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
^ PRO^°
For
U.S. Environmental Protection Agency
Region II
and
U.S. Army Corps of Engineers
Kansas City District
Book 3 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
# A s
For
U.S. Environmental Protection Agency
Region II
and
U.S. Army Corps of Engineers
Kansas City District
Book 3 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: PRELIMIN ARY MODEL CALIBRATION REPORT
VOLUME 2C: DATA EVALUATION AND INTERPRETATION REPORT
DECEMBER 1998
TABLE OF CONTENTS
BOOK 3 OF 3
IV. USEPA REVIEW AND COMMENTARY ON THE GENERAL ELECTRIC/QEA
REPORT, MARCH 1998
A. REVIEW AND COMMENTARY ON THE GE/QEA REPORT
B. GE/QEA REPORT: THOMPSON ISLAND POOL SEDIMENT PCB
SOURCES, MARCH 1998
December 22. W8
TAMS l.H TeiraTcch MCA
-------
IJSEPA Review and Commentary
on the GE/QEA Reports
-------
Review and Commentary on
General Electric/Quantitative Environmental Analysis, LLC
Thompson Island Pool Sediment PCB Sources, Final Report, March 1998
December 1998
For
U.S. Environmental Protection Agency
Region II
and
U.S. Army Corps of Engineers
Kansas City District
Prepared by
TetraTech, Inc.
and
TAMS Consultants, Inc.
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
Review and Commentary on
General Electric/Quantitative Environmental Analysis, LLC
Thompson Island Pool Sediment PCB Sources, Final Report
TABLE OF CONTENTS
Table of Contents i
List of Tables ii
List of Figures ii
Executive Summary 1
Introduction 3
1. Presence of a Sampling Bias at TID-West 3
1.1 Bias Corrections for Total PCBs 4
1.2 Bias Corrections for 2 Tri+ 9
1.3 Summary'of Bias Estimates 12
1.4 Interpretation of the Apparent Bias 13
1.5 Evidence of Loading from TIP Sediments 14
1.6 Re-evaluation of Thompson Island Pool Load 14
1.6.1 Analytical Bias 15
1.6.2 Sampling Bias 15
1.7 Implications of Alleged Sample Bias 15
2. Signature and Origin of the TIP Load 17
2.1 Characteristics of TIP Summer Load and TIP Surface Sediments 17
2.2 QEA Analysis of Sediment Source 20
2.2.1 Modeling Framework 20
2.2.2 Model Calibration 24
2.2.3 Depletion Rate of TIP PCB Inventory 28
2.2.4 Analysis of Ground Water Seepage Flux 29
2.3 Reanalysis of Thompson Island Pool Sediment Source Congener Signature . . . 30
2.3.1 Pore Water Source 32
2.3.2 Alternative: Mixed Pore Water and Bulk Sediment Loading 34
2.3.3 Influence of Advection and Dispersion on Pore Water Concentration . 37
3. Uniform Areal Flux of PCBs 43
4. Relative Contribution of Sediments below Thompson Island Dam 44
5. Hot Spot Versus non-Hot Spot Sources 45
5.1 Surface Sediment Concentrations in the TIP 45
5.2 PCB DNAPL Loading 48
5.3 Flood Pulse Loading of PCBs 50
5.4 Mass Loading of Contaminated Sediment Following the Allen Mill Collapse . . 51
December 22 1998
i
TAMS'l-TLTctra I cch MCA
-------
Review and Commentary on
General Electric/Quantitative Environmental Analysis, LLC
Thompson Island Pool Sediment PCB Sources, Final Report
TABLE OF CONTENTS
6. Summary 54
References 55
LIST OF TABLES
Table l-l Summary Statistics for Total PCB Loads at the TI Dam 9
Table 1-2 Summary Statistics for ZTri+ Loads at the TI Dam 12
Table 1-3 Correction Factors for the TI Dam PCB Loads 13
Table 2-1 NEA Peaks and Associated Congeners Used in Pattern Analysis 18
Table 2-2 Revised Three-Phase Partition Coefficient Estimates for PCBs in Sediment
in the Freshwater Portion of the Hudson River 32
Table 2-3 Physical Characteristics Assumed for TIP Surface Sediments 34
Table 5-1 Surface PCB Concentrations in NYSDEC 1984 Data Compared
to Texture Class 48
Table 5-2 Comparison of 0-2 cm and 2-^4 cm Aroclor 1242 Equivalent Concentrations
in Fall 1992 High Resolution Cores in the Thompson Island Pool 53
LIST OF FIGURES
Figure l-l Relationship Between Ft. Edward Flow and the Ratio of TID-West/TID-
enter for Total PCB Concentrations 5
Figure 1-2 Relationship Between Ft Edward PCB Concentration and the Ratio of TID-
West/TID-Center for Total PCB Concentrations 6
Figure 1 -3 Conceptual Model of PCB Loads Near the TI Dam 8
Figure 1-4 Relationship Between Ft. Edward Flow and the Ratio of TID-West/TID-Center
for LTri+Concentrations 10
Figure 1-5 Relationship Between Ft. Edward ZTri- Concentration and the Ratio of TID-
West/TID-Center for £Tri+ Concentrations 11
Figure 1 -6 Apparent PCB Load Gain across the Thompson Island Pool using GF. Data
with Corrections for Analytical Bias 16
Figure 2-1 PCB Homologue Shift Across the TIP. Summer 1997 19
Figure 2-2 Summer 1997 Water Column Relative PCB Congener Concentrations near the
Thompson Island Dam Compared to Aroclor 1242 21
Figure 2-3 Congener Pattern in TIP Sediment Compared to Aroclor 1242 22
Figure 2-4 Average Monthly Load and Concentration of Monochlorobiphenyls at the
TID-West Station. 1991-1997 31
Figure 2-5 Relative Percent Patterns in Water Column Gain at I IP-18C. Surface Sediment.
and Surface Sediment Pore Water 33
December IW8
ii
TAMS/LTlTeifaTech/MCA
-------
Review and Commentary on
General Electric/Quantitative Environmental Analysis, LLC
Thompson Island Pool Sediment PCB Sources, Final Report
TABLE OF CONTENTS
Figure 2-6 Sediment Congener Pattern Derived from Summer 1997 Gain at TIP-18C
Attributed to Pore Water Flux 35
Figure 2-7 Sediment Relative Concentrations Required to Support Observed
Water Column Concentrations via Pore Water Flux 36
Figure 2-8 Concentrations at TID-West Predicted as a Mixture of Pore Water and
Sediment Exchange 38
Figure 2-9 Concentration Gain at TIP-18C Predicted as a Mixture of Pore Water and
Sediment Exchange 39
Figure 2-10 Influence of K
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
Review and Commentary on
General Electric/Quantitative Environmental Analysis, LLC
Thompson Island Pool Sediment PCB Sources, Final Report
EXECUTIVE SUMMARY
In March 1998, Quantitative Environmental Analysis, LLC (QEA). on behalf of the General
Electric Company (GE), submitted to U.S. EPA Region 2 a report on the Hudson River PCBs NPL
site entitled "Thompson Island Pool Sediment PCB Sources. Final Report" (QEA, 1998). This
report summarizes, interprets, and refines a number of earlier data analysis and interpretation reports
relative to Thompson Island Pool (TIP) PCB dynamics written for GE by HvdroQual and O'Brien
& Gere. Rather than comment on each of the individual interim reports, this review of the "Final
Report", compiled on behalf of EPA Region 2. w ill serve as a review of all the GE reports submitted
to date on the TIP sediment source.The interpretive reports referenced in the final 1998 report are
as follows:
HvdroQual. 1995a. Anomalous PCB Load Associated with the Thompson Island pool:
Possible Explanations and Suggested Research:
HydroQual. 1995b. The Erosion Properties of Cohesive Sediments in the Upper Hudson
River; and
• HydroQual. 1997. Hudson River PCB DNAPL Transport Study.
QEA raises a number of important and interesting issues regarding the TIP PCB source. In
many instances, addressing these issues has required substantial new analyses and careful re-
examination of conclusions reached in the Phase 2 Reassessment Remedial Investigation/ Feasibility
Study (RRI/FS) Data Evaluation and Interpretation Report (USEPA. 1997). The net result of
combining and assessing the interpretations of QEA and the Phase 2 team is an improved
understanding of the TIP sediment PCB source. The conclusions presented by QEA are. however,
in many cases overstated, and. in some instances, not supported by the data. Numerous other data
compilation reports are sited in the 1998 report. See the 1998 QEA report for the complete list of
references.
This review addresses primarily the data evaluation and interpretation aspects of the QEA
report. QEAs modeling effort, which is not complete, is described in the report, but is addressed
here primarily in context oi its use in interpretation of PCB loading mechanisms.
The major conclusions of this review are as follows:
• There does appear to be a high sampling bias associated with the GE TID-West station near
Thompson Island Dam: however. GE "s attempt to compensate for this bias is incorrect and
frequently results in an underestimation of the actual load at the TI Dam. Additionally.
December 22 1998
I
TAMS/lTLTeiraTech/MCA
-------
sampling bias is markedly less for spring high flows, most likely as a result of more energetic
mixing conditions.The bias is also less pronounced for trichlorinated and higher congeners
relative to total PCBs.
The analytical bias corrections provided by GE more than compensate for the sampling bias,
resulting in load gain estimates for the TIP which are higher than those presented in the
DEIR (USEPA, 1997).
The congener signature of the TIP load is consistent with a weathered, partially-
dechlorinated PCB source—although not as fully dechlorinated as some buried hot spot
sediments. The assumption that pore water flux is the only summer loading pathway appears
to be incorrect. Instead, new analyses conducted for this review suggest that the summer TIP
load is a mixture of pore water flux andimlk loading of fine sediment, perhaps driven by
bioturbation.
QEA's modeling effort is incomplete, and some aspects of model calibration are
unsatisfactory or poorly documented. The QEA PCB fate and transport model is not
sufficiently refined to be useful for quantitative analysis of hypotheses of PCB loading
sources.
Analyses of sediment PCB inventory depletion rates presented by QEA are flawed, and do
not result in a constraint on interpretation of the TIP load.
PCB loading occurs throughout the TIP. This loading is consistent with the distribution of
known hot spot sediments, and suggests no need to invoke an unknown, '"anomalous" PCB
source.
Sediment PCBs both within and downstream of the TIP contribute PCB load to the water
column. The presence of a sampling bias at TID-West requires some reanalysis of relative
contributions above and below TID. On a load-per-mile basis, however, TIP sediments
remain the major concentrated source of loading from sediment. The TIP appears to
contribute PCBs at a rate per mile of between two and four times that of downstream
sediments.
Hot spot sediments are a major source of PCB load, although non-hot spot sediments also
contribute. QEA's argument that all sediments contribute equally to the PCB load appears
incorrect.
There is no credible evidence for extensive loading of PCB DNAPL or mass flux of highly
contaminated sediment bedload into the TIP after the Allen Mill failure. It does appear,
however, that elevated water column PCB concentrations entering the TIP in 1991-1993
have resulted in a general increase in surface sediment PCB concentrations in several
depositional areas. (See Table 5-2.)
22. 1998
TAMS/LTlTetraTech/MCA
-------
INTRODUCTION
The QEA report (1998. p. 1) presents four major bulleted conclusions, plus a fifth summary
conclusion in (he text. Each of these will be addressed in turn, and used to organize other comments
on the document.
The five conclusions presented by QEA are:
1. During the I990's, the amount of PCBs leaving the TIP was significantly
overestimated due to a sampling bias at the routine sampling station located at the
downstream limit of the TIP.
2. The composition of water column PCBs attributed to the TIP sediments indicates that
relatively undechlorinated PCBs are the principal source and that surface sediment
pore water is the principal point of origin.
3. PCB levels in the water column increase in a near linear fashion as water passes
through the TIP. indicating a nearly uniform areal flux from sediments within the
TIP
4. Sediments downstream of the Thompson Island Dam (TID) contribute PCBs to the
water column in a manner consistent with the TIP sediments (i e . transfer from
surface sediment pore water), increasing the water column loading by approximately
50% between TID and Schuylerville.
5. Surface sediments within ull areas of the river contribute PCBs to the water column,
not simply PCBs residing in "hot spot" areas Comparison of dry weight sediment
PCB concentrations, either at depth or at the sediment surface, gives a false
impression of the relative importance of various sediments within the river The
surface sediment pore water PCB concentrations, and. hence, the diffusive sediment
PCB flux is controlled by PC B concentrations associated with the organic carbon
component of the sediments, .-i.v these average organic carbon normalized PCB
concentrations are similar within "hot spot" and non- "hot spot" areas, these areas
contribute similarly to the water column PCB load.
1.0 PRESENCE OF A SAMPLING BIAS AT TID-WEST
During the 1990's. the amount of PCBs leaving the TIP was significantly
overestimated due to a sampling bias at the routine sampling station located at the
downstream limit of the TIP (QEA. 1998. p. 1)
As summarized in QEA (1998). HydroQual (Rhea. 1997) conducted monitoring and
produced a memo documenting apparent consistent differences between PCB measurements at the
TID-West sampling station and center channel measurements at TIP-18C. Rhea's memo presents
results which "indicate that PCB concentrations uithin TID-west samples are unrepresentative of
the average concentration passing the TID. PCB concentrations measured in samples collected from
this station consistently exceed those in samples collected in the center channel immediately
December 22 199#
3
r AMS tTLTeira Tech. MCA
-------
upstream and downstream of the dam. This bias appears to be responsible for the excess loading
observed from the TIP since 1991." The ev idence for the "bias" seems strong, at least during some
low tlow periods. The bias does not. however, account for all the "excess" loading from the TIP.
Most of the observations available from EPA and GE near the Thompson Island Dam were
collected at or near the TID-West station, and these data will need to be used in modeling. QEA
suggests "correcting" TID-West observations downward to reflect TIP-18C observations, but does
not propose a specific correction factor. It should be noted that this sample bias correction factor
is more than offset by GE's upward correction for analytical biases (HydroQual. 1997a). Both
corrections were proposed after release of the DEIR (USEPA, 1997). The net result of the two
corrections is that inferences regarding total PCB load generation from the TIP presented in the
DEIR based on analysis of GE data remain appropriate, and may in fact be underestimated; however,
estimates of relative loading from the TIP and downstream sediments may need some revision.
QEA's initial examination of the apparent bias in TID-West sample observations estimated
that center-channel and downstream concentrations were only about 63 percent of concentrations
observed at the TID-West station. These results were based on analysis of the ratio of TID-Center
to TID-West samples collected from 9/18/96 through 10/16/97. Subsequently, QEA determined
that TID-Center samples were essentially equivalent to samples obtained downstream of the dam
("TIDPRW" samples), and has continued to collect samples at TID-West and TIDPRW for
comparison. Samples through 9/15/98 (total of 51 samples) are now available in the most recent
(10/13/98) update to the GE database. Some significant revisions have occurred in the first few
data points through 6/17/97; only minor differences were detected in later data. Note that the
earliest data points do not have a TID Center (TIP-18C) result reported in the database, but
equivalent samples are available as float survey (FS) or TIP samples. One data point which was
non-detect at TID-West (12/29/97) was omitted from calculations. For the period of 8/13 through
10/16/97 in which both TIP-18C and TIDPRW samples are available, the TIP-18C results have
been used.
Over the full set of 1996-1998 samples, the average ratio of TID-Center or TIDPRW total
PCB results to TID-West total PCB results is 0.86, much closer to unity than the original ratio of
0.62 proposed in J. Rhea's memo of 9/30/97 (Rhea, 1997). Samples collected in 1996-1997 had
an average ratio of 0.72 (including samples after 9/30/97). while samples collected in 1998 had
an average ratio of 0.92. A wide range of ratios is seen in individual sample pairs, including
results where the ratio is greater than 1 {i.e., the TID-West result is less than center channel
result).
1 1 Bias Corrections for Total PCBs
In revisiting this analysis, it is first appropriate to enquire why the ratio in the GE 1998
results is higher than in 1996-1997 results. Two important characteristics distinguish this
sampling period from the full set of samples. First, flows were low (less than 5.000 cfs), whereas
higher flows occur in the 1998 sampling Second, the 1996-1997 samples all were taken during
a period in which the upstream concentrations at Fort Edward were very low or non-detect,
whereas increased upstream loads reoccurred during the 1998 sampling Figures 1-1 and 1-2 show
the relationship between observed ratios and (1) flow, and (2) total PCB concentration at Fort
Edward.
December 21. 1998
4
r AMS/L riTctra Tcch/MCA
-------
TO
oa
"ro
•«—«
o
y-
16
14 —
1.2 —
1
0 .8 4-
0.6
0.4 —
0 2 —
0
&
' '4^
5000
10000 15000 20000
Flow at Fort Edward (cfs)
25000
30000
I'etraTech/TAMS
Figure 1-1 Relationship Between Ft. Edward Flow and the Ratio of TID-WestATID-Center for Total PCBConcentrations
-------
16
1.4
12
1
o
m
* 0 8
o
H 0 6
0.4
0.2
0
0
A
-A "
10
20
30 40
C, Fort Edward
50
60
70
TetraTech/TAMS
Figure 1-2 Relationship Between the Ft. Edward PCB Concentration and the Ratio of TID-West/TID-Center for Total
PCBConcentrations
-------
From these figures, it will be noted that the ratio between center channel and TID-West
observations appears to approach unity as either flow or upstream concentration increases. The
relationship to flow is fairly obvious: increased flow implies greater lateral mixing potential,
which should make concentrations more uniform across a channel section.
To understand the relationship to upstream concentration, consider the extremely simplified
conceptual mode shown in Figure 1-3, in which downstream flow through the TIP is indicated by
arrows. Discrepancy between shore concentrations (C,) and mixed concentrations at the dam (Q)
presumably arises because there is an additional load in the nearshore area (L), which is not
immediately mixed laterally. Consider a case in which transport is laterally mixed at some point
(say, the end of Griffin Island). At this point, there is a flow of magnitude Q, with a concentration
of C0. Downstream (i.e., in the areas of the TID-West sampling station) full lateral mixing does
not occur, and an additional load. L. is introduced. For simplicity, assume that the flow is split
into two portions, with a flow of Q, going through the nearshore portion, and a flow of QrQ,
going through the main channel. These flows then mix and recombine at the dam. It is important
to realize that the concentration in the nearshore area is determined by both the upstream
concentration and the local loading. L. Under these conditions, the concentration in the nearshore
area (TID-West) would be given by
C - C + L/Q
i o ^ 1
while the mixed center channel concentration at the dam would be given by
C2 ^ C0 ~ L/Q0
The ratio would then be
eye,
c0 - l/q„
c0 * L'Q,
This ratio depends on the relative magnitude of Q, to Q(), indicating that lateral mixing intensity
presumably increases with the magnitude of Q,. As Q, increases toward Q0 (implying instant
lateral mixing of L), the ratio should approach 1. The ratio also depends on the relative magnitude
of C„ versus L. As the upstream concentration increases, the ratio should again increase toward
1 because the contributions from the nearshore area are swamped by upstream loads.
Thus, the high bias seen in initial GE sample comparisons is a joint result of low flows and
low upstream concentrations. The bias results from incomplete lateral mixing of what is likely
(to a first approximation) a fixed local load. If this load is small relative to the upstream load, or
if mixing is high, the bias is reduced. Thus, it is entirely inappropriate to apply the apparent bias
correction observed in 1996-1997 to the entire observed time series at TID-West In particular,
a much smaller bias correction should apply during conditions prior to 1995 in which much higher
upstream loads were observed.
December 22. I«W8
rAMS-l.TlTeiraTecK.MCA
-------
Vs
0* Vo
yQ I
fo-Q,
c, = c0 < L/Q, C
C2 = C0 + L/Q0
T
C..Q.
Figure 1-3 Conceptual Model Of PCB Loads Near the TI Dam
TelraTech/TAMS
-------
Note that the mixed upstream concentration relative to TID-West. Q,. is presumably the
concentration near Griffin Island, which is not known for most sampling events. This
concentration includes the load at Fort Edward plus any incremental load within the pool above
Griffin Island. Nonetheless, a high load at Fort Edward would tend to reduce the ratio between
TID-West and center channel observations.
The graphs presented above suggest that high bias for total PCB measurements at TID-
West occurs at conditions of flow less than about 4,000 cfs at Fort Edward and concentrations less
than about 17 ng/1 at Fort Edward. If we segregate the observations based on these criteria, using
the full set of GE data through 9/15/98, the following results are obtained for the ratio of center
channel to TID-West observations:
Table 1-1
Summary Statistics for Total PCB Loads at the TI Dam
Total PCB Results
Flow at Fort Edward
< 4.000 cfs
> 4.000 cfs
Concentration at Fort
Edward
< 17 ng/1 total
PCBs
average = 0.64
median = 0.60
count = 23
average significantly
different from 1
average = 0.78
median = 0.75
count = 8
average significantly
different from 1
> 17 ng/ total PCBs
average = 0.80
median = 0.78
count = 8
average significantly
different from 1
average = 0.90
median = 0.77
count = 11
average not
significantly
different from 1
In three of the four cells the average ratio is significantly less than 1 at the 95% confidence level.
As predicted, the ratio increases with both increasing upstream flow and increasing upstream
concentration. At high flow and high concentration, the average is not significantly different from
1, and no bias correction should be used.
1 2 Bias Corrections for £ Tri +
The bias was also analyzed for the £Tri+ parameter, that is the sum of trichlorinated and
higher homologues. The ratio for STri+ also shows relationships to upstream concentration and
flow, as shown in Figures 1-4 and 1-5. Based on these figures, a flow of 4,000 cfs at Fort Edward
was again selected as a breakpoint, while a £Tri+ concentration of 15 ng/1 at Fort Edward was
selected as the concentration breakpoint. Results of the analysis are shown in Table 1-2.
December 22 1 ')Q8
9
MMS.l. IlTciraTech.MCA
-------
2
15
ro
a 1
+
0.5
0
0
4 r*
'a '
.N
•A
A-
. ^
*' /
5000
10000 15000 20000
Flow at Fort Edward (cfs)
25000
30000
Tel mTech/T A MS
Figure 1-4 Relationship Between Ft. Edward Flow and the Ratio of TID-West/TID-Center for ITri+Concentrations
-------
2
1.5
ro
OC 1
+
0.5
i *-
0
10
20
30
C, Fort Edward
40
50
60
Tetr«Tech/TA MS
Figure 1-5 Relationship Between the Ft. Edward ZTri+ Concentration and the Ratio of TIDWestmD-Center for
IT ri+Concentrations
-------
Table 1-2
Summary Statistics for £Tri+ Loads at the TI
Dam
STrl-f
Flow at Fort Edward
< 4.000 cfs
> 4.000 cfs
Concentration at Fort
Edward
< 15 ng/1 STri +
average = 0.69
median = 0.65
count = 24
average significantly
different from 1
average = 0.97
median = 0.83
count = 10
average not
significantly
different from 1
> 15 ng/ £Tri +
average = 0.88
median = 0.88
count = 7
average significantly
different from 1
average =1.13
median = 1.00
count = 9
average not
significantly
different from 1
In two of the four cells, the average is significantly less than 1—but only in the low flow, low
upstream concentration cell does the average approach the large bias correction factor which has
previously been proposed.
1.3 Summary of Bias Estimates
An empirical analysis of the full set of comparisons between TID-West and TIP-18C or TIDPRW
samples shows that the observed bias is dependent on both flow and upstream concentration.
When upstream concentration is high (as was the case in the early 1990 s), the bias will be small.
The bias will also be small or non-existent at high flows. As a result, the apparent bias should
have only a small effect on calculations of annual load.
Stratifying the analysis by flow and concentration removes much of the apparent seasonal
component of the bias. While there does appear to be a seasonal cycle of load generation in the
nearshore sediments, peaking around June, this may have little effect on the ratio between
nearshore and center channel concentrations. These results indicate that any bias correction of
TID-West observations must be conditional on both concentration and flow at Fort Edward.
Estimated empirical bias correction factors are summarized in Table 1-3:
December 22 IW8
12
r AMS.l. Il'Tetra fcch/MCA
-------
Table 1-3
Correction Factors for the TI Dam PCB Loads
Empirical Bias Correction Factors
Total
PCBs
LTri +
Low Flow,
Low Upstream
Concentration
Fort Edward Flow < 4000 cfs
Fort Edward Concentration < 17 ng/1 total
PCBs or < 15 ng/I 2Tri +
0.64
0.69
Low Flow,
High Upstream
Concentration
Fort Edward Flow < 4000 cfs
Fort Edward Concentration > 17 ng/1 total
PCBs
or > 15 ng/1 2Tri +
0.80
0.88
High Flow,
Low Upstream
Concentration
Fort Edward Flow > 4000 cfs
Fort Edward Concentration < 17 ng/I total
PCBs or < 15 ng/1 2Tri +
0.78
1.0
High Flow.
High Upstream
Concentration
Fort Edward Flow > 4000 cfs
Fort Edward Concentration > 17 ng/1 total
PCBs
or > 15 ng/1 2Tri +
1.0
1.0
1.4 Interpretation of the Apparent Bias
While evidence of a difference in concentrations between the TID monitoring stations under
various conditions is clear, the interpretation is not. These correction factors serve to match the
center and west station at the TI Dam but do not indicate which value is more correct. A more careful
examination of the data leads to the follow ing conclusions:
• Even if the estimate of bias is correct, this does not account for all the "excess loading"
observed from the TIP: instead, the evidence continues to suggest that there is a summer gain
of at least 0.5 kg/d (and more likely 0.7 to 1 kg/dav) total PCBs from the Thompson Island
Pool.
• While there is a difference between summer concentrations at stations near the Thompson
Island Dam. the conclusion that TID-West observations are biased high, and that this
constitutes the entire difference, is only one among a number of possible explanations.
• HydroQual observations on lateral variability of PCB concentrations in the Thompson Island
Pool appear to provide evidence that near-shore contaminated sediments are a significant
source of PCB load to the Pool.
While there are consistent differences in low-How concentrations between TID-West and
TIP-18C. homologue patterns at the two stations are generally similar and reflect an increase in
mono-, di-. and trichlorobiphenyl loads relative to the Rogers Island station. Concentrations in the
December 22 IW8
13
TAMS/LTlTctraT«lvMCA
-------
center channel just downstream of the Thompson island Dam (available for August through October
1997 only) are similar to those at TIP-18C. suggesting that TIP-18C (rather that TID-West) provides
a good estimate of load exiting the TIP. These observations are consistent with a theory that the
PCB load exiting the TIP represents a simple dilution of concentrations originating in nearshore hot
spot areas, as discussed in Section 1.1 above.
There are also alternative explanations, other than simple bias in TID-West observations,
which may account for part or all of the difference between TID-West and center channel
observations at TIP-18C: The difference between TIP-18C and TID-West samples might suggest
that TID-West is biased high relative to the average concentration leaving the TIP, or that TIP-18C
is biased low . or some combination of the two. During four of the sampling events. HydroQual also
sampled PCB concentrations at TID-East and at several stations below the Thompson Island Dam.
During each of these events. TID-East concentrations were similar to those at TIP-West and greater
than those at TIP-18C. indicating that either both wing-wall stations are biased high, or TIP-18C is
biased low . Samples 200 feet downstream of the dam were generally within 10 ng/1 of TIP-18C
samples, although higher than TIP-18C in three out of four summer events despite any volatilization
losses during transport over the dam. suggesting that TIP-18C was approximately representative of
concentrations going over the dam at the time of observations. However, on the one occasion
(8/13/97) on which samples were also taken two miles further downstream at Fort Miller a different
picture emerges. On 8/13. total PCB concentration at TID-West was 90.2 and T1P-18C 49.6 ng/l.
Concentration at Fort Miller on this date was 76 ng/1. or a little greater than the average of TID-West
and TIP-18C concentrations, while concentration at Schuylerville was 74.2 ng/1. Measurements at
Fort Miller and Schuylerville presumably average out short-term diurnal and lateral variability in
TIP loads relative to observations just above or below the dam. This suggests the possibility that
the actual daily load transported downstream may be an average of TID-West and TIP-18C
observations.
1.5 Evidence of Loading from TIP Sediments
The QEA bias study results demonstrate that, under some conditions. PCB concentrations
are higher in shallow nearshore areas above Thompson Island Dam than in the main channel. The
strong concentration differential suggests that the increased PCB concentrations nearshore must anse
from nearby sources (e.g., hot spots 15 through 20). thus allowing limited time for lateral mixing.
An unintended consequence of the bias study w ould thus appear to be a demonstration that these hot
spots do indeed constitute a significant source of PCBs to the water column. Even if the nearshore
concentrations are biased high relative to total load, it is these shallow nearshore concentrations
which are most relevant to biological exposure.
1.6 Re-evaluation of Thompson Island Pool Load
Since the release of the DEIR (USEPA. 1997). GE has released corrections for analytical bias
in their PCB analyses, and proposed a correction for sampling bias. Revised estimates of load from
the TIP need to account for both factors. The re-evaluation below first considers the effect ot the
analytical bias corrections, without any correction for sampling bias, on the estimates of load gain
between the station at Rogers Island and TID-West. then adds the effect of potential sampling bias.
December 22 l«WK
14
T AMS 1.11 Tctra I ech. MC A
-------
1.6.1 Analytical Bias
GE's recent analytical corrections resulted in a significant increase in the apparent load gain
between the Rogers Island (Rt. 197) and the TID-West stations. The older, uncorrected data led to
an estimate of approximately 0.56 kg/d load gain across the TIP (USEPA. 1997, Table 3-21); using
the corrected data gives an estimate of about 0.82 kg/d over the 1991 -1997 period of record. During
summer (June-August) of 1996 and 1997 the re-calculated gain from Rogers Island to TID-West
after analytical corrections appears to have been about 1.26 kg/d. As shown in Figure 1-6, the
estimates of load at Rogers Island (River Mile 197) and TID-West (River Mile 188.5) diverge,
showing a consistent increase in apparent load between the two sampling stations. Over 90% of this
apparent load gain is in mono-, di-, and trichlorobiphenyls, with the largest load gain in
dichlorobiphenyls. Note that the load estimates are not corrected for the apparent sampling bias.
1.6.2 Sampling Bias
//"the QEA conclusion that the center channel observation is more representative of transport
through the Pool is assumed to be correct, and the correction factors developed above are applied,
this would still not eliminate the load gain. Instead, the apparent load gain for 1991-1997 (after
correction for both analytical and sampling biases) would be a value of about 0.62 as an annual
average (note that the value exceeds the DEIR estimate of 0.56 kg/d) by 10 percent. This value
assumes the 1991 - 1997 bias correction is based on five years with Rogers Island concentratins
greater than 17 ug/L (1991 - 1995) (correction factor = 0.8) and two years with concentrations less
than 17 ug/L (1996 - 1997) (concentration factor = 0.64). This yields a correction factor of 0.75. This
correction factor assumes that all TIP loads are generated at low flow conditions, thereby yielding
maximum correction and minimum estimate of the 0.82 average annual TIP load. It is very important
to note that the application of the proposed sample bias correction would still not cancel out the
increase in estimated load which resulted from GE's analytical recalculation—and would continue
to identify the Thompson Island Pool as a significant source of PCB load.
1.7 Implications of Alleged Sample Bias
QEA (p. 11) states that the DEIR "concluded that PCBs passing the TID during low flow
conditions were the major source of PCBs to the freshwater Hudson*', and implies that this
conclusion is incorrect in light of the sample bias. However, the DEIR did not claim that low flow
conditions dominated load or that TIP sediments were dominant source of total PCBs during 1993
observations (USEPA, 1997, pp. 3.90-91):
...the GE Hudson Falls source contributes the majority of the PCBs to the water
column on an annual basis due to its large contribution during the spring runoff
period. The TI Pool source is estimated to be the primary source of PCBs to the
water column for 11 months of the year (i.e.. the low flow period) and it contributes
approximately 32 percent of the annual PCB load.
December 22. 1998 I? TAMS/LTLTetraTech/MCA
-------
Load across the Thompson Island Pool
Total PCBs, GE Data
5000
4000
I 3000
o 2000
o>
1000
0
W1 w w w %
River Mile 188 5
River Mile 197
River Mile 194 2
Figure 1-6.
Only
Te!r»Tech/TAMS
Apparent PCB Load (iain across the Thompson Island Pool using CJE Data with Corrections for Analytical Bias
-------
The presence of a sampling bias in USEPA Phase 2 data would, however, require a
reassessment of the relative percent contribution of TIP loads. If a sample bias correction factor of
0.80 is applied only to low flow load estimates at the Thompson Island Dam. the net effect would
be a revision downward of the percent contributed by the TIP in 1993 from 32 percent to
approximately 27 percent of total load. The value of 0.8 is used since the Ft. Edward concentration
never fell below 17 ng/L during the 1993 sampling events. Because of the high total loads seen in
1993 this still represents a significant PCB loading source. A more accurate estimate of the TIP's
contribution is expected from the upcoming Baseline Modeling Report. Nonetheless, it is clear that
the load revisions serve to re-emphasize (and not detract from )the importance of the TIP load.
