HUDSON RIVER PCBs REASSESSMENT RI/FS
RESPONSIVENESS SUMMARY FOR
VOLUME 2A: DATABASE REPORT
VOLUME 2B: PRELIMINARY MODEL CALIBRATION REPORT
VOLUME 2C: DATA EVALUATION AND INTERPRETATION REPORT
DECEMBER 1998
«eo sr.,.
(&)
PROt^0
For
U.S. Environmental Protection Agency
Region II
and
U.S. Army Corps of Engineers
Kansas City District
Book 2 of 3
TAMS Consultants, Inc.
Limno-Tech, Inc.
TetraTech, Inc.
Menzie-Cura & Associates, Inc.
-------
HUDSON RIVER PCBs REASSESSMENT RI/FS
RESPONSIVENESS SUMMARY FOR
VOLUME 2A: DATABASE REPORT
VOLUME 2B; PRELIMINARY MODEL CALIBRATION REPORT
VOLUME 2C: DATA EVALUATION AND INTERPRETATION REPORT
DECEMBER 1998
TABLE OF CONTENTS
BOOK 2 Of 3
III. COMMENTS ON THE PHASE 2 REPORTS
A. COMMENTS ON THE DATABASE REPORT AND DATABASE
General Electric (DB-1)
B. COMMENTS ON THE PMCR
Federal (PF-1)
Local (PL-1)
Community Interaction Program (PC-1)
Public Interest Groups and Individuals (PP-1)
General Electric (PG-1)
C. COMMENTS ON THE DEIR
Federal (DF-2)
State (DS-2)
Local (DL-1)
Community Interaction Program (DC-1 through DC-4)
Public Interest Groups and Individuals (DP-1 through DP-5)
General Electric (DG-1)
December 17, 1998
i
TAMS/LTI/TetraTech/MCA
-------
Comments
-------
General Electric
(Database - l)B)
-------
db-
Joim G. Haggara G*nm m £J&ctne Company
Cft^ntrmg frv/metManagwr Corpormta Envkmmanxat Program*
Hudmon Bivmr Profmct
OmGcmpumr Onvm Seutn
AJbmny, Nmrn York 122BS
Phonm: *6+4*19; FAX 4SS.1014
VIA OVERNIGHT MAIL
May 29,1996
Douglas J. Tomch.uk
Hudson River Remedial Project Manager
290 Broadway
20th Floor
New York, New York 10007-1366
BE: COMMENTS ON THE U-S. EPA HUDSON RIVER DATABASE
With, your letter of April 18,1996 you transmitted to GE copies of the CD-ROM
containing' the data compiled by the U.S. EPA for the Hudson River Reassessment RI/FS
(RRI/FS) project. You requested comments on the database by May 29,1996.
The database clearly represents a significant effort to compile analytical data from
the Hudson River. The database appears to capture the majority of the daca that GE is
aware of. However, a number of large data sets have not been included. Attachment 1 is a
listing of data sets that should be included in the database since they are relevant to the
Hudson River RRI/FS. The major omission in the database is for information from the
lower river. During the March 17,1993 presentation by GE to the EPA project team and
my subsequent letter to you dated March SO, 1993 we identified the data sets generated by
GE on the Hudson, river, including chose GE daca sets listed in Attachment 1. This data
should have been provided to vou in electronic or hard copy form. However, if you find
you do not have this data in your files we will be giad to make it available to you
La addition to adding the important historical data sets to the database, GE
encourages EPA to update and distribute the database as new information becomes
available. As you are aware, GE is collecting water samples on a weekly basis for PCS
analysis. Additionally, the New \brk DEC obtains PCB levels in fish on a yearly basis in
the entire Hudson river. This fish and water data collected over the next few years will be
crucial to understanding the recovery rate in the river that will occur in the future as a
result of the declining impact of the elevated loading from the Allen Mill in 1991 and
1992 as well as from the fiirther anticipated reductions in PCB loading from ground
water in the vicinity of Baker Falls. GE believes that this data will be critical to the
calibration and validation of the EPA PCB model of the Hudson River.
-------
Even if the EPA chooses not to update the database with the information
described in the Attachment I and the data vet to be collected. GE requests that ail of
these data be considered in the Hudson River RJKI/FS ana be placed into the site
Administrative Record as it is made available to the agencv.
If you have any questions or disagree with the recommendations please Jet me
know.
attachment
cc Walter Demxck, NYDEC
Anders Carlson. NYDOH
Ron Sloan, NYDEC
Paul Simon, U. S. EPA
Yours truly.
-------
ATTACHMENT
DATA SETS MISSING FROM THE U.S.EPA HUDSON RIVER DATABASE
MAY23.1995
Lower River, New
Yoii. Harbor,
Long-Island
Sound
GE/Haxza
1988-1991
sediment,
biota
PCB, pest, lipid.
dBase IV files
TIP TSS Survey
GE/OBG
1991
water
TSS
report, dBase IV
files
Polygon (3)
GE/OBG
1990
sediment
PCB
dBase IV
EPA Lower River
Helicopter
Survey
EPA. 1976
sediment
PCS
EPA report
EPA Lower River
Survey
EPA, 1981
sediment
PCB
EPA report
NYU Lower River
Biota
NW, pre-
1982
biota
PCB
NYU Report
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
Federal
(PMCR - PF)
-------
PF-1
U.S. DEPARTMENT OF COMMERCE
National Oceanic and Atmospheric
Administration
National Ocean Service
Office or Ocean Resources Conservation and Assessment
Hazarcous Materials Response ana Assessment Division
Coastal Resources Cooromation Brancn
Room 3137-C
26 Federal Fiaza
New York. New York 10278
December 3,1996
Doug Tomcnuk
U.S. EPA
Emergency and Remedial Response Division
Special Projects Branch
NYSCB1
290 Broadwav
New York, NY 10007
Dear Doug:
Thank you for the opportunity to review the Phase 2 Further Site Characterization and Analysis
Preliminary Model Calibration Report (Volume 2B) for the Hudson River PCB Reassessment
Remedial Investigation/Feasibility Study (RI/FS). The following comments are submitted by the
National Oceanic and Atmospheric Administration (NOAA).
The Phase 2 Preliminary Model Calibration Report was prepared as part of the overall Phase 2
Reassessment RI/FS activities currently ongoing to provide further characterization and analysis of
the Hudson River PCB Site which extends from Hudson Falls, NY to the Battery in New York
Harbor. The Preliminary Model Calibration Report provides a description of these PCB-modeling
efforts and preliminary results from calibrations of the mathematical models to available field data.
The modeling work proposed is designed to answer three questions: 1) When will PCB levels in
fish recover to acceptable human health and ecological risk levels under the current no action status
of the site? 2) Can remedial activities in the Hudson River shorten this recovery period? 3) Will
contaminated sediments currently buried become "reactivated" following a major flood and result in
increased contamination to fish? The stated goal of the proposed modeling activities is to develop
and field validate mass balance models for evaluating and comparing the impacts of continued no
action, various remedial scenarios and hydromenrical events in terms of PCB concentrations in the
water column, sediment and fish. The approaches proposed to achieve this goal include transport
and fate modeling and fish body burden modeling of the upper and lower Hudson River.
The proposed transport and fate models include : 1) the Upper Hudson River PCB Mass Balance
Model, 2) the Thompson Island Pool (TIP) Hydrodynamic Model, 3) the Thompson Island Pool
Depth of Scour Model and 4) the Lower Hudson River PCB Mass Balance Model. The proposed
fish body burden models include: 1) the Bivariate Statistical Model for Fish Body Burdens (Upper
Hudson River and Lower Hudson near Albany), 2) the Probabilistic Bioaccumulation Food Chain
Model (Upper and Lower Hudson River near Albany) and the 3) the Thomann Food Chain Model
for striped bass and white perch (Lower Hudson River). The overall conceptual approach to
modeling conducted to date in Reassessment RI/FS involved the development and application of a
suite of individual models to describe hydrology, solids dynamics and PCB dynamics in the
environment, and fish body burdens.
-------
NOAA Conuae«s on Hudson River Preliminary Model Caiibraaoo Report 110^6)
12/3/96
The individual models were then coupled within an integrated modeling framework. Tne contents
of the Phase 2 Preliminary Model Calibration Report are limited to descriptions of the individual
models, descriptions of dkabases used for model applications, ana preliminary calibration results
for each modeL As such, all results were deemed preliminary and with the exception of the TIP
Depth of Scour Model were not meant for predictive purposes.
Specific Comments '
1 Introduction
P- 1-1 A more detailed listing of commercial fishing bans and advisories for the Hudson river
should be provided. Also recreational fish advisories should be addressed. As of Nov
26,1996, the following NYSDOH advisories/bans were in effect and are based on
PCBs: ^7)
Hudson Fall to Troy Dam - Can fish but recommend you eat none (all fish)
Troy Dam to the Catskills - Eat none except American shad
Catskilk to Upper Bay - Eat no more than one meal per month of American eel, Atlantic
needlefish, bluefish. Carp, goldfish, large and small mouth bass, rainbow smelt, striped
bass, walleye, white perch and white catfish. For any other freshwater fish not
included on the one meal per month list, the advisory is to consume no more than one
meal per week. There is a commercial ban on American eel and striped bass.
2 Summary and Preliminary Conclusions
p. 2-4 Item 6: What are the three cases where segment-average values for the model output ( 2 J
were significantly different than observed values? Were the predicted values ^
significantly higher or lower than the observed values?
©
Item 8: Why was the percent gain in water column solids mass greater during periods
of low flow than high flow?
p. 2-5 Item 13: The last sentence doesn't make sense. Lower chlorinated congeners are pan
of the total PCB value. Do you mean higher chlorinated congeners instead of total
PCBs? Also this sentence should be relocated to item 12 or made into a separate item.
p. 2-8 Item 3: Will differences in PCB tissue residues by fish sex, age or season collected be
adressed in the Baseline Modeling Report?
Item 5: The contributions of the water and sediment pathways appear to be determined
by the model structure. For example, the model for brown bullhead assumed that , g \
sediment was die only exposure pathway.
2
-------
NOAA Commons on Hudson River Preliminary Model Calibration Report (10/96) 12/3/96
p.2-9 Item 3: "...the estuarine portion of the Lower Hudson River is influenced primarily by
direct external Loadings and loadings from the vicinity of NY city." This appears to
contradict the conclusions of the Draft Data Evaluation and Interpretation Report
(DEHt. in prep), which estimates that the Upper Hudson River PCB load represents
about half of the total PCB loading in New York/New Jersey Harbor.
p.2-9 Item 4; The conclusion that striped bass net PCB uptake occurs primarily between RM
18.5 and 78.5 is an artifact of the model, since the model did not consider the
distribution of striped bass in the river above RM 80.
p. 3-lC Section 3.62. first sentence: Insert (vertically) after "u and v".
p.3-13 Section 3.7.1: Weren't the Phase 2 DEIR High resolution sediment cores selected
because the locations/material were depositionai in nature? These seem like
inappropriate cores to include in a modeling effort to determine risk of resuspension or
erosion of PCB contaminated sediments during high flow events. Their depositionai O7
nature would suggest that they would provide an underestimate of the actual depth of
scour within the TIP. This model should include non-cohesive sediments and
sediments failing between these two categories.
p.3-17 There should be an indication that Thomann is revising the food chain model, what ^
those revisions might or will entail and how such modification may impact model (11 )
output v—
p. 4-1 Section 4.1 Para. 2: A description of the physico-chemical properties for the five
selected PCB congeners should be provided. Explain the relevancy of the five selected /TT\
congeners to the food chain model and ecological risk assessment \^y
Section 4.2: Is there a concomitant increase in sediment total PCB concentrations to those in the
water column between Ft Edward and Thompson Island Dam (TED)? Also is there any
explanation why this is not true year round as total PCB concentrations at Ft Edward \
are equal to or greater than at HD at certain times of the year (i.e fall 1991, winter and
spring 1993).
ii-
Figure 5 The legend to this figure could be made clearer. Add "PCB" after transect flow
average and GE. Insert "daily" before "flow".
p4-4 In the revised validated dataset many of the BZ#138 values in water and sediment woe
qualified as below detection, resulting in Value2 = 0. What effect will these zero values .
have on the model output?
p.4-6 Why wasn't an expanded dataset (1993 + historical) from USGS used for TSS and (16)
flow? —y
p.4-7 Para 1; Isn't this inconsistent with using 1993 USGS data for estimating upstream
TSS loading? Why not use historical + 1993 or historical only when 1993 data was
unavailable? (see p.4-6 comment).
14
1
-------
NOAA Comments on Hudson River Preliminary Model Calibration Report (10/96)
12/3/96
Para 4 and 5: TSS data from Batten Kill was insufficient to define a a me series, ^ 18
therefore measurements were averaged to yield a median TSS of 5.0 mg/1. Values are
averaged to yieid a mean not a median- In Para 4 the ungaged sources do not contribute
to the overall mass balance of TSS but in Para 5 modeling indicates that Batten Kill and
Fish Creek contribute significantly to the solids loads. Am I missing something?
p. 4-8: Para 1: Figure 4-6(b) depicts total mass not percent. ( 19
Para 3: Change "other tributaries" in last sentence to "Mohawk and Hoosic Rivers"
Para 4: It is stated that PCBs are better correlated with TSS than with flow and that
higher chlorinated congeners are better correlated with TSS than the lower chlorinated
ones. Include the correlation coefficients in the text or refer to the table that contains ^ >
them.
p.4-9 Para 1: The high PCB measurement reported by GE for January 1993 was excluded
from all PCB loading estimates because it was believed to be an outlier. An explanation (22s)
should be provided as to this determination rather than citing a pending document ^—s
Were there qa/qc problems with the data or was it identified as an outlier by statistical
methods?
Para 2: How was the 10 ng/1 PCB concentration derived? ( 23
Para 4: Table 4-5 suggests that BZ#4 from upstream sources accounts for 48.8% not N
27% of the load during spring high flow. In addition, 76.2% not 68% of the total PCB (24 )
external load to the Upper Hudson occurs during spring high flow. "—
p.4-10: Para 3: A statement is made about the negligible inputs of atmospheric PCB to the
Upper Hudson. This statement should be supported with documentation. NOAA has ^25
requested that Bruce Brownawell. MSRC. SUNY Stony brook, provide them
information on local and regional PCB atmospheric inputs. This material will be
forwarded upon receipt. It may be useful for pending modeling tasks.
Para 4: Was temperature measured in surface or bottom water? Does water column
stratification occur and if so is it accounted for in the model?
p.4-11 Bruce Brownawell has more localized PCB atmospheric inputs (See p.4-10 above).
They should be compared to the Green Bay estimates utilized in the modeling effort and ( 27J
a decision should be made as to which is the more appropriate value.
p. 4-12 Para 2: The GE 1991 sediment data represent an average both vertically and (28
horizontally.
Para 3: Is bulk density on a wet or dry weight basis? (29
A
-------
NGAA Comments on Hudson River Preliminary Modei Calibration Report (10/96)
12/3/96
p.4-13 The modei appears to be sensitive to PCB sediment concentrations. How reasonable
are the interpolations of GE 1991 PCB capillary column peak measurements to specific
PCB congener concentrations? What affect does a change in concentration (i.e. 2x, 5x,
lOx) of total or individual PCBs have on the model outputs? How valid is it to use one
approach for BZ#4 ana another approach for the other congeners? Is the assumption
thai the 10-25 cm layer is representative of the deeper 25-50 and 50-100 cm layers
defensible based on sedimentation rates or dated cores?
p.4-14 Section 4.5.2: What affect does median (vs. mean vs. maximum) parameter values for
total PCBs have on the output?
p. 6-8 The largest depth of scour predicted by the model was 23 cm or 1 inch and that the
median depth of scour for a 100 year flood event was 0.16 cm. The conclusion drawn
was that flood events will not erode PCB contaminated cohesive sediments to any large
degree. As the input data was based on depositional sites experiencing little shearing,
is this conclusion valid?
p.7-1 Section 7.0. The Lower Hudson River modeling should be conducted on the updated
version of the Thomann modeL
p.7-4 Do the cited references (Waldman 1988a,b) reflect the most current knowledge about
striped bass migration patterns, ix. above RM 80?
p.7-9 Section 7.6.2 Para 2: River Segment 2 is "not sensitive" rathe- than "slightly sensitive"
to loading according to the referenced table. Segment 2 is also not sensitive to
volatilization. In the second to last sentence insert "quite" between "very" and
"sensitive".
p.7-10 Fig 7-7 is discussed as depicting the original Thomann model. Either it is not
appropriately marked or missing from the figure.
p. 8-18 Para 1: The model assumes that most water-column PCBs are associated with
particulate organic carbon. This appears to contradict the Draft DEIR findings that (a)
water-column PCB transport occurs largely in the dissolved phase, (b) the dissolved
phase represents 80% of the water column PCB inventory in the Upper Hudson during
most of the year (10 to 11 months) and (c) the majority of the Upper River PCB input
to the water column is introduced upstream of RM 181.3 under low-flow summer
conditions. This load is transported through the Upper Hudson to Troy with minor
alterations and additions. This is particularly important because the summer low-flow
period coincides with the summer feeding period for fish, which is their period of
maximum exposure. Given the high bioconcentration factors for PCB congeners, any
water-column dissolved PCB exposure could be significant, and all species, regardless
of trophic levei or feeding strategy, would have comparable exposure to any dissolved
PCBs. This would be in addition to any dietary input.
Table 9-2 The table should indicate whether the lipid and PCB concentrations are on a wet or a
dry weight basis. It is assumed they are wet for the modeling exercise.
Table 9-7 Values in the R2 column are two decimal places off.
-------
NQAA Comments on Hudson River Preliminary Model Calibrauon Report s 10/96)
\wm
Fig 9-11 and 9-12 should inciudc "0" for the y-axis. -3 (j*®)
Fig 9-8 thru 9-13 should have 1:1 regression line drawn in.
41
p .9-14 A more detailed description of the various model outputs for PCB fish burdens would
have been usefuL Note that the text incorrectly emphasizes that "the model appears
generally to underestimate burdens [of Aroclor 1254 in largemouth bass J at River Mile (42)
175, [and] overestimate^] those downstream at River Miles 142-155. Fig 9-12
actually demonstrates that the model underestimates RM 142-155 and 189-193 at PCB
concentrations of > 500 ug/g lipid and overestimates PCBs at RM 160, RM 175 and
RM 189-193 at concentrations <500ug/g lipid. In Fig 9-9, there were some over
estimates and some underestimates for PCBs in large mouth bass at RM 189-193 while
at RM 175 the model over predicted PCB tissue residues, i In Fig 9-10, one out of five
RM 189-193 PCB concentrations were overestimated by model. Fig 9-11 correlation is
worse at higher concentrations where predictions were underestimates except at RM
175. Below 150 ug PCB/g lipid, the model prediction was a slight overestimate. Fig
9-16 and Fig 9-17 showed how the model overestimated largemouth bass Arocior 1016
in 1979 and Arocior 1254 in 1979,1991-1992 respectively while Fig 9-19 showed the
overestimate of PCBs in brown bullhead 1982-1992 (exceotl986) and underestimate in
1979 and 1980.
p.9-15 58% should read 59%. (43)
p.9-16 Para 1. Reference should also be made to potential differences by sex.
p. 10-2 Is the data from the chironomid short-term study presented in this document? What /""n
information do they provide on the short-term relationship between water-column
invertebrates and water-column sources?
p. 10-3 Section 10,2, Bullet 2; It should state that tissue is lipid normalized and sediment is
organic carbon normalized. ^
p. 10-5 Para 2: The report states that "...the appropriate statistic for use in the BSAF
calculations is a geometric mean sediment concentration." If benthic organisms have an
equal opportunity to be exposed to a given sediment concentration, the arithmetic mean
is the appropriate statistic, regardless of the distribution of sediment concentrations.
Because the BSAF is assumed to be the same for all concentrations, each sample
concentration should have equal weight in the calculation of the mean accumulation
factor. Using a geometric mean value for sediment will likely result in over-estimation
of the BSAF. This would also be true for the other estimated BAFs (e.g., FFBAF p.
10-20).
p.10-5 Para 1: What affect does the assumption that forage dish feed on benthic organisms
indiscriminately have on model outputs? How sensitive is the model to dietary input
from within a given compartment? Benthic invertebrates may be sediment consumers,
detritivores, omnivores, filter feeders or predators that could result in different patterns
and concentrations of PCBs in invertebrate tissues.
200
6
-------
N'OAA Comments on Hudson River Preliminary Model Calibrauan Report (10/96)
12/3/96
p. 10-8... Section 10.2.3.: It was stated that "[t]he modeled and observed percentiles compare
favorably." It is not clear what criteria are used to make this statement Figures 10-16
and 10-36 arc examples of comparisons that do not appear particularly favorable.
p. 10-7 Para 1: It appears that major peaks were more representative of the model maxima and
the low peaks were bracketed by the model 50 and 90 percentile. It would be useful for
the authors to include the 75 percentile in the figure as it may have characterized the
smaller peaks better than the 50 or 90 percentiles.
Para 2.: The means were from 0.2 to 0 J not 0.5. to U.
Para 5: Chironomids have different feeding ecologies. Some are associated with the
- sediment but are meriatnrs ? Ahlahes-mvia. Crvmochironomom. Pmrtafln^ Some
filler feed (e.g. FolYPCdilum, Rticmanyiarstts), some spin nets to trap panicles
(Glvmorendinest some eat Aufwuchs (e.g. NflnrelatilUS) or detritus (e.g.
Qrthociflriiml Therefore temporal changes in PCB water-column concentrations
should be more closely correlated with purely filter-feeding or herbivorous chironomids
than with sediment, sediment/epiphytic or predatory species. Have different congener
patterns been observed for chironomids pursuing different feeding strategies?
p.10-8 Biota to sediment goodness of fit plot described for BZ#4 is in a different format than
Fig 10-16 and 10-20 for BZ#28 and BZ#52 respectively.
The greatest differences in BSAF were observed at RM 100, RM 189 and RM 189.5.
BZ#52 RM 100 similar to BZ#28. BZ#52 RM 189 and RM 189.5 pattern similar to
BZ#4 but not BZ#28. Is this partially a result of benthic organisms collected at these
stations and their feeding behavior? For example, the highest BSAFs were for
chironomids, isopods and gastropods. Did these organisms dominate sites with the
highest BSAFs on a river mile basis?
p 10-19 NOAA is concerned about combining the data from all fish samples with an average
length of less than 10 cm to calculate a forage fish BAF. The majority of the fish
samples in that size category were young-of-year spottail shiners. Other species, such
as tessellated darters, smallmouth bass, etc. were represented by only a few samples
and most likely have very different feeding behaviors. Using spottail shiners and other
cyprinid species would provide data from all stations except river miles 47 and 26. In
addition, it is likely that the feeding preferences for adult spottail do not apply for
young-of-year fish.
p. 10-21... Tessellated darters and spottail shiners observed concentrations of FCBs (calibrated
congeners, Aroclor 1016,1254) were the highest. Is their food source more heavily
contaminated? Do the congener patterns for forage fish spottail shiner and
pumpldnseed sunfish reflect the concentrations found in their diet? What about for
largemouth bass and white and yellow perch? Do they reflect the concentrations found
in forage fish and invertebrates?
p. 10-22 Para 1: "...with QUE spottail shiner at river mile 194.1..." Most of the fish samples
collected in 1993 (probably all of the forage fish samples) represented composites of
multiple individual fish.
7
-------
NOAA Comments oo Hudson River Preliminary Model Calibrauon Report (10/96) 12/3/96
p.10-23 Para 3 second to last sentence: Insert Fig 10-63 through Fig 10-65 after "A second set
of figures'*
p.A-11 Para 1: In the discussion about water depths, should the units be feet instead of
meters?
p. A-24. Under spotxail shiner, the estimate dietary consumption is 50% water vs 50% sediment
source while Table A-15 lists 75% from water and 25% from sediment \ 60y
P-A-25 Para 1: "The data show that forage fish diet is primarily from water column organisms
when averaged over the entire Hudson River." The information on feeding habits of
forage fish (Table A-15) only showed data between Lock 7 and the Troy Dam.
p.A-25 Para 3: The following statement raises a major concern about the development of this
model: "...it is possible that the derived BAPs are artifacts of the model. „.(M]odel
application can only confidently be accomplished through a greater understanding of
the water column invertebrate box..." What can be done to reduce the uncertainty in the
water column invertebrate compartment, "which impacts all subsequent compartments"
and directly affects the application of the model?
p.A-25 Para 5: Forage fish diet is 33 % benthic to 67% water column invertebrates but in
Table A-16 the breakdown is 31% vs 69%.
Please contact me at (212)637-3259 or Jay Field at (206)526-6404 should you have any questions
regarding these comments.
Sincerely,
•, 0^5
Lisa Rosman
NOAA Associate Coastal Resource Coordinator
cc Shaxi Stevens, DESA/HWSB
Gina Feneira, ERRD/SPB
Robert Hargrove, DEPP/SPMM
Charles Merkel. USFWS
Ron Sloan, NYSDEC
n no no
8
-------
Local
(PMCR • PL)
-------
PL-1
ENVIRONMENTAL MAM
*ETEn 5AuE~"
* - 4(h?MAN
AGEMEMT COUNCIL
;ECRG=HCDGSCN
*6C*C~
'.Jov ember 25, 1396
Mr. ^ougias Tcr.cr.uk
7.S. Environmental Protection Agency
230 Broadway, ZZth Floor
New York, 'Jew rcrk 1C007
Tear Doug:
Enclosed you will find Saratoga County EMC'j comments or. EPA's Phase 2, Volume
"3, "Preliminary Model 'lalibration Report" for "he Hudson River PCS
Reassessment Kl/FS catea October 1996. These comments were prepared by Dave
Adams and endorsed by the EMC at its November 12, 1996 meeting.
As I mentioned ir. our telephone conversation, the aforementioned report was r.c
received in the Saratoga County EMC office until Novemcer 4, 1996. Due to the
highly technical nature or -.he report and its late arrival in EPA's
information repositories, the Saratoga County EMC recommends that EPA's public
comment period oe extended beyond the Noveirtoer 22, 1396 comment deadline t~
allow the public adequate time for review and comment.
The Saratoga T.unty EMC would aiso appreciate a response to the enclcsec
specific comments/questions so cne Council might ratter understand the USEPA'i
rationale on tr.ese matters contained within tne Preliminary Model Caiibraticr
Report. The Saratoga County EMC 13 disappointed by tr.e Lacx cf response :r™
EPA to our previous comment submittals.
Thank vcu
fincerelv,
\~«t c
George Hodgson Jr
Director
c: All EMC "emoers
Dave Adams
Garryi CecKer
Judy Dean
10.0204
wp<;t -iGH s~~5
3ALLS70N SPA N Y ';320
15185 885 5381 SXT -
-------
November 22, 1996
Comments on EPA Phase 2 Report
Volume 23 - 'Preliminary Model
Calibration Report
Hudson River PCB's Reassment P.I/FS
October 1996
Provided by David D. Adams
Member Saratoga County £MC ana
Government Liaison Committee
Overall Comments
1. The organization cf the report is awkward which makes review difficult
ana less efficient than it needs to be. Separating the tables ana figures in |
a separate book from the text causes excess effort and Lost time going back
and forth between the two volumes. It would be mucn better to follow the more
convential practice of integrating the tables and figures in appropriate
places in the text. Also, the inclusion of specific information about each of
the models in Section 2, the description cf the overall apprcacn, again makes
review difficult as this model information in Section 2 is pertinent to the
review of the subsequent sections in the report. This causes lost time and
ixtra effort in looking cac.< to locate tne applicable information in Section
2. Section 3 shouid have ended with Section 3.4 ana the subsequent model
iiscussions integrated with the model discussions in the following section,
.-lease consider these suggestions in the preparation of future reports.
2. £?A has made public participation in the Reassessment process a major
part cf the reassessment effort. However, most C if not ail!) of the interested
public does not have the technical training to understand this report as
written. It is recognized that reports of the type Volume 23 represents are
fundamental and necessary to the reassessment process. It is not intended
that these reports should be replaced by Less technical reports. Rather, the
recommendation is that a "concept summary" version be added which would be
written in a mannner that would still convey the basic information that's in
:he '"-.echnical" version (e.g. Volume 23 as it now stands) but in a way that
"he .ess technically trained citizen couid better understand.
Presentation cf the H'JDTCX model is ir.adecuate ir. that .".one of the
zcrmuias or mathematical relationships used ir. the model are included.
Without that, it is not possible to get an adequate understanding cf the model
-------
particularly the reiaticnsnips Between constants ana variables, Reference tc
:ome ctr.er report cuch Hi "Ambrore, et.ai." is r.ct adequate as accessibility
to tr.ese reports is difficult if not impossible for the general public. This
comment also applies tc some areas cf the Fish 5odv Burdens modeling
discussion. A good example in the report cf the correct way tc present a
model is the presentaticn of the TIP Hydraulic Model. The presentation of the
Lower Hudson Transport and Fate Modei section also represents a better moaei
presentation than that for HUDTOX.
4. The discontinuity ir. the early 1990' s of the PCB concentration data vs
time snows a definite need for the HUDTOX model to incorporate the effects of
the PCB releases at Bakers Falls as no other reason appears reasonable to
account "for" the "sudden PCB increase in the early 1990's. It is not clear from
the discussion in this report that the HUDTOX model is able to incorporate the
Bakers Falls situation ana/or that the EPA has the necessary data to include
the effects of Bakers Fails. The need to consider the effects of Bakers Falls
is especially urgent ir. view of the Fish Body Euraen models being of a quasi
oteaay state nature with a time span interval average of about 1 year and in
view of the fail-off ir. PCB levels m the years leading up to 1992-93 and then
the increase when the Bakers Falls sources came into play. Because of this
oehavior of the PCB concentrations, it is recommended that the study of PCB
1 eveis using the HUDTOX model be separated into two time periods. The first
time period would be that up to 1992-1993 and would evaluate the HUDTOX and
Fish Body Burden models ability to predict how PCB levels decrease when new
PCB sources are zero cr negligible. Then the HUDTOX model could be applied to
the years after i992-1993 ana in the future to predict how PC3 levels would
decrease after the Bakers Falls source is eliminated by the remedial actions
now underway. The model would have to incorporate the initial inputs of PCB's
from 3akers Falls ana then the sucsequent drop of PCB input as the remedial
actions take (or have taken effect). Assumptions would have to be made as to
the timetable for elimination of the Bakers Falls source - perhaps it could be
a parameter in the study. This approach should give a better prediction of
the future tnan trying to maxe the HUDTOX model ;ump througn the evident
discontinuity that occurred in 1992-93.
5. The stated tendency of fish to accumulate higher chlorinated PCB's (see
page 3-3 of Book 1) empnasizes the need to do the modeling on a congener basis
and to include terms in the PCB transport and fate models for Diodegradation
of PCB's. Consideration of this biodegradaticn effect is necessary to
accurately estimate the total uptake of PCB's in the fish, which so far, is
the basis for EPA's Health Risk Analysis. The importance cf considering the
effect is shown by the stated sensitivity shown to biodegraaation ir. the Lower
Hudson River Model (page 7-9 of Book 1). As an aside, it is hoped that some
day scon EPA will recognize that the health risk of ail PCB's is not the same
and modify the Health Risk Assessment procedure accordingly.
6. The discussions and formulas presented ir. Book 1 are based on the
assumption that there is a direct one-to-one relationship between PCB
concentrations in water and sediment tc PCB levels in fish. The tendency of
fish to accumulate higher chlorinated PCB's ; see comment No.5) raises
questions about this assumption. This tendency suggests that it may be
necessary to include in the Fish Body Burden model terms to reference the time
rate of PCB congener removal from fish (whether by excretion or metabolism)
and tc consider the age of the fish sampled. What has been done to evaluate
-------
:r.e need fcr including such effects ir. tr.e moaei? If data are not available,
-.t. laast a parameter .study could be ~=ae .3ing cest estimates from the
scientific literature to see if these effects are significant.
7. 5ook 1 should include a discussion as to why it was fait necessary to
;eveiop a new model fcr the Upper Hudson instead or applying the existing
lower Hudson model (or conversely, if a new model is needed for the Upper
Hudson, wny isn't it also used for the Lower Hudson?).
Specific Comments
1. Section 3
a. Fig. 3-1: Explanation of the symbols used is needed for the Solids
Balance & PCB Balance. While the arrows are self evident, the meaning of ^
the symbol within the circle between coxes and the "omega looking" symbol
at the sediment-water and sediment-sediment interfaces are not clear.
b. Solids submodel, page 3-6: The discussion should include the rationale
for the organic camon fraction assigned to TSS to represent particulate ^
organic carbon being a constant. Also, why do not tables 4.9 & 4.12
incluae this parameter? Some explanation as to why values for biotic
solids loading due to primary production from the Lower Hudson are
satisfactory to use in the Upper Hudson (even after temperature
correction) should be provided.
c. Toxic chemical submodel, page 3.S: It is not satisfactory to say that
the Phase 2 database doesn't distinguish DOC bound PCB's from truly ( c
dissolved PCB's out it is important to do so and then drop the subject.
Some explanation as to why this omission in the data is not significant
zo this reassessment is necessary.
\
d. Toxic cnemicai submodel, page I-"7: The ''ether enhancements" made to
simplify application of the model snoulc be spelled out r.ere as access to ^
the reference cited is not readily available to the general public.
e. Scour Model, page 3-13: Will the inability to model "subsequent trans- ;
port ana redistribution" of eroded sediments be of significant impact to \ ^
the modeling results? If not, why not and if yes, how will the Reassess-
ment accommodate this failing?
Section 4
a. Page 4-11: The applicability of Green Bay data for PCB concentration
in the air is questioned. If voiatization is important, this assumption (l \
should be revisited and either data collected for the Hudson or the PC3 " i
air concentration made a oaramenter in the studv.
b. Page 4-13: Why are no tables provided fcr PC3 concentration in
sediment layers S-1G and 1C-25 cm?
c. Page 4-18: Values of V aiven here range from .25 to 3.05 so some
5
0
( !)
-------
explanation is needed as to wnv a value of 2.0 is the correct one for the
jdsor..
i. Page 4-19: A more complete explanation or discussion of the basis for
"he "professional judgement " value of .22cm/vs for is needed.
©
•=. Table 4-10: 'J., is listed as a parameter for HUDTOX but is not used. /^\
The decision not to use shouid be explained. J
?. Tables 4-13 througn 4-17: The iarge variances shown in these tables
are disturbing and some discussion of their significance is needed. For
example, the t-test uses the assumption that the variances, T. and T ,
are equal. It is usual to test this assumption using the "F" t^est. Heis —S
this been done and if so, what conclusions were reached? Should the
analysis of variance method be used in addition to or in place of the'
t-test?
Section 7
Page 7-3: Explain why Hydroscience vaiues for horizontal dispersion
coefficients are good vaiues to use. Also explain the rationale for (18'
using t 30% to adjust these coefficients for high and low flow years.
Section 10
1. Page 10-23 and following: Why are not equations defining FFBAF,
calculations of FFBAF, ana FFBAF tables and/or figures provided as
other BAF's?
for (^}
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
PL-2
FN VI PON MENTAL MANAGEMENT COUNCIL
"ES 3A-.E"
•*«MAN
GEORGE
'fovHmber 25, *396
Mr. Douglas Tcmenuk
U.S. Environmental Protection Agencv
230 Broadway, I3th Floor
Mew York, New Yorx 13007
.'ear Doug:
Enclosed you will find Saratoga County EMC'.: comments on EPA's Phase 2, Voiuzie
IB, "Preliminary Model Calibration Report" for the .-iuason River ?CE
Reassessment Rl/FS aated October 1996. These comments were prepared by 'Dave
Adams and endorsed by the EMC at its November 13, 1996 meeting.
As I mentioned in our telephone conversation, the aforementioned report was r.c
received in the Saratoga County EMC office until November 4, 1996. Due to the
highly technical nature or -.he report and its Late arrival in EPA';
information repositories, the Saratoga County EMC recommends that EPA's puDli:
comment period be extended beyona the Novemcer 22, 1996 comment deadline t
allow the public adequate time for review and comment.
The Saratoga County EMC would also appreciate a response tc the enclose
specific comments/questions so the Council .Tight better understand the USEPA'
rationale on tr.ese matters contained within the Preliminary Model Calibratic
Report. The Saratoga County EMC is disappointed by the iaoc ci response :rc
EPA to our previous comment submittals.
Thank you.
Sincereiv,
-
George Hodgson Jr.
Director
inc.
cc: All EMC Members
Dave Adams
Carryi CecKer
Judy Dean
SO WEST NiGH STREET
BALISTGN SPA NY '2020
(518) 885 530T EXT
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
Community Interaction
Program (PMCR - PC)
-------
PC-1
tx : z.r.r. -nu .-umosoneric 2..:r.ces {{TTr - ' ;' S/-U2—"r -'
: s«.;encc% 5; j'-U2-5SI5 .*r —
J> -r <* •*• • '*.}
•»-* t: islcne'aunosaicanv .e
< t\ ,IZZ2 " ¦*€$¦'J.£. r |eojO'|>«scnsunix.aicsRN s
' "-y^' , :uto:«u.i.vw jtmos.aioans ,
* 5 i
University at Albany
STATE UNIVERSITY OF NEW YORK
November 15. 1996
Mr. Douglas Tomcnuk
l.'S EPA - Region 2
190 Broadway - 20th Floor
\e\vYork.NY 10007-1S66
RJE: PMCR Comments
Dear Mr. Tomchuk:
The following comments are submitted in reference to the Upper Hudson Mass Balance
Model of the PMCR (Sections 3.5 and 4. and I emphasize are based only on a brief review for
consistency with my prior report to you s March 1994).
1) External soiids loadings and resuspenston [Sections 4.2. 4.4.2. 4.7. 4.8") TSS cannot be
related simply to flow or discharge data. "Hie correlation in Figure 4 13 is. poor
because of the sediment source-discharge-ilow event timing relations described in my
1994 report for the upper Hudson. These relations are weil displayed in the t'SGS
data for the 1993 calibration period, v.nic'n provided representative high discharge fx
events. For the most part. TSS is in a depositional mode during events, i.e. the
sediment concentration peak precedes the discharge peak, and sediment concentrations
further are a function of event timing and Sequence rather than absolute discharge.
Resuspension of prior year deposited sediment is. of course. Ukelv during a spring
event but not on a continuous or timing'event independent basis as embodied in 4.6.2
items 4 and 5. Figure 4-4 shows that TSS has declined sharply during high discharge. .
but much more close interval sampling (< 1 day) is needed to adequately assess this
feature. A careful analysis of the TSS-discnarge decay curves and timing
(instantaneous values) across many events might provide the proper relationship^h
The Solids Model calibration (4.7.1) appears to relate resuspension to a iow-flow (%
baseline condition, and an empirical factor based on mean river velocity. Apparently
this is the basis )'11 for the TSS mass balance of Figure -i-3~. especially part !b). This
is in turn critical to the PCB balance of Figure 4-40 ana others derived from it. i It
certainly does not follow from an examination of Figures 4-2. 4-4. 4-7. 4-10). The
application of this calibration to the TIP. especially under high flow conditions, is
-------
questionable. The seaiment provenance area ana sediment or' the TIP are uniike that at
Stillwater ana Waterioru. i: is aiso iikeiy mat tr.e TSS composition under iow flow
conditions is not the same as at high flow. Further. TSS under iow flow is not
necessarily resuspended (Figure 4-57. non-event.'). Taken together wuh the assumed
TSS vs. discharge relation. 1 Jo not think this moael has made a vaiia mass balance
distinction between TSS loading via tributary and runoff inputs vs. true resuspension
of river bed sediment deposited during the last spring eventisi.
In the 1994 report'I noted the very erratic nature of PCB loading during high flow
events and the problem of inhomogeneous PCB flow distributions for sampling,
especially at Ft. Edward or the Rogers Island locality. Figure 4-14. segment 3. aptly
depicts this first aspect, and it is more fully amplified in the O'Brien and Gere reports
(1993a.b.c.d; pg. R-SY PCB transport wiil be reiated to high discharge events merely
by mass movement, but PCB water column concentrations at Ft. Edward remain
erratic at any flow rate i Figures 4-14. a.5. a.2. Note paragraphs 5 and 4. p. 4-2:
paragraph 4. p. 4-8: paragrapn i ana 2. p. 4-9. The outlier point is typical of episodic,
erratic PCB release from the area between Bakers Fails and Rogers Island). For
example, note the following PCB concentration data (L'SGS except as noted) for the
Rogers Island station:
USGS 1993 Water Year
Pat? Total PCB as i24? fnpn)
Aprii 13 <.01. 0.8 iame date
April 29 0.10. 1.1. 11.0 same date
1992 Water Year
Apni 22 0.125 iGE)
Apni 23 3.0
There is perhaps more chance of PCB "".spikes ' during nigh discharge, but
overall this is unpredictable.
The model PCB calibrations of Figure 4-!4 suggest further that adding peaks for
high flow events will overestimate PCB concentrations, and hence total (integrated1)
loadings for the spring event. Flow averaged sampling appears to reduce the
variability among samples, but does not change the iack of correlation between PCB.
TSS. ana flow noted by G.E. (P. 4-8. 4-9; Fig. 4-14. segment 3Y The apparent
increase in PCB concentrations between Ft. Edward ana the T.I. dam has been noted
previously (NYSDEC data; Bames. L'SGS 1987) but may aiso be partly an artifact of
sampling procedures. Random sampling of a time variable distribution with sharp
high value peaks and longer !ow-vaiue intervals wiil be biased towards
underestimation of the integrated total (mass) when the sample exposure is very short,
and "outlier" points are discarded. T"o this can cc added river transect variation or flow
inhomogeneity. and additional bias can arise due to a fixed sample point position.
-------
Downstream now nomogenization effects would tend to lessen these biases te.g. T.I.
aam) which are most acute at. ana upstream of. Rogers islana.
Flow inhomogeneitv at Ft. Edward has been documented by O'Brien and Gere
i I993a.b.c: p. R-8). ana previously by Tofflemire 11984. Northeastern Environmental (.J^
Science, v. 3. p. 202-208. Why isn't the latter referenced here - it is certainly
relevant?) As a result of the above factors I do not consider the Ft. Edward PCB data
adequate for reliable upstream tor input) mass-loading calibration. The T.I. dam data
are better as a basis for Table 4-5 ("p. 4-9), for example, but these data do not warrant
the implication of Figure 4-40(B) in which the resuspended TIP PCB mass (less
congeners BZ #4) approximates the Ft. Edward input. In this case the caveat of #2 (p.
4-26") is the operative factor, but not the oniv problem.
j'i TIP Depth of Scour Model (Section 6.2.2) Resuspension Experiments. How do we
know that sediment samples sent to I'CSB wiil recompact or otherwise approximate
their physical state in the River when tested? How do we know that the act of core
collection does not influence the subsequent shaker shear stress observations?
tecommendations
1) The time sequence curves for TSS and flow at Ft. Edward and the TI dam in high
discharge events need to be determined as a basis for meaningful mass balance
estimates for the TIP. Probably more observations within the TIP in order to estimate
external sediment and flow inputs (e.g. Moses Kill and spring runoff) will be needed.
Regardless of the PCB loading, this information is needed as a direct check on the TIP
erosion/resuspension estimates provided by the TI hvdrodynamic model. It may also
be necessary to sample the lower water column at elevated discharge to assess bealoaa
transport if this component is not normally pan of TSS determinations.
Another approach is to examine the TIP sediment cores in areas of predicted greatest
PCB "hot spot"' erosion for a similar effect, i.e. an erosionai hiatus, during the last
actual -100 year event (1975-76). If erosion occurred a visual demarcation may be
present: otherwise truncation of the Cs 137 time/stratigraphy scale should be noted.
Deposition, on the other hand, was evident by a Csl37 reversal (arrival of former Ft.
Edward dam sediment) as documented by Tofflemire and Quinn 11979. NYSDEC .
Tech. Paper No. 56; studies of numerous cores, sediment samples, and other data.
Again, why isn't any of this body of work referenced?)
2) The uncertainties of PCB loading and water sample representation at Ft. Edward need
to be resolved in order to obtain meaningful input data for examining PCB mass
balance :n the TIP. Integration of the PCB concentration vs. time relation is needed,
but attempts to correlate this with TSS or discharge should be abandoned at Ft.
Edward. Flow averaged sampling can improve integration, but the question of cross
channel inhomogeneitv remains. This needs to be evaluated before proceeding with
any comparison of PCB loading at Ft. Edward vs. the TI dam. If resuspension is a
-------
factor tn TIP PCB loading, then a reiauonsnio with TSS can oe iookea for at the TI
Jam. However the same time sequence detail is neeaed as under recommendation =1.
At the present time n appears the TIP data is too unconstrained for the fullest use of
other potential insights such as changes in congener makeup - -easonaiiy. annually
cyclical, or high riow-iow rlow < .""i; or "tracking" PCB concentration spikes at Ft.
Edward downstream,
[ hope these comments are of some help. Please cail if further discussion is desired. Also
piease indicate the distribution of these and other PMCR comments.
Very truly yours.
George W. Putman
-------
Public Interest Groups & ,
Individuals (PMCR - PP) K
-------
pp-1
JOHN E. SANDERS
33 Sherman Avenue
Dobbs Ferry, NY 10522
24 October 1996
Mr. Douglas Tomcnuk
US EPA - Region 2
290 Broadway, 20th Floor
New York, NY 10007-1866
Attn: PMCR Comments
Dear Doug,
Herewith some comments about the PMCR draft copy that I received
on 16 October 1996. I include my usual exercises that seem to
fall into the category of tilting at the wrong-hyphen-usage
windmills plus some other editorial-type notes.
My main comment about the modelling work deals with the use of /
Julian Days along the time axis on many of the graphs, starting
with Figure 4-3. The range of days for the model-calibration
period (01/01/93 through 09/30/93) is shown as zero through 300
and the label is "Julian Date." The only correct label for these
numbers is "Julian Date (-2448989 ) " .
If the plan for later is to pick up other intervals on the same
time-string base, then I think it would be better to use abbrevi-
ated forms of the correct Julian numbers, not any old arbitrary
0-300 with a subtraction of 2448989. Using the conversion factor
for Julian Day = the DOS serial number for a date + 2415019, the
Julian Day for 01 January 1993 becomes 33970 + 2415019 = 2448989.
Correct-date Julian graphs for the model-calibration period thus
start with 2448989 and end with 2449289 (the basis for the re-
quired subtraction mentioned above). This could be shortened to
989 through 1289 [with the horizontal axis label reading "Julian
Date (+2448000)"}. Also, users should not forget that Julian
Days begin at High Noon, not at midnight.
Other figures needing the above change are: 4-4, 4-5, 4-7, 4-8,
4-10 (each of the 8 little graphs), 4-11, 4-12, 4-14 through 4-31
(each of the 8 little graphs on each figure), 4-51 through 4-62
(ditto),
I enclose an example of a graph I worked up for Hudson River
discharge expressed as yearly mean flows on a Julian-Date time
axis. This graph was drawn by Quattro Pro from a spreadsheet
that I compiled from U. S. Geol. Survey data + earlier years from
the NYC Board of Water Supply. Since I drew this graph, I have
updated the file to 1994. If you can use it, I can provide the
file on a disk.
1
-------
Mr. Douglas Tomcnuk
Page 2
21 October 1996
I recently got hold of the proceedings volume of the 1963 Federal
Inter-Agency Sedimentation Conference that contains Panuzio's
oft-cited (but possibly little read) paper on lower Hudson River
siltation. On p. 518-519, he gives the results of daily sampling
of the river for suspended sediment at the Poughkeepsie water
intake for the period 01 Sep 59 through 31 Aug 60. I never heard
about this data set before. I do not recall ever seeing the data
volume {TAMS/Gradient Oct. 1995 of the PMCR) , so I do not know
whether or not this is in your system. (The Panuzio 1963 refer-
ence is not in the PMCR list.) This data set might be a useful
thing to have for model hindcasting.
I'm just home from a week in the hospital where I underwent major
abdominal surgery. I hope to be recovered sufficiently to attend
the meeting on 28 October 1996 in Albany.
Cordially yours
John E. Sanders
JES/s
End : PMCR comments
>/
r
-------
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
COMMENTS ON PMCR BY JOHN E. SANDERS (20 NOV 96)
Item Location Comment
01 Title, L. 2, last word An example of space and hyphen
needed before a word to indicate
word(s) left out. In this case,
the left-out words are "Further" &
"Site" so the correct usage is
"Further Site Characterization
and -Analysis"
02 Title, L3, middle
03 Contents, 1.3
04 Contents, 2.1.2
05 Contents, 2.1.3
Hyphen needed between two words
that modify- and precede a noun.
In this case, "Model" and
"Calibration" modify and precede
"Report" so the correct usage is
"Model-Calibration Report"
No. 01 hyphen rule applies before
"Organization." Should read:
"Report Format and -Organization"
Hyphen and space needed after a
word to indicate left-out words,
cf. No. 01; also another example
of No. 02 hyphen needed. Thus:
"Water Column and Sediment Models"
should be "Water-Column- and
Sediment Models" (to indicate
the meaning for models of the
water column and of the sediment).
Hyphen error of No. 02 type:
"Fish Body Burden Models" should
be "Fish-Bodv-3urden Models."
06 Contents, 2.2.2
07 Contents, 2.2.4
Hyphen-needed error cf. No. 01
(If "Sediment Erosion" refers to
the Thompson Island Pool, i.e.
words left out, then the correct
usage is "Thompson Island Pool
Hydrodynamics and -Sediment
Erosion." Because of geographic
name, "Thompson Island Pool" needs
no hyphens.
A double whammy; No. 02 hyphen
needed between "Striped" and "Bass"
before "Accumulation"; space and
hyphen before "Striped" to indi-
cate left-out words "Lower Hudson
River." Thus to be: "Lower Hudson
River PCB Mass Balance and -Striped-
Bass Bioaccumulation"
1
-------
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
08
Conrenrs, 2.2
09
Contends, 2.5
10
11
12
13
14
Contents, 3.5.3
Contents, 3.6.3
Contents, 3.7
:ontents, 2.7.3
Contents, 3.8
15
16
Contents, 3.8.3
Contents, 4.2
17
Contents, 4.3
If meaning is "Modeling Goals and
Modeling Objectives" then the
correct form is "Modeling Goals
and -Objectives" (cf. No. 01).
Hyphen needed between "Mass" and
"Balance;" type No. 02; should be
"Upper Hudson River Mass-Balance
Model"
"Spatial-Temporal Scales" First
correct-hyphen usage! Hooray!
Gold -star for correct usage
cf. No. 10.
No. 02 hyphen blunder; should read:
"T I Pool Depth-of-Scour Model"
What happened? Must be:
"Spatial-Temporal Scales"
as in Nos. 10 & 11.
Here's a real hyphen challenge; the
correct usage not being immediate-
ly clear. The question is, how
many models? Is it two? A trans-
port model and a fate model? Or
only one? (i. e., model of trans-
port and fate). If it's two
models, then the correct usage is:
"Transport- and Fate Model." If
it's only one model of transport
and fate, then the correct usage
is: "Transport-and-Fate Model."
You see, hyphens are important
and I am not just making this
stuff up. There are rules.
Back on track again with
"Spatial-Temporal Scales"
Hyphen needed between "Water" and
"Quality" (No. 02 type): should
read "Historical Trends in
Water-Quality Observations"
Hyphen needed between "Preliminary"
and "Calibration" before "Dataset*'
(No. 0 2 type): should read
"... Preliminary-Calibration
Dataset"
2
-------
Contents,
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
Hyphen- needed betweeen "Model" and
"Input" before "Data" (No. 02
again): should read
"Model-Input Data." (Note,
this one is debatable; if
"Input Data" are considered
as a single word, then no
hyphen is needed. I argue
for the hyphen to contrast with
possible "node1-output data."
19
Contents, 4.4.1
"System-Specific Physical Data"
is correct. Score another hyphen
for the home team!
20
Contents, 4.8
21 Contents,
Another No. 02, but with some
possible debate. If "Component
Analysis" is a single entity,
then the correct form is"
"Mass-Balance Component Analysis"
(But if it is analysis of the
components of the mass balance,
then the correct version is:
"Mass-Balance-Component Analysis"}
Same as 20, with result depending
on status of "Sensitivity Analy-
sis ." Minimum-correct-hyphen
usage is: "PCB-Model-Calibration
Sensitivity Analysis"
22
Contents, 5.2
No. 02 again; to be "Model-Input
Data"
23
Contents, 5.2.1
2 4 Contents, 5.6.1
See No. 19. [But I wonder if
somebody believes that whenever
"system" ana "specific" are used
they should be hyphenated (as
contrasted with hyphen required
because they modify- and precede
"Physical Data").]
No. 02 again; should be "Rating-
Curve Velocity Measurements."
I also question use of "velocity"
here. "Velocity" is a vector
consisting of "celerity" along a
particular direction. My guess is
that "Celerity" is what is being
discussed.
25
Contents, 5.7
Multiple No. 02: should be:
"100-Year-Flood-Model Results"
(but with poss. option on last
hyphen if "Model Results" are
3
-------
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
2 6 Contents, 5.8.2
27 Contents, 5.9
2 8 Contents, 6
29 Contents, 6.2.1
3 0 Contents, 6.3
31 Contents, 6.4
3 2 Contents, 6.5
3 3 Contents, 7.
3 4 Contents, 7.2
3 5 Contents, 7.2.1
3 6 Contents, a
37 Contents, 8.1
38 Contents, 8.4
an integral term). If it is the
results from the model of the
100-year flood, then the way I
show it above is the only correct
form.
No. 02 to be "Turbulent-Exchange
Coefficient"
See No, 24 re; "Velocity."
See No. 12.
No. 02 requires "Bottom-Sediment
Distribution" here.
If the "Uncertainty" refers to
"Model," then No. 01 comes into
play and the correct form must be
"Model Parameterization and
-Uncertainty" (Otherwise
the only correct version is
"Model Parameterization and
Model Uncertainty.")
A double-header bonanza for No. 02:
Make it: "Depth-of-Scour
Predictions at Selected Locations
in Cohesive-Sediment Areas."
No. 02 requires "Cohesive-Sediment
Areas"
See Comment No. 14.
See Comment No. 18.
Gold star again for correct
hyphen usage of "System-Specific
Physical Data"
OK as written. "Fish Body
Burdens" gets no hyphens just
because these words are used.
But, this changes later.
See Comment No. OS.
No. 02 hyphen rule; hyphen is
needed between "Bivariate" and
"Statistical" modifying- and
preceding "Model." Thus:
"Bivariate-Statistical Model"
4
-------
39
40
41
42
43
44
45
46
47
48
49
50
51
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
Contents, 3.4.1
Contents, 3.4.2
Contents, 3.5
Contents, 3.5.1
Contents, 9.
Contents, 9.1
Contents, 9.1.3
Contents, 9.2
Contents, 9.3
Contents, 10.
Contents, 10.1.2
Contents, 10.2
Contents, 10.3
No. C4 hyphen rule for './orris
left cut; hyphen ana space needed
after "Rationale" (assuming this
word applies to the model). Also
No. 23 applies here, zoo. Should
be: "Rationale- and Limitations
for Bivariate-Statistical Model"
See No. 38.
Hyphen error of type No. 02; should
be: "Probabilistic-Bioaccumula-
tion-Food-Chain Model" (assuming
the reference is to a probabi-
listic model of bioaccumulation
in the food chain).
"Rationale and Limitations" are
OK as written here, but cf. No.
39.
No. 02 hyphen; as in No. 33.
No. 0 2 hyphen, plus I would use
the words before the abbreviation
BAF (it to be included in
parentheses immediately
following). Thus, the correct
heading should read: "Data
Used for Development of
Bivariate-Bioaccumulation-Factor
(BAF) Models"
No. 02 hyphen, to read: "Water-
Column Data"
No. 02 hyphen, to read: "Results
of Bivariate-BAF Analysis"
Same as No. 46.
No. 0 2 hyphens to be added:
"Calibration of Probabilistic-
Bioaccumulation-Food-Chain Model"
No. 0 2 hyphen for "Water-Column
Invertebrates" (cf. No. 4 5)
Multiple No. 02; to be:
"Benthic-Invertebrate:Sediment-
Accumulation Factors"
Multiple No. 02; to be: "Water-
Column-Invertebrate:Water-
Accumulation Factors (BAFs)"
5
-------
52
= 3
34
55
56
57
p
58
59
60
61
62
»1
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
Contents, 10.3.2
Contents, 13.4
Contents, 10.4.2
Contents, 10.4.3
Contents, 10.5
Contents, 10.6.1
Contents, 10.7
Contents, 10.8
Contents, 10.9
p. E-l, par oi, 1.02
p. E-l, par 02, l.oi
See No. 49 for " . . .Water-Column
Invertebrates"
Cf. Nos. 5 0 & 51; should be
"Forage-Fish:Diet-Accumulation
Factors..."
See No. 49 for "Water-Column
Concentrations..."
No. 02; to be "Forage-Fish
Body Burdens..."
Cf. Nos. 50, 51, & 53: to be:
"Piscivorous-Fish:Diet-Accumulation
Factors..."
Hyphen No. 04 needed after "Ap-
roach" to read: "Approach- and
Calculations..."
Hyphen No. 02 to read: "Summary
of Probabilistic-Food-Chain
Models"
Hyphen No. 02; notice inclusion
of "Model" in the string (modifies
and precedes "Application"); to
be: "Illustration of Food-Chain-
Model Application"
Hyphens Nos. 02 & 04: to be:
"Comparison of Bivariate-
Statistical- and Food-Chain
Models." Can you guess the
left-out word requiring the
hyphen after "Statistical"?
(Hint: its first letter is M.)
At last, done with the Contents I
"No Action" before "decision"
requires No. 02 hyphen. If you
reject the hyphen, then the
only correct thing to do is
put a "(sic)" after "Action" to
show you are quoting verbatim a
wrong usage.
Need No. 02 hyphen after "mathemat-
ical" to read "mathematical-model-
ing efforts..."
6
-------
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
63 p. E-l, par 02, 1.02
64. p. E-l, par C2 , 1.06
65 p. E-l, par 03, 1.01
66 p. E-l, par 04, 1.01
67 p. E-l, par 04, 1.02
68 p. E-l, par 04, 1.02
69 p. E-l, par 04, 1.02
70 p. E-l, par 05, 1.01
71 p. E-l, par C7, i'.Ol
72 p. E-l, par 07, 1.02
73 p. E-2, par 01, 1.04
74 p. E-2, par 01, 1.06
7 5 p. E-2, par 01, 1.08
Need No. 04 hypnen after "prelimi-
nary" to read: "....neanc as a
preliminary- or interim report,"
Sentence to be recast :a show
completed past action. Replace
"When" with "After" and in
next line, change "are" to
"have been."
