REPORT ON THE PEER REVIEW OF THE DATA EVALUATION AND
INTERPRETATION REPORT AND LOW RESOLUTION SEDIMENT
CORING REPORT FOR THE HUDSON RIVER PCBs SUPERFUND SITE
—Final Report—
Preparedfor:
U.S. Environmental Protection Agency, Region n
Emergency and Remedial Response Division
290 Broadway, 18th Floor
New York City, NY 10007-1866
EPA Contract No. 68-W6-0022
Work Assignment No. 3-12
Prepared by:
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02421
June 3, 1999
-------
REPORT ON THE PEER REVIEW OF THE DATA EVALUATION AND
INTERPRETATION REPORT AND LOW RESOLUTION SEDIMENT
CORING REPORT FOR THE HUDSON RIVER PCBs SUPERFUND SITE
—Final Report—
Preparedfor:
U.S. Environmental Protection Agency, Region II
Emergency and Remedial Response Division
290 Broadway, 18th Floor
New York City, NY 10007-1866
EPA Contract No. 68-W6-0022
Work Assignment No. 3-12
Prepared by:
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02421
June 3, 1999
-------
NOTE
This report was prepared by Eastern Research Group, Inc. (ERG), an EPA contractor, as
a general record of discussion for the peer review meeting. This report captures the main points
of scheduled presentations and highlights discussions among the reviewers. This report does not
contain a verbatim transcript of all issues discussed during the peer review. Additionally, the
report does not embellish, interpret, or enlarge upon matters that were incomplete or unclear.
EPA will evaluate the recommendations developed by the reviewers and determine what, if any,
modifications are necessary to the current modeling approach. Except as specifically noted, no
statements in this report represent analyses or positions of EPA or of ERG.
-------
TABLE OF CONTENTS
LIST OF ABBREVIATIONS iii
EXECUTIVE SUMMARY v
1.0 INTRODUCTION 1-1
1.1 Background 1-2
1.2 Scope of the Peer Review 1-3
1.2.1 Selecting the Reviewers 1-3
1.2.2 Briefing the Reviewers 1-4
1.2.3 The Peer Review Meeting 1-5
1.3 Report Organization 1-6
2.0 RESPONSES TO SPECIFIC QUESTIONS REGARDING THE DEIR 2-1
2.1 Responses to Question 1 2-1
2.2 Responses to Question 2 2-4
2.3 Responses to Question 3 2-5
2.4 Responses to Question 4 2-7
2.5 Responses to Question 5 2-8
2.6 Responses to Question 6 2-9
2.7 Responses to Question 7 2-10
3.0 RESPONSES TO SPECIFIC QUESTIONS REGARDING THE LRC 3-1
3.1 Responses to Question 1 3-1
3 .2 Responses to Question 2 3-2
3.3 Responses to Question 3 3-3
3.4 Responses to Question 4 3-4
3 .5 Responses to Question 5 3-5
3 .6 Responses to Question 6 3-6
3 .7 Responses to Question 7 3-8
4.0 RESPONSES TO GENERAL QUESTIONS REGARDING THE DEIR
AND LRC 4-1
4.1 The Usefulness of the Data Set for Understanding Fate and Transport
of PCBs in the Upper Hudson River 4-1
4.2 Recommended Additional Data Analyses 4-3
-------
TABLE OF CONTENTS (Continued)
5.0 REVIEWERS' OVERALL RECOMMENDATIONS 5-1
5.1 Peer Reviewers'Final Statements 5-1
5.2 Summary of Peer Reviewers' Final Recommendations 5-6
6,0 REFERENCES 6-1
APPENDIX A List of Expert Peer Reviewers
APPENDIX B Charge to Expert Peer Reviewers
APPENDIX C Premeeting Comments, Alphabetized by Author
APPENDIX D List of Registered Observers of the Peer Review Meeting
APPENDIX E Agenda for the Peer Review Meeting
APPENDIX F Summaries of Observers* Comments
APPENDIX G Minutes from the January 1999 Briefing Meeting
ii
-------
LIST OF ABBREVIATIONS
DEIR
Data Evaluation and Interpretation Report
DNAPL
dense, nonaqueous phase liquid
DOC
dissolved organic carbon
EPA
U.S. Environmental Protection Agency
ERG
Eastern Research Group, Inc.
GE
General Electric Company
HRC
high-resolution sediment coring
LRC
Low Resolution Sediment Coring Report or low-resolution sediment coring
MDPR
molar dechlorination product ratio
MW
molecular weight
NYSDEC
New York State Department of Environmental Conservation
PCA
principal component analysis
PCB
polychlorinated biphenyl
PPm
parts per million
RPD
relative percent difference
TID
Thompson Island Dam
TIP
Thompson Island Pool
iii
-------
EXECUTIVE SUMMARY
Six independent peer reviewers critiqued the following reports prepared as part of the U.S.
Environmental Protection Agency's (EPA's) reassessment of the Hudson River PCBs Superfund
site, the "Data Evaluation and Interpretation Report," the "Low Resolution Sediment Coring
Report," and Responsiveness Summaries for both documents. After thoroughly discussing and
debating the scientific rigor of the main conclusions of these reports, the reviewers unanimously
agreed that the reports were acceptable. Four of the six reviewers found the reports "acceptable
with minor revisions;" the other two reviewers found the reports acceptable, but they were unsure
if their recommended revisions were "minor" or "major."
When answering the questions in the charge, the reviewers generally agreed with the major
conclusions of the DEIR and LRC (e.g., the sediments in the Thompson Island Pool act as a
source of PCBs to the water, the data suggest that most hot spots have lost PCBs, widespread
burial of PCBs is not occurring, and so on), but they suggested that some conclusions should be
modified to more accurately reflect the supporting data. At the close of the peer review meeting,
every reviewer listed his major findings and recommendations. Following is a list of specific
recommendations that at least two reviewers made during their closing statements. Specific
examples of other suggested revisions and recommendations made by the reviewers can be found
throughout this report.
• The reviewers unanimously agreed that the reports should have included multivariate
statistical analyses to identify and quantify trends and patterns among the data, but
especially for evaluating the large volume of congener-specific data.
• Every reviewer thought the reports should have more prominently acknowledged the
uncertainty associated with some major findings. The reviewers were particularly
concerned with reporting estimated PCB mass losses from hot spots as firm numbers. The
reviewers suggested that reporting a range of estimated mass losses might have been more
appropriate.
• The reviewers agreed that the DEIR's original finding on anaerobic dechlorination of PCBs
was not supported by the data. The reviewers thought a more accurate conclusion would
v
-------
indicate that dechlorination is predictable at higher PCB concentrations, but this should not
be taken as evidence of lack of dechlorination at lower concentrations.
• Several reviewers recommended that EPA publish a concise summary of the main findings
of the DEIR, the LRC, and the Responsiveness Summaries.
• Several reviewers recommended that EPA validate selected conclusions in the DEIR with
the results from more recent water column sampling data.
• Several reviewers agreed that the DEIR and LRC did not fully characterize the fate of
PCBs in the Hudson River. Two reviewers indicated that EPA should have considered
evaporative losses, photochemical degradation, and aerobic degradation in the reports.
vi
-------
1.0 INTRODUCTION
This report summarizes an independent peer review by six experts of the following
documents the U.S. Environmental Protection Agency (EPA) released as part of its reassessment
of the Hudson River PCBs Superfiind site:
• The February 1997 "Data Evaluation and Interpretation Report" (DEIR) (TAMS et al.,
1997)
• The December 1998 "Responsiveness Summary" for the DEIR (TAMS et al., 1998a)
• The July 1998 "Low Resolution Sediment Coring Report" (LRC) (TAMS et al., 1998b)
• The February 1999 "Responsiveness Summary" for the LRC (TAMS et al., 1999)
To facilitate their evaluations of these reports, the reviewers also were given copies of the
"Hudson River Reassessment Database," which contains all of the sampling data used to prepare
the above reports.
The six reviewers attended two meetings, which were both open to the public. The first
meeting, which took place in Albany, New York, on January 11-12, 1999, included several
presentations and a tour of the Upper Hudson River to familiarize the reviewers with the site and
its environmental history. The second meeting, which took place in Albany on March 16-18,
1999, was the forum in which the reviewers critiqued the above documents. Eastern Research
Group, Inc. (ERG), a contractor to EPA, organized the expert peer review and prepared this
summary report.
This introductory section provides background information on the Hudson River PCBs
Superfiind site, the scope of the peer review of the DEER, and LRC, and the organization of the
report.
1-1
-------
1.1 Background
In 1983, EPA classified approximately 200 miles of the Hudson River in the state of New
York as a Superfund site, due to elevated concentrations of polychlorinated biphenyls (PCBs) in
sediments. The sediments are believed to have been contaminated by discharges of PCBs over
approximately 30 years from two General Electric (GE) capacitor manufacturing plants, one in
Hudson Falls and the other in Fort Edward. The superfund site runs from Hudson Falls to New
York City. After an initial site assessment, EPA issued an "interim No Action decision" in 1984
for the contaminated sediments of the Hudson River PCBs site.
Since 1990, EPA has been reassessing its earlier decision to determine whether a different
course of action is needed for the contaminated sediments in the Hudson River. EPA is
conducting this reassessment in three phases: compiling and analyzing existing data for the site
("Phase I"), collecting additional data and using models to evaluate human health and ecological
risks ("Phase II"), and studying the feasibility of remedial alternatives ("Phase in"). As part of
Phase II, EPA's contractors conducted field studies to characterize levels of PCBs in the water
and sediments of the Hudson River to better understand the factors that affect the fate and
transport of PCBs in this system. The original findings of these studies are documented in the
DEIR and LRC. Since EPA released these reports, several parties submitted comments during
the designated public comment periods, after which EPA's contractors prepared Responsiveness
Summaries to address the comments.
To ensure that the assumptions, methods, and conclusions of the DEIR, the LRC, and their
Responsiveness Summaries are based on sound scientific principles, EPA decided as per policy to
obtain an expert peer review of the documents. The remainder of this report describes the scope
and findings of this independent peer review.
1-2
-------
1.2 Scope of the Peer Review
ERG managed every aspect of the peer review, including selecting reviewers, briefing the
reviewers on the site, and organizing the peer review meeting. The following subsections describe
what each of these tasks entailed.
1.2.1 Selecting the Reviewers
To organize a comprehensive peer review, ERG selected six independent peer reviewers
who are engineers or senior scientists with demonstrated expertise in any combination of the
following technical fields.
• River sedimentology
• Low and high resolution sediment coring
• Hydrology and water column fate and transport
• Geochemistry
• Analytical chemistry of PCBs
• Anaerobic dechlorination of PCBs
Appendix A lists the six reviewers ERG selected for the peer review meeting; brief bios
that summarize each reviewer's areas of expertise can be found in Appendix C. Recognizing that
few individuals specialize in every technical area listed above, ERG ensured that the collective
expertise of the selected peer reviewers covers the six technical areas (i.e., at least one reviewer
has expertise in analytical chemistry of PCBs, at least one reviewer has experience in river
sedimentology, and so on).
To ensure the peer review's independence, ERG considered only individuals who could
provide an objective and fair critique of EPA's work. As a result, ERG did not consider in the
reviewer selection process individuals who were associated in any way with preparing the DEIR
or the LRC or individuals associated with GE or any other specifically identified stakeholder.
1-3
-------
1.2.2 Briefing the Reviewers
Given the large volume of site-specific information in the DEIR and LRC and the fact that
none of the reviewers had extensive experience with the Hudson River PCBs site, ERG organized
a 2-day meeting prior to the actual peer review to provide the reviewers with background
information on the reports and to tour the Upper Hudson River. The purpose of the meeting was
to familiarize the reviewers with the site; the reviewers did not provide technical comments on
EPA's reports during this briefing. A copy of the minutes from this briefing can be found in
Appendix G.
To focus the reviewers' evaluations of the documents, ERG worked with EPA to develop
written guidelines for the technical review. These guidelines (commonly called a "charge") were
presented during the briefing meeting and asked the reviewers to address at least the following
topics: whether the main conclusions of the DEIR and LRC are well supported by the data; if the
data presented in these reports is sufficient for understanding fate and transport mechanisms in the
Upper Hudson River; and if additional analyses should be performed to verify certain findings of
the reports. A copy of this charge, which includes many additional topics and questions, is
included in this report as Appendix B.
In the weeks following the briefing meeting, ERG requested that the reviewers prepare
their initial evaluations of the DEIR, the LRC, and the Responsiveness Summaries. ERG
compiled these "premeeting comments," distributed them to the reviewers, and made copies
available to observers during the peer review meeting. These initial comments are included in this
report, without modification, as Appendix C. It should be noted that the premeeting comments
are preliminary in nature and some reviewers' technical findings might have changed based on
discussions during the meeting. As a result, the premeeting comments should not be considered
the reviewers' final opinions.
The peer reviewers were asked to base their premeeting comments on the written materials
distributed by ERG: the DEIR, the LRC, and the Responsiveness Summaries. Though not
1-4
-------
required for this review, some reviewers might also have researched site-specific reports they
obtained from other sources.
1.2.3 The Peer Review Meeting
The peer review meeting, which was held at the Albany Marriott Hotel in Albany, New
York, on March 16-18, 1999, was attended by the six expert reviewers and at least 30 observers.
Appendix D lists the observers who confirmed their attendance at the meeting registration desk.
The schedule of the peer review meeting generally followed the agenda, presented here as
Appendix E. As the agenda indicates, the meeting began with introductory comments both by the
designated facilitator and by the designated chair of the peer review meeting. (These and other
introductory comments are summarized below.) For the remainder of the meeting, the reviewers
discussed and debated several technical issues when answering the questions in the charge.
During the technical discussions, the reviewers provided many comments, observations, and
recommendations. The agenda included two time slots for observer comments, which are
summarized in Appendix F of this report. An ERG writer attended the meeting and prepared this
summary report.
On the first day of the meeting, Jan Connery of ERG—the designated facilitator of the peer
review—welcomed the six reviewers and the observers to the 3-day meeting. In her opening
remarks, Ms. Connery introduced Dr. Ken Reimer (a peer reviewer and the technical chair of the
meeting), stated the purpose of the peer review meeting, and identified the documents under
review. To ensure the peer review remained independent, Ms. Connery asked the reviewers to
discuss technical issues among themselves during the meeting and to consult with EPA only for
necessary clarifications. Ms. Connery explained the procedure observers should follow to make
comments. Finally, she reviewed the meeting agenda.
Following Ms. Connery's opening remarks, the peer reviewers introduced themselves,
noted their affiliations, identified their areas of expertise, and stated that they had no conflicts of
interest in conducting the peer review. Selected representatives from EPA and from EPA's
1-5
-------
contractors then introduced themselves and identified their roles in the site reassessment. To
orient the peer reviewers and observers to EPA's ongoing site reassessment efforts, Mr. Doug
Tomchuk (EPA) then gave a presentation describing the history, current status, and planned
future activities for the Hudson River PCBs site. Mr. Tomchuk also discussed the importance of
peer review in the ongoing site reassessment efforts. Mr. Tomchuk then reviewed the four major
conclusions of the DEIR and the LRC, but he did not interpret, or expand on, the assumptions
and findings documented in the reports.
As a transition into technical discussions, Dr. Ed Garvey of TAMS Consultants gave a
presentation on the main findings of the Responsiveness Summaiy for the LRC—the only report
that was not available prior to the January briefing meeting. Dr. Garvey clarified several findings
documented in this Responsiveness Summary, but he focused on several topics: the precision of
the data; the use of radioactive isotopes to "date" the sediments; approaches used to quantify the
extent of anaerobic dechlorination; the significance of wood chips in the sediment cores; and the
general findings of the appendices to the LRC.
Following Dr. Garvey's presentation, Dr. Reimer began to chair the technical discussions
of the peer review meeting. Dr. Reimer first identified several common themes among the
reviewers' premeeting comments, and then worked with the peer reviewers to answer the
questions in the charge, following the agenda. The remainder of this report summarizes the peer
reviewers' discussions and documents their major findings and recommendations.
1.3 Report Organization
The structure of this report reflects the order of questions in the charge to the reviewers:
Section 2 of this report summarizes the reviewers' discussions on specific questions regarding the
DEIR; Section 3 summarizes the discussions on specific questions regarding the LRC; Section 4
summarizes the discussions on general questions that apply to both documents; and Section 5
highlights the discussions that led to the reviewers' final recommendations. Section 6 of this
1-6
-------
report lists all references cited in the text. In these sections, the initials of the reviewers are used
to attribute technical comments and findings to the persons who made them.1
As mentioned earlier, the appendices to this report include a list of the peer reviewers
(Appendix A), the charge to the reviewers (Appendix B), the premeeting comments organized by
author (Appendix C), a list of the observers who confirmed their attendance at the meeting
registration desk (Appendix D), the meeting agenda (Appendix E), summaries of the observers'
comments (Appendix F), and minutes from the January briefing meeting for the reviewers
(Appendix G).
1 The initials of the reviewers are: RB (Dr. Reinhard Bierl), PL (Dr. Per Larsson), KM (Dr. Keith Maruya), RM
(Dr. Ron Mitchum), KR (Dr. Ken Reimer), and BR (Dr. J. Bruno Risatti).
1-7
-------
2.0 RESPONSES TO SPECIFIC QUESTIONS REGARDING THE DEIR
The peer reviewers opened their discussions by addressing the seven questions in the
charge that related to the DEIR. In answering these questions, each reviewer presented his initial
thoughts and comments, which the reviewers as a group then further discussed. At the end of the
discussion on a given question, the chair summarized the common themes expressed by the
reviewers and indicated areas where reviewers had differing opinions. A general record of the
peer reviewers' discussions on the DEIR, organized by question, follows. The reviewers' final
conclusions and recommendations for the DEIR are presented in Section 5 .0.
Note: Throughout this section, the initials of the reviewers are used to attribute comments to the
individuals who made them: RB=Dr. Reinhard Bierl, PL=Dr. Per Larsson, KM=Dr Keith
Maruya, RM=Dr. Ron Mitchum, KR=Dr. Ken Reimer, and BR=Dr. J. Bruno Risatti.
2.1 Responses to Question 1
The first question in the charge relating to the DEIR asked the reviewers: "Is the
documented PCB load, which originated from the TI Pool [the Thompson Island Pool], consistent
with a source consisting of historically deposited PCB-contaminated sediments?" The reviewers
made the following comments and observations when responding to this question:
• The Thompson Island Pool (TIP) sediments act as a source of PCBs. The six reviewers
unanimously agreed the data reported in the DEIR indicate sediments in the TIP act as a
source of PCBs to the water column in the Hudson River, but the reviewers made several
caveats in reaching this conclusion. Two reviewers, for example, noted that some of the
water column transect data presented in the DEIR provide evidence of other PCB sources,
particularly upstream sources, in addition to sediments of the TIP (RM,KM). Two
reviewers emphasized, however, that changes in PCB loads and congener profiles during
the summer low-flow conditions quite clearly indicated that the TIP sediments act as a
source of PCBs (KM,KR). To put this finding into perspective, one reviewer commented
that sediments downstream of the Thompson Island Dam (TED) likely also act as a source
of PCBs, though he still agreed that sediments in the TIP are a source as well (KM).
• Questions regarding whether "historically deposited" sediments act as a source. Though
the reviewers agreed that the TIP sediments acted as a source of PCBs, several reviewers
did not think the water column transect data were sufficient for determining the extent to
which recently deposited sediments and sediments buried at depth contributed to the PCB
2-1
-------
loading (KM,KR,PL). One reviewer noted that this distinction was particularly difficult to
resolve because the terminology is vague (i.e., exactly what should be considered as
"historically deposited" sediments?) and because he did not think the PCB congener
profiles differed enough with depth to determine conclusively which sediment layers act as
the predominant sources (KM). One reviewer modified his earlier findings by noting that
the TIP sediments clearly act as a source of PCBs, despite the uncertainties as to when
these PCBs were originally deposited (PL). In short, one reviewer thought, and others
agreed, conclusions on exactly what layers of sediments contributed to the PCB loading
were speculative (RM).
The reviewers revisited this finding towards the end of the meeting, but they did not modify
their original finding: the TIP sediments act as a source of PCBs, but the relative
contributions of recently deposited and historically deposited sediments is not known.
• Questions regarding the mechanisms by which PCBs enter the water column. Two
reviewers indicated that many physical, chemical, and biological mechanisms (e.g.,
resuspension, partitioning, bioturbation) might contribute to the source of PCBs (RB,PL).
One reviewer indicated that laboratoiy studies, rather than strict data collection and
analysis, are ultimately needed to understand these mechanisms in the Hudson River; he
also noted that the peer reviewers were not asked to determine the extent to which
different mechanisms affect PCB transport in the Upper Hudson River (PL). After brief
discussions, the reviewers agreed that the data collected for the DEIR did not determine
exactly how PCBs move from the sediments to the water column, but this shortcoming did
not modify their primary conclusion: regardless of what mechanisms are most important,
the sediments in the TIP act as a source of PCBs to the water column.
• Discussions of upstream sources of PCBs. Two reviewers discussed at length the extent
to which releases of PCBs as dense, nonaqueous phase liquids (DNAPL) from GE's
upstream facilities might act as a source in the Upper Hudson River (RM,BR). These
reviewers indicated that locating and quantifying releases from DNAPL upstream sources
would be extremely difficult. All six reviewers considered whether partitioning of PCBs in
the form of oil droplets might explain trends in the water column transect data, and one
reviewer indicated that the congener profiles of the PCBs, particularly the presence of
relatively large amounts of mono- and di- substituted PCBs, were inconsistent with an oil
droplet source of PCBs in the TIP (KM). After a lengthy discussion on upstream sources,
the reviewers agreed that DNAPL sources of PCBs at upstream locations, if any, do not
change their general response to the original question (i.e., that the sediments in the TIP
act as a source of PCBs to the water column).
• Recommendations that this conclusion be verified by analyzing additional monitoring
data. Noting that the conclusions in the DEIR are based primarily on 1 year of water
column transect data, one reviewer thought the role of TIP sediments should be further
investigated by analyzing water column monitoring data from more recent years (KM).
2-2
-------
The reviewer thought this verification was particularly important for evaluating river
conditions during the winter months, because the DEIR data that was collected during the
winter was confounded by an upstream release (the Allan Mill event) (KM). Two
reviewers thought GE might have more recent water column data available for verifying
this conclusion (KM,KR).
• An improved statistical approach would have strengthened this conclusion. Several
reviewers found the statistical analyses in the DEIR difficult to follow and overly simplistic.
One reviewer felt strongly that the DEIR should have included a clear framework that
outlined the statistical analyses in the report (PL). This reviewer clarified that he thought
the statistical tests used in the report were properly applied, but he found the analyses
difficult to follow since a clear framework was not presented.
Commenting further on the statistical approach, several reviewers thought the DEIR relied
too heavily on qualitative comparisons (e.g., similarity between diagrams of congener
profiles) in reaching its conclusions. These reviewers thought conclusions would have
been more convincing had they been better supported with quantitative, multivariate
statistical tests (RB,RM,KR). One reviewer was largely unconvinced by simple plots
showing that certain parameters might have "increased" or "decreased," without any
comments on whether changes were statistically significant (KR). Another reviewer
thought EPA's contractors should have adopted statistical approaches to identify outliers
among the sampling data (RM).
• Other comments regarding the treatment of analytical data. When commenting on the
role of TIP sediments, several reviewers offered general comments on the presentation of
data in the DEIR One reviewer noted that the DEIR included very little information,
quantitative or qualitative, on analytical variability of the PCB measurements (KR). This
reviewer indicated that EPA should have more prominently acknowledged in the DEIR the
analytical variability of the water column transect data and sediment coring data. Noting
that the analytical laboratory had quality assurance criteria that automatically excluded
from consideration any samples that did not meet certain precision criteria, one reviewer
thought the report should have clearly stated these criteria and the number of samples that
were excluded as a result (RM).
• Comments on data quality. Since the quality of the water column transect and sediment
coring data were relevant to every question in the charge, the reviewers decided to state
their general findings on data quality when responding to Question 1. Two reviewers
commented that the quality of the monitoring data, as a whole, appeared to be acceptable
(KM,RM). Another reviewer agreed with this general statement, but he again suggested
that the DEIR should have clearly documented measurement precision for each PCB
congener (KR).
2-3
-------
2.2 Responses to Question 2
The second question asked the reviewers: "Are the two-phase and three-phase partitioning
coefficients, derived in the DEIR, appropriate and do they properly address the physical
parameters of the system (e.g., temperature)?" The reviewers made the following comments and
observations in response:
• Comments on the two-phase partition coefficients. The reviewers unanimously agreed that
the derivation and calculation of two-phase partition coefficients, including their
corrections for temperature, were scientifically sound. As a qualitative check on the
calculated values, one reviewer noted that the calculated two-phase partition coefficients
generally increased with reported values of octanol-water partition coefficients, as is to be
expected (KM). Another reviewer indicated that estimated partition coefficients for many
congeners had widely variable values (RB), but a reviewer indicated that such variability is
typical for deriving PCB partition coefficients from field measurements (KM).
• Comments on the three-phase partition coefficients. Several reviewers thought the water
column transect data were insufficient for calculating reliable three-phase partition
coefficients (RB,KM,RM). As evidence of this finding, one reviewer mentioned that three-
phase partition coefficients for some congeners appeared to have unrealistic values, when
compared to the coefficients for other congeners (KM). This reviewer thought the three-
phase partition coefficients might include errors of an order of magnitude or greater and
should not have been reported to two decimal places, as was done in the DEIR. The
reviewers did not take exception with how mathematical expressions for the three-phase
partition coefficients were derived (RM), but they thought additional data that characterize
concentrations of dissolved organic carbon (DOC), including colloids, in the water column
are needed for more accurate estimates of the three-phase partition coefficients (RB).
• Use ofpartition coefficients in future modeling studies. Two reviewers thought the
partition coefficients should be used to develop empirical models of PCB transport
mechanisms (RB,PL). These reviewers indicated that such modeling could quantify how
temperature and other relevant parameters affect partitioning of PCBs in the Hudson River,
which, in turn, would be useful for understanding underlying mechanisms of PCB transport
(RB,PL).
• Consideration of nonequilibrium partitioning and other "compartments "for equilibrium.
Noting that sorption and desorption kinetics affect partitioning of PCBs in the water
column, one reviewer suggested that nonequilibrium effects might need to be considered in
future modeling exercises (RB); other reviewers did not comment further on this topic.
2-4
-------
Regarding other "compartments" to consider for partitioning, one reviewer noted several
times that volatilization and photolysis of PCBs should have been addressed in the reports
(RM). Other reviewers, however, were not convinced of the need to address these
compartments in the analysis of equilibrium partitioning: one reviewer explained that
photolysis is a nonequilibrium process (KR) and another reviewer acknowledged that EPA
should eventually address volatilization in its reassessment, but not necessarily in these
partitioning models (PL).
2.3 Responses to Question 3
The reviewers discussed at length the third question on the DEIR: "Are the conceptual
models based on the transect sampling consistent with the data?" A summary of these discussions
follows:
• General agreement that the conceptual models were supported by the data and illustrated
important aspects of PCB transport in the Hudson River. Though the reviewers expressed
several concerns about the conceptual models used to interpret the water column transect
sampling data (these concerns are summarized below), they agreed that the models were
generally consistent with the data and provided useful insight into PCB transport in the
Hudson River. One reviewer felt, and other reviewers agreed, that the conceptual models
presented in the Responsiveness Summary offered a much more defensible account of the
water column transect data than did the models presented in the DEIR (KR).
Some reviewers identified what they considered to be particularly useful findings of the
conceptual models. Two reviewers, for example, indicated that the conceptual models
helped depict seasonal changes in PCB levels in the water column (KM,BR). They noted
that the models clearly illustrated how PCBs in the water column, particularly those bound
to suspended solids, increased during high-flow events and how levels of lower molecular
weight PCBs tended to decrease with downstream distance during the wanner summer
months, whether by volatilization, photolysis, or degradation. Another reviewer indicated
that the models were useful for illustrating congener-specific trends (BR).
• Models should have been supported by more sophisticated statistical analyses. Almost
every reviewer indicated that a more rigorous statistical analysis would have provided more
compelling evidence of the models' findings than did the simple visual comparisons of
congener profiles in the DEIR. One reviewer noted that he had conducted a principal
component analysis (PCA) on a subset of the water column transect data to verify the
conclusions drawn in the conceptual models (KR). This reviewer thought PCA or similar
multivariate statistical analyses should have been conducted to quantify notable, but
possibly subtle, trends among the large volume of monitoring data. Several reviewers
2-5
-------
agreed and indicated that multivariate statistics would have particular utility in interpreting
the large volume of congener-specific data (RM,BR,KR).
• Concern about corrections made to the river flow data. Several reviewers thought the
conceptual models were consistent with the data, but were concerned about corrections
that were made to the river flow data originally presented in the DEIR (RB,RM,BR).
More specifically, river flow data for some stations presented in the Responsiveness
Summaries were roughly 40 percent higher than the corresponding data presented in the
DEIR. One reviewer found it difficult to verify whether this correction was made correctly
and noted that the magnitude of the flow correction has a notable impact on the calculated
PCB loads to the water column (RB). Another reviewer, however, explained that the
magnitude of the flow correction has no bearing on the relative changes in PCB
concentrations from one sampling station to the next (KM). This reviewer thought the
conceptual models of the water column transect data provided insight into PCB transport,
regardless of whether the flow corrections were correctly or incorrectly applied.
• Consideration of parameters other than PCB concentrations in the conceptual models.
Several reviewers thought applying the conceptual models to pollutants other than PCBs
might lead to a greater understanding of fate and transport of chemicals in the Upper
Hudson River. For instance, one reviewer thought the models should be applied to
measured levels of metals and chlorophyll, if such data are available (RB). In support of
this recommendation, another reviewer noted that the U.S. Geological Survey has used
metals and other contaminants to gain greater insight into physical processes in other rivers
(BR). Another reviewer indicated that examining levels of chlorophyll might be
worthwhile because in-situ production might be an important factor to consider in the
relatively quiescent TIP (KM). Though these three reviewers recommended evaluating
data trends and patterns for other parameters as part of the ongoing reassessment efforts
on the Hudson River, none of these reviewers listed this recommendation among their
major findings for the peer review meeting.
• Miscellaneous comments. When discussing the conceptual models, the reviewers made
several comments that do not fall under the categories listed above. One reviewer, for
example, noted that the database of sampling results was extremely difficult, and almost
impossible, to use (KR). Further, some reviewers thought the conceptual models should
have more prominently acknowledged the analytical variability of the laboratory
measurements (KR) and the difficulties associated with quantifying congeners of lower
PCB homologues in environmental samples (BR). Another reviewer thought the term
"model" applies more to a mathematical construct that has predictive capabilities, and that
the "conceptual models" in the DEIR were more simply "conceptual reasoning" (PL).
2-6
-------
2.4 Responses to Question 4
The fourth question in the charge asked the reviewers: "Does the sampling at the TI Dam-
West location impact EPA's conclusion that the sediments of the TI Pool are the major source of
PCBs to the freshwater Hudson during low flow conditions considering the analytical corrections
made to GE's PCB data? What are the other implications of finding higher concentrations along
the shoreline than in the center channel?" The reviewers discussed these two questions at length
and offered several general responses:
• Cancellation of sampling and analytical biases. The reviewers understood that
corrections made for a sampling bias altered the quantitative findings of the DEER, but they
did not think these corrections affected the general conclusion that the TIP sediments are a
primary source of PCBs to the Hudson River (RB,PL,RM,KR). More specifically, a
reviewer noted that the corrections made for the sampling bias were almost entirely offset
by other corrections made to laboratory analytical data (KM). Assuming both corrections
were made correctly, this reviewer thought the sampling bias had little impact on the
DEIR's findings. Two reviewers commented that the algorithm, or "box model," EPA
used to derive the corrections appeared to be valid (PL,KM).
• Comments on the clarity of the question. Several reviewers thought the first part of
Question 4 did not clearly indicate which water column sampling data was corrected and
how this correction was made (KM,BR,KR). At the reviewers' request, EPA's contractors
identified the three sampling locations in the vicinity of the TID—GE's "west wing wall"
location, GE's "center channel" location, and EPA's location about V* mile upstream from
the dam—and explained the sampling bias and the corresponding data corrections. An
observer offered to present additional data to clarify this issue, but the meeting facilitator
noted that presenting such information would be more appropriate during the observer
comments.
• Other comments regarding potential sampling biases. One reviewer thought the use of a
different sampling technology, such as one that pumps water from different depths of the
river, might have provided a more accurate account of concentrations of PCBs in the water
column (PL). Nonetheless, this reviewer believed EPA's corrections for the sampling bias
were appropriate.
• Implications of PCB concentrations in near-shore areas being higher than those in the
center channel. The reviewers raised and discussed several implications of the spatial
variations of PCB concentrations: they agreed that the greatest implications pertain to
calculating PCB load to the water column and estimating the inventory of PCBs in the
sediments. Regarding PCB loads, one reviewer explained, and the others agreed, that load
2-7
-------
estimates would be biased if water column samples were collected in a channel with
artificially high or low PCB concentrations (KM). More specifically, they thought
uncorrected sampling results from a near-shore "hot spot" location might lead to a different
estimate of PCB loads than sampling results from the center channel. Regarding PCB
inventories, two reviewers indicated that the inventory might be understated if relatively
high concentrations of PCBs in near-shore sediments have not been adequately
characterized (KM,RM). The reviewers agreed to revisit the issue of PCB inventories
during their discussions on the LRC, summarized in Section 3 of this report.
The reviewers briefly discussed several other implications of the spatial variations in PCB
concentrations. These implications include, but are not limited to, a hypothesis that PCB
loads to the water column might actually be lower than expected if near-shore
contaminated sediments are not submerged during seasonal low-flow conditions (KM) and
an observation that local river flow patterns, which affect sediment deposition and
resuspension, also change considerably from the center channel to the shoreline (BR).
After answering the specific questions in the charge, the reviewers revisited their response
to this question later in the meeting. One reviewer expanded on his earlier discussions
about the implication of higher PCB concentrations in near-shore sediments: he noted that
an undersampling of near-shore sediments might have biased the geostatistical analysis of
the 1994 PCB inventory to lower levels (KM). He explained that, in cases where near-
shore cores were not collected, the kriging and polygonal declustering analyses would use
PCB concentrations measured in deeper sections of the river to estimate PCB
concentrations in near-shore sediments. Other reviewers did not comment on this
observation and indicated that their earlier summary statements were sufficient.
2.5 Responses to Question 5
The reviewers answered the fifth question: "Are the geostatistical techniques (polygonal
declustering and kriging) correctly applied?" as follows:
• General agreement that the geostatistical techniques were correctly applied. The
reviewers agreed that the findings from the geostatistical analyses gave a reasonable
approximation of the PCB inventory and that EPA's contractors appeared to have applied
the techniques correctly. Since most of the reviewers did not have extensive experience
using these geostatistical techniques, however, they did not comment in detail on this topic.
• Concerns about the selected geostatistical techniques. Though he agreed that EPA's
contractors had applied kriging and polygonal declustering analyses correctly, one reviewer
thought the spatial heterogeneity of PCBs in the sediments necessitated the use of more
sophisticated analyses of the PCB inventory (RB). This reviewer recommended nonlinear
statistical techniques for this purpose, but he did not specify a particular test or method that
2-8
-------
would be best suited for such analyses.2 He also recommended gathering more data to
characterize the spatial variations in PCB concentrations more thoroughly, but another
reviewer commented that the results of EPA's side-scan sonar study already offered insight
into the spatial heterogeneity of the river sediments (PL).
• Concerns about presenting inventory estimates without addressing uncertainty. Though
the reviewers thought the geostatistical analyses were valid, one reviewer noted, and
several reviewers agreed, that EPA's reports did not acknowledge the uncertainty
associated with estimating PCB inventories from a finite number of sediment cores (KR).
Given the uncertainty in making this estimate, this reviewer recommended the reports
acknowledge that the calculated PCB inventory is only an estimate of the actual inventory.
He also suggested that EPA consider presenting a range of inventory estimates, rather than
presenting a single value.
2.6 Responses to Question 6
The sixth question on the charge asked the reviewers: "Are the methods applied in the
DEIR (change in molecular weight (MW) and evaluating concentrations of BZ#s 1, 4, 8, 10 and
19 (MDPR)) appropriate standards for determining extent of dechlorination? Are there any
significant problems with this approach, or more appropriate approaches?" The reviewers
discussed these two questions at length and offered several general responses:
• Agreement that the MDPR is an approximate measure of the extent of anaerobic
dechlorination. The reviewers agreed the MDPR provides a useful characterization of
dechlorination, though they identified several potential shortcomings with the MDPR.
These shortcomings relate to the fact that the MDPR is calculated from concentrations of
several PCB congeners from the lower homologues. Noting that the lower homologues
are the most difficult to measure, one reviewer thought the MDPR might be biased by the
analytical method (BR). Furthermore, because lower homologue PCBs are more likely to
be removed from sediments than higher homologue PCBs (whether by pore water
diffusion, aerobic degradation, or some other mechanism), several reviewers indicated that
the sediment coring data do not characterize the amounts of dechlorination products that
have actually been formed (KM,BR,KR). The reviewers noted that the DEIR did
acknowledge these potential shortcomings of the MDPR.
2 When reviewing the draft peer review report, this reviewer indicated that EPA could have used "disjunctive
kriging" or "kriging in terms of projections." The reviewer indicated that these more complex approaches may help gain
accuracy in non-linear estimators. The reviewer recommended that EPA consult the following software library: "Glayton
V. Deutsch and Andre G. Journel: GSLIB: Geostatistical Software Library and User's Guide. Oxford University Press,
1997.
2-9
-------
The reviewers also discussed the implications of the MDPR being calculated from
concentrations of both "near final" and "terminal" dechlorination products. The reviewers
thought this calculation was defensible, provided that ortho dechlorination of PCBs in the
Hudson River does not occur (as is stated in the DEIR). During this discussion, two
reviewers indicated that their own research has observed ortho dechlorination in sediments
(BR,KR); however, another reviewer noted that several research projects on Hudson River
sediments have not provided much evidence of ortho dechlorination (KM). The reviewers
did not comment further on this topic.
• Alternative measures for quantifying the extent of dechlorination. The reviewers
identified alternative measures for characterizing dechlorination in the Hudson River
sediments, but they were not certain whether these alternative measures would provide any
greater insight into the issue. One reviewer suggested that EPA could have examined
"parent-daughter" dechlorination pairs to characterize overall levels of dechlorination, but
this reviewer noted that this approach would suffer from some of the same shortcomings as
the MDPR (KM). Another reviewer suggested that EPA quantify dechlorination strictly
from data trends for heavier PCB congeners, which are not as difficult to measure and are
not as likely to partition to the water column (BR). When discussing these alternatives, a
reviewer asked whether EPA's contractors had considered variations of the MDPR to
estimate the extent of dechlorination. As a point of clarification, EPA's contractor
indicated that the Responsiveness Summary for the LRC contains such an analysis.
• Other comments on estimating the extent of dechlorination. Two reviewers offered other
insights when discussing the appropriateness of the MDPR. One reviewer did not think the
DEIR acknowledged the uncertainty associated with estimating the extent of
dechlorination: he thought presenting point estimates of dechlorination ratios without
including error bounds or appropriate caveats did not reflect the associated uncertainties
(KR). Another reviewer emphasized that dechlorination has no bearing on the total mass
of PCBs in the river sediments, since dechlorination merely transforms PCBs and does not
remove them entirely from the system (PL). This reviewer thought the transformation of
PCBs was notable since dechlorination products are generally more mobile than the
higher-chlorinated PCBs (PL).
