Evaluation of Dredging Effectiveness: What Has Experience Taught Us?
Over the last 20 years, various contaminated sediment sites have been remediated in whole or in part through dredging. The committee examined 26 dredging projects to evaluate whether dredging was able to meet short-term and long-term goals. Short-term goals are defined as cleanup levels that can be measured during or immediately post-dredging to verify the effective implementation of the remediation. The ability to maintain cleanup levels in the long term is ideally linked to the achievement of long-term risk-based goals or remedial action objectives. Appendix C presents the various sites’ cleanup levels and remedial action objectives and describes whether they were achieved at individual sites. Taken as a whole, the projects indicate what can and cannot be achieved with dredging and the conditions that favor or discourage the use of dredging.
Evidence that dredging projects led to the achievement of long-term remedial action objectives and did so within expected or projected time frames is generally lacking. It was often not possible to evaluate long-term remedy performance relative to remedial action objectives because of insufficient post-remediation data, quality, or availability or because of lack of an equivalent pre-remediation dataset. Post-remediation
conditions are always influenced by long-term natural attenuation processes and ongoing sources of contaminations if they exist, so long-term monitoring—over decades—may be needed to establish effectiveness; few of the sites reviewed by the committee have reached that level of maturity. Not counting the 5 pilot studies or hot-spot removal actions, about one half of the sites apparently did not achieve remedial action objectives or had inadequate monitoring to judge performance relative to remedial action objectives. Insufficient time has elapsed to judge achievement of remedial action objectives in approximately one quarter of the sites. The remaining sites apparently met remedial action objectives although the extent to which those remedial action objectives achieve long-term risk reduction may not be known.
There were often sufficient data to evaluate performance relative to cleanup levels or short-term implementation goals, but the relationship of these measures to long-term risk reduction was often not clear. An examination of Appendix C shows that many sites achieved cleanup levels; however, many were operational goals (mass removal or dredging to elevation) rather than contaminant-specific goals.
Natural processes are always modifying conditions at a site; their influence can be difficult or impossible to separate from the remedial action, particularly when control or reference sites are not monitored before and after remediation. Conditions also are often influenced by the implementation of combined remedies, such as dredging and capping, which complicate the assessment of the performance of dredging alone. Thus, the committee was unable to evaluate the effectiveness of dredging alone at most sites.
Experiences at the sites can nevertheless inform remedial project managers as to what may be achievable with dredging and what site and operational factors may limit dredging effectiveness or contribute to its success. Experience is especially useful in identifying factors that contribute to success or failure of dredging to meet short-term cleanup levels because monitoring has been conducted at most sites to judge performance relative to these standards. The ability to meet chemical-specific cleanup levels, however, does not in itself mean the ability to meet long-term risk-reduction targets or indicate the time frame over which any such targets might be met. This chapter discusses the lessons learned from sites where dredging was conducted and uses specific examples to
illustrate them. It also provides recommendations regarding implementation of successful remediation with dredging.
DATA AVAILABILITY AND ACCESSIBILITY
The potential utility of a review of remedial effectiveness is governed by the availability of pre-remediation and post-remediation monitoring data. That issue has three components: whether data were acquired at a site, whether the data are available for review, and whether the data are sufficient to support conclusions about effectiveness. The goal of acquiring data for this type of analysis appears relatively simple: collect and evaluate pre-remediation and post-remediation monitoring data on concentrations and effects from Superfund sites. However, obtaining this information is surprisingly difficult.1,2
The amount, frequency, and type of data collected at dredging sites are highly variable. Some of the earlier sites had very little post-dredging monitoring. For example, Bayou Bonfouca, LA, and Outboard Marine Corporation, IL, did not sample sediment concentrations immediately after dredging (see footnotes 1 and 2). Marathon Battery, NY, is an exception in that sediment and biota concentrations were collected and bioaccumulation was tested, but obtaining monitoring data proved difficult, requiring several iterations, and ultimately the committee could not access the full range of reports. At some more recent sites, dredging is supported and guided by chemical confirmation sampling, and the resulting data are accessible. For example, the U.S. Environmental Protection Agency (EPA) provided the committee concentration and location data on the recently completed dredging in the Grasse River, NY, and Hylebos Waterway in Commencement Bay, WA, including date, location coordinates, and chemical concentration information. Information on the operations and sampling and the monitoring results at pilot projects (such as conducted at the Grasse River, NY; Fox River, WI; and Lavaca Bay, TX, sites) are often well documented, as would be expected from studies specifically intended to document remedial effectiveness on a smaller scale.
Data and reports from remediation sites are often held by various entities (including EPA, consultants, states, and responsible parties), and
this complicated the compilation of information. Committee requests for data on sites were sent to Superfund Headquarters, which did not have them and forwarded the requests to project managers in the EPA region who were responsible for a particular site (although these managers may or may not have been in that position during the remediation). The data might not be held by the EPA region, but instead may reside with the contractors that performed the work or the responsible parties that funded the work. Some data at sites where remedial actions had been completed are archived or not readily retrievable. Thus, even when information was available to EPA it might have been inaccessible. In some cases, reports containing monitoring data and interpretations were held by the responsible parties but EPA wished not to have them released because sensitive negotiations were under way.
When reports and data were available, they may have been reproduced only on paper although they were originally produced electronically. Such conditions severely limit distribution and faithful replication of information, because many site documents rely on large-scale maps in color. The ability to access reports and data via public Web sites was generally extremely limited, but there were exceptions. The mid-Atlantic EPA Region 3 has each site’s administrative records on line, and this permits the public and researchers to access site files electronically (although typically these files are in a scanned, nonnative format).3 Public information is available on all Superfund sites via the CERCLIS database, which frequently contains a site’s record of decision and 5-year reviews, if available. It was presumed that a site’s 5-year review would contain explicit statements of the sampling that had been conducted and provide, at a minimum, concentration and location data on sampling, but the committee was surprised to see that that was not necessarily the case.1,2
Comments regarding the ready public accessibility of electronic data may seem trivial. However, pre-remediation and post-remediation information is the end result of massive planning, implementation, and
data collection efforts that typically have involved large expenditures of public resources (whether the expenditures result from remediation itself or from the establishment of agreements with responsible parties who conduct it). Provision of pre-remediation and post-remediation data on chemicals of concern in an accessible, intuitive manner that defines collection efforts and results is a prerequisite to reviewing and understanding the results of remedial projects.
An issue related to the availability (or lack) of complete sampling data is the need to rely on data summaries and various site reports to evaluate pre- and post-remediation results. At times the committee relied on reports and summaries that did not convey the necessary raw data to confirm summary statistics. That is because the committee did not have access to the primary sources or the resources to complete an ad-hoc reassembly and evaluation of all the information. A note of caution relevant to this and other studies that summarize site information is that data on concentrations and effects should be collected consistently over time (for example, from the same locations, media, depth interval, and developed with similar techniques and protocols) to be most useful. Summary statistics may not be derived from similar datasets and reports may have incomplete annotation on sample location (for example, the relation of the samples to the dredging footprint), sampling protocols, and chemical analyses. Over time, analytical methods, contractors, sampling locations, and sampling methods can change. These changes complicate pre- and post-remediation comparisons. When possible, the committee provides information on these issues.
Dredging to Remove Contaminant Mass
The direct effect of dredging is the removal of sediment and its associated contaminant mass. Experience at a variety of sites has shown that dredging is effective at removing contaminant mass. Where sediments are subject to scour by storm or other high-flow events, buried contaminated sediment may be the source of future exposure and risk. In such cases, mass removal may result in risk reduction because the future
exposure and transport of sediment have been thwarted (see Chapter 2 for additional discussion).
For example, a demonstration dredging project was conducted to remove a deposit (Deposit N) contaminated with polychlorinated biphenyls (PCBs) in a high-velocity reach of the Lower Fox River, WI, in 1998 and 1999. PCB contamination in sediments of the river is the result of historical wastewater discharge from the manufacture and recycling of carbonless copy paper incorporating Aroclor 1242. The objective of the Deposit N demonstration was to remove contaminated sediment and leave no more than 3-6 in. of residual material in place while minimizing resuspension and offsite loss of sediment. Dredging to target elevations in 1998 and 1999 resulted in the removal of 112 kg of PCBs, or 78% of the pre-dredging inventory (Foth and Van Dyke 2000). Mass removal may have been an appropriate cleanup objective if there was the potential for future mobilization and transport of the PCBs.
Simple mass removal, however, may not reduce risk. For example, the non-time-critical removal action conducted in 1995 in the Grasse River in Massena, NY, had the objective of removing much of the PCB mass that was in the vicinity of an outfall. PCBs had been in use at the adjacent Alcoa facility and were introduced into the river through the outfall and from other sources. It was estimated that this localized removal of about 2,500 cy removed 27% of the PCB mass from the entire study area, consisting of several miles of river (BBL 1995). Despite removal of as much as 98% of the targeted contaminant mass (Thibodeaux and Duckworth 1999), no measurable reduction in water-column or fish concentrations of PCBs was noted. Site characterization and assessment efforts have led to the conclusion that water-column PCB concentrations are related, at least during low-flow periods, to surficial sediment concentrations of PCBs throughout the river and that removal of buried mass does not have a major influence on water-column concentrations (Ortiz et al. 2004). The removal may still have been warranted to avoid potential scouring during high-flow conditions, but risk reduction was not achieved during base flow conditions.
Dredging to Reduce Risk
A more complete assessment of dredging effectiveness would include evaluation of long-term risk reduction in addition to mass removal
or performance relative to cleanup levels. Although few sites have sufficiently complete datasets, dredging has apparently resulted in risk reduction in some cases, including at some sites with long-term datasets.
Lake Jarnsjon, Sweden
The Lake Jarnsjon site was remediated in 1993 and 1994. The pre-dredging surface sediment (0-40 cm) PCB concentrations had a geometric mean of 5.0 mg/kg (n=12; range 0.4 to 30.7 mg/kg) in 1990. Following remediation, the surface sediment (0-20 cm) concentrations in 1994 were significantly reduced and had a geometric mean of 0.060 mg/kg (n=54; range 0.01 to 0.85 mg/kg) (Bremle et al 1998a). Out of 54 defined subareas, one exceeded 0.5 mg/kg dry weight (set as the highest acceptable level to be left in the sediment); 20% of the sediment areas had PCB levels higher than 0.2 mg/kg dry weight, the remediation objective was set at 25% (Bremle et al 1998b).
Fish-tissue PCB concentrations declined after remediation although post-dredging monitoring did not take place until 2 years after dredging. Concentrations did not, however, decrease to those upstream of the contaminated area. In their report, Bremle and Larsson (1998) compare fish concentrations in the remediated lake to fish in upstream areas and conclude that “fish from all the locations in 1996 had lower PCB concentration than in 1991 [dredging occurred in the summers of 1993 and 1994]. The most pronounced decrease was observed in the remediated lake, where levels in fish were halved. The main reason for the reduced levels was the remediation.” However, the authors also state that “the reason for the decline of PCB in fish could be decreased atmospheric deposition and thus lowered loadings of PCB to the freshwater.”
The comparisons in the study benefited from the use of a reference site that indicated background declines in fish-tissue concentrations; these declines have been seen elsewhere as well (Stow et al. 1995). In that regard, Bremle and Larsson (1998) state that
…the results show that if a remedial action is to be evaluated and the process is extended over several years, changes in background contamination must be taken into account. After a remedial action, the results need to be followed over several years to show if it has
been successful, which has not yet been the case in the present study. It also stresses the importance of using reference sites, to compare the results from the remedial area. A decrease in overall background contamination could otherwise well be interpreted as a result of the remedial action only.
Black River, Ohio
At the Black River, in Lorain, OH, sediment was contaminated with polycyclic aromatic hydrocarbons (PAHs) as a result of effluent from a steel-plant coking facility. In 1983, the coking facility closed. Dredging occurred from late 1989 to early 1990 below the Kobe Steel outfalls at river miles 2.83-3.55 (EPA 2007a). In the early 1980s, PAH compounds were detected in sediments at high concentrations, and the brown bullhead population had high rates of liver cancer and pre-cancerous lesions. Since closure of the facility and dredging, PAH concentrations in surface sediments, fish PAH residues, and neoplasm frequencies in fish have declined (Baumann 2000). As shown in Figure 4-1, a decrease in cancer at the site was noted immediately after the plant closure. An increase in cancer was also noted immediately after dredging and was probably due to the exposure of fish and their prey to higher concentrations of PAHs in sediment and water during dredging. Later sampling, however, showed decreases in cancer, suggesting that the increase during dredging was a short-term phenomena. Within 5 years after remediation, the cancer incidence was lower than the pre-dredging data, presumably as a result of the dredging. However, it is unclear to what extent continued natural attenuation, as evidenced by the reduction in observed cancer after plant closure but before dredging, could have reduced cancer incidence in the same time frame.
Marathon Battery, New York
The ability to achieve remedial action objectives and long-term risk reduction with dredging was demonstrated at Foundry Cove of the Marathon Battery, NY, site. Foundry Cove is a small body of water adjacent to the Hudson River about 85 km north of New York City. The
Marathon Battery Company discharged cadmium, nickel, and cobalt during the manufacture of batteries through the plant’s outfalls, located beneath the Cold Spring pier and in the East Foundry Cove Marsh. About 50 metric tons of nickel-cadmium waste is estimated to have been discharged from 1953 to 1971 (Levinton et al. 2006).