"Model-Calibration Report" (cf. No.
02)
Add hyphen after last word (No. 02)
Add hyphen & space after "health"
(cf. No. 04)
Add No. 02 hyphen after "ecologi-
cal" the phrase to read:
"...meeting human-health- and
ecological-risk criteria..."
Note: "continued No Action?"
is OK as written; no hyphen here.
Same as No. 69. No hyphen needed
in "No Action"
No. 04 hyphen & space needed after
"useful"
No. 02 hyphen needed after "mass"
the phrase to read: "...validate
useful- and scientifically cred-
ible mass-balance r.odels..."
See No. 14; should read "transport-
and fate of PCBs..."
Ditto; No. 04 hyphen & space need-
ed. to read: "...transport- and
fate of PCBs"
Insert comma after interface. This
is the first case of a systematic
adherence to leaving out the comma
before the final "and" in a string
of more than two items. I see
this usage a great deal, but think
that even if some expert says it's
OK, it tends to introduce ambiguity
I recommend changing throughout.
7
-------
7 6
77
78
79
30
81
32
83
84
35
36
37
88
89
90
91
92
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
p. E-2, first EF heading Needs hyphens as in Nos. 14 & 59;
to read: "Transport-ana-Fate-
Model Development"
p. E-2, par 02, 1.02 No. 7 5 comma needed after "solids
dynamics"
p. E-2, par 02, 1.03 No. 01 space & hyphen needed before
"sediments" (if the ref. is to
"river sediments")
p. E-2, par 02, 1.04 Need a No. 04 hyphen & space after
"diverse"
p. E-2, par 02, 1.04 Ditto after "developing"
p. E-2, par 03, 1.03 Insert No. 75 comma after (GE) ;
this is a good example to show
the ^.eed for my preferred
comma usage. A comma before the
first "and" eliminates any confu-
sion with the second "and"
p. E-2, par 03, 1.03 Need a No. 04 hyphen & space after
"private;" I would change this
line to read: "...(GE), and pri-
vate- and academic research..."
p. E-2, par 04, 1.01 Ne-'d No. 02 hyphen after "mass";
See also No. 09.
p. E-2, par 04, 1.02 Recommend No. 7 5 comma after
"solids"
p. E-2, par 04, 1.02 No. 01 space & hyphen needed before
"sediments"; See No. 73.
p. E-2, par 04, 1.06 No. 04 hyphen & space needed after
"physical"; to read "physical- and
chemical properties."
p. E-2, par 04, 1.08 Need No. 02 hyphen after "Phase"
to read: "Phase-2 monitoring..."
p. E-2, par 05, 1.01 "velocities;" See No. 24.
p. E-2, par 05, 1.04 "Depth-of-Scour Model" (cf. No. 12)
p. E-3, first line See No. 39.
p. E-3, par 01, 1.04 See No. 39.
p. E-3, par 01, 1.05 Insert No. 02 hyphen after "high"
to read: "high-flow events."
8
-------
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
93 p. E-3, par 01, 1.01
94 p. E-3, par 01, 1.04
95 p. E-3, par 01, 1.06
96 p. E-3, 3rd BF hdg
97 p. E-3, par 04, 1.04
97A p. E-3, 3rd BF hdg
97B p. E-3, par 05, 1.01
97C p. E-3, par 05, 1.04
98 p. E-3, 4th BF hdg
99 p. E-3, par 06, 1.02
100 p. E—4, par 01, 1.01
101 p. E-4, par 01, 1.04
102 p. E-4, par 02, 1.02
103 ?. E-4, 1st BF Hdg
104 p. E-4, par 03, 1.01
105 p. E-4, par 04, 1.01
106 p. E-4, par 04, 1.03
First sentence to read: "An existing
mass-balance model....was used for
hydrology-, solids-, and PCSs in
Lower Hudson River water and
-sediments."
"individual PCB homologues" means
what here? Were the FCBs expressed
as Aroclors?
Change "re-calibration of" to
"recalibrating."
To read: Development of Fish-Body-
Burden Models
Insert space and hyphen before
"sediments" (to indicate reference
to left-out words Hudson River).
See No. 3 8 for needed hyphens.
See No. 97A.
See NO. 97A.
See No. 41 for needed hyphens.
Same as No. 98.
Needs No. 04 hyphen after "historical."
Needs some No. 02 hyphens to read;
"average-body-burden estimates."
Insert No. 75 comma after "range."
To read: "Thomann Food-Chain Model."
See No. 103? + add hyphens (Nos. 02, 04)
to read: "PCB-transport- and fate
model."
Change "Since" to "Because." (Reserve
"since" for time usage—since 1974...)
Change "have" to "has;" the subject
of this sentence is "number" (singular)
not the object of the preposition of
(i.e. "conclusions"). You can avoid
this by finessing the "A-singular-of-
plural" usage by "several preliminary
conclusions have been drawn."
9
-------
107
108
109
110
111
112
113
114
114
115
116
117
118
119
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
p. E-4, par 05, 1.01 Must be "PCB-mass-balance model" (No.
02 hyphen fault).
p. E-4, par 05, 1.02 Insert: No. 7 5 comma after "dynamics."
p. E-4, par 06 Huge question: What is the timing of
the HUDTOX simulation with respect to
the recent, much-belated GE cleanup
of the oozing PCBs from the old mill?
p. E-4, par 08, 1.01 No. 2 hyphens needed to read "water-
column concentrations" and "dissolvea-
phase PCBs."
p. E-4, par 08, 1.03 Use of the words "appear to be" is
totally inappropriate. Unless you
have been hiding something, the
words "must be" should be substituted.
p. E-5, par 01 This point about pore-water (must add a
No. 02 hyphen here) PCBs being flushed
upward out of the sediments into the
river is very important. What is being
planned to test this hypothesis? Have
the individual-congener "fingerprints"
been investigated?
p. E-5, par 01, 1.04 No. 02 hyphen needed after "pore" to
read "pore-water advective flux."
p. E-5, par 02, 1.02 No. 02 hyphen needed after "river"
to read "river-flow velocities."
See also No. 24 about "velocities."
p. E-5, par 03, 1.01 Need 2 No. 02 hyphens to read:
"Thomoson Island Pool Depth-of-Scour
Model"."
p. E-5, par 03, 1.03 Need No. 2 hyphen after "cohesive."
p. E-5, par 04, 1.01 See No. 114.
p. E-5, pars 03 & 04 I recommend transposing these two
paragraphs to put the predicted depths
of scour ahead of the total predicted-
sediment/PCB scour.
p. E-5, par 05, 1.01 Need No. 2 hyphen after "Bivariate"
(See No. 38.)
p. E-5, par 05, 1.04 Need No. 2 hyphen after "water" to
read: "water-column concentrations."
10
-------
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
120 p. E-5, par 0 4, 1.04 No hyphen after adveros; thus none
needed after "highly" before
"chlorinated."
121 p. E-5, par C4, ..09 See No, 119.
122 p. E-5, par 0-4, 1.10 Change "while" (a time term) to
"whereas."
123 p. E-5, par 04, 1.10 Delete hyphen after bottom,' change to
read "bottom feeders." No basis for
any hyphen here except duplication of
wrong usage found in the literature.
124 p. E-5, par 05, 1.12 Need No. 2 hyphen after "water" and No.
04 hyphen after "column" to read:
"water-column- and sediment pathways."
(No. 04 hyphen needed to indicate that
"pathways" was left out after "water-
column." otherwise, must read "water-
column pathways and sediment pathways
II
125 p. E-5, par 06, 1.01 Need No. 2 hyphens; See nos. 41. 98, 99.
126 p. E-5, par 06, 1.06 Need No. 2 hyphen after Bivariate; See
nos. 38 & 97A.
127 p. E-6, BF heading Needs No. 02 hyphen after "Baseline"
to read "Future Baseline-Hodeling Ef-
forts."
128 p. E-6, par 02, 1.01 Need 2 No. 02 hyphens to read:
"Preliminary-Model-Calibration Report."
129 p. E-6, par 02, 1.02 Need a No. 01 hyphen before "fate" to
indicate connection to left-out "PCB;"
thus to read: "PCB transport, -fate..."
(or else it must read: "PCB transport,
PCB fate...")
130 p. E-6, par 02, 1.02 Need a No. 75 comma after "fate."
131 p. E-6, par 02, 1.03 Need a No. 01 hyphen before "bioaccumu-
lation;" (as in No. 129 to indicate th
connection to left-out "PCB;" thus to
read: "PCB transport, -fate, and
-bioaccuaulation" (or else it must read
"PCB transport, PCB fate, and PCB bioac
cumulation").
132 p. E-6, par 02, 1.04 Need a No. 02 hyphen after "more" to
read: "more-definitive conclusions"
(modifying- and preceding rule).
11
10.0227
-------
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
133 p. E-6, par C2, 1.07 See No. 14 hyphen discussion. I suggest
it read; "transport- ana fate mass-
balance models."
134 p. E-6, par 12, 1.07 Need a No. 32 hyphen after "fish" and
"body" to read "fish-body-burden
models."
135 p. E-6, par 02, 1.09 Need 2 No. 02 hyphens to read:
"preliminary-model-calibration work."
136 p. E-6, par 03, 1.01 No hyphen after "finely." (See No. 120.
137 p. E-6, par 03, 1.03 Need a No. 02 hyphen after "suspended"
to read "suspended-solids data."
138 p. E-6, par 03, 1.05 A gold star for the hyphen in "long-
term" modifying and preceding. But
probably a No. 02 hyphen is needed
before "hindcasting" (all precede
"calibration."
139 p. E-6, par 04, 1.01 See No. 115.
140 p. E-6, par 05, 1.01 See No. 05.
141 p. E-6, par 05, 1.03 Insert No. 75 comma after "inverte-
brates . "
142 p. E-6, par 06, 1.05 See Nos. 05 and 140.
143 p. 1-1, par 02, 1.04 Need No. 04 hyphen after "manufactured '
144 p. 1-1, par 02, 1.08 • No hyphen after "highly" (See No. 120.)
14 5 p. l-l, pars 3/4 Shame on you for ignoring all the work
HITS DEC did before EPA started doing
anything constructive! You cite some
this work on p. E-S (par. 3 last line)
as 1984 NYSD1C Survey (same as EPA
1984 Feasibility Study mentioned here?)
Many problems may exist in using some
of the NYBDEC work, but at tha time
when it was done, it was a major
achievement.
[I have started to skip through pages; it has taken me too long to la
out my extensive hyphens tutorial. You'll probably ignore it anyway
146 p. 1-2, par 01, 1.05 You get 50% for hyphens here; need
a No. 02 hyphen after "water" to read
"water-column sampling" and you get
a rare gold star for the correct
hyphenation of "flow-averaged" (modify
ing and preceding "composites").
12
-------
147
148
149
150
151
152
153
154
155
156
157
158
159
Comment numbers for PP-1.1 to PP-1.159 are given on the left of the page.
p. 2-1, par 06, 1.04 "homologues?" See No. 34.
p. 2-2, par 04, 1.04 "homologues?" See Nos. 34 & 146,
o. 2-6, par 05, 1.02 Delete hyphen after "vertically" See
No. 120.
p. 3-5, par 02, 1.05 Please spare us the use of "geometry*
where you should employ "configuration;"
"geometry" is a specific mathematical
discipline. You could use it as an
adjective as: "geometric configuration."
p. 3-5, par 02, 1.08 A good place far a review of many hyphen
faults, should read: "Particle gross-
settling-, resuspension-, and net-burial
celerities." (Note insertion of a No.
75 comma after "resuspension-," and
the change from "velocities" to
"celerities" (No. 24).
p. 3-5, par 03, l.oi "geometry;" (See No. 150.)
p. 3-6, par 05, 1.02 "Since" (See No. 105.)
p. 3-6, par 05, 1.05 Change "unbound" to "nonbound." The
prefix "un" implies that something was
in one state, then changed to an op-
posite state, as in a tied shoelace
that becomes untied. Thus "unbound"
correctly means it was bound and then
was bound no more. "Non" means
it never was that way. A "nontied"
shoelace is what is in the shoebox at
the shoes-core.
p. 3-7, par 03, 1.02 "Since" (See Nos. 105 & 153.)
p. 3-7, par 05, 1.01 "geometry;" (See Nos. 150 & 152.)
p. 3-7, par 05, 1.07 Delete hyphen after "finely." See
NOS. 120 & 149.
Fig. 1-3 Lower R corner Rensselaer County is spelled with
one N and two S's, not as shown.
Fig. 1-4 Upper R corner See No. 158.
13
-------
CO
H—
a>
iU
o
cc
<
X
o
CO
a
<
3
<
Z
<
LU
CO
"O
c
co
CO
D
O
jc
H
MEAN ANNUAL H.R. DISCHARGE
Green Island, NY (1929-1988)
25
20
15
10
5
425
1941 Jan 05
1968 May 23
~T"
430 435 440 445
TIME (JD-T2000000), THOUSANDS
450
-------
1ooooo
Jan 05
1941
May 23
1968
Oct 09
1995
> 10000
1000
425 430 435 440 445 450
TIME (JD+2000000), THOUSANDS
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
General Electric
(PMCR - PG)
-------
PG-1
CHICAGO
Dallas
I.OS ANGELES
S i d l e y & Austin
A P ARTSEKSItP ISCLVDIS'6 PROFESSIONAL CORPORATIONS
172-J Eve Street. \ U*
Washington. D C joooo
TraEPHONH ^02 7^C> SOOO
Facsimile :J02 Tie 8711
Founded 18S<5
VKW YORK
LONDON
SINGAPORE
TOKYO
1VHITWI DtKKCT Ni MDF.H
(202) 726-1271
November 21, 1996
Mr. Douglas Tomchuk
Emergency and Remedial Response Division
U.S. Environmental Protection Agency
Region 2
290 Broadway - 20th Floor
New York, NY 10007-1866
Re: Hudson River PCBs Superfund Site: PMCR Comments
Dear Mr. Tomchuk:
Attached are General Electric Company's ("GE's") comments on EPA's "Phase 2
Report - Review Copy, Further Site Characterization and Analysis, Volume 2B - Preliminary
Model Calibration Report, Hudson River PCBs Reassessment RI/FS" (October 1996) ("Report").
Please place this letter and the attached comments in the administrative record for the Hudson
River PCBs Superfund Site ("Site").
The Report provides an overview of the current status of EPA's modeling effort
for the Site, which is designed to assess the effect of possible remedial actions, including no
action, addressing the PCB-contaminated sediments in the Upper Hudson River. GE applauds
EPA's decision to provide an opportunity to comment on its modeling effort while it is still a work
in progress and is in fundamental agreement with the Agency's stated goals for preparing and
using these models, the the principles which guide the development of these models, and the
Agency's intent to validate them against existing data.
GE has a number of concerns about the models that EPA is developing. The solids
mass balance underestimates solids loading from tributaries to the Upper River, overestimates
resuspension and deposition rates, and improperly decouples net sedimentation from resuspension
and deposition, all of which lead to an overstatement of the transfer of PCBs from solids to water.
The PCB mass balance uses sediment data from 1991 to represent 1993 conditions, ignoring the
substantial release of sediments and PCBs to the River from the Allen Mill during the interim.
These and other problems with the fate and transport model will become more apparent as EPA
-------
Sidley & Austin
Washington, D.C.
Mr. Douglas Tomchuk
November 21, 1996
Page 2
attempts to complete its "hindcasting" against the historical data. In addition, although EPA's
steady-state, statistically-based bioaccumulation models may provide some useful information, the
Agency should not use them as predictive tools because they ignore variability in the relationships
among PCBs in water, sediment and biota, a time-variable, mechanistic food-web model, such as
the Gobas Model, is more appropriately used for predictive purposes. Finally, although EPA's
depth of scour model uses the right approach to analyze the resuspension of cohesive sediments
during flood conditions, EPA has made several errors in developing this model and must take care
in selecting an appropriate formulation for estimating resuspension of non-cohesive sediments.
All these issues, as well as several others, are set out in detail in the attachment.
There are three issues that we emphasize in this letter. First, as the Agency
acknowledges in the Report, its PCB mass balance cannot calibrate to the water column PCB data
in the Thompson Island Pool ("TIP") without resorting to an untested hypothesis. With no
supporting data, EPA assumes the existence of a spatially-limited groundwater influx through the
TIP sediments that purports to flush a sufficient quantity of PCBs from the sediment to account
for the mass imbalance of PCBs across the TIP. EPA must recognize, however, that there are
other, equally plausible hypotheses that can account for this mass imbalance. For example, the
release of a large volume of PCB-contaminated sediments from the Allen Mill between 1991 and
1993 could have deposited fresh PCBs into the TIP sediments and, combined with biodegradation
of these PCBs, could provide the source of the "excess" PCBs found in the water column at the
Thompson Island Dam. Alternatively, it is possible that the water column sampling stations are
not identifying the true amount of PCBs that move through the TIP, either under-quantifying the
amount of PCBs entering the pool or over-quantifying the amount leaving the pool.
All these hypotheses must be considered and tested if EPA is to rely on its model
with any confidence to make predictions about potential courses of action. The mass imbalance
of PCBs across the TIP affects PCB levels in the TIP and further downstream. Understanding the
source of these excess PCBs is critical to understand the effects of various potential courses of
remediation. Without a factual grounding for and understanding of the cause of the mass
imbalance, the model will not accurately predict the fate and transport of PCBs into the future. If,
for example, EPA assumes an untested groundwater influx, the model may suggest that
remediation addressed to deeply buried sediments will reduce the bioavailability of PCBs in the
Upper River. If, however, the apparent imbalance were the result of being unable to measure the
full amount of PCBs passing Rogers Island, an intrusive remedy aimed at deep sediments would
have no real benefit and could possibly worsen conditions in the River. As a result, EPA cannot
use its model to make predictions until the issue of the mass imbalance of PCBs across the TIP is
resolved.
As EPA is aware, GE is undertaking studies aimed at answering this question and
identifying which of the possible hypotheses is the cause of the TIP imbalance. We intend to
-------
Sidley 8c Austin
Washington. D.C.
Mr. Douglas Tomchuk
November 21, 1996
Page 3
provide the results of these studies to EPA as they become available and look forward to working
with the Agency to resolve this critical issue.
Second, to ensure that one can have confidence in the models when they are used
to predict the effects of various remedial alternatives, the models must closely match available
data. To date, EPA has only calibrated its models against a temporally limited data set between
January and September 1993 If EPA's models provide a close fit against the more extensive
historical data set, then arguments for their use as predictive tools will be stronger. There are a
number of tests to validate the models:
Solids Balance: comparison of model predictions with data on the spatial patterns
of TSS during low flow, temporal and spatial patterns of TSS and water column
PCBs during flood events, and annual average solids loading passing Schuylerville,
Stillwater and Waterford will all verify the solids balance.
• PCB Fate comparison of model predictions with data on spatial patterns of water
column PCBs during low flow and spatial changes in water column PCB
composition will verify PCB flux from pore water and PCB loss by volatilization.
• PCB Loss: comparison of model predictions with data on long-term changes in
surface sediment PCB levels, vertical profiles of PCBs in sediments, and PCB
inventory in sediment will verify the loss of bioavailable PCBs from sediment.
Overall Test of Model comparison of model results with data on the annual
average flux of PCBs passing Schuylerville, Stillwater and Waterford will provide
an overall test of the model
Effect of Allen Mill Release: comparison of model results with data on the
apparent increase in the PCB flux from Fort Edward to Thompson Island
Dam/Schuylerville that occurred between the mid- to late-1980s and the 1990s will
verify that the model reflects the effects of the Allen Mill release.
• Bioaccumulation Models comparison of model results with data on temporal
changes in predatory and forage fish at the TIP and Stillwater over a 15-year
period and the response of the fish to the short-term changes in water column PCB
levels in the early 1990s will verify the predictive power of the bioaccumulation
models
These comparisons will uncover any apparent biases and are essential to have confidence in the
predictive power of the models
-------
Sidley & Austin
Washington. D.C.
Mr. Douglas Tomchuk
November 21, 1996
Page 4
Third, the proposed use of the Thomann model, which analyzes and predicts the
body burden of PCBs in striped bass in the Lower Hudson River, raises fundamental issues about
the scope and focus of the reassessment. The Report states, "The purpose of the Reassessment is
to determine an appropriate course of action for the PCB contaminated sediments in the Upper
Hudson River in order to protect human health and the environment." (Report at E-l) The scope
and focus does not include any consideration of remedial action in the Lower Hudson. It is
undisputable that there are a number of significant PCB discharges into the Lower Hudson River
which affect PCB levels in fish, such as the striped bass, in the Lower River and that EPA has not
identified any parties responsible for those Lower River discharges as PRPs in this matter.
Presumrlly, the justification for this is found in the purpose of the Reassessment: its remedial
scope and t^cus are confined to the Upper River. This scope is appropriate because,
notwithstanding EPA's claim to the contrary in the Report (Report at E-l), the Site is confined
entirely to tlie Upper River. See Administrative Record, NPL-UI-2-29 (EPA).
EPA must accept the constraints that are imposed as a result of the limited
geographical reach of the Site and the Agency's choice to limit its review to the Upper River. It is
reasonable to look at the effect of potential remedial measures in the Upper River to assure that a
possibh remedial course of action will not have adverse remedial effects on the Lower River or,
at most, if there are adverse effects, that they are acceptable when weighed against other benefits.
Justifying Upper River remedial action on the basis of benefits to Lower River fish
is an entirely different matter. If benefits to the Lower River are to be used to justify remedial
action in the Upper River, there must be an investigation and evaluation of remedial alternatives,
such as source control, in the Lower River, and a congruent recognition that responsibility for
achieving these benefits falls on a much wider group of parties than the present PRPs; that wider
group must be classified as PRPs and treated as parties to this proceeding.
EPA must be clear what its objectives in the Lower River are in this reassessment.
If the objective is simply to avoid any increase in risk in the Lower River, the scope of the
reassessment need not address sources of PCB discharge in the Lower River. If the objective is
to decrease risk or attain a human health protection or an ecological risk reduction goal in the
Lower River, EPA must address the Lower River sources of PCB discharge. EPA can not defend
as cost-effective a remedy for the Lower River which examines only Upper River sources. It may
well be that the cost effective remedy for the Lower River is control of Lower River sources, and
the failure of the Agency to consider and analyze that obvious and plausible possibility will render
clearly arbitrary any Upper River course of action which is justified on the basis of Lower River
benefits. In addition, fundamental fairness is involved: the costs of a Lower River remedy should
not be bome entirely by Upper River sources, thus providing a remarkable windfall to Lower
River sources.
-------
— SlDLEY 8c
Austin
Washington. D.C.
Mr. Douglas Tomchuk
November 21, 1996
Page 5
It follows that if EPA intends in any way to justify and support a remedial course
of action in the Upper River by reference to benefits in the Lower River, it must investigate the
sources of PCB discharge to the Lower River that contribute to the human health or ecological
risk in the Lower River; and determine whether the cost effective remedy to obtain that beneficial
reduction in risk is remedial action in the Upper River or the Lower River or a combination of the
two. Such an investigation and determination calls for identifying the Lower River dischargers as
PRPs so that, like GE, they can fairly express their views and provide the benefit of their expertise
and analysis.
GE looks forward to working with the Agency to address the issues we have
identified in these comments. In light of GE's own experience developing an integrated PCB fate,
transport and bioaccumulation model for the Upper Hudson River, we are available to discuss our
comments and the related issues with the Agency and to work with EPA to improve the predictive
power of the models to ensure that a factually-based remedy is selected.
^Sincerely yours.
Angus Macbeth
Thomas G. Echikson
attachment
-------
Sidley 8e Austin
Washington, D.C.
Mr Douglas Tomchuk
November 21,1996
Page 6
cc: Richard Caspe (USEPA Region II)
Michael Zagata (NYSDEC)
Paul Simon (USEPA Region II)
Ann Rychlinski (USEPA Region II)
John C ah.il! (NYSDEC)
Ronald Sloan (NYSDEC)
William Pons (NYSDEC)
Walter Demick (NYSDEC)
Steven Hammond (NYSDEC)
Jay Field (NOAA)
Ann Secord (USFWS) /
A1 D'Bernardo (TAMS) /
Victor Bierman (Limnolech)
Charles Menzi (Menzi-Curie)
Jon Butcher (Tetra Tech)
-------
COMMENTS OF GENERAL ELECTRIC COMPANY ON
PHASE 2 REPORT - REVIEW COPY
FURTHER SITE CHARACTERIZATION AND ANALYSIS
VOLUME 2B - PRELIMINARY MODEL CALIBRATION REPORT
HUDSON RIVER PCBs REASSESSMENT RI/FS
OCTOBER 1996
November 21,1996
Melvin B. Schweiger John Connolly
John G. Haggard HydroQual, Inc.
General Electric Company 1 Lethbridge Plaza
Corporate Environmental Programs Mahwah. NJ 07430
1 Computer Drive South (201) 529-5151
Albany, NY 12205
(518)458-6646
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
TABLE OF CONTENTS
EXECUTIVE SUMMARY ii
I. INTRODUCTION 1
II. F.PA's FATF. AND TRANSPORT MODF.T 4
A. Solids Mass Balance 5
Solids Loading from the Tributaries 6
Solids Deposition and Resuspension 9
Sedimentation Rate 12
B. Calibration of PCR Mass Balance 13
Relationship Between Solids Transport
Components and PCB Fate Components 13
Initial Sediment Conditions 15
Analytical Issues 16
Dechlorination/Biodegradation Issues 18
C. PCB Mass Imbalance in the Thompson Island Pool 20
D. Long-Term PCB Mass Balance 25
ffl. BTOACCUMULATION MODFT.S 30
A. The BSM and PFCM Models 31
Structure of the Models 31
Validation of the Models 34
B. The Gobas Model 35
C. Recommended Approach 36
IV. DF.PTH OF SCOUR MODF.T 36
V. PRF.DICTTVE POWF.R OF TTTF MODFT. 40
VI. I.OWF.R HUDSON PCB TRANSPORT AND FATE MODF.T 43
VII. CONCLUSIONS AND RECOMMENDATIONS 45
LIST OF REFERENCES 50
-------
EXECUTIVE SUMMARY
General Electric Company ("GE") submits these comments on the United States
Environmental Protection Agency's ("EPA") "Phase 2 Report - Review Copy, Further Site
Characterization and Analysis. Volume 2B - Preliminary Model Calibration Report, Hudson
River PCBs Reassessment RI/FS" (October 1996) ("Report"). The Report describes the current
status of EPA's modeling effort to predict the levels of PCBs in sediment, water and fish in the
Upper Hudson River under various remedial scenarios, including no action.
It appears that GE and EPA agree on several important aspects of the modeling
effort for the Hudson River PCBs Superfund Site. We agree generally with the objectives and
focus of the modeling effort identified in the Report, as well as the Agency's conclusion that
modeling is the appropriate way to address the questions of PCB fate, transport and
bioaccumulation in the Upper Hudson. We also agree with EPA that the principle of "mass
balance" should be the basis of its models and on the need to assure that model predictions are
calibrated against and are consistent with the available data.
Notwithstanding this general agreement, there are several fundamental problems
with EPA's models that should be corrected as EPA moves forward. We urge EPA to correct
these problems and look forward to assisting the Agency in this endeavor. Given that the
modeling effort is a "work in progress." we request that the Agency provide an opportunity to
comment on its work in refining the models before they are used to make predictions about the
ii
-------
appropriateness of remedial action. The problems that EPA must address as it develops its
model more fully are set out in detail in these comments and are summarized below:
1. There are three significant problems with the solids balance in EPA's fate and
transport model. EPA has underestimated the solids loadings to the Upper River from the Snook
and Moses Kills. EPA's deposition and resuspension rates, particularly during low flow, are too
high and cannot be calibrated against the other solids parameters. Finally, the sedimentation rate
is not integrated into the model as the net of deposition and resuspension. violating the principle
of mass balance.
2. The problems with the solids mass balance affect EPA's estimates of the PCB
mass balance. By overestimating resuspension and deposition, underestimating tributary solids
loadings, and decoupling sedimentation from the other solids parameters, the model overstates
the transfer of PCBs from sediment to water, which, in turn, implies greater benefits from
remedial actions aimed at sediments than will be the true case.
3. GE has several other concerns with the PCB mass balance in EPA's model. The
model's estimate of initial conditions of PCBs in the sediment are based on data that do not
reflect the significant loadings of sediments and PCBs from the Allen Mill in 1991 to 1993.
EPA's model also fails to consider the effect of PCB dechlorination in the sediments of the Upper
Hudson. The manner in which EPA "corrected" GE's PCB data also appears to contain errors.
-------
Finally, the limited time period for which EPA has attempted to calibrate the model is
insufficient to test the model's ability to represent the long-term fate of sediment PCBs.
4. EPA's model is unable to achieve a PCB mass balance across the Thompson
Island Pool ("TIP") without resort to hypothesized mechanisms, particularly the introduction of
groundwater flux of buried PCBs, for which there is no factual demonstration. There are other
plausible hypothesized mechanisms that can explain the mass imbalance of PCBs across the TIP,
including (1) inaccurate estimation of the PCB load across the TIP by GE's monitoring program,
(2) increased surface sediment concentrations resulting from the Allen Mill release. (3) external
load from dredge spoil sites, or (4) resuspension of surface sediments at low flows. Until the
cause of the mass imbalance of PCBs across the TIP is understood, model predictions of
remedial scenarios will not be sufficiently fact-based to be useful in addressing the key
reassessment questions. GE is working to collect data to test all these hypotheses and will
provide these data to EPA as they become available.
5. The shortcomings in EPA's preliminary fate and transport model will become
more apparent as EPA attempts to calibrate it against the historical PCB data in fish, water and
sediments. For example, EPA's proposed groundwater inflow would result in greater quantities
of PCBs moving into the water column from sediments than the historical water monitoring data
show. Similarly, this mechanism would result in depletion of the PCB inventory at a rate that is
much higher than the sediment data show.
iv
-------
6. EPA should not use the two steady-state statistically-based bioaccumulation
models — the Bivariate Statistical Model and the Probabilistic Food Chain Model — to make
predictions because they ignore the short and long-term variability in the relationships among
PCB levels in the water column, sediment and fish and do not attempt to describe or respond to
the mechanisms by which fish bioaccumulate PCBs. Instead. EPA should use a time-variable,
mechanistic food web model, such as the Gobas model, which explicitly incorporates variability
in exposure, uptake and depuration of PCBs in fish and reflects real world bioenergetic and
toxicokinetic mechanisms. GE has been developing such a model and offers to share this work
with EPA as the Agency develops its food web model.
7. EPA's depth of scour model to predict the effect of a 100 year flood on solids and
PCBs in the Upper Hudson River is sound in its application of principles and its analytic
approach. The model properly applies the Lick equations to the dynamics of cohesive sediment
resuspension. but we believe there are some errors in the application of these equations. In
addition, we recommend that EPA use a modified van Rijn model to model the resuspension
properties of non-cohesive sediments in the Upper Hudson.
8. EPA must test its models against the extensive data set for the Upper Hudson
River to have any confidence in their predictive powers. Specifically, EPA should validate its:
• Solids balance model against (1) spatial patterns of TSS during low flow, (2)
temporal and spatial patterns of TSS and water column PCBs during flood events,
and (3) annual average solids loading passing Schuylerville, Stillwater and
Waterford.
-------
PCB fate model against (1) spatial patterns of water column PCBs during low
flow and (2) spatial changes in water column PCB composition.
Estimate of PCB loss against long-term changes in surface sediment PCB levels,
vertical profiles of PCBs in sediments, and PCB inventory in sediment.
Fate and transport model against the annual average flux of PCBs passing
Schuylerville. Stillwater and Waterford.
Estimate of the effect of the Allen Mill release against the apparent increase in the
PCB flux from Fort Edward to Thompson Island Dam/Schuylerville that occurred
between the mid- to late-1980s and the 1990s.
Bioaccumulation models against temporal changes in predatory and forage fish at
the TIP and Stillwatei over a 15-year period and the response of the fish to the
short-term changes in water column PCB levels in the early 1990s.
9. EPA has not clearly defined its objectives in using the Thomann model of PCB
fate, transport and bioaccumulation in the Lower Hudson River. Given the present scope of the
reassessment, EPA should only use this model to assess whether remediation in the Upper River
would have an unacceptable adverse impact on the Lower River, and only after Thomann and
Farley have completed their update of the model.
GE commends EPA for providing this opportunity to comment on its modeling
effort. As EPA recognizes, some of the topics in the Report are closely related to EPA's yet-to-
be-issued Data Evaluation Report. As a result, we may revisit some of these topics as they are
developed more fully in future reports.
vi
-------
I. INTRODUCTION
In 1989, Region II of the U.S. Environmental Protection Agency ("EPA") decided
to reassess its 1984 decision under the Comprehensive Environmental Response. Compensation,
and Liability Act ("CERCLA" or "Superfund") that no action should be taken with regard to
sediments at the Hudson River PCB site (U.S. EPA, 1984). This reassessment is to determine
what CERCLA action, if any, should be taken with regard to the PCB-contaminated sediments in
the Upper Hudson River. The reassessment is focused on answering three central questions
(U.S. EPA, 1996b; pg. 3-1):
1. When will PCB levels in fish populations recover to levels meeting human
health and ecological risk criteria under continued No Action?
2. Can remedies other than No Action significantly shorten the time required
to achieve acceptable risk levels?
3. Are there contaminated sediments now buried and effectively sequestered
from the food chain that are likely to become "reactivated" following a
major flood, possibly resulting in an increase in contamination of the fish
population?
As a first step, the reassessment reviewed existing data (U.S. EPA, 1991) and
collected new data. EPA has now turned to developing a fate and transport model and three
associated bioaccumuiation models for the Upper Hudson River to help answer these questions.
In addition, the Agency has developed a separate sediment model for flood conditions and
examined an existing model of PCB concentrations in striped bass in the Lower River. If
successful, the modeling effort will replicate data reflecting known past conditions in the River -
water flows, total suspended solids ("TSS") and PCB concentrations in water, and PCB
concentrations in sediment and through the aquatic food chain up to the fish that people, birds,
and other animals may eat. If successfully developed. EPA can then use the models to predict
10.0245
-------
the course of natural recovery with no intrusive remedial activity, as well as the future PCB
concentrations in environmental media and aquatic biota under assumed remedial scenarios. In
October 1996, EPA issued for review and comment its Preliminary Model Calibration Report
("Report") (U.S. EPA. 1996b).
These comments, prepared with the aid of HydroQual. Inc. (experts in modeling
the behavior of contaminants in surface water and biota) set out General Electric Company's
("GE's") views regarding the Report. As the Agency recognizes, some of the topics addressed in
the Report are not fully developed, and some are intertwined with issues to be addressed in the
Data Evaluation Report, which the Agency has not yet issued. Consequently, as these topics are
more fully developed and discussed in future reports. GE is likely to return to many of the issues
addressed in these comments.
GE generally agrees with the fundamental objectives and focus that are the
foundation for the present Report. GE largely agrees with EPA on the central questions which
the reassessment should address (GE believes, however, that EPA needs not only to consider
what impact a large flow will have on PCB levels in the River, but also whether and to what
extent remedial efforts would mitigate these impacts). GE also agrees with EPA that modeling is
the appropriate methodology for addressing the questions of PCB fate, transport and
bioaccumulation in the Upper Hudson River and is the appropriate way to achieve the
reassessment objectives. Moreover, GE agrees with EPA that mass balance is the appropriate
principle on which to base the fate and transport model.
2
-------
Based upon the information provided in the Report, we believe EPA and GE also
agree on the basis by which a model should be judged: (1) its congruence with accepted scientific
laws and principles: (2) the internal consistency of the physical, chemical, and biological
mechanisms that are reflected in the model: (3) the plausibility and reasonableness of the
professional judgments and assumptions used at points where scientific principle or relevant data
do not constrain the modeler; and, most important, (4) its ability to replicate observations of the
physical system being modeled over appropriate time and space scales.
Modeling provides a method for describing thr relationship of elements in a
complex natural system, such as the Upper Hudson River. It requiu" the modeler to analyze and
describe the relations between the elements in the system. It develops knowledge of which
elements have the greatest influence over the results of interest and which elements have the
greatest uncertainty associated with them. This process allows a refined focus on those areas
where data collection, laboratory experiments or refinement of judgment will be of greatest value
- and in some cases will be essential - to produce modeling results that will be useful for
decision-making. The validation of the model against known data provides an acid test for the
ultimate value of the model and defines the degree to which one's confidence in the models'
predictive power is warranted.
In its present iteration, many aspects of EPA's models appear to meet the
evaluation criteria set out above, but there are others where the model clearly falls short. This
may be the result of the EPA's models being works in progress, and we commend EPA for
3
-------
seeking input during the development phase to help improve the quality of the Agency's effort.
This is the spirit in which we submit these comments. Our comments address these central
aspects of modeling and reflect, as well, the experience of GE and its consultants in attempting to
construct GE's own PCB fate, transport and bioaccumulation model for the Upper Hudson River.
II. F.PA's FATE AND TRANSPORT MODEL
EPA's fate and transport model attempts to model the mass balance of solids and
PCBs in the Upper River. Four factors primarily control the calibration of this model: solids or
sediment transport, PCB partitioning, upstream loading of PCBs, and initial sediment PCB
conditions. Because these factors also control PCB dynamics in the river and the operation of
the model, it is critical that the model accurately define them. GE and EPA have no material
disagreement about PCB partitioning or what the data show as the upstream loading of PCBs
(although a question remains as to the accuracy of that data, particularly after the Allen Mill
discharges of 1991 to 1993). We do have clear differences of opinion on EPA's assumed initial (\
sediment PCB conditions because the model uses 1991 sediment PCB data to reflect conditions
in 1993, ignoring the large amount of PCB-containing sediment that entered the River upstream
of the model boundary between September 1991 and March 1993 from the failure of the Allen
Mill. This discharge probably affected surficial sediment conditions in the Thompson Island
Pool ("TIP"). We also have several areas of significant disagreement with regard to solids (or
sediment) transport issues and the associated behavior of PCBs. These are the central issues on
which we focus our comments on EPA's fate and transport model.
4
-------
A. Solids Mass Balance
The physical-chemical fate and transport model that EPA has developed for the
Upper Hudson River appropriately relies on the principle of mass balance. As both EPA and GE
have recognized, solids transport is critical to the fate of PCBs in the River because of the
affinity of PCBs for solids, the release of PCBs to the water column through sediment
resuspension, and the burial of PCBs through net solids deposition. Development of a solids
mass balance model that accurately simulates the loading, resuspension. deposition and transport
of suspended solids and sediments in the Upper Hudson River is thus necessary to the calibration
of the PCB model and to the model's ability to evaluate remedial alternatives. Without a detailed
understanding of sediment transport, the model will not accurately predict the fate and transport
of PCBs in the Upper Hudson River and thus will not materially assist EPA in its evaluation of
alternatives.
The solids mass balance described in the Report has several significant
shortcomings:
1. Solids loading to the River from Snook Kill and Moses Kill are underestimated:
2. Deposition and resuspension rates, particularly during low flow, are too high; and
3. The sedimentation rate is not integrated into the model as the net of deposition
and resuspension. violating the principle of mass balance.
The solids mass balance is constrained by data on TSS concentrations and mass
flux in the water column at various points in the River. From reach to reach, the TSS levels are
controlled by solids loading from upstream and tributaries and by the sedimentation rate. The
5
-------
sedimentation rate, in turn, is the net difference between solids deposition and solids
resuspension. Thus, solids loading, deposition and resuspension, and sedimentation rate are all
closely related in achieving the solids mass balance.
There also is an important link between the solids and PCB mass balances.
Changes in the internal working of the solids mass balance model can have major impacts on the
PCB mass balance. The PCB mass balance provides a check on the solids mass balance and vice
versa. The effect of changes made in one aspect of the model must be considered in evaluating
both aspects of the model. Consequently, many of the issues that arise in examining the solids
balance estimates also arise in the context of the PCB mass balance estimates.
Solids l oading from the Tributaries
EPA has underestimated solids loadings from tributaries to the Upper Hudson
River. Although EPA uses the proper approach for estimating loads from the upstream and
major tributary sources, the method used to estimate solids loads from minor tributaries,
particularly into the TIP. results in a significant underestimation of actual loadings. This is
confirmed by data that EPA collected in 1994 and reported in the Database Report (U.S. EPA,
1995), but which EPA has not yet incorporated into its model.
The standard and accepted method for estimating solids loading is to use available
TSS and flow rate data to develop a solids rating curve, which can then be used to predict solids
loading as a function of flow rate. This is the method EPA used to develop solids loadings
6
-------
relationships for the Upper Hudson River at Fort Edward (the upstream boundary), the Hoosic
River, and the Mohawk River, and GE generally agrees with EPA's approach with respect to the
loadings estimates for these sources. GE also agrees with EPA's use of the minimum variance
unbiased estimator ("MVUE") to correct for log-linear regression analysis bias during
development of the solids rating curve because this will provide increased accuracy in estimating
solids loads.
Nevertheless, we believe that further analysis of the larger data set will improve
the accuracy of these loadings estimates. First, with respect to the loadings estimate at Fort
Edward, EPA should include all the available TSS data in the MVUE analysis. By relying only
on the limited EPA Phase 2 1993 TSS data set, the validity of EPA's analysis is uncertain. EPA
rejected both the earlier TSS data, claiming that the TSS-flow relationship had changed over
time, and the TSS data GE collected in 1993, but the Agency failed to conduct a proper statistical
analysis to ensure that it was appropriate to exclude these data. We are confident that such an
analysis would show that EPA should include a large portion of the data set in the TSS-flow
relationship, which will increase the power and accuracy of the regression. Second. EPA should
validate its loadings estimates for all the tributaries and the upstream boundary with data
collected in April, 1994 and other high flow periods. This validation is important to determine
whether EPA's solids loadings estimates are correct.
EPA correctly recognized that Batten Kill and Fish Creek contribute significant
solids loads to the Upper Hudson River (U.S. EPA, 1996b; pg. 4-7). The Agency estimated the
7
-------
loadings from these tributaries by comparing them to and developing a ratio of the TSS-flow
relationship developed for the Hoosic River. This is not an unreasonable first approximation, but
EPA should test the vaiidity of the approach by comparing predicted with observed loads in the
Batten Kill during April. 1994.
EPA's method for estimating TSS loads for Snook Kill and Moses Kill, on the
other hand, is fundamentally flawed. EPA improperly assumed a constant TSS value of 5 mg/1
for these tributaries, ignoring the evidence that they contribute substantial solids. EPA should
use the data from April, 1994 to establish the relationship of solids loading from these tributar es
to loadings in the Hoosic River, similar to the approach used to estimate loadings from Batten
Kill and Fish Creek. Although these data are limited, they do show that these tributaries can
contribute significant loadings into the TIP, with peak values of over 200 mg/1.
EPA's assumed loadings value for the Snook and Moses Kills results in an
implausible loading per unit drainage area (i.e.. sediment yield) for these tributaries. During the
1993 calibration period. EPA's assumed rate results in a sediment yield of 1.1 metric tons/mi: for
the Snook and Moses Kills, 100 times less than the calculated yield of 111 metric tons/mr for the
Hoosic River. This vast discrepancy cannot be accounted for simply by differences in soil type
and land use along the Snook and Moses Kills and the Hoosic River and thus demonstrates the
invalidity of the 5 mg/l assumption for the Snook and Moses Kills.
8
-------
Solids Deposition and Resuspension
The formulations that EPA presently uses to simulate deposition and resuspension
in the solids transport model are unsupported by generally accepted theory and will not produce
accurate predictions of solids or PCB fluxes across the sediment-water interface. EPA must
include resuspension and deposition formulations that are physically consistent with the observed
behavior of sediments in both laboratory and field studies in its modeling framework before it
uses the solids transport model to make predictions.
There are four fundamental problems with EPA's simulation of resuspension and
deposition: (1) resuspension during low flows; (2) constant resuspension rate at high flow; (3)
constant settling velocity; and (4) assumed resuspension rates based on judgment and not
calibrated with supporting data. These invalid assumptions result in a model that overestimates
resuspension of solids and PCB movement from the sediment bed into the water column.
First, laboratory and field studies on cohesive and non-cohesive sediment erosion
properties show that resuspension only occurs when the bed shear stress exceeds some critical
value, which depends upon the bed properties (Galiani, et al., 1991; Hawlev, 1991; Hayter and
Mehta, 1986; Parehure and Mehta, 1985; van Rijn, 1984; Araturai and Krone, 1976).
Examination of bottom shear stresses predicted by a two-dimensional hydrodynamic model of
the Upper Hudson River that GE is developing indicates that critical shear stresses are not
exceeded at typical low flow rates. As a result, resuspension will be negligible in both cohesive
and non-cohesive bed areas during low flows in the Upper Hudson River. This conclusion is
9
-------
supported by modeling results, using calibrated and validated sediment transport models, in other
riverine systems (Ziegler and Nisbet, 1994).
Second, use of a constant resuspension rate during high flows ignores the
observed phenomenon of sediment bed armoring for both cohesive and non-cohesive sediments.
Rather than continual resuspension at high flow, resuspension occurs only until the surface is
depleted of resuspendable particles, leaving larger or cohesive particles at the surface that form
an armoring layer for the panicles below (Ziegler and Nisbet, 1994; Karim and Holly, 1986; van
Niekerk. et al.). The assumption of a constant resuspension rate is also inconsistent with the
resuspension formulation used in EPA's depth of scour model, which accounts for bed armoring
of cohesive sediments. Neglecting bed armoring causes the fate and transport model to
overestimate the resuspension of solids and PCBs during high flow events.
Third, EPA has assumed a constant settling velocity of 2 m/day to describe
deposition rates in the Upper Hudson River. This assumption, however, does not accurately
represent the dynamics of sediment deposition in the River. Laboratory studies on cohesive and
non-cohesive sediments show that deposition rates vary with particle size and shear stress at the
sediment-water interface, both of which change with river flow (Ziegler and Nisbet, 1994; van
Rijn, 1984; Mehta and Partheniades, 1975). EPA should include these effects in the formulations
used to simulate deposition in the solids transport model.
10
-------
Fourth, specifying deposition and resuspension rates based on judgment and
model calibration, without supporting data, can produce highly inaccurate solids fluxes between
the water column and sediment. The accuracy of the flux rates is dependent on the accuracy of
the solids loading estimates used in the solids mass balance. If tributary solids loadings are
underestimated, as we demonstrated earlier for the TIP, the model will overestimate net
resuspension in order to account for observed TSS levels.
This is illustrated by examination of the solids balance presented in the Report
(U.S. EPA, 1996b; Figure 4-35). Net erosion of 9,100 MT is calculated during the simulation
period. However, if Snook and Moses Kills have solids yields similar to the Hoosic River, as we
believe they do, the solids loading would increase by about 22,000 MT. Instead of net erosion,
the mass balance would indicate net deposition of about 13,000 MT.
The flawed deposition and resuspension rates used in the model are evident in
EPA's attempts to calibrate the model. Only the nsl result of deposition and resuspension. and
not the absolute values of deposition and resuspension, affect the water column solids balance.
Thus, in the absence of an independent assessment of deposition or resuspension, efforts at
calibration are circular: one parameter is played off against the other. For example, the Fort
Edward to Stillwater solids balance during the low flow or "Non-Event" portion of the model
calibration includes resuspension and settling fluxes that exceed the solids flux passing Stillwater
(11,000 to 15,000 MT versus 10,000 MT). Because the presumed solids loading to the River
between Fort Edward and Stillwater during this period approximately equaled the solids flux
11
-------
passing Stillwater, even eliminating deposition and resuspension would not appreciably affect
the solids calibration. Thus, the solids balance provides no basis for calibration of these
processes.
In contrast, the PCB calibration is very sensitive to the absolute values of
deposition and resuspension and does provide a basis for calibration of the solids balance. The
comparisons of model and data in Figure 4-32 of the Report show that the model overestimates
water column PCBs at low PCB concentrations. This demonstrates that the model overestimates
the transfer of PCBs to the water column and, hence, the solids fluxes between the sediment and
water column at low flow.
Sedimentation Rate
Another fundamental problem with the solids modeling framework is that the
model defines the sedimentation rate independently of the net transport of solids across the
sediment-water interface. The model assumes an external net sedimentation rate of 0.22 cm/yr,
derived solely from profession judgement (U.S. EPA, 1996b). This rate is unrelated to the
resuspension and deposition rates used in the model. Because sedimentation is tied directly to
the net transport of solids across the sediment-water interface, EPA should restructure the model
to calculate the sedimentation rate as the net difference between resuspension and deposition.
This will bring greater internal consistency to the model.
12
-------
The result of EPA's approach is that the solids transport model predicts net /—n
nc
erosion during the calibration period. Although net erosion might be plausible for a short time
period, it is inconsistent with data showing net sedimentation in the Upper Hudson River
(HydroQual. 1995a). The model's prediction of net erosion also demonstrates that EPA has
underestimated the solids loadings to the River.
B. Calibration of PCB Mass Balance
GE has a number of concerns with respect to the calibration of PCBs in EPA's
model. First, the problems with the solids balance - namely, the underestimation of tributary
loadings, the overstatement of the resuspension and deposition rates, and the decoupling of the
sedimentation rate from the other solids parameters - cause the model to overstate the transfer of
PCBs from sediments to water. Second, the initial conditions of PCBs in sediment are based on
data that do not reflect the significant loadings of sediments and PCBs from the Allen Mill in
1991-1993, thus adding substantial uncertainty to the model's estimates. Third. EPA failed to
consider or incorporate the documented occurrence of PCB dechlorination or degradation within
the sediments. Fourth, there appear to be errors in EPA's "correction" of the GE PCB data to be
consistent with EPA data. We address each of these issues in turn.
Relationship Between Solids Transport Components and PCB Fate Components
The inaccuracies in the solids transport model have significant implications to the
PCB model. They result in overestimation of the transfer of sediment PCBs to the water column
and underestimation of the reduction of surface sediment PCBs by burial. As explained in
-------
Section II.A. we believe that the EPA model underestimates the solids loading from tributaries
and overstates the resuspension and deposition rates. This, in turn, results in excess movement
of PCBs from the sediment to the water column by way of desorption through instantaneous
equilibration between resuspended particulate PCB and water column dissolved PCBs. Based on
available information and generally accepted theory, we believe that the Report's conclusion that
sediments have a dominant influence on water column PCBs is incorrect.
EPA can take the following actions to improve the accuracy of its PCB fate .
model:
Use the 1994 TSS study to refine the solids loading estimates for all of the
tributaries. Particular attention should be paid to the solids yields from Snook
Kill and Moses Kill.
Eliminate resuspension at low flows unless there is conclusive evidence that this /ue)
phenomenon occurs in the Upper Hudson River.
• Incorporate the concept of bed armoring into the description of high flow
resuspension. This is best accomplished by using the theory incorporated in the
scour model.
Calculate sedimentation as part of the soiids mass balance.
Incorporate the variation in solids load composition and deposition velocity with
river flow. 'v—'
• Calibrate and validate the sediment transport model against data for multiple
floods.
• Validate the balance between solids loading and resuspension by comparing
computed and observed water column PCBs during floods.
14
-------
Initial Sediment Conditions sllG
EPA's model relies entirely on [991 sediment sampling data to establish the initiai
composition and concentration of PCBs in sediments in 1993. The release of substantial
quantities of fresh PCBs and sediment to the River between the fall of 1991 and the spring of
1993, combined with alteration of these fresh PCBs, undoubtedly resulted in a significant change
in the composition and concentration of surface sediments in the TIP and perhaps elsewhere in
the Upper River. GE estimates that the collapse of the Allen Mill in 1991 resulted in the release
of substantial quantities of PCB-containing sediment from scouring of sediments in the Mill's
tailrace tunnel. Since this occuned during a period of low flow in the Hudson River, it is likely
that a significant portion of the sediment-bound PCB was deposited upstream of the Thompson
Island Dam ("TID"). As a result, relying on data that do not reflect this significant event creates
substantial uncertainty in the calibration of EPA's model. We recommend that the starting date
for initial sediment conditions in the model calibration be at a point in time for which sediment
data exist, such as 1991.
The impact of the changed sediment conditions may be evident in PCB
concentrations in fish in the TIP. Comparisons of the changes in PCB content in brown
bullhead, largemouth bass, and pumpkinseed in the TIP and at Stillwater following the 1991 and
1992 PCB releases shows a greater increase in the TIP, as shown in Figure 1. Given that water
column concentrations were similar at the two locations, the impact in the TIP would only be
greater if exposure concentrations in the sediments increased to a greater extent in the TIP than at
Stillwater. The higher sediment concentrations might be reflected in the increased PCB levels in
-------
brown bullhead and largemouth bass (both of which derive some of their PCBs from sediments)
caught in the TIP compared to those caught near Stillwater; and the lack of any discemabie
difference in the PCB levels in pumpkinseed (which derives most of its PCBs from the water
column) caught at these two locations. This suggests an increase in surface sediment
concentrations of PCBs in the TIP following the discharges from the Allen Mill.
Analytical Issues
An important issue identified in the Report is an apparent discrepancy between the
GE PCB data and the EPA data (U.S. EPA, 1996b; pg. 4-13). EPA believes that the GE data set
significantly underestimates the amount of congeners BZ4 and BZ10. The GE methodology,
which was based upon the EPA analytical method used in the Green Bay Mass Balance study,
utilizes a DB-1 capillary column. With this column, PCB congeners BZ4 and BZ10 coelute in
peak 5. The method used by EPA for the Hudson River, in contrast, is able to distinguish these
peaks.
Knowledge of PCB congener composition is useful in evaluating the sources of
PCBs that are responsible for the measured PCB load in the TIP water column because the
various PCB sources may have different congener compositions. Studies have shown that the
sediment PCBs have been subjected to extensive anaerobic dechlorination (Abramowicz. 1991;
Brown, et ai„ 1987} and that the sediment PCBs are enriched in congeners with fewer chlorine
atoms than the PCBs entering the pool at Rogers Island, which appear to be mainly
undechlorinated Aroclor 1242.
-------
Additionally, recent work {Fish and Principe, 1994) demonstrates that
dechlorination may occur on relatively short time scales (6-12 months) and in surface sediments,
not just deeply buried sediments. Congener BZ4 is produced during this process and high levels
have been found in sediments. Because the EPA-developed analytical technique used by GE
underestimates the amount of this congener present in the water column, the PCB load in the TIP
does not appear as dechlorinated in reports of the GE data as it does in EPA's own data. In fact,
the composition of PCBs in the TIP water in the GE data closely resembles unaltered Arocior
1242 that has partitioned from sediment panicles to the water column, whereas EPA's data
suggest that the PCBs in the TIP water derive from dechlorinated PCBs.
GE is working to understand why the EPA Green Bay method underestimates
these congeners. The 1994 Cook memorandum EPA provided to GE describes the differences in
the EPA and GE method for many Arocior standards. Based on our initial review, it appears that
the Green Bay method used by GE was based on flawed assumptions concerning the amounts of
certain congeners in the tested standards. Recent work by GE provides more accurate
assessments of which congeners are present in each Arocior standard. This information should
allow a correction to be made to the existing GE data, and preliminary recalculations of congener
levels in the various standards compared to those reported in the 1994 Cook memorandum are in
much better agreement. When this analysis is completed, we will document our findings and any
corrections applied to the data.
17
-------
It is unfortunate that EPA did not communicate to GE its knowledge of this
important discrepancy in a more timely way. For the last two years GE has continued to generate
data using the Green Bay method under agreement with EPA as part of the Remnant Deposits
Monitoring Program. If notified earlier, we could have altered our analytical method to be
consistent with that used by EPA, thus eliminating, at least in part, the need to "correct' for
different analytical methods.
Dechlorination/Biodevradation Issuer
Work performed by GE (Abramowicz. 1991; Brown et al, 1987) and others
(Quenson, et al., 1990) clearly demon' trates that naturally occurring microorganisms in the
sediments of the Upper Hudson River have extensively altered PCBs in the sediment.
Biodegradauon in the anaerobic sediments has resulted in the extensive loss of chlorine, a result
confirmed by the EPA high resolution coring data (U.S. EPA, 1995). This chlorine loss causes a
reduced bioaccumuiation potential compared to that of unaltered Arocior 1242 and also a
potential reduction in PCB toxicity. Additionally, researchers have found that in aerobic
sediment, aerobic bacteria destroy the more lightly chlorinated PCBs. This process has generally
been found to occur in aerobic surface sediments, but some researchers have also detected
metabolites of this process in subsurface sediments (anaerobic) of the Upper Hudson Riven
EPA's models neither incorporate nor acknowledge these potentially important
fate processes. To use this information, one needs an estimate of the rates for these processes.
Recent laboratory work conducted by Ken Fish of GE (Fish, 1996) demonstrates that the
18
-------
introduction of unaltered Aroclor 1242 in Hudson River sediments under expected River
conditions can result in extensive dechlorination and degradation in less than one year. Because
EPA will presumably use its model to project PCB levels decades into the future, these processes
should significantly impact PCB levels in surface sediments, even if the rates of dechlorination
and degradation seen in the laboratory are faster than those that occur in the River. As a result.
EPA's models should account for dechlorination and degradation of PCBs.
These processes may also be important in understanding the fate of PCBs entering
the river near Hudson Falls and their relationship to the PCB imbalance across the TIP. One
hypothesis for the high PCB water column loading in the TIP (see Section II.C) is that PCBs may
be entering the pool undetected and then settling into the surface sediments. According to EPA
water data, the PCB load in the TIP appears to be derived from dechlorinated sediments. It may
be that newly deposited PCBs are undergoing rapid alteration as shown by Fish's laboratory
studies.
Lastly, one of the key objectives of the reassessment is to determine what levels
of PCBs in fish are acceptable and when they will be achieved. Dechlorination affects which
congeners are available to consumers of fish and the concentration or mass of those congeners.
EPA's recently released PCB Toxicity Reassessment (U.S. EPA. 1996a) recognizes a reduction
in toxicity with lower chlorination levels. Also, the PCB congeners of primary concern in the
Hudson River, BZ77 and BZI26 (coplanar congeners), which are thought to exhibit dioxin-like
toxicity, have been found to be dechlorinated in the Upper Hudson River. Thus, dechlorination
19
-------
and biodegradation affect both the degree of exposure and the chemical to which people are
exposed. Because these processes have been demonstrated to be occurring in the Upper Hudson.