2.7 Responses to Question 7
The reviewers discussed at length the final question in the charge related to the DEIR,
which asked: "The DEIR finds that the degree of anaerobic dechlorination is primarily a function
of original concentration rather than time, and accordingly that there is not significant predictable
dechlorination in sediments containing less than approximately 30 mg/kg PCB. Is this
reasonable?" The reviewers addressed the following topics when answering this question:
2-10
-------
• Comments on the wording of the question. Two reviewers commented that this question in
the charge was open to several interpretations. One reviewer, for instance, explained that
he addressed simply whether dechlorination could occur at PCB concentrations below
30 mg/kg (30 parts per million [ppm]) while other reviewers might have answered a
different question: whether predictable dechlorination occurs at these levels (KM).
Another reviewer noted that people might have differing opinions on what constitutes
significant predictable dechlorination (KR). Despite the ambiguities in the question, the
reviewers focused their discussions on whether thresholds for anaerobic dechlorination, in
a general sense, are scientifically plausible and whether the 30 ppm threshold reported for
the Hudson River is supported by the available data. The following bullets summarize
these, and other, discussions relevant to Question 7.
• Discussion on whether concentration thresholds for dechlorination are plausible. The
reviewers talked at length about whether concentration thresholds for dechlorination are
consistent with fundamental physical and biological mechanisms. One reviewer noted that
he has observed concentration thresholds for dechlorination and other biological
phenomena in his own research (BR), but another reviewer indicated that other studies
have observed dechlorination occurring at levels considerably lower than 30 ppm (KR).
Two reviewers were not surprised by this discrepancy, noting that conditions that affect
biological activity in river sediments are different from those in controlled laboratory
conditions (BR) and real-life river conditions often vary notably from river to river (PL).
Several reviewers confirmed these comments by identifying the many parameters affecting
biological processes (e.g., temperature, nutrients, inhibitors, organic carbon) that might
have unique levels in the Upper Hudson River.
The reviewers then identified fundamental biological and physical processes that might
explain thresholds. One reviewer commented that, under certain conditions (e.g., severely
limited diffusion or unavailable nutrients), dechlorination kinetics can conceivably become
imperceptibly slow, so as to give the appearance of a concentration threshold for
dechlorination (KM). Another reviewer agreed, but had difficulty believing the findings in
the DEIR because the report failed to offer a mechanistic explanation for the apparent
concentration threshold (PL). The remainder of the reviewers' discussion on thresholds
focused specifically on the likelihood that they apply for dechlorination in the Upper
Hudson River.
• Discussion on whether the DEIR and LRC data support a concentration threshold for
dechlorination. The reviewers unanimously agreed that the sediment coring data from the
DEIR and the LRC do not support the reported 30 ppm threshold for dechlorination, and
one reviewer went further in stating that the data do not support a threshold occurring at
any concentration (RM). The reviewers gave several reasons for rejecting this finding.
Noting that a large subset of the sediment cores were not considered in the dechlorination
calculations, for example, one reviewer wondered whether this selective use of data might
have masked more general trends (KM). Another reviewer did not think enough samples
2-11
-------
with PCB concentrations lower than 30 ppm were available to reach a firm conclusion tha
dechlorination does not occur below this threshold (KR).
Other reviewers offered different perspectives on this topic: one reviewer thought EPA's
sediment coring data supported a predictive empirical relationship between the extent of
dechlorination and PCB concentrations greater than 30 ppm; however, he did not think thi
data implied that dechlorination ceases at lower concentrations (PL). Agreeing with this
sentiment, another reviewer emphasized that no conclusion should be drawn about
dechlorination in sediments with PCB concentrations below 30 ppm (KR). Finally, one
other reviewer highlighted some exceptions to the basic trend reported in the DEIR, for
example, a small subset of sediment cores with relatively high PCB concentrations had ver
little evidence of dechlorination (BR). Later in the meeting, this same reviewer listed three
sediment cores with PCB concentrations lower than 30 ppm that showed evidence of
dechlorination. After thoroughly reviewing these arguments, the reviewers all agreed the
data provided in the DEIR and LRC do not support the 30 ppm dechlorination threshold.
Based on this finding, one reviewer thought a summary statement in the DEIR ("PCBs in
sediments with less than 30 ppm are largely left unaffected by the dechlorination process")
should be qualified (KR).
• Agreement that dechlorination is predictable at "higher " PCB concentrations. After
answering the specific questions in the charge, the reviewers revisited Question 7 to furthe:
debate whether the extent of dechlorination is predictable. Several reviewers commented
that the figures in the DEIR clearly demonstrate a relationship between the extent of
dechlorination and PCB concentration, at least among the sediment cores with relatively
high PCB levels (PL,BR,KR). The reviewers did not specify the lowest PCB concentratioi
at which the extent of dechlorination appears to be predictable, but one reviewer did not
think predictable dechlorination occurred at levels near 30 ppm (KR). Based on these
discussions, the reviewers unanimously agreed with the summary statement: "There is
predictability of dechlorination at higher PCB concentrations, but this should not be taken
as evidence of lack of dechlorination at lower concentrations."
• Comments on whether dechlorination might be a function of time (i.e., age of sediments).
The reviewers briefly discussed the possibility that the extent of PCB dechlorination varies
as a function of time. They indicated that available data provide conflicting answers to this
question: some studies by other researchers have reported considerable dechlorination in
freshly deposited sediments (KM), yet many of the Hudson River cores showed little
evidence of dechlorination in some of the older sediments (BR). Focusing on the Hudson
River sediments, another reviewer commented that the coring data clearly show that the
extent of dechlorination is more dependent on PCB concentration than it is on time (KR).
This reviewer cautioned, however, that the greater dependence on PCB concentration does
not imply that dechlorination is totally independent of time, as he documented in his
premeeting comments. The reviewers did not discuss this topic further.
2-12
-------
Comments on the presentation of data. One reviewer thought presenting dechlorination
data on a logarithmic scale, as was done in the DEIR, was inconsistent with the
mathematical derivation of the MDPR (RM). This reviewer asked EPA's contractors to
clarify several issues related to the presentation of the data, after which he still concluded
there was no scientific basis for using logarithmic scales to depict the dechlorination
results. He thought EPA's contractors chose to use logarithmic scales simply to fit the
data to a trend.
2-13
-------
3.0 RESPONSES TO SPECIFIC QUESTIONS REGARDING THE LRC
The peer reviewers continued their discussions by addressing the seven questions in the
charge that related to the LRC. The scientific chair followed the same format as used in the
previous discussion about the DEIR in facilitating these discussions: individual reviewers were
asked to present initial thoughts on the questions; the reviewers as a group then further discussed
and debated these initial comments; and finally the chair summarized the common themes
expressed by the reviewers and indicated areas where reviewers had differing opinions. A general
record of the peer reviewers' discussions on the LRC, organized by question, follows. The
reviewers' final conclusions and recommendations for the LRC are presented in Section 5.0.
Note: Throughout this section, the initials of the reviewers are used to attribute comments to the
individuals who made them: RB=Dr. Reinhard Bierl, PL=Dr. Per Larsson, KM=Dr. Keith
Maruya, RM=Dr. Ron Mitchum, KR=Dr. Ken Reimer, and BR=Dr. J. Bruno Risatti.
3.1 Responses to Question 1
As the charge in Appendix B shows, the first question specific to the LRC asked: "In the
LRC, EPA compared sediment data from cores taken in 1977, 1984 and 1994, which had the
PCB analysis conducted by different laboratory methods. How valid are the methods used to
establish a consistent basis for comparison?" The reviewers' comments and main findings on this
topic follow:
• Comments on comparing cores collected in 1984 to those collected in 1994. The
reviewers unanimously agreed that EPA's contractors used a reasonable method to
compare sediment coring results between 1984 and 1994. Several reviewers thought no
other defensible methods could have been used, given the difficulties laboratories had
measuring levels of mono- and di-substituted PCBs (PL,KM,BR). Individual reviewers
made several other observations regarding the data comparisons. For instance, one
reviewer felt confident in the data comparison, partly because the majority of PCB releases
to the Hudson River were reportedly Aroclor 1242, which likely produced consistent peaks
among the chromatograms; he said he would have been less confident in comparisons
involving complex mixtures of Aroclors (KM). Two reviewers thought the comparison
between the 1984 and 1994 data had greater uncertainty than the LRC acknowledged. As
a result, they thought the comparison should have been presented as an approximation of a
trend, rather than as a concrete estimate (RM,KM). Finally, one reviewer added, and
3-1
-------
several other reviewers agreed, that the data comparison would be better supported by a
detailed review of the 1984 sampling effort (e.g., analyzing archived samples and extracts,
inspecting chromatograms) (RB).
• Comments on comparing cores collected in 1977 to those collected in 1994. The
reviewers had no confidence in quantitative comparisons between the 1977 and 1994
sediment coring data sets. Moreover, two reviewers thought the LRC did not describe the
1977 sampling effort in detail nor did it propose a method for comparing the 1977 and
1994 data (KM,KR). Based on the lack of confidence in the quality of the 1977 data, one
reviewer concluded that any comparison between the 1977 and 1994 data sets would be
speculative (KM). Two reviewers indicated they would be more confident in the 1977 data
set if samples could be reanalyzed and chromatograms examined (RB,KM).
• Discussion on the quality of the 1977 and 1984 data. While reviewing the methods used
to compare the data sets, the reviewers expressed several concerns about data quality for
the previous coring studies. The main concern was that very little information was
provided on the extraction procedures, precision estimates, use of internal standards, and
other quality assurance measures that were used in the 1977 and 1984 sampling and
analytical programs (RB,PL,RM). On the other hand, some reviewers offered reasons to
believe the data quality from the past sampling efforts, particularly from 1984, was
acceptable. Based on his experience with EPA's oversight of laboratory quality assurance
in the 1980s, for example, one reviewer was satisfied that the 1984 data were likely of a
reasonable quality, though he was less confident in the quality of the 1977 data (RM).
Agreeing with this sentiment, another reviewer noted that he did not think analytical
variability for PCB measurements had changed dramatically between 1984 and 1994 (PL).
As noted above, several reviewers suggested that the best way to gain greater confidence
in the past data is by carefully reviewing chromatograms and reanalyzing archived samples
or extracts, if such information is available.
3.2 Responses to Question 2
The reviewers discussed at length the second question in the charge on the LRC, which
asked: "In the Upper Hudson River system, it has been well established that there is significant
lateral heterogeneity in sediment concentrations. While it was attempted to reoccupy previous
locations, some uncertainty is added with respect to the actual sampling location. While the
statistical techniques help compensate for this, how does the sediment heterogeneity affect the
comparison of cores from two different years? Given the spatial variability, is the finding that
there is loss from most of the locations supported by the data?" The reviewers addressed the
following topics when answering this question:
3-2
-------
Evaluation of the techniques used to compare sediment concentrations from two different
years. The reviewers agreed the combined "point-to-point" and "area-to-area" comparison
was a reasonable approach for examining changes in PCB mass loading between two
different years, but they had several comments on how these approaches were applied.
First, given the heterogeneity of the sediments, several reviewers indicated that the "area-
to-area" comparisons presented in the Responsiveness Summary are much more defensible
than the "point-to-point" comparisons originally reported in the LRC (KM,BR). Second,
several reviewers emphasized that statistical techniques alone cannot compensate for
heterogeneous sediments, as implied by the question in the charge; these reviewers
explained that only larger sample sizes can effectively reduce uncertainty in the sediment
core comparisons (PL,KM,BR). Third, two reviewers noted that EPA used acoustical
techniques to characterize sediment properties and heterogeneity in areas where cores had
not been collected—an issue that was discussed in greater detail later in the meeting (see
Section 3 .7 of this report) (PL,KM). Finally, one reviewer thought the LRC should have
included more information on the factors that contribute to the spatial heterogeneity of
PCB concentrations (e.g., is the heterogeneity caused by historical deposition areas,
differing sediment characteristics, or other factors?) (RB).
Comments on the reported loss of PCBs from sediments in most sampling locations. The
reviewers agreed the sediment coring data indicate a general trend of PCB loss from
sediments in most locations. Several reviewers added, however, that estimated amounts of
PCB loss should be interpreted with caution due to the uncertainty inherent in comparing
sediment cores collected in different years (RB,PL,KM). Another reviewer noted that the
analytical variability in the measurements alone complicates efforts to quantify PCB losses
(RM). The reviewers discussed the implication of uncertainty further when answering
Question 3, as summarized in the next section.
3.3 Responses to Question 3
The reviewers continued their discussion on the estimates of PCB loss from river sediments
when answering the third question in the charge: "What is the impact of the difference between
replicate samples in the 1994 sampling effort (36 percent average variability) on the finding that
there was a 40 percent loss of PCB inventory from the highly contaminated sediments in the TI
Pool?" The reviewers' responses to this question focused on the following issues:
• Recommendations for acknowledging the uncertainty in the reported PCB loss. The
reviewers unanimously recommended that EPA's reports not present discrete estimates of
the PCB inventory loss without caveats about the uncertainty associated with the
calculation. More specifically, one reviewer suggested that point estimates of PCB loss
3-3
-------
could be bracketed by estimates of uncertainty (KM), while other reviewers thought the
loss estimate should simply be reported as a range of values without a point estimate (RB).
The reviewers noted that the estimate of 40 percent loss of PCBs has considerable
uncertainty, but they added that the sediment coring data do support a loss of PCBs from
most areas of the TIP (see response to Question 3, above).
• Comments on how the analytical variability affects the PCB loss estimates. To focus their
discussions on analytical variability, the reviewers asked EPA's contractors to clarify how
they calculated relative percent difference (RPD) and whether the RPD accounts for
sampling variability. The contractors responded that they calculated RPDs from "true
splits," which, in theory, strictly characterize analytical variability. Some reviewers were
surprised that the average analytical variability was as high as 36 percent (RM,BR,KR):
one reviewer noted that his laboratory routinely generates data with better precision (KR).
Other reviewers found it difficult to comment on analytical variability, because little
information was provided on the RPDs for the 1984 data set (KM,RM).
Despite these concerns about data variability, the reviewers agreed that the sediment cores
provide a basis for evaluating changes in PCB inventory from 1984 to 1994. As noted
above, however, the reviewers emphasized that quantitative comparisons are highly
uncertain. Citing a figure in the LRC that presented congener-specific RPDs, one reviewer
noted that the analytical variability among the 1994 data seemed to be random and not
systematic (KM). This reviewer felt more comfortable with the PCB inventory
comparisons due to the apparent absence of a systematic bias in the analytical data, but he
emphasized that the reports should more prominently acknowledge the uncertainty
associated with the estimated inventory loss.
3.4 Responses to Question 4
The reviewers then discussed the fourth question in the charge: "In the LRC, it was found
that Hot Spot 28 contained much more mass than previous estimates. Is the conclusion that this
'gain' is primarily due to incomplete characterization in 1977 valid?" A summary of their
responses follows:
• Agreement that the apparent gain in PCB mass for Hot Spot 28 was not a validfinding.
The reviewers unanimously agreed that the apparent increase in PCB mass for Hot Spot 28
did not represent a true gain in mass, but merely resulted from the 1977 coring study failing
to characterize Hot Spot 28 completely. One reviewer offered two reasons for questioning
the validity of the 1977 mass loading estimates (KM). First, noting that the 1977 study did
not sample an area of Hot Spot 28 that the 1994 study found to have relatively high PCB
concentrations, this reviewer indicated that the 1977 study might have underestimated the
spatial extent, and hence the mass loading, of the hot spot. Second, the reviewer explained
3-4
-------
that the cores used in the 1977 study were not long enough to characterize the depth of the
hot spot—a shortcoming that also caused an underestimation of the mass loading. For
these and other reasons, the reviewers concluded that the quality of the 1977 data was
unknown, but they thought the 1994 characterization of Hot Spot 28 seemed adequate.
• Lack of other logical explanations for the apparent gain in PCBs. Several reviewers
could not envision any other logical reason (except for the incomplete characterization
during the 1977 study) that could adequately explain the considerable increase in PCB
mass in just one hot spot, while the PCB mass in other hot spots apparently decreased
(RB,KM,RM).
3.5 Responses to Question 5
Continuing their discussion on losses in PCB inventory, the reviewers answered the fifth
question in the charge, which asked: "Does the data set and its interpretation support the
conclusion that significant losses have occurred from hot spots below TI Dam?" The reviewers
addressed the following issues in their response:
• Comments on the wording of the question. Two reviewers thought this question was open
to several interpretations, due to ambiguity in the term, "significant losses" (PL,KR). For
instance, one reviewer indicated that he could answer whether a loss of PCBs is significant
from the perspective of downstream ecosystems, from the perspective of total inventory, or
from the perspective of statistics (PL). This reviewer explained further that a 1 percent
loss of PCBs from the sediments might be significant in terms of the implication on
downstream ecosystems, but such a loss might not be significant when compared to the
total PCB inventory in the sediments. Given these concerns, the reviewers decided to
answer a more direct question. "Does the data set support the conclusion that losses have
occurred from hot spots below the TID?" Responses to this question, which omits the
word significant, are summarized below.
• Agreement that PCB losses seem reasonable, but the amounts are difficult to quantify.
The reviewers unanimously agreed that the data presented in the LRC support the
conclusion that sediments downstream from the TID have lost PCBs, but they thought
estimates of the actual mass loss would be difficult, if not impossible, to quantify. The
reviewers thought PCB losses seemed reasonable based on data reported in the LRC.
noting that approximately 50 percent of the PCB inventory in the downstream hot spots
appeared to lie within the top 9 inches of sediments, one reviewer thought it was
conceivable that losses could have occurred (KM). Another reviewer agreed, stating that
PCBs in the top 9 inches of sediment are probably available for transport to the water
column in some manner, though the exact mechanism might not be known (KR). Yet
-------
another reviewer argued against this reasoning, because he thought sediment cores
collected by GE provided evidence that maximum PCB levels typically occurred at depths
between S and 9 inches (BR). Nonetheless, this reviewer agreed with the basic summary
finding: PCB losses have likely occurred from sediments downstream from the TID. (The
reviewers decided to address the issue of how PCB concentrations vary with sediment
depth when responding to the sixth question in the charge [see Section 3.6].)
The reviewers were concerned about attempts to quantify PCB losses from this stretch of
the river since many of the hot spots were characterized only by the 1977 and 1994
sampling efforts. As summarized in Section 3.4, the reviewers questioned the quality of
the data from the 1977 sampling.
• Comments on the mechanisms contributing to PCB losses. Two reviewers indicated
mechanistic explanations for the loss of PCBs from sediments downstream of the TID
(RB,PL). They agreed that particle transport (sediment resuspension) could have
accounted for the PCB losses in this stretch of the Hudson River, but they were skeptical
that either pore water diffusion or bioturbation were the primary mechanism of PCB
transport to the water column (RB,PL). One of these reviewers recommended that future
work on the site focus more on mechanistic explanations for observed data trends (RB).
3.6 Responses to Question 6
The reviewers debated several issues pertaining to the sixth question in the charge: "The
LRC found that the historically contaminated sediments in the TI Pool were not universally being
buried and sequestered from the environment. How much confidence would you place in the
LRC evidence against widespread burial?" A summary of their discussion follows:
• Comments on the wording of the question. Several reviewers thought Question 6 was
open to several interpretations, largely due to the terms "widespread" and "burial." More
specifically, one reviewer noted that different people might have different conceptions of
what "widespread" actually means (KM). On a similar note, another reviewer indicated
that he had difficulty answering this question because he was not sure how to interpret
"burial" (i.e., exactly how many inches of sediment must deposit for "burial" to occur?)
(RB). Due to these concerns, the reviewers carefully worded their responses to the
question, which are summarized below.
• Agreement that widespread burial of PCBs is not occuring. The reviewers offered many
different opinions on whether PCBs are being buried in the TIP, after which they agreed
that the data in the LRC suggest that widespread burial does not appear to occur. One
reviewer based this finding on how PCB concentrations varied with depth in the low
3-6
-------
resolution sediment cores: for the hot spots in the TIP, he calculated that approximately
60 percent of the PCB inventory lies within the upper 9 inches of sediments (KM). This
reviewer used this evidence to argue against burial of PCBs to depths of 9 inches or
deeper. Another reviewer added that the water column transect data are inconsistent with
widespread burial (RB). He explained that the water column data from the DEIR, which
indicated that PCBs enter the water column from the TIP sediments (see Section 2.1),
suggest that PCBs likely remain in the upper layers of the sediments and that widespread
burial probably does not occur. Yet another reviewer agreed with both of these arguments
and concluded that the weight of the evidence from EPA's reports is against deep burial of
PCBs (KR).
During these discussions, one reviewer stressed that PCBs are likely being buried in certain
parts of the Upper Hudson River (BR). Other reviewers agreed with this statement, but
noted that "deep" burial does not appear to be widespread (PL,KM). All six reviewers
eventually agreed that burial might occur in some places, but it does not appear to be
widespread.
After answering the specific questions in the charge, the reviewers revisited Question 6,
focusing primarily on whether "deep" burial of PCBs occurs. One reviewer explained that
the depth of burial can have significant implications on the bioavailability of PCBs (KM).
Another reviewer agreed, but noted that future modeling exercises will have to determine
whether or not the PCBs are, in fact, bioavailable (KR). The reviewers then discussed
basic data trends of the LRC, as summarized in one reviewer's premeeting comments
(KM), and eventually agreed with their original summary statement: "There does not
appear to be widespread burial."
Caveats on drawing conclusions from data collected over a 10-year period Though he
agreed that widespread burial of PCBs does not appear to occur, one reviewer thought
debating the evidence of burial from 1984 to 1994 might be a moot point, particularly
because sediment deposition trends might easily be reversed during flood events (PL).
Another reviewer agreed, citing his personal experience working with other rivers where
flood events considerably alter the river sediments (BR). These reviewers asked EPA to
clarify how flood events have historically affected the Hudson River. Representatives from
EPA explained that the Hudson River has a relatively controlled flow, due in part to
upstream reservoirs, and 100-year floods might not have as great an impact on sediment
transport as one might expect. The reviewers did not discuss issues pertaining to flood
events further.
On a related topic, however, another reviewer suggested that the conclusion of no
widespread burial should be revisited in later years (KM). Noting that much of EPA's
sampling data was collected during a release of PCBs from an upstream source, which
might not be characteristic of PCB sources over the long term, this reviewer recommended
3-7
-------
that EPA verify the finding of "no widespread PCB burial" during times when upstream
sources of PCBs have been considerably reduced.
• Comments on the importance of future modeling efforts. One reviewer commented several
times that modeling sediment deposition and resuspension might be the best approach for
determining whether widespread burial of PCBs likely occurs (RB). Another reviewer
agreed, but explained that such modeling was not included in the scope of the DEER or the
LRC (KM).
3.7 Responses to Question 7
The reviewers answered the seventh question, "Is the interpretation of the sidescansonar
data appropriate and supported by the analysis of the associated sediment properties?", as
follows:
Agreement that interpretations of the sidescansonar-data seem appropriate. The .. .
reviewers unanimously agreed that the interpretation of the sidescansonar data seemed
reasonable. One reviewer based tins finding on his personal experiences with this
acoustical technique (KR), and others based the finding on consultations with colleagues
who have used the technique (KM.BR). Two reviewers commented that the sidescansonar
data seemed to complement many other findings presented in EPA's reports (KM,RM).
Miscellaneous comments. The reviewers made several miscellaneous comments when
discussing this topic. For instance, one reviewer indicated that sidescansonar studies were
particularly useful for differentiating fine-grained and coarse-grained sediments (BR): This
reviewer thought the sidescansonar data might help researchers identify sediments that
likely contain PCBs, but he cautioned that the data cannot be used as an absolute indicator
of where PCB-contaminated sediments occur Another reviewer suggested "ground
penetrating radar" data, which might be available from the U.S. Geological Survey, also
could be useful for understanding the properties of the bedrock that underlies the river bed _
(RM). Finally, yet another reviewer thought the reports should have documented the
operative details of the sidescansonar study more extensively (e.g., describing how the
geometry of the river bed might have affected the sonar reflectivity) (KR).
3-8
-------
4.0 RESPONSES TO GENERAL QUESTIONS REGARDING THE DEIR AND LRC
After answering the 14 questions in the charge that were specific to the DEIR and LRC,
the reviewers then discussed two general questions that addressed issues documented in both
documents and their Responsiveness Summaries. When answering these questions, the reviewers
reiterated many of the findings they had presented earlier in the meeting and offered additional
comments for debate and discussion. A general record of the peer reviewers' discussions on the
two general questions follows. The reviewers' final conclusions and recommendations for the
meeting are listed in Section 5.0.
Note: As was done in previous sections, the initials of the reviewers are used to attribute
comments to the individuals who made them. RB=Dr. Reinhard Bierl, PL=Dr. Per
Larsson, KM=Dr. Keith Maruya, RM=Dr Ron Mitchum, KR=Dr. Ken Reimer, and
BR=Dr. J. Bruno Risatti.
4.1 The Usefulness of the Data Set for Understanding Fate and Transport of PCBs in the
Upper Hudson River
The first general question asked the reviewers: "Is the data set utilized to prepare the
DEIR, LRC and Responsiveness Summaries sufficient to understand the fate and transport of
PCBs in the Upper Hudson?" A summary of their responses, and the discussion that led to these
responses, follows:
• Agreement that the general conclusions of the reports are supported by the data. After
lengthy discussions on the question, and different interpretations of the question, the
reviewers eventually agreed that the conclusions in the DEIR and LRC are generally
supported by the data. In reaching this summary statement, one reviewer emphasized that
the collective weight of evidence in the EPA reports supported the main conclusions and
illustrated where PCBs generally originate and transport along the Hudson River, though
this reviewer thought a lesser emphasis should have been placed on selected quantitative
findings (KR). Most of the reviewers agreed, and indicated that the data collected by
EPA's contractors were extremely thorough (RB,PL,RM,BR). One reviewer added that
the compilation of data collected by EPA GE, and USGS generated a very comprehensive
database (RB). Despite these areas of agreement, the reviewers had differing opinions on
certain aspects of this question, as summarized below.
4-1
-------
• Comments on transport mechanisms. Though the reviewers generally agreed that the
DEIR and LRC provided a basic understanding of PCB transport in the Hudson River, two
reviewers suggested additional analyses of transport mechanisms. (Note, the reviewers
identified other recommended analyses when responding to General Question 2.) First,
noting that EPA collected only 1 year of water column transect data, one reviewer
recommended validating the results of the transect study with data collected in more recent
years (KM); this reviewer also noted that EPA's transect study did not fully characterize
PCB transport between sampling locations (e.g., did PCBs transport conservatively
between two locations? or was there trapping and resuspension?). Another reviewer
thought EPA could have provided greater insight into PCB transport by conducting studies
on sediment dynamics, even if only in the TIP, by more thoroughly characterizing the
dissolved phase, and by analyzing data trends for other pollutants in the system (BR).
• Comments on fate mechanisms. The reviewers generally agreed that the DEIR and LRC
did not provide sufficient data for understanding the fate of PCBs in the Hudson River, but
several reviewers did not necessarily view this as a shortcoming of the reports. More
specifically, several reviewers noted that EPA's study did not characterize the extent to
which certain mechanisms remove PCBs from the Hudson River, such as evaporative
losses, aerobic degradation, uptake by biota, and photolysis (PL,KM,RM). Nonetheless,
other reviewers commented that losses by some of these mechanisms are not only very
difficult to measure (RB) but also might not have been included in the scope of EPA's
study (PL). Moreover, a reviewer suspected that EPA's future modeling efforts will
address bioavailability and other phenomena related to the fate of PCBs (KR).
• Discussion on the objectives of the study. Two reviewers focused their responses on
whether the reports, particulaiiy the DEIR, met their stated objectives (RM,BR). Both
reviewers were concerned that EPA's reports might not have identified "the major factors
affecting the long term recovery of the Hudson"—an issue specified on page 1-3 of the
DEIR. The reviewers did not discuss the study objectives further, but rather agreed to
determine whether the general conclusions stated in the reports (and as modified in the
Responsiveness Summaries) were supported by the data. That discussion is summarized in
the previous bullet items.
• Other comments on the data collected for the DEIR and LRC. The reviewers raised, but
did not discuss in detail, several general issues while responding to this question. For
instance, one reviewer noted that a complete congener-specific mass balance could not be
performed on the historical data, since the sampling effort in 1984 did not characterize the
lower homologue PCBs (KM). Another reviewer recommended that EPA perform a more
complete mass balance to characterize fate and transport of PCBs more completely (BR).
This reviewer also thought the reports should have included representative chromatograms
from sediment samples collected in different stretches of the river, such that readers can
better understand the composition of PCBs in the sediments. Finally, one reviewer
4-2
-------
suggested that future work should focus specifically on understanding the fate of coplanar
PCBs, since these congeners might be important to distinguish due to their toxicity (RB).
4.2 Recommended Additional Data Analyses
The second general question asked the reviewers: "Are there any additional analyses that
should be done to verify certain findings of the DEIR and LRC?" Since reviewers had identified
additional analyses throughout the peer review meeting, they did not discuss and debate this
question in detail. Rather, they compiled a list of recommended data analyses from their
responses to earlier questions. The following bullet items present the recommendations and
identify the reviewers who made them:
• All of the reviewers thought use of multivariate statistical analyses to quantify trends and
patterns among the data would have strengthened the documents' conclusions.
• Two reviewers thought studies of sediment dynamics, at least for the TIP, concurrent with
water column sampling might better illustrate PCB transport (BR,PL). One reviewer
suggested that EPA should perform these studies during different seasons to characterize
high-flow and low-flow conditions (BR).
• Noting that the "air compartment" for a PCB mass balance has not been quantified, one
reviewer recommended further analysis of evaporative losses and photochemical
degradation of PCBs (RM). This reviewer indicated that these issues could be addressed in
many ways, such as by reviewing the scientific literature, modeling the processes, or
actually measuring them. Other reviewers agreed that evaporative losses should be
considered in EPA's future work on the site (PL,BR).
• Two reviewers recommended that the findings of the conceptual models presented in the
DEIR should be validated against more recent water column sampling data (KM,KR).
• Several reviewers offered recommendations pertaining to interpretation and presentation of
PCB analytical data in the reports. One reviewer suggested that EPA exhaust all possible
methods for relating the 1977 sediment coring data to the 1994 data, such as analyzing
archived samples and reviewing chromatograms, if any of this information is available
(KM). Another reviewer agreed and added that the DEIR should clearly state the
analytical variability of the water column and high resolution sediment coring
measurements (KR).
4-3
-------
• One reviewer suggested that EPA use more sophisticated statistical analyses, including
nonlinear analyses, when calculating certain data trends (RB).
The reviewers reiterated some of these recommended data analyses, and added others,
when presenting their final thoughts on the DEIR, the LRC, and the Responsiveness Summaries
(see Section 5.0).
4-4
-------
5.0 REVIEWERS' OVERALL RECOMMENDATIONS
After answering the specific and general questions in the charge, and after listening to the
second set of observer comments, the reviewers reconvened to provide their final findings on
EPA's reports.3 The reviewers decided to offer these findings as individual statements, during
which other reviewers did not discuss or debate each reviewer's final recommendations. Section
5.1 summarizes each peer reviewer's final statements, and Table 5-1 in Section 5.2 identifies
common themes among these final recommendations.
5.1 Peer Reviewers' Final Statements
The peer review meeting concluded with each peer reviewer providing closing statements
on the reports, including an "overall recommendation" in response to the final question in the
charge: "Based on your review of the information provided, please identify and submit an
explanation of your overall recommendation for both the DEIR and LRC.
1. Acceptable as is
2. Acceptable with minor revision (as indicated)
3. Acceptable with major revision (as outlined)
4. Not acceptable (under any circumstance)"
A detailed summary of the peer reviewers' final statements, in the order they were given,
follows:
• Dr. Keith Maruya. Dr. Maruya indicated that he accepted the main conclusions of the
reports, though he did have suggestions and recommendations for improving them. First,
he suggested that EPA publish a concise summary of the information provided in the
DEIR, LRC, and the Responsiveness Summaries. He recommended the use of multivariate
statistical analyses to make certain conclusions in these reports more convincing. Dr.
Maruya also recommended the reports more prominently acknowledge the uncertainty in
some key findings, like the estimated mass loss of PCBs.
3 Due to unforseen circumstances, one reviewer (Keith Maruya) had to leave the peer review meeting at the end
of the second day. He gave his final recommendations before the second set of observer comments.
5-1
-------
Focusing specifically on the DEIR, Dr. Maruya first reiterated a recommendation he had
mentioned earlier in the meeting: EPA should validate the findings of the conceptual
models with more recent water column sampling data. He thought such validation would
better quantify PCB sources between Roger's Island and Waterford during times when
upstream sources of PCB s are negligible. Dr. Maruya then suggested that EPA consider
the limnology of the TIP and other pools in the Hudson River for a better understanding of
PCB transport (e.g., how primary production affects partitioning, fate, and transport of
PCBs). On the topic of partition coefficients, Dr. Maruya recommended that EPA only use
the two-phase coefficients derived in the DEIR until sufficient data are available to estimate
the three-phase coefficients. Dr. Maruya did not think the data in the DEIR supported a 30
ppm threshold below which PCB dechlorination reportedly does not occur.
Commenting on the LRC, Dr. Maruya first concluded that the comparisons between the
PCB inventories in 1984 and 1994 were reasonable and the data from 1977 were not
sufficient for inventory estimates. He thought the analytical variability contributed to
considerable uncertainty in the inventory estimates, which the LRC did not acknowledge.
Dr. Maruya thought EPA should further consider how elevated PCB concentrations in
near-shore sediments might have affected the inventory estimates. Finally, Dr. Maruya
maintained that the sampling data suggest that widespread burial of PCBs does not occur.
Overall, Dr. Maruya thought the DEIR and LRC were both "acceptable with minor
revisions."
Dr. Ken Reimer. Dr. Reimer concluded that the weight of evidence of the data presented
in the DEIR and LRC generally support the reports' main conclusions, especially as they
were modified in the Responsiveness Summaries. He thought the data collected for the
reports provided an adequate basis for EPA to proceed with its reassessment.
Dr. Reimer then listed several suggestions and recommendations. First, noting that the
public might have difficulty identifying the basic messages of the DEIR and LRC, Dr.
Reimer recommended that EPA prepare a succinct summary of the major findings of these
reports. Second, he strongly recommended that EPA's reports present quantitative findings
in appropriate context, particularly with respect to uncertainty. Dr. Reimer suggested that
EPA consider presenting ranges of data when the actual values are not known. He
cautioned EPA about "over interpreting" data.
Focusing on the main conclusions of the reports, Dr. Reimer indicated that they were
generally supported by the data, but with a few caveats. He thought the conceptual models
used to interpret the water column transect studies could be improved, for example, with
the use of multivariate analyses to "fingerprint" sources of PCBs. Further, Dr. Reimer
suggested that the reports not infer that anaerobic dechlorination of PCBs does not occur
at PCB concentrations less than 30 ppm. He added, however, that dechlorination is "a
5-2
-------
very minor issue" in terms of the overall problem of contaminated sediments. Dr. Reimer
then discussed the issue of estimating PCB mass loss in the sediments: he thought the
1984 and 1994 were sufficient for making these estimates; he cautioned against presenting
firm estimates of the mass loss; and he also cautioned against using the 1977 sediment
coring data for this purpose. Finally, Dr. Reimer concluded that the data suggest that
widespread burial of PCBs does not occur in the TIP and that the TIP sediments act as a
source of PCBs to the water column.
Overall, Dr. Reimer found the DEIR and LRC to be "acceptable with minor revisions."
Dr. Reinhard Bierl. Dr. Bierl opened his final statements by indicating that the data
reported in the DEIR and LRC are sufficient for EPA to proceed with its reassessment, but
he identified several aspects of the reports that should be improved to make them more
convincing. Regarding the statistical methods used in the reports, Dr. Bierl recommended
the use of multivariate analyses to quantify certain trends and additional statistical analyses
to calculate changes in PCB inventories. Dr. Bierl then suggested that EPA qualify its
quantitative estimates of PCB mass loss to put these figures into perspective. Dr. Bierl
added that he wanted to see more information in the reports on the PCB analytical methods
(e.g., quality assurance plans and standard operating procedures). He thought this
information was particularly lacking for the previous sediment coring studies.
Noting the time gaps between the various sediment coring studies, Dr. Bierl recommended
that EPA consider reviewing more recent sampling data and possibly even consider
implementing ongoing monitoring studies. He thought future studies should focus on
characterizing how PCBs partition between the suspended and dissolved phases, among
other research topics.
Overall, Dr. Bierl found the DEIR and LRC to be acceptable with revisions, but he was not
sure whether his recommended revisions should be considered "minor" or "major."
Dr. Per Larsson. Dr. Larsson concluded that the data summarized in the DEIR and LRC
identified major source areas of PCBs in the Hudson River and characterized the extent of
contamination in these areas. Dr. Larsson found that the data indicate a loss of PCBs from
the river sediments, but he thought the exact amount of losses are difficult to quantify. He
reminded the reviewers, however, that even "a very small percentage" loss of PCBs might
have very serious consequences on downstream ecosystems.
Dr. Larsson then reviewed his responses to selected questions in the charge. First, he
found that the river sediments in the TIP undoubtedly act as a source of PCBs to the water
column; he recommended that EPA include a basic model in the final report to estimate the
source loading of the sediments. Second, Dr. Larsson commended EPA's work on
differentiating dissolved phase PCBs from suspended phase PCBs—a distinction he
thought would be important for future analyses of bioavailability. Third, Dr. Larsson noted
5-3
-------
that he and other reviewers had questions about the mechanisms that cause PCBs to enter
the water column; he suspected that particle transport (rather than bioturbation or pore
water diffusion) is probably the primary mechanism affecting PCB transport. Finally, Dr.
Larsson addressed the findings of PCB mass loss and sediment burial. He was convinced
that PCBs are gradually transporting with the sediments, and he speculated that the river
sediments will continue to redistribute in the future. Noting that the Hudson River is a
dynamic system, Dr. Larsson cautioned against assuming data trends from a 10-year time
frame are representative of river conditions over the longer term.
Based on his review of the documents, Dr. Larsson thought two specific revisions were
necessary. He recommended the use of multivariate statistics for identifying and
quantifying trends and patterns among the large volume of congener-specific data. He also
recommended the reports thoroughly describe the data analysis methodology, such that the
statistical analyses are transparent and easier to follow.
Overall, Dr. Larsson thought the DEIR and LRC were "acceptable with minor revisions."
Dr. Ron Mitchum. Dr. Mitchum split his comments into those specific to the DEIR and
those specific to the LRC. Beginning with the DEIR, Dr. Mitchum noted that many of the
report's original conclusions had been "softened" in the Responsiveness Summary. He
then offered several suggestions for future work on the site and improving the DEIR. He
first recommended that EPA include in its ongoing analysis some assessment of
evaporative losses and photochemical degradation of PCBs. Dr. Mitchum then suggested
that EPA use multivariate statistical analyses to verify many of the findings in the report.
He also suggested that the report's conclusions include discussions about uncertainty,
particularly in regard to sampling and analytical variability. Dr. Mitchum thought the
DEIR's original conclusion of a concentration threshold for anaerobic dechlorination was
not well founded.