One major topic for which the sample bias would have a major effect is on the estimation of
relative loads from the TIP and from downstream segments. This is because the analytical bias
corrections apply to all GE measurements, whereas the sample bias correction would apply only to
estimates at Thompson Island Dam. This topic is covered in Section 4. The issue of relative loads
is also affected by the revised flow estimates discussed in the corrections to Chapter 3.
As to load estimates based on GE data, the net effect of analytical corrections and the
suggested corrections for sampling bias result in a higher estimate of total PCB load generation from
the TIP relative to that estimated in the DE1R based on the earlier version of the GE data. The
apparent sampling bias does not appear to affect the homologue pattern of the TIP load gain, only
its magnitude.
2. Signature and Origin of the TIP Load
The composition of water column PCBs attributed to the TIP sediments indicates that
relatively undechlorinated PCBs are the principal source and that surface sediment
pore water is the principal point of origin. (QEA. 1998, p. 1)
QEA's summary statement is misleading and not strongly supported by the evidence. The
statement contains two parts. It is first concluded "that relatively undechlorinated PCBs are the
principal source." This is misleading. It is true that the TIP load is relatively undechlorinated
compared to buried, more highly dechlorinated sediments. However, the load is strongly
dechlorinated relative to raw Aroclor 1242, and. indeed, the surface sediments in the TIP on average
show significant dechlorination. It is secondly stated "that surface sediment pore water is the
principal point of origin." This conclusion is not fully supported by the data: indeed, the congener
pattern of the TIP load shows consistent differences relative to the congener composition from
surface sediment pore water. Reanalysis of the available data suggest that the TIP load most likely
originates from a mixture of pore water flux and resuspension of fine sediment.
2.1 Characteristics of TIP Summer Load and TIP Surface Sediments
GE monitoring of summer water column PCB homologue concentrations from 1991 through
1997 shows a consistent shift in homologue pattern between Rogers Island and Thompson Island
Dam (as measured at the TID-West station). In all years monitored, average summer homologue
patterns shift from a tri- and tetrachlorobiphenvl dominated pattern at Rogers Island to a mono-, di-.
and tri-chlorobiphenyl dominated pattern at TID-West. A similar shift is seen in the Phase 2 data.
The significance of the apparent shift in the GE data was strengthened by GE's recent "'corrections
December 2.2 )W8
17
T AM S/l. Tl Tetra T cc K.MC A
-------
for analytical biases" (HydroQual. 1997a). and is greater than was reported in the Phase 2 DEIR
(USEPA. 1997). Based on data collected in 1997. the pattern shift is also immune to any potential
spatial biases in the TID-West versus TIP-18C (center channel) sampling stations.
QEA chose to base their analysis on summer 1997 data (June through August), in part
because the 1997 data contain observations from both TID-West and TIP-18C, allowing them to
choose to use the supposedly unbiased center-channel observations. The choice of summer 1997
data is also fortuitous because upstream loads and concentrations at Rogers Island were very low
during this period, enabling a more direct interpretation of the TIP signal. Homologue patterns at
Rogers Island (Rt. 197) and TID-West (Thompson Island Dam) during summer 1997 are shown in
Figure 2-1. and show the usual strong shift to a mono-, di-, tri-chlorobiphenyl dominated pattern.
A more informative comparison can be made by examining the relative percent composition
of a set of key congeners. For this and subsequent analyses the following GE/NEA capillary column
peaks and associated congeners were chosen for comparison because (1) they are environmentally
significant, and (2) three-phase partition coefficient estimates are available. For each peak the
congener of most environmental significance in upper Hudson River sediments is listed first.
Table 2-1. NEA Peaks and Associated Congeners Used in Pattern Analysis
NEA Peak
Homologue Group
Congeners
Peak 2
Monochlorobiphenyl
BZ #1
Peak 5
Dichlorobiphenyl
BZ#4. B7.#10
Peak 8
Dichlorobiphenyl
BZ#8. BZ#5
Peak 14
Di/Trichlorobiphenyl
BZ#15. BZ#18
Peak 24
Tri/Tetrachlorobiphenyl
BZ#28. BZ#50
Peak 23
Trichlorobiphenvl
BZ#31
Peak 37
Tetra/Pentachlorobiphenyl
BZ#44. BZ# 104
Peak 31
Tetrachlorobiphenyl
BZ#52. BZ#73
Peak 47
Tetrachlorobiphenyl
BZ#70. BZ#76. BZ#61
Peak 48
Penta/Tetrachlorobiphenyl
BZ#95. BZ#66. BZ#93
Peak 53
Pentachlorobiphenyl
BZ#101. BZ#90
Peak 69
Penta/Hexachlorobiphenyl
BZ#118. BZ#149. BZ#106
Peak 82
Hexachlorobiphenyl
BZ#138. BZ#163
Peak 75
Hexachlorobi phenyl
BZS153
Across this set of peaks, the congener pattern at the TID is remarkably similar during summer 1997
whether we examine raw concentrations at TID-West. concentrations at the "unbiased" center
December 22 IW8
18
TAMS-T-TlTctra Tech/MCA
-------
TetraTech/TAMS
-------
channel station TIP-18C. the difference in concentration between Rogers Island and TIP-18C
(TIPC-Gain). or the concentration at TID-West normalized to solids concentration (F igure 2-2). The
pattern, however, is distinctly different from that of unaltered Aroclor 1242 (based on Aquatec
analyses). The similarity between the different water column measures, when evaluated as relative
percentages, coupled with the near lack of upstream load, removes a number of confounding issues
(such as whether the TIP represents a net addition or a replacement of the upstream load) and greatly
simplifies the analysis.
Measures of congener concentration in the TIP sediments are also available in a number of
variations. Figure 2-3 compares, for the selected GE peaks, the congener pattern found in the surface
0-2 cm layer of Phase 2 cores 18, 19, and 20 (analyzed as sum of quantitated congeners associated
with each GE/NEA peak); the pattern found in the top 0-5 cm layer of the GE 1991 composite
sediment samples; and. as an example of a more extensively dechlorinated pattern, the 8-12 cm layer
of Phase 2 core 18. The unweathered Aroclor 1242 pattern is also shown in this figure.
In this figure, the patterns in the Phase 2 and GE surface sediments are similar, except that
the relative contribution of BZ#4+BZ#10 appears elevated in the GE results. This probably reflects
the much more extensive spatial coverage of the 1991 GE data, plus potential re-contamination of
surface sediments prior to the collection of Phase 2 cores in fall 1992 (see Section 5.4). The 8-12
cm layer of Core 18 is clearly more dechlorinated. as shown by the depletion of BZ#5+BZ#8,
BZ#15^BZ#18. and BZ#28 relative to BZ#1 and BZ#4+BZ#10. More noticeable, however, is the
fact that all the sediment patterns appear to be significantly dechlorinated relative to unweathered
Aroclor 1242.
2.2 QEA Analysis of Sediment Source
QEA undertook a modeling approach to estimate the characteristics of a sediment which
would result in the observed PCB gain across the TIP. The results of this analysis, although credible,
are largely determined by the initial assumptions, which are not fully constrained by available data.
Further, although this is stated to be a "Final Report", the modeling which is presented is preliminary
and clearly in the process of further development.
2.2.1 Modeling Framework
It is not the intention of this review to provide a detailed critique or commentary on GE"s
evolving PCB modeling framework. This modeling framework is still under development, and is
not fully documented in QEA's (1998) report. Brief comments are. however, appropriate related to
the ability of the modeling framework to represent and assess sources of PCB load.
The modeling framework consists of four linked components: hydrodynamic models,
sediment transport models. PCB fate models, and PCB bioaccumulation models. Of these, the
bioaccumulation component is not relevant to the topics of this review . The other three components
are summarized below.
To represent hydrodynamics in the upper Hudson River. QEA has developed two separate
models: A finely-segmented two-dimensional vertically-integrated model which is used to estimate
December WH
:o
TAMSl.TLTetralcch/MCA
-------
Summer 1997 Water Column Concentrations
0 6
0.5
04
0.3
K 0.2
1
0
BZ#138
B2#70
BZ#153
BZ#52
BZ#31
Aroclof 1242 —Q— TIDWest ^ TIP-18C —^— TIPC-Gain ^ Cone on Solids
TetraTech/TAMS
Figure 2-2. Summer 1997 Water Column Relative PCB Congener Concentrations near the Thompson Island Dam Compared
to Aroclor 1242
-------
Thompson Island Pool Sediment Congener Concentration
04
0.3
c
3)
O
«
0)
cr
BZ#138
BZ#118*149
BZ#52
Aroclor 1242 —_ Core18-20Surf A GE Q-5cm Sedl HR 018-0812
Figure 2-3. Congener Pattern in TIP Sediment Compared to Aroclor 1242
TetraTech/TAMS
-------
shear stress at the sediment-water interface, and a simpler one-dimensional model designed to drive
long-term PCB fate simulations.
Sediment resuspension. deposition, and transport is also addressed by two separate model
components, one for cohesive sediment and one for non-cohesive sediment. There are two
significant limitations to the sediment component of the model, as currently configured, which may
limit its usefulness as a tool to evaluate the Thompson Island Pool PCB load. First, the model used
for the report does not consider resuspension of non-cohesive sediment, which constitutes a majority
of the surface area in the TIP. Preliminary results developed for USEPA (LTI, 1996) indicate, as
we would generally expect, that the non-cohesive sediments are subject to greater shear stress and,
potentially, greater amounts of erosion and potential loading of PCBs to the water column than the
relatively stable cohesive sediment areas.
A second limitation of the QEA approach to sediment modeling is that hydrodvnamic shear
stress is the only mechanism considered for mobilization of cohesive sediments. Given the low rates
of predicted resuspension from shear stress acting on cohesive sediment surfaces, other mechanisms
may be more important in mobilizing or freeing sediment mass from cohesive areas. Mechanisms
that might cause large-scale disturbances of cohesive sediment beds include destabilization of
undercut areas adjacent to the canal channel; mechanical abrasion by bedload and other debris; and
ice scour at spring ice breakup. Other localized, non-hvdrodvnamic scour disturbances which may
introduce sediment into the water column from cohesive sediment areas include; bioturbation by
benthic organisms, bioturbation by demersal fish, mechanical scour by boats (prop wash) and
floating debris in shallow/nearshore areas, and uprooting of macrophytes by flow. ice. wind, or
biological action.
QEA's PCB fate and transport model is generally similar to the approach used by USEPA:
It considers three-phase partitioning (water. POC, DOC) and models PCB transport from the
sediment to water column via diffusion, advective seepage, and sediment resuspension. It should
be noted that the QEA approach contains only a weak and indirect linkage between the PCB model
and the hydrodvnamic and sediment models. Typically, calibration constraints are placed on the
solids and PCB models simultaneously, a process made much mre difficult by weak linkage.
Additionally, sediment resuspension is used by the PCB model only on a reach-averaged basis
(QEA. 1998. p. 18): "The calibrated sediment transport model was used to generate a relationship
between the mass of sediment resuspended and flow rate for each of the eight reaches from Fort
Edward to the Troy Dam. These relationships were then used in the PCB fate model to determine
erosion rate in each model segment for a specified flow rate." It is not stated how non-cohesive
sediment scour or bedload movement were accounted for in these relationships. Further, QEA has
emphasized the importance of lateral variations in flow velocity and hence shear stress, but any
lateral variability represented in their sediment transport model is essentially lost when the results
are laterally averaged within each reach. Applying a reach average sediment flux to a reach average
PCB concentration will yield incorrect results, as the highest PCB concentrations are found in
depositional areas which, by definition, tend to experience lower shear stress.
Diffusive flux is similarly spatially aggregated across model reaches (p. A-9): "...the
diffusive loading equation can be used with area-weighted averages for organic carbon normalized
PCB concentrations to calculate the net TIP flux." This spatial aggregation implies that the model
is only appropriate for making inferences about net fluxes at the reach scale, and should not be used
December 12 1998
n
T AMS/LTlTetra T ech/MC A
-------
to make inferences about loading patterns within a model reach. Further, the approach of applying
a reach-average flux rate (of sediment or pore water) to a reach-average PCB concentration is likely
to yield biased estimates of loading if there is any correlation between PCB concentration and rate
of sediment or pore water flux. It is important to note that the model calculates resuspension from
cohesive sediments only, but surface concentration averages are apparently obtained over the whole
Thompson Island Pool. As PCB concentrations are likely to be higher in fine-grained cohesive
sediments, use of the average concentration with a sediment flux rate from cohesive sediments will
underestimate the total PCB flux by cohesive sediment erosion.
2.2.2 Model Calibration
The hydrodynamic model was calibrated to observations of 28-29 November. 1990, at which
time the flow at Fort Edward was steady at around 7.860 cfs. A validation test was performed using
data from the May 1983 flood, with a peak flow at Fort Edward of 34,100 cfs. using stage heights
reported at Champlain Canal staff gauges. (These were the same data used for calibration of a one-
dimensional hydrodynamic model in USEPA's Phase I effort (USEPA, 1991)). The hydrodynamic
calibration provides a reasonable fit. but is only matched to water surface elevation at two locations.
True calibration of hydrodynamics should include comparison to stage and velocity at multiple
stations. QEA emphasizes lateral variability in flow velocity in their interpretations of model results,
but provides no information to show that this lateral variability is correctly represented in the model.
For the sediment model. QF.A presents only a calibration to April 1982 USGS data, with no
validation results. This is not a very satisfactory data set for calibration because data are not
available on the loads from many tributaries, nor are data available from the TIP or below Thompson
Island Dam. In addition, only limited information is available on size class distributions during this
event. Neither the basis for assigning particle size classes to tributary loads nor the assumed settling
velocities are adequately documented. The sediment model described by QEA addresses erosion
of cohesive sediment only, yet the calibration data would include scour from both cohesive and non-
cohesive sediments. Model results appear to underpredict. by about one-third, peak suspended solids
concentrations at Waterford. while concentrations at Stillwater were also underpredicted by a small,
but consistent amount. QEA suggests that the under prediction "is likely due to an underestimation
of solids loading from the Hoosic River", and states, without providing details, that "More recent
calibrations of the sediment transport model using new tributary solids loading data confirm this
assessment of the preliminary model calibration." Note that calibration with incorrectly specified
tributary solids loads may lead to an incorrect estimate of the rate of resuspension of sediment w ithin
the river. It is not clear why calibration/validation to more detailed solids data collected by USEPA
in the Phase 2 effort or to the 1991 Thompson Island Pool suspended solids data collected by CiE
(O'Brien & Gere. 1993b) are not presented.
The PCB model was applied for long term simulation over the period 1977 to 1996. The
approach taken here is cause for considerable concern. First, the state variable was taken as total
PCBs. and assigned a single partition coefficient to organic carbon (in a three phase model) ot
40.000 I, kg (log K
-------
partitioning behavior of mono-, di-. and tri-chlorobiphenyls. which dominate water column
concentrations in the TIP and downstream. This would not present a major problem as long as the
congener composition remained relatively constant in space and time. Observations from the 1990's
show, however, that there is a strong shift in congener pattern across the Thompson Island Pool
under low flow conditions, with the mono- and di-chlorobiphenvl components largely absent at
Rogers Island. Modeling total PCBs with a single partition coefficient w ill reduce accuracy of the
model in predicting PCB fate and transport across the Thompson Island Pool. The major source of
error would come from lumping mono- and dichlorobiphenyls, which generally show weaker
partitioning, with more highly chlorinated homologues. The approach proposed by USEPA of
modeling the sum of tri- and higher-chlorinated homologues. instead of the sum of all congeners,
is both more consistent with historical analytical methods and reduces the influence of variability
in partitioning behavior among congeners. There is also a strong possibility that congener patterns
have changed over time, although little or no data to resolve this issue are available for the earlier
time periods.
Application of the same partition coefficients to water column particulates and the sediment
matrix is also questionable, as the physical availability of binding sites may be very different within
a compacted sediment. As presented in Section 2.3. the effective K(K- for the lightest congeners
(BZ#1. BZ#4+10) may be significantly lower in the sediment than in the water column.
For KD(X. QEA assumed a value equal to 10 percent of K
-------
concentrations is likely to miss the point for long term modeling, however. What the long term
model really needs is an accurate representation of upstream seasonal mass loading, not daily
concentrations. It would therefore appear advisable to establish the upstream boundary condition
by forming a best estimate of PCB mass load (see USEPA. 1997. Section 3.3.5). then apportioning
this load based on the flow volume over a time-scale that is appropriate to the mass balance model
application.
An additional problem in using the USGS data is that what was measured is not equivalent
to total PCBs determined by modern capillary column methods, resulting in a disjunction between
model input data coincident with the start of the GE monitoring effort in 1991. Instead, the packed
column results appear to approximate the sum of tri- and higher-chlorinated PCB homologues. NEA
conducted split sample experiments to compare the USGS packed column method results (based on
the description in Schroeder and Barnes. 1983) to capillary column analyses, using individual or
mixed standards composed of Aroclor 1242. 1254. and 1221. A regression of split sample results
for USGS-method total PCBs on the capillary column sum of tri- and higher-chlorinated homologues
results in a good linear fit. with an intercept not significantly different from zero and a slope not
significantly different from one. Thus the USGS packed-column results can be used as a measure
of the tri- and higher-chlorinated sum. but not as a measure of total PCBs.
The USGS laboratory switched to a capillary column analysis beginning in November 1987
(personal communication from Ken Pearsall. USGS/Troy, based on letter received from Brooke
Connor at USGS Denver laboratory ). USGS capillary column results are also believed to
approximate the sum of tri- and higher-chlorinated PCB homologues. rather than total PCBs.
although this issue is still under investigation by QEA and TAMS.
QEA (1998. p. 20) recognizes the issue of "analytical bias" in the USGS monitoring data,
but has not incorporated any correction into their modeling effort. The resulting discontinuity in the
upstream boundary condition associated with the switch to GE capillary column results in 1991
suggests the existing calibration of the PCB model should be regarded with a high degree of
scepticism.
QEA does not present any attempt to match the PCB model predictions to water column
observations. Instead, their •"calibration" of the long-term PCB model is based on matching
"observed surface sediment (0-5 cm) PCB concentrations in TIP and downstream in the vicinity of
Schuylerville. Stillwater, and Waterford." Data for 1991 are available for each of these locations
(O'Brien & Gere. 1993a). Relatively sparse data for the whole upper Hudson were also collected
in 1977 (O'Brien & Gere. 1978). Finally, detailed sediment sampling for 1984 is available from the
Thompson Island Pool only (Broun el al.. 1988). The result is that three separate sediment sampling
events are available for "calibration" in the Thompson Island Pool, while only two are available for
each of Schuylerville. Stillwater, and Waterford. It is not hard to fit a curve through two points, but
it is difficult to guarantee that the fit represents a realistic and unique interpretation of the data. To
obtain the results presented in the report. QEA also found it necessary to add "an empirically
defined, exponentially decreasing load ...to the TIP in the period between 1977 and 1983."
Similar to the water column daia. additional problems are occasioned by the fact that total
PCB measurements by the methods used in 1977. 1984. and 1991 are not equivalent. The 1991
sediment data (O'Brien & Gere. 1993a) were analyzed by modern capillary column methods and
December 22 |W8
26
TAMS/LTlTctraTcclvMCA
-------
represent an estimate of the sum of all PCB congeners for an 0-5 cm depth. The 1977 and 1984
analyses generally do not have a 5 cm slice. Indeed, in the 1984 analyses the "surface" core section
has an average length of about 10 inches, and concentration in the top 5 cm can only be guessed at.
The 1977 and 1984 analyses were also by packed column methods which are believed to miss most
mono- and di-chlorobiphenyls. Our analysis of the 1984 sediment data suggests that these results
approximate the sum of tri- and higher-chlorinated congeners (representing on average 93 .4% of this
sum). The 1977 sediment data are also suspected to approximate a sum of tri- and higher-chlorinated
congeners, but may have a small upward bias relative to the 1984 results due to the use of an Aroclor
1016 standard rather than an Aroclor 1242 standard. The 1977 sediment data also serve as initial
conditions for sediment in the model. Unfortunately, surviving documentation of this analytical
effort does not appear to be sufficient to definitively establish exactly what was measured in 1977.
QEA also compared its model predictions to estimates of annual PCB load derived from the
USGS data for 1977-1991, and imply a general agreement, although QEA's predicted loads at
Waterford are overestimated in the early I980's, and loads from 1983 through 1991 show only
limited variability. QEA's estimates of annual loads, however, appear to differ significantly from
those presented in the DE1R (USEPA, 1997, Table 3-23). For instance, the QEA estimate of 1983
calendar year loading is about 2300 lbs, while the Phase 2 estimate is about 3200 lbs. QEA does not
document how their annual mass load estimates were obtained for Figure 3-7; however, the method
was presumably that described on p. 27 for estimation of load across the TIP. This method
calculates annual loads based on the average daily load in observations. As demonstrated by Preston
et al. (1991). this is not an advisable approach to estimating annual loads from sparse data. The
problem is that load is typically correlated with flow (even if concentration is not). Therefore, if the
available data do not constitute an unbiased sample of annual flows calculating annual load from
observed daily loads will result in a biased estimate. Estimators which take into account the
relationship between loads and flows usually provide better results. As documented in the DEIR
(USEPA, 1997. pp. 3-132 through 3-133) a stratified version of a ratio estimator of annual loads
(Cochran. 1977) appears to be a good choice for PCBs in the Hudson, while a seasonal averaging
method (Dolan et al.. 1981) can be used as a check.
QEA also compared its PCB model predictions to weekly water column monitoring at the
Thompson Island Dam and found that the model underpredicted observed loads by 300 to 500
percent. Apparently, the Thompson Island Dam estimates were not corrected for the presumed
sampling bias at the TID-West station, but this would appear likely to account for only a small
portion of the under-prediction. because the correction factor is no more than 20 percent (0.8) for
the period prior to 1996.
The net result of these data issues is that neither the upstream boundary conditions, the
sediment initial conditions, nor the calibration targets are correctly specified in the existing QEA
PCB model. Further, the model calibration was not successfully validated against recent Thompson
Island Dam loads. It thus appears that the existing QEA PCB model should be considered an
incomplete experimental tool, and should not be used to test quantitative hypotheses regarding PCB
load from the Thompson Island Pool or other areas of the Hudson River.
10.0574
December 22 1998
TAMS/LTVT«raT«WMCA
-------
2.2.3 Depletion Rate of TIP PCB Inventory'
QEA (1998, Section 4.1.1) uses mass balance calculations to "test the hypothesis that the
anomalous PCB loading could be attributed to sediment surface PCB transport processes", and
concludes that "the PCB loadings observed from the TIP between 1993 and 1996 cannot be
representative of long-term surface sediment-water exchange processes" because they would result
in rapid depletion of the observed mass of monochlorobiphenyls (by 1995), dichlorobiphenyls (by
1996), and trichlorobiphenyls (by 2000). The analysis does not depend on the long-term PCB mass
transport model, but rather is based on simple mass balance calculations. Flux rates of PCB
homologues from TIP sediments were calculated from GE monitoring and compared to estimated
1984 surface layer concentrations to estimate time to depletion.
This analysis is flawed on a number of grounds, and should not be regarded seriously. Key
issues include the following:
1. The estimates of load generation from the TIP are based on uncorrected TID-West
concentration observations. According to QEA, these estimates are biased high, and the
estimates of TIP load generation, and rate of depletion, should be correspondingly lower. In
fact, it is likely that the actual loads are higher than those estimated by QEA, as discussed
in Section 1.
2. The analysis assumes that the monochlorobiphenyl and dichlorobiphenyl fractions of the
1984 sediment inventory may be estimated by application of the observed fraction in USEPA
1994 data to the 1984 total PCB estimate. In fact, as noted above, the 1984 quantitations
substantially do not account for the mono- and dichlorobiphenyl fractions, and instead
provide an approximation of the sum of tri- and higher-chlorinated homologues. No
evidence is available as to the inventory of monochlorobiphenyls and dichlorobiphenyls in
1984. Further, assuming that part of the 1984 inventory consists of monochlorobiphenyls
and dichlorobiphenyls results in an under-estimation of the surface sediment inventory of
higher-chlorinated congeners.
3. The analysis assumes that there is no replenishment of surface sediment PCB homologue
inventories. In fact, diffusion and pore water advection would both move dissolved and
DOC-bound PCBs into surface sediment from deeper, more highly contaminated sediment
reservoirs (see Section 2.3.3). Further, no accounting is made for erosive and mass wasting
processes which may mechanically move buried PCBs to the surface, particularly in unstable
non-cohesive sediments. Finally, there is some evidence suggesting replenishment of surface
sediment inventories by the Bakers Falls source during the early 1990's (see Section 5.4).
4. Estimates of loading L.re based on an average of observed loads. As noted in the previous
section, this is not an appropriate method for estimating annual loads from sparse point-in-
time data, and may result in significantly biased estimates of load.
In sum. the analysis of depletion rates is not supported by the data, and does not represent a
constraint on possible mechanisms of PCB load generation within the TIP.
December 22. 1998
28
TAMS/LTlTetraTech/MCA
-------
2.2.4 Analysis of Ground Water Seepage Flux
In Section 4.1.2. QEA provides an analysis of PCB loading by ground water seepage
(advective) flux, and concludes that this represents an insignificant source of PCB loading to the
TIP. This analysis also is flawed and cannot be used to draw firm conclusions.
The analysis presented by QEA involves estimation of an average seepage rate, and
application of this rate to mean surficial pore water PCB concentrations calculated from total PCB
concentrations in the 0-5 cm layer in 1991 (O'Brien & Gere, 1993a) by application of equilibrium
partitioning assumptions. A single Koc estimate of 1054 L/kg was applied to total PCBs, and
was assumed equal to 10% of K^ , which may result in inaccurate predictions, as described above
in Section 2.2.2. The seepage rate was estimated as 0.04 L/nr-hr, and the estimated seepage flux
was 11 kg/yr total PCBs (0.03 kg/d). However, analysis on a congener basis, with congener-specific
partition coefficients, would likely result in a greater estimate of ground water seepage flux. As a
check on QEA's rough estimates, the seepage rate can also be applied directly to pore water
concentrations observed in 1991. This provides a similar order of magnitude estimate of 6.4 kg/yr.
It thus appears that seepage advection could not be an important PCB loading source, if the
assumptions used by QEA are appropriate.
QEA's analysis of seepage rate is based on observations obtained from two replicate seepage
meters deployed at five locations within the TIP and one location downstream in May-June, 1997.
Such seepage meters have been used with considerable success in lake environments. Their use to
draw inferences within riverine environments is, however, fraught with difficulty. It is well known
that sediment texture within the TIP is highly heterogeneous, while the sediment is underlain by
fractured rock. In such circumstances, ground water seepage can be expected to flow via
preferential, localized pathways. Deployment of seepage meters at five locations is likely to miss
these preferential outlets, and thus underestimate total seepage rate. This is one of the reasons that
USEPA decided against deploying seepage meters during the Phase 2 sampling effort. In addition,
the seepage meter results were highly variable, and apparently subject to large uncertainties. A
better estimate of total gain from ground water flow across the Thompson Island Pool could be
obtained from careful monitoring of flow in the mainstem and tributaries. In addition, localized
seepage outlet springs or boils could provide a mechanism for resuspension of fine sediment during
quiescent low flow conditions.
A focus on ground water seepage may also miss important components of advective loading
from sediment, including interflow and drainage of exposed nearshore sediments. Interflow refers
to the fact that a portion of the total flow in a river may proceed through lateral flow within
permeable surface sediments. This could provide a mechanism for PCB advection from non-
cohesive sediments, but is unlikely to be significant in clays or other cohesive sediments. In
nearshore sediments there is a seasonal cycle of saturation and drainage, in which spring high stages
pump water into the sediment, which is subsequently drained, by both surface and subsurface
pathways, as stage recedes. L'nfortunately. the PCB inventory in shallow nearshore areas is very
poorly characterized due to limitations on boat accessibility.
In fact, empirical evidence suggests that advective flux may constitute a significant portion
of PCB loading within the TIP. During every year from 1991 through 1997. GE data suggest that
the rate of PCB load gain across the TIP declines from early to late summer. Because the hydraulic
December 22 IW8
29
T AMS-1. TLTerra Tech/MCA
-------
gradient between near-river ground water and the river also typically declines across this time period,
one possible explanation is that the seasonal decline in PCB gain is attributable to seasonal variations
in advective flux.
Figure 2-4 shows the average loads and concentrations of monochlorobiphenyls observed
at the TID-West station from 1991 through 1997. Monochlorobiphenyls were selected because loads
of this homologue group appear to arise almost entirely within the TIP, avoiding the difficulty of
calculating gain. Concentration peaks after the spring high flows, then declines across the summer
months. Monochlorobiphenyl load peaks during the spring flow, but also shows a similar decline
across the summer. Such a pattern would be consistent with a significant advective flux, which
would be highest immediately after spring flows and would decline over the summer. A second peak
is seen in the early fall, which is a period in which flows typically increase, and might be associated
with flushing of PCBs out of senescent macrophyte beds and other nearshore areas.
2.3 Reanalysis of Thompson Island Pool Sediment Source Congener Signature
A reanalysis of the potential characteristics of a sediment source to account for the summer
1997 TIP load was undertaken for this review. This reanalysis used three-phase partitioning in the
sediment, based on in situ partition coefficient estimates obtained from the GE 1991 data (O'Brien
& Gere. 1993a). Because of the analytical corrections made to the GE congener data in mid-1997
(HydroQual. 1997a). the three phase sediment partition coefficient estimates reported in the DEIR
(USEPA. 1997) are no longer valid, and were re-estimated for this work. Three different methods
of fitting these coefficients were used in the DEIR. For application to the TIP sediment pattern
matching it appeared desirable to use estimates obtained by a consistent method. Accordingly,
optimization method 3 (USEPA. 1997) was applied for all congeners (conditional optimization based
on estimated two-phase Koc J. The resulting estimates are shown in Table 2-2. As has been noted
previously, three-phase sediment partition coefficient estimates from the GE data are highly
uncertain, due to problems with the sample handling and compositing procedures. It is believed,
however, that the estimates of in situ partitioning provide the best available basis for attempting to
match water column concentrations to sediment.
These partition coefficients can be used to estimate absolute and relative concentrations of
congeners in pore water given a total sediment concentration. They also may be used to back-
calculate a total sediment concentration from water column gain, given assumptions about the
transfer mechanism from sediment to the water column. Results are presented below for (1) source
originating from pore water, and (2) source originating from a mix of pore water and bulk sediment
transfer to the water column.
December 22. 1998
30
TAMS/LTLT ctr a Tech/MCA
-------
40
O)
15
f 10
Load —m— Concentration
TetraTech/TAMS
Figure 2-4. Average Monthly Load and Concentration of Monochlorobiphenyls at the TID-West Station, 1991-1997
-------
2.3.1 Pore Water Source
QEA has focused on diffusive transfer from
sediment pore water as the main source of PCB
loading from TIP sediments. Using the partition
coefficient approach and pattern matching, the case
of a pure pore water source, whether loaded to the
water column via diffusion or advection. is readily
examined.
At first glance, the relative concentration
gain measured at TIP-18C looks quite similar to
the relative concentrations in surface sediment pore
water (Figure 2-5). The apparent agreement is.
however, largely due to the fact that both patterns
are dominated by BZ#4+10. For other congeners,
there is much less agreement, as there is a
substantially higher proportion of BZ#1 in pore
water than in surface water, while the more highly
chlorinated congeners have a relative percent of
21% in the TIP-18C gain, but only 5% in pore
water. Further, the tetra- and higher-chlorinated
congeners show a pattern which looks more like
sediment than pore water.
As noted above, congener concentration in
pore water consists of both a truly-dissolved and a
DOC-complexed phase. Together these represent
the apparent dissolved phase, denoted CPW a. For
a pure pore water source, the congener pattern in
the water column should be equivalent to the
pattern in CPWa. Equation (3-29) in the DEIR
(USEPA. 1997) states the equilibrium relationship
between CPWa and the particulate concentration.
CP—which is a close approximation to the total
concentration within the sediment matrix:
Table 2-2. Revised Three-Phase
Partition Coefficient Estimates for
PCBs in Sediment in the Freshwater
Portion of the Hudson River
PCB
Congeners
(BZ#)
logKoc
(L/kg)
l°g K-DOC
(L/kg)
1
4.46
3.63
4+10
4.73
3.60
5+8
5.78
4.03
15+18
5.95
4.23
22+51
6.14
4.48
28+50
6.49
4.36
31
6.17
4.33
44+104
6.98
5.78
52+73
5.98
4.32
66+93+95
6.09
4.53
61+70+76
6.01
4.10
101+90
5.98
4.68
118+149+106
6.10
4.91
138+163
6.31
5.12
153
6.28
5.25
C,
f<
oc oc
C
PW.L
6(1
mDoc^noc)
where
/(x is the fraction of organic carbon in the solid phase:
/C(K is the partition coefficient to organic carbon:
0 is the saturated porosity, or volume of water per volume of wet sediment.