EPA should attempt to incorporate them into its modeling effort.
C. PCB Mass Imbalance in the Thompson Island Fool
As we pointed out in the Introduction, one of the valuable aspects of modeling is
the identification of areas requiring further data collection or refinement of judgment. Balancing
the PCB mass across the TIP is such a case. It is apparent that EPA has been unable to achieve a
PCB mass balance across the TIP without resort to mechar'sms that can be hypothesized but not
factually demonstrated. GE has met the same obstacle in its mo>ling effort (HydroQual.
1995b). In EPA's model, PCBs are introduced into the TIP water column from four sources: 1)
PCBs are diffused from sediment-pore water: 2) PCBs desorb from sediments that are
resuspended: 3) PCBs enter the pool from upstream; and 4) pore water containing PCBs is driven
into the water column by groundwater discharge. Diffusion from pore water is not in dispute.
GE questions, to one degree or another, the values that EPA has used for other PCB transport
mechanisms, and as discussed in Section II.A, we believe that the resuspension rate EPA used is
too high. While we are in substantial agreement with EPA on the values used for upstream
loadings, one must question whether those measurements have been able to capture the total
loading source(s) because GE's upstream monitoring station may not be detecting discharges of
dense oil. Finally, although we agree with EPA that groundwater discharge is a plausible
hypothesis, the Agency must recognize both that it is only one of several hypotheses and that it
20
-------
has not been verified by field data. Further factual development and data collection is necessary
before one can select the most likely hypothesis with confidence.
Consequently. GE is undertaking investigations to test all reasonable hypotheses
for the PCB mass imbalance in the TIP. The PCB load in the TIP dominates PCBs in the water
column downstream of the TIP. Knowledge of the source of the PCB load in the TIP is thus
essential to determining whether and how the PCB load can be controlled. Regardless of GE's
success in testing the hypotheses. EPA must be able to select a factually supported hypothesis.
Otherwise EPA will lack a sound factual basis for its remedial decision, and any selected course
of action will be arbitrary, with a very real probability that it will not have its intended
consequences.
The hypotheses that must be tested in order to determine the cause of the mass
imbalance of PCBs in the TIP include:
1. The mass and concentration of PCBs entering the TIP are markedly
greater than the mass and concentration measured at the upstream Rogers Island monitoring
station. This hypothesis - that PCBs escape detection at Rogers Island - is plausible because the
known releases of PCB oil from the Hudson Falls plant area are denser than water, and the
current monitoring program was not designed to detect or quantify dense oil phase PCBs. To the
extent that PCB oil escapes detection at Rogers Island, either because it travels as part of an
21
-------
unquantified bed load or as undetected pulse loadings, the present monitoring would
underestimate PCB transport past Rogers Island. GE has conducted a number of monitoring
studies this fall to address the issue of the representativeness of the current monitoring program.
For these and the other hypothesis testing activity described in this section. GE will share the
data and its analysis with EPA as the results become available.
2. The mass and concentration of PCBs passing the TID are markedly less
than the mass and concentration measured at the TID monitoring station. Since we do not
understand the mechanism by which excess PCBs enter the TIP, we can not be sure that the PCB
monitoring conducted from the single point at the TID accurately represents PCB transport over
the dam. Monitoring studies conducted this fall, including a longitudinal transect study
following a single mass of water through the TIP, and water column monitoring conducted
across the river near the TID, should provide insights into the representativeness of the TID
monitoring station.
3. Groundwater inflow within the TIP is causing substantial release of
PCBs from the buried sediments into the water column in the TIP. This hypothesis appears
plausible because it can account for a portion of the excess PCB loading observed across the
pool. The hypothesis has several weaknesses, however. First, as tested in the EPA model
calibration, ground water advection is a spatially limited mechanism; it is invoked only in the
TIP. Given the similarities between the TIP and downstream reaches, if groundwater were
important to PCB releases in the TIP, one would expect it to contribute as well to PCB transport
22
-------
in downstream reaches. Sediment diffusive flux alone, however, can account for PCB loading
observed downstream of the TIP.
Second, the groundwater flux hypothesis cannot account for the temporal
variability in the excess PCB loading from the TIP. The excess loading only appeared after the
substantial discharges from the Allen Mill between 1991 and 1993. If groundwater advective
flux is a significant contributor to PCB loadings in the TIP, then water column monitoring
conducted prior to the event should contain some evidence of its existence. GE plans to evaluate
the groundwater flux hypothesis by evaluating the groundwater system near the TIP and possibly
deploying groundwater seepage meters within the TIP sediments.
4. There are markedly greater PCB concentrations in the surface
sediments of the TIP (resulting from the 1991-1993 Allen Mill discharges) than reflected in
the surface sediment data used in the model This hypothesis is plausible as it accounts for the
coincidence in timing with the Allen Mill discharges. Two of the EPA high'resolution cores
collected from the TIP in 1992 (after the initial Allen Mill discharges) show an increase in PCB
concentrations near the sediment-water interface. These surficial sediment layers reflect recent
PCB deposits. As stated in Section II.B, comparison of PCB content in brown bullhead,
largemouth bass, and pumpkinseed in the TIP and at Stillwater before and after the 1991-93
releases shows a greater increase in the TIP than at Stillwater (see Figure 1). The brown bullhead
23
-------
and largemouth bass derive more of their diet from the sediment than does the pumpkinseed.
This suggests an increase in surface sediment concentrations of PCBs in the TIP following the
discharges from the Allen Mill.
Higher surface sediment concentrations could increase the driving force for
diffusive transport from the sediment to the water column and may account for some of the
excess PCB observed in the River. This mechanism alone, however, cannot account for all of the
excess PCB because the PCB congener pattern of excess loading reflects some degree of
dechlorination. Therefore, a post-depositional process of PCB dechlorination is required to
produce the PCB congener pattern of the loading. Although recently conducted laboratory
experiments (Fish, 1996) indicate that dechlorination can occur at a rate sufficient to contribute
to the excess loading as described above, extrapolation of these results to the field is not yet
complete. These studies suggest, however, that the combined process of deposition of PCBs into
surficial sediments followed by dechlorination is a possible cause for the excess PCB loading
from the TIP.
5. A substantial mass of PCBs enters the TIP between Rogers Island and
the TID from sources outside the Pool, such as dredge spoil sites. While this hypothesis is
possible, it is unlikely because there is no physical explanation why the external load from these
sources would be correlated in time with the discharges from the Allen Mill. The spatial pattern
24
-------
of PCB loading within the pool as discerned from the longitudinal transect studies described
above should provide further insight into whether the dredge spoil sites might contribute to the
TIP PCB loading.
6. Resuspension of surface sediments introduces a substantial mass of
PCB into the water column of the TIP. The EPA model invokes sediment resuspension during
low flow periods to account for a portion of the excess PCB loading observed in the TIP. While
this hypothesis may account for the temporal correspondence between the Allen Mill discharges
and the excess PCBs emanating from the pool (the resuspended surface sediments may contain
elevated PCBs originating from the Mill), it is counter to the generally accepted understanding of
sediment transport processes. Under current theory and experience, surface sediments are not
appreciably resuspended until river flow velocities produce 3 critical shear stress at the sediment-
water interface. This critical shear stress occurs at river flow velocities substantially higher than
those that occurred during the periods of excess loading. The marginal increase in TSS observed
during the excess PCB loading periods can be accounted for by tributary loadings from the
Snook Kill and Moses Kill. Indeed, the longitudinal transect studies conducted this fall
substantiate this conclusion because TSS did not increase in the TIP until downstream of these
tributaries.
0. Long-Term PCB Mass Balance
The current EPA model does not calibrate to the existing data for 1993. matching
neither the low flow PCB data nor the PCB load at the TID. We have discussed the potential
25
-------
reasons for this in prior sections. Indeed, the limited calibration period is insufficient to test the
model's ability to represent the long-term fate of sediment PCBs. in particular the impacts of
resuspension. sedimentation, dechlorination and biodegradation. Based on GE's own modeling
and data evaluations, this problem will become more significant when EPA compares the model
results with the vast array of historic PCB data in fish, water, and sediment (EPA refers to this as
"hindcasting").
Two critical tests of the match between the historic data and the results of the
current EPA model are likely to show unacceptable incongruity between the data and the model.
The first test of the model is its congruence with the amount of PCBs present in the water
column at Schuvlerville in the mid- to late-1980s. If the EPA model were used, incorporating
EPA's assumed low flow resuspension and groundwater inflow to move PCB from the sediment
into the water column, one would expect approximately 100 ppt of PCB in the water column at
Schuvlerville during low flow in the summer months. In fact, the average value measured by the
U.S. Geological Survey in 1988 ana 1989 is 30 ppt. This is close to the amount one would
estimate from diffusion alone.
The second test of the EPA model is its congruence with the inventory of PCB
mass in the sediments over time. If significant amounts of PCB were moved into the water
column from the sediments, as the EPA model assumes, there should be a discernible change in
the PCB reservoir in the sediments or the depth profile of PCBs in the sediment. Three large
scale sediment sampling events were carried out in the Upper Hudson in 1977,1984. and 1991
26
-------
which can be used to estimate the mass of PCBs in the sediments of the TIP. EPA has estimated
that the TIP sediments contained about 32.000 pounds of PCBs in 1984 (U.S. EPA. 1996b. pg. 6-
2). The EPA model estimates that 800 pounds of PCBs are lost from the sediments into the
water column over the 270 days in 1993. If the same rate of loss were projected into the past
{presumably an underestimate because PC B levels in the sediment would have been higher
earlier), the total mass of PCB mass lost from 1984 to 1991 would be 8,000 pounds, or
approximately 25% of the total inventory. This is inconsistent with the historic sediment data
which show little change in PCB inventory over fifteen years.
Another way to compare the model results to the historic levels of PCBs in the
sediment is to compare PCB depth profiles over time. Unfortunately, the current EPA model has
decoupled sediment transport and net sedimentation rates (see Section II.A), making such a
comparison difficult. EPA should link the sedimentation rate and the solids mass balance
portions of the model. If the PCBs were being flushed out of the sediment, as the EPA model
predicts, then the loss of bioavailable PCBs due to long term burial would be less significant than
if less interaction between the sediment water column were occurring. Comparing the PCB
profiles in the sediment over time to the model predictions should provide a check on the model
assumptions.
Long term burial as the dominant mechanism for loss of bioavailable PCBs is
consistent with the available data reviewed to date. As an illustration, consider the vertical
profiles of PCB concentrations in TIP sediment shown on Figure 2. Each panel shows measured
-------
sediment PCB concentrations, averaged over selected depth intervals, for the years 1977, 1984
and 1991. In the panel on the left, the PCB concentrations for each depth interval are plotted
against the depth of the slice. The surface to 5 cm slice is plotted at a depth of 5 cm.
corresponding to the depth at which material would be buried below a mixed "active layer" 5 cm
deep. The other concentrations are plotted at the depth below the sediment-water interface of the
mid-point of the associated sediment core slice. When plotted in this manner, the only obvious
pattern in the data is that the surface concentrations decreased from 1977 to 1984 to 1991, and
the deeper concentrations (50 to 70 cm) are relatively higher than the surface sediment
concentrations.
The middle and right panels of Figure 2 show these same sediment PCB profiles,
but with the measured profiles modified in the following manner. First, the surficial (0-5 cm)
concentrations are reduced by 20%, corresponding to an upper bound limit to the potential
degradation of PCBs associated with anaerobic dechlorination processes. (This decrease is only
applied to the mixed surface layer, since this is the only layer likely to have relatively recently
deposited, undechiorinated PCBs). Second, the 1977 and the 1984 PCB profiles are shifted
downward to correspond to the depth of burial that would occur from the time of sampling (1977
or 1984) to 1991. This vertical translation corresponds to a long term average net sedimentation
rate of 0.5 cm/year on the middle panel and 1.0 cm/year on the right hand panel. With these
modifications applied to the data, the blurred image described by the untranslated profiles on the
left panel comes into focus and a relatively distinct, unified profile emerges. This is consistent
with the conclusion that buriai has simply moved the PCB profiles measured in 1977 and 1984
-------
down in the sediment, modified by some degree of dechlorination, possibly in combination with
other loss processes, in the surface layer. The actual burial rate is difficult to discern with
precision, but this analysis suggests that it is on the order of 0.5 to 1 cm/year. Since the PCB
data are reach average results, this burial rate is indicative of a reach average long term net
sedimentation rate. These data do not support EPA's hypothesis that PCBs are being "flushed"
from the sediment at the high rate calculated by the EPA model.
• * ~
Based on GE's evaluation of the EPA model, it is clear that the model will need
significant revisions to calibrate properly to the historic data. The current EPA model greatly
overestimates the amount of PCBs contributed to the water column by the old sediments. If left
unchanged, future projections will show exaggerated benefits from sediment remediation
projects if the projection focuses only on the present loss rate from buried sediments. If the
predicted loss rate were continued into the future, the sediments would be largely depleted of
PCBs in the near future, which implies that remediation is unlikely to shorten significantly the
period required to achieve acceptable risk levels.
To complete the modeling effort and to answer the key questions for the
Reassessment, the source of the TIP load imbalance must be understood. The Reassessment
29
-------
cannot be completed in a defensible way until this is done. This wiil be a major focus of GE's
efforts in 1997, and GE looks forward to working closely with EPA to unravel this important
technical issue.
The Bivariate Statistical Model ("BSM"), Probabilistic Food Chain Model ("PFCM"), and the
Gobas Model ("GM"). Each has its strengths and weaknesses. The BSM and PFCM are
essentially steady-state statistically-derived models that rely on examinations of historic PCB
levels in sediment, water column and biota to ascertain the relationship among these natural
compartments. Because they ignore the short and long term variability in the relationships
among PCB levels in water column, sediment, and fish and do not attempt to discern the
mechanisms by which fish bioaccumulate PCBs, these models will have limited utility for
predicting PCB levels in fish in the future. The GM, on the other hand, is a time-variable,
mechanistic food web model, which explicitly incorporates variability in exposure, uptake and
depuration of PCBs and which, because it reflects real world bioenergetic and toxicokinetic
mechanisms, provides a useful and easily checked predictive tool. For these reasons, GE urges
EPA to develop and use a time-variable mechanistic food web model, such as the GM, using the
statistical relationships developed through the BSM and PFCM, as well as the performance of the
model in other systems, as external checks on its dynamics and output.
III. RTOACCUMULATION MODELS
EPA is developing three bioaccumulation models to predict PCB levels in fish:
30
-------
A. The BSM and PFCM Models
Stream Qfth? Hotels
The BSM and PFCM are models that attempt to derive quantitative relationships
between PCB levels in sediment, water, and fish through statistical analysis of Hudson River
data. The BSM examines these data using multiple regression analysis to estimate average PCB
levels in fish from PCB concentrations in sediment and water. The PFCM is similar, except that
it involves characterizing the historic data and food web transfers within the River to estimate
trophic transfer factors ("TTFs") between each link. Using Monte Carlo techniques, the model is
intended to be used to estimate the mean and variation in PCB levels in top predators, given
average exposure levels in sediment and water. Regardless, like the BSM, it is a statistically-
based model.
The statistical, steady-state nature of these models limit their utility as predictive
tools. Moreover, the inconsistency in their use of sediment PCB data undermines their validity.
First, both models assume that PCB levels in the biota are near steady-state with regard to levels
in the sediment and water. PCB levels in biota in the Upper Hudson River, however, have
exhibited relatively slow long-term declines, as well as significant short-term changes. Lipid
content of some species has also changed dramatically over time, exhibiting both long-term
trends and year-to-year variation. Because PCBs tend to accumulate in lipid, the variability in
lipid content results in changes in excretion rates, which, in turn, cause variation in PCB body
burden on the scale of one to a few years. Ignoring these temporal changes in the system lends
significant uncertainty to the validity of these models.
-------
Second, both models ignore the causal mechanisms by which fish bioaccumulate
PCBs. The primary advantage of the regression approach used in these models is that it is
relatively simple. Only the site-specific data are required; no ancillary information on
bioaccumulation processes is needed. This simplicity is also their primary disadvantage. While
the PFCM, for example, explicitly incorporates variability in PCB levels in sediment, water, and
biota to derive the TTFs, the model does not allow one to discern the cause for that variability.
By failing to incorporate available information about the biological mechanisms for the variation
in contaminant levels in fish, such as growth rate, size, lipid content, feeding behavior, and
exposure levels, these regression models do not allow one to assess, for example, how PCB body
burdens will change over time as the relative importance of sediment and water column PCB
levels change.
Third, another flaw in the PFCM is that it requires having the answer to solve the
problem. The observed variability in a fish population is input into the model in the form of the
variability in the TTF, and the model then calculates the variability in the fish population. The
model is circular.
Fourth, the model is improperly structured. It attempts to calculate variability
(population variance) from uncertainty (standard errors of the mean TTFs). These metrics are
incompatible, and the model results have no physical meaning.
32
-------
Fifth, these models, as applied by EPA, incorrectly assume that surface sediment
concentrations of PCBs have remained constant over time. In addition, the models rely on
different and inconsistent sediment data: both models rely on a single year's sediment data - for
the BSM. the 1991 GE data: for the PFCM. the 1993 EPA data. The inconsistency in data raises
questions about the comparability of results from the two models. The assumption that temporal
changes in sediment PCB concentration will not affect PCB levels in fish is flawed in light of the
generally accepted understanding that surface sediment PCBs can contribute significantly to PCB
levels in certain fish species. (The assumption is odd in a reassessment focused on whether or
how remediation of sediments might reduce PCB concentrations in fish). Yet, EPA used a single
set of sediment data to compare to all the historical data on PCB levels in water column and fish,
thus compromising the validity of the statistical analysis. The appropriate approach is to couple
the bioaccumulation modeling to the water and sediment exposure concentrations computed by
the PCB fate model.
For the PFCM. EPA estimated BSAF values for benthic invertebrates using a
limited number of cores and data from several species of invertebrates. Each species of
invertebrate may feed in a different manner, leading to differences in exposure concentration.
The distribution of invertebrates sampled may differ from the distribution of invertebrates
actually consumed by forage fish. In addition, EPA estimates the trophic transfer from water
column particulates to water column invertebrates using the historical muitiplate and caddisfly
data. EPA assumed that the fine fraction of material on the multiplates represents water column
particulates, but this material may not represent particulates that caddisflies foraging just above
-------
the sediment bed consume. Finally. EPA failed to consider the uncertainty associated with its
assumption that the caddisflies are representative of the water column invertebrates in the diet of
the forage fish.
Validation of the Models
EPA has not adequately validated the models:
(1) The Report presents temporal trends for pumpkinseed, brown bullhead and
largemouth bass at river mile 175 but not for the TIP. where the BSM and data do not match as
well.
(2) The Report compares BAF values estimated by the BSM with values calculated
using the GM at other sites. While it is appropriate to compare values with other sites, EPA
should make these comparisons against data from field populations of similar species in similar
ecological settings, rather than values calculated using another model for another system.
Moreover, because both the sediments and the water column are important sources of PCBs to
the fish, EPA should compare both BAF and BSAF values with other systems, not just the BAF.
(3) The analyses used by EPA to test the BSM demonstrate that it will not be
useful for predicting fish PCB levels. The Report gives values of the coefficients of
determination of the regressions (r). The r value is a measure of the proportion of variation in
the data that is accounted for by the regression. In addition, the Report presents scatter plots of
-------
observed versus predicted PCB levels. The Report does not present any other statistical analyses
to evaluate goodness of fit. In addition, the scatter plots show problems with the predictive
ability of the models:
• The slopes of these values are often less than one. indicating that there is a bias in
their predictions.
• Predicted values are often higher than observed at low concentrations. (See
Figures 9-8 to 9-13). This high bias will be critical when making predictions of
future PCB levels in fish, when exposure levels should be reduced. Thus, the
models' projections will be biased high.
• The pattern of observed versus predicted values differs among reaches. For
example, within the TIP, predicted values vary little temporally while observed
values have considerable variability. This suggests that the models do not provide
accurate predictions of PCB levels within each pool.
B. The Goha* Model
A time-variable bioenergetics-based food web model such as the GM overcomes
the major failings of the BSM and PFCM. The GM can represent key time-dependent features of
the historical data: short term exposure changes and variations in lipid content and changes in
sediment and water column concentrations over time. This results in a calibration that more
accurately represents the relationships among PCBs in water, sediment and biota. Also, the
validity of the model's coefficients can be evaluated because they have biological meaning.
Finally, because the model is mechanistic, causes for differences between model and data can be
explored, leading to further field measurements or experimental work to improve the model
parameterization and predictive capability.
35
-------
In the Report. EPA suggests that poor knowledge of many of the important model
parameters hampers the GM approach (U.S. EPA. 1996b; pg.8-13). Actually a significant
amount of information exists to provide adequate constraints on the model. All of the
informational requirements for development of the GM can be met. Food web structure and fish
natural history have been evaluated in the PFCM. The uptake and depuration rates of PCBs can
be estimated from field and laboratory data and other modeling studies. Uncertainty and
variability associated with these parameters can be used to generate predicted uncertainty and
variability in fish PCBs. As a result, lack of data is not a valid reason not to develop the GM.
C. Recommended Approach
We recommend that EPA use a time-variable bioenergetics-based food web
model, such as the GM, to estimate mean PCB levels, and that this model be coupled to the PCB
fate and transport model to integrate the predictions of exposure and bioaccumulation. GE is
developing such a model and believes that it would provide a solid basis for work in this area
We recommend that this model be further developed in a cooperative effort that takes advantage
of the knowledge and expertise of both GE and EPA. Further, we recommend that EPA use the
observed coefficients of variation in conjunction with the calculated mean PCB levels to describe
the distribution of PCB levels for use in the risk assessment.
IV. DEPTH OF SCOUR MODFT
GE is in basic agreement with the approach EPA used to model the 100 year
flood. The model is consistent in its characterization of the resuspension properties of cohesive
36
-------
sediments and appropriately incorporates field data from the TIP. Nevertheless. EPA should
correct several errors in the model.
First, the EPA model uses an improper method to calculate bottom shear stress
distribution in the TIP: outputting predicted current velocities from the RMA-2V hydrodvnamic
(ub)
model and using an external formula (Equation 5-3) to calculate shear stresses. This approach is v
inconsistent because Equation 5-3 predicts different bottom shear stresses than does from RMA-
2V. The correct approach is to output the bottom shear stresses calculated by RMA-2V and use
those values to calculate cohesive sediment resuspension.
Second. EPA did not properly use the Lick equation for calculating resuspension
potential. The Lick equation calculates resuspension potential (e), which is expressed as eroded
sediment mass per unit area; grams/cm2. Converting predicted resuspension potential at a
particular location to a scour depth (ZKOUf) requires application of Equation 3-7; Z,couf = (e) /
C^|„. In this equation. C9u,k is the dry bulk density (grams/cm3), where
Cbultdry = 0 * P) Pied
and P = porosity and = sediment particle density (= 2.65 grams/cm3). The dry bulk density
represents the dry sediment mass per unit bed volume and is also referred to as the bed solids
concentration. Examination of EPA bulk density and solids data suggests that the EPA database
lists Cbult w not dry bulk density. It is inappropriate to use wet bulk density to
calculate Z^:
Qwltwei = P Pv» 0 * P) Pjed
37
-------
where pw = water density (= 1 gram/cm3). The wet bulk density is the mass of water and
sediment per unit bed volume. If EPA did use wet bulk density, it should convert those wet bulk
density values to dry bulk density as follows:
Cfetjlk.dry ~~ Psed (^bulk^wet ~ Pw )/(P«d" Pw)
Third, as previously discussed, an inconsistency exists between the treatment of
cohesive sediment resuspension in EPA's solids transport model and in the depth of scour model.
The depth of scour model uses a formulation, the Lick equation, that accounts for bed armoring ^130
effects and also utilizes Upper Hudson River resuspension data to determine site-specific
parameter values. Eliminating the inconsistency between the two models and incorporating the
Lick equation into the solids transport model is necessary to achieve credible solids transport
simulations using EPA's solids transport model.
Finally, a major uncertainty remains for the depth of scour model: how will EPA
simulate non-cohesive resuspension? Research on suspended load transport of non-cohesive
sediments has a long history, and researchers have proposed a wade variety of formulas and
methods. The various formulas have been tested on a number of different data sets, from
laboratory and field studies, and found to produce accurate results under a wide range of
conditions. Thus, EPA must carefully screen these formulations to find one that is appropriate
for the TIP. GE requests that it be informed of EPA's choice in sufficient time to comment
before EPA develops its final product.
13E ;•
38
-------
One formulation that has produced reliable results is the van Rijn model for
suspended load transport of sands (van Rijn, 1984). This model would be appropriate for the TIP
with some modifications. The van Rijn model was developed for an ungraded bed: i.e..
relatively uniform-size sand particles. The non-cohesive bed in the TIP consists of a wide range
of panicle sizes, with a significant fraction of coarse sand and gravel. Under these conditions.
EPA needs to consider the effect of bed armoring due lo heterogeneous bed composition and
modify the van Rijn equations appropriately. Extensive research on non-cohesive bed armoring
has resulted in the development of credible formulations which have been successfully applied in
modeling studies on other rivers (Ziegier and Nisbet. 1994).
Realistically simulating non-cohesive suspended load transport in the TIP depends
not oniy on the model formulations, but also on specification of bed property parameters and the
flood hydrograph. EPA must determine the distribution of non-cohesive bed parameters (e.g..
grain size distribution) in the TIP with extreme care because model results are very sensitive to
input parameters. EPA also needs to include the effect of the flood hydrograph on non-cohesive
erosion in the 100 year flood simulation. The steady-state flow assumption presently used is
valid for approximating cohesive resuspension. Non-cohesive erosion, even with bed armoring,
is rate-dependent, however, and total scour in the non-cohesive bed will depend on the flood
hydrograph; a steady-state assumption will not yield credible results.
EPA will need to calibrate the depth of scour model after adding the non-cohesive
component because of the uncertainty in the non-cohesive bed property parameters. EPA
-------
collected an excellent TSS data set during April 1994 and should use it to demonstrate that the
non-cohesive component of the depth of scour model is functioning with reasonable accuracy.
Without some form of model calibration, predictions of non-cohesive sediment erosion during
high flow events will be very uncertain and can not be used with confidence in evaluating the
effects of the 100 year flood.
complex relationships of the natural elements operating in the Upper Hudson River aaa on that
basis to predict the effect of possible future action and no action. EPa. has calibrated its fate and
transport model to a temporally limited data set collected between January 1 and September 30.
1993. There is a very extensive array of data for some parameters for ten years or more prior to
1993 and for the period since September 30, 1993. This presents a crucial and significant test of
the predictive power of the EPA model. If the EPA model is able to provide a close fit to the
data points before and after 1993. the arguments for its use as a predictive tool will be very
powerful. If it is not possible to validate the model by a close fit to the data, its lack of
usefulness for predictive purposes will be apparent. If this is the case, it should follow that
further significant analysis, data collection, and model development will be necessary before it
can be used by decision makers in the reassessment.
V. PRF.DTCTTVF POWF.R OF THF. MODEL
The purpose of developing models in the reassessment is to understand the
40
-------
In determining whether the model meets an acceptable standard for predictive use.
the model's ability to match the data closely in the following instances will be the acid test of
success:
1. To validate the components of the solids balance, EPA should compare model
results with data on:
a. Spatial patterns of TSS during low flow periods, which will evaluate the
balance between low flow solids loading and deposition:
b. Temporal and spatial patterns of TSS and water column PCBs during
flood events, which will evaluate the balance between high flow solids and
resuspension: and
c. Annual average solids loading passing Schuylerville. Stillwater and
Waterford, which will evaluate the balance between solids loading and
sedimentation.
2. To validate the flux of PCBs from pore water and PCB loss by volatilization. EPA
should compare model results with data on:
a. Spatial patterns of water column PCBs during low flow periods; and
b. Spatial changes in water column PCB composition I based on the five
congeners modeled).
3. To validate the mechanisms by which bioavailable PCBs are lost from the
sediments, particularly the balance between losses to water column (diffusion and resuspension)
and losses by burial, EPA should compare model results with data on:
a. Long-term changes in surface sediment PCB levels;
b. Long-term changes in the vertical profiles of PCBs in sediments; and
41
-------
c. Long-term changes in PCB inventory in the sediments.
4. To provide an overall test of the model. EPA should compare model results with
data on the annual average flux of PCBs passing Schuylerville. Stillwater and Waterford.
5. To test the model's ability to describe the impact of recent PCB releases from
Hudson Falls, EPA should compare model results with data on the apparent increase in the PCB
flux from Fort Edward to Thompson Island Dam/Schuylerville that occurred between the mid- to
late-1980s and the 1990s.
6. Finally, to evaluate the food web structure and toxicokinetics of the
bioaccumulation model, EPA should compare model results with data on:
a. Temporal changes in PCB concentrations in a predatory fish (largemouth
bass) and in a forage fish (pumpkinseed) at the TIP and Stillwater over a
15-year period. In addition to overall fit between model and data, patterns
in the quality of fit should be explored (e.g.. differences among species,
locations, time periods); and
b. Response of the fish to the short-term changes in water column PCB levels
in the early 1990s, which will evaluate the relative contributions of water
column and sediment PCBs and the uptake and loss ratio.
In assessing the capability of the model to match data, care must be taken to
uncover any apparent biases. The current calibration exhibits several biases that are not
discussed in the Report. For example, water column PCB concentrations are generally over-
predicted at low flow and under-predicted at high flow. The over-prediction is likely due to the
42
-------
inclusion of resuspension at low flow. The cause of the under-prediction at high flow is
uncertain. It may be a real bias, or simply a slight mistiming. EPA should examine this issue as
pan of its recalibration efforts. The model also underestimates the solids loading passing
Stillwater and Waterford during the non-event period. For the calibration period, the model is
low by about 5,000 MT at Stillwater (U.S. EPA 1996b; Figure 4-11) and 30.000 MT at
Waterford (U.S. EPA 1996b; Figure 4-12). These differences between model and data suggest
an underestimation of low flow tributary solids loading. The Report presents the hypothesis that
construction activities at Lock 1 temporarily increased solids loading, but gives no data
supporting this hypothesis. Given the importance of the solids balance to long-term predictions,
EPA needs to examine the possibility of and reasons for model bias. Finally, as discussed in
Section III. fish PCB concentrations are overestimated at low levels. As with the other biases,
cause must be determined, and the bias eliminated before EPA uses the model as a predictive
tool.
VI. f OWFR HUDSON PCB TRANSPORT AND FATF. MODF.I. 0
The cover letter to these comments addressed the question of what consideration
of conditions in the Lower River is appropriate in a reassessment directed to what action, if any,
should be taken with PCB-contaminated sediments in the Upper Hudson. As we noted, it is
important to ensure that any remedial action in the Upper River has no adverse impact, or. at
most, an acceptable adverse impact on the Lower River. It is not appropriate, however, to justify
remedial action in the Upper River on the basis of benefits to the Lower River. If EPA is to
consider benefits to the Lower River, it must examine remedial action in the Lower River, such
43
-------
as source control, to assure that a remedy designed to achieve Lower River benefits is cost-
effective. If the Agency follows the course of seeking benefits for the Lower River, it must also
identify Lower River dischargers as PRPs. Thus, assuming EPA maintains the present limited
focus of the reassessment, it must limit the examination of impacts on the Lower River and its
fish to assuring that a remedy in the Upper River will not have an unacceptable adverse impact
on the Lower River.
With regard to the Thomann model. GE does not believe that it indicates that a
remedy in the Upper River would adversely impact the Lower River and therefore does not
object to its use for that purpose. From the point of view of technical accuracy and soundness in
model development, there are a number of comments that could be made. We recognize,
however, that Thomann and Farley are in the midst of a thorough review of the model and
believe it appropriate to wait for the conclusion of that review to determine whether there are any
disputed aspects of the model that are of relevance to use for the limited purposes appropriate for
this reassessment. We believe that EPA should also wait for completion of the revision of the
Thomann model before using it in this reassessment.
EPA should also consider other analyses in addressing whether a remedy in the
Upper River would have an unacceptable adverse impact in the Lower River. In particular,
Chilrud (1996) has completed a quite different sort of analysis of PCBs in striped bass in the
Lower Hudson. EPA should review Chilrud's work in the same context in which it has examined
Thomann's model.
-------
Finally, the Report states that EPA intends to extrapolate from stnped bass
modeling results (derived from the Thomann model) to estimate impacts on the short-nose
sturgeon (U.S. EPA 1996, p. 8-3). It is not clear how EPA would carry out this extrapolation,
and GE requests the opportunity to comment on whatever method EPA may propose. We are
unaware of any data showing PCB levels in short-nose sturgeon in the Hudson, or data or
analyses relating Upper River sources to PCB body burdens in Hudson River short-nose
sturgeon. We are unaware of any data or analyses showing adverse impacts in the short-nose
sturgeon population as a result of PCB exposure and uptake. Finally, we are unaware of what
sort of adverse impact as a result of what sort of PCB exposure is claimed to occur in the short-
nose sturgeon population. In fact, recent reports suggest that the short-nose sturgeon is now
present in such abundance in the Hudson that its continued status as an endangered species may
soon come to an end. All of these relationships need to be explicated before any defensible
relationship between PCB-contaminated sediments in the Upper River and adverse impacts on
the short-nose sturgeon population can be propounded.
sediments of the Upper Hudson River, EPA appropriately determined that it had to develop
objective, quantitative tools to predict the future levels of PCBs in fish, water and sediment under
various remedial scenarios, including no-action. EPA has made substantial progress in
developing a physically- based mechanistic model that will allow it to make such predictions.
This is a complex and resource intensive project, but a necessary one.
VII. CONCLUSIONS AND RFCOMMF.NDATIONS
To determine what action, if any, is necessary to address the PCB levels in the
45
-------
GE commends EPA for seeking input on the preliminary modeling effort before
completing construction of its models or attempting to utilize its models to make predictions.
Based on our review, we agree with much of EPA's approach, including;
1) The overall goals of the reassessmerft and the role of the model in meeting
these goals;
2) The guiding principles for model development; and
3) The general criteria by which to judge the reliability of a model.
The current iteration of EPA's models, however, will not provide a reasonable
representation of PCB fate in the Upper Hudson River and cannot be used to address the key
reassessment objectives. EPA must make modifications to the structure and calibration of its
models. Additionally, the model effort has highlighted gaps in data that add significant
uncertainty to the models. These need to be addressed before EPA can complete its modeling
effort.
The two primary short-comings of the models are: i) the assumed interaction
between PCBs in the sediment and the water column, and 2) lack of knowledge concerning the
source of a large portion of the PCBs entering the water column within the TIP. The first
problem is largely a result of only using data from a 270 day period in 1993 to calibrate the
model. The current model configuration results in a significant movement of PCBs from the
sediment to the water column that is not consistent with current knowledge of processes affecting
PCBs or with the historical data. Recalibration of the model to the large historic database will
46
-------
provide further constraints in determining the long term interaction between the sediments and
water column, and will show the impact of sediment burial as a PCB loss mechanism.
The second major problem is that EPA assumes that PCB loading to the water
column in the TIP occurs by processes that either are not physically reasonable or are entirely
speculative. The assumed high resuspension and deposition rates, particularly at low flows, are
not supported by current sediment transport theory. This results in an unreasonably large amount
of sediment bed/water column interaction and an exaggerated flux of PCBs from the sediments to
the water. In addition, EPA makes an untested assumption that 30 cfs of groundwater moves
through the sediments in the TIP forcing PCBs from the deeply buried PCBs into the water
column.
EPA acknowledges that the source of the PCB load imbalance in the TIP is not
known. This load imbalance is a significant portion of the PCB water column load in the Upper
Hudson River, and the source of this load must be determined if predictions of future conditions
are to have any validity.
Ultimately, the PCB levels in fish are of concern. EPA's work to date has focused
on the statistical correlation between PCB levels in fish, water and sediment, but EPA's models
do not describe the physical basis for this relationship. GE encourages EPA to develop a time-
variable mechanistic food web model because of its far greater explanatory power and its
47
-------
improved utility as a predictive tool compared to the statistical correlations EPA has prepared so
far.
Before EPA uses the Thomann model, it must clearly define the objectives of this
modeling effect and await the revisions that Thomann and Farley are preparing.
Based on these concerns, GE recommends that EPA take the following actions to
complete its modeling efforts:
1. Develop and test by data collection, if necessary, the range of hypotheses
that might explain the TIP load imbalance;
2. Refine the solids loading estimates for all tributaries, particularly for the
Snook and Moses Kills;
3. Eliminate low flow resuspension and incorporate the concept of bed
armoring into high flow resuspension. Also, incorporate solids
composition into depositional velocity estimates;
4. Calculate, do not arbitrarily specify, net sedimentation using the difference
between resuspension and deposition in the models solids balance;
5. Recalibrate the model using the full historical data set and the revised
processes specified above. Specifically utilize:
Historic water column PCB levels;
• High flow TSS and PCB values for multiple events (floods);
• 1994 TSS data for tributaries;
• PCB levels in sediment (total inventory as well as depth profiles)
from the 1977, 1984 and 1991 surveys;
48
-------
Develop and calibrate a time-variable bioenergetics-based food web model
for fish in the Upper Hudson River:
Test the model by comparison to known data points to assure that each of
its central elements is able to replicate the natural system the model is
designed to imitate.
Clearly define the use of the Lower River model, and if appropriate, apply
the revised model being developed; and.
Before using the revised models for predictive purposes, reissue the
calibration report for additional comments.
49
-------
LIST OF REFERENCES
Abramowicz. D.A. and M.J. Brennan. 1991. In: Biological Remediation of Contaminated
Sediments with Special Emphasis on the Great Lakes. "Aerobic and anaerobic
biodegradation of endogenous PCBs." C.T. Jafvert and J.E. Rodgers (eds), EPA/600/9-
91/001, pp. 78-86.
Arathurai, R., and R.B. Krone, 1976. "Finite Element Model for Cohesive Sediment Transport."
J. Hydr.Div., ASCE. 102(3), 323-338.
Brown, J.F., Jr., D.L. Bedard, M.J. Brennan. J.C. Carnahan, H. Feng, and R.E. Wagner. 1987.
"Environmental dechlorination of PCBs," Environ. Toxicol. Chem., 6, 579-593.
Chillrud. S., 1996. "Transport and Fate of Particle Associated Contaminants in the Hudson River
Basin." Columbia University. Ph.D. Thesis.
Fish, K.M., 1996. "The Influence of Aroclor 1242 Concentration on PCB Biotransformation in
Hudson River Test Tube Microcosms." Appl. Environ. Microbiol., 62(8), 3014-3016.
Fish, K.M. and J.M. Principe, 1994. "Biotransformation of Aroclor 1242 in Hudson River Test
Tube Microcosms." Appl. Environ. Microbiol., 60(12), 4289-4296.
Gailani, J.C., C.K. Ziegler. and W. Lick. 1991. "Transport of Suspended Solids in the Lower
Fox River," J. Great Lakes Res., 17(4), 479-494
Hawiev, N., 1991. "Preliminary Observations of Sediment Erosion from a Bottom Resting
Flume," J. Great Lakes Res., 17(3), 361-367.
Hayter. E.J. and A.J. Mehta. 1986. "Modeling Cohesive Sediment Transport in Estuariai
Waters." Appl. Math. Modeling, 10 (Aug.), 294-303.
HydroQual. 1995a. "Quantification of Sedimentation in the Upper Hudson River." HydroQuai
technical report.
HydroQual. 1996b. "Anomalous PCB Load Associated with the Thompson Island Pool:
Possible Explanations and Suggested Research." HydroQual technical report.
Karim. M.F. and F.M. Holly, Jr., 1986. "Armoring and Sorting Simulation in Alluvial Rivers,"
J. Hydr. Engrg., ASCE, 112(8), 705-715.
Mehta. A.J. and Partheniades, 1975. "An Investigation of the Depositional Properties of
Flocculated Fine Sediments," J. Hydr. Res., 12(4), 361-381.
50
-------
Parchure. T.M. and A.J. Mehta. 1985. "Erosion of Soft Cohesive Sediment Deposits." J Hydr.
Engrg., ASCE. 111(10), 1308-1326.
Quenson. J.F.. Ill, S.A. Boyd and J.M. Tiedje, 1990. "Dechlorination of four commercial
polychlorinated biphenyl mixtures (Aroclors) by anaerobic microorganisms for
sediments," Appl. Environ. Microbiol.. 56. 2360-2369.
U.S.EPA. 1984. "Superfund Record of Decision: Hudson River PCBs Site, NY."
EPA/ROD/R02-84/004.
U.S.EPA, 1991. "Phase I Report - Review Copy Interim Characterization and Evaluation
Hudson River PCB Reassessment RI/FS." Tarns Consultants, Inc., Gradient Corporation.
EPA-68-S9-2001.
U.S.EPA, 1995. "Phase 2 Report - Review Copy Further Site Characterization and Analysis.
Volume 2A-Database Report. PCB Reassessment RI/FS." EPA-68-S9-2001.
U.S.EPA. 1996a. "PCBs: Cancer Dose-Response Assessment and Application to Environmental
Mixtures." EPA/600/P-96/001.
U.S.EPA. 1996b. "Phase 2 Report - Review Copy Further Site Characterization and Analysis
Volume 2B - Preliminary Model Calibration Report. PCB Reassessment RI/FS."
Limno-tech, Inc, Menzie Cura & Associates, The CADMUS Group, Inc.
van Niekerk, A., K..R. Vogel, R.L. Slingerland, and J.S. Bridge. "Routing of Heterogeneous
Sediments over Movable Bed: Model Development." J. Hydr. Engrg., ASCE. 118(2),
246-279.
van Rijn. L.C., 1984. "Sediment Transport, Part II: Suspended Load Transport." J Hydr.
Engrg., ASCE. 110(11), 1613-1638.
Ziegler. C.K.. and B. Nesbet. 1994. "Fine-Grained Sediment Transport in Pawtuxet River. Rhode
Island." J. Hydr. Engrg., ASCE, 120(5), 561-576.
51
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
3.0,
PKSD BB LMB
2.0 .
1.0
O.OL
SW T9 SW TO $W
River Location
Figure 1.
Ratio of Post-Release to Pre-Release Fish PCB Concentrations for
Pumpkinseed (PKSD). Brown Bullhead (BB), and largemouth Bass (LMB) in
Thompson Island Pool (TIP} and Stillwater Pool (SW)
-------
4
W. - 0.0 cm/yr
E
3
x
|m
a.
ui
a
0 25 60 75 100
TOTAL PCB (ug/g)
0
10
20
30
40
SO
60
70
80
90
W4 «• 0.5 cm/yr
A
,v'
a
W. - 1.0 cm/yr
I
0 5 cm at 20%
Dechlorination
0 25 60 75 100
TOTAL PCB lug/g)
0
10
20
30
40
50
60
70
80
90
1
/
0-5 cm at 20%
Dechlorination
0 25 50 75 100
TOTAL PCB (ug/g)
Figure 2.
TIP Sediment PCB Profile Cleft} with Projected 1991 Profiles with
Sedimentation and Dechlorination Mechanisms (center & right).
A 1991
• 1984
v 1977
-------
Federal
(DEIR - DF)
-------
U.S. DEPARTMENT OF COMMERC'
National Oeaanlc and Atmospha rtc i
Administration 1
National Ocoan Ssrvicc
Office of Ocun Raaourcas Oonsarvatbn and Assassmant
Hazardous Malariaia Raaponaa and Assassmant Division
Coastal Rasourcas Coordination Branch
290 Broadway, Rm 1831
N«w York. Now Yorti 10007
June 3,1997
Doug Tomchuk
U.S. EPA
Emergency and Remedial Response Division
Sediment Projects/Caribbean Team
290 Broadway
New York, NY 10007
Dear Doug;
Thank you for the opportunity to review the February 1997 Phase 2 Report, Further Site
Characterization and Analysis, Volume 2C - Data Evaluation and Interpretation Report (DEIR) for
the Hudson River PCB Reassessment Remedial Investigation/Feasibility Study (RI/FS). The
following comments are submitted by the National Oceanic and Atmospheric Administration
(NOAA).
Summary
The Phase 2 DEIR Report was prepared as part of the overall Phase 2 Reassessment RI/FS
activities currently ongoing to provide further characterization and analysis of the Hudson River
PCB Site which extends from Hudson Falls, NY to the Battery in New York Harbor. The
Reassessment RI/FS Work Plan, completed in September 1992, identified various data collection
activities to support the reassessment effort The February 1997 document presents geochenrical
analyses of water column and sediment data collected during die Phase 2 assessment and data from
other sources including New York State Department of Environmental Conservation (NYSDEC),
United States Geological Survey (USGS) and General Electric (GE).
The Phase 2 objectives were as follows: 1) the current and recent PCB source
contributions exclusive of the Upper Hudson River sediments 2) characterize the sources,
movement and distribution of water column and sediment associated PCBs, and 3) examine PCB
distribution and inventory within the Upper Hudson sediments!
General Comments
NOAA commends the authors of this report for a generally well thought-out site characterization
and analysis effort Overall, the report covered appropriate subjects and addressed them in a
credible manner. The authors should be complemented for an executive summary which clearly
highlights and explains the major conclusions of the Phase 2 reassessment and provides a
conceptual model for the factors affecting the fate and transport of PCBs from the Upper Hudson
to New York Harbor. NOAA's more specific concerns with the report are presented below.
Principal Congeners
The DEIR would have benefited from an early iH»ntifirarinn of congeners that are important for
understanding the fate of PCBs in the system, including congeners that are important in fish and
that represent the major contribution to each of the primary PCB homoiogues (Le., di-hepta).
Appendices A and B refer to the 12 principal target congeners as being the major focus of the
-------
NOAA Comments on Hudson River DEIR (February 1997) (6/3/97)
chat represent the major contribution to each of the primary PCB homologues (i.e., di-hepta).
Appendices A and B refer to the 12 principal target congeners as being die major focus of the
project. However, the report text primarily focuses on total PCBs or homologues, without much
discussion of the 12 "principal" target congeners. Including the 12 "principal" target congener
findings or representative congeners from each of the homologues in the report text might
contribute to the overall interpretation by providing more detailed congener-specific analysis based
on good quality data (Le. exclusion of sample where data for important congeners were rejected
or qualified as below detection). These congeners should be the focus of the discussions of
congener-specific and homoiogue concentrations throughout the repot This would also make it
possible to more easily identify samples with quality problems and exclude them from
analyses, where appropriate.
Implications of Data Quality Issues
The implications of data quality issues in the analyses of homoiogue compositions and congener
patterns, especially below detection limit and blank contaminated data, should be explicitly
addressed.
The treatment of below detection limit (BDL) missing or rejected data, and outliers, which
may have affected the results and interpretations, was not discussed. It is difficult to determine
how the qualifiers may have affected the results, including the calculation of totals and the
interpretations of observed patterns. The interpretive report did not adequately refer back to the
quality of the data, particularly as it might have affected rigta consistency and inter-comparability. 2A
For example, how were BDL values and rejected values treated in the homoiogue compositions
and pattern analysis? Some of these values were for principal congeners and the laboratory-
reported concentrations were quite high (and often these values were consistent with the data from
other samples). Were data for the entire sample or only the individual congener excluded from the
homoiogue and congener pattern analyses? What are the implications for the analysis, particularly
for presentations of homoiogue compositions and congener patterns?
All of the target congeners identified in the main report should be discussed in die usability
appendices. For example, BZ#44 is not discussed in Appendix A or B, although it is used as an 7
important example in Volume 1 Chapter 3.
p. A-25-28. Most of the congeners had a statement that "The detection limit goal of 0 J ppb was
met for nearly all samples." However, several congeners (particularly the higher chlorinated
congeners (BZ# 118, 138, 180) had a high percentage of samples qualified as below detection limit 2C\
values due to blank contamination. Many of these were in samples with elevated total PCBs, and, "
based on the congener composition in other samples, these congeners were undoubtedly present.
The implications for the data analysis and interpretation should be considered.
Congener-Specific Loading
Analyses of PCB loading and PCB patterns ^hn^ld include a discussion of congener-specific
concentrations as well as congener or homoiogue composition.
In many cases, the report relies primarily on the use of total PCBs and/or homoiogue PCBs to
describe trends without any discussion of representative individual congeners. The presentation of 3^
data for a consistent suite of important and representative congeners would make die description of -- - ¦
trends and comparisons of samples much clearer and easier to follow.
BZ#11 & was used to identify the presence of Aroclar 1254. However, according to the data from ^
the analysis of Arocior standards. BZ#118 is also a constituent of Aroclar 1242, though at a lower
concentration. If the PCBs in the source areas were determined to consist of approximately 85%
2
-------
NOAA Comments on Hudson River DEDt (February 1997)
imm)
Aroclor 1242 and 10-15% Arocior 1254, the contribution of BZ#1 IS from the two Axoclars would
be essentially the same.
Reliance on PCB Composition Data to Estimate Loading
Much of the trend and PCB composition analyses relied on PCB congener patterns expressed
primarily in relation to the concentration of BZ#52. Whik this approach is useful far comparison
of congener patterns, and BZ#52 is known to be one of the more persistent congeners, the
concentration of BZ#S2 is also changing, which makes it necessary to also present the data on a
concentration basis (preferably for selected individual congeners) to understand the changes in
composition. For example, are the-relative contributions of some congeners appearing to increase
due to additional loading or due to a more rapid decrease of BZ#52?
Presentation of the PCB composition data alone without concentration data can be misleading and
may lead to erroneous conclusions. Far in the paragraph on p. 3-122, the following
statement was made based on a comparison of congener patterns: "-the mixture at RM-1.9
contains substantially higher concentrations of the more-chlorinated congeners relative to RM
177.8." This statement would be correct if it referred to percent composition rather than
concentration. Based on our analysis of surface sediment sections from the high resolution core
data, the individual congener concentration data actually show that higher concentrations of these
higher chlorinated congeners are found at RM 177.8 than at RM -1.9. For example, the avenge
concentrations of BZ#101 and BZ#138 in the surface section of the two cores at Stillwater were a
factor of 6.4 and 3.5 greater than their respective concentrations at RM-1.9. While some of the
observed change in composition may reflect additional loading below Stillwater, the amount of
additional loading cannot be determined without taking into account congener concentrations and
the limitations of the conservative transport hypothesis.
Conservative Transport Assumption
The potential effect of phyricochemical weathering processes on congener patterns should be (
discussed and evaluated with die available data. A relative loss of lower chlorinated PCBs with
distance from a source area has been observed in New Bedford Harbor (Pruell et aL 1990) and
Lake Hartweil, South Carolina (Farley et aL 1994). Similarly, summer water column transects
"...show a gradual loss of mono- and dichlorohomologues as a water parcel travels from the TI
Dam to Waterford " (p. 3-148). This is particularly important to consider for any assumptions of
"conservative transport" and the analysis of congener and homologue patterns.
Weathering was identified on p. 3-138 as potentially an important factor to consider in tenxxs of
estimating local loads: To some extent weathering of the PCB homologue pattein in the water
column or soon after deposition may be responsible far the apparent local loadinp since loss of
lighter congeners would give the appearance of an additional local PCB load. However, on pp3-
162-164 an estimate of the percentage of contribution to NY Harbor (RM-1.9) sediments from the
Upper Hudson soiree and NYC sewage effluent is based cm a comparison of congener
compositions assuming conservative transport (no changes m PCB composition due to
weathering). Making the more realistic assmqption that at least part of the change in composition
can be accounted for by weathering of the material from the Upper Hudson, die percent
contribution of PCBs in harbor sediments from the Upper Hudson would be higher than the 50%
estimated.
Use of Physical Data
The report does not make good use of die physical ***»» presented in Chapter4 in the
of the PCB data. For example, information on pain size and organic carbon together with the field
descriptions can clearly indicate when die deposition of sediments has been episodic and allow far
3
10.03
-------
NOAA Comments on Hudson River DEIR (2/97)
the arguments when the PCBs came from the same source by demonstrating a further level of
consistency (sediment characteristics) among the samples.
Because the report presents little of the physical and conventional data, the reader has no
independent means of assessing the reasonableness of the data. For example, the particulate
organic carbon (POQ data were generated by applying a relationship developed for sediments. The
resulting estimates of POC concentration in rime and space and their comparison to suspended load
are not presented. Do the data make sense compared to other data from the Hudson River or other
systems?
Evaluation of PCB Sources above Rogers Island Thompson Island Dam (TO))
The discussion of the Hudson Falls Plant site and the Remnant Deposits as potential source areas
rely on the reports prepared by O'Brien & Gere in their reports far GE with little independent
analysis. For example, on p. 2-20 the conclusions of O'Brien & Gere about the "insignificance"
of the contributions from the Remnant Deposits are presented verbatim. As the DEIR points out,
there are several potential sources of PCB loading to the TIP and beyond in addition to the Hudson
Falls source including:
• High concentrations of PCBs in sediment above Bakers Falls adjacent to the old Hudson Falls
plant outfall. The reported total PCB concentrations in high resolution care 28 (RM197.1) was
greater than 100 ppm, even in the surface section. According to O'Brien & Gere (1994),
deeper sediment PCB results were as high as 22,000 ppm.
• PCB concentrations measured up to 44,800 ppm in soils at the base of a very steep cliff along
approximately 1200 feet of shoreline in the vicinity of the Ft Edward 004 outfall (Dames and
Moore 1994). Up to 2,700 ppm PCBs were detected in surface (0-6") shoreline sediment
samples in the same area (O'Brien &Geie 1995).
• The Remnant Deposit area including Remnant Deposit 1 (not capped), Remnant Deposits 2-5
(capped), and other sediment in the Remnant Deposit pool area
Considering the potential importance of this information for remedial decisions, NOAA strongly
recommends that EPA conduct an independent critical analysis using the available data and also
evaluate the adequacy of these data for determining the current and projected PCB loading from all
of these potential sources.
The PCB composition of the Hudson Falls source was derived from a single high concentration
water sample collected from the Remnant Deposit area during winter low-flow, low-temperature
conditions (Transect 1). This assumption of compositional consistency over time was not
evaluated, although the PCB composition in water column samples from the Thompson Island
Pool (TCP) differed considerably between winter and summer low flow periods (see p. 3-88).
The Hudson Falls source has been characterized by GE as unaltered Arocior 1242. Has an
independent evaluation determined whether the congener composition of the sample is consistent
with unaltered Arocior 1242? The composition of the surficial sediment in the TIP is consistent
with a mixture that also includes a smaller but significant amount of Arocior 1254 type material.
Reference to PCBs in the source areas as Arocior 1242 (as in "unaltered Arocior 1242") is
misleading, particularly due to the importance of the higher chlorinated PCBs in fish throughout
the river (including the TIP). Considering the fact that multiple potential sources may be
contributing to the observed loading at Rogers Island, congener analysis of samples of the seeping
oil could provide useful information on the composition of that source material.
4
-------
NOAA Comments on Hudson River DEIR (February 1997)
(6/3/97)
Evaluation of Upper River Loading Below TID
Although the DEIR states that a "consistent gain and homologue pattern change shown in all three
summer events strongly support the occurrence of an additional sediment-based load below the TI
Dam," there is no substantial evaluation of sediment-based loading below the TID. The importance
of a PCB sediment source below the TID should be evaluated, as it potentially contributes
additional exposure to fish, which would be reflected in both the composition and concentration of
PCBs in fish.
Additional Data Needs ~
There are two major reasons that EPA should consider limited additional collections at this time: \
1) to reduce uncertainty in the modeling projections; and, 2) to develop the baseline for long-term
monitoring.
Both the DEIR and the Model Calibration Report acknowledge the existence of important areas of
uncertainty, particularly in terms of understanding the dynamics of PCB loading to the Thompson
Island Pool. Understanding the dynamics of PCB loading to the Thompson Island Pool from all
potentially significant sources, which may include sources associated with the Hudson Falls and
Ft. Edward plant sites. Remnant Deposit 1 and any other remaining sediment deposits above
Rogers Island, and Remnant Deposits 2-5, is necessary for the model to make credible long-term
projections and therefore to be useful in the remedial decision-making process. Documenting the
impact of on-going remedial actions associated with the Hudson Falls and Ft Edward plant sites
on the loading to the TIP and the PCB body burdens in fish will provide important supplemental
information.
One of the ultimate goals of the Reassessment RI/FS is to evaluate the effect of remediation of
PCB -contaminated sediments in Thompson Island Pool on the concentrations of PCBs in fish in
the upper and lower Hudson River. Therefore, establishing a congener-specific baseline
monitoring program of resident fish species from selected areas is essential. Monitoring fish
provides a direct measure of the effectiveness of any remedial action and may help resolve some of
the uncertainty associated with the assessment of PCB loading at different locations in the river.
Because of the complexity of the system and the number of factors affecting PCB loading
dynamics in the river, the monitoring baseline should include sampling from multiple years.
Limited additional data collection should not cause any delay in the process but may help avoid
future delays resulting from high levels of uncertainty in the model projections that could be
reduced by additional information. NOAA would be to assist EPA in developing plans for
limited additional data collection efforts designed to reduce major sources of uncertainty and to
provide the basis for long-term remedial-effectiveness monitoring.
Thank you for your continual efforts in keeping NOAA apprised of the progress at this site. Please
contact me at (212) 637-3259 or Jay Held at (206) 526-6404 should you have any questions or
would like further assistance.
Sincerely,
j
Lisa Rosman
NOAA Associate Coastal Resource Coordinator
-------
NOAA Comments on Hudson River DEIR (Z/97)
References
Dames and Moore. 1994. The Outfall 004 Investigation Report, October 28,1994, General
Electric Co., Ft Edward, New York
Farley, KJ., G.G. Germann, A.W. Elzerman. 1994. Differential weathering of PCB congeners
in Lake Hartwell, South Carolina. Advances in Chemistry Series, V. 237: 575
O'Brien & Gere Engineers, Inc. 1994. Bakers Falls Remedial Investigation Operable Unit 3,
January 1994, General Electric Co., Albany, New York
O'Brien & Gere Engineers, Inc. 1995. The Fort Edwards Facility Outfall 004 Sediment
Investigation and Shoreline Protection IRM, November 1995, General Electric Co., Ft Edwards,
New York
Pruell, RJ., C.B. Norwood, R.D. Bowen, W.S. Boothman, PJ. Rogerson, M. Hackett, and
B.C Butterworth. 1990. Geochemical study of sediment contamination in New Bedford Harbor,
Massachusetts. Mar. Environ. Res. 29: 77-101.
cc Shari Stevens, DESA/HWSB
Gina Ferreira, ERRD/SPB
Robert Hargrove, DEPP/SPMM
Charles Merkel, USFWS
Ron Sloan, NYSDEC
Anton Giedt, NOAA/GC
6
-------
State
(DEIR - DS)
-------
DS-2
New York State Department of Environmental Conservation
50 Wolf Road, Albany, New York 12233-7010
John P. Cahill
Acting Commissioner
APR 25 I997
Mr. Douglas Tomchuk
Project Manager
Emergency and Remedial Response Division
United States Environmental Protection Agency
Region II
290 Broadway, 20th Floor
New York , New York 10007-1866
Dear Mr. Tomchuk:
This letter transmits the New York State Department of Environmental Conservation's
comments on the Hudson River PCBs Reassessment RI/FS Phase 2 Report-Review Copy Further
Site Characterization and Analysis Volume 2C - Data Evaluation and Interpretation Report (DEIR)
dated February 1997. The Specific Comments are arranged by the order they are found in the report
and general comments are included at the end.
