lab Analysis Time
LabOata
LabOata
LabOata
LabData
Tracking ******************************************************************************
Sample
(2)
NUM
NUM
NUM
NUM
NUM
NUM
NUM
8
8
8
8
8
8
8
LabOata Gen Date
LabOata Gen Time
DB Add Date
DB Add Time
DB Most Recent Update Date
DB Most Recent Update Time
DB Update Count
LabOata
LabData
IntDBBld
IntDBBld
IntDBBld
IntDBBld
IntDBBld
Log i n Data **********************************************************************
CHAR
CHAR
Injection Data
(2)
(2)
(2)
CHAR
CHAR
CHAR
NUM
Interpretation
(8,3)
(8)
(8,3)
(5)
(3)
(8,3)
(8,3)
(5)
(5)
CHAR
CHAR
CHAR
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
10
3
(For
15
8
15
8
Data
12
1
1
8
8
8
8
8
8
8
8
Sample Id for CC/PC1 and PC2
Analysis Lab Id
LabData
LabData
CC/PC1 and PC2) ****************************************************
GC/MS Instrument Id
GC/MS Analyst, by 2 Analyses
GC/MS Datafile Id, by 2 Analyses
Injection Volume (ul), by 2 Analyses
LabData
LabData
LabData
LabData
********************************************************************
GC/MS Shift Results File Name
For CC/PC1 ONLY:
Quant i tat ion Method Flag, by 8 LCICs,
by 3 Ions
LCIC Quantisation Ion Selection, By 8 LCICs
Peak Height, by 8 LCICs, by 3 Ions
Peak Height, by 5 Int. Stds. (Primary Ion)
Peak Height Pyrene-010 (Secondary Ion)
Peak Height, by 3 Surrogates (Primary Ion)
Scan for Peak Height, by 8 LCICs, by 3 Ions
Retention Time for Peak Height, by 8 LCICs,
by 3 Ions
Scan for Peak Height, by 5 Int. Stds.,
Retention Time for Peak Height, by 5 Int.
Stds.
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
CC.F12
CC.F13
CC.F14
CC.F15
*******
CC.G01
CC.G02
CC.G03
CC.G04
CC.G05
CC.G06
CC.G07
CC.G08
CC.G09
CC.G10
CC.G11
CC.G12
CC.G13
CC.G14
CC.G15
*******
CC.H01
CC.H02
CC.H03
CC.H04
CC.H05
CC.H06
CC.H07
CC.H08
CC.H09
CC.H10
CC.H11
CC.H12
NAME
CCHPYRS
CCHPYRR
CCHSSS
CCHSSR
Correction
CCCDATE
CCCTIME
CCCANAL
CCCLCA
CCCLCS
CCCLCR
CCCISA
CCCISS
CCCISR
CCCPYRA
CCCPYRS
CCCPYRR
CCCSSA
CCCSSS
CCCSSR
ARRAY TYPE
NUM
NUM
(3) NUM
(3) NUM
Flags (For CC/PC1
CHAR
CHAR
CHAR
(8,3) CHAR
(8,3) CHAR
(8,3) CHAR
(5) CHAR
(5) CHAR
(5) CHAR
CHAR
CHAR
CHAR
(3) CHAR
(3) CHAR
(3) CHAR
Analysis Raw Data For CC/PC1
CCLCA
CCLCS
CCLCR
CCISA
CCPYRA
CCISS
CCPYRS
CCISR
CCPYRR
CCSSA
CCSSS
CCSSR
(2,8,3) NUM
(2,8,3) NUM
(2,8,3) NUM
(2,5) NUM
(2) NUM
(2,5) NUM
(2) NUM
(2,5) NUM
(2) NUM
(2,3) NUM
(2,3) NUM
(2,3) NUM
LENGTH
8
8
8
8
ONLY)
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
(After
8
8
8
8
8
8
8
8
8
8
8
8
DESCRIPTION
Scan for Peak Height for Pyrene-D10
Retention Time for Peak Height for
Pyrene-D10
Scan for Peak Height, by 3 Surrogates
Retention Time for Peak Height, by 3
Surrogates
SOURCE
LabData
LabOata
LabData
LabData
******************************************************************
Analysis Lab Analysis Date
Analysis Lab Analysis Time
Analyst
Area, by 8 LCICs, by 3 Ions
Scan, by 8 LCICs, by 3 Ions
Retention Time, by 8 LCICs, by 3 Ions
Area, by 5 Int. Stds. (Primary Ion)
Scan, by 5 Int. Stds. (Primary Ion)
Retention Time,
by 5 Int. Stds. (Primary Ion)
Area Pyrene-DlO (Secondary Ion)
Scan Pyrene-D10 (Secondary Ion)
Retention Time Pyrene-D10 (Secondary Ion)
Area, by 3 Surrogates (Primary Ion)
Scan, by 3 Surrogates (Primary Ion)
Retention Time,
by 3 Surrogates (Primary Ion)
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
Corrections, if any) and PC2 ************************************
Area, by 2 Analyses, by 8 LCICs, by 3 Ions
Scan, by 2 Analyses, by 8 LCICs, by 3 Ions
Retention Time, by 2 Analyses, by 8 LCICs,
by 3 Ions
Area, by 2 Analyses, by 5 Int. Stds.
(Primary Ion)
Area Pyrene-D10 (Secondary Ion),
by 2 Analyses
Scan, by 2 Analyses, by 5 Int. Stds.
(Primary Ion)
Scan Pyrene-D10 (Secondary Ion),
by 2 Analyses
Retention Time, by 2 Analyses,
by 5 Int. Stds. (Primary Ion)
Retention Time Pyrene-D10 (Secondary Ion),
by 2 Analyses
Area, by 2 Analyses, by 3 Surrogates
(Primary Ion)
Scan, by 2 Analyses, by 3 Surrogates
(Primary Ion)
Retention Time, by 2 Analyses,
LabData
LabData
LabData
LabData
LabData
LabData
LabOata
LabData
LabData
LabData
LabData
LabData
-------
DATA
ELEMENT
ID
NAME
ARRAY
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
TYPE LENGTH DESCRIPTION
by 3 Surrogates (Primary Ion)
SOURCE
******* LabData Computations
CC.I01 CCRFA (11,3)
CC.I02 CCRFPH
CC.I03 CCR1R
(11,3)
(11)
for I
NUM
NUM
NUM
for Continuing Calibration *****************************************************
CC.I04
CC.I05
CC.I06
CC.I07
CC.I08
*******
CC.J01
CC.J02
CC.J03
CC.J04
CCPDRF
CCPDC
CCPDF
CCPDOUT
CCRR
(11)
(11)
(11)
(8,3)
NUM
NUM
CHAR
NUM
NUM
8 Response Factors using Area, by 8 LCICs
and 3 Surrogates, by 3 Ions
8 Response Factors using Peak Height,
by 8 LCICs and 3 Surrogates, by 3 Ions
8 Target Ion Ratios for Id Criteria using
Theoretical Values, by 8 LCICs and
3 Surrogates
8 % Dev. From Mean of Response Factors,
by 8 LCICs and 3 Surrogates
8 % Dev. Criteria, by 8 LCICs
and 3 Surrogates
1 Flag for % Dev out of criteria,
by 8 LCICs and 3 Surrogates
8 Number of % Dev Out of Criteria
8 Relative Retention Time,
by 8 LCICs, by 3 Ions
*LabOata
*LabData
LabOata
LabData
LabOata
LabData
LabData
*LabOata
LabOata Performance
PCIR (2,9)
PCIRLC (2,9)
PCIRUC (2,9)
PCIRF (2,9)
Check Data **********************************************************************
CC.J05 PCIROUT
(2)
NUM
NUM
NUM
CHAR
NUM
8
CC.J06
CC.J07
CC.J08
CC.J09
CC.J10
CC.J11
PCBLSEP
PCPV
PCPVF
PCPVC
PCSNR
PCSNRF
(3)
(3)
(3)
(2,2)
(2,2)
CHAR
NUM
CHAR
NUM
NUM
CHAR
1
8
1
8
8
1
CC.J12
PCSNRC
(2,2)
NUM
Ion Ratios, by 2 analyses (PC1, PC2), *LabData
by 8 LCICs and Pyrene-D10
Ion Criteria lower value, by 2 analyses LabData
(PC1, PC2), by 8 LCICs and Pyrene-010
Ion Criteria upper value, by 2 analyses LabOata
(PC1, PC2), by 8 LCICs and Pyrene-D10
Flag for Ion ratio Out of Criteria, *LabData
by 2 analyses (PC1, PC2),
by 8 LCICs and Pyrene-D10
Number of PC Vals Out of Criteria, LabData
by 2 analyses (PC1, PC2)
Baseline Separation LabData
% valley (PC1/CNP, PC2/CNP, PC2/BHC) LabData
Flag for % valley (PC1/CNP, PC2/CNP, PC2/BHC) *LabData
Out of Criteria
% Valley Criteria (PC1/CNP, PC2/CNP, PC2/BHC) LabData
Signal to Noise Ratio(PC1/BHC,PC2/BHC,PC1/TeBB,PC2/TeBB) LabData
Signal to Noise Ratio Out of Crit. Flag *LabData
(PC1/BHC,PC2/BHC,PC1/TeBB,PC2/TeBB)
Signal to Noise Ratio(PC1/BHC,PC2/BHC,PC1/TeBB,PC2/TeBB) LabData
Criteria
******* Data Validation Flags *******************************************************************************
CC.K01 DVDELIV
CC.K02 DVMISS
CC.K03 DVRES
CC.K04 DVPC1A
CHAR 1 Are all deliverables present
(7) CHAR 40 Missing deliverables
(7) CHAR 8 Dates resolved
CHAR 1 PC1 Compare percent valley SICPS
DataVal
DataVal
DataVal
DataVal
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
CC.K05
CC.K06
CC.K07
CC.K08
CC.K09
CC.K10
CC.K11
CC.K12
CC.K13
CC.K14
CC.K15
CC.K16
CC.K17
CC.K18
CC.K19
CC.K20
CC.K21
CC.K22
CC.K23
CC.K24
CC.K25
CC.K26
CC.K27
CC.K28
CC.K29
CC.K30
CC.K31
CC.K32
CC.K33
CC.K34
CC.K35
NAME ARRAY
DVPC1B
DVPC1C
DVPC1D
DVPC1E
DVPC1F
DVPC1DT
DVPC1TM
DVPCMDP
DVHALF
DVPC1CM (3)
DVPC2A
DVPC2B
DVPC2C
DVPC2D
DVPC2E
DVPC2F
DVPC2DT
DVPC2TM
DVPC2CM (3)
DVEPAA
DVEPAB
DVEPAC
DVEPAD
DVEPASMP
DVEPACM (3)
DVCCCALA
DVCCCALB
DVCCCALC
DVCCDT
DVCCTM
DVCCCOM (3)
TYPE
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
LENGTH
1
1
2
2
2
8
8
1
12
60
1
1
1
2
2
2
8
8
60
2
2
2
2
10
60
1
1
2
8
8
60
DESCRIPTION
PC1 Compare signal to noise ratios with raw data
PC1 Check ion ratio criteria
PC1 Check chromatography
PC1 Check quant i tat ion reports for evidence of editing
PC1 Miscellaneous
PC1 Date
PC1 Time
Is this a PC2 midpoint for 16 hour analytical run?