The site comprises six separate regions. West Foundry Cove borders the eastern shore of the Hudson River and is connected to East Foundry Cove via an opening in a railroad trestle. The most contaminated sites are East Foundry Cove Marsh (13 acres), East Foundry Cove (36 acres), East Foundry Pond (3 ac), and the Cold Spring Pier area (~5 acres) that borders the Hudson River to the north. Constitution Marsh (281 acres) is to the south and is less contaminated (see Figure 4-2). As summarized in the record of decision and summary to the committee, core sediment samples collected from East Foundry Cove during the remedial investigation ranged from 0.29 to 2700 mg/kg cadmium and had
a mean of 179.3 mg/kg (median = 5.6 mg/kg).4 Samples collected in the Pier area (a much larger area than what was actually dredged) ranged from 1.2 to 1,030 mg/kg for cadmium and had a mean of 12.6 mg/kg
(median = 3.9 mg/kg)5 (EPA 1989a; 2006a [Marathon Battery Superfund Site, May 10, 2006]). A human health risk assessment that considered fish and crab consumption and exposure to suspended sediment in water concluded that achievement of a sediment cadmium concentration of 220 mg/kg would be protective. Sediment bioassays indicated that 10-255 mg/kg would be protective of aquatic life. A cleanup level of 10 mg/kg was set and believed to be achievable by removing the top 1 ft of sediment (EPA 1989a). About 80,000 cy of sediment was dredged from the contaminated areas of East Foundry Cove, East Foundry Cove Pond, and the Cold Spring Pier area from 1993 to 1995. In contrast, East Foundry Cove Marsh was dry excavated to a 100 mg/kg limit and then capped with a bentonite and geotextile blanket followed by 1 ft of sandy marsh planting material. No active remediation was implemented in the Constitution Marsh6 or the West Foundry Cove (EPA 1995; 2006a [Marathon Battery Superfund Site, May 10, 2006]).
Post-dredging verification sampling was conducted to establish whether cleanup levels had been met. EPA states “In the Hudson River and East Foundry Cove, an average of 10 mg/kg cadmium remained, which was consistent with the ROD requirement that at least one foot of sediment and 95% of the contamination be removed” (EPA 1998a). The record of decision also required long-term monitoring at the site for thirty years after completion of the remedial action (AGC 2001). Figure 4-3 presents median cadmium concentrations7 at the East Foundry Cove portion of the site before and after remediation from this monitoring program. These data indicate that surficial sediment concentrations were reduced as a result of dredging, and the concentrations have not re-
These data are for all depths. The ROD describes these data as being from 85 locations covering 465 acres with cores down to a depth of 137 cm. This is a greater area than the approximate 5 acres that was dredged.
“Although cadmium-contaminated sediment hot spots were identified in Constitution Marsh, to remediate these sediments would have had a significant adverse impact on the marsh’s sensitive ecosystem. In addition, the cadmiumcontaminated sediments would eventually be covered with clean sediments following the remediation of the cadmium contaminated sediments in East Foundry Cove marsh. Therefore, long-term monitoring was selected for Constitution Marsh” (EPA 1995).
For information on the distribution of data, see Figure 4-4.
turned to pre-dredging levels. Note that although cleanup levels were achieved immediately after dredging, median concentrations have since fluctuated above and below the 10 mg/kg cleanup level. Figure 4-4 presents the frequency of occurrence of sediment cadmium concentrations in the period 1995-2000. The figure shows a reduction in surficial concentration and also shows that many individual sediment samples show concentrations greater than 100 mg/kg (log10 = 2). Concentrations in some sediment samples are indistinguishable from the original sediment concentration distribution. Independent sampling and analysis by Mackie et
al. (2007) had similar results. Their 2005 sampling of the dredged area of East Foundry Cove showed a median cadmium concentration of 39.2 mg/kg in the top 5 cm of sediment (16 samples ranging from 2.4 to 230.4 mg/kg, mean 59.7, SE 16.8).
The significance of occasional high residual concentrations after dredging can be evaluated by examination of ecologic exposure before and after dredging. Figure 4-5 is a summary of long-term monitoring data collected for 5 years post-dredging (AGC 2001) and shows the ratio of pre-remediation to post-remediation tissue concentrations in various plants and animals. The data most relevant to the dredging (which occurred only in East Foundry Cove, East Foundry Pond, and the Cold Spring Pier area) are the water chestnut in East Foundry Cove, which show improvement and the benthic invertebrates in East Foundry Cove, which show an increase after remediation and then a decrease to pre-remediation levels after 5 years.9 Bioaccumulation studies (using in-situ
enclosures) were also conducted at the site. Data from East Foundry Cove and the Cold Spring Pier area generally indicate declines in accumulated cadmium body burdens compared with pre-remediation values (AGC 2001; EPA 2003b).
The data suggest that in at least some cases dredging can achieve and maintain reductions in sediment concentrations and body burdens of contaminants although occasional measurements of elevated concentrations complicate the interpretation of the results. In addition, there may be short-term increases in body burdens in species directly affected by the remediation (such as the benthic organism data). The fact that individual sediment samples exhibited elevated concentrations emphasizes that evaluation of the performance of any remedy requires adequate monitoring of key indicators before and after remediation to fully characterize the distribution of concentrations so decisions are not driven by low probability events.
Dredging remains one of the few options available for the remediation of contaminated sediments and should be considered, along with other options, for managing the risks that they pose. The need to remove contaminated sediments can be particularly acute at sites where navigational channels need to be maintained or where buried contaminated sediment deposits are likely to be subjected to erosion and transport from high flows or changes in hydrologic conditions. As shown at the Grasse River and other sites, dredging can achieve removal of sediments and much of the contaminants they contain. Mass removal alone, however, may not achieve risk-based goals, which should be the basis for remedy selection.
The results at the Marathon Battery and other sites outlined above show that under some conditions dredging can achieve cleanup levels and aid recovery of biota at contaminated sediment sites. As indicated previously, it is often difficult to evaluate the effectiveness of dredging alone as a result of reductions in ongoing sources of contamination (for example, outfalls and atmospheric deposition), the use of combination remedies, and natural burial of existing contamination. As illustrated in the examples above, measurement of time trends in sediment and water contaminant concentrations prior to and after dredging can help identify changes due to dredging as well as evaluate the risks of not dredging. Reference sites are also useful in contrasting the contaminant dynamics and risk reduction due to dredging with that caused by these other processes. To assist in evaluating the performance of dredging, it is important to monitor a range of dredging performance metrics linked to risk reduction, as was done at Marathon Battery.
FACTORS AFFECTING DREDGING EFFECTIVENESS
A variety of factors and site characteristics influence dredging effectiveness and can limit the ability to achieve cleanup levels and remedial action objectives. However, it is generally not possible to definitively identify the specific conditions or factors that determined success or failure in a particular project. The committee strived to identify the condi-
tion or conditions that appeared to significantly contribute to the project’s outcome and summarized those herein.
Foremost among the factors that influence dredging effectiveness are whether a site has been adequately characterized and whether ongoing sources of contamination have been controlled. Site characterization and source control require a firm understanding of the nature and distribution of the contamination, any potential sources that contribute to the contamination burden in the watershed, and the processes influencing site risks and their attenuation. A strong understanding of the extent of contaminants in place and the contribution of outside sources is essential to developing an effective conceptual site model and remedial plan to eliminate or lessen contaminant exposures and risk. The influence of source control and site characterization on remedy effectiveness will be illustrated through experience at particular sites in later sections. These factors, however, influence the success of all sediment remediation efforts, not just dredging.
Destruction of the benthic community and removal of habitat is unavoidable with all dredging projects and represents an immediate negative effect to the existing benthic community. This effect also occurs with other active remedial efforts such as capping. As such, the ecologic benefit of the current habitat needs to be an important part of the decisions in determining whether or not to dredge. In a net risk reduction framework, the habitat destruction will be compared to the benefits of removing contaminated sediment, bearing in mind the post-dredging substrate’s desirability as a habitat (or the substrate created following backfilling).10 Recovery after disturbance is typically relatively rapid with estimates of benthic recovery rates ranging from several months to several years (Qian et al. 2003; USACE 2005). Immediately after destruction of the habitat, hardy, opportunistic organisms such as polychaetes (oligochaetes in freshwater) and small bivalves colonize surficial sediments. Subsequently the population increases in diversity and abun-
dance. Recovery occurs when the site returns to pre-disturbance conditions or does not differ significantly from a reference area.
In contrast to the above factors which affect many or all remedial approaches, dredging projects are specifically influenced by two additional processes that weigh heavily on the effectiveness of dredging:
Resuspension of sediments and release of contamination during dredging.
Generation of residual contamination giving rise to potential long term exposure after dredging.
As described in Chapter 2, resuspension, release, and residuals occur to some extent with all dredging projects. Resuspension refers to sediment that is disturbed during dredging and transported out of the dredging area. Exposure to resuspended sediments is generally transitory and ends soon after the completion of dredging. Residuals are the contaminated sediments exposed after conclusion of the dredging and can lead to longer term exposure and risks to organisms. Release of contaminants to the liquid phase (for example, solubilized PCBs) can occur from both resuspended and residual sediments. In the case of strongly sorbing contaminants, it is often assumed that the fraction of sediment resuspended corresponds to the fraction of contamination released and transported down current. Sediment resuspension and contaminant release, however, may not be closely related if there are large dissolved or separate phase releases or if release from residuals is substantial.
Figure 4-6 summarizes the amounts of resuspension and residual contamination that have been observed in a variety of dredging projects (Patmont 2006). The sediment resuspension data points are the fraction of sediment resuspended during dredging. (The fraction is the mass of suspended solids measured at some distance downstream of the dredge [typically less than 100 ft] divided by the mass dredged on a dry weight basis.) These data are from a variety of sources including consultant reports and the open literature. As such, sampling methods and approaches used to estimate these fractions can vary depending on the objectives of the study, the nature of the project, and site conditions. The residual data in the figure are the fraction of contaminant mass (not sediment mass) remaining post-dredging compared to the estimated
contaminant mass removed in the last production dredging cut at 11 sites (residual fractions are determined based on a chemical mass-balance approach [see Patmont and Palermo 2007]). The projects span a range of physical settings, operating conditions, and data collection methods. Despite variations among sources of the resuspension and residuals data, the distribution shown is a useful compendium of existing data and likely to indicate at least the magnitude of expected residuals and resuspension rates.
The figure suggests that about half the dredging projects have resulted in resuspension that amounts to 1% or less of the mass of sediments dredged.11 About half the dredging projects have resulted in a residual contaminant mass that amounts to 5% or less. Although the resuspension losses and residuals are small relative to the total mass
dredged, the availability of the contaminants to organisms may be higher than prior to dredging because of exposure to contaminants in the water column (resuspension) or at the sediment surface (residual). The residuals may be especially problematic in that the concentration in the residual is similar to the average concentration in the dredged sediment (Reible et al. 2003; Patmont 2006) and directly accessible to organisms that live at or interact with the sediment-water interface. Because of the presence of the residuals, surface concentrations may not change or may even increase when compared with pre-dredging conditions.
Patmont and Palermo (2007) used a similar (though not identical) dataset to investigate the influence of site-specific factors on residual contamination after dredging. In this analysis, they concluded that
Similar generated residual percentages were observed for both mechanical and hydraulic dredges. The available data suggest that multiple sources contribute to generated residuals, including resuspension, sloughing, and other factors. However, on a mass basis, sediment resuspension from the dredgehead appears to explain only a portion of the observed generated residuals, suggesting that other sources such as cut slope failure/sloughing are likely quantitatively more important. The available mass balance data also indicate that the presence of hardpan/bedrock, debris, and relatively low dry density sediment results in higher generated residuals.
Figure 4-7, from Patmont and Palermo, shows the influence of debris and/or hardpan and sediment bulk density on the estimated residual at 11 dredging project sites. (Figure 4-7 has one less case study than Figure 4-6.) Higher amounts of debris, the presence of hardpan, and low sediment bulk density all contribute to higher generated residuals.
An examination of dredging at the various sites included in Appendix C indicated that resuspension, release, and residuals can all be influenced by site conditions such as those discussed by Patmont and Palermo (2007) and by the manner in which dredging is implemented. The next section discusses the role that each of those may play in limiting dredging effectiveness on the basis of experience at specific sites where dredging was used; taken together, the experience illustrates site-specific conditions or activities that contribute to or limit dredging effectiveness.
The objective is to identify conditions under which dredging might be effectively implemented and conditions that could discourage the use of dredging because of its inability to meet desired cleanup levels or remedial action objectives.
Resuspension and Release of Dredged Contaminants During Removal
Resuspension and release of contaminants during dredging are among the most important adverse effects of dredging. As shown in Figure 4-6, up to 10% of the mass of sediment dredged can be resuspended during dredging. Early illustrations of the potential for dredging to give rise to at least short-term increases in adverse effects on organisms can be found in the previously discussed Black River, OH, and Marathon Battery, NY, sites. Those increases were probably due to exposure to more highly contaminated sediment or resuspension of sediment and contaminants during dredging. In the next two sections, the effect of sediment resuspension during dredging and the duration of those effects will be illustrated by experiences at other sites.