Specific Comments:
P. E-l to E-2: (Major Conclusions)
Conclusion 1: The area of the site upstream of the Thompson Island Dam represents the primary
source of PCBs to the freshwater Hudson. This includes the GE Hudson Falls and Ft. Edward
facilities, the Remnant Site area and the sediments of the Thompson Island Pool.
We agree with this conclusion. Additionally, it is important to understand that this Thompson Island
Pool PCB load is altered by interactions between the sediments and the water column, as it is
transported downstream. This gradual alteration to the water column load results in an increase in
the proportion and overall amount of higher chlorinated PCBs. These higher chlorinated PCBs are
more bioaccumulative and thus more important in impacting fish PCB levels.
It is logical to assume the same process responsible for releases of PCBs in the Thompson Island
Pool are occurring in the rest of the Upper Hudson down to the Troy Dam. This would suggest that
Executive Summary
1
-------
the sediments in the Upper Hudson, downstream of the Thompson Island Dam, may be an additional
significant factor in generating the PCB leveis in the fish of this portion of the river, and the lower
Hudson River. If this is the case, then the impact of the sediments in the entire Upper Hudson needs
to be understood in order to evaluate a meaningful remedy. Otherwise, only addressing the
Thompson Island Pool sediments, while potentially very beneficial, may be too limited an action to
effectively address the PCB levels in the entire river's environment to an adequate level.
Conclusion 2: The PCB load from the Thompson Island Pool has a readily identifiable homologue
pattern which dominates the water column load from the Thompson Island Dam to Kingston during
low flow conditions (typically 10 months of the year).
We generally agree with this conclusion. It is important to recognize that there is a gradual shift in
the water column PCB load as the water passes downstream, which is indicative of the interaction
processes discussed above.
Conclusion 3: The PCB load from the Thompson Island Pool originates from the sediments within
the Thompson Island Pool.
We agree with this conclusion and the Department believes that there is an important question to be
answered related to this conclusion. Which sediments (the newly contaminated, or older sediments)
are contributing the load, and in what proportion?
Conclusion 4: Sediment inventories will not be naturally "remediated" via dechlorination. The
extent of dechlorination is limited, resulting in probably less than a 10 percent mass loss from the
original concentration.
We agree with this conclusion. The major corollary to this conclusion is that dechlorination will
have little or no effect other than in the highly concentrated PCB areas in the Upper Hudson. As a
result, the water column downstream of the Thompson Island Dam carries a greater proportion of
higher chlorinated PCBs. These types of PCBs are more readily bioaccumulative. The historical
aroclor fish data and the 1993 & 1995 congener fish data support this tenet. PCB fish concentrations
decrease with the distance downstream but the composition shifts to favor the more highly
chlorinated portion of the PCB spectrum.
Section 2
P. 2-3: The last paragraph should be changed to read from the "significance of sediments" to
significance of the site. The remediation of the shoreline has been completed and preliminary fish
monitoring data indicates improvement. Additional fish data should be available in the future to
evaluate conditions over time.
P. 2-14: The report discusses estimating seepage rates. The estimated flow rate of "200 gpm"
2
-------
is higher than the mean monthly maximum values cited on page 2-15 and could be the permitted
discharge rate for the wastewater (SPDES) outfall. We recommend that EPA use the 174 gpm value
cited on page 2-15 for the estimate used in the report.
P. 2-12: The General Electric Fort Edward Plant Site is currently listed as a dump in the New ~
York State Department of Environmental Conservation Listing of Inactive Hazardous Waste Sites
(Site No.: 5-58-004). The referenced listing is somewhat misleading and the DEIR should recognize
that the site is an existing capacitor manufacturing facility.
P. 2-18: "Because it was not part of the containment program, any remaining contaminated
sediment from Remnant Deposit 1 will be considered for possible remediation in the Phase 3 (4
Feasibility Study."
Does this mean that Remnant Sites 2-5 will not be considered for remediation in the Feasibility
Study? It is our understanding that EPA would evaluate remedial alternatives for the remnant
deposits if action is chosen for the river sediments. We recommend EPA determine if adequate
information exists to evaluate remedial alternatives for the remnant deposits. Additional time and
work may be required to obtain sufficient information in order to allow for evaluation and selection
of remedy.
Section 3
P. 3-81 "Water column may also serve to remove a fraction of the PCB load but this appears /
unlikely in light of the near-conservative transport observed from the TI(Thompson Island) Dam to \
Waterford noted below."
What process is/may be ongoing in the water column to account for this fractional load removal?
Degradation? Volatilization?
P. 3-84: (Discussions of Flow-Averaged Events 5 and 6) "Both events indicate an increase in ~
the water column load of approximately 18 to 39 percent (0.15 to 0.25 kg/day) between the TI Dam \
and Waterford."
This is the first indication in the report that the segments of the Upper Hudson below the Thompson
Island Dam contribute significant loads to the water column of the river under low flow conditions
(see Conclusion 1 comment). Given that, on p. 3-81, there is introduction of the concept of some
PCB load loss in the water column (presumably of the lighter congeners), the increase in load in
these two events is (1) made up of heavier congeners than the Thompson Island Dam load; and (2)
indicative of less altered, ie. lower concentration sediments, which have been found in the portion of
the river between the Thompson Island Dam and Waterford.
P. 3-85 to 3-86: "The close match of the water column homologue pattern to that of the
sediment identifies the sediments as the likely source of the water column load. The data suggest
that the total PCB load at the TI Dam is not the result of a simple addition by the sediment to the
3
-------
Rogers Island load. Rather it appears that some, if not ail of the upstream load is stored within the
Pool and that processes within the Pool serve to yield a load at the TI Dam principally derived from
the sediments, either from dechlorinated sediment directly or from sediment porewater or both."
This statement raises multiple issues:
First, is it possible to predict/identify the fate of the newly stored Rogers Island load in the (7 J
Thompson Island Pool?
Second, since the statement strongly implies that there are physical/chemicai/sedimentological
processes that result in an exchange of the Rogers Island load for a new, readily identifiable load (the
dechlorinated Thompson Island Pool load), are these physical/chemical/sedimentological processes
unique to the Thompson Island Pool? That seems very unlikely. The process or processes that
result in the load exchange seen in the Thompson Island Pool are almost certainly occurring in the
other pools in the Upper Hudson. The process or processes may vary in magnitude due to physica'
parameters such as contaminated sediment distribution and magnitude, channel
depth/shape/variability, length of pool, water velocities and velocity variability, and other
parameters which can control the sedimentation in the pools and the interactions between the water
column and the sediments.
The above leads to the third issue, which is a question "to what extent is the water column load seen
in the Hudson River downstream of the Thompson Island Dam influenced by the contaminated
sediments downstream of the Thompson Island Dam?" Is it possible that, while the Thompson
Island Pool water column/sediment interactions set the initial load, that the appearance of
conservative ("pipeline") transport in the Upper Hudson is actually indicative of a near equilibrium
between load loss and gain across the other pools in the Upper Hudson River downstream of the /"~x
Thompson Island Dam? While the mass load does seem conservative, the congener makeup
changes. The relative proportion of higher chlorinated PCBs increases as you go downstream.
P. 3-88: "In Transect 3, as the result of the onset of spring-flood conditions, scour provided
more than 94 percent of the total water column PCB load seen at Waterford (a gain of 1554 percent).
As noted in the detailed analysis in Subsection 3.2.5, this scour appears to be directly related to the
increase in the Hoosic River flow and represents erosion of Hudson River PCB-bearing sediments
within or near the Hoosic River delta and does not represent a PCB load originating in the Hoosic
River."
According to the NYS Canal Corporation, which has recently (1996) performed sampling and
dredging of Hoosic River delta sediments, the sediments of the Hoosic River delta contain very low
levels of PCB. Scour of the delta itself, therefore, is likely not the source of the loadings seen in
Transect 3. Rather, the scour of the sediments of the main stem of the Hudson itself, driven by the
increase in Hoosic River flow, is the likely origin of the load increase seen in Transect 3.
The observed increase in load in Transect 3, which originated downstream of the Stillwater
monitoring location, is also a clear indicator of the ability of the sediments downstream of the
4
-------
Thompson Island Pool to contribute to the PCB loads carried by the Upper Hudson, as discussed
above consequently affecting the fish regardless of the Thompson Island Pool actions.
P. 3-88: "Essentially, the net TI Pool loading is consistent at around 0.65 kg/day excluding
spring high flow. Only under spring high flow conditions is there no substantive TI Pool
contribution."
Does this mean that, under spring high flow conditions in 1993, that the load increase across the
Thompson Island Pool was not substantive as compared to the very high Rogers Island load, or that
the increase in load was not significant given the ability to measure the load? Also, if the load
increase was truly not substantive during spring high-flow conditions, does this mean that the load
may be groundwater discharge driven, and was minimized by the high river stages present over this ^
time period (which would minimize groundwater discharge to the river due to a lower head
differential from the groundwater to the river)? or do the results indicate that this equates to
residence time or dilution?
P. 3-101: "These coring sites are not typical of the general sediment type found in the Hudson
River; they are more consistently fine-grained and therefore represent only a small portion of the
Hudson river sediments."
This brings up an important question. Are the sediments in the Thomson Island Pool, that are
theorized to generate the Thompson Island Pool load, in areas where one could generate a core
chronology? If not, then this would indicate that there are times when either there is no
sedimentation, or events that result in the removal of part of the sediment column.
P. 3-108: "The total PCB analysis of the high-resolution cores yields two major findings. First,
there was a substantial decline in PCB levels between 1975 and 1982 followed by a plateau to 1990
in all regions of the Hudson River studied."
This agrees well with the historical fish PCB body burden data set, which adds credence to the
finding.
P. 3-142: "It should be noted, however, that the difference in Rogers Island conditions relative
to locations downstream may also be the result of the form of PCB release from the GE Hudson
Falls facility."
Is the EPA confident that their water column data generated from the Rogers Island monitoring point
is representative of the entire PCB load of the river at this point? May there have been transport of /^\
PCB oil past Rogers Island that is not represented in the data set?
P. 3-147: "As indicated in Section 3.2, it appears likely 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. This implies that some if not all of the unaltered PCB load originating above Rogers Island
is stored within the sediments of the TI Pool and is replaced and augmented by older PCBs released
5
-------
from the TI Pool sediments."
Again, this begs the question "what is the fate of the Rogers Island PCB load being stored in the
Thompson Island Pool?"
P. 3-151 Regarding the explanation of events that were occurring in April and May 1993. The
combined sewer overflow (CSO) was fixed in the middle of May by Washington County and
Niagara Mohawk. The significance of this is that water from the CSO was still flowing through the
Allen Mill and discharging out the Tailrace Tunnel until this problem was fixed. This helps explain
the observed PCB loading conditions.
P. 3-152 to 3-153: (Paragraphs on the relative importance of recent plant site related loadings
versus TI Pool sediment related loadings)
These paragraphs touch on several important points.
(1) Is it the current interpretation of the EPA that the relatively large GE plant site related
loadings from September 1991 to June 1994 are not related to the increase in Thompson Island Pool
loading over that time period? Or are the plant site loadings related, but it is not possible to quantify
their relative importance?
(2) The last sentence of this paragraph seems to state that the Thompson Island Pool
sediments that were contaminated years ago, combined with the recently contaminated (by the recent
plant site related releases) sediments, will be the primary source of PCB to the water column of the
river. Is that happening now?
(3) The statement is made on p. 3-153 that "...although the TI Pool load increased at about
the same time as the Rogers Island load, the net TI Pool contribution merely doubled and has since
dropped to its 1991 levels in spite of the roughly ten fold increase and decrease in the mean monthly
Rogers Island Loading for the same period."
What would be the result of this comparison of the Rogers Island load and net Thompson Island
Pool loadings on the basis of magnitude of load, rather than on the basis of proportional increase and
decrease at each of the two monitoring points? In other words, why not discuss, and take into (is*)
account in the interpretation, the magnitude of the loads at one point as compared to the other?
P. 3-170: (Water column conclusion summary)
The Department generally agrees with the conclusion presented at the close of Chapter 3, with a few
questions or comments.
Conclusion 7: "The source of the TI Pool load appears to be PCBs stored within the sediments.
N19
Which sediments? Is this old sediments, newly contaminated sediments from the source(s) above
6
-------
Roger's Island, or both?
Conclusion 11: "Hudson River PCB transport appears to be relatively conservative from RM
188.5 at the TI Dam to RM 88.5 near Kingston with loses and gains amounting to no more than
roughly 25 percent of the load at the TI Dam."
As discussed above, the Department believes that the same processes which result in the exchange of
the Rogers Island load for the Thompson Island Dam load within the Thompson Island Pool are also
going to result in some exchange of the Thompson Island Pool load within the other pools of the
Upper Hudson downstream of the Thompson Island Dam. The increase in load across the rest of the
Upper Hudson downstream of the Thompson Island Dam, coupled with a gradual loss of
dechlorination products, seems to indicate some exchange in load between the Thompson Island
Dam load and the progressively less dechlorinated sediments downstream of the Thompson Island
Pool. Some effort should be expended to evaluate the potential for the sediments from the rest of the
Upper Hudson River beyond the Thompson Island Pool to impact the water column PCB load in the
Hudson River. This is needed because a key result of the ongoing modeling effort will be predicting
the result of various remedial measures on future PCB fate and transport in the river. If the
Thompson Island Pool is not the only pool in the Upper Hudson in which sediments significantly
impact the water column PCB load of the river, then we need to know this and take it into account in
the remedy selection process. We view this as an important concept coming from the review of the
report.
Section 4
P. 4-80: When discussing the relationship (in the freshwater lower Hudson) between the water
column homologue pattern and that of the sediments, the statement is made that "...it appears that
the close agreement between water and sediment is again the result of sediment-water exchange..."
This again supports the importance, stated above, of the exchange between the water column and the
sediments throughout the river, not just in the Thompson Island Pool.
P. 4-81: (First full paragraph) "During warm, low flow conditions, the Thompson Island Pool
sediments are still locally important but, upon transport downstream, sediment-water exchange such
as porewater transport and/or possible in situ water column processes modify the water column
congener mixture yielding a less dechlorinated result."
We agree.
P. 4-82: (In discussing the newly deposited contaminated sediments, impacted by the recent
GE plant site related loads) "It is more likely that these materials, in combination with the existing
TI Pool sediment inventory, are responsible for the TI Pool source, i.e., the source results from a
combination of altered sediment and freshly deposited sediment whose net result is a mixture whose
properties closely resemble those of 1984 sediment."
7
-------
We agree. However, it will be important to quantify the magnitude and relative importance of the s~-
two loads for the purposes of remedy selection. We believe it may be necessary to perform
additional sampling to determine the magnitude of each source.
P. 4-82 to 4-83: "In this case, the greater the proportion of the water column load derived from
freshly deposited, unaltered PCBs, the more altered (and therefore concentrated) the remaining
sediment source must be in order to vield the properties of the water column mixture seen at the TI
Dam."
Would the understanding of the relative magnitude of the loads out of the two potential source
materials (newly deposited sediments vs. the old sediments) significantly change if there was rapid
alteration of the freshly deposited PCB-bearing sediments? Or would this only allow the /^\
contribution from the old sediments to be, on average, from somewhat less altered sediments? «—'
P. 4-83: "The sediments of the Pool are clearly responsible for the Thompson Island load
although thj mechanism for transfer from sediment to water column is unclear/'
"Downstrei*m of the Pool, sediment-water exchange via porewater may be an important mechanism
for the additional PCB loading noted during the summer. Given the complexities of sediment water
exchange, it is likely that the TI Pool load results from a combination of resuspension settling and
porewater exchange, involving recently deposited PCBs as well as PCB deposits that are ten years
old and older. In light of the large existing PCB inventory whose viability is suggested by the
geoph, sical and geochemical data presented here, it is likely that these sediments will continue to be
the major PCB source to the freshwater Hudson for the foreseeable future. How long these
sediments will continue to impact the Hudson on this scale is unclear but given the continual
sediment release for at least 3 years after the remedial controls were installed at the Hudson Falls
facility, it appears likely that this load will continue for several years, perhaps a decade or more."
Does this mean that the TI Pool load could go away (without remediation or other radical changes to
the system) in less than ten years? That seems highly unlikely, given the trends in PCB water /—n
column load, fish flesh body burdens, and sediment PCB history in the dated cores. These all
indicate that the PCB load carried by the river has been relatively stable (except for the increase
associated with the 1991-93 event related to the GE plant sites) since around 1984. We find this
conclusion to be questionable without knowing the basis for making it. It appears that once the
mechanism or mechanisms are understood, EPA should have additional information to estimate a
time frame.
Does this mean that the porewater driven exchange is the only exchange mechanism responsible for
the additional PCB loading seen downstream of the TI Dam? This also seems unlikely, given that
the TI Pool load is theorized to be a result of "...resuspension settling and porewater exchange...'V-^
Why would resuspension and settling be limited to only the TI Pool? These same processes are (25)
probably occurring in the downstream areas.
We caution EPA's interpretation that" ... continual sediment release for at least 3 years after the
8
-------
remedial controls were installed at the Hudson Falls facility..." Remedial work continues to be
active at this facility. The work completed in 1994 addressed some of the problems that were known
at that time. The Department has not selected final remedy at this site and therefore remedial /
controls have not been installed other than the Interim Remedial Measures and pilot remedial work
that has been completed.
General Comments:
• Data Needs to Verify the Model:
EPA's report, and their consultant's presentation at the Scientific and Technical Committee meetings
on March 25 and 26, 1997, acknowledge that the mechanism(s) and specific sources of PCB in the
load observed coming out of the Thompson Island Pool sediments have not been determined. The
report states, "these issues will be further addressed in the modeling efforts subsequent to this
report." EPA's modelers indicated that computer modeling would be used to understand the
significance of the various potential PCB sources to the water column. Thereby, EPA hopes to '
ascertain what sources are most important for causing the impacts to the environment. From this
understanding EPA would then evaluate remedial options that most effectively deal with the
important PCB sources. This modeling work will be qualified with an analysis of the uncertainty of
this proposed work. However, EPA's modeling consultants are primarily relying on the data
collected in 1993 to identify the sediments and processes responsible for the water column PCB load
in the river, and in turn perform predictive model runs for various remedial options.
EPA's proposed method to address this important issue raises a number of questions that warrant
careful consideration. We propose that, rather than primarily relying on the 1993 data, EPA
consider gathering some limited additional data to reduce the uncertainty of data being fed into the
computer modeling effort. This information could also be helpful to identify the potential
mechanisms that cause the observed PCB loading leaving the Thompson Island Pool. We are
willing to meet with EPA to discuss gathering additional data.
The advantage with this approach is that it attempts to resolve an important issue through direct
measurement while potentially complementing the modeling effort through direct verification. We
realize that gathering this information may slightly alter the reassessment schedule or components of
the schedule. However, we believe such data could reduce the uncertainties in the evaluation of
remedial options under consideration and provide a more supportable basis for the decisions to be
made regarding the sediments.
• Long-term monitoring:
a
We all recognize that the monitoring of PCBs in the Hudson River will have to be performed v
regardless of which remedial alternative is chosen. Therefore, we believe it is important that such a
monitoring program begin now. Further EPA data gathering within such a monitoring program
could also address a number of the known data gaps identified to date.
9
-------
The primary period of data collection for the Reassessment was 1993. There were dynamic
conditions involved during this period, and it is generally acknowledged that large inputs of PCBs to
the Hudson River above Rogers Island were still occurring. We believe the data gathered by EPA
was important to help discern the processes involved during the period of data collection . This
information further points to questions regarding which sources and mechanisms of PCB fate and
transport in the River are important. While remedial progress continues at the General Electric
Hudson Falls and Fort Edward Plant Sites, it is important to measure any observed changes to the
loading conditions on a river-wide basis. We anticipate that ongoing remedial work will change the
PCB loadings to the Hudson River observed at Rodgers Island. Measured changes in the PCB
-loadings, and the observed response, would certainly be important to predicting future conditions in
the Hudson River. The long-term monitoring program could also be used to verify and evaluate the
plant site loadings and response that would be measured, and also aid in the modeling of the impacts
of various remedial alternatives.
Therefore it is important that monit.. ring begin now and address obtaining useful data for specific
purposes. This r, pe of monitoring should address the key areas where additional data would reduce
the uncertainty with the modeling w^rk and aid in continuing the important trends that have been
monitored in the past. We are confluent in the value of additional monitoring and we request EPA
and its team of expert consultants design and implement the sampling and data collection program.
We are prepared to actively participate in the effort to assist EPA in performing this work.
Incorporate Ecological and Low Resolution Coring Data f ^
The data evaluation and interpretation report should eventually incorporate the ecological data( fish
and macro invertebrate) and low resolution coring data gathered as part of the Reassessment At a
minimum the conclusions drawn from this information should be documented and reported.
Information on Bioturbation
Enclosed are copies of papers regarding information on bioturbation. The Department of Health
raised this as a potential mechanism that is worth researching to determine if it could explain the
observed PCB measurements in the Thompson Island Pool.
The Department's technical staff would like to meet with you and your consultant to discuss
these comments and understand EPA's view of any additional work that is necessary. The data
demonstrates the impact the PCB contaminated sediments have on the Hudson River, especially
those above Troy. We believe the work completed by EPA has significantly narrowed the focus of
the remaining information needed to adequately identify and evaluate remedial alternatives.
10
-------
Please contact Mr. Stephen B. Hammond at 518-457-5677 to arrange such a meeting.
Enclosures
cc: Robert Montione DOH w/encl.
John Davis DOL w/encl.
Lisa Rosman NOAA w/encl.
Anne Secord USF&WS w/encl.
Sincerely,
David Sterman
Deputy Commissioner
11
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
Local
(DEIR - DL)
-------
DL-1
SARATOGA COUNTY
ENVIRONMENTAL MANAGEMENT COUNCIL
PETER BALET GEORGE HODGSON
CHAIRMAN DIRECTOR
April 8, 1997
Mr. Douglas Tomchuk
U.S. Environmental Protection Agency
290 Broadway, 20th Floor
New York, NY 10007
Dear Mr. Tomchuk:
Enclosed you will find comments on the Phase 2 "Data Evaluation and
Interpretation" report prepared by Saratoga County Environmental
Management Council member David Adams.
Since the comments in the enclosure to this letter call into question the
entire approach and conclusion of the "Data Evaluation and Interpretation"
report, the Saratoga County EMC has decided to refrain from providing
additional detailed comments.
The Saratoga County EMC is very concerned about the approach used by
EPA in this report in drawing conclusions that can only be viewed by the
general public as calling for dredging of sediments in the Hudson River if
EPA's future health risk environment shows the need for remedial action.
Such conclusions are most premature considering the work which remains to
be done, and the comments made on various EPA reports, and most recently
the serious and major concerns raised about the recently released "Data
Evaluation and Interpretation" report.
The Saratoga County EMC has been dismayed by the lack of feedback from
EPA on EMC comments made previously to EPA. EPA's statement in the
cover letter to the "Data Evaluation and Interpretation" report that the
Phase 2 report was divided into sections "to allow interested parties to
comment on the reports prior to the incorporation of the work" into
subsequent work would have real meaning and substance if EPA were to
provide some feedback on previous comments, at least to indicate whether
they have been rejected or whether they have been factored into subsequent
work.
While the Saratoga County EMC is still looking for EPA responses to its
previous comments, the EMC " believes it is absolutely essential that EPA
respond to the comments in the enclosure due to the fundamental nature of
these concerns and their potential effects on the future study of the PCB
problem in the Hudson River.
Siiwerely,
Peter M. Balet
Chairman
Enc.
cc: Mr. Richard Caspe, Region II USEPA
Mr. Albert DiBernardo, TAMS
Mr. Darryl Decker, Government Liaison Committee
SO WEST HIGH STREET
BALLSTON SPA. N.Y. 12020
(518) 884-4778
-------
April 7, 1997
Comments on EPA Phase 2 Report
Volume 2C - Data Evaluation and Interpretation Report
Hudson River PCB's Reassessment RI/FS February, 1997
David D. Adams, Member Saratoga County EMC and
Government Liaison Committee
1, The EPA report concludes the data show a PCB load at Thompson Island
dam greater than the PCB load entering at Rogers Island and the sediments
in the Thompson Island pool have been and are the major source of the PCB
load at the Thompson Island dam. Further, the EPA report concludes the
dechlorinated, buried sediments in the Thompson Island pool have been and
still are a likely source of the PCB's in the water column based on homolog
patterns and the level of PCB's in the water will not substantially decline
until the sediments are depleted (likely to be decades per EPA) or
remediated. The Saratoga County EMC strongly believes these conclusions
must be revisited for the following reasons:
o At the April 1, 1997 meeting of the liaison Committee members, GE
stated that the use of homolog patterns to pinpoint the source of
PCB's is not sufficiently definitive and that congener
"fingerprinting" is required. GE further stated that their analysis
of the congener "fingerprints" in the water samples at Rogers Island
and Thompson Island dam agreed and showed the PCB's in the water
to be directly related to a non-dechlorinated Aroclor 1242 source. GE
also stated the congener "fingerprint" of the dechlorinated sediments
in the Thompson Island pool did not agree with the congener
"fingerprints" of the water samples.
o It is not clear if the data available relative to the PCB loading
at Rogers Island are sufficient to completely define the PCB
loading input to the Thompson Island pool from the GE Hudson
Falls facility/Bakers Falls area. For example, information
obtained by GE in the last six to nine months has shown drops of
PCB oil entering the river. These drops could be carried along
the river bottom into the Thompson Island pool and thus not be
detected by the water samples. Therefore, liPA's conclusion that
there must be a net source of PCB's from the sediments in the
Thompson Island pool is suspect.
o EPA's data shows the PCB load in the water column at the
Thompson Island dam to be about the same as the PCB loading at
Waterford despite the indicted presence of PCB "hot spots" in
sediment deposits between the Thompson Island Dam and
Waterford. Therefore, EPA's conclusion regarding sediments in
the Thompson Island pool as the major source of PCB's to the
Hudson River requires that somehow the Thompson Island pool
sediments behave differently than the sediments below the
Thompson Island dam. However, no rationale or data are
presented by EPA to justify such different behavior.
o EPA's approach predicts that the source of the PCB's in the
Thompson Island pool is from buried sediments deposited in 1983
and prior years and not from more recent deposits. This source
of the PCB's defies logic in that it is not clear how these deep
deposits can both be the source of PCB's in the water and yet
-------
-2-
remain as "hot spots" over an extended time period. Logic would
seem to dictate that the sediments most likely to interact with the
water are the surfical sediments and not the buried sediments.
Also, EPA's scouring analysis indicates very little of the "hot
spot" sediments (none in the more highly contaminated deposits)
are resupended even in a 100 year flood which would mean much
less or none during normal river flows.
2. The discussion in item 1 suggests that a likely scenario for the source
of the PCB's going over the Thompson Island dam is as follows:
PCB's, primarily Aroclor 1242 are continuing to flow into the Hudson
upstream of Rogers Island from the GE Hudson Falls site/Bakers Falls
area. These PCB's, being largely hydrophobic, attach to surficial
sediments in the Thompson Island pool. The lower chlorinated PCB's
from the surficial sediments are preferentially transferred to the
water column as would be expected from the partitioning coefficients
for the various homologs. These PCB's then become the source of
the predominately mono and di-chloro PCB's seen in the water at the
Thompson Island dam, consistent with the "fingerprinting" work done
by GE. In this scenario, the buried sediments in the Thompson
Island pool became not a source of PCB's to the water but rather a
sink for PCB's, removing the PCB's from active contact with the
water column as sedimentation continues over time. This scenario is
consistent with PCB "hot spots" remaining over long periods of time
in the sediment and with the lack of contribution of PCB's to the
water column from sediment "hot spots" below the Thompson Island
dam.
3. As previously noted in our comments on Volume 2B, it is not at all clear
as to how EPA's model can handle the significant changes (one might say
even a discontinuity) in PCB output to the Hudson River above Rogers
Island starting in 1991. Drawing conclusions about the future water
concentration of PCB's from water data that is post- 1991 and sediments data
that pre-dates 1991 seems to be very difficult, if not impossible, and likely
to lead to erroneous conclusions. Repeating our previous comment, EPA
should divide its modeling effort into pre & post 1991 periods.
4. Based on the discussion in comments 1,2,&3, it is absolutely necessary
that EPA evaluate the possibility that continuing PCB inputs from areas
above Rogers Island are the principal source of PCB's to the water column at
Thompson Island dam and not the buried sediments in the Thompson Island
pool If the source is from- PCB's above Rogers Island, then the conclusions
at the bottom of page 4-91 of Book 1 of Vol. 2C that the water column PCB
level downstream of the Thompson Island dam will not substantially decline
unless the sediments in the Thompson Island pool are depleted or remediated
is completely wrong. If the source of the PCB's is from above Rogers
Island, the correct approach to reducing PCB levels below the Thompson
Island dam is to eliminate or significantly reduce the PCB inputs to the
Hudson River as GE is now working to do. In fact, there is recent evidence
that the GE effort to stop the PCB inputs above Rogers Island is
succeeding in reducing water column PCBs at the Thompson Island dam.
Certainly the statement on page 4-91 of Book 1 about the rate of depletion of
PCBs in the sediment implies sediment removal if remediation is deemed
necessary. If the PCB source is not the sediments in the Thompson Island
pool, removal of the sediment by dredging would not achieve the desired
result of reducing PCB's in the water column and indeed may cause just the
-------
-3-
opposite effect as well as possibly cause major ecological damage to the
Hudson River.
5. EPA's statement that dechlorination is not effective in reducing the mass
of PCB's in the Hudson River misses the point about the importance of
dechlorination. Obviously the removal of a few chlorine atoms from the PCB
molecule is not going to significantly reduce the PCB mass. What is
important is the difference in behavior between mono and di-chloro PCB's
and higher chlorinated PCB's. First is the evidence that the lower
chlorinated PCB's are not retained as much in fish as the higher chlorinated
PCB's thus reducing the PCB input to people eating the fish. Second,
EPA's recent re-evaluation of the cancer risk from PCB's shows a major
reduction in cancer risk from the lower chlorinated PCB's. Since EPA data
shows the water column to be dominated by the lower chlorinated PCB's,
both these factors serve to reduce the cancer risk from eating fish. On
page 4-51 of Book 1, EPA cites literature which purports to show that less
chlorinated congeners cause neurological impairment and developmental
damages. Information presented by GE at the April 1, 1997 Liaison
Committee Meeting indicated that subsequent studies have either not
confirmed the effects suggested by EPA or have shown observed effects to
be attributable to reasons other than PCBs. EPA should produce evidence
that has been accepted by the scientific community that PCB's produces
neurological or developmental effects before implying such effects can be
caused by less-chlorinated PCB's.
6. The use of congener "finger printing" should be applied to PCB data
(water columns, sediment and fish data) from the Lower Hudson and the
NY/NJ Harbor. The finger printing is necessary to either confirm EPA's
contention about the significant PCB input to these areas from the Upper
Hudson or show that lower Hudson and NY/NJ harbor sources are the reason
for the PCB contamination in these areas.
-------
Community Interaction
Program (DEIR - DC)
-------
DC-1
O.A. Borden & Sons, Inc.
Dairy & Fruit Farm
RD# 1 Box 153 Schaghticoke. NY 12154 (518) 692-2370 or 753-4186
April 11, 1997
Doug Tomchiik, RPM
USEPA, Region 2; Attn:DEIR Comments
290 Broadway, 20th floor
New York, NY 10007
Dear Sirs;
©
©
As chairman of the Agricultural Liaison Committee, I have been involved with this
process for some time. I firmly believe that w are well on the way to correcting the PCB
problem in the Hudson River. Unfortunately I'm not sure that the EPA or the Superfund
have had much to do with it except for encouraging GE to take action. Remedial action
taken by GE at the Hudson Falls plant site and Allen Mills site seem to be making a
substantial difference in the PCB levels in the water column.
My comments on these reports at this time are mainly procedural and are briefly as
follows:
1. Public participation seems to be of little importance in this project; we had almost no
time to view the documents before the EPA meeting in Albany and only TAMS view of
data was presented.
2. No indication of when and what conclusions would be drawn in these reports. For
instance, we could conclude first that the water column PCB load is decreasing before we
look at where the remaining load is coming from.
3. This is obviously a controversial project: why do we avoid looking at all opinions?
Why do public comments come due before we are exposed to differing opinions that arose
in the Science and Technical Committee? v
4. Why do obvious firesh sources of PCB contamination get ignored in this process?
5. How can you possibly put so much importance on 1993 data when so much remedial —_
work has been done since then? ( 5 j
Sincerely,
dZ z&CZifte'
i "
Thomas A. Borden
Chairman
Agricultural Liaison Group
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
J'1 U } JC -i' -f'
-/ — r*J , . DC-2
cctc<5x&(,
- - ijSXZ
!
V/5
7^
i
Vf I ' / ^
9 *y
, '" • - ~~)
f ££/<_£ £cY^?-?c •
/ 7
J
[.(iS, £¦
~/°fi //^c.
://
/*7 ~LX.~ri<~-
J , 'w'V z'
jr ,y
..to-
^ Z>^rzs
/ ) w^/
7 J
' £"/"?"*- "**-*
T
<2
D, A'dT" .
V*
,/-
©
7
/oLJaMa^^^cC dX~
C Cw^f
^£^/"C.<3-.L~ CS-'c-<- £—rr,-C^ •p^JL-T-rJ^^X-—
-^_CC CJC - -> >x <*1
V— ^"^*S-cy-Z-c
/ /•" , *v.
"-V
r+~*Z?
" /*'
' s /
dx<
/?
: ci-CC
v ^-7<
—T-.."
Of
<¦¦ i-
' vd-—&?~f~' C
/
t «&C£st -•><. iGi c-z-;. f?Zc<-c
-------
.••OCv. —c$ -y.O- u.«x'
/
~
_ss£— JH7€-
*l "?'~ ~—' C. £ (- . - ?Q . _ 7-7
,' S-'f i_ !_ (J-
'*n _3fcr.
.^wt 7 "*',
¦ *— (.
-/
WC4-1
_-7 Z. rr- 7 c_i
- /^t-v dd t,r>-c,
- y t'S - _
xuccl<_ „ /?{ o. cx_r
•7' ^Xc^ccl^-l.
. / T' A C CdT." 7^.6c2_<.'-CC_-
' */
-/Z*2
¦&*-¦- Vc 'z s*. j£*s
'£" .3 w-d. . cc—c-t,
7 •
' —c
. ^ C Zf- t—L
y
/
'_/
¦ <-I CCLcs
'C<2^-
? c':o,cctc
:oe_-
_Tc ^CcJT
7*±
-------
DC-3
15 Burgoyne Avenue
Hudson Falls, New York 12839
April 9, 1997
Douglas Tomchuk
US EPA - Region 2
290 Broadway - 20th Floor
New York, NY 10007-1866
Dear Doug:
Given the information contained in the Data Evaluation and Interpretation
Report as well as the information discussed at the most recent Science and
Technical Committee meeting, I offer these comments for the record. On page
4-90, Volume 2C Book 1 of 3, it is stated, "Given the large inventory of
existing PCBs as well as the possibility of additional PCB inventory from the
recent releases from the Bakers Falls area, it is not possible to strictly
define the exact nature of the sediments responsible for the TI Pool load or
the exact mechanism or mechanisms for sediment to water column transfer."
This statement goes to the heart of the reassessment. I believe that we have
all agreed that the failure of the gate at the Allen Mills was an event that
had a tremendous impact on the PCB levels in the upper Hudson. It is
imperative that we distinguish between the non dechlorinated PCBs coming from
the Allen Mill site and the older dechlorinated PCBs buried in the sediments.
There must be a quantification of the amount of PCBs released as a result of
the gate failure and a quantification of the results of all actions taken by
General Electric, with New York State's Department of Environmental
Conservation oversight, since 1991.
I have a real concern for the credence give the pore water (ground water flux)
theory. As I have read through volumes of material generated over the last 20
years, this is a phenomenon which has never before been identified. How can
we possibly believe that the ground water flux theory is unique to the
Thompson Island Pool and not to other areas of the Hudson? I question if this
ground water flux theory has been invented in order to make numbers fit in
areas where there are amounts that just cannot be accounted for.
I also feel compelled' to say that we need responses to comments in a timely
fashion. Meetings are scheduled to discuss the documents as they are
released, but we have no way of knowing what comments are being received by
EPA and how those comments are being addressed.
In conclusion, since the Thompson Island Pool has become the focus of study,
the precise location of all sources and the magnitude of their contribution
must be known in order to lead us to an informed ultimate decision.
Sincerely yours,
JJla
Sharon Ruggi
©
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
epartment of Earth and Atmosphenc Sciences
ollege ot Arts and Sciences
Science 351
.ny. New York 12222
University at Albany
STATE UNIVERSITY OF NEW YORK
518/442-4466 or 4556
Fax: 518/442-5825 or 4468
eilen@atmos.albany.edii
or geology@cnsunix.albany.edu
http://www.atmos.aibany.edu
DC-4
April 11, 1997
Mr. Douglas Tomchuk
US HP A P.eaion 2
290 Broadway - 20th Floor
New York, New York 10007-1866
Re: DEIR Comments
Dear Mr. Tomchuk:
Unfortunately I was away on business during the recent Hudson River STC meeting and
unable to attend. At the risk of repeating some points possibly raised at the meeting, I will
exercise the option to comment further as offered in the letter of February 13 accompanying Vol.
2C of the Phase 2 Report.
In my view the most serious shortcomings of the present stage of the Phase 2 investigation
are the mass balance calibrations obtained and applied to the Thompson Island (TI) pool, and the
consequent process interpretations derived therefrom. I have noted two of the major data
inconsistencies (suspended solids and PCB mass loadings) in my previous letter of November
13. ana add the following comments:
I. Assumed mass balance PCB loadings, and to a lesser extent suspended solids (TSS)
loadings, to and within the TI pool are in error, and result in an unrealistic interpretation of
TI pool processes and dynamics. A conclusion, based on the Phase 2 calibration, that the
TI pool generates a significant increase in total annual PCB loading beyond that passing
Ft. Edward is almost certainly wrong because of inconsistencies with other data and
observations — both historical, and some of those cited in Phase 2. In brief, some of these
are as follows.
1) A comparison of the PCB concentrations of TSS at higher flow rates (>300 m /s) with
the tops of sediment cores (e.g. Rogers Island and TID RM 188.5) would indicate that ( 1
the coexisting water column PCB concentrations at Rogers Island are higher (median
and maximum by 3x) than at the TID (Table 3-19).
-------
2) Two dated cores in the TIP establish an annual sedimentation rate of ~ 1 cm/vr.. which
agrees well with other estimates of net sedimentation. A net or prevailing state of (2)
scour cannot exist in the TIP. and a net loss of resuspended sediment cannot be argued
as a PCB loading dynamic, especially at low flow. (Table 3-18).
3) An inference of a significant pore water diffusion "contributionv to PCB loading driven
by groundwater flow is not supported by any evidence. Diffusion alone is incapable of
such an effect since other studies indicate a mm range at best; a comparison of
discharge estimates at various River stations provides no indication that groundwater
recharge to the River in the TIP is any more a factor there than elsewhere downstream
-- including other stretches with PCB "hotspots". A major fault zone, potentially
capable of transmitting groundwater, underlies the Champlain Canal terminus at Ft.
Edward, but this is upstream of most of the TIP. Finally, any continuing PCB mass
net discharge flux from TIP sediment older than 1-2 years would progressively lower
TIP hot spot concentrations. No evidence of this exists in comparisons of sediment
core profiles of different ages.
©
4) If PCB loading increases in transit of the TIP, why is no similar effect observed below
the TID where other substantial hot spots exist? The concept of a PCB loading
"pipeline"' between the TID and Waterford is inconsistent with the observations of
water column total PCB-TSS relations at high flow, a condition of net annual
sedimentation in the upper Hudson, and observed PCB concentrations of core tops.
©
5) Historical data comparisons of PCB concentrations in water samples at Rogers Island
versus Schuylerville. and at Rogers Island versus the TID (also Phase 2. vol. 2B, Fig.
4-2, 4-5), reveal many exceptions to a presumed trend of an increase in concentrations
or PCB loading along the TIP. Such a trend is best displayed in data at low flow, and
is very dubious at high flow (e.g. Vol. 2C, Table 3-16; Transect 4. Flow Average #1).
Even at low flow, exceptions are common (e.g. Table 3-16; Transect 1, Flow Average
#3; Vol. 2B, Fig. 4-2). Clearly, sample representation relative to flow rate and event
timing is a problem with all the data, and a basic weakness in estimating annual f5)
loadings at any station. For example in Table 3-23, note that estimated annual PCB ^
loadings at Schuylerville (below TID) are greater than Ft. Edward 1977-1985, and then
abruptly reverse 1986-1993 (1990-93 relative to Waterford). This shift may merely
reflect a change in USGS sampling procedure (personal communication, C.R. Barnes)
rather than any change in the TIP or River processes. Another illustration of the
problem is the very large variance of the observed (USGS and GE) PCB concentration
data at Ft. Edward, at both low and high flow. This variance reflects erratic PCB pulse
or "spike" discharges from or near the source; all too often high values have been
discarded as anomalous in estimates of mean or annual loadings, which can lead to
large errors at high flow rates.
II. My letter of November 13 noted the problem of representative sampling for estimating PCB /-—-x
mass balance in calibration of the TIP. Upon reviewing Vol. 2C this problem looms even
more forcefully; my view now is that much or all of the Phase 2 PCB loading data for the Ft.
-------
Edward - Rogers Island sample site is unreliable and cannot be used for meaningful model
calibration. This view derives from: 1) a consideration of the variance shown in analyses
from this site, as outlined above; 2) available information as to the sampling methods
employed in Phase 2 (flow averaged and transect sampling); and 3) known physical sources
of variance at the site which make representative sampling difficult.
Variance due to the latter includes cross channel and vertical inhomogeneitv in PCB
distribution in the River at any point in time, and any effect of changes in flow rate and
water temperature on the distribution. This variance is separate from that of the variation in
PCB source (Hudson Falls) discharge, and both are included in the observed variance of (1)
above.
A more or less persistent cross channel variation effect at Ft. Edward was described by
Tofflemier (references in November 13 letter), and further demonstrated in dye injections by
G.E. The sampling studies of O'Brien and Gere also showed that non-persistent and
unpredictable vertical and cross channel variations were present relative to both (east and
west) margins of the River, as well as the Rogers Island channels.
Some idea of the combined variance is found in the historical data, especially that of the
USGS when both channels at Rogers Island were sampled (water column) from the Rte. 197
bridge. It should be noted that this is instantaneous sample point site variance; time
dependent variance (e.g. 15 observations, March 31 - April 29, 1993 high flow event) of
PCB loading is an additional component of the total variance. Relative to total variance, the
Phase 2 Ft. Edward calibration data in essence assumes a knowledge of the time dependent
PCB loading without any knowledge of the various components of variance referred to
above.
In short, any competent statistician would be in agony if confidence limits were required
on the Phase 2 PCB data simply because no analysis of the components of total variance has
ever been made. Without such knowledge, the relevance of individual sample results is
obscure, whether transect, water column, or flow averaged: and the reliability is unknown.
In respect to Phase 2, a scenario whereby TSS from the spring high discharge event is
annually deposited in the TIP and subsequently desorbs partially altered PCB to the TIP at
low flow (any net scour being restricted to erodable or unstable sites of prior year deposited
sediment in the first spring high discharge event), seems equally compatible with the cited
data, including a possible TID PCB net discharge > Ft. Edward at low flow.
III. The problem of relating TSS to discharge was also noted in my November 13 letter. The
Phase 2 model calibration assumes a linear regression, which is incorrect, and until an
algorithm is developed to incorporate flow event peak and sequence timing, such calibration
is also in error. In addition, the mass balance model for PCB vs. TSS in the TIP assumes the
two may be related as an estimator of scour or resuspension of sediment (Vol. 2C: 3.2:4).
.An analysis of all the historic data for TSS vs. PCB concentration (water column) shows no
relationship at any station, especially at high flow, therefore PCB loading cannot be used to
-------
infer scour or resuspension even if such is present. Further, possible shifts in PCB
homologue or congener distributions which might bear on this issue are non diagnostic in
Phase 2 (Vol. 2C, book 1. p. 3-48. 49).
I would be happy to discuss any of the above in more detail at the next STC meeting. In
the meantime, it might be helpful if I could be provided with the physical details of the on-site
water sampling procedures beyond the descriptions in the Phase 2 Report.
Very truly yours.
George W. Putman
cc: W. Nicholsen
J. Haggard
J. Davis
-------
Public Interest Croups &
Individuals (DEIR - DP)
-------
DP-1
Hudson River Sloop
CLEARWATER.
112 Market Street, Poughkeepsie, NY 12601-4095 • 914-454-7673 • Fax: 914-454-7953
e-mail: office@maii.clearwater.org htcp://www.cleanvater.org
April 9, 1997
Douglas Tomchuk
US EPA - Region 2
290 Broadway - 20th Floor
New York, NY 10007-1866
Attn: DEIR Comments
Dear Mr. Tomchuk,
Hudson River Sloop Clearwater, Inc. is a 501(c)(3) organization, founded in 1966, with
the mission to preserve and protect the Hudson River and its watershed. We currently
have some 10,000 members, most from the Hudson Valley, but also from 44 states and 9
countries.
Clearwater has been involved with the Hudson Rivei1 PCB contamination issue for the last
two decades. We are grateful for the opportunity to participate in the reassessment
process and are confident it will lead to a favorable Record of Decision. We applaud the
EPA for taking the time and energy needed to collect and analyze the vast amount of data
presented in the Phase 2 report.
In general, we support the conclusions presented by the Data Evaluation and
Interpretation Report (Phase 2). They are: (1) the area of the site upstream of the
Thompson Island Dam represents the primary source of PCBs to the freshwater Hudson;
(2) the PCB load from the Thompson Island Pool has a readily identifiable homologue
pattern which dominates the water column load for the Thompson Island Dam to Kingston
during low flow conditions (typically 10 months of the year); (3) the PCB load from the
Thompson Island Pool originates from the sediments within the Thompson Island Pool;
(4) sediment inventories will not be naturally "remediated" via dechlorination. These four
major conclusions come as welcome news and are a confirmation of our long-held
convictions. It was also encouraging to have General Electric's rhetoric exposed as the
"myths" that they are.
To restore and protect the Hudson River, its shorelines and related waterways
-------
Clearwater does disagree with the EPA on one point: The time to depletion of the
Thompson Island Pool PCB inventory (last sentence of Executive Summary).
According to this EPA report, there exists in the pool between 19.6 and 14.5 metric tons
of PCBs. An average of those two numbers yields an estimate of 17 mt, or 37,500
pounds. Using averages from EPA's data in table 3-24, it appears that approximately 3
pounds of PCBs go over the Thompson Island Dam every day. This results in an annual
estimate of 1,100 pounds. If one were to assume linear depletion rates with no
"diminishing returns," the time required for depletion would be almost 4 decades
instead of the single decade suggested by EPA in the last sentence of their Executive
Summary. We hypothesize, however, that the actual depletion of Thompson Island Pool
PCBs is likely to be characterized by diminishing rates of contamination, with the effect of
prolonging for an indefinite time the constant (though diminishing) recontamination of the
lower Hudson, with the probable loss of fisheries and extension of human health risks
for additional decades. One must also remember that the "armoring" of the remnant
deposits, a temporary solution which already allows PCBs to leach into the river, may
begin to break down, with resulting elevated PCB released over time. Furthermore, there
remains as unquantified "very large amount" (DEC 1996) of PCBs in inventory in the
shale bedrock below GE's Hudson Falls plants, which has been contributing approximately
1/2 pound per day to the river's burden (about 30%). Remediation efforts at Hudson
Falls, conducted by GE contractors under the oversight of DEC, may be slowing the rate
of seepage, but the prospect of complete stoppage are highly uncertain.
We suggest that it would not be unreasonable to contemplate a time horizon of PCB
contamination in the Hudson River a century or longer in duration. The only prospects for
curtailing this threat to biological and human health lie in dredging and treatment of
sediment from the upper Hudson hot spots, removal and treatment of the remnant
deposits, and continued remediation of the Hudson Falls seeps.
In conclusion, Clearwater is pleased with the majority of the Data Evaluation and
Interpretation Report (Phase 2) and would request that the time to depletion of the
Thompson Island Pool PCB inventory be reexamined for possible errors in calculation.
We are looking forward to the next phase of the reassessment process. If you have any
questions regarding our comments, please call. Thank you for your time.
Sincerely,
Andre Mele
Environmental Director
-------
DP-2
13 Roveland Avenue
Delmar, New York 12054-3037
November 11, 1994
Doug Tomchuk
Regional Project Manager
PCB's REASSESSMENT
US En». Prot. Agency Region 2
26 Federal Plaza
New YOrk, New York 10278
Dear Mr. Tomchuk:
I am a resident of the Town of Bethlehem, Albany County,
New York.
Our town water supply serves approximately 20,000 residents
and several large industries.
In 1995, an additional new source of supply (up to 6 MGD)
will supplement our upland source. The new source will
be an infiltration gallery approximately 1000' long
20* wide 30' deep and located parallel to within 30' of
the west side of the Hudson River. The infiltration
gallery is located 17 miles south of the Troy Dam opposite
the Village of Castleton.
I understand that EPA is currently evaluating the Human
Health Risk from PCB's in the Hudson River.
Our new water source will serve our community sometime
in 1995. Very limited sampling of the new source indicates
that concentrations of PCB's are within acceptable
limits. A local citizen's group "Clearwater For Bethlehem'
believes that the potential Health Risks from PCB's should
be more thoroughly evaluated with particular reference
to our new water source.
Attached is a letter dated 11/29/93 from Mr. John Dunn
of the NYS Dept. of Health to Mr. Fred Sievers of
NYS Conservation Department. Mr. Dunn notes that the
avenue of water travel is vertically through the river i 1
bottom and then horizontally via sand/gravel. Our new
water source could result in PCB exposure of our residents.
Is it possible that water travel through Hudson River
bottom sludge containing PCB's will increase our level
of exposure that would not be encountered from a direct
Hudson River intake?
-------
A limited number of samples have been collected over
a relatively short time frame. Significant changes in
the water chemistry between river water and infiltration
gallery water have been noted. The ammonia concentrations
exceed 2.0 ppm and phosphate and carbon dioxide levels
increased significantly. The total organic carbon following
chlorination ranges from 7-13 ppm.
The town consultants have recommended an ozone dose of
2 to 3 ppm to treat the water. This dose is, in part,
to remove iron (over 5 ppm) and manganese (o.5 ppm).
Would this ozone dose produce toxic ozone by-products
in the presence of PCB's?
You have scheduled a report due May 1995 on the Human
Health Risk Assessment of Hudson River PCB's. Can you
include an evaluation of the risk from PCB's in connection
with withdrawal of water via the infiltration gallery which
will serve as a source of water for Bethlehem?
You can get additional information on source, sampling
results and the schedule for operation from-
Ms. Sheila Fuller, Supervisor
Town of Bethlehem
445 Delaware Avenue
Delmar, New York 12054
I would appreciate a reply
Very truly yours
Sherwood Davies
SD/mbd
c.c. Ms. Sheila Fuller, Supervisor, Town of Bethlehem
Ms. Linda Burtis, Clearwater For Bethlehem
-------
13 Roweland Avenue
Delmar, Mew York. 12054-3037
April 5, 1997
Douglas Tomchuk
LrS EPA - Region 2
290 Broadway - 20th Floor
New York, Nev York 10007-1866
Re:
DEIR Comments
USEPA Data
Evaluation and Interpretation
Report (DEIR) on Hudson River
RCB's Superfund site
Dear Mr. Tomchuk:
I offer the following comments on the above referenced
report. The DEIR fails to provide data on.any interpretation
on porewater migration into the Hudson River ground water
aqui fer.
In 1996, the Town of Bethlehem, Albany County began with-
drawing 2.0 to 3.0 MGD from a horizontal well (infiltration
gallery) adjacent to the Hudson River. This ground water
source is under the influence of a surface water (Hudson River)
The veil is located on the west side of the Hudson River
opposite the Village of castleton. Attached is a copy of a
letter to you dated November 11, 1994 furthur describing
the Bethlehem supply.
The following data and conclusions from the DEIR•indicates
that PCB's in the pore-water of bottom sediments exceed drinking
water limits and that migration into ground water can
contaminate potable water supplies.
* PCB's can transfer to the water column involving
porewater exchange i.e. the transport of PCB's to the
water column via interstitial water found within the
river sediments (E-4) .
* The flux of PCB's from the sediment porewater must
also be considered as a potential PCB source to the
water column (3-49).
* On the basis of this PCB "fingerprint" it was
concluded that the Thompson Island Pool sediments
represented the major source to the water column...(E-4.
-------
* Porevater samples from the Great Lakes were found to
be 10 to 500 times greater than the water column
concentrations (3-49).
* Porevater results yielded a total median composited
PCB concentration of 6.43 ug/L (ppb) from a sediment
with a median value of 17,000 ug/Kg )(ppb) (3-4$).
~Total PCB's in the 3.5 foot depth of bottom sediments
at the Albany Turning Basin R.M. 143.5 ranged from
1,000 ug/Kg to 18,000 ug/Kg with an average value of
4,000 ug/Kg (Figure 3-61).
* PCB's are hydrophobic and tend to bind preferentially
to organic carbon present in suspended solids in the
water column, or in the sediments in the river bottom
(3-46).
* Based on a 1 liter sample, PCB Congeners in the water
column are detected at levels of 0.5 to 1.0 ng/L (ppt)
and total PCB's are detected at levels of 10 ng/L
(ppt) (B-24).
Extrapolating data from the DEIR, PCB's in porevater from
sediments in the Albany area ranges from 380 ng/L to
6,800 ng/L (ppt). New York States Maximum Contaminant Level
(MCL) for PCB's in potable water is 500 ng/L with a
USEPA goal of zero.
The Water Quality Regulations (Surface and Groundwater'
Classification and Standards) of the New York State Department
of Environmental Conservation has established standards for
PCB's in groundwater. Section 703.5 establishes a standard
of 100 ng/L in groundwater. Section 703.6 establishes a
graundwater effluent standard of 100 ng/L which would apply
to a point source such as Hudson River bottom sediments.
The laboratory reporting results on samples collected from
the Bethlehem well failed to report total PCB's. Thus no
comparison can be made as to PCB's in porevater reported by
the USEPA and the Town's PCB analysis of the well water.
High concentrations of organic carbon and nitrogen and a
variable turbidity in the well water indicates the influence
of water "filtering" through Hudson River bottom sediments.
The DEIR confirms the environmental persistance of PCB's
in the bottom sediments and a concentration of PCB's in the
porewater much greater than that found in the water column.
Bethlehem's well is within the zone of influence of the
PCB's bottom sediments.. These sediments represent a point
source of. contamination.
-------
The Hudson River PCB's Superfund site represents a health
risk to residents of the Town of Bethlehem. The USE PA Human
Health Risk assessment report to be released in December 1997
will be incomplete if it fails to evaluate the health risk
from PCB contamination of groundwater sources adjacent to
the Hudson River.
Very-tj?uly yours, .
Sherwood Davies
SD/mbd
Enc.
c.c. Richard L Brodsky, New York State Assemblyman
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
DP-4
PACE ENVIRONMENTAL LITIGATION CLINIC, INC.
PACE UNIVERSITY SCHOOL OF LAW
78 NORTH BROADWAY
WHITE PLAINS, N.Y. 10603
SUPERVISING ATTORNEYS
MATTHEW R. ATKINSON
KARL S. COPUVN
ROBERT F KENNEOi'. JR.
914-422-4343
FAX: 914-422-4437
ADMINISTRATOR
CONSTANCE HOUGH
ADMINISTRATIVE ASSISTANT
MARY BETH OlSTEFANO
April 11, 1997
Mr. Douglas Tomchuk
US EPA Region II
290 Broadway, 20th Floor
New York, NY 10007-1366
Re: Hudson River FCBs Data Evaluation and Interpretation Report
Dear Mr. Tomch.uk:
We represent Hudson Riverkeeper Fund, Inc. (Riverkeeper), a
not-for-profit conservation organization whose purpose is to
conserve and enhance the beauty, quality and life of the Hudson
River, its tributaries and the New York City watershed.
Riverkeeper supports the major conclusions found in the Data
Evaluation and Interpretation Report (DEIR).
We believe that the significant source of Hudson River PC3s
is the Upper Hudson River. The Upper Hudson River, particularly
the GE sites at Fort Edward and Hudson Falls, acts as a faucet
for the PCB contamination in the Lower Hudson River. EPA's data
demonstrates that the source of PCBs from the Mohawk and Hoosic
Rivers was less than 20% of the total PCB load, even during 100-
year flood events.
Additionally, we support the EPA's evaluation that the PCBs
located in sediment will not be naturally remediated by
dechlorination. We agree that the "hot spot" sediments are a
potential and recurring source of PCBs in the water by
resuspension.
Riverkeeper believes that the EPA is correct in determining
that the water column PCB levels in the Lower Hudson River will
not substantially decline beyond current levels until the active
FCB-laden sediments are remediated.
-------
Hudson Riverkeeper
conclusions in the DEIR.
"aken :o remedy the PC3
Fund, Inc., supports Che major
We encourage remediation efforts
concentrations found in the Hudson
be
River.