Dr. Mitchum then summarized his major findings pertaining to the LRC. First, he
concluded that EPA did "the best job possible" in comparing the 1984 and 1994 sediment
coring data. Dr. Mitchum added, however, that sampling and analytical variability limited
the confidence he had in the estimated PCB inventories. Regardless of the uncertainty, Dr.
Mitchum believed the 1984 and 1994 data sets support EPA's conclusion that the hot spots
in the river have lost PCBs. He cautioned EPA against using the 1977 sediment coring
data in the ongoing reassessment. Finally, Dr. Mitchum suggested that use of multivariate
statistical analyses was needed to verify conclusions in the LRC.
Overall, Dr. Mitchum thought the DEIR and LRC were both "acceptable with minor
revisions."
Dr. J. Bruno Risatti. During his final statements, Dr. Risatti provided general comments
about both reports, followed by comments specific to the individual reports. Dr. Risatti
5-4
-------
thought the data collected for the reports provide a background for a better understanding
of PCB transport in the Hudson River, but he did not think the reports should be
considered as an "all encompassing" study. In general, Dr. Risatti was uncertain about
some findings in the reports, due largely to the analytical variability in the data. He thought
the PCB transport processes could be further characterized by conducting sedimentological
studies concurrent with water column sampling. Though he found the reports extensive,
Dr. Risatti thought they should have more thoroughly addressed the fate of PCBs by
considering aerobic degradation and evaporative losses.
Focusing specifically on the DEIR, Dr. Risatti's primary finding was that EPA should
reconsider its conclusions regarding anaerobic dechlorination, particularly the finding of a
30 ppm threshold below which dechlorination does not occur. He then reiterated that the
MDPR might underestimate actual dechlorination, since the MDPR is calculated from
concentrations of lower homologue PCBs that are more likely to transport from the
sediments.
When presenting his comments on the LRC, Dr. Risatti suggested that the study had some
evidence of cross contamination of the "vibracore" samples, and he recommended that
EPA conduct a basic study to quantify the potential extent of this cross contamination.
Noting that he had difficulties reading the LRC (and the DEIR), Dr. Risatti also
recommended that EPA develop guidelines for writing technical reports in a format similar
to articles in scientific journals.
Overall, Dr. Risatti found the DEIR and LRC to be acceptable with revisions, but he was
not sure whether his recommended revisions should be considered "minor" or "major."
5-5
-------
5.2 Summary of Peer Reviewers' Final Recommendations
The reviewers' final recommendations, which are detailed in Section 5.1, are summarized
by peer reviewer in Table 5-1. (Note that this table does not incorporate any additional
recommendations the reviewers made during earlier portions of the meeting.)
5-6
-------
Table 5-1
Peer Reviewers' Final Recommendations
1
1 Recommendation
Reinhard Bierl
Per
Larsson
Keith
Maruya
Rem
Mitchum
Ken
Reimer
J. Bruno
Risatti
Overall DEIR recommendation
Acceptable, with
revisions that fall
between "minor"
and "major"
Acceptable
with minor
revisions
Acceptable
with minor
revisions
Acceptable
with minor
revisions
Acceptable
with minor
revisions
Acceptable, with
revisions that fall
between "minor"
and "major"
Overall LRC recommendation
Acceptable, with
revisions that fall
between "minor"
and "major"
Acceptable
with minor
revisions
Acceptable
with minor
revisions
Acceptable
with minor
revisions
Acceptable
with minor
revisions
Acceptable, with
revisions that fall
between "minor"
and "major"
Use multivariate statistical analyses of data
~
/
/
~
/
More prominently acknowledge uncertainty
in conclusions; use data ranges to present
findings that might be highly uncertain
/
/
~
~
~
Publish a concise summary of the
information in the DEIR, LRC, and
Responsiveness Summaries
~
~
Consider the limnology of the TIP and other
pools in the Hudson River (e.g., how primary
production affects PCB fate and transport)
~
Validate the conceptual models and other
findings with more recent water column
sampling data
~
~
Note: This table summarizes recommendations made during the peer reviewers* final statements; some reviewers might have made additional recommendations during
their earlier discussions.
-------
Table 5-1 (Continued)
Peer Reviewers' Final Recommendations
| Recommendation
Rcinhard Bierl
Per
Larsson
Keith
Maruya
Ron
Mitchum
Ken
Reimer
J. Bruno 1
Risatti H
| Use two-phase partition coefficients until
| enough data are available to derive three-
| phase coefficients
~
fl Modify the conclusion regarding the 30 ppm
H threshold for anaerobic dechlorination
~
~
~
~ |
| Further consider how elevated PCB
H concentrations in near-shore sediments might
| affect inventory estimates
/
Use more sophisticated statistical analyses to
estimate PCB inventory
~
Provide additional details on the analytical
methods used in the various sediment coring
studies
~
Better characterize exchange of PCBs
between the suspended and dissolved phase
~
Describe the data analysis methodology in
the reports
/
Consider other compartments in the PCB
mass balance (e.g., evaporative losses,
photochemical degradation, aerobic
degradation)
/
~ I
Note: This table summarizes recommendations made during the peer reviewers' final statements; some reviewers might have made additional recommendations during
their earlier discussions.
-------
Table 5-1 (Continued)
Peer Reviewers' Final Recommendations
| Recommendation
Reinhard Bierl
Per
Lars son
Keith
Maruya
Ron
Mitchum
Ken
Reimer
J. Bruno
~
1 Conduct sedimentological studies concurrent
H with water column sampling
1 Conduct an experiment to characterize the
1 extent of cross contamination in
H "vibracoring" samples
~
H Establish guidelines for writing future reports
~
Note: This table summarizes recommendations made during the peer reviewers' final statements; some reviewers might have made additional recommendations during
their earlier discussions.
-------
6.0 REFERENCES
TAMS et al., 1997. "Data Evaluation and Interpretation Report." Volume 2C of the Hudson
River PCBs Reassessment RI/FS. Review copy. Prepared by TAMS Consultants, Inc.,
The CADMUS Group, Inc., and Gradient Corporation. February, 1997.
TAMSetal., 1998a. "Responsiveness Summary for: Volume 2A: Database Report; Volume 2B.
Preliminary Model Calibration Report; Volume 2C: Data Evaluation and Interpretation
Report." Prepared by TAMS Consultants, Inc., The CADMUS Group, Inc., and Gradient
Corporation. December, 1998.
TAMS et al., 1998b. "Low Resolution Sediment Coring Report." Volume 2C-A of the Hudson
River PCBs Reassessment RI/FS. Review copy. Prepared by TAMS Consultants, Inc.,
The CADMUS Group, Inc., and Gradient Corporation. July, 1998.
TAMS et al., 1999. "Responsiveness Summary for: Volume 2C-A: Low Resolution Sediment
Coring Report." Prepared by TAMS Consultants, Inc., and TetraTech, Inc. February,
1999.
6-1
-------
APPENDIX A
LIST OF EXPERT PEER REVIEWERS
-------
A EPA
Unitod States
Environmental Protection Agency
Region 2
Peer Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Data Evaluation and Interpretation Report
Low Resolution Sediment Coring Report
Albany Marriott
Albany, New York
March 16 -18,1999
Peer Reviewers
Dr. Reinhard Bierl
Senior Lecturer
University of Trier
FB VI - Hydrology
Universitdtsring 15
D-54286 Trier GERMANY
49-651-201-2229
Fax: 49-651-201-3978
E-Mail: bieri@uni-trier.de
Dr. Ronald Mitchum
President
Data Analysis Technologies, Inc.
6385 Shier Rings Road
Dublin, OH 43016
614-791-8008
Fax: 614-791-8007
E-Mail: datlab@infinet.com
Dr. Per Lars son
Assistant Professor
Lund University
Ecotoxicology, Department of Ecology
Ecology Building
Sdlvegatan 37
Lund, SWEDEN 223 62
46-46-222-3779
Fax: 46-46-222-3790
E-Mail: Per.Larsson@ecotox.lu.se
Dr. Ken Reimer
Director, Environmental Sciences Group
The Royal Military College of Canada
Building 62
P.O. Box 17000 Stn Forces
Kingston, Ontario CANADA K7K 7B4
613-541-6161
Fax: 613-541-6596
E-Mail: reimer-k@rmc.ca
Dr. Keith Maruya
Assistant Professor
Skidaway Institute of Oceanography
10 Ocean Science Circle
Savannah, GA 31411
912-598-2315
Fax: 912-598-2310
E-Mail: kam@skio.peachnet.edu
Dr. J. Bruno Risattl
Senior Organic/Microbial Geochemist
Illinois State Geological Survey
615 East Peabody Drive
Champaign, IL 61820
217-333-5103
Fax: 217-244-2785
E-Mail: risatti@isgs.uiuc.edu
Printed on Recycled Paper
-------
APPENDIX B
mADrp Tfi PYDrDT nrrD DPvnru/rDC
J«\Fu * frfam A A iKk JKiJErf * ¦ m.*t
-------
a unitod state*
B—Environmental Protection Agency
IhI #m Region 2
Peer Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Data Evaluation and Interpretation Report
Low Resolution Sediment Coring Report
Albany Marriott
Albany, New York
March 16-18,1999
Charge
Members of this peer review will be tasked to determine whether the scientific analyses
conducted for U.S. EPA's Data Evaluation and Interpretation Report (DEIR) and the Low
Resolution Sediment Coring Report (LRC) are credible, the conclusions valid, and whether
the findings are appropriate to use to support the decision-making process for the Hudson
River PCBs Site Reassessment study. The peer reviewers will base their assessment on the
review of the DEIR and LRC, as well as additional information found in the Responsiveness
Summary issued in December 1998 (responding to several documents including the DEIR)
and the Responsiveness Summary for the LRC (which will be released in February 1999).
The peer reviewers will also have available for their review the Hudson River Reassessment
database, which contains all of the data used in the preparation of the above documents,
along with other data.
The DEIR and LRC present the results of the geochemical analyses conducted on the
water-column and sediment data collected for the Reassessment, as well as data collected b
a number of other agencies and General Electric. It should be noted that there have been
several changes in the available data since the time of the release of the DEIR in February
1997. These changes include a better estimate of flow for several reaches of the river, a
recalculation of GE's PCB data due to an analytical bias, and the discovery of a sampling bia
at the Thompson Island Dam monitoring station. These changes are discussed in the
Responsiveness Summaries, and the analyses in the Responsiveness Summaries should
supersede those conducted in the reports, as appropriate.
It is important to realize that the geochemical analysis conducted in the DEIR and LRC will b«
complemented by mass balance modeling and human health and ecological risk assessment:
to provide a thorough understanding of the fate and transport and impacts of PCBs in the
Upper Hudson River. These other reports will address questions regarding the mechanisms
that release PCBs from the sediment, toxicity, and bioavailability/biouptake. A peer review
was previously conducted for the approach proposed to conduct the modeling for the
Reassessment. After the modeling and the risk assessment reports are released, the Agency
will also have those documents peer reviewed.
PrinM on RacycM Pipw
-------
Specific Questions
Data Evaluation and interpretation Report (DEIR)
1. Is the documented PCB load, which originated from the Tl Pool, consistent with
source consisting of historically deposited PCB-contaminated sediments?
2. Are the two-phase and three-phase partitioning coefficients, derived in the
DEIR, appropriate and do they properly address the physical parameters of
the system (e.g., temperature).
3. Are the conceptual models based on the transect sampling consistent with the
data?
4. Does the sampling at the Tl Dam-West location impact EPA's conclusion tha
the sediments of the Tl Pool are the major source of PCBs to the freshwater
Hudson during low flow conditions considering the analytical corrections
made to GE's PCB data? What are the other implications of finding higher
concentrations along the shoreline than in the center channel?
5. Are the geostatistical techniques (polygonal declustering and kriging)
correctly applied?
6. Are the methods applied in the DEIR (change in molecular weight (MW) and
evaluating concentrations of BZ#s 1, 4, 8,10 and 19 (MDPR)) appropriate „
standards for determining extent of dechlorination? Are there any significant
problems with this approach, or more appropriate approaches?
7. The DEIR finds that the degree of anaerobic dechlorination is primarily a
function of original concentration rather than time, and accordingly that there
is not significant predictable dechlorination in sediments containing less than
approximately 30 mg/kg PCB. Is this reasonable?
Low Resolution Sediment Coring Report (LRC)
1. In the LRC, EPA compared sediment data from cores taken in 1977, 1984
and 1994, which had the PCB analysis conducted by different laboratory
methods. How valid are the methods used to establish a consistent basis for
comparison?
2. In the upper Hudson River system, it has been well established that there is
significant lateral heterogeneity in sediment concentrations. While it was
attempted to reoccupy previous locations, some uncertainty is added with
respect to the actual sampling location. While the statistical techniques help
compensate for this, how does the sediment heterogeneity affect the
comparison of cores from two different years? Given the spatial variability, ir
the finding that there is loss from most of the locations supported by the
data?
-------
3. What is the impact of the difference between replicate samples in the 1994
sampling effort (36 percent average variability) on the finding that there wa
a 40 percent loss of PCB inventory from the highly contaminated sediment!
in the Tl Pool?
4. In the LRC, it was found that Hot Spot 28 contained much more mass than
previous estimates. Is the conclusion that this "gain" is primarily due to
incomplete characterization in 1977 valid?
5. Does the data set and its interpretation support the conclusion that signifies
losses have occurred from hot spots below Tl Dam?
6. The LRC found that the historically contaminated sediments in the Tl Pool
were not universally being buried and sequestered from the environment.
How much confidence would you place in the LRC evidence against
widespread burial?
7. Is the interpretation of the sidescansonar data appropriate and supported b
the analysis of the associated sediment properties?
General Questions
Is the data set utilized to prepare the DEIR, LRC and Responsiveness
Summaries sufficient to understand the fate and transport of PCBs in the Uppei
Hudson?
Are there any additional analyses that should be done to verify certain findings
of the DEIR and LRC?
Recommendations
Based on your review of the information provided, please identify and submit an explanation
of your overall recommendation for both the DEIR and LRC.
1. Acceptable as is
2. Acceptable with minor revision (as indicated)
3. Acceptable with major revision (as outlined)
4. Not acceptable (under any circumstance)
-------
APPENDIX C
PREMEETING COMMENTS, ALPHABETIZED BY AUTHOR
Note:
Dr. Reinhard Bierl's premeeting comments were submitted on the first day of the peer
review meeting. These comments are included at the end of this appendix.
-------
Peer Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Data Evaluation and Interpretation Report
Low Resolution Sediment Coring Report
Premeeting Comments
Albany, New York
March 16-18, 1999
-------
Notice
The U.S. Environmental Protection Agency (EPA) strives to provide accurate, complete, and
useful information. Neither EPA nor any person contributing to the preparation of this
document, however, makes any warranty, expressed or implied, with respect to the usefulness
or effectiveness of any information, method, or process disclosed in this material. Nor does
EPA assume any liability for the use of, or for damages arising from the use of, any
information, methods, or process disclosed in this document.
Any mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
11
-------
Table of Contents
Charge for Peer Review 2 1
Peer Reviewers
Dr. Reinhard Bierl 5
Dr. Per Larsson 9
Dr. Keith Maruya 21
Dr. Ronald Mitchum 47
Dr. Ken Reimer .* 59
Dr. James Risatti 83
Note: Premeeting comment materials have been reproduced as received.
in
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
IV
-------
Hudson River PCBs Site Reassessment RI/FS
Data Evaluation and Interpretation Report (DEIR) and
Low Resolution Sediment Coring Report (LRC)
Peer Review 2
Charge for Peer Review 2
Members of this peer review will be tasked to determine whether the scientific analyses
conducted for USEPA's Data Evaluation and Interpretation Report (DEIR) and the
Low Resolution Sediment Coring Report (LRC) are credible, the conclusions valid,
and whether the findings are appropriate to use to support the decision-making
process for the Hudson River PCBs Site Reassessment study. The peer reviewers will
base their assessment on the review of the DEIR and LRC, as well as additional
information found in the Responsiveness Summary issued in December 1998
(responding to several documents including the DEIR) and the Responsiveness
Summary for the LRC (which will be released in February 1999). The peer reviewers
will also have available for their review the Hudson River Reassessment database,
which contains all of the data used in the preparation of the above documents, along
with other data.
The DEIR and LRC present the results of the geochemical analyses conducted on the
water-column and sediment data collected for the Reassessment, as well as data
collected by a number of other agencies and General Electric. It should be noted that
there have been several changes in the available data since the time of the release of the
DEIR in February 1997. These changes include a better estimate of flow for several
reaches of the river, a recalculation of GE's PCB data due to an analytical bias, and the
discovery of a sampling bias at the Thompson Island Dam monitoring station. These
changes are discussed in the Responsiveness Summaries, and the analyses in the
Responsiveness Summaries should supercede those conducted in the reports, as
appropriate.
It is important to realize that the geochemical analysis conducted in the DEIR and
LRC will be complemented by mass balance modeling and human health and
ecological risk assessments to provide a thorough understanding of the fate and
transport and impacts of PCBs in the Upper Hudson River. These other reports will
address questions regarding the mechanisms that release PCBs from the sediment,
toxicity, and bioavailability/biouptake. A peer review was previously conducted for
the approach proposed to conduct the modeling for the Reassessment. After the
modeling and the risk assessment reports are released, the Agency will also have those
documents peer reviewed.
1
-------
Specific Questions
Data Evaluation and Interpretation Report (DEIR)
1. Is the documented PCB load, which originated from the TI Pool, consistent
with a source consisting of historically deposited PCB-contaminated
sediments?
2. Are the two-phase and three-phase partitioning coefficients, derived in the
DEIR, appropriate and do they properly address the physical parameters of the
system (e.g., temperature).
3. Are the conceptual models based on the transect sampling consistent with the
data?
4. Does the sampling at the TI Dam-West location impact EPA's conclusion that
the sediments of the TI Pool are the major source of PCBs to the freshwater
Hudson during low flow conditions considering the analytical corrections
made to GE's PCB data? What are the other implications of finding higher
concentrations along the shoreline than in the center channel?
5. Are the geostatistical techniques (polygonal declustering and kriging) correctly
applied?
6. Are the methods applied in the DEIR (change in molecular weight (MW) and
evaluating concentrations of BZ#s 1, 4, 8, 10 and 19 (MDPR)) appropriate
standards for determining extent of dechlorination? Are there any significant
problems with this approach, or more appropriate approaches?
7. The DEIR finds that the degree of anaerobic dechlorination is primarily a
function of original concentration rather than time, and accordingly that there
is not significant predictable dechlorination in sediments containing less than
approximately 30 mg/kg PCB. Is this reasonable?
Low Resolution Sediment Coring Report (LRC)
1. In the LRC, EPA compared sediment data from cores taken in 1977, 1984 and
1994, which had the PCB analysis conducted by different laboratory methods.
How valid are the methods used to establish a consistent basis for comparison?
2. In the upper Hudson River system, it has been well established that there is
significant lateral heterogeneity in sediment concentrations. While it was
attempted to reoccupy previous locations, some uncertainty is added with
2
-------
respect to the actual sampling location. While the statistical techniques help
compensate for this, how does the sediment heterogeneity affect the
comparison of cores from two different years? Given the spatial variability, is
the finding that there is loss from most of the locations supported by the data?
3. What is the impact of the difference between replicate samples in the 1994
sampling effort (36 percent average variability) on the finding that there was a
40 percent loss of PCB inventory from the highly contaminated sediments in
the TI Pool?
4. In the LRC, it was found that Hot Spot 28 contained much more mass than
previous estimates. Is the conclusion that this "gain" is primarily due to
incomplete characterization in 1977 valid?
5. Does the data set and its interpretation support the conclusion that significant
losses have occurred from hot spots below TI Dam?
6. The LRC found that the historically contaminated sediments in the TI Pool
were not universally being buried and sequestered from the environment.
How much confidence would you place in the LRC evidence against
widespread burial?
7. Is the interpretation of the sidescansonar data appropriate and supported by
the analysis of the associated sediment properties?
General Questions
1. Is the data set utilized to prepare the DEIR, LRC and Responsiveness
Summaries sufficient to understand the fate and transport of PCBs in the
Upper Hudson?
2. Are there any additional analyses that should be done to verify certain findings
of the DEIR and LRC?
Recommendations
Based on your review of the information provided, please identify and submit an
explanation of your overall recommendation for both the DEIR and LRC..
1. Acceptable as is
2. Acceptable with minor revision (as indicated)
3. Acceptable with major revision (as outlined)
4. Not acceptable (under any circumstance)
3
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
4
-------
Dr. Reinhard Bierl
5
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
6
-------
Comments not available at time of print
7
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
8
-------
Dr. Per Larsson
9
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
10
-------
1
Lund February 1999-02-23
Review of the "Data evaluation and interpretation report (DEIR) and
low resolution sediment coring report (LRC).
Per Larsson
Ecotoxicology, Department of Ecology
University of Lund Sweden
1. Introduction and major conclusions
The PCB problem in the Hudson River system is in my view a serious one, and
the extent of the problem has several aspects that cover such areas as exposure
and effect of aquatic biota; transport of the compounds from the source areas to
the lower Hudson river and ultimately to the sea; volatilization from river water to
the atmosphere (with a consequent exposure to terrestrial environments or further
atmospheric transport); contamination of groundwater and river bed soils. The risk
for human exposure of PCB in the area from e.g. fish and by lower extent by
(drinking) water is obvious. The list of problem areas can, furthermore, be
expanded substantially.
As I see it, a few simple and straightforward questions or objectives can be drawn
from the two reports cited above, and if these are answered or fulfilled, I find that
the reports can be considered as acceptable. It is then obvious that the studies
answering the questions need to be carried out in a scientifically acceptable way
and that the conclusions drawn are appropriate. The questions or objectives are:
*The major source areas (sediment) for PCB contamination can be identified.
*What is the extent of the PCB contamination in the source areas?
*Has the extent of the PCB contamination in the sediment (concentration, mass)
changed over time?
*PCB in the sediment of the source areas affects concentration in the water (i.e.
there is a PCB transport over the sediment/water interface).
*What is the extent of the sediment to water transport of PCB?
*PCB from the source areas are transported downstream.
*How does PCB in the river water speciate between the dissolved and particulate
phase? The answer to this question is interesting in two ways;
i. the transport form of PCB in the river water (and possible
resedimentation, readsorption, etc.)
11
-------
ii. for exposure to biota.
These are the major points that need to be evaluated, in order to establish a base
for understanding of the fundamental behaviour of PCB in the upper Hudson
River. Within the questions or objectives lie understanding of the potential
variables of importance, like the effect of water discharge, water temperature,
microbial activity etc. In addition to these points, other items of interest can be
identified that cannot be considered to cover the main objectives, but may lead to
understanding of the underlying mechanisms like:
*ls a potential change over time in PCB concentration in the sediment at one site
due to transport to the water?
* Is a potential change over time in PCB concentration in the sediment at one site
due to microbial, anaerobic dechlorination?
*ls a potential change over time in PCB concentration in the sediment at one site
due to transport of contaminated sediment downstream?
" Is a potential change over time in PCB concentration in the surface sediment at
one site due to burial by new, uncontaminated sediment?
The question of microbial, anaerobic dechlorination of PCB in the sediment is
addressed thoroughly throughout the report, but for me this question has nothing
to do with the main objectives of the studies. As anaerobic dechlorination as a
process only dechlorinate congeners containing meta- and/or para positions and,
consequently, reduce more chlorinated congeners to less chlorinated, it cannot
be considered as a sink process for PCB. No ring cleavage or break-down of PCB
occurs through this mechanism (e.g. Bedard et al. 1986). It will, however, have
effects on the transport of PCB across the sediment/water interface (possibly
increase the transport as low-chlorinated congeners have a higher transport-rate
than high-chlorinated ones), and on toxicity and bioaccumulation.
In conclusion, I find that the important issues raised above have
been addressed in the reports and in a majority of cases also have
been satisfactory answered. I find that the two reports, DEIR and
LRC, are acceptable. As no studies or reports are perfect, some minor
revisions can be made as stated below, on parts of the material. My main
negative comments are:
a) Multivariate data are best treated by multivariate statistics, as by
Principal Component Analysis (PCA), Cluster analysis or Regression
Tree. PCA would have been suitable for the data sets, resulting in a
more comprehensive and objective analysis of the data (Zitko 1989.
Bremle and Larsson 1997). The results are now concluded from
individual regression or correlation analysis, similarities of curves, tests
12
-------
of individual data sets, congener pattern etc. I do not conclude that
these analysis are carried out in a erroneous way, they would probably
be good instruments to combine with e.g a PCA approach. They do not,
however, give a good overview in analysis of the whole data set. The
PCA could have been especially useful for the congener "fingerprinting"
in source identification, comparisons between matrixes, and indications
of processes as anaerobic dechlorination as well as for determining
variable importance (like discharge, water chemistry, etc. (Zitko 1989,
Bremle and Larsson 1997, Bremle and Larsson 1998).
b) The sampling of river water for PCB is not as advanced as the sediment
sampling program. It would have been better to have introduced more
permanent sampling stations and pump up water from a defined water
depth (10-20 cm below the surface in the centre of the river), and taken
integrated, continuous water samples over time (Bremle and Larsson
1997). The water volumes taken in the sampling program, 17 L, is quite
sufficient for PCB analysis, while 1 L is not (I L sampling will lead to
substantial errors in the quantification process). The separation in
dissolved and particulate phase PCB by filtering is a good approach,
that facilitate later transport and exposure estimates of PCB.
c) Incomplete homogenisation of sediment samples, which could have
been the case in the LRC program, may lead to errors in conclusions of
PCB concentration and PCB mass in large sediment slices. It's tricky to
homogenise a large amount of sediment containing low percentages of
water. Note that I state may, it's hard to conclude from the reports that
this was really the case.
2. Specific questions
2.1 DEIR
2.1.1. Historically deposited PCB contaminated sediment has been shown to
contaminate the water of rivers (Brown et al 1985, Chevreuil and Granier 1991,
Bremle and Larsson 1997). In principal, PCB transport from sediment to water of
river systems is determined by i) water discharge, where high discharges
(flooding) results in turbulence of the river sediment, and a consequent
downstream transport of PCB contaminated particles (Chevreuil and Granier
1991). This transport does not change the PCB fingerprint in the water column (as
stated in the report and Bremle and Larsson 1998) ii) temperature, where
desorption (partitioning) from sediment is enhanced with increasing temperature
and changing the PCB pattern to a low-chlorinated one (Larsson et al 1990). The
process is enhanced by processes such as bioturbation by benthic invertebrates,
and by microbial processes that mineralise organic matter in the sediment
(Jeremiason et al. 1998, Larsson et al 1990). These processes are affected
positively by concentration of PCB in the sediment (a higher sediment to water
13
-------
transport) and the latter three negatively to the organic carbon content in the
sediment (e.g. Steen et al 1978, Lara and Ernst 1990).
The present study shows that the PCB load into Thompson Island pool (Tl) is less
than the transport out of the pool, than revealing an addition of PCB from within
the pool, a PCB transport from sediment to water. Further, water parcels is
followed downstream by simultaneous sampling revealing some dilution of PCB
by increasing water discharge and some smaller addition to the PCB load (or
constant PCB mass transport depending on sampling station). Fingerprinting of
PCB congeners within the Tl pool water and downstream show that the PCB
originated from the pool. The results are convincing, and very similar to results
obtained for a PCB contaminated river system in Sweden (Larsson et al. 1990,
Anon 1998), where one single source (a small lake with PCB contaminated
sediments within the river system, water residence time about 4 h for the lake)
affected the whole river system downstream.
2.1.2. The two-phase and three-phase models are appropriate to use for
predicting PCB partitioning between the dissolved and particulate phase in the
water, and the models can be scrutinised in detail by using the extensive
international literature on this subject (e.g. Horzempa and DiToro 1983, Baker et
al. 1986). I find the approaches a bit out of focus for the objectives of this study.
Instead, it would have been useful to include an empirical modelling work on, for
instance, Tl pool using variables as sediment surface (as determined by coring
and side scan-sonar) and sediment concentration of PCB, water discharge and
temperature to predict transport and mass-balances over time (Larsson et al.
1990, Bremle and Larssson 1997). In this way an empirical model for PCB in the
upper Hudson River can be constructed and applied for different situations. For
prediction on bioavailability of PCB the proposed models seem to be appropriate
(dissolved versus particulate PCB).
2.1.3. The conceptual reasoning of PCB fate in the upper Hudson River is
convincing and show that the authors of the report know the literature. The
discussion on how the sources (PCB containing sediment) affect PCB in the water
column, how concentrations of PCB decrease downstream as a result of
volatilisation, and adsorption/settling, dilution by increasing water discharge in
the river as the catchment area increases, are logic and can be understood by the
reader. The figures underlying the reasoning could have been simplified, given a
logic system of location numbering etc., but this is just a technical matter.
2.1.4. I don't find that the sampling location would affect the conclusion that Tl
pool is the major source of PCB to the upper Hudson River system. As far as I
understand, corrections have to be made only under low flow conditions. As the
PCB load follow water discharge, the major PCB loads occur under higher flow
situations. As stated in my introduction, however, water sampling near the
riverbanks (sides) or just upstream of large objects should be avoided, as there is
a risk of sampling turbated, nearshore sediment or upwelling sediment. The
14
-------
5
possible erroneous sampling can be detected in the results by studying filtered
particle amounts in the water (too high amounts of particle dry mass), as stated by
the report authors.
2.1.5. The applied techniques for evaluating sediment-coring analysis, and to
calculate PCB masses for larger sediment areas (i.e. hot spots) using polygonal
clustering and kriging seem correct to me. A further development suggested by
the report authors to use results from side scan-sonar to identify primary
accumulation areas containing small particles high in PCB and organic content,
and combine that with polygonal clustering (clustering coring points within an
accumulation area), seems even better.
2.1.6. In order to evaluate if microbial, anaerobic dechlorination of PCB has
occurred in the sediment, the authors use enrichment in congener 1, 4, 8, 10 and
19 versus the total sum of all congeners as an index. The index is quite
appropriate, and will reveal if dechlorination has occurred for the sample. There
are other ways of constructing similar indexes, the important factor to consider is
that the proportion of the dechlorination products (congener 1, 4, 8, 10 and 19)
can be compared with the same congeners in a sample not subjected to
dechlorination (or a standard) and the difference revealed. This is simple to
perform for a limited data set of congeners but when a larger data set is to be
evaluated, indexes have limitations. It is then more appropriate to evaluate the
whole PCB pattern by a multivariate method, like principal component analysis
(PCB, Bremle and Larsson 1997). If significant dechlorination has occurred in
samples, this will be revealed by clustering of these data in a PCA plot.
2.1.7. I fully agree that the degree of anaerobic dechlorination is a function of the
original PCB concentration in the sediment. The predictive model (figure)
showing that the extent of dechlorination start to be significant at ca 30 ppm PCB
is elegant, and show scientific skill. As stated in the introduction, I don't find that
the subject of anaerobic dechlorination of PCB is a central objective of the study.
The study by Williams and May (1997) suggesting that low-temperature microbial
degradation was significant for di- and trichlorobiphenyls in laboratory model
systems, and possibly connected to the reduction/oxidation of the iron cycle in the
sediment surface seems to have higher relevance (although I find objections to
parts of this study).
2.2. LRC
2.2.1. It is obvious that comparisons between concentrations of PCB in sediment
cores, taken in similar areas from 1977, 1984, and 1994, will show variations due
to i) the sampling methods used ii) the improvement of the analytical methods
used, like capillary columns iii) the use of surrogate and internal standards. I do
not, however, consider this a serious problem since you have to expect variation
due to the analytical methods used in a time period >15 years. I find the quality
control of the present study satisfactory, as well as the use of dual GC-columns,
15
-------
6
and use of surrogate standards,(it is a bit confusing though, that the use of
octochloronaphtalene (OCN), did not work out. We have used it extensively when
analysing sediment for PCB at the ecotoxicology laboratory at Lund University),
and confirmation on GC/MS for some of the samples.
To my experience, there is a much greater variation in the field
sampling strategies between different studies and years, that lead to variability in
the data and make comparisons hard to do. This kind of problems are hard to
overcome, due to lack of information.
2.2.2. It is indeed so that the spatial heterogeneity is large for sediment in rivers,
and thus concentration of PCB (e.g. Bremle and Larsson 1998 and references
herein) in a 10-15 year period, accumulation areas within a river may change due
to flow events and man-made measures upstream. It is possible that the core
sampling locations in 1984 or earlier are not the same (i.e. that they don't show
identical conditions) as in 1994. There are no statistical techniques that help to
compensate for this (as stated in peer review questions). The techniques
(statistical analysis) used by the authors generally seem to be adequate. It is,
however, impossible for me to follow the statistical template used. The approach
should be very simple; data on PCB concentration in the sediment (e.g. pg/g dry
weight) or PCB mass per surface area of sediment (ug/dm2) from the defined area
are compared in two populations, one from 1984 and the other from 1994. The
comparison is limited by the number of cores taken at each sampling occasion,
and nothing else. As the populations are log/normally distributed, PCB
concentration data are log-transformed. A simple comparison test will now reveal
if the populations differ significantly or not, and the direction the difference (PCB
concentration 1984 > PCB concentration 1984). If significant (and only then), the
possible decrease between years can be calculated, as carried out in the present
report. I cannot elucidate if this was the case, as the statistical
approach/calculation pattern is not transparent (I cannot follow it from step to
step). A flow scheme of the statistical tests used would have been a great help.
Further, high spatial variability (or any other high variability) can only be
described properly by using a larger number of samples, reflecting the variability.
3. 2.3. The results from the "replicate" samples in 1994 from Tl pool show a 36%
average variability. At the same time, conclusions are reached that a 40% loss of
PCB has occurred from the Tl pool sediment from 1984 to 1994. I did a very
simple simulation, using the average 10 g PCB m'2 in sediment for Tl pool in
1984, a 40% decrease to 1994 and simulated all other data, obtaining a standard
deviation around 36% for the two data sets (the variation thus defined by the
standard deviation and n=19 for each data set). Understanding that this is a
major simplification, the results revealed that the decrease in PCB concentration
was significant (Student's t, p<0.001). So I cannot see any problems with the
conclusion, assuming that the number of samples underlying the analysis is large
enough (again I cannot follow the statistical testing).
16
-------
7
Table 1. A simulated data set for changes of PCB concentrations in Tl pool from
1984 to 1994 (N=19 for each sampling occasion in time).
Sample No Sediment 1984 Sediment 1994
(g/nr2) (g/m2)
1
10
6
2
15
11
3
8
4
4
7
3
5
14
10
6
9
5
7
8
4
8
9
5
9
12
8
10
12
8
11
11
7
12
13
9
13
6
2
14
6
2
15
8
4
16
9
5
17
17
13
18
4
0
19
8
4
Average 9.8 5.8
Standard 3.4 3.4*
deviation
*= higher than the 36% variation mention earlier
2.2.4. In the study of 1994, the calculated mass of PCB in the sediment of Hot
Spot 28 was 20 metric tons. Previous estimations resulted in a mass of 2 - 7
metric tons. In the present study there was no evidence of overall burial of "old"
sediment, <50% of the sediment core profiles. The only possible transport that
would result in a transfer of >10 metric tons of PCB in a sediment in a river is a
very large resuspension event in the upstream river system, transferring
contaminated sediment downstream from one hot spot and depositing the
sediment in Hot Spot 28 (with no similar transport to other areas). This is highly
unlikely. Another unlikely explanation is a direct PCB discharge to Hot Spot 28.
Ruling out these explanations the proposed one seems likely; the previous
calculations underestimated PCB mass.
17
-------
8
2.2.5. I agree to the conclusion that losses of PCB have occurred from Tl pool to
the river system downstream. Highly PCB contaminated sediment in rivers will
have a major impact on PCB concentration in water and aquatic biota within the
area of deposition and downstream (Brown et al 1985, Larsson et al. 1990). This
impact is caused by a transport of PCB over the sediment/water interface, i.e. a
loss of PCB from the contaminated area. The PCB loss from sediment need not to
cause any significant decrease of PCB in the sediment (decrease in
concentration), to cause substantial increase of PCB in water and biota. I do
agree that that losses have occurred from the Tl pool and from hot spots
downstream. If the magnitude of these losses over a 10 year-period exceed 10%
or more, this is caused by particle transport. Contaminated sediment particles are
resuspeded and transferred downstream at high discharge events. These events
are frequent in rivers. I do not conclude losses of this magnitude to be caused by
desorption of PCB from sediment to water, bioturbation or anaerobic
dechlorination.
2.2.6. Within a river, sediment is subjected to resuspension, transport and
resedimentation. As pointed out earlier, the extent of these processes depends on
the water discharge. As discharge follow a seasonal cycle, sediment transport
events are likely to occur during spring flooding. As the magnitude of flooding
may vary greatly in a decade and even more in a longer time span, the
transport/resuspension events occur irregularly. Being a dynamic system, there
are no true sediment accumulation areas in a river, all sediment mav be
transported downstream. Therefore, any burial of contaminated sediment is just
temporary. This is also shown in the LRC study.
2.2.7. I find the results from the side-scan sonar and the connection to particle
size distribution very convincing. It's a good approach for determining the extent
of temporary accumulation areas of organic sediment and thus areas with high
PCB concentration.
3. References
Anon 1998. The Jarnsjon Project, Sweden. Remediation of PCB contaminated sediments.
AMBIO 27, No 50, 373-426.
Baker, J.E., Capel, P.D., Eisenreich, S.J. 1986. Influence of colloids on sediment-water
coefficients of polychlorinated biphenyls in natural waters. Environmental Science and
Technology 20, 1136-1143.
Bedard, D.L. Unterman, R. Bopp, L.H., Brennan, M.J.Haberl,, M.L. and Johnson C. 1986. Rapid
assay for screening and characterization microorganisms for the ability to degrade polychlorinated
biphenyls. Applied and Environmental Microbiology 51, 761-768.
Berglund, O., Larsson, P., Bronmark, C.t Greenberg, L., Eklof, A. and Okla, L. 1997. Factors
influencing organochlorine uptake in age-0 brown trout (Saimo trutla) in lotic environments.
Canadian Journal of Fisheries and Aquatic Sciences 54, 2767-2774.
18
-------
9
Bremle, G., Okla. L. and Larsson. P. 1995. Uptake of PCBs in fish in a contaminated river system -
Bioconcentration factors measured in the field. Environmental Science and Technology
,29, 2010-2015.
Bremle; G. and Larsson, P. 1997. Long-term variation of PCB in the water of a river in relation to
precipitation and internal sources. Environmental Science and Technology 31, 3232-
3237.
Bremle, G., Larsson, P., Hammar, T., Helgee, A. and Troedsson, B. 1998. Fate of polychlorinated
biphenyls in a river system during sediment remediation. Air, Soil, and Water Pollution
107, 237-250.
Bremle, G. and Larsson, P. 1998. PCB in the air during landfilling of a contaminated lake
sediment. Atmospheric Environment 32, 1011-1019.
Bremle, G., Okla, L. and Larsson, P. 1998. PCB in water of a lake after remediation of
contaminated sediment. Ambio, 27, 398-403.
Bremle, G. and Larsson, P. 1998. PCB in Em3n River ecosystem. Ambio, 27, 384-392.
Bremle, G. and Larsson, P. 1998. PCB concentration in fish in a river system after remediation of
contaminated sediment. Environmental Science and Technology, 32, 3491-3495.