First half shift result file name if exists
PC1 Comments
PC2 Compare percent valley SICPS
PC2 Compare signal to noise ratios with raw data
PC2 Check ion ratio criteria
PC2 Check chromatography
PC2 Check quantitation reports for evidence of editing
PC2 Miscellaneous
PC2 Date
PC2 Time
PC2 Comments
EPA Chk. Std. Recoveries within acceptance windows
EPA Chk. Std. Check chromatography
EPA Chk. Std. Check quantitation reports
EPA Chk. Std. Miscellaneous
EPA Chk. Std. Sample Id
EPA Comments
Continuing Calibration Internal standard areas
Continuing Calibration Check retention time differences
Continuing Calibration Miscellaneous
Continuing Calibration Date
Continuing Calibration Time
Continuing Calibration Comments
SOURCE
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
******* Integrated DB Audit System Flags ********************************************************************
CC.L01
CCRECALC
CHAR
1 Flag for calculation descrepancies
IntDBBld
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID NAME ARRAY TYPE LENGTH DESCRIPTION SOURCE
******* Keys ************************************************************************************************
CHAR 23 Continuing Calibration/Performance Check 1 Id LabOata
(Lab Id + Lab Analysis Id of CC/PC1)
CA.A01 CCANALID
******* Data Transfer Tracking ******************************************************************************
CA.B01
CA.B02
CA.B03
CA.B04
CA.B05
CA.B06
CA.B07
*******
CA.C01
CA.C02
CA.C03
CA.C04
CA.C05
CA.C06
CA.C07
CA.C08
CA.C09
CA.CCO
CA.CC1
CA.CC2
CCGENDTE
CCGENTME
CCADDDTE
CCADDTME
CCUPDDTE
CCUPDTME
CCUPDCNT
Data Replaced
CCLCA
CCLCS
CCLCR
CCISA
CCPYRA
CCISS
CCPYRS
CCISR
CCPYRR
CCSSA
CCSSS
CCSSR
by
<2,
(2,
(2,
(2,
(2)
(2,
(2)
(2,
(2)
(2,
(2,
(2,
NUM
NUM
NUM
NUM
NUM
NUM
NUM
Corrections '
8,3) NUM
8,3) NUM
8,3) NUM
5) NUM
NUM
5) NUM
NUM
5) NUM
NUM
3) NUM
3) NUM
3) NUM
8
8
8
8
8
8
8
k**V
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
LabOata Gen Date
LabOata Gen Time
DB Add Date
DB Add Time
DB Most Recent Update Date
DB Most Recent Update Time
DB Update Count
LabData
LabOata
IntDBBld
IntDBBld
IntDBBld
IntDBBld
IntDBBld
************************************************************************
8
8
8
8
8
8
8
8
8
8
8
8
Area, by 2 Analyses, by 8 LCICs, by 3 Ions
Scan, by 2 Analyses, by 8 LCICs, by 3 Ions
Retention Time, by 2 Analyses, by 8 LCICs,
by 3 Ions
Area, by 2 Analyses, by 5 Int. Stds.
(Primary Ion)
Area Pyrene-DlO (Secondary Ion),
by 2 Analyses
Scan, by 2 Analyses, by 5 Int. Stds.
(Primary Ion)
Scan Pyrene-010 (Secondary Ion),
by 2 Analyses
Retention Time, by 2 Analyses,
by 5 Int. Stds. (Primary Ion)
Retention Time Pyrene-D10 (Secondary Ion),
by 2 Analyses
Area, by 2 Analyses, by 3 Surrogates
(Primary Ion)
Scan, by 2 Analyses, by 3 Surrogates
(Primary Ion)
Retention Time, by 2 Analyses,
by 3 Surrogates (Primary Ion)
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID NAME ARRAY TYPE LENGTH DESCRIPTION SOURCE
******* Keys ari(:j status**************************************************************************************
QC.A01 QCANALID CHAR 23 QC Lab Analysis Id - LabOata
(Lab Id + Lab Analysis Id)
QC.A02
QC.A03
QC.B01
QC.B02
QC.B03
QC.B04
QC.B05
*******
QC.C01
QC.C02
QC.C03
QC.C04
QC.C05
QC.C06
QC.C07
QC.D01
QC.D02
QC.D03
QC.D04
QC.D05
QC.D06
*******
QC.E01
QC.E02
QC.E03
QC.E04
*******
QC.F01
QC.F02
QC.F03
QC.F04
*******
QC.G01
QC.G02
QC.G03
QC.G04
QC.G05
QCSTATUS
CLEANHS
QCALDTE
QCAEDTE
QCAADTE
QCAATME
QCDBDTE
Data Transfer
QCGENDTE
QCGENTME
QCADDDTE
QCADDTME
QCUPDDTE
QCUPDTME
QCUPDCNT
Analysis Lab
QCSMPL
QCALID
QCRERUN
QCRRSTAT
QCORIG
QCTYPE
Analysis Lab
QCTWTEXT
QCWTEXT
QCMOIST
QCCONCDF
Analysis Lab
QCINSTID
QCANLST
QCFILE
QCINJVOL
Analysis Lab
QCQFILE
QCQMTH
QCQION
QCLCPH
QCISPH
CHAR
CHAR
NUM
NUM
NUM
NUM
NUM
1
8
8
8
8
8
8
Status Flag
IntDBBld
Original Project Id (HS Number) IntDBBld
Analysis Lab Login Date LabData
Analysis Lab Extract Date
Analysis Lab Analysis Date
Analysis Lab Analysis Time
Data Validation Date
LabData
LabData
LabData
DataVal
Tracking ******************************************************************************
NUM
NUM
NUM
NUM
NUM
NUM
NUM
8
8
8
8
8
8
8
Sample Login Data ««»«'
CHAR 8
CHAR
CHAR
CHAR
CHAR
CHAR
3
2
1
1
6
Sample Extraction Data
NUM
NUM
NUM
NUM
Injection Data
CHAR
CHAR
CHAR
NUM
Interpretation
CHAR
(8,3) CHAR
(8) CHAR
(8,3) NUM
(5) NUM
8
8
8
8
LabData Gen Date
LabData Gen Time
DB Add Date
DB Add Time
DB Most Recent Update Date
DB Most Recent Update Time
LabData
LabData
IntDBBld
IntDBBld
IntDBBld
IntDBBld
DB Update Count IntDBBld
QC Sample Id from LabData (the Project Sample Id) LabData
Analytic Laboratory Id
Re-run Flag currently
Re- run Type currently
Orig. Found Flag currently
QC Sample Type
LabData
not used
not used
not used
LabData
*****************************************************************
Target Weight to Extract (gm)
Weight Extracted (gm)
Percent Moisture
Concentration Dilution Factor
LabData
LabData
LabData
LabData
*************************************************************************
15
8
15
8
GC/MS Instrument Id
GC/MS Analyst
GC/MS Datafile Id
Injection Volume (ul)
LabData
LabData
LabData
LabData
Data ********************************************************************
12
1
1
8
8
GC/MS Shift Results File Name
Quant i tat ion Method Flag, by 8 LCICs, by 3 Ions
LCIC Quant i tat ion Ion Selection, by 8 LCICs
Peak Height, by 8 LCICs, by 3 Ions
Peak Height, by 5 Int. Stds. (Primary Ion)
LabData
LabData
LabData
LabData
LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
QC.G06
QC.G07
QC.G08
QC.G09
QC.G10
QC.G11
QC.G12
QC.G13
QC.G14
QC.G15
*******
QC.H01
QC.H02
QC.H03
QC.H04
QC.H05
QC.H06
QC.H07
QC.H08
QC.H09
QC.H10
QC.H11
QC.H12
QC.H13
QC.H14
QC.H15
*******
QC.I01
QC.I02
QC.I03
QC.I04
QC.I05
QC.I06
QC.I07
OC.I08
QC.I09
QC.I10
QC.I11
QC.I12
NAME
QCPYRPH
QCSSPH
QCHLCS
QCHLCR
QCHISS
QCHISR
QCHPYRS
QCHPYRR
QCHSSS
QCHSSR
Correction
QCCDATE
QCCTIME
OCCANAL
QCCLCA
QCCLCS
QCCLCR
QCCISA
QCCISS
QCCISR
QCCPYRA
QCCPYRS
QCCPYRR
QCCSSA
QCCSSS
QCCSSR
ARRAY
(3)
(8,3)
(8,3)
(5)
(5)
(3)
(3)
TYPE
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
LENGTH
8
8
8
8
8
8
8
8
8
8
DESCRIPTION
Peak Height Pyrene-D10 (Secondary Ion)
Peak Height, by 3 Surrogates (Primary Ion)
Scan for Peak Height, by 8 LCICs, by 3 Ions
Retention Time for Peak Height, by 8 LCICs, by 3 Ions
Scan for Peak Height, by 5 Int. Stds.
Retention Time for Peak Height, by 5 Int. Stds.
Scan for Peak Height for Pyrene-D10
Retention Time for Peak Height for Pyrene-D10
Scan for Peak Height, by 3 Surrogates
Retention Time for Peak Height, by 3 Surrogates
SOURCE
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
Flags ************************************************************************************
(8,3)
(8,3)
(8,3)
(5)
(5)
(5)
(3)
(3)
(3)
Analysis Raw Data
QCLCA
QCLCS
QCLCR
QCISA
QCPYRA
QCISS
QCPYRS
QCISR
QCPYRR
OCSSA
QCSSS
QCSSR
(8,3)
(8,3)
(8,3)
(5)
(5)
(5)
(3)
(3)
(3)
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
(After
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Analysis Lab Analysis Date
Analysis Lab Analysis Time
Analyst
Area, by 8 LCICs, by 3 Ions
Scan, by 8 LCICs, by 3 Ions
Retention Time, by 8 LCICs, by 3 Ions
Area, by 5 Int. Stds. (Primary Ion)
Scan, by 5 Int. Stds. (Primary Ion)
Retention Time, by 5 Int. Stds. (Primary Ion)
Area Pyrene-010 (Secondary Ion)
Scan Pyrene-010 (Secondary Ion)
Retention Time Pyrene-DIO (Secondary Ion)
Area, by 3 Surrogates (Primary Ion)
Scan, by 3 Surrogates (Primary Ion)
Retention Time, by 3 Surrogates
(Primary Ion)
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
Corrections if any) ******************************************************
8
8
8
8
8
8
8
8
8
8
8
8
Area, by 8 LCICs, by 3 Ions
Scan, by 8 LCICs, by 3 Ions
Retention Time, by 8 LCICs, by 3 Ions
Area, by 5 Int. Stds. (Primary Ion)
Area Pyrene-010 (Secondary Ion)
Scan, by 5 Int. Stds. (Primary Ion)
Scan Pyrene-010 (Secondary Ion)
Retention Time, by 5 Int. Stds. (Primary Ion)
Retention Time Pyrene-D10 (Secondary Ion)
Area, by 3 Surrogates (Primary Ion)
Scan, by 3 Surrogates (Primary Ion)
Retention Time, by 3 Surrogates
(Primary Ion)
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
******* LabData Computations for Field and QC Samples *******************************************************
QC.J01 QCRR (8) NUM 8 Relative Retention Time, by 8 LCICs *LabData
(Quantitat ion Ion)
QC.J02 QCRRC (8) NUM 8 • Relative Retention Time Criteria, by 8 LCICs LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
QC
QC
QC
QC
QC
QC
QC
QC
QC
QC
QC
QC
QC
.J03
.J04
.J05
.J06
.J07
.J08
.J09
.J10
.J11
.J12
.J13
.J14
.J15
*******
QC
QC
QC
QC
QC
.KOI
.K02
.K03
.K04
.K05
QC.K06
QC.L01
QC.L02
QC
QC
QC
QC
QC
QC
QC
QC
QC
QC
.L03
-L04
.L05
.L06
.L07
.L08
.M01
.M02
.M03
.M04
*******
QC
QC
QC
QC
QC
QC
.N01
.N02
.N03
.N04
.N05
.N06
NAME
QCRRF
QCCONCB
QCCONCA
QCIRAT
QCIONF
QCMINS
QCMAXS
QCRANG
QCRANGF
QCIDEVT
QCIDEVF
QCPRESCR
QCSULFUR
ARRAY
(8)
(8)
(8)
(8)
(8)
(8)
(8)
(8)
(8)
(8)
(8)
LabData Computations
QCPR
QCPRF
QCSSADD
QCPRCL
QCPRCU
(3)
(3)
(3)
(3)
(3)
QCPROUT
LabData Computations
QCISADD
QCRD (5)
QCAD
QCRF
QCAF
QCRDC
QCADCL
(5)
(5)
(5)
(5)
(5)
QCADCU (5)
LabData Computations
QCSICON (8)
QCTICON
QCMAXCON
QCCONF
(8)
(8,3)
LabData Matrix Spike
QCSADDED
QCSRECV
QCSPRCU
QCSPRCL
QCSPRL
QCSPR
(8)
(8)
(8)
(8)
(8)
TYPE
CHAR
NUM
NUM
NUM
CHAR
NUM
NUM
NUM
CHAR
NUM
CHAR
CHAR
CHAR
LENGTH
1
8
8
8
1
8
8
8
1
8
1
3
1
for Surrogate
NUM
CHAR
NUM
NUM
NUM
8
1
8
8
8
NUM 8
for Internal
NUM 8
NUM 8
NUM
CHAR
CHAR
NUM
NUM
8
1
1
8
8
DESCRIPTION
Flag for Rel. Ret. Time Out of Criteria, by 8 LCICs
PrelD-criteria Concentration, by 8 LCICs
(i.e., primary ion equivalent concentration)
Id Criteria applied Concentration, by 8 LCICs
Ion Ratio, by 8 LCICs
Flag for All Ions Being Present, by 8 LCICs
Minimum Scan Number of 3 Ions, by 8 LCICs
Maximum Scan Number of 3 Ions,
by 8 LCICs
Scan Range (Max • Min), by 8 LCICs
Flag for Scan range >2, by 8 LCICs
Ion Ratio % Dev. from Theoretical Values, by 8 LCICs
Flag for Ion ratio % Dev. Out of Criteria, by 8 LCICs
Flag for Analysis Pre-screen
Flag for Sulphur Cleanup Performed
SOURCE
. *LabData
•LabData
•LabData
•LabData
•LabData
LabData
LabData
•LabData
•LabData
•LabData
•LabData
LabData
LabData
Standards ********************************************************
% Recovery by 3 Surrogates
Flag for % Rec. Out of Criteria, by 3 Surrogates
Amount of added(ng), by 3 Surrogates
% Recovery Criteria lower limit, by 3 Surrogates
% Recovery Criteria upper limit, by 3 Surrogates
Number of % Rec. Out of Criteria