Sincerely,
tonette Boehlje
Legal Intern
-------
SCENIC
DP-5
HUDSON
INC
April 11, 1997
Officers and Directors
Chairman Emeritus
Mrs. Willis Khm
C^atnnan
David N Redden
Vice Chairman
Frederick Osbora HI
Treasurer
Manone L Hart
Assistant Treasurer
J«h V. Johnson. FAJA
Secretary
Elizabeth B Pugh
Alice W Bamberger
Mrs. Francis H Cabot
Christopher C. Davis
Kathleen Durham
William M Evjru. /r
B Hamsori Frinkei
Anna Carlson Gannett
Gen. Pal Carve v
vnthia H. Gibbons
J. Thurston Greene
lemson H. Heckscher
Lowell Johnston
Richard Kdev
Frank Martuca
Hamilton W Meserve
Anthony I Mcneilo
David H Mortimer
Francis I Murray, |r
Warne L.S Pnce
Rudoloh S Rauch
FredencC Rjch
David S Sampson
H. Claude Shostal
Whwlock Whitriev in
lohri P. Wort
Alexander E. Za^oreos
Advisory Board
Mash Castro
Stephen P Duggan
William H Ewrn
John French ID
George W Gowm
Bamabas McHenrv
Charles P Naves 111
Mrs. Frederick M. Osborn, |r.
La u ranee Rockefeller
David Sivt
Mrs. Thomas M Waller
William H Whvte
Honorary Directors
Robert Bovte
Rjchard H Psugh
Executive Director
"lata B Sauer
Mr. Doug Tomchuk
Emergency and Remedial Response Division
U.S. Environmental Protection Agency
290 Broadway, 20th Floor
New York, NY 10007-1866
Re: Data Evaluation and Interpretation Report for
the Hudson River PCBs Site
Mr. Tomchuk:
This letter constitutes Scenic Hudson's comments on the Data Evaluation
and Interpretation Report (DEIR) of EPA's Phase 2 Reassessment for the
Hudson River PCBs Site.
As you know, Scenic Hudson has a Technical Assistance Grant (TAG)
from EPA to aid our participation in the Reassessment. It is our intention to
reserve the bulk of the TAG for analysis of human health and ecological issues
as well as remedy evaluation and selection. With the limited TAG funds
allocated to the DEIR, we are working with our advisor to prepare interpretive
materials to help the public understand the DEER beyond simply the key
findings.
The DEIR is an impressive document. It is without question the most
thorough and authoritative analysis of the PCB contamination in the Hudson
River to date. In documenting the nature and extent of contamination for this
site, EPA has vastly surpassed typical Superfund investigations in the amount
and quality of data, scope of analysis, and scientific rigor.
The DEIR is an important step forward because it effectively clears much
of the confusion and controversy regarding sources and dynamics of PCB in
the Hudson River. In particular, it establishes that the Thompson Island Pool
sediment "hot spots" are a significant source - in fact the dominant source ~
of PCBs to the river. In addition, dechlorination has reduced the PCB mass in
sediments less than 10 percent and, therefore, "no action" is not a viable means
of ameliorating human health and ecological risks.
~ Vassar Street
Paughkeepsie. NY 12601-3091
914-473-4440
FAX 914-473-2648
email: scenschuSmhv.net Q
-------
Mr. Doug Tomchuk
April 11, 1997
Page 2
The conclusions of the DEIR reflect a considerable weight of evidence. Several lines of
investigation using numerous sources of data (including abundant data supplied by GE)
combine to a form cohesive and consistent understanding of the dynamics of PCBs in system.
There are still aspects of the system that are understood incompletely. However, the central
DEIR conclusions derive from robust empirical datasets. In carrying out the investigation,
EPA cooperated with the New York State Department of Environmental Conservation,
members of the Scientific and Technical Committee, and General Electric, which provided
extensive data and exhaustive scrutiny.
It is noteworthy that the data show patterns over time and over distance moving downriver
from the sources of contamination. Because the study period encompasses several seasonal
cycles and one-time events such as floods, increased releases from the plant sites in the early
1990s, and attenuating releases from the plant sites in 1993, EPA has been able to gain a
better understanding of the history and dynamics of the system. Moreover, analysis of the
types of PCBs enables EPA to distinguish between releases from the sediment hot spots, the
plant sites, tributaries, and New York Harbor sources.
Please call me or Josh Cleland at (914) 473-4440 with questions of comments. Thank
you.
Sincerely,
Cara Lee
Environment Director
/ rmm
-------
General Electric
(DEIR-DG)
-------
Sidley & Austin
A PARTNERSHIP INCLUDINO PROFESSIONAL CORPORATIONS
CHICAGO
DALLAS
LOS ANGELES
1722 Eye Street. Nt.W.
Washington. D.C. 20006
Telephone 202 736 8000
Facsimile 202 736 8711
Founded 1866
NEW YORK
LONDON
SINOAPORE
TOKYO
WRITER S DIRECT NUMBER April 11, 1997
(202) 736-8271
, ¦ i .327
Mr. Douglas Tomchuk
U.S. Environmental Protection Agency _ ; •
Region II
290 Broadway, 20th Floor
New York, New York 10007-1866
RE: DEIR Comments
Dear Mr. Tomchuk
General Electric Company ("GE") is pleased to submit these comments on EPA's February
1997 Data Evaluation and Interpretation Report ("Report") for the Hudson River PCBs
Reassessment Remedial Investigation/Feasibility Study.
Recent technical meetings between the Agency and its contractors and GE and its
contractors have been extremely helpful and productive in clarifying and resolving critical
issues. Our comments would have been more extensive but for this dialogue. We believe there
needs to be an ongoing dialogue to both share and test various technical positions developed as
the reassessment continues.
The reaction of the public to the release of the Report - and particularly to its Executive
Summary - suggests that it was taken to be a completed Remedial Investigation ("RI"). Of
course, it does not meet that description. There are important conclusions that are stated so
broadly that they are not helpful in the context of remedial analysis. For instance, a major source
of PCBs at the Thompson Island Dam is identified as the sediments of the Thompson Island
Pool, but the particular class or category of sediments is not identified as it must be for useful
analyses in the remedial context. Other conclusions focus on one aspect of an issue while
neglecting other equally or more important aspects. The remedial importance of dechlorination
is limited to discussion of reduced PCB mass, ignoring reduced toxicity and bioaccumulation.
The half of the annual load of PCBs which is perceived to originate in the Thompson Island Pool
sediments is analyzed at length, the half originating near the Hudson Falls plant site is given
short shrift. Data issues essential to remedial analyses are left unaddressed or glibly brushed
-------
—Sidley 8c Austin
Washington, D.C.
Mr. Douglas Tomchuk
April 10, 1997
Page 2
aside: what is the present and future load coming from recent releases near the Hudson Falls site
and what is the fate of such releases in the Thomspon Island Pool?
In short, despite the very considerable effort that clearly has gone into the development of
this Report, a great deal of very important work remains to be done. Most important,
interpretation of data must be tested in the context of rigorous fate and transport modeling which
will constrain interpretations with consistent and plausible mechanisms.
We are continuing to review the Report and encourage the Agency to consider any
additional comments we may have in the context of our continuing dialogue. We believe that
with ongoing cooperation and exchange of data analyses and modeling approaches, the final
outputs can result in an RI that forms a sound basis for selecting and testing remedies that are
based on the realities of PCB fate and transport in the Upper Hudson.
Please place a copy of these comments in the Administrative Record for the site.
cc: Richard Caspe (USEPA Region II)
John Cahill (NYSDEC)
Paul Simon (USEPA Region II)
Ann Rychlinski (USEPA Region II)
Frank Bifera (NYSDEC)
Ronald Sloan (NYSDEC)
William Ports (NYSDEC)
Walter Demick (NYSDEC)
David Sterman (NYSDEC)
Steven Hammond (NYSDEC)
Jay Field (NOAA)
Ann Secord (USFWS)
A1 DiBernardo (TAMS)
Victor Bierman (Limno Tech)
Charles Menzie (Menzie-Cura)
Jon Butcher (Tetra Tech)
Sincerely,
FOR GENERAL ELECTRIC COMPANY
igus Macbeth
-------
COMMENTS OF GENERAL ELECTRIC COMPANY ON
PHASE 2 REPORT - REVIEW COPY
FURTHER SITE CHARACTERIZATION AND ANALYSIS
VOLUME 2C -
DATA EVALUATION AND INTERPRETATION REPORT
HUDSON RIVER PCBs REASSESSMENT RI/FS
FEBRUARY 1997
April 11, 1997
Melvin B. Schweiger
John G. Haggard
General Electric Company
Corporate Environmental Programs
I Computer Drive South
Albany, New York 12205
(518) 458-6646
Mahwah, NJ 07430
(201) 529-5151
1 Lethbridge Plaza
John Connolly
HydroQual, Inc.
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
TABLE OF CONTENTS
Pass.
I. INTRODUCTION AND EXECUTIVE SUMMARY I
II. NEITHER DIFFUSION OF PCBS FROM AGED SURFACE SEDIMENTS NOR
RESUSPENSION OF HIGHLY DECHLORINATED, HIGHLY CONCENTRATED
PCBS CAN ACCOUNT FOR THE PCBS THAT PASS THE THOMPSON ISLAND
DAM AT LOW FLOWS 10
A. Diffusion of PCBs from surface sediments deposited in the Thompson Island Pool
before 1991 cannot account for the increase in PCB load across the Thompson Island
Pool 11
B. Resuspcnsion of highly dechlorinated, highly concentrated PCBs from the Thompson
Island Pool sediments is implausible and cannot account for the increase in PCB load
across the Thompson Island Pool 13
1. Resuspcnsion of dechlorinated PCBs is inconsistent with the Thompson
Island Pool bathymetry 14
2. Resuspcnsion of dechlorinated PCBs is inconsistent with the minimal
resuspension of PCBs in the Thompson Island Pool at low flows 16
3. The congener fingerprint of the PCBs at the Thompson Island Dam shows
that dechlorinated PCBs are not the source; rather the source is likely
relatively unaltered Aroclor 1242 18
III. THE PCB LOAD AT HIGH FLOW ACCOUNTS FOR MORE THAN ONE THIRD
OF THE ANNUAL LOAD OF PCBS THAT PASS THE THOMPSON ISLAND DAM
AREA. THIS LOAD ORIGINATES NEAR GE'S HUDSON FALLS PLANT SITE
22
A. There is no dispute that discharges from the vicinity of the Hudson Falls site
contributed at least half of the annual PCB load to the Thompson Island Dam and
downstream areas in the form of undechlorinated Aroclor 1242 22
B. GE has undertaken extensive remedial work at the Hudson Falls site but the extent
of the source reduction and control is presently unknown 23
i
-------
TABLE OF CONTENTS
(Continued)
Page
DURING BOTH HIGH AND LOW FLOW, A SUBSTANTIAL PORTION OF THE
PCBS PASSING THE THOMPSON ISLAND DAM ORIGINATES FROM NEAR
THE HUDSON FALLS PLANT SITE 25
A. At low flow, there is a PCB load at the Thompson Island Dam that is unaccounted
for by the sum of the measured load entering the Thompson Island Pool and the load
that would be diffused from aged Thompson Island Poo] sediments 26
1. Calculation of unaccounted-for load by mass balance 26
B. The fingerprint of the unaccounted-for load at the Thompson Island Dam indicates
the likely source is undechlorinated Aroclor 1242 30
C. The PCB fingerprint in the TIP fish is consistent with PCBs recently entering the
river above the Thompson Island Pool 31
D. The behavior of the unaccounted-for load is consistent with the Allen Mill collapse
and seeps from the bedrock 34
E. The monitoring at Rogers Island appears to understate the PCB load entering the
Thompson Island Pool .36
1. Flushing of DNAPL during high flow is likely to escape detection at Rogers
Island ~ 37
2. Hydro plant operation likely causes flushing from Baker's Falls plunge pool.
39
3. GE is working to resolve the Rogers Island measurement issue 40
THE CONTRIBUTION OF PCB SOURCES DOWNSTREAM OF THOMPSON
ISLAND DAM MUST BE RECOGNIZED AND QUANTIFIED 42
A. PCBs passing the TID are decreased downstream by volatilization and deposition
42
B. In the Freshwater Lower River, external sources contribute significantly to the PCB
load 46
ii
-------
TABLE OF CONTENTS
(Continued)
Page
VI. SEDIMENTATION AND DECHLORINATION ARE IMPORTANT REMEDIAL
PROCESSES 50
A. Dechlorination is an important mechanism in reducing the bioaccumulation and
toxicity of PCBs 51
B. The Report's Dechlorination Indices are Flawed and Insensitive 52
C. Dechlorination Occurs at Concentrations Less Than 30 ppm 54
D. Dechlorination Has Not Stopped in the Hudson River 56
VII. CONCLUSIONS AND RECOMMENDATIONS 59
iii
-------
LIST OF APPENDICES
Page
APPENDIX A
SUMMARY OF ANALYTICAL BIAS CORRECTIONS FOR
USGS AND GE DATABASE A-l
APPENDIX B
DNAPL SOURCES IN THE VICINITY OF HUDSON FALLS B-l
APPENDIX C
APPLICATION OF INTRA-HOMOLOGUE PEAK RATIOS TO
CHARACTERIZE PCB SOURCES C-l
APPENDIX D
BENEFITS OF PCB DECHLORINATION •. D-l
APPENDIX E
ADDITIONAL COMMENTS AND CLARIFICATIONS E-l
1. Dechlorination Pattern H/H' E-l
2. Partitioning E-l
3. Volatilization E-4
4. Analytical Issues E-5
5. Miscellaneous Issues E-9
iv
-------
LIST OF TABLES
Table 1. Surface sediment PCB reservoir depletion under 1993-1996 average TIP loading.
Table 2. Average PCB loading across TIP from 1993 to 1996.
Table 3. Magnitude and composition of the unaccounted-for summer PCB load from TIP.
Table 4. Magnitude and composition of the unaccounted-for monthly 1996 PCB load from
TIP.
Table 5 Information sources for homologue-specific parameters of the bioaccumulation
model of TIP.
Table 6. Exposure sources for homologue-based bioaccumulation model of TIP.
v
-------
LIST OF FIGURES
Figure 1. Areas of TIP with surface sediment PCB concentrations greater than 100 ppm.
Figure 2. Average 1984 sediment PCB profile for areas with surface sediment PCBs greater
than 100 mg/Kg.
Figure 3. Annual sediment scour depths required to achieve water column concentrations
necessary to maintain the TIP load.
Figure 4. Predicted cohesive sediment bed erosion in the TIP under low flow conditions.
Figure 5. Comparison of TID water column PCB signature with TIP deep dechlorinated
sediment.
Figure 6. Comparison of TID water column PCB signature with TIP surface sediments.
Figure 7. Molar dechlorination product ratio versus fractional molecular weight change relative
to Aroclor 1242.
Figure 8. Location of EPA sampling events relative to 1993 hydrograph.
Figure 9. Summer 1991 congener peak TIP PCB loading - a) water column load; b) calculated
diffusive sediment load; c) unaccounted-for load; d) negative congener loadings.
Figure 10. Summer 1991 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted-for load; d) sediment composition
required to yield unaccounted-for load.
Figure 11. Summer 1992 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted-for load; d) sediment composition
required to yield unaccounted-for load.
Figure 12. Summer 1993 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted-for load; d) sediment composition
required to yield unaccounted-for load.
Figure 13. Summer 1994 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted for load; d) sediment composition
required to yield unaccounted-for load.
vi
-------
LIST OF FIGURES
(Continued)
Figure 14. Summer 1995 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted for load; d) sediment composition
required to yield unaccounted-for load.
Figure 15. Summer 1996 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted for load; d) sediment composition
required to yield unaccounted-for load.
Figure 16. Comparison of PCB congener composition between average TIP surface sediments
and unaccounted-for TIP load (summer 1991 - 1996).
Figure 17. Comparison of PCB congener composition between average TIP deep dechlorinated
sediments and unaccounted-for TIP load (summer 1991 - 1996).
Figure 18. Steady state food web model simulation of TIP (exposure: realistic water, top 0-2 cm
of sediment, see Table 6).
Figure 19. Steady state food web model simulation of TIP (exposure: realistic water, top 0-5 cm
of sediment, see Table 6).
Figure 20. Steady state food web model simulation of TIP (exposure: realistic water, heavily
dechlorinated sediments, see Table 6).
Figure 21. Steady state food web model simulation of TIP (exposure: heavily dechlorinated
sediments and partitioned water, see Table 6).
Figure 22. Relationship between PCB congener dechlorination ratios and number of chlorines
per biphenyl in EPA Phase II high resolution sediment cores.
Figure 23. Probability distribution of PCB congener dechlorination ratio in 1993 NOAA fish
samples.
Figure 24. Annual average TIP region loading (1980-1996).
Figure 25. Cumulative PCB DNAPL oils collected from seep 13 (1996-1997).
Figure 26. PCB and solids transport during 1992 spring high flow.
vu
-------
Figure 27.
Figure 28.
.Figure 29.
Figure 30.
Figure 31.
Figure 32.
LIST OF FIGURES
(Continued)
PCB DNAPL transport study fluorescent particle mass balance.
Predicted average June-August 1993 Hudson River PCB loading profile (river mile
195- 155).
Predicted settling and volatilization components of average June-August 1993
Hudson River PCB loading profile (river mile 195-155).
Absolute amount of chlorines removed from PCBs as a function of concentration.
Intra-homologue PCB dechlorination peak ratio as a function of PCB concentration.
Intra-homologue PCB dechlorination peak ratio as a function of sediment depth for
core segments with total PCB less then 30 ppm.
viii
-------
COMMENTS OF THE GENERAL ELECTRIC COMPANY
TO THE UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
CONCERNING THE PHASE 2 DATA EVALUATION AND INTERPRETATION
REPORT FOR THE HUDSON RIVER PCBs SUPERFUND SITE (FEBRUARY 1997)
I. INTRODUCTION AND EXECUTIVE SUMMARY
The General Electric Co. ("GE") is pleased to submit these comments to the United States
Environmental Protection Agency ("EPA") on the "Phase 2 Report - Review Copy, Further Site
Characterization and Analysis, Volume 2C - Data Evaluation and Interpretation Report, Hudson
River PCBs Reassessment RI/FS" (February 1997) ("Report"). The Report presents EPA's analysis
of the sediment and water column data collected during Phase 2 of the Remedial Investigation ("RI")
of the Hudson River PCBs Superfund Site ("Site") Reassessment, as well as historical data collected
by GE, the United States Geological Survey ("USGS") and others.
The Report is part of EPA's Reassessment, which seeks to determine the sources, transport
and fate of PCBs in the Upper Hudson River for the purposes of remedial analysis. The Report
analyzes the data collected in the RI and presents conclusions concerning the sources, transport and
fate of PCBs in the Hudson River. If the Agency finds that the PCBs pose a risk that it believes
would be prudent to abate, the evaluation and interpretation of the data should aid in determining
where the PCBs come from, how they move through the river system, and how they leave the
system. This, in turn, should form the basis for addressing the remedial questions of what will
1
-------
happen to the system without further intrusion and whether any particular remedy would abate any
perceived risk more quickly than natural recovery.
Fundamental to these determinations is a technically sound analysis of the data. The Report
does not provide a persuasive account of the data, and GE respectfully submits that several primary
conclusions of the Report are incorrect as a result of flawed, incomplete or incorrect analyses. The
Report is notably unhelpful in answering the central questions that the Reassessment must address —
the questions of source, transport, fate, and remedial alternatives.
We start with the four major conclusions that the Agency drew from the data analysis and
set forth the basis for our disagreement:
1. The area of the site upstream of the Thompson Island Dam represents the primary
source of PCBs to the freshwater Hudson. This includes the GE Hudson Falls and
Fort Edward facilities, the Remnant Deposit area and the sediments of the Thompson
Island Pool. Report at E-2.
The Report describes the PCB load that passes the Thompson Island Dam ("TID") as moving
in "pipeline" fashion, with little or no loss of PCBs from the water column, to the freshwater Hudson
downstream of Troy. We disagree with this conclusion in two regards. First, the sediments
downstream of the TID contribute significantly to the water column load as measured at Waterford.
During low flow, this contribution is on the order of 33%. Second, external sources contribute to
the PCB load in the freshwater Hudson, particularly downstream of Troy. The Albany core analyzed
in the Report indicates that 15% to 25% of the PCB load found at Albany originated downstream
2
-------
of the TED and not in the aged sediments of the river. Both facts are inconsistent with the theory of
a conduit between the TIP and the freshwater Hudson. This correction is important. The remedial
analysis must recognize that elimination of the PCB load at the TID will not eliminate the PCB
loadings downstream of the TID.
2. The PCB load from the Thompson Island Pool has a readily identifiable homologue
pattern which dominates the water column from the Thompson Island Dam to
Kingston during low flow conditions (typically 10 months of the year). Report at E-3.
This conclusion suffers from two misleading omissions. First, the Report shows that
approximately 36% of the total annual PCB load passing the TID occurred in the two-month high-
flow period. Second, the high-flow load does not show the homologue pattern seen during low flow
and clearly originates upstream of the Thompson Island Pool ("TIP").
These additions are important. The remedial analysis must recognize that eliminating the
low-flow PCB load at the TID or eliminating the PCB load originating in the TIP will not eliminate
50% of the annual PCB load passing the TID. Only if conditions during the sampling period are
unrepresentative of present conditions will this not be the case. The Report suggests, without factual
demonstration from the data, that loading above the TIP has been substantially reduced since the
sampling period. To conduct a remedial analysis with confidence, one must know whether that is
in fact the case. This determination cannot be made until the effects of the remedial work GE has
conducted can be fully evaluated.
3
-------
3. The PCB load from the Thompson Island Pool originates from the sediments within
the Thompson Island Pool. Report at E-4.
This conclusion is too vague to be useful in the remedial context. The Report postulates that
the TIP load originates in highly dechlorinated sediments deposited before 1984 or in
undechlorinated sediments more recently deposited. In order to know which of these distinct
sediment classes might be a candidate for remedial action, one needs to know which class
contributes the PCBs to the TIP load. This is important because each class supports distinct
remedies. The Report is of no help on this issue.
More fundamentally, there is no evidence to support the conclusion that the TIP load
originates from PCBs in highly dechlorinated. highly concentrated sediments. No realistic
mechanism exists to resuspend these PCBs from the sediments into the water column, and the
congener pattern of these sediments does not match the pattern of the TIP load.
GE believes that the increase in PCB load across the TIP originates partially from the
surficial sediments of the TIP (particularly from PCBs deposited from upstream of Rogers Island
in the recent past) and partially from the PCB load that passes Rogers Island undetected and only
later detected up at the TID after reprocessing through the surface sediments. This limits the
contribution from the sediments to amounts congruent with mechanisms that transfer PCBs from
sediments to the water column; provides a more persuasive match to the congener pattern at the TID;
and takes into account the major releases to the river following the 1991 collapse of the Allen Mill.
4
-------
The source of the PCB load at the TID is of central importance to the remedial analysis. If
current and recent releases from upstream of Rogers island are the source of the load at the TID, the
remedial analysis must focus on sources upstream of Rogers Island. If the source of the load at the
TID is PCBs deposited in the sediments of the TP several years ago, the remedial analysis is likely
to focus on the sediments of the TIP.
4. Sediment inventories will not be naturally "remediated" via dechlorination. The
extent of dechlorination is limited, resulting in probably less than a 10 percent mass
loss from the original concentrations.
This conclusion focuses on the wrong issue, mass. Dechlorination reduces the potential
toxicity and bioaccumulation of the affected PCBs. Dechlorination will reduce the carcinogenicity
of the PCBs; it can reduce the toxic equivalency of the PCBs by more than 90%; and it will reduce
the bioaccumulation of the PCBs between 4 and 35 fold. Because of these effects, dechlorination
makes a very substantial contribution to remediation and must be considered in the food-web
modeling and risk assessment. Finally, the major conclusions fail to address sedimentation that
buries PCBs and effectively removes them from the food chain and is an important remedial process
in the dynamics of the river.
The arrangement of the Report's conclusions makes it difficult to grasp clearly EPA's view
of PCB sources, transport and fate in the Upper Hudson. We attempt here to array what we believe
are the Report's central positions in an order that reflects sequential movement in the river:
5
-------
1. During periods of low flow, PCBs enter the TIP from above Rogers Island and are
stored in the TIP, despite the fact that this stored PCB load averages approximately
one-third of those PCBs that pass the TID during low flow.
2. During low flow, PCBs from relatively undechlorinated, aged surface sediments
(which do not include PCBs entering the TIP in the immediate past) or PCBs from
dechlorinated and highly concentrated sediments deposited before 1984 are the
source of the PCBs passing the TID.
3. During low flow, the PCBs that pass the TID dominate the freshwater Hudson to
Kingston.
4. During high flow, PCBs originating upstream of Rogers Island pass through the TIP
and dominate the freshwater Hudson to Kingston.
5. Approximately 36% of the annual PCB load passes the TID during high-flow events.
The Report suggests that this load may now be substantially reduced as a result of
source-control remediation projects at GE's plant sites.
Apart from the points made in the analysis of the Report's major conclusions, this is not a
plausible account of the behavior of PCBs in the river for the following reasons:
6
-------
* The Report offers no persuasive account or explanation of the fate of the PCBs entering the
TIP from above Rogers Island during low flow.
* The Report ignores established mechanisms of deposition and volatilization in describing the
fate of PCBs below the TID.
* The Report treats similar reaches of the river in a dissimilar fashion. For instance, the report
claims that sediment conditions immediately above the TID significantly contribute PCBs
into the water column but similar sediment conditions below the TID do not.
* The Report assumes that the conditions observed in 1993 are representative of long-term
conditions in the river and ignores the atypical impacts of the large loading of PCBs to the
river in the 18 months following the collapse of the Allen Mill in September 1991.
In these comments, we offer our own account of the behavior of PCBs in the Upper Hudson,
based on our analysis of the data.
Our account makes the following improvements on the one provided in the Report:
* It incorporates plausible mechanisms for the movement of PCBs through the river.
* It treats portions of the river with similar conditions in a similar manner.
-------
* It recognizes the major known changes in PCB loading to the river over time.
* It addresses all the major processes, including sedimentation, at work in the river.
* It uses a more comprehensive array of data to test and constrain our evaluation of the data
and the conclusions derived therefrom.
These comments focus on testing the analyses and conclusions in the Report against these
benchmarks, and emphasize the importance to remedial analysis of the issues raised herein.
It is apparent that the most significant unresolved matter is the source of the PCB load at the
TID at low flow which cannot be accounted for by PCBs measured at Rogers Island and PCBs
expected to be diffused from the aged sediments of the TIP. GE has been and is engaged in the data
collection and evaluation essential to reaching a sound answer as to the source of the unaccounted-
for load. It is likely that undechlorinated Aroclor 1242 from the Allen Mill collapse and the bedrock
seeps near the GE Hudson Falls plant site has entered the TIP undetected, particularly during higher
flow events, and has been deposited in the Pool, contributing substantially to the unaccounted-for
PCB load at the TID. This interpretation takes account of the large-scale Allen Mill release,
addresses the fate of the PCBs measured at Rogers Island, avoids implausible mechanisms for PCB
mobilization within the TP and is supported by a variety of lines of additional evidence, such as the
match with the congener fingerprint of the PCBs at the TID and the congener fingerprint of PCBs
found in TIP fish.
8
-------
Downstream of the TIP it is also important to recognize the effects of deposition and
volatilization as well as external sources. While the size of the PCB load may not be substantially
altered between the TED and Troy, PCBs are lost and other PCBs added to the load from sediments.
In addition, the significant contribution of external sources, particularly in the tidal Hudson, must
be recognized, a point we have emphasized with the Agency in the past.
The importance of the high-flow load to remedial analysis must also be recognized and
addressed. The high-flow load described in the Report underscores the importance of determining
the magnitude and duration of PCB releases upstream of the TIP now and over the past several years.
Yet the Report is unable to address this significant issue due to lack of data. To answer the question,
GE has been engaged in collecting and analyzing data following the major remedial projects at its
plants.
In order to complete a technically defensible remedial analysis, EPA must develop a
consistent and physically plausible explanation of the data and then test that explanation against a
calibrated and validated model to ensure that the true sources of PCBs to fish, wildlife and humans
are identified. The explanation of the data needs to take into account all the processes at work in
the river that are relevant to remedial analyses. Where questions central to remedial analyses cannot
be answered with the data presently at hand, additional data must be obtained to resolve the issues
so that we can have full confidence in the conclusions reached on the basis of data interpretation and
evaluation.
9
-------
II. NEITHER DIFFUSION OF PCBS FROM AGED SURFACE SEDIMENTS NOR
RESUSPENSION OF HIGHLY DECHLORINATED, HIGHLY CONCENTRATED
PCBS CAN ACCOUNT FOR THE PCBS THAT PASS THE THOMPSON ISLAND
DAM AT LOW FLOWS.
The Report concludes that the PCBs passing the TID at low flow are a major source of PCBs
to the freshwater Hudson. Consequently, determining the source of these PCBs is essential to the
remedial analysis of the Reassessment.
The Report concludes that the sediments in the TIP provide most, if not all, of the PCBs
passing over the TID during low flow. The Report, however, does not specify what sediments are
believed to be the source of these PCBs. Whether PCBs are derived from surface sediments (0-1
cm), near surface sediments (0-8 cm), deep sediments, hot spots or other areas is critical to
understanding the ultimate source of PCBs to the water column in the TIP and downstream of the
TID. The Report hypothesizes two possible sediment sources: (1) porewater diffusion of relatively
undechlorinated PCBs at low concentrations in the surface sediments or (2) resuspension of
extensively dechlorinated PCBs deposited before 1984. Neither source can account for all the PCBs
at the TID for at lest three reasons. First, there is an insufficient mass of PCBs in the aged surface
sediment to account for the increased PCB load measured across the TIP. Second, there is no erosive
mechanism to resuspend a sufficient quantity of dechlorinated aged sediments to provide the
increased load across the TIP and, if such erosion had occurred, sediment bed elevations would be
very different from what has been measured. Finally, the congener pattern of the extensively
dechlorinated sediments does not match the pattern of the TID load.
10
-------
A more careful analysis shows that there is a load passing the TID at low flow which cannot
be accounted for by the loads measured at Rogers Island and diffusion from surface sediments in the
TIP. This unaccounted-for load is similar to undechlorinated Aroclor 1242 and is probably related
to PCB loadings from the vicinity of GE's Hudson Falls plant. The identification of and explanation
for this unaccounted-for load is perhaps the most fundamental difference between GE's and EPA's
view of what is happening in the TIP.
A. Diffusion of PCBs from surface sediments deposited in the Thompson Island Pool
before 1991 cannot account for the increase in PCB load across the Thompson Island
Pool.
One of the Report's hypothesized sources for the increase in PCB load across the TIP is the
diffusion of PCBs from porewater in aged surface sediments to the overlying water column. This
process can account for only a portion of the load of PCBs measured at the TID. A simple mass
balance calculation demonstrates that if porewater diffusion were providing all the PCBs apparently
coming from the TIP (the net increase in PCBs between Rogers Island and the TID), the reservoir
of PCBs in surface sediments (as measured in 1984) would be nearly depleted by now. The high
resolution cores, however, do not reflect any significant depletion of PCBs from these sediments.
In this mass balance calculation, we assume that the net increase in PCBs between Rogers
Island and the TID (the increased load across the TIP) comes from PCBs in the surface sediments
of the TIP. This calculation uses measured annual paired loadings from Rogers Island and the TID
11
-------
from 1993 to 1996 using corrected GE data (see Appendix A)1 and employs a conservative measure
of the active surface layer (0-8 cm). Our analyses indicate that the active surface layer is 0-5 cm.
The mass of PCBs in the surface sediments was estimated using the results of the Report's analysis
of the 1984 sediment data. The depletion of surface sediment PCB homologues was based on the
following calculation:
Inventory
Year in which the surface sediment reservoir is depleted = + 1984 (1)
Flux rate
The surface sediment inventory was computed as follows:
c
Surface sediment inventory = —-D-A-10"6 (2)
w
S
in which:
css = surface sediment PCB homologue concentration
ws = average specific weight for sediments in the Upper Hudson = 0.77 g/cm3 (Report at
4-30);
D = depth of the surface layer = 8 cm
A = area of TIP = 2,000,000 m2 (Report, Table 4-7)
The results of this mass balance calculation are presented in Table 1 and show that, by now,
all monochloro and dichlorobiphenyls would have been depleted from the surface sediments. This
1 We have not used the period from 1991-1993 in light of the unusually high loadings to the
River during this period resulting from the releases from the Allen Mill.
12
-------
is particularly significant since current water column measurements show a continuing source of
mainly mono- and dichlorobiphenvls from the TIP, the same congeners that would have been
depleted without continued loading from the upstream source. The significant reserves of PCBs
remaining in the surface sediments of the TIP in 19912 and in the samples collected by EPA indicate
that PCBs fluxed from the surface sediments must comprise a relatively minor component of the
total increase in PCB load across the TIP in the 1990s.
B. Resuspension of highly dechlorinated, highly concentrated PCBs from the Thompson
Island Pool sediments is implausible and cannot account for the increase in PCB load
across the Thompson Island Pool.
EPA's other hypothesis — that highly dechlorinated PCBs at concentrations greater than 120
ppm and deposited prior to 1984 are the source of the increased PCB load across the TIP — is also
implausible. The mass of PCBs in these sediments is insufficient to maintain the increased load of
PCBs across the TIP during low flow, and there is insufficient erosion in the TIP to expose and
resuspend such sediments. In any event, the composition of PCBs in these sediments does not match
the TID load on a congener basis.
2 HydroQual has calculated that the average PCB concentration in surface sediments (0-5 cm) of
the TIP was about 30 ppm in 1984 and 20 ppm 1991.
13
-------
1. Resuspension of dechlorinated PCBs is inconsistent with the Thompson
Island Pool bathymetry.
The Report's hypothesis that "dechlorinated" sediments may be the source of the increased
load across the TIP relies on the mechanism of "resuspension" to place these PCBs into the water
column. As resuspension is a surface sediment process, this hypothesis requires that the PCBs
originate from areas with surface sediments containing PCBs in sufficient concentrations to match
the homologue pattern of PCBs found at the TID. The Report concluded that these PCB
concentrations had to be greater than 120 ppm. Simple mass balance calculations, however, show
that the increased load could only have been provided by erosion to depths of over 75 cm on average.
Such erosion is implausible and unreasonable. Had it occurred, it would have resulted in changes
in sediment bed elevation levels between 1984 and 1991 that are not seen in measured levels.
To evaluate whether the resuspension of PCBs from these deposits is plausible, we made an
estimate of the surface area in the TIP containing PCB concentrations exceeding a conservative value
of 100 ppm (A100). The kriging analysis of the NYSDEC 1984 data presented in the Report provides
an estimate of approximately 69,000 m2 (Figure 1) of river bottom in the TIP where such deposits
occur. The average vertical profile of PCBs within these areas was constructed using results from
the 1984 NYSDEC cores (Figure 2). The average TIP loading increase over the 1993 and 1996
period would require an estimated loading from the sediments of 254 kg/yr since 1984 (W^) (Table
2).
14
-------
The following equation provides a calculation of total sediment mass scoured from these
areas on an annual basis:
w V,
pcb
Wsed = 7T^ (3)
C
100
where Wsed is the mass of sediment loading required on a yearly basis and Qoo is the actual average
surficial sediment PCB concentrations in areas with concentrations greater than 100 ppm. The depth
of scour required to achieve these PCB loading estimates can then be calculated as follows:
w „
D = !*- (4)
' Ps Aioo
where D, is the depth of scour in year t (cm) and ps is the bulk density of the sediment (g/cm3).
Both C,oo and ps vary as simulated scour removes surface sediments and exposed sediments of
varying PCB concentration and bulk density.
Figure 3 presents the depth of scour required on an annual basis to achieve the concentrations
necessary to maintain the TED load. As can be seen, approximately 75 cm of sediments within these
areas would have to have been eroded between 1984 and 1993 to provide the increased load across
the TIP. EPA's analysis of its geophysical data indicates that such massive scour has not occurred.
Indeed, these data indicate that the aged dechlorinated sediments are largely intact (report at 4-91).
15
-------
2. Resuspension of dechlorinated PCBs is inconsistent with the minimal
resuspension of PCBs in the Thompson Island Pool at low flows.
The Report's hypothesis that highly dechlorinated surface sediments are the source of the
increased load across the TIP is implausible because it requires significant erosion of these sediments
during low flow, when such erosion is known not to occur.
Laboratory and field studies on the resuspension properties of cohesive sediments from the
TIP show that a critical shear stress exists below which erosion does not occur (HydroQual. 1995).
Similarly, resuspension of non-cohesive sediments will begin once the bottom shear stress exceeds
a certain critical value, which is typically greater than the cohesive critical shear stress (van Rijn,
1984). Bed armoring processes, in both cohesive and non-cohesive bed areas, will also limit the
amount of sediment eroded at a particular flow rate (Karim and Holly, 1986; Rahuel et al., 1989;
Ziegler and Connolly, 1995). The result of these well-established sediment processes and observed
measurements is that sediment resuspension under low flow conditions is very limited, and once the
flow rate is below a particular value, no resuspension occurs.
Sediment transport studies in various riverine systems have shown that the concept of no or
negligible erosion of the sediment bed during low flow conditions is valid. An effective method for
quantitatively evaluating resuspension and deposition processes in a river is to use a calibrated and
validated sediment transport model to predict solids fluxes across the sediment-water interface under
various flow conditions. A sediment transport model of the TIP has been developed, calibrated and
validated by HydroQual for GE. This model has also been successfully used by EPA in
16
-------
contaminated sediment studies of the Fox River in Wisconsin (Gailani et al.. 1991), Saginaw River
in Michigan (Cardenas et al.. 1995) and Buffalo River in New York (Gailani et al., 1996), in addition
to other riverine applications by HydroQual (Ziegler and Nisbet, 1994, 1995). These past studies
have shown that this model, if properly calibrated, can simulate sediment transport processes with
sufficient accuracy to use it as a diagnostic tool to study resuspension and deposition fluxes in the
Upper Hudson River under low flow conditions.
The sediment transport model was thus used to predict resuspension and deposition fluxes
in the TIP under low flow conditions. Model simulations show that negligible gross resuspension
of the cohesive sediment bed occurs for flow rates less than 5,000 cfs (Figure 4). To further
investigate resuspension and deposition dynamics during low flow conditions, a simulation was
performed for a constant flow rate of 3,200 cfs at Rogers Island. This flow rate corresponds to the
mean value at Fort Edward during the EPA sampling periods in May and June of 1993. The data
from this period were the primary basis for the low-flow resuspension hypothesis proposed in the
Report. Model results show that net deposition, with only a minimal amount of resuspension, occurs
in the TIP at this flow rate. The gross resuspension flux in the TIP is about 1,000 times smaller than
the gross deposition flux, and the sediment that is eroded comes from a thin, surficial layer
(approximately 10 |im thick) that is composed of loosely-consolidated, recently-deposited sediment
(Figure 4). These results, combined with the observed resuspension properties of sediments,
demonstrates the implausibility of the Report's hypothesis that relatively high gross deposition and
resuspension rates, producing a small net change in water column solids load, caused the observed
changes in PCB loading through the TIP during the 1993 low-flow periods.
-------
3. The congener fingerprint of the PCBs at the Thompson Island Dam shows
that dechlorinated PCBs are not the source; rather the source is likely
relatively unaltered Aroclor 1242.
The Report relies on homologue fingerprinting, combined with its two "dechlorination"
indices, to identify pre-1984 sediments as a possible source of the increased load of PCBs across the
TIP. There are two significant problems with the Report's analysis. First, while a crude composition
match of PCB TIP load to dechlorinated, aged sediments on a homologue basis can be made, a much
better match can be made on a congener basis or with the relatively undechlorinated PCBs in surface
sediments. Second, the Report relies on a simplistic "geochemical" analysis that purports to show
that all PCBs within the freshwater portion of the Hudson River can be derived from a mixture of
fresh and biologically dechlorinated Aroclor 1242. This analysis fails to recognize that partitioning
can also account for changes in PCB composition observed in the river, leading to a better
understanding of the PCBs seen at the TID.
For the first point, we have examined the similarity between the congener composition of
the total water transect samples and the sediment samples that represent potential sources to the
water. The TIP high-resolution core slices with more than 100 ppm of PCB were used to represent
the more highly contaminated and dechlorinated sediments of the TIP. The surface slices of the TIP
high-resolution cores were used to represent the surface layer. The percent weight of each congener
in each total water column sample at TID was plotted against its average percent weight in the
dechlorinated sediments (Figure 5) and against its average percent weight in the surface sediments
(Figure 6). In these figures all congeners that were not detected were placed on the axes. In both
18
-------
cases, there is a positive relationship between water and sediment. However, the scatter about the
relationships is much greater for the dechlorinated sediments. That is, the congener composition of
the water column is more similar to relatively undechlorinated surface sediments than to more
contaminated dechlorinated sediments. Accordingly, contrary to the Report's hypothesis, the TID
load cannot originate from the more contaminated, dechlorinated TIP sediments.
To the second concern, when partitioning between solid and dissolved phases is incorporated
into the analysis, as can be seen from a graphical analysis, the dechlorination indices ("indices") used
in the Report indicate that particulate and dissolved PCBs sampled at the TID originate from
different mixes of sources, and both include significant contributions of relatively unaltered Aroclor
1242.
Index values for sediment samples, dissolved and particulate water column samples, and
Aroclor standards are plotted in Figure 7, similar to the index plots provided in the Report. The
index values of the particulate PCBs sampled at Rogers Island and at the TID (filled squares and
circles on Figure 7) center on Aroclor 1242 (large open square symbol at the position (0.14,0.00),
demonstrating that the source of these PCBs is unaltered or very slightly altered Aroclor 1242.
Partitioning of PCBs between the solid and the dissolved phases can alter the composition
of PCBs and therefore the values of the indices. To demonstrate this, the original Aroclors (large
open square symbols) and the dissolved material that would result from partitioning from original
Aroclor sorbed to particulate material (large open triangles) are plotted on Figure 7. The
-------
composition of dissolved partitioned material was computed using partition coefficients derived
from data set forth in the Report. Computed dissolved partitioned Aroclor samples are always to the
right and above the original Aroclors. In addition, the dissolved partitioned Aroclor 1242 and 1016
samples lie close to the scatter of sediment samples, demonstrating that in the Upper Hudson, simple
partitioning can result in dissolved water column samples with index values similar to dechlorinated
sediments. That is, when the indices employed in the Report are applied to water samples, they do
not necessarily characterize dechlorination.
Computed dissolved PCBs partitioned from dechlorinated sediments also appear within the
scatter of the sediment data, to the right of the sediment source. For example, the large inverted
triangles represent the average aged surface sediment concentration in the TIP computed from the
1991 GE data (0-5 cm depth; filled: sediment, open: computed dissolved partitioned material). A
comparison of the positions of computed partitioned Aroclor 1242 and 1991 TIP surface sediments
suggests that the degree of dechlorination of the source material for dissolved PCB samples can be
characterized from the position of a dissolved sample on this plot. However, in contrast to the
method used in the Report, the measured dissolved samples must be compared with the computed
positions from fresh and dechlorinated particulates.
The dissolved PCBs sampled at Rogers Island are located within the scatter of the sediment
data, slightly to the right of their respective particulate samples. In general, the Rogers Island
samples do not lie as far right as partitioned Aroclor 1242 and lie slightly below dissolved
partitioned Aroclor 1254. This suggests that dissolved material at Rogers Island may be a
-------
combination of partitioned Aroclors 1242 and 1254 and that this material does not originate from
dechlorinated sediment.
The dissolved PCBs sampled at the TID are also located within the scatter of the sediment
data, between the computed dissolved materials partitioned from the 1991 surface sediment data and
from Aroclor 1242. Thus, dissolved PCBs at TID are not likely to originate entirely from
dechlorinated sediments; they are more likely to originate from a combination of sources that on
average is less dechlorinated than the partially altered 1991 surface sediments. As a result, unaltered
or very slightly altered Aroclor 1242 must be an important contributor to dissolved PCBs at the TID.
a fact that the Report does not recognize.
21
-------
III. THE PCB LOAD AT HIGH FLOW ACCOUNTS FOR MORE THAN ONE THIRD
OF THE ANNUAL LOAD OF PCBS THAT PASS THE THOMPSON ISLAND DAM
AREA. THIS LOAD ORIGINATES NEAR GE'S HUDSON FALLS PLANT SITE.
A. There is no dispute that discharges from the vicinity of the Hudson Falls site
contributed at least half of the annual PCB load to the Thompson Island Dam and
downstream areas in the form of undechlorinated Aroclor 1242.
The Report's analysis of water column PCB monitoring data collected during the 1993 spring
high flow event showed that an estimated 250 kg of PCBs originating upstream of the Roger's Island
monitoring station were transported through the system. This represents approximately 36 percent
of the total PCBs which passed TID for the entire year and all the TID load at high flow. This PCB
load differs from the loading from the TIP in that it is largely in the particulate matter phase and
exhibits a PCB congener distribution of non-dechlorinated Aroclor 1242, similar to that observed
within the Allen Mill and entering the river from bedrock fractures.
More specifically, the spring high flow event data collected by the EPA demonstrate that a
major portion of the total PCB loadings to the system originate from the Hudson Falls plant site area
and occur over a very short time frame. Such time variable PCB loading from Hudson Falls is
further supported by the observation that the flow weighted average PCB loading from upstream of
Roger's Island was approximately 50% of that measured during a transect monitoring event collected
under similar flow conditions. The Report attributes this difference to the variable loading dynamics
associated with the GE Hudson Falls source. That is, short term PCB loadings captured by the
transect sampling were not present in the flow-averaged event samples. Moreover, as the transect
4 data were collected after the river reached an initial peak of approximately 20,000 cfs (Figure 8),
22
-------
we believe that the EPA data underestimates the actual loading from the Hudson Falls Plant site
area. This is because we believe that PCBs are mobilized at lower flows along the rising limb of the
spring event hydrograph than that sampled.
Although the EPA water column data suggest that the spring high flow event loading passes
through the system, the EPA high resolution sediment cores show recent deposition of non-
dechlorinated Aroclor 1242 similar in composition to that mobilized during the spring high flow
event. These observations are consistent with current understanding of sediment depositional
processes: sediments are generally deposited during elevated flow events. This is significant
because it suggests a link between the Hudson Falls Plant Site area loadings and surface sediment
PCB levels, which control the sediment water column interaction and biota PCB exposure levels.
B. GE has undertaken extensive remedial work at the Hudson Falls site but the extent
of the source reduction and control is presently unknown.
In September 1991, elevated river water levels of PCBs were detected by GE upstream of
the TIP. Intense subsequent investigations localized the source area to the eastern shoreline near
river mile 196.8, in the vicinity of the GE Hudson Falls plant site. The results of these investigations
revealed the presence of active seeps of Dense Non-Aqueous Phase Liquids (DNAPL) along the
eastern cliff face and the rock face of the eastern raceway within the Allen Mill. In addition, free
phase PCB oil (Aroclor 1242) and oil contaminated sediments (up to 70,000 ppm) were found within
the Mill and the tailrace tunnel.
23
-------
A number of different remedial measures have been implemented to mitigate the seepage of
PCBs from the vicinity of the plant site to the River: 1) DNAPL seepage from the rock face of the
eastern raceway is now routinely captured; 2) hydraulic control of conduits through the Mill was
achieved in 1993; 3) a slurry wall was constructed within the eastern raceway in 1994 to reduce
seeps from this region; 4) removal of DNAPL and oil-contaminated sediments from the Allen Mill
containing 50 tons of PCBs was completed in 1995; 5) DNAPL-recovery wells have been installed
in the vicinity of the plant site that have recovered more than 8000L of DNAPL to date; and 6)
barrier wells utilizing hydraulic control to further reduce DNAPL transport through subsurface
fractures are being installed. These remedial efforts have reduced the PCB loading of
undechlorinated Aroclor 1242 to the Hudson River, but it is not yet possible to determine the degree
of control that has been achieved or to predict how much PCB is still likely to enter the River.
24
-------
IV. DURING BOTH HIGH AND LOW FLOW, A SUBSTANTIAL PORTION OF THE
PCBS PASSING THE THOMPSON ISLAND DAM ORIGINATES FROM NEAR
THE HUDSON FALLS PLANT SITE.
The Report demonstrates that during the Phase 2 sampling period. Aroclor 1242 recently
entering the river from the area of the Hudson Falls plant site makes up almost all of the PCB load
at the TIP during high flow and that high flow loads make up more than one third of the annual PCB
load at the TIP. The data show that this Hudson Falls source also contributes a substantial portion
of the PCB load at the TIP during low flows. First, some PCBs are measured at Rogers Island
entering the TIP during low flow. Next, mass balance shows that a PCB load passes the TID during
low flow that cannot be accounted for by the sum of the load entering the TIP and the load attributed
to diffusion from the aged sediments of the TIP as they were prior to 1991. Next, the PCB congener
fingerprint of this unaccounted-for load indicates that its likely source is undechlorinated Aroclor
1242. The PCB fingerprint of the fish in the TIP corroborates this, also indicating an
undechlorinated Aroclor 1242 source. The fluctuation from year to year of the unaccounted-for load
at the TID since 1992 clearly suggests a close connection between the releases from the Hudson
Falls site area and the magnitude of the unaccounted-for load. Monitoring results at Rogers Island
do not reflect a PCB load congruent with identifying the Hudson Falls sources as the origin of the
unaccounted-for load; but EPA's 1993 data, 1992 GE data, and a comparison of load behavior in
1995 and 1996 all indicate that unaltered fresh PCBs are flushed into the TIP as flows rise at Hudson
Falls. Such time-variable flows are likely to pass Rogers Island undetected by a weekly sampling
program. GE is presently collecting data to evaluate this proposition.
25
-------
A. At low flow, there is a PCB load at the Thompson Island Dam that is unaccounted
for by the sum of the measured load entering the Thompson Island Pool and the load
that would be diffused from aged Thompson Island Pool sediments.
1. Calculation of unaccounted-for load by mass balance.
The data show a PCB load at the TID that is not accounted for by the sum of the measured
PCB load entering the TIP and the load that would be diffused from the aged sediments. The
existence and magnitude of the unaccounted-for load at the TID can be determined by subtracting
from the water column PCB load at the TED (1) the load attributable to sediment porewater diffusion
and (2) the water column load of PCBs entering the TIP at Rogers Island. The diffusive fluxes of
PCB congeners from the surface sediment were calculated using GE's 1991 surficial sediment PCB
data and EPA's equilibrium partitioning concepts, including the use of temperature-corrected
partitioning coefficients. The water column PCB load increase across the TIP was calculated by
subtracting water column PCB loads at Rogers Island from those at the TID using paired Rogers
Island and TID water column samples from the corrected GE database (Appendix A) for 1991-1996
and daily average USGS flow measurements.
The following mass flux equation calculates the diffusive flux of individual PCB congeners
from TIP sediment porewater to the water column:
J = K A (C' - C ) (5)
s f s v d wc' ' '
where Js is the diffusive mass flux of individual PCB congeners (kg/day), Kf is the sediment/water
exchange coefficient (m/day), As is the sediment surface area of the TIP (m2), Cd' is the mean
26
-------
surfkial sediment porewater PCB concentration (kg/m3) calculated from the 0-5 cm section of
sediment cores collected in 1991, and Cwc is the water column PCB concentration (kg/m3).
Total dissolved porewater PCB concentrations (Cd') contain two components: freely
dissolved PCBs (Cd) and that adsorbed onto dissolved organic carbon (Cdoc):
Cd = Cd + Caoc <6)
Freely dissolved PCBs are in equilibrium with PCBs sorbed to sediment organic carbon. This
relationship is described by:
C
c, = — (7)
d f K '
oc oc
where Cs and are the mean surficial sediment (0-5 cm) PCB concentration (mg/kg) and fraction
organic carbon calculated from the 1991 sediment survey data, respectively, K0[. is the organic
carbon-based PCB partition coefficient (L/kg) calculated using EPA water column partitioning data
and corrected for temperature using temperature correction functions appearing in Appendix A. Cdoc
is in equilibrium with Cd and can be calculated as;
C = C m KJ (8)
doc d dot doe * '
where mdoc is the porewater dissolved organic carbon concentration (mg/L) calculated as the mean
surficial sediment (0-5 cm) TIP dissolved organic carbon measurements from the 1991 sediment
survey, and (L/Kg) is the equilibrium constant describing partitioning between freely dissolved
27
-------
PCBs and PCBs adsorbed to dissolved organic carbon, which was assumed equal to 0.1 Koc.
Substituting Equation 7 and 8 into Equation 6 yields the following expression for porewater PCB
concentrations:
C = (1 + m K )
d doc doc £ j£
c
s
(9)
Using Equations 5 and 9, the sediment diffusive flux equation becomes:
J = K A
V
wc
(10)
Equation 10 allows calculation of sediment diffusive flux of PCB congeners from known sediment
PCB congener concentrations using principles of equilibrium partitioning.
The sediment water exchange coefficient (Kf) can be estimated by substituting the water
column PCB congener flux estimates for the summer low flow period of 1991 into Equation 10 and
adjusting Kf to minimize the sum of differences in individual PCB congener loading estimates.
Negative PCB congener loadings due to detection limits for the higher PCB congeners in water
column samples are disregarded (Figure 9; panel d).
28
-------
Based upon 1991 sediment PCB measurements including PCB congener concentrations, the
sediments of the TIP can contribute through diffusive mechanisms an estimated 200 to 300 g/day
ofPCBs to the water column for 1991 to 1996 (panel b; Figures 10 through 15)\ The PCB congener
distribution of this load is shifted toward the lighter end of the congener spectrum due to the
relatively low values for the lightly chlorinated PCBs.
When the actual PCB load increase across the TIP is calculated for periods after 1991, more
PCB is present than can be explained by the diffusion calculation. The water column PCB congener
load increase across the TIP (J^c) can be calculated for the summer low flow periods (June -August)
of 1991-1996 using the following expression:
J = Q (C -C ) (U)
we ri nd ri \ >
where Qn is the daily average flow recorded at the Rogers Island monitoring station on the day of
sampling, and C„d and Cn are the individual corrected PCB congener concentrations (Appendix A)
at the TID and Rogers Island monitoring stations, respectively.
} Sediment diffusion load varies due to differences in mean summer surface water temperatures
and their effect on K,,,..
29
-------
The unaccounted-for load at the TID (JJ can be calculated on a PCB congener basis by
subtracting the estimated diffusive flux of PCBs (Js) from the difference in water column PCB
loadings between the TID and Rogers Island as follows:
Ju = " J, (12)
The results of this calculation for the summer low-flow period (June-August) of 1991 to 1996 are
presented on a homologue basis in Figures 10 through 35 (Panel c). The 1992 unaccounted-for PCB
load was 559 g/day. This is twice the surface sediment diffusive flux calculated using the 1991
surface sediment PCB data. This increased to approximately 1000 g/day in 1993 and 1994 and
declined in 1995 to less than 250 g/day. In 1996 the unaccounted-for PCB load was approximately
1200 g/day.
B. The fingerprint of the unaccounted-for load at the Thompson Island Dam indicates
the likely source is undechlonnated Aroclor 1242.
The composition of this unaccounted for load indicates its probable source. On a homologue
basis, this unaccounted-for load was dominated by dichlorobiphenyls and generally resembled the
homologue mass loadings attributable to diffusion from surface sediments (Figures 10 through 15;
panel c). Assuming the unaccounted for PCB load is derived exclusively from a surficial sediment
diffusional process, one may calculate the expected mean surface sediment and porewater PCB
composition which would produce this load by diffusion.
30
-------
The sediment phase PCB homologue distributions required to produce the unaccounted for
PCB load were calculated on a yearly basis for the summer low flow period (June-August) for 1992
through 1996 and monthly for 1996. These results are summarized in Tables 3 and 4 and presented
as PCB homologue distributions in Figures 10 through 15, panel d.
As can be seen in these Tables and Figures, the homologue composition of the unaccounted-
for load closely resembles Aroclor 1242. The congener composition of the unaccounted-for load
is markedly similar to surface sediment PCB congener distributions (Figure 16) and deviates
considerably from deep dechlorinated sediments on the TIP (Figure 17). These data indicate that
the unaccounted-for load from the TIP originates from an undechlorinated Aroclor 1242 source.
C. The PCB fingerprint in the TIP fish is consistent with PCBs recently entering the
river above the Thompson Island Pool.
Analysis of PCBs in fish supports the conclusion that the source of the unaccounted-for load
at the TID is relatively undechlorinated PCBs, and the homologue and congener composition of the
fish provides a way to identify PCB sources to the food web.
Two analyses were performed to test whether fish body burdens are representative of PCBs
originating in sediments containing dechlorinated or relatively undechlorinated PCBs. First, using
the GE bioaccumulation model, the relationship between the homologue composition in fish and the
homologue compositions of surface sediments and the water column can be explored. One may
assume that varying degrees of dechlorination in exposure sources result in differing homologue
31
-------
distributions in the fish. These compositions can be computed and compared with the observed
composition in TIP fish in order to characterize the likely composition of the exposure sources.
Second, congener-based fingerprinting indices can be used to assess the similarity in PCB
composition between surface sediments and fish.
The total PCB bioaccumulation model developed by HydroQual was modified to compute
bioaccumulation on a homologue-specific basis. Table 5 identifies the sources of the parameter
values for individual homologues. To characterize the homologue composition of the exposure
sources, several model simulations were performed using exposure concentrations and distributions
derived from TIP data. Table 6 identifies the basis for the derivation of the exposure values.
Figures 18 through 21 computed and observed homologue distributions in fish. Model
results for pumpkinseed were compared with pumpkinseed data collected in the TIP after 1989.
Model results for largemouth bass were compared with largemouth and smallmouth bass data
collected in the TIP after 1989 using fish greater than 400 g. The dashed lines represent the spread
of the mean of the data +/- two standard errors.
The data best match pumpkinseed homologue distributions computed using the surface slice
of the high-resolution cores and the measured water column distribution (Figure 18). The
largemouth bass distributions computed using either surface slices, 0-2 cm (Figure 18) or the 0-5 cm
layer (Figure 19) best reproduce the observed distributions. As the degree of dechlorination in the
exposure sources increases, the model results diverge from the measured homologue distributions
32
-------
(Figures 18 through 21). Thus, the PCB composition of the fish is most consistent with a relatively
unaltered source.