Brown, M.P., Werner, M.B., Sloan, R.J., Simpson, K.W. 1985. Polychlorinated biphenyls in the
Hudson River. Environmental Science and Technology, 19, 656-661.
Chevreuil, M. And Granier, L. 1991. Seasonal cycle of polychlorinated biphenyls in the waters of
the catchment basin of the River Seine (Frantee). Air, Soil, and Water Pollution 59, 217-
229.
Horzempa, L.M. and DiToro, D.M. 1983. The extent of reversibility of polychlorinated biphenyl
adsorption. Water Research 17, 851-859.
Jeremiason, J.D., Eisenreich, S., Baker, J.E. and Eadie, B.J. 1998. PCB decline in settling
particles and benthic recycling of PCBs and PAHs in Lake Superior. Environmental Science
and Technology 32, 3249-3256.
Lara, R. And Ernst, W. 1990. Sorption of polychlorinated biphenyls on marine sediments. 1. The
role of organic carbon content. Environmental Technology 11, 83-92.
Larsson, P., Okla, L., Ryding, S.-O. and Westob, B. 1990. Contaminated sediment as a source of
PCBs in a river system. Canadian Journal of Fisheries and Aquatic Sciences. 47, 746-
754.
Steen, W.C., Paris, D.F. Baughman, G.L. 1978. Partitioning of selected polychlorinated
biphenyls to natural sediments. Water Research 12, 655-657.
Williams. W.A. and May, R.J. 1997. Low-temperature microbial degradation of polychlorinated
biphenyls in sediment. Environmental Science and Technology 31, 3491-3496.
Zitko, V. 1989. Characterization of PCBs by principal component analysis (PCA of PCB). Marine
Pollution Bulletin 20, 26-27
19
-------
Dr. Keith Maruya
21
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
22
-------
Maruva
Specific Questions
Data Evaluation and Interpretation Report (DEIR)
I. Is the documented PCB load, which originated from the 77 Pool, consistent with a source
consisting of historically deposited PCB-contamwated sediments'
PCB contaminated sediments in the TI Pool were the most likely source of the water column
loads described during summer low flow periods in the 1993 Phase 2 Transect study. A look
at the corrected homolog distributions for the Transect study (Figs. 3-38. 3-40. 3-43 and 3-47 in
Appendix C, LRC Responsiveness Summary) shows that for various flow conditions, total and
dissolved PCBs greatly increase at the Tl Dam relative to Rogers Island In the absence of some
undiscovered source, this sharp increase in dissolved PCBs strongly suggests a local source.
Corrected estimates of PCB loading in the Upper Hudson are less consistent, w ith the bulk of the
loading occurring above the TI Pool in the early transects of 1993 (Figs C-6 and C-12
corresponding to Transects 1 and 4). switching clearly to the Tl Pool during later, low flow
transects (Figs C-14 and C-16 corresponding to Transects 5 and 6). Both instantaneous
("transect") and 15d flow-averaged data reveal consistent patterns in loadings.
Load is the product of concentration and flow rate Since it is assumed that flow rate increases
with decreasing river mile (i.e. as one heads downstream), it follows that water column PCB
concentrations in the river must decline after the TI Pool during summer low flow conditions
Conversely. PCB concentrations upstream of the TI Dam. and in particular in the TI Pool itself,
must increase more rapidly than flow rate to sustain the consistent increase in loading observed
Plotting of total and dissolved PCB data for transect 6 illustrates this clearly (Fig Dl-1 below).
The origin of the PCBs w ithin the TI Pool sediment inventory and what is meant by "historically-
deposited" is less clear From the LRC study, it is clear that "shallow " layers (< 10 in depth) still
contain a large PCB inventory at many TI Pool locations Whether the water column PCBs
originated from deeper, historically contaminated layers or recently deposited sediments cannot
easily be determined My ow n analysis of the LRC data indicates that sediment PCBs are split
roughly 50:50 between shallow (0-8cm) and deep (>8 cm) layers (see also my comment on LRC
Question 6) Each layer is dechlonnated. deeper layers probably being more so
Another confounding factor that complicated interpretation of likely PCB sources, especially for
the early 1993 transects (winter low flow and early spring high and transition flow), was the bulk
release of Aroclor 1242-like PCBs from the Allen Mills source, reported to have ceased sometime
during the middle of 1993. A major source of PCBs during this period was clearly the stretch
above Rogers Island Post 1993 water column data would have been extremely helpful, however,
they were never put into a single, coherent presentation that this reviewer could comment on
The only alternative explanation for these profiles is selective leaching of mono- and dichloro
homologs from relatively immobile free product (oil droplets) upstream of the TI Pool This
would also require enhanced or accelerated "dispersion" of these homologs away from the source
and a corresponding attenuation of trichloro- and higher homologs ("Tn+") in sediments
upstream of the TI Pool. Tins scenario, however, would have been observed as a gradual shift in
the homolog profile betw een the upstream Allen Mills source and the head of the TI Pool (Rogers
Island) Since the sediments and water column samples collected at Rogers Island had a very
different homolog pattern than that w ithin the TI Pool (Figs C-2 through C-4. and 3-38. 3-40. 3-
43 and 3-47). this scenario is not a very likely one
23
-------
Review of Phase 2 DEIR
Maruya
Total
Dissolved
Part.
o _j
^ o>
CD £
I
FB Rl TID Sch Wtfd GIB
Sample location
Fig. D1-1. Water Column PCB Concentrations for Summer-Low Flow Conditions (Transect 6)
-------
Maruya
2. Are the two-phase and three-phase partitioning coefficients, derived in the DEIR. appropriate
and do they properly address the physical parameters of the system (i.e. temperature)?
Based on its magnitude and trend relative to the octanol-water partition coefficient (Kow),
estimated values of the particulate organic carbon partition coefficient (Kpoc) appear to be
reasonable, but estimates of partitioning to dissolved organic carbon (KDoc) should be
viewed with skepticism and used with caution. The background discussion on two- and three-
compartment partitioning theory and pertinent relationships/equations presented in section 3 I of
the DEIR are scientifically and mathematically sound. It is obvious, however, that in situ water
column partitioning ratios reported in the Phase 2 study varied by orders of magnitude for the
same congener, even when normalized to particulate organic carbon (POC) (see DEIR. Table 3-
6a). A temperature correction was invoked to further help explain and account for this
variability, the effect of which was consistent with experimental determinations (Warren et al.
1987). Estimates of Kpoc and Kdoc were thus made based on attempts to fit available in situ data
into equilibrium, mass balance and temperature correction relationships
These estimates are presented for partitioning in the water column in Table 3-8 of the DEIR. and
appear again as corrected values for bedded sediments (i.e. sediment and porewater distributions)
in Table 2-2 of Book 3 of the DEIR Responsiveness Summary In Tabic 2-2. K values are given
for coeluting congener combinations with Kpoc > Knoo w ithout exception. My own analysis of
the data show that log K's are positively correlated with log Kow (/><0 01; sec Fig. D2-I and Table
D2-1 below). For New Bedford Harbor sediments, positive Kp0c-KoW associations were also
noted by Burgess et al. (1996) and Brownawell and Farrington (1986). However, in Table 3-8.
one curious exception is noted for BZ#8. where the estimated KPOc (5.19) is less than Kdoc
(5 43). No explanation is given for this unique trend reversal, other than the fact that footnote "c"
in Table 3-8 indicates some sort of blank problem. It is not clear if the congener specific results
in Table 3-8 will be used in the future. If they are. I suggest that values for BZ#8 be revisited. If
the data for BZ#8 turn out to be of questionable quality, interpolation or regressions based on
aqueous solubility or K<,w. should be considered to estimate K values for this congener For any
future predictions of K, a quick check of the trend with increasing chlorines (or log Kou,) can be
performed as a quick sanity check of the model.
On a more general note, it is not clear why all the effort was made to model the effects of DOC.
especially since it w as noted in the original version of the DEIR that DOC w as fairly constant at
-5 mg/L in the water column. If these data were accurate, then one can conclude that DOC in the
water column would exert a relatively constant and predictable influence in terms of partitioning.
A case can be made that DOC in porewater of fine-grained sediment, although higher, w ould be
relatively constant as well In fact, in the example given in the DEIR Responsiveness Summary,
Book 3. DOC was estimated (a) 34 mg/L. -10 times higher than that measured/assumed in the
water column. The high degree of uncertainty in modeling the effects of DOC make it more of a
mental exercise and less of a practical tool for modeling The danger also exists that "ov er-
manipulation" of model parameters such as partitioning constants can be abused to better fit in
situ values.
I w ould thus recommend simplifying the partitioning model back to a 2-phasc model instead of a
3-phase model. In the 2-phase model. DOC would be included as a more-or-less constant
influence in the apparent or operationally defined dissolved phase.
25
-------
Review of Phase 2 DEIR
Maruya
M
as
o>
O)
o
o
Kdoc
Kpoc
ideal
o o
ft*
; •• • --
#
* %
<9 °°° °
5 6
log Kow
Fig. D2-1. Estimated Phase 2 log K's show increasing trend with log Kow (Hawker and Connell 1988)
-------
Review of Phase 2 DEIR
Maruya
SUMMARY OUTPUT, log Koc vs. log Kow (H&C)
Regression Statistics
Multiple R 0.653364
R Square 0.426885
Adjusted 0.382799
Standard 0.490518
Observatio 15
ANOVA
df SS MS F gnificance F
Regressio 1 2.329824 2.329824 9.683055 0.008258
Residual 13 3.127909 0.240608
Total 14 5.457733
Coefficient andard Err t Stat P-value ower95% pper95%ower95 0 pper95 0
Intercept 2.76122 1.0368 2.663213 0.019515 0.52135 5.001091 0.52135 5.001091
X Variable 0.55038 0.176871 3.111761 0.008258 0.168274 0.932487 0.168274 0.932487
SUMMARY OUTPUT, log Kdoc vs. log Kow (H&C)
Regression Statistics
Multiple R 0.733528
R Square 0.538063
Adjusted 0.50253
Standard 0.416725
Observatio 15
ANOVA
df SS MS F gnificance F
Regressio 1 2.629623 2.629623 15.14239 0.001856
Residual 13 2.257577 0 17366
Total 14 4.8872
Coefficient andard Err t Stat P-value ower 95% pper95%ower95 0 pper95 0
Intercept 1.088098 0.880824 1.235317 0.238572 -0.81481 2.991003 -0.81481 2.991003
X Variable 0.58472 0 150263 3.891322 0.001856 0.260098 0.909343 0.260098 0.909343
Table D2-1. Linear regression indicates statistically significant (p<0.01)
positive correlations between log K and log Kow
27
-------
Manna
3. Are the conceptual models based on the transect sampling consistent w ith the data7
The overall conceptual model invoked to explain the 1993 transect data is consistent with
the corrected data/profiles contained in the LRC Responsiveness Summary, Book 1,
Appendix C. This basic model, characterized by the reviewer in Fig. D3-1 below, translates the
PCB load picked up in the upper reaches of the Upper Hudson via advection Translation of
PCBs into the w ater column of the T1 Pool is expressed by a combination of sediment porewater
exchange and advection/diffusion across the sediment-water interface, and exchange between
resuspended sediment and the water column. This combination of processes is purported in the
DEIR Responsiveness Summary, Book 3. to be consistent with the mono-, di- and tnchloro
homolog distribution observed for dissolved phase PCBs Sediment porewater exchange was also
found to be an important mechanism for PCB loading into the lower Hudson (Achman et al
1996) Losses dow nstream of Rogers Island, and in particular downstream of the TI Dam. are
suggested to be the result of gas exchange and/or aerobic degradation
Because of their phvsicochemical properties including aqueous solubility and Henry s Law
constants (see Fig. D3-2 below), mono- and diCBs will preferentially partition into the dissolved
phase and into the atmosphere relative to heavier homologs If air above the Hudson River is
undersaturated. a constant flux of PCBs to the atmosphere can act to continually pull PCBs from
the source (sediment) into the water column so that a concentration gradient is maintained
More limnological data (e g residence times; chlorophyll a. temperature gradients/stratification
during summer low flow conditions) w ould likely improve our understanding of the processes at
work in the TI Pool For example, this stretch would seemingly act much more like a lake than
upstream and/or narrow er stretches of the Upper Hudson As such, the processes acting on PCBs
that are enhanced under lake-like conditions will be most important The larger surface area to
volume, increased residence time and quiescent flow regimes would likely result not only in
greater fluxes of PCBs out of sediments and into the water column, but also into the air! Mass
transport betw een phases is dependent both on the magnitude of the concentration gradient and
surface area The increased dissolved phase PCBs thus provide a larger driving force for
transport out of the w ater column into the air This flux would increase under summer, low -flow
conditions w hen both air and water temperatures are at their annual maximum
28
-------
c
r\
•
> f
d
>
¦3
">
0
~T3
c5
ST
c5
O
C
C
o
A
ci &
cD
O r--
S
oj £
~§ I
©
*5"" Tj
I
rT
Q
a
29
-------
Review of Phase 2 DEIR
Maruya
u»
O
~ HLC
« Cw,sat
6.E-04 ^
re 2 4.E-04 *
~ * V
E
O co
_1 £
~ o
~
2F-04
¦ hi
0.E+00
0
o o ^
50 100 150 200
IUPAC No.
1.E+02
1.E-01
O)
(0
1.E-04 »
4 :«t
o
1.E-07
Fig. D3-2. Aqueous solubilities and Henry's Law constants (from Dunnivant et al. 1992)
-------
Maruya
4. Does the sampling at the 71 Dam-West location impact EPA 's conclusion that the sediments
of the TJ Pool are the ma/or source ofPCRs to the freshwater Hudson during low flow-
conditions considering the analytical corrections made to GE s PCB data'' What are the
other implications of finding higher concentrations along the shoreline than in the center
channel?
Clearly, accurate estimation of PCB loads, and changes thereof, in the fashion adopted by EPA.
relies on representative sampling of River conditions. Lateral and vertical heterogeneity in water
column PCB concentrations result in estimations with greater uncertainties If the sampling
station at the TI Dam renders artificially high concentrations, loading will be exaggerated:
conversely, if the station is isolated in a mainstream channel or at a depth w here concentrations
are artificially low. loadings will be underestimated.
As thoroughly stated in Book 3 of the Responsiveness Summary for the DEIR. any sampling
bias imparted due to systematically high PCB concentrations measured at TI Dam West
location were more than offset by the underestimation of water column PCB
concentrations The analytical bias was deemed to be 40% on average whereas the sampling
bias was 40% maximum (low-flow, low concentration '<33 Rogers Island), a condition that was
deemed to exist in 2 of the 7 years that data was collected (1991-97). These biases appear to be
mostly a wash, and loadings stated in the original DEIR report appear to be correct in
relative, if not absolute magnitude.
If water column (and sediment) PCB concentrations are higher in nearshore areas, several
implications to both modeling and monitoring efforts can be envisioned First, spatial coverage
and resolution criteria for estimating PCB mass inventory need to be revisited Specifically, are
"hot spot" near shore areas represented w ith sufficient spatial resolution'1 If the higher nearshore
PCB concentrations were not accounted for in the Phase 2 kriging analysis, an underestimate of
total PCB inventory may have resulted Second, what is the degree of mixing between shallow
and center-channel regimes of the TI Pool? Third, w hat are the airborne losses associated with
shallow vs. deep portions of the Pool? Fourth, what is the net suspended sediment deposition rate
in these areas? Are these fine grained near shore deposits/sediments subsequently scoured and/or
resuspended during "first flush" Spring melt flow events?
31
-------
Maruva
5. Are the geostatistical techniques (polygonal declustermg and kriging) correctly applied7
I am not qualified to give a detailed critique of these methods; however, I will offer the
following general comments.
The polygonal declustermg method described in section 4 .2 .3 of the DEIR resulted in the creation
of Thiessen polygons of vastly unequal area (Plates 4-5 through 4-9 in the DEIR) Also, spatial
correlation between adjacent sampling points is complicated by the high degree of heterogeneity
in PCB levels, presumably also coupled to profound differences in sediment texture. Total mass
inventory estimates from the LRC based on this first order method was 19 6 MT (DEIR. p 4-34).
16% less than the original 1984 assessment
To better account for uncertainty associated with large unsampled areas, a geostatistical method
known as kriging was applied. A semi-vanogram function is used to represent the degree of
"continuity" between PCB concentrations of adjacent sampling points This approach was not
successful for the full dataset but was deemed successful for various sub-reaches ("chopped up
segments") of the T1 Pool using a "block kriging" approach. An estimate of 14 .5 MT resulted
from this analysis. 38% less than the original 1984 estimate.
Clearly, estimates based on polygonal declustermg are considered "conservative" (i.e. upper
bound) whereas the lower estimates based on the kriging analysis are probably more accurate
However, the importance of nearshore areas with higher than expected PCB concentrations is
unclear Perhaps, a detailed assessment should be made on a short, spatial scale in two or three
different reaches with (a) varying sediment texture and (b) significant nearshore PCB inventories,
to "verify" the results of the kriging analy sis (see also comments to DEIR Question 4 and General
Question 2)
32
-------
Manna
6. Are the methods applied in the DEIR (change in molecular weight (MW) and evaluating
concentrations of BZ-s I. 4. 8. 10 and 19 (MI)PR) appropriate standards for determining
extent of dechlorination? Are there any significant problems with this approach, or more
appropriate approaches
The change in MW (AMW) and molar dechlorination product ratio (MDPR) appear to be
appropriate quantities for tracking the extent of dechlorination in Upper Hudson River
sediment samples. The accumulation of orf/jo-substituted mono- and dichloro congeners (i.e.
BZ#U 4. 10. and 19) is consistent with the pattern of (reductive) dechlorination observed in this
environment as reported in several papers (e g Bedard et al. 1996) and summarized in Bedard
and Qucnsen III (1995) For example, it was noted that dechlorination patterns B. B" and C in
Hudson River sediments were enriched in 2-chlorobiphenyl (BZ#1). 2.2 - and 2.6-
dichlorobiphenyl (BZ#4 and #10), and 2.2".6-and 2.3".6-trichlorobiphenyl (BZ#19 and #27)
BZ#8 is not a strictly ortho substituted congener so its inclusion in this ratio may be superfluous
That AMW and MDPR track well together is an additional indication that their representation of
dechlorination is consistent
There appear to be no significant problems associated with this approach for Upper Hudson River
sediments There are. however, alternate approaches as outlined in Quensen 111 & Tiedjc (1997)
that also give measures of the extent of meta- and para dechlorination. These methods require
congener specific data which for the most part are available for the Phase 2 study. The first of
these plots the average number of meta and para chlorines vs. ortho chlorines. This is compared
to the unaltered source material (in this case e.g. 90-95% A1242; 5-10% A1254) and the vertical
distance between the altered and unaltered source data point is a direct indication of the extent of
dechlorination The second method is the creation of charts, subtracting mole fractions of
individual congeners in a dechlorinated sample from the original or starting mixture Individual
molar increases and decreases should balance if dechlorination is the only transformation process
involved
It is worth noting that although EPA/TAMS justified in detail the use of AMW and MDPR these
parameters as overall good indicators of dechlorination, they chose to disregard a large portion of
the LRC data where these parameters appeared to indicate "widespread" dechlorination in
samples w ith lower total PCBs that were primarily from deeper core segments (see also
comments on DEIR Question 7)
33
-------
Maruva
7. The DEIR finds that the degree of anaerobic dechlorination is primarily a fmcl ion of
original concentration rather than time, and accordingly that there is not significant
predictable dechlorination in sediments containing less that approximately 30 mg kg PCB Is
this reasonable?
Based on academic investigations into PCB dechlorination, and the selectivity with which a
good deal of Phase 2 data was discarded from dechlorination analysis, I do not agree with
this conclusion. The DEIR and LRC both show evidence that suggests that rates and extents of
anaerobic reductive dechlorination in Upper Hudson River sediments are a function of PCB
concentration However, data from deeper sediments w ith lower concentrations that also
appeared to show evidence of dechlorination were thrown out of the analysis. The reason given
for excluding these data (constituting -40% of all core data) was cross-contamination resulting
from the core slicing procedure. Inspection of the entire LRC core data (see Figs. 3-2 and 3-3
of the LRC), however, suggests that there is no relationship between MDPR or AMW and
total PCBs.
Abramowicz et al. (1993) showed that measurable dechlorination occurred in Upper Hudson
River sediments collected near Ft. Edward (RM 194) that contained total PCB concentrations as
low as 20 mg/kg Although most laboratory investigations into PCB dechlorination utilize
relatively high concentrations due to time constraints, there is no clear scientific basis for a single
threshold
In fact, there are many chemical and biological factors which are thought to impact rates and
extents of in situ dechlorination. These include absence of oxygen, nutrient, carbon and electron
donor availability and quality, PCB availability, organic carbon content and quality, and the
presence of the appropriate microbes (Mohn & Tiedje 1992) Temperature has also been shown
to affect not only rates and extents, but also patterns resulting from PCB transformation under
anaerobic conditions (Wu et al 1997) Whereas there is no doubt that higher PCB concentrations
will increase the likelihood of PCB availability through porewater. given equal TOC. the order of
magnitude heterogeneity observed for in situ partition coefficients reported in the DEIR (see also
comments for DEIR Question 2) suggests that this generalization cannot be made with a high
degree of confidence
34
-------
Maru\a
Low Resolution Sediment Coring Report (LRC)
1 In the LRC. EPA compared sediment data from cores taken in 1977. 1984 and 1994. which
had the PCB analysis conducted by di fferent laboratory methods. How valid are the methods
used to establish a consistent basis for comparison 's
The method described to correlate 1984 "total Aroclor" PCB concentrations with 1994
Phase 2 "congener specific" data appears to be valid. However, this is true only for £Tri+
homologs. No satisfactory method was given to correlate 1977 data with other years or to
estimate mono- and dichlorohomologs for earlier periods. Clear biases exist among data sets
arrived at using different PCB analytical methods. The 1976-78 and 1984 data sets tracked the
response of a limited number (3) of packed column peaks in Aroclor standards, which ignored
mono- and diCBs and thus underestimated their contribution. The 1994 LRC was based on
congener specific analysis, the preferred and most accurate method to date. In Appendix E of the
LRC. a linear correlation method was described to estimate "tri+" values from 1984 total Aroclor
concentrations For the "as if' analysis of 1994 Phase 2 high resolution core data using the
Aroclor method. Fig 2 of Appendix E show s that a strong correlation exists for Itri+
concentrations (r: = 0.983).
If PCB congener profiles had changed drastically between the 1984 and 1994 sampling, this
correlational method would probably not yield adequate results. Since the composition of the
major PCB contaminant. Aroclor 1242 in this case, is dominated by lower to mid-range
congeners in term of chlorines per biphenyl molecule, the resultant degraded mixtures arc
potentially less complex and/or variable than would be for heavier Aroclor sources (e g 1254 and
1260) Another indication that this estimation scheme did not impart a significant bias is the fact
that both PCB mass inventory losses and gains were determined
Concerning 1976-78 data for hot spots below the TI Pool, it w as stated on p.27 of Book 3 of the
DEIR Responsiveness Summary that "The 1977 (USGS) sediment data arc also suspected to
approximate a sum of tri- and higher-chlorinated congeners, but may have a small upward bias
relative to the 1984 results due to the use of Aroclor 1016 standard rather than an Aroclor 1242
standard Unfortunately, surviving documentation of this analytical effort does not appear to be
sufficient to definitively establish exactly what was measured in 1977 " However, on p 4-27 of
the DEIR. it was stated "The three peaks used were the same ones used by O'Brien and Gere for
Aroclor 1016 in the 1978 analysis." Based on this discrepancy and the general lack of
information/analysis, I cannot comment on the comparability of 1977 data.
Clearly, a major limitation of earlier PCB datasets is the absence of mono- and dichlorobiphenyl
homolog data Based on Table A-7 in the Responsiveness Summary for the LRC. these
homologs account for - half of the entire inventory on a molar basis. In terms of assessing
changes in PCB inventory, it is presumed that mono- and diCBs would be most mobile and losses
to compartments not accounted for in Phase 2 DEIR measurements would be potentially the
greatest. Then there is the impact of dechlorination which over a 16-18 year period might be
expected to be significant. If a substantial fraction of the 1976-78 ZTri+ PCB mass was
subsequently dechlorinated (and not desorbed). the "non-change"' would be reflected only in the
1994 total PCB estimates, and not the STri+ This no-change in truth would would be detected as
a loss based on comparison of £Tri+ Losses between 1976-78 and 1994 were in fact reported in
the LRC general conclusions (see also comments to LRC Question 3) As repeatedly asserted
throughout Phase 2 reports, this dechlorination change w ould be limited to 25% of the total PCb
mass for Aroclor 1242; however this is still a significant fraction that could be unaccounted for
35
-------
Manna
2. In the upper Hudson River system, it has been well established that there is significant lateral
heterogeneity in sediment concentrations. How does sediment heterogeneity affect the
comparsion of cores from two different years? Given the spatial variability, is the finding
that there is loss from most of the locations supported by the data?
Based on the general agreement between the loss amounts stated in the original LRC
analysis ("point-to-point comparison") and the Reassessment analysis ("duster area"), the
losses appear to be supported by the data. The original LRC point-to-point comparison
resulted in a 39% mass loss for sediments with higher PCB inventories (> 1 Og PCB/m2) The
Reassessment included regression and ratio-based analyses to estimate mass loss using the mass
per area (MPA) parameter for PCB inventory. The regression analysis resulted in a mean mass
loss of 59 + 19 percent; the ratio-based analysis resulted in a mean mass loss of 45 percent (95%
confidence range: -59 to -4 percent) A small correction (-5%) was also made for
dechlorination
Although spatial and temporal variability in sediment PCB concentrations are of concern, it is
prudent to utilize as much of the collected data set as possible w hen making conclusions
Sediment texture (grain size) and total organic carbon (TOC) data are two classic examples of
parameters that are typically correlated with the concentrations of particle reactive, hydrophobic
contaminants like PCBs (LRC. Figs 3-20 and 3-21). The grouping of core data into 14 cluster
areas served to eliminate sampling location bias caused by short scale spatial heterogeneity that
could lead to very large errors when comparing data on a "point-to-point"" basis, as was done in
the original LRC analysis. In fact. 11 of the 14 cluster areas analyzed in the Reassessment were
determined to be largely fine-grained sediments (LRC Reassessment. Book 1. Appendix A. p A-
5)
Thus, in all likelihood, the effect of spatial heterogeneity of PCBs is associated with differences
in sediment types as measured by these parameters In other words, as long as the basic
sedimentological parameters were similar in cores collected from the same location. I would not
expect sediment heterogeneity to impart significant comparative errors, as supported by the
general agreement of mass change estimates from point-to-point and cluster area estimates
36
-------
Maruva
3. What is the impact of the difference between replicate samples in the 1994 sampling e ffort
(36 percent average variability) on the finding that there was a 40 percent loss of PCB
inventory from the highly contaminated sediment in the 77 Pool9
The LRC (p 2-18) reports an average relative percent difference (RPD) of 36 for total PCBs in
core field splits The formula for calculating RPD is also given on p.2-18. Total PCB
concentrations in a given core segment arc then used to calculate mass per unit area (MPA)
estimates and length weighted averages (LWA) (LSR, eqs.4.1-1 and 4.1-2. p.4-3) Thus any error
in the concentration term (C.) are directly translated into these estimates. In fact. EPA/TAMS
found that absolute changes could not be assessed because of this uncertainty. As a result, they
switched to relative measures of mass change The relative percent change in PCB inventory (A)
is calculated according to Eq 4.1-4 (p.4-6):
A = KMPAw - MPA84)/MPA84| * 100%
In this representation, uncertainties in MPA estimates are now applied in both the numerator and
denominator
A problem with accepting a 40 percent decrease in sediment-associated PCB mass (actually 39%.
p.4-17) would arise if one believed that the 36% uncertainty reported w ere primarily of a
systematic nature In other w ords, if the 36 percent uncertainty was consistently applied as an
underestimation of 1994 data relative to the 1984 estimates, then little or no change in PCB
inventory could be concluded However, there appeared to be little/no evidence of extreme
systematic bias in the 1994 data and so it can be assumed/concluded that these uncertainties
would be expected to be somewhat random (i.e. an equal chance for underestimation and
overestimation) This is supported in this case by inspecting the regression slopes for field split
pairs for BZ#52 shown in Fig 2-6 of the LRC w here 11 slopes were less than unity and 10 w ere
greater than unity This suggests that on average field duplicates were biased high with the same
frequency that they were biased low As such, the tendency for mass losses to be real is not
compromised
I would like to point out that the confidence intervals around the reported 40-50 percent mass
iosses arc substantial and reflect the uncertainty in the mean estimates. Reporting single values
without mention of their rather large uncertainties in this case is misleading and should be
avoided Thus, I do not feel that the 36 percent variability between replicates invalidates the
conclusions of mass losses; however, the predictions of mass losses should be represented as
ranges instead of single "mean" values to avoid misrepresentation/misinterpretation.
37
-------
Marina
4. In the LRC. it was found that Hot Spot 28 contained much more mass than previous
estimates. Is the conclusion that this gam is primarily due to incomplete characterization in
1977 valid?
This conclusion appears to be the only plausible explanation for the substantial inventory
increase and is supported by at least 2 pieces of evidence. First, there were widely divergent
PCB concentrations for co-located cores (LRC. Plate 4-22). For example, the 1994 core
exhibiting a length weighted average (LWA) PCB concentration of 1184 mg/kg was co-located
next to the NYSDEC 1976-78 sediment grab sample whose reported mean PCB concentration
was 176 mg/kg. almost one order of magnitude less Second, several 1994 cores with significant
PCB levels were collected in areas apparently not sampled in the 1976-78 survey Two examples
of this are the cores with LWA PCB concentrations of 680 and 267 mg/kg near the center of the
large area of fine-grained sediments It is also possible that accretion over the past 15 years has
greatly increased the mass inventory at this location, or that inventories have shifted within the
Hot Spot itself These scenarios are not likely to have occurred based on the ' stability" of
location of most other major hot spots (see core profiles in the LRC, Appendix D)
38
-------
Maruva
5 Does the data set and its interpretation support the conclusion that significant losses have
occurred from hot spots below T1 Dam''
Based on the significant inventories in shallow sediments and the potential difference/bias in
PCB quantification, I am more inclined to believe the "losses" reported from these hot spots
more than I would "gains". PCB inventory losses were reported for Hot Spots 31, 34 and 37:
gains were reported for Hot Spots 28 and 39 (LRC. Table 4-10) Based on the PCB concentration
profiles in Appendix D. I created a simple spreadsheet to estimated relative PCB inventories in up
to 3 core horizons, i.e. the "shallow" (roughly 0-1 Oin); "second" (8-20in) and "third" (>20in) core
segments (see also LRC. Fig. 2-2). Based on my calculations, which simply sum the product of
average core segment concentrations and the corresponding core length. 48% of the PCB
inventory in sediments from these Hot Spots resides in the surface (0-1 Oin) layer (Table L5-1)
Thus it can be concluded that a large fraction of the PCB inventory associated with bedded
sediment is not buried "deeply " and is available for resuspension and advection downstream,
resulting in a net mass loss if replenishement from upstream sources did not keep up w ith losses
to the water column (and bey ond)
How ever, many of the cores collected were incomplete, particularly for Hot Spots 34. 37 and 39
In the LRC. it is acknowledged "PCB estimates derived from these incomplete cores probably
underestimate the actual sediment inventory in the affected cores by less than 50 percent" (LRC.
p 2-17). The other factor to consider for all PCB mass change estimates (between 1984 and
1994) is the presumed underestimation of PCB sediment inventory in the 1976-78 NYSDEC
study This underestimation is due to the use of packed column GC analysis and quantitation
based on total Aroclors and would serve to increase confidence in mass loss estimates relative to
those that concluded a mass increase
39
-------
Review of Phase 2 LRC
HS#
25
28
31
34
35
a
64
108
179
4225
b
186
270
3840
46
1100
c
64
14050
41
532
d
51
183
45
50
264
e
1500
17500
1430
392
4
f
40
9500
44
7
g
950
1250
693
40
h
5900
15960
20
57
I
860
12380
290
290
j
2700
2655
130
3520
k
3890
556
I
1032
672
m
16660
51
n
9830
0
sum
12315
105160
6600
5901
6125
mean
1232
8089
733
454
1225
0-10
9772
51656
3936
2134
5317
10-20
2503
42796
2324
3277
716
>20
13
10683
1373
446
88
check sum
12288
105134
7633
5857
6121
%top
80
49
52
36
87
%interm
20
41
30
56
12
%bottom
0
10
18
8
1
o
Maruya
39
41
42
43
44 sum %
430
5
10430
194
198
8
919
4620
84
22
116
36
444
345
20
1000
276
1080
3905
604
2665
2355
2396
1070
2160
418
251
16
18474
960
15770
623
240
184936
1232
320
3943
208
80
1698
3555
887
1472
377
82
89016
48
9280
73
13014
218
84
76868
41
5639
1285
28
72
19981
11
18474
960
15770
622
238
185865
100
sum %
19
92
9
61
34
50
8
83
35
35
31
0
8
4
30
37
359
120
212
200
150
20
1650
168
56
63
9770
12768
1161
9831
2582
354
12767
77
20
3
Table L5-1. Spreadsheet summary of PCB inventory in low resolution cores from downstream Hot
Spots.
-------
Maruva
6. The LRC found that the historically contaminated sediments in the 77 Pool were not
universally being buried and sequestered from the environment. How much confidence
would place in the LRC evidence against widespread burial'>
Since J don't know what is meant by "historically contaminated sediments". I chose to answer
instead "Are significant concentrations/inventories of PCBs available in the shallow layers of
sediments in the TI Pool?" And my answer to this question based on the data provided in
the LRC is yes. Based on the PCB concentration profiles in Appendix C. I created a simple
spreadsheet to estimate relative PCB inventories in up to 3 core horizons, i.e. the "shallow"
(roughly 0-1 Oin): "second" (8-20in) and "third" (>20in) core segments (see also LRC. Fig 2-2)
Based on my calculations, which simply sum the product of average core segment concentrations
and the corresponding core length. 58% of the PCB inventory in TI Pool sediments resides in the
surface (0-10in) layer (Table L6-1) Thus I conclude that a large fraction of the PCB inventory
associated with bedded sediment is not buried "deeply"
41
-------
Review of Phase 2 LRC
Core zon
1
2
3
4
5
a
400
320
0
800
2400
b
200
1040
1120
5500
1100
c
1360
0
8800
3860
d
1120
3500
9200
e
12500
T
sum
3080
1360
1120
18600
29060
mean
770
680
373
4650
5812
0-10
1400
320
370
15200
9209
10-20
1700
1040
370
3400
17892
>20
0
0
370
0
1959
check su
3100
1360
1109
18600
29060
Core zon
10
11
12
13
14
a
3400
6840
207
330
72
b
2200
2160
138
80
60
c
2430
1800
90
90
244
d
1530
660
220
e
f
182
¦
sum
9560
10800
1277
500
596
mean
2390
3600
255
167
149
0-10
6365
2750
538
386
474
10-20
3195
594
668
56
122
>20
0
7434
68
56
0
check su
9560
10778
1274
498
596
Maruya
7
8
9
10
11
240
130
611
3400
6840
324
264
0
2200
2160
1280
1050
1060
2430
1800
108
90
200
1530
252
450
290
1952
1786
2611
9560
10800
488
357
435
2390
3600
672
45
740
6365
2750
1280
1741
1871
3195
594
0
0
0
0
7434
1952
1786
2611
9560
10778
16
17
18
sum %
243
1400
960
490
360
6200
70
176
1125
476
1360
7200
259
0
2820
1538
3296
18305
111116
308
659
3661
1538
2596
17743
63809
58
0
700
563
37295
34
0
0
0
9991
9
1538
3296
18305
109873
101
6
450
4000
0
4450
1483
2450
2000
0
4450
15
290
25
200
710
1225
306
1014
104
104
1222
Table L6-1. Spreadsheet summary of PCB inventory in low resolution cores from the Tl Pool.
-------
Maruva
General Questions
/ Is the data set utilized to prepare the DKIR. LRC and Responsiveness Summaries su fficient to
understand the fate and transport of PCBs in the Upper Hudson'''
The simple answer to this ambitious question is no. Understanding the "fate and transport' of
PCBs in the Upper Hudson requires an extensive multimedia investigation An potentially
important medium that has been ignored (i.e. no data collected or presented) is the gas phase I
would suspect airborne measurements of the lighter, more volatile PCBs might help with closure
of the mass balance (see also comments to DE1R Question 3) Because these homologs are
expected to be "more mobile" as the Phase 2 reports repeatedly point out. they are also more
prone to partition into the gas phase, and leave the aquatic system entirely. This would be
especially pertinent during warmer (low-flow) months, and possibly in slower-moving, larger
surface area to volume quiescent pools such as the TI Pool Evidence supporting the importance
of this mechanism are the revised water column loads in Appendix C of the LRC Responsiveness
Summary ("However, both of these events show a marked decline in the fraction of monochloro-
homologue(s) between the two stations, representing about a 50 percent loss", p C-l 1) The 2
events referred to are transects 2 (May) and 3 (June) and the 2 stations are the TI Dam and
Waterford Loss of PCBs to the air is never estimated or quantified.
Another problem with Phase 2 comparisons of PCB mass inventory is the lack of mono- and
dichloro homolog data for earlier years. According to McNulty (1997) as reported in the LRC
Responsiveness Summary (Table A-7). roughly 40-50 percent (molar or mass basis) of PCBs in
fine-grained TI Pool sediments are mono- and diCBs. However, for analytical reasons,
comparisons could only be made for "Tri+" homologs. There are obviously competing processes
that would determined the net shift, if any. in congener patterns over several years. Among these
are dechlorination (shift toward lighter PCBs) and degradation (shift toward heavier PCBs).
selective "weathering" (dissolution, diffusion/advection away from sediments; shift toward
heavier PCBs). air-water partitioning (mentioned above; shift toward heavier PCBs'') and
particulate-mediated transport (resuspension. scouring; no shift) These competing mechanisms
are consistent w ith the advertised loss of PCBs from the sediment inventory, and also the
maintenance of a relatively stable or declining water column load, if that is indeed w hat has
happened since 1993
Another limitation of this study was the collection and analysis of water column loading
data for a single year (1993), a year that unfortunately was subject to transient upstream
inputs Clearly, loadings resulting from the Allen Mills source influenced PCB loading prior to
June of 1993. and probably for at least several months thereafter assuming a stepw ise transport
downriver. Thus, it was difficult to assess the nature and degree of loading in the TI Pool, at least
during the early months (winter low flow and spring flood conditions). Thus, water column
monitoring data from post 1993 years would clearly be helpful (see also comments for General
Question 2)
These limitations aside, I do believe that the data presented in the Phase 2 DEIR and LRC
reports are adequate to (I) identify stretches of the Upper Hudson where major PCB
loading to the water column occurred under summer low flow conditions, including as an
example the TI Pool; and (2) suggest mass losses from many of the Upper Hudson hot spots,
including the TI Pool. The major questions arc thus shifted from "Where are water column
PCBs coming from''" to "How long will these Hot Spots persist '" and "What is the ultimate fate
of the PCBs introduced into the water column from these locations'.'"
43
-------
Maruva
2. Are there any additional analyses that should be done to verify certain findings of the DEIR
and LRC?