Internal Standard Quantity Added (in nanograms)
Ret. Time Difference From CC Val, by 5 Int Stds
Area Difference % From CC Val, by 5 Int Stds
Flag for Ret. Time Out of Criteria, by 5 Int Stds
Flag for Area Out of Criteria, by 5 Int Stds
Retention Time Difference Criteria, by 5 Int Stds
Area % Diff. Crit. lower limit, by 5 Int Stds
NUM 8 Area % Diff. Crit. upper limit, by 5 Int Stds
NUM 8 Concentration, by 8 LCICs (secondary ion)
NUM
NUM
CHAR
8
8
1
Concentration, by 8 LCICs (tertiary ion)
Maximum Cone. Criteria
Concentration Out of Criteria Flag,
by 8 LCICs, by 3 Ions
•LabData
•LabData
LabData
LabData
LabData
•LabData
LabData
•LabData
•LabData
•LabData
•LabData
LabData
LabData
LabData
•LabData
•LabData
LabData
•LabData
Data ***************************************************************************
NUM
NUM
NUM
NUM
NUM
NUM
8
8
8
8
8
8
Amount of Spike Added (ng)
Amount of Spike Recovered (ng), by 8 LCICs
% Recovery Upper Limit Criteria, by 8 LCICs
% Recovery Lower Limit Criteria, by 8 LCICs
% Recovery Relative % Dev. Limit, by 8 LCICs
% Recovery, by 8 LCICs
LabData
•LabData
LabData
LabData
LabData
•LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
QC.N07
*******
QC.001
QC.002
QC.003
QC.004
QC.005
QC.006
QC.009
QC.010
QC.011
QC.012
QC.015
QC.016
QC.017
QC.018
QC.021
QC.022
QC.023
QC.024
QC.025
QC.026
QC.029
QC.030
QC.031
QC.032
QC.035
QC.036
QC.037
QC.038
QC.039
QC.042
QC.P01
QC.P02
NAME
QCSPDD
ARRAY TYPE
(8) NUM
LENGTH
8
DESCRIPTION
Relative % Dev. of MS/MSD % Recoveries,
by 8 LCICs
SOURCE
*LabOata
Osts Vs I i Q3 t i on F I 393 find Comments
DVCHKA
DVCHKB
DVCHKC
DVCHKD
DVSAMP
DVCOM
QCSURRA
QCSURRB
QCSURRC
QCSURCM
QCINTA
QCINTB
QCINTC
QCINTCM
QCIDA
QCIDB
QCIDC
QCIDD
QCIDE
QCIDCOM
QCQTA
OCQTB
QCQTC
QCQTCOM
QCGENA
QCGENB
QCGENC
QCGEND
QCGENCM
QCDQ
Integrated
QCF1FLAG
QCRECALC
CHAR
CHAR
CHAR
CHAR
CHAR
(3) CHAR
CHAR
CHAR
CHAR
(3) CHAR
CHAR
CHAR
CHAR
(3) CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
(3) CHAR
CHAR
CHAR
CHAR
(3) CHAR
CHAR
CHAR
CHAR
CHAR
(3) CHAR
(8,5) CHAR
DB Audit System
CHAR
CHAR
2
2
2
2
10
60
1
1
1
60
1
1
1
60
1
1
1
2
2
60
1
1
2
60
2
2
2
2
60
2
Flags **
1
1
Recoveries within acceptance criteria
Check chromatography
Check quantisation reports
Blind QC Samples Miscellaneous
Sample Id
Blind QC comments
Check surrogates spiked into all samples
Recoveries within criteria
Miscellaneous
Surrogates Comments
Internal Standards RT Criteria
Internal Standards Area Criteria
Internal Standards Miscellaneous
Internal Standards Comments
Id All ions maximize simultaneously
Id Appropriate flags used
Id All peaks reported that meet Id criteria
Id Low level peaks examined
Id Miscellaneous
Id Comments
Quantisation Appropriate RRF's used if nonstandard
Quantisation Check integration parameters if manual
Quantisation Miscellaneous
Quantisation Comments
General Review case narrative and address all problems
Examine SICPS
Examine quantisation reports
Miscellaneous
General Comments
Data Qualifiers 8 Analytes by 5 samples
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
DataVal
Flag for Forml Match Results IntDBBld
Flag for calculation descrepancies
IntDBBld
-------
LOVE CANAL HABITAB1LITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
*******
QA.A01
*******
QA.B01
QA.B02
QA.B03
QA.B04
QA.B05
QA.B06
QA.B07
*******
QA.C01
QA.C02
QA.C03
QA.C04
QA.C05
QA.C06
QA.C07
QA.C08
QA.C09
QA.C10
QA.C11
QA.C12
NAME
ARRAY
TYPE
LENGTH
DESCRIPTION
SOURCE
Keys ************************************************************************************************
QCANALID
CHAR
23
QC Lab Analysis Id -
(Lab Id + Lab Analysis Id)
LabData
Data T ransf er T rack i ng ******************************************************************************
QCGENDTE
QCGENTME
QCADDDTE
QCADDTME
QCUPDDTE
QCUPDTME
QCUPDCNT
Analysis Raw
QCLCA
QCLCS
QCLCR
QCISA
QCPYRA
QCISS
QCPYRS
QCISR
QCPYRR
QCSSA
QCSSS
QCSSR
Data
(8,3)
(8,3)
(8,3)
(5)
(5)
(5)
(3)
(3)
(3)
NUM
NUM
NUM
NUM
NUM
NUM
NUM
(After
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
NUM
8
8
8
8
8
8
8
LabData Gen Date
LabData Gen Time
DB Add Date
DB Add Time
DB Most Recent Update Date
DB Most Recent Update Time
DB Update Count '
LabData
LabData
IntDBBld
IntDBBld
IntDBBld
IntDBBld
IntDBBld
Corrections if any) ******************************************************
8
8
8
8
8
8
8
8
8
8
8
8
Area, by 8 LCICs, by 3 Ions
Scan, by 8 LCICs, by 3 Ions
Retention Time, by 8 LCICs, by 3 Ions
Area, by 5 Int. Stds. (Primary Ion)
Area Pyrene-D10 (Secondary Ion)
Scan, by 5 Int. Stds. (Primary Ion)
Scan Pyrene-D10 (Secondary Ion)
Retention Time, by 5 Int. Stds. (Primary Ion)
Retention Time Pyrene-D10 (Secondary Ion)
Area, by 3 Surrogates (Primary Ion)
Scan, by 3 Surrogates (Primary Ion)
Retention Time, by 3 Surrogates
(Primary Ion)
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID NAME ARRAY TYPE LENGTH DESCRIPTION SOURCE
******* Keys and Status *************************************************************************************
F1.A01
F1ANALID
F1.A02 CLEANHS
CHAR 20 Analysis Lab Sample Id
(Lab Id + Lab Analysis Id)
CHAR 10 Original Project ID (HS Number)
LabData
LabOata
*******
F1.801
F1.B02
F1.B03
F1.B04
F1.B05
F1.B06
F1.B07
F1.B08
F1.B09
F1.B10
F1.B11
F1.B12
F1.B13
F1.-B14
F1.B15
Sample Information
F1SMPL
F1TYPE
F1GENDTE
F1GENTME
F1LDVER
F1MOIST
F1PRESCR
F1SULFUR
F1ANLST
F1WTEXT
F1AADTE
F1AATME
F1CONCD
F1QFILE
F1DFILE
*********
CHAR
CHAR
CHAR
CHAR
CHAR
NUM
CHAR
CHAR
CHAR
NUM
CHAR
CHAR
NUM
CHAR
CHAR
****<
10
6
8
8
5
8
3
1
8
8
8
8
8
12
15
**********************************************************************************
Project Sample Field Id (HS #)
Type of sample
LabData Generation Date
LabData Generation Time
LabData Software Version
Percent Moisture
Flag for Analysis Pre-screen
Flag for Sulphur Cleanup Performed
GC/MS Analyst
Weight Extracted
Analysis Date
Analysis Time
Concentration Dilution Factor
GC/MS Shift Results File Name
GC/MS Data File Id
LabData
LabData
LabData
LabOata
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
LabData
******* summary Sample Data Validation Qualifiers ***********************************************************
F1.C01 LCICUSE (8) CHAR 2 Data validation usability flags, by 8 LCICs IntDBBld
******* EMSL-LV Sample Specific Data Validation Qualifiers **************************************************
F1.D01 S_FLAG (4) CHAR 2 Sample specific qualifiers (1 through 4) FormlDE
F1.D05 S FLAGS
*******
F1.E01
F1.E02
F1.E03
F1.E04
F1.E05
F1.E06
F1.E07
F1.E08
F1.E09
F1.E10
F1.E11
F1.E12
Analyte Specific
E_FLAG1A
E_FLAG1B
E_FLAG1C
E_FLAG1D
E_FLAG1E
E_FLAG2A
E_FLAG2B
E_FLAG2C
E_FLAG2D
E_FLAG2E
E_FLAG3A
E FLAG3B
LabData and I
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
.abor
2
2
2
2
2
2
2
2
2
2
2
2
ato
1
1
1
1
1
1
1
1
1
1
1
1
CHAR 2 Sample specific qualifier 5 - EMSL-LV data validation FormlDE
rating
LabData and Laboratory Qualifiers **************************************************
1,2-Dichlorobenzene qualifier A
1,2-Dichlorobenzene qualifier B
1,2-Dichlorobenzene qualifier C
1,2-Dichlorobenzene qualifier D
1,2-Dichlorobenzene qualifier E
1,2,4-Trichlorobenzene qualifier A
1,2,4-Trichlorobenzene qualifier B
1,2,4-Trichlorobenzene qualifier C
1,2,4-Trichlorobenzene qualifier D
1,2,4-Trichtorobenzene qualifier E
1,2,3,4-Tetrachlorobenzene qualifier A
1,2,3,4-Tetrachlorobenzene qualifier B
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
-------
LOVE CANAL HABITAB1LITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID
F1.E13
F1.EU
F1.E15
F1.E16
F1.E17
F1.E18
F1.E19
F1.E20
F1.E21
F1.E22
F1.E23
F1.E24
F1.E25
F1.E26
F1.E27
F1.E28
F1.E29
F1.E30
F1.E31
F1.E32
F1.E33
F1.E34
F1.E35
F1.E36
F1.E37
F1.E38
F1.E39
F1.E40
F1.E41
F1.E42
NAME ARRAY
E_FLAG3C
E_FLAG3D
E_FLAG3E
E_FLAG4A
E_FLAG48
E_FLAG4C
E_FLAG4D
E_FLAG4E
E_FLAG5A
E_FLAG5B
E_FLAG5C
E_FLAG5D
E_FLAG5E
E_FLAG6A
E_FLAG6B
E_FLAG6C
E_FLAG6D
E_FLAG6E
E_FLAG7A
E_FLAG7B
E_FLAG7C
E_FLAG7D
E_FLAG7E
E_FLAG8A
E_FLAG8B
E_FLAG8C
E_FLAG8D
E_FLAG8E
CONC (8)
EXT_DATE
TYPE
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
CHAR
NUM
CHAR
LENGTH
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
8
8
DESCRIPTION
1,2,3,4-Tetrachlorobenzene qualifier C
1,2,3,4-Tetrachlorobenzene qualifier D
1,2,3,4-Tetrachlorobenzene qualifier E
2-Chloronaphthalene qualifier A
2-Chloronaphthalene qualifier B
2-Chloronaphthalene qualifier C
2-Chloronaphthalene qualifier D
2-Chloronaphthalene qualifier E
Alpha-BHC qualifier A
Alpha-BHC qualifier B
Alpha-BHC qualifier C
Alpha-BHC qualifier D
Alpha-BHC qualifier E
Delta-BHC qualifier A
Delta-BHC qualifier B
Delta-BHC qualifier C
Delta-BHC qualifier D
Delta-BHC qualifier E
Beta-BHC qualifier A
Beta-BHC qualifier B
Beta-BHC qualifier C
Beta-BHC qualifier D
Beta-BHC qualifier E
Gamma -BHC qualifier A
Gamma -BHC qualifier B
Gamma -BHC qualifier C
Gamma -BHC qualifier D
Gamma-BHC qualifier E
Concentrations, by LCIC
Order of the LCICs:
1 , 2 - D i ch I orobenzene
1,2,4-Trichlorobenzene
1,2,3,4- Tet rachorobenzene
2-Chloronaphthalene
Alpha-BHC
Delta-BHC
Beta-BHC
Gamma-BHC
Date Sample Extracted
SOURCE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
FormlDE
LabOata
FormlDE
******* Analyte Specific Flags ******************************************************************************
F1.F01
F1.F02
F1.F03
F1.F04
ALLIONS
SCANRNG
IONRATI
RRT
(8)
(8)
(8)
(8)
CHAR
CHAR
CHAR
CHAR
1
1
1
1
All Ions Flags, by 8 LCICs
Scan Range Flags, by 8 LCICs
Ion Ratio Flags
Relative Retention Time Flags
LabOata
LabData
LabData
LabData
-------
LOVE CANAL HABITABILITY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID NAME ARRAY TYPE LENGTH DESCRIPTION SOURCE
******* pOpm 1 Comments *************************************************************************************
F1.G01 COMM (10) CHAR 80 Form 1 Comments • containing explanation of manual LabOata
LabData quantitat ion override
******* integrated DB Audit System Results
F1.H01 F1TRACK CHAR 1 Sample not found in Sample Tracking
-------
LOVE CANAL HABITABIL1TY STUDY
DATA ELEMENT DICTIONARY
DATA
ELEMENT
ID NAME ARRAY TYPE LENGTH DESCRIPTION SOURCE
******* Key *************************************************************************************************
BQ.A01 QCSMPL CHAR 8 Sample Id BlindQC
******* Analyte Spiking Level Data **************************************************************************
BQ.B01 QCBOTTLE CHAR 3 Bottle Number BlindQC
BQ.B02 QCSPIKE (8) NUM 8 Spike Levels, by 8 LCICs in ppb BlindQC
-------
APPENDIX P
Summary of Health Study
on Dioxin Level of Concern
! Soil Assessment--2, 3, 7, 8-TCDD
i
! I
I !