Wtxx- is the mass of DOC per volume of pore water: and
^doc>s partition coefficient to dissolved organic carbon.
December 22. 1998
TAMS/LTLTctraTcch/MCA
-------
Summer 1997 TIP-ISC versus Sediment and Porewater
0.6
0.5
0.4
0.3
*0.2
1
0
BZ#138
BZ#1
BZ#52
BZ#153
BZ#31
!—[ T IPC Gam y GE 0 5cm PW 4) GE 0-5cm Sedl
TelraTech/TAMS
Figure 2-5. Relative Percent Patterns in Water Column Cain at TIP-I8C, Surface Sediment, and Surface Sediment Pore
Water
-------
This equation may be used to calculate
a congener pattern in a sediment source given
a congener pattern in the assumed pore water
flux to surface water. To apply the equation,
physical characteristics for the sediment are
assumed to the average from 0-5 cm sections
within the Thompson Island Pool (Reach 8) in
the 1991 GE sediment data (O'Brien & Gere,
1993a), as shown in Table 2-3.
Figure 2-6 shows the congener pattern
for sediment concentrations driving a pore
water source, as computed from the gain in concentration at TIP-1BC in summer 1997, and compares
this pattern to the pattern found in the 0-2 cm layer in Phase 2 Cores 18-20 and unweathered
Aroclor 1242.
The computed sediment concentration pattern appears to be quite different from that seen in the 0-2
cm layer of Phase 2 cores 18-20 (and the difference is greater when compared to the 0-5 cm layer
of 1991 GE cores from the Thompson Island Pool). While there are some similarities in pattern,
BZ#52 and BZ#28 appear to be elevated in the water column relative to the derived sediment pattern,
while BZ#1 through BZ#10 are depressed. The relative importance of these congeners, which tend
to have lower partition coefficients and a greater concentration in the water phase relative to
sediment phase, is lowered by the fact that large sediment concentrations of congeners above BZ#28
are required to account for the water column gain by a purely pore water mechanism.
As noted above, during summer 1997 there is little difference in congener pattern (despite
absolute differences in concentration) between observations at TID-West, TIP-18C, and the gain at
TIP-18C relative to Rogers Island. As a result, the derived sediment concentration is similar
regardless of which measurement is used as a basis for the analysis (Figure 2-7).
In sum, the available evidence contained in congener patterns does not appear to support a
theory of pore water flux (either diffusion or advection) as the sole source of PCB load gain in the
TIP-unless the congener pattern is strongly shifted in the water column by some unspecified
mechanism. Clearly, pore water constitutes part of the source of PCBs to the TIP. but apparently
not the only source. It also does not appear that unweathered Aroclor 1242 makes up the missing
part of the source.
2.3.2 Alternative: Mixed Pore Water and Bulk Sediment Loading
During a typical summer period there appears to be insufficient shear stress at the sediment-
water interface to scour significant quantities of PCB-contaminated sediment. Lack of significant
erosion of pool sediments during summer is also consistent with observed solids concentrations.
Nonetheless, the congener pattern observed in the water column is consistent with a source partially
composed of PCBs on bulk sediment, rather than PCBs partitioned from sediment into pore water.
An alternative mechanism to hydrodynamic scour for introducing PCBs on sediments into
the water column would be through localized disturbances which result in temporary introduction
Table 2-3. Physical Characteristics
Assumed for TIP Surface Sediments
0 (unitless)
0.386
foe (unitless)
0.01788
Wdoc (mg/L)
33.68
December 22. 1998
34
TAMS/LTCTetraTech/MCA
-------
Summer 1997 Derived Sediment Concentration from Gain at TID-West
Compared to HR Cores 18-20 and Aroclor 1242
0.2
0.05
BZJK28
BZ#138
BZ#1
BZ#31
BZ#153
Aroclor 1242 (=g— Corel8-20Surf TIP-C(Gain)
TetraTech/TAMS
Figure 2-6. Sediment Congener Pattern Derived from Summer 1997 Gain at TIP-18C Attributed to Pore Water Flux
-------
Summer 1997 Derived Sediment Concentrations
0 25
P 0.15
2 0 1
0)
o:
BZ#1 BZ^5'8 BZ&28 BZM44 BZ#66»95 BZ#101*90
BZ#4»10 BZ#15«18 B2«31 BZ#52 BZ#70 BZ«118*149 BZ#153
Summer97(To!al)
Summer97(Gain)
TIPC(Gam)
TelraTech/TAIVIS
Figure 2-7. Sediment Relative Concentrations Required to Support Observed Water Column Concentrations via Pore Water
Flux
-------
of contaminated sediment into the water column, followed by equilibration and exchange of PCBs
between sediment and the water column. This sediment may either settle out locally, or replace
influent solids to the TIP. such that there is little net increase in solids load. Localized, non-
hydrodynamic scour disturbances which may introduce sediment into the water column from either
cohesive or non-cohesive sediment areas during summer low flow periods include: bioturbation by
benthic organisms, bioturbation by demersal fish, mechanical scour by propwash in shallow areas,
mechanical scour by boats and floating debris in shallow/near shore areas, and uprooting of
macrophytes by flow, wind, or biological action.
To test the reasonableness of this theory experiments were performed to see if the observed
sediment concentrations could be reproduced by a weighted combination of surface sediment and
surface sediment pore water concentrations. Direct combination—which would be consistent with
net solids loading from TIP sediments to the water column, coupled with pore water exchange—does
not yield a close fit to the observed congener pattern. However, a very close fit can be obtained
under an assumption of sediment resuspension, exchange with the water column, and settling.
To provide a gross representation of the fractionation that occurs during the exchange process
it is simply assumed that, within the water column, sediment-sorbed PCBs re-equilibrate to
reproduce the average water column phase distribution shown in Table 3-8 of the DEIR (USEPA,
1997). following which the POC fraction settles back out while the dissolved and DOC fractions
remain in the water column. This fractionation would result in 91% of resuspended BZ#4 remaining
in the water column, but only 22% of BZ#118.
Using these assumptions, water column concentrations at TID-West can be fairly closely
predicted as a mixture of pore water and water column exchange with suspended sediment, using
average 0-5 cm concentrations in the TIP for sediment and pore water from the GE 1991 data
(Figure 2-8). In contrast, pore water alone provides a much poorer fit. Very similar results are
obtained by fitting a mixture to the estimated gain at TIP-18C in ng/L (Figure 2-9). In the case of
TID-West. the best fit coefficient on pore water concentration (ng/L) is 0.0034 and that on sediment
concentration (|ig/kg) is 0.0058: for gain evaluated at TIP-18C the coefficient on pore water
concentration is 0.0011 and that on sediment concentration is 0.0038.
In sum. observation of congener patterns in the TIP load gain suggests that this load is driven
by a mix of pore water flux (advection plus dispersion) and direct exchange of sediment with the
water column.
2.3.3 Influence of Advection and Dispersion on Pore Water Concentration
As has been noted above, concentrations in both surface sediment pore water and the TIP
load source are enhanced in the lightest congeners (BZ#1. BZ#4+10^ relative to unweathered Aroclor
1242. At first this appears somewhat surprising, as anaerobic dechlorination is not expected to be
significant in the surface sediment layer, and diffusion alone is not likely to be responsible for the
enhancement. The enhancement can. however, be inferred to represent differential transport in
ground water seepage, and thus supports the idea that seepage loading may be significant. Different
PCB congeners have partition coefficients that differ by orders of magnitude, and this affects the
speed with which they are transported to the sediment water interface. Because the dechlorination
end-products are among the congeners w ith the lowest partition coefficients, they are transported
December 21 1998
37
TAMS l.TLTctra FcclvMCA
-------
Summer 1997 Water Column Concentrations at TID-West
Predicted as a Mix of Porewaterand Surface Sediment
70
60
o>
£ 30
10
0
8Z#1
BZ#52
BZ#31
Predicted w-— Observed PW Only
TetraTech/TAMS
Figure 2-8. Concentrations at TID-West Predicted as a Mixture of Pore Water and Sediment Exchange
-------
Summer 1997 Water Column Concentration Gain at T1P-18C
Predicted as a Mix of Porewater and Surface Sediment
40
30
O)
ro 20
10
0
BZ#28
BZ#138
BZ#1
BZ#118*149
BZ#70
BZ#31
BZ#52
BZ#153
Predicted v Observed
TetraTech/TAMS
Figure 2-9. Concentration Gain at TIP-18C Predicted as a Mixture of Pore Water and Sediment Exchange
-------
more quickly to the surface. Thus, the observation of an increase in the molar percent of these
congeners is a natural consequence of the process of flux out of the sediments.
Under ultimate equilibrium conditions with an unlimited buried sediment source, partitioning
would not affect the relative concentration of congeners at the sediment water interface. However,
the time to concentration breakthrough at the surface from even a small depth (e.g., 1 foot) in
undisturbed sediments could well be on the order of hundreds of years or more, given the partition
coefficients observed, while concentration breakthroughs of the lightest congeners will be much
quicker. Since most of the contaminated sediments in the Thompson Island Pool have been in place
for 25 years or less, it is reasonable to expect an enhanced flux of the dechlorination end-product
congeners into the water column.
This can be seen via some simple numerical experiments with a one-dimensional advection-
dispersion ground water transport model. Consider the case where a substantial deposit of highly-
contaminated sediment was laid down following the removal of the Fort Edward dam, then covered
by a layer of less-contaminated sediment. Atop this there may be a layer of transient muck which
controls interfacial transport; however, for simplicity of the example we will consider only transport
within the in-place sediment and ignore the process of transport across the sediment-water interface.
For the example, consider that there is a "substantial" mass of contaminants at depth, which
provides a constant-concentration boundary at 10 cm depth, defined as x=0 (in this depth range, the
solution is not sensitive to the choice of the constant-concentration boundary depth). Initially, the
same concentration is assumed to apply from 10 to 3 cm depth, while the overlaying surficial 3 cm
is assumed to have an initial concentration one-tenth that in the layer below. Ignoring processes
directly at the sediment-water interface and examining only transport within the sediment to near
this level, the initial conditions are:
c(x,t= 0) =
C,
0 < x < 7
1 < x
(3-10cm)
(0 -3cm)
C. = 10
C\
while the boundarv condition is:
c(x=0,r) = C.
The solution to the one-dimensional advection-dispersion equation under these conditions (van
Genuchten and Alves. 1982, simplified version of solution #A5.) is:
c(x,() = C + (C - C ) — erfc
where
R(x-x ) - vf
2\[DRt
+ — exp(vx/D) erfc
~>
R(x+xx) + vt
2^/DRt
erfc is the complementary error function.
v is the interstitial or pore-water velocity.
/ is time.
D is the dispersion coefficient, assumed to be 106 cm:/s. and
R is the retardation coefficient, or rate of movement of water relative to rate of
movement of the pollutant.
December 22. 1998
40
TAMS/LTLTctra Tech/MCA
-------
The retardation coefficient. R, is defined as
R = 1 + p k/Q
where
p is the matrix dry bulk density.
k is the distribution coefficient, equal to ' foo and
0 is the volumetric moisture content.
Given a typical TIP organic carbon fraction in the sediments of 0.0179, matrix bulk density
of 1.35 g/cm3 and porosity of 38.6 %, predicted retardation coefficients for PCB congeners in
Hudson sediments are approximately 6% of the Kqc coefficient.
For the purposes of the example, consider a case in which there is seepage advection through
the sediments at an interstitial velocity of 3 m/vr. Figure 2-10 shows the predicted concentrations
(as a fraction of the concentration in the more contaminated buried sediments) near the sediment-
water interface after 25 years. Under these conditions, congeners with a log(Koc) greater than about
4.8 essentially show no influence of the more contaminated sediments below, and reflect the initial
concentration in the surface layer (specified to be 1/10 of C,). However, for log(Koc) equal to 4.6,
the surface concentration is expected to be 54% of C,, or five times the concentration present in the
initial surface concentration. In other words, over a 25 year time frame the less strongly sorbing
congeners may be mobilized from more contaminated sediments at depth, while more strongly
sorbing congeners will not.
As shown in Table 2-2, BZ#1 and BZ#4+10 have estimated values of log less than 4.8,
while other congeners tested have values greater than 5.7. This indicates that there is indeed a good
probability that seepage transport/retardation processes in the sediment can account for an
enrichment of BZ#1 and BZ#4+10 relative to other congeners in the spectrum.
A second point of interest in this analysis is that for most congeners the time to concentration
breakthrough may be very long for undisturbed sediments buried at even a small depth. Where more
contaminated sediments are buried and subject to an advective tlux. but are not disturbed by erosion,
this implies that the flux into the water column may still be rising, with breakthrough of many
congeners into the surface layer not yet achieved. For instance, for the situation described above
with a burial depth of 3 cm and an seepage velocity of 3 m/yr. surface concentrations of a congener
with a log (Koc) 5-6 would take 240 years to reach 50% of the buried sediment concentration,
while a log (K^) of 6 would require 600 years. For advection from buried sediments we might thus
expect to see a continued increase in loading to the water column over time. Further, concentrations
observed in the flux from the sediment should spike upwards in order of increasing partition
coefficients.
December 22. 1998
41
TAMS/LTlTetraTech/MCA
-------
100%.
60%
40% -
20%
i r
0%
5.2
5.4
5.6
5
4.8
5.8
4.6
6
44
4 2
4
Log (Koc)
TelraTech/l'AMS
Figure 2-10. Influence of K,k on Advective T ransport from Buried Sediment (3 cm Depth) to Surface Sediment after 25 Years
-------
3. Uniform Areal Flux of PCBs
PC B levels in the water column increase in a near linear fashion as water passes
through the TIP. indicating a nearly uniform areal flux from sediments within the
TIP (QEA. 1998. p. I)
PCBs appear to be loaded to the water column throughout the TIP. In other words, there are
no hidden major sources apart from the contaminated sediments known to be present in the Pool.
This is an expected result, as hot spot sediments are also distributed throughout the TIP.
QEA's presentation of a "near linear" flux of PCBs from sediments throughout the TIP seems
intended to downplay the importance of hot spots, with a contention that equal rates of flux occur
from the entire TIP sediment area. The primary evidence cited for this theory consists of four time-
of-travel studies conducted under approximately 4,500-5.100 cfs flow conditions in September 1996
and June 1997, which consisted of vertically composited sampling along three-point lateral transects
every 0.25 to 0.5 miles between Rogers Island and the TID. Within the TIP. however, there are so
many hot spots that the argument that all sediments contribute equally is unconvincing.
Part of QEA's argument is based on an observation that organic carbon normalized PCB
concentrations in surface sediment are similar in hot spot and non hot spot areas. This is contended
to result in equal pore water flux: however, as discussed in Section 5.1, this argument is invalid
unless the correlation between organic carbon content and sediment type (as shown, for instance, by
porosity) is also taken into account.
What the time-of-travel survey results do show is that, during low flow conditions, the
highest water column PCB concentrations tend to be associated with low-velocity, nearshore areas.
This finding is also consistent with the bias study results comparing near shore station TID-West to
center channel observations at TIP-18C (Section 1). Elevated concentrations in near shore low-
velocity areas are consistent with a pore water flux loading mechanism, which would result in higher
concentrations where dilution flow is lowest. These low flow areas are. however, precisely the areas
where sediment deposition and accumulation of PCBs is expected. QEA points specifically to high
concentrations observed in the backwater on the east shore opposite Snook Kill. This area is.
however, coincident with NYSDEC hot spot 8. and serves only to show that hot spots can generate
high concentrations.
For the 1996 time of travel studies there was at least one hot spot area between all
consecutive sampling locations except for stations 5 and 6. just below Rogers Island. The average
of the three lateral samples declined between these stations. For the 1997 time of travel studies the
only pair of sampling stations which did not encompass a hot spot were stations 15 and 15 A. A
small increase in average PCB concentration occurred between these stations due to an increase in
center channel concentrations. No increase was seen in nearshore concentrations between these
stations.
In sum. the time-of-travel results suggest that PCBs accumulate to the water column at a
fairly steady rate across the TIP. with a net increase of 0.4 to 0.6 kg d during average How rates of
4.500-5.100 cfs present during the time of travel surveys (QEA. 1998. p. 45). Given higher loads
with spring flows, this would appear to be consistent with an annual average load gain of 0.79 kg/d.
December 22 1998
43
rAMS1.TlTfftrjTech.MC A
-------
as stated in Section 1. Observ ations of a fairly steady rate of PCB gain do not necessarily negate the
importance of the hot spots as sources of PCB load.
4. Relative Contribution of Sediments below Thompson Island Dam
Sediments downstream of the Thompson Island Dam (TID) contribute PCBs to the
water column in a manner consistent with the TIP sediments (i.e. transfer from
surface sediment pore water), increasing the water column loading by approximately
50% between TID and Schuylerville. (QEA. 1998, p. 1)
The DEIR (USEPA. 1997) suggested that the TIP was the primary instream source of PCB load in
the Hudson, and that, under low flow conditions, this load was greater than that derived from
sediment betweeft Thompson Island Dam and Waterford. QEA (p. 55) states there is "an
approximately linear increase in PCB loadings with river mile" between Fort Edward and
Schuylerville, and "low flow loading estimates developed from USEPA water column transect data
produce a spatial pattern of PCB loading that is inconsistent with spatial patterns of sediment PCB
levels and our understanding of sediment-water interactions." (This latter conclusion is referenced
to USEPA. 1997. Figure 4-28. which is not directly relevant: Table 3-16 is more appropriate.)
If there is indeed a sampling bias associated with observations by EPA and GE at near-shore
stations near the Thompson Island Dam ( see Section 1), then relative load contributions from the TIP
and downstream segments need to be re-evaluated. The QEA (1998) conclusions regarding relative
loading, are. however, based on very limited data which are insufficient to reach final conclusions,
as no regular monitoring has been conducted downstream of the TID in recent years. In fact, the
conclusion of a "near-linear" increase with river mile (QEA Figure 4-27) is based on only four
samples, one from August and three from October 1997. These limited samples from late summer
and fall are not necessarily representative of either early-summer or spring high-flow loading
patterns.
Although the GE/QEA data are inconclusive, the presence of a potential sampling bias in TID
estimates may help in explaining some apparent anomalies in the Phase 2 data: In six of nine Phase
2 low-flow analyses (Transect 1. Transect 2. Transect 5. Flow Average 2. Flow Average 3. Flow
Average 4) the load at Thompson Island Dam appeared to be greater than the load at Waterford
(USEPA. 1997. Table 3-16). This might be explained by dilution by and settling of clean sediment;
however, presence of a sampling bias at Thompson Island Dam may provide a more intuitive and
parsimonious answer. The corrections to the river flow estimates noted for Chapter 3 of the DEIR
(See Book 1 of this responsiveness summary) also effect this issue.
For the Phase 2. 1993 low flow observations as reported in the DEIR, the mean load at
Rogers Island was 0.49 kg/d. w'.ile mean loads at Thompson Island Dam and Waterford were both
1.16 kg/d. This represents a load gain of 0.67 kg/d between Rogers Island and TID. However, the
load at Waterford is believed to be overestimated due to overly high flow estimates. (See corrections
to Chapter 3 in book I of this responsiveness summary). The actual load at Waterford is expected
to be roughly 40 percent lower at low flow conditions. (/' c .0.70 kg/day) yielding a net loss relative
to the TID. If it is assumed that a sampling bias correction factor (Section I) of 0.8 is appropriate
for Thompson Island Dam load estimates, the Phase 2 load at the TID would be decreased to 0.93
kg/d. and the adjusted gain within the TIP would be 0.44 kg/d (about 0.07 kg per mile per day),
December 22. I
-------
while the loss between the TID and Waterford would be 0.23 kg/d (or about 25 percent of the TID
load). Thus. TID sediments would appear to contribute significant amounts of PCBs to the water
column while downstream reaches lose PCBs from the water column, implying little if any net
contribution from the downstram sediments, even after the maximum likely correction for potential
sample bias at the TfD station. In fact. QEA's analysis of the time of travel surveys (QEA. 1998,
p. 45) suggests that TIP sediments contribute between 0.4 and 0.6 kg/d PCBs during flow conditions
around 5.000 cfs. nearly identical to the corrected value given above.
Of course, downstream hot spots are expected to contribute PCB loads to the water column
to some degree, and by a mechanism similar to that found in TIP sediments. This flux is not
necessarily via pore water only, as described above in Section 2.3.2. The difference in per-mile
loading rates above and below the TID reflects the lower areal coverage of hot spots, on average, in
reaches below the TID. Focus thus far has been on the TID sediments because these are a more
concentrated source for which more data are available. It is suspected that it will be possible to
model PCB fluxes from sediment both in the TIP and in downstream hot spot areas without the
necessity of invoking any "anomalous" special mechanism to explain the TIP load.
5. Hot Spot versus non-Hot Spot Sources
[S]urface sediments within all areas of the river contribute PCBs to the w ater
column, not simply PCBs residing in "hot spot" areas. Comparison of dry weight
sediment PCB concentrations, either at depth or at the sediment surface, gives a false
impression of the relative importance of various sediments w ithin the river. The
surface sediment pore water PC 'B concentrations, and. hence, the diffusive sediment
PCB flux is controlled by PCB concentrations associated with the organic carbon
component of the sediments .1.9 these average organic carbon normalized PCB
concentrations are similar within "hot spot" and non- "hot spot" areas, these areas
contribute similarly to the water column PCB load. (QEA. 1998. pp. 1 -2)
QEA implicitly sets up the hypothesis that known mechanisms of flux from "old" hot spot sediments
in the TIP (considered to be hydrodynamic erosion, diffusion, and pore water advection) are not
sufficient to account for the "anomalous" TIP load. Therefore, additional mechanisms are needed
to provide a newer, enhanced PCB load to surficial sediments in the TIP. Three additional
mechanisms are postulated:
1. PCB DNAPL loading in bedload along the sediment-water interface
2. Pulse loading of PCBs due to periodic flooding of the Baker Falls plunge pool
3. Transport of oil-soaked sediment into the TIP at the time of the Allen Mill collapse.
As an implied result of these "additional mechanisms". QEA claims that organic-carbon normalized
PCB surface sediment concentrations are similar across the TIP. and that these active sediment
concentrations are disconnected from buried hot spots.
5.1 Surface Sediment Concentrations in the TIP
QEA (1998. Table 4-6) presents information showing that mean PCB concentration in
surface sediments, w hen normalized to organic carbon concentration, is similar in the hot spot and
December 22 1998
45
rAMSl-TLTetra Tech/MCA
-------
non-hot spot areas, and is similar for fine and coarse sediment. They then state (p. 48): "The flux
of PCBs from surface sediments to the water column depends on the organic carbon normalized PCB
concentration, the sediment-water exchange coefficient, and the PCB partition coefficient as
described using Equations A-10 to A-15 (Appendix A), Regions of the river with equal surface
sediment organic carbon normalized PCB concentrations and composition contribute equally to the
water column PCB load."'
This argument is flawed. Suppose PCB concentrations on organic carbon are everywhere
the same, but location A has a high weight percent of organic carbon, while location B has almost
no organic carbon. Obviously, location A has a much greater mass of PCBs per volume of sediment
and is likely to contribute more PCB load to the water column, even if similar pore water
concentrations are calculated for each location under equilibrium conditions. What QEA's argument
primarily reflects is that hot spot areas are "hot" because they have more fine-grained sediment with
high organic carbon concentrations.
QEA's argument is invalid for any source mechanisms that involve bulk sediment movement
(scour, bioturbation. etc.), and only partly valid for consideration of a purely pore water source from
sediments. It is true that equilibrium partitioning assumptions imply that the observed apparent pore
water concentration. C,.u(including both dissolved and colloidally-sorbed PCBs) should be
proportional to the organic-carbon normalized PCB concentration, but this is not the only factor.
Rearranging Equation 3-29 (USEPA. 1997) yields
where CP is the paniculate concentration.
0 is the saturated porosity.
m(KjC is the mass of DOC per volume of pore water.
KmK is the partition coefficient to dissolved organic carbon, and
is the partition coefficient to sediment organic carbon
Inspection of this equation show s that the apparent pore water concentration depends not just
on the organic-carbon normalized sediment concentration but also on 0 and mlxx . As both porosity
and the concentration of dissolved organic carbon tend to increase in fine-grained, organic
sediments, the pore water concentration should also be higher in hot spot areas,
Analysis of the 1991 GE data from the 0-5 cm layer in the TIP reveals w ide ranges in TOC
concentration (from 4.961 to 69.474 ppm) and in porosity (from 16 to 70 percent). With a few
exceptions. TOC concentration increases with porosity (Figure 5-1). This correlation indicates that
inferences of pore water source strength cannot be based on organic carbon normalized PCB
concentrations alone.
In Phase 2 results (USEPA. 1997. p. 4-20) it was noted that "locations with...finer-grained
sediments have consistently higher median and mean PCB levels." The 1984 NYSDEC data also
show a strong relationship between sediment texture class and total PCB concentration, with the
highest concentrations tn the finest grained sediments. Table 5-1 shows the averages of NYSDEC
top core section and grab sample results for the near-surface layer. These results show a clear
increase in average PCB concentration for sediments with finer texture and higher organic content.
C
Cf 0(1
m DOC ^DOC-
K
oc
December 22 IW8
¦46
IAMS1 UlciraTechMCA
-------
100,000
80,000
60,000
g 40,000
20,000
0
40 60
Porosity (%)
100
TetraTech/TAMS
Figure 5-1. Correlation of TOC Concentration and Porosity in TIP Surface Sediments
-------
Results are similar for sample medians, except in the case of sediments classified as clay. A portion
of these samples are believed to include intact, uncontaminated glacial clays. In any case, it appears
clear that it is inappropriate to compare sediment concentrations as a source of pore water flux unless
both organic carbon fraction and porosity are taken into account.
Table 5-1. Surface PCB Concentrations in NYSDEC 1984 Data Compared to Texture Class
Texture
Class
Interpretation
Average
Total PCBs
(mg/kg)
Median Total
PCBs
(mg/kg)
Median
Specific
Weight (g/cc)
Sample
Count
FS-GRV
Fine sand
and gravel
14.7
9.1
0.9
7
CS-WC
Coarse sand,
wood chips
16.9
10.7
1.1
9
GRAVEL
Gravel
19.8
14.1
~
127
CS-SND
Coarse sand
25.0
13.8
1.25
22
GR-WC
Gravel, wood
chips
29.9
29.3
—
19
FS-WC
Fine sand,
wood chips
47.3
25.7
0.9
79
CLAY
Clay
54.9
6.7
1.0
10
FN-SND
Fine Sand
80.8
31.1
0.8
290
MUCK
Muck
121.1
103.8
0.5
14
It should be noted that it is reasonable to expect a smoothing out of surface concentrations relative
to buried hot spot concentrations. However, such a general smoothing of surface sediment
concentrations does not indicate that the surface PCB inventory is unconnected to buried hot spots.
PCBs introduced into the water column by erosion or other disturbance of bulk sediment would be
subject to local-scale settling, spreading concentrations. Some settling may also occur of PCBs
loaded to the water column via pore water advection. following partitioning to solids in the water
column, while lateral interflow could also "smear" the pore water signal.
5.2 PCB DNAPL Loading
.Another theory advanced by QEA/HydroQual is that transport of PCBs from the Bakers Falls
area into the Thompson Island Pool may proceed through bedload movement of droplets of dense
non-aqueous phase liquids (DNAPL}. Because they are denser than water, droplets of pure-product
PCBs would sink and remain near the bottom (if there was insufficient vertical mixing), and so
might move past the Fort Edward/Rt. 197 sampling station without detection. PCB DNAPL loading
could contribute a fresh supply of unweathered Aroclor 1242 to the surface sediment layer in the
Thompson Island Pool, and offer an alternative to in-place sediments as the source for TIP PCB load.
December 22. 1998
48
TAMS/LTLTeira Techy MCA
-------
If such a phenomenon did exist, it would imply that load estimates obtained from
concentration measurements at the Rt. 197 station at the head of the Thompson Island Pool are
biased low— resulting in a further diminishment of the importance of load generated trom in-place
Thompson Island Pool sediments. It is well established from GE field observations that PCB
DNAPL seeps occur in the area of the Bakers Falls plunge pool. What is not established is whether
any significant amounts of PCB transport past Fort Edward occur as bedload DNAPL.
The main objection to such a theory is that no observational evidence has been collected to
support the transport of PCB DNAPL droplets into the Thompson Island Pool. In addition, the
following should be noted:
Average surface sediment concentrations in 1991 and 1992, as well as the pattern of a
derived sediment which would account for the TID load via pore water flux, show a
significant elevation in concentration of monochlorobiphenvls and dichlorobiphenyls relative
to Aroclor 1242 which is consistent with a source driven by weathered PCBs, and not
consistent with significant replenishment of surface PCB concentrations by PCB DNAPL.
1992 core samples did not provide any evidence of accumulation of unweathered Aroclor
1242 in depositional areas of the Thompson Island Pool. They do suggest accumulation of
additional PCBs (Section 5.4), but only after significant fractionation in the water column.
Energetic hydrodynamics at and below the plunge pool suggest that any PCB droplets which
moved out of the pool would likely be broken up and mixed throughout the water column,
although no quantitative analysis has been performed. A water column cross section obtained
by O'Brien and Gere for GE in 1997 confirms the general homogeneity of the water column.
(QEA, 1998).
• The condition in which PCB DNAPL is most likely to be swept out of the plunge pool is
during spring high flows: however, comparison of load estimates at Rt. 197 and TID-West
suggests that the PCB load during high flows is likely transported through the TIP with little
mass loss. (It is assumed likely that the apparent high bias associated with TID-West
samples would not apply during energetic high flow conditions.)
HvdroQual (1997c) conducted a study from September 18-20. 1996 with fluorescent resin
particles (of approximately the same density as Aroclor 1242) to investigate the possibility of
transport of DNAPL droplets. The resin particles were released in slurry form into the fish bypass
line at AHDC's hydroelectric plant, and recovery monitored by passive filtration at a station at Fort
Edward 300 feet upstream of the north end of Rogers Island (river mile 194.9). at the Rt. 197 bridge
at Rogers Island (river mile 194.2). and 500 feet upstream of the Thompson Island Dam (river mile
188.8). Samples were collected at three depths and analyzed for resin particles as well as PCBs.
Released particle size ranged from 19 to 380 nm. The experiment suffered from a number of
methodological and analytical problems which resulted in difficulties in completing the resin particle
mass balance. It was estimated, however, that 28% of the particle mass (primarily in the size range
greater than 190 (im) was retained upstream of Fort Edward. 18% was retained between Fort Edward
and Rogers Island, and 44% was retained within the Thompson Island Pool, with only 1% of particle
mass (entirely in the 19-38(im size class) detected at the Thompson Island Dam station (this estimate
December 22 1998
49
TAMS/l.TlTetra Tech/MCA
-------
does not. however, account for delayed transport of particles which may have been retarded in eddies
and backwaters beyond the three days of the experiment).
The estimates of particle retention should be used with extreme caution, as they appear to be
subject to considerable uncertainty. Of particular concern is the possibility that much of the mass
of smaller particle size classes (<114 jam) may have passed through the in situ filtration devices.
This would result in an over estimate of the percent of particles retained in each reach.
HydroQual's study is somewhat informative as to the trapping patterns of rigid fluorescent
resin particles, but may not tell us much about the transport of PCB droplets. In contrast to the resin,
PCB droplets are liquid and may (1) deform and break into smaller particles, (2) gradually dissolve
into the water column, (3) sorb to organic sediment particles (whereas the resin particles may tend
to form hydrostatic bonds with clay particles), and (4) infiltrate into bottom sediment (if of sufficient
mass). Perhaps the most notable result of the experiment is that it failed to show preferential bedload
transport of the resin particles.
In their report, HydroQual (1997c, Figure 38) makes much of the fact that, in observations
of 18 September 1996 at the Fort Edward station (upstream of Rogers Island), the bulk of PCB mass
in the water column was detected near the sediment interface. This was due to high TSS
concentrations near the sediment interface on this date, probably representing localized scour. The
same condition does not seem to be found downstream at Rogers Island on 18 September, and the
vertical distribution of PCB concentrations showed no clear trend in observations taken on
September 19 and September 20, 1996 at the two stations, while the flourescent particle mass
(HydroQual 1997c. Figure 20) shows little or no vertical trend on all three dates.
In sum, there is no evidence to suggest that DNAPL bedload transport past Rt. 197 is a
significant component of the annual PCB mass balance above Thompson Island Dam. It is clear that
releases of pure-product Aroclor 1242 at Bakers Falls represent much of the PCB concentration
observed at Rt. 197; however, no evidence has been presented which indicates that the Rt. 197
observations are biased low. Consistent with this conclusion. QEA (1998. p. 39) states: "it is unclear
whether this mechanisms [PCB DNAPL loadings] has been contributing to the anomalous loading
observed from the TIP during the 1990s.
5.3 Flood Pulse Loading of PCBs
PCB DNAPL seeps to the Bakers Falls plunge pool have clearly contributed to the PCB load
entering the TIP. although it is not at all clear they have had a significant impact on surface sediment
concentrations within the TIP.