Several indices of dechlorination have been developed (Appendix C). These are ratios
between the proportions of individual congeners in a sample. The numerator is the proportion of a
congener that is dechlorinated and the denominator is the proportion of a congener that is left
relatively unmodified by dechlorination. For each ratio, congeners within the same homologue
group are used, so that partitioning and bioaccumulation differences are minimized. The degree to
which these ratios indicate dechlorination can be tested by comparing their values to other indices
of dechlorination in sediments; for example, the number of chlorines per biphenyl. They are found
to be indicators of dechlorination in sediments. The ratios can be used in a diagnostic fashion to
compare samples collected from sediment, water and fish.
These ratios decline as dechlorination proceeds. For example, four of these ratios are plotted
against the number of chlorines per biphenyl (Cl/BP) using high-resolution core sediments from the
Upper Hudson (Figure 22). The dashed lines indicate values of the ratios and Cl/BP for Aroclor
1242. The ratios in the sediments approach the Aroclor 1242 value in samples in which Cl/BP
approaches the Aroclor 1242 value. In addition, values of the ratio decline in dechlorinated
sediments.
Figure 23 presents the values of these four ratios in fish collected from the TIP by NOAA
in 1993. Comparison of the fish ratios with the sediment ratios in Figure 22 indicates that slightly
33
-------
altered Aroclor 1242 is the source of PCBs to these fish. Similar results are obtained with fish
caught in earlier years.
D. The behavior of the unaccounted-for load is consistent with the Allen Mill collapse
and seeps from the bedrock.
Having established that the source of the unaccounted-for PCB load at the TID appears to
be relatively undechlorinated Aroclor 1242, possible sources for these PCBs must be explored. The
releases of DNAPL PCB oil from the Allen Mill and/or the bedrock seeps near GE's Hudson Falls
facility are the likely sources and the temporal parallels between the behavior of the unaccounted-for
PCB load and the sources near Hudson Falls suggests a close connection between the two.
The USGS and GE data sets provide a 20-year record of water column PCB data that can be
used to assess long-term spatial and temporal patterns of PCB loadings in the upper Hudson River.
Historical loadings from the TIP were estimated from the corrected USGS PCB data. Low flow
PCB loading from the TIP was calculated as the difference in PCB loadings based on paired flow
and PCB data at Schuylerville and Fort Edward during periods of flow less than 10,000 cfs at Fort
Edward from 1980-1989 (Figure 24).4 Loads for the period 1991-1996 were calculated using paired
flow and PCB data collected as part of GE's water column monitoring program.5
4 The USGS record did not contain data from the Fort Edward or Schulerville Station at flows
less than 10,000 cfs for 1990 - 1995. Large loadings calculated for 1977 - 1979 period were
excluded in this figure to highlight observed changes between the late 1980s and early 1990s.
5 GE collected data in 1991 to provide preliminary estimates of the bias in the USGS
methodology and to "correct" historical USGS water column PCB records (see Attachment 1).
34
-------
The PCB loads from the TIP declined steadily from approximately 1.2 lbs/day during the
early 1980s to an estimated 0.5 lbs/day in the late 1980s (Figure 24).6 These data document the
recovery of the system from the impacts of process discharges and the erosion of PCB-contaminated
sediments after the removal of the Fort Edward Dam in 1973. The principal processes contributing
to this recovery likely included the reduction of PCB deposition from sources within the remnant
reach of the River and the deposition of clean solids from the tributaries.
Summer low flow loadings in 1991, prior to the September 1991 Allen Mill collapse, were
comparable to the loadings calculated using the USGS data for late 1980s (Figure 24). In 1992,
low flow PCB loadings from the TIP increased by a factor of 4 from an estimated 0.5 lbs/day in the
late 1980s to approximately 2 lbs/day (Figure 24).
Thus, the unaccounted-for load from the TIP is temporally correlated with PCB releases from
the Allen Mill. Considering its estimated magnitude, the Allen Mill release is likely the cause of
this increased loading. PCBs discharged from the Mill were likely transported downstream and
deposited within the surficial sediments of the TIP and subsequently released to the water column
5 (continued...) The GE data have been corrected for the analytical bias in quantitation of peaks 5,
8, and 14, which contain coeluting congeners BZ#4 and 10, BZ#5 and 8, and BZ#15 and 18,
respectively.
6 PCB levels monitored at Schuvlerville were used to infer PCB levels at the TID for the periods
in which data were unavailable.
35
10.0389
-------
as a dissolved phase loading through the processes of partitioning between sediment and oils and
sediment porewater, with subsequent diffusion into the overlying water.
Under this scenario, loadings from the TIP should decline as the 1991-1993 loads are eluted
from the surficial sediment porewater, buried by the deposition of clean solids, dechlorinated and
transported out of the TIP. The unaccounted-for PCB load originating in the TIP declined from
approximately 1.0 kg/day in 1993 to less than 0.25 kg/day in 1995 (Figure 12 through 14, panel c).
This reduction suggests that the system was recovering from loadings from the Allen Mill between
1991 and 1993. The 1996 unaccounted-for PCB loadings increased from 1995 levels to
approximately 1.2 kg/day. This suggests that between 1995 and 1996 additional PCBs originating
upstream were deposited on surficial sediments in the TIP.
E. The monitoring at Rogers Island appears to understate the PCB load entering the
Thompson Island Pool.
It is clear that the unaccounted-for load of PCBs in the TIP resembles unaltered Aroclor
1242. The most likely source of this material is located upstream in the vicinity of Hudson Falls.
However, for this source to explain this unaccounted-for loading requires that the PCB pass the
monitoring station at Rogers Island undetected. Given that the PCBs from this source are dense,
nonaqueous phase liquids (DNAPL) and the limited monitoring, this is likely the case.
36
-------
First, the presence of DNAPL in the vicinity of the Hudson Falls Site is well established.
Numerous river bed seeps were exposed upon the dewatering of Baker's Falls when Adirondack
Hydro Development Corp.'s hydroelectric facility was being constructed along the western shore
of the Falls. Recent work at GE's Hudson Falls plant has shown that PCB DNAPL oils are
transported through bedrock fractures and enter the Hudson River from river bed seeps adjacent to
the Hudson Falls plant site. These seeps represent a significant source of PCB DNAPL to the River.
At one of these seeps (Seep 13), discovered within the Baker's Falls plunge pool during an
underwater inspection by divers, an estimated 16 liters (22.5 Kg) of PCB DNAPL oils were collected
over a period of 3-1/2 months beginning in September 1996 (Figure 25).7 Second, the weekly
monitoring at Rogers Island will not detect events of limited duration that might mobilize PCBs for
this region of the river, such as high flow events.
1. Flushing of DNAPL during high flow is likely to escape detection at Rogers
Island.
Oil-phase loads from the Allen Mill and the bedrock seeps introduce indeterminate errors in
water column monitoring designed to measure particulate and dissolved phase PCB loads. The
behavior of DNAPLs within natural aquatic systems is not well understood. It is likely that DNAPL
oil droplets from bedrock fractures will behave in a manner similar to particles possessing the same
diameter and density and likely will settle onto the river bed during low flows. As flow velocities
7 Localized groundwater pumping efforts appear to have mitigated PCB losses through Seep 13
(Figure 25).
-------
along the sediment/water interface increase during periods of elevated flow, the DNAPL droplets
will become resuspended in the water column and will be transported downstream. Such
resuspension occurs almost instantaneously at the point when critical sheer velocities are reached
at the sediment bed surface. It is possible that oil discharges near the Hudson Falls Plant Site
accumulate within quiescent regions of the river adjacent to the site and are mobilized during high
flow events.
There is evidence for flushing of significant amounts of PCBs from the River in the EPA
high flow water column transect study. PCB loading from the region upstream of the TIP during
this event contributed approximately 18 Kg/day PCB, primarily in the particulate phase. This
particulate phase PCB transport occurred in the absence of sediment resuspension. These data
generally support the hypothesis that oil phase PCB loading may be occurring during periods of high
flow. EPA monitoring did not occur on the rising limb of the event hydrograph (Figure 8) and
probably missed the peak PCB load, because critical sheer velocities for oil resuspension were likely
reached prior to the EPA sampling. Similar high flow data collected by GE in 1992 showed elevated
PCB loadings from upstream of the TIP with little or no evidence of sediment scour (Figure 26).
Finally, neither the EPA water column monitoring program nor the GE high flow monitoring
program collected sediment bed load samples. Such loads are likely to represent a significant
portion of PCB DNAPL transport, particularly during periods of high flow. As observed in the EPA
transect data, particulate PCB loads from upstream are preferentially deposited within the TIP. This
38
-------
indicates surface sediment PCB contamination mechanisms which are consistent with the increased
loads following the Allen Mill collapse and DNAPL PCB seeps from the Hudson Falls area.
Seasonal patterns in the unaccounted-for PCB loading at the TID in 1996 also suggest that
PCB oils collect near the Hudson Falls Plant Site and are mobilized during high flow events.
Monthly calculations of the unaccounted-for TIP load for 1996 indicate that, prior to the April high
flow event, PCBs within the water column could largely be derived from sediment diffusive
mechanisms considering the 1991 surficial sediment PCB concentrations (Table 4). Following the
spring high flow period, an unaccounted-for load was apparent, varying in magnitude from 0.6 to
2.0 Kg/day during April through August and decreasing steadily to less than 0.2 Kg/day by October
(Table 4). This seasonal variability and correlation with spring high flow suggests that PCB
loadings from the vicinity of the Hudson Falls plant site are associated with high flow events. This
is supported by the observation that the unaccounted-for loadings in 1995, a low flow year in which
spring high flows never exceeded 15,000 cfs, were considerably lower than the 1994 and 1996
unaccounted-for loads.
2. Hydro plant operation likely causes flushing from Baker's Falls plunge pool.
Another possible method for PCBs to enter the TIP from sources in the vicinity of Hudson
Falls undetected is related to the operation of the newly constructed hydroelectric facility at Baker's
Falls. In front of the turbine intakes this facility has trash racks which are cleaned every few days.
39
-------
During the cleaning process, a water bypass structure is used which discharges significant amounts
of water into the plunge pool at the base of Baker's Falls. This has been observed to transform the
plunge pool from calm to turbulent.
Divers have observed PCB oils seeping into the plunge pool and accumulating on the river
bed. The flushing of water into the plunge pool during the trash rack cleaning is likely to move
PCBs downstream in pulses that would not be detected by the weekly monitoring at Rogers Island
unless coincidentally synchronized with the trash rack cleaning. In the Fall of 1996, one round of
monitoring was conducted to coincide with this cleaning and wc»intended to monitor the potential
movement of a pulse of PCBs downstream. PCB levels at Rogers Island increased from 15 ppt to
42 ppt. Since flow conditions are generally low during the cleaning, it is likely that PCB oil would
be deposited in to the pool and would only later be detected in "dissolved" form at the TID. GE is
conducting additional sampling this spring and summer to determine the importance of this transport
mechanism.
3. GE is working to resolve the Rogers Island measurement issue.
Preliminary results of the PCB DNAPL transport study conducted by GE in the Fall of 1996
j
(HydroQual, 1996) suggest that PCB DNAPL oils emanating from Hudson Falls are retained within
the reach of the river between Hudson Falls to the TID. This is based on the observation that only
2 percent of the fluorescent particles (selected to represent PCB DNAPL) were transported
40
-------
downstream of the TID (Figure 27). To the extent that these fluorescent particles simulated DNAPL
transport, these data suggest that DNAPL originating from Hudson Falls is largely retained within
the river upstream of the TID. Moreover, preliminary analysis of fluorescent particle size
distribution data indicates that particles with an average diameter of approximately 100 um constitute
the majority of the mass of particles retained within the Hudson Falls to Fort Edward reach of the
river. These larger particles were not transported downstream under the flow conditions during the
study (est 800-8000 cfs) suggesting that oil droplets with mean particle diameter of greater than
approximately 100 um are retained near Hudson Falls at flows less than 8,000 cfs and are likely
transported downstream during higher flow periods. Additional analysis of these data is underway,
and a full report will be submitted to EPA.
Additionally, as reported to EPA, GE is undertaking an extensive data collection program
focused on this potential loading mechanism. Due to the importance of the unaccounted-for TIP
loading in evaluating remedial options, it is imperative that the source be determined.
41
-------
V. THE CONTRIBUTION OF PCB SOURCES DOWNSTREAM OF THOMPSON
ISLAND DAM MUST BE RECOGNIZED AND QUANTIFIED
The Report concludes that PCBs in the water column are conservatively transported
downstream from the TID to the freshwater tidal Hudson with'little or no loss or gain. This implies
that sediments downstream of the TED or other external sources of PCBs, such as a point source or
tributary, are insignificant. This is inaccurate. Consequently, the significance of the load passing
the TID is overstated and the benefits of reducing that load will be overstated. A more careful
analysis demonstrates that sediments in the reaches of the Upper River below the TED are important
PCB sources and that the contribution of the PCBs passing the TID decreases downstream. In the
freshwater portion of the lower Hudson, EPA's own analysis shows that external sources of PCBs
contribute significantly to the sediment — for instance, 25 percent at Albany.
A. PCBs passing the TID are decreased downstream by volatilization and deposition.
The Report contends that the PCBs passing the TID are transported downstream from Reach
7 through Reach 1 and into the Lower River with little or no loss in PCB mass: the data indicate "the
occurrence of quasi-conservative transport of water column PCBs (i.e., no apparent net losses or
gains) throughout the Upper Hudson to Troy" (Report at E-3), and that the PCBs "pass relatively
unaltered, as through a conduit, through the length of the Upper River during winter and spring
conditions" (Report at 3-87). GE agrees that the net load of PCBs in the water column does not vary
markedly through these reaches of the upper Hudson River; however, we disagree that the region
42
-------
above the TID sets water column PCB concentrations and loads downstream of the TED to Kingston.
The Report's conclusion rests on an incomplete consideration of the physical transfer processes
which affect the fate and transport of PCBs in the Upper Hudson River.
The fate and transport mechanisms relied on in the Report to explain spatial patterns in PCB
loadings in the various reaches of the Upper Hudson River are inconsistent. The Report
hypothesizes various mechanisms to describe PCB dynamics within the TIP during the different
water column monitoring events. For example, sediment deposition, sediment resuspension,
porewater diffusion, and groundwater advection are cited as possible causes of changes in PCB
loading patterns across the TIP. In contrast, reaches downstream of the TID are described as a
pipeline in which upstream loads are transported downstream with very little sediment/water
interaction. Interpretation of the spatial and temporal patterns observed in the data should be
described from a consistently applied mechanistic perspective. There is no sound explanation for
why sediments within the TIP would be highly reactive while sediments downstream of TIP
containing similar PCB concentrations and subjected to similar physical forces would behave
differently. Moreover, the more plausible account for both the TIP and the downstream reaches is
that which the Report implies downstream of the TID: aged, dechlorinated PCBs in sediments
deposited several years ago make a very limited contribution to the PCBs found in the water column.
Several transport mechanisms change water column PCB loads in the Upper Hudson River,
including particulate settling, volatilization, dilution due to tributary solids, and inputs from local
-------
sediments. Deposition zones and several so-called PCB "hot spots" are found in a number of
locations downstream of the TID, indicating that settling of PCB-contaminated particles occurs in
this region. Volatilization of dissolved-phase PCBs will occur at all locations in the river and at all
times, with the transfer rate across the air-water interface varying spatially and temporally,
depending upon local conditions. Addition of tributary solids and flow will dilute water column
PCB concentrations in the main stem of the Upper Hudson; however, the additional tributary solids
will also reduce the PCB load due to partitioning of dissolved PCBs onto uncontaminated tributary
sediments and subsequent deposition of these solids. Tributary sediment loadings to the River occur
downstream of the TID (HydroQual, 1995), are significant, and result in reduced PCB transport.
These processes reduce water column PCB loads downstream of the TID. These losses are offset
to some degree by the addition of PCBs to the water column from the sediments in Reaches 7
through 1.
The impacts of fate processes can be evaluated in the context of a mass balance model.
EPA's report does not contain such an analysis. The GE model, however, has been used to examine
the fate processes acting upon PCBs in the upper Hudson River: transport with the river flow;
adsorption-desorption among dissolved, particulate and colloidal phases; settling and resuspension
of the particulate phase; diffusion between the water column and the surface sediment and within
the sediment; volatilization from the water column dissolved phase to the atmosphere; and burial of
sediment-associated PCBs through sedimentation. This model has been compared to water column
and sediment data over a 14-year period at locations throughout the Upper River. Based on the
-------
model's ability to reproduce these data using well-accepted descriptions of the fate processes, it can
be used to examine PCB dynamics within the system.
Figure 28 compares the three EPA estimates of loading through the Upper Hudson River for
the June to August 1993 period with the model simulation for the same period. The dashed-line
profile represents the conservative transport of PCBs that EPA claims exists from the TID to Troy.
The model shows a 19% loss of PCBs in the downstream direction that is the net result of
volatilization, net deposition of solids and sorbed PCBs, and the addition of PCBs to the water
column from downstream sediments and, to a lesser degree, by downstream tributary inputs. This
net loss is small enough to be within the uncertainty bounds of the data. The EPA data collection
in the summer of 1993 found an approximately 20% increase in the PCB load between the TID and
Waterford. This was a period during which the increase in the PCB load across the TIP reflected the
impact of the Allan Mill release. Conditions downstream of the TID may also have been influenced
by the same event.
Figure 29 provides further clarification of the origins and fate of the TID PCBs in the river
during this period. Here, the lower limit of the shaded region represents the profile of water column
PCB load without a sediment source in Reaches 1-7. PCB load in the river decreases by 45.1%
between Thompson Island Dam and Waterford. This decrease is a result of the combined effects of
settling (the unshaded region, 17.4%) and volatilization (the shaded region, 27.6%). Considering
the results of Figures 28 and 29 together, the sediments in Reaches 7 - 1 contribute 32.3% to the
45
-------
downstream load passing Waterford with upstream sources and TIP sediments contributing the
remainder.
B. In the freshwater Lower River, external sources contribute significantly to the PCB
load.
The Report uses a simple dilution model (PCB/137Cs) to estimate the contribution of Upper
River PCBs to PCBs in the sediments of the freshwater Lower River and to support its contention
that PCBs are conservatively transported from the TID to Kingston. The methodobgy is flawed
because it fails to account for the increasing solids yield from the drainage basin below the f ID and
the losses of PCBs to volatilization and deposition, as described in the previous section, and cannot
explain the variability of cesium data upstream of Stillwater. EPA should have recojnized these
deficiencies and abandoned this approach when it could not describe the change in PCB/U7Cs
between the TIP and Stillwater.
The Report's PCB/mCs model is based on an assumption of uniform distribution of cesium
in sediments throughout the Hudson River. While this uniformity of cesium inputs from tributaries
may be true at any given point in time (cesium levels are known to be decreasing over time), the
demonstration of this spatial uniformity using data on Figure 3-63 from the Report is wrong because
the cesium data upstream of Stillwater are inexplicably variable and frequently at levels much higher
than downstream. This is an instance where the Report's analysis is not consistently applied
throughout the river system. The dilution analysis did not work in the segment of the river upstream
46
-------
of Stillwater; therefore it was only employed downstream of Stillwater, ignoring what is arguably
the most important part of the River from a PCB source standpoint.
Further, the data downstream from Stillwater should not necessarily be expected to be
uniform, over a fixed depth interval This is because spatial variation in deposition rate would make
the sediments in this fixed 0-2 cm layer representative of different time periods at different locations,
and hence different cesium levels. Moreover, the presence of a mixed surface layer with varying
mixing depths at different locations would also complicate the assignment of sediments to a known
time period. These difficulties are compounded when PCB and the (PCB/'"Cs) are considered,
because again, the 0-2 cm layer which is assumed to represent sediment deposited between 1991 and
1992 is not necessarily representative of this period. The analysis also neglects the decrease in PCB
concentrations in the water column resulting from net deposition and volatilization from the water
column, processes which are partially offset by sediment sources of PCB. Accordingly, simple
dilution by tributary solids cannot account for all of the decrease in downstream PCB sediment
concentrations.
Significant increases in the solids load due to tributary inputs as one moves downstream also
invalidate the Report's simplified dilution analysis. The sediment yield, i.e., annual sediment load
per square mile of drainage area, increases by about a factor of five between Fort Edward and the
Federal Dam at Troy (Phillips and Hanchar, 1996). The large sediment load from tributaries in the
Upper Hudson River enhances the deposition of PCB-contaminated solids in the river and modifies
-------
the rate of change of the PCB/cesium ratio from that which would occur by simple dilution. Both
of these effects complicate the PCB/cesium analysis and undermine the Report's dilution hypothesis.
EPA used comparisons between the PCB composition in the high resolution core collected
at RM 177.8 near Stillwater and cores in the lower River to imply the contribution of upper River
PCBs to the PCBs in the lower River. The Agency characterizes this contribution as that of the
"combined TI Dam load" (report at. 3-120), although, as discussed earlier, the load from the Upper
River to the Lower River reflects contributions from sediments below the TIP as well. Using cores
collected at Albany and Kingston, the Agency concludes that the importance of the Upper River
PCB source has varied over time, being most important during the period between 1975 and 1981
and less important more recently. Comparison of the congener patterns in the top sections of the
Albany and Stillwater cores reveals differences that were attributed to the addition of other Aroclors
between Stillwater and Albany. Simple mixing of the Stillwater core composition with the PCB
compositions of Aroclors 1016 and 1260 was used to imply that about 22 percent of the PCBs in the
core were derived from non-Stillwater sources.
The EPA analysis of the Albany core probably underestimates the contribution of sources
other than the Upper River. Any Aroclor 1242 (the most widely-used mixture) entering the river
between Stillwater and Albany was attributed to the Upper River. Further, the composition of the
Albany core top is biased toward the upper River because of the high loadings from the upper River
that occurred in the fall of 1991 and summer of 1992. Thus, it is likely that sources other than the
-------
upper River contribute substantially to the PCBs in the sediments of the lower River. The
conclusions of the report fail to cite the evidence of other loadings that it found in its own analysis
and thus mischaracterized the importance of upper River sources to PCBs in the lower River.
49
-------
VI. SEDIMENTATION AND DECHLORINATION ARE IMPORTANT REMEDIAL
PROCESSES
The Report understates the importance and benefits of natural recovery processes, including
sedimentation and PCB dechlorination/biodegradation. These processes combine to reduce the
availability of PCBs to the water column and biota and reduce the toxicity and bioaccumulation of
the PCBs that remain. GE has explained to EPA the importance of sedimentation in previous
submittals (HydroQual, 1995, GE/HydroQual, 1996), and we will not repeat these here.
The Report fails to recognize dechlorination as an important risk-reduction process and this
failure is reflected in the Report's misplaced emphasis on mass rather than toxicity and
bioavailability. The importance of dechlorination is notmass reduction but its effect in reducing the
toxicity and bioaccumulation potential of PCBs. The indices used in the Report to measure
dechlorination fail to capture the complexity and variability of dechlorination processes and are
insensitive measures of dechlorination.
The Report also overlooks microbial PCB dechlorination in the sediment of the Hudson
River as a critical tool for distinguishing the source of PCBs to specific receptors. Dechlorination
produces unique, congener-specific changes in PCB congener distributions that permit precise source
identification, as is described in detail in Appendix C. Briefly, this Appendix identifies several peak
ratios that can "fingerprint" PCB sources in more than 30 species of Hudson River fish collected
during a 16-year period. This analysis shows that PCBs in the fish of the TIP have not undergone
-------
dechlorination because they have been continually exposed to a supply of fresh PCBs reaching that
section of the river from the Hudson Falls source. This finding underscores that the aged, buried,
dechlorinated PCBs in "hot spots" are not the predominant source to Upper Hudson River fish.
Instead, the fish are accumulating recently deposited PCBs with a composition very similar to
unaltered Aroclor 1242 (and with bioaccumulation and toxicity properties similar to those of Aroclor
1242), consistent with a known source of undechlorinated Aroclor 1242 in the vicinity of the GE
Hudson Falls plant.
A. Dechlorination is an important mechanism in reducing the bioaccumulation and
toxicity of PCBs.
There are environmentally important benefits to dechlorination:
Dechlorination reduces the tendency of PCBs to bioaccumulate.
Dechlorination sharply reduces the levels of the particular PCB congeners that appear
responsible for producing potential risks to wildlife and humans.
Dechlorination of the more heavily chlorinated PCB congeners to lightly chlorinated
congeners facilitates biodegradation and provides a route for the ultimate destruction
of PCBs.
The benefits of dechlorination are discussed at greater length in Appendix D. In summary,
dechlorination reduces both the total chlorine level of the PCB mixture and the concentration of
specific coplanar congeners. Reduced chlorine level results in significant reductions in PCB
carcinogenic potential (USEPA, 1996). For mixtures containing only mono- through
51
-------
tetrachlorobiphenyls, carcinogenic potential has been reduced 100 fold. Coplanar congeners can
determine acute toxicity, and dechlorination in Hudson River sediments has been shown to reduce
the concentration of these congeners by up to 97 percent (Quenson, et al, 1992b). Dechlorination
also dramatically reduces the potential for environmental receptors to be exposed to PCBs in PCB-
contaminated Upper Hudson River sediments. Dechlorination reduces the bioaccumulation potential
of this mixture four to 35 fold. It also facilitates aerobic biodegradation by converting the mixture
to readily degradable congeners.
The insensitive dechlorination indices developed in the Report, which only measure the final
phases of dechlorination, are incapable of measuring these benefits because potential toxicity,
carcinogenicity and exposure are reduced by even modest levels of dechlorination as the initial
stages of dechlorination provide disproportionate reductions in these endpoints.
B. The Report's Dechlorination Indices are Flawed and Insensitive.
The Report characterizes PCB dechlorination in terms of two indices: a so-called "molar
dechlorination product ratio" and a "fractional change in molecular weight". PCB homologue
distributions can be affected by other known physical processes, such as the selective extraction into
the water column, in addition to microbial dechlorination and by different mixtures of Aroclors.
Laboratory experiments show that when water is passed over Aroclor 1242, the extracted PCBs are
generally enriched in the lower homologues, particularly BZ 1, 4, 8,10 and 19 that were used in the
52
-------
Report in calculating molar dechlorination product ratios, and these same congeners are depleted in
the other phases. The enrichments observed in the water phase, relative to the tetrachlorobiphenyls,
are about 40-fold for the monochlorobiphenyls, 10-fold for the dichlorobiphenvls and 3-fold for the
trichlorobiphenyls. Thus, no simple index of homologue distribution, whether expressed as
"product" ratios or as mean molecular weight, can provide definitive information about the
dechlorination state of PCBs formed in a water-extract, and all statements regarding the extent of
dechlorination in media susceptible to such extraction based on these indices are questionable.
An additional serious shortcoming of the "molar dechlorination product ratio" is its
insensitivity, due to the selection of only "terminal" dechlorination products, to assess the extent of
dechlorination. Due to this insensitivity, it is capable only of detecting extensive dechlorination in
sediments containing the dechlorination activity that carries out nearly complete removal of meta
and para chlorines from congeners with 2-4 chlorines per biphenyl (described above as the activity
limited to the upper Hudson River). It would barely register dechlorination in sediments containing
only moderate dechlorination activity, and it would completely miss those sediments containing the
dechlorination activity found throughout the upper and lower Hudson that carries out partial
dechlorinations of higher PCB congeners (i.e., those with 4-7 chlorine atoms per biphenyl), but
producing very little of the lower homologues (with only 1 -2 chlorines). As this other dechlorination
activity still produces significant reductions in toxicity and exposure, its benefits as well as its
detection are completely missed by the analysis in the Report.
53
-------
An illustration of the insensitivity in this flawed dechlorination index is the impact of
dechlorination of BZ 8. The most abundant congener in Aroclor 1242, BZ 8 is dechlorinated to BZ
1. However, this activity would be completely missed in EPA's molecular dechlorination product
ratio because the sum of these congeners never changes due to dechlorination. Moreover, many of
the other most abundant PCB congeners in Aroclor 1242 would be dechlorinated to BZ 1 via BZ 8,
Therefore, this methodology would not detect the final chlorine removal step.
C. Dechlorination Occurs at Concentrations Less Than 30 ppm.
The Report relies on the analysis shown in Figure 4-22 to conclude that dechlorination does
not occur predictably at PCB concentrations <30 ppm. This misrepresents the data. The data clearly
show that the majority of upper river sediments samples register as dechlorinated, even with the
insensitive dechlorination index used in the Report. Moreover, the analysis shows that the majority
of the lower river samples lie below the line, demonstrating that the approach is inappropriate for
the lower Hudson. For the upper Hudson data at <30 ppm (0.8 to 30 ppm), nearly 80% of the
samples displayed on the graph are above the molar dechlorination product ratio (MDPR) for
unaltered Aroclor 1242. This fraction would only increase as more sensitive dechlorination indices
are utilized. For the lower Hudson, -75% of the samples <30 ppm lie below the MDPR for
unaltered Aroclor 1242. It has been well established that the contribution of more highly chlorinated
Aroclors increases in the lower Hudson, particularly in the estuary region (EPA Phase 1 Report).
The addition of higher Aroclors would both invalidate the application of the MDPR analysis and
54
-------
predictably drive this measure below that of Aroclor 1242. The peak ratio methods described in
Appendix C overcome both of these limitations, as they are a more sensitive index to identify
dechlorination (able to detect a variety of dechlorination processes, including Process B, B', C, H,
and FT dechlorination), and they are relatively insensitive to partitioning and variable Aroclor
compositions.
Numerous studies in the laboratory and the field have detected anaerobic PCB dechlorination
over a broad range of concentrations (reviewed in Bedard and Quensen, 1995). Figure 30 shows that
anaerobic PCB dechlorination has been observed in controlled laboratory studies at concentrations
as low as 10 ppm (Abramowicz et. al., 1993; Fish, 1996, Rhee et. al. 1995). Taken together, the
studies demonstrate that there is a linear relationship between PCB concentration and dechlorination
rate without a threshold concentration.
Although field studies of PCB dechlorination are limited by analytical detection limits at low
concentrations, longer incubation times in the field have permitted the detection of PCB
dechlorination at even 5 ppm (Table 1, Abramowicz et. al., 1996). These researchers noted that even
at the lowest concentration range analyzed (5-10 ppm), 63% of the samples still met the established
criteria for extensive dechlorination.
Additional support for environmental dechlorination at low concentrations has been obtained
through direct comparisons of surface PCB congener profiles in the Hudson River to Aroclor 1242
-------
and completely dechlorinated Aroclor 1242 congener profiles. A sediment sample was collected
from the TIP site known as H7. A fraction of this sample was extracted and analyzed for PCB
content and the measured concentration was 6.2 ppm. From the PCB distribution, the number of
chlorines per biphenyl (Cl/BP) was 2.71. Since Aroclor 1242 contains 3.26 Cl/BP, even at this low
PCB-concentration some dechlorination took place in the environment.
The use of intra-homologue peak ratios to assess dechlorination at various PCB
concentrations is shown in Figure 31. Four peak ratios are utilized (BZ 56/49, 23-34-/24-25-CB;
BZ 60/49, 234-4-/24-25-CB; BZ 66/49, 24-24-/24-25-CB; BZ 74/49, 245-4-/24-25-CB). These
ratios represent the change in the dechlorination sensitive tetrachlorobiphenyls (mono-or/Ao
substituted) to the more resistant tetrachlorobiphenyl congener 24-25-CB (di-ortho substituted).
The result of this peak ratio analysis demonstrates that significant dechlorination occurs at all
concentrations, even at sub-ppm levels. There is also a clear trend demonstrating more extensive
dechlorination at higher PCB concentrations, consistent with laboratory experiments.
D. Dechlorination has not stopped in the Hudson River.
The Report incorrectly states that dechlorination has stopped in the Hudson River, based
upon the analysis of high resolution cores (Report at 4-70). This conclusion is flawed since the
indices used to monitor dechlorination are insensitive to some dechlorination processes and since
it ignores the ongoing dechlorination of fresh Aroclor 1242 in surficial sediments. Evidence for
56
-------
dechlorination in surficial sediments exists in several forms. First, surficial sediments collected from
the "fluff' layer from the TIP display Pattern A dechlorination (described in Appendix C). Second,
microcosms that simulate the fate of fresh Aroclor 1242 in Hudson River sediments display Pattern
A dechlorination at early time points (4-6 weeks, Fish, private communication). Third, fish in the
TIP display a PCB distribution consistent with Pattern A dechlorination. This pattern is likely the
result of the initial stages of dechlorination, with more extensive dechlorination occurring over
longer time periods, when additional burial sequesters this material.
These surficial biotransformations (Process A initially and Process Y later) and subsurface
dechlorinations (Process C) which were observed in microcosm experiments (Fish and Principe,
1994 and Fish, 1996) demonstrate rapid dechlorination and degradation of Aroclor 1242 in Hudson
River surface sediments. The changes observed in these physical river models closely correspond
to changes observed in the environment.
Additional evidence to demonstrate that dechlorination is still occurring in the upper Hudson
is found in Figure 32. This Figure represents the application of the intra-homologue peak ratios to
the EPA Phase 2 high resolution sediment cores. To minimize the impact of concentration on the
rate of dechlorination, only a narrow concentration range was utilized (all 3-30 ppm core segments
from all Upper Hudson high-resolution cores). Note that the extent of dechlorination continues to
change in a smooth continuum with increasing depth in the core. This data demonstrates that the
extent of dechlorination strongly correlates with increasing depth and increasing age of the sediment,
57
-------
inconsistent with the Report's claim that dechlorination has stopped in the upper Hudson. These
analyses also demonstrate the strength of peak ratios as effective indices of anaerobic PCB
dechlorination in environmental media, assessing concentration and temporal effects even under
conditions when partitioning, variable Aroclor compositions, or modest level of dechlorination are
present, conditions that can confound other dechlorination indices.
58
-------
VII. CONCLUSIONS AND RECOMMENDATIONS
In order to evaluate remedial options properly, EPA must understand the sources, transport
and fate of PCBs in the Upper Hudson River. This Report provides a geochemical analyses of the
data, based on several inconsistent hypothesis, that do not provide a realistic and accurate view of
the river system.
The Report offers various hypotheses to explain the spatial and temporal patterns in the PCB
data. Many of the hypotheses are incompatible, and the Report chooses from among them to reach
overall conclusions. In most cases, the choices are based on qualitative arguments that are not
rigorously evaluated. In these comments, we have presented technical arguments that refute the
primary hypotheses that form the basis of the Report's conclusions. Our arguments are based on
geochemical fingerprinting techniques similar to those used by EPA; a quantitative determination
of the PCB fate mechanisms required by the hypothesis; and PCB mass-balance calculations. The
last two types of analyses are a requirement of hypothesis-testing because they examine the
plausibility of the stated hypothesis.
Because EPA's approach to data interpretation is restricted to a geochemical examination of
the data, it is sufficient for developing hypotheses but not for testing them. The Agency must
acknowledge this limitation and conduct further analyses to test the hypotheses within the
framework of the mass-balance model currently under development. As we have done in our
59
-------
evaluation of the Report conclusions, the Agency must use the model to examine the consistency
of each hypothesis with the estimated rates and magnitudes of relevant fate processes and with
historical information regarding the spatial and temporal distributions of PCBs in water, sediment
and fish. An integrated interpretation that accounts for all the sources and losses of PCBs is
necessary to develop conclusions about the relative importance of the various PCB sources and the
rate of recovery.
The utility of the Report and its conclusions are fundamentally undermined by its numerous
inconsistent statements. An exhaustive review of each of such statements is beyond the scope of
these comments, but the examples set out below demonstrate this shortcoming:
* The Report hypothesizes resuspension of dechlorinated sediments from the TIP
despite its own analysis that explains that resuspension is not occurring. On pages 3-62 and 3-63,
the Report concludes that resuspension of TIP sediments is of limited importance to the load
measured at the TID because there is no evidence of resuspension during low flow and during high
flow, when resuspension would be expected, the load above Rogers Island is transported relatively
unaffected through the TIP. Notwithstanding this sound conclusion, the Report later hypothesizes
that a possible source of the increased load across the TIP seen most prominently during low-flow
periods is resuspension of sediments containing highly concentrated, highly dechlorinated PCBs.
This hypothesis, presented as one of the primary conclusions of the Report, is obviously inconsistent
60
-------
with the observation made at 3-62 and 3-63 that resuspension within the TIP is not an important
process.
The Report emphasizes the importance of sediments as PCB contributors to the TIP
and de-emphasizes its own analysis that shows that the current sources near Hudson Falls are more
important. On page 3-90, the Report describes the mass of PCBs provided by the different internal
and external sources in the Upper River in 1993: 370 kg from the sources above Rogers Island; 225
kg from the TIP; 25 kg from the "Schuvlerville" source; and 83 kg from scour from the Hoosic-
driven scour of surficial sediments during Transect 3. Using these figures, the sources above Rogers
Island provide approximately 50% of the annual load during 1993. This fact, however, is ignored
in the Report's primary conclusions, which, in particular, attribute increased loads to the TIP
sediments):
"The PCB load from the Thompson Island Pool... dominates the water column from
the Thompson Island Dam to Kingston during low flow conditions" (Report at E-3);
"The sediments of the Thompson Island Pool strongly impact the water column,
generating a significant water column load (as documented in Chapter 3) whose
congener pattern can often be seen throughout the Upper Hudson" (Report at 4-91).
61
-------
The Report dismisses the importance of the more recent load originating near Hudson
Falls attributed to 1991 - 1993-era loads from the Allen Mill, which the Report concludes have been
essentially eliminated:
"Recent remedial efforts by GE have greatly decreased the PCB loads originating
above Rogers Island. As a result, the total annual loads to the water column have
decreased but the importance of the TI Pool load has increased." (Report at 3-91).
Not only are there no data to support this conclusion, it ignores the likelihood that the large
releases from the Mill were deposited above and within the TIP, where they are now contributing
to the water column.
• The Report is inconsistent in its reliance on different mechanisms above and below
the TID. On the one hand, the Report posits a number of possible physical mechanisms within the
TIP -- settling or volatilization of the load entering the TIP and porewater diffusion or resuspension
of TIP sediments — to explain the increased loading of PCBs across the pool. At the same time, the
Report appears to ignore or discount these same mechanisms in the area below the TID in reaching
the conclusion that the load at the TID is transported without significant loss of PCBs to the water
column through the rest of the freshwater Hudson. Yet, the sediments in the areas above and below
the TID are very similar, and the Report itself identifies an increase in PCB load downstream of the
TID during low flows (Report at 3-84). It is unreasonable to invoke these processes where they tend
62
-------
to support one's conclusions, but ignore them when they do not comport with those same
conclusions.
In order to complete a technically defensible remedial analysis, EPA must develop a
consistent and physically plausible explanation of the data and then test that explanation against a
calibrated and validated model to ensure that the true sources of PCBs to fish, wildlife and humans
are identified. The explanation of the data needs to take into account all the processes at work in
the river that are relevant to remedial analyses. Where questions central to remedial analyses cannot
be answered with the data presently at hand, additional data must be obtained to resolve the issues
so that we can have full confidence in the conclusions reached on the basis of data interpretation and
evaluation.
63
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
"D i? iri? "D tr xr ci? c
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
REFERENCES
Abramowicz, D.A., M.J. Brennan. and H.M. Van Dorl, 1989. Anaerobic biodegradation of
polychlorinated biphenyls, in Extended Abstracts: 198th American Chem. Soc. National Meeting,
published by Div. Environ. Chem., Vol. 29(2). ISSN# 0093-3066, p. 377-379,
Abramowicz, D.A., 1990. Aerobic and anaerobic biodegradation of PCBs: A review, in CRC
Critical Reviews in Biotechnology, G.G. Steward and I. Russell (eds.), CRC Press, Inc., Vol. 10, p.
241.
Abramowicz. D.A., J.F. Brown Jr., M.K. O'Donnell, 1991. Anaerobic PCB dechlorination in
Hudson River sediments, in General Electric Company Research and Development Program for the
Destruction of PCBs. Tenth Progress Report, General Electric Corporate Research and
Development Schenectady, NY. Chapter 17.
Abramowicz, D.A.. M.R. Harkness. J.B. McDermott, and J.J. Salvo, 1992. 1991 in situ Hudson
River research study: a field study on biodegradation of PCBs in Hudson River sediments, in
General Electric Company Research and Development Program for the Destruction of PCBs.
Eleventh Progress Report, General Electric Corporate Research and Development, Schenectady, NY,
Chapter 1.
Abramowicz, D.A.. M.J. Brennan, H.M. Van Dort, E.L. Gallagher. 1993. Factors influencing
the rate of polychlorinated biphenyls dechlorination in Hudson River sediments. Environ. Sci.
Technol. 27:1125-31.
Abramowicz. D.A., 1994. Aerobic PCB Biodegradation and Anaerobic PCB Dechlorination in the
Environment. Research in Microbiology, 145:42-46.
Abramowicz, D.A., J.F. Brown Jr., M.R. Harkness. and M.K, O'Donnell. 1996. In situ anaerobic
PCB dechlorination and aerobic PCB biodegradation in Hudson River sediments, in Biotechnology
in Industrial Waste Treatment and Bioremediation, ed. R.F. Hickey and G. Smith, Lewis Publishers,
CRC Press.
Ahmed, M. and D.D. Focht. 1973. Degradation of polychlorinated biphenyls by two species of
Achromobacter. Can. J. Microbiol, 19:47-52.
Alder, A.C., M.M. Haggblom,S.R. Oppenheimer, and L.Y. Young, 1993. Environ. Sci. Technol,,
27:530-538.
1
-------
Bedard, D.L., R.E. Wagner. M.J. Brennan, M.L. Haberi, and J.R. Brown, Jr. 1987.
Extensive Degradation of Arociors and Environmentally Transformed Polychlorinated Biphenyls
by Alcaligenes eutrophus H850. Appl. Environ. Microbiol., 53:1094-1102.
Bedard, D.L, 1990. Bacterial transformations of polychlorinated biphenyls, in Biotechnology and
Biodegradation, D. Kamely, A. Chakrabarty and G.S. Omenn (eds.), Adv. Appi. Biotechnol. Series,
Vol. 4, Portfolio Publishing Co., The Woodlands, TX, p. 369.
Bedard, D.L., and J.F. Quensen, III, 1995. Microbial reductive dechlorination of polychlorinated
biphenyls, in Microbial Transformation and Degradation of Toxic Organic Chemicals, eds. L.Y.
Young and C. Cemiglia, John Wiley & Sons, Inc.
Brown, J.F., R.E. Wagner, D.L. Bedard, M.J. Brennan, J.C. Carnahan, and R.J. May, 1984.
PCB Transformations in upper Hudson sediments. Northeast. Environ. Sci., 3:167-179.
Brown, J.F., D.L. Bedard, M.J. Brennan, J.C. Carnahan, H. Feng, and R.E. Wagner, 1987a.
PCB dechlorination in aquatic sediments. Science. 236:709-712.
Brown. J.F., Jr., R.E. Wagner. H. Feng, D.L. Bedard, M.J. Brennan. J.C. Carnahan, and R.J.
May, 1987b. Environmental dechlorination of PCBs, Environ. Toxicol. Chem., 6:579-593.
Brown. J.F., Jr, and R.E. Wagner. 1990. PCB movement, dechlorination, and detoxication in the
Acushnet Estuary. Environ. Toxicol. Chem., 9:1215 -1233.
Cardenas, M., J. Gailani. C.K. Ziegler, and W. Lick, 1995. Sediment Transport in the Lower
Saginaw River, Mar. Freshwater Res., 46:337-347.
Carroll, K.M., M.R. Harkness, A.A. Bracco, and R.R. Balcarcel. 1994. Application of a
permeant/polymer diffiisional model to the desorption of polychlorinated biphenyls from Hudson
River sediments. Environ. Sci. Technol., 28:253-258.
Coates, J.T., and A.W. Elzerman. 1986. J. Contam. Hydrol. 1:191-210.
Code of Federal Regulations, 1995. CFR 40, pans 700-789, section 761.3, pg. 392.
Erickson, M.D., 1992. Analytical Chemistry of PCBs. Lewis Publishers.
Fish, K.M., 1996. Influence of Aroclor 1242 concentration on polychlorinated biphenyl
biotransformations in Hudson River test tube microcosms, Applied and Environmental
Microbiology 62:3014-6.
Fish, K.M., and J.M. Principe, 1994. Biotransformations of Aroclor 1242 in Hudson River test
tube microcosms, Appl. Environ. Microbiol60:4289-96.
2
-------
Flanagan, W.P. and R.J. May, 1993. Metabolite formation as evidence for in situ aerobic
biodegradation ofpolychlorinated biphenyis, Environ. Sci. Technol., 27:2207-2212.
Furukawa, K., 1982. Microbial degradation of polychlorinated biphenyis (PCBs), in
Biodegradation and Detoxification of Environmental Pollutants, A.M. Chakrabarty (ed.), CRC
Press, Inc., Boca Raton, FL, p. 33.
Furukawa, K., 1986. Modifications of PCBs by bacteria and other microorganisms, In PCBs and
the Environment, Vol. 2, J.S. Waid (ed.), CRC Press, Inc., Boca Raton, FL, p. 89.
Gailani, J., C.K. Ziegler, and W. Lick, 1991. The Transport of Suspended Solids in the Lower
Fox River,/. Great Lakes Res., 17:479-494.
Gailani, J., W. Lick, C.K. Ziegler, and J. Endicott, 1996. Development and Calibration of a
Fine-Grained Sediment Transport Model for the Buffalo River, J. Great Lakes Res., 22:765-778.
General Electric, HydroQual. Inc. 1996. Comments to the Environmental Protection Agency
Preliminary Model Calibration Report.
Harkness, M.R., J.B. McDermott. D.A. Abramowicz, J.J. Salvo, W.P. Flanagan, M.. Stephens,
F.J. Mondello. R.J. May, J.H. Lobos, K.M. Carroll, M.J. Brennan. A.A. Bracco, K.M. Fish,
G.L. Warner, P.R. Wilson, D.K. Dietrich, D.T. Lin, C.B. Morgan, and W.L. Gately, 1993. In
situ stimulation of aerobic PCB biodegradation in Hudson River sediments. Science. 259:503-507.
HydroQual. Inc. 1995. Quantification of Sedimentation in the Upper Hudson River. Prepared for
General Electric. Submitted to the Environmental Protection Agency.
HydroQual. Inc. 1996. DNAPL Transport and Water Column Monitoring Study. Sampling and
Analysis Plan. Prepared for the General Electric Company. Albany, New York.
Karim. M.F. and F.M. Holly. 1981. Armoring and sorting simulation in alluvial rivers, ASCE J.
Hydr. Engr., 112:705-715.
Lake, J.L., RJ. Pruell, and F.A. Osterman, 1992. An examination of dechlorination processes and
pathways in New Bedford harbor sediments, Marine Environ. Res., 33:1.
Lobos, J.H., 1994. PCB Biodegradation Under Microaerophilic and Aerobic Conditions With and
Without Stimulation by Biphenyl. ASM Annual Meeting, Las Vegas, NV.
Lobos. J.H., 1995. Isolation and Characterization of Novel PCB-Degrading Bacterai Unable to
Grow on Biphenyl. ASM Annual Meeting, Washington, DC.
3
-------
Mousa. M.A., J.F. Quensen, Ill, K. Chou. and S.A. Boyd, 1996. Microbial dechlorination
alleviates inhibitory effects of PCBs on Mouse gamete fertilization in vitro. Environ. Science
Technol.. 30:2087-2092.
Nadim. L., M.J. Schocken. F.J. Higson. D.T. Gibson, D.L. Bedard, L.H. Bopp, and F.J.
Mondello. 1987. In: Proceedings of the 13 th Annual Research Symposium on Land Disposal,
Remedial Action, Incineration, and Treatment of Hazardous Waste. EPA/600/9-87/015. U.S.
Environmental Protection Agency, Cincinnati, Ohio.
Phillips, P. J. and D.W. Hanchar, 1996. Water-Quality Assessment of the Hudson River Basin in
New York and Adjacent States - Analysis of Available Nutrient, Pesticide, Volatile Organic
Compound, and Suspended-Sediment Data. 1970-90, U.S. Geological Survey, Water-Resources
Investigations Report 96-4065.
Quensen, J.F, III, J.M. Tiedje, S.A, Boyd. 1988. Reductive dechlorination of PCBs by anaerobic
microorganisms from sediments. Science. 242:752-754,
Quensen. J.F. Ill, S.A. Boyd, and J.M. Tiedje, 1990. Dechlorination of four commercial
polychlorinated biphenyi mixtures (Aroclors) by anaerobic microorganisms from sediments. Appl.
Environ. Microbiol.. 56:2360-2369.
Quensen, J., S.A. Boyd, J.M. Tiedje, R.F. Lopshire. and C.G. Enke, 1992a. Expected dioxin-like
toxicity reduction as a result of the dechlorination of Aroclors, in General Electric Company
Research and Development Program for the Destruction of PCBs. Eleventh Progress Report,
General Electric Corporate Research and Development. Schenectady, NY, Chapter 17, p. 189.
Quensen. J.F. Ill, J.M. Tiedje, S.A. Boyd, C. Enke, R". Lopshire, J. Giesy, M. Mora, R.
Crawford, and D. Tillitt. 1992b. Evaluation of the suitability of reductive dechlorination for the
bioremediation of PCB-contaminated soils and sediments, in Preprints: International Symposium
on Soil Decontamination Using Biological Processes, Karlsruhe/D. 6-9 December 1992, pp. 91-100.
ftahuel. J.L., F.M. Holly, J.P. Chollet, P.J. Belleudv, and G. Yang. 1989. Modeling of riverbed
evolution for bedload sediment mixtures. ASCE J. Hydr. Engr.. 115:1521-1542.
Rhee. G.-Y., B. Bush, C.M. Bethoney, A. DeNucci, H.-M. Oh, R.C. Sokol. 1995. Reductive
dechlorination of Aroclor 1242 in anaerobic sediments: Pattern, rate and concentration dependence.
Environ. Toxic. Chem.. 12:1025-32.
Risatti, J.B., 1992. Rates of Microbial Dechlorination of PCBs in Anaerobic Sediments from
Waukegan Harbor, Illinois Hazardous Waste Research and Information Center, Champaign, IL,
HWRIC Project HWR 86-010. Report RR-061,
4
-------
Safe, S.. 1992. Toxicology, structure-function relationship, and human and environmental health
impacts of polychlorinated biphenvls: Progress and problems. Environ. Hlth. Perspect, 100:259-
268.
Schroeder and Barnes, 1983. Trends in Polychlorinated Biphenyl Concentrations in Hudson River
Water Five Years after the Elimination of Point Sources. U.S.G.S. Water Resources Investigation
Report. 83-4206. Albany, New York.
Sokol, R.C., O-S. Kwan, C.M. Bethoney, and G-Y. Rhee, 1994. Reductive Dechlorination of
PCBs in St. Lawrence River Sediments and Variations in Dechlorination Characteristics. Env. ScL
Technol.. 28:2054-2064.
US Environmental Protection Agency, 1996. PCBs: Cancer Dose-Response Assessment and
Application to Environmental Mixtures. EPA/600/p-96/001, Washington DC, September 19%.
US EPA Great Lakes National Program Office (prepared by Swackhamer. D.L.), 1987. Quality
Assurance Plan: Green Bay Mass Balance Study I. PCBs and Dieldrin.
Memorandum from: M.D. Mullin of the U.S. EPA Environmental Research Laboratory - Duluth,
Large Lakes Research Station, to: G. Frame of GE Corporate Research and Development, dated 21
November 1994.
van Rijn, L.C. 1984. Sediment transport. Part II: Suspended load transport. ASCE J. Hydr. Engr.,
110:1612-1638.
William A. Williams and R.J. May. Low Temperature Aerobic Degradation of Polychlorinated
Biphenvls (PCBs) in Sediment, submitted.
Williams W.A., and R.J. May. Low Temperature Microbial Aerobic Degradation of
Polychlorinated Biphenvls (PCBs) in Sediment. Submitted Environmental Science and Technology.
Witkowski, P.J., P.R. Jaffe, and R.A. Ferrara. 1988. Sorption and desorption dynamics of
Aroclor to natural sediment. J. Contam. Hydrol. 2:249-269.
Ziegler, C.K. and J.P. Connolly. 1995. The impact of sediment transport processes on the fate of
hydrophobic organic chemicals in surface water systems. Proceedings of the WEF Toxic Substances
in Water Environments Conference, pp. 1-13 to 1-24.
Ziegler. C.K. and B. NisbeL 1994. Fine-Grained Sediment Transport in Pawtuxet River, Rhode
Island, ASCE J. Hydr. Engr., 120:561-576.
Ziegler. C.K. and B. Nisbet. 1994. Long-Term Simulation of Fine-Grained Sediment Transport
in Large Reservoir, ASCE J. Hydr. Engr., 121:773-781.
5
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
TABLES
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
Table 1.
Surface Sediment PCB Reservoir Depletion under 1993-1996 Average
Thompson Island Pool Load
Homologue
Mass of PCBs in TIP
Surface Sediments in
1984(U)
(MT)
Load from TI
Pool(3)
(MT/year)
Time to deplete the
sediment reservoir
(Years)
1
0.58
0.055
1995
2
1.4
0.117
1996
3
1.0
0.062
2000
4
0.41
0.016
2009
5
0.13
0.002
2040
Sum
3.52
0.25
= 0.69 kg/day
(1) Mass of total PCB in surface sediments = surface sediment concentration* x specific weight of
sediments* x 8 cm depth x area of TI Pool*
* Values based on EPA analysis.
(2) Homologue mass based on homologue composition of EPA low resolution cores
(3) Load from TI Pool = Load at TI Dam - load at Rogers Island; all GE data, 1993-1996.
-------
Table 2.
Average PCB Loading Across Thompson Island Pool from 1993 to 19961
Year
Number of Paired Samples
Average PCB Load
[kg/year]
1993
49
202.2
1994
34
296.9
1995
45
84.3
1996
57
406.6
Overall
185
253.9
'Loadings calculated from GE database (corrected for analytical bias), based on daily average flows
measured at Fort Edward and differences between paired water column PCB concentrations from
samples collected at Fort Edward and Thompson Island Dam.
-------
Table 3.
Magnitude and Composition of the Unaccounted for Summer PCB Load from
Thompson Island Pool
Year
Unaccounted-
for PCB Load
(kg/dayj
Unaccounted-for Solid Phase PCB Homolog Distribution [WT%]
Mon
Di
Tri
Tet
Pen
Hex
Hep
Oct
Non
Dec
1991
0.05
5.9
0.1
38.8
32.1
22.8
0.3
0.0
0.0
0.0
o.c
1992
0.56
11.8
19.9
42.1
23.4
1.8
1.1
0.0
0.0
0.0
o.c
1993
0.99
7.5
17.8
28.8
31.7
12.1
2.2
0.0
0.0
0.0
o.c
1994
0.97
9.3
22.4
33.8
2S.0
6.4
0.2
0.0
0.0
0.0
. o.c
1995
0.23
2.3
14.8
29.1
35.4
15.0
3.3
0.0
0.0
0.0
0.0
1996
1.18
•14
18 3
36.9
29.4
9.4
1.6
0.0
0.0
0.0
o.c
-------
Table 4
Magnitude and Composition of the Unaccounted-for
Monthly PCB Load from Thompson Island Pool During 1996
Month
of 1996
Unaccounted-
for PCB Load
[kg/day]
Unaccounted-for Solid Phase PCB Homolog Distribution [WT%]
Mon
Di
Trf
Tet
Pen
Hex
Hep
Oct
Non
Dec
Jan
0.54
3.6
17.4
43.1
31.9
3.7
0.2
0.0
0.0
0.0
0.0
Feb
0.13
0.6
26.2
36.9
28.6
7.5
0.1
0.0
0.0
0.0
0.0
Mar
0.60
3.4
15.6
33.5
35.3
11.2
1.0
0.0
0.0
0.0
0.0
Apr
2.05
4.3
11.6
38.3
38.2
6.7
0.7
0.0
0.0
0.0
0.0
May
1.80
5.6
19.6
34.7
31.5
7.8
0.8
0.0
0.0
0.0
0.0
Jun
2.06
4.9
15.6
35.9
30.9
11.1
1.6
0.0
0.0
0.0
0.0
Jul
0.68
4 9
28.4
35.1
21.2
8.7
1.7
0.0
0.0
0.0
0.0
Aug
0.92
3.9
19.1
40.3
30.4
4.7
1.7
0.0
0.0
0.0
0.0
Sep
0.56
2.3
17 9
33.0
32.3
13.7
0.8
0.0
0.0
0.0
0.0
Oct
0.27
23.9
42.0
12.7
16.0
4.9
0.5
0.0
0.0
0.0
0.0
Nov
0.38
19.2
40.1
19.1
12.4
8.3
0.8
0.0
0.0
0.0
0.0
Dec
0 19
00
35 6
38.2
10.2
13.7
2.2
0.0
0 0
0.0
0.0
1
-------
Table 5.
Information Sources for Homolog-Specific Parameters of the Bioaccumulation Model
Parameter
Data Source
Benthic invertebrate/sediment accumulation
factors
EPA invertebrate data, unsorted total, all
samples
Water column invertebrate/water column
particulate trophic transfer factors
Green Bay zooplankton/phytoplankton
trophic transfer factors
Water column particulates/dissolved partition
coefficient
EPA partitioning data as analyzed by
HydroQual
Assimilation efficiencies at the gut and gill
Values applied in the Green Bay model
-------
Table 6.
Exposure Sources for Homologue-Based Bioaccumuiation Model for TIP
Figure #
(Simulation #)
18
(h24)
19
(H23)
20
0*25)
21
0*26)
Exposure basis:
Realistic water,
top 2 cm of
sediment bed
Realistic water,
top 5 cm of
sediment bed
Realistic water,
heavily
dechlorinated
sediments
Heavily
dechlorinated
sediments and
water
Water column
dissolved tPCB
concentration
20 ng/L
(TIP late 1980s)
20 ng/L
(TIP late 1980s)
20 ng/L
(TIP late 1980s)
computed from
water column
particulates01
Water column
dissolved
homo log
composition
Avg of summer
data 91 -96 at
Ft. Edward and
TIDam
Avg of summer
data 91 -96 at
Ft. Edward and
TIDam
Avg of summer
data 91 -96 at
Ft. Edward and
TIDam
Computed from
water column
particulates0'
Water column
particulate
tPCB
concentration
Computed from
dissolved0'
Computed from
dissolved*"
Computed from
dissolved0'
Same as
sediment bed
particulates
Water column
particulate
homo log
composition
Computed from
dissolved0'
Computed from
dissolved0'
Computed from
dissolved0'
Same as
sediment bed
particulates
Sediment
particulate
tPCB
concentration
400 ug/gOC
(TIP late 1980s,
fate model)
400 ug/gOC
(TP late 1980s,
fate model)
400 ug/gOC
(TIP late 1980s,
fate model)
400 ug/gOC
(TIP late 1980s,
fate model)
Sediment
particulate
homolog
composition
TIP EPA hires
cores, top slice
TIP 1991 GE
data, top slice
TIP EPA hires
core slices with
<2 Cl/BP
TIP EPA hires
core slices with
<2 Cl/BP
(1) Water column dissolved composition was computed from particulate composition (and vice versa)
using partition coefficients based on analysis of EPA data.