Several additional analyses may help verify the general conclusions of the DEIR and LRC. The
first has to do with "validating" the 1993 water column/PCB loading study. Reference was made
throughout parts of the Phase 2 reports and the Reassessment Summaries concerning post 1993
water quality monitoring data, mostly compiled by GE. However, no coherent analysis and/or
summary of this data was included. If enough data exist, it would be extremely useful to see if
the same general water column PCB concentration and loading trends, particularly during winter
low flow conditions, are observed for years where transient upstream inputs of PCBs (e g. the
Allen Mills source of the early 1990s) were absent/minimized.
The second analysis has to do with assessing the significance of elevated "neashore" PCB
contamination. It was not clear to me whether these areas were considered in the DEIR
geostatistical analyses They certainly were underrepresented in sheer numbers of coring
locations (roughly 20 of 200 or < 10%). The concern here would be whether PCB inventories
were underestimated because these areas were 'ignored" or incorrectly weighted in the analysis.
The third analysis would combine the major conclusions of the Phase 2 analysis to see if sediment
PCB inventory losses are consistent with annual water column loadings and estimates of
downstream transport The difference between the two could then be investigated as losses via
evaporation and/or in situ degradation in future monitoring and/or modeling exercises. This mass
balance check was mentioned somewhere in the Reassessment Summary, but should be elevated
into the summary of major Phase 2 conclusions.
A final analysis would reassess the compatibility of 1976-78 and Phase 2 PCB data. In my
experience, total PCB estimates based on Aroclor or congener-specific data are generally strongly
correlated, unless major congener/homolog shifts are present. This was done convincingly for the
1984 NYSDEC and Phase 2 data in Appendix E of the LRC. Statements made concerning
Aroclor standards used for the 1976-78 data set were not consistent throughout the Phase 2
documentation, but if a consensus could be reached on what approach was used, a correlational
analysis would help determine if these data were compatible.
44
-------
Maruya
Literature Cited
Abramowicz, D.A.. M.J. Brennan. H M Van Dort and E.L. Gallagher. 1993 Factors influencing
the rate of polychlorinated biphenyl dechlorination in Hudson River sediments. Environ Sci.
Technol. 27:1125-1131.
Achman, D.R., B J Brownawell and L. Zhang. 1996. Exchange of polychlorinated biphenyls
between sediment and water in the Hudson River estuary. Estuaries 19:950-965.
Bedard. D.L.. S C. Bunnell and L A. Smullen 1996 Stimulation of microbial para-dechlorination
of polychlorinated biphenyls that have persisted in Housatonic River sediment for decades
Environ. Set. Technol. 30:687-694.
Bedard, D.L. and J F Quensen III 1995. Microbial reductive dechlorination of polychlorinated
biphenyls. In: Microbial transformation and degradation of toxic organic chemicals. Young.
L Y and C.E. Cerniglia. eds.. Wiley Liss, NY. pp. 127-216
Brownawell. B J and J W Farrington. 1986. Biogeochemistry ofPCBs in interstitial waters of a
coastal marine sediment. Geochim. Cosmochim Acta 50:157-169.
Burgess, R.M., R.A. McKinney and W.A. Brown 1996. Enrichment of marine sediment colloids
with polychlorinated biphenyls: trends resulting from PCB solubility' and chlorination. Environ.
Sci. Technol. 30:2556-2566.
Dunnivant, F.M.. A W. Elzerman, P C. Jurs and M.N. Hasan. 1992. Quantitative structure-
activity relationships for aqueous solubilities and Henry's Law constants of polychlorinated
biphenyls. Environ. Sci Technol. 26:1567-1572.
Hawker. D.W. and D W. Connell. Octanol-water partition coefficients for polychlorinated
biphenyls. Environ. Sci. Technol. 22:382-387
Mohn. W W and J.M Tiedje. 1992. Microbial reductive dehalogenation Microbiol Rev. 56:482-
507
Quensen III. J F and J.M Tiedje. 1997. Evaluation of PCB dechlorination in sediments. In:
Methods in biotechnology. Vol. 2: Bioremediation protocols. D. Shechan. ed.. Humana Press.
Totowa. NJ. p257-273.
Warren. S.D . R.F. Bopp and H.J. Simpson 1987. Volatilization ofPCBs from contaminated
sediments and water Final Report, NYS Contract C001263. 49pp.
Wu, Q , D.L. Bedard and J. Wiegcl. 1997. Temperature determines the pattern of anaerobic
microbial dechlorination of Aroclor 1260 primed by 2,3.4.6-tetrachlorobiphenyl in Woods Pond
sediment. Appl. Environ. Microbiol. 63:4818-4825.
45
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
46
-------
Dr. Ronald Mitchum
47
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
48
-------
Peer Review - DEIR and LRC - Hudson River System
Provided by: Ronald K. Mitchum, Ph.D.
February 26, 1999
This peer review is provided in response to the "Charge for Peer Review 2" given to the
review panel on January 11 and 12, 1999. The following addresses the specific questions
before the panel.
I. Data Evaluation and Interpretation Report (DEIR)
1. "Is the documented PCB load, which originated from the TI Pool,
consistent with a source consisting of historically deposited PCB-
contaminated sediments?"
The TI sedimentary material was the result of several events, which
deposited PCB. The first known event would have been associated with
Federal Dam structure which was removed in 1973 and the sediment
resulting from more recent and continuing releases associated with the GE
Baker Falls Plant.
Response:
The determination of the source of PCB originating from the TI inlet pool
can be formulated by comparing the congener distribution of the aroclor
1242 and that of biodegraded 1242 found in the sediment. The use of
congener patterns will provide the most information where the use of
homolog patterns will provide the least.
Figure 3-53 distinctly shows the redistribution of sediments caused by the
destruction of the Federal Dam up river from the TI pool which occurred
about 1973. This establishes that the sediment contained in the pool
predominantly occurred via redistribution caused by that event.
The data has been treated using both congener and homolog patterns. The
homolog data in figure 4-34 indicates that at normal to low flow
conditions the load is consistent with the TI pool load. At high flow
conditions where equilibrium was not established in the TI pool the down
river load approximates the new material being released at Baker Falls.
A comparison of congener data in the same way the homolog data was
compared in 4.34 can not be readily done. The use of multivariate
statistical technique would have allowed a similar comparison.
49
-------
In conclusion the data supports the thesis that the PCB load originates
from the TI pool. Deviations from the model may occur during high flow
upsets.
2. "Are the two-phase and three-phase partitioning coefficients, derived in
the DEIR, appropriate and do they properly address the physical
parameters of the system (e.g. temperature)?"
To define the porewater derived equilibrium concentration, the
partitioning coefficients must contain the most important physical
parameters. The equilibrium may be described as the ratio of the
absorption and the desorption rate constants. The rate constants have a
fundamental temperature term, which must be included. The discussion
on page 3-15 and the derivation of the temperature term appears to be
correct. The three-phase system would seem more appropriate, however,
there was not enough data to apply the model consistently.
In addition to the three phases used in the model some PCB was lost to
aerosol formation in the falls (dams), ripples, edge eddy currents and
surface evaporation. This source may be minor, however, significant
pesticide transport has been observed in fog water in the San Joaquin
Valley located in California. The use of Henrys Law to explain away the
importance of the process may not be warranted. PCB present at the
surface may not behave as an ideal gas. If the mechanism is via aerosol
formation, Henrys Law can not be invoked except to explain how PCB
may be lost from the particle. Dechlorination by photolysis is a significant
chemical degradation process (see Erickson pp. 37). This factor may be
significant in the near surface water of a stagnate system, such as, the TI
pool and the river flow.
The adequate representation by the two or three phase models is supported
by the data. The comparison of the data with the model may suffer due to
non-equilibrium events such as temperature currents, gradients and
stratification within the system.
3. "Are the conceptual models based on the transect sampling consistent with
the data?"
The conceptual models suggest that the TI pool is the source of the PCB
down river either through resuspension of fine sediments or porewater
equilibrium. The data is consistent with the above. The transect sampling
events support the conceptual model.
4. "Does the sampling at the Tl-West location impact EPA's conclusion that
the sediments of the TI-Pool are the major source of PCBs to the
50
-------
freshwater Hudson during low flow conditions considering the analytical
corrections made to GE's PCB data? What are the other implications of
finding higher concentrations along the shoreline than in the center
channel?"
The total net load resulting from the TI pool was calculated as the net
increase observed from a sample point up river at mile 194.6 and at the
dam at mile 188.5. The GE data as adjusted by EPA does show a gain
across the pool, see figure 3-83. The EPA data set should be recalculated
based upon the GE target sampling and the estimates revised. The GE
data consisted of enough data points to fully characterize the gain due to
the TI-Pool.
The flow through a system such as the TI pool is not uniform across the
pool. The flow will be slower near the edge of the pool and faster near the
center. A current will be established during high flow conditions, which
will influence the concentration of PCBs in the water column. A
concentration gradient across the dam should exist if equilibrium
dynamics are used as the mechanism for developing the TI pool
contribution to the down river PCB load. However, during low flow
conditions the influence will be minimal and the over the dam
concentration of PCB may reflect the equilibrium concentration in the
pool.
5. No Comment - This is out of my area of expertise. The discussion
presented a logical argument for the use of each.
6. "Are the methods applied in the DEIR (change in molecular weight (MW)
and evaluating concentrations of BZ#s 1,4,8,10 and 19 (MDPR))
appropriate standards for determining extent of dechlorination? Are there
any significant problems with this approach, or more appropriate
approaches?"
The approach provides a quantifiable method to represent the
dechlorination on a per sample basis and to compare it to other samples in
the set. This appears to be a brut force method and the linearity of the plot
in figure 4-21 simply represents the derivation of the equation 4-13, which
is the equation for a straight line. The difference in the intercept between
the regression line and the theoretical line may be due to the initial
assumptions regarding the concentration of the total PCB rather than the
PCB containing only ortho chlorines.
The MDPR approach looses the information, which may be present if each
congener were treated. Use of a technique such as SIMCA or other
51
-------
multivariate statistical packages may provide fine detail information
regarding the dechlorination of the PCBs found in the Hudson River
system. In addition information regarding possible changes to the
dechlorination pattern due to further dechlorination or changes due to
further sediment equilibration may be observed.
7. "The DEIR finds that the degree of anaerobic dechlorination is primarily a
function of the original concentration rather than time, and accordingly
that there is not significant predictable dechlorination in the sediments
containing less than approximately 30 mg/kg PCB. Is this reasonable?"
The rate of anaerobic dechlorination contains both a concentration, term
and a time term according to:
d[dechlorination product concentration]/dt = k [original concentration]
Therefore, the dependence upon the original concentration would be
expected. No experiment was performed which would determine the time
dependence. The time dependence may be on the order of days, weeks or
months rather than years.
The 30 mg/kg was deduced from a plot of fractional dechlorination vs log
total PCB. The basis for the plot comes from rearrangement of equation
4-13 to:
(.86/. 14)*AM = MDPR ~ 6.43 -.223
This equation predicts the intercept will be -.223 and if .86/. 14 AM were
plotted against MDPR the resultant slope of the line would be 6.43. There
are no provisions for the log relationship. The use of a log relationship
simply allows what would be a discontinuous relationship observed in
figure 4-21 to now appear continuous.
The meaning of the data is as follows: The points which lie below the
aroclor 1242 in 4-21 and those which are referred to as being below 30
ppm in figure 4-22 belong mostly to the class referred to as being derived
from the fresh water lower Hudson River and a few representatives of the
upper Hudson. This represents a change in congener pattern (ratio) in the
lower Hudson. A multivariate analysis may have indicated that this group
of samples belongs to a separate congener composition representing the
lower fresh water Hudson.
52
-------
II. Low Resolution Sediment Coring Report (LRC)
1. "In the LRC, EPA compared sediment data from cores taken in 1977,
1984 and 1994, which had the PCB analysis conducted by different
laboratory methods. How valid are the methods used to establish a
consistent basis for comparison?"
Some assumptions were made based upon the various methods used
during these eras. In 1977 PCB analysis consisted mainly of packed
column ECD analysis following drinking water methods developed by
EPA or FDA. Only the significant homolog groups were separated.
Typically the data was reported as the closest matching aroclor and
reported as a total. Access to the data could not have resulted in the
deciphering of homolog groups. The samples may or may not have been
preserved and quality assurance programs were not in place to assure the
integrity of the data.
In 1984 data were generated again using packed column technology and
followed the NYSDEC program methodology. The data would have been
reported as totals or homologs. The aroclor would have been identified.
The data was most likely useable. There was no indication that the data
had been validated. In 1994 the data collected consisted of congener
specific data collected using capillary chromatography. The quality
assurance protocols reflected a mature QA program. The data should be
useable.
There are a couple of problems with the old data using packed column
technology. First, one may want to interpret the groupings in packed
column data as pure homolog groups, such as, Cl-3, Cl-4, Cl-5 etc.
Although distinct peaks are seen in the chromatogram these are not pure
homolog groups. Therefore, time travel of the data to 1994 was not
possible. Using the data presented in the LRC, E-4, for the congener
distribution within the individual quantification peaks for the various
aroclors, reverse extrapolation to packed column data was made.
The data is shown to be self-consistent when the tri + congeners are used
in the extrapolation.
There are several assumptions, which have been made which detract from
the use of the estimate. First, the data set has been corrected by NYSDEC
to reflect an apparent oversight by Versar in the calculation of PCB
concentration for aroclor 1242. Second, the work of Gauthier-TAMS may
not have reproduced the column or conditions used for the NYSDEC
study. Third, the data is not of known defensible quality. That is, the
same quality standards were not in place in 1984 both in the laboratory or
the field sampling.
53
-------
The 1984 data at best should be used for estimation purposes only. In
addition, since the old data does not reflect the lower homologues which
play an important part in the assessment of the PCB containing sediments,
its use may detract from conclusions regarding fate and transport.
2. "In the upper Hudson River system, it has been well established that there
is significant lateral heterogeneity in sediment concentrations. While it
was attempted to reoccupy previous locations, some uncertainty is added
with respect to the actual sampling location. While the statistical
techniques help compensate for this, how does the sediment heterogeneity
affect the comparison of cores from two different years? Given the spatial
variability, is the finding that there is loss from most of the locations
supported by the data?"
Sampling errors associated with homogeneity, sample transects, and
sample numbers and the statistical design associated with the sampling
plan represent the single largest error associated with assessments of this
nature. Comparison of sampling events provides the opportunity to
propagate these errors. If the sampling plan covers an area with 300 ft
transects, then the error associated with comparison of core samples 10
years apart may be large if the sediment has significant lateral
heterogeneity. If the sampling plan included more samples with smaller
transects than the error would be smaller.
The sampling used a grouping around the 1984 sample site. This will tend
to average out sampling error associated with position.
Question 2.
Given that the data set for 1984 is internally consistent and that the data
set for 1994 is internally consistent but that there is no common ground
between the methods makes it likely that some bias may be introduced.
Given the estimates of PCB concentration in the 1984 study and the error
associated with sampling, an error estimate should be established which
will define the likelyhood of the data supporting loss of PCB from most
locations.
3. "What is the impact of the difference between replicate samples in the
1994 sampling effort (36% average variability) on the finding that there
was a 40 percent loss of PCB inventory from the highly contaminated
sediments in the TI Pool?"
The variability of the 1994 and the 1984 data must be considered together.
If the deviation, change from the mean is ±18%, this is the only data
available and the 1984 data is considered to have the same variability, then
54
-------
an observed 40%, in 1994 may not be significantly different from the 1984
data.
4. "In the LRC, it was found that HOT Spot 28 contained much more mass
than previous estimates. Is the conclusion that this "gain" is primarily due
to incomplete characterization in 1977 valid?"
The insufficient characterization in 1977 could have resulted in the
discrepancy observed. More likely the analytical and sampling
methodology contributed to the apparent underestimate. Since the 1977
data is not of known quality.
5. "Does the data set and its interpretation support the conclusion that
significant losses have occurred from hot spots below the TI dam?"
The comparison of the 1984 and the 1994 data sets indicate that most of
the hot spots lost mass during the 10 years. The sampling design is robust
enough to have located the 1984 sediment sampling sites. The chemistry
comparison may produce a bias due to uncertainties in the 1984 data set.
6. "The LRC found that the historically contaminated sediments in the TI
pool were not universally being buried and sequestered from the
environment. How much confidence would you place in the LRC
evidence against widespread burial?"
There is a preponderance of evidence in the DEIR and the LRC, which
suggest the sediment borne PCB, are being redistributed to the water
column. High-resolution cores supported the low-resolution core data. If
the data under penning the conclusions are sound, widespread burial does
not appear to be occurring.
7. "Is the interpretation of the sidescan sonar data appropriate and supported
by the analysis of the associated sediment properties?"
I can not draw a conclusion due to my lack of knowledge of SSS.
General Questions
1. The data set has addressed many of the variables necessary to assess the
fate and transport of PCB in the Upper Hudson.
2. New deposition from the GE Bakers Falls plant site appears to be
occurring. Due to the high loading of PCB in the sediment, NAPL, may
be of concern. Although none was reported in the LRC, sampling
methods to determine NAPL were not used. The equilibration of the
water column appears to be associated with the dissolution of NAPL from
55
-------
the GE plant site. The loss of PCB either photochemically or by aerosol
formation resulting from the stripping occurring below each dam on the
Hudson may be factors, which influence the PCB load. Recalculation of
the water column PCB load resulting from the apparent overestimate
introduced due to sample inhomogenity near the TID in the DEIR data
needs to be performed.
Development of a multivariate statistical treatment of the data needs to be
performed. Much of the information has been lost due to the data
treatment using bivariate statistics.
56
-------
Reviewer: Ronald K. Mitchum, Ph.D.
Review of the DEIR
March 4, 1999
Recommendation: Acceptable with minor revision
Review of the DEIR and associated Responsiveness reports indicate that
the objective of the reassessment has been accomplished. Inclusion of the
fate compartments consisting of air transport, resulting from the dams
along the Upper Hudson, and photochemical degradation complete the
overall conclusions. Perhaps of more concern is the bivariate statistical
treatment of the data. It is understood that this may stem from a historical
approach in which continuity of presentation was desired. The use of
multivariate statistics will express obscured trends in the data. The
congener specific analysis offers a rich data base from which to reassess
and expand upon the conclusions drawn to date.
The plot representing the fractional dechlorination vs. the log of the
dechlorination ratio appears to have no theoretical basis. The log
transform removes the skewness in the data set and has led to an
inadequate conclusion regarding the apparent 30 ppb dechlorination limit.
A recalculation of the PCB load leaving the TIP requires using the new
GE data from the transect sampling across the dam. Inclusion of a model,
which addresses the edge effects, may serve to explain the apparent
sampling discrepancies observed.
57
-------
Reviewer: Ronald K. Mitchum, Ph.D.
Review of the LRC
March 4, 1999
Recommendation: Acceptable with major revision
The LRC provides a current 5-year-old update to the 1984 sediment study
performed by NYSDEC. The sampling design was adequate to assess the
1984 sampling points. The underlying data variability must be assessed to
determine if the conclusions, which were drawn, are valid. The joint
variability of the 1984 and the 1994 analysis events along with the
sampling variability due to spatial variations need to be addressed. The
conclusions then need to be reassessed. Although the data trend is present
which suggests there has been a loss of hot spot PCB load below the TID
are the measurements precise enough to define the loss.
The inclusion of data, which was derived from 1977, or earlier events has
so much analytical uncertainty as to be unusable. Any conclusions drawn
from these comparisons should be qualified or removed from the
document.
The use of a bivariate data treatment rather than a multivariate statistical
treatment may have obscured trends in the data. The congener specific
analysis offers a rich data base from which to reassess and expand upon
the conclusions drawn to date.
58
-------
Dr. Ken Reimer
59
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
60
-------
Hudson River PCBs Site Reassessment (RJ/FS)
Date Evaluation and Interpretation Report (DEIR) and
Low Resolution Sediment Coring Report (LRC)
Peer Review 2: Pre-meeting Comments
Dr. K.J. Reimer
Director, Environmental Sciences Group
The Royal Military College of Canada
Kingston, Ontario
Canada
March 1, 1999
61
-------
K.J. REIMER
Hudson River PCBs Site Reassessment (RI/FS)
Date Evaluation and Interpretation Report (DEIR) and
Low Resolution Sediment Coring Report (LRC)
Peer Review 2: Pre-meeting Comments
A. Introduction
The comments that follow are based on an evaluation of the DEIR and LRC reports as well as the
corresponding Responsiveness Summaries. As noted in the charge for the peer review, the analyses in
the responsiveness documents were considered to supersede those in the original reports. The Hudson
River Reassessment Database (Release 4.1) was also used in preparing the following remarks.
Reference was also made to the scientific literature as found in refereed reports, but not to other
documents or reports dealing with this Hudson River issue.
Comments are divided into:
• Responses to the specific charge questions for the DEIR - Part B.
• Responses to the charge questions for the LRC - Part C.
• General question responses - Part D.
• Recommendations - Part E.
B. Response to DEIR Questions
1. Is the documented PCB load, which originated from the TI Pool, consistent with a source
consisting of historically deposited PCB- contaminated sediments?
There seems to be little disagreement that the combined TI Dam (TID) load - i.e. the PCB load
originating above the TI Dam and including Hudson Falls, the Remnant Deposit area and the
Thompson Island Pool (TIP) - is the major source of PCBs to the freshwater Hudson. It further
62
-------
K.J. REIMER
seems reasonable that the PCBs in the water column at the TID originate from the sediments of the
TIP. The question remains - "what portion of the sediments?'
In order to assess if it is historically contaminated sediments or some other source it is necessary to
examine several lines of evidence - foremost among these is a change in homologue pattern across
the TIP (DEIR, page 3-171).
Section 3.4.2 of the DEIR examines the nature of the PCB loading to the TIP. The results of the
March 1993 Phase 2 and GE monitoring programs show similar homologue patterns at the Rogers
Island station. Both analyses demonstrate (Figure 3-102) that the load largely consists of tri- and
tetra-chlorinated congeners - suggestive of an Aroclor 1242-like source. It is suggested that this
similarity, together with the highly variable loading, is due to the GE Hudson Falls releases. [It is
interesting that the loading is higher at Rogers Island relative to the station at RM 195.5 near the
Remnants Deposit although the congener patterns at the two stations are frequently the same. This
raises the question of a PCB source from the Remnants Deposits, although GE has apparently
concluded that they are only important as a source of secondary mobilization of PCBs from the
Bakers Falls source (page 2-20).]
Section 3.4.3 indicates that the homologue patterns of the net TI Pool loading during two Transects
(3 and 6) are characterized by a higher loading from the Pool with a homologue pattern dominated by
dichlorobiphenyls. The data seems reasonable (compare Figures 3-102 with 3-103) but the case for
temporal variations (Figure 3-103 March versus August) at the TI Dam is less persuasive as the
reader has no knowledge of the analytical variability of the individual congeners. Indeed the lack of
information regarding analytical precision is a major problem in reading the DEIR. Much is made
about trends and visual comparisons but the reader cannot independently evaluate their significance.
Book 3 of the Responsiveness Summary provides additional evidence of a change in the homologue
pattern. An argument is made (page 18) that, at a time of low load (summer 1997) and low flow, the
data show "the usual strong shift to mono-, di- and tri-chlorobiphenyl dominated pattern" for the TID
relative to Rogers Island. Unfortunately, this is difficult to see from the provided Figure (2-1). Figure
2-2 is more illustrative in that it shows, at a time of low upstream load (summer 1997), a shift to a
pattern enriched in the mono- and di-chlorinated congeners relative to Aroclor 1242. It is noted that
these data are similar regardless of whether the TID-West or center station data are used. This
reviewer cannot, however, see that this plot makes the case for enrichment of the trichlorinated
63
-------
K.J. REIMER
congeners; the only apparently significant changes are in BZ#1 (mono) and BZ#4+10 (di). It is
unfortunate that there is no comparison to the Rogers Island data in this instance as a direct
comparison of the two patterns would have strengthened the argument. Nevertheless, as will be
shown later in this document (Part B.3), an alternative analysis of the Transect 6 data by the reviewer
confirms the difference in congener fingerprint between Rogers Island and the TI Dam.
Given that there is a distinctive change in the congener pattern, it remains to determine if this is
derived from the sediments of the TIP. Section 3.2 of the DEIR makes the case that the "load at the
TI Dam is derived almost entirely from the sediment given the consistency of the total TI Dam load
and its homologue pattern." The Responsiveness Summary provides the most direct evidence by
making a comparison between Aroclor 1242 and the composition of the surface sediments from the
TIP. The sediment patterns (Figure 2-3, page 22) appear to be significantly dechlorinated relative to
unweathered Aroclor 1242 - less obvious are the suggested differences between the cores within a
reasonable limit of analytical variability.
The enrichment in the lightly chlorinated PCBs in the water column could be a result of resuspension
of dechlorinated sediment. An alternative mechanism is diffusion and pore water advection - which
could move the more soluble congeners (dissolved and DOC-bound) into the surface sediments. The
Responsiveness Summary makes the case (Sections 2.3.1 and 2.3.2) that the congener patterns can
only result from a mix of pore water and direct exchange of sediment with the water column. It
further argues (Section 2.3.3) that the less strongly sorbed lightest congeners will be more easily
mobilized from depth relative to the more strongly sorbing congeners.
Section 5.4 of the Responsiveness Summary examines the potential effect of the Bakers Falls area
releases on six sediment cores. Comparison of the Aroclor 1242 equivalent concentrations in the
surface sections, 0-2 and 2-4 cm. suggests that surface layer PCB concentrations had been increased
by recent upstream loadings. The evidence is not strong as many of the changes must be close to
analytical variability (no comment is made as to what is significant or not). It is unfortunate that
congener fingerprints were not used. In a reanalysis of some of the data by this reviewer (described
later in this report), the Rogers Island East Core (core 26) appears to have a different composition
than the other cores. It would be interesting to know if this is due to a greater proportion of recent
input.
64
-------
K.J. REIMER
In conclusion, the PCB load can be reasonably associated with historically contaminated sediments
but the contribution of recent inputs must also be considered. It would appear that USEPA has
reached a similar conclusion, at least according to a statement in the DE1R Responsiveness Summary
(Book 1, page DEIR-47) - "it is highly unlikely that either PCB type (i.e. old or recently-
contaminated sediments) is solely responsible for the water column load generated by the sediments.
Most likely, the PCB load is a combination of both recently deposited and older PCBs." I concur with
this statement.
2. Are the two-phase and three-phase partitioning coefficients, derived in the DEIR,
appropriate and do they properly address the physical parameters of the system (e.g.
temperature)?
The calculation of the constants appears to have been conducted according to normal procedure but
rigorous analysis is left to those more expert in this area. Comments are. therefore, restricted to
concerns regarding the degree to which one can accept the conclusions drawn from these constants.
It is presumed that 126 vice 130 congeners were used in the analysis.
Figures 3-13, -14 and -15 plot Kp.a for various congeners with River Mile for different Transects.
Various conclusions are drawn - such as partitioning at Waterford (RM 156.6) is very similar to that
in the freshwater portion of the lower Hudson (at least for Transects 1 and 6). Given at least the
variability of the constants (not obvious and not shown on the plots), one must question the
conclusions as well as the common sense in attempting an interpretation of any kind.
J. Are the conceptual models based on the transect sampling consistent with the data?
In general, the conceptual models are consistent with the data obtained from the transect sampling.
The discussion put forth in Appendix C, Book I of the Responsiveness Summary for Volume 2C-A
Low Resolution Sediment Coring Report is, for the most part, much more convincing than that in the
DEIR itself.
65
-------
K.J. REIMER
Concurrence with several aspects of the general construct of the conceptual model for water column
PCB transport (section 3.2.4. pages 3-59,60 of the DEIR) is straightforward and it is useful to briefly
summarize these:
• It is clear from the data that there is no significant PCB load generated above Bakers Falls
although it is optimistic to conclude that background = 0. It is unlikely that there is any region of
the earth, including the poles, where there are zero PCBs.
• The principal source of PCBs to the freshwater Hudson is undoubtedly the GE facilities as there
are no other apparent sources such as tributaries and atmospheric input is expected to be small.
Direct input and re-release via sediment suspension and/or porewater exchange are the only
remaining pathways.
• The congener patterns are consistent with a mixture of Aroclors with 1242 dominating.
• As noted in response to an earlier question, the T1 Pool is a substantive source of PCBs.
consistent with the framework for the conceptual model.
The conclusions of the "Revised Estimates of PCB and Suspended Solids Loads in the Upper Hudson
River" (LRC Responsiveness Summary. Appendix C, Book 1, pages C-14. 15) indicate that the
revisions do not require a major modification to the main conclusions of the DEIR. It is noted that
concept of year-round conservative transport has been abandoned and that (page C-13) "Low
flow/low temperature or high flow conditions yield near conservative transport. During late spring
and summer conditions, the total PCB load is not conservative and declines downstream of the TI
Dam." This seems reasonable in light of the revised load data.
I have less confidence in the congener specific arguments even though the presentation in the Low
Resolution Coring Responsiveness Summary is much improved over that found in the DEIR. Visual
comparison of homologue patterns (e.g. Figures 3-36 to 3-49, DEIR) is not very convincing -
especially as it is not obvious as to how analytical variability would impact on the conclusions.
Indeed, analytical variability is not mentioned -at least often - and the reader must ask whether it was
forgotten. It is noted that several sections of the Responsiveness Summaries address this point and
this is a distinct improvement. Nevertheless, it is very difficult for the reader to determine what is
statistically significant or not.
Amongst Figures C-6 through C-31 (LRC Responsiveness Summary, Appendix C, Book 1) are plots
showing the PCB load for mono-, di-, tri- and tetra-chlorinated congeners with River Mile.
Considerable interpretation is made of the changes in these plots but it is not apparent what the
66
-------
K.J. REIMER
variation is about each point. In one instance (page C-l 1) a comment is made about a 50% loss in
monochlorinated congeners and that this change is beyond the analytical uncertainty. This begs the
question of what analytical uncertainty was considered when interpreting any of the plots. One can
make some assumptions based on the comments regarding Figure C-31 (which shows a downward
slope for all congeners from the TI Dam to Waterford) as it is stated (page C-9) that there is a
"similarity of the homologue pattern between the TI Dam and Waterford" thereby suggesting that the
slope is not significant. This raises questions about the interpretation of other plots, particularly those
representing low PCB loads (lower concentrations near detection limit may be less reliable) from just
the Figures. Statements (page C-ll) such as "these changes are quite substantial... and are well
beyond any analytical variability" are a distinct improvement over the DEIR but it should be stressed
that there is no way that the reader can independently verify such conclusions with the information
provided.
More worrisome is the use of homologue patterns in this section and throughout the report. In view
of the rigor of the analytical program (i.e. 126 congeners in most cases) it is surprising that a more
rigorous statistical approach was not used - such a principal components analysis (PCA) to interpret
congener specific data.
In order to examine the conclusions made in the DEIR, three of the Transects (1,4 and 6) were
examined using PCA. Data were obtained from the Hudson River Reassessment Database (Release
4.1). Plots are appended as Annex A to this report. It should be stressed that this was a cursory
attempt to see if the DEIR conclusions could be confirmed by another approach. Some interesting
points can be noted and these are summarised for each of the Transects examined. Comparisons are
made to the conclusions found in the DEfR and associated Responsiveness Summary.
Transect 4
Plot 1 presents the preliminary PCA for dissolved and particulate phases of the water column samples
collected during this Transect (April; spring flood). Samples are labelled with d and p prefixes
indicating dissolved and particulate phases. The first number denotes the sampling station number
and the second the transect; e.g. the dissolved fraction collected at Rogers Island during Transect 1 is
labelled d4-1. As time only permitted a cursory examination of the data, this discussion and those
that follow will be similarly restricted. In simplest terms, samples that project in the same region of a
PCA plot may be assumed to have similar compositions.
67
-------
K.J. REIMER
It can be noted that the dissolved and particulate samples generally project differently. The samples
collected from Glens Falls and the Fenimore Bridge (stations 1 and 2 respectively) appear at the right
side of the plot: tributaries are widely scattered throughout suggesting different and variable inputs of
PCBs. These observations are consistent with the argument that the tributaries do not contribute to
the congener composition of the Upper Hudson. Most pertinent is the behaviour of samples
associated with stations at the Remnant Deposits (stn 3), Rogers Island (stn 4) and the series of
samples from the Thompson Island Dam (stn 5), Schuylerville (stn 6), Stillwater (stn 7) and
Waterford (stn 8). All samples from the Remnant Deposits to Waterford cluster tightly together. This
observation is consistent with the conclusion (page C-S. Book I) that total PCBs are transported to
Waterford in a conservative manner.
Transect I
Plot 2 displays the results of the sampling for this Transect (Jan/Feb). Fenimore Bridge and some
tributaries project to the right side. The dissolved samples for the Tl Dam. Schuylerville and
Waterford cluster tightly - indicating similar composition - but are very different from the Remnant
Deposits. The corresponding particulate samples are not as tightly grouped but appear to the left side
of the plot. It was noted in the DEIR that the Rogers Island sample was suspect and this is confirmed
by the PCA - both the dissolved and particulate samples project tightly and to the right side. These
observations are reasonably consistent with the conclusion that the water column load originating
above the Tl Dam is transported in a near-conservative manner, for all homologues.
Transect 6
Plot 3 presents the results of the PCA for this transect (August). Once again, background samples
(Glens Falls, Fenimore Bridge), both particulate and dissolved, project to the right side of the plot
and the tributaries are widely scattered. The dissolved samples for the Tl Dam, Schuylerville and
Waterford are clustered as are the particulate samples (although these project to the left of the
dissolved samples). The dissolved samples for the station at the Remnant Deposits and Rogers Island
are tightly grouped but this is not the case for the particulates. This plot would suggest that the
congener composition is maintained from the Tl Dam to Waterford, not showing a loss in mono- and
di-chlorinated congeners as discussed in the Responsiveness Summary. This difference may be due to
the insensitivity of the PCA to the loss of the lightly chlorinated congeners (the PCA was run using
all congeners) but it does raise questions.
68
-------
K.J. REIMER
The PCA plots do indicate that at low flow, sources above the Thompson Island Pool do not
significantly influence the congener profile at the TI Dam through to Waterford. This provides
additional verification of the importance of the TI Pool.
The surface sediments from the high resolution coring study were also examined by PCA (Plot 4). In
general, background (s27 = core 27 at RM 202.9) projects lo the right of the plot and tributaries and
lower Hudson River samples are widely scattered. Most of the remaining samples project to the lower
left with the exception of core 26 from Rogers Island East which is displayed to middle lower section
of the plot.
It is recommended that the significance of the congener profile changes from the TI Dam to
Waterford be discussed at the Peer Review.
4. Does the sampling at bias of samples collected at the TI Dam- West sampling location
impact EPA's conclusion that the sediments of the TI Pool are the major source of PCBs
to the freshwater Hudson during low flow conditions considering the analytical
corrections made to GE's PCB data? What are the other implications of finding higher
concentrations along the shoreline than in the center channel?
The arguments put forward in Section 1.0, Book 3 of the DEIR Responsiveness Summary as well as
Appendix C, Book 1 of the Responsiveness Summary for Volume 2C-A Low Resolution Sediment
Coring Report appear reasonable. Evidence for the bias is persuasive but much of the effect appears
to be mitigated by the analytical corrections.
The ratio between center channel and TID-West appears to approach unity as either flow or upstream
concentration increases and this is consistent with the explanation provided - i.e. that increased flow
creates greater lateral mixing and that as the upstream concentration increases it will overwhelm the
effect from the nearshore areas. An intriguing argument is made (section 1.4, Book 3) that the actual
daily load transported downstream may be an average of the TID-West and TIP-18C observations.
The correction factors and the revised load calculations are consistent with EPA's conclusion that the
sediments of the TI Pool are the major source of PCB to the freshwater Hudson during low flow
conditions. The conclusion (Appendix C, Book 1) that the net result of the revisions (including flow
69
-------
K.J. REIMER
corrections) is =20% lower overall loads in the Upper Hudson under low flow conditions appears
warranted.
Furthermore, the presence of a bias is consistent with the argument that hot spots within the TI Pool
are a source of PCBs to the water column. The point made in the Responsiveness Summary (page 43)
- "Elevated concentrations in near shore low velocity areas are consistent with a pore water flux
loading mechanism, which would result in higher concentrations where dilution flow is lowest'* - is
quite reasonable.
5. Are the geostatistical techniques (polygonal clustering and kriging) correctly applied/
The techniques appear to have been correctly applied. I would prefer to see a range of estimates for
the PCB inventory. Reporting the calculated inventory to a decimal place strikes me as overly
optimistic and potentially misleading to the public - considering the variability in the results (not
stated) and the two estimates - 19.6 and 14.5 - that have been determined.
6. Are the methods applied in the DEIR (change in molecular weight (MW) and evaluating
concentrations of BZUs 1, 4, 8, 10 and 19 (MDPR) appropriate standards for determining
extent of dechlorination? Are there any significant problems with this approach, or more
appropriate approaches?
The molar dechlorination product ratio (MDPR) relies on the measurement of five specific congeners
in order assess the degree of dechlorination in sediments. The congeners used (BZtt 1, 4, 8, 10 and
19) all possess chlorines in the ortho positions on the assumption that anaerobic dechlorination
processes only remove meta and para chlorines.
The DEIR notes, correctly, that the less chlorinated congeners are more soluble and more susceptible
to aerobic degradation processes and may be lost from the sediments more readily, in which case the
MDPR will underestimate the actual degree of dechlorination. It might also be expected that the less
chlorinated congeners could be lost, by similar mechanisms, from the sediments prior to the
establishment of the anaerobic conditions that are essential to dechlorination - in such a case the
MDPR would be reduced.
70
-------
K.J. REIMER
The MDPR will also underestimate the degree of dechlorination if ortho-substituted chlorines were
removed. There is evidence in the literature of ortho-dechlorination [Van Dort and Bedard. Appl
Environ Microbiol. May 1991, 1576-1578; Berkaw, Sowers and May. Appl Environ Microbiol. July
1996, 2534-2539 (marine sediments)]; if this is significant in the Hudson River sediments, the
reliability of the MDPR will be compromised.
The change in mean molecular weight, AMW, is less susceptible to the points noted above. However,
loss of the lighter congeners - a likely possibility - would increase the molecular weight of the
mixture and reduce AMW.
Low values of MDPR and AMW found for the sediment samples from the Lower Hudson River are
interpreted (DEIR, page 4-63; Figures 4-19.20) to be representative of lower levels of dechlorination
with only some loss of lighter congeners. The linear relationship between MDPR and AMW (DEIR.
Figure 4-21) is the most convincing evidence that contributions from the above-mentioned processes
are minimal and that the approach is appropriate.
Both the DEIR and the LRC normalize data to BZ#52 (2.2"-5. 5'-tetrachlorobiphenv). Considering
the dechlorination argument, it is surprising that this congener does not undergo removal of the meta-
chlorines.
7. The DEIR finds that the degree of anaerobic dechlorination is primarily a function of
original concentration rather than time, and accordingly that there is not significant
predictable dechlorination in sediments containing less than approximately 30 mg/kg PCB.
Is this reasonable?
No. 1 do not agree with the conclusion as originally suggested in the DEIR (page 4-68) in discussing
Figure 4-22; namely, that "the distribution of the data strongly suggests that virtually all samples with
PCB concentrations greater than 30ppm are measurably altered with respect to Aroclor 1242. More
convincing are the conclusions stated in the Responsiveness Summary (Book 1, page DE1R-62 and
elsewhere): "Below 30ppm, the occurrence of dechlorination is not predictable using AMW as a
measure, because data fall above and below the initial AMW of Aroclor 1242. It is possible that
samples with AMW values less than that of an Aroclor 1242 have undergone dechlorination and
preferentially lost the mono- and di-chlorobiphenvls."