-------
Appendix P
SUMMARY OF HEALTH STUDY ON DIOXIN LEVEL OF CONCERN
SOIL ASSESSMENT—2,3,7,8-TCDD
The 1.0 ppb level of concern for 2,3,7,8-TCDD in residential
surface soil was developed based on extrapolations from
animal toxicity experiments. The extrapolations are
reported in "Health Implications of 2,3,7,8-Tetrachloro-
benzo-p-dioxin (TCDD) Contamination in Residential Soil" by
Kimbrough, et al. The following quote and figures from the
article summarize several of the assumptions on which the
level of concern is based.
To estimate human TCDD intake after exposure to
TCDD-contaminated soil in residential areas, we
calculated estimates for dermal, ingestion, and
inhalation doses. With these estimates (the
assumptions on which they are based are outlined in the
text, the best estimate of a daily dose at 1 ppb in
residential soil (assuming uniform distribution of TCDD
in soil at 1.0 ppb) is calculated to be 44.6 pg/d (or
636.5 fg/kg b.w.d for a person weighing 70 kg). In
consideration of the range of the estimated VSD and
because of the unlikelihood that all of the
conservative exposure assessment assumptions will be
realized on a continuous or lifetime basis, we have
concluded that residential soil levels greater than
1.0 ppb TCDD pose a level of concern. The appropriate
degree of concern for which management decisions are
made should also consider an evaluation of the specific
circumstances at each contaminated site.
Exposure in contaminated residential areas would be
greater than in only occasionally frequented commercial
areas. In residential areas, levels at or above
1.0 ppb TCDD in soil cannot be considered safe and
represent a level of concern. In certain commercial
areas, higher levels may present an acceptable risk to
nonoccupationally exposed individuals. On ranges and
pastures, however, lower soil levels may still be of
concern, since TCDD accumulates in the tissues of
grazing cattle and rooting swine. (Kimbrough, et al.)
WDR361/032
P-l
-------
REFERENCES
Kimbrough, Renate D.; Henry Falk; Paul Stehr; and George
Fries. 1984. "Health Implications of
2,3,7,8-Tetrachlorobenzo-p-dioxin (TCDD) Contamination
in Residential Soil." Journal of Toxicology and
Environmental Health, 14, pp. 47-93.
WDR361/032
P-2
-------
APPENDIX Q
Summary of Assumptions for 1.0 ppb
Level of Concern
Soil Assessment-2, 3, 7, 8-TCDD
WDC.63394.T1 ;SA)
-------
Appendix Q
SUMMARY OF ASSUMPTIONS FOR 1.0 PPB LEVEL OF CONCERN
SOIL ASSESSMENT—2,3,7,8-TCDD
The assumptions regarding exposure to 2,3,7,8-TCDD (dioxin)
in residential soil as presented in the article summarized
in Appendix P are summarized in The National Dioxin Study.
The following is taken from The National Dioxin Study
report:
MAJOR EXPOSURE ASSUMPTIONS FOR THE RECOMMENDED LEVELS
Recommendations Exposure Assumptions
CDC Level of Concern for Residential Soil
General Assumptions
CDC Exposure Pathway Specific Assumptions
Ingestion
o TCDD has a half-life in soil of
12 yrs.
o All soil assumed contaminated at
same level (100% contamination)
o Exposure occurs during 6 months of
the year and is averaged over a
70-year lifetime.
o Human body weight—70 kg
o An increased lifetime risk for
cancer of approximately 2.5x10
would result, assuming an initial
soil concentration of 1 ppb and the
lower bound for the virtually
safe dose.
o Daily soil consumption varied with
age:
0 to 9 months 0 g
9 to 18 months 1 g
1-1/2 to 3-1/2 years 10 g
3-1/2 to 5 years 1 g
>5 years 0.1 g
o Absorption in GI tract 30%
Q-l
-------
MAJOR EXPOSURE ASSUMPTIONS FOR THE RECOMMENDED LEVELS
(continued)
Recommendations Exposure Assumptions
Dermal Daily soil contact with skin
varied with age:
0 to 9 months 0 g
9 to 18 months 1 g
1-1/2 to 3-1/2 years 10 g
3-1/2 to 15 years 1 g
>15 years 0.1 g
Absorption through skin 1%
Inhalation o Concentration of TCDD on airborne
dust is equivalent to soil.
o 15 m of air exchanged/day
o 100% of TCDD inhaled is absorbed.
(The National Dioxin Study, U.S. EPA, 1987)
WDR363/011
Q-2
-------
APPENDIX R
Guide to Love Canal Dioxin Soil Sampling
Study Quality Assurance Project Plan
Soil Assessment-2, 3, 7, 8-TCDD
WDC 63394 T' (SA)
-------
Appendix R
GUIDE TO LOVE CANAL DIOXIN SOIL SAMPLING STUDY
QUALITY ASSURANCE PROJECT PLAN (QAPP)
SOIL ASSESSMENT-2,3,7,8-TCDD
To make the dioxin QAPPs more user-friendly, this guide has
been prepared. The two QAPP documents for which this guide
is provided are the following:
o Love Canal Dioxin Soil Sampling Quality Assurance
Project Plan dated November 3, 1986
o Revisions to Love Canal Dioxin Soil Sampling Study
Quality Assurance Project Plan dated April 6, 1987
The background for the two QAPP documents is discussed
below. Specific QC criteria for the laboratory analyses are
given, and flow charts in Figures R-l and R-2 provide an
overview of the dioxin field and laboratory QA/QC programs,
respectively.
BACKGROUND
The Love Canal Dioxin Soil Sampling Study QAPP, dated
November 3, 1986, was the document prepared for use by field
and laboratory personnel in the collection and analysis of
the soil samples collected in November and December 1986.
Because of the onset of winter, this document was con-
ditionally approved without a Project Plan from the sample
preparation laboratory (Appendix C of the dioxin QAPP), so
that the field work could begin. The sample preparation
laboratory, UNLV, provided the Project Plan on November 26,
1986. Because of the high moisture content of the samples,
the homogenization procedure described in the UNLV Project
Plan was ineffective. (Using an electric blender caused
clumping of the soil.) UNLV made operational revisions to
the Project Plan to describe the revised, "hand-mixing"
technique. (The original UNLV Project Plan (11/26/86) and
the operational revisions (12/11/86) are now included in the
dioxin QAPP as Appendix C: Exhibits A and B, respectively.)
The project's goal was to collect all of the dioxin samples
in 1986; however, because of inclement weather, sampling was
stopped and was started again in May 1987. Sampling could
not begin earlier because the winter thaw left soil with a
moisture content too high for study purposes.
Because of the change to the homogenization procedure and
because of other minor changes relating to the actual
collection of the samples, it was decided that the QAPP
should be revised. The change in the homogenization
R-l
-------
procedure meant that the samples no longer needed to be
mixed in a controlled laboratory environment. Hence, the
1987 samples were mixed in the field. The "hand-mixing"
procedures described in the revised UNLV Project Plan were
included in Appendix A, "Dioxin Soil Sampling Plan."
Appendix C, which contained procedures for use at UNLV, no
longer applied, and was not included in the revised QAPP.
Consequently, only the actual body of the QAPP—Appendix A
and Appendix H, "Dioxin Soil Sampling Work Plan," were
affected by these changes, and only those portions of the
original QAPP were included in the Revisions to Love Canal
Dioxin Soil Sampling Study Quality Assurance Project Plan,
dated April 6, 1987.
The following summarizes the chapters of both of the QAPPs
and their appendices:
o Chapter 1—This chapter discusses the necessity
for and the contents of the QAPP.
o Chapter 2—This chapter introduces the Love Canal
site and provides background information on why
the study was needed.
o Chapter 3—This chapter presents the project
organization and explains which organization is
.responsible for which duties.
o Chapter 4—This chapter defines the quality
assurance objectives which, for the dioxin study,
were measured by accuracy, completeness,
representativeness, and comparability.
o Chapter 5—This chapter references Appendix A, the
"Dioxin Soil Sampling Plan," which outlines the
procedures for collection of the soil samples.
The sample design, which resulted in the
identification of locations where the samples were
to be collected, is referenced in Chapter 5 and
reported in Appendix B. The procedures for sample
homogenization at the University of Nevada at Las
Vegas are referenced and reported in Appendix C
(used during November and December of 1986) .
Appendix F is referenced which includes portions
of the "Users Guide to the Contract Laboratory
Program." This User's Guide describes how the
samples were to be shipped from the field to the
analytical laboratories. Appendix H is referenced
which is the "Dioxin Soil Sampling Work Plan," and
outlines daily procedures to be followed for
sample collections.
R-2
-------
o Chapter 6—This chapter describes the sample
custody procedures and references Appendix A, the
"Dioxin Soil Sampling Plan," for specific details.
o Chapter 7—This chapter references Appendix D, the
"Site Safety Plan," and discusses the equipment
needing calibration. Equipment used for collected
samples did not require calibration.
o Chapter 8—This chapter references Appendix E, the
"Statement of Work—Dioxin Analysis," which
describes the analytical procedures to be used for
the study.
o Chapter 9—This chapter describes data reduction,
validation, and reporting, and discusses in detail
how information relating to collection and
homogenization was to be reported. The chapter
references Appendix E for the methodology used to
report analytical results.
o Chapter 10—This chapter discusses the QC checks
used in the study and explicitly defines the field
QC checks. The chapter references Appendix C for
the QC checks that were used during the analytical
work. Table R-l is a summary of the dioxin
analytical QC requirements used during the study.
o Chapter 11—This chapter describes the parties
responsible for performing audits on the
procedures used in the study.
o Chapter 12—This chapter references Appendix D,
the "Site Safety Plan," which describes the
preventive maintenance procedures for the
analytical instruments. Field equipment for this
study, however, did not require preventive
maintenance.
o Chapter 13—This chapter references Appendix I for
the data assessment procedures that were to be
performed by EPA Region II, and the subsequent
analytical results.
o Chapter 14—This chapter describes how corrective
measures were to be taken, if necessary.
o Chapter 15—This chapter describes the quality
assurance reports.
