Extraordinary estuary

Mapping the Hudson Estuary's submerged lands, JW Ladd et al.

PCBs in the Hudson River: Role of sediments, EA Garvey et al.

Hudson River restoration: Role of natural resources trustees, TM Brosnan et al.

Ecosystem indicators for Lake Ontario, F Luckey et al.

Effect of lower chlorine dosage at Buffalo WWTP, KN Irvine et al.

Livestock waste management and lake rehabilitation, CB Lind

President's message

Executive director's message

Plant profiles
   Southtowns STP
   Big Sisters STP
   Lacawana STP
   Holland TP

People and places

Spring 2002 — Vol. 32, No. 1

Abstract

In 1990 USEPA began a reassessment of its 1984 interim no-action decision concerning PCB contamination in the sediments of the Hudson River. As part of this reassessment, an extensive survey of environmental media was conducted and analyzed for PCBs on a congener-specific basis. These data have provided a wealth of information on the transport and transformation of PCBs in the environment from both geochemical and biological perspectives. These data also represent an important component supporting the recent USEPA's Record of Decision requiring the removal of over 2.6 million yd³ of sediment with an associated 70,000 kg (150,000 lb) of PCBs. The data document the continued release of PCBs from the sediments of the Upper Hudson to the overlying waters, despite the control measures implemented by General Electric at the upstream plant sites.

By looking at sediment cores representing over 200 miles of the Hudson, it has also been possible to document the importance of the GE releases relative to other potential sources. In particular, these cores document the history of PCB releases throughout the Hudson as well as the mixture of PCB congeners present at the time of deposition. Finally these cores were used to estimate the extent of dechlorination occurring within the sediments. The results show the GE releases to comprise essentially all PCB contamination in the freshwater Hudson, extending a distance of 135 miles downstream of the GE facilities.

Figure 1. Hudson River monitoring locations for (a) water column and (b) sediment.

Background

The Hudson River has a long and controversial history of PCB contamination extending back nearly 50 years. (See Sanders' extensive review of the history of PCB discharges before 1988.) PCB contamination in the Hudson is highest in the area above Troy, NY, which has been the primary area of focus for the investigation (Figure 1a). PCB contamination in the Hudson River stems chiefly from the use of PCBs at two capacitor manufacturing plants located in Hudson Falls and Ft. Edward, NY.

These plants were owned and operated by the General Electric Corporation (GE). Between 1947 and 1977, GE discharged PCB-bearing wastewater from the plants directly to the river. It is estimated that between 209,000 and 1.3 million lb of PCBs were discharged (TAMS/Gradient 1991). Direct discharges from the GE plants were effectively halted in 1977 under an agreement with New York State.

In addition to the direct discharge of PCBs, PCB-bearing oil has migrated from the Hudson Falls site through bedrock; however, the extent and magnitude of this release is not well known. This release through bedrock continues to the present, but it has diminished. Despite some evidence for its existence before 1991 (Tofflemire 1984), this leakage was not identified until the partial failure of an abandoned mill structure near GE's Hudson Falls plant site in 1991.

The mill's failure caused a large release of what were probably PCB-bearing oils and sediments that had accumulated in the structure, and it may also have augmented PCB migration from the bedrock beneath the plant to the river until remedial measures by GE from 1993 to 1996 greatly reduced the release rate. Once released to the river, PCB interactions among sediment, water, and biota have distributed PCBs throughout the Hudson River.

The Hudson River between Ft. Edward and Troy is part of the Lake Champlain canal system and contains a series of run-of-the-river dams and associated locks. These dams create large slowly flowing pools where sediments can accumulate. Before 1973, a large portion of the PCBs released from the GE facilities was sequestered behind a dam located at Ft. Edward. Concern over a possible failure due to the dam's age and weakened condition resulted in its removal in 1973.

Unfortunately, before a second dam could be built in its place, major spring floods in 1974 and 1976 redistribute the sediments downstream with a significant portion of them settling in the next river pool downstream, the Thompson Island (TI) pool.

In response to the extensive PCB contamination generated by the GE plants as well as its redistribution in mid-1970s, both state and federal agencies have attempted to assess the contamination and select a remedial action. This process has been a difficult one for many reasons, both technical and political.

As part of the USEPA's Hudson River PCBs Reassessment, TAMS and its associates have conducted several major sample collection efforts. Two of these efforts provide the main data set for this discussion:

In addition to these sampling efforts, GE has been conducting a federal-mandated water column-monitoring program in the Upper Hudson. Some of these data are examined here as well.