71
-------
K.J. REIMER
In our own work (unpublished) we have found no such threshold for dechlorination of even the more
recalcitrant Aroclors (such as 1260). Experiments with concentration ranges from 5 to 500ppm gave
essentially the same percent dechlorination (e.g. 80% removal of hexa-chlorinated biphenyls). Rates
were low at low PCB concentrations but there was no threshold concentration in the range tested.
Inhibition was noted above 500ppm for Aroclor 1260.
Although not part of this charge, the argument for lack of correlation of dechlorination with age is not
totally convincing. Figure 4-23 in the DEIR appears to suggest this but the results could be
overwhelmed by the relationship to total PCB concentration. Figure 4-24 could be interpreted as a
correlation with age when analytical variability about each data point is included.
C. Response to LRC Questions
Note: The Responsiveness Summary for the LRC is quite extensive and includes numerous
corrections as well as an alternative calculation for the comparison of sediment inventories in the
Thompson Island Pool. The Summary was received later than expected and it was not possible to
conduct a detailed review prior to the submission of pre-meeting comments Accordingly, the
following comments are intentionally brief. A more detailed examination will be completed prior to
the peer review meeting.
1. In the LRC, EPA compared sediment data from cores taken in 1977, 1984 and 1994, which
had the PCB analysis conducted by different laboratory methods. How valid are the
methods used to establish a consistent basis for comparison?
The correction factor developed in Appendix E of the LRC and the arguments supporting it seem
quite reasonable. Using the I trichloro to decachloro homologues puts both sets of data on an equal
footing. It is interesting to note (Responsiveness Summary, page LRC-41) that this approach is being
reviewed - presumably this information will be made available before the LRC is considered
complete.
Given the effort to make meaningful comparisons between the 1994 and 1984 data, it is surprising
that there is little discussion of the potential problems associated with the 1977 information. It is
noted (page 4-21, LRC) that "the simple sum of the reported Aroclor values yields an estimate for
72
-------
K.J. REIMER
total PCB concentration, which is probably the best that can be done to create a value for comparison
to the low resolution coring results." It is agreed that this is probably the best that can be done but it
does suggest that any trends derived from comparison of these data should be qualified.
2. In the Upper Hudson River system, it has been well established that there is significant
lateral heterogeneity in sediment concentrations. While it was attempted to reoccupy
previous locations, some uncertainty is added with respect to the actual sampling location.
While the statistical techniques help compensate for this, how does the sample
heterogeneity affect the comparison of cores from two different years? Given the spatial
variability, is the finding that there is a loss from most locations supported by the data?
Sample heterogeneity is always a concern in such an investigation. The Responsiveness Summary
appears to provide a more convincing argument in favour of the validity of comparing the 1984 and
1994 data than does the LRC.
The sampling locations were reoccupied with quite good accuracy (Responsiveness Summary page
LRC-5). Equally important is the fact that, within the clusters, the sampling density matched that of
the NYSDEC study - thereby strengthening the ability to make comparisons. It is stressed that the
intention of the study was to assess the direction and approximate magnitude of change of the PCB
inventory over the intervening period. The reanalysis of the data - presented in Appendix A, does
suggest that there has been a general loss; it further emphasizes the difficulty in assigning an absolute
value to this loss. This seems more reasonable than the impression one might get from the LRC even
though the word approximately is used in the discussion.
3. What is the impact of the difference between replicate samples in the 1994 sampling effort
(36 percent average variability) on the finding that there was a 40 percent loss of PCB
inventory from the highly contaminated sediments in the Tl Pool?
The point made in the preceding response is emphasized by this question. There is a tendency once a
number - i.e. 40% - is cited to forget the 'approximately' qualifier.
A quick read of the Responsiveness Summary (Appendix A) suggests that the estimated loss has been
revised but that the value is in agreement with that presented in the LRC. A case is also made (LRC-
11 to 19) that the low-resolution cores have uncertainties closer to 20% vice the 37% originally
73
-------
K.J. REIMER
proposed. No mention is made in either report as to the uncertainty associated with the 1984 data, but
one could assume that it is about the same - i.e. approximately 20-30%. The point (LRC-71) that
dividing by the 1984 concentration has the effect of normalizing the 1994 and 1984 data to account
for analytical uncertainty is a good one. Again, the data appear to be consistent with a loss of PCB
inventory from the highly contaminated sediments of the Tl Pool; a comment on the magnitude of
this change is left to those who can more adequately review Appendix A.
4. In the LRC, it was found that Hot Spot 28 contained much more mass than previous
estimates. Is the conclusion that this "gain " is primarily due to incomplete
characterization in 1977 valid?
Yes. The argument that the 1977 samples underestimated the amount of PCBs whereas a more
complete characterization was achieved in the recent program is reasonable. Several lines of evidence
seem particularly persuasive. These include the 1 "Cs data (Figure 4-25) which show that in 1994 the
bottom of the core represented true background as well as the argument that there would be
insufficient quantity of PCBs to so dramatically raise the sediment inventory between 1977 and 1994.
The general statement that the earlier studies probably underestimated the PCB inventory provides
additional support.
5. Does the data set and its interpretation support the conclusion that significant losses have
occurredfrom hot spots below the TI Dam?
The data set is consistent with a statistically significant loss of 50 to 80% for hot spots 31, 34 and 37
whereas hot spots 35 and probably 25 are unchanged. It is presumed that the term "significant" in this
question relates to statistical significance. Caution should be used in calculating overall mass changes
given the challenges in comparing the analytical data from the two eras.
6. The LRCfound that the historically contaminated sediments in the TI Pool were not
universally being buried and sequestered from the environment. How much confidence
would you place in the LRC evidence against widespread burial?
The evidence provided in the LRC - in particular the loss of sediment inventory and that the depth of
contamination has decreased or remained the same - is consistent with the water column information
74
-------
K.J. REIMER
described in the DEIR. Accordingly, the weight of evidence argues against widespread burial - at
least deep burial.
7. Is the interpretation of the sidescan sonar data appropriate and supported by the analysis
of the associated sediment properties?
The comparison of the sonar images with the 1976-1978 NYSDEC sediment survey seems
reasonable.
D. General Questions
1. Is the data set utilized to prepare the DEIR, LRC and Responsiveness Summaries
sufficient to understand the fate and transport of PCBs in the Upper Hudson?
It would appear that the data set lays out the overall understanding of the fate of PCBs in the Upper
Hudson. The debate that appears in the Responsiveness Summaries - namely the relative influence of
the TIP and releases from the Hudson Falls facility - could, however, be dealt with by direct
comparison of current data showing the relative loads at Rogers Island and the Thompson Island
Dam. Several points allude to this comparison but I could find no direct evidence.
2. Are there any additional analyses that should be done to verify certain findings of the
DEIR and LRC?
I would have liked to see more discussion of the effect of analytical variability in the DEIR
documents and the application of techniques such as principal components analysis.
E. Recommendations
I w ill finalize my opinion at the Peer Review when I have heard the comments of the other reviewers
but my preliminary opinion is that the DEIR and the LRC are acceptable.
The question of revisions hinges less on the need to make a more substantive argument than on what
the reports will be used for next. If they are to be used as a basis for the next report and the new
¦75
-------
K.J. REIMER
conclusions and statements that appear in the Responsiveness Summaries are to be incorporated at
that point, there is little use in making revisions to the DEIR and LRC. If these conclusions are to be
publicly accessible, however, a succinct summary document would be very useful.
76
-------
K.J. REIMER
ANNEX A
Principle component analysis (PCA) was carried out to explore patterns in the PCB congener data. In
this indirect gradient analysis, all of the variables (in this case congeners) are considered
simultaneously; variance in the congener data is explained by fitting a straight line through a
multidimensional normal curve, using a converging iterative ordination algorithm, such that the
residual sum of squares is minimized. This line is the first ordination axis, or first principal
component. Further axes are constructed in the same way, with the constraint that they are
uncorrelated. This technique is thus a convenient way to summarize multivariate data in two
dimensional space.
PCA was carried out using the statistical program SYSTAT (version 6.0.1). Twenty-eight surface
sediments, and 83 water samples (separated into dissolved and particulate phases) were normalized
using standard techniques and then ordinated according to their congener profiles (based on 126
congeners). PC axes 1 and 2 explained 44.3% and 7.0% of the total variance in the data indicating
that PCB congeners explain the bulk of variation in the samples collected.
77
-------
^ lot \
Transect 4 - Particulate and Dissolved PCBs
-P11-4
•09-4
•01-4
(117-4
1 0 1
FACTOR(1)
Particulate
Dissolved
78
-------
VWV L
Transect 1 - Particulate and Dissolved PCBs
1.0
_ 0.5
S
of
K 0.0
O
-0.5
1-0
-2-10 1 2
FACTOR(1)
- Particulate
• Dissolved
I I
I
—p11-1
P14-t- p6-V-
p8-1-
di-idi 3-1
•<114-1
—p2-1
—
pfM—
p 16-1—
—p12-1
•08-1
•06-1
•<35-1
—
p3-V
I I
I
—
79
-------
cu
Transect 6 - Particulate and Dissolved PCBs
3i , r
p13-6-
plO-6-
p14-6-
P17-6-
K
•dt-6
di7-®
d15-8>
d!0-»
FACTOR(1)
- Particulate
• Dissolved
80
-------
Surface Sediment PCBs - High Resolution Cores
1
s5*
s24«
I
I
s12*
—
s6«
s7*
s8«
s9*
• s27
• s4
—
- S2s^r23^3
^•s2f3M6Sl4sl9
sf£
s26 .
• s1
• s17
I
1
0
-1
-2
-1 0
FACTO R(1)
1
81
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
82
-------
Dr. James Risatti
83
-------
This Page Was Intentionally Left Blank For Pagination Purposes.
84
-------
Dechlorination Index.
The dechlorination index,used in this report (MDPR), is based on a ratio derived from the sum
of the molar concentrations of congeners BZ-l,4,8,10,and 19 found in the sample, divided by
the sum of 126 congeners identified in the study (V:2C,bkl,pg 4-56). The degree of in situ
dechlorination is determined relative to Aroclor 1242 by calculating a similar ratio from
Aroclor 1242 and using the Aroclor 1242 index as a base value. This value (0.14) is subtracted
from the sample ratio to give the degree of in situ dechlorination.
Harkness et al.(1993), characterized the extent of in situ dechlorination in a Hudson River
sample by noting the amount of mono- and dichlorobiphenyls present relative to the
predominant Aroclor originally released. The product of complete PCB dechlorination is an
unchlorinated biphenyl molecule but in Hudson River sediments the dechlorination scheme
favors removal of chlorines at the meta and para positions which leaves congeners containing
ortho chlorines such as BZ 1,4, 8, 10, and 19 to accumulate as ultimate or penultimate
products. For Hudson River samples the suite of congeners selected for the MDPR would
appear to reasonable but there are other issues inherent with the MDPR which reduce its
effectiveness as an indicator of dechlorination.
The major problem associated with a using a PCB dechlorination index based on final product
accumulation is diminished confidence that the measured product is a true representation of the
original compound mixture. In addition to lower ECD response (particularly BZ 1), the less
chlorinated PCB congeners are subject to aerobic degradation and to physical-chemical
movement out of the sample environment. Aerobic degradation has been recognized and has
been well studied for some time (Ahmed and Focht,1973, Furakawa, et al. 1978 and Beddard
et al., 1986, among others); although some isolated cultures can degrade specific highly
chlorinated biphenyls, most aerobic PCB degrading bacteria favor the less chlorinated mono-,
di- and tr- chlorinated congeners as substrates. Furakawa, et al. (1978) also found that in
addition to more rapid degradation of the lower chlorinated congeners,the non-chlorinated ring
was preferentially degraded. A screening study by Beddard et al., (1986) indicated that BZ 4
and BZ 8,both of which are used for the MDPR, were rapidly degraded by the environmental
isolates used in the study.
The mono-, di- and tri- chlorinated congeners are also more readily lost from the sample pool
by sorbtion and partitioning into the sediment pore water and into the water column than are
the more chlorinated congeners. In the report (pg 3-31), it has been recognized that the
movement of BZ 1,4,and 8 out of the Thompson Island pool sediments "may be facilitated by
binding to dissolved organic carbon" and that PCBs in general "may move from the porewater
to the water column by diffusion and groundwater advection of dissolved and DOC-sorbed
phases, "(pg 3-31). Lastly, the report notes that because of their lower partitioning
coefficients, the " concentration of mono- and dichlorobiphenys is enhanced in porewater
relative to more highly chlorinated congeners, which may facilitate loading of these congeners
from the sediment to the water column " (pg3-39) and that the sediments of the TI Pool
85
-------
released less-chlorinated congeners typical of dechlorinated sediments found in the Pool (pg. 3-
2). Under ideal conditions (temperature, pH, DOX ), biological and partitioning processes may
become strongly interactive as biodegradation of the less chlorinated congeners maintains a
disequilibrium in the sediment porewater and near the sediment surface.
Loss of the lighter congeners (BZ 1,4, 8, 10, and 19) directly affects the sensitivity of the
MDPR by underestimating the amount of dechlorination. The underestimated difference
depends on the MDPR estimate and the amount of BZ 1, 4, 8, 10, and 19 lost from the sample.
Underestimates can range from about 2 percent in samples with a MDPR of 0.2 and 10 percent
loss of the index suite to as much as 17 percent in samples with a MDPR of 0.6 but which has
had 50 percent of the index congeners lost from the dechlorinated sample.
As an index to compare dechlorination or other PCB changes, it is better to use a ratio
indicating decrease in concentration of two or more peaks selected from chromatograms of the
PCBs in the system being studied. Murphy (1989) correlated changes in sediment PCBs from
within Waukegan Harbor and with Lake Calumet by changes in the BZ18 / BZ17 ratio.
Ideally, however, ratios of more chlorinated congeners having similar chemical characteristics
should be used as they are less susceptible to aerobic biodegradation and partitioning into the
aqueous phase and at lower concentrations still give a measurable ECD response ( see Brown
and Wagner ,1990 for ratios used in a study of the Acushnet Estuary sediments).
Also, for both MDPR and MW, Aroclor 1242 is considered as the as the only commercial
PCB mixture in the sediments but in some areas, as much as 19 percent Arolor 1254 was found
(pg. 2-19) and Brown et al., (1988) indicate that in the area of RM 194.8, their "hot spot"
analysis indicates an average of 94 % Aroclor 1242, 5 % Aroclor 1254 and 1 % Aroclor
1260/1268. Both of the indices should be weighted if they are to be used.
Limits of Dechlorination
The suggestion that dechlorination of PCBs is limited by PCB concentration has been
suggested by laboratory dechlorination experiments with natural sediments (Quensen et al.
1988, Risatti,1992, Rhee et al. 1993 and Fish, 1996) and discussed by Brown and Wagner
(1990) in reference to field observations. Brown and Wagner (1990), state that there are no
examples of PCB dechlorination at isolated low level (1-3 ppm ) sites but that they have seen
dechlorination in low concentration PCB samples collected near sites of active dechlorination.
Quensen et al. (1988), working with Hudson River sediments found active dechlorination at
PCB levels of 700 ppm, less active at 140 ppm and none at 14 ppm. In experiments with
Waukegan Harbor sediment cultures (no PCBs added),that there was very little, if any,
dechlorination at PCB concentrations of 150 ppm, some dechlorination at 1,500 ppm and very
rapid dechlorination at 17,000 ppm (Risatti,1990 and unpublished data). Fish (1996) observed
86
-------
rapid dechlorination rates in Hudson River sediments spiked with approximately 248 ppm to 25
ppm Aroclor 1242 and slow dechlorination rates at concentrations of about 9.9 ppm. A wide
range of dechlorination limits observed have been observed in laboratory cultures and in field
studies.
The authors state (page 4-68) that "the distribution of the data strongly suggests that virtually
all samples with PCB concentrations greater than 30 ppm are measurably altered with respect
to Aroclor 1242.". That is, dechlorination is limited at about 30 ppm PCB concentrations. The
trend in Figure 4-22 suggestss a trend of increasing dechlorination with increased PCB
concentration but from the figure the dechlorination threshold limit could be anywhere from 20
ppm to 40+ppm. Also, Figure 4-23 doesn't support the 30 ppm statement. Sub-sample 10
has approximately 55 ppm PCB but has undergone only 3 percent dechlorination. On the other
hand, at horizons 1 and 2, 9 ppm and 6 ppm respectively, dechlorination is 16 percent and 21
percent. The middle horizons seem to be relatively consistent with extensive dechlorination
and high PCB levels. The trend in Figure 4-24, indicates, for the most part, dechlorination
increasing with both PCB concentration and depth (age), and that sample 1( approx. 26 ppm
PCB) has been dechlorinated by about 21 percent.
Although there is evidence indicating that PCB concentration levels limit the degree of PCB
dechlorination in sediments, the threshold concentration seems to vary with the sites examined
as well as within the stations sampled including the ones investigated for this report. There
does not appear to be a universal PCB dechlorination threshold limit. Given slight changes in
conditions (which are as yet unknown), dechlorination may occur rapidly and at lower PCB
concentrations; Beddard (1996) has managed to "prime" dechlorination in Housatonic River
sediments which had lain dormant for years. However, the factors that induced 98%-99%
dechlorination of specific congeners within 30 days from unammended, 15 year old Waukegan
Harbor sediment cultures are still unknown. What is interesting about Figure 4-27 is not that it
illustrates the contention of poor correlation between time and the dechlorination ratio but that
most of the dechlorination in the Upper Hudson occurs within the envelope bounded by
approximately 20 percent to 80 percent dechlorination and about 5 cm to 45 cm depth levels.
This suggests there is a common factor in the system which influences dechlorination rates. It
may be available organic carbon, reducing potential, or flushing of inhibitors or nutrients into
or out of the system.
Most of the data presented in the report supporting PCB dechlorination limits is based on the
molar dechlorination product ratio (MDPR). This index, as stated above, is not a useful
measure of dechlorination and must be calibrated with another set of ratios determined from
chromatograms of the samples and which avoid the problems inherent with the MDPR as
derived in the report. Consequently, until the degree of error associated with the MDPR data
are determined the usefulness of data derived from the MDPR are limited.
-------
References
Ahmed, M., and D.Focht, 1973. Degradation of polychlorinated biphenyls by two species of
Achromobacter. Can. J. Microbiology. V.19, pp.47-52.
Beddard, D., R. Unterman, L.Bopp, M.Brennan, M.Haberl, and C. Johnson, 1986. Rapid
assay for screening and characterizing microorganisms for the ability to degrade
polychlorinated biphenyls. Appl. Environ. Microbiology, V.51, pp.761-768.
Brown, J. and R. Wagner, 1990. PCB movement,dechlorination and detoxification in the
Acushnet Estuary. Environ. Toxicology and Chem., V.9, pp. 1215-1233.
Brown, J., R.Wagner and D. Beddard, 1988. Technical Comments: PCB Dechlorination in
Hudson River sediment. Science,V. 240, pp. 1674-1676.
Brown, J.F., D.L. Bedard, M.J. Brenan, J.C. Carnahan, H. Feng, and R.E. Wagner, 1987.
Polychlorinated Biphenyl Dechlorination in Aquatic Sediments. Science. V. 236, pp.709-712.
Fish, K., 1996. Influence of Aroclor 1242 concentration on polychlorinated biphenyl
biotransformations in Hudson River test tube microcosms. Appl. Environ. Microbiology,V.
62, pp.3014-3016.
Frame, G., J. Cochrane, S. Boewadt, 1996. Complete PCB congener distributions for 17
Aroclor mixtures determined by 3 HRGC systems optimized for comprehensive, quntitative,
congener-specific analysis. J. High Res. Chromotagraphy, V.19, pp.657-668.
Furakawa, K., K. Tonomura, and A. Kamibayashi, 1978. Effects of chlorine substitution on
the biodegradability of polychlorinated biphenyls. Appl. Environ. Microbiology, V.35,
pp.223-227.
Harkness, M., J.MeDermott.D.Abramowicz, J.Salvo, W. Flanagan, M.Stephens, F.
Mondello, R. May, J.Lobos, K. Carroll, M. Brennan, A. Bracco, K. Fish, G.Warner, P.
Wilson, D. Dietrich, D. Lin, C. Morgan and W. Gately, 1993. In situ stimulation of aerobic
PCB biodegradation in Hudson River sediments. Science, V.259, pp. 503-505.
Mullen, M., C. Pochini, S. McCrindle, M. Romkes, S. Safe, and L. Safe, 1984. High-
resolution PCB analysis: Synthesis and chromatographic properties of all 209 PCB congeners.
Environ. Sci. Technol., V.18, pp.468-476.
Murphy, T., D. Galinis and C. Arnold, 1989. The activity of PCBs in sediments and water
from Lake Calumet and Waukegan Harbor. Hazardous Waste Research and Information Center
Report RR-039. Savoy, IL pp. 1-51
88
-------
Quensen, J.F., J.M. Tiedje and S.A. Boyd, 1988. Reductive Dechlorination of
Polychlorinated Biphenyls by Aerobic Microorganisms from Sediments. Science. V.242, pp.
752-754.
Rhee, G-Y., R. Sokol, B.Bush, and C. Bethoney, 1993. Long-term study of the anaerobic
dechlorination of Aroclor 1254 with and without biphenyl enrichment. Environ. Science and
Technology, V.27, pp.714-719.
Risatti, J. B. 1992, Rates of microbial dechlorination of polychlorinated biphenyls (PCBs) in
anaerobic sediments from Waukegan Harbor. Hazardous Waste Research and Information
Center Report RR-061. Champaign, Illinois.
Schulz, D., G. Petrick, and J.C. Duinker, 1989. "Complete Characterization of
Polychlorinated Biphenyl Congeners in Commercial Aroclor and Clophen Mixtures by
Multidimensional Gas Chromatography-electron Capture Detection." Environ Sci. Technol.,
V.23, pp.852-859.
89
-------
It is much better to use ratios involving the more chlorinated congeners
The less chlorinated congeners used in the MDPR are more subject to aerobic degradation than
the
Anaerobic degradation may occur but I'm not convinced that degradation rates are significant
in situ.
Anaerobic degradation (alteration of the molecule by ring cleavage) here has been some is
products is valid but the major problem
ortho-substituted chlorines are more resistant to dechlorination (Brown, 1987, Abramowisz..)
and congeners containing ortho-chlorines, such as the suite chosen for the MDPR also can be
final products, associated with using mono-, di- and tri-chlorinated congeners is loss into the
water column from the sediments or by aerobic degradation. The loss of "light" congeners
from the sediment pool of congeners will give a MDPR that underestimates the degree of
dechlorination.
Figure 1. Changes in the MDPR at specific dechlorination indices with postulated losses
from the sediment of the "lighter" congeners used to determine the index.
Additional comments forthcoming
90
-------
Chpt. 1)
COMMENTS
PI-2, Sect 1-2: The format should have been organized according to work plan
tasks.
Should have a diagram showing tasks in the report and an outline for each
chapter.
This chapter should also have for the report a tasks accomplished plot similar to
a Gant plot.
1.4.4 - Analytical ChemProgram:
How about a flow sheet showing which samples were analyzed and methods for
analyses.
How many congeners were actually identified as occuring in the study ?
Particularly in the sediments ? - chromatograms would have been useful .
Chpt. 2.
P2-3: Air Monitoring for VOCs—were any PCBs detected ?
P2-5: Purchases by GE from 1955-1971 were 97.4% Aroclor 1242,
(50.6.106kg) and 2.6% Aroclor 1254 (1.4.106kg)
(Brown et al., 1988-Science, V240,p. 1675 )
Chpt. 3:
P3-7: Samples should be specified as water samples - grab samples most often
imply solid samples and it's hard to visualize grabbing "dissolved PCBs and ...
This paragraph is confusing - if these samples are not "appropriate" (does this
mean not useful or not unbiased) why are they discussed and then how can they
be "important to reveal possible non-equilibrium... "?
p3-8: Same paragraph as above. State the number of samples used (10 of 18
collected) and refer to the table (Table 1-1). I think the samples from tributaries
since they were collected and analysed for PCBs could also have been used.
Last paragraph. "Appear to represent" or are they a representative set ?
Figure 3-2.: A line graph showing change may have been better for POC
P3-10: Perhaps because theoretical values assumed equilibrium conditions -
deviations from theoretical should be discussed
91
-------
P3-12: Last paragraph. "Dissolved concentration (DOC) should be dissolved
organic carbon
Figures 3-7 to 3-10: Do these represent single water samples taken at a point in
the river at the RM given? Also, it would be helpful to have dates of collection
below transect numbers -particularly since figure represents seasonal data
P3-17: Samples were held for four days- were they kept at 4° C ? Were
measures taken to inhibit biological activity ?
P3-18: Maybe it should be assumed that disequilibrium in nature is the norm-
regarding PCBs: if dechlorination, as an example, is occurring the PCBload is
changing-generally from more chlorinated to less chlorinated congeners which in
turn would alter the pore water concentration, etc. The process is probably not a
slow, continuous reaction throughout the year, but probably goes in spurts as
favorable conditions occur.
Figure 3-13: Why is BZ 52 on this figure?
Figure 3-16: At approximately RM196.8~does this suggest that more Mono, Di
and Tri-s are being put into the water column from sediments? Could this be
related to biological activity in the sediments?
Is this due to input of sediments from Rogers Island?
P3-24: Was this information regarding DOC (4 mg/1) etc. obtained during this
study? I don't recall a prior discussion.
P3-27: What is the relationship of Hudson River [POC] and[DOC] to fitting data
at Green Bay ?
P3-28: Why was BZ 151 used? It occurs in Aroclor 1254 but not at all in A
1242- Yet the emphasis particularly with MDPR and MW is put on Aroclor
1242.
P3-31 Sect 3.1.3: Makes case for flux of PCBs out of sediments from porewater
to H20 column by diffusion and groundwater advection of DOC and dissolved
phases.
P3-35: What specific analytical differences were used by GE and could some of
the Congener analyses from the GE study be used?
P3-38: Could reasons for non-equilibrium be due to addition of PCBs from (a)
other than sediments and (b) movement of specific congener from the sediments
to water column.
92
-------
P3-39: Lower partition coefficients (Koc and Kpoc) mono/di CBs concentration
is enhanced in porewaters relative to higher CI-congeners facilitating loading to
water column from sediments.
P3-40: Were water samples kept on ice or biological activity inhibited in some
way?
P3-53: Regarding suspension of fine grained cohesive sediments-could these
sediments or a fraction of them, go into suspension (before flows reach the
necessary shear stress levels) by impacts and scouring from saltating and
suspended particles?
P3-59: Also, Aroclors 1260 and 1268 approximately 1%; Aroclor 1254
approximately 5% (Brown et al 1988)
P3-124: Were water and sediment samples examined by GE or just water
samples?
P3-125: Report needs chromatograms.
P3-125: Coelution of BZ 4/10 common with DB-1 and DB-5- but why not group
this peak and use in a ratio with other peaks in the chromatogram to measure
change?
Figure 3-83: Follow text (p3-128) and put total PCB's on y axis label.
Figure 3-84 to 3-87: I think the y axis labels need to be corrected to fit the
figure's legends.
P3-128: From these plots (Fig. 3-83 to 3-87) it would be interesting to determine
rates of change between Roger's Island and TI Dam.
Why is there an increase through the winter months. It seems that the curve
would flatten out as biological activity decreased as a result of lower
temperatures. Also, eventually the curve must become asymptotic - could this
possibly be used to get an estimate of the dechlorination rate.
P3-119: Why not use ratios of these congeners (BZ 56, 60, 70 and 74) as a
measurement of dechlorination?
Chpt. 4)
P4-9/4-13: The occurrence (persistence) of wood and wood chips at surface and
to 30cm depth is interesting-does the wood show any signs of degradation and
have PCBs been extracted and analysed from any of the wood debris ?
93
-------
P4-5: 8. Should this read finer-grained sediments with and without higher total
PCB inventories...?
Figure 4-7: Higher PCB levels do correlate with lower DN values, but there is a
lot of scatter- In reality, DN values correspond to sediment type and not to PCBs
- PCBs are more likely to be associated with finer grained sediments - it is a good
approximation tool to help increase the probability of finding sediments with
higher PCB levels.
P4-21: Last two lines: "finer sediments tend to be more easily eroded...".
Would this also be true with clays or clay rich sediments?
Table 4-3: Check the natural log values.
Table 4-9: Notes: b. Refers to Eq 4-7 and 4-8 for definition of terms. These
are not the correct equations (see pg 4-44).
P4-50: 4-3.1: Last line. Dechlorination "reduces" PCB to biphenyl-not
"destroys" it because molecule is still intact -although it's nbo longer a PCB.
P4-51: Last paragraph-J. Brown, 1993 is not in reference list.
P4-54: J. Brown, 1987 reports the occurrence of BZ 54 in Hudson River
sediments, probably a dechlorination product as according to it doesn't occur in
Aroclors 1016, 1242, 1254, 1260, or in Clophen A30, A40, A50, and A60
(Schulz et al 1989).
P4-56: Underestimation could be greater than 5% depending on the amount of
light congeners lost and the MDPR ratio.
P4-57: Aroclor 1242 is not primarily a tri-chlorinated biphenyl mixture but
instead a tetra- and penta-chlorinated mixture. Tetra-CBs comprise about 31%
and penta-CB about 29% of the congeners found in Aroclor 1242 by Frame et
al.,(1996) See attached Table I . Table II indicates the Aroclor 1242 congeners
listed in the report- (Table 4-8) the differences in the congeners found in the same
commercial mixture (Aroclor 1242) Particularly the larger number of Hexa-
and hepta chlorinatedbiphenyls.
P4-57-P4-65: See discussion of MDPR.
Figures 4-23. 4-24: the RMs in the text differ from those on in the figures.
P4-71: Brown and Wagner (1990) have found dechlorination in marine sediments
of the Acushnet Estuary.
94
-------
Vol. 2C, Bk 3/3-Data Usability Report
A.5.2.6: Why was OCN used as a recovery standard? Hexachlorobenzene
(HCB), among others, would have been a good choice. It gives a good ECD
response, and with an HP-5 or Rtx-5 column shouldn't co-elute with any
congener. OCN would have better served as an internal standard. This would
have facilitated gc peak identifications by enabling comparisons relative retention
times determined by Mullin et al.(1984) for all 209 congeners on a similar
column. It seems strange that OCN would breakdown during extraction. Was the
OCN standard chromatographed to determine if it was pure ?
Holding times: Were sediment and water samples maintained at 4°C after
collection and prior to extraction ?
Pg.A-10; B-ll: Why was BZ18 used to differentitate Aroclor 1016 from Aroclor
1242? Why not a ratio of BZ 56 Aroclor 1242 to BZ 56 sample? There is about
30X more BZ 56 in A1242 than in A1016.- Even with extensive dechlorination
(90% ) the ratio would still work and could be used to measure dechlorination.
Table I. Congeners in Arochlor 1242 (wt.% ^ 0.05). Compiled
from Frame et al. (1996).
nCl
IUPAC NUMBER
1
1,3
2
4, 5, 6, 7, 8, 9, 10, 12, 13,15
3
16, 17, 18, 19, 20, 22, 24, 25
26, 27, 28, 29, 31, 32, 33, 35, 37
4
40, 41, 42, 43, 44, 45,46, 47,48, 49, 51, 52,
53, 55, 56, 59, 60, 63, 64, 66, 70, 71, 74, 76, 77
5
82, 83, 84, 85, 86, 87, 89, 91, 92, 95, 97, 99,
101, 102, 105, 109, 110, 114, 115, 118, 119
6
138, 149, 153
95
-------
Table II. Congeners (wt.% ^ 0.05) in Arochlor 1242 from
Report Table 4-8
nCl
IUPAC NUMBER
1
1,3
2
4, 6, 8, 9, 10, 12, 15
3
16, 17, 18, 19, 20, 22, 23, 25
26, 27, 28, 29, 31, 32, 33, 37
4
41,42, 44, 45,47,48, 49,51,52
53, 56, 60, 63, 64, 67, 70, 74, 75, 77
5
82, 83, 84, 85, 91, 92, 95, 96, 97, 99, 101
105, 107 110, 114, 115, 118, 119, 122, 123
6
128, 129, 136, 137, 138, 141, 149, 153, 156, 157, 158, 167
7
170, 178
96
-------
Premeeting Comments Submitted by Reinhard Bierl
-------
Review of the Data Evaluation and Interpretation Report (DEIR)
and Low Resolution Coring Report (LCR) - Hudson River
Reassessment Study
R. Bieri 01/03/99
Hydrological Department, University of Trier, Germany
1. Specific Questions (DEIR)
(1) Is the documented PCB load, which originated from the Tl Pool,
consistent with a source consisting of historically deposited PCB-
contaminated sediments?
I think the data support the assumption that the whole Thompson Island Pool
which includes the upper areas are the main source to the water column load
during low flow periods. It is not clear which parts of the sediments deliver most
of the PCBs, what is the difference between ..historically" deposited PCBs and
recent loadings and what happens during high flow events. Additionally we
have rise the question whether one year of sampling is representative for the
hydrological and geochemical situations in the Hudson River system.
(2) Are two-phase and three-phase partitioning coefficients, derived in the
DEIR, appropriate and do they properly address the physical
parameters of the system (e.g. temperature)?
The theoretical background represented in the report expresses mainly the
state-of-the-art in the scientific literature. The estimated particulate organic car-
bon partition coefficients seem to be reasonable. It would have been necessary
to see the variability of the constants to comment some of the conclusions
-------
which have been made.
Applying three-phase partitioning coefficients would require a much more
detailed analysis of the dissolved organic carbon and colloid contents and
properties of pore and surface water. I think this is far beyond the task of this
study and should be avoided.
(3) Are the conceptual models based on the transect sampling consistent
with the data?
The conceptual models applied to explain the transect data of the 1993
sampling campaign are mainly supported by the data of the Responsiveness
Summary. As in some other parts the discussion in the Responsiveness
Summary is much more detailed and convincing as in the original DEIR report.
Inspite of the quite detailed analysis there remains a kind of unsatisfactory
feeling. First most of the corrections concerning flow has to believed. A detailed
analysis of the flow data would require much more time. Second much effort
has been spent to present and discuss the transect results but no hint is given
towards the variability of the data. Third due to the tremendous variety of data a
multivariate approach would have been a useful approach to interpret congener
specific data.
(4) Does the sampling at bias of samples collected at the Tl Dam-West
sampling location impact EPA's conclusion that the sediments of the Tl
Pool are the major source of PCBs to the freshwater Hudson during
low flow conditions considering the analytical corrections made to
GE's PCB data? What are the other implications of finding higher
concentrations along the shoreline than in the center channel?
As stated in the DEIR Responsiveness Summary, „the net result of combining
-------
and assessing the interpretations of QEA and the phase 2 team is an improved
understanding of the TIP sediment PCB source", the arguments put forward in
the report appear reasonable. The problem of the sampling bias seems to be
compensated by the analytical corrections. Therefore the loadings reported in
the DEIR appear to be in the right order.
The implications of higher concentrations along the shoreline would stress
different areas: the PCB inventory needs probably be revisited. That means a
more detailed geostatistical analysis of the present data emphasizing the
shoreline in a more detailed way. Sampling in the future should also consider
these areas with a more detailed resolution.
Concerning the transport and the exchange mechanisms it would be necessary
to answer a few more questions:
• how are the differences in deposition and resuspension rates between
shoreline and center channel,
• is the biological activity very high,
• will you find forms of biofilms during the months with higher temperatures
with an influence on sorption processes and sediment stability,
• what are the volatilization losses in the shallow parts of the pool.
(5) Are the geostatistical techniques (polygonal clustering and kriging)
correctly applied?
I do not feel as a real expert on geostatistical techniques to say the methods
were not correctly applied. Anyway I have a few comments. Fact is that we
have data of unequal quality, with a spatial variance of the variograms and with
spatial and temporal unfavorable distributed data. In that case it would be
useful to use more advanced geostatistical methods, that means unlinear
methods. The program GeoEAS which has been used for the analysis does not
offer such options. It would be necessary to quantitate the uncertainty of the
estimated values. The variograms which are represented have mostly the
-------
character of white noise which reinforces the need for analysis of the estimation
uncertainty. I think it would be useful, to express the parameters of the
variograms as a function of the spatial coordinates. As far as I understood are
the kriging results not only a matter of estimating the PCB-amount in the Tl
Pool but are also intended to support a redevelopment in the future. Therefore
some additional effort would be quite helpful.
(6) Are the methods applied in the DEIR (change in molecular weight (MW)
and evaluating concentrations of BZ#1,4,8,10 and 19 (MDPR)
appropriate standards for determining extent of dechlorination? Are
there any significant problems with this approach, or more appropriate
approaches?
The approach to take the change in molecular weight (MW) and molar
dechlorination product ratio (MDPR) appears to be reasonable. It gives similar
results concerning the accumulation of ortho-substituted mono- and dichloro -
congeners as has been reported in several papers working with samples of the
Hudson River.
A major restriction to the use of this kind of index is that it depends on the
knowledge of the pure original Aroclor-mixture. I'm not sure if this point is
completely addressed. A second restriction is that the different mobility of the
congeners is neglected that means in reality you will have a chromatographic
effect in sediments.
(7) The DEIR finds that the degree of anaerobic dechlorination is primarly
a function of original concentration rather than time, and accordingly
that there is not significant predictable dechlorination in sediments
containing less than approximately 30 mg/kg PCB. Is this reasonable?
-------
Even when it is reasonable that the adaptation of microorganisms is connected
to certain amounts of PCBs ((1) how comes the PCB-moiecule to the
miroorganism or vice versa; 2) selectivity advantage), there is no indication in
the literature that there exists a single threshold. The factors which control the
dechlorination process are numerous.
A second point which has influenced my opinion is that I could not proof why
quite a huge amount of data have been excluded from this analysis.
2, Specific Questions (LCR)
(1) In the LRC, EPA compared sediment data from cores taken in 1977,
1984 and 1994, which had the PCB analysis conducted by different
laboratory methods. How valid are the methods used to establish a
consistent basis for comparison?
Analysis of environmental samples over a period of nearly 20 years will cause
differences. This is hardly avoidable due to development of analytical
techniques but also due to different sampling techniques and different teams
doing the work. There has been much effort to establish a comparable basis
especially between the 1984 and 1994 data. Despite the quite detailed work of
Butcher (1997) to get a comparison of the non-resolved PCBs I'm not
completely convinced about the correctness of the analytical data. I can
imagine that much of the variance in the analytical data could be a result of
different quantitation methods (e.g. how to draw the baseline in chromatograms
which are a result of different matrix of samples etc.) and of problems with the
use of the surrogate and internal standards. In the Final Phase Working Plan
(1992) it was planned to do some comparable analysis of the older sediment
-------
and water extracts. I haven't seen results to that. When samples (extracts or
original sediment material) had been stored in a suitable way (that means at a
minimum of -20°C, better -40° to -80°C) I would suggest to do this comparison
at least for some of the samples. It would be very interesting to get at least an
impression of the lower chlorinated congeners in the 1984 samples.
Concerning the internal standards I'm aware that for reasons of keeping
methods comparable and perhaps for reasons of cost, OCN and TCMX has
been used but why are not at least some samples analysed by using 13C-
standards to validate the results.