Effects of Resuspension and Release
Grasse River, New York
Monitoring of dissolved PCB concentrations in the water column during the 1995 non-time-critical removal in the Grasse River, Massena, NY indicated water concentrations above the 2 μg/L PCB action level at several time points and as high as 13.3 μg/L PCBs (BBL 1995).12 These concentrations were detected along the perimeter of the project, beyond three lines of silt curtains that were used in an effort to contain suspended sediment. The interior curtain was extended to the bottom, and, as stated in the documentation report (BBL 1995), “the lowering of both the boulder and the inner, secondary curtains to the River bottom greatly enhanced TSS [total suspended solids] containment.” This site also used caged fish before, during, and after dredging to monitor bioaccumulation of PCBs (BBL 1995). The study concluded that the increases in caged fish exposed for six weeks during dredging had increases in PCB concentrations 20 to 50 times higher than those observed in the pre-dredging time frame and increases of that magnitude suggest that uptake of PCBs was affected by the release of PCBs to the water column during dredging. However, the report also states that some of the increases in caged fish may be attributable to higher water temperatures during the dredging exposures.13
A second dredging project took place in the Grasse River in 2005. This demonstration project was intended to remove approximately 64,000 cy of PCB-contaminated sediment from 3 work zones, but ultimately removed about 24,600 cy of sediments from approximately 40% of the targeted area. Water column sampling during this demonstration project showed about one-fourth of the 100-odd measurements taken over the course of the project exceeded the 2 μg/L PCB action level in the most downstream sampling site adjacent to the silt curtain (EPA, unpublished data, 4/18/2006). All measured water column concentrations throughout the river prior to the demonstration dredging project were less than 0.065 μg/L total PCBs.
As shown in Figure 4-8, increases in water-column PCB concentration were noted downstream of the dredging operation. TSS increased in concert with PCBs at the sampling locations adjacent to the site, but downstream PCB concentrations remain high when TSS decreased back to upstream values. The presence of dissolved PCBs that would not settle may account for the different behavior downstream. Because dredging is a technology designed to remove solids, dissolved contaminants are contained much less effectively. Similar behavior would be expected for remedial dredging of sediments containing fluid contaminants, such as nonaqueous-phase liquids. In addition, operational controls on resuspension of sediment particles are expected to have less effect on dissolved-phase or fluid-phase contaminants. Silt curtains, for example, are designed to provide additional residence time to encourage particle settling and would have less influence on non-settleable contaminants.
The increase in suspended solids and their flux down river was not associated with high-flow events and therefore was likely due to dredging-related resuspension processes. A horizontal auger dredge was used at this site for most of the demonstration dredging. A horizontal auger of the size used is limited in its ability to dredge effectively in the presence of stone or debris 4 in. or larger. A separate debris-removal operation was implemented to eliminate larger stones and debris. As is typical, an open bucket was used to allow sediment captured with the debris and stone to redeposit on the bottom. Such operations can increase the entry
of suspended solids into the water column and thus increase the contaminant burden in the water column. Similar problems occur in the use of clamshell-bucket dredging of sites laden with debris. The occasional inability to close the bucket completely because of debris interference can increase resuspended solids and thus resuspension of contaminants.
A more detailed depiction of the PCB concentrations in water seen approximately 0.5 miles downstream of the dredge site is presented in Figure 4-9. The pre-dredging baseline PCB concentrations are low; during dredging activities, these concentrations generally increase and occa-
sionally exceed the 2 μg/L PCB action level; after dredging, concentrations decrease back to baseline concentrations (see Figure 4-9). It is estimated that during dredging activities about 3% of the PCBs removed from the river bottom were released downriver, largely as PCBs that had desorbed from resuspended sediments (Connolly et al. 2007).
The overall effect of resuspension and release during the dredging operation can be seen more clearly by examining PCB concentrations in fish at the Grasse River site. Figure 4-10 shows PCB concentrations in spottail shiner measured every fall from 1998 to 2006.
Spottail shiners are useful indicators because they forage only over a limited area (Becker 1983) and, being small, respond quickly to increases in PCB concentrations (Connolly et al. 2006). During 2005, the fish sampling coincided with dredging activities and PCB concentrations in the spottail shiner increased dramatically. The following year, PCB concentrations in shiners decreased to levels seen prior to dredging. There was a statistically significant increase in the downstream locations, but there is insufficient information to evaluate trends associated with dredging, because only a single post-dredging monitoring period was
available for PCB body-burden analysis (see Box 4-1). Additional data collection and detailed analyses of the Grasse River data are ongoing by EPA and Alcoa.
Duwamish Diagonal, Washington
In the Duwamish/Diagonal CSO, WA, early-action sediment dredging project, high PCB concentrations (above the pre-dredging surface concentrations) were found in sediments in and outside the dredge prism during a pre- and post-dredging sampling program (EcoChem Inc. 2005). During dredging, several complaints were logged about poor dredging practices that may have contributed to resuspension and release of contaminants. The dredged areas were capped, and to address the unexpected contamination created outside the dredge prism, 6-9 in. of sand was added to areas adjacent to the dredging area. Water quality monitoring during dredging indicated that turbidity standards were exceeded on several occasions (particularly in the first 2 weeks). Total PCBs and dissolved mercury (measured only during the first 8 days of dredging) were below water quality standards even at the highest turbidity values (EcoChem Inc. 2005).
As part of the overall RI/FS of the Lower Duwamish River Superfund site and other studies, fish tissue samples were collected both before and after the Duwamish/Diagonal dredging project. These data suggest that fish-tissue PCB concentrations were greater after dredging activities (Figure 4-11). While there is only one data point for the targeted species collected several years prior to dredging, other fish data collected in the project area in the years prior to the early action (that is, between 1992 and 2000) indicate that tissue levels remained steady during this period (J. Stern, King County Department of Natural Resources and Parks, personal commun., April 20, 2007). The exposure dynamics of resident fish are complex, so monitoring data alone are unable to directly implicate dredging as the cause of the apparent “spike” in tissue PCB concentrations. However, the timing of the increased fish concentrations, the rapid decrease after dredging, and corroborating fish bioaccumulation modeling (Patmont 2006; Stern and Patmont 2006; Stern et al. 2007) are suggestive of a dredging-related release.
Statistical Analysis of Fish PCB Body Burdens, Grasse River, New York
The data analyzed for this report (EPA, unpublished data, 4/18/2005) included percent lipid and total PCB concentrations in fish tissue for smallmouth bass (fillets), brown bullhead (fillets), and spottail shiner (whole body) sampled from the years 1993-2004 (pre-dredging), 2005 (during dredging) and 2006 (post-dredging). Brown bullheads and smallmouth bass were sampled from 4 areas: Background (upstream of dredging) and the Upper, Middle, and Lower stretch of the river (increasing distance downstream from dredging). Spottail shiners were sampled from the Background, Near Outfall 001, Near Unnamed Tributary, River Mouth areas (see Figure 4-10 legend for spottail shiner location details).
Temporal trends in fish tissue PCB concentrations and region specific effects were established based on linear regression models using monitoring year, percent lipid content, and sampling region as independent variables. PCB concentrations were centered upon their sampling region mean. The Box-Cox transformation (Box and Cox 1964) was parameterized in the regression model likelihoods to allow the data to optimally choose possible transformations. Analyses were stratified by fish species. Nonlinear trends in time were considered (Stow et al. 1995), the results of which led to interpretations that were qualitatively similar.
Different detection limits were reported for these data. Regression model inference was based on maximum likelihood treating the below detection limit data as left censored (Helsel 2005). (Data are left censored if their numeric value only indicates they are less than some given threshold such as the detection limit.) This method has been suggested as an alternative to substitution based techniques such as replacing non-detects with half their detection limit.
Fish tissue PCB concentrations were transformed using natural logarithm for all analyses. Lipid adjusted fish tissue PCBs sampled in the Background region were significantly lower than those sampled in other regions, smallmouth bass (p<0.01), brown bullhead (p<0.01), spottail shiner (p<0.01). Region specific significant decreasing temporal trends based on pre-dredging data (1993-2004) were established for all species.
We explored whether PCB concentrations during dredging (2005) were significantly greater than that expected from the temporal trend established on the pre-dredging data (1993-2004). The post-dredging (2006) data were also compared to the established pre-dredging temporal trend.
For the brown bullhead, the during dredging lipid adjusted average PCB concentrations were larger than the upper 95% limit of the time trend prediction interval for all regions. Results for the Background stretch region is interpreted with caution due to small sample size. For the smallmouth bass, the during dredging lipid adjusted average PCB concentrations were larger than the upper 95% limit of the time trend prediction interval for all regions, except for the Background stretch. For the spottail shiner, the during dredging lipid adjusted average PCB concentrations were larger than the upper 95% limit of the time trend prediction interval for “Mouth of river” and “Near unnamed tributary” areas. Concentrations during dredging in the background and “Near outfall” stretches were within these time trend based prediction limits. Due to small sample sizes results for spottail shiners are interpreted with caution.
The post-dredging (2006) lipid adjusted average PCB concentrations decreased significantly compared to those measured during dredging (2005) for all fish species and all sites, except the background regions for smallmouth bass and brown bullhead. The lipid adjusted fish tissue PCB concentrations post dredging (2006) were within the range (95% prediction intervals) predicted by the temporal trend established on the pre-dredging data (1993-2004) for all species and regions except for smallmouth bass in the lower and middle regions which remained above the 95% prediction intervals.
Additional post-dredging sampling points will be needed to evaluate long-term dredging effectiveness. Longitudinal monitoring (encompassing pre- and post-dredging time frames) can be used to statistically compare trends in contaminant concentrations before and after dredging and better associate changes with dredging.
Fox River, Wisconsin
A further illustration of the influence of sediment resuspension on dredging effectiveness can be found in various demonstration projects conducted at the Fox River, WI. At Deposit N, silt curtains were used in 1998 to contain any resuspended sediment, and downstream turbidity
was found to be no greater than that at upstream sampling locations. Upstream and downstream turbidity levels were also very similar in 1999, when silt curtains were not used (Foth and Van Dyke, 2000). Nevertheless, during 1998 dredging, water-column PCB concentrations averaged 11 ng/L downstream of dredging and 3.2 ng/L upstream. Prior to dredging, the average PCB water column concentrations were similar upstream and downstream of the dredging area. Similar increases (24 ng/L downstream vs 14 ng/L upstream) were observed in 1999. Overall, about 4% of the PCB mass removed from the deposit was released to the water column by dredging (Malcom-Pirnie and TAMS, 2004). The demonstration conducted in Sediment Management Units 56 and 57 (SMU 56/57) in 1999 and 2000 targeted a deposit of about 80,000 cy near the outfall of a recycling mill in Operable Unit (OU) 4, a relatively low-energy estuarine reach of the river. During dredging in 1999, in which
31,000 cy were removed, about 2.2% of the PCBs dredged were estimated to have been resuspended and transported downstream (Steuer 2000).
Direct estimates of contaminant release can rarely be used to guide day to day dredging operations because water column samples may take several days to analyze. Suspended-solids measurements are sometimes used to provide real-time feedback for the dredging operation. As indicated in Box 4-2, however, the correlation between suspended solids and even strongly sorbing contaminants, such as PCBs, may not be adequate to guide operations appropriately.
Correlations between Suspended Solids and Contaminant Concentrations
Although resuspension of sediment is largely viewed as the source of contaminant losses, any contaminant that partitions rapidly from the sediment to the water column will quickly cease to be related to resuspended-sediment concentrations. As shown by the Grasse River 2005 data (Figure 4-8), the water-column concentrations did not generally correlate with suspended-solids concentrations. Turbidity is typically used as a rapid surrogate measure of suspended solids and can be useful to indicate suspended solids if a site-specific correlation between the two quantities can be found. As suggested in the text, however, turbidity and contaminant concentration might not correlate. In an effort to understand the nature of the PCB releases from the Grasse River site, samples were taken from adjacent areas in and outside the silt-containment device surrounding the dredge area. Analysis indicated that although TSS concentrations were about 2 times greater inside the curtain, the dissolved-PCB concentrations were the same (Connolly et al. 2006). At the downstream sampling site, about 0.5 mile from the dredging area, about 75% of the PCBs was dissolved (this is operationally defined as passing a 0.45-μm filter).
At the Fox River SMU 56/57 sites, dredging-related releases resulted in an increase in downstream dissolved-PCB concentrations of about 59%. However, little or no difference in turbidity or TSS concentrations between upstream and downstream locations was detected during dredging, and turbidity and TSS did not correlate with water-column PCB concentrations (Steuer 2000). Those results indicate that turbidity and TSS were of little value as indicators of water-column PCB release. Dissolved and fluid contaminants, such as nonaqueous-phase liquids, will not be well characterized by TSS or turbidity monitoring. A good correlation between suspended solids and contaminant resuspension would be expected if the contaminant remained strongly associated with the solid phase.
Duration of Effects from Resuspension and Release
Increases in contaminant concentrations in the water column and fish during or immediately after dredging may be short-lived. As shown in the Black River, cancer prevalence at the site increased following remediation, but then declined dramatically shortly afterward (Figure 4-1) (Baumann 2000). At Marathon Battery, cadmium concentrations in benthic invertebrates increased following dredging, before declining again (Figure 4-5). In addition, PCB concentrations in the fish in the Grasse River declined substantially a year after concentrations spiked during dredging in 2005 (Figure 4-10).