GE investigated two different mechanisms of pulse transport of DNAPL out of the plunge
pool: spring high flow loading and short-term pules during hydropower operations. The experiments
were conducted during 1997. which is a period in which the loading upstream of Rt. 197 appears to
have been minimal, apparently due to control of seeps in the plunge pool, so the results may not be
very informative as to loading mechanisms during the early 1990's.
High flow sampling was conducted during the spring event of April 6-9. 1997 (O'Brien &
Gere. 1998), during which maximum flow at Fort Edward was approximately 19.400 cfs. No
December 22. I <>98
50
TAMS/LTVTelraTech/MCA
-------
significant elevation of PCB concentrations was found in bedload sediments at Rogers Island, and
QEA (1998. p.41) concludes: "Based upon the high flow data collected in 1997, flow event driven
water column and sediment bed load PCB transport do not appear to be significant mechanisms for
continued pulse loadings of PCB from the plant site and into the TIP."
GE also conducted monitoring to assess shorter period loading of PCBs from the plunge pool
associated with fluctuations in flow through the AHDC hydroelectric facility, which cause periodic
"flushing" of the plunge pool (O'Brien & Gere, 1997). Surprisingly, they found that flushing of the
plunge pool resulted in elevation of concentrations in the plunge pool, probably due to collection of
small DNAPL seeps in the bedrock outcrop of the falls, rather than removal of mass downstream.
QEA (1998, p. 44) concludes that "periodic inundations of Bakers Falls provides relatively
insignificant PCB loads into the TIP."
Taken together, these findings suggest that PCB DNAPL from the Bakers Falls plunge pool
is transported downstream primarily through either dissolution and/or emulsification of very small
droplets into the water column. Enhanced transport during spring high flows may then be associated
with increased velocity and turbulence of DNAPL resident in the plunge pool, increasing the rate
of interfacial PCB transfer and suspension of small droplets. One interesting possibility is
suspension of small, non-settling-size droplets which tend to become associated with or coat
particulate matter in the water column. This might help explain why PCB congener partition
coefficients observed at Fort Edward during the Phase 2 sampling effort appear to be out of
equilibrium toward the particulate phase during transects 2 through 5 (USEPA, 1997, Figure 3-10),
with the apparent disequilibrium not evident during the winter low flow (transect 1) and late summer
low flow conditions (transect 6).
If Bakers Falls PCBs are transported into the TIP primarily mixed into the water column this
has two important implications: (1) the Rt. 197 sampling station is likely to be approximately
unbiased, and (2) PCBs from the Bakers Falls area are likely to have re-contaminated surface
sediment within the TIP only to the extent allowed by general settling of water column particulate
matter. Potential contributions of Bakers Falls PCBs to surface PCB concentrations in the TIP are
discussed further in Section 5.4.
5.4 Mass Loading of Contaminated Sediment Following the Allen Mill Collapse
QEA (1998. p. 36) also suggests the possibility that the Allen Mill collapse resulted in the
bulk movement of highly contaminated sediment into the TIP. although little further discussion is
provided. Thus far. no evidence seems to be available to support this theory, such as discovery of
surface deposits in the TIP with an elevated concentration closely resembling Aroclor 1242. Instead,
it appears likely that any contaminated sedimented mobilized from the Allen Mill collapse was
transported into the TIP more gradually as part of the general water column PCB transport process.
Monthly PCB loads estimated at the Rt. 197 station are shown in Figure 5-2, based on
calculations using a monthly averaging estimator (Dolan el ai, 1981). which has been demonstrated
by Preston et al. (1989) to provide relatively accurate estimates of load for samples obtained on a
fixed time schedule. Superimposed on the bars showing monthly load estimates is a 12-point
moving average line, which indicates long term trends. The loads first spike upwards in September
1991, following the failure of the gate structure in the Allen Mill. Intermittent high loads continued
December 22. 1998
51
TAMS/LTlTeuaTech/MCA
-------
PCB Load at Rt 197
Monthly Load and Moving Average
300 ——¦
250
1 -c 200
c
o
| 150
-------
through the spring of 1993, when flow through two of the three water ways in the mill was stopped.
Removal of contaminated material from the mill continued through fall of 1995, but average loads
(on an annual basis) remained relatively stable through 1994—1996. In September 1996 an apparatus
was installed to collect PCB DNAPL seepage at the base of Bakers Falls, while in January 1997 a
ground water production well was installed to hydraulicallv limit PCB seepage. (QEA. 1998, p. 7
discusses hydraulic control of the DNAPL source in general, but no detailed information on this
effort has been provided.) Since these latest remedial actions, PCB loading above Rt. 197 appears
to have remained low.
From Figure 5-2, it appears that the bulk of PCB mobilization from the Bakers Falls area
occurred between September 1991 and May 1993. Did this loading result in additions to the surface
PCB inventory in the TIP? If so, evidence should be seen in the surface layers of the Phase 2 High
Resolution Cores, collected in depositional areas of the Thompson Island Pool in the fall of 1992.
These cores were cut in two centimeter slices, and comparison of the 0-2 cm layer with the 2-4 cm
layer should help reveal the effects of 1991-1992 loading from the Bakers Falls source.
Comparisons on the basis of total PCBs and Aroclor 1242 equivalents contained in the Phase 2 data
base (USEPA. 1997) are similar. As the Bakers Falls source was unweathered Aroclor 1242. results
in terms of Aroclor 1242 equivalents are presented in Table 5-2. Five cores taken within the TIP are
included in this table, as well as one core taken a few miles below the TID, above Lock #5. All six
cores were collected in October or November of 1992.
Table 5-2. Comparison of 0-2 cm and 2—4 cm Aroclor 1242 Equivalent Concentrations in Fall
1992 High Resolution Cores in the Thompson Island Pool
Core
River
Mile
Location
Aroclor
1242
(Hg/'kg)
0-2 cm
Aroclor
1242
(Hg/kg)
2-4 cm
% Increase.
Surface Layer
HR-018
185.8
Above Lock #5
8.886
5.752
54%
HR-019
188.5
Thompson Island Dam
23,922
30,904
-22%
HR-020
191.2
Thompson Island Pool
26.046
20.652
26%
HR-023
189.3
Thompson Island Pool
2.952
1.141
159%
HR-025
194.2
Rogers Island West
6,149
8.717
-29%
HR-026
194.1
Rogers Island East
97.529
113.419
-14%
Three out of six core tops show an elevation in PCB concentrations relative to the 2—4 cm layer. The
results in the two Rogers Island cores may suggest that the Bakers Falls load passed Rogers Island
predominantly in the water column, rather than as bed load. Three out of four of the other stations
showed an increase. Results are difficult to interpret, however, because of the highly varying basis
of comparison in 2-4 cm slices. In cores 18. 20. and 23 surface concentrations are greater than 2-4
cm concentrations, but then increase again over depth from 4 cm to the PCB maximum, which is
December 22. 1998
53
TAMS/l-Tl/TeiraTech/MCA
-------
between 12 and 30 cm in depth. The occurrence of a minimum in the 2-4 cm layer in these cores
suggests that surface-layer PCB concentrations had been increased by recent upstream loadings past
Rogers Island. Ability to replicate these temporal changes should be a key test of the PCB fate and
transport model.
6. Summary
In this review, a number of major flaws have been found in the GE/QEA analysis. GE's
criticism of the DEIR's finding that a significant PCB load originates from TI Pool sediments is
based primarily on the assumption that there is a high degree of sampling bias at the TI Dam-West
station over the period from 1991 to 1997. However, the analyses conducted in this report show that
the degree of sampling bias is less than implied by GE/QEA and that the findings of the DEIR
regarding PCB loads are still valid after the correction for the analytical bias in the GE data. In
addition, the GE/QEA analysis depends on a sediment transport model for the TI Pool which
assumes that all areas of contaminated material are being buried, a condition which is highly unlikely
in a river setting such as this one. Finally. GE/QEA assumes that there is an oil-phase-based transfer
of PCBs to the TI Pool, despite the absence of evidence for it. While the GE/QEA analysis does
provide some insights into the Upper Hudson River system, the conclusions presented in the report
are frequently overstated and not supported by the data. USEPA's review of the GE/QEA report is
summarized in the Executive Summary at the front of this document.
December 22. 1998
<4
TAMS LTlTerra I ech. MCA
-------
References
Brown. M P.. M B. Werner. C.R. Carusone. and M. Klein. 1988. Distribution of PCBs in the
Thompson Island Pool of the Hudson River: Final Report of the Hudson River PCB Reclamation
Demonstration Project Sediment Survey. NYSDEC. Albany, New York.
Burgess, R.M., R.A. McKjnney, and W.A. Brown. 1996. Enrichment of marine sediment colloids
with polychlorinated biphenyls: Trends resulting from PCB solubility and chlorination. Environ.
Sci. Techno!., 30: 2556-2566.
Cochran, W.G. Sampling Techniques (3rd ed.) John Wiley & Sons, Inc., New York, NY.
Dolan, D.M., A.K. Yui, and R.D. Geist. 1981. Evaluation of river load estimation methods of total
phosphorus. J Great Lakes Res.. 7(3): 207-214.
HydroQual. 1997a. Development of Corrections for Analytical Biases in the 1991 -1997 GE Hudson
River PCB Database. Report to General Electric Company. Albany, NY. HydroQual, Inc.,
Camillus, NY, June 1997.
HydroQual. 1997b. Modeling Suspended Load Transport of Non-Cohesive Sediments in the Upper
Hudson River. Report to General Electric Company. Albany. NY. HydroQual. Inc.. Mahwah. NJ.
June 1997.
HydroQual. 1997c. Hudson River PCB DNAPL Transport Study. Report to General Electric
Company. Albany. NY. HydroQual. Inc.. Camillus. NY, June 1997.
LTI. 1996. Phase 2 Report—Review Copy—Further Site Characterization and Analysis. Volume
2B—Preliminary Model Calibration Report. Hudson River PCBs Reassessment RI/FS. Report to
USEPA Region II. Limno-Tech. Inc.. Menzie Cura & Associates. Inc., and The Cadmus Group. Inc.
October 1996.
O'Brien & Gere. 1998. Data Summary Report. Hudson River Project. 1996-1997 Thompson Island
Pool Studies. Report to General Electric Company, Albany. NY. O'Brien & Gere Engineers. Inc..
Syracuse. NY, February 1998.
O'Brien & Gere. 1997. 1997 Hydro Facility Operations and Thompson Island Pool Monitoring.
Report to General Electric Company. Albany, NY. O'Brien & Gere Engineers. Inc.. Syracuse. NY.
O'Brien & Gere. 1993a. Data Summary Report. Hudson River Project. Sampling and Analysis
Program. 1991 Sediment Sampling and Analysis Program. Report to General Electric Company.
Albany. NY. O'Brien & Gere Engineers. Inc.. Syracuse. NY, May 1993.
O'Brien & Gere. 1993b. Data Summary Report. Hudson River Project. Sampling and Analysis
Program. 1991 Thompson Island Pool Total Suspended Solids Study. Report to General Electric
Company. Albany. NY. O'Brien & Gere Engineers. Inc.. Syracuse. NY. May 1993.
December 22. IW8 55 TAMSXTI/TetraTeclvMCA
-------
O'Brien & Gere. 1978. PCB Analysis Final Report. O'Brien & Gere Engineers. Inc.. Syracuse.
NY. July 1978.
Preston. S.D.. V.J. Bierman. and S.F.. Silliman. 1989. An evaluation of methods for the estimation
of tributary mass loads. Water Resources Res.. 25(6): 1379-1389.
QEA. 1998. Thompson Island Pool Sediment PCB Sources. Final Report. Prepared for General
Electric Company. Albany. NY. Quantitative Environmental Analysis. Inc.. March 1998.
Rhea. J. 1997. Memorandum Re: TID Monitoring to M. Schweiger and J. Haggard. GE.
HydroQual. Inc.. September 30. 1997.
Schroeder. R.A. and C.R. Barnes. 1983. Trends in Polychlorinated Biphenyl Concentration in
Hudson River Water Fiv e Years after Elimination of Point Sources. Water-Resources Investigations
Report 83-4206. U.S. Geological Survey. Albany. NY.
USEPA. 1997. Phase 2 Report - Review Copy - Further Site Characterization and Analysis. Volume
2C - Data Evaluation and Interpretation Report. Hudson River PCBs Reassessment RI/FS. Report
to USEPA Region II and US Army Corps of Engineers Kansas City District. TAMS Consultants.
Bloomfield. NJ. February 1997.
USEPA. 1991. Phase 1 Report-Review Copy-Interim Characterization and Evaluation. Hudson
River PCB Reassessment RI/FS Report to USEPA Region II. TAMS Consultants. Inc.. Bloomfield.
NJ. August 1991.
Turk. J.T. and D E Troutman. 1981. Polychlorinated biphenyl transport in the Hudson River. New-
York. Water-Resources investigations Report 81-9. U.S. Geological Survey. Albany. NY.
van Genuchten. M. Th. and W.J. A Ives 1982. Analytical Solutions of the One-Dimensional
Convective-Dispersive Solute Transport Equation. Technical Bulletin No. 1661. U.S. Department
of Agriculture.
December I
-------
GE/QEA Report:
Thompson UUnd Pool Sedintut PCB Sources
Murch 1998
-------
OEIV
Thompson Island Pool Sediment
PCB Sources
Prepared for General Electric Co.
Albany, New York
MARCH 1998
FINAL REPORT
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
FINAL REPORT
CENhud 131
THOMPSON ISLAND POOL SEDIMENT PCB SOURCES
Prepared for:
General Electric Company
Corporate Environmental Programs
Albany, New York
Prepared by:
Quantitative Environmental Analysis, uc
March 19, 1998
mjwd- GENhud reports .upreport98 0398nprpi.wpd
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
TABLE OF CONTENTS
1 INTRODUCTION 1
2 BACKGROUND 3
2.1 History 3
2.2 Hudson Falls and Allen Mill Remediation 5
2.3 Upper Hudson River Water Column PCB Sources 7
2.3.1 Upstream of the plant sites 7
2.3.2 Plant sites, Allen Mill, and remnant deposits 8
2.3.3 Contaminated sediment deposits 9
2.4 USEPA's Analysis of Water Column PCB Data 11
3 QUANTITATIVE MODELING OF TIP SEDIMENT-WATER INTERACTIONS ... 13
3.1 Modeling Framework 13
3.2 Model Calibration 15
3.2.1 Hydrodynamics 15
3.2.2 Sediment transport 17
3.2.3 PCB fate 18
3.2.4 Summary of preliminary PCB fate and transport modeling 21
4 EVALUATION OF ALTERNATIVE HYPOTHESES
FOR ANOMALOUS PCB LOADING WITHIN TIP 23,
4.1 Additional Mechanism of PCB Exchange Between
Sediments and Water Column 25
4.1.1 Effect of long-term high flux on sediment PCB inventory 26
4.1.2 Measurement of ground water seepage rates 30
4.1.3 Estimates of low to moderate flow sediment bed resuspension 33
4.2 Additional PCB Sources 36
4.2.1 Simulation of PCB oil transport 37
4.2.2 High flow water column and sediment bed loading 40
4.2.3 Pulse loadings during periodic flooding of Bakers Falls plunge pool .... 42
4.2.4 Localized PCB source areas within TIP 44
4.3 Erroneous Estimates of PCB Flux Due to Biased Sampling 50
4.3.1 Route 197 Bridge in Fort Edward 50
4.3.2 Thompson Island Dam 51
4.3.3 Possible mechanism for the observed bias at TID-west 56
4.3.4 Composition of the sediment PCB source 56
4.4 New Paradigm for Sediment-Water PCB Exchange in the TIP 59
QEA
ii
March 19, 1998
-------
5 SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS 60
5.1 Summary 60
5.2 Conclusions 60
5.3 Recommendations 62
REFERENCES 64
APPENDIX A: CONCEPTUAL MODEL OF TIP PCB DYNAMICS A-l
A. 1 PCB Mass Balance A-l
A. 1.1 Partitioning A-l
A. 1.2 External loadings A-3
A. 1.3 Sediment sources A-3
A. 1.4 Settling A-4
A. 1.5 Advection A-4
A. 1.6 Volatilization A-5
A. 1.7 Governing equation A-6
A.2 Sediment PCB Source Loading Mechanisms A-7
A.2.1 Diffusive flux A-7
A.2.2 Sediment resuspension A-10
A.2.3 Groundwater advection A-11
QEA iii March 19, 1998
-------
LIST OF TABLES
N& Table Title Page
4-1. Average Annual Total PCB Loading Across TIP from 1993 to 1996 27
4-2. Surface Sediment PCB Inventory Depletion Under Average 1993-1996
TIP PCB Loadings 29
4-3. Parameters Used to Calculate Groundwater Induced Advection of PCBs from
Surface Sediments to the Water Column 33
4-4. Estimates of TIP Sediment and PCB Erosion as a Function of River Flow 35
4-5. Relationship between river flow rate and PCB concentration considering a constant
sediment flux rate 47
4-6. 1984-85 Organic Carbon Normalized PCB Concentrations Both Inside and Outside of
1976 NYSDEC "Hot Spots" within TIP 49
4-7. Paired Center Channel and TID-west Total PCB Concentrations 53
4-8. Parameters Used in the Calculation of Surface Sediment PCB Source Signature 58
QEA
IV
March 19,1998
-------
LIST OF FIGURES
^ Figure Title
2-1 Map of Hudson River from Glens Falls to Thompson Island Dam.
2-2 Temporal Trend in Water Column PCB Concentrations at Fort Edward.
2-3 Water Column PCB Loading at Fort Edward during Spring High Flow Events.
2-4 Temporal Trends in Mean Annual Low Flow Water Column PCB Loading from
Thompson Island Pool (1980-1997).
2-5 Temporal Trends in Mean Monthly Low Flow Water Column PCB Loading from
Thompson Island Pool (1993-1997).
3-1 Models, State Variables, and Kinetic Processes for PCB Dynamics.
3-2 Sediment Transport Model Grid for Thompson Island Pool.
3-3 I-D Hydrodynamic Model Predictions and Data: Stage Height During 1983 Flood.
3-4 2-D Hydrodynamic Model Predictions and Data: Stage Height During 1983 Flood.
3-5 Model-Predicted and Observed TSS Concentrations at Schuylerville, Stillwater, and
Water ford.
3-6 Comparison of Computed and Observed Surface (0-5 cm) Sediment PCB Levels in Four
Areas of the Upper Hudson River,
3-7 Comparison of Estimated of the Annual PCB Load Passing Waterford, NY Computed
from the USGS Data (symbols) and from the Daily Model Results (lines).
3-8 Comparison of Computed and Observ ed Water Column PCB Levels at Schuylerville for
the Years 1989- 1991.
3-9 Comparison of Computed and Observed Water Column PCB Levels at Thompson Island
Dam for the Years 1993 - 1996.
4-1 Schematic of Groundwater Seepage Meter.
4-2 Locations of Spring 1997 Groundwater Seepage Monitonng Stations.
4-3 Temporal Trends in Measured Groundwater Seepage into Thompson Island Pool.
4-4 Spatial Trends in Measured Groundwater Seepage into Thompson Island Pool.
4-5 Model-Predicted Sediment Bed Resuspension as a Function of Flow Rate.
4-6 Sampling Locations for Thompson Island Pool Time of Travel Surveys.
4-7 Spatial Profile of TSS Concentrations for 1997 Time of Travel Surveys.
4-8 Epifluorescent Photograph of Fluorescent Particles within Natural Sediment at
Approximately lOOx Magnification.
4-9 Schematic of In-Situ Particle Filtration Device.
4-10 Fluorescent Particle Mass Balance for PCB DNAPL Transport Study.
4-11 Fluorescent Particle Size Distribution for PCB DNAPL Transport Study,
4-12 Schematic of Passive Bed Load Sampling Device.
4-13 Temporal Trends in TSS and PCB Concentration and Loading During the 1997 Spring
High Flow Period.
QEA v March 19,1998
-------
4-14 Particulate PCB Concentrations for the Fort Edward Station.
4-15 Water Column PCB Concentrations at Bakers Falls Plunge pool and Fort Edward
from Hydrofacility Monitoring Program.
4-16 Center Channel PCB Concentrations from the 1996 Time of Travel Surveys.
4-17 Center Channel PCB Concentrations from the 1997 Time of Travel Surveys.
4-18 Comparison of West, Center, and East Channel PCB Concentrations at Select Transects
from the 1996 and 1997 Time of Travel Surveys.
4-19 Flow Velocity and Normalized Conservative Sediment Tracer Concentration Predictions
for the Snook Kill Vicinity of TIP.
4-20 1984 Organic Carbon Normalized Surface Sediment PCB (0-2.5 in.) Concentration
within the TIP.
4-21 Water Column Sampling Locations Within the Vicinity of the Routine Monitoring
Station at Fort Edward, NY.
4-22 Water Column PCB Concentrations Within the Vicinity of Fort Edward from the 1995
River Monitoring Test.
4-23 Water Column Sampling Locations Within the Vicinity of the Routine Monitoring
Station at the West Wingwall of Thompson Island Dam.
4-24 Comparison of PCB Concentrations Upstream of Thompson Island Dam and at the West
Wingwall.
4-25 Comparison of TSS Concentrations Upstream of Thompson Island Dam and at the West
Wingwall.
4-26 Comparison of PCB Concentrations Within the Vicinity of Thompson Island Dam.
4-27 Spatial Profile of Mean Upper Hudson River PCB Loading (Aug - Dec 1997).
4-28 Spatial Profile of Upper Hudson River PCB Loading During Summer Low Flow Period
for EPA (August 1993) and GE Data (August 1997).
4-29 Flow Velocity and Normalized Conservative Sediment Tracer Concentration Predictions
for the Thompson Island Dam Vicinity of the TIP.
4-30 PCB Homolog Distribution of Water Column Delta Load Across the TIP and Calculated
Sediment Source Required to Produce Water Column Load by Equilibrium Partitioning.
4-31 Comparison of PCB Peak Compositions for Calculated Diffiisional Sediment Source
(1997 Summer Average) with Sediments from 1992 EPA High Resolution Cores
Collected from TIP.
A-l Conceptual Model of PCB Dynamics in the Hudson River.
QEA
VI
March 19,1998
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
TIP Sediment PCB Sources
SECTION 1
INTRODUCTION
Since 1990, the U.S. Environmental Protection Agency (USEPA) has been performing a
reevaluation of the 1984 Superfiind no-action decision for the PCB-containing sediments within the
upper Hudson River. One principal objective of the reassessment is to determine the relative
importance of the varied sources contributing to water column PCB loadings. This report provides
a quantitative analysis of water column PCB sources within the TIP based on the extensive historical
database of water and sediment PCBs generated by the state and federal governments and the more
recent data sets generated by the General Electric Company (GE).
As a result of this work the following major conclusions can be drawn:
• During the 1990s, the amount of PCBs leaving the TIP was significantly overestimated due
to a sampling bias at the routine sampling station located at the downstream limit of the TIP;
• The composition of water column PCBs attributed to the TIP sediments indicates that
relatively undechlorinated PCBs are the principal source and that surface sediment pore
water is the principal point of origin;
• PCB levels in the water column increase in a near linear fashion as water passes through
the TIP, indicating a nearly uniform areal flux from sediments within the TIP; and
• Sediments downstream of the Thompson Island Dam (TID) contribute PCBs to the water
column in a manner consistent with the TIP sediments (i.e., transfer from surface sediment
pore water), increasing the water column loading by approximately 50% between TID and
Schuylerville.
The analyses presented in this report demonstrate that surface sediments within all areas of
the river contribute PCBs to the water column, not simply PCBs residing in "hot spot" areas.
Comparison of dry weight sediment PCB concentrations, either at depth or at the sediment surface,
gives a false impression of the relative importance of various sediments within the river. The surface
QEA
1
March 19,1998
-------
TIP Sediment PCB Sources
sediment pore water PCB concentrations and, hence, the diffusive sediment PCB flux is controlled
by PCB concentrations associated with the organic carbon component of the sediments. As these
average organic carbon normalized PCB concentrations are similar within "hot spot" and non-"hot
spot" areas, these areas contribute similarly to the water column PCB load. This finding has
important implications for the development and evaluation of remedial strategies for the river.
The conclusions of this report are in many cases inconsistent with those reached by the
USEPA in the Data Evaluation and Interpretation Report (USEPA, 1997). The differences are
primarily due to the results of additional data collection since the release of the USEPA report and
the application of a rigorous, quantitative PCB fate and transport modeling effort sponsored by GE.
USEPA is in the process of developing a similar model.
This report has been prepared by Quantitative Environmental Analysis, LLC. (QEA) on the
behalf of the GE to document the results of numerous field research, data analysis, and modeling
efforts investigating the origin, fate, and transport of PCBs within the upper Hudson River. Section
2 provides a historical background of the Hudson River PCB problem and describes significant
events that have impacted the observed temporal changes in water column PCB loadings. Section
3 describes the basic physical and chemical processes affecting PCBs in aquatic environments and
their incorporation into a state-of-the-science PCB fate and transport model. Section 4 presents the
results of field research, data analysis, and modeling studies conducted on the nver over the last
several years that are the basis for the conclusions presented above. Section 5 presents the summary,
conclusions, and recommendations drawn from the analysis presented.
QEA
i
March 19, 1998
-------
TIP Sediment PCB Sources
SECTION 2
BACKGROUND
2.1 History
Over an approximate 30 year period, ending in 1977, two GE capacitor manufacturing
facilities in Fort Edward and Hudson Falls, New York discharged PCB-containing wastewaters into
the upper Hudson River. Much of the PCBs accumulated in sediments upstream of the former Fort
Edward Dam located approximately 2 miles downstream of the Hudson Falls capacitor plant (Figure
2-1). Removal of this dam in 1973 by the owner (Niagara Mohawk Power Corporation) and
subsequent high flow events resulted in the movement of large quantities of PCB-containing
sediments downstream. Some of these sediments deposited further downstream in pools formed
by dams along the Champlain Canal, which is coincident with the Hudson River channel (USEPA,
1984).
In the late-1970s, the New York State Department of Environmental Conservation
(NYSDEC) undertook a number of studies to determine the concentration and distribution of PCBs
in the water column, sediments, and biota of the upper Hudson River. As a result, they identified
sediment "hot spot" areas defined as regions of the river containing sediments with PCB
concentrations exceeding 50 parts per million (ppm). Forty of these "hot spots" were identified in
the 40 mile stretch of the upper Hudson River between Fort Edward and Troy, N.Y. Twenty "hot
spots" were located in the TIP, a six mile section of the river formed by the TID, which is the first
dam downstream of the former Fort Edward Dam. In the early I980's, the NYDEC proposed that
the sediments from the TIP "hot spots" be removed and placed in a landfill in Ft. Edward, New York.
Due to community opposition, the NYSDEC was unable to proceed with the project.
QEA
3
March 19,1998
-------
TIP Sediment PCB Sources
In 1984, the USEPA placed the upper Hudson River on the Superfund National Priorities List
and issued a Record of Decision (ROD). The ROD determined that the approximately 60 acres of
shoreline PCB deposits upstream of the former Fort Edward Dam, formed when the pool elevation
dropped approximately 20 feet due to the removal of the dam, were to be capped in-place to
minimize direct contact with the exposed PCB-containing sediments. For the PCB-containing
sediments within the TIP and downstream, an interim no-action decision was reached for a number
of reasons, including: 1) declining PCB levels in water and fish as a result of source control
measures on the plant sites and natural attenuation processes in the river, and 2) the unproven status
of contaminated sediment removal technology (USEPA, 1984).
After the 1984 ROD, GE entered into agreements with the Federal government to implement
the remnant deposits capping program. This was carried out between 1988 and 1991 (JL
Engineering, 1992). In addition , GE implemented a water column monitoring program beginning
in 1989 (Harza, 1990) to monitor the construction activities on the remnant deposits and to
demonstrate that the remedy was functioning as intended1. The NYSDEC continued to pursue a TIP
"hot spot" dredging and landfill program, and in 1987 began the process of siting a local landfill,
which ended in 1989.
In 1990, the USEPA reopened the 1984 no-action decision on the PCB-containing sediments
of the upper Hudson River and initiated a reassessment remedial investigation and feasibility study
(RRI/FS). Although GE was only one of two named potentially responsible parties (PRPs; the other
being Niagara Mohawk Power Corp.), the USEPA decided to complete the RRI/FS using
government contractors and funds. The complexity of the technical issues associated with assessing
the origin, fate, and transport of PCBs in the system has delayed the original schedule of the RRI/FS,
which is now scheduled for completion sometime after the year 2000.
'This program and later variants provided much of the water column PCB data presented in this report.
QEA 4 March 19, 1998
-------
TIP Sediment PCB Sources
Although GE was not permitted to perform the RRI/FS for the USEPA, the company
collected relevant scientific data that would enable: I) a better understanding of PCB dynamics
within the system, and 2) the development of a state-of-the-science PCB fate and transport model.
The data collection program started in earnest during the spring of 1991 (O'Brien & Gere, 1993a,
1993b). A key component of the program was the routine (at least weekly) monitoring of water
column PCB concentrations at a number of stations in the upper Hudson River, including (Figure
2-1):
•Route 27 Bridge in Hudson Falls (background station),
•Route 197 Bridge in Fort Edward (downstream of the plant sites and remnants deposits and
upstream of the TIP), and
•the TID (downstream of the TIP).
This monitoring has continued and now provides a valuable data set to evaluate the temporal trends
in water column PCBs in the upper Hudson River.
2.2 Hudson Falls and Allen Mill Remediation
During the routine monitoring performed by GE, a significant increase in water column PCB
loading was detected after mid-September 1991. This loading originated upstream of the Fort
Edward and downstream of the Route 27 Bridge stations (Figure 2-1). Within a weeks time, PCB
levels within the river increased from less than 100 ng L 1 to approximately 4000 ng L 1 (O'Brien
& Gere, 1993a). After significant investigation, the source of the increased water column PCB
loading was attributed to the collapse of a wooden gate structure within an abandoned paper mill
(Allen Mill) located adjacent to the Hudson Falls capacitor plant on Bakers Falls (O'Brien & Gere,
1994a; Figure 2-1, inset). The gate had kept water from flowing through a tunnel cut into bedrock
below the mill, presumably since the mill's closure in the early 1900s. The tunnel contained oil
phase PCBs that migrated there via subsurface bedrock fractures.
QEA
5
March 19,1998
-------
TIP Sediment PCB Sources
In January 1993, with the cooperation of the Bakers Falls Hydroelectric Dam owner and the
NYSDEC, the water flow through the mill was largely controlled. By Spring 1993, two of the three
water ways within the mill were isolated from the river and planning for the removal of PCB
containing material from within the Allen Mill commenced. Removal continued until the fall of
1995. Approximately 45 tons of PCBs were contained in the 3,430 tons of sediment removed from
the Allen Mill (O'Brien & Gere, 1996a).
As part of the investigation and clean-up of the Allen Mill, dense non-aqueous phase liquid
(DNAPL) seeps of PCBs were discovered within the exposed bedrock of the falls. In 1994, during
the construction of the new dam at Bakers Falls. PCB DNAPL seeps were observed in the portion
of the falls adjacent to the Hudson Falls plant site. A number of actions have been taken to contain
and control these PCB seeps including grouting of bedrock fractures, manual collection of PCB oils,
when accessible, and the operation and installation of pumping wells to hydraulically control the
seeps. The release of PCB DNAPL through these bedrock seeps has declined in response to
mitigation efforts, but has not ceased. While efforts are made to collect the material, uncollected
oils are released into the river when the falls are inundated during elevated river flow events.
Sediments and debris from the vicinity of the original wastewater outfall located immediately
upstream of the dam and the area where the seeps are concentrated are being removed in an
additional effort to control the seeps. This removal is scheduled for completion during the Summer
of 1998.
In September 1996, divers discovered an additional area of PCB DNAPL seepage at the base
of the Bakers Falls adjacent to the eastern shore in an area referred to as the plunge pool. A
subaquatic collection system was installed to arrest the flow of the PCBs into the river. This seep
produced approximately 0.5 pounds per day of PCBs. In January 1997, a ground water production
well was installed on the shoreline upgradient from this seep in an effort to hydraulically control
QEA
6
March 19, 1998
-------
TIP Sediment PCB Sources
PCB discharges from the seep. This well produces significant quantities of PCB DNAPL and
appears to have controlled discharges from the seep as PCB DNAPL has not been observed in the
subaquatic collection system since the installation of the on-shore recovery well.
In addition to the activities to control riverbed PCB seeps and PCB movement from the Allen
Mill, GE has conducted an intensive investigation and remedial program at the Hudson Falls plant
site. DNAPL PCBs have been discovered in the fractured bedrock below the site. To date, over
3,000 gallons of oil have been removed from the subsurface. A series of 26 ground water pumping
wells have been installed to create a hydraulic barrier between the site and the river, not only to
collect PCB-containing ground water but also PCB-oil (GE, 1997). The effectiveness of this system
in reducing PCB flux from the site to the river is being monitored by measuring PCB levels in the
river, and through an assessment of the hydraulic capture zone created by the groundwater pumping
system.