(2) In the upper Hudson River system, it has been well established that
there is significant lateral heterogeneity in sediment concentrations.
While it was attempted to reoccupy previous locations, some
uncertainty is added with respect to the actual sampling location. While
the statistical techniques help compensate for this, how does the
sample heterogeneity affect the comparison of cores from two different
years? Given the spatial variability, is the finding that there is a loss
from most locations supported by the data?
There has been much effort to reoccupy previous sampling locations which
indeed has worked out in many locations. But we have to keep in mind the
governing factors which are responsible for lateral heterogeneity. Depends it
merely on historical" deposition, dynamic exchange or predominately on
sediment parameters like texture (grain size) and total organic carbon content
(TOC). Based on the data in the report and of literature it is obvious that PCB
are mostly connected to fine grained sediments especially in the silt fractions
with second maxima in coarse fractions. This is due to differences in the
properties of the organic matter. The data on organic carbon and nitrogen are
neither complete nor are they precise enough but a careful look on the C/N-
ratio shows at least a trend that low ratios are followed by high contents of
PCBs and high ratios by low contents. Low C/N-ratio expresses organic matter
-------
with less polar hydroxy! and phenolic groups and a more hydrophobic
microenvironent which in fact has consequences on partitioning of hydrophobic
micropollutants like PCBs.
In the LRC Reassessment analysis most of the cluster areas were determined
to be fine-grained sediments. Therefore some of the uncertainty of comparing
data on a „point-to-point" basis was ruled out. Based on the agreement of the
mass change estimates between the two methods the data seem to support
the general loss of PCBs. But I do not believe that a certain amount of loss can
be stated.
(3) What is the impact of the difference between replicate samples in the
1994 sampling effort (36 percent average variability) on the finding that
there was a 40 percent loss of PCB inventory from the highly
contaminated sediments in the Tl Pool?
The relative measures of mass change as calculated by the equation 4.1.4
A = [(MPA^ - MPAm)/ MPAm] * 100%
in the LRC is an accectable means to compensate for some of the uncertainty.
As stated before the uncertainty of the estimated mean values does not allow
to represent mass losses as single values. It's definitely a range and should be
presented as a range.
(4) In the LRC, it was found that Hot Spot 28 contained much more mass
than previous estimates. Is the conclusion that this „gain" is primarly
due to incomplete characterization in 1977 valid?
-------
I think no other explanation is plausible enough. Comparison of the data from
the NYSDEC 1976-78 sediment samples and the TAMS 1994 samples show
considerable differences even at near located sampling points. A shift in the hot
spot itself is certainly possible. A mass loss at other „hot spots" of the pool and
a mass gain exactly at this location below the dam is not very likely. These
would be fundamental events which are not supported by the data of most of
the other hot spots.
(5) Does the data set and its interpretation support the conclusion that
significant losses have occured from hot spots below the Tl Dam?
I do agree with this conclusion that significant losses have occured from hot
spots below the dam. Again I think it is not possible to calculate overall mass
changes but it is sufficient to estimate trends. A more rigorous (geo)statistical
analysis of the data would be perhaps more persuasive. The question however
what mechanism has caused this losses has still to be answered. Losses of
more than 10% in my opinion can only occur due to resuspension and
advective transport during the major runoff events.
(6) The LRC found that the historically contaminated sediments in the Tl
Pool were not universally being buried and sequestered from the
environment How much confidence would you place in the LRC
evidence against widespread burial?
The aspect of burial depends to a certain part on the values for deposition and
resuspension rates which are not known exactly in the different zones.
Depending on the PCB concentration profiles given in the reports there is no
indication that large fractions of the PCB inventory are buried very deeply. This
is supported by the high-resolution cores, by the loss of PCB inventory and by
-------
the fact that a redistribution of sediment-bound PCBs is consistent with the
water-column data.
(7) Is the interpretation of the sidescan sonar data appropriate and
supported by the analysis of the associated sediment properties?
I have no personal experience with sidescan sonar data but the results are very
convincing and the comparison of the SSS-data and the other grain-size data
show that valuable spatial data could be gained.
-------
» V
3. General Questions
(1) Is the data set utilized to prepare the DEIR, LRC and Responsiveness
Summaries sufficient to understand the fate and transport of PCBs in
the Upper Hudson?
The data set is very comprehensive and I think it is enough to get a quite good
impres-sion of the transport behavior of PCBs in the Upper Hudson.
Nevertheless some points - already discussed in the Responsiveness
Summary - concerning for example the exact influence of the Thompson Island
Pool to the overlying water still remain. It would be interesting to see how good
the transport models are working.
Concerning the fate of PCBs I would state the data set it is not sufficient.
USEPA/TAMS suggest some additional work in the modeling and
(eco)toxicology part of the Hudson Reassessment study but these data are not
the basis for this review. There are some compartments which seem to be
addressed only theoretically in the DEIR. To determine exact volatilization rates
of PCBs is certainly a very difficult task but some measurements of
atmospheric PCB contents along the Hudson River would be very helpful. Due
to the frequent situations with pools (very low flow velocity) and dams (spray)
volatilization could be an essential process of PCB removal. Photochemical
destruction of PCBs (both in the surface water and in the atmosphere) is an
important fate mechanism but the overall rate limiting steps are diffusion from
sediments and the transfer at the water/atmosphere boundary.
Another point of concern is the fate of the coplanar congeners. In respect of the
enourmous amounts of PCBs in the Hudson River system the toxicological
importance of the coplanar PCBs can not be ignored. Are there any data on
these congeners and perhaps some recent data on dioxins and furans besides
-------
the data given in Brown et al. (1988)?
Nevertheless I have the feeling that the important issues have been
addressed and mostly answered. Combining all the available and most
recently data from USEPA/TAMS, GE, USGS and the other involved
organizations the database should serve as a qualified basis for the
further reassessment study. So to my opinion the DEIR and LRC are
acceptable.
(2) Are there any additional anaiyses that should be done to verify certain
findings of the DEIR and LRC?
Some of the points have already been addressed in the preceding questions. I
would have liked some additional work on the geostatistical techniques and the
statistics as a whole. In certain parts a multivariate approach would have been
the better choice to come to qualified results. Particularly it could have been
useful to trace sources of the PCBs by a multivariate fingerprinting approach as
well as to analyze the dechlorination patterns.
In general it's a pity that the study didn't use a more multivariate approach
determining for example the main nutrients and metals which could serve as
additional indicators of the system behavior.
I would suggest to do some additional analysis in the estimation of the
sampling and analytical error. The described sampling program for river water
rises some questions. How are the 17 L water results comparable to the 1 L
results taken by GE? How big are the deviations concerning the filtering of
water samples in a dissolved and particulate phase? Is the colloidal phase of
minor importance to transport and is it adequately addressed by the DOC-
content?
-------
As stated earlier I would also suggest, if the storage of the extracts and
samples has been done in an adequate manner, to do some re-analysis of the
old extracts or samples. It could be (or not) a confirmation of the results done
by estimation of the packed column analysis and it could give some indication
of the low-chlorinated congeners in the 1984 samples.
-------
APPENDIX D
LIST OF REGISTERED OBSERVERS OF THE PEER REVIEW MEETING
-------
A United States
P*l^£j Environmental Protection Agency
Region 2
Peer Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Data Evaluation and Interpretation Report
Low Resolution Sediment Coring Report
Albany Marriott
Albany, New York
March 16-18, 1999
Final Observers List
B. Bush
State University of New York
1 University Plaza
Albany, NY 12214-2345
E-mail: bbush2@nycap.rr.com
Jonathan Butcher
Principal Engineer
Tetra Tech
3200 Chapel Hill-Nelson Boulevard
Cape Fear Building - Suite 105
P.O. Box 14409
Research Triangle Park, NC 27709
919-485-8278
Fax: 919-485-8280
E-mail: jo3n@msn.com
John Connolly
Senior Managing Engineer
Quantitative Environmental Analysis, LLC
305 West Grand Avenue
Montvale, NJ 07645
201-930-9890
Fax: 201-930-9805
E-mail: jconnolly@qeallc.com
Kathleen Cooke
Program Associate
Office of Fiscal Research & Policy Analysis
Office of the State Comptroller
A.E. Smith Office Building - 5th Floor
Albany, NY 12236
518-486-5433
Fax: 518-473-1900
E-mail: kcooke@osc.state.ny.us
John Davis
Environmental Scientist
Environmental Protection Bureau
State of New York
Office of Attorney General Eliot Spitzer
120 Broadway
New York, NY 10271
212-416-8482
Fax: 212-416-6007
E-mail: epnjd@oag.state.ny.us
Printed on Recycled Paper
*£RG
-------
James Eldred
Field Assistant
New York State Department of
Environmental Conservation
50 Wolf Road
Albany, NY 12233
Douglas Fischer
Office Regional Counsel
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
212-637-3180
E-mail: fischer.douglas@epa.gov
Kenneth Fish
Staff Chemist
Corporate Remediation Division
General Electric Company
P.O. Box 8
Schenectady, NY 12065
518-387-5990
Fax: 518-387-7611
E-mail: fishkm@crd.ge.com
Ed Garvey
TAMS Consultants, Inc.
300 Broadacres Drive
Bloomfield, NJ 07003
973-338-6680
Fax: 973-338-1052
E-mail: egarvey@tamsconsultants.com
Bob Gibson
Project Manager
General Electric Company
1 Computer Drive, S
Albany, NY 12205
518-458-6633
Fax: 518-458-9247
E-mail: bob.gibson@corporate.ge.com
Elizabeth Grisaru
Assistant Attorney General
Environmental Protection Bureau
State of New York
Office of Attorney General Eliot Spitzer
The Capitol
Albany, NY 12224-0341
518-474-1968
Fax: 518-473-2534
E-mail: elizabeth. grisaru@oag. state, ny. us
John Haggard
Engineering Project Manager
Hudson River Project
Corporate Environmental Programs
General Electric Company
1 Computer Drive, S
Albany, NY 12205
518-458-6619
Fax: 518-458-1014
E-mail: john.haggard@corporate.ge.com
Melvin Hauptman
Emergency & Remedial Response Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
212-637-3952
E-mail: hauptman.mel@epa.gov
Alison Hess
Emergency & Remedial Response Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
212-637-3959
E-mail: hess.alison@epa.gov
Scott Hinz
Environmental Engineer
Limno-Tech, Inc.
5909 Cary Drive
Austin, TX 78757
512-371-1973
Fax: 512-371-1783
E-mail: hinz@limno.com
George Hodgson
Director
Saratoga County Environmental
Management Council
50 West High Street
Ballston Spa, NY 12020
518-884-4778
Fax: 518-885-2220
-------
Damien Hughes
Emergency & Remedial Response Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
212-637-3093
E-mail: hughes.damien@epa.gov
Michael Kane
Biologist
New York State Department of
Environmental Conservation
50 Wolf Road - Room 576
Albany, NY 12233
518-457-6179
Fax: 518-485-2484
E-mail:mwkane@gw.dec.state.ny.us
Bill McCabe
Deputy Director
Emergency & Remedial Response Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
212-637-4405
E-mail: mccabe.bill@epa.gov
Robert Montione
Public Health Specialist
New York State Department of Health
2 University Place - Room 205
Albany, NY 12203
518-458-6316
Fax: 518-458-6372
E-mail: rjm04@health.state.ny.us
William Ports
Senior Engineer
Environmental Remediation
New York State Department of
Environmental Conservation
50 Wolf Road
Albany, NY 12233-7010
518-457-5637
Jim Rhea
Senior Managing Engineer
Quantitative Environmental Analysis, LLC
290 Etwood Davis Road
Liverpool, NY 13088
315-453-9009
Fax: 315-453-9010
Chandler Rowell
Research Scientist
New York State Department
of Environmental Conservation
50 Wolf Road - Room 392
Albany, NY 12233
518-457-7626
Ann Rychlenski
Communication Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
212-637-3672
E-mail: rychlenski.ann@epa.gov
John Santacrose
Counsel
ASNY/AI
P.O. Box 3705
Albany, NY 12203
518-489-9945
Melvin Schweiger
Manager, Hudson River Project
Corporate Environmental Programs
General Electric Company
1 Computer Drive, S
Albany, NY 12205
518-458-6648
Fax: 518-458-1014
William Tate
Environmental Engineer
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
202-260-7052
Fax: 202-260-9830
E-mail: tate.william@epa.gov
Doug Tomchuk
Emergency & Remedial Response Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
212-637-3956
E-mail: tomchuck.douglas@epa.gov
-------
Marty Torray
District Director
Congressman Sweeney's Office
285 Broadway
Saratoga Springs, NY 12866
518-587-9800
Fax: 518-587-1228
C. Kirk Ziegler
Senior Managing Engineer
Quantitative Environmental Analysis, LLC
305 West Grand Avenue
Montvale, NJ 07645
201-930-9890
Fax: 201-930-9805
E-mail: kziegler@qeallc.com
-------
APPENDIX E
AGENDA FOR THE PEER REVIEW MEETING
-------
&EPA
United States
Environmental Protection Agency
Region 2
Peer Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Data Evaluation and Interpretation Report
Low Resolution Sediment Coring Report
Albany Marriott
Albany, New York
March 16-18, 1999
Agenda
Meeting Facilitator Jan Connery, Eastern Research Group, Inc.
Meeting Chair: Ken Reimer, Environmental Sciences Group, The Royal Military College of Canada
TUESDAY. MARCH 16, 1999
7:15AM Registration/Check-in
8:15AM Welcome Remarks and Panel Introduction
Jan Connery, Eastern Research Group, Inc.
8:40AM EPA Overview and Background Remarks
Doug Tomchuk, U.S. Environmental Protection Agency
8:55AM Observer Comments
9:45AM BREAK
10:00AM Presentation on Responsiveness Summary for the Low Resolution
Sediment Coring Report
Ed Garvey, TAMS Consultants, Inc.
11:00AM Charge to the Panel/Summary of Pre-meeting Comments
Ken Reimer, Chair
11:30AM LUNCH (on own)
12:45PM Discussion on Data Evaluation and Interpretation Report (DEIR)
Questions 1 & 2
nrndointcrcMPiptr
%ERS
-------
TUESDAY, MARCH 1 6, 1 9 9 9 (C o n t i n u e d)
2:15PM BREAK
2:30PM Discussion of DEIR Questions 3, 4 & 5
4:00PM Adjourn
WEDNESDAY, MARCH 17, 1999
8:30AM Discussion of DEIR Questions 6 & 7
BREAK
Summary of Discussion of DEIR Questions
L U N C H (on own)
10:30AM
10:45AM
11:45AM
1:00PM
Discussion of Low Resolution Sediment Coring Report (LRC)
Questions 1 & 2
2:30PM BREAK
2:45PM Discussion of LRC Questions 3 & 4
4:15PM BREAK
4:30PM Discussion of LRC Questions 5 & 6
5:15PM Adjourn
THURSDAY, MARCH 18, 1999
8:00AM Discussion of LRC Question 7
8:45AM Summary of Discussion on LRC Questions
9:30AM BREAK
9:45AM Discussion of General Questions 1 & 2
10:45AM Summary of Discussion on General Questions
11:15AM BREAK
11:30AM Observer Comments
12:15PM L U N C H (on own)
1:30PM Recommendations and Chair's Summary
3:15PM Closing Remarks
3:30PM Adjourn
-------
APPENDIX F
SUMMARIES OF OBSERVERS* COMMENTS
-------
List of Observers Who Made Comments
Day 1 (March 16,1999):
George Hodgson, Saratoga County Environmental Management Council
John Connolly, Quantitative Environmental Analysis
Jim Rhea, Quantitative Environmental Analysis
Kirk Ziegler, Quantitative Environmental Analysis
John Haggard, General Electric
Day 3 (March 18,1999):
George Hodgson, Saratoga County Environmental Management Council
Marilyn Pulver, Town of Fort Edward
William Ports, New York Department of Environmental Conservation
Mel Schweiger, General Electric
John Connolly, Quantitative Environmental Analysis
Jim Rhea, Quantitative Environmental Analysis
John Haggard, General Electric Company
The remainder of this appendix summarizes the comments made by the observers listed above.
Comments are summarized in the order in which they were presented As the meeting agenda in
Appendix E shows, observer comments were scheduled only for the first and third day of the
meeting.
-------
Appendix F—Summaries of Observers' Comments
Day 1, Comments from George Hodgson, Saratoga County Environmental Management Council
Mr. Hodgson opened his comments by stating that the Saratoga County Environmental
Management Council is very supportive of the peer review process. He noted that the council
had recommended that EPA conduct an independent peer review for all scientific aspects of the
site reassessment project. Mr. Hodgson then commented that EPA's responses, as presented in
the reports and their responsiveness summaries, should be carefully reviewed because he thought
some findings were "not well-founded and sometimes misleading." Mr. Hodgson then provided
the following examples of findings that he encouraged the reviewers to evaluate critically. Most
of these examples are from the LRC.
First, Mr. Hodgson commented that EPA did not adequately justify that the number of
sediment core samples collected in the 1994 sampling effort were sufficient for estimating changes
in PCB inventory. Mr. Hodgson noted that comparing results from "matched pairs" of sediment
cores was not an acceptable approach for evaluating changes in inventory among the 1,200
samples collected by NYSDEC and the 60 samples collected by EPA Rather, Mr. Hodgson
advocated using a statistical comparison of the means of these sampling efforts (e.g., by
conducting an analysis of variance using an F-test). Mr. Hodgson indicated that such statistical
analyses are critical for determining whether the amount of PCBs in the Hudson River sediments
have truly changed between 1984 and 1994.
Second, Mr. Hodgson noted that EPA did not adequately address concerns raised about
spatial variability of PCB concentrations in the river sediments, particularly in "hot spots." Mr.
Hodgson explained that EPA had identified "large variations" in PCB concentrations in one hot
spot (area H7) but had concluded that this area did not represent most of the other hot spots. To
support the claim that the hot spots are relatively homogeneous, he recommended that EPA
conduct sampling on a fine grid (1- to 2-foot spacing). Mr. Hodgson also noted that EPA did not
respond to comments on the large spatial variations in PCB concentrations depicted in plates 4-21
through 4-28 of the LRC, particularly in plate 4-23.
Third, Mr. Hodgson felt that EPA did not provide a convincing argument to support that
burial of PCBs is not occurring. Mr. Hodgson noted that low resolution sediment coring samples,
which mix the top 9 inches of sediments, are incapable of characterizing how PCB concentrations
vary with sediment depth. Nonetheless, Mr. Hodgson noted that EPA when responding to
comment 4-1.7 on page LRC-20, stated that peak PCB concentrations are "only a few inches"
below the surface Mr. Hodgson did not think the LRC data could support such a finding. He
continued by stating that the high resolution coring actually shows peak PCB concentrations at 6
inches or more below the surface—a depth Mr. Hodgson thought was "likely well below the
active surface layer." Mr. Hodgson also noted that EPA disputes the use of high resolution
coring data to characterize PCB concentrations with depth in areas of the river without fine-
grained sediments. After expressing his concerns about the concentration profiles with depth, Mr.
F-l
-------
Appendix F—Summaries of Observers' Comments
Hodgson recommended that EPA obtain data proving that the peak PCB concentrations occur
within the "active surface layer" of the sediments, rather than inferring this finding from the low
resolution sediment coring results.
Fourth, Mr. Hodgson commented on the mechanisms EPA proposed in the DEIR for how
PCBs transport from the sediments to the water column. Regarding advection of groundwater
through hot spots as a possible mechanism, Mr. Hodgson thought EPA's sample calculations (i.e.,
those that showed "breakthrough" occurring in 25 years) used an assumed advection flow that is
much higher than actual field data generated by GE. Mr. Hodgson noted that the calculated
"breakthrough" time would be an order of magnitude different had EPA used GE's data. Though
he acknowledged that EPA considered GE's field data to be "too meager" for this calculation,
Mr. Hodgson still wondered why EPA chose not to use the only data set that is available on
groundwater advection. Mr. Hodgson concluded by recommending that EPA collect data to
support its sample calculations on groundwater advection.
Day I, Comments from John Connolly, Quantitative Environmental Analysis
Mr. Connolly opened his comments by introducing himself as a consultant for GE and by
noting that he has worked in the field of contaminated sediments for more than 20 years—some of
this experience was gained as an EPA employee. Mr. Connolly then listed several sites on which
he has worked and continues to work. Regarding the Hudson River PCBs site, Mr. Connolly first
congratulated EPA on completing a "very thorough" study of the river and sediments. Mr.
Connolly indicated, however, that some of the conclusions in the DEIR and LRC are incorrect.
Mr. Connolly explained that he and two of his colleagues from Quantitative Environmental
Analysis (QEA) would explain what these incorrect conclusions are.
Mr. Connolly stated that EPA attempted to address "four major issues" in the two reports
under review: (1) identifying the sources of PCBs that pass over the Thompson Island Dam;
(2) determining the fate of PCBs that pass over the Thompson Island Dam; (3) determining the
fete of PCBs in the fine-grain river sediments; and (4) attributing the PCBs in the freshwater
portion of the lower Hudson River to specific sources. Mr. Connolly's comments addressed the
second of these four major issues. In giving his comments, Mr. Connolly referred to a pie
diagram that Doug Tomchuk (EPA) had used during the opening remarks. Mr. Connolly
commented that the pie diagram suggests that the PCB concentration passing Thompson Island
Dam constitutes the "vast majority of the PCBs passing through the freshwater Hudson." Mr.
Connolly noted that this diagram implied that most of the PCBs moving through the freshwater
Hudson could be eliminated by removing the sources of PCBs upstream of the TTD. Mr.
Connolly offered several reasons why he thought such a finding is incorrect.
F-2
-------
Appendix F—Summaries of Observers' Comments
First, Mr. Connolly noted that the pie diagram was based on 1991 sampling data—the last
time a comprehensive sediment survey was conducted. He explained that this data set found, on
average, that the surface sediments in the 5 .9 miles of the TIP contained 19 parts per million
(ppm) of PCBs; he then explained that the same data set found, on average, that surface sediment
in the 34 miles immediately downstream of the TID contained 5 ppm of PCBs. Based on these
average values, Mr. Connolly presented a simple analysis that weighted PCB concentrations by
the lengths of the river over which they were measured. Using this analysis, Mr. Connolly
suggested that 5 ppm of PCBs for 34 miles (5 x 34 = 170) contributes 1.5 times the amount of
PCBs to the water table as 19 ppm of PCBs for 5 .9 miles (19 x 5 .9 = 112). Therefore, Mr.
Connolly concluded that contaminated sediments in the 34 miles downstream of TID could
contribute more PCBs to the water column than the contaminated sediments in the TIP. Though
he acknowledged the shortcomings of this simple analysis of PCBs in the Hudson River, Mr.
Connolly stated that the simple model shows that remediating sediments in the TIP would
probably not solve the PCB contamination problem for the entire river system.
Second, Mr. Connolly commented that the pie diagram used in EPA's opening remarks at
the peer review meeting is based upon "simple accounting." The pie diagram evaluates PCB
concentrations at TID and at Waterford, notes that the concentration profiles of PCB congeners
are similar, and then assumes that the PCBs observed in Waterford must be the PCBs that left the
TID. He compared this reasoning to examining a bank account and concluding that no
transactions had occurred simply because the ending balance is similar to the beginning balance.
Mr. Connolly emphasized that this is flawed logic and explained that one must look at all of the
sources and sinks to understand the fate and transport of PCBs in the Hudson River system. Mr.
Connolly noted that it is a "naive conclusion" to suggest that PCBs transport conservatively from
the TlD to Waterford.
Third, in commenting on the transport of PCBs in the freshwater portion of the Hudson
River, Mr. Connolly again questioned the idea of conservative transport of PCBs. He noted a
contradiction in the reasoning of EPA's reports: Mr. Connolly explained that PCBs could not
both be conservatively transported through the river and be responsible for contaminating the
river sediments.
At the end of his comments, Mr. Connolly noted that two of his colleagues would address
other findings in the DEIR and LRC that QEA questioned. He also noted that a statistician from
Stanford University who was unavailable to attend the peer review meeting provided comments
on the statistical analyses in EPA's reports. These comments were distributed later in the
meeting.
F-3
-------
Appendix F—Summaries of Observers' Comments
Day J, Comments from Jim Rhea, Quantitative Environmental Analysis
Mr. Rhea first acknowledged the difficult task the reviewers face in evaluating the reports,
given their limited experience with this particular site, unlike many of the observers at the meeting
who have spent around 10 years working on the problem of contaminated sediments in the
Hudson River. Mr. Rhea then added that the sediments of the Hudson River are probably the
"most studied" sediments of any site in the United States, by virtue of the various EPA and GE
sampling efforts. Mr. Rhea then commented that he has some concerns regarding the "broad-
based conclusions" drawn by EPA. The remainder of Mr. Rhea's comments focused on the source
of PCBs that pass TID. (This was the first of the four issues raised by Mr. Connolly.)
Referring to a pie chart that EPA had displayed earlier in the meeting, Mr. Rhea stated
that 1.7 parts per billion (ppb) of PCBs were measured in the water passing the TID (or in the
water flowing through the TIP). Mr. Rhea then questioned whether PCBs passing the TID were
associated with recently deposited sediments or with sediments that had been in the TIP for a long
time period. Mr. Rhea indicated that the similar PCB congener profiles between the water
column and the sediments suggests that the sediments likely act as a source of PCBs. He
indicated further that EPA's reports postulated two mechanisms that might account for the PCBs
in the water column: pore water diffusion and resuspension of contaminated sediments.
Regardless of whether the underlying mechanisms are ever fully understood, Mr. Rhea noted that
a more "relevant question" is determining whether the PCBs detected in the water column
originated from contaminated sediments deposited in the last couple of years or from sediments
deposited more than 20 years ago. Mr. Rhea indicated that the answer to this question could have
"tremendous implications" on the effectiveness of source control and sediment remediation in the
Hudson River.
Mr. Rhea noted that the available data suggest that upstream sources have been "a major
factor" in contaminating surface sediments with PCBs. He explained that much of the data are
consistent with "large-scale, external" loads of PCBs to the Hudson River. Mr. Rhea identified
some of these external loads, such as the Allan Mill event in the early 1990s and more recent
loadings of dense nonaqueous phase liquids (DNAPL) from the GE plant to the river, despite
GE's remediation efforts to eliminate such releases. Mr. Rhea indicated that EPA's water column
transects indicate that "plant site loadings" account for half of the total PCBs entering the Hudson
River system. He indicated further that EPA's study shows that most of the PCB loadings are as
"particulate phase PCBs." Mr. Rhea noted that the particulate phase PCBs that enter the TIP
settle in that stretch of the river. Mr. Rhea also indicated that a study conducted by GE found
that DNAPL releases from the GE facilities also would remain confined within the TIP.
In concluding his comments, Mr. Rhea noted that the congener profile of PCBs in the
water column throughout the TIP closely matches the congener profile of the surface sediments.
He emphasized that, because of this, surface sediments are critical for understanding the source of
F-4
-------
Appendix F—Summaries of Observers' Comments
PCBs in the river, regardless of whether PCBs enter the water from diffusive sources or
resuspension. Mr. Rhea maintained that the extent to which upstream sources have impacted
levels of PCBs in surface sediments has important implications for the "final remedy" for the
contaminated sediments.
Day J, Comments from Kirk Ziegler, Quantitative Environmental Analysis
Mr. Ziegler opened his comments by indicating that his expertise is in the fate and
transport of cohesive sediments and that he has worked in this field for more than IS years and
has studied more than 20 river systems. He noted that he has worked on the contaminated
sediments in the Hudson River for 8 years, during which he focused on sediment transport within
the TIP. Mr. Ziegler indicated that his comments would primarily address EPA's findings
regarding the fate and transport of PCBs in the fine-grained sediment areas. (This is the third of
the four issues raised by Mr. Connolly.)
Before critiquing EPA's findings, Mr. Ziegler first restated several of the main conclusions
of the DEIR and LRC (e.g., that PCBs were "somewhat unstable" in the fine-grained sediments
with 40 to 80 percent of the PCB mass lost from selected hot spots over a 10-year period). Mr.
Ziegler then noted that he and his colleagues had done a "tremendous amount of analysis" that
showed that some of the main conclusions are incorrect. He then presented arguments to support
this statement.
First, Mr. Ziegler addressed the stability of the sediments in the Upper Hudson River. Mr.
Ziegler noted that "a very good" side-scan sonar study had been conducted by EPA contractors in
1993. He then indicated that the results of this 1993 bed mapping study were largely consistent
with a 1978 study by NYSDEC to identify hot spots in the river. Mr. Ziegler indicated that the
similarity between these studies, which were conducted IS years apart, suggests that areas of
fine-grained sediments in the Upper Hudson River are "fairly stable."
Second, Mr. Ziegler addressed EPA's finding that "widespread burial" of PCBs is not
occurring in areas of the Hudson River with cohesive sediments. He stated that this finding is
counterintuitive because the Upper Hudson River is a "reservoir" system that has many dredged
channels. He noted further that no "strong perturbations" had occurred in the Upper Hudson
River for many years. Mr. Ziegler thought these observations were inconsistent with a hypothesis
that net burial of sediments is not occurring. Mr. Ziegler then reviewed some of the data
presented in the DEIR and LRC to refute EPA's finding regarding net deposition of sediments.
For example, Mr. Ziegler thought cesium profiles in high resolution sediment cores indicated that
sediments were depositing at a rate between 0.S and 1 centimeters per year in some areas.
Furthermore, Mr. Ziegler indicated that the peak PCB concentrations in many of the high
F-5
-------
Appendix F—Summaries of Observers' Comments
resolution cores "are buried at depth." He also noted that about 70 percent of the sediment cores
that detected certain beryllium isotopes (Be7) were consistent with sediment burial.
Mr. Ziegler concluded by stating that reviewing water column and sediment sampling data
is not sufficient for understanding the fate and transport mechanisms in the Hudson River.
Rather, he suggested that EPA evaluate the data in conjunction with a mass balance modeling
approach that includes the various physical processes as constraints. By this approach, Mr.
Ziegler claimed that he and his colleagues found that net burial is occurring in the river,
particularly in the TIP. Their studies estimated a net sedimentation rate of approximately 0.8
centimeters per year among the fine-grained sediments of the TIP. Mr. Ziegler noted that his
analyses have suggested that approximately "85 percent of the net sedimentation" occurs within
the TIP—a result that he thought was consistent with the behavior of cohesive sediments in "low
energy" areas of rivers.
Day 1, Comments from John Haggard, General Electric Company
Mr. John Haggard, an engineering program manager with GE, began his comments by
noting that EPA had denied his request for making a "lengthy presentation" during the peer
review meeting. Mr. Haggard indicated that he and his colleagues would be available throughout
the peer review meeting to answer any questions the reviewers might have.
The remainder of Mr. Haggard's comments focused on the conclusion from the LRC that
there had been a 40 percent loss of PCBs from the sediments in the TIP over a 20-year period.
Mr. Haggard indicated that EPA based this conclusion on statistical and inferential arguments. To
comment on these arguments, GE hired a statistician, Dr. Paul Switzer of Stanford University, to
review the report. Mr. Haggard indicated that Dr. Switzer basically found that the statistical basis
for EPA's conclusion was not supported by the data. Though Mr. Haggard acknowledged that he
believes a loss of PCBs from the TIP has occurred over the last 20 years, he questioned EPA's
estimates of this loss (40 percent) and wondered how this finding would be used to predict future
conditions in the Hudson River. He indicated that the methods used by TAMS Consultants to
estimate loss of PCBs are not useful for evaluating how PCB levels might change in the future.
Mr. Haggard suggested that the best method for evaluating changes in inventories would be to
use modeling or additional data collection to test hypotheses drawn from the existing data. Mr
Haggard reiterated that he thought the estimate of 40 percent loss of PCBs in the TIP was
incorrect.
Mr. Haggard continued by reading written comments attributed to Dr. Switzer, who was
unable to attend the meeting. Mr. Haggard noted that he would not read the entire set of Dr.
Switzer's comments, but he quoted some passages, such as the "responses to my earlier
comments were disappointing" and "some responses that invoke statistic concepts are inarticulate
F-6
-------
Appendix F—Summaries of Observers' Comments
and meaningless as understood by statisticians, suggesting that responsibility for replying to my
earlier questions and criticisms may not have been entrusted to a professional statistician." Mr.
Haggard stated that Dr. Switzer seemed frustrated that many times his criticisms "were waved
away" when EPA responded to his original comments. Mr. Haggard submitted Dr. Switzer's
written comments to the peer reviewers.
Day 3, Comments from George Hodgson, Saratoga County Environmental Management Council
Mr. Hodgson opened his comments by commending the peer reviewers for their work
during the meeting. Mr. Hodgson's comments, which he made on behalf of the Saratoga County
Environmental Management Council, primarily addressed the peer review process and the
adequacy of the data in the DEIR and LRC for understanding PCB dynamics. Regarding the peer
review, Mr. Hodgson first noted that he was disappointed that the EPA charge did not allow
reviewers to "wrap their arms around" the main issues of PCB dynamics in the Hudson River,
particularly in the TIP. Mr. Hodgson thought understanding the mechanisms of PCB fate and
transport is critical for selecting appropriate remediation alternatives. He felt strongly that no
remedial decisions should be made until these mechanisms are understood.
Mr. Hodgson then addressed several issues regarding data interpretations in the DEIR and
LRC. First, he questioned the reliability of estimates that 40 percent of the PCBs in the TIP were
lost to the water column over a 20-year period. He thought the uncertainties associated with
upstream sources of PCBs, groundwater advection, and depositional and scour areas of the
riverbed prevented such a firm estimate of PCB loss from the sediments. Mr. Hodgson indicated
that the Hudson River is "too important a resource" to make remedial decisions without "having
all the facts" regarding PCB dynamics in the river.
Mr. Hodgson then stated that he and the Saratoga County Environmental Management
Council advocate collecting additional data to understand the system dynamics better. He noted
two specific examples where additional data might be useful: reviewing chromatograms to put
the historical data into perspective; and using "congener fingerprinting" to determine the extent to
which surface sediments, buried sediments, and upstream inputs act as sources of PCBs to the
water in the TIP. Mr. Hodgson commented that these examples raise important questions that
need to be answered.
Mr. Hodgson then returned to his earlier comments on the peer review process. He again
noted that the peer reviewers had "a very strict charge" that limited their evaluations of the DEIR
and LRC. Mr. Hodgson thought this was unfortunate since his organization had recommended
that EPA's peer reviews consider all available data, not just a "snapshot or one side of the
picture." Mr. Hodgson believed that the limited scope of the peer review did not serve the public
interest well. To determine the extent of interaction between GE and EPA on the peer review
F-7
-------
Appendix F—Summaries of Observers' Comments
process and other issues, Mr. Hodgson asked Bill McCabe (EPA) if meetings between GE and
EPA, as discussed in January 1998, had been scheduled. [Mr. McCabe responded that EPA and
General Electric were trying to schedule meetings, but none had been conducted to date ] Mr.
Hodgson responded that he believed "an open dialogue" between the two parties was essential for
reaching scientifically defensible solutions.
As a final comment, Mr. Hodgson stated that the Saratoga County Environmental
Management Council does not think the available data for the Hudson River provide convincing
evidence that the major source of PCBs to the water column is the hot spots in the TIP. He
emphasized that additional data or scientific studies are needed to prove this point to the Council.
Day 3, Comments from Marilyn Pulver, Town of Fort Edward
Ms. Pulver introduced herself as a councilwoman from the town of Fort Edward and as a
farmer. She indicated that she has "been directly involved" with the Hudson River PCBs site for
20 years. Ms. Pulver began her comments by commending the independent reviewers for their
efforts during the meeting and noting that the peer reviewers focused much of their discussions on
what she thought was "the greatest weakness" in EPA's reports: the data uncertainty.
Ms. Pulver continued her comments by indicating that many people who live near the
Upper Hudson River believe the peer reviewers offer the only truly independent evaluation of
EPA's work. She urged the reviewers to make strong recommendations to EPA about reporting
firm conclusions and acknowledging associated uncertainties, instead of misleading the public
with unsupported claims. Ms. Pulver thought such strong recommendations would help others
make sense of the many conflicting studies published to date. She also thought that EPA still has
not provided compelling evidence to support a "20-year-long dredging project."
Ms. Pulver then expressed several concerns regarding the peer review process. After
noting that the reviewers evaluated only those data collected by EPA's contractors, she indicated
that the public would prefer to know "the facts" about the Hudson River, regardless of whether
they were derived from data collected by EPA, GE, or NYSDEC. Ms. Pulver thought a "truly
informed interpretation" of the Hudson River PCBs site must consider data from all available
relevant studies. Ms. Pulver then indicated that she thought the reviewers might have felt
obligated to answer only the questions asked by EPA Ms. Pulver thought the community
members in the Upper Hudson River area deserved a more open peer review process, and she
indicated that the consequences of remediation were too great to have anything less than a
completely open peer review. Ms. Pulver ended her comments by confirming with Bill McCabe
(EPA) that the observers' comments would become part of the peer review record.
F-8
-------
Appendix F—Summaries of Observers' Comments
Day 3, Comments from William Ports, New York State Department of Environmental
Conservation
Mr. Ports introduced himself as a project manager from NYSDEC, and he then
commented on the reviewers' discussions about the comparison of results from a 1984 sediment
survey to EPA's 1994 coring studies. According to Mr. Ports, NYSDEC asked a contractor with
extensive experience with the Hudson River to review EPA's comparison of the two studies. This
contractor reportedly reviewed the two data sets and examined in detail one sediment core from
1983. Mr. Ports noted that some the contractor's interpretations were limited since they were
based on only one sediment core. Nonetheless, Mr. Ports indicated that NYSDEC's contractor
recommended a more detailed evaluation of chromatograms from 1984—a recommendation that
the peer reviewers had made during the second day of the meeting. Mr. Ports then submitted the
contractor's written comments and thanked the reviewers for their efforts during the meeting.
Day 3, Comments from Mel Schweiger, General Electric
Mr. Schweiger introduced himself as the manager of the "Hudson River research and
remediation project" for GE. Mr. Schweiger then noted that he and his colleagues were
frustrated during the peer review meeting because they were not given the opportunity to take
part in the deliberations among the reviewers. Nonetheless, he commended the reviewers for their
efforts and began to offer technical comments.
Mr. Schweiger indicated that GE has been asking EPA to conduct scientific peer review of
its research for the last 8 years. Mr. Schweiger considered independent peer review to be an
important facet of scientific studies. He then praised the review process by providing an example
of how the reviewers identified critical flaws in the LRC. More specifically, Mr. Schweiger noted
that EPA, shortly after releasing the LRC, claimed that 40 percent of the PCBs had "washed out
of hot spots" in the TIP over a span of 10 years. Mr. Schweiger indicated that EPA said this 40
percent loss was a "rock solid" estimate and that the loss of PCBs to the water column might
necessitate immediate emergency actions. On the other hand, Mr. Schweiger noted that two
scientists hired by GE concluded that EPA had "no factual basis" for its estimated loss of PCBs
from the river sediments—a finding that was presented to EPA. Mr. Schweiger then
acknowledged that the reviewers reached a similar conclusion as GE: some PCBs were lost from
the sediments, but the 40 percent loss estimate was unfounded.