There is also evidence that contaminant releases from dredging can be reduced. At the GM Massena project in the St. Lawrence River, a sheet-pile wall was erected around the dredging zone because silt curtains were unable to withstand the river currents. The interlocking steel sheet piling enclosed the area to reduce offsite migration of sediment (EPA 2005a). Sheet piling was possible, however, because isolation was required for only a small portion of the river bottom. During dredging, turbidity and PCB and PAH concentrations were measured downstream of the sheet piling to monitor for potential releases into the river. PAH results were consistently below detection limits, and sampling ceased after 19 days. PCBs were monitored over about 3 months. All samples at the monitoring stations were well below the action level of 2 μg/L; the maximum was 0.32 μg/L, and most of the samples were below the detection limit (BBL 1996).14
The examples cited indicate that resuspension and contaminant release during dredging can limit at least short-term dredging effectiveness. Large dredging projects that may continue for years or decades are more likely to exhibit more serious problems associated with resuspension and contaminant release than the projects of shorter duration that were examined. For example, dredging could continue for over 25 years
at New Bedford Harbor (Dickerson and Brown 2006) and thus any effects of resuspension would be at least of similar duration. There is no experience with such large projects, although some have been initiated (for example, New Bedford Harbor, MA, and Fox River, WI) or planned (for example, Hudson River, NY). The rate of recovery and time to achieve remedial goals after long-term exposure to remedial dredging are not known. Careful design of a monitoring program is needed to separate short-term from longer-term performance of a remedy.
Generation and Exposure of Residual Contamination
Potentially the most serious limitation to dredging effectiveness is residual contamination that is left after dredging. As described in Chapter 2, there are two general types of residuals: generated residuals, contaminated sediment that is resuspended during dredging and later redeposited; and undisturbed residuals, contaminated sediments found at the post-dredge sediment surface that have been uncovered but not fully removed as a result of the dredging operation (Bridges et al. in press). A portion of the generated residual may be unconsolidated and potentially more susceptible to transport. As such, this portion may not be accounted for by confirmation sampling conducted to define post-dredging residuals, depending upon the timing of that sampling. The presence of such residuals directly limits the ability to meet cleanup levels and may also reduce or eliminate opportunities to achieve long-term remedial action objectives. Findings from several of the studied sites on the extent of residual contamination after dredging are provided below.
Lavaca Bay, Texas
A dredging demonstration project was conducted in August 1998 to evaluate the use of a full-size hydraulic dredge to remove mercury-contaminated sediment near the outfall of a chloro-alkali manufacturing facility on the northwest shore of Lavaca Bay, TX.15 This hot-spot area
had the highest sediment mercury concentrations in Lavaca Bay, which has widespread mercury contamination. Six acres of very soft plastic clay sediment was dredged in 20 days; 2,300 lbs of mercury was removed with 60,000 to 80,000 cy of sediment, and placed in a confined disposal facility (Alcoa 2000). Extensive data on sediment contaminant concentrations (including post-dredging residual sediment in the dredge area), water quality (including TSS, turbidity and mercury concentrations), and mercury accumulation in caged oysters were collected at the site. Low TSS and insignificant mercury were mobilized beyond the curtained-off zone surrounding the dredge unit. Elevated mercury concentrations in oysters were within the range of those observed in oysters native to Lavaca Bay (Alcoa 2000).
The demonstration project is informative because of the efforts to control sediment residuals. Multiple passes of the dredging operation were conducted with sampling between passes to define the residual present after each pass. There was a notable increase in residual concentration between passes 2 and 3, as shown in Figure 4-12, apparently reflecting exposure of more highly contaminated sediment. Overall, the pass-to-pass concentration changes were not statistically significant, although analyses stratified by subarea did reveal some pass-to-pass concentration changes that were statistically significant (see Box 4-3 and Figure 4-12).
Grasse River, Massena, New York
At the previously discussed 1995 non-time-critical removal action in the Grasse River, Massena, NY, the average PCB concentrations in surficial sediments (upper 8 in.) were reduced by only 53% despite removal of as much as 98% of the PCB mass from the sediment column (Thibodeaux and Duckworth 1999). The site contains numerous rocks and boulders that contributed to residual contamination and contained high concentrations of PCBs near the bottom of the sediment column that could not feasibly be dredged, because of underlying bedrock and glacial till (hardpan) (see Box 4-4).
Statistical Analysis of Mercury Concentrations in Surficial Sediment, Lavaca Bay, Texas
Surficial sediment mercury concentrations in Lavaca Bay (Alcoa 2000) before any dredging and after four sequential dredging passes were analyzed. Sample locations were identified in four subareas: Capa, North Capa, AA, and the Trench Wall.
Statistical comparisons of surficial mercury concentrations were based on a linear regression model with indicators of dredge pass (before and after dredging passes 1, 2, 3, and 4) as the independent variables. The Box-Cox transformation (Box and Cox 1964) was parameterized in the regression-model likelihoods to allow possible transformations to be chosen optimally. Regression-model inference was based on maximum likelihood. Stratified analyses were performed for different subareas on the basis of the Kruskal-Wallis rank sum test (Hollander and Wolfe 1973) because the samples were small.
Figure 4-12 (left) displays mean surficial mercury concentrations prior to dredging and after each dredging pass and their corresponding 95% confidence intervals on the log base 10 scale. Mean surficial concentrations after each dredging pass were not significantly different from the means sampled before dredging. Figure 4-12 (right) displays the distribution of surficial mercury concentrations and corresponding sample sizes stratified by subarea and dredge pass. For Capa, mean surficial mercury concentrations before dredging after dredging passes 1 and 2 did not differ significantly (p = 0.28). For the Trench Wall, mean surficial mercury concentrations increased sequentially after dredging passes 1, 2, and 3, however these changes were not statistically significant. The mean concentration after dredging pass 4 was significantly lower than after dredging pass 3 for the Trench Wall (p = 0.04). Before dredging, mean surficial mercury concentrations differed significantly among the four subareas (p = 0.06).
Exploratory and followup stratified analysis suggested the geographic subareas of Capa, North Capa, AA, and the Trench Wall as a potential source of surficial mercury variation. However, sample sizes were insufficient to include this spatial effect in the regression model while considering variation across dredging times.
Description of the Substrate Topography of the Grasse River During Dredging
“The river bottom, we believe, contains boulders and rock outcrops that account for these features [seen in the side scan sonar images]. Soft sediment is intermixed with these features. The river was dredged in early 1900s using equipment and techniques that may have included blasting, which left a bottom littered with rocks and boulders, perhaps some outcrops and glacial till. Soft sediment began to settle on top of this bottom beginning in 1958 when the power canal ceased contributing flow to the river. All the accumulating soft sediment contains PCBs because PCBs were discharged from the late 1950s on, and this contaminated sediment fell in and amongst the rocks and boulders and finally covered them. So what we encountered is a bottom littered with rock debris intermixed with soft sediment below which is glacial till. The [horizontal auger dredge] captured some material but the productivity was very low because the auger couldn’t get down in between the rock debris, so I think that was part of the problem.”
Source: Connolly et al. 2006.
Immediately following dredging at the 2005 demonstration project at the Grasse River, residual surficial sediment PCB concentrations (0-3 in.) averaged 150 mg/kg (dry weight), compared to a pre-dredging average of 4.1 mg/kg (Connolly et al. 2006). The increase occurred despite the fact that more than 80% of the PCB mass in the dredging footprint was estimated to have been removed by the dredging operation. Residuals (generated and undisturbed) were measured at this site and an average of about 16 in. of contaminated sediments remained after dredging (range from 3 to 32 in.) (Connolly et al. 2006). Following dredging, the dredged area was capped with an average of 1.5 ft of a sand and topsoil mix. At this site, surficial sediment concentrations increased due to dredging, although there was a large removal of PCB mass from the river. The analysis illustrates that dredging can achieve substantial mass removal but may not reduce surficial sediment concentrations. The data were also analyzed for spatial correlation to identify and account for any form of spatial variation in surficial PCB concentrations or mass (see Box 4-5).
Statistical Analysis of Surficial Sediment PCB Concentrations, Grasse River, New York
PCB concentrations in sediment (mg/kg dry weight) from samples taken at three times—before dredging, after dredging, and after capping—and from two areas—Main Channel and Northern Near Shore—were analyzed (EPA, unpublished data, April 18, 2006). Longitude and latitude spatial coordinates for samples were also available.
Linear regression models with an indicator of monitoring time (pre-dredging, post-dredging, and post-capping) were used to statistically compare the sediment PCB concentrations before and after dredging and after capping. The spatial sample design (data coordinates) was not sufficiently consistent between times to consider a repeated-measures-based approach. The Box-Cox transformation (Box and Cox 1964) was parameterized in the regression-model likelihoods to allow possible transformations to be chosen optimally. Regression-model inference was based on maximum likelihood. Analyses were stratified by geographic region and considered PCB concentrations. The nonparametric Kruskal-Wallis Rank Sum Test (Hollander and Wolfe 1973) was used to compare surficial PCB concentrations from the Northern Near Shore because the sample sizes were small.
Figure 4-13 displays the region-specific distribution and sample sizes for log10 surficial PCB concentrations. For sediment in the Main Channel, there was a significant increase in average PCB concentrations after dredging (p <0.01) and post-capping concentrations were not significantly different from averaged pre-dredge concentrations. In the Northern Near Shore, the decline in average PCB concentrations between pre-dredging and post-dredging was not statistically significant. After capping the Northern Near Shore area, the average PCB concentration in surficial sediments was significantly lower than before dredging (p <0.01). Regression analyses were on the natural-logarithm-transformed data.
The geographic layout of sample locations was not sufficient to allow statistical models to identify and account for any form of spatial variation (either as a regression
trend or residual dependence) in PCB concentration. Residual spatial dependence is known as potentially biasing tests of significance (Cressie 1991). As described in this report, the entire main channel area was not able to be dredged. For the Main Channel analysis, pre-dredging, post-dredging, and post-capping data were compiled for the dredged area only (referred to as “extended work zone 1”). Pre-dredging data were compiled from three pre-dredging sampling periods for the dredged area (termed “Phase II,” “January 2004,” and “Pre-ROPS,” generally top 3 in.). The post-dredging (collected 10/22/2005; generally top 3 in.) and post-capping data (collected 11/28-29/2005; top 2 in.) from that area were also compiled. Northern Near Shore samples were compiled from pre-dredging (9/9/2004, top 3 in.), post-dredging (8/19/2005, top 3 in.), and post-capping (11/29/2005, generally top 2 in.) sampling efforts.
General Motors Central Foundry, Massena, New York
In 1995, General Motors dredged an area of about 10 acres in the St. Lawrence River near Massena, NY, that was contaminated with PCBs as a result of the release of hydraulic fluids. The action removed over 13,000 cy of sediment and over 99% of the PCB mass in the sediment (EPA 2005a). However, it did not meet the cleanup level of 1 mg/kg in all locations, because there was residual contamination after dredging. As described in the remedial action completion report (BBL 1996), boulders and debris were excavated mechanically, and sediment was removed later with a horizontal auger dredge. The contaminated sediment was underlain with dense glacial till that made it impossible to use overdredging to increase sediment-removal efficiency. In areas in which initial concentrations exceeded 500 mg/kg, 15-18 dredge passes were required to reduce sediment concentrations to below 500 mg/kg. In one area that initially exceeded 500 mg/kg, eight additional attempts, including multiple dredge passes, were conducted to reduce sediment concentrations. Ultimately, the contractor concluded, with EPA concurrence, that attainment of target cleanup levels in this quadrant was not possible with dredging alone, and capping was instituted (BBL 1996). Without capping, high residual PCB concentrations would have remained at the sediment surface and limited the effectiveness of the remediation.
Manistique Harbor, Michigan
The presence of debris and bedrock limited the effectiveness of the 1995 to 2000 dredging operations to remove PCB contaminated sediment at Manistique Harbor, MI (Nadeau 2006; Weston 2002). The primary remediation goal of the project was the long-term protection of Lake Michigan by removal of the potential PCB source in Manistique Harbor. A secondary goal was reducing risks to people and wildlife that consume fish from the harbor (EPA 1994). As described in Weston (2002), “Initially, the goal of the removal action was to remove sediments within the dredge area with total PCB mass concentrations of more than 10 ppm PCBs …The objectives of the removal action were further clarified and restated …that the “objective of 95% removal of the total PCB mass from within the AOC [Area of Concern] and an average concentration of not
more than 10 ppm throughout the sediment column within the AOC shall be verified.” Remediation of PCB contaminated sediments began in the fall of 1995 and was originally expected to continue through the fall of 1997 and remove a total of 104,000 cy of sediment (MIDEQ 1996). Ultimately, the project required 6 seasons of dredging from 1995-2000 and removed approximately 190,000 cy of contaminated sediment (EPA 2006a [Manistique River and Harbor Site, May 10, 2006]). The estimated mass of PCBs removed by the end of the project was 82-97% of the initial mass (Weston 2002). Information on pre-remediation surface sediment concentrations varies.16 Post-dredging average concentrations throughout the river and harbor (including dredged and non-dredged areas) reported for the top 1 ft were 9.0 mg/kg (sampled in 2000) and 7.3 mg/kg (sampled in 2001) (Weston 2002). Nadeau (2006) summarized the pre-
and post-dredging surface PCB concentrations as increasing slightly over initial surface concentrations (concentrations increased from 5.2 to 7.3 mg/kg in the whole area17 and 15.1 to 18.8 mg/kg in the dredged area18). Since completion of remediation, sediment from upstream has deposited in the harbor area, burying sediments with elevated PCB concentrations. A bathymetric analysis by EPA shows that in a five year period after dredging (between fall 2000 and fall 2005), approximately 83,000 cy of sediment deposited in the harbor from upstream sources, in some places between 10-16 ft deep (EPA 2006a [Manistique River and Harbor Site, May 10, 2006]). Surface sediment samples collected in 2004 (using a ponar dredge) had a mean concentration of 0.88 mg/kg PCBs in the area of interest (Weston 2005a). This example illustrates both the difficulty of eliminating residual sediment concentrations in the presence of debris and bedrock and the inability to achieve long-term risk reduction because of the residual unless other processes, such as sedimentation, intervene to reduce surficial sediment concentrations.