-------
FTCTTRFS
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
t* "¦ |
Al
4832
A2
539^4
A3
65993
AS I <7595
A6
129309
A4
6206
A7
166642
A8
41450
A9
5936
A10
223165
Total Area = 69,409 sq. m
(Total TIP ¦ 1.93E+06 sq. m)
Legend
MS EPA PCB > 100 ppm
/\/ Mile Markers
/\J Shore Line
Area of Interest
A
A9«H A10
0,5 Miles:
HydroQual, Inc.
Figure i. Areas of TIP with surface sediment PCB concentrations greater than 100 ppm.
-------
10 -
a
30
0
20
10
30
70
80
50
60
90
40
Total PCBs Jppm dry]
Figure 2. Average 1984 sediment PCB profile for areas with surface sediment PCBs greater
than 100 mg/Kg.
-------
\ 1 1 1 1 1 1 1 — h
1 >
1 >
.... 1
•
'
>
1
~
J
y
f
;
r*
-
o
i i—
Mtt
P
1
Scoured Depth In:
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
10 20 30 40 SO 60 70 80 90 100
Total PCB Ippm dry]
(Average from TIP Historic Cores)
Figure 3.
Annual sediment scour depths required to achieve water column concentrations
necessary to maintain the TIP load.
-------
1.50
1.25
to
»
to w
«5 C
C
o
E
'o
a>
a.
®
O
n
c
o
-------
Transect 4
••
'• •
0.001
0.0005
O.OOODD1 0.01 0.1
1
10 60
Transect 1
iiiiii i
••1
0-1.
0.0*
0.001
0.0005
WWiliWtim i
0.1
0.0008)1 0.01
10 50
1
Transect 5
50
0.1
• •
o.oi;
0.001^
0.0005
O.OOOB1 0.01 0.1
10 50
1
Transect 2
50
100 ppm
Figure 5. Comparison of TID water column PCB signature with TIP deep dechlorinated
sediment.
-------
Transect 1
im i 11ilia i iihib 11 mm iiiiib^ i \t
0.001
0.000!
0.00CQD1 0.01 0.1 1 10 SO
Transect 4
ll« I lllllll 'ITTIIIII I lllllli I IIIIW l >
0.001
0.000!
O.0OOS1 0.01 0.1 1 10 GO
Transect 2
UU 1 IH»H IDDiB 1 ) HUH ""1 11I 11
c *
S*
SI
4. (B
>«Q
'%l '*"1
0.001
0.000:
Transect 5
0.001
0.0005,
111 ma 111 im 11 una i nma i
0.01
'« 0 *m*
fcHtt'lh ' '
O.OOOBDI 0.01 0.1
10 60
0.00BB1 0.01 0.1 1 10 50
Transect 3
0.01
0.001
0.000!
Q.OOdSI 0.01 0.1 1 to so
Avg, Surface Sediment
BZ Weight % in TIP
Transect i
i uiii i nnm i i
0.01
0.001
0.000!
O.OOOB1 0.01 0.1 1 10 80
Avg. Surface Sediment
BZ Weight % In TIP
ToUl Water Column Oata Source; EPA Phase 2 Transact Studies at TID
Surface Sediment Data Sourca: EPA Phase 2 High Res Cora aiieaa in TIP
Figure 6.
Comparison of TID water column PCB signature with TIP surface sediments.
-------
0.30
0.20
0.10
0.00
«« -0.10
C «
oQC
-0.20
-0.301
-0.1
A
A- A.-
I^VvBpk. #
• A A _ A
ts'o"
Legend;
,O0® O
•\'b
0 .
.0
Rogers Island Water Column, iSss and part
Thompson Island Dam Watar Column, diss and part
a Hl-ras cor* data - Uppar Hudson
v Hl-ras cora data • Lower Hudson - freshwater only
^ Aroclors on particles & computed pore w
T 1991 TIP surfssd data & compd porew
X
_L
0.0
0.1
0.2 0.3 0.4 0.5 0.6 0.7
Molar Dechlorination Product Ratio
0.8
0.9
1.0
Molar dechlorination product ratio versus fractional molecular weight change relative
to Aroclor 1242.
-------
T3 T4
F2
T2
T6
40000
30000
20000
10000
A
J
J
M
S
A
0
M
F
N
J
D
J
Figure 8. Location of EPA sampling events relative to 1993 hydrograph.
-------
>»
ts
[a
¦a
a
o
_j
a
a
a.
1000
100
Conactad for Analytical BUi in NEA Pitki
6. 0. and 14 for W«t»r and Sadbnantl
i i i i i i
Summer 91 Water Column Load
Total m
304 g/day
20 30 40 SO 60
70
SO
M
100
110
1000
too
1
I I I I I
Calculated Diffusive Sediment I
from 1991 Sadimant Survey Average!
Total ¦
1_L
70
•0
90
294 g/day
100
110
200.
100.
10.
M
111
Li
i i
Summer 91 Load Difference
Total
45 g/day ;
10
20
30
40
so
CO
70
BO
90
100
110
0.1,
inr
ir^nr
i i
ii
10 I
100
1000
Total = -46 g/day
Summer 91 Load Difference^
10 20 30 40 50 60 70
NEA Peak Number
90
90
100
110
Kf - 0.005 m/day. Note, negative weter column loads are not Included.
Figure 9. Summer 1991 congener peak TIP PCB loading - a) water column load; b) calculated
diffusive sediment load; c) unaccounted-for load; d) negative congener loadings.
-------
600
500 .
400.
300.
200.
100.
OL
600-
500.
400 .
300 .
200 .
100 .
OL
600
500.
400.
300 .
100.
100.
0 .
60
50.
40 .
30.
20 .
10
0
;ure
Convcttd for Analytical Bias tn NEA Paaki 8. 8, and 14 for Watar and Sadimant
Conganar Kocs Corractad for Tamfwratura Oapamlanca
1 1 i l '" i l 1 )
Summer 91 Water Column Load
Total " 304 g/day
i i i > i i i
Calculated Diffusive Sadimant Load
from 1991 Sediment Survey Averages.
Total = 234 g/day
LAIUlt
10
Summer 91 Unaccounted for Load
Total
48 g/day
10
"I' ' I I > I 1 1 I 111 I 1 1
Sadimant Composition Needed to
Yield Summer 91 Unaccounted for Load.
Homolog Group
Summer 1991 TIP PCB homoiogue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted-for load; d) sediment composition
required to yield unaccounted-for load.
-------
•00
SCO.
400 .
300.
200.
100 .
OL
Comctad let AnitytieaJ But ki NEA P«»k» f. I, and 14 tm Waiar and Saimam
Conganar Koo Conacta* for Tamparatun Dapandanca
—i ~ i ' i
Summer 92 Watar Column Load
Total
769 g/day
800-
500 .
400 .
300 .
200 .
100.
OL
«00_
500.
400.
300 .
200.
100 .
OL
#0 f
so.
40.
30.
20.
10.
0.
10
Calculated Diffusive Sediment Load
from 1991 Sadimant Survey Average a.
Total
237 g/day
10
1 - — — ) 1
Summar 92 Unaccounted for Load
Totat
SS9 g/day
10
i 1 1 1
Sedbnent Composition Needed to
Yield Summar 92 Unaccounted for Load.
18
Homolog Group
Figure 11. Summer 1992 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted-for load; d) sediment composition
required to yield unaccounted-for load.
10.0442
-------
500.
400 .
300 .
200 .
100.
0L_
600.
500 .
400 .
300 .
ZOO .
100 .
oL_
600
500 .
400.
300.
200 .
100 .
oL
60
50 .
40 .
30 .
20
10
0
Corraetad for Analytical Blat in NEA Paaka S. 8. and 14 for Watar and Sadlmant
Conganar Koca Corraetad for Tampatatura Oapandanca
Summer 93 Water Column Load
Total =• 1217 g/day
_essl
10
Calculatad Diffusive Sediment Load
from 1991 Sediment Survey Averages-
Total
230 g/day
10
Summer 93 Unaccounted for Load
Total =
994 g/day
10
Sediment Composition Needed to
Yield Summer 93 Unaccounted for Load-
.MM
10
Homolog Group
Summer 1993 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted-for load; d) sediment composition
required to yield unaccounted-for load.
-------
*4
600
SOO
>»
(0
-Q
400
o>
«o
300
<9
O
200
CO
o
a.
100
0
600
SOO
>»
CO
"O
400
CD
"O
300
(O
o
-J
200
GO
o
a.
100
0
>.
a
§aR
OH
I*
X C
-.2
c «s
a a
E-°
.5 -c
CO Q
600
500
400
300
200
100
0
60.
50 .
40 .
30 .
20 .
10 .
ol
Comcted for Analytical Biat in NEA Paaka 5. 8. and 14 for Water and Sadimant
Conganar Koca Corractad for Tamparature Oapandanca
Summer 94 Water Column Load
Total
1185 g/day
JSBQ-
. '«¦
10
-i 1 r
; Calculated Diffusive Sediment Load
from 1991 Sediment Survey Averages-
Total
226 g/day
10
Summer 94 Unaccounted for Load
Total =
967 g/day
IWV1
10
-» 1 i
Sediment Composition Needed to
Yield Summer 94 Unaccounted for Load-
10
Homoiog Group
Figure 13. Summer 1994 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted for load; d) sediment composition
required to yield unaccounted-for load.
10.0444
-------
500 .
400.
300.
200 .
100.
OL
600—.
500 .
400 .
300 .
200 .
100 .
oL
600
500.
400.
300 .
200.
100 .
oL
60 _
50 .
40 .
30 .
20 .
10
0
Corrected (or Analytic*! Bias In NEA Paaks 5. 8, and 14 for Watar and Sacfimant
Conganar Kocs Com clad for Tamparatura Dapandanca
i i i i i . ¦ i i
Summer 95 Water Column Load
Total = 504 g/day
10
i i i i
Calculated Diffusive Sediment Load
from 1991 Sediment Survey Averages-
Total = 288 g/day
^SB3L
10
Summer 95 Unaccounted for Load
Total = 228 g/day
J3S ps*
10
I I 1 I
Sediment Composition Needed to
Yield Summer 95 Unaccounted for Load
Homolog Group
Summer 1995 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted for load; d) sediment composition
required to yield unaccounted-for load.
-------
600
SOO .
400 .
300 .
200.
100 .
OL
600—
500 .
400 .
300 .
200 .
100 .
oL
600
500 .
400 .
300 .
200 .
100 .
oL
60
50 .
40 .
30 .
20 .
10 .
0
Corractad for Analytical Bias in NEA Paak* 5. 8. and 14 (or Watar and Sadimant
Conganar Koca Corractad for Tamparatura Dapandanca
Summar 96 Watar Column Load
Total ¦ 1441 g/day
M.
10
1 1 1
Calculated Diffusiva Sacfimarrt Load
from 1991 Sadimant Survay Avars gas.
Total
273 g/day
10
Summar 96 Unaccountad for Load
Total
1176 g/day
_E85L
10
I I . I
Sadimant Composition Naadod to
Yield Summar 96 Unaccountad for Load.
1 23456789 10
Homoiog Group
Summer 1996 TIP PCB homologue loading - a) water column derived load; b)
calculated diffusive sediment load; c) unaccounted for load; d) sediment composition
required to yield unaccounted-for load.
-------
Not*: Psak* > 15% by weight ir* not plotttd
15
ffl
3
CO
c
o>
CO
9
e
3
o
o
u
o
e
3
a.
12 15
a
c
a>
25
TJ
O
+*
c
3
O
a
o
CO
©
c
3
O
o
u
tt
c
3
0 3 6 9 12 15
Avg TIP Surface Sediment
(1992 EPA HR Cores 0-2 cm)
3 6 9 12 15
Avg TIP Surface Sediment
(1992 EPA HR Cores 0-2 cm)
Fieure 16. Comparison of PCB congener composition between average TIP surface sedimen
and unaccounted-for TIP load (summer 1991 - 1996).
-------
Not*: P«*k* > 15% by weight ar* not plotted
15
12 IS
12 15
12 15
1995
• •
9
12
15
0
6
3
3 6 9 12 15
Avg TIP Deep Sediment
(EPA HR Cores tPCB >100 pprn)
3 6 9 12 15
Avg TIP Deep Sediment
{EPA HR Cores tPCB >100 ppm)
Figure 17. Comparison of PCB congener composition between average TIP deep dechlorinated
sediments and unaccounted-for TIP load (summer 1991 - 1996).
-------
1.0
0.1
0.8
0.4
0.2
0.0
Urgamouth Bat*
L
liTii
1 234S6789 10
1.0
1.0
PumpUnsMd
PM
(uj/g Bp)
0.8
0.8
0.8
0.8
0.4
A
0.4
0.2
f V
0.2
I I I 1
0.0
,1 1 1 V
0.0
.1111
1 23456789 10
1234S6789 10
1.0
0.8
0.6
0.4
0.2
0.0
Surfaca Sadiment
Daposit Faidiri
il
Ptnphyton
tuygOCl
1 23486789 10
12 3 466789 10
1.0
OA
0.6
0.0
Water Column
h
1 23456789 10
123456789 10
Figure 18. Steady state food web model simulation of TIP (exposure: realistic water, top 0-2 cm
of sediment, see Table 6).
-------
Largamoutn B*ss
0
1234587B9 10
Pumpklnsttd
(uo/gHpl
hi
123458789 10
123466789 10
1.0
0.8
0.6
0.4
0.2
0.0
Surfact S«dim«nt
Dapotit Ft«dm
il
Ptriphyton
Cug/gOCi
Lu
123466789 10
123466789 10
1.0
0.6
0.6
0.4
0.2
0.0
"" ' ' ' ' 1 ¦ ' ¦ '
S»dlmani
1.0
Water Column
OA
Dissolved PCB
(ng/l)
0.6
0.4
¦
11.
OJ
1 I 1 1 ¦ -
1
0.0
123456789 10
123466789 10
Figure 19. Steady state food web model simulation of TIP (exposure: realistic water, top 0-5 cm
of sediment, see Table 6)
-------
1.0
o.e
0.8
0.4
0.2
0.0
Largtmouth But
z
' / N *•%
Ht£
JtujM
123*58789 10
c
o
'€
o
Q.
O
PumpkktaMtf
(ug/g Hp)
lLli
1 23456789 10
123456789 10
OJ
*5
£
1.0
0J
o.e
0.4
0.2
0.0
Surface Sediment
Deposit Feeder*
1_I
1 2 3 4 6 8
7 8 9 10
to
OA
Sedbnent
o.e
0.4
. I
0.2
0.0
III..
Periphyton
(ug/gOC)
111
1.0
OA
0.«
04
0J
Oil
123468789 10
Water Column
Dissolved KB
(nfl/U
_L
123456789 10
1234SBJ«»10
Figure 20. Steady state food web model simulation of TIP (exposure: realistic water, heavily
dechlorinated sediments, see Table 6).
-------
Largamouth Bats
y
12346*789 10
Pumpktruaed
(ua/fl Bp)
123456789 10
123488789 10
1.0
0.8.
0.8
0.4
0.2
0.0
Surfaca Sadbnant
~•posit Faadari
I »
panpnytnit
(ug/gOCl
123466789 10
123468789 10
Sadmant
1.0
0.8
0.8
0.4
0.2
0.0
Watar Column
Oiaaohrad PCB
(nfl/LJ
123468789 10
123460789 10
Figure 21. Steady state food web model simulation of TIP (exposure: heavily dechlorinated
sediments and partitioned water, see Table 6).
-------
10 0
60:49
<
oe
<
z
cc
o
u
UJ
Q
o
H
<
CC
Z
o
p
<
z
cc
o
_l
X
o
UJ
Q
56:49
.
66:49
74:49
NUMBER CHLORINES
PER BIPHENYL
NUMBER CHLORINES
PER BIPHENYL
Data are Geometric Means +1-2 Standard Errors (0.5 CL/BP bins)
Horizontal Dashed Line Represents Ratio in Aroclor 1242
Figure 22. Relationship between PCB congener dechlorination ratios and number of chlorines
per biphenyl in EPA Phase II high resolution sediment cores.
-------
10 0
86:49
60:49
'1-r-r-
TT-T
e
0*
£
z
•
o
1 10'
r
z
o
-
-
—1
•
z
m
o
11J
lil
o
-
.
10 2
i , .
i r i
10
io •'
r-T-r-
Jlllll 1 1 JM1I1,
i
•
1 1 1
1 1 1
66:49
74:49
io o
<
cc
<
2
cc
o
X
o
LU
Q
10 •'
¦ niim )ilium
"im' '
iii
jrwgjypomrr
l jiiin| mi mm
"I"' *
¦" r
|iiiiii j "i pun
10 ®
7
•
**
-
•
¦
-
-
to •«
-
-
-1,,! 1
mid i iiiiiiJ
i i i
« * >"'l
0.1 1 10 20 50 80 90 99 99.9
PROBABILITY
0.1 1 10 20 50 80 90 99 89.9
PROBABILITY
Upper Hudson (RM > - 153)
Rgure 23. Probability distribution of PCB congener dechlorination ratio in 1993 NOAA fish
samples.
-------
All Data Corractad for Analytical Biaa
10
3 .
¦ USGS data batwaan Fort Edward and Schuytorvflto
• GE data batwaan Fort Edward and Thompson Island Dam
2 .
T
¦
i
80
¦
81
82 83 84 85 86 87
T
¦
1
! L_
88 89
Year
90
T
1
91 A 92
r
93
94
T
_1
95 96
Flow at Fort Edward < 10.000 cf«. High load* from 9/91 to 12/91 axchided
-------
GENERAL ELECTRIC COMPANY
HUDSON FALLS PLANT SITE
Seep 13 DNAPL Recovery
o J
MONTH (1996-1997)
Figure 25. Cumulative PCB DNAPL oils collected from seep 13 (1996-1997).
-------
20000
16000
12000
8000
4000
30
25
35
4
S
15
20
10
Julian Day (Starting 4/1/92)
PCB data are corrected for analytical bias
Figure 26. PCB and solids transport during 1992 spring high flow.
-------
Fluorescent Particle Mass Balance
Transported Downstream of
Thompson Island Oam
2%
Retained between Hudson Falls to
Fort Edward
30%
Retained in Thompson Island Pool
51*
Retained between Fort Edward and
Roger* Island
17%
Figure 27. PCB DNAPL transport study fluorescent particle mass balance
-------
3.5,
Thompson
Island Pool
Batten Kill
r
Hoosic River
T
ID
> 1.0
0.5
_L
Conservative Load
Run x0297-1
• EPA estimates
J '
196
190
18S 180 175
Hudson River Mile
170
16S
160
165
Figure 28. Predicted average June-August 1993 Hudson River PCB loading profile (river mile
195 - 155).
-------
Thompson
island pom
Batten Kill
Hoos
c River
3.5
3.0
2.5
5
¦O
n
m
o
2.0
O
a
1.0
—.Conservative Load
Bun x0397-1
— Run *0397-4
0.S
0.0
1S5
ISO
176
170
10S
200
190
19S
ISO
1SS
Hudson River Mile
Figure 29. Predicted settling and volatilization components of average June-August 1993
Hudson River PCB loading profile (river mile 195-155).
-------
(0
c
'C
o
xz
o
200
-------
0
1
z
o
p
<
z
S
o
—I
X
o
Ui
Q
56:49
10
10 -1
10
-2
Hit
^ ^ .I.II-J
10 1
60:49
10 0
10
10 -2
<
'Hi
.
i
:
-
*
66:49
74:49
o
P
<
GC
z
o
I-
<
z
cc
o
-J
X
o
m I 10 1
'Miinq Miinj i Miiiq i imq iinnq iiinnq iimq liiiii^ imiiq iiimq iiniiq inui^ i i mi iiiiw
10 0
. ...uJ . ...mJ J J
1 ¦
10 •'
10 -*
-1 -> ...1^ MIlJ .I.IMJ II
10"* 10 3 10 2 1010° 101 102
10
TOTAL PCBs
(ppm dry)
TOTAL PCBs
(ppm diy)
USEPA Phase II High Resolution Sediment Core Data (UPPER HUDSON)
Data are Geometric Means +1-2 Standard Errors (0.5 log unit bins)
Vertical Dashed Line Represents Total PCBs = 30 ppm dry
Horizontal Dashed Line Represents Ratio in Aroclor 1242
Figure 31. Intra-homologue PCB dechlorination peak ratio as a function of PCB concentration.
-------
X
fc
a 6
. o
a"
%
a
$
s.§
x
fc
q|
10
30
30
40
56:49
fgq llll^ I Hill
60:49
63:49
3
m
i—»—<
to"* 10 *" 10 ¦* 10 •' 10* «' 10
66:49
111)^1 rum
10
»
38
to'1 to"* 10¦* to"' 10* 10 ' 10
77:49
iiiM^np^rnn| iiiim
10
20
30
mt
t-p-l
»
!•
40
10"* 104 104 10"' 10* 10* 101
DECHLORINATION RATIO
16
20
30
40
i nil nu
mm him
t- ~- t
1
10"* 10 * 10 •* io'1 10 8 10* 10 '
67:49
°- ii»i imi ii"i hiw
10
20
30
40
ii
AiiiJf nilai-
10"* 10 "® 10 ¦' 10"1 10* 10 ' 10
105:110
o
10
20
30
40
I itnq I nil I liMf
i
mri i—1 ""¦* '*"¦
10"* 10* 10 '* 1010* 101 10 1
DECHLORINATION RATIO
10
20
30
mil IfII
74:49
mli I >!¦» AW
10 10 10 10 10 10 10
114:110
10
20
30
10"* 10"* 10 10" 10* 10' 10'
DECHLORINATION RATIO
Figure 32. intra-homoiogue PCB dechlorination peak ratio as a function of sediment depth for
core segments with total PCB less then 30 pom.
-------
a ppi?ivrriTr,T?Q
rV.X I JtLi i mJm. xL»J
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
APPENDIX A
SUMMARY OF ANALYTICAL BIAS CORRECTIONS FOR
USGS AND GE DATABASE
Corrections for analytical biases in both the USGS and GE water column databases were
developed to support the data analyses presented in this document. Summaries of the rationale and
data used to develop these corrections are presented below. More detailed reports documenting the
development of these corrections will be presented in the future.
A. Quantification of The Analytical Bias in The USGS Data
The USGS data contains analytical biases, however, we believe that these biases can be
bounded and used to "correct" the existing database. The analytical biases within the USGS
database relate to the packed column methodology for PCB separation employed during the period
of 1977 to 1989. This separation and quantitation method failed to resolve the mono-chlorinated
biphenyls as they elute with the solvent peak on the packed column. Based upon the PCB loading
patterns observed in the Hudson, this bias would tend to underestimate the PCB loading from
sediment porewater, which is enriched in mono-chlorinated biphenyls.
In anticipation of these potential biases, GE analyzed a subset of the water column data
collected over a five month period during 1991 by both congener methods (NEA608CAP) and
A-l
-------
packed column USGS methods (Schroeder and Barnes. 1983). These paired data provide a means
of quantifying the low bias in the USGS data.
Based upon linear regression analysis of the difference between paired USGS and capillary
column PCB data and the weight percent of homologues quantified by the capillary column method,
the low bias in the USGS data (Figure A-l) appears to be strongly correlated with the concentration
of mono-chlonnated biphenvls in the sample (Figure A-2). Addition of dichlorinated biphenvls to
the regression strengthens the correlation suggesting that a portion of the bias may also be
attributable to an underquantification of the dichlorinated biphenvls (Figure A-3). Therefore, it
appears that the bias associated with the USGS data is directly related to the concentration of mono-,
and to a lesser extent, the dichlorinated biphenvls. Since there is no direct measure of the mono-
chlorinated biphenvls in the historical USGS data, a straight linear regression between capillary
column and USGS total PCB concentrations was performed to assess the analytical bias in the USGS
data. This regression was performed for each sampling station due to differences observed in mono-
and dichlorinated PCB concentrations at the different stations in subsequent congener PCB
monitoring programs.
The results of the regression analyses for the different stations appear in Figure A-4. At the
Fort Edward station, where water column samples contain only a small amount of mono- and di-
chlorinated biphenvls, the low bias in the USGS methodology is estimated at 29 percent (bias
determined by the slope of the relationship). Similarly, PCB measurements by USGS methodologies
underestimate Schuvlerville PCB concentrations by an estimated 40 percent. Stillwater and
-------
Waterford correction factors were similar to those of Schuylerville. Biases at downstream stations
are greater than those calculated for the Fort Edward station due to the higher proportion of mono-
chlonnated biphenvls in samples from these downstream stations. These correction factors were
applied to the USGS data for the entire period of record (1977-1995). The corrected data were then
used to analyze long term spatial and temporal patterns in PCB loading in the upper Hudson River,
specifically, the historical loading observed from the TIP region of the river.
B. Quantification of the Analytical Bias in NEA Peaks 5, 8. and 14
Comparison of 1993 water column PCB concentrations measured at the Fort Edward and
Thompson Island Dam stations by GE with those measured in the EPA Phase II study suggested that
analytical biases were present in the GE data set. Although total PCB levels and homologue
distributions in the two data sets exhibited consistency in magnitude and temporal trends,
examination of specific dechlorination products suggested that analytical biases are manifested in
individual PCB congeners. Differences between GE and EPA data for capillary column peak 5
(PK.5), which contains PCB congeners 2.2' dichlorobiphenvl (BZ4) and 2.6 dichlorobiphenvl
(BZ10), are especially evident in the 1993 data from TID (Figure A-5).
Biases in individual congeners may significantly affect data analyses used in developing an
understanding of PCB fate and transport mechanisms in the Thompson Island Pool (TIP). For the
case of PK5, the low bias of the GE data (Figure A-5) causes the TIP loading to be underestimated.
A-3
-------
and may significantly affect conclusions drawn from data analyses regarding the relative importance
of sediment diffusive flux and dechlorination as PCB fate and transport mechanisms.
The primary cause of analytical biases in the GE data is related to the capillary column (DB-
1) method used to separate PCB congeners. Coelution of congeners with differing relative response
factors (RRFs) causes DB-1 results to be sensitive to the assumption made regarding their relative
amounts within the given peak. Also, coeluting congeners can cause the shape of a chromatograph
peak to deviate from an ideal Gaussian distribution, resulting in area calculation errors. Biases in
the GE data set were also attributed to errors in the original Mullin calibration of the PCB standard
used in DB-1 analyses. The weight percent of peak 5 components within the 25:18:18 mixture of
Aroclors 1232. 1248. and 1262 of the Green Bay Mass Balance Study mixed Aroclor standard (US
EPA, 1987) was apparently miscalculated.
Preliminary attempts to "correct" the GE database for analytical biases were focused on the
revised Mullin calibration of the Green Bay Standard (Mullin. 1994). However, temporal trends in
1993 GE water column data recomputed with the revised calibration did not compare well with EPA
data. Remaining differences were attributed to differences in RRFs among individual congeners
within a given DB-1 peak containing coeluting congeners, as the ratio of coeluting congeners can
be altered upon dechlorination or by other processes.
To further explore the coelution issue, peaks with coeluting congeners were first ranked
based on their potential for analytical bias. Archived extracts from Hudson River water column
A-4
-------
samples were then reanalyzed in the laboratory to separate out coeluting congeners from selected
target peaks. Regression analyses were used to quantify single peak analytical biases by relating
DB-1 measurements to congener sums.
Peaks targeted for reanalysis were ranked by a surrogate parameter chosen to reflect their
contribution to PCB loadings in the TrP and the effects of coelution. For peak /, containingy=/...,H
coeluting congeners, the potential bias index. 0, was defined as the product of its relative range in
congener RRFs and its average weight percent (IVTID) in 1991-96 summer low flow GE water
column PCBs measured at TID:
$ = WTID
RRF.n. - RRF.-,
ric to T(t>eo
t
RRF
»e n
*100%
The DB-1 peaks with the five highest 's are listed in Table A-l.
A-5
-------
Table A-l.
DB-1 PCB Peaks with the highest potential for analytical bias in the
GE water column dataload. Congener RRFs obtained from Table 7-1V
(Erickson, 1992).
DB-1
Peak#
IUPAC Congener Us
Congener RRFs
Relative
Range in
Avg
TID
d>
1st
2nd
3rd
4th
1st
; 2nd 1
1 1
3rd !
4th
RRFs (%)
WT%
5
4
10
0.037
0.262
150.03
10.15
1523
S
5
. 8
0.119
0.206
53.54
7.90
423
14
15
18
0.107
0.313
98.10
3.68
361
25
20
21
33
53
0.724
1.060
0.361
0.447
107.93
3.16
341
17
16
32
0.447
0.278
46.62
5.68
265
26
22
51
1.094
0.600
58.28
2.82
164
] 993 water column data for the five peaks listed in Table A-l were compared with measured
values from the EPA data set and the largest biases were found to be in peaks 5 (low). 8 (high), and
14 (low) Therefore, these peaks were selected as target peaks for further analyses.
Laboratory separation of the congeners within target peaks was performed on a CP-SIL
5/C18 (CI8) gas chromatograph column, manufactured by Chromopack. Inc. This column was
selected primarily on its ability to resolve low molecular weight PCB congeners including those
coeluting in peaks 5, 8, and 14
Historical archived GE Hudson River water column sample extracts selected for laboratory
reanalvsis included recent samples (1995-1996) with total PCBs greater than 40 ng/L collected from
FE. TrD. and the Hudson Falls plunge pool, and historical paired samples collected from FE and TrD
A-6
-------
during the summer low flow periods of 1991-1996. The paired samples chosen for reanalysis
exhibited a strong TIP loading signal (i.e., large difference in total PCBs between FE and TID). The
selected data set enabled examination of the variation in single peak correction factors for different
time periods and sampling locations, and for data that were expected to have significantly different
PCB compositions.
Comparison of results from DB-1 analyses of archived extracts with original data indicated
that the laboratory achieved good ana I vie recovery (i.e., storage loss was not significant). Extracts
were reanalyzed on the CI8 column, and linear regression analyses were performed to relate CI8
congener sums with DB-1 results. Regression analyses for the three target peaks (PK5. PK8. PK.14)
are plotted in Figure A-6. Results from the regression plots suggest that the analytical bias in the
target peaks is systematic and independent of sample time and location (i.e., correlation coefficients
close to unity and small y-intercepts). Based on these results, correction factors were developed to
account for analytical biases in peaks 5 (3X). 8 (0.5X), and 14 (1.5X). Regression statistics for the
three target peaks are summarized m Table A-2.
Table A-2
Statistics for regression of DB-1 and CI8 data.
DB-1
3eak
#
Structure •
of PCB
Congeners i
Reanalyzed Extract Data
Regression Statistics
Number Max Cone.: Min Cone., Slope y*lntercept
[ng/LJ [ng/L] [ng/L]
R1 Standard
y-Error [ng/L]
Significance F
(P-value)
5
2.2'+ 2,6
38 30.7 0.0 2.94 >
1.4
0.931 5.8
1.7E-22
8
2.3 + 2.4'
38 102.7 0.0 0.47
0.3
0.995 0.9
9.9E-44
14
4.4' + 2.2'.5
38 83.8 0.9 1.53
-1.2
0.996 1 6
3.0E-45
A-7
10.0471
-------
The correction factors developed from regression line slopes were applied to the 1993 GE
water column data set for comparison with GE data from the same period. 1993 total and target peak
PCB water column concentrations are plotted in Figures A-7 and A-8 for the FE and TID sampling
stations, respectively. Inspection of Figures A-7 and A-8 suggests that application of the correction
factors to the GE data set dramatically improved its comparability with the EPA data. This
improvement is most notable for PCB data collected at TID.
A-8
-------
APPENDIX B
DNAPL SOURCES IN THE VICINITY OF HUDSON FALLS
Source Identification:
In September 1991, elevated river water levels of PCBs were detected by GE upstream of
the contaminated sediments in the TIP. Intense investigations localized the source area to the eastern
shoreline near river mile 196.8, in the vicinity of the GE Hudson Falls plant site. Access to this area
was hampered due to complex site geography, the presence of a steep cliff and the Baker's Falls, a
150 year old mill (Allen Mill) in poor structural condition and several hydraulic conduits.
The results of these investigations revealed the presence of active seeps of Dense Non-
Aqueous Phase Liquids (DNAPL) along the eastern cliff face and the rock face of the eastern
raceway within the Allen Mill. In addition, free phase PCB oil (Aroclor 1242) and oil- contaminated
sediments (up to 70,000 ppm) were found within the Mill and in the tailrace tunnel (the tailrace
tunnel, a 200 foot tunnel discovered below the Mill in the fall of 1992. outlets into the plunge pool
at the base of Baker's Falls).
As part of the reconstruction of the Baker's Falls dam by Adirondack Hydro Development
Corporation, the eastern portion of the Falls was dewatered in the first quarter of 1996. revealing
additional seeps in the river bottom.
B-l
-------
Remediation:
A number of different remedial measures have been implemented to mitigate the seepage of
PCBs from the vicinity of the plant site to the River. It is now recognized that DNAPL is present
in fractured bedrock below the site. Remedial efforts are briefly summarized below: 1) DNAPL
seepage from the rock face of the eastern raceway is now routinely captured; 2) hydraulic control
of conduits through the Mill was achieved in 1993 to allow access and additional investigation of
the Mill; 3) a slurry wall was constructed within the eastern raceway in 1994 to reduce river seeps
from this region; 4) removal of DNAPL and oil-contaminated sediments from the Allen Mill
containing 50 tons of PCBs was completed in 1995 (this material represents approximately twice
the amount of PCBs estimated to reside in all of the TIP sediments); 5) construction of a WWTP to
allow expansion of recovery efforts at the site in 1995; 6) installation of DNAPL-recovery wells in
the vicinity of the plant site that have recovered >8000L of DNAPL to date; and 7) the ongoing
installation of barrier wells utilizing hydraulic control to further reduce DNAPL transport through
subsurface fractures. Clearly these remedial efforts have reduced the PCB loading of
undechlonnated Aroclor 1242 to the Hudson River, but it is not yet possible to determine the degree
of control that has been achieved.
Recent Efforts:
Additional chemical characterization of DNAPL fluids from seeps and subsurface wells has
recently been undertaken to aid DNAPL control efforts at the site. Chemical characterization of
these fluids by GC/MS has identified the major chemical components of the DNAPL in the bedrock
at Hudson Falls to be Aroclor 1242 (PCBs). phenyl-xvlvl ethane (PXE). bis-(2-ethvlhexyl) phthalate
B-2
-------
(BEHP) and tnchlorobenzene (TCB) The level of PCBs in the DNAPL ranged from 12 to 100
weight percent. The other components also varied considerably. Although there was considerable
variation in the composition of the DNAPL recovered from different locations at Hudson Falls, the
composition of the DNAPL falls into four distinct categories.
The four categories of DNAPL recovered from 38 different locations, including wells and
in-river seeps, can be defined as !) PCBs only (primarily PCBs), 2) all components (PCBs. PXE,
BEHP and TCB), 3) all components less TCB (PCBs. PXE and BEHP), and 4) low PCBs
(containing less than 50% PCBs. and variable quantities of PXE. BEHP and TCB). The presence
of these four categones of DNAPL suggest that there are at least four unique DNAPL reservoirs
present in the vicinity of the plant site These DNAPL reservoirs are located as follows and are
shown in Figure B-l:
TABLE B-l
Chemical characterization of DNAPL reservoirs in the
vicinity of the Hudson Falls Plant Site.
DNAPL composition
Location
PCBs only
Seeps I & 5; vertical seep control
borings in tunnel; Eastern Raceway
All components
Seep 13; RW-104; angled seep
control borings in tunnel
Ail components less TCB
RW-100& east of RR tracks
Low PCBs
Bldg I & south of John St.
B-3
-------
Note that the river seeps fall into two distinct reservoirs. Seep 13 was discovered in the fall
of 1996 by divers at the base of Baker's Falls. DNAPL collected from this river seep totaled over
16 liters by early January 1997. This represented -0.5 lbs/day of potential PCB loading to the river
that was being captured since its discovery in late 1996. The location and chemical characterization
of this seep was distinct from the other river seeps, suggesting additional controls would be
necessary to capture this material. Recovery well RW-104 was installed to capture this material and
it quickly began capturing DNAPL with the same chemical composition as that previously collected
from Seep 13. Moreover, production ceased from Seep 13 in nearly the same time frame. The new
recovery wells in the Eastern Raceway, installed as a portion of the barrier wells near the River's
edge, may also be capturing DNAPL that is impacting Seeps 1 and 5.
These recent results suggest that targeted remedial activities at the Hudson Falls site are
currently reducing the upstream source responsible for the contamination of surface sediments in the
TIP. As these surface sediments represent a source of PCBs to the biota and water column from the
TIP. impacting the upstream source (undechlorinated Aroclor 1242) should have a direct benefit on
water and fish PCB levels. The negative impact on both of these media was observed after the Allen
Mill releases in 1991-1993. We would expect correspondingly positive impacts on both media due
to the recent controls implemented on the Hudson Falls plant site.
B-4
-------
APPENDIX C
APPLICATION OF INTRA-HOMOLOGUE PEAK RATIOS TO
CHARACTERIZE PCB SOURCES
ABSTRACT
The origin of the persistent PCB levels in Hudson River fish has remained controversial:
primarily, we believe, for lack of chemical "fingerprinting" procedures that would permit
distinguishing between alternative sources for the fishes' PCBs. Past attempts to provide such
fingerprinting via descriptions of PCB congener distribution or principal components analysis have
been generally unproductive: largely, it now appears, because of data confounding by variabilities
in such processes as elutnation. bioaccumulation. and depuration. Since these processes impact
much more heavily on inter-homologue ratios than on intra-homologue. or isomer, ratios, we have
explored the use of isomer ratio data sets as information indicators of the environmental alteration
state, and hence environmental pathways, taken by the fishes' PCBs. Examination of over 300 such
data sets, determined for the PCBs in Hudson River fish belonging to 30 species. 21 genera, and 11
families, collected over a 200-mile stretch of the river over a 16-vear period, showed that the resident
fishes' isomer ratio "fingerprints" have generally corresponded to those of the local surficial
sediments in all sections of the river, except as altered by metabolic processes that were found
characteristic of 9 of the 21 genera studied. Since 1977, the PCBs of the fish of the Thompson
Island Pool (upper Hudson River Reach 8) have exhibited surficial sediment Pattern A, indicative
of recently deposited Arocior 1242. Those of fish taken a hundred miles downstream in the mid-
estuarv have instead exhibited subsurfictal sediment dechlorination Pattern H\ indicative of PCB
C-l
-------
compositions, such as hydraulic fluids, that had long been present in the sediments. In between,
there has been a smooth transition in pattern, indicating a decrease in the extent of fish PCB
dechlorination with decreasing distance from the known source of undechlonnated Aroclor 1242
input at Hudson Falls.
INTRODUCTION
The Hudson River is the major waterway draining eastern New York State (Figure C-l). Its
fish have been known to be carrying elevated levels of PCBs (polvchlorinated biphenvls) since 1970
(1) All industrial uses of PCBs ceased in the 1970's and the levels of PCBs in Hudson River Fish
have declined only slowly since the mid 1980s. Continuing controversies as to the sources of the
fishes' PCBs revolve around such questions as: whether the PCBs in upper Hudson fish are coming
from old local high level sediment deposits ("hot spots") or from ongoing drainage from rock
fractures under a contaminated plant site; whether the PCBs in lower Hudson resident fish are
coming from old deposits in the local sediments or from ongoing inputs from the upper Hudson: and
whether the PCBs in lower Hudson migrator.' fish are coming from etiher of these sources, or from
the sediments and sewers of the New York metropolitan area.
The commercial PCB products (e.g. Aroclors) that were released into the environment were
complex mixtures of isomers (PCBs of the same Cl-Number) and homologues (PCBs of different
Cl-Number), which are genericallv referred to as congeners. The original distributions of such
congeners can be altered by biological processes in each of the environmental compartments through
which the PCBs may pass, e.g., by aerobic microbial biodegradation near the sediment surface
C-2
-------
(4,5.6); by anaerobic microbial dechlorination in subsurficial sediments (7,8.9.10); by limited
microsomal metabolism in some fish species (11); and by more extensive microsomal metabolism
in crustaceans, piscivores, and man (11. 12). Since each of these processes alters the PCB congener
distribution in a different way. it should, in principle, be possible to identify the set of niches through
which any environmental PCB has passed via observable alterations in congener distribution.
In practice, this has generally proved difficult, primarily because K^-dependent phenomena
such as evaporation, elutriation. bioaccumulation. and depuration, as well as variations in Aroclor
proportions in the original release, can produce variations in homologue distributions large enough
to obscure the effects on congener distribution produced by niche-specific biological activities. It
occurred to us. however, that this problem could be minimized by simply using isomer ratios rather
than congener levels as indices of chemical composition. Accordingly, we undertook to determine
enough PCB isomer ratios on enough types of fish samples to determine whether an isomer ratio data
set could indeed provide a robust indicator of PCB source and alteration history.
MATERIALS AND METHODS
Site Description. - The lower 156 mi. (251 km) of the Hudson River, from Troy to New York
City, is a tidal estuary herein referred to as the lower Hudson (see Figure C-1). The next ca. 80 mi.
(ca. 125 km), i.e.. the lower part of the upper Hudson, consists of a series of dammed stillwaters
called "reaches," numbered in order starting from the Federal Dam at Troy. Distances along the
Hudson are measured as "river miles" (RM) starting from the Battery at the southern end of
C-3
-------
Manhattan Island (New York City). Several descriptions of the contamination of the Hudson River
by PCBs have been published (13, 14. 15, 16, 17).
Fish Data Used. - The fish samples or analytical data used in this study came from
collections made by seven other investigators:
1. From R.J. Sloan of the New York State Dept. of Environmental Conservation (DEC) we
obtained about 800 archived analytical extracts of fish from his 1977-82 collections (2) that had been
returned by his analyst after low resolution packed column analysis. We selected 75 specimens that
reflected a variety of fish species. PCB levels, and Aroclor ratios and submitted them to Northeast
Analytical Services of Schenectady. XY (NEA) for DB-1 capillary gas chromatographic (GC)
analysis by described procedures (18). These analyses revealed that a few of the extracts, notably
those of the goldfish and eels, had been allowed to dry out and lost lower congeners, but that most
still exhibited homologue distributions in accord with the original "Aroclor" determinations.
2. From P.A. Jones, also of DEC. we obtained splits of the fathead minnow samples
collected in his 1985 study (19) of Hudson River PCB uptake by caged minnows, which were
analyzed at GE (18).
3. From J.M. O'Connor, then of New York University, we obtained the original packed
column GCs of the gammarns collected in 1980 as a part of the NYU Hudson River Survey (20).
C-4
-------
He also supplied us with a dozen lower Hudson striped bass, collected in 1985. that were analyzed
at GE (18).
4. From B.K. Shephard. then of Harza Engineering Co., we obtained both NEA GCs and
data for samples of sediments, Hester-Dcndy (periphyton) plates, dialysis bags, invertebrates, and
fish collected during his 1988-1998 survey of PCBs in the lower Hudson River, New York Harbor,
and western Long Island Sound.
5. From J .G. Haggard of General Electric we obtained 90 frozen fish that had been collected
from upper Hudson Reaches 1-11 by Law Environmental Services in 1990. These were submitted
to NEA for a 118-peak DB-1 analysis (18). along with a separate analysis for congener 77. which
is not well resolved from PCB 110 on DB-1.
6. From W.A. Avling of O'Brien and Gere Engineers. Inc. of Syracuse. NY we obtained
NEA chromatograms and data for fish, invertebrates, and sediment surface scrapings collected from
the Thompson Island Pool in May, 1992. eight months after the major PCB loading event of
September, 1991 (21).
7. From L.J. Field of NOAA. Seattle, WA. we received 145-congener dual column PCB data
files for 115 fish samples that he had collected in collaboration with RJ Sloan (DEC) at ten
collection stations between RM 40 and 200 on the upper and lower Hudson in the autumn of 1993.
C-5
-------
The sediment reference samples for A. B, C, H and H' alterations were taken from individual
core sections that exhibited these patterns as previously described (7-91.
Data Processing. From each PCB congener data set we calculated, if not already provided,
the total PCB level, the PCB/lipid ratio, the levels of the homologue groups, the ratios between
successive homologue groups, and the ratios of about 40 of the stronger single congener peaks to
those of a selected isomeric reference congener, as well as site and species averages. The selected
reference congeners) were, for the tn-CB (hereinafter CB(s) = chlorobiphenvl(s)). the sum of PCBs
28 + 31 (these are normally the highest and second highest level tn-CBs. respectively; but they elute
so closely on a DB-1 column that we were concerned about the reliability of the peak splitting
calculation); for most tetra-CBs. PCB 49 (which maintains a relatively constant level during the
early stages of dechlorination), for the tetra-CB PCB 70. which is readily metabolized by the AP-
ICT activity, the non-metabolized PCB 74 (which is also more similar in to PCB 70 than PCB
49); for most penta-CBs. the rather slowly dechlonnated PCB 110. with PCB 99 as a non-
metabolizeable alternate; and for the hexa-CBs. PCB 153.
Adjustment for Reference Congener Depuration - We noted that the ratio between isomers
74 and 49. which differ somewhat in water solubility, became elevated in individual fish that were
heavily depurated, as indicated by low levels of di- and tn-CB's. In such fish the elevation in log
(PCB 74/PCB 49) averaged about 0.2 times that in log (tetra-CB/tri-CB). Accordingly, a possible
depuration adjustment to the 74/49 ratio was calculated on that basis.
C-6
-------
RESULTS
Table C-l lists the fish species examined, abbreviations used, and metabolic alteration
patterns observed. Table C-2 presents the mean values of the upper and lower Hudson River fish
PCB homologue levels and selected PCB isomer ratios for the 1977-78, 1990, 1992, and 1993 fish
collections, along with reference values for a 90:10 Aroclor 1242:1254 mix and Hudson River
sediments exhibiting alteration patterns A. B, C, H. and H'. The variability of all tetra-CB and some
penta-CB homologue levels, and the upper Hudson PCB isomer ratios involving penta- and hexa-
CBs was low (5 - 20% Relative Standard Deviation (RSD)). Tri- and hexa-CB homologue levels,
and the other isomer ratios involving tetra-. penta-, or hexa-CBs displayed somewhat greater
variability (20 - 40% RSD). Generally, however, the % RSD's were only about half as great for the
isomer ratios as for the homologue levels. Much of this remaining variance in the Table C-2 isomer
ratio data, which arose from measurements of the PCBs in many different species of fish, taken over
large geographical ranges, could be correlated with specific variables. These will be considered in
tum.
Variations arising from PCB depuration. - Some of the individual fish collected from the
upper Hudson in November, 1990 showed levels of tri-CBs that were reduced to as little as 10% of
their usual values, and displayed even greater reductions in di-CBs. Such reductions in lower
homologue levels occurred most frequently in walley (WAL), smallmouth bass (SWM), and
largemouth bass (LMB), and significantly influenced the average lower homologue levels reported
for 1990, since SMB had been selected as the species to be measured in triplicate in every reach of
the upper Hudson. The depurative losses could have resulted from either a late-season cessation of
C-l
-------
feeding, or from periods of feeding in uncontaminated tributary streams. These losses were much
less prominent in the 1977-1978. 1992, and 1993 collections. However, the observation of the
effects they might have on the PCB 74/49 isomer ratio prompted the inclusion of a possible 74/49
ratio adjustment for depuration in Table C-2.
Variations arising from atypical Aroclor inputs. Occasional fish in most collections
exhibited congener profiles clearly divergent from the majority. Thus, most of the very lightly
contaminated (
-------
elutriative. evaporative, or depurative losses of lower congeners. Evidently, the original source of
the lower Hudson River fish PCBs had been an Aroclor mix containing higher proportions of more
heavily chlorinated Aroclors than that contaminating the upper Hudson.
Variations correlatable with fish species or genus. The PCBs in certain of the fish species
showed consistent depletions of particular groups of congeners, thus defining an alteration pattern
(AP), presumably arising from a species-specific PCB metabolizing activity (Table C-l). The
commonest. Pattern AP-ICT (for Ictalums. the first genus in which noted) was previously designed
"P450-1 A-like" (11); however, that term now seems better restricted to the somewhat different
pattern seen in higher animals (12). AP-ICT shows a marked reduction in PCB 70 (and hence the
70/74 ratio) and lesser reductions in PCBs 16. 17. 18, 22. 27. 33, 40. 49. 56, 91, 97. and possibly
101, 110. and 174; all of them congeners with adjacent unsubstituted 4-positions. By contrast. AP-
PET (for Petromvzon) showed reduced levels of the 4,4'-substituted PCBs 28, 74, 118, 105. 128,
167. and 156. and also of PCBs 49, 52. and 174. leaving the peak given by the coeluting pair. PCBs
64 - 71. as the strongest in the chromatogram. Pattern AP-ESX (for Esox where it appeared
occasionally) showed clear reductions in every resolved congener carrying a 2, 3-dichlorophenyl
group, i.e., PCBs 22, 40, 42, 44, 56, 82, 84, 97, and 129. Pattern AP-LEP (for Lepomis) showed
clear reductions in just two of the above, namely, PCBs 40 and 44. Finally in AP-CAT (for
Catostomns) the only clear reduction was in congener 52. Thus, the only observed fish alteration
patterns that would affect a PCB 74/49 isomer ratio were AP-ICT and AP-PET.
C-9
-------
A very different set of variations was observed in the anadromous (migratory) fish of the
lower Hudson River. The STB (for Striped Bass) and AMS (for American Shad) all showed
substantial levels of DDE. sometimes accompanied by DDD or DDT; rra/w-nonachlor, sometimes
accompanied by a- and y-chlordane; and other pesticides as well, generally producing enough
interfering peaks in the tri- through hexa-CB range to make calculation of isomer ratios from GC-
ECD data problematical. The observed pesticide/PCB ratios generally corresponded to those seen
in the sediments of the New York metropolitan area, including western Long Island Sound, which
is where these species overwinter. Conversely, the two Atlantic tomcod (ATT) examined, both
collected at RM 41 in January, 1978, showed only low levels of Aroclor 1242-like PCB
contamination, without any pesticides, not even the low level of DDE present in the lower Hudson.
Variations due to biodegradation/dechlorination state. Congener distributions in Hudson
River subsurface PCB dechlorination Patterns B and C (7,8,10) and H and H' (9,10) have been
previously described. Generally speaking, Pattern C dechlorination, which gives the most extensive
conversion to mono- and di-CBs. is seen in the most heavily contaminated sediments of the upper
Hudson; Patterns B and B' are seen in somewhat less heavily contaminated sediments as far
downstream as Albany; and the rather selective Patterns H and H' are uncommon in the upper
Hudson, but dominant in most of the more lightly contaminated lower Hudson (9). Bedard has
argued that these patterns may all result from the dechlorination activities of just three microbial
strains, all separable in anaerobic laboratory cultures, designated M, Q, and H or H' (10), with most
dechlorination of the higher congeners coming from the H/H' activity. This could explain why the
patterns of higher congener loss are essentially identical for the observed alterative patterns B, C,
C-10
-------
H, and H', even though the distributions of the more lightly chlorinated PCB congeners formed are
quite different.
The columns on the left side of Table C-2 present the homologue distributions and selected
PCB isomer ratios for some representative specimens of subsurface sediment PCBs exhibiting those
patterns, along with comparable data for a 90:10 Aroclor 1242:1254 mix, selected as a representative
example of an unaltered Aroclor'release. The marked compositional changes effected by anaerobic
dechlorination are evident.
Geographical variations in dechlorination state. Surficial PCB alteration Pattern A was noted
as far back as 1984 (7), but was not seen free from admixed Pattern B until recently. It has now been
observed in the 0-1 cm. sediment layers and "fluff' layers collected in the Thompson Island Pool
at the same time as the 1992 fish sampling, and repeatedly reproduced in the upper (0-5mm.),
presumably microaerobic, sediment layers in laboratory microcosms where upper Hudson sediments
were spiked with fresh Aroclor 1242 (21). The Pattern A alteration appeared in the microcosms
within six weeks. Its microbiological basis is uncertain; one speculation is that it arises from a
combination of an oxygen-tolerant, mera-selective dechlorination process followed by an aerobic
biodegradation of the most of the dechlorination products. In sediments, it appears to effect limited
removal of PCBs 17, 18, 33,97, 99, 101, 153, and 167 without attacking 40, 44, 56, 60, 66, 70, 74,
87, 105, 114, or 128, and to result in increases in 47, but not in 19 or 27. Its most sensitive indicator
is a depression in the ratio 33/28+31, or if separately measurable, just 33/28. Such depressions.
C-ll
-------
usually paralleled by increases in the 47/49 ratio, are almost universally observed in PCBs recovered
from aquatic environments.
Table C-2 shows that in general the isomer ratios observed in upper Hudson fish fell between
the values for fresh Aroclor and Pattern A altered PCB, with no obvious contribution from the
subsurface Pattern B and C PCBs, with their high levels of dechlorination product PCBs 19 and 27,
and low levels of readily dechlorinated PCBs 74, 87, 97, and 105. By contrast, the Table C-2 data
for isomer ratios in lower Hudson River fish generally fell between the values for dechlorination
patterns H and H', indicating as much dechlorination as in the local sediments.
The 1993 NOAA analyses by a dual column procedure permitted resolution of a number of
congeners that were less readily quantified by the NEA single column analyses. Table 3 presents
mean isomer ratios, calculated from the NOAA data, for 17 PCB isomers known (9) to be sensitive
to Pattern H/H' dechlorination at various stations between the Thompson Island Pool (RM -192)
and lona (RM 40). The levels of all of the dechlorination-sensitive congeners were found to
decrease .with distance downstream: most rapidly in the case of the toxic coplanar PCB 77; quite
slowly for congeners 118 or 138; and with the major mono-ortho tetra-CBs 56, 60, 66, 74, and
penta-CB 105 in between.
The 1993 EPA water and sediment analyses by the same procedure, as presented graphically
in Figs. 3-73 to 3-80 of a recent report (17), were noted to show the same changes for congeners 56.
60, 66. 70, and 74. Their levels in upper Hudson water at RM 177.8 were similar to those ir
C-12
-------
undechlorinated Aroclor 1242, while those in lower Hudson Pattern H/H' - dechlorinated sediments
at RM 143.5, 88.5 or 43.2 were only half as great.
Figure C-2 shows the trend for unadjusted 74/49 ratios not only for the 1993 NOAA fish, but
also for the 1990 and 1992 fish collections, the 1985 caged minnow study, and the 1989 gammarus,
periphyton, and dialysis bags. In all cases there appeared a smooth transition between values
characteristic of the Pattern A sediment surface of the Thompson Island Pool and the Pattern H/H'
dechlorinated sediments and water of the lower Hudson.
DISCUSSION
Utility of isomer ratio analysis. The above results show that PCB isomer ratio analysis can
be used to identify and quantify the niche-specific biological alteration processes to which an
environmental PCB release may be subjected. These processes include the ubiquitous but enigmatic,
possibly microaerophilic, microbial alteration process A at the sediment surface; the well-
characterized subsurficial anaerobic microbial dechlorination processes leading to alteration patterns
B, C, H or H'; the genus-dependent fish PCB alteration processes leading to the patterns AP-ICT,
AP-PET, AP-CAT, AP-ESX, and AP-LEP described above; and the microsomal P4502B-like
alteration process exhibited by many crustaceans (which are frequently prey of the fish examined
here) as well as by higher vertebrates (12). Characterization of such processes can be useful in
defining the set of environmental niches through which a PCB composition has passed on its way
from point of release to accumulation in a fish.
C-13
-------
One complication of isomer ratio analysis is interferences between the effects of different
processes, especially as they effect the reference congeners used. For example, reference tri-CB 31,
reference tetra-CB 49, and possibly penta-CB 110 are all potentially subject to metabolism by the
AP-ICT system, and the first two of these also to losses under conditions of heavy depuration. We
have presented here a procedure for correcting the 74/49 ratio for such losses of PCB 49 as a possible
adjustment. An alternative would be to simply avoid the use of heavily depurated individuals, or
of lampreys, eels, icterids or goldfish, in making quantitative determinations of local PCB alteration
state via isomer ratio analysis.
Previously, the most popular approach to handling environmental PCB congener distribution
data has been by principal components analysis. This defines PCB composition in terms of two or
three enigmatic "principal components." These may permit the grouping of samples into related
sets, but do little to explain the chemical nature of the differences. It is now evident why this
happens: there are simply many more significant alteration processes affecting PCB composition
than there are mathematically resolvable "components." so that the resolved "components"
inevitably represent combinations of the effects of multiple alteration processes in various
proportions.
Sources of the PCBs in Hudson River fish. The PCB (and pesticide) "fingerprinting"
provided by isomer ratio analysis, along with data from other environmental studies, shows that the
PCBs in Hudson River fish originate from four readily distinguished sources.
C-14
-------
The first, and least important of these, is atmospheric deposition. This PCB source is
characterized by homologue and pesticide distributions that are very different from those of the other
sources, and is responsible for the low level PCB (and DDE) contamination seen in the fish and
sediments of the upper reaches of the Hudson, the Mohawk River, and presumably other tributaries
as well.
The second identifiable source consists of the sewers and sediments of the New York
Metropolitan area, which is where two important anadromous fish of the lower Hudson, the striped
bass (STB) and the American shad (AMS) spend the winter before migrating upriver to spawn.
These sediments are known to contain substantial levels of DDT-derived, chlordane-derived, and
other pesticide residues (13), as well as PCB mixtures reflecting heavy contributions from Aroclors
1254 and 1260, which were particularly extensively used in railroad and substation transformers in
that area. The 1978, 1982, and 1985 STB and AMS in our collection generally showed PCB
m
homologue distribution and pesticide/PCB ratios comparable to those of the sediments of New Yort
Harbor and western Long Island Sound, indicating the significance of those sources.