2.3 Upper Hudson River Water Column PCB Sources
Numerous upper Hudson River water column PCB sources have been identified and
quantified using water column PCB data collected from four primary monitoring locations: the
Route 27 Bridge, the Route 197 Bridge at Fort Edward, the western abutment at TID, and the Route
29 Bridge at Schuylerville (located approximately six miles downstream of the TIP).
2.3.1 Upstream of the plant sites
The background station at the Route 27 Bridge typically yields water column PCB
concentrations of less than the method detection limit of 11 ng/1 (ppt). While there are known PCB
sources upstream of this sampling station, most notably Niagara Mohawk Power Corporation's
QEA
7
March 19,1998
-------
TIP Sediment PCB Sources
Queensburv site, they do not appear to be significant sources of PCBs to the water column of the
upper Hudson River. Water column PCBs at the Route 27 Bridge station are likely present at
quantities between 1 and 11 ppt (USEPA, 1997).
2.3.2 Plant sites, Alien Mill, and remnant deposits
Potential external PCB sources between the Route 27 Bridge and the Route 197 Bridge in
Fort Edward (Figure 2-1) include: the Hudson Falls capacitor site, the Allen Mill, the remnant
deposits (including the site adjacent to the former Fort Edward Plant outfall area referred to as the
004 site) and the Fort Edward capacitor manufacturing site. The steep river bed grade in this reach
of the river produces flow velocities that inhibit sediment deposition. Therefore, there are only
limited areas of sediment accumulation in this portion of the river, and water column PCB loadings
observed at the Fort Edward station generally reflect the activity of the external sources. This
activity is illustrated in Figure 2-2 which presents the results of water column PCB measurements
made at the Fort Edward station since the 1970s. Additionally, Figure 2-3 shows the PCB loading
observed at this station during three spring high events in the 1990s. The following observations
can be made from these data:
•PCBs have been present in samples collected from this station since the late-1970s,
•PCB levels declined between the late 1970s and late 1980s,
•PCB levels increased dramatically in September 1991 as a result of inputs from the Allen
Mill,
•Remediation of Allen Mill and efforts to control PCB releases to the Hudson River reduced
the large PCB loading observed during the 1991-1993 period, and
•PCB levels during the annual high flow period have declued in response to source control
measures implemented at the mill and plant site.
QEA
8
March 19,1998
-------
TIP Sediment PCB Sources
These data indicate that a non-sediment PCB source has been active for many years. Even
before the failure of the Allen Mill gate, a base load of PCB was entering the river, presumably from
fractures in bedrock near the Hudson Falls site. Only recently has this base load of PCB been
controlled. Although plant site sources still exist, it appears that remedial measures at the Hudson
Falls plant site have reduced water column PCBs in this segment of the river to levels below those
observed in the late-1980s. The current flux from the site is still being evaluated. Finally, these data
indicate that the Allen Mill event, while transitory, represents the largest external PCB loading event,
both in duration and magnitude, seen in this section of the river since the late-1970s.
2.3.3 Contaminated sediment deposits
The contaminated sediments within the upper Hudson River represent a source of PCBs to
the water column. Within a given reach of the river, this source can be estimated as the difference
in the product of PCB concentrations and flow between an upstream and downstream station2.
Figure 2-4 presents PCB loading between either Ft. Edward and Schuylerville (12 rrule length of the
river) or between Ft. Edward and T1D (6 mile length of river), for the period 1980 to 1997. The
earlier data (1980 to 1991) depicts loading over the longer reach (12 miles) and the later data (1991-
1997) depicts loading over the TIP region only. Figure 2-5 presents the loading from the TIP alone
between 1993-1997. These data indicate that:
•Through the 1980s, PCB loading from the contaminated sediments decreased from
approximately 1 pound per day to approximately 0.25 pounds per day (although significant
year-to-year variation is apparent);
2This ignores any losses from the water column due to settling or volatilization which have been judged to
be minor at low to moderate nver flows.
QEA
9
March 19, 1998
-------
TIP Sediment PCB Sources
•An increase in PCB loading, from approximately 0.25 pound per day to between 1 and 2
pounds per day occurred between 1989 and the early 1990s;
•The loading exhibits a seasonal pattern with the highest loading observed following the
annual spring high flow period and the lowest loading observed in the winter;
•The PCB loading through the 1990s has not showed significant declines although the data
contains significant year-to-year variations; and
•The lowest PCB loading since 1993 was observed in 1995, a year in which Spring flows
were significantly lower than in other recent years.
While the decrease in PCB loading through the 1980s is consistent with natural recovery of
the system through the burial of contaminated surface sediments with clean material, the cause of
the apparent increase in loadings observed from this region of the river in 1991 is unclear. Several
changes, both in the river monitoring program and the activity of external PCB sources occurred
during this period. First, a monitoring station was added at the TID to assess PCB loadings directly
from the TIP. Second, the PCB analysis scheme was changed from an Aroclor-based scheme that
failed to detect the lowest chlorinated congeners to one that quantified the full spectrum of PCB
congeners. Finally, over an approximately 18 month period beginning in 1991, the river experienced
the largest external PCB loading since the late 1970s. Each of these changes may have exerted some
influence on the observed PCB loadings from the TIP in the 1990s.
Estimates of PCB flux from TIP sediments, based on surface sediment conditions measured
in the summer of 1991, cannot account for the PCB loadings observed from the TIP (HydroQual,
1995a)3. Possible causes for this apparent increase in PCB loading were presented in earlier reports
and formed the basis for the data collection programs undertaken over the last few years (HydroQual,
JThc TIP anomaly is defined as the excess PCB loading observed from the TIP since approximately 1992
that can not be accounted for by known PCB fate processes given the known PCB concentranon of surficial
sediments I HydroQual. 1995a).
QEA 10 March 19, 1998
-------
TIP Sediment PCB Sources
et al., 1997a, 1997b, O'Brien & Gere 1997a, 1997b). The results of this work will be discussed in
Section 4 of this report.
2.4 USEPA's Analysis of Water Column PCB Data
In association with the on-going RRI/FS of Hudson River PCBs, the USEPA issued a report
in February 1997 that documented their interpretation of water column and sediment data collected
in 1992 and 1993 (USEPA, 1997). The USEPA concluded that PCBs passing the TID during low
flow conditions'4 were the major source of PCBs to the freshwater Hudson. Additionally, the USEPA
contended that sediments within "hot spot" areas of the TIP contribute the majority of PCBs passing
the TID during low flow periods.
The USEPA's interpretation of the data did not recognize that the loading observed from the
TIP could not be explained via known PCB fate and transport mechanisms given the level of PCBs
within surface sediments (the TIP anomaly; GE, 1997). Moreover, the agency did not fully consider
the temporal correspondence between the appearance of the excess loading, the upstream PCB
loadings from the plant site areas, and the change in sampling and analytical methods. Based upon
a qualitative assessment of the data, the agency offered three possible mechanisms for transfer of
PCBs from the sediment to the water column:
1) sediment pore water diffusion of relatively undechlorinated PCBs partitioned from the
particulate to the pore water phase,
2) groundwater-induced advective flux of sediment pore water PCBs within the TIP, and
3) resuspension of sediments contaminated with extensively dechlorinated PCBs deposited
prior to 1984.
4Less than 10,000 cfs at the USGS Fort Edward gauging station.
QEA 11
March 19,1998
-------
TIP Sediment PCB Sources
The US EPA did not conduct a quantitative mass balance evaluation to test these hypothesized
mechanisms, but simply offered them as possible explanations for the observed loading from the TIP
(US EPA, 1997). They deferred rigorous analysis of these mechanisms to the PCB fate and transport
modeling phases of the project (USEPA, 1997).
The apparent impact of recent plant site loadings on PCB dynamics in the river, and the
uncertainties expressed by the USEPA over mechanisms controlling such dynamics, underscores the
need to develop a quantitative understanding of PCB origin, fate, and transport in the Hudson River
system. It is only after such understanding is achieved that a technically defensible analysis of
remedial alternatives for PCBs in Hudson River sediments can be developed. Recognizing this need,
GE has conducted an extensive field research program and data analysis effort to identify and
quantify the principal sources of PCBs in the system and the mechanisms controlling PCB fate and
transport. Of particular concern was the anomalous PCB loading observed from the TIP.
QEA
12
March 19,1998
-------
TIP Sediment PCB Sources
SECTION 3
QUANTITATIVE MODELING OF TIP SEDIMENT-WATER INTERACTIONS
To allow objective, quantitative evaluation of potential remedial measures in the Hudson
River, GE has sponsored the development of state-of-the-science PCB fate, transport, and
bioaccumulation models. This section describes the developmental state of these models, how well
the model comports to existing data , and how model applications aided in the identification of the
source of the PCB loading referred to as the TIP Anomaly.
3.1 Modeling Framework
A series of models have been developed to forecast changes in water column, sediment, and
biota PCB levels in the upper Hudson River. Given initial sediment PCB concentrations and a time
series of daily flows, total suspended solid (TSS), and PCB concentrations in the river at Fort
Edward and each of the major tributaries, these models predict a time series of PCB concentrations
in the water column, sediment, and biota. Four models are used: hydrodynamic, sediment transport,
PCB fate, and PCB bioaccumulation (Figure 3-1). Sediment-water interactions of the TIP are driven
by processes described in the hydrodynamic, sediment transport, and PCB fate models. Therefore,
these models are the principal focus of this modeling discussion.
Hydrodynamics refers to the movement of water through the river and the friction or shear
stress that this movement causes at the interface between the water and the sediment bed. A
hydrodynamic model computes the velocity and depth of the river, as well as the shear stress at the
sediment-water interface, in response to upstream flows and flows entering the river from tributaries.
Sediment transport includes the movement of suspended and settled solids within the river and the
settling and resuspension of solids that occurs at the sediment-water interface as a result of the shear
caused by the moving water. A sediment transport model computes the concentration of solids in
QEA
13
March 19,1998
-------
TIP Sediment PCB Sources
the water column and the rate at which sediment accumulates in the bed. PCB fate includes the
transport of PCBs dissolved in the water or sorbed to solids, transfer between the dissolved and
sorbed phases, transfer between the water and atmosphere, and degradation that occurs chemically
or biochemically. A PCB fate model computes the concentrations of PCBs in the water column and
sediment in general accordance with the equations presented in Appendix A.
The models are equations developed from the basic principles of conservation of mass,
energy and momentum from laboratory and field studies of individual phenomena (§ A.l). The
equations are general and can be applied to any river system. The application of the equations to a
specific system such as the upper Hudson River involves the determination of appropriate values for
each of the parameters in the equations. Site-specific data are the basis for assigning values, either
directly or by the process of model calibration. Each of the models was calibrated and validated
using a data record that extends from 1976 to the present. The extensive database that is available
makes the Hudson River uniquely suited for the application of these models.
Two hydrodynamic models have been developed and calibrated and validated in order to
provide the necessary hydrodynamic input for the sediment transport and PCB fate models. A two-
dimensional, vertically-integrated hydrodynamic model is needed to define the distribution of shear
stresses at the sediment-water interface that controls sediment transport. By contrast, a one-
dimensional hydrodynamic model is sufficient to define the average transport of PCBs within the
water column. The one-dimensional model is also more computationally efficient and was therefore
used to drive long-term PCB fate simulations.
The sediment transport model uses the results of laboratory and field studies to describe the
resuspension and deposition processes of cohesive and non-cohesive sediments. The model
described here does not consider the resuspension of non-cohesive sediments. This process is
QEA
14
March 19, 1998
-------
TIP Sediment PCB Sources
included in ongoing modeling work. Results of the sediment transport model in the form of
resuspension and deposition fluxes are used directly by the PCB fate model.
The PCB fate model includes mechanistic descriptions of the transport, transfer and reaction
processes occurring in the river as described in § A.l and presented in Figure 3-1. PCBs are
assumed to partition between dissolved and particulate phases, with partitioning assumed to be rapid,
such that equilibrium conditions are generally well approximated. The dissolved phase is composed
of freely dissolved PCBs and PCBs sorbed to dissolved and colloidal organic matter. Freely
dissolved PCBs are transferred from the water column to the atmosphere by volatilization across the
air-water interface. Particulate-phase PCBs settle from the water column to the sediment bed, and
are resuspended from the sediment bed into the water column. Dissolved PCBs are exchanged
between the water column and sediment bed in accordance with the laws of diffusion, that is, from
a region of higher concentration to one of lesser concentration, with the rate of transfer controlled
by a mass transfer coefficient.
3.2 Model Calibration
3.2.1 Hydrodynamics
Applying the one- and two-dimensional hydrodynamic models to the upper Hudson River
requires that the river be divided into discrete segments or grid elements. The eight dams on the
nver make it necessary to construct a separate grid system for each reach. The eight distinct
hydrodynamic models, one for each reach, are linked together by running the system from Reach 8
(TIP) downstream to Reach 1. The downstream output of one reach provides the inlet boundary
condition information for the adjacent downstream reach. The two-dimensional grid for the TIP is
shown in Figure 3-2. While the two-dimensional model utilizes a variable, curvilinear grid, the one-
QEA
15
March 19,1998
-------
TIP Sediment PCB Sources
dimensional model uses a constant grid spacing of 762 m (2,500 ft). Both models extend from
Rogers Island at Fort Edward to the Troy Dam.
The hydrodynamic models were calibrated and validated using two sets of data. The first
data set consists of water surface elevations measured at two locations in Reaches 1 to 7 and three
locations in Reach 8 on November 28 and 29, 1990 (O'Brien & Gere, 1991). The mean flow rate
at Fort Edward during this period was 7,860 cfs, with a maximum variation of less than 2 percent.
One measurement was taken at the dam and the other was measured at an upstream location. Model
calibration in each reach was conducted by fixing the dam stage height to the measured value and
then adjusting model parameters until good agreement was achieved between the predicted and
measured stage heights at the upstream location. In the one-dimensional model, Manning's
coefficient (n) was the adjustable parameter; in the initial two-dimensional model the horizontal eddy
viscosity (AH) was the calibration variable and bottom friction coefficient (cf) was assumed to have
a constant value of 0.0025 in all reaches. The results of the calibration exercise demonstrated that,
for a given flow rate, water surface elevation can be predicted with average errors of 8 and 1 percent
for the one- and two-dimensional models, respectively. The two-dimensional model has been
recalibrated using a variable friction coefficient related to sediment type.
Both models were validated using a second set of data consisting of stage height
measurements collected in the TIP during the May 1983 flood. This flood had a peak flow at Fort
Edward of 34,100 cfs, which corresponds to a recurrence interval of approximately 10 years. The
stage heights were measured by NYSDOT personnel at staff gages 118 and 119 on the Hudson
River/Champlain Canal. These staff gages approximately correspond to river stage heights at river
miles 190.0 and 193.7. Values of the calibration parameters ( i.e., n and AH) were not changed
during the model validation, the results of which are shown on Figures 3-3 and 3-4 for the one-
dimensional and two-dimensional models, respectively.
QEA
16
March 19,1998
-------
TIP Sediment PCB Sources
3.2.2 Sediment transport
The sediment transport model used the same grid as the two-dimensional hydrodynamic
model to describe the upper Hudson River. The particle size distribution of suspended solids was
approximated as two particle size classes in the model. Class 1 represents cohesive sediments (i.e.,
clays and silts with particle diameters of less than 62 urn) while Class 2 is composed of coarser, non-
cohesive sediments, primarily fine sands with diameters between 62 and 250 (im. The deposition
rate of the Class 1 particles was a function of shear stress and particle concentration. The deposition
rate of the Class 2 particles was the product of particle concentration and an assumed settling
velocity. Erosion potential parameters were determined from upper Hudson River data on the
relationship between mass of resuspended sediment per unit of surface area and applied shear stress
(HydroQual, 1995b).
The sediment transport model was calibrated using suspended solids data from the April
1982 flood. This flood had a peak flow rate of 27,700 cfs at Fort Edward, which corresponds to a
return period of three to four years. The settling velocity of Class 2 sediments was set at 24 mm s'1,
which corresponds to a particle diameter of 200 jim, and the tributary sediment loads were assumed
to be composed of 35 percent Class 1 and 65 percent Class 2 sediments. Comparisons of predicted
and observed TSS at Schuylerville, Stillwater, and Waterford for the April 1982 flood are presented
in Figure 3-5. Predicted TSS concentrations at Schuylerville and Stillwater are in close agreement
with measured values. However, the model under predicts TSS concentrations at Waterford during
the peak of the flood. This under prediction is likely due to an underestimation of solids loading
from the Hoosic River. More recent calibrations of the sediment transport model using new tributary
solids loading data confirm this assessment of the preliminary model calibration presented in Figure
3-5.
QEA
17
March 19,1998
-------
TIP Sediment PCB Sources
The calibrated sediment transport model was used to generate a relationship between the
mass of sediment resuspended and flow rate for each of the eight reaches from Fort Edward to the
Troy Dam. These relationships were then used in the PCB fate model to determine erosion rate in
each model segment for a specified flow rate. In a similar manner, a relationship between the
effective settling velocity and flow rate was developed from results of the sediment transport model.
3.2.3 PCB fate
The one-dimensional hydrodynamic model grid was used to model PCB fate. Daily values
for river flow and water depth for the period from 1977 to 1996 were obtained from the
hydrodynamic model. Rates of resuspension and deposition were obtained from the sediment
transport model.
The sorption partition coefficient was determined from an analysis of dissolved and
particulate PCB measurements taken by the USEPA as part of the Phase 2 field data collection
program (USEPA, 1997). A 20°C value of 40,000 L kg'1 dry sediment was used in the model. This
value corresponds to an organic carbon normalized partition coefficient (K^.) of 105 4 L kg"1 organic
carbon.
Dissolved organic carbon in sediment pore water was included as a competitive sorptive
phase. The partition coefficient for DOC was fixed at 10 percent of K^., based on an analysis of
1991 field data (O'Brien & Gere, 1993b).
The volatilization rate constant was calculated from two film theory using a Henry's Law
constant of 3x104 atm-m3 mola liquid film mass transfer coefficient calculated using the
QEA
18
March 19,1998
-------
TIP Sediment PCB Sources
O'Connor-Dobbins reaeration equation and a gas film mass transfer coefficient fixed at 100 m day 1
(Equation A-8).
The vertical diffusion of PCBs between the pore waters of adjacent sediment segments was
modeled using a diffusion coefficient of 1 cm2 day'1. A temperature dependent mass transfer
coefficient with a value at 20°C of 2 cm day1 was used to model the exchange of PCBs between the
pore water and the water column (Equation A-15).
Two external sources of PCBs were considered in the model. First, PCBs entering from
upstream prior to 1991 were estimated from a correlation of PCB concentration with flow at Fort
Edward based upon USGS data. Daily flows assigned at the upstream boundary were used to
evaluate the associated PCB concentration, except on days when data were available; then the actual
measured values were used. The correlation was modified over time to reflect the decrease in
upstream PCB levels. From 1991 through 1996, the monitoring data at Fort Edward were used
directly to define the upstream boundary concentration (Figure 2-2).
The second external PCB source was an empirically defined, exponentially decreasing load
that was added to the TIP in the period between 1977 and 1983. The source of these PCBs has not
been determined, but may have been related to leaching from dredge spoils deposited along the
shoreline or a consequence of dredging activities.
Model calibration results for the March 1977 through September 1996 period are shown in
Figures 3-6 through 3-9. Figure 3-6 compares temporal profiles of calculated and observed surface
sediment (0-5 cm) PCB concentrations in TIP and downstream in the vicinity of Schuylerville,
Stillwater, and Waterford. The declines in concentration between 1977 and 1991 that are predicted
by the model are in general agreement with the observed declines. The model predicts a slightly
QEA
19
March 19, 1998
-------
TIP Sediment PCB Sources
greater decline in the TIP than the data (68% versus 60%). The model also underestimates the
average concentration measured in 1984. This underestimation may be due to a sampling bias
because the 1984 sampling program targeted areas of higher concentration. The methods used to
average the data are currently under review to determine if alternate averaging methods should be
employed.
Figure 3-7 compares the annual PCB load passing Waterford that has been estimated from
the USGS PCB data with that computed by the model. The model picks up the overall trend in the
data, as well as the year-to-year variations due to variations in river flow and associated
resuspension.
Since the calibration of this model, an analytical bias has been identified in the water column
PCB data appearing in Figure 3-7 (Tetra-Tech, 1997; HydroQual, 1998). This bias is associated with
the analytical methods employed by the USGS. A preliminary analysis of the survey's laboratory
records suggests that the historical water column data are biased low as the technique does not
account for the entire compliment of mono- and dichlorinated PCBs within the samples (HydroQual,
1998). Preliminary estimates suggest the bias ranges from 10-40 percent and depends on the
relative proportion of mono- and dichlorinated PCBs in the samples. Since, mono- and dichlorinated
PCBs account for a significant portion of the total water column PCB loading occurring across the
TIP, additional efforts are underway to more fully characterize this bias, and possibly correct a
portion of the USGS database. Considering these limitations of the data, the absolute water column
concentrations predicted by the model are of less importance than the model predicted change in
water column levels over the 15 year monitoring period. The model accurately predicts the factor
of five decline in measured water column PCB levels between 1976 and 1991.
QEA
20
March 19,1998
-------
TIP Sediment PCB Sources
From 1991 to 1996, the data for calibration are largely restricted to the results of weekly
monitoring of the water column at TID; although limited data are available from Schuylerville5. The
comparison of the model to these data is less favorable than to the historical USGS water column
data, even considering the bias in the USGS data. Figure 3-8 compares computed and observed
water column PCB levels at Schuylerville for the period from 1989 through 1991. The model and
data closely correspond in 1989, but the model underpredicts the observed levels in 1991. The
comparison at TED for 1993 through 1996 demonstrates a consistent low bias by the model (Figure
3-9). The computed concentrations at TID are 300 to 500 percent lower than those measured. In
contrast, the preliminary analysis of the bias in the USGS data appears to be less than 50 percent.
Therefore, it is unlikely that the differences observed between model predictions and monitoring data
can be solely attributed to the bias in the USGS data.
This difference between model projections and observed data was unexpected given the
favorable comparison of model predictions to the data from 1977 to 1991, even considering the
potential bias in the USGS data. Efforts to alter the model calibration to achieve water column levels
consistent with the TID data were unsuccessful. No combination of reasonable rates of sediment-
water interaction were able to reproduce both the long-term trends in sediment PCB levels and the
TID water column PCB levels.
3.2.4 Summary of preliminary PCB fate and transport modeling
State-of-the-science hydrodynamic, sediment transport, and chemical fate models were
developed to describe the PCB dynamics within the Hudson River system and to provide a means
5 Limited additional data collected by GE in 1991 and 1992 and the USEPA in 1993 are available for
stations located downstream of the TID.
QEA
21
March 19,1998
-------
TIP Sediment PCB Sources
of predicting PCB concentrations in the different media into the future. The favorable comparison
of the model predictions to observed field data between the late 1970s and 1991 indicated that the
models provided a consistent and accurate assessment of the mechanisms controlling PCB fate in
the system over this period. The degradation of the model calibration to water column data collected
after 1991 suggested that the models did not accurately account for the varied sources and processes
affecting PCB dynamics within the TIP region of the river. Another PCB source(s), loading
mechanism(s), or data inadequacy(s), not accurately represented by the models, was controlling PCB
loading observed from the TIP region of the river.
A number of hypotheses were developed to explain these observations and were tested
through a rigorous analysis of existing field data and the development and execution of a field
research program (HydroQual, 1996, 1997a, 1997b, O'Brien & Gere, 1995, 1997b). These efforts
are describe in Section 4.
QEA
22
March 19,1998
-------
TIP Sediment PCB Sources
SECTION 4
EVALUATION OF ALTERNATIVE HYPOTHESES
FOR ANOMALOUS PCB LOADING WITHIN TIP
During 1996 and 1997, GE conducted an extensive field research program and data analysis
effort to evaluate different hypotheses for the anomalous PCB loading observed within the TIP. As
described in Section 3, known and understood PCB fate and transport mechanisms could not account
for the entire loading observed from the TIP region of the river. An alternative PCB source, loading
mechanism, or data inadequacy was required to account for this anomalous loading. The hypotheses
considered to explain the loading anomaly fell into three general categories:
• additional mechanism of PCB exchange between sediments and water column,
• additional PCB sources, and
• erroneous estimates of PCB flux due to biased sampling.
The USEPA advanced the hypothesis of alternative mechanisms for PCB exchange between
sediments and the water column in their Data Evaluation and Interpretation Report (DEIR; USEPA,
1997) and Preliminary Model Calibration Report (PMCR; USEPA, 1996) developed as part of their
ongoing RRI/FS. In the DEER, the USEPA hypothesized that either groundwater induced advective
flux or resuspension of dechlorinated sediments in addition to diffusive flux mechanisms may
account for the loading observed at TID. During preliminary model calibration, the USEPA invoked
all of these mechanisms to transport PCBs from surface sediments into the overlying water column
to account for the TID loading. Data analysis conducted by GE and documented in formal
comments on these reports does not support these mechanisms as possible explanations for the
observed anomalous loading (GE, 1997). GE undertook a field research program designed to
evaluate the plausibility that these mechanisms can contribute significantly to the observed loading
from the TIP.
QEA
23
March 19,1998
-------
TIP Sediment PCB Sources
The hypothesis that additional PCB sources may have been introduced into the TIP and were
responsible for the anomalous loading was considered in light of recent PCB DNAPL loadings to
the river. DNAPL PCBs within fractured bedrock underlying the GE Hudson Falls Plant site
(O'Brien & Gere, 1996a) is believed to have migrated through bed rock fractures and accumulated
in waterways within the 150 year old Allen Mill (O'Brien & Gere, 1994a). Collapse of a wooden
gate structure within the mill is believed to have resulted in the transport of PCB DNAPL into the
Hudson River during September 1991 and until flow through the waterways was controlled in
January 1993 (O'Brien & Gere, 1994a). Although these sources were controlled by remedial
measures (O'Brien & Gere, 1996a), PCB DNAPL from the plant site continued to enter the river
directly through fractures in the river bed until remedial measures on the plant site mitigated these
sources. The temporal correspondence of the mill loadings and the increase in PCB loadings from
the TIP suggested the mill loadings as the causative factor. For this hypothesis to be true, PCBs
must have passed the Fort Edward sampling station (Figure 2-1) undetected and then been deposited
within the pool. This could occur if PCBs enter the river between sampling events or are transported
as part of the bed load passing under sampling devices. PCB DNAPL transport was evaluated in
a field research program sponsored by GE (HydroQual, 1997c).
The hypothesis that biased sampling may have resulted in erroneous estimates of PCB flux
into or out of the TIP was considered as a possible cause of the TIP anomaly. For this hypothesis
to be true, PCBs must have either; 1) entered the headwaters of the TIP undetected by the routine
water column monitoring program and, following transport through the TIP, been detected within
samples collected at the sampling station downstream of the TEP at TID, or 2) been
unrepresentatively elevated within samples collected from the shore-based station located at TID.
Data from limited sampling conducted during the early 1990s from the eastern and western shore
areas at TID were in general agreement, supporting the representativeness of the western shore-based
sampling location (O'Brien and Gere, 1996c). This hypothesis was further tested during extensive
QEA
24
March 19,1998
-------
TIP Sediment PCB Sources
field efforts conducted in 1996 and 1997 and appears to be the principal cause of the anomalous
loadings.
The results of specific field research programs and data analysis efforts evaluating these
different hypotheses for the observed anomaly are presented below.
4.1 Additional Mechanism of PCB Exchange Between Sediments and Water Column
The hypothesis that additional mechanisms of PCB exchange between the sediments and the
overlying water column were responsible for the anomalous loading was evaluated through an
intensive data evaluation effort as well as field research. The effect of long-term elevated PCB flux
from the sediments either as a result of surface sediment erosion or ground water advection, was
assessed within a mass balance framework. The results of these analyses were presented in
comments to the USEPA on their DEER (GE, 1997). Additionally, in-field groundwater advection
measurements were made and the resulting groundwater velocities were compared to those required
to sustain the anomalous loading as presented in the USEPA PMCR (1996). Moreover, low flow
water column TSS measurements were made through the TIP to assess the possibility that sediment
resuspension may be contributing to the observed loading under low flow conditions. Finally,
quantitative sediment transport modeling, based on state-of-the-science cohesive sediment transport
theory, was conducted to estimate low-flow sediment resuspension and test the plausibility that this
mechanism is contributing to the TIP anomaly.
QEA
25
March 19,1998
-------
TIP Sediment PCB Sources
4.1.1 Effect of long-term high flux on sediment PCB inventory
A mass balance calculation was performed to test the hypothesis that the anomalous PCB
loading could be attributed to surface sediment PCB transport processes, either surface sediment
resuspension or ground water advection. In this calculation, the net increase in PCBs between
Rogers Island and the TID was assumed to originate from PCBs in the surface sediments of the TIP,
defined conservatively as 0-8 cm. No vertical mixing was assumed between surface sediments and
deeper sediments. The inventory or mass of PCB homo logs within the surface sediments (MJSS) was
estimated using the results of USEPA's reanalysis of the 1984 sediment data (USEPA, 1997) as
follows:
M = C o z A C4_n
J. SS J,SS ss ss tip \ /
where:
Cj „ is the average concentration of PCB homolog j within the surficial sediments (0-8
cm) as calculated from 1984 data (M M"1),
pss is the density of surface sediments (M L"3),
zss is the depth of the surface layer (L), and
Anp is the surface area of the TIP (L2).
The surface sediment area of the TIP (2.0x106 m2) and the density of surface sediments (0.77 g cm'3)
were developed from information provided by the USEPA (USEPA, 1997). Annual total loadings
of PCB homologs (j) across the TIP (Wtipj) were calculated using measured annual paired loadings
from Rogers Island and the TID from 1993 to 1996 (O'Brien & Gere, 1994b, 1995, 1996b, 1997c)
in accordance with the following expression:
QEA
26
March 19, 1998
-------
TIP Sediment PCB Sources
I Q
W
/--i
/«.
(C . -c )
v lid, i, / n, i. y'
"P. 1
x 365
(4-2)
where:
Qfej is the daily average flow at the USGS Fort Edward gauging station for day i (L3 T"1),
Ctjig is the concentration of PCB homolog j on day i at the TID station (M L'3),
Cn ij is the concentration of PCB homolog j on day i at the Rogers Island station (M L'3),
and
n is the number of paired samples collected at the Rogers Island and TID stations for
the year in question.
The calculated average annual total PCB loadings across the TIP, calculated as the sum of homolog
loadings, are presented in Table 4-1. The average annual load ranged from a low of 84 kg yr 1 in
1995 to a high of 407 kg yr 1 in 1996. The four year average loading is 248 kg yr'1.
Table 4-1. Average Annual Total PCB Loading Across TIP from 1993 to 1996.
Year
No. of Paired Samples
Average PCB Load
(kg yr'1)
1993
49
202
1994
34
297
1995
45
84
1996
57
407
Average
46
248
QEA
27
March 19,1998
-------
TIP Sediment PCB Sources
The depletion of 1984 surface sediment PCBs was estimated simply by dividing the
estimated 1984 surface sediment mass of PCB homologs by the annual flux rate as calculated above
using paired Rogers Island and TID data and projecting into the future. The year in which the
surface sediments would become depleted of homolog j (Yrj) was calculated as follows:
M
Yr = -Aii + 1984 (4-3)
j W v '
i. "p
The results are presented in Table 4-2 below6.
6This calculation is conservative since historical flux rates were likely greater than those measured in the
1990s because the higher surface sediment PCB concentrations in the 1980s would have resulted in higher flux rates
than those observed in the 1990s.
March 19,1998
QEA
28
-------
TIP Sediment PCB Sources
Table 4-2. Surface Sediment PCB Inventory Depletion Under Average 1993-1996 TIP PCB
loadings.
Homolog
Mass of PCBs in TIP
Surface Sediments in 19841
(MT)
Load from
TIP
(MT yr1)
Year in Which
Surface Sediment
Reservoir Depleted
Mono
0.58
0.055
1995
Di
1.40
0.117
1996
Tri
1.00
0.062
2000
Tetra
0.41
0.016
2009
Penta
0.13
0.002
2040
Total
3.52
0.25
-
1) The mass of PCB homologs was calculated by multiplying the average PCB homolog distribution of the
1994 low resolution cores (USEPA, 1995) and the estimates of TIP PCB mass obtained by statistical analysis
of the 1984 NYSDEC data (USEPA. 1997).
This mass balance calculation indicates that, if surface sediments of the TIP were the sole
source of PCBs contributing to the apparent loading increase observed over the TIP, then the mono,
di, and tri homologs present within the surficial sediments in 1984 would be entirely depleted by the
year 2000. This is particularly significant since the current water column measurements show a
continuing source of mono- and dichlorinated PCBs from the TIP. Moreover, sediment sampling
by both GE in 1991 (O'Brien & Gere, 1993b) and the USEPA in 1992 (USEPA, 1997) and 1994
(USEPA, 1995) indicate that significant reserves of PCB remain within the surface sediments of the
TIP. Therefore, on a mass balance basis, the PCB loadings observed from the TIP between 1993 and
1996 cannot be representative of long-term surface sediment-water exchange processes. Another
source of PCBs, possibly related to upstream sources, or data inadequacies as discussed below must
be contributing to the observed loading from the TIP in the 1990s.