Table R-l, which follows, provides specific QC requirements
for the laboratory analyses.
WDR363/012
R-3
-------
Table R-l
SUMMARY OF QC REQUIREMENTS
Soil Assessment—2,3,7,8-TCDD
Prepared by EPA EMSL-LV
QC Requirement
Initial Calibration:
1. Column Performance
Criteria
The percent valley between 2,3,7,8-TCDD
and peaks from all other isomers must
be < 25%
Action Required if
Out of Criteria
ANALYSIS MAY NOT PROCEED
UNTIL ALL CRITERIA ARE MET
The ratio of m/z 320 to m/z 322 for
2,3,7,8-TCDD must be > 0.67 and < 0.90
The ratio of m/z 332 to m/z 334 for
C -2,3,7,8-TCDD must be _> 0.67 and
< 0.90
2. Calibration Criteria
The signal to noise ratio (S/N) must be
> 2.5 for m/z 259, 320, and 322 for
unlabeled 2,3,7,8-TCDD and 328 for
Cl -2,3,7,8-TCDD and >10 for m/z 332
and 334 for C -2,3,7,8-TCDD
J. ^
The percent RSD for the RRF of the
triplicate analysis of each standard
concentration must not exceed 10%
The percent RSD of the mean of the four
concentrations must not exceed 10%
Continuing Calibration:
1. Column Performance
Must be performed at the beginning and
end of each 12-hour period
Same as for Initial Calibration
ANALYSIS MAY NOT PROCEED
UNTIL ALL CRITERIA ARE MET
WDR36I/013/1/DRAFT/7-7-88
-------
Table R-l
(continued)
QC Requirement
2. Calibration Criteria
Criteria
S/N same as for Initial Calibration
The RRF of the unlabeled 2,3,7,8-TCDD
relative to 1 C -2,3,7,8-TCDD must be
within ± 10% of the grand mean RRF
from the initial calibration
Action Required if
Out of Criteria
FFB (Matrix Blank)
Method Blank
Duplicate Analysis
Recovery must be between 60 and 140%
Must not contain any signal at m/z 320,
322, or 259 which is greater than 2% of
the m/z 332 response within ± 5 scans
of the m/z 332 peak maximum
The RPD must be no greater than
50 percent
Re-extract and reanalyze the
FFB
Associated positive samples
must be re-extracted and
reanalyzed
Contact SMO or the DPO for
resolution Report All
Values
Blind Quality Control:
1. Uncontaminated
EMPC
EMSL-LV Performance
Evaluation
Must not contain 2,3,7,8-TCDD (False
Positive)
Must fall within the performance based
acceptance windows
Less than 1 ug/Kg
The entire associated batch
of samples must be
re-extracted and reanalyzed
The entire associated batch
of samples must be
re-extracted and reanalyzed
The sample must be
re-extracted and reanalyzed
WDR361/013/2/DRAFT/7-7-88
-------
WDR361/013
Table R-l
(continued)
QC Requirement
Criteria
Samples:
1. Identification
2. Internal Standard
S/N Ratio
3. Internal Standard
m/z 332/334 ratio
4. Native 2,3,7,8-TCDD
Present
5. Internal Standard
Recovery
6. Surrogate S/N Ratio
All criteria except ion ratios are met
Must be greater than 10:1
Must be > 0.67 and < 0.90
Concentration must not exceed
calibration range
Limits not established
Limits not established
Action Required if
Out of Criteria
Sample must be re-extracted
and reanalyzed
Sample must be re-extracted
and reanalyzed
Sample must be re-extracted
and reanalyzed
Re-extract and reanalyze a
smaller aliquot
WDR361/013/3/D RAFT/7-7-88
-------
W63394.T1(SA)/V.5
QUANTITATIVE QA/QC
ACTIVITY
Decontaminate
Sampling Equipment
(App. A: Sec. 3.1,3.4)
SSB/MSB/PE
(Ch. 10 and App. A:
Sec. 3.5)
Pack Coolers &
Ship to Field with
Empty Jars and Blank
Blank Samples
(App. A: Sec. 3.0)
Survey Sample
Locations
(App. A: Sec. 3.2)
FR
(Ch. 10 and App. A:
Sec. 3.5)
May & July
1987
Obtain
Core Samples
(App. A: Sec. 3.2)
Fall 1986
Ship to
Homogeniration Lab
(App. A: Sec. 4.0)
Homogenize
Sample
(App. C or A : Sec. 3.3)
Archive Sample
(App. A: Sec. 2.1)
Split Sample
(App. A: Sec. 2.1)
Ship to
Analysis Lab
(App. A: Sec. 3.7)
Abbreviations
App. Appendix
Ch. Chapter
FR Field Replicate
MSB Matrix Spike Blank
PE Performance Evaluation Sample
Sec. Section
SSB Shipping and Storage Blank
QUALITATIVE QA/QC
Sample Equipment
Cleaning Form
(Ch. 9)
Survey Logs (App. H)
Photographs
Field Notebooks
Sample Collection Forms
Traffic Report Forms
Chain-of-Custody Forms
Custody Seals
(App. A: Sec. 4.0, 4.8
and App. H)
Sample Prep Forms
Chain-of-Custody Forms
Traffic Report Forms
Sample Tracking Numbers
Sample ID Labels
Custody Seals
(App. C, App. H)
Figure R-1
SOIL ASSESSMENT-^, 3, 7, 8-TCDD
SCHEMATIC REPRESENTATION
OF THE FIELD QA/QC PROGRAM
WITH REFERENCE TO QAPP SECTION
-------
W63394.T1(SA)/V.5
QUANTITATIVE QA/QC
ACTIVITY
Field QC
Samples: FR
PE
SSB
MSB
(Ch. 10 and
App. A: Sec. 3.5)
Sample Receipt
(App. E: Exh. A)
Lab QC Sample: MB
(App. E: Exh. E)
QUALITATIVE QA/QC
QAP
SOPs (App. E)
Documentation (App. E: Exh. B)
Store
(App. E: Exh.G)
Lab QA/QC
Measures: Surrogates
MS/MSD
BQC
(App. E: Exh. D)
Extract/Cleanup
(App. E: Exh. D)
Lab QA/QC
Measures: 1C
PC
RC
(App. E: Exh. D)
GC/MS/SIM
Analysis
(App. E: Exh. D)
Internal Review
Data Require Reanalysis?
(App. E: Exh.C)
Chain of Custody (App. E: Exh. B)
Traffic Report Forms (App. E: Exh. B)
Abbreviations
No
Data Package/
Magnetic Tapes
(App. E: Exh. A)
App. Appendix
BQC Blind Quality Control
Exh. Exhibit
FR Field Calibration
GC/MS/SIM Gas Chromatograph/Mass
Spectrometer/Selected Ion
Monitoring
1C Initial Calibration
MB Method Blank
MS Matrix Spike
MSB Matrix Spike Blank
PE Performance Evaluation Sample
PC Performance Check
OAP Quality Assurance Plan
RC Routine Calibration
Sec. Section
SOP Standard Operating Procedure
SSB Shipping and Storage Blank
External Review
(App. I)
V
Data Validation Report
Figure R-2
SOIL ASSESSMENTS, 3, 7, 8-TCDD
SCHEMATIC REPRESENTATION
OF THE LABORATORY QA/QC PROGRAM
WITH REFERENCE TO QAPP SECTION
-------
APPENDIX S
Method Validation Report
2, 3, 7, 8-Tetrachlorodibenzo-p-Dioxin in Soil and
Sediment by Low Resolution Mass Spectrometry
Prepared by
U.S. EPA Environmental Monitoring
Systems Laboratory
Lockheed Engineering and Management
Services Company
Las Vegas, Nevada
WDC 63394.Tl (SA)
-------
May 1988
METHOD VALIDATION REPORT
2,3,7,8-TBTRACHLORO-DIBENZO-p-DIOXIN
IN SOIL AND SEDIMENT BY
LOW RESOLUTION MASS SPECTROMBTRY
F.C. Garner and G.L. Robertson
Environmental Programs
Lockheed Engineering and Management Services Company, Inc.
P.O. Box 15027
Las Vegas, Nevada 89119
Contract Number 68-03-3249
Job Order 70.02
J.D. Petty and D. W. Bottrell
Quality Assurance Division
Environmental Monitoring Systems Laboratory
944 B. Harmon
Las Vegas, Nevada 89119
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
OFFICE OP RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NEVADA 89119
-------
NOTICE
Although the research described in this report has been funded
wholly or in part by the United States Environmental Protection Agency
through contract number 68-03-3249 to Lockheed Engineering and
Management Services Company, it has not been subjected to Agency review
and therefore does not necessarily reflect the views of the Agency and
no official endorsement should be inferred. Mention of trade names or
commercial products does not constitute endorsement or recommendation
for use.
ii
-------
ABSTRACT
The United States Environmental Protection Agency's superfund
contractor Laboratories use the method found in USEPA IFB WA84-A002 to
analyze soil/sediment and water samples for the' specific isomer
2,3,7,8-tetrachloro-dibenzo-p-dioxin. This method uses isotope dilution
gas chromatography mass spectrometry. The precision, accuracy, and
other parameters are important in evaluating performance of laboratories
and methods. The observed performance of this method in the
production-line analysis of more than two thousand soil and sediment
samples by four contractor laboratories is presented in this paper.
iii
-------
iv
-------
TABLE OF CONTENTS
Page
Abstract Lii
Tables vi
Acknowledgment vii
Introduction 1
Conclusions 1
Method Description 2
Limitations 4
Performance Data 4
Precision 5
Accuracy 6
False Positive Rate 7
False Negative Rate 7
Specificity 8
Quality Control Performance 8
References 12
-------
TABLES
Number Page
1 Precision 5
2 Accuracy 6
vi
-------
ACKNOWLEDGMENT
The authors are grateful for the assistance of K. Huang of
Computer Sciences Coorporation in writing programs to summarize the data.
vii
-------
INTRODUCTION
The Environmental Protection Agency's (EPA) Environmental
Monitoring Systems Laboratory - Las Vegas (EMSL-LV) is responsible for
conducting a quality assurance program in support of the Agency's
Superfund Contract Laboratory Program (CLP). The EMSL-LV quality
assurance support program includes providing analytical calibration
standards and quality control materials, maintaining a quality assurance
data base, conducting performance evaluation studies, performing on-site
laboratory evaluations, and conducting technical audits of CLP data.
Through these activities the EMSL-LV continuously evaluates and
documents CLP laboratory and method performance. Analytical methods are
continually evaluated and periodically documented. Where these
assessments indicate a need for further knowledge about the performance
of methods, the BMSL-LV designs and conducts performance evaluation
studies to acquire this information. Through these studies areas for
improvement of analytical methodology are identified. The objective of
this paper is to document the performance of the CLP analytical method
for the analysis of 2,3,7,8-tetrachloro-dibenzo-p-dioxin (2,3,7,8-TCDD)
in soils and sediments. The method-has also been adapted for analysis
of water samples for 2,3,7,8-TCDD by using the extraction procedure in
EPA Method 613. However, an insufficient number of water samples has
been analyzed to make any statement of method performance in routine use.
CONCLUSIONS
The precision of the method can be characterized by a percent
relative standard deviation of 7.4 percent for fortified field blanks
and 12.6 to 23.1 percent for performance evaluation samples. Little, if
any, bias exists. The mean percent bias for the fortified field blanks
- 1 -
-------
was -4.9 percent, while that of the performance evaluation samples was
-27.6 to + 6.3 percent. Qualitatively, the method performance is
characterized by a false positive rate of one to four percent, and a
false negative rate of two to three percent at a concentration of 0.8 to
1.5 yg/kg. The required isomer specificity, which is based on a
percent valley criterion of 25 percent, is achieved about 90 percent of
the time with the contractually recommended columns. The three
sample-specific quality control procedures, the maximum possible
concentration, the surrogate signal-to-noise ratio, and the percent
recovery of the internal standard were all found to be useful in
identifying false negatives.