Identification of PCB sources

Identification of the sources of PCBs to the Hudson was a major focus of the Reassessment early in the investigation. Given the large number of potential sources in the Hudson watershed (cities, factories, and so forth), it was necessary to examine both current and historical conditions to assess the optimal remedial measures and their likely effect.

The identification of PCB sources had two perspectives:

Current sources to the water column

To examine the first question, a series of water column samples of various types were collected in 1993. In each case, the samples represented essentially “snap shots” of conditions in the Hudson. Two types of samples were collected by TAMS for the USEPA. Dissolved and suspended phase samples were analyzed for PCB congeners in both sampling techniques.

The first type were large volume (16-L) samples that were obtained during a time-of-travel survey of the first 40 miles of the Upper Hudson (Hudson Falls to Troy). These samples were obtained sequentially from upstream to downstream to track a single parcel of water through the Upper Hudson. Figure 1a shows the locations of the water monitoring stations.

The second type of water column sampling involved a flow-weighting collection technique. During these events, samples were obtained at four stations in the Upper Hudson. The sample volume collected each day was proportional to the river flow on that day, based on a staff gauge maintained by the U.S. Geological Survey at the upstream end of the study area. Samples were obtained at the four stations every other day over a 15-day period. At the end of the 15-day period, the eight individual samples for each station were composited into a single sample for the station, approximately 16 L in volume, creating four composite samples for every 15-day period.

In both approaches, the sampling attempted to isolate temporal variations among monitoring points by tracking the movement of water through the Upper Hudson. The first techniques minimized the effect of day-to-day variations by sequencing the sample collection effort. In this manner, changes in PCB concentration and load between stations could be ascribed to events that had affected the water parcel during its transit between the stations. Thus load gains between stations were the result of additions between the stations and not an artifact of sample timing and temporal variations in sources located upstream.

In a similar manner, the flow-weighted technique avoided this issue by compositing samples over a time sufficient to dampen the daily variations. By sampling proportional to flow, the technique obtained an estimate of PCB flux over the period, effectively weighting each day's concentration by its flow.

The third type of sample examined for this study was that obtained by GE under directive from the USEPA. These samples consisted of regular weekly monitoring at a limited number of locations in the Upper Hudson, covering just the first 15 miles downstream of the GE facility. These samples consist of 1- to 2-L samples analyzed for PCB congeners relative to an Aroclor standard.

Figure 2. Water column PCB loads show a distinct pattern change due to sediment releases.

Sampling results of current sources

The results of the various sampling techniques yielded similar conclusions. The first two techniques clearly documented the release of PCBs from the sediments of the Upper Hudson to the water column early in the investigation when the results of the weekly monitoring were somewhat ambiguous. The change in pattern of the water column-borne PCBs provided further clues to their origin. Congener-specific analysis of the PCB water column concentration conducted for USEPA revealed a dramatic change as a result of passage through the Thompson Island Pool. Figure 2 presents the supporting data. It shows the mean summer load conditions at two stations from the TI Pool for 1993 and 1999. The effect of the sediment release is most readily observed during this season.

The results for the Rogers Island station, at the upstream end of the pool, represent the load primarily generated by the leakage from the GE Hudson Falls facility. The results for the TI Dam station show a significant increase in PCB load as a result of passage through the TI Pool. Further, the results showed a dramatic change in the homologue pattern, yielding a load at the TI Dam comprised of lighter homologue groups. The transformation and load gain are attributable to the release of partially dechlorinated PCBs from the sediments, a phenomenon that has been observed every summer from 1993 to 2000. Data obtained before this period were confounded by the large releases occurring at the GE Hudson Falls facility.

Figure 3. The magnitude of the release of PCBs is largely unchanged despite upstream controls Notes: 1. No data were available for Schuylerville before 1998. In 1999 and 2000 the load at Schuylerville was essential identical to that at Ti Dam. 2. The loads for 1991 are based on average loads from April to December.

Figure 3 illustrates the importance of the sediment contribution; it shows the annual loads in the Upper Hudson at several locations in the TI Pool area. These data, obtained by GE, clearly show the continued release of PCB from the sediments of the Upper Hudson despite the great reduction in the upstream load over the last 10 years. This is illustrated by the relatively constant load increase (the dark portion of each bar) for each year of sampling. The sediment contribution from the TI Pool represents more than 200 kg of PCB annually. Much of this contamination is transported downstream to the tidal Hudson.