Mr. Schweiger used the following quote from the LRC Responsiveness Summary as
evidence that EPA concurs with the findings of GE and the peer reviewers regarding the loss of
PCBs from contaminated sediments: "EPA acknowledges that there is considerable uncertainty
surrounding the loss values in these estimates, but stresses that there is statistically significant loss
despite this uncertainty." Mr. Schweiger paraphrased the statement, finding that what EPA
F-9
-------
Appendix F—Summaries of Observers' Comments
presented as fact previously, it now presents as very uncertain. Mr. Schweiger then emphasized
that he presented this example not as a criticism of EPA, but as praise of the peer review process
in causing EPA to revisit its conclusions.
Mr. Schweiger concluded his comments by asking the peer reviewers to consider carefully
the comments that his colleagues (John Connolly and Jim Rhea) were about to present. He noted
in particular that his colleagues would address the extent to which the TIP sediments contribute to
PCBs in the freshwater Hudson and whether widespread burial of PCBs occurs in the TIP. Mr.
Schweiger indicated that EPA's own data, as well as data not yet presented at the meeting,
suggest that widespread burial of PCBs does occur.
Day 3, Comments from John Connolly, Quantitative Environmental Analysis
Mr. Connolly's comments addressed what he considered one minor issue (uncertainty in
EPA's data) and one major issue (whether "widespread burial" of PCBs occurs). Regarding the
data uncertainty, Mr. Connolly acknowledged that the reviewers discussed this issue extensively,
but he wanted to clarify another aspect of the issue. His overall point was that EPA's reported
value of PCB loss from the TIP did not fully account for the uncertainty among the data. To
illustrate this point, Mr. Connolly detailed step by step the method EPA used to estimate losses of
PCBs. He explained that EPA's estimates of PCB loss did account for the uncertainty associated
with calculating mean inventory levels from individual sediment coring measurements; he also
noted, however, that EPA's estimates did not consider the variability inherent in each individual
measurement. By ignoring this uncertainty, Mr. Connolly thought EPA's range of estimated PCB
loss (4 to 59 percent) likely understated the actual range of PCB loss indicated by the data. Mr.
Connolly indicated that, when all uncertainties are considered in calculations, changes in PCB
inventory over time might actually range from 100 percent loss (complete loss of PCBs in the
sediments) to a SO percent gain (an increase in PCBs in the sediments). He thought this broader
range underscored an important finding: EPA's reports "considerably underestimate the
uncertainty" associated with estimates of PCB mass loss.
Regarding EPA's conclusion that "widespread burial" of PCBs did not occur in the TIP,
Mr. Connolly provided a series of arguments to refute this finding. Mr. Connolly addressed this
issue by first defining two criteria that he thought must be met to reach conclusion in scientific
documents: a conclusion must be stated as clearly as possible to avoid misinterpretations, and a
conclusion must be supported by the available data. On the topic of the clarity of EPA's
conclusions, Mr. Connolly indicated that the terms "widespread" and "burial" are both subjective
and open to many different interpretations. Using an analogy of flu epidemics, Mr. Connolly
indicated that "widespread" could be carefully defined (e.g., 25 percent of a population is
affected) or the term could be vaguely defined, thus causing observers to have different
interpretations. He reiterated that EPA's use of the term "widespread" in the LRC was open to
F-10
-------
Appendix F—Summaries of Observers' Comments
multiple interpretations. Mr. Connolly then explained several possible interpretations of "burial":
new sediments gradually bury old sediments with time or new sediments have moved PCBs to
depths below "some bioavailable layer." Noting that these two interpretations have considerably
different implications, Mr. Connolly concluded that EPA used a subjective term ("burial") in
presenting a major conclusion of the LRC. Overall, Mr. Connolly indicated that the vaguely
defined terms in EPA's conclusions violated his first criterion for making conclusions in scientific
documents.
Mr. Connolly then provided two arguments to suggest that the conclusion of "widespread
burial" is not supported by the data—his second criteria for accurate, scientific conclusions. First,
Mr. Connolly critiqued EPA's interpretations of the beryllium measurements. He explained that
these measurements can indicate deposition of sediments, but only when beryllium is detected; he
noted that no conclusions regarding sediment deposition can be made when beryllium is not
detected. Mr. Connolly continued by indicating that beryllium was detected in the sediments at 70
percent of the locations that EPA sampled. Since beryllium is an indicator of sediment deposition,
Mr. Connolly thought this frequency of detection offered "definitive evidence" that sediments are
depositing in the Upper Hudson River.
Second, Mr. Connolly reviewed EPA's interpretations on how PCB concentrations varied
with sediment depth and offered an alternative interpretation that he thought was better supported
by the data. To address this issue of PCB "burial," Mr. Connolly first provided an overview of
the LRC results: in 60 percent of the samples, the highest PCB concentration was observed to be
in the top 9 inches of sediment; in the remaining 40 percent of the samples, the highest PCB
concentration was observed in deeper sediments. Mr. Connolly asked the reviewers whether they
could truly infer that "widespread burial" occurs when roughly half of the samples had maximum
PCB concentrations at depths more than 9 inches below the surface. Using the profile of PCB
concentrations for a single high resolution sediment core, Mr. Connolly showed how EPA's low
resolution cores might not be sufficient for commenting on burial: a high resolution core might
show a maximum PCB concentration between 8 and 9 inches below the surface, but a low
resolution core taken at the same location would only show that the maximum occurred within the
top 9 inches. In short, he emphasized that the low resolution cores cannot distinguish sediments
with maximum PCB concentrations that occur up to 9 inches below the surface from those where
maximum concentrations occur at the surface layer. Mr. Connolly thought this shortcoming was
very important to future evaluations on the bioavailability of the PCBs. He concluded this
discussion by presenting what he thought was a more accurate summary of the LRC data: in 60
percent of the samples collected, the highest concentrations of PCBs, which presumably were
deposited in the 1970s, occurred within the upper 9 inches of the sediments.
As a final point, noting that EPA's finding of "no widespread burial" has potentially
important implications, Mr. Connolly indicated that additional studies are first needed to verify
EPA's findings. He thought such studies might include collecting additional high resolution
F-l 1
-------
Appendix F—Summaries of Observers* Comments
sediment cores, evaluating the depth over which bioturbation affects mixing of sediments, and
more thoroughly characterizing the stability of the sediments. Mr. Connolly concluded his
comments by urging the reviewers to consider his many concerns regarding EPA's conclusion that
"widespread burial" of PCBs does not occur.
Day 3, Comments from Jim Rhea, Quantitative Environmental Analysis
Mr. Rhea's comments focused on clarifying two interpretations of the water column
sampling data: the extent to which the sediments of the TIP act as a source of PCBs to the
freshwater Hudson River and the impact EPA's correction factors had on monitoring data
collected at stations with sampling biases. Mr. Rhea indicated that he tried to offer the following
remarks during the peer reviewers' discussions.
First, Mr. Rhea noted that EPA based its conclusion that the sediments of the TIP were a
primary source of PCBs to the freshwater Hudson strictly on the water column transect data from
1993. He noted further that a large volume of water column data have since been collected,
including weekly sampling results from stations at Fort Edward, the TID, and another
downstream location. Mr. Rhea said the more recent data collected by GE suggest a different
spatial profile of PCB concentrations than reported in the DEIR, and he suspected this difference
was largely due to a sampling bias in EPA's work (which he explained in greater detail later in his
comments). In short, Mr. Rhea thought the GE data show that the sediments downstream of the
TIP are just as much a source of PCBs to the water column as are the sediments within the TIP
itself. To support this claim, Mr. Rhea presented a chart summarizing a subset of monitoring
results from GE's sampling stations, which he indicate are operated as part of a consent decree
with oversight by EPA The chart showed a roughly linear increase in PCB concentrations in the
water from the Fort Edward station all the way to the Schuylerville station. Mr. Rhea reiterated
that this gradual increase suggests that the sediments within the TIP as well as the sediments
immediately downstream of the TIP act as comparable sources of PCBs to the water. Mr. Rhea
did not find this result surprising, since river sediments from the TIP through Schuylerville are
known to be contaminated with PCBs. Based on his arguments, Mr. Rhea asked the reviewers to
clarify in their final comments the extent to which sediments in the TIP act as a source of PCBs.
Second, Mr. Rhea discussed the corrections that EPA made to its transect study, as
presented by Ed Garvey during the peer review meeting. Before commenting in detail on the
correction factors, Mr. Rhea first presented a map of the various sampling stations that had been
operated in the vicinity of the Thompson Island Dam. He pointed out the location of GE's former
sampling station at the dam's ''Western Wing Wall" (a station that was found to have a sampling
bias) as well as GE's current sampling station at a downstream location (a station that is believed
to be unbiased). Mr. Rhea then explained how EPA's contractors derived a correction factor
based on data from the two GE sampling stations and applied this factor to data collected at
F-12
-------
Appendix F—Summaries of Observers' Comments
EPA's sampling station, which was approximately one-quarter mile upstream from the TTD. Mr.
Rhea thought an implicit assumption in applying this correction factor was that the PCB
concentrations at the EPA sampling station must be similar to those at GE's former sampling
station. After presenting limited sampling data indicating that PCB concentrations at these two
locations are indeed different, Mr. Rhea concluded that EPA's contractors had not appropriately
applied the correction factor that they derived. He noted further that he did not think enough data
were available to derive an appropriate and defensible correction factor for the EPA sampling
station. To summarize the significance of the correction factors, Mr. Rhea presented a slide
showing an EPA water column transect based on the correction factor and he then showed a
different slide showing water column PCB concentrations that were "unbiased." The data
presented on the slides had notably different implications as far as what sediments acted as
sources of PCB s to the water column.
In closing his comments, Mr. Rhea commended the peer reviewers for their efforts in
critiquing EPA's reports. He acknowledged that the reviewers had a difficult task, especially
considering that many scientists at the meeting have worked on the Hudson River PCBs site for
more than 10 years.
Day 3, Comments from John Haggard, General Electric
Mr. Haggard introduced himself as the technical program manager for a GE team that
conducts research on the Hudson River. As a general comment on the peer review process, Mr.
Haggard expressed his frustration that GE was not allowed to present material during the peer
reviewers' discussions. He indicated that many people associated with GE had potentially
valuable contributions to the peer review by virtue of their many years of experience working on
the Hudson River. Mr. Haggard thought GE's exclusion from the peer review discussions hurt
the overall process.
Before providing his comments on the DEER and LRC, Mr. Haggard pointed out that
some of the reviewers' findings were consistent with "key issues" that GE had raised with EPA.
As an example, he indicated that both GE and the reviewers had concerns about the statistical
techniques used to compare sediment inventories from different years and the conclusions stated
in the reports (particularly the estimates of PCB loss from the sediments) without "proper
qualifications." The remainder of Mr. Haggard's comments focused on three technical issues.
First, Mr. Haggard addressed the issue of PCB loss from sediments downstream of the
TID. He noted that only two sets of data are available (data from 1977 and data from 1994) for
estimating PCB loss for this stretch of the river. Mr. Haggard recalled that the reviewers were
skeptical about the reliability of the 1977 data set, and he thought the reviewers should comment
further about what this unreliability means in terms of estimating losses of PCBs from sediments.
F-13
-------
Appendix F—Summaries of Observers' Comments
In short, Mr. Haggard felt that estimated losses of PCBs for the area downstream of the T1D
might be "somewhat unreliable," given the concerns the reviewers had about the 1977 data set.
Mr. Haggard requested the reviewers to reconsider this issue in their final deliberations.
Second, Mr. Haggard addressed EPA's conclusions regarding losses of PCBs from the
sediments in the TIP. Mr. Haggard acknowledged that even GE's reports have concluded that
surface sediments in this part of the river clearly add PCBs to the water column; however, he
found EPA's quantitative estimates of PCB losses from sediments to be unconvincing. Since the
reviewers suggested that EPA's estimates should be reported as ranges instead of point estimates,
Mr. Haggard wondered whether a range of 5 to 59 percent would be really meaningful. More
specifically, he explained that this range spans virtually no PCB loss (S percent loss) to more than
half of the PCBs being lost (59 percent loss). Rather than relying on this range, Mr. Haggard
instead recommended an analysis of fate and transport processes and water column measurements
to complete a "real mass balance" for the PCBs. He thought this type of analysis would generate
a less uncertain estimate of PCB losses from the river sediments.
Third, Mr. Haggard asked the reviewers to consider in their final discussions how the
conclusions of the LRC and DEIR will be used in the future—an issue that was raised by Dr. Ron
Mitchum (a peer reviewer) during the meeting. Mr. Haggard recalled that the peer reviewers did
not discuss this topic thoroughly during the meeting because it was not part of the charge. Noting
that the reviewers' conclusions might eventually be used to make press statements and remedial
decisions, Mr. Haggard urged the reviewers to state their conclusions with appropriate qualifiers.
Mr. Haggard closed his comments by commending the reviewers for evaluating EPA's
reports and managing to complete the peer review during the 3-day meeting.
F-14
-------
APPENDIX G
MINUTES FROM THE JANUARY 1999 BRIEFING MEETING
-------
Minutes from the Briefing and Site Visit for the Peer Review of the
Data Evaluation and Interpretation Report and the Low Resolution
Sediment Coring Report for the Hudson River PCBs Superfund Site
On January 11-12, 1999, Eastern Research Group, Inc. (ERG), conducted a meeting to provide
six independent reviewers with background information on two reports that were prepared for the
Hudson River PCBs Superfund site, and ERG organized a site visit of the Upper Hudson River to
familiarize the reviewers with the site. The reports of concern were the Data Evaluation and
Interpretation Report (DEIR) and the Low Resolution Sediment Coring Report (LRC). A copy
of the agenda of the meeting and site visit is among the Attachments to these minutes. The
meeting, which was open to the public, was facilitated by ERG and attended by the reviewers,
representatives from the U.S. Environmental Protection Agency (EPA), representatives from
EPA's contractors, and approximately 20 observers. The Attachments to these minutes list the
meeting attendees. Though observers were invited to attend the site visit of the Upper Hudson
River, none did so. The remainder of these minutes briefly summarizes the presentations made
during the meeting and site visit, in the order that the presentations were given.
Ms. Jan Connery (ERG), meeting facilitator, welcome remarks and introduction. Ms. Jan
Connery opened the meeting by welcoming the observers and describing the purpose of the
briefing and site visit: to provide background and context for the reviewers such that they
understand the site history and the scope of the reports. Ms. Connery emphasized that the
briefing was not the actual peer review of the documents; she noted that the peer review was
scheduled to take place in March 1999. Ms. Connery then reviewed the agenda for the two-day
meeting. During Ms. Connery's presentation, the reviewers, representatives from EPA, and
representatives from EPA's contractors introduced themselves.
Mr. Doug Tomchuk (EPA), overview presentation. Mr. Doug Tomchuk's presentation
outlined the history of EPA's involvement with the Hudson River PCBs Superfund site. As
means of introduction, Mr. Tomchuk presented a series of slides that showed the setting of the
Upper Hudson River. He then discussed historical releases of PCBs to the river, as well as
controls that have been implemented to minimize them. Mr. Tomchuk explained that EPA has
been involved with this site for many years and indicated that, in 1984, the Agency reached an
interim "No Action" decision for the contaminated sediments of the Upper Hudson River. He
noted that current work on the site is part of EPA's Reassessment of the previous decision.
Mr. Tomchuk then gave an overview of the purpose and organization of the Reassessment. He
explained that the Reassessment is being conducted in three phases and that the DEIR and LRC
were prepared as part of Phase 2. To provide additional context for the reviewers, Mr. Tomchuk
described the scope of the six reports that comprise Phase 2 of the Reassessment. Focusing
specifically on the DEIR and LRC, Mr. Tomchuk reviewed the four major conclusions of both
reports. He also reviewed the schedule for completing the Reassessment.
-------
Minutes from the Briefing and Site Visit for the Peer Review of the
Data Evaluation and Interpretation Report and the Low Resolution
Sediment Coring Report for the Hudson River PCBs Superfund Site
—Continued—
Mr. Tomchuk closed his presentation by describing relevant aspects of the Superfund process,
such as EPA's criteria for selecting remedies for a given site and EPA's general decision making
process for Superfund. The reviewers did not have any questions about the presentation.
Mr. Damien Hughes (EPA), overview of the charge to the reviewers. Mr. D ami en Hughes
stepped through the charge to the reviewers, a copy of which is included in the Attachments to
these minutes. Mr. Hughes explained the purpose of the peer review and the charge to the
reviewers, and he indicated that EPA will likely not collect more data for this site as part of the
Reassessment. During his presentation, Mr. Hughes reviewed every question in the charge and
answered several of the reviewers' questions regarding the charge. These questions addressed the
meaning of "conceptual models" (see Question 3 in the charge), the need to address data quality
during the peer review, and the process by which reviewers should ask questions in the time
between the briefing meeting and the actual peer review.
Dr. Ed Garvey (TAMS Consultants, Inc.), presentation on the DEIR and LRC. Dr. Ed
Garvey, an author of the DEIR and LRC, provided a detailed overview of the content of the
reports. Starting with the DEIR, Dr. Garvey first reviewed the four major conclusions of the
report. For background purposes, he listed the various data collection efforts (i.e., water column
monitoring, sediment coring, geophysical surveying) that were considered in the Phase 2 reports.
Dr. Garvey then gave an overview of six subject areas covered by the DEIR. A summary of this
overview follows:
• Water column sampling data. Dr. Garvey noted that EPA, General Electric Company
(GE), and the United States Geological Survey (USGS) have all collected water column
samples in the Hudson River. Dr. Garvey then explained the scope and monitoring
locations of the transect sampling and flow-average sampling performed for the DEIR,
during which he provided background information on the sampling and analytical
methods. Dr. Garvey then showed selected figures from the DEIR to provide an overview
of the water column sampling results. Dr. Garvey informed the reviewers that some of the
data presented in the DEIR has been modified in the Responsiveness Summary—a
document that was distributed to the reviewers later in the meeting.
• High resolution sediment coring. Dr. Garvey indicated that high resolution sediment
coring was performed to establish a time history of PCBs in the river sediments and to
characterize spatial variations in PCB concentrations. Dr. Garvey then described the
sampling and analytical methods used to measure sediment concentrations and explained
how selected radionuclides were analyzed to "date" the sediments. To illustrate the main
findings derived from the high resolution sediment coring results, Dr. Garvey presented a
series of figures from the DEIR and commented on selected data trends. In response to a
-------
Minutes from the Briefing and Site Visit for the Peer Review of the
Data Evaluation and Interpretation Report and the Low Resolution
Sediment Coring Report for the Hudson River PCBs Superfund Site
—Continued—
reviewer's question, Dr. Garvey indicated that he did not have a slide that showed PCB
concentration profiles for sediments in the Thompson Island Pool.
• Equilibrium partition coefficients. Dr. Garvey indicated that part of the scope of the
DEER, was to derive equilibrium partition coefficients for PCBs in the water column. He
presented some of the coefficients that were calculated from the water column monitoring
data, and indicated that the DEIR presents additional sets of coefficients, including those
with corrections for temperature and those with corrections for dissolved organic carbon.
• Geophysical analyses. Dr. Garvey reviewed the geophysical analyses (the sidescansonar
characterization of the river bed) that were documented in the DEIR. He explained how
these analyses helped generate a two-dimensional map of the river sediments and how
sediment "confirmation cores" were collected to verify the findings of the sidescansonar
data. Dr. Garvey presented some results from the geophysical analyses and explained how
the acoustical signal from the sidescansonar study relates to PCB concentrations in
sediments.
• Anaerobic dechlorination. Dr. Garvey opened his discussion on anaerobic dechlorination
by briefly describing the microbial processes that are believed to occur in the Hudson
River. After defining the two measures used in the DEIR to characterize the extent of
anaerobic dechlorination, Dr. Garvey showed a series of plots from the report that
illustrate its main findings with respect to anaerobic dechlorination.
• Geostatistical analyses. Dr. Garvey indicated that geostatistical analyses, namely
polygonal declustering and kriging, were used to estimate PCB inventory from sediment
samples collected in 1984 by the New York State Department of Environmental
Conservation (NYSDEC). Dr. Garvey then gave some background information on the
techniques and how they were applied to the NYSDEC sampling effort. Finally, he
presented some results of the geostatistical analyses that were cited in the DEIR.
After elaborating on these six topics, Dr. Garvey reiterated the four major conclusions of the
DEIR. Dr. Garvey informed the reviewers that some of the analyses in the report had been
revised in a Responsiveness Summary—a topic that he would revisit later in the meeting.
Following the presentation on the DEIR, Dr. Garvey provided background information on the
LRC. Dr. Garvey first explained that the LRC focuses primarily on estimating PCB inventory in
the river sediments. Dr. Garvey then reviewed the four major findings of the LRC and described
salient features of the sampling and analytical methods used to collect and analyze low resolution
sediment cores. He also reviewed some aspects of the data quality from this study, such as
relative percent differences (RPDs) from the coring samples, and answered a reviewer's questions
-------
Minutes from the Briefing and Site Visit for the Peer Review of the
Data Evaluation and Interpretation Report and the Low Resolution
Sediment Coring Report for the Hudson River PCBs Superfund Site
—Continued—
about how the RPDs were calculated. Dr. Garvey then gave an overview of three subject areas
covered by the LRC. A summary of this overview follows:
• Consistency with findings of the DEIR. Dr. Garvey presented two examples of how the
findings documented in the LRC were, to a certain extent, consistent with those
documented in the DEIR. First, he presented a graph which showed that the extent of
anaerobic dechlorination observed in the low resolution sediment cores was consistent, in
a general sense, with that observed in the high resolution sediment cores. Second, he
presented some data that suggested that the low resolution sediment coring data were
generally consistent with the sidescansonar data.
• Change in PCB inventory in the sediments in the Thompson Island Pool. Dr. Garvey
described the methods by which changes in PCB inventory were calculated for the
sediments in the Thompson Island Pool; he also indicated that the Responsiveness
Summaries will document additional methods for calculating these changes. Dr. Garvey
then presented a series of slides illustrating the calculation methods and their findings.
Though he noted that the Responsiveness Summary which had not yet been released might
modify the findings, Dr. Garvey indicated that the LRC reported an estimated 40 percent
loss of PCBs from hot spots in the Thompson Island Pool, and he noted that this estimate
has considerable uncertainty associated with it.
• Change in PCB inventory in the sediments downstream of the Thompson Island Pool.
Dr. Garvey gave an overview of the procedure documented in the LRC for comparing the
1977 sediment coring results to the 1994 sediment coring results. Dr. Garvey presented
several figures that indicated a statistically significant loss of PCBs from several hot spots
downstream of the Thompson Island Pool. Dr. Garvey also presented data suggesting
that hot spot 28 gained a large mass of PCBs between 1977 and 1994; however, he noted
that the LRC provides evidence that this apparent gain is likely due to incomplete
characterization of the hot spot during the 1977 sampling effort.
After reviewing these topics addressed in the LRC, Dr. Garvey identified several subjects in the
reports that will likely be revised in the Responsiveness Summary expected to be released in
February 1999. Dr. Garvey then answered reviewers' questions regarding the availability of data
characterizing grain size distribution for the river sediments, measurements of biphenyl or other
compounds in the sediment cores, and consideration of aerobic degradation in the reports.
Mr. Doug Tomchuk (EPA), additional comments. Following the first half of Dr. Garvey's
presentation, Mr. Tomchuk informed the reviewers that some conclusions in the DEIR and LRC
were slightly modified, and these modifications were documented in Responsiveness Summaries
for the reports. For additional background on the documents, Mr. Tomchuk also informed the
-------
Minutes from the Briefing and Site Visit for the Peer Review of the
Data Evaluation and Interpretation Report and the Low Resolution
Sediment Coring Report for the Hudson River PCBs Superfund Site
—Continued—
reviewers of EPA's extensive "community interaction program" for the Hudson River PCBs site.
As examples, he noted that all documents go through both extensive internal reviews and reviews
by other agencies (e.g., NYSDEC), as well as public comment periods, and he explained that
"peer input" has occurred throughout the process through use of a "scientific and technical
committee." Mr. Tomchuk indicated, however, that these various reviews and peer inputs are not
necessarily independent.
Mr. Damien Hughes (EPA), additional comments. Prior to Dr. Garvey's final presentation,
Mr. Hughes indicated that independent peer review was an important part of the Reassessment,
since the peer review would help ensure that the reports prepared for EPA were based on sound
scientific principles. Mr. Hughes reminded the reviewers that they should feel free to comment on
all aspects of the reports during their peer review, including topics that may not be addressed in
the specific questions in the charge. Finally, Mr. Hughes urged the reviewers to maintain their
independence during the review process and to contact ERG directly with any inquiries they might
have prior to the peer review meeting.
Dr. Ed Garvey (TAMS Consultants, Inc.), presentation on the Responsiveness Summaries
for the reports. Dr. Garvey's final presentation addressed revisions that were made, or that were
being made, to the DEIR and LRC in the documents' Responsiveness Summaries. He discussed
several reasons why revisions were necessary: analytical corrections that were made to GE's data
set; corrections for sampling biases at a sampling location near the Thompson Island Dam;
corrections to the river flow data that were originally used in the DEIR; and revisions to selected
statistical analyses presented in the reports. Dr. Garvey noted that the overall impact of these
revisions is to be documented in the two different volumes of Responsiveness Summaries. Dr.
Garvey closed his presentation by answering reviewers' questions, which addressed quality
assurance, methods for collecting water samples, sedimentation rates, consideration of air
sampling, annual loads of PCBs in the water column, and ranges of suspended solid
concentrations.
Mr. Doug Tomchuk (EPA), site tour of the Upper Hudson River. The briefing meeting
closed with Mr. Tomchuk outlining the itinerary for the site visit of the Upper Hudson River. Mr.
Tomchuk identified the six locations that the reviewers would see. Observers were invited to
follow on this site visit, but none did so. After Mr. Tomchuk's brief presentation, the reviewers
boarded a bus and visited the following six locations along the Upper Hudson River:
• An observation point immediately downstream from Bakers Falls and directly across the
Hudson River from GE's Hudson Falls plant.
• An overlook of the Hudson River, near a former outfall from GE's Fort Edward plant.
-------
Minutes from the Briefing and Site Visit for the Peer Review of the
Data Evaluation and Interpretation Report and the Low Resolution
Sediment Coring Report for the Hudson River PCBs Superfund Site
—Continued—
• An overlook of the Hudson River, directly across from capped remnant deposit #4, and
upstream from the former Fort Edward Dam and Rogers Island.
• The northern tip of Rogers Island.
• The western wall of the Thompson Island Dam.
• Lock #5 on the Hudson River.
The briefing and site visit for the peer review of the DEIR and LRC ended after the reviewers
visited these six locations.
301263
-------
Minutes from the Briefing and Site Visit for the Peer Review of the
Data Evaluation and Interpretation Report and the Low Resolution
Sediment Coring Report for the Hudson River PCBs Superfund Site
ATTACHMENTS
1.
Meeting agenda
2.
Peer reviewers
3.
ERG onsite conference support staff
4.
EPA and contractor participants
5.
Observers
6.
Charge to the reviewers
-------
Mk United States
ImMII Environmental Protection Agency
#% Region 2
Briefing and Site Visit for the
Peer Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Holiday Inn Turf on Wolf Road
Albany, New York
January 11-12, 1999
Agenda
MONDAY, JANUARY 11, 1999
11:00AM Lunch (on own)
12:00N Registration/Check-In
12:30PM Welcome Remarks and Introduction
Jan Connery, Eastern Research Group, Inc. (ERG)
12:45PM Overview Presentation
Doug Tomchuk, U.S. Environmental Protection Agency (EPA)
1:45PM Presentation on Data Evaluation and Interpretation Report (DEIR) and the
Low Resolution Coring Report (LRC)
Ed Garvey, TAMS Consultants, Inc.
3:00PM
BREAK
3:15PM
Presentation Continues
4:30PM Adjourn
TUESDAY, JANUARY 12, 1999
8:30AM Presentation on Responsiveness Summaries: DEIR and LRC
Ed Garvey
11:30AM Adjourn
11:35AM Lunch (on own)
12:30PM Site Tour and Briefing (Albany to Hudson Falls to Albany)
4:30PM End of Site Tour at Holiday Inn Turf, Albany
-------
United States
laUll Environmental Protection Agency
Region 2
Briefing and Site Visit for the Peer
Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Holiday Inn Turf on Wolf Road
Albany, New York
January 11-12, 1999
Peer Reviewers
Dr. Reinhard Bieii
Senior Lecturer
University of Trier
FB VI - Hydrology
UniversitStsring 15
D-54286 Trier GERMANY
49-651-201-2229
Fax: 49-651-201-3978
E-Mail: bier1@uni-trier.de
Dr. Ronald Mitchum
President
Data Analysis Technologies, Inc.
6385 Shier Rings Road
Dublin, OH 43016
614-791-8008
Fax: 614-791-8007
E-Mail: datlab@infinet.com
Dr. Per Larsson
Assistant Professor
Lund University
Ecotoxicology, Department of Ecology
Ecology Building
Sdlvegatan 37
Lund, SWEDEN 223 62
46-46-222-3779
Fax: 46-46-222-3790
E-Mail: Per.Larsson@ecotox.lu.se
Dr. Ken Reimer
Director, Environmental Sciences Group
The Royal Military College of Canada
Building 62
P.O. Box 17000 Stn Forces
Kingston, Ontario CANADA K7K 7B4
613-541-6161
Fax: 613-541-6596
E-Mail: reimer-k@rmc.ca
Dr. Keith Maruya
Assistant Professor
Skidaway Institute of Oceanography
10 Ocean Science Circle
Savannah, GA 31411
912-598-2315
Fax: 912-598-2310
E-Mail: kam@skio.peachnet.edu
Dr. James Risatti
Senior Organic/Microbial Geochemist
Illinois State Geological Survey
615 East Peabody Drive
Champaign. IL 61820
217-333-5103
Fax: 217-244-2785
E-Mail: risatti@isgs.uiuc.edu
Printed on Recycled Paper
-------
AEPA
United States
Environmental Protection Agency
Region 2
Briefing and Site Visit for the
Peer Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Holiday Inn Turf on Wolf Road
Albany, New York
January 11-12, 1999
ERG Onsite Conference Support Staff
Facilitator.
Jan Connery
Eastern Research Group
110 Hartwell Avenue
Lexington, MA 02421
781-674-7322
Fax; 781-674-2906
E-mail: jconnery@erg.com
Technical Writer:
John Wilhelmi
Eastern Research Group
110 Hartwell Avenue
Lexington, MA 02421
781-674-7312
Fax: 781-674-2906
E-mail: jwilhelmi@erg.com
Printed on Recycled Paper
*ER9
-------
United States
Environmental Protection Agency
Region 2
Briefing and Site Visit for the
Peer Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Holiday Inn Turf on Wolf Road
Albany, New York
January 11-12, 1999
EPA and Contractor Participants
Emergency & Remedial Response Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
Damien Hughes
Emergency & Remedial Response Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
Ed Garvey
TAMS Consultants, Inc.
New York, NY
Ann Rychlenski
Communication Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
Melvin Hauptman
Emergency & Remedial Response Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
Doug Tomchuk
Emergency & Remedial Response Division
U.S. Environmental Protection Agency
290 Broadway
New York, NY 10007-1866
Alison Hess
Printed on Recycled Paper
-------
l-nA United States
Environmental Protection Agency
Vial #m Region 2
Briefing and Site Visit for the
Peer Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Holiday Inn Turf on Wolf Road
Albany, New York
January 11-12, 1999
Observers
David Adams
Member-At-Large Saratoga County EMC
216 Stage Road
Charlton, NY 12019
518-399-1690
Fax: 518-399-1690
1 Computer Drive, S
Albany, NY 12205
518-458-6614
Michael Elder
General Electric Company
Fax: 518-458-1014
Donald Aulenbach
24 Valencia Lane
Clifton Park, NY
518-376-7592
Fax: 518-276-2080
Ken Fish
Staff Chemist
Corporate Remediation Divison
General Electric Company
P.O. Box 8
Schenectady, NY 12301-0008
518-387-5990
Fax: 518-387-7611
E-mail: fishkm@crd.ge.com
Jean Bordewich
Town Council Member
P.O. Box 655
Red Hook, NY 12571
914-758-1042
Fax: 914-758-1043
E-mail: parvin@epix.net
Leigh Foster
Arbor Hill Environmetnal Justice Corporation
200 Henry Johnson Boulevard
Albany, NY 12210
518-463-9760
E-mail: staff@w-haywoodbums.org
Kathy Cooke
Program Research Specialist
Office of Fiscal Research and Policy Analysis
NYS Comptroller's Office
5th Floor (AESOB)
Albany, NY 12236
518-486-5433
Fax: 518-473-1900
Joan Gerhardt
Community Affairs
Hudson River Project
General Electric Company
1 Computer Drive, S
Albany, NY 12205
518-458-6618
E-mail: kcooke@osc. state, ny. us
Printed on Recycled Paper
(over)
-------
John Haggard
General Electric Company
1 Computer Drive, S
Albany, NY 12205
518-458-6619
Fax: 518-458-1014
E-mail: john.haggard@corporate.ge.com
George Hodgson
Director, Saratoga County EMC
SCEMC
50 West High Street
Ballstow Spa, NY 12020
518-884-4778
Fax: 518-885-2220
Kevin Hogan
Reporter
Troy Record
501 Broadway
Troy, NY
518-465-9795
Fax: 518-270-1202
William Forts
Environmental Engineer
Environmental Remediation
New York State Department of
Environmental Conservation
50 Wolf Road
Albany, NY 12233
518-457-5637
Mel Schweiger
General Electric Company
1 Computer Drive, S
Albany, NY 12205
518-458-6648
Fax: 518-458-1014
Maren Stein
Associate Professor
Russell Sage College
141 Cushman Road
Melrose, NY 12121
518-663-5230
E-mail; stein@sage.edu
Marty Torrey
District Director
Congressman Sweeney
22nd Congress District
Saratoga Springs, NY
518-587-9800
Lloyd Wilson
Research Scientist
New York State Department of Health
2 University Place - Room 205
Albany, NY
518-458-6316
E-mail: lrwo3@health.state.ny.us
Jim Rhea
Quantitative Environmental Analysis, LLC
290 Efwood Davis Road
Liverpool, NY 13088
315-453-9009
Fax: 315-453-9010
John Santacrose
Counsel
Chair of Environmental Liaison Committee
P.O. Box 3705
Albany, NY 12203
518-458-9945
-------
4% United States
Environmental Protection Agency
m\ Region 2
Peer Review of Hudson River PCBs
Reassessment RI/FS Phase 2 Reports
Data Evaluation and Interpretation Report
Low Resolution Sediment Coring Report
Albany Marriott
Albany, New York
March 16-18,1999
Charge
Members of this peer review will be tasked to determine whether the scientific analyses
conducted for U.S. EPA's Data Evaluation and Interpretation Report (DEIR) and the Low
Resolution Sediment Coring Report (LRC) are credible, the conclusions valid, and whether
the findings are appropriate to use to support the decision-making process for the Hudson
River PCBs Site Reassessment study. The peer reviewers will base their assessment on the
review of the OEIR and LRC, as well as additional information found in the Responsiveness
Summary issued in December 1998 (responding to several documents including the DEIR)
and the Responsiveness Summary for the LRC (which will be released in February 1999).
The peer reviewers will also have available for their review the Hudson River Reassessment
database, which contains all of the data used in the preparation of the above documents,
along with other data.
The DEIR and LRC present the results of the geochemical analyses conducted on the
water-column and sediment data collected for the Reassessment, as well as data collected by
a number of other agencies and General Electric. It should be noted that there have been
several changes in the available data since the time of the release of the DEIR in February
1997. These changes include a better estimate of flow for several reaches of the river, a
recalculation of GE's PCB data due to an analytical bias, and the discovery of a sampling bias
at the Thompson Island Dam monitoring station. These changes are discussed in the
Responsiveness Summaries, and the analyses in the Responsiveness Summaries should
supersede those conducted in the reports, as appropriate.
It is important to realize that the geochemical analysis conducted in the DEIR and LRC will be
complemented by mass balance modeling and human health and ecological risk assessments
to provide a thorough understanding of the fate and transport and impacts of PCBs in the
Upper Hudson River. These other reports will address questions regarding the mechanisms
that release PCBs from the sediment, toxicity, and bioavailability/biouptake. A peer review
was previously conducted for the approach proposed to conduct the modeling for the
Reassessment. After the modeling and the risk assessment reports are released, the Agency
will also have those documents peer reviewed.
Printed ctt BicycM Ptptf
%ERG
-------
Specific Questions
Data Evaluation and Interpretation Report (DEIR)
1. Is the documented PCB load, which originated from the Tl Pool, consistent with
source consisting of historically deposited PCB-contaminated sediments?
2. Are the two-phase and three-phase partitioning coefficients, derived in the
DEIR, appropriate and do they properly address the physical parameters of
the system (e.g., temperature).
3. Are the conceptual models based on the transect sampling consistent with the
data?
4. Does the sampling at the Tl Dam-West location impact EPA's conclusion thf
the sediments of the Tl Pool are the major source of PCBs to the freshwater
Hudson during low flow conditions considering the analytical corrections
made to GE's PCB data? What are the other implications of finding higher
concentrations along the shoreline than in the center channel?
5. Are the geostatistical techniques (polygonal declustering and kriging)
correctly applied?
6. Are the methods applied in the DEIR (change in molecular weight (MW) and
evaluating concentrations of BZ#s 1, 4, 8, 10 and 19 (MDPR)) appropriate
standards for determining extent of dechlorination? Are there any significant
problems with this approach, or more appropriate approaches?
7. The DEIR finds that the degree of anaerobic dechlorination is primarily a
function of original concentration rather than time, and accordingly that there
is not significant predictable dechlorination in sediments containing less thar
approximately 30 mg/kg PCB. Is this reasonable?
Low Resolution Sediment Coring Report (LRC)
1. In the LRC, EPA compared sediment data from cores taken in 1977,1984
and 1994, which had the PCB analysis conducted by different laboratory
methods. How valid are the methods used to establish a consistent basis for
comparison?
2. In the upper Hudson River system, it has been well established that there is
significant lateral heterogeneity in sediment concentrations. While it was
attempted to reoccupy previous locations, some uncertainty is added with
respect to the actual sampling location. While the statistical techniques help
compensate for this, how does the sediment heterogeneity affect the
comparison of cores from two different years? Given the spatial variability, ir
the finding that there is loss from most of the locations supported by the
data?
-------
3. What is the impact of the difference between replicate samples in the 1994
sampling effort (36 percent average variability) on the finding that there was
a 40 percent loss of PCB inventory from the highly contaminated sediments
in the Tl Pool?
4. In the LRC, it was found that Hot Spot 28 contained much more mass than
previous estimates. Is the conclusion that this "gain" is primarily due to
incomplete characterization in 1977 valid?
5. Does the data set and its interpretation support the conclusion that significant
losses have occurred from hot spots below Tl Dam?
6. The LRC found that the historically contaminated sediments in the Tl Pool
were not universally being buried and sequestered from the environment.
How much confidence would you place in the LRC evidence against
widespread burial?
7. Is the interpretation of the sidescansonar data appropriate and supported by
the analysis of the associated sediment properties?
General Questions
1. Is the data set utilized to prepare the DEIR, LRC and Responsiveness
Summaries sufficient to understand the fate and transport of PCBs in the Upper
Hudson?
2. Are there any additional analyses that should be done to verify certain findings
of the DEIR and LRC?
Recommendations
Based on your review of the information provided, please identify and submit an explanation
of your overall recommendation for both the DEIR and LRC.
1. Acceptable as is
2. Acceptable with minor revision (as indicated)
3. Acceptable with major revision (as outlined)
4. Not acceptable (under any circumstance)
------- |