- 2 -
-------
METHOD DESCRIPTION
The USEPA has a standard method for use by contractor laboratories
in analyzing for 2,3,7,8-TCDD in soils and sediments. This method is
described in detail in USEPA IFB WA84-A002^ and by Kleopfer . The
method is intended to achieve a detection limit of one yg/kg and to
separate 2,3,7,8-TCDD from all other TCDD isomers. Surrogate
(37C1 -2,3,7,8-TCDD) and internal standard ("c -2,3,7,8-TCDD) are
added before sample extraction. The sample is then subjected to a jar
extraction technique using 150 mL hexane and 20 mL methanol in a shaker
for at least 3 hours. The extract is then subjected to cleanup using
silica gel, alumina, and Carbopak C. The eluate is concentrated by
evaporation under a gentle stream of dry nitrogen before GC/MS SIM
analysis. Peak areas are obtained from reconstructed ion current
profiles for m/z 257, 320, 322, 328, 332, and 334. The quantity of
native TCDD is estimated by calculating the ratio of the sum of the peak
areas for m/z 320 and 332 (native TCDD) to the sum of the peak areas for
m/z 332 and 334 (internal standard). This ratio is then multiplied by
the quantity of internal standard added to the sample and divided by the
sample weight and the mean response factor from the 12-point initial
calibration to obtain an estimate of the concentration of native TCDD in
the sample. Because of the isotope dilution technique, any bias due to
incomplete extraction or cleanup losses should be corrected by
equivalent losses of the internal standard, provided that enough of each
dioxin is recovered to enable identification and quantitation.
The following quality control analyses are routinely performed.
Routine Calibration - The lowest concentration calibration standard
(corresponding to 1.0 vg/kg in a 10 gram sample) must be analyzed once
at the beginning of each 12-hour period.
- 3 -
-------
Method Blank Analysis - One analysis per case (approximately 24
samples) is performed on laboratory reagents to determine if
contamination is occurring.
Surrogate Signal-to-Noise Ratio - The signal to noise ratio of the
surrogate is computed for each sample analyzed. The amount added
corresponds to 0.14 yg/kg in a 10 gram sample. The signal-to-noise
ratio is intended to demonstrate adequate low-level detectability.
Fortified Field Blank Analysis - One sample per case is fortified
with 1.0 yg/kg of native TCDD and analyzed to determine if matrix
effects are present.
Duplicate Analysis - One sample per case is analyzed twice to
determine within-laboratory precision.
Column Performance Check Analysis - A standard solution containing
several closely eluting isomers of TCDD is analyzed initially and at the
end of each 12-hour period to demonstrate adequate isomer resolution.
Performance Evaluation Sample Analysis - One sample per case is a
performance evaluation sample with a known amount of TCDD and
interferences. The concentration of this sample is not known by the
laboratories. Four well-characterized materials of 1 to 10 yg/kg
native TCDD in dry soil are used routinely, as are two blank materials.
Internal Standard Recovery - In each sample the response of the
internal standard (added before extraction) is compared to that of the
recovery standard (added just before injection). This provides a
- 4 -
-------
measure of losses of the internal standard during extraction and
cleanup. It is inferred that this provides information on the recovery
and detectability of native 2,3,7,8-TCDD.
Maximum Possible Concentration - For each sample in which
2,3,7,8-TCDD is not detected, the maximum possible concentration is
calculated. This is an estimate of the highest concentration of
2,3,7,8-TCDD that could be present based on the strength of the observed
instrumental response.
LIMITATIONS
The range of calibration corresponds to sample concentrations of
1.0 to 100 wg/kg 2,3,7,8-TCDD. Samples exceeding 100 yg/kg must be
reanalyzed using a smaller (1 g) sample aliquot to ensure accuracy, thus
effectively extending the calibration range to 1000 yg/kg.
PERFORMANCE DATA
The following statistics and tables are the results of several data
reductions of a comprehensive system of databases maintained at the
BMSL-LV. These databases include information from 4 contractor
laboratories analyzing thousands of real environmental samples from
hazardous waste sites across the United States under production-line
conditions in commercial laboratories. The information presented
represents a realistic appraisal of the performance of the method under
routine operating conditions.
- 5 -
-------
PRECISION
The precision of the method is expressed by the percent relative
standard deviation (% RSD). This quantity is estimated by the standard
deviation of 2,3,7,8-TCDD quantifications divided by the mean, and
multiplied by 100 to obtain a percentage. The precision is presented in
Table 1 for the fortified field blank analysis, the performance
evaluation sample analysis.
TABLE 1. PRECISION
Sample
Type
Fortified
field Blank
Performance
Evaluation
Performance
Evaluation
Performance
Evaluation
Target
Concentration
1.0 wg/kg
0.8 wg/kg
1.2 vg/kg
1.5 yg/kg
Number
of data
131
63
49"
34
% RSD
7
12
14
23
.4%
.6%
.2%
.1%
* Two high outliers were removed by Grubb's test.
Inspection of Table 1 reveals that the precision is poorer for the
performance evaluation samples than for the fortified field blank
analysis. This may be due to the laboratories knowing the target
concentration of the fortified field blanks samples, but not the
performance evaluation samples. Another possible contributing factor is
the much longer time that dioxin is present in the performance
evaluation materials.
- 6 -
-------
In a similar study by Garner, Hornsher, and Pearson , it was found
that the precision of the method could be characterized by % RSD values
between 10 and 20 percent. It was also found that there was no
statistically significant long-term interlaboratory component of
variance. For this reason, within-laboratory precision was not
displayed separately in Table 1.
ACCURACY
The accuracy is expressed as the percent bias, which is estimated
by the difference between the observed concentration and the target
concentration, divided by the target concentration, and then multiplied
by 100 to obtain a percentage. The accuracy is presented in Table 2 for
the fortified field blank analysis and the performance evaluation
samples analysis.
TABLE 1. ACCURACY
Sample
Type
Fortified
Field Blank
Performance
Evaluation
Performance
Evaluation
Performance
Evaluation
Target
Concentration
1.0 yg/kg
0.8 yg/kg
1.2 yg/kg
1.5 yg/kg
Number
of data
131
63
49»
34
Mean
Bias
- 4.9%
+ 6.3%
-27.6%
- 3.5%
* Two high outliers were removed by Grubb's test.
Inspection of Table 2 reveals that substantial bias is not
apparent. The 1.2 yg/kg performance evaluation sample shows
substantial negative bias (-27.6%), but may simply be due to anomalous
preparation of the performance evaluation material. A similar study
- 7 -
-------
found that the percent bias could be characterized by values between
zero and -20 percent, and that performance evaluation sample analyses
tended to be negatively biased.
FALSE POSITIVE RATE
A false positive occurs when native 2,3,7,8-TCDD is identified in a
sample which contains no native 2,3,7,8-TCDD. The frequency of
occurrence of false positives may be estimated by using the results of
the method blank and field blank analyses. Pour out of 175 method blank
analyses were reported to have identified native 2,3,7,8-TCDD. While
this represents a false positive rate of 2.3 percent, it must be noted
that none of these samples was reported to contain more than 0.18 yg/kg
native TCDD. The maximum permissible concentration of 2,3,7,8-TCDD that
may be identified in method blank analyses is 0.3 yg/kg. Five of 127
field blank analyses were reported to have identified native
2,3,7,8-TCDD. Three of these were quantified above 0.3 yg/kg, with
values of 0.61, 0.93, and 0.93 yg/kg. The false positive rate of 3-9
percent for field blanks may be more representative of performance for
routine environmental samples because the Identity of these samples is
not revealed to the laboratory. A previous study estimated the false
positive rate to be approximately one percent for method blank analyses
only.
FALSE NEGATIVE RATE
A false negative occurs when native 2,3,7,8-TCDD is not identified
in a sample which actually contains native 2,3,7,8-TCDD. The frequency
of false negatives may be observed for the fortified field blank
analysis and for the. performance evaluation sample analysis. Three out
of 134 fortified field blank analyses failed to identify native
2,3,7,8-TCDD, giving an estimated false negative rate of 2.2 percent.
Five of the 153 performance evaluation sample analyses failed to
- 8 -
-------
identify native 2,3,7,8-TCDD, giving an estimated false negative rate of
about 3.3 percent. The performance evaluation sample analysis probably
yields the most realistic information pertaining to false negatives, as
the laboratories are not informed in advance that native 2,3,7,8-TCDD is
actually present in these samples.
SPECIFICITY
The identification criteria for native 2,3,7,8-TCDD are clearly
defined in the IFB, and are designed to prevent any other compound from
being mistaken for native 2,3,7,8-TCDD. The criteria are based on
relative retention time, which must match the internal standard, and on
ion ratios. Peaks at mass 259, 320, and 322 must be observed and meet
signal-to-noise ratio requirements, and the ratio of integrated ion
current at mass 320 to that at mass 322 must be between 0.77 and 0.90.
The authors do not know of a single analysis where another compound or
isomer was mistaken for 2,3,7,8-TCDD. It was previously shown that
the GC columns recommended in the IFB could adequately resolve
2,3,7,8-TCDD from the nearest eluting isomer approximately 90 percent of
the time when analyzing the column performance check solution.
QUALITY CONTROL PERFORMANCE
The quality control procedures are an inherent part of the
analytical method and, as such, should be described and evaluated with
the method. The quality control procedures include estimation of the
maximum possible concentration of native 2,3,7,8-TCDD (formerly called
the detection limit), estimation of the signal-to-noise ratio of a
low-level isotopically-labelled "surrogate" compound, estimation of the
percent recovery of the internal standard, duplicate analysis, field
blank analysis, fortified field blank analysis, method blank analysis,
- 9 -
-------
and performance evaluation sample analysis. Bach of these analyses has
an important implied role in ensuring the quality of the final results.
Comparison of the actual performance of these procedures to the implied
quality control objectives will demonstrate their effectiveness.
The maximum possible concentration (MFC) is estimated for each
sample in which native 2,3,7,8-TCDD is not identified. The MFC is
intended to represent the greatest concentration of 2,3,7,8-TCDD which
could be present in a given sample. It is determined by integrating the
instrumental response over the appropriate retention time for mass 320
or mass 322, whichever is less, quantifying this result through the
calibration response factor, then multiplying by 2.5. The smaller peak
is used because potential interferences at the larger peak may have been
the cause of failure to identify native 2,3,7,8-TCDD. The multiplier
2.5 was chosen to be sufficiently large to ensure a small probability of
the true concentration of native 2,3,7,8-TCDD exceeding the MFC.
Approximately 2,500 values of the MFC were observed for analysis of
environmental samples. The median value was 0.05 yg/kg, while the 95th
percentile was 0.39 yg/kg, and the largest observed value was 80.5
yg/kg. More importantly, it was also observed that for performance
evaluation samples in which native 2,3,7,8-TCDD was not identified
(presumably through some laboratory error) the MFC usually exceeded the
true concentration of native 2,3,7,8-TCDD. Hence the MFC is a useful
piece of quality control information.
An isotopically labelled surrogate compound, 3?C1 -2,3,7,8-TCDD is
added to every sample before extraction. The surrogate is not
quantified. A previous work showed the quantification of the
surrogate to be poorly correlated with the quantification of native
TCDD, which led to the USEPA altering the method. The surrogate is
currently added at a very low level, 0.14 yg/kg, to represent
- 10 -
-------
the approximate strength of the tertiary ion, mass 259, of native TCDD
at i yg/kg. The signal-to-noise ratio of the surrogate is measured and
reported for every sample. Approximately 2,700 samples had a median
reported surrogate signal-to-noise ratio of 28, a 95th percentile of 97,
and a maximum of 270. All of the values reported were greater than or
equal to 1.0, and 95 percent exceeded 4.3. Values of the surrogate
signal-to-noise ratio were examined for the performance evaluation
samples in which the laboratory failed to identify native 2,3,7,8-TCDD.
Below average signal-to-noise ratios were observed, with a median of
about 3.2 and a maximum of 6.7. Hence the surrogate signal-to-noise
ratio is a useful piece of quality control information.