QEA
29
March 19,1998
-------
TIP Sediment PCB Sources
4.1.2 Measurement of ground water seepage rates
Direct field measurements of ground water seepage rates provided data with which to
evaluate the hypothesis that ground water advection may be responsible for the anomalous PCB load
detected in the TIP. This effort was prompted by the USEPA's invocation of a groundwater
mechanism in the PMCR to account for the anomalous PCB loadings (USEPA, 1996). The
mechanism of groundwater advection of PCBs from the sediments to the water column is described
in detail in Appendix A, and a complete description of the groundwater investigation is documented
elsewhere (HSI GeoTrans, 1997).
Direct measurement of groundwater seepage has been widely employed as a means of
assessing the hydraulic and chemical interactions between groundwater and surface water, and to
examine spatial and temporal patterns of groundwater seepage (Lee, 1977; Lee and Cherry, 1978;
and Woessner and Sullivan, 1984; Gallagher et al., 1996). The seepage meters employed to monitor
groundwater seepage within the Hudson River were modeled after the original design by Lee (1977),
with modifications to reduce the potential for measurement biases that have been documented in the
literature (e.g., Belanger and Montgomery, 1992, Shaw and Prepas, 1989). Seepage meters consisted
of a cylindrical stainless steel vessel equipped with two Va inch Teflon bulkhead fittings, a Teflon
air sampling bag equipped with a release valve, and % inch Teflon tubing (HydroQual, 1997b;
Figure 4-1). Two seepage meters were deployed at each of the six locations depicted in Figure 4-2.
Measurements were taken at multiple sites within the TIP and one downstream site to allow
delineation of spatial trends in groundwater seepage rates. Multiple seepage rate measurements
were conducted over approximately a one month period between late May and late June 1997. This
period was immediately after the annual spring high flows and snow melt, when hydraulic gradients
between the groundwater system and the river were expected to be at their greatest.
QEA
30
March 19,1998
-------
TIP Sediment PCB Sources
The seepage meter study produced pronounced temporal and spatial patterns in groundwater
seepage (HSI GeoTrans , 1997). Average seepage rates declined over the monitoring period from
a high of 0.18 L m2 hr1 in late May to a low of -0.03 L m2 hr 1 in mid June (Figure 4-3). A
decreasing temporal trend occurred in measurements collected within the headwaters of the HP at
Site SI (Figure 4-2). This observation is consistent with the reduction in hydraulic gradient observed
in piezometers installed adjacent to the seepage meters (HSI GeoTrans, 1997) and that expected in
response to seasonal changes in surface water and groundwater elevations.
Within the TIP, ground water seepage increased with distance upstream of the TID (Figure
4-4). This is expected since the hydraulic gradients near the TED would be affected by the artificial
increase in surface water elevation produced by the dam. Seepage measurements were generally
positive (flux of groundwater into the Hudson River) at sites SI through S3 (Figure 4-4) located 3-5'
miles upstream of the TID (Figure 4-2). In contrast, groundwater flow was consistently negative at
site S5 (Figure 4-4) located just one mile upstream of the TID (Figure 4-2).
The groundwater seepage investigation produced temporal and spatial patterns in
groundwater seepage that were consistent with both independent measurements of hydraulic
gradients between the surface water and groundwater systems and our understanding of the Hudson
River system. Piezometers installed adjacent to the seepage meters generally yielded hydraulic
gradients indicative of water movement in the same direction measured within the seepage meters
(HSI GeoTrans, 1997). Moreover, spatial and temporal patterns in groundwater seepage were
consistent with those expected in response to both seasonal changes in surface water and
groundwater elevation and the artificially elevated surface water condition at the downstream limit
of the TIP as a consequence of the TED. However, the ground water seepage rates were, on average,
approximately an order of magnitude lower than the value assumed during preliminary calibration
of the USEPA PCB fate and transport model (1.3 L m : hr USEPA, 1996; Figures 4-3 and 4-4).
QEA
31
March 19,1998
-------
TIP Sediment PCB Sources
Ground water induced PCB flux from the sediments to the water column of the TIP were
estimated as the product of groundwater seepage flow developed from the ground water seepage
measurements and estimates of sediment pore water PCB concentrations in accordance with
Equation A-19. The total groundwater flow was estimated as the product of the average volumetric
seepage flux and the total area of the TIP. The mean surficial sediment pore water PCB
concentration was calculated from the 0-5 cm section of sediment cores collected in 1991 based upon
equilibrium partitioning concepts described in Equations A-13 and A-14. The organic carbon-based
PCB partition coefficient was calculated using USEPA water column partitioning data (USEPA,
1997) and corrected for temperature using temperature correction functions (GE, 1997). The pore
water dissolved organic carbon concentration was calculated as the mean surficial sediment (0-5 cm)
TIP dissolved organic carbon measurements from the 1991 sediment survey (O'Brien & Gere,
1993b), and the equilibrium constant describing partitioning between freely dissolved PCBs and
PCBs adsorbed to dissolved organic carbon was assumed equal to 0.1 K^.
Applying the parameter values in Table 4-3 to Equations A-13, A-14 and A-19 yielded
groundwater induced PCB flux measurements of approximately 30 g day'1. Assuming these seepage
measurements represent an average seepage flux for the entire year, groundwater induced PCB
loading contributes an estimated 11 kg yr'1 of PCBs to the water column. This represents
approximately 4% of the average PCB loading observed from the TIP between 1993 and 1996 (Table
4-2). These estimates are conservatively high due to the assumption that the spring 1997
measurements are representative of groundwater flux for the entire year, even though they were
collected during a period in which the hydraulic gradient between surface water and ground water
was expected to be at its greatest. Based on these measurements and calculations, groundwater
induced PCB flux cannot account for the anomalous loading observed from the TIP.
QEA
32
March 19,1998
-------
TIP Sediment PCB Sources
Table 4-3. Parameters Used to Calculate Groundwater Induced Advection of PCBs from
Surface Sediments to the Water Column.
Parameter
Description
Value (units)
organic carbon-based PCB
partition coefficient
1054 (L kg"1)
K
-------
TIP Sediment PCB Sources
weighted average of the 0-5 cm sections collected in 1991 (21.8 mg/kg; O'Brien & Gere,
1993b).
Additionally, field measurements of TSS through the TIP during the elevated loading period were
collected and analyzed to test the predictions of the sediment transport model.
At flow rates less than 10,000 cfs, sediment, and consequently PCB, resuspension is minimal
(Table 4-4 and Figure 4-5). At the average annual river flow rate of approximately 5,000 cfs at Fort
Edward, the estimated mass of sediment erosion is approximately 6 kg7. Using the average 0-5 cm
PCB data collected in 1991 (O'Brien & Gere, 1993b), this corresponds to an estimated PCB erosion
of only 0.12 grams. It is only after river flow rates approach 10,000 cfs that sediment bed erosion
significantly contributes to water column PCB loading. This is consistent with our understanding
of sediment erosion processes, which predict no resuspension at bottom shear stresses less than the
critical shear stress as described in § A.2.2.
The critical shear stress established for the cohesive sediments of the TIP is 1 dyne cm2
(HydroQual, 1995b). The lack of significant bed erosion at flows less than 10,000 cfs indicates that
there are only limited areas within the TIP where shear stresses exceed 1 dyne cm"2 at these flows.
This is reflected in the data presented in Table 4-4. At 5000 cfs, less than 0.5% of the cohesive
sediment bed area within the TIP is subject to sheer stresses greater than 1 dyne cm'2. This increases
to approximately 1% at flows of 10,000 to 20,000 cfs (Table 4-4). Moreover, negligible
resuspension from the non-cohesive sediment bed occurs at flow rates below 10,000 cfs; the non-
cohesive bed is not mobilized and fine sands cannot be resuspended because of bed armoring effects
caused by coarse sands and gravels.
7Sediment bed erosion is not represented as a rate (MT1) since erosion occurs instantaneously (see
Appendix A).
QEA
34
March 19,1998
-------
TIP Sediment PCB Sources
Table 4-4. Estimates of TIP Sediment and PCB Erosion as a Function of River Flow.
River Flow1
(cfs)
Cohesive Bed
Eroded2
(%)
Mass of
Sediment
Eroded3
(kg)
Mass of PCBs
Eroded4
(g)
2500
.05
1.09e-03
2.37e-05
5000
.32
5.92e+00
1.29e-01
6000
.75
1.64e+02
3.57e+00
7000
.80
l.78e+02
3.88e+00
8000
.84
3.25e+02
7.08e+00
9000
.87
4.92e+02
1.07e+01
10000
.89
1.32e+03
2.88e+01
15000
.95
1.32e+04
2.88e+02
20000
.96
5.61e+04
l.22e+03
1) Flow at the headwaters of the TIP at Fort Edward, N.Y.
2) Percent of TIP sediment surface area subject to erosion under different river flows as calculated using the
hydrodynamic and sediment transport model descnbed in Section 3.
3) Mass of sediment eroded under the different flow conditions as calculated using the hydrodynamic and
sediment transport model descnbed in Section 3.
4) Estimates of PCB erosion calculated by multiplying the mass of sediment eroded by the 1991 area-weighted
mean surface sediment (0-5 cm) PCB concentration within the TIP (21.8 mg/kg; O'Brien & Gere, 1993b).
The lack of significant sediment bed erosion at low to moderate river flows was also
observed in field studies conducted on the river in 1996 and 1997 (O'Brien & Gere, 1998). Time
of travel surveys, consisting of sampling at stations located along lateral transects established every
0.25 to 0.5 miles between Rogers Island and TED (Figure 4-6), yielded TSS concentrations that were
QEA
35
March 19,1998
-------
TIP Sediment PCB Sources
generally less than 5 mg L'1 and did not produce patterns indicative of sediment bed erosion (Figure
4-7). During these studies, water column samples were collected from upstream to downstream so
as to correspond with the flow of river water as it traversed the TIP and should have detected regions
of the river subject to erosive conditions at the sampled flows.
The hypothesis that low flow sediment resuspension is contributing to the TIP anomaly can
not be supported by sediment fate and transport theory or field data. The application of state-of-the-
science hydrodynamic and sediment transport models predicts insignificant sediment bed erosion
under the low to moderate flow conditions under which the TIP anomaly has been observed. Field
measurements of TSS support these predictions. Therefore, the USEPA hypothesized mechanism
of low-flow sediment resuspension cannot explain the TIP anomaly.
4.2 Additional PCB Sources
The second general hypothesis to explain the anomalous TIP PCB loading considers the
possibility that additional PCB DNAPL loadings from the Allen Mill and Hudson Falls Plant site
are entering or have entered the TIP without being detected at the Rogers Island sampling stations.
Potential DNAPL loading mechanisms include:
• preferential transport of PCB laden sediments and PCB DNAPL along the sediment-water
interface,
• pulse loading of PCBs associated with the periodic flooding of the Bakers Falls plunge pool
as a result of the operation of the Adirondack Hydro Development Corporation's (AHDC)
turbines or during elevated flow events, and
• transport of oil-soaked sediment into the TIP at the time of the Allen Mill collapse.
QEA
36
March 19, 1998
-------
TIP Sediment PCB Sources
These hypothesized PCB sources were the subject of an extensive field research program sponsored
by GE in 1996 and 1997.
4.2.1 Simulation of PCB oil transport
The hypothesis that PCB DNAPL loadings may be transported from the plant site areas into
the TIP was examined in a field research program that simulated the fate of PCB DNAPL in the
river. The program included the direct discharge of a conservative tracer with properties similar to
PCB DNAPL into the river near Hudson Falls and tracking of the tracer downstream. The details
of this study have been documented elsewhere (HydroQual, 1997c). In summary, the study
included:
• injection of 20 pounds of fluorescent particles (Figure 4-8) with a density similar to that of
Aroclor 1242 into the river from the AHDC Hydroelectric Plant,
• collection of daily composites of water column and bed load particle samples in specially
designed sampling devices (Figure 4-9) at or near routine water column monitoring stations
for three days following fluorescent particle injection,
• analysis of water column and bed load particle samples for fluorescent resin particle
concentration,
• calculation of the total mass of fluorescent particles passing each station over the three day
period by scaling up the mass of particles trapped within the sampling devices to reflect the
entire river cross section, and
• development of fluorescent particle mass balances to evaluate particle transport and, by
inference, the transport of PCB DNAPL w ithin the Hudson River.
The results of the three day fluorescent particle mass balance appear in Figure 4-10. Of the
9.1 kg of particles injected into the river near the Hudson Falls Plant site (RM 196.9), an estimated
73% (6.6 kg) was transported downstream to the Fort Edward station (RM 194.4). These
QEA
37
March 19,1998
-------
TIP Sediment PCB Sources
calculations suggest that an estimated 27% (2.5 kg) of the fluorescent particle mass released into the
river was retained between the particle injection point and the Fort Edward station. This pattern of
particle retention continued downstream, as approximately 18% of that injected (1.6 kg) was retained
within the river between the Fort Edward and Rogers Island sampling stations. Over the three day
study, only an estimated 2% (0.1 kg) of the particles were transported downstream of the Thompson
Island station (Figure 4-10). These data indicate that 98% of the particles injected in the river near
the Hudson Falls plant site were retained in the river upstream of Thomson Island.
Fluorescent particles retained upstream of the Fort Edward station consisted predominantly
of the smallest particle size class (19-38 jim) and the two size classes greater than 190 fim (Figure
4-11c). This distribution was calculated as the difference between the mass of particles injected
(Figure 4-1 la) and the mass of particles passing the Fort Edward station (Figure 4-1 lb), on a size
class basis. The apparent retention of the smaller particles between the injection point and the Fort
Edward station may be the combined result of: 1) under sampling of smaller particles by the 100 u.m
mesh of the in situ filtration devices and, 2) loss of particles near the injection point. The larger
particles retained upstream of the Fort Edward station likely settled within the river near the injection
point as they were never detected downstream.
Several inferences with regard to the transport and fate of PCB DNAPL within the Hudson
River may be drawn from the fluorescent particle data. First, PCB DNAPL droplets in excess of
190 nm will likely be sequestered near the discharge point where they would be subject to
dissolution. Mobilization of these droplets downstream may be limited at flows less than the 7000
cfs observed during this study, but may occur under higher flow conditions. Such temporary storage
is demonstrated by the presence of fluorescent particles in sediment bed load samples collected
during the spring high flow event of April 1997 (HydroQual, 1997c). Second, PCB DNAPL
existing in the river over the particle size range tested (19-380 urn) would be deposited upstream of
QEA
38
March 19,1998
-------
TIP Sediment PCB Sources
the TID. That is, little, if any DNAPL would be transported downstream of the TIP. Once within
these sediments, DNAPL would be subject to other fate determining processes such as dissolution,
diffusion, advection, and partitioning onto sediment solids.
The PCB DNAPL transport study provides a unique data set from which to infer the fate of
PCB DNAPL loadings within the Hudson River system. The fluorescent particles employed during
this study possessed a density similar to that of PCB DNAPL oils found on the Hudson Falls plant
site and a particle size distribution believed to be representative of DNAPL oil droplets within the
river (HydroQual, 1997c). As such, the behavior of these particles was considered to represent PCB
DNAPL fate and transport in the system. Several conclusions regarding PCB DNAPL were drawn
from the results of this study:
• PCB DNAPL with droplet sizes greater that approximately 200 p.m entering the river under
low river flow conditions will be sequestered near the point of entry into the system,
• PCB DNAPL sequestered in the river may be mobilized during high flow events, possibly
as part of the sediment bed load, and
• PCB DNAPL transported into the TIP will be deposited within the surface sediments of the
TIP.
The results of the PCB DNAPL study generally support the hypothesis that PCB DNAPL
loadings from the Hudson Falls plant site (§1.2) may have contaminated the surface sediments of
the TIP. This may have been occurring throughout the 1980s. However, it is unclear whether this
mechanism has been contributing to the anomalous loading observed from the TIP during the 1990s.
QEA
39
March 19,1998
-------
TIP Sediment PCB Sources
4.2.2 High flow water column and sediment bed loading
The results of the DNAPL study suggest that PCBs may be transported from the vicinity of
the Hudson Falls plant site and into the TIP as a pulse loading within the water column or within the
sediment bed load during periods of high flow. To evaluate this hypothesis, high flow water column
and sediment bed load sampling was conducted on the Hudson River during the spring high flow
period of 1997 (O'Brien & Gere, 1998). The approach included:
• water column sampling and analysis for PCBs and TSS from the Fort Edward and TID
stations along the rising and falling limb of the spring high flow event hydrograph between
April 6 and 9, 1997, and
•sediment bed load sampling and analysis for PCBs within the east and west channel of
Rogers Island (Figure 2-1) during the high flow event using a specially designed bed load
sampling device (Figure 4-12).
Water column samples were collected as vertically integrated composite samples consisting of
discrete samples collected from three depths in the east and west channels of Rogers Island at Fort
Edward and as discrete grab samples collected in a stainless steel vessel at TID.
During the 1997 spring high flow period, instantaneous flows at the Fort Edward gauging
station increased from approximately 9,000 cfs on April 6 to a maximum flow rate of approximately
19,400 cfs on April 8, 1997 (Figure 4-13a). These flows produced only modest increases in TSS
levels (Figure 4-13b), as TSS concentrations never exceeded 12 mg/L, indicating that the event did
not produce bottom sheer stresses capable of causing significant sediment resuspension. Water
column total PCB concentrations also remained low during the high flow event, ranging between
10 and 30 n(Figure 4-13c). At peak flow, water column concentrat )ns represent a PCB loading
of 1.3 kg/day at the Fort Edward station (Figure 4-13d).
QEA
40
March 19,1998
-------
TIP Sediment PCB Sources
The PCB loadings observed during the 1997 high flow period represent a significant
reduction in high flow event driven PCB transport in the system compared to similar events sampled
in 1992 and 1993 immediately following external PCB loadings to the system (Figure 2-3). The
1992 and 1993 spring flood events produced maximum PCB loadings of approximately 50 lbs/day
and approached the loadings observed in the late 1970s. These observations indicate:
• high flow events were an important mechanism transporting PCBs downstream from the
plant site regions of the river and into the TIP during the early 1990s, and
• remedial measures conducted on the plant site (§1.2) appear to have mitigated PCB
discharges to the river and significantly reduced high flow PCB transport in 1997 (Figure 2-
3).
Flow event-driven transport of sediments and associated PCBs along the sediment-water
interface (sediment bed loading) does not appear to be a significant mechanism by which PCBs are
transported into the TIP. Particulate phase PCB concentrations of sediment bed load samples
collected from both the east and west channel of Rogers Island contained less than 15 mg/kg PCBs
(Figure 4-14), and two of the three samples collected contained less than 5 mg/kg PCBs. These
concentrations are significantly lower than the water column particulate phase PCB concentrations
measured at Fort Edward by the USEPA in 1993 and the average surface sediment (0-5 cm) PCB
concentrations measured in 1991. These data indicate that sediment bed loading in 1997 was not
a significant contributor to the PCB loading into the TIP.
Based upon the high flow data collected in 1997, flow event driven water column and
sediment bed load PCB transport do not appear to be significant mechanisms for continued pulse
loadings of PCB from the plant site regions and into the TIP. However, the high flow events in the
early 1990s did mobilize significant PCB loads into the system. These loadings may have
contributed to PCB DNAPL transport into the TIP as the results of the DNAPL transport study (§
QEA
41
March 19, 1998
-------
TIP Sediment PCB Sources
4.2.1) indicate that PCB oils transported downstream of the plant site would be deposited in the TIP.
This mechanism may have contributed to the elevated TIP loadings observed following the Allen
Mill loading event by elevating surface sediment PCB concentrations. Additionally, PCB loading
via this mechanism may have contributed to TIP surface sediment PCB contamination prior to the
mill event. However, the results of the 1997 high flow study indicate that remedial measures
conducted on the Hudson Falls plant site and the Allen Mill have mitigated these sources to the river
and greatly reduced the transport of PCB into the TIP. Hence, to the extent that flow event driven
pulsed loadings contributed to the TIP load in the early 1990s, their effect should be greatly
diminished in the future.
4.2.3 Pulse loadings during periodic flooding of Bakers Falls plunge pool
Pulse loadings during periodic flooding of the Bakers Falls plunge pool as a result of the
operation of the AHDC hydroelectric facility is another possible source of PCBs to the TIP. The
trash rack assemblies that protect the turbines from debris transported through the intake raceway
require cleaning every few days. During this process, flow through the facility ceases and the racks
are pneumatically cleaned of debris, which is carried by water flow through a bypass structure along
the western shore of Bakers Falls and into the plunge pool. Due to the reduced flow through the
hydroelectric facility, the surface water elevation of the pool upstream of Bakers Falls increases and
spills over the dam. This inundates the falls and provides additional waters for flushing of PCBs
downstream.
The periodic flooding of Bakers Falls was considered as a possible source of PCBs to the TIP
due to the PCB DNAPL seeps located within fractures in the bed rock outcroppings of the falls and
within the plunge pool. A specific monitoring program was designed to assess the relative
contribution of this PCB source to the TIP loading anomaly (O'Brien & Gere, 1997a).
QEA
42
March 19, 1998
-------
TIP Sediment PCB Sources
The approach used to monitor the impact of hydro facility operation on PCB transport in the
system included (O'Brien & Gere, 1997a):
• release of rhodamine WT dye into the plunge pool prior to trash rack washing activities at
the hydroelectric facility,
• monitoring of the dye-front and collection of samples representing water flushed from the
pool at three locations: the plunge pool, Fort Edward, and the TID, and
• analysis of collected samples for PCBs and TSS.
Three hydrofacility operation monitoring events were conducted; one in September 1996 and two
in June 1997.
The periodic flushing of Baker Falls appears to have a significant effect on the PCB
concentrations found within the plunge pool (Figure 4-15). During two of the three sampling events,
PCB concentrations within the plunge pool increased substantially from near the method detection
limit of 11 ng/1 before inundation of the falls to approximately 400 ng/1 (June 9, 1997) and 130 ng/l
(June 23, 1997) after falls inundation. These data suggest that PCB DNAPL that accumulates on
the bedrock outcrops of the falls, is transported into the plunge pool as water flows over the falls.
The magnitude of the release from the falls is difficult to assess from the plunge pool data due to
uncertainties over flow characteristics of the pool. Therefore, the impact of the loading from the falls
was assessed by examining the transport of these PCBs downstream at the Fort Edward station.
Periodic loading of PCBs as a result of hydroelectric facility operations had little effect on
PCB loadings into the TEP. Although water column PCB concentrations at the Fort Edward station
increased in response to the loadings from the plunge pool, these increases were relatively small
(Figure 4-15). Moreover, there was no evidence of any correlation between the PCB levels observed
QEA
43
March 19,1998
-------
TIP Sediment PCB Sources
within the plunge pool and those observed at Fort Edward. The largest increase in PCB
concentrations in the plunge pool was observed during the June 9, 1997 sampling event. In contrast,
PCB concentrations from the Fort Edward station on this date increased only slightly. Therefore,
the total mass of PCB transported downstream as a result of this loading mechanism is not sufficient
to appreciably increase PCB loading observed at the Fort Edward station.
Data collected in association with the hydrofacility operations monitoring indicate that
periodic inundation of Bakers Falls provides relatively insignificant PCB loads into the TIP. Hence,
this mechanism is not likely responsible for anomalous loadings from the TIP.
4.2.4 Localized PCB source areas within TIP
PCB DNAPL loadings from the Hudson Falls plant site area and the Allen Mill during the
early 1990s (§ 2.2) may have contaminated surface sediments within localized regions of the TIP.
This hypothesis was generally supported by the PCB DNAPL study which indicated that oil phase
PCBs entering the river within the vicinity of the Hudson Falls plant site would be transported
downstream and deposited in the TIP (§4.2.1). To assess the importance of this potential cause of
the TIP anomaly, time-of-travel surveys were conducted through the TIP. These surveys were
designed to monitor a single mass of water as it traveled through the pool. In this way, localized
areas potentially contributing a disproportionate quantity of PCBs to the water column load could
be detected.
Detailed information regarding the methods employed for the TIP time-of-travel studies is
provided elsewhere (O'Brien & Gere, 1998). In summary, the surveys consisted of sampling along
lateral transects established every 0.25 to 0.5 miles between Rogers Island and TID, with sampling
stations at three positions across each transect: east shore, west shore, and center channel (Figure 4-
QEA
44
March 19, 1998
-------
TIP Sediment PCB Sources
6). Transects were sampled from upstream to downstream so as to correspond with the flow of river
water. Stations along each transect were sampled simultaneously. Time-of-travel between each
transect was estimated from flow information retrieved from the USGS gauging station located in
Fort Edward (1996) and by monitoring a pulse of dye injected into the river (1997). A total of four
time of travel surveys were conducted: two in September 1996 and two in June 1997. Samples from
each station consisted of vertically stratified composite samples collected from three depths and were
analyzed for PCBs and TSS.
The four TIP time of travel surveys exhibited similar spatial trends in total PCB
concentration within the center channel (Figures 4-16 and 4-17). PCB concentrations were
generally at or near the method detection limit of 11 ng/L at the Rogers Island sampling station and
increased gradually to approximately 30 ng/1 over the first 2 miles of the TIP, to river mile 193.
Over the four mile section of the TIP between river mile 193 and 189, center channel PCB
concentrations increase by approximately 40 to 60 ng/L. At average flows of approximately 4,000
cubic feet per second (cfs) observed during the surveys, this increase represents a mass loading rate
of 0.4 - 0.6 kg day4. These mass loading rates represent sediment areal flux rates of approximately
0.3 to 0.4 mg m 2 day1 across this region of the TIP. This mass loading rate is generally consistent
with the load expected from observed 1991 surface sediment PCB concentrations. It does not appear
that any additional load, other than that attributed to surface sediments, is required to achieve the
observed water column PCB concentrations between river miles 193 and 189.
The TIP survey results indicate elevated PCB concentrations in waters along the eastern and
western shoreline. These occasional high values do not necessarily indicate the presence of an area
of elevated PCB flux from the sediments. Rather, they appear to be the result of lateral variations
in river flow. For example, pronounced increases in water column PCB concentrations along the
eastern shore across from the Snook Kill (Figure 4-18; Transect 12) can be attributed to a change in
QEA
45
March 19,1998
-------
TIP Sediment PCB Sources
hydrodynamics in this region of the river. The elevated concentrations occur downstream of a group
of small islands that impede river flow along the eastern shore (Figure 4-6). Field measurements of
flow velocities in this region of the river indicate the high concentrations were measured in
backwater areas (O'Brien & Gere, 1998). Under these conditions, surface sediments at the same
PCB concentration as upstream areas and exhibiting the same areal PCB flux would produce higher
water column PCB concentrations. This phenomenon was observed along several of the near shore
areas (Figure 4-18).
To illustrate the backwater effect, consider a section of the river having a sediment area A,
(L2). Water flows into and out of this section of the river at a rate of Q (L3 T '). Assume water
flowing in does not contain PCBs and the only water column source is diffusion from contaminated
sediments (Js: M L'2 P1). At steady state, the PCB concentration in water leaving this area (Cout :M
L'3) can be calculated as:
/ A
Given a uniform areal PCB flux rate of 0.4 mg m 2 day 1 and a sediment area of 100,000 m2 (the
approximate area of the eastern river channel between transects 10 and 12), the PCB concentration
in water traveling over this sediment would increase in inverse proportion to the river flow rate, as
shown in Table 4-5.
QEA
46
March 19, 1998
-------
TIP Sediment PCB Sources
Table 4-5. Relationship between river flow rate and PCB concentration considering a constant
sediment flux rate.
Flow Rate
Cout
(cfs)
(ng/L)
10
1635
50
327
100
164
500
33
1000
16
To further demonstrate the importance of river hydrodynamics in determining spatial patterns
of water column PCB concentrations, the two-dimensional hydrodynamic model described in
Section 3 was used to estimate river flow velocities within the TIP. These results are presented in
Figure 4-19 for a total river flow rate of 4380 cfs8. The model predicts the greatest river flow
velocities within the center channel, with lower velocities along the shorelines, a pattern consistent
with the field measurements described above. The impact of spatially varying flow velocities on
observed water column PCB concentrations was simulated by:
•applying a spatially uniform flux of a conservative substance from the sediments to the
water column,
•calculating water column concentrations for each of the model grid elements, and
•normalizing water column concentrations to the average concentration passing TID9.
8The average flows for the TIP time-of-travel surveys.
9In this way the influence of hydrodynamics on the predicted water column concentrations can be observed
independent of the actual flux used in the calculation.
QEA
47
March 19,1998
-------
TIP Sediment PCB Sources
The results of these calculations appear in Figure 4-19 and demonstrate that given a uniform
sediment flux rate, water column concentrations are dependant on river hydrodynamics. The largest
concentrations occur in regions with the slowest flow velocities. This simplified representation of
a uniform sediment source underscores the importance of understanding small-scale differences in
river hydrodynamics when interpreting the spatial patterns in water column PCB loading observed
during the time of travel surveys.
In addition to river hydrodynamics, spatial patterns in water column PCB concentrations
depend upon spatial variations in sediment PCB flux. The flux of PCBs from surface sediments to
the water column depends on the organic carbon normalized PCB concentration, the sediment-water
exchange coefficient, and the PCB partition coefficient as described using Equations A-10 to A-15
(Appendix A). Regions of the river with equal surface sediment organic carbon normalized PCB
concentrations and composition contribute equally to the water column PCB load. Data gathered
by the NYSDEC in 1984 indicate that mean organic carbon normalized PCB concentrations are
similar inside and outside the sediment "hot spot" areas (Table 4-6; Figure 4-20)'°. Moreover,
organic carbon normalized PCB concentrations were similar for both coarse grained and fine grained
sediments collected in 1991 from the TIP (Table 4.6). Therefore, coarse grained and fine grained
sediment areas and "hot spot" and non-"hot spot" areas are expected to have similar sediment pore
water PCB concentrations and, through the process of sediment diffusion, similar areal PCB fluxes.
Such conditions would produce the pattern of gradually increasing water column PCB concentrations
observed within the center channel during the time-of-travel surveys conducted in 1996 and 1997
(Figures 4-16 and 4-17). Thus, the differences in water column PCB concentration between the
center channel and the near-shore zones are not evidence that "hot-spots" dominate the PCB flux.
10In this analysis, 1984 organic carbon concentrations were estimated as 40% of the reported volatile solids
concentration.
QEA
48
March 19,1998
-------
TIP Sediment PCB Sources
In fact, the sediment data suggest that the non-"hot spot" areas dominate because they constitute the
vast majority of the river bottom. Localized variations in river hydrodynamics are the likely cause
of the concentration variations observed during the time-of-travel surveys.
Table 4-6. TIP Organic Carbon Normalized Surface Sediment PCB Concentrations: 1) Both
Inside and Outside NYSDEC "Hot Spots" in 1984 (0-2.5 In.), and 2) for Coarse Grained and
Fine Grained Sediments Collected in 1991 (0-5 cm).
Sediment Survey
Location/
Sediment Type
U
Observations
Mean PCB
Concentration
(mg/kg oc)
Std. Deviation
(mg/kg oc)
1984 NYSDEC
Inside
"Hot Spots'"
155
2045
2069
Outside
"Hot Spots"
177
2030
1827
1991 GE
Coarse
Sediments
16
2941
1824
Fine
Sediments
41
.
2185
2265
1) These statistics excludes one sample collected in 1984 which contained 331,000 mg PCB/kg oc.
The TIP time-of-travel surveys did not reveal any localized regions of elevated surface
sediment PCBs within the pool that are disproportionately contributing to the water column PCB
load. Elevated water column PCB concentrations along several of the near shore areas have been,
at least partially, attributed to localized changes in the hydrodynamics. PCB loadings characterized
using center channel data uninfluenced by localized hydrodynamics depict an approximately uniform
increase in PCB mass loading that is consistent with surface sediment exchange processes and the
1991 surface sediment PCB concentrations. These observations are discussed further in §4.3 below.
QEA
49
March 19,1998
-------
TIP Sediment PCB Sources
4.3 Erroneous Estimates of PCB Flux Due to Biased Sampling
The hypothesis that biased sampling may have resulted in erroneous estimates of PCB flux
into or out of the TIP was considered as a possible cause of the TIP anomaly. Biased low estimates
of PCBs transported into and/or biased high estimates of PCB transported out of the pool could have
produced the anomaly. Initial assessments of the routine monitoring stations located along the Route
197 Bridge in Fort Edward and the western wing wall of the TID suggested that samples collected
from these stations provided reasonably representative estimates of PCB loading into and out of the
TIP, respectively (O'Brien & Gere, 1993a; HydroQual, 1995a). Nonetheless, this hypothesis was
further tested during extensive field efforts conducted in 1995, 1996, and 1997.