Cumberland Bay, Lake Champlain, Plattsburgh, New York
Debris and a heterogeneous substrate caused dredging problems at the PCB-contaminated Cumberland Bay, NY, site, where logs, wood chips, and large rocks were encountered. Dredging began in July 1999 and ended in December 2000. After the initial dredging of the 34-acre site, divers found many areas where PCB removal was incomplete, apparently because of the presence of debris. As described by the New York State Department of Environmental Conservation (NYSDEC 2001), one area originally thought to have been dredged to a hard bottom was found by divers to be a hard crust that covered 4 ft of sludge containing PCBs at 54 ppm. Hand-held hydraulic dredge lines were used by divers to remove contaminated sediment from areas where dredging with only the large hydraulic dredge was difficult. Another difficulty encountered near a dock area was the bubbling up of gas during dredging that floated sludge to the surface. On the completion of dredging, residual PCB con-
centrations averaged 6.8 mg/kg on the basis of analyses of 51 samples taken from 42 cores. That is lower than pre-remediation PCB concentrations. A dock area had previously averaged 430 mg/kg, with maximum of 13,000 mg/kg, and other dredged areas had averaged 33 mg/kg. However, the reduction in risk cannot be quantified, because no risk-based numeric remediation goals were selected for the site.
Fox River, Wisconsin
Residuals were noted at three dredging projects on the Fox River. At the 1998-1999 removal at Deposit N, sediment rested on a fractured bedrock surface, so it was not possible for a dredge to cut into a clean underlying layer. Sediment PCB concentrations after dredging averaged 14 mg/kg, similar to the average pre-dredging concentration of 16 mg/kg. It is estimated that of the pre-project 142 lb of PCBs measured at Deposit N, 111 lb was removed, and about 31 lb remained in the residual sediment on the completion of the project (Foth and Van Dyke 2000).
At the 1999 Fox River SMU 56/57 demonstration dredging project, steep side slopes, debris, and underlying clay made it difficult to remove contaminated residuals. Final cleanup dredging passes were performed in four subareas before termination of the 1999 dredging, and post-dredging surface concentrations in three of the four areas were less than pre-dredging concentrations, although the 1-mg/kg target was not generally achieved even with cleanup passes. The overall post-dredging average surficial sediment PCB concentration at the end of 1999 dredging was 73 mg/kg, compared with 4.4 mg/kg before dredging (Montgomery Watson 2001). The surficial sediment concentration in the areas not subject to overdredging exhibited an average concentration of 116 mg/kg and a median of 45 mg/kg, very close to the initial 53 mg/kg average concentration in the deposit (Reible et al. 2003). Dredging of the remaining volume was completed by the responsible party as a removal action in 2000 and achieved an average surficial sediment PCB concentration of 2.6 mg/kg; this was followed by backfilling with a minimum of 6 in. of clean sand.19 The outcome exceeded closure requirements for the re-
moval action, and surficial concentrations after the incorporation of backfill were not measured (Fort James Corporation et al. 2001).
Early reports of the 2005 full-scale dredging of the upper reaches of the Fox River have indicated problems in achieving cleanup targets (Fox et al. 2006). The reports indicate that dredging did not remove all sediment with PCB exceeding 1 mg/kg and that sand backfilling will be necessary to meet the 0.25-mg/kg surface-area-weighted average concentration end point. High-concentration deposits in thin soft sediment layers overlying stiff clay have made residual contamination difficult to remove. A pilot test conducted in a portion of the dredged area indicated strongly diminishing returns for redredging: doubling the volume removed with the goal of removing all soft sediment above native clay was not sufficient to meet the remedial goal of PCB at 1 mg/kg. Dredging results available at the time of the study were from three subunits of OU 1 (Subunit A, C/D2S, and POG1). The preliminary pre-remediation and post-remediation results—PCB mass removal and surficial (upper 4 in.) PCB concentration—from verification sampling are presented in Table 4-1.
More recently, a specialized dredge (a cutter-less head suction dredge) developed by the remedial contractor was used during the 2006 dredging at the site. The dredge was designed for very thin deposits of sediments over a clay or hard till bottom (including generated residuals). Preliminary results for three dredge management units (about 2 acres combined) show that concentrations well under the 1 mg/kg remedial action level were attained even when initial concentrations ranged from 20 mg/kg to above 50 mg/kg (Green et al. 2007). The conditions (thin deposits over clay or hard till bottom) are considered to be among the most difficult for attaining target cleanup levels.
Commencement Bay, Washington
The combination of the ability to overdredge into clean sediment and the presence of sediment that has minimal debris or other obstacles
TABLE 4-1 Summary of Pre-dredging and Post-dredging Verification Sampling Results (2005) from Three Subunits of Operable Unit 1 in the Lower Fox River
to dredging has led to more manageable residual concentrations during the cleanup of several waterways in Commencement Bay, Tacoma, WA. The 1993-1994 Sitcum Waterway cleanup in Commencement Bay was combined with a redevelopment project by the Port of Tacoma designed to create a capacity to handle deep-draft vessels in its facility. As a result of the desire to increase navigable depth, the dredging plan included removal of sediment to a bottom elevation that exceeded the depth of contamination in open-water areas by at least 2 ft. The ability to overdredge facilitated removal of contamination in those areas and helped to reduce the impact of contaminated residuals on final sediment quality. Immediately after dredging in 1994, sediment quality objectives (SQOs) had not been achieved in all areas. An additional 2 ft of sediment was dredged from one of the areas, and sampling indicated that concentrations in the area were below the SQOs. In the other areas above the SQOs, natural recovery was determined to be sufficient to meet the remedial action objectives; these areas achieved SQOs in 2003. In 2004, EPA approved the Port of Tacoma's request to end further sediment monitoring (EPA 2006a [Commencement Bay–Sitcum Waterway, April 26, 2006]).
The 2004-2006 Head of Hylebos dredging project also indicated the effectiveness of overdredging to reduce residual contamination. The Hylebos Waterway was originally cut into a broad river delta consisting of native sediments composed of clean and fairly compact silts and sands. After the waterway was established, industry developed along the waterway, and this resulted in industrial-chemical discharges into it. No river was feeding the waterway, so it slowly shoaled in with very fine-grained sediment in the form of “soft black muck” over the natural or native sediment. Characterization of subsurface sediment with core samples showed that the contaminants from the industrial discharges were restricted to the fine-grained surface sediment, whereas the immediately underlying native sediment was not contaminated (Dalton, Olmsted & Fuglevand, Inc. 2006). There was a clear visual difference between the contaminated sediment and the underlying native sediment (compact silts and sands). During dredging, each bucket of material was examined visually by onboard inspectors to ensure that all fine-grained sediment had been removed before moving on to the next area. In that manner, overdredging into clean sediment could indicate that residual contamination was minimal. The ability to differentiate clearly between contaminated and uncontaminated sediment, dredging into the uncontaminated sediment, and the relative lack of debris combined to minimize the residuals.
Pre-dredging (dates unspecified) and post-dredging (August 2004-January 2006) surficial-sediment (top 10 cm) total PCB concentrations from the Head of Hylebos and a few samples identified as post-capping (January 2006) samples were available for analysis (unpublished data; Paul Fuglevand; Dalton, Olmsted & Fuglevand, Inc.; August 22, 2006). Linear regression with an indicator of monitoring time (pre-dredging or post-dredging) was used to statistically compare the sediment PCB concentrations before and after dredging. There was a significant decrease in PCB concentrations in surficial sediment after dredging (p <0.01); the pre-dredging mean was 685.9 μg/kg dry weight (n = 135), and the post-dredging mean was 74.7 μg/kg dry weight (n = 400). Only six samples were identified as post-capping samples, and they ranged in concentration from 36.4 to 847.0 μg/kg dry weight. This site remains one of the few where cleanup levels were obtained by dredging alone (except in a few areas). As stated by EPA, “dredging to expose clean native sediment was successfully completed throughout the entire project area, with the ex-
ception of an under-dock cap (completed in 1998) and a shoreline subtidal cap completed early this year at a location of groundwater discharge with elevated arsenic concentrations. The sediment remediation project successfully achieved the project SQOs with no residual sediments exceeding the SQOs (except at the two noted capping areas which were not driven by generated residuals from dredging)” (EPA 2006a [Commencement Bay - Head of Hylebos, May 17, 2006]). The comparatively low initial concentrations in the Hylebos Waterway (and many of the Puget Sound sites) and the smaller magnitude of difference between contaminated sediment concentrations and cleanup levels decrease the potential effect of dredging residuals on achieving cleanup levels. However, generated residuals are derived at least partly from the contaminated material being removed so residuals management remains a critical issue at Pacific Northwest sediment sites.
Harbor Island, Duwamish River, Washington
Harbor Island is another site in the Puget Sound area whose stratigraphy is conducive to dredging (a clear separation between contaminated and native, largely uncontaminated sediment). Residual contamination after dredging may be more important than observed at the Head of Hylebos or Sitcum waterways, because of extensive debris. The Lockheed Shipyard is on the eastern bank of the West Waterway in the lower Duwamish River in the Harbor Island Superfund site. The site was the location of a bridge-building company and then a ship-building and maintenance facility. At the time of remediation, the site comprised a failing bulkheaded shoreline; almost the entire nearshore area was covered by docks or marine railways, and an open-water area was immediately adjacent to the federal shipping channel. Extensive debris was present in the underpier and open-water area immediately adjacent to the pier face. Surface debris included consolidated machine turnings and other metal debris, cable, concrete blocks, and wood. Much of the site was covered with thousands of deeply embedded creosote piles that were slated for removal as part of the remedy. The remedy selected for the site included dredging and capping of the nearshore area (not all contaminated sediment would be removed) and removal of the contaminated sediment layer in the open-water area including debris that might
limit dredging effectiveness. Habitat restoration was a major component of the remedy (EPA 2006a [Lockheed Shipyard Sediment OU, May 12, 2006]). Additional subsurface debris was encountered once dredging began, including large concrete pier blocks, broken piles, and the original willow cribbing that was used to contain dredged material during the construction of the West Waterway and Harbor Island around 1900. Debris removal had an important effect on the duration, timing, and cost of the project, and the remedy had to be implemented in two phases over two seasons (because of restricted in-water work periods for the protection of endangered species in the Duwamish River). Phase 1 consisted of pier and railway demolition, bulkhead replacement, and initial debris removal (by dredging). Dredging to complete the debris removal and achieve cleanup levels was conducted as phase 2 from November 2003 to March 2004 and from October 2004 to November 2004 (EPA 2006a [Lockheed Shipyard Sediment OU, May 12, 2006]). Because cleanup levels were not achieved after the first phase of dredging, a thin layer of sediment (6-12 in.) was placed over the dredged area to stabilize the residuals until the next season of dredging, when it was removed as part of the dredged inventory.
The available project data indicates that sediment resuspension and the generation of residuals represent a nearly universal problem in connection with dredging of contaminated sites. Resuspension can be more of a problem in the presence of debris or other site conditions that interfere with normal dredging operations. In addition, readily desorbable contaminants and fluid contaminants, such as nonaqueous-phase liquids, are unlikely to be effectively captured by the dredge or by common operational controls on resuspension. Nor will such contaminants be adequately characterized by measuring the suspended solids, such as TSS or turbidity.
Low sediment bulk density and the presence of debris and hardpan or bedrock all tend to increase resuspension and residuals. Available data indicate that dredging is most likely to be successful when dredges penetrate into clean sediment layers reducing the amount of generated residuals. At sites where structures, debris, hardpan, or bedrock limit
dredging effectiveness, the desired cleanup levels, if based on the attainment of specified chemical concentrations, are unlikely to be met by dredging alone. The inability to attain cleanup levels would presumably translate into an inability to meet both short-term and long-term remedial goals and objectives.
Resuspension appears to result in at least short-term negative impacts on water quality and organisms. Residuals may give rise to longer term negative impacts, but at most sites, there has been insufficient monitoring to evaluate the long-term impact of residuals or capping has been used to manage residuals.
MANAGEMENT OF DESIGN AND IMPLEMENTATION TO MAXIMIZE DREDGING EFFECTIVENESS
Although the factors affecting dredging effectiveness outlined in the preceding section are operative at all sites, their influence can be minimized, although not eliminated, by active management of the dredging process, that is, managing design and implementation to maximize effectiveness. Through experience gained at dredging sites, a number of actions have been identified that can help to maximize the effectiveness of dredging in particular situations. However, that experience also suggests that successfully overcoming the limitations of dredging requires both site conditions conducive to dredging and the implementation of some or all of those actions. Sites that exhibit extensive debris, hardpan or bedrock immediately below contamination, or other factors that limit the ability to control residuals or resuspension will continue to be problematic for dredging even if all the actions discussed below are implemented.
Ensure Adequate Site Characterization
Central to the successful implementation of any remedial action is site characterization sufficient to define a conceptual site model. A comprehensive conceptual site model should define the contaminants of concern at a site, the spatial distribution of contamination, the processes that describe the change in contamination over time, the human and ecologic exposure routes, and the significance of exposure and risk. Only when
these aspects of the model are developed can a remedial effort be designed to respond to risk appropriately and achieve remedial goals. Adequate site characterization can identify potential sources of contamination and provide the data necessary to design an effective remedial program.