The surrogate compound was previously used to demonstrate the
accuracy of the analysis for each sample. However, previous work
showed that the surrogate accuracy was not closely related to accuracy
of the analysis of native TCDD. The surrogate was also much easier to
detect than native TCDD at the same concentration, because the surrogate
had no ion ratio criteria, no tertiary ion criteria, and the response
was concentrated at one molecular mass. Subsequently, it was decided
not to use the surrogate compound as an accuracy check. Instead, the
current use is as a low-level detectability check.
The internal standard, C -2,3,7,8-TCDD, is also added to every
sample before extraction. The percent recovery of the internal standard
is measured against another isotopically-labelled TCDD, the recovery
standard, which is added to the extract just before injection.
Approximately 2,700 reported values of the percent recovery of the
internal standard indicated a median of 48 percent, a 95th percentile of
85 percent, and a maximum of 139 percent. The smallest reported value
was 0.0 percent while 95 percent of the reported values exceeded 17
percent. False negatives on performance evaluation samples are
- 11 -
-------
generally associated with lower than average percent recoveries of the
internal standard, with a median reported value of 9 percent and a
maximum of 26 percent. Thus, the percent recovery of the internal
standard is a useful piece of quality control information.
Clearly there is some redundancy in the information produced by
these quality control procedures. In the authors' opinion, the low
level surrogate has the least value of the quality control procedures
used and the cost and the benefits of its continued use should be
examined. The authors also feel that quality control limits should be
placed on the KPC and the percent recovery of the internal standard. A
reasonable upper limit for the MFC would be about 1.0 yg/kg, or perhaps
a bit lower. A reasonable lower limit for the percent recovery of the
internal standard would be somewhere in the range of 10 to 20 percent.
- 12 -
-------
References
1. Invitation for Bid, Solicitation No. WA84-A002, United States
Environmental Protection Agency, 400 M Street SW, Washington, D.C.,
20460, 1984.
2. Kleopfer, R. 1986. "Dioxin Analytical Methods-General Description
and Quality Control Considerations, "Quality Control in Remedial
Site Investigation: Hazardous and Industrial Solid Waste Testing.
Fifth Volume. ASTM STP 925. C.L. Perket, Ed., American Society for
Testing and Materials.
3. Garner, F.C., Homsher, M.T. , and Pearson, J.G. 1986. "Performance
of USBPA Method for Analysis of 2,3,7,8-Tetrachloro-dibenzo-p-dioxin
in Soils and Sediments by Contractor Laboratories," Quality Control
in Remedial Site Investigation: Hazardous and Industrial Solid
Waste Testing. Fifth Volume. ASTM STP 925. C.L. Perket, Ed.,
American Society for Testing and Materials.
- 13 -
-------
This document was produced on the Lockheed-EMSCO word processing system
and can be identified for the purpose of reproduction by the control
number WP-2278C.
- 14 -
-------
APPENDIX T
Retrospective Probability of Detecting the
Target-Size Locally Contaminated Area
Soil Assessment-2, 3, 7, 8-TCDD
-------
Appendix T
RETROSPECTIVE PROBABILITY OF DETECTING
THE TARGET-SIZE LOCALLY CONTAMINATED AREA
SOIL ASSESSMENT--2,3,7,8-TCDD
In response to the written preliminary comments from
Dr. Janick Artiola, a simulation was conducted to determine
the retrospective probability of detecting the target-size
locally contaminated area using the points sampled during
the sampling program for 2,3,7,8-TCDD. In summary, the fact
that 14 of the 2,274 points sampled had no valid analytical
results could create a high bias. Paved areas larger than
the target-size locally contaminated area, which will give a
low bias, exist within the EDA. The simulation conducted
does not correct for these considerations; thus, the
estimate of the retrospective probability is probably lower
than that actually achieved. The gross estimate of the
retrospective probability of detecting the target-size
locally contaminated area is 88 percent.
The following discussion of the simulation is organized in
the following sections: "Background," "Scope," and
"Results."
BACKGROUND
The sampling program was designed to have a 95 percent
probability of detecting a locally contaminated elliptical
area approximately 126 feet long by 66 feet wide. This
ellipse has an area of approximately 6,500 square feet,
which is the median lot size in the EDA.
During the implementation of this sampling plan, several
sample locations had to be moved because of the
infeasibility of sampling the assigned location (pavement,
structure, etc.).
SCOPE
An estimate of the retrospective probability of detecting a
locally contaminated area of 6,500 square feet was developed
by use of Monte Carlo simulation. The coordinates of the
EDA neighborhood boundaries and the points sampled were used
as a data base by the simulation. The computer program
estimated the probability of detecting a locally contam-
inated area by generating a simulated locally contaminated
area and placing it at a random location and orientation on
the map of the EDA. If the simulated area was within the
EDA and contained a sampled location, it was counted as a
T-l
-------
hit. If the simulated area was within the EDA but did not
contain a sampled location, it was counted as a miss.
The software for performing the Monte Carlo simulations was
implemented on a personal computer-AT (PC-AT) with high-
resolution graphics. Graphical displays of maps of the EDA,
actual dioxin sampling sites, and the locally contaminated
area ellipses were used in the software development process
for visual QA/QC. The following sequence of operations was
performed by the simulation software:
1. The X-Y coordinates of the boundaries of the 13 EDA
neighborhoods were read and loaded into the memory.
2. A map of the EDA was displayed to ensure that the
neighborhood boundary data were loaded correctly.
3. The X-Y coordinates of the dioxin sampling sites were
read and loaded into the memory.
4. The dioxin sampling sites were displayed on the map of
the EDA to provide a visual check of the site data.
5. An elliptical locally contaminated area was
analytically placed on the map of the EDA by taking an
ellipse with a centroid at the origin, rotating that
ellipse a random number of degrees, and then
translating the rotated ellipse to the EDA area a
random number of feet up and a random number of feet
over.
6. The locally contaminated area placed in step 5 was
classified as to whether or not it fell totally within
the EDA by using a 16-point representation of the
ellipse and performing a check on the 16 points to
determine if all 16 points lay within any of the
13 neighborhoods of the EDA. If all 16 points lay
within one or more EDA neighborhoods, then the locally
contaminated area was classified as being within the
EDA.
7. If the locally contaminated area was classified as
being within the EDA in step 6, then the dioxin
sampling sites were checked to see if any sites lay
within the locally contaminated area, then, if so, the
locally contaminated area was classified as "sampled";
otherwise it was classified as "not sampled."
Analytically, this was done by using the equation for
an ellipse:
T-2
-------
X2/a2 + Y2/b2 = 1, where (1)
a = one-half of the length of the major axis
b = one-half of the length of the minor axis
X,Y = the ellipse border centered at the origin
Given values for X and Y, if the left-hand side of
equation 1 was less than or equal to 1, then the
X-Y point lies inside or on the border of the ellipse.
The proper X-Y coordinate transformations were
performed and applied to equation 1 to determine if a
site was inside the locally contaminated area.
8. Steps 5 through 7 were repeated many times to get a
good estimate of the probability of sampling a locally
contaminated area.
At all steps in the development, graphical displays were
used to verify that the procedures were correct. For
instance, graphical displays of the randomly placed ellipses
with the classification of that ellipse as "in" or "out" of
the EDA were used to verify steps 5 and 6.
RESULTS
A simulation run of 10,000 randomly placed ellipses was used
to estimate the probability that a locally contaminated area
would have been sampled. The results of this run were:
Number of ellipses simulated = 10,000
Number of ellipses in the EDA = 4,144
Number of ellipses with sites = 3,634
Number of ellipses without sites = 510
Percentage of ellipses with sites = 88 (3,634/4,144)
The ratio of hits to misses is an empirical estimate of the
probability of detecting a contaminated area. This
88 percent estimate could be biased slightly high because
14 of the 2,274 points sampled had no valid results;
however, this estimate is biased low because the EDA
contained areas of 6,500 square feet or larger that did not
have accessible soil (i.e., the area was paved) but were not
removed from the simulated EDA.
WDR356/053
T-3
-------
APPENDIX U
Letter:
Summary of Preliminary Research on Sampling
Point with Results above Level of Concern,
March 31 1988
Soil Assessment-2, 3, 7, 8-TCDD
Prepared by
New York State Department of Health
Albany, New York
WDC 63334 .T1 (EAI
-------
STATE OF NEW YORK
DEPARTMENT OF HEALTH
Corning Tower The Governor Nelson A. Rockefeller Empire State Plaza Albany, New York 12237
David Axelrod M D
Commissioner
OFFICE OF PUBLIC HEALTH
Linda A Randolph. M D . M P H
Director
William F Leavy
Executive Deputy Director
March 31, 1988
Doug Garbarini
USEPA-Region II
26 Federal Plaza
New York, New York 10278
Dear Doug:
RE: Vacant Lot with Dioxin
The following information has been uncovered regarding the lot where
approximately 20 ppb 2378-TCDD was found in soil:
Lot number - 161.15-2-53 (called Lot C on CH2M Hill maps)
Lots in
fenced area
661 100th St
657 100th St
Lot C
645 100th St
LCARA House Previous
purchased demolished owner
2/18/81 9/4/85 LeMaster
4/3/81 4/2/84 Starr
owned by City of Niagara Falls
12/1/80 9/4/85 Stevens
Lot C is located on the southern margin of a "swale". A soil core taken from the lot
at 657 100th St. found "fill" to at least 3 feet in depth, and undisturbed soil was found
at 6 feet. At 661 100th St. fill extended to at least 4 feet with undisturbed soil at 5
feet. Data for intervening depths was not available.
Sincerely,
Edward G. Horn, Ph.D.
Environmental Scientist
Division of Environmental Health
Assessment
cc:
J. Liddle
M. O'Toole
W. Stasiuk
T. VanEpp
-------
APPENDIX V
Letter:
Results ofFollowup Investigation,
July 6, 1988
Soil Assessment—Indicator Chemicals
; Prepared by
I
i
! Ecology and Environment, Inc.
i Lancaster, New York
WOC 63394.T1 ISA)
-------
ecology and environment, inc.
Ill BUFFALO CORPORATE CENTER
368 PLEASANTVIEW DRIVE, LANCASTER, NEW YORK 14086, TEL. 716/684-8060
International Specialists in the Environment
July 6, 1988
Mr. Doug Garbarini
USEPA - Region II
26 Federal Plaza, Room 737
New York, NY 10278
RE: 2,3,7,8-TCDD Concentrations in Soil Samples Collected in the
Vicinity of Lot C, 100th St., Love Canal EDA, Niagara Falls, NY
Dear Mr. Garbarini:
This letter report summarizes the results of 2,3,7,8-tetrachlorodi-
benzo-p-dioxin (2,3,7,8-TCDD) analysis of soil samples collected by
Ecology and Environment, Inc. (E & E), on April 11 and 12, 1988, in and
around Lot C, 100th Street, Love Canal EDA, Niagara Falls, New York.
HISTORY
During 1986 and 1987, over 2,000 soil samples were collected in the Love
Canal Emergency Declaration Area (EDA) to determine the concentrations
of 2,3,7,8-TCDD as part of the Love Canal EDA Habitability Study. These
results are summarized in Volume IV of the Love Canal EDA Habitability
Study report (CH2M Hill, March 1988).
In the 1986-87 study, 2,3,7,8-TCDD was found at concentrations above the
study target level of 1 part per billion (1 ppb) at only one sample
location (Station 4256), located in Lot C, 100th Street, where only
surface soil was sampled. In several analyses, the 2,3,7,8-TCDD was
found at concentrations ranging from 17.3 to 21.2 ppb.
Based on these data, USEPA-Region II requested E & E to sample surface
and subsurface soils in the vicinity of Station 4256 to determine the
extent of dioxin contamination. The sampling strategy was developed by
NUS Corporation, Superfund Division, and employed a radially stratified
random sample design (NUS 1986).
E & E determined sampling locations and developed the soil sampling plan
(E & E 1988). Surface (0 to 2 inches) and subsurface (2 to 7 inches,
and 7 to 12 inches) samples were collected at 33 locations (see Fig-
ure 1). The surface sample was analyzed from all locations. Subsurface
samples were analyzed from selected locations. One-hundred-sixty-six
samples were collected, including 32 QA/QC samples. S^mpl* :g pro-
cedures, chain-of-custody procedures, documentation and the Health and
Safety plan are presented in the soil sampling plan (E & E 1988) and the
sampling letter report appended to this letter.
recycled paper
-------
Mr. Doug Garbarini
July 5, 1988
Page 2
ANALYTICAL LABORATORIES
Thirty-five surface soil samples and eight QA/QC samples were shipped
for analysis to Compuchem Laboratories, Research Triangle Park, NC, on
April 12, 1988. Sixteen subsoil samples were sent for analysis to
Environmental Testing Corporation on May 13, 1988. The results of these
analyses are presented herein.