4.3.1 Route 197 Bridge in Fort Edward
The approach for assessing the representativeness of the Fort Edward monitoring station
involved the simultaneous collection of water column samples from the routine monitoring station
and at stations across a transect perpendicular to river flow located approximately 0.5 miles upstream
(Figure 4-21). The transect was located in a region of the river characterized by shallow, vertically
well mixed, and swift moving waters to minimize the potential for vertical stratification of water
column PCBs due to particle size sorting or sediment bed loading. Nonetheless, samples were
collected from two depths: near the air-water interface (0-3 inches of water column) and the
sediment-water interface (3-6 inches from the sediment bed; O'Brien & Gere, 1996c). Samples
along the transect were collected as temporal composites with equal volume aliquots collected every
hour over a six hour period. Samples were collected from the routine monitoring station on the same
day as the transect samples as an equal volume composite of samples collected at three depths from
both the east and west channel of Rogers Island (O'Brien & Gere, 1996c). Sampling was performed
twice during the fall of 1995 (September 17 and October 3, 1995).
QEA
50
March 19, 1998
-------
TIP Sediment PCB Sources
The routine monitoring station located at the Fort Edward station provides reasonably
representative data for assessing the PCB loading into the TEP. Results of the transect study indicate
that PCB concentrations of samples collected from the routine monitoring station agree well with
samples collected across the transect located 0.5 miles upstream (Figure 4-22). PCB concentrations
in transect samples were generally within 25% of the concentrations found in samples collected from
the routine station. Furthermore, the transect monitoring indicates that PCBs within this reach of
the river, and under the flow conditions sampled, are both vertically and laterally mixed (Figure 4-
22). There was no significant difference between PCB concentrations within the shallow or deep
samples collected at the transect stations nor any significant trend in PCB concentration across the
river (Figure 4-22).
These data indicate that routine monitoring at the Fort Edward station provides reasonable
data upon which to base estimates of PCB loading into the TIP. Therefore, it is not likely that biased
sampling at the Fort Edward station contributed to the TIP anomaly.
4.3.2 Thompson Island Dam
The approach for assessing the representativeness of the TID monitoring station involved the
simultaneous collection of water column samples from the center channel of the river at a location
approximately 1000 feet upstream of the dam and from the routine station at the western wing wall
of the dam. The results of the time-of-travel surveys indicated that samples from this center channel
station accurately represent average PCB concentrations within this section of the river and are
uninfluenced by localized changes in hydrodynamics that may bias samples collected along the
shoreline (§4.2.4). Additional sampling was conducted from the eastern wing wall of the dam and
from stations located immediately downstream of the dam within the western and eastern channels
of the river at Thompson Island (Figure 4-23). Sampling and analysis methods generally followed
QEA
51
March 19,1998
-------
TIP Sediment PCB Sources
the protocols described within the sampiing and analysis plans (O'Brien & Gere, 1997a, 1997b).
Generally, where water column depth permitted, samples consisted of vertically integrated
composites made up of discrete aliquots collected from three depth intervals (0.2, 0.5 and 0.8 times
the total depth) using a stainless steel Kemmerer Bottle sampler. Where water depth restricted the
use of the Kemmerer Bottle, grab samples were collected using a stainless steel beaker. Several of
the sampling rounds also consisted of temporal composites consisting of discrete aliquots collected
over a several hour period and composited. Finally, the sampling occurred from upstream to
downstream with the timing corresponding to the estimated time-of-travel of a parcel of water
between the stations. Water column samples were analyzed for PCBs and TSS.
The TID monitoring program found that the routine shoreline sampling station at TID-west
(Figure 4-23) consistently yielded PCB concentrations in excess of those observed from the center
channel station. Fifteen pairs of samples were collected from the center channel of the river and
TID -west between September 1996 and November 1997. In all pairs, the samples from the TID-
west station contained higher PCB concentrations (Table 4-7 and Figure 4-24). The difference
between the samples ranged from 3 to 167 ng/L representing between a 6 and 163% increase (Table
4-7 and Figure 4-24). The increase observed between the two stations does not appear to be the
result of resuspension of contaminated sediments since there does not appear to be any significant
bias in TSS concentrations between the two stations (Figure 4-25).
QEA
52
March 19,1998
-------
TIP Sediment PCB Sources
Table 4-7. Paired Center Channel and TID-west Total PCB Concentrations
Date
Center Channel
(ng L"')
TID-west
(ng L ')
Difference
(ng L ')
%
Difference'
18Sep96
54
142
88
163
25Sep96
50
53
3
6
290ct96
50
102
52
104
4Jun97
84
113
29
35
17Jun97
105
272
167
159
30Jun97
175
271
96
55
14Jul97
92
190
98
107
28Jul97
67
116
49
73
13Aug97
50
90
40
80
9Sep97a
64
107
43
67
9Sep97b
70
90
20
29
10Sep97
52
94
42
81
01Oct97
65
72
7
11
10Oct97
74
82
8
11
160ct97
83
87
4
5
Mean
76
125
50
66
Std. Dev.
32
67
46
52
1 Percent difference calculated (TID-west - center channel)/center channel * 100.
The difference in PCB levels between the two stations suggested that either: 1) one or both
sampling stations were biased and unrepresentative of average PCB concentration in water passing
QEA
53
March 19,1998
-------
TIP Sediment PCB Sources
the TID, or 2) the sediments between the center channel station located approximately 1000 feet
upstream of the dam and the dam were contributing, on average, approximately half the total PCB
load observed over the entire TIP (Table 4-7). A second phase of the monitoring program was
conducted to evaluate this.
Phase 2 of the TID monitoring program involved the collection of water column samples
from numerous locations both upstream and downstream of the TID during four sampling events in
August and September, 1997. As with the other sampling events, samples from the TED-west station
contained higher PCB concentrations than those collected upstream at the center channel station
(Figure 4-26). The center channel samples produced PCB concentrations consistent with the
generalized PCB loading pattern observed throughout the TIP as observed during the time-of-travel
surveys (§4.2.4). Similarly, water column samples collected downstream of the dam in both the
western and eastern channels were consistent with center channel samples collected upstream of the
dam, and were significantly lower than concentrations along the shoreline at the dam. PCBs in
samples downstream of the dam within the western and eastern channels were, on average, 34%
lower than in samples collected from the dam. These data clearly indicate that the routine samples
collected from the TED-west station are not representative of average concentrations passing the TED.
Water column monitoring conducted by the USEPA from the western shoreline upstream of
the TID likely contains a bias similar to that of the TED-west station. On October 1, 10, and 16,1997
water column samples were collected from a western shoreline station upstream of the TID from a
location close to that sampled by the USEPA during their water column transect and flow averaged
sampling studies. These samples produced PCB concentrations significantly higher than those
collected from the more representative center channel stations upstream and downstream of the dam
(Figure 4-26). These data provide strong evidence that the TIP PCB flux estimates developed by
the USEPA are based upon biased data.
QEA
54
March 19,1998
-------
TIP Sediment PCB Sources
Water column monitoring downstream of TID at Fort Miller" and Schuylerville, NY
provides further evidence that the routine TID-west station and the USEPA TID station produces
biased high PCB concentrations. Samples from the Fort Miller and Schuylerville stations contained
PCB concentrations consistent with both the measurements at the stations downstream of the TID
and our understanding of PCB dynamics in the river (Figure 4-27). The sediments within river
reaches between TID and Fort Miller, and Fort Miller and Schuylerville contain PCBs at levels that
should produce water column PCB loadings through sediment-water exchange mechanisms under
low flow conditions (O'Brien & Gere, 1993b). The monitoring conducted since August 13, 1997
between Fort Edward and Schuylerville produces an approximately linear increase in PCB loadings
with river mile (0.1 lb mi1 day"1 ; Figure 4-27), indicating that the sediments of the TIP are
contributing no more PCBs than adjacent reaches downstream.
In contrast, low flow loading estimates developed from USEPA water column transect data
produce a spatial pattern of PCB loading that is inconsistent with the spatial patterns of sediment
PCB levels and our understanding of sediment-water interactions (USEPA, 1997; Figure 4-28).
Samples collected by the USEPA during August 1993 produced a spatial pattern of PCB loading that
suggested the loading from the TIP was elevated compared to adjacent reaches of the river, as the
calculated loading at the TID exceeded that measured at the Schuylerville station (Figure 4-28).
This is inconsistent with spatial patterns of PCB loading observed during the same season in 1997
using data from stations considered to be free of sampling bias (Figure 4-28). This analysis provides
further evidence that the loading estimates developed by the USEPA and presented in the DEIR
(USEPA, 1997) overestimate PCB loading from the TIP.
' 'The Fort Miller sampling station is located approximately two miles downstream of the TID.
QEA
55
March 19, 1998
-------
TIP Sediment PCB Sources
4.3.3 Possible mechanism for the observed bias at TID-west
The observed bias at TID-west may be the result of incomplete lateral mixing. The region
immediately upstream of the TID along the east and west shorelines consist of emergent aquatic
vegetation beds that may be hydraulically isolated from the main stream of the river. PCB
concentrations in these waters are likely elevated in comparison to PCBs in the center channel
samples as the diffusive flux from sediments is integrated into a smaller volume of water. Shear
forces along the boundaries of these water masses may promote the transport of waters containing
higher PCB concentrations within a thin band along the shorelines. This thin band of water may be
what is sampled from the shoreline locations at the TID and what was sampled by the USEPA during
its transect and flow averaged sampling studies of 1993 (USEPA, 1997). This hypothesis is
supported by two-dimensional hydrodynamic model estimates of river flow velocities (described in
§3.2.1) which identify a region of river flow immediately upstream of the TID that is lower than that
in the main channel of the river. Additionally, application of a spatially uniform flux of a
conservative substance from the sediments to the water column (§4.2.4), produces normalized
concentrations at the TID west station that are in excess of that observed across the face of the dam
(Figure 4-29). These data demonstrate that river hydrodynamics play an important role in the
representativeness of the samples collected from the TID-west station. Nonetheless, it is apparent
that the routine sampling station located at the western wing wall of the TID produces PCB
concentrations that are not representative of the average PCB concentration across the TID.
4.3.4 Composition of the sediment PCB source
The composition of the summer low-flow (June - August 1997) average TIP load was
calculated as the difference in water column derived PCB congener peak loading across the TIP
using unbiased data collected from Fort Edward and the vicinity of the TID. The source of this
QEA
56
March 19, 1998
-------
TIP Sediment PCB Sources
loading was assessed by calculating the required composition of a surface sediment source, assuming
equilibrium partitioning between sediments and pore waters and a diffusive mass transport
mechanism. Specifically, the approach included:
1) Calculation of TIP water column PCB peak (based on a DB-1 capillary column) loadings
from paired Fort Edward (C,c) and unbiased TID water column PCB data (Ctld) in accordance
with the following equation:
W = O (C - C ) 4-5
wc */ty lid ft' H J
where:
Qfe is Fort Edward flow (L' T '), and
2)Calculation of the sediment-phase PCB composition assuming the load calculated using
Equation 4-5 originates from surface sediments and is transported to the water column via
diflusional processes. This load was calculated by substituting Wwc for Wd in Equation A-15
(Appendix A), solving for C, on a DB-1 peak basis, and calculating the congener peak and
homolog distributions. The parameters used in the calculation are summarized in Table 4-8.
QEA
57
March 19,1998
-------
TIP Sediment PCB Sources
Table 4-8. Parameters Used in the Calculation of Surface Sediment PCB Source Signature.
Parameter
Description
Value
Units
Source
Kf
Sediment-water
Exchange Coefficient
2
cm day
GE Model
Calibration
A,
Surface Sediment
Area
2xl06
m2
GE Hudson River
GIS
"W
Pore Water DOC
33.7
mg L 1
GE 1991
Sediment Survey
Temperature
Corrected Partition
Coefficient
Varies w/
Temp, and
Congener Peak
L kg"1
USEPA Phase 2
Data as calculated
in GE (1997)
foe
Fraction Organic
Carbon
1.82
%
GE 1991
Sediment Survey
K.OC
DOC Partition
Coefficient
0.1
L kg"1
_
The homoiog distribution of the summer low flow TIP load appears in Figure 4-30a. On
average, the homoiog distribution of the TIP load consists of approximately 55% mono- and
dichlonnated PCBs (Figure 4-30a). Back calculating the particulate-phase PCB concentration of
surface sediments yields the average homoiog distribution in Figure 4-30b. This PCB source best
matches the surface sediment PCB composition as represented by the 0-2 cm sections of the USEPA
high resolution cores collected from the TIP in 1992 (Figure 4-3 la). In contrast, the source of the
TIP load does not appear to match the composition of PCBs found at depths greater than 8 cm
(Figure 4-3 lb). This analysis indicates that the source of the TIP PCB load is surface sediments as
expressed through a diffusive flux mechanism.
QEA
58
March 19,1998
-------
TIP Sediment PCB Sources
4.4 New Paradigm for Sediment-Water PCB Exchange in the TIP
The discovery that a sampling bias at the TID-west station was responsible for the
anomalously high PCB loading estimates from the TIP sets up a new paradigm for sediment-water
interactions in the upper Hudson River: PCB loading patterns within the river are consistent both
with conventional sediment-water exchange mechanisms and PCB concentrations and compositions
found within the surface sediments. In response to loadings emanating from sediments throughout
the upper Hudson River, primarily by way of diffusion, water column PCB concentrations increase
approximately linearly with distance downstream. The observations of elevated loadings from the
TIP following the release of PCBs from the plant site areas appear primarily to be the result of biased
high PCB levels in samples collected from the TID-west station. This discovery now allows
calculations of PCB fate and transport in the river to proceed without invoking extraneous PCB
sources or unsupported sediment-water exchange mechanisms to account for the observed loadings.
The discovery of the sampling bias at the TID-west station invalidates conclusions drawn by
the USEPA regarding TIP sediment loadings. In the DEIR, the USEPA concluded that the
measured TIP load originated from the TIP sediments (USEPA, 1997). Based upon the unbiased
sampling conducted since the fall of 1997, it appears that a significant portion of this loading was
due to the sampling bias. The USEPA also stated that PCB transport downstream of the TIP was
conservative, with little or no change in water column PCB loads with distance downstream. This
pattern was produced by the biased high PCB data collected at the TID station. Unbiased data
collected since the fall of 1997, produces a gradual increase in water column PCB loading between
Fort Edward and Schuylerville as water flows over downstream PCB deposits (Figure 4-27).
QEA
59
March 19,1998
-------
TIP Sediment PCB Sources
SECTION 5
SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS
5.1 Summary
A state-of-the-science model of PCB fate and transport has been configured to represent the
upper Hudson River system and calibrated using extensive field data from the study area. The ability
of the model to represent the short-term and long-term changes in water column and sediment PCB
levels over the period from 1977 to 1991 indicates that the model provides a reliable, quantitative
representation of the significant mechanisms in the upper Hudson River that affect the fate and
transport of PCBs. However, comparison of the model to water column monitoring data from the
TDD for the period from 1991 to 1996 suggested that the model underestimated the increase in PCB
tevels between Rogers Island and the TID. Efforts to alter the model calibration to achieve water
column levels consistent with the TID data were unsuccessful. This failing of the model led to the
formation of hypotheses regarding additional PCB sources and biased sampling.
5.2 Conclusions
An extensive field sampling program and data analysis effort designed to address the various
hypotheses regarding TIP PCB loading sources revealed five major conclusions:
The water column concentrations measured at the TID overestimated the average PCB
concentration in water passing this location. The shoreline sampling location is influenced by a
quiescent backwater immediately upstream that tended to result in higher PCB concentrations than
the cross-sectional average concentration. Unbiased data collected in the main channel upstream and
downstream of the TID indicate that cross-sectional average concentrations are approximately 1.5
QEA
60
March 19,1998
-------
TIP Sediment PCB Sources
times lower than those measured at the shoreline stations. The unbiased data indicate that sediments
within the TIP contribute between a 0.5 and 1 lb/d of PCBs to the water column.
2. PCB levels increase in a linear fashion as water passes through the TIP, indicating a nearly
uniform areal flux from sediments within the pool. The spatial patterns in water column PCB
loading indicate that the diffusive flux of PCBs from sediments is similar across the TIP. This is due
to the similarity of surface sediment organic carbon-normalized PCB concentrations that produce
spatially invariant areal PCB flux.
3. The composition of the TIP PCB load is consistent with the surface sediment PCB
composition considering equilibrium partitioning and sediment pore water exchange processes.
During the summer low flow period, the composition of the TIP load closely resembles that which
would result from equilibrium partitioning and pore water exchange with the surface sediments of
the TIP. The composition is inconsistent with pore water exchange with sediments containing
extensively dechlorinated PCBs such as those buried within the hot spot regions of the over.
<*• Water column PCB loadings increase as water travels downstream of the TIP. The spatial
patterns in water column PCB loading developed from unbiased data are consistent with known and
understood sediment-water exchange mechanisms and surface sediment PCB concentrations. The
conclusions drawn by the USEPA regarding the origin and fate of the TIP sediment source are not
supported by the unbiased data. The biased data overestimate the magnitude and importance of the
TIP sediment source.
5. While a significant portion of the TIP anomaly can be explained bv sampling bias at the
TIP station, it is still possible that the Allen Mill event increased PCB levels in surface sediments
within the TIP. Care must be taken when calibrating the long term fate and transport models to even
QEA
61
March 19,1998
-------
TIP Sediment PCB Sources
the unbiased data currently being collected as they may be affected by elevated surface sediment
concentrations resulting from the mill loadings. Calibration to the corrected USGS data may provide
a means to determine if the Allen Mill event did increase the surface sediment concentrations within
the TIP. This would allow more accurate estimates of the sediment to water mass transfer coefficient
and yield more reliable estimates of future water column PCB concentrations.
6. Based on observations of DNAPL in the river bed at Bakers Falls, the extent of DNAPL
presence at the Hudson Falls site, and the results of the DNAPL transport study, it is possible that
PCB DNAPL from the plant entered the river throughout the 1980s and was deposited in surface
sediments of the TIP. Since this mechanism and PCB loading is not represented in the PCB fate and
transport models, the models may not accurately describe what is occurring in the river. For
example, surface sediment mixing, sediment water exchange, and PCB partitioning may be sensitive
to an underestimation of surface sediment loading. However, the 1997 high flow data indicate that
remedial activities at the site have successfully controlled the movement of DNAPL from the plant
site into the river. This suggests that recovery rates may be accelerated over those observed in the
late 1980s. The on-going water column monitoring program will provide data from which to
evaluate the recovery rate of the river. Moreover, as it has been seven years since the last extensive
survey, additional sediment sampling within the TIP may yield important information on the impact
of the Allen Mill event and subsequent recovery on surface sediment PCB concentrations.
5.3 Recommendations
1. The analytical bias in PCB concentrations reported bv the USGS for monitoring stations
south of Ft. Edward during the mid to late 1980's should be assessed and corrected hefore the data
are used in model calibration. If this data set is not corrected, the impact of the Allen Mill PCB
QEA
62
March 19,1998
-------
TIP Sediment PCB Sources
loadings on surface sediment PCB concentrations within TIP may be indeterminable, resulting in
considerable uncertainty in model projections.
2. Model calibration needs to be based on at least 1 year of data from the unbiased sampling
station located immediately downstream of TIP. GE began collecting this data as well as data from
Schuylerville last September. This is necessary given the strong seasonal variability in the PCB
loading from the TIP observed in the existing data set.
3. Consideration should be given to performing additional sampling and analysis for PCBs
in TIP sediments. Extensive surveys were conducted in 1977, 1984, and 1991. Given the
uncertainty over the impact of the Allen Mill event on surface sediment PCB concentrations and the
amount of time that has transpired since the last survey, this data would be useful in model
calibration and increasing the reliability of the projections.
QEA
63
March 19, 1998
-------
TIP Sediment PCB Sources
REFERENCES
Belanger, T.V. and Montgomery, M.T., 1992. Seepage Meter Errors. Limnology and Oceanography,
37(8)1787-1795.
Gallagher, D.L., Dietrich, A.M., Reay, W.G., Hayes, M.C., and Simmons, G.M., 1996. Groundwater
discharge of agricultural pesticides and nutrients in estuarine surface water. Ground Water
Monitoring and Remediation, Winter issue, 118-129.
General Electric Company, 1997. 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.
Harza Engineering Company, 1990. Fort Edward Dam PCB Remnant Deposit Containment,
Environmental Monitoring Program, Baseline Studies. Report of 1989 results. February,
1990.
HSI GeoTrans, 1997. Investigation of Groundwater Seepage in the Upper Hudson River, prepared
for General Electric Company. October, 1997.
HydroQual, 1998. Memorandum from J. Rhea of HydroQual, Inc. to M. Schweiger and J. Haggard
of GE Corporate Environmental Programs, Albany, NY dated 29 January, 1998.
HydroQual, 1997a. Sampling and Analysis Plan, 1997 High Flow Monitoring Program, Upper Hudson
River. Prepared by HydroQual, Inc. for the General Electric Company Corporate Environmental
Programs, Albany, N.Y. March 1997.
HydroQual, 1997b. Sampling and Analysis Plan, Investigation of Groundwater Seepage in the
Thompson Island Pool Section of the Upper Hudson River. Prepared by HydroQual, Inc. for
the General Electric Company Corporate Environmental Programs, Albany, NY. February
1997.
HydroQual, 1997c. Hudson River PCB DNAPL Transport Study, Prepared by HydroQual, Inc. for
the General Electric Company, Corporate Environmental Programs, Albany, NY. June, 1997.
HydroQual, 1996. Hudson River PCB DNAPL Transport and Water Column Monitoring Study.
Sampling and Analysis Plan. HydroQual, Inc. in association with O'Brien & Gere Engineers,
Inc. August 1996.
QEA
64
March 19,1998
-------
TIP Sediment PCB Sources
HydroQual, 1995a. Anomalous PCB Load Associated with the Thompson Island Pool: Possible
Explanations and Suggested Research. Prepared by HydroQual, Inc. for the General Electric
Company, Corporate Environmental Programs, Albany, N Y. October 1995.
HydroQual, 1995b. The Erosion Properties of Cohesive Sediments in the Upper Hudson River. Prepared
by HydroQual, Inc. for the General Electric Company, Corporate Environmental Programs,
Albany, N.Y. October, 1995.
J.L. Engineering. 1992. Final Remedial Action Report: PCB Remnant Deposit Sites 2. 3, 4 and 5. Fort
Edward, New York. Report of Remediation and Construction Activities Oct. 1989-Sept. 1990).
September 25, 1992.
Lee, D R., 1977. A device for measuring seepage flux in lakes and estuaries. Limnology and
Oceanography, 22(1)140-147.
Lee, D R. and Cherry, J.A., 1978. A field Exercise on Groundwater Flow using Seepage Meters and
mini-piezometers. Journal of Geological Education, 27. 6-10.
O'Brien & Gere Engineers, Inc. 1998. 1996-1997 Thompson Island Pool Studies. Prepared for
General Electric Company, Corporate Environmental Programs, Albany, NY. January, 1998.
O'Bnen & Gere Engineers, Inc. 1997a. 1997 Hydro Facility Operations and Thompson Island Pool
Monitoring. Prepared by O'Brien & Gere Engineers, Inc. for the General Electric Company
Corporate Environmental Programs, Albany, N.Y.
O'Bnen & Gere Engineers, Inc. 1997b. 1997 Thompson Island Pool Monitoring. Sampling and
Analysis Plan Addendum. O'Bnen & Gere Engineers, Inc. Prepared for the General Electnc
Company, Corporate Environmental Programs, Albany, N.Y. August, 1997.
O'Bnen & Gere Engineers, Inc. 1997c. Fort Edward Dam PCB Remnant Deposit Containment. 1996
Post-Construction Monitoring Program. O'Brien & Gere Engineers, Inc., May 1997.
O'Brien & Gere Engineers, Inc. 1996a. Hudson Falls. Operable Unit 3. Interim Remedial Measures
Report. O'Bnen & Gere Engineers, Inc. Prepared for the General Electric Company Corporate
Environmental Programs, Albany, N.Y. July 1996.
O'Bnen & Gere Engineers, Inc. 1996b. Fort Edward Dam PCB Remnant Deposit Containment. 1995
Post-Construction Monitoring Program. O'Brien & Gere Engineers, Inc., July 1996.
QEA
65
March 19, 1998
-------
TIP Sediment PCB Sources
O'Brien & Gere Engineers, Inc. 1996c. Hudson River Project, River Monitoring Test. O'Brien and
Gere Engineers, Inc., Syracuse, NY. January, 1996.
O'Brien & Gere Engineers, Inc. 1995. Hudson River Project, River Monitoring Test, Sampling and
Analysis Plan. Syracuse, N.Y. O'Brien & Gere Engineers, Inc. August 1995.
O'Brien & Gere Engineers, Inc. 1994a. Bakers Falls Operable Unit 3, Remedial Investigation Report:
Syracuse, N.Y. O'Brien & Gere Engineers, Inc. Prepared for the General Electric Company
Corporate Environmental Programs, Albany, N.Y. January 1994.
O'Brien & Gere Engineers, Inc. 1994b. Fort Edward Dam PCB Remnant Deposit Containment, 1993
Post-Construction Monitoring Program. O'Brien & Gere Engineers, Inc., May, 1994.
O'Brien & Gere Engineers, Inc. 1993a. Temporal Water Column Monitoring Program Report. Hudson
River Project, 1991-1992 Sampling and Analysis Program. Prepared for General Electric
Company Corporate Environmental Programs, Albany, NY. May, 1993.
O'Brien & Gere Engineers, Inc. 1993b. 1991 Sediment Sampling and Analysis Program. Prepared for
General Electric Company Corporate Environmental Programs, Albany, NY. May, 1993.
O'Brien & Gere Engineers, Inc. 1991. Hudson River Project, Fall 1990 Sampling and Analysis
Program. Prepared for General Electric Company, Corporate Environmental Programs,
Fairfield, CT.
Parchure, T.M. and A.J. Mehta. 1985. Erosion of Soft Cohesive Sediment Deposits, ASCE J. Hvd,
Engr.. 111:1308-1326.
Shaw, R.D., and Prepas, E.E., 1989. Anomalous, short-term influx of water into seepage meters.
Limnology and Oceanography, 34(7) 1343-1351.
Tetra-Tech, Inc. 1997. Draft Memorandum from J.B. Butcher of Tetra-Tech to E. Garvey of Tams
Consultants, Inc. and V. Bierman and S. Hinz of Limno-Tech, Inc. regarding PCB data issues
• related to modeling. April 18, 1997.
Tsai, C.H. and W. Lick. 1987. Resuspension of Sediments from Long Island Sound, Wat. Sci. Tech..
21(6/7): 155-184.
QEA
66
March 19,1998
-------
TIP Sediment PCB Sources
U.S. Environmental Protection Agency, 1997. Phase 2 Report - Review Copy, Further Site
Characterization and Analysis, Volume 2C - Data Evaluation and Interpretation Report,
Hudson River PCBs Reassessment Rl/FS developed for the USEPA Region 2 by TAMS
Consultants et al. February 1997.
U.S. Environmental Protection Agency, 1996. Phase 2 Report - Review Copy. Further Site
Characterization and Analysis. Preliminary Model Calibration Report. Hudson River PCBs
Reassessment RI/FS. U.S. EPA, October, 1996.
U.S. Environmental Protection Agency, 1995. Phase 2 Report - Review Copy. Further Site
Characterization and Analysis. Database Report. Hudson River PCBs Reassessment RI/FS.
U.S. EPA, October, 1995.
U.S. Environmental Protection Agency, 1984. Feasibility Study. Hudson River PCB Site. New York.
Volume 1. Prepared by NUS Corporation.
Woessner, W W. and Sullivan, K.E., 1984. Results of Seepage Meter and Mini-Piezometer Study, Lake
Mead, Nevada. Ground Water, 22(5), 561-568.
QEA
67
March 19,1998
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
FIGURES
mv
mmmmm—mmmmmmmm¥c
Quwttatiw WmmmldkalmM
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
Hudson Falls Plant Sits
& Allan Mill
Thompson Island Pool
LEGEND:
Ǥ>
APPRO* RIVER MILE
ROM BATTERY HNVC
•
ROUTINE WATER COIUMN
MONITORING STATION
<£¦» CAPACITOR MANUFACTURING
FACUT1ES
OIRECTION OF
RIVER FLOW
~
APPRO* LOCATION
» REMNANT OEPOSITS
ormar Fort Edward
Dam Site
Thompson Island Pool
Thompson
iskjnct
pmpson
Island
Dam
W
INDEX MAP
Figure 2-1
Map of Hudson River from Glens Falls to Thompson Island Dam.
-------
o>
c
c
o
~3
s
*•»
c
m
o
c
o
o
CQ
0
a.
1
1500
1400
1300
1200
1100
1000
900
800
700
600
500
400
300
200
100
0
- GE Data
- USGS Data
I
I
•
•
II
. M
* II
1 11 .
I It
III
II I
II I 1
• 'i
AJI
_Ll
• < 'M
V i \«ii
* 1 !!*
j i
.1
A t,
' V
t. . i I " A i1
J Li i 11 |1
A
*f «|t
l| l/ l
* ¥ K
, I
!\
I
*
II
4 41
V'V V
•
ii
ii
H
ii
-1
1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
Year
Allan Mill
Loading Event
Figure 2-2.
Temporal Trend in Water Column PCB Concentrations at Fort Edward.
-------
( 1992 (93 1997 1 ( '
300uu
24000
18000
!- 12000
6000
JUUUU
24000
18000
12000
6000
30000
24000
18000
12000
6000
¦E 640
100.0
100.0
10.0
March
April
0.1
100.0
10.0 -
March April
March April
May
Figure 2-3. ... , c
Water Column PCB Loading at Fort Edward during Spring High Flow Events.
Note Daily averages plotted. Non-deteet PCBs plotted as open symbols a, MP'-
-------
6.0r
4.5
-i 1 r
—i 1 i 1 1 1 1 1 r 'i"" ¦
¦ USGS data between Fort Edward and Schuylarville
• GE data between Fort Edward and Thompson Island Dam
4,0
>
3.5 .
3.0.
2.5.
2.0.
1.5.
1.0.
80 81 82 83 84 85
86 87 88 89 90 91 92 93 94
Year
95
96 97
Temporal'Trends in Mean Annual Low Flow Water Column PCB Loading from
Thompson Island Pool (1980-1997).
Note: Data for flow < 10.000 cfs at Fort Edward plotted. High loads from J/)! to IJJI an>
xcluded
-------
50000
40000
30000
20000
10000
0
- r —|-
i | i —|-~r -r
J » 1— T " 1—J I I J—1 »
' ' I 1 ' 1 » t | 1 1
—r 1 ;—i—1—1—1—1—]—1—1—
"1 1 1—1 I 1—1—1—1—1—1—1—
Rlvar Row
at Fort Edward
i 1
-r
1 I 1 1 ,1 1 1 1 1 . 1 1.
i i 1 •i T
.1, i
1 1 I 1 I I . 1 1 1
_.»
¦ i t "i i i • ™ • i
w
u
Q
<
O
_j
CQ_
a-B
<=t
gwptfyy/nf.q j/-,».»
1993
1994
1995
1996
1997
Figure 2-5.
Temporal Trends in Mean Monthly Low Flow Water Column PCB Loading from
Thompson island Pool (1993-1997).
Noli-. Ihiltt (or(low < 10.000 cfs at h'ort Edward plotted
-------
MODELS
AIR
WATER
BED
Hydrodynamics
Interfacial
Bed Layer
Intermediate
Layer
Deep Bed
Sediment
Transport
Burial to
Deep Bed
I
Physical/
Chemical
Volatilization
Food Chain
Bioaccumulation
Dissolved
Organic
Carbon
Partitioning
Hydrology
dynamics
Dissolved
Component
Partitioning
Suspended
Solids
Particulate
Component
Diffusion
r
r
r
Invertebrates
-
Forage
Fish
—~
Predatory
Fish
Predation
Groundwater
Advection
Partitioning
DOC
PART
HIS
Benthic
Invertebrates
1
Diffusion to
Deep Bed
Figure 3-1.
Models, State Variables, and Kinetic Processes for PCB Dynamics.
-------
3eneral Electric Company
Hudson River Project
November, 1997
, -
f
1000 0 1000 2000 Feet
Match Line
Champlain Canal
Moses Kill
Snook Kill
Thomson Island Dam
Match Line
Figure 3-2.
Sediment Transport Model Grid for Thompson Island Pool.
-------
130
4-1
4-1
E-<
X
CD
H
W
X
m
o
<
m
128
126
124
122
Q _ = 34, 100 cfs
FE
Q = 34,300 cfs
10
~ - Observed at Fort Edward
— - Predicted at Fort Edward
O - Observed at Crockers Reef
- - Predicted at Crockers Reef
120
24
26 28
30
10
12
14
16
18
April 24 - May 18, 1983
Figure 3-3.
1-D Hydrodynamic Model Predictions and Data; Stage Height During 1983 Flood.