At the Reynolds Metals Superfund site on the St. Lawrence River near Massena, NY, pre-dredging site characterization was not adequate to delineate the distribution of the chemicals of concern at the site. Dredge design was based on the assumption that the PCBs were collocated with the other chemicals of concern, PAHs, and total dibenzofurans. However, post-dredging sampling indicated that this was not the case (EPA 2006c). Following dredging, which included redredging several of the areas, it was determined that PCBs were not collocated with PAHs and that about one-third of the 22 acre dredged area contained PAH concentrations above the cleanup level (EPA 2006c).20 Future remedial activities at this site are currently being decided.
The Head of the Hylebos Waterway was adequately characterized prior to dredging. Historical surface and core samples were used in conjunction with planned studies to determine the horizontal and vertical distribution of contaminants. As described in the remedial action construction report (Dalton, Olmsted & Fuglevand, Inc. 2006), the historical U.S. Army Corps of Engineers (USACE) post-dredging surveys were used to map the interface between the soft black muck and the native bed sediments and to refine the dredge plan. However, core samples were used to confirm the interfaces, and care was taken not to composite core samples across the muck-native interface. Over 100 cores and over 500 surface samples were used to delineate the area and depth for remediation in this approximately 45 acre site. The coring studies were also used to establish that the recent and native sediments were physically, visually, and chemically different from each other and that chemical ex-
ceedances of the SQOs were only found in the recent sediments (EPA 2006a [Head of Hylebos Waterway, Commencement Bay, May 17, 2006]). During implementation, the designers viewed the deepest historical dredging as a general guide but relied on observations during dredging to establish successful removal of the impacted sediment.
The design of a dredging plan requires interpolation of depth-of-contamination data from sediment core samples. The upstream portions of the Fox River (OU 1), where dredging is currently being conducted, is a challenging site in that regard because it consists of multiple discrete contaminated sediment deposits arising out of local differences in flow regime and relationship to contaminant sources. The method applied by the remedial design team in OU 1 was to develop deterministic interpolations of sediment PCB concentrations for each deposit and then to connect the interpolations at deposit boundaries (CH2M Hill 2005). When that method is applied, the result is a surface of predicted depth of contamination, which can be expected to be most accurate in the neighborhoods of samples used in the interpolation and most uncertain in un-sampled locations and at boundaries of deposits.21 The spatial density of cores is a key component of adequately characterizing sediment distribution (particularly cores that penetrate through the entire deposit). However, there is no single optimum spacing between core samples because the necessary density depends on the heterogeneity of the site deposit and is site specific.
Accurate characterization is particularly challenging in areas where contaminated sediments are underlain by uneven sub-bottom (for example, furrows, gulleys, or depressions). In these areas, deposition of contaminated sediment over time will often fill in low spots to create a relatively flat sediment-water interface, but with marked differences in the underlying depth of contamination (for example, see description in Box 4-5). At the afore-mentioned Cumberland Bay site, variations in the uncontaminated sub-bottom characteristics and topography proved difficult to characterize prior to dredging. According to the NYSDEC, “site characterization and pre-design studies included bathymetric surveys
and sludge probing to define the top of the sludge bed, its thickness, and horizontal extent. In addition to sludge coring, divers confirmed the outer extent of the sludge bed in areas where it was too thin to measure by coring.” However, following dredging in 1999, it was determined that in one area “originally believed to have been dredged to a hard bottom since the sampling device encountered refusal” the sediment “consisted of a hard crust underlain by up to four feet of [contaminated] sludge.” In another area, PCB-contaminated sludge was found in 1-6 ft deep depressions scattered along the bottom of the lake following dredging. Further dredging targeted both of these contaminated areas (NYSDEC 2001).
In the previously described 2005 pilot study in the Grasse River, it also proved difficult to accurately define the thickness of contaminated sediments using available sampling techniques and protocols. Prior to dredging, multibeam bathymetry, sediment probing, sediment coring, and acoustic sub-bottom profiling were used to characterize the site. Depth to hard bottom was estimated using sediment probing on a 25-ft by 25-ft grid with PCBs expected to be present in sediments above the hard bottom (Connolly et al. 2007). Following dredging, vibracore sampling indicated that in some areas the estimated thickness of contaminated sediments was wrong and significant contaminated sediment remained (see Figure 4-14). These results indicated that at this site “sediment probing is not a reliable indicator of the depth of sediment and manual pushcore sample collection is not a reliable indicator of the full depth of PCBs contaminated material and below the deepest contaminated layer” (EPA 2006a [Grasse River Site, April 18, 2006]). Acoustic subsurface sampling at this site was also not successful for detailing sub-bottom characteristics.22
A similar situation existed at the Manistique River and Harbor site. During site characterization, the samples taken before the dredging were
thought to be taken to bedrock, but were not. This is apparently because wood debris under the sediments was thought to be the bedrock harbor bottom (EPA 2006a [Manistique River and Harbor Site, May 10, 2006]). At this a site, a subbottom profiling device was used to estimate sediment thickness. However, the wood pulp and debris in the sediments contained large amounts of gases that rendered the subbottom profiling device useless. As a result, EPA indicated that the dredging depth could not be predetermined in each area (EPA 2006a [Manistique River and Harbor Site, May 10, 2006]).
The influence that incomplete characterization has on the success of dredging points to the importance of accurate site characterization. However, characterization activities are resource intensive and can consume time and funds otherwise available for remedial activities. As a result, decisions on whether to proceed with further characterization should seek to ascertain whether additional characterization will benefit remedial effectiveness and the point at which the additional efforts provide diminishing returns. These considerations will be site-specific. Obviously, areas with complex and heterogenous sub-bottom (such as the
Grasse River) will benefit from greater characterization than those with less heterogeneity. The variety of subsurface characteristics at these sites also indicates that the most useful characterization technologies will be site specific. Sediment sampling coupled with an understanding of the fluvial and geologic nature of an area will shed light on the attributes of the sediment deposit, but not all site conditions can be completely understood prior to beginning work. As a result, verification samples and progress cores taken during dredging are useful for indicating whether operations are succeeding or modifications need to be made (see Chapter 5 for further discussion).
Defining ongoing sources of contaminants to the waterway through site characterization is of critical importance for determining appropriate cleanup responses and for eliminating recontamination of remediated areas. This issue is addressed further in the next section.
Implement Source Control
As pointed out by the National Research Council Committee on Remediation of PCB-Contaminated Sediments, “the identification and adequate control of sources of PCB releases should be an essential early step in the site risk management” (NRC 2001). If contaminant sources are not controlled, dredging cannot be effective in managing risk. In the Hylebos Waterway, the combined efforts of EPA and the Washington State Department of Ecology to achieve source control before dredging contributed to the ability to implement successful remedies. Dredging of the Hylebos Waterway was also at least partially successful because of the efforts to directly address inaccessible areas that could not be dredged. Much of the shoreline of the waterway is modified with over-water structures, such as docks, piers, and wharves. The remedy selected for the head of the waterway included dredging accessible areas, excavation or capping in isolated intertidal and under-pier areas, and natural recovery. Without control of the contaminated sediments under piers and along the shoreline, the project would likely not have been considered successful. Dredging beneath the Arkema dock during 2005 used a long-reach excavator to remove most of the impacted sediment followed by a diver-deployed hydraulic dredge to remove the loose residual material that accumulated during mechanical dredging. The 2005 dredging
activities achieved the SQO cleanup objectives beneath both the Ace Tank and Arkema docks. Some areas along the shore were inaccessible to dredging and were either capped or determined to be suitable for monitored natural recovery.
In contrast, both potentially important source areas and exposed sediment contributing to risk were missed, or not appropriately evaluated, in the characterization of the Lauritzen Channel at the United Heckathorn site. Dredging did not achieve remedial action objectives at this site despite achieving cleanup goals immediately after dredging for DDT, the contaminant of concern—apparently because of a failure to address contaminant sources or mass contributing to exposure and risk at the site. Confirmation sampling after dredging seemed to confirm the cleanup goal of 0.59 mg/kg in the Lauritzen Channel of the site, but an investigation a year later, in 1998, found DDT concentrations as high as 30.1 mg/kg. Year 1 biomonitoring showed that pesticide concentrations in the tissues of mussels exposed at the site were higher than those observed before remediation; these values decreased slightly in later years. Anderson et al. (2000) noted remaining toxicity in amphipods after dredging. In 2002, a buried outfall visible during low tide was identified as a persistent source of DDT in the channel; it was plugged by EPA in 2003. Investigations in 2002 and 2003 found sediment concentrations greater than 1,000 mg/kg in the vicinity of a dock on the eastern side of the channel. Remedial objectives were not met in the Lauritzen Channel for a number of reasons, including the decision not to dredge side slopes completely or to remove material from under piers. Some of those areas were intended to be capped with sand, but because of the steepness of slopes or inaccessibility, this was not done (Chemical Waste Management 1997).
A recent analysis of recontamination of completed sediment remedies based on publicly available reports, such as 5 Year Reviews, indicated that 20 areas where dredging or capping remedies had been completed were recontaminated from outside sources, primarily by combined sewer outfalls (CSOs), unremediated upland areas, and adjacent and upstream unremediated areas (Nadeau and Skaggs 2007). The potential for recontamination at a site underscores the importance of identifying and controlling sources before undertaking a sediment remedy.
Monitor Appropriate Indicators of Effectiveness
Adequate site characterization should provide an understanding of the key sources of exposure and risk at a site and of how to intervene effectively to control risk. It should allow the definition of appropriate remedial action objectives and of cleanup levels that will lead to their achievement. It should also therefore identify appropriate indicators of successful implementation and, ultimately, effectiveness. A baseline of appropriate indicators of effectiveness must be established before dredging to make possible a comparison with post-dredging data. Monitoring should continue until effectiveness can be evaluated.
At the Outboard Marine Corporation—Waukegan Harbor site, success of the dredging remedy was monitored by using, among other indicators, fish-tissue data. The site is in Lake County, IL, 50 miles north of Chicago, and consists of industrial, commercial, municipal, and open or vacant lands at the mouth of the Waukegan River and North Ditch drainage basins. It is estimated that 300,000 lb of PCBs (Aroclors 1242 and 1248) were released into the harbor and that sediment concentrations were up to 25,000 ppm (EPA 2000a). The harbor was dredged in 1992 and contaminated sediments were placed in an abandoned boat slip. Only areas exceeding 50 ppm were remediated; as a result, some areas in the harbor are expected to have relatively high residual concentrations, and further remediation is being considered (EPA 2002; EPA 2007b).
Fish-tissue data from Waukegan Harbor were analyzed to evaluate the hypothesis that PCB contamination has decreased (see Box 4-6). To evaluate dredging effectiveness accurately on the basis of pre-remediation and post-remediation fish body burdens, data sufficient to estimate both a pre-dredging time trend and a post-dredging time trend are needed from representative samples of fish collected from exposure areas that are the subject of cleanup levels and remedial action objectives. Effectiveness can then be shown if the post-dredging trend is lower than would be expected from simple extrapolation of the pre-dredging natural-recovery trend. For Waukegan Harbor, sampling was conducted at only two times before dredging. That pre-dredging time trend was inadequate for comparing to the post-dredging time trend.
A pre-dredging sediment toxicity test found relationships between toxicity and sediment-contaminant concentrations, with toxicity ranging
Statistical Analysis of PCB Concentrations in Fish, Waukegan Harbor, Illinois
Analyzed data (T. Hornshaw, Illinois EPA, written commun., August 3, 2006) included percent lipid and total PCB concentrations in carp tissue (fillets) that were caught before dredging (1981 and 1983) and after dredging (1996-2001 and 2005). Dredging activity was conducted during 1991-1992.
Temporal analysis of fish trends was based on linear-regression models of PCB concentrations in fish samples with monitoring year and percent lipid content as independent variables. The Box-Cox transformation (Box and Cox 1964) was parameterized in the regression-model likelihoods to allow possible transformations to be chosen optimally. Nonlinear trends in time were considered (Stow et al. 1995), and their results led to interpretations that were qualitatively similar. The two-sample Wilcoxon Rank Sum Test (Hollander and Wolfe 1973) was used to compare percent lipid-normalized PCB body burdens before and after dredging.
Figure 4-15 displays the log base 10 lipid-adjusted total PCB in carp fillets for the available monitoring years. There was no significant difference between mean pre-dredging (1981 and 1983) lipid-normalized PCB and mean post-dredging (1996-2001 and 2005) lipid-normalized PCB (p = 0.34). Despite the scatter of the data points, there is a statistically significant 12% decline in lipid-adjusted PCB per year for 1996-2005 (p = 0.03).
The temporal trend shown in Figure 4-15 was established only on the post-dredging data. Comparisons with the pre-dredging monitored fish (total sample size, 7) were insufficient to establish any conclusions on dredging effectiveness. Improved longitudinal monitoring (before and after dredging) could provide data sufficient to establish a time trend that could be used to associate changes with dredging. Monitoring during dredging would provide insight into the effects of sediment release and resuspension and inform statistical models as to when to postulate post-dredging effects.