ANALYTICAL RESULTS
Surface soil analyses in the current study revealed 2,3,7,8-TCDD at only
one location (Station 1033) at a concentration of 35.1 ppb. This
location was within several feet of Station 4256 of the 1986-87 study.
Subsoil samples were also collected at Station 1033. Analyses for
2,3,7,8-TCDD revealed 32.7 ppb in the 2 to 7 inch interval (sample
1033-2) and 5.9 ppb in the 7 to 12 inch interval (sample 1033-4).
All analytical results are presented in Table 1. Horizontal and
vertical coordinates shown were calculated relative to the center point
(1033). Radial distances from the center point to each sample station
are also shown.
Maximum possible concentrations are listed for samples which were below
the analytical detection limit.
DISCUSSION
The compound 2,3,7,8-TCDD was found at only one location (Station 1033),
in close proximity to Station 4256, the only location where 2,3,7,8-TCDD
was found in the 1986-87 study. The surface soil concentration in the
current study was the same range as was found in the 1986-87 study.
Decreasing concentrations were found at progressively greater depths.
No other detectable concentrations were found. No statistical data
analysis is presented, since there is no evidence of any lateral
concentration gradient at the site.
Sincerely,
Joseph T. Angley, Ph.D.
JTA/j f
Enclosure
LI
-------
Table 1
SUMMARY OF ANALYTICAL RESULTS
E t E
SAMPLE
NO.
SECTOR
DEPTH
INCHES
HORIZONTAL
COORDINATE
VERTICAL 2
COORDINATE
CONCENTRATION
PP8
WC3
RADIAL DISTANCE
FROM CENTER (FT)
1988 DATA
1033-1
1833-3
1833-4
1001-1
1001-3
1681-4
1602-1
1882-3
1882-4
CENTER POINT
CENTER POINT
CENTER POINT
Al
Al
Al
A2
A2
A2
e-2
2-7
7-12
e-2
2-7
7-12
8-2
2-7
7-12
e
8
8
-6.688
-6.688
-6.688
-9.385
-9.385
-9.385
8
1
8
+ 8.772
+ 8.772
+ 8.772
+ 8.561
+ 8.561
+ 8.561
35.18
32.7
5.9
ND«
ND
ND
ND
ND
ND
. 5
-
-
8.46
8.13
8.12
8.13
8.28
8.13
8
8
8
6.65
6.65
6.65
12.64
12.64
12.64
1803-1
1808
1842-1
1842-3
1042-4
ieie
- 5.332
+12.122
+ 3.799
+ 8.788
+ 8.788
+ 8.788
-5.387
-5.641
-5.582
-5.582
-5.582
-18.364
ND
8.87
ND
ND
ND
ND
M>
8.52
8.49
8.17
8.12
8.27
13.24
1004-1
1004-3
1004-4
1005-1
1005-6
1006-1
1006-3
1006-4
1007-1
1007-3
1087-4
A4
A4
A4
A5
A5
A6
A6
A6
A7
A7
fl7
8-2
2-7
7-12
8-2
8-2
8-2
2-7
7-12
8-2
2-7
7-12
+ 1.485
+ 1.485
+ 1.485
+18.343
+18.343
+ 4.216
+ 4.216
+ 4.216
+ 6.835
+ 6.835
+ 6.835
+ 2.896
+ 2.896
+ 2.896
+ 6.439
+ 6.439
+ 2.118
+ 2.118
+ 2.118
-3.115
-3.115
-3.115
ND
ND
ND
ND
REJECTED 6
ND
ND
ND
ND •
ND
ND
8.72
8.17
8.18
8.43
'
8.77
8.17
8.888
8.72
8.16
8.12
3.25
3.25
3.25
12.18
12.18
4.71
4.71
4.71
7.51
7.51
7.51
6.8
5.63
5.63
5.63
11.68
1011-1
1011-3
1PI1-4
1012
1613
1014
1015
1016
1017
All
All
nn
A12
Bl
B2
B3
B4
B5
0-2
2-7
7-12
0-£
8-2
8-2
8-2
8-2
8-2
-18.863
-18.863
-J6.8U
-8.885
-15.458
-14.248
- 2.668
+ 8.966
+22. 749
-8.K7
- 8.827
- 8.827
- 2. 198
+ 3.819
+ 9.498
+17.965
+15.747
+15.753
ND
ND
ND
ND
ND
ND
ND
ND
ND
0.a
8.11
8.23
8.55
8.48
8.57
8.33
8.62
8.7l
12.87
12.87
12.87
9.15
15.92
17.12
18.16
18.12
27.67
-------
Table 1 (Cont.)
E « E
SAWLE
NO.
SECTOR
DEPTH
INCHES
HORIZONTAL
COORDINATE
VERTICAL *
COORDINATE
CONCENTRATION
PPB
MPC 3
RADIAL DISTANCE
FROM CENTER (FT)
1018-1
1031-1
1020-1
1021-1
1022-1
1023-1
1024-1
1025-1
1025-5
1059-1
1027-1
1028-1
1029-1
1030-1
1032-1
B6
B7
B7
68
B9
610
Bll
B12
Cl
Cl
C2
C3
C4
C5
C6
C8
0-2
0-2
0-2
0-2
0-2
0-2
0-2
0-2
0-2
0-2
0-2
0-2
0-2
0-2
0-2
0-2
+29.137
+15.422
+ 8.505
+20.870
+ 9.648
- 5.067
-13.592
-15.241
- 1.333
- 1.333
+12.335
+52.630
+39.840
+26.611
+48.016
-15.681
+ 5.814
- 8.674
-32.496
-14.824
-18.802
-30.171
-12.383
-6.005
+37.379
+37.379
+31.986
+37.303
+ 3.610
-5.560
-£9.260
-32.806
ND
ND
ND
ND
ND
REJECTED
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
0.64
1.40
0.33
0.29
0.35
0.40
0.40
0.66
0.26
0.23
0.65
0.50
0.2B
0.74
0.21
29.71
17.69
33.53
25.60
21.13
30.59
16.39
16.38
37.40
37.40
34.28
64.51
40.00
27.19
56.23
36.11
1986-1987 DATA
1778
1779
1780
3298
4256
0-5
0-2
- ' 0-2
0-5
0-2
- 13.53
-112.45
+ 38.86
-24.98
+ 2.56 •
+ 55.51
-154.22
- 27.79
- 19.51
+ 0.52
ND
ND
ND
ND
17.3-21.2
57.14
190.86
47.77
31.70
2.61
1. Values represent distance in feet fro* the center point, perpendicular to the datui line in Figure 1. Minus values indicate westward
direction. Positive values indicate eastward direction.
2. Values represent distance in feet froa the center point, parallel to the datui line in Figure 1. Minus values indicate southward directior.
Positive values indicate northward direction.
3. HPC: Baxiiu* possible concentration.
4. ND: not detected.
5. - not applicable
6. Rejected: indicates data did not pass Qfl review by USEPA Region II Environmental Services Division.
-------
1122678N
100-19AA
1
1
CO
•c
4*
C)
B1
1033
100-18 A A
B12-*
3
: In feet
0 5 10 20 30 40 SO
KEY:
(i) Coordinate Reference Point
1986-1987 Sampling Station
• 1988 Sampling Station (Surface So lit Analyzed)
A Surveying Control Point
-•• 1988 Sampling Station (Surface and Su*
Soils Analyzed) ' -- <•
Figure 1 ACTUAL DIOXIN SAMPLING POINTS, 100TH STREET,
LOVE CANAL EDA, NIAGARA FALLS, NEW YORK
recycled paper
lu£} and environment
-------
APPENDIX W
Memorandum:
Health Consultation Based on Results of the So/7
Assessmenf for 2, 3, 7, 8-TCDD,
June 17, 1988
Prepared by
U.S. Department of Health & Human Services/
Agency for Toxic Substances and Disease Registry
Atlanta, Georgia
W2C63394.T1 (SA1
-------
Public Health Service
DEPARTMENT OF HEALTH & HUMAN SERVICES Agency for Toxic Substances
and Disease Registry
Memorandum
Date -June 17, 1988
From lexicologist
Emergency Response Branch
Subject Health Consultation: Love Canal Emergency Declaration Area Habitability
Study, Niagara Falls, New York
To Mr. William Q. Nelson
Public Health Advisor
EPA Region II
Through: Chief, Emergency Response Branch, OHA, ATSDR
Region II of the Environmental Protection Agency (EPA) asked the Agency
for Toxic Substances and Disease Registry (ATSDR) to reviev the results of
the Love Canal Emergency Declaration Area Habitability Study investigation
of surface soil for dioxin contamination.
STATEMENT OF PROBLEM
Public concern for chemical contamination in the area around the Love
Canal prompted the Federal Government to provide funds to purchase the
houses from those who wished to sell. Criteria for habitability of this
area have been developed. A portion of this criteria is an assessment of
dioxin contamination of the area surface soil. The results of an
extensive surface soil sampling investigation are the subject of the
primary document reviewed.
DOCUMENTS REVIEWED
1. "Love Canal Emergency Declaration Area Habitability Study," Final
Report, CH2MHill Southeast, Inc., Reston, Virgina, March 1988.
2. "Soil Sampling Plan for Love Canal EDA Station NO. 4256," Ecology
and Environment, Inc., April 5, 1988.
DISCUSSION
Both the Centers for Disease Control (CDC) and the ATSDR participated in
design of the dioxin sampling plan for the Emergency Declaration Area
(EDA). We believe that the execution of the designed plan provided data
sufficient to define the surface soil 2,3,7,8-TCDD contamination within
the EDA.
-------
Page 2 - Mr. William Q. Nelson
In a 1984 paper, Kimbrough et. al., (J. Tox. & Envir. Health, 14: 47-93)
stated that 1 ppb of 2,3,7,8-TCDD in soil is a reasonable level at which
to begin consideration of action to limit human exposure for contaminated
soil. The results from 2274 surface soil samples show that the general
2,3,7,8-TCDD contamination within the EDA is below the action level for
surface soils. Chemical analysis showed positive 2,3,7,8-TCDD results
below the action level for 48 samples. Of these only three sample
locations had results greater than 0.5 ppb. These results show that
2,3,7,8-TCDD is not present in the surface soil of the EDA at a
concentration of human health concern.
Only one sample had a 2,3,7,8-TCDD concentration exceeding the action
level. The results for this sample ranged from 17.3 to 21.2 ppb. The
location was directly across the street from the gate to the Love Canal
dewatering facility. The positive results prompted analysis of additional
aliquots of the sample in order to substantiate the results. In response
to these results EPA installed a fence around the entire property
associated with the sample location in order to restrict human access.
They then had a contractor develop a point source sampling plan to
determine the extent of contamination around this sample point. The
results from this sampling are now undergoing quality assurance/quality
control (QA/QC) evaluation. Preliminarily, the results from a repeat
sample within 6 inches of the initial sample range from about 25 to 35 ppb
(private communication from CDC's member on the Technical Review
Committee). Areas based on concentric circles divided into 12 sectors
each and centered on the initial sample location defined the sampling plan
locations. Reportedly, results from randomly selected discrete sampling
locations within each sector are in the range from nondetect to low tenths
of a ppb. Of these, only those from 2 sectors are positive for
2,3,7,8-TCDD, slightly above the detection limits. Further evaluation of
these results awaits the completion of the QA/QC evaluation.
CONCLUSIONS AND RECOMMENDATIONS
Based upon the results of an extensive surface soil sampling for
2,3,7,8-TCDD within the EDA, it is the opinion of ATSDR:
1. That the investigation for 2,3,7,8-TCDD contamination is adequate
to define the potential exposure of humans posed by that
chemical's presence in the area.
2. That there is no evidence of widespread 2,3,7,8-TCDD *
contamination in the surface soil within the EDA which might
present any threat to human health for the residents. •>:--
"*
-------
Page 3 - Mr. William Q. Kelson
3. That the only soil requiring remediation because of the evidence
of 2,3,7,8-TCDD prior to residential use is at sample location
4256.
/ S1 /
Mark A. McClanahan, Fh.D
------- |