-------
130
Q
w FK
-P
4-1
En
K
0
H
w
K
w
0
<
fr*
W
128
126
124
122
120
Q
10
= 34,100 cfs
= 34,300 cfs
Observed at Fort Edward
Predicted at Fort Edward
Observed at Crockers Reef
Predicted at Crockers Reef
~ o /
o o
24 26 28 30
April 24 - May 18, 1983
Figure 3-4.
2-D Hydrodynamic Model Predictions and Data. Stage Height During 1983 Hood.
-------
IT,
"O
o
in ~
*c
c
OJ
a
CO
o
LO
ai
E
to
*D
O
"O
GJ
C
a
D.
tn
(75
*c
o
en
•o
QJ
"O
C
o
a
cn
3
cn
CT)
£
350
300
350
300
250
200*-
l
f
150L-
100 _
50 _
0
350
16
Schuy 1 erv: 1 le
o - Measures Concentration
St i11 water
¦fi SL
300
-
0
1
o
Water f ore!
-
250
-
-
200
0
-
150
-
-
100
N. c o
-
50
-
^ 0
-
0
1
i
19 20
April, 1382
Figure 3-5.
Model-Predicted and Observed TSS Concentrations at Schuylerville, Stillwater, and Waterford.
-------
1980
THOMPSON ISLAND POOL
' ' 1
1985
1990
1995
SCHUYLERV LLE
1980
1985
1990
1995
70
1
1 1 1 1 1 I 1 I
i i i i
I 1
60
—
STILLWATER
—
- 50
—
—
•c 40
—
—
1 30
c.
-•
—
— 20
—
.
—
—
0
1
¦ i i i i i i i
1 .
1980 1985 1990 1995
CJ
c_
>-.
L.
~c
c.
a.
T,
WATERFORD
1980 1985 1990 1995
Figure 3-6. TIME (>'ears)
Comparison of Computed and Observed Surface (0-5 cm) Sediment PCB Levels in Four Areas
of the Upper Hudson River.
-------
ANNUAL PCB LOADING PAST WATERFORD
10000
9000
LEGEND: AVERAGE AND RANGE
¦ WATER YEAR ESTIMATES
O CALENDAR YEAR ESTIMATES
8000
Run *1197-9
Run x1197-8
7000
01
-------
Schuylerville
10000
1000
100
O 0
1990
1989
YEAR
Figure 3-8.
Comparison of Computed and Observed Water Column PCB l evels at Schuylerville for the
Years 1989 - 1991.
-------
Thompson Island Dam Calibration, xll97-9
toooo
Base run :
x1197-8 -
1 a •rf5tTo •°otT
100
CD tfij °
: a
<
1996
1995
1997
1993
TIME (years)
Figure 3-9.
Comparison of Computed and Observed Water Column PCB Levels at Thompson Island Dam
for I he Years 1993 - 1996.
-------
1/4 IN. TEFLON -
BULKHEAD FITTING
1/4 IN. TEFLON "
BULKHEAD FITTING
WITH VALVE FOR AIR
RELEASE
I2QT. STAINLESS
STEEL VESSEL
SED1MEN'
N()I l<) SI Al I
Figure 4-1,
Schematic of Groundwater Seepage Meter.
1/4 IN. TEFLON TUBING
" 4.7 L. TEFLON AIR
SAMPLING BAG WITH
VALVE
-------
Locator Map
New York State
Upper
Hudson River
Rogers
Island
Site
S2
Champlain
Canal
Snook Kill
Site
S3
0.5 0 0.5 1 1.5 Miles
Griffin /
Island/
Site
S5
Moses Kill
Thompsc
Island
fDam
Legend
Instrument Locations
m Seepage Meter
• Piezometer
• Piezom. & Data Logger
/\/ Mile Points
/\/ Dams and Locks
/\/ Shore
Lock6
Figure 4-2.
Locations of Spring 1997 Groundwater Seepage Monitoring Stations.
-------
Temporal Profile of Seepage Flux Measured in Seepage Meters
I—
-4-
-OH
II
-•—I
-•-i
H-0—I
I—*—I
~ Site S1
A Site S3
O Site S5
——LTI Estimate
-B-
¦ Site S2
~ Site S4
• Site S6
Note: Horizontal bars represent
duration of seepage measurement
1.00
5/28/97
5/31/97
6/3/97
6/6/97
6/9/97
Date
6/12/97
6/15/97
6/18/97
6/21/97
Figure 4-3.
Temporal Trends in Measured Groundwater Seepage into Thompson Island Poo
-------
S1
S2
~ 5/29/97
¦ 5/30/97
• 6/5/97
~ 6/18/97
O 6/19/97
— LTI Estimate
Q
o
o
~
o
~
I
B
S3
S4
1
~
9
S5
~
~
0
o
S6
Thompson
Island
Dam
194
193
192
191
190 189
River Mile
188
187
186
185
Figure 4-4.
Spatial Trends in Measured Groundwater Seepage into Thompson Island Pool.
-------
Sediment Mass Eroded as a Function of Flow Rate
60
50
S 30
w
n
M
2
20
0
2000
8000 10000 12000 14000 16000 18000 20000
4000
6000
Flow (cfs)
PCB Mass Eroded as a Function of Flow Rate
1 50
1 25
1 00
0 75
0 50
0 25
ooo 4
0
2000
8000 10000 12000 14000 16000 18000 20000
4000
6000
Flow (cfs)
Figure 4-5.
Model-Predicted Sediment Bed Resuspension as a Function of Flow Rate,
-------
Port Etftvaro
General Electric Company
Hudson River Project
Sample Point Numbering Example
1.5 Miles
Notes
A/
/V
Legend
Sample Locations
Transect Locations
NOAA Bouys
Shore
Mile Markers
1976 NYSDEC Hotspots
Dams & Locks
NQAA 8ouy toeaoon* aepronmait Scued from
a rweoucbon of NOAA Cftwt »147W (WGS84). wrtfi
cnangn* **l cornacftom rmoti tivou^ tt&mmxm 1994
f*odu«d by ifiemaeort* Sartmg S^cry Wf»Owrt»tt
10.0698
Figure 4-6.
Sampling Locations for Thompson Island Pool Time of Travel Surveys.
-------
8
¦ i ' i i i i i i i i i
Data from Juna 4, 1997
• •
• • • •
• •
• * H
Snook KM
MoimKU
19S
8.
194
193
192
191
190
189
188
Data from Jura 17, 1997
Snook KID
Moaaa KID
195 194 193 192 191
River Mils
190
189
188
Figure 4-7.
Spatial Profile of TSS Concentrations for 1997 Time of Travel Surveys.
Note: Lateral averages plotted, open squares represent PC RDM P samples at west wtngwall ofTID
-------
Figure 4-8. ,
Epi fluorescent Photograph of Fluorescent Particles within Natural Sediment at Approximately
lOOx Magnification.
-------
Slotted Angle
Steel Frame
River Bed
100 mm Nylon
Mesh Bag
SIDE ELEVATION VIEW
NOT TO SCALE
Slotted Angle Steel Frame
Filter Bag
Orifice
River Bed
FRONT ELEVATION VIEW
NOT TO SCALE
Figure 4-9.
Schematic of In-Situ Particle Filtration Device.
-------
PCB DNAPL Transport Study
Fluorescent Particle Mass Balance
Transported
Downstream of
Thompson Island Dam
2%
Retained in Thompson
Island Pool
53%
Retained between
Hudson Falls and Fort
Edward
27%
Retained
etween Fort Edward
and Rogers Island
18%
Figure 4-10,
Fluorescent Particle Mass Balance for PCB DNAPL Transport Study.
-------
(a) Fluorescent Particles Injected
o>
M
19-38 39-114 1 15-190 190-380
Particle Size Class Osm)
> 380
v>
V)
<0
2
(b) Fluorescent Particles Captured at Fort Edward Station
(3 Day Total)
19-38
39 - 114 115 - 190 190 - 380
Particle Size Class (//m)
> 380
o>
*
Iff
5
(c) Fluorescent Particles Retained Upstream of Fort Edward
(3 Day Total)
4 0
3 5
3 0
2 5
2 0
1 5
1 0
0 5
00
•0 5
¦1 0
19-38
39 114
115- 190
190 - 380
> 380
Particle Size Class (//ml
Figure 4-11.
Fluorescent Particle Size Distribution for PCB DNAPL Transport Study
-------
Front Elevation
Plan View
1/2" Steel Plate
Lifting
Ring (
Angle
Frame
Angle
Frame
Gate
Filter Bag
(to shore)
Removable Gate
///////////////////////////////////////////Am^/////y//////////////
-------
30000
w 24000
T3 ^
uo
p 18000
W*fl
*2 12000
o>
E*-«
6000
¥ 20
» 16
W 12
-J 200
m
c
c
o
«
t-
e
v
o
C
o
u
CD
a
a.
160
120
5
March April May
Figure 4-13.
Temporal Trends in TSS and PCB Concentration and Loading During the 199" Spring High
Flow Period.
\»/t- D*sih >r. enters pi-K'ai \. PCB> pioncti ;;j <>/v»: s-.n:f>ois a: : X1DL
-------
Comparison of Particulate PCB Concentrations at Fort Edward
25
20
0>
East Channel Bed West Channel Bed West Channel Bed
Load (4/8/97) Load (4/8/97) Load (4/9/97)
1993 EPA Water 1991 TIP Surface
Column Particulate Sediment Avg
Avg
Figure 4-14.
Particulate PCB Concentrations for the Fort Edward Station.
-------
OJ
£
c
o
5
re
c
©
o
c
o
o
CQ
O
a.
450
400
350
300
250
200
150
100
50
0
Hydrofacility Monitoring
Plunge Pool
¦ Before
~ During/After
9/4/%
6/9/97
6/23/97
Non-Detects Plotted at Detection Limit
O)
£
c
o
'<5
re
c
0)
o
c
o
o
m
o
Q.
100
90
80
70
60
50
40
30
20
10
Hydrofacility Monitoring
Fort Edward
yj
4
9/4/96
6/9/97
6/23/97
Non-Detects Plotted at Detection Limit
¦ Before
~ During/After
Figure 4-15.
Water Column PCB Concentrations at Bakers Falls Plunge pool and Fort Edward from
Hydrofacility Monitoring Program
-------
if
CO
¦C CO
O CO
sfc
£ «
0 «-<
o O
300
270.
240 .
210.
180.
150 .
120 .
90.
60.
30.
0 _
195
i i i
Flow = 4,600 eft
HUDSON RIVER PROJECT
1996 Time of Travel Survey
• • •
• • • • •
September 24, 1996
194
193
192
191
190
189
188
<0 —
-C M
oca
e a.
§5
oo
300
270.
240 .
210 .
180 .
150.
120.
90 .
60.
30.
i i i
Flow = 5,200 cfs
i i i
September 25, 1996
195
194
193
192
191
190
189
188
Figure 4-16.
Center Channel PCB Concentrations from the 1996 Time of Travel Surveys.
-------
II
«*—
¦c «
o to
® o.
_
£ «
• t;
o o
Sf
<0
.C M
(J CO
**
£ «
o «-
O o
300
270 .
240 .
210.
180.
150.
120 .
90 .
60 .
30 .
HUDSON RIVER PROJECT
1997 Time of Travel Survey
i i i
Row = 4,500 cf»
Jurw4, 1997
•• • •
• •
195
300
194
193
192
191
190
189
188
270
240
210
180
150
120
90
60
30
Flow = 3.600 eft
T I I
Jutm 17. 1997
• •
• • •
• •
1 ¦ '
195 194 193 192 191 190 189
River Mile
188
Figure 4-17.
Center Channel PCB Concentrations from the 1997 Time of Travel Surveys.
S'ote Open squares represent PCRDMP samples at nest mng\ui/I of T/D
-------
September 24,1996
300
250
200
0>
c
150
IS
o
Cl
100
50
0
^
10
n
12 14
Transect No.
18
¦ WEST
HCENTER
~ EAST
September 25,1996
300
250
200
o>
c
150
m
0
Cl
100
50
0
¦ssn ^
10 12
I WEST
HCENTER
~ EAST
14
18
Transect No.
300
250
- 200
o>
- 150
to
0
1 100
50
0
10
June 4,1997
12
14
Transect No.
¦ WEST
HCENTER
~ EAST
18
10
June 17,1997
100
12 14
Transect No,
¦ WEST
HCENTER
~ EAST
18
Comparison of West, Center, and Hast Channel FCB Concentrations at Select Transects from the
1996 and 1997 Time of Travel Surveys.
-------
Snook Kill
0 2 Miles
/ Shoreline
Normalized Concentration
¦I < 0 00
0.80 - 0 85
0 85-090
0 90 - 0 95
0.95- 1 00
> 1 00
11
Figure 4-19.
Flow Velocity and Normalized
Conservative Sediment Tracer
Concentration Predictions for the
Snook Kill Vicinity of TIP.
/\/ Shoreline
Velocity (m/s)
¦¦ 0 00-005
¦I 0.05-0.10
¦I 0.10 -0 15
0.15-0 20
3 0 20-025
mi > 0 25
-------
0.2 0 0.2 0.4 Miles
Legend pcb/foc (mg/kg)
• 0-1000
9 1000-2000
9 2000 - 3000
O 3000 - 5000
O 5000 - 7000
• > 7000
Figure 4-20.
1984 Organic Carbon Normalized Surface Sediment PCB (0-2.5 in.) Concentration within the
TIP.
/\/ Mile Markers
¦¦I 1976 Hot Spots
A/Match Line
/\y Shore
-------
Fort Edward
HRM 194.9 Transect
Route 197 Bridge (West)
Routine Station - HRM 194.4
Route 197 Bridge (East)
, Routine Station - HRM 194.4
\ \ \ \ Rogers
v N1 1 Island
Champlain
Canal
Lock 7
/
A
/ /
/ /
/
/
Figure 4-21.
New York State
Upper Hudson
1
I "M Remnant Deposits
1976 NYSDEC Hot Spots
"/ \J Shoreline
ftfDams and Locks
500 0 500 1000 Feet
The General Electric Company
Water Column Sampling
Locations Within the Vicinity
of the Routine Monitoring
Station at Fort Edward, NY
10.0713
March, 1998
-------
a
e
eo
o
a.
o
t-
100
90
80
70
60
50
40
30
20
10
0
~ SHALLOW
BDEEP
¦ ROUTINE
I
1
West
LI
9/17/95
HRM 194.9 Transect
East
HRM 194.4
a
c
CO
u
Bl
100 =====
~ SHALLOW
90 - BDEEP
QO ¦ ROUTINE
OU
70 -
60 -
50 -
S 40
30 -
20 -
10 -
0 -
1
West
1
10/3/95
I.
|
HRM 194.9 Transect
East
HRM 194.4
Figure 4-22.
Water Column PCB Concentrations Within the Vicinity of Fort Edward from the 1995 River
Monitoring Test.
-------
Y
Above Dam
Shorelina^^Bm (EPA)
Shoreline at Dam (GE)
Below Dam (West Channel) $
Thompson
Island
Below Dam (East Channel)
Figure 4-23.
1976 NYSDEC Hot Spots
Shorel
/V Dams
/"\J Shoreline
100 0 100 200 300 400 Feet .
New VorV State Upper Hudson
x
! ¦
1
The General Electric Company
Thompson Island Dam
Monitoring Study
Sample Locations Near
Thompson Island Dam
February. 1998
-------
300
260
200
ihO
100
Comparison of PCB Concentrations from
Above Thompson Island Dam and at tha West Wingwall
~ Above the Dam
¦ Shoielme al Dam
{West WingwalQ
Note 9/9/97
tfau *rt sn
tretag* of 2
I illll n 11
09/18/96 09/25/96 10/29/96 06/04/9? 06/17/97 06/30/97 07/14/97 07/28(97 08/04/9? 08/13/97 09/09/97 09/10/97 10/01/97 10/10/97 10/16/97
Difference in PCS Concentrations between Samples Collected
Above Thompson Island Dam and at the West Wingwall
180
160
140
S ; 120
-?i
5 3 ioo
c •
I i
= I 80
5 I
3 I 60
1
40
20
0
tut* ste m
tvents
Bvetage ot 2
09/18/96 09/25/96 10/29/96 06/04/97 06/17/97 06/30/97 07/14/97 07/28/97 08/04/97 08/13/97 09/09/97 09/10/97 10/01/97 10/10/97 10/16/97
Figure 4-24.
Comparison of PCB Concentrations Upstream of Thompson Island Dam and at the West
Wingwall.
-------
7
¦ TID-PRW2
OTIP-18C
6
5
0>
4
3
2
1
0
0
6
5
2
3
TSS at TID-West [mg/L]
Figure 4-25.
Comparison of TSS Concentrations Upstream of Thompson Island Dam and at the West
Wingwall.
-------
400
350
300
250
200
150
100
50
0
~ Above Dam
¦ Below Dam (West Channel)
¦ Below Dam (East Channel)
¦ Shoreline at Dam (GE*)
¦ Shoreline at Dam (EPA**)
H'est Wingwall of Dam
** Approximate Location of
1991 US El'A Sampling
Note 9,9>97
data are an
average of 2
events
\
/
A
O1
A
£
A
X
JsT lNb'
&
V
-------
2.5
Below Thompson
Island Dam
Schuylerville
Fori Edward
196
194
192
190
188
River Mile
Figure 4-27.
Spatial Profile of Mean Upper Hudson River PCB Loading (Aug - Dec 199 ).
186
184
182
180
-------
Spatial Profile of Average Low Flow PCB Loading for 1993 EPA Data and 1997 GE Data
¦U.S. EPA Transect Study 6
(August, 1993
August 1997 GE Data
196
194
192
Note: Flows at Thompson Island Dam and
Schuylerville were calculated from Fort
Edward Flow IUSGS Daily Average) based
on drainage area proration
190
188
River Mile
186
184
182
180
Figure 4-28.
Spatial Profile of Upper Hudson River PCB Loading During Summer Low Flow Period for EPA
(August 1993) and GE Data (August 1997).
-------
N
I
0.08 0 0 08 Miles
Moses Kill
Thompson Island Dam
^orm
Shoreline
lormalized Concentration
<0.80
0.80 - 0 85
0.85 - 0.90
0.90-0 95
0.95-1.00
> 1.00
Figure 4-29.
Flow Velocity and Normalized
Conservative Sediment Tracer
Concentration Predictions for the
Thompson Island Dam Vicinity
of the TIP.
&
Shoreline
el'ocity (m/s)
¦I 0.00 - 0.05
¦ 0.05-0.10
¦I 0.10-0.15
0.15-0.20
3B 0 20 - 0.25
¦ > 0.25
-------
Water Column Delta Load
S
V
w
u
0.
U
'Z
50%
40%
30%
20%
10%
0%
Mono Di Tn Tetra Penta Hexa
Homolog Group
Hepta
Octa
Nona
Deca
Calculated Diffusional Sediment Source
50%
40%
2 30%
u 20%
z
z
10%
0%
i
1
i
rh
Mono Di Tn Tetra Penta Hexa Hepta Octa Nona Deca
Homolog Group
Mean +/'- 95% Confidence Interval for June to August 1997
Upstream Station: Fort Edward
Downstream Stations: TIP 18C and TID PRW-2
Figure 4-30.
PCB Homolog Distribution of Water Column Delta Load Across the TIP and Calculated
Sediment Source Required to Produce Water Column Load by Equilibrium Partitioning.
-------
\.l\ A\ £
4(1
(a) Surface Sediments
I «< 1997 Calculated Summer Average
1992 Hi Res 0-2 cm
60 SO 100 )2u
DB I PC B l'r«k
(b) Deep Sediments
_ i& JS. J!
** *x „ **
A_ :SSW»#—
****/!*
20
40
* 1997 Calculated Summer Average
-1992 Hi Res >8 cm
60
DB I PC B Hcnk
«»««»« Mil.I.lllllll j
80 100 120
Figure 4-31.
Comparison of PCB Peak Compositions for Calculated Diffusional Sediment Source (1997
Summer Average) with Sediments from 1992 EPA High Resolution Cores Collected from TIP.
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
r\.JT JT .LrlA r\
-------
THIS PAGE LEFT BLANK INTENTIONALLY
-------
TIP Sediment PCB Sources
APPENDIX A
CONCEPTUAL MODEL OF TIP PCB DYNAMICS
A.l PCB Mass Balance
The conceptual model of PCB dynamics in TIP is presented graphically in Figure A-l. PCBs
in the water column are present in three phases: 1) freely dissolved, 2) sorbed to particulate matter,
and 3) bound to dissolved organic carbon (DOC). The relative distribution among these water
column PCB phases is described by equilibrium partitioning concepts. TIP water column PCB
concentrations are affected by external loadings from the upstream plant site areas, loadings from
sediment sources, advective transport to downstream reaches, and exchange with the atmosphere via
volatilization. A brief description of these mechanisms and processes with respect to their
importance in TIP PCB dynamics is described below.
A.l.l Partitioning
Total water column PCBs are expressed as the sum of the dissolved, particulate-bound, and
DOC-bound fractions. Equilibrium partitioning, with local linear sorption is used to describe the
distribution among these phases. Particulate phase PCBs are bound to the organic carbon fraction
of the water column suspended solids and are in equilibrium with the freely dissolved phase. The
organic carbon partition coefficient is used to characterize the distribution between these two phases
as follows:
C = C K f m (A-l)
p a oc oc ss v '
QEA
A-l
March 19,1998
-------
TIP Sediment PCB Sources
where:
Cp is the water column particulate PCB concentration (M L"3),
Cd is the water column dissolved PCB concentration (M L"3),
K^. is the PCB organic carbon partition coefficient (LJ M"1),
f^ is the organic carbon fraction of water column particulates (M M'1), and
mss is the water column suspended solids concentration (M L'3).
PCBs sorbed to water column DOC (Cdoc; M L'3) are in equilibrium with the freely dissolved phase,
as described by the following equation:
Kj()C is the PCB dissolved organic carbon partition coefficient (L3 M'1), and
mdoc is the water column dissolved organic carbon concentration (M L"3).
Total TIP water column PCBs can be written in terms of the dissolved phase concentration:
doc J doC doc
(A-2)
where:
(A-3)
where:
C„p is the total PCB concentration in the TIP water column (M L'3).
QEA
A-2
March 19,1998
-------
TIP Sediment PCB Sources
A.1.2 External loadings
External PCB loadings to the TIP potentially exist anywhere water flows into the system.
The magnitude of an external PCB loading depends on the flow rate and PCB concentration of the
contributing source:
Wc is the external PCB loading (M T"'),
Qc is the volumetric flow rate of the external source (L3 T'1), and
Ce is the PCB concentration of the external source (M L ?).
External loadings result from sources such as tributaries, industrial discharges, and sewer outfalls.
A.1.3 Sediment sources
PCBs within the TIP sediments contribute to water column PCBs through three mechanisms:
1) bed resuspension, 2) pore water diffusion, and 3) groundwater advection. The sediment PCB
load is the product of the mass flux due to each mechanism listed above and the surface area of PCB-
contaminated sediments. The total loading from TIP sediment sources (W,; M T ') is therefore the
sum of the flux from these three loading mechanisms taken over the sediment area (As; L:)
W = Q C
€ e
(A-4)
where:
W
(A-5)
QEA
A-3
March 19, 1998
-------
TIP Sediment PCB Sources
where:
J, is the sediment PCB resuspension flux (M L"2 T),
Jd is the sediment PCB pore water diffusive flux (M L"2T), and
Jgw is the sediment PCB groundwater advective flux (M L'2 T).
The physical and chemical processes that govern the sediment PCB flux ascribed to these
mechanisms are described in Section A.2.
A. 1.4 Settling
PCBs are lost from the water column via settling. In this process, particulate phase PCBs
settle from the water column and are deposited on surficial sediments. PCB mass loss from the
water column due to settling is parameterized with the mean settling velocity:
Settling Loss = C A s (A-6)
where:
vs is the mean particulate settling velocity (L T'1).
A. 1.5 Advection
Water column PCBs within the TIP are affected by advection from the upstream to the
downstream reaches as a consequence of water movement through the TIP. Advective mass
transport is the product of the PCB concentration and the river discharge. Upstream loadings and
downstream transport are summed to produce the net advective PCB mass transport:
QEA
A-4
March 19,1998
-------
TIP Sediment PCB Sources
S'et Advectwn = QC - QC
*- up c Hp
(A-7)
where:
Q is the Hudson River discharge within TIP (L3 T '), and
C„p is the water column PCB concentration flowing into TIP (M L
In the modeling framework discussed in Section 3, the measured PCB load passing the Fort
Edward station was treated as a boundary condition. This loading was calculated as the product of
the flow and PCB concentration of water passing the Fort Edward station. In this manner, external
PCB loadings to the river from sources upstream of the TIP including the remnant deposits, Allen
Mill loadings, river bed DNAPL seeps, and sediment sources were incorporated into the modeling
assessment.
A.1,6 Volatilization
Volatilization is the net mass exchange across the air-water interface and is driven by a
concentration gradient between the air and water phases. Since atmospheric concentrations are
considerably lower than the TIP water column concentrations, volatilization represents a PCB loss
mechanism. The volatilization flux is expressed in general terms as the product of the dissolved
phase PCB concentration and the volatilization mass transfer velocity. The net volatilization mass
transfer is the product of the volatilization flux and the surface area of the air-water interface:
Volatilization Loss = v C, A
v J
(A-8)
QEA
A-5
March 19, 1998
-------
TIP Sediment PCB Sources
where:
vv is the volatilization mass transfer velocity (L T'1), and
Aw is the surface area of the air-water interface (L2).
A.1.7 Governing equation
The governing mass balance equation to describing PCB dynamics in TIP can be expressed
as: the time rate of change of PCB mass is equal to the sum of PCB sources less the sum of PCB
sinks within the TIP. PCB sources include external loadings, internal sediment sources, and
advection from upstream. PCB sinks include advection to downstream, settling, and volatilization.
Assuming constant volume (V), the governing mass balance equation can be expressed as:
oC
= Y,C Q + -4 iJ + J + J ) - v C A + QC - QC - v C A (A-9)
e Mr ^ S* I s p s up ^ tip v d w
where:
t is time (T), and
V is the TIP volume (L3).
The overall mass balance equation contains concentrations in terms of total, dissolved, and
particulate PCBs. Using the partitioning relationships presented in section A. 1.1, the mass balance
may be expressed in terms of total water column PCBs. The mass balance equation presented above
is coupled with similar expressions for upstream and downstream reaches, resulting in a system of
equations for the entire river reach being modeled. Furthermore, the particulate phase and sediment
source terms require coupling of the water column and sediment solids and PCB mass balance
equations.
QEA
A-6
March 19,1998
-------
TIP Sediment PCB Sources
A.2 Sediment PCB Source Loading Mechanisms
As discussed above, sediment loading mechanisms play an important role in TIP PCB
dynamics. Sediment PCB loading mechanisms include diffusive flux, sediment resuspension, and
groundwater advection.
A.2.1 Diffusive flux
Diffusion contributes PCBs to the water column due to a concentration gradient between the
water column and surface sediment pore water. Diffusion is traditionally described using Fick's
First Law, in which the diffusive flux is proportional to the concentration gradient:
Jj = -»J, j (A-10)
where:
ip is the surface sediment porosity (L3 L '),
Ds is the diffusion coefficient (L2 T '), and
dC/dx is the surface sediment pore water PCB concentration gradient (M L 4).
The diffusion equation is simplified by expressing the concentration gradient as the difference
between the pore water and water column PCB concentrations over a finite characteristic mixing
depth:
QEA
A-7
March 19, 1998
-------
TIP Sediment PCB Sources
dC
<3*
c„' ' c<
d tJ, up
Ax
(A-ll)
where:
Cd' is the pore water PCB concentration (M L'3);
Ax is the surface sediment mixing depth (L).
Grouping the porosity, diffusion coefficient, and mixing depth into a bulk exchange
coefficient, and expressing the sediment flux in terms of mass loading, results in the following
expression:
W t = k A C' - C
d f s \ d d.tip
(A-12)
where:
and
Wd is the water column PCB load from sediment pore water diffusion (M T"1),
kf is the sediment diffusion exchange coefficient (L T'1).
Equilibrium kinetics with local linear sorption is assumed to describe the partitioning
between the freely dissolved pore water PCBs, and PCBs sorbed to sediment organic carbon:
C
C = — (A-13)
K f V
oc oc
QEA A-8 March 19,1998
-------
TIP Sediment PCB Sources
where:
Cs is the dry weight surface sediment PCB concentration (M M ').
Sediment pore water PCBs are the sum of freely dissolved and DOC-bound fractions:
c; = c„(' - "vkJ (A-|4>
where:
mdoc is the sediment pore water dissolved organic carbon concentration (M L ?).
Since the sediment pore water PCB concentration is typically much larger than the water column
concentration, the water column concentration can be neglected in the resulting sediment diffusive
loading equation:
C
Wi = k A ( I - « K ) — (A-15)
J f I \ Joe JoC I f v
' e\r t\r
The equation shown above states that the diffusive loading is proportional to the surface sediment
concentration and surface area of PCB-containing sediments. Although surface sediment PCBs
within the TrP are not spatially homogeneous, the diffusive loading equation can be used with area-
weighted averages for organic carbon normalized PCB concentrations to calculate the net TIP flux.
QEA
A-9
March 19, 1998
-------
TIP Sediment PCB Sources
A.2.2 Sediment resuspension
Resuspension or bed scour is the process by which surface sediments are mobilized and
resuspended into the water column in response to shear forces produced by water movement over
the bed. Only a finite amount of material can be resuspended from a cohesive sediment bed that is
^exposed to a constant bottom shear stress. This phenomenon, referred to as bed armoring, has been
observed and quantified in numerous laboratory studies (Tsai and Lick, 1987; Parchure and Mehta,
1985). The amount of fine grained sediment that is resuspended (s; M L'2) at a given shear stress
(t; F L"2) is given by the following empirical expression:
a
< -0
T > T„
(A-16)
where:
a is a system constant dependent on time since material was deposited as determined
from field studies,
T0 is the critical shear stress below which sediment is not subject to resuspension (F L'2),
At I < t0, e is equal to zero. Equation A-16 determines the net resuspension at a given shear stress.
The flux of sediment resuspended from the TIP at a given flow rate is dependent on the
spatially varying bottom shear stress (T,0) as calculated using a two-dimensional hydrodynamic
model. Using estimates of e within each of the model grid elements, the maximum total mass of
PCBs resuspended at a given flow (Wr) can be calculated as follows:
QEA
A-10
March 19,1998
-------
TIP Sediment PCB Sources
W
£ A C
i-i
(A-17)
where:
e, is the mass of sediment eroded from hydrodynamic grid element i (M L2),
A, is the area of grid element t (L2), and
C, is the surface sediment PCB concentration within grid element i (M M ').
This loading is distributed evenly over an assumed I hour resuspension period.
A.2.3 Groundwater advectioa
Groundwater advection occurs due to a hydraulic gradient within a porous media. One
dimensional groundwater flow is described by Darcy's Law, in which the flow is related to the
hydraulic gradient and the hydraulic conductivity of the media:
where:
Qgv, is the upward groundwater flow (L3 T '),
A, ts the area perpendicular to flow (L2),
K is the sediment hydraulic conductivity (L T '), and
ch/cx is the vertical hydraulic gradient (L L ).
The upward flow of groundwater through surface sediments and into the TIP water column
results in a net PCB loading. As groundwater travels through PCB-contaminated sediments, the
interstitial pore water reaches an equilibrium with the solid-phase PCBs. Therefore, the net
QEA
A-11
March 19, 1998
-------
TIP Sediment PCB Sources
groundwater advective PCB loading to the TIP water column is the product of the upward
groundwater flow and the surface sediment pore water PCB concentration:
W = Q Cj (A-19)
gw g» d v '
where:
Wgw is the groundwater advective PCB loading to the water column (M T"1).
The total internal sediment PCB loading within the TIP is the sum of the loadings from
surface sediment pore water diffusion, sediment resuspension scour, and groundwater advection.
Since the total sediment PCB loading varies both spatially and temporally, it is important to gain an
understanding of the factors that influence the relative importance of these three loading
mechanisms. These processes were considered in the quantitative modeling framework described
in Section 3, and several field studies described in Section 4 were conducted to examine each loading
mechanism.
QEA
A-12
March 19,1998
-------
Figure A-l.
Conceplual Model of FCB Dynamics in the Hudson River.
External Loads
Water
Column
Volatilization
Total Water Column PCBs
Upstream
Loading
Downstream
Transport
Particulate
Dissolved
Bound
Bound
i
Surface
Sediment
Layer
.•.r.sTP-T
'im'' < -
£ »<
Sediment
Bound
Porewater
Dissolved
Porewater
DOC
Bound
Total Sediment PCBs
, *r • - j '
I
i Mixing
fiifffel
*!. r'
Groin
Acfto
¦ ¦¦:
vS'.VO. .r'.
1 *
fc'Jti'l
-------
THIS PAGE LEFT BLANK INTENTIONALLY
------- |