The comparison of pre- and post-remediation data is further complicated by likely improvements in analytical procedures between 1985 and 2005. Other issues, for example, the failure to segregate fish by age or size, are also likely to affect this analysis.
from 0 to 100% survival for Ceriodaphnia dubia and Daphnia magna in whole-sediment laboratory assays. Hyalella azteca survival was much better in 48-hr exposures with survivals of 73.3 to 100 % (Burton et al. 1989). A post-dredging study by Kemble et al. in 2000 suggested that the PCB concentrations were lower (less than 10 mg/kg) and that sediments were generally not lethal to amphipods, but there were sublethal effects. The Kemble et al. (2000) and EPA (1999) studies suggested that because post-dredging PAH concentrations in sediments exceeded sediment quality guidelines based on probable effect concentrations, they may be contributing to the observed toxicity. The presence of PAH-related toxicity at a site with cleanup levels based on PCBs points to the need to focus on the full range of chemicals that may be causing toxicity at a site.
At the PCB-contaminated Cumberland Bay NPL, NY, site where 34 acres was dredged, quantitative remedial goals were not set. Rather, according to the site’s ROD (NYSDEC 1997), “The goals selected for this site are: mitigate the immediate threat to the environment posed by the PCB contaminated sludge bed; rapidly and significantly reduce human health and environmental risks; [and] prevent further environmental
degradation resulting from this known source of PCB contamination.” Without the guidance of risk-based quantitative criteria, it is difficult to judge whether the remedy achieved the goals that were set. As mentioned previously, PCBs averaging 6.8 mg/kg were still present on completion of dredging.
At the Fox River, a baseline monitoring plan is under development and is expected to establish the framework for long-term monitoring. However, the plan is being developed after remedial dredging has begun. A wealth of data is available from the remedial investigation and from scientific studies of the Lower Fox River and Green Bay that have been conducted over the decades. Nevertheless, the initiation of the dredging remedy before the establishment of a baseline—which ideally would use media, methods, and locations consistent with the long-term monitoring to follow—will probably complicate the evaluation of remedy effectiveness. In particular, it will be especially difficult to infer the initial 5-year effects of dredging on contaminant exposures by comparing years 0 and 5, because the apparent baseline will also be affected by dredging rather than reflect true pre-dredging conditions. Whether dredging causes an immediate drop in exposures by removing contamination or causes an increase in exposures to contaminated residuals or water-column releases, the collection of baseline data after the beginning of dredging will confound the comparison of long-term monitoring data with the true baseline. That will be true not only of the upstream portion of the Fox River, where dredging is being implemented, but also of downstream locations because upstream PCB releases to the water column can affect downstream conditions.
A common problem in monitoring for effectiveness is focusing on the meeting of cleanup levels, especially if operationally defined, and not on the long-term remedial action objectives. For example, cleanup for the Christina River was based on sediment removal. Sediment contaminant concentrations or biologic responses were not reported immediately following dredging or after backfilling (URS 1999). 5 years after dredging, data on the status of the benthic community was collected (EPA 2005b); however, this information is unable to indicate whether the remediation was effective.23 Although the cleanup requirements based on dredging to
a specified elevation may have been met, attainment of short-term and long-term risk-reduction goals has not been demonstrated. Similarly, cleanup requirements were met at the Naval Shipyard in Newport, RI, by removal of sediment to bedrock in many locations (TetraTech 2004). However, dredging to the specified depth (bedrock) is not an appropriate indicator of risk reduction. There was no verification sampling in these locations prior to capping (backfilling), so residual contamination may remain. As a result, contaminants may also be available for transport through the sand cap to surface sediments. The observation of continued toxicity to sea urchins during long-term monitoring suggests that risk-reduction goals were not achieved (TetraTech NUS 2006).
Evidence from other sites shows how meeting of cleanup levels, even if based on post-dredging concentrations, might not achieve desired risk-reduction goals. The 1995 remediation at GM Massena, NY, was designed to ultimately reduce exposure of fish and other wildlife to PCBs in the sediment. As indicated previously, cleanup levels were achieved only after capping of a portion of the site where reduction in residual concentrations below the cleanup level was not achievable solely through dredging. Examination of monitoring designed to evaluate performance relative to long-term remedial action objectives, however, has not shown expected reductions in fish concentrations. As shown in Table 4-2, spottail shiner, a fish with a limited foraging range, showed no obvious increasing or decreasing trends in PCB concentrations at the site even 5 years after the end of remediation (EPA 2005a).
Use Cleanup Levels Appropriately in Determining Remedy Effectiveness
When comparing post-remediation concentration data to cleanup levels, risk managers sometimes treat the cleanup levels as concentrations that should never be exceeded. However, this approach is not nec-
essarily appropriate or consistent with the evaluation of human and ecologic exposure conducted in the baseline risk assessments and, more importantly, with the derivation of cleanup levels. EPA guidance (EPA 1989b) recommends use of arithmetic mean concentrations within each exposure area to quantify exposures to chemicals of concern over time. While concentrations can vary significantly within an exposure area, the arithmetic mean is the appropriate statistic based on the assumption that a receptor integrates its exposure by moving about within the exposure area. Surface area weighted average concentrations, for example, at the Fox River, have also been used as appropriate indicators of exposure to surficial sediments (WI DNR/EPA 2002). The Marathon Battery site illustrates the importance of comparing cleanup levels to the appropriate statistic. At this site, occasional sediment samples had concentrations that were indistinguishable from the pre-remediation concentration distribution, yet these samples apparently are not reflected in the cadmium body burden in the benthic community, which has decreased.
Because a sampling program provides an imperfect measure of the arithmetic mean, EPA recommends use of the 95% upper confidence limit (UCL) of the mean. Therefore, cleanup levels ideally should be compared to the 95% UCL for monitoring data representative of the exposure area of concern that incited establishment of the cleanup level. EPA used 95% UCLs calculated from surface-sediment samples collected after completion of dredging to determine whether cleanup levels had been achieved at the Sitcum Waterway in Puget Sound (EPA 2006a [Commencement Bay–Sitcum Waterway, April 26, 2006]). For the Sitcum
TABLE 4-2 Spottail Shiner PCB Concentrations After Remediation of the GM Massena, NY, Site in 1995
Number of Samples
Total PCBs-Whole Body Concentration (mg/kg)
Lipid-Normalized PCB (mg/kg-lipid)
Source: EPA 2005a.
Waterway site, EPA reported that the “general approach [was to] redredge if [concentrations] exceed SQOs, but [EPA] also looked at surrounding data, historical data, and 95th percentile UCL of the mean sediment concentration for a chemical in a subarea to recommend whether additional sampling should be done, re-dredging, natural recovery, and/or nothing.” In cases in which monitoring data are biased such that they are not representative of exposure areas of concern, one can perform spatial weighting of the data before calculating 95% UCLs. For example, EPA used an interpolation method called inverse distance weighting to spatially weight floodplain soil data affected by contaminated sediment in the Housatonic River before calculating 95% UCLs for use in the human health risk assessment (Weston 2005b, Attachments 3 and 4).
Consider Using Pilot Tests
As described above, adverse site conditions may significantly limit the ability of dredging to achieve cleanup levels and remedial action objectives. Pilot testing can assist in identifying and characterizing potential limitations to dredging effectiveness, planning and responding to unexpected factors that may arise, and in defining the degree of effectiveness that might be obtainable through dredging. At the Lockheed Shipyard in Puget Sound, delays, additional costs, and limitations of dredging effectiveness were encountered owing to the unexpected quantity of debris. In retrospect, the construction manager for the remediation project stated that a pilot dredging program would have been able to inform him of the extent of the debris and the implications for dredging production rates and rehandling issues (G. Gunderson, TRC Solutions, personal commun., July 7, 2006). Similarly, the dredging remedy at Manistique Harbor, MI, was much more expensive, took twice as long, and required the dredging of nearly twice the expected sediment volume because of the unexpected problems associated with the presence of debris and thin sediment layers over bedrock. A pilot dredging program would have been advantageous in efficiently characterizing the scale and costs of the dredging remedy before full-scale implementation.
Pilot testing has been successfully used at a number of sites in identifying potential limitations to remedial effectiveness and allowing the
development of appropriate responses. For example, pilot testing has been used at the Grasse River in Massena, NY, in which the effectiveness of different dredges and different remedial technologies was explored. The previously described demonstration dredging project at Lavaca Bay, TX, was designed to evaluate the ability of dredging to effectively and economically address mercury-contaminated bed sediment at the outfall of a former chloro-alkali manufacturing site (Alcoa 2000). Results of the study indicated that hydraulic dredging could be readily implemented at this site, offsite transport of mercury on tidal flows moving through and around the curtained-off dredging unit were minimal, a large mass of mercury (2,300 lbs) was extracted from the hot spot, and increased mercury concentrations in oysters above the historical observed background in the wider bay did not occur. Residual surface sediment (generally, 0-5 cm) mercury concentrations were reduced in areas with high surface concentrations and lower subsurface concentration, while areas with highly contaminated buried sediments and low pre-dredging surface sediment contaminant concentrations typically showed increased surface sediment concentrations post-dredging (Alcoa 2000). The pilot study was judged to be a successful undertaking in that the data collected were key in the evaluation of the role of dredging in the remedial activities to be undertaken at this site.
Several dredging demonstration projects were conducted during the remedial investigation and feasibility study at the Fox River. The previously-described demonstration projects conducted in Sediment Management Units 56 and 57 (SMU 56/57) in 1999 and 2000 provided valuable information on dredging and dewatering productivity and operations. The 1999 demonstration removed 31,000 cy which was much less than the 80,000 cy objective. During that project, hydraulic dredging equipment was upgraded three times in an effort to increase the solids content of the dredged slurry. Dewatering of solids proved to be a limiting constraint on production rate and required installation of additional filter presses. The average production rate was 294 cy/day, compared with a desired rate of 900 cy/day (Montgomery Watson 2001). Further adjustments in dredging and dewatering equipment, beyond those made by the 1999 project team, were needed in 2000 to remove the remaining 50,000 cy of targeted sediments. The resulting average production rates exceeded a project target of 833 cy/day and reached a peak production
rate of 1,599 cy/day on a single day near the completion of the removal action (Fort James Corporation et al. 2001).
Pilot studies also assisted in the success of the Head of Hylebos dredging project. Two pilot studies were used to help to define the scope of such problems as debris, provide large samples for additional testing, and help with selection of equipment and development of the dredging operation plan. One study concentrated on how to remove the “soft black muck drainage water” from the mechanically dredged material at an upland storage facility. About 5 gal of water per cubic yard of sediment, or about 40 lb per 2,200 lb, was associated as free water. The second study focused on rail transport and placement into the offsite landfill and helped to refine the rail-transport program. During the pilot study, the contractor observed the dissociation of physical integrity (strength) of the soft fine-grained sediment after handling, both in the barge and on the bottom of the waterway. The loss of strength was seen as a contributor to formation of a flowable residual layer (fluidized soft mud) on the bottom during dredging that needed to be captured (P. Fuglevand; Dalton, Olmsted & Fuglevand, Inc.; personal commun., July 7, 2006).
In some cases, pilot testing is not required, because of the scale of the dredging project or because of other conditions. For example, the relatively small scale of the U.S. Naval Shipyard site in Newport, RI, and the ability to dredge much of the contaminated material from land combined to make the implementation of dredging favorable. Of 30,000 cy dredged, the vast majority was from the nearshore area (TetraTech NUS 2006). The nearshore materials were removed by a long reach excavator, which was operated on a bay haul road constructed for the project, and were loaded directly into offroad dump trucks (Tetra Tech 2004). Dredging of the remaining elevated offshore material was performed from a barge with a crane equipped with a clamshell bucket and loaded onto an adjacent haul barge. The excavator was much faster and less expensive than the barge-mounted crane, and direct loading into haul trucks minimized handling of material (EPA 2006a [Newport Naval, May 17, 2006]).
In most contaminated sediment megasites, however, the scale and complexity of the sites suggest that pilot studies are appropriate and will assist in reducing limitations of dredging effectiveness. Adaptive man-
agement24 and even pilot testing during implementation may still be necessary to respond to unforeseen problems during implementation. Pilot testing alone will not ensure success; to maximize the project’s usefulness, the scope and objectives need to be clearly communicated and monitoring needs to be capable of establishing whether objectives were achieved and the factors that influenced the project’s performance.
Implement Best Management Practices
Although it is not a guarantee, the adoption of best management practices (BMPs) will help to ensure appropriate implementation of a remedial project. Best management practices are defined on an activity-specific basis and will depend upon the type of dredging and transport equipment used, the environment in which the dredging takes place, and the process “train” or sequencing of the remedial activities. There are no standardized BMPs for environmental dredging, although “lessons learned” from environmental dredging projects to date suggests that there are BMPs that will likely be applicable to many dredging projects. These BMPs are primarily designed to minimize the loss or transport of contaminated sediment or debris from the dredging footprint and minimizing the generation and runoff of leachate from dredged material to the receiving water during transport or rehandling of dredged sediment. BMPs that may be useful for minimizing loss or off-site transport of sediment and debris include (this list is not considered comprehensive):
Use of silt curtains to reduce the transport of suspended solids.
Use of floating and/or absorbent booms to capture floating debris or oil sheens.
Reduction of the impact speed of the dredge bucket with the bottom and/or reduction of the rate of ascent of a filled bucket; reduction of the swing rate of cutter-head dredge.
Prevention of overfilling buckets through accurate and controlled placement of bucket.
In general, adaptive management is the testing of hypotheses and conclusions and re-evaluation of site assumptions and decisions as new information is gathered. See Chapter 6 for further detail.
Use of environmental or sealed buckets, where sediment characteristics will allow.
Protection of the overwater swing path of a filled bucket (by placing an empty barge or apron to catch lost material).
Eliminating bottom stockpiling of dredged material or sweeping with the dredge bucket/head.