The GE data have also been used to examine the seasonality of the PCB loads originating from the sediments. For the past 5 years, since the partial remediation of the leakages at the GE Hudson Falls facility, water column observations have indicated that about half of the sediment release of PCB occurred under low flow conditions, during late spring, summer and fall (see Figure 4). Although the mechanism for this process is not known, it is clearly not hydrologic, since maximum flow rates occur in March and April.

Figure 4. Nearly half of the 1999 sediment release of PCBs occurred under low-flow conditions (< 10,000 ft³/sec).. Click for larger image in new window.

More likely, these loads are biologically mediated, perhaps the result of sediment disturbances associated with the emergence of plants and invertebrates from the river bottom. This is also consistent with the cessation of PCB release in late fall which has been observed during some years—for example, 1997—when biological activity in the river is greatly reduced.

External sources

The sediments clearly represent the major source of PCBs under current conditions, but several issues remain concerning PCB contamination throughout the Lower Hudson (the 150 miles of river between Troy and New York City). Although the GE facilities were suspected of being the only significant source to the Upper Hudson, other sources could have contributed PCBs to the Lower Hudson. The Hudson watershed supports many industrial areas, any one of which could have been a significant source of PCBs. To investigate this issue, a series of sediment cores were obtained from locations throughout the watershed.

The use of sediment cores to examine historical transport of human-made contaminants is well established, particularly in the Hudson River (Bopp and Simpson 1989). The technique relies on the fact that suspended matter that settles from the water column in quiescent areas of the Hudson reflects the contaminant load carried by the river at the time the particles were deposited. In most areas of the Hudson sediment accumulates sporadically with episodes of both deposition and scour; some areas of the river, however, tend to collect sediments nearly continuously. In these areas, carefully collected sediment cores provide a record of the river's transport history.

As part of the sediment core collection study, cores were collected in 1992 from twenty-eight locations throughout the Hudson (Figure 1b). Each core was sliced into 2- to 4-cm slabs and analyzed for PCB congeners as well as other parameters. Typically fifteen samples were obtained from each core, with one core from each of the coring sites. Using radiocesium (¹³⁷Cs) as a time marker (Bopp and Simpson 1989), it is possible to establish the age of a given sediment layer and thus establish the history of river transport.

Several studies conducted in the 1980s documented the PCB dechlorination in the sediments of the Upper Hudson (for example, Brown et al. 1984). Therefore, PCB contamination within the sediments would not be readily identified using an Aroclor-based quantitation.

Additionally, it was recognized that PCB congeners have different geochemical properties and thus would redistribute themselves among sediments, water, and biota—tending to confound the original distribution. On this basis, a congener-specific analytical method was employed (TAMS et al. 1997) to quantitate individually the 126 to 145 PCB congeners in each sample. Thus, the variations in the congener mixtures could be used to identify PCB sources as well as important geochemical transformations. TAMS used the same technique on the water samples discussed above.

Figure 5. Congener loadings for principal Components 1 and 2 Hudson River sediment and water column samples.. Click for larger image in new window.

Using the PCB samples from the dated sediment cores, a principal components analysis (PCA) was performed on the PCB results for the sediment samples. The PCA was optimized for the most important congeners by selecting just congeners that contributed at least 0.01% to any of the first five principal components. This approach yielded fifty-six congeners for the purpose of the PCA. The loadings for the first two principal components are illustrated in Figure 5. These diagrams show the relative contributions of each congener to the two main components of variance in the data set. The components themselves are readily interpreted.

The first diagram in Figure 5, representing the first principal component, highlights the difference between the congeners BZ#1, 4, 8, 10, and 27 and the rest of the congeners present in the sample. In fact, this component closely approximates the ratio between the products of dechlorination and the rest of the congeners present.

These congeners originally represented only a small fraction of the original PCB mixtures that GE released. However, they are produced in the sediment under anaerobic conditions when sediment concentrations of PCB are high (generally greater than 30 mg/kg [TAMS et al. 1997]). Thus the first component is a measure of the extent of dechlorination in the PCB mixture. Highly dechlorinated samples will have a negative value for component 1; unaltered mixtures will have a positive value.