The third distinguishable source consists of the moderately Pattern H/H'-dechlorinated PCB
of the sediments of the lower Hudson, which exhibit an isomer ratio "fingerprint" closely matchin
that of the lower Hudson resident fish. These PCBs have been there a long time. Radionuclid
dating has shown that most were deposited in the I950's and 1960's (14) and elevated levels in fis
were seen in 1970 and 1973 (1), all before the removal of the Ft. Edward dam and sedimei
scouring/'redeposition events of 1974-76 caused the heavy PCB contamination of the upper Hudsc
C-15
-------
(16). One original source is the formerly extensive industrial usage of PCB-based hydraulic systems
in many of the riverside communities. Available Monsanto sales records for 1957-1977 document
purchases by such users of over 3 x 106 lbs. of PCB products, especially the hydraulic fluid Pydraul
A-200, an Aroclor 1242-1248 blend. Releases of such compositions into the river would result in
the moderate elevations in higher homologue levels exhibited by lower Hudson sediments and fish
(TabJe C-2). A less plausible source would be inputs of dissolved PCBs eluted from upstream
deposits, since these are depleted, rather than enriched, in the higher homologues (17), and
substantially undechlorinated, as noted above.
The fourth fmgerprintable source of the 1977-1993 Hudson River fish PCBs consists of
Aroclor 1242-like compositions that have been on the surface of the sediments of the Thompson
Island pool only long enough to have undergone a limited Pattern A alteration, thus indicating a
continuing deposition. The ultimate source of this PCB input may be Aroclor 1242 seeping from
*
the fractured bedrock near the former capacitor manufacturing plant at Hudson Falls. Seepage from
this reservoir is now known to have been entering the River as droplets of undechlorinated oil-phase
Aroclor 1242 that contaminate surface sediments of the Thompson Island Pool. There, the PCBs
soon undergo Pattern A alteration and partial extraction into the water column and its biota, leading
to the appearance of undechlorinated Pattern A PCBs in the fish. Eventually, of course, ongoing
sedimentation covers each increment of PCB and allows anaerobic microbial dechlorination to the
Pattern B- or C-dechlorinated PCBs of the local subsurface accumulations.
C-16
-------
NYSDEC (New York State Dept. of Environmental Conservation) 1975. Monitoring of
PCBs in fish taken from the Hudson River. Albany, NY.
Sloan, R.J., Simpson, K.W., Schroeder, R.A. and Barnes, C.R., Bull. Environ. Contam.
Toxicol. 1983, 31: 377-385.
Armstrong, R. and Sloan, R.J., 1988. PCB Patterns in Hudson River Fish I, Resident-
Fresh Water Species, Chap. 12 in Fisheries Research in the Hudson River, Smith, C.L,
Ed., SUNY Press, Ithaca, NY.
Bedard, D.L., Unterman, R.D., Bopp, L.H., Brennan, M.J., Haberl, M.E. and Johnson,
C.J. Appl. Environ. Microbiol. 1986, 51: 761-768.
Harkness, M.R., McDermott, J.B., Abramowicz, D.A., Salvo, J.J., Flanagan, W.P.,
Stephens, M.L., Mondello, F.J., May, R.J., Lobos. J.H., Carroll. K.M., Brennan, M.J.,
Bracco, A.A., Fish, K.M., Warner, G.L., Wilson, P.R., Dietrich, D.K., Lin, D.T., Morgai
C.B. and Gately, W.L. Science 1993, 259: 503-507.
Flanagan, W.P. and May, R.J. Environ. Sci. Technol. 1993, 27: 2207-2212.
Brown, Jr., J.F., Wagner, R.E., Bedard, D.L., Brennan, M.J., Carnahan, J.C. and May,
R.J. Northeast. Environ. Sci. 1984, 3: 166-178.
Brown, Jr., J.F., Wagner, R.E., Feng, H., Bedard, D.L., Brennan, M.J., Carnahan, J.C. ai
May, R.J. Environ. Toxicol. Chem. 1987, 6: 579-593.
Brown, Jr., J.F. and Wagner, R.E. Environ. Toxicol. Chem. 1990. 9: 1215-1233.
Bedard, D.L. and Quensen, J.F. Ill, 1996. Microbial Reductive Dechlorination of
Polychlorinated Biphenyls. in Microbial Transformation and Degradation of Toxic
Organic Chemicals. Young, L.Y. and Cemiglia, C., Eds. Wiley, New York.
Brown Jr., J.F. Marine Environ. Res. 1992, 34: 261-266.
Brown, Jr. J.F. Environ. Sci. Technol. 1994, 28: 2295-2305.
Bopp, R.F., Simpson, H.J., Olson, C.R., Trier, R.M. and Kostyk, N. Environ. Sci.
Technol. 1981, 15: 210-216.
Bopp, R.F., Simpson, H.J., Olson, C.R., Trier, R.M. and Kostyk, N. Environ. Sci.
Technol. 1982, 16: 666-672.
C-17
-------
15. Brown, M.P., Werner. M.B., Sloan. R.J. and Simpson, K.W. Environ. Sci. Technol.
1985, 19: 656-661.
16. Sanders, J.E. Northeast. Environ. Sci. 1989, 8: 1-86.
17. USEPA, 1997. Region II. Hudson River PCBs Reassessment Rl/FS, Phase 2 Report,
Vol. 2C- Data Evaluation and Interpretation Report. New York.
18. Frame, G.M. II, Wagner, R.E., Carnahan, J.C., Brown, J.F. Jr., May, R.J., Smullen, L.A.
and Bedard, D.L. Chemosphere 1996, 33: 603-623.
19. Jones, P.A., Sloan, R.J.. and Brown, M.P. Environ. Toxicol. Chem. 1989, 8: 793-803.
20. O'Connor. J.M., Califano. R.J., Pizza. J.C.. Lee, C.C. and Peters, L. S. 1982. Final
Report. The Biology of PCBs in Hudson River Zooplankton, New York University, New
York.
21. O'Brien & Gere Engineers. Inc. 1993. Data Summary Report. Hudson River Sampling
and Analysis Program. J 992 Food Chain Study. General Electric Co., Albany, NY.
22. Fish, K.M.. and Principe, J.M. Appl. Environ. Microbiol. 1994, 60: 4289-4296.
23. Bokuniewicz, H.J. and Arnold, C.L. Northeast. Environ. Sci. 1984,3:185-190.
C-18
-------
APPENDIX D
BENEFITS OF PCB DECHLORINATION
Reduced Toxicity. The toxic effects produced by PCBs in inbred strains of laboratory
rodents, and possibly in a few wildlife species, are now generally recognized to be mediated by
binding to a particular cytoplasmic protein, called the Ah-receptor (AhR), which has the ability to
induce expression of cytochrome P450 isozymes 1A1, 1A2 and 1B1, and several Phase 2 drug-
metabolizing enzymes as well. The magnitude of this AhR-agonist activity is most commonly
determined by measuring the ability of a PCB congener (or other toxicant of concern) to induce the
expression of cytochrome P4501A1 in a test animal or cell culture, as indicated by ethoxyresorufin-
O-deethylase (EROD) activity, and generally reported as "toxic equivalency," e.g., the ratio of the
EROD activity exhibited by the test congener to that exhibited by dioxin. which is the strongest
known AhR agonist. Using this test, and direct measures of toxic response, the particular PCB
congeners that are the most active of Ah-receptor agonists have been found to be those lacking ortho
substituents. such as the "coplanar" congeners having substitution patters 34-34, 345-34. and 345-
345 (Safe, 1992). Several "near coplanar" analogs of these congeners, i.e., mono-ortho substituted
analogs such as 234-34, 245-34, and 2345-34, may also have some AhR-agonist activity [Safe,
1992], but this is much weaker than that of the coplanar congeners in rodents, and often undetectable
in other species.
The particular types of PCB dechlorination activities present in the subsurface "hot spots"
of the upper Hudson have been found to result in dramatic reductions in the levels of the toxic
D-l
-------
coplanar and near coplanar congeners (Quensen et al. 1992a; Quensen et al. 1992b). The net effect
of these reductions, as measured by the EROD activity of the dechlonnated mixture, was a 97% loss
of toxic equivalency. In general, the percent toxicity decrease was much greater than the percent
decrease in meta and para chlorine level. It has also recently been shown that microbial
dechlorination markedly reduced or eliminated the adverse effects observed with Aroclor 1242 on
mouse gamete fertilization (Mousa et al., 1996). The reductions in concentrations of coplanar PCB
congeners in environmental samples due to microbial anaerobic dechlorination has now been
documented in several sediments, including the Hudson River, Sheboygan River, Waukegan Harbor,
and Lake Ketelmeer (reviewed in Bedard and Quensen, 1995).
In the case of the lower Hudson River fish PCBs, Table C-3 of Appendix C shows that the
relative levels of the one coplanar congener measured (PCB 77; 34-34 CB) were only 10-20% of
those in the PCBs of Thompson Island Pool fish, and those of several "near coplanar" congeners
between 10 and 35% of such values, indicating a >80% overall loss of toxic equivalency, despite the
very modest level of dechlorination in the local sediments.
Reduced Carcinogenicity. Currently, the EPA employs a default assumption, that any
positive finding in a high dose rodent bioassav implies a proportionate human cancer risk. In the
most recent statement of application of this policy to PCB risk assessment (USEPA, 1996) the
Agency recognizes that Aroclors 1016, 1242, and the higher Aroclors 1254/1260 can differ
considerably in their calculated carcinogenic potency, with upper-bound cancer slope factor (CSF)
values of 0.07. 0.4. and 2.0 per mg/kg-dav. respectively. However, the Aroclor 1242 CSF value of
D-2
-------
0.4 is applicable only for Aroclor 1242 uptake by inhalation. For sediments or fish contaminated
by Aroclor 1242, EPA uses the CSF value (2.0) as it would if they were contaminated with Aroclor
1254 or 1260. Likewise, the guidance document recognizes that the CSF may be sharply increased,
and in proportion to toxic equivalency, by the presence of coplanar or near coplanar congeners with
AhR-agonist activity, but does not allow for a reduction in the CSF when the levels of toxic
congeners have been reduced. Thus, although the current guidelines permit the risk assessor to
ignore the beneficial effects of PCB dechlorination upon presumed cancer risk, they also clearly
indicate the presumed cancer risk is dependent upon both the overall degree of PCB chlorination
(which may be moderately reduced by dechlorination) and the content of toxic congeners (which is
sharply reduced by dechlorination, as indicated above).
Reduced Exposure via Aerobic Degradation. The aerobic bacterial biodegradation of
PCBs is widely known and has been well studied (Abramowicz 1990; Bedard, 1990; Alder, 1993;
Bedard and Quensen, 1995; Furukawa. 1986). Numerous microorganisms have been isolated that
can aerobically degrade a wide variety of PCBs. although the more lightly chlorinated congeners are
preferentially degraded. These organisms attack PCBs via the well known 2,3-dioxygenase pathway,
converting PCB congeners to their corresponding chlorobenzoic acids. These chlorobenzoic acids
can then be readily degraded by indigenous bacteria, resulting in the production of carbon dioxide,
water, chloride, andbiomass. (Harkness, 1993).
The ability of native microbes to aerobically metabolize PCBs has been demonstrated in a
field test in the Hudson River. There are two lines of evidence that strongly indicate that natural
D-3
-------
aerobic PCB biodegradation ts occurring in the upper Hudson River. First, significant aerobic PCB
biodegradation was observed in the field without the addition of microorganisms, nutrients, or
supplemental oxygen (although mixing was performed) in the field test. (Harkness. 1993) This
result suggested that these sediments contain all the necessary elements for in situ aerobic activity.
To prove this hypothesis, a sensitive analytical method was developed to detect chlorobenzoic acids,
the intermediate products of aerobic PCB biodegradation, in undisturbed cores taken from the River.
PCB metabolites were found in all PCB contaminated samples, but not in any of the uncontaminated
sediments from further upstream (Flanagan and May, 1993). Moreover, the concentrations and
congener distributions of the observed chlorobenzoic acids closely matched the predicted
degradation products from the PCBs mixture in the samples .
The detection of chlorobenzoic acids of aerobic PCB biodegradation in contaminated upper
Hudson River sediments provides persuasive evidence that aerobic PCB biodegradation occurs
naturally in the environment. This finding is consistent with previous studies indicating that aerobic
PCB-degradmg bacteria with broad congener specificities are widely distributed in contaminated
soils and sediments. It could be argued that the PCB biodegradation metabolites observed by the
Flanagan and May (1993) study represent evidence of ongoing aerobic biodegradative activity,
remnants of past activity, or both. There is evidence to suggest that the metabolites are formed in
ongoing biodegradative activity since in a microcosm study, designed to mimic unperturbed Hudson
River conditions, the same chlorobenzoic acids are formed and then degraded in the course of
approximately 3 months. (Fish, 1996).
D-4
-------
Reduced Bioconcentration. The lightly chlorinated PCB congeners resulting from
dechlorination in Hudson River sediments (e.g., 2-CB and 2-2-CB) display an approximately 450-
fold reduction in their tendency to bioconcentrate in fish, as compared to the more highly chlorinated
tri- and tetra-chionnated PCBs present in the original Aroclor 1242 mixture. (Abramowicz, 1994).
Thus, natural anaerobic PCB dechlorination reduces the potential risk associated with PCBs via
direct reductions in carcinogenic potency, dioxin-like toxicity, and exposure.
Reduced Bioavailability. An additional reduction in PCB exposure results from long-term
contact of PCBs with sediment particles, and consequent reductions in bioavailability. It is well
established that the desorption of many nonionic organic compounds from sediment display bimodal
kinetics; a "labile fraction" of the contaminant desorbs readily, while a "'resistant fraction" desorbs
orders of magnitude more slowly (Kanckhoff and Morris, 1985). This phenomenon has been
observed with PCBs both in spiked and environmentally contaminated sediments (Carroll et al.,
1994. Coates and Elzerman, 1986. Witkowski. et al., 1988). The desorption kinetics of PCBs from
environmentally-contaminated Hudson River sediment and spiked sea sand using XAD-4
(polystyrene bead resin) as a "PCB sink" is shown in Figure D-l. PCB levels on the y-axis are
normalized to the starting PCB levels of the sand and sediment before desorption (13 and 25 ppm,
respectively). PCBs from Aroclor 1242 spiked sand were readily desorbed (85% in 8 hours). In
contrast, roughly half of the PCBs from H-7 sediment desorbed within the first 8 hours (the labile
fraction), with little additional desorption observed over the remaining 162 hours (Carroll et al.,
1994). The slowly desorbing fraction represents the proportion of PCB molecules that have diffused
into the organic material of the sediment over an extended period of time. Under these conditions
D-5
-------
PCB molecules are not available to bacteria and other river biota making them resistant to uptake
and degradation. Longer-term desorption experiments demonstrated that resistant fraction PCBs
desorb from Hudson River sediment with a half life of approximately 1 year (Carroll et al., 1994).
The bioaccumulation model developed by HydroQual for GE was used to compute total PCB
concentrations in fish for a variety of sediment and water column homologue compositions and
concentrations. Depending on the relative concentrations in the sediment and water, as well as on
the structure of the food web. dechlorination can lead to reductions in total PCB concentration of
between 4 and 35-fold lower than with relatively undechlonnated exposure sources.
D-6
-------
APPENDIX E
ADDITIONAL COMMENTS AND CLARIFICATIONS
1. Dechlorination Pattern H/H'
The Report recognizes that a variety of more highly chlorinated PCB congeners are very
susceptible to losses (pg. 3-119), but instead of recognizing this as a widespread example of
anaerobic dechlorination (pattern H and H'), it hypothesizes an unknown selective degradation
process that favors these tetra-chlorinated congeners (BZ 56, 60, 70, and 74). In fact, the data
displayed to demonstrate this unknown selective degradation (Figures 3-73 to 3-75) are further
evidence that dechlorination occurs at low concentrations (down to 1 ppm). Due to the insensitivity
of the dechlorination indices used by EPA. the Report fails first to identify this process as process
H/H' PCB dechlorination, and more important, fails to recognize us widespread occurrence.
Although this pattern is uncommon in the upper Hudson, it is widespread in areas of low
contamination in the lower Hudson. Process H/H' dechlorination does not produce significant levels
of the terminal PCB dechlorination products, but even such modest dechlorination significantly
reduces potential exposure, toxicity, and carcinogenicity, as the initial stages of dechlorination
provide disproportionate reductions in these endpoints.
2. Partitioning
One major failing of the analysis of equilibrium partitioning performed by the
TAMS/CADMUS/Gradient group is the inclusion of the Remnants and Rogers Island Stations in
the determination of global partition coefficients. The stations above the TI Dam show' much
E-l
-------
different behavior from those below the dam. Figure E-l shows the average and range of total PCB
partition coefficients for Transects 002-006 vs. river mile. Note the relatively constant values for
KP beginning at the TI Dam of about 50000 1/kg and the distinctly higher values upriver. This
difference cannot be attributed to a change in organic content of the solids. Figure E-2 shows
estimated log Koc values plotted against log Kow (as reported by Hawker and Connell). For the
Thompson Island Dam, Schuylerville, and Waterford stations, the Kqc pattern is relatively uniform
throughout all transects, and is reasonably well represented by KoW. The Fort Edward station
estimates show a much different pattern, being generally higher than the other stations and exhibiting
a higher degree of variability.
The Report states on pp. 3-13 and 3-14: "Noticeable in all transects are the generally
consistent values for KP 3 and Kp^ a estimates for most congeners within a given transect beginning
at Station 5, the TI Dam (RM 188.5). This suggests that approximate equilibrium conditions are
established within the Pool and remain consistent throughout the remainder of the freshwater
Hudson. The results for Rogers Island. Station 4. are distinctly different from those downstream and
probably reflect its proximity to the Hudson Falls source resulting in a lack of water column
equilibrium partitioning." The Report further states on p. 3-20, "All congeners tend to show
increased estimates of KP0Ca at RM 196.8 (Rogers Island) [note: Rogers Island is actually at RM
194.6], which may represent presence of non-equilibrated sediment in these samples."
This observation of markedly different partitioning above and below the TI Dam is apparent
and the conclusion of partitioning non-equilibrium above the TI Dam and equilibrium below is valid.
-------
With this conclusion, it is incorrect to include Remnant and Rogers Island data in the determination
of equilibrium partition coefficients. Estimates including this data will yield partition coefficients
well above equilibrium.
The analysis of the temperature dependence of partition coefficients (p 3-16) is based upon
historical data reported in Warren, et al (1987). It is not clear why this data was used to determine
the temperature dependence of partition coefficients when the Phase 2 data collected covers a
sufficient range in temperature to determine temperature dependence directly. Figures E-3 and E-4
show log vs 1/T for congeners 10 and 27.
It should be noted that these values of temperature have been corrected to ambient river
temperature. Figures E-3b and E-4b show log vs 1/T for congeners 10 and 27 based on the
temperature reported in the Phase 2 data. That this temperature is not ambient is evident by the large
temperature differences within transects as well as the reporting of high (-20 degC) temperatures
in the early Spring surveys. Figure E-5 shows a comparison of ambient TI Dam water temperature
data collected by OBG as compared to those reported in the Phase 2 transect data. In all transects
the reported Phase 2 temperature is higher than ambient. It is not clear when and where the Phase
2 temperatures were taken, especially in relation to the time of filtration of water samples. If the
filtration was performed before samples reached equilibrium at the higher temperatures, values of
1^(20) will be biased high.
E-3
-------
3. Volatilization
The equation used to determine PCB volatilization during low-flow conditions is
inappropriate for river systems, Eq. 3-33 (Report at 3-55) is an empirical model developed by
Hartman and Hammond from their studies of San Francisco Bay. This model is driven solely by
wind shear and is only appropriate for large open water bodies such as lakes and bays where water
velocities are minimal and there exists sufficient fetch to generate appreciable surface shear forces.
Even under extreme low-flow conditions in the Hudson River (-lOOOcfs) flow-induced shear
dominates gas exchange. The equation developed by O'Connor and Dobbins (Eq. 3-34) is an
appropriate model for rivers.
Additionally, these models only estimate the liquid film transfer coefficient. Prevailing
#
theory holds that gas exchange across a gas-liquid interface is subject to both a liquid and gas film
resistance. For substances with a high Henry's constant, such as oxygen, the liquid film resistance
dominates and it can be assumed that the overall transfer coefficient is equal to the liquid film
transfer coefficient. This assumption is incorrect for PCBs and other chemicals with much smaller
Henry's constants. Gas film resistance must be considered for PCB volatilization.
The conclusion at 3-56 in the Report that there exists no seasonal dependence on gas
exchange is not accurate. Both ice cover and temperature variations play a major role in
volatilization, whereas wind does not. As the Report states at 3-55, "during the winter months when
ice cover is extensive, the effective gas-exchange rate is reduced to near zero." At 3-56, it further
states, "K, increases by approximately a factor of two between 0"C and 25°C as a result of the
-------
temperature dependence of water viscosity." Both these observations are accurate. Wind, however,
will not appreciably affect gas exchange in the Hudson River for reasons discussed above. The net
effect is a significant seasonal variation in PCB volatilization.
4. Analytical Issues (EPA Appendix C)
Pg 1 - 14.
The congener-specific fCB analysis method lists 126 congeners which can be measured, as
listed in Table 1-4. Inspection of the list reveals some whose possible reported presence should be
regarded with suspicion: PCBs 12 and 126 should appear in Aroclors at very low levels, and their
ECD response factors are low. making any reported detection suspect, especially as they have no
ortho- chlorines and would not be expected to build up as bacterial dechlorination products of other,
more abundant congeners. PCB 20 coelutes with PCB 33 on the HP-5 column and with PCB 28 on
the octyl column, and would therefore appear difficult or impossible to quantify in the system
described.
PCBs 23, 58, 69, 96, 140, 143, 169, and 184 have been found to be present at only trace
levels in Aroclors, and are unlikely dechlorination products, (Frame, G.M., Cochran. J.W., and
Bowadt. S.S., J. High Res. Chromatogr., 19, (1996) 657 - 668), so any reported values of these
should be viewed with skepticism.
E-5
-------
Volume 2C. Book 1
Pg. A-l
Principal target congeners are listed as 1, 4, 5,1Q, IS, 19, 2S, 52. 101, 118. 138. and 180.
25B
Those in bold and underline have some analytical problems acknowledged in the report and
additional ones noted here below.
Those underlined only are tneasurable only with the octyl/Apiez«n system, and some of the
values might be affected by the problems of retention time instability noted both by Aquatec and
GE-CRD in 1993 with octyl columns.
PCB 1. This is a critical congener contributing substantially to the dechlorination index
defined by PCBs 1,4,8,10 and 19 using dechlorination products only. PCB 1 has a very low ECD
relative response factor, and interferences or substandard detector operation can introduce large
errors into its quantitation. In section A.5.4. mention is made that SDG 169803 samples did not
display peaks for BZ-1 on the octyl column, but the results from the HP-5 column were accepted on
the grounds that the peak is expected. The octyl non-detect probably represents substandard
operation of the detector on that column at that time rather than a column problem, and highlights
the potential for detection and quantitation problems for this congener when using ECD quantitation.
PCB 8. This congener was separable from PCB 5 on octyl/Apiezon L only, and considerable
problems in quantitation confirmation were encountered until the coelution criterion with PCB 5 was
relaxed.
E-6
-------
PCB 18. This is a critical congener assessing the amount of Aroclors 1016/1242 present.
Data had to be requalified for this congener (see pg A-26) based on the known general presence of
this in these sediments, and GC-ITD confirmation. Note however that quantitation of this congener
is suspect, even if its presence is reasonable on the grounds of its general appearance in these
sediments and the GC-ITD confirmation. In section A.5.2.7 it is stated that "TAMS/Gradient
considered quantitative differencs between the GC-ITD and GC-ECD results less than a factor of
5 acceptable, while differences greater than five times were considered unacceptable and associated
results rejected." Quantitative uncertainty of this magnitude may not require rejection of a finding
of detection of this key congener, but it renders its use in quantitative modeling highly suspect.
PCB 118. When the shift was made from octyl to Apiezon L. the potential for a small
difference due to coelution with PCB 122 on Apiezon which does not occur on octyl was ignored.
This will likely cause only a minor quantitative error.
PCB 138. The fact that PCB 138 coelutes on both octyl and HP-5, but not on Apiezon L.
with substantial amounts of PCB 163 when they come from Aroclors 1254 or 1260, was not
recognized. This could cause discrepencies between data collected before and after 1993, when the
shift from octyl to Apiezon L was made. In some fish the relative rates of metabolic clearance of
138 and 163 differ strongly, so failure to realize that both could be present in the peak in varying
amounts can result in quantitative difference errors if calibration was only against a 138 standard.
E-7
-------
In the listing of 126 congeners measurable in the dual column system in Table 1-4, it is not
clear why PCB 46. 2,2',3,6-tetrachlorobiphenyl, (present at -0.4 weight% in Aroclor 1242, and
cleanly resolvable from all other Aroclor congeners on both the HP-5 and octyl columns) was
dropped from the list of non-target congeners as noted at the top of page A-12. This congener should
have been clearly iidentified and reported in this study.
Section A.5.2.2. pg. A-12.
In the corrections to relative response factors obtained by measurement of the actual
congeners in 1993 and 1994 for congeners previously quantified only against the PCB 52 response,
there is no indication of how stable and comparable the previous responses being corrected actually
were. Individual congener's ECD response factors can easily vary by as much as a factor of 2 on
i
different detectors at different times, especially if temperature and/or carrier make-up flows are
changed. One needs to know how well these variables were controlled over the whole period of data
gathering to assess how consistent the quantitation and the quantitative correction which was
retroactively applied to the non-target congeners is.
On pg A-3 it is stated for core tops collected in 1992 that "RPI dried and archived core tops
(0-2 cm) from these cores for eventual PCB congener analysis." Aquatec subsequently analyzed a
small subset. The behavior and congener distributions of PCBs in the topmost layers of Hudson
River sediments is critical to evaluation of competing models of their fate and transport. How were
these samples dried? Excessive drying, especially in air with heat, can result in uncontrolled losses
of mono- and di- chloro PCBs which are major components of EPA's dechlorination index.
E-8
-------
On pg. A-4 appears the statement "Aquatec extracted sediment samples with hexane. and
performed applicable cleanup procedures prior to analysis by GC/ECD, as detailed in Appendix A3
of Phase 2A SAP/QAPP." Reference to the latter document reveals that the protocol for sediments
requires Soxhelet extraction with hexane/acetone. Which solvent was used? Hexane extraction
alone is inadequate to remove all PCBs from sediments unless they are previously dried so
strenuously as to risk losing by evaporation substantial amounts of lower congeners, which are
critical to assessing dechlorination processes. A call to Kurt Young of ITS Environmental in
Colchester, VT (successor company to Aquatec Chemistry Division) on 3/14/97 elicited the
statement that the information to answer this question was stored in a warehouse, not instantly
accessible, and in any event would require EPA authorization to be made available. A request for
information to resolve this question was made orally by phone to Douglas Tomchuck of the EPA
on the same date. As of 4/1/97 no reply has been received by GE-CRD.
5. Miscellaneous Issues
Report at 4-49: The document states that remobilization by sediment resuspension or
porewater displacement can serve to return PCBs to the water column long after any point source
contributions have been eliminated.
Response: EPA failed to determine the depths of the scour and porewater displacement
26A
contribution to the water column. Answering these questions will give an index of relative sediment
PCB importance as a function of depth.
E-9
-------
Report at 4-50: The document states that in general, aerobic processes affect only the lightest
congeners, and are ineffective at altering heavy congeners under environmental conditions.
Response: It should be pointed out that several microorganisms including MB 1, H850 and
L400 have been enriched from the Hudson River and are capable of degrading PCBs containing as 26B
many as six chlorines. In addition, there are several later remarks suggesting significant amounts
of degradation may have taken place in some instances. It should be pointed out that Flanagan and
May (1993) showed that PCB metabolites have been found throughout the river even in relatively
deep sediments and that these metabolites are short lived, with lifespans on the order of weeks (Fish.
1996) in physical models of the Hudson River. This points to the possibility that biodegradation
does take place even in non-surficial sediments.
Report at 4-50: The document states that in the absence of oxygen, the only
biotransformation possible is dechlorination.
Response: It has been repeatedly found in surface sediment microcosm sediments that both
aerobic and anaerobic PCB biodegradation takes place (Fish, 1994; 1996). This has a significant 26C
impact because the surface sediments are where the newest releases of PCBs are settling and where
the fish are getting their PCBs from. There is the potential for decrease in chlorine content and mass
reduction in this layer.
E-10
-------
Report at 4-51: The document states that the dechlorination process is more effective on the
heavier PCB homologues.
Response: No generalizations about PCB dechlorination are valid except that there is no
/
conclusive evidence for ortho- dechlorination in the Hudson River. A good review of the literature 26D/
offered in Bedard and Quensen (1995).
Report at 4-51: The document states that there is no reduction in the total number of PCB
molecules.
Response: There is -3 mole% of PCBs that can be reductivelv dechlonnated to biphenyl and
then easily degraded, namely the dioxin-like congeners including 34-4-CB. 4-4-CB, 34-34-CB, 345 - 26E
34-CB. The dechlorination of these congeners would significantly reduce the risk associated with
toxicological effects.
Report at 4-51: More chlorinated congeners are often associated with carcinogenic endpoints
while the literature suggests that less chlorinated congeners are more likely to produce neurological
impairment and developmental damage.
Response: The Battelle rat bioassay demonstrated that chlorine content does affect potential
carcinogenic potential in rats, and that potency is markedly reduced as compared to previous 2(
estimates. This study served as a basis for EPA's recent reassessment of PCB cancer slope factors.
E-U
-------
Furthermore there have been questions regarding the validity of the neurological impairment studies
as these studies are generally considered flawed and they have not been reproduced in six separate
attempts.
Report at 4-51: The document states that little evidence has been demonstrated for anaerobic
degradation in sediments.
Response: As discussed above, the presence of metabolic PCB biodegradation products in
anaerobic sediments is indicative of degradation processes even at low oxygen concentrations. The
transient nature of these metabolites also indicates that at least some low level of the degradation
process occurs.
Report at 4-52: The document states that the issue of anaerobic dechlorination will be
revisited in a Phase 3 report, incorporating the results of both high and low resolution coring.
Response: This provides EPA and its contractors with an opportunity to develop and use
more sensitive dechlorination indices.
Report at 4-88: The document concludes that the degree of in situ PCB dechlorination is not
a function of time but rather dependent upon the total PCB concentration within the sediment.
E-12
-------
Response: This conclusion is based in part on the use of an insensitive index of
dechlorination, and the lack of understanding dechlorination biochemistry. A more detailed analysis
of the data using a more appropriate index of dechlorination would show that concentration effects
the rate as has been shown in lab studies (Abramowicz et al., 1993; Rhee et al., 1995) and
microcosm studies (Fish, 1996). Sokol et al. (1994) and Rhee et al. (1995) indicate that the belief
that in situ dechlorination may not occur in areas with relatively low PCB contamination is based
on "dechlorination potential" and by definition is flawed. The phrase dechlorination potential is
defined as relating to the length of the lag phase before which dechlorination occurred. This
definition is not an indication of the dechlorination potential of a congener at a given concentration,
but an indication of the acclimation time for anaerobic microorganisms to respond in laboratory
studies. Note that this investigation only addressed BZ 138 and BZ 21. In the case of BZ 138 which
is 0.15-0.54 wt % of Aroclor 1242, no dechlorination was detected below 35 ppm, but the
investigators probably did not run the experiment long enough to overcome the lag time for
acclimation observed in these experiments. In comparison, these investigators looked at the
concentration effects of dechlorination of BZ 21, which does not exist in Aroclor 1242. but they did
show a 70% decrease in concentration in 7.5 months of incubation. Even at 4 ppm in the latter
paper, the concentrations used to evaluate the concentration were somewhat high, but complement
rates found by Abramowicz et al. (1993) who used a mixture of Aroclors 1242, 1254 and 1260, and
Fish (1996) who used much lower concentrations of Aroclor 1242 in test tube microcosms. When
all of these data are combined, they produce an interesting correlation that passes through the origin
as discussed above.
E-13
-------
Report at 4-52: The document introduces the concept of measuring the degree of
dechlorination resulting from PCB storage based on PCB sources and usage.
Response: This dramatically overestimates the usefulness of simply comparing final
products to the starting congener mix and underestimates the complexity of the task. It ignores the
fact that there are a variety of selectivity processes and sets the stage for using an inappropriate index
of dechlorination. Even if this methodology did work, it would rely upon the homogeneous mixing
of the Aroclors throughout the Hudson River with a single pair of reference points, namely Aroclor
1242 and completely dechlonnated Aroclor 1242. Earlier in the document, it is stated that as much
as 25% of the load south of Waterford is Aroclor 1254. Thus, their estimate of molecular weight is
underestimated by -6%. This may at first glance appear to be trivial, but is most likely pan of the
reason for the unusually low (<0% dechlorination) indices calculated for the lower Hudson River.
There is also an extensive discussion of PCB partitioning to porewater and the water column on a
congener basis. After such an extensive discussion on these principles, it is surprising that the
indices of dechlorination do not account for "washout" of the lighter congeners from sediment and
porewater. In fact, congeners used to calculate MDPR are up to lOx more likely to be lost to the
water column than are the heavier ones.
Report at 4-49 to 4-56. The use of BZ 8 as an indicator peak itself poses some serious
problems. First of all it is the most abundant congener in Aroclor 1242. Thus even at the onset of
dechlorination, small changes in a large peak of the chromatogram will meet with uncertainty. In
addition, if BZ 8 is dechlonnated to BZ 1 there would be no change in the MDPR.
E-14
-------
Report at 4-56: The document states that MDPR is a measure of the number of affected PCB
molecules.
Response: MDPR is a measure of the last one or two dechlorination steps for mixtures
containing 3.26-3.7 chlorines/biphenyl. Dechlorination via processes H' and H that are active in the
lower Hudson River (where the PCB concentration is generally lower) primarily attacks higher
homologues and will not be picked up by this index of dechlorination. Therefore, a significant
percentage of PCB molecules could be affected by these processes and still be missed by EPA's
analysis.
Report at 4-57: The document states that due to the lack of orr^o-chlorine removal, the
dechlorination process is theoretically limited in its ability to reduce the PCB sediment inventory.
Response: The dechlorination processes that are known to occur in the Hudson River serve
to reduce the ability of PCBs to bioaccumulate an important risk-reduction benefit. This also ignores
the possibility of other loss mechanisms, such as photo-destruction and biodegradation. In fact, the
latter two are mentioned later to account for the <25% of the PCB alterations in the lower river. If
the Agency believes that these are the processes responsible for <25% of the PCB alterations in the
lower river, there is no reason to believe that they are not also occurring in other parts of the river.
Furthermore, Fish (1994, 1996), has shown that in surface sediments, combined reductive
dechlorination and aerobic degradation serve to reduce a significant amount of mass in as short as
140 days. Both indices of dechlorination will tend to account for this as either no net mass loss or
E-15
-------
even as mass gain (in the case of aerobic degradation with little dechlorination) as shown in MDPR
vs. AMW at low concentrations in the lower Hudson River samples.
Report at 4-59: The mean molecular weights of Phase 2 sediment samples with low
concentrations of dechlorination products have been found to be close to that of Aroclor 1242,
indicating that processes other than dechlorination have not greatly modified the sample PCB
content.
Response: Alternatively, both higher and lower homologues are lost due to process H' and
H or moderate process C. M or Q (as their chromatograms indicate) plus loss of lighter congeners
from degradation and partitioning. The Agency's analysis could result in insignificant changes in
MDPR and AMW.
Report at 4-60: The document states that the sensitivity of MDPR has a larger range (relative
to AMW), and thus is more sensitive to changes in the PCB congener composition.
Response: The congeners used to measure the MDPR have the lowest response factors by
electron capture detection and are the most insensitive measure and thus most susceptible to
uncertainty, especially for BZ 1. In addition, any dechlorination due to para attack on BZ 8 will
not be evident from the calculation. This is 7.65 weight percent of the total PCBs in Aroclor 1242.
E-16
-------
Report at 4-62: The document states: "The Lower Hudson sample MDPRs tend to cluster
just below the Aroclor 1242 value of 0.14. The mean MDPR for the Lower Hudson is 0.11,
suggesting the presence of a minor contribution by heavier Aroclors. or more likely, possible loss
of BZ 1. 4, 8, and 19 prior to deposition due to their generally greater solubility and degradability.
The congener pattern comparisons made in Chapter 3 (Subsection 3.3.3), suggest that both processes
probably occur to some degree. It is important to note the absence of any significant degree of
dechlorination in the sediments'of the Lower Hudson. Based on this observation, it would appear
that dechlorination will not decrease the sediment PCB inventory of the Lower Hudson.
Response: The paragraph immediately discusses the Upper Hudson and ignores the
possibility that same loss mechanisms (partitioning and degradation) would reduce mass. It is
scientifically invalid to have one set of paradigms for one reach of the river and another set for other
parts.
Report at 4-65: The document states that there is a maximum decrease of 26% mass due to
dechlorination.
Response: This does not account for any of the degradation that was suggested earlier in the
document, for which Flanagan and May (1993) have found evidence. Furthermore, Fish (1994,
1996) have shown this in a physical model of the Hudson River in the laboratory.
E-17
-------
Report at 4-69: The document states that no significant change in AMW occurs in PCB
concentrations less than 30000 ug/kg.
Response: If there were slow steady-state dechlorination with steady washout of mono-
trichlorobiphenyls, one would expect the plots of AMW vs. PCB concentration to look as they do. 26Q
Report at 4-70: This presents a discussion on the age of sediments and degree of
dechlorination as a function of sediment age. The document does not address the upstream source 26R
and low-level recent contamination of the newest sediments. Also the "sampling" is skewed: there
are >2x the number of "new sediment samples" than "old sediments" used to construct Figure 4-28a.
Furthermore, the discussion of concentration effects on dechlorination should take into account the
different processes and end products expected as a function of concentration. In particular,
processes B.B' and C are common in the Upper Hudson and can lead to non-selective extensive
dechlorination, whereas processes H' H and A are more selective with different end products.
E-18
-------
Table C-l. Species-Characteristic PCB Alteration Patterns Observed in the Hudson
River,
waH>r. S yeo'o n«-«rx. AlT'n. PaTfc-rh
Lampreys (Petromyzontidae)
Sea Lamprey
SLP
Petromyzon marrnus
AP- per
Freshwater Eels (Anguillidae)
AP- \CT
American Eel
AME
Anguiila rostada
Herrings (Clupidae)
American Shad
AMS
Alosa sapidissima
fione.
Bullhead/Catfish (Ictaluridae)
AP-icr
Brown Bullhead
BRB
Ictalurus nebulosvs
Yellow Bullhead
YBH
Ictalurvs natalus
A?- *cT
White Catfish
WCF
Ictalurvs catus
Af- itT
Suckers (Catostomidae)
A f - CAT
Northern Hogsucker
NHS
Hypentelium nigricans
White Sucker
WSR
Colostomas commersoni
A f» -CAT
Minnows (Cyprinidae)
f4
Af«
Carp
CAR
Cyprinus carpio
Goldfish
GLF
Carassius auratue
Golden Shiner
GSH
Notemigoms crysoleucas
«
"Minnows"
MMM
Cyprmia spp.
M *>•««•
Fathead Minnow
FHM
Dimephales promelas
H t> n •€
Pikes (Esocidae)
AP- EStfC
Chain Pickerel
CHP
Esox mger
Northern Pike
NOP
Esox iucius
rioyt-*.
Codfish (Gadidae)
Af- fcr
Atlantic Tomcod
ATT
Microgadus tomcod
Temperate Bass(Moronidae)
hi eiM. C
Striped Bass
STB
Morone saxatilis
White Perch
WPR
Morone americana
pi t b)
Sunfishes (Centrarchidae)
N oruu ( b)
H t>*« £>)
Largernouth Bass
LMB
Micropterus salmoides
Smailmouth Bass
SMB
Microptervs dolomieui
Rock Bass
RKB
Amblophites ropestris
f<4 oti%.
Black Crappie
BLC
Pomoxis nigromaculatus
White Crappie
WCR
Pomoxis annularis
M ©to.
BluegiU
BLG
Lepomis macrochirus
Longear Sunfish
LSF
Lepomis megalotis
/4F- 1-6 P
Pumpkinseed
PKS
Lepomis gibbosus
AP * »-£P
/ » \
Redbreast Sunfish
RBS
Lepomis auritus
Af - uef(b)
Perches (Perchidae)
Ner<-
Yellow Perch
YPR
Percha flara
Walleye
WAL
Stizostedion vitreum
bio**-
Tesselated Darter
TES
Etheostoma olmstedi
/si On*.
P»»Tmetion 3.lhrjti**% 5<#n ih bt.T >1 m"t »//, inWif iSSv*^i
(^h) Saw**- • n"e51 v / ^ ^ *)± ak«w«^ flP-^T-hfa
cLcprttji ffy\ % |*n p£0 7® 6CC +.[Jy pV05^*//A*
d*Fre-tW Peg I/O, prtsvmmb}^ ajhr^ ,i, frff S f*.c/'S •
-------
ParamBtBr
Rafaranca Distribution*
Uppar Hud*on RJvar Maan* I
own H.A. Maono
kiwhr
OocNortnataon Pattam*
977-1978
1990
1992
1993
977-1978
1993
10:10
A
8
C
H
H'
4a an *
Uaan
*
Maan *
4* an * k
Aaan * k
taan *
Wt*
Wt*
Wt*
Wt*
Wt*
Wt*
Wt* RSO
Wt* RSO
Wt* RSO
Wt* RSO
Wt* RSO
Wt* RSO
Footnoui —
1
2
2
2
2
2
3
4
5
6
7
6
No. of FUh In M««n •
<¦18
*•29
¦ 15
<¦36 r
¦ IS n
¦ 46
c
1
X
i
•
f
i
« i a
.40
8.61
7 20
20 94
90
1 03
.20 *•
.37
• •
.21 **
47 ••
.04 ••
.01 ••
* 2 a
12.98
13.89
33 76
48 01
16 77
10 79
3.88 *
1*0
• •
2.S2 •
2.26 •
2.41 ••
1 44 •
* 3a
42.14
31.72
33 38
19 39
39 38
42.72
23.49
18.39
24.35
19.79
18 66
16.93
* 4 a
30.95
32.11
16 02
5 95
23.88
28.75
46.44
48.19
46.28 .
46.90
43.79
38.84
%ia
9.05
8.79
6 61
2 68
11.99
11.26
15.85
18.54
16 95
21.53
20.02
24.35
*8a
3.88
3.83
3 03
2.01
8.60
3.84
8 42
10.26
7.88
7.04
11.25
13.95
*7a
.81
.85
.55
.70
1.27
1.22
2.07
2.88
•
2.23 *•
1.61
3.28
4.88 *
* sa
.01
.19
.IS
18
.24
26
.40
.56
• •
.80 ••
.38 •
.48 •
1.33 *
* ta
.00
06
.10
16
.10
.31
.17 •
.12
• •
.15 ••
.06 ••
.17 ••
.23 *
Moan *RSO al 2090
18
64
68
38
46
42
lionw Ratio* PCD/PCB
Maan
Maan Ma*n M**n Maan Maan
Maan RSO
Maan RSO
Maan RSO
Maan RSO
Maan RSO
Maan RSO
T4cM*rabfpti»w»<»
19/28 ~ 31
.08
10
49
1 76
09
04
.07 ••
.06
• •
.02 •*
09 •
.08 *•
.10
18/2801
.68
.29
43
97
93
55
.15
.12
•
.13
.15
.27 •
.34
17/28 ~ 31
.31
.23
42
27
46
.33
.21
.13
•
.13
.12
.22 *
.26
27/28 ~ 31
06
09
.28
its
.10
08
.10
.06
• •
03 •
09 *
.11 ••
.12
28/28 ~ 31
08
.14
32
13
21
18
.18
.12
10
.16
.16
24
33/28 ~ 31
.43
35
22
16
20
.12
.11
.25
.24
21
.12
.20
22/28 ~ 31
19
13
07
10
15
07
.11
26
27
17
11
11
T atracMOfoWphanyt*
52/49
1 40
1 36
1 32
1 BO
1 24
1 00
99
1.29
1.24
1 01
1 06
97
47/49
39
56
74
1.12
.70
.79
1.06
.72
.64
51
85
56
44/49
1 26
1 37
08
27
68
26
.48
.76
96
SI
.64
.44
40/49
.36
14
04
07
25
19
.13
.15
.16
06 *
.11
.02 ••
74/49
61
73
04
J5
11
22
.70
94
63
79
.50
.18
dof**'* ad|. 7v/W
61
71
06
19
18
26
.52
.59
62
.52
.34
.26
77/49''
.13
.12
•
.10
.03 *
70/74
2 21
2 76
1 86
1 15
1 S2
1 64
.76
1 89
2.30
1 41
72 *
1.17
PantocMoioMpfcanyt*
101/99
1.69
1 93
3 07
3 04
1 80
1.88
1.31
1.69
1.76
1 SO
1.40
1.42
87/101
.67
1 03
53
15
33
22
.49
.80
.62
48
.42
.61 •
101/110
.91
53
49
55
.76
.99
1.15
.71
.69
.96
.99
1.00
99/110
.54
26
17
IS
42
S3
.88
.42
.41
64
67
.75
97/110
.46
.29
06
08
21
19
.36
31
.28
36
25
.26
87/110
61
54
26
19
.25
22
55
57
55
46
41
63 »
118/110
96
87
61
72
89
113
1.69
1.28
1.15
1.19
1 22
1 26
105/110
.59
74
17
21
25
31
89
91
83
63
19
49
HomacMapaWphanyl*
148/1 S3
.14
.26
60
73
.15
.26
.22
.24
.27
34
.21
.36
141/153
.11
41
.00
.00
.12
07
.12
.26
.29
.16
.13
.10
138 ~ 163/153
1.1C
1.43
3.83
1 91
1.04
1.17
1.29
1.27
1.34
1.38
1.15
1 04
128/153
• 2S
.25
.17
.09
.19
.18
.22
.22
.21
.26
.15
.15
187/153
.11
03
00
08
.09
07
.09
.03
.03
.09
.07
.10 ••
158 ~ 171/1S3
.1"
42
.16
12
.14
.15
.23
.32
39
.19
.14
12
Maan *RS0 <29 Ratio*)
1
7 H
0 2
6 2
2 2
9 19
1. Mirtura of Aroclor* 1242:1254 in ratio of 10:10
2. Moan* 70*
Table C-2. Mean Values of PCB Homologue Levels and Isomer Ratios in Upper and
Lower Hudson River Resident Fish Collected 1977-1993.
-------
Isomer CI Level -
No. of Ortho- CIs -
1
Tetrachlorobiphenyls
11111
0
Pentachlorobiphenyls
1111
2
Hexachlorobiphenyls
2 112
1
BZ0/BZ#
o>
5
to
m
CT>
O
ID
J
cn
(O
o)
ft
¦—
(O
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
180.
1«0 .
140 .
120.
100 .
•0 .
• •• .
•7 •*
* ••• •
20 .
#
%
Split Sample Data
1:1 Lina
14«
190
Total PCBs (ng/Ll
USGS Packed Column Method
Figure A-l.
Comparison of water column PCB concentrations calculated by the NEA
capillary column method versus the USGS packed column method.
-------
110
100
• •
/ •
K .
u
• «
••
u
• SpSt Sampta Data
- - Ragraition Una
R2 - 0.82
Skip* . 2.09
Intarctpt - 11.7
* 1
30 » 40 4
Mono CBs (ng/L)
NEA Capillary Column Method
Figure A-2.
The correlation of differences between NEA and USGS total PCBs and
monochlorobiphenyls.
-------
too.
76
•0
1©
20
10
¦to
S
• <
#
• r'
•
• * *v.#
* -i*
,•* x/
*
Spit Sampta Data
Ragrattion Una
R2 - 0.90
Slop* • 1.08
Inttrctpt =* 2.8
100
110
Mono + Di CBs (ng/L)
NEA Capillary Column Method
Figure A-3. The Correlation of differences between NEA and USGS total PCBs and
monochlorobiphenyls plus dichlorobiphenyls.
-------
Regression Analysis:
Total PCBs at Fort Edward
140
Linear Regression
~/- Prediction Umits
(95% Confidence!
120
O
u
100
a
2
Z
•0
40
m
%
to
10
60
M
90
0
20
30
40
SO
PCBa (ngn.)- U5CS IMhM
Regression Analysis:
Total PCBs at Schuylervllle
250
Linear Regression
*/¦ Prediction Limts
(95% ConfidenceI
O
u
150
a
2
z
0
o
m
©
*-
100
120
0
20
SO
40
60
ToUl PCB* (ng/1). USGS M«hod
I
Regression Analysis:
Total PCBs at Stillwater
200
Unear Rtgrwsston
~/- Pwdtctton Umits
(95% ConMBnc*)
140
• •
r «o
too
40 50 60
Total PCfta -USCS MMtod
Figure A-4.
Comparison of total water column PCB concentrations calculated by the
NEA capillary column method versus the USGS packed column method at
Fort Edward, Schuylerville and Stillwater.
-------
140
120
100
SO
60
40
20
Fort Edward Station
/\ • •
^ * f 1 | ' - - '
' m '
¦
JFMAMJ JASOND
1993
140
120
100 .
Thompson Island Oam Station
Figure A-5. Comparison of GE and EPA analysesof 1993 water column DB-1 capillary
column peak 5 components (BZ#4 plus BZ#10) collected at Fort Edward and
the Thompson Island Dam.
• EPA Data
GE Data
-------
i-
It
9 ¦
S-s
im
l*
*3
it
_ V
=f|©
Mr*
O®
Figure A-6.
R2 ¦ oj»
Slops « »•»*
Intercept » i-M
OB I Anatyili PkS |ng/L]
Intercept
DB-1 Andyti> Pk8 (ng/L)
Intercept
DB-1 AnalYtit P*1« Ing/LI
Comparisons of top: CP-Sil-ClE BZ#4 plus BZ#10 analysis to DB-1 peak
5 analysis; middle CP-Sil-C18 analysis of BZ#5 plus BZ#8 to DB1 peak 8
analysis; bottom, CP-Sil-C18 analysis of BZ#15 plus BZ#18 to DB1 peak
14 analysis.
-------
o>
00
a.
i i
-------
uo
120.
100.
80.
60.
40.
20.
0.
i i i i
i i i i i i
/\ \
J F M
50
40
30
20
10
0
50
40
30
20
10
0
1—
>11111
1 '
1 1 F
-
• 1
- L 1 1 1— _ 1 _
_ j 1 i i
JFMAMJ JASOND
1993
# EPA Data GE Data — — Corractad GE Data
Figure A-8.
Comparison of EPA, GE and corrected GE temporal 1993 water column
concentrations of Total PCBs, peak 5 (BZ#4 plus BZ#8), peak 8 (BZ#8) and
peak 14 (BZ#15 plus BZ#18) at the Thompson Island Dam.
-------
Figure B-1 Map of I liaison Falls plant indicating locations of DNAPL and
grouiulwaici recovery locations winch are color coded for composition.
JV R,
DNAPL Reservoirs
Legend
PCBs only
All components
All less TCB
Low PCB
UWCE IU#*L^ //
JOt'i
IHF-23B
¦l i -
NF-IOOB
0 f
CL-r£ .^aeo^ ^ ti^9B
rt~"\ Is- J 1 _ffl ®- - r®HF-r5S—
-'yf /'"T I fai£w
7//S^ "iHHfir'"p®27
-S1BD
HF^SBO
U^-105
Ko\:
r \\
\
GftAPHIC SCALE
DNAPL ANO GROUNDWATER
RECOVERY LOCATIONS
GENERAL ELECTRIC COMPANY
Mricruj rii i c uru Ynov
tntHAL. tLtL IK 11. uUMrAPi
HUDSON FALLS, NEW YORK
A DAMBS & MOORB
—— i^Mi.,iy,>«i
¦S, fl D4J
20171 -181
AS SHOWN
M» <*>
t1/17/96
L5GSND
HF-5B 9- DNAPL RECOVERY LOCATION
RV-2A GROUNDWATER RECOVERY HELL
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
^Hudson Falls
Fort Edward
Thompson Island
Upper
Hudson
River
RM 191-4
Stillwater
Mohawk River
RM 175
0 RM 153
Albany «r ,
' J*-RM 143 /
RM 123-6-*" I
RM 112
New York State
RM 88
Lower
Hudson
River
'•v/k
Afc xH'fi, f V «• RM 40
ew York CI
25 0 25 50
RM indicates Fish Collection Site River Mile
Figure C-!. Hudson River Drainage Basin,
10.0531
-------
fa
ec
0)
6
o
m
m
r-
ffl
O
a.
TJ
O
~->
(0
3
-------
H-7 Sediment
CL 04
m a
Desorption Time (Hours)
Figure D-l,
Desorption of PCBs from H7 sediment compared to spiked sea sand.
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
200
195
190
185
180
'
175
River Mile
170
165
160
155
150
Figure E-1. Partition coefficients as a function of river mile calculated from EPA Phase
2 water column monitoring transect studies 2-6.
-------
Roger* liland (North Tip) 01,28/93
Thompion liUnd Dim
01/30/83
SchuylarvUlt
01/30/93
TSS-1.2
OOC-4« •
TaMPCB-TlM
Ftn«f*M»
02/11/»3
TSJ-0 7 .
DOC»4.7 tfk C .
T^rci-UMV"^!
RRm'IM
03/20/93
TtSal.l
DOC. 41
Taut Kt
RflM)
04/12/93
TS«-i7.a
DOC-4 J
T»mfC»-3»0%<
ren«..«ny>»r»
LoalKo*)
1 1
^
TSS-1.4
- DOC-G 2
- S*
Tot>4 PC( - M 00
FIFkw-tOOO
a, '
02/20/93
i 1 1 1 ~7\
m-1.3
DOC-It
Total PC*-M *» ,
nnn-aio
so#
Ili'M*
DOC-4 4
Ft How-20300 w ^
03/27/93
TM-»7
OOC-S.f
T*Mrca-i2«s«
n n»w-woo
04/12/93
S •
UfiltCowI
M7I
ft Flow.4000
02/20/93
TSS-01
- OOC-4J »
y/ "
Total PCS-24 <2 .
Ft now-4§10
-
a
•
(
TSS-2«t
OOC-4.S
Total K*-M4 41
Ft Fkm-20300 *
03/27/93
TSS w 14.1 #
• DOC-i.4
y' *
T«t«f Id* TS s? ^
fEFfc>w<0300 jJP
• c fT
-
04/12/93
S • 7
Log(Kowr)
Waterford
02/01/93
TSS-1 »
' COC-4.2
T«airC#-**.99 •
, FlPWw-MOO i
• y*
-
^ «
* *
02/22/93
TM«I.2
DOC-4J
TMKI-M.05
R Flow-4*20 | 4^'
03/30/93
TU-JVOJ
DOC-4.1
T*WfCS-1t3 4}
RdMt'TWO
•
04/13/93
Tit "14.0
DOC-4.1
ra«iKa»2M
RKwtlW
LogiKowl
Figure E-2.
The relationships between and K^w at Rogers Island, the Thompson Island
Dam, Schuylerville and Waterford from Transects 1-8; Tss in mg/L, Total
PCBs in ny/L and Fori Ldward in cfs.
-------
Roger* lil«nd (North Tlpl 0U2M3
"T"
Tss-ta
DOC • §.4
TM*i KB-H2*
ft Flow-2*10
08/19/83
TSS-1.S
- DOC-SJ •
TsMrca-soai
. Fl Flow. 22(0
b •
•jCr
/ 1 ' «
¦
08/06/93
TfS
DOC
TaulPCt
rc new-moo
Thompson Itland Dam
06/26/93
TSS-1 i
- OOC-tJ
TMitrca-239 ?i 0
FCno«-2i(0 _0k
»
• '•**
» /
08/20/83
TM-1.2
- ooc-so
Total PCB-112 fiS
. Fl Flow-2320
x/*
i
t
05/05/93
Schuylarvllta
' 1 ' F ""'I
T»I-
- eoc-
Towirea-
FI Flow-11100
i
06/27/93
TSS-2.7
DOC-(3
Tm*1 PCS-1(2.7%
FEFk»«,-26JO *
08/21/13
TSI-12
DOC-1 0
TM«IPCS-att7
F£FVm.21»0 .
II
"" i i
TSS*
1
1
" DOG-
•
t«« i>ca-
. Kflm>
;
/ t 1
t
—i. ,
Watarford
TIS-4,7
DOC" 4.1
T*Mrca-ao.os .
FIRCW-2MO
TO-1.4
DOC-tO
T«MKt«(7.ai
nnm-nu %
mnsd-tiioo
LoglKow)
LoglKow)
LoglKow)
LoglKow)
Figure E-2 The relationships between K,k and K„w at Rogers Island, the Thompson Island
(continued). Dam, Schuylerville and Waterford from Transects 1-8; Tss in mg/L, Total
PCBs in ng/'L and Fort Edward flows in cfs.
-------
Log Koc as a function of inverts Temperature
Congener 10
BOO
0 0033
0.0034 0 0036 0.0036
irTemperature (Kelvin)
0.0037
CO
a*
u
o
•*.
Dl
O
8 00
7.00
6.00
6 00.
4 00
0.0033
0.0034 0 0036 0.0036
1/Temperature (Kelvin)
0.0037
Figure £-3. Log K^of BZfllOas a function of inverse temperature. Only Transect studies
I -6 from the Thompson Island Dam to Waterford; Left, with temperature
correction; Ri^ht, without temperature correction.
-------
Log Koc as a function of inverse Temperature
Congener 27
0.0036
0.0033
0.0034
0.0036
1/Temperature (Kelvin)
0.0037
o>
*
u
o
o»
o
8 00
7.00 .
S.00.
6.00 ..
4 00
0.0033
0.0034
0.0036
0.0030
1/Temperature (Kelvin)
0.0037
Figure E-4 Log K«of BZ#27as a function of inverse temperature. Only Transect studies
1-6 from the Thompson Island Dam to Waterford: Left, with temperature
correction; Rit^ht, without tcmpcruturc concction.
-------
30.0
25.0 _
20.0
a
OJ
©
T3
15.0
a
E
a>
H-
10.0
5.0
0.0
J
J
F
M
0
M
A
J
J
A
N
S
0
Month
Figure E-5. Temporal temperature profile in 1993 at the Thompson Island Dam: Line GE
data, O EPA pH meter, ~ EPA DO probe.
------- | |