The second diagram shows an inverse relationship between congeners BZ#1 to BZ#82 and those greater than BZ#82. This split represents the boundary between congeners with four or fewer chlorine atoms on each molecule (mono- to tetra-chlorohomologues) and congeners with five or more chlorine atoms per molecule (penta- to deca-chlorohomologues). This component can be thought of as a measure of the molecular weight of the mixture. It is also approximates the proportion of Aroclor 1242 to 1254 present in the mixture. Higher fractions of Aroclor 1242 yield higher concentrations of congeners BZ#1 to BZ#82 and a positive value for this component. Higher fractions of Aroclor 1254 yield higher fractions of the congeners greater than BZ#82 and a negative value for this component.

Figure 6. Results of a principal components analysis for Hudson River sediments showing the importance of GE-related contamination.. Click for larger image in new window.

Figure 6 shows the data are plotted against the two principal components. Data in the diagram are annotated for the general areas of the Hudson they represent. The PCA serves to separate and identify clearly the tributary samples (labeled with capital letters) from the samples of the mainstem Hudson. On this basis, it is apparent that the tributaries do not contribute significantly to the PCB burden of the mainstem Hudson because their patterns of contamination are distinct from those of the Hudson.

The samples of the mainstem Hudson itself form an inverted arc with highly dechlorinated samples forming the left side and samples from the saline portion of the river forming the right side. At the top of the arc are the samples closest in pattern to that of the GE releases, a pattern documented by the collection of water samples just downstream of the facility. Its homologue pattern can be seen in the mean summer homologue pattern at Rogers Island (Figure 2).

For the entire freshwater Hudson (indicated by the open symbols on Figure 6), the pattern of contamination varies from that of freshly released material to that of a highly dechlorinated mixture (as noted by the similarity to Aroclor 1221). The near-linear trend between the GE source characteristics and those of a highly dechlorinated mixture indicate that no other source is present.

These results indicate that the GE source represents the major source of PCBs to the entire freshwater Hudson, a distance of 135 miles.

The right arc of the sediment distribution represents the saline portion of the river. In this area at least two sources are evident:

• That originating from GE
• A second from New York Harbor.

Samples from Newtown Creek, a receiving water body for a major sewage treatment plant, are used to characterize the New York City source. From this portion of the diagram, it is clear that PCB contamination in the saline portion of the Hudson is well represented as a mixture of the GE discharges and the New York discharges. Cores from New York City harbor contain both Newtown Creek and GE patterns as indicated in the PCA analysis, demonstrating that even in this region, nearly 200 miles downstream, the contribution from the GE releases is still evident.

Conclusions

The investigation of water column PCB concentrations and loads clearly identifies the sediments as the most significant source of PCBs to downstream regions of the Hudson. Sediment-based releases have been documented through both intensive studies and long-term monitoring. PCB congener analyses have added additional supporting evidence to identify the sediment as the main source to the water column.

The sediment inventory of the Upper Hudson developed over many years of PCB discharges from the GE facilities. This inventory now represents the main reservoir of contamination affecting regions downstream. Sediment releases to the water column appear largely unabated despite significant declines in the upstream discharges since 1993. Thus, in the absence of remedial measures dealing directly with the sediment, these releases are expected to continue. Given the continued release of PCBs from the TI Pool and their continued presence in the sediments throughout the Hudson, it is likely that, in the absence of remediation, there will be measurable and potentially unacceptable levels of PCBs in Hudson River fish for the foreseeable future.

Acknowledgments

As with any program of this size, there are many people who have contributed significantly to the work represented here. In addition to the contributions of the members of the authors' organizations, the authors are also grateful for important contributions by members of the USEPA, NYSDEC, and NOAA.

The opinions expressed here are those of the authors and do not necessarily reflect those of the USEPA or the U.S. Army Corps of Engineers. This work was paid for in part by the USEPA and U.S. Army Corps of Engineers under several contracts. Support of this work does not constitute an endorsement by the USEPA or by the U.S. Army Corps of Engineers.
____________

Edward Garvey, PhD, PG is a senior geochemist with TAMS. He has over 20 years of experience in geochemistry and contaminant transport analysis involving estuaries, rivers, lakes and ground water sites throughout the Eastern U.S. Phone 973-338-6680. Click here for additional information (opens new browser window).

Juliana Atmadja, ScD is an environmental engineer and a hydrogeologist at TAMS Consultants, specializing in transport modeling of chlorinated compounds and heavy metals in surface water and ground water settings.
Jonathan B. Butcher, PhD, is a professional hydrologist and a principal engineer with Tetra Tech, Inc. specializing in fate and transport modeling.

References

Further details on the analysis of the data discussed in this article can be found in TAMS et al. 1997, TAMS/TetraTech 1999, and TAMS 2000.