BMPs that may be useful for controlling production or runoff of leachate include:
Maximization of the “bite” of a dredge bucket (that is, avoiding thin lifts).
Allowance for draining a sediment-filled bucket before breaking the water’s surface.
Use of filtration cloth, hay bales, curbing, or other physical baffles (similar to stormwater BMPs) to control runoff from barges or rehandling areas.
Additional BMPs that may be used to minimize environmental impacts of dredging include (but are not limited to):
Scheduling dredging during periods when sensitive species or populations are not present at the site.
Daily construction oversight and progress surveys.
Water quality monitoring during dredging activities.
BMPs for control or prevention of resuspension or loss of contaminated material in the waterway were implemented during the remediation of Todd and Lockheed Shipyards. BMPs were specified for overwater demolition, pile removal, dredging, barge dewatering, vessel management, sediment offloading, capping and fill placement, and overwater construction. During demolition, pile removal, and overwater construction, an absorbent boom with 4- to 6-ft silt curtains was deployed to contain floating debris or sheen caused by the removal of creosoted piles. Entrainment of water during dredging was minimized by taking complete “bites” with the dredge bucket whenever possible. Each full bucket was held just at the water’s surface to allow water to drain before the bucket was swung to the barge. Dredged sediment was pas-
sively dewatered on an onsite flat-deck barge through straw bales and filter fabric before being discharged to the waterway. During offloading, the clamshell bucket was prevented from swinging over open water by placement of a spill-collection platform under its path. Asphalt curbing surrounded the transloading area to prevent sediment, sediment drainage water, and contact stormwater from migrating offsite. Water collected from the transloading area was not allowed to enter the waterway but was collected and treated on site by a process of settling, multimedia filtration, and carbon filtration. Treated water was discharged to the sanitary sewer.
Implementation of BMPs for control of produced water was also important in the success of the Head of Hylebos dredging project. All the water entrained with the dredged sediment was placed in the barge rather than being released back to the water. Overall, the enclosed mechanical buckets placed more water than sediment in the barges. This water can contain an important load of sediments and contaminants. During the 2005 season, the water-management system captured about 4,000 cy of sediment. It is estimated that if that material had been released back to the dredge area, it would have generated a layer of impacted sediment an average of 3-4 in. thick over the dredged area. Capture of the solids contributed to the ability to meet the cleanup goals (Dalton, Olmsted & Fuglevand, Inc. 2006).
Remediation contractors for the Todd Shipyard found benefit in working with dredging companies that were able to mobilize an array of equipment from their inventory to meet project needs and respond to changing site conditions and schedules. The project engineer for Todd Shipyard included the dredging contractor as a consultant during the design stages of the remediation—an important step that ensured a smooth transition from design to implementation (EPA 2006a [Todd Shipyards Sediment OU, May 12, 2006]). Experienced environmental dredgers also have the capabilities to operate within the typical regulatory restrictions and an understanding of the difficulties associated with environmental dredging. The importance of using contractors experienced in environmental dredging was emphasized by the remediation contractors at the Fox River SMU 56/57 dredging project:
Most large dredging contractors in the United States have little or no experience with contaminated sediment projects, working pre-
dominantly on navigational dredging projects. Navigational dredging projects typically have no environmental controls, resulting in higher production rates and lower unit costs. Larger-scale projects may also limit the available temporary water treatment and dewatering equipment unless planned well in advance, as well as onshore land space, that are necessary to complete the work in a timely fashion (Montgomery Watson 2001).25
Since 1999, the first year of the SMU 56/57 dredging project, the numbers and experience of firms experienced with environmental dredging has increased. However, the need for contractors familiar with the challenges of environmental dredging remains, particularly at large, multi-year megasite projects.
Use Appropriate Contracting Arrangements
The nature of the contracting vehicle used to conduct the work can drive behavior of the contractor and ultimately impact project results. The contracting terms and approaches provide incentives for contractor performance so the contracting approach needs to be aligned with the project’s risk reduction goals. For example, the implementation of BMPs can be encouraged or discouraged by the contracting mechanisms used in a remedial project.
EPA described the importance of the contracting mechanism at the Lockheed Shipyard cleanup (EPA 2006a [Lockheed Shipyard Sediment OU, May 11, 2006]):
The primary keys to success for a complex project such as the LSSOU cleanup include (1) a contract where the dredging contractor was not taking the risk and 2) the dredger was guaranteed a daily rate for each activity. Frequently, in the past, environmental dredging contractors have followed the “navigational dredging
model” in terms of contract style and dredging methods. Under a Unit Rate contract the dredger is in a "production dredging mode, which does not work for environmental dredging. Under the Time and Material [contract] the dredger is not penalized for taking the appropriate time to accomplish the task at hand. Because of this the dredger is more likely to comply with BMPs and to take care to minimize loose of material back into the waterway or to cause resuspension.
Phase 1 dredging at the site resulted in incomplete debris removal, the presence of undredged inventory, and a large amount of residuals at the end of the in-water work window. On review of the approach, the remediation project manager revised the contract mechanism for phase 2 dredging (second season) to use a time-and-materials approach and requested new bids for the work (EPA 2006a [Lockheed Shipyard Sediment OU, May 11, 2006]).
The Todd Shipyard project engineer found that a cost-plus-incentive-fee form of contract worked well by motivating contractors to complete every aspect of the construction in accordance with defined quality objectives at the lowest overall cost. The contract reimbursed the contractor for all direct costs of the work. That removed the financial risk to the contractor and thereby reduced bid costs to cover unknown or un-quantifiable risks (EPA 2006a [Harbor Island Todd Shipyards Sediment OU, May 12, 2006]).
In the Head of Hylebos dredging project, a cost-plus-fee contract assisted in the reduction of residual contamination. The engineers and contractors determined that it was in their best interest to minimize the extent of the residual layer by slowing down the production to match “good housekeeping” on the bottom. That contracting mechanism provided the opportunity for the owner to work with the contractor to adjust construction and operations to achieve the project objectives. Contractor oversight was provided by onsite dredge inspectors during each shift. That oversight was used to make decisions regarding whether the target elevations were met (that is, finding the underlying native material) before dredging moved to the next cut. The cab on each dredge was actually expanded to provide a place for the dredge observer to sit side by side with the operator day and night throughout the dredging operation (Dalton, Olmsted & Fuglevand, Inc. 2006).
Overall, there is no single best contract type. These decisions will depend on site conditions and necessary equipment and materials. However, contracting terms and approaches that encourage contractors to focus on achieving cleanup goals and remedial action objectives are best suited for environmental dredging. These arrangements should create incentives for reducing resuspension and residual production, using best management practices, and adjusting the dredging approach to improve chances of meeting cleanup levels and result in cost savings.
Use Operational Controls to Improve Dredging Accuracy
In addition to appropriate design, implementation, and monitoring, technologic approaches can improve dredging efficiency and effectiveness. Some of them have been used successfully at contaminated sediment sites.
A one-of-a-kind specially designed high-technology dredge outfitted with innovative sensors and controls to achieve a 6-in. excavation-cut tolerance was used to extract creosote-contaminated sediment at Bayou Bonfouca, LA. A cutline to the depth associated with a total PAH concentration of 1,300 ppm was established and programmed as an absolute elevation along a 4,000-ft length by using borehole concentration profiles. Maximum contamination depth was 17 ft (average, 10 ft). Logs, concrete, metal objects, and so on were removed with grapple hooks before excavation (EPA 2006a [Bayou Bonfouca Superfund site, May 12, 2006]). The pre-dredging operation and low tidal fluctuations and low stream flow rate (13 ft3/sec) provided a stable dredging platform (spud barge) in the loose, high-organic-matter layer over “harder” unconsolidated inorganic substrate, which all aided in the implementation of the new precision dredging technology. No post-dredging measurements (such as bottom elevations) were taken to evaluate achievement of the bottom cut-line target programmed into the excavator, nor were any sediment analyses performed to verify achievement of a total PAH concentration of less than 1,300 ppm. Targets for volume of dredged material were achieved, however, and that was the primary goal of controlling the excavation depth.
The dredge used at Todd Shipyards was equipped with a positioning system with 20-cm (GPS-controlled) horizontal accuracy. It provided
real-time display and tracking of the horizontal and vertical position of the dredge bucket. Digital GPS receivers and a gyrocompass were used to determine real-time horizontal (X and Y) positioning of the derrick barge and the dredge bucket. An electronic tide gauge was used to allow the operator to determine the proper dredge elevation below the water surface accurately. The vertical position of the bucket was combined with the electronic tide-gauge data to determine the bucket elevation (Z). In addition, the dredge-bucket wires were painted in 1-ft increments to provide for a check on the electronically calculated vertical position. The information generated by the positioning system was electronically stored and used to create maps that showed dredging progress, including the degree of overlap between bucket deployments (EPA 2006a [Harbor Island Todd Shipyards Sediment OU, May 12, 2006]). That approach to navigation and positioning of the dredge bucket has been used at a number of sites in Puget Sound.
Precision positioning systems are seeing increasing use. Although the basic technology of dredging has changed little in recent decades, the ability to position the dredge accurately has improved dramatically; in principle, this can improve our ability to remove contaminated sediment accurately and efficiently if sufficient site-characterization data are available. Regardless of the improvements in dredging methods and equipment, the reliability of the equipment and the availability of skilled operators capable of processing and interpreting the data remain challenges.
Consider Backfilling and Capping to Control Residuals
As indicated previously, the factor limiting dredging effectiveness that is the most difficult to manage is high residual contaminant concentrations. Residuals are always detected after dredging and can be relatively high in concentration and typically of the same order as the average concentration in the dredged material (Reible et al. 2003). The magnitude of residuals can be higher in the presence of debris or when site conditions make it infeasible to overdredge into clean material. Even in favorable dredging conditions, however, some degree of residual control is usually necessary to achieve site cleanup standards and to address site remedial action objectives. Generally, control of residuals is achieved by adding backfill or thin-layer capping; this has clear advantages in
achieving bulk sediment contaminant concentration targets even if the backfill layer is intermixed with the residual sediments. Although backfill can effectively manage bulk sediment concentrations, the effectiveness of backfill for aiding long-term risk reduction is less well understood. In addition, the advantages of dredging with backfill for residual control relative to a complete capping remedy need to be assessed during remedy evaluation.
At Bayou Bonfouca in Slidell, LA, backfilling was a necessary phase of the overall remedial operation. Excavation to up to 3 m compromised the stability of the unconsolidated material forming the banks along the bayou. Gravel (about 1 ft) over sand (about 1 ft) and additional fill gave support to the sheet-piling-reinforced banks (5,000 ft long) and served to cap the residual PAH contamination of 1,300 ppm (target concentration). Backfilling to the original bottom grade maintained the historical water flow rates and levels needed for recreational and other boating traffic and allowed the bayou to begin natural recovery.
The remediation of the former Ketchikan Pulp Company facility, in Ward Cove in Ketchikan, AK, also involved some backfilling of dredged areas. The facility operated as a dissolving sulfite pulp mill from 1954 until 1997 and discharged untreated sulfite waste liquor (magnesium bisulfite), pulping solids, and bleaching waste (chlorine caustic) into Ward Cove until 1971, with increasing wastewater treatment after that. Mill operations affected sediment by releasing large quantities of organic material (up to 10 ft thick) as byproducts of wood pulping. The organic material altered the physical structure and chemistry of the sediments and thus the type and abundance of benthic organisms. Degradation of the organic-rich pulping byproduct led to anaerobic conditions in the sediment and production of ammonia, sulfide, and 4-methylphenol in quantities that were potentially toxic to benthic organisms (EPA 2006a, Ketchikan Pulp Company; April 26, 2006). Remedial action objectives included reducing toxicity of surface sediments to benthic life and enhancing benthic recolonization. The selected remedy included thin-layer (6-12 in.) placement of clean sand over dredged areas (to less than the full depth of contamination) and undredged areas and monitoring of natural recovery where thin-layer placement was not practicable. Sand backfilling was expected to achieve the remedial objectives by diluting contaminants and organic matter, both of which are associated with benthic toxicity on this site (EPA 2000b).
Remediation was completed in 2001. The long-term monitoring program includes sediment chemical analysis, toxicity testing, and assessment of the benthic macroinvertebrate community. The first round of long-term monitoring in 2004 found that the remedy appeared to have met its objectives in backfilled areas; chemical concentrations were generally below sediment cleanup levels as determined from pre-remediation toxicity testing, survival of benthic test organisms was high, and benthic species diversity and abundance increased relative to the pre-remediation baseline and are similar to reference areas (Exponent 2005). Additional monitoring is planned for 2007, and equally favorable or improved results may lead to a reduction in required monitoring of backfilled areas. In contrast, 2004 monitoring results showed that only one of four natural recovery areas had improvements comparable to those found in the backfilled areas (Exponent 2005; Herrenkohl et al. 2006).
Backfilling has been used at a variety of sites, including the Fox River and several sites in the Puget Sound area. It has been proposed for many sites that have not met or are unlikely to meet cleanup levels after dredging alone. Backfill that is at least about 6-12 in. thick probably forms an effective separation between much of the benthic community that might colonize the top of the backfill layer and the underlying sediment. Thin sand backfill layers (less than 6 in.), however, are of uncertain effectiveness because the low sorptivity of sand means that the benthic community may be exposed to pore water contaminant concentrations similar to that of uncapped sediment. Exposure and risk are often more closely related to pore water concentration than to bulk sediment concentration, and further research or field monitoring is needed to confirm the appropriateness of thin-layer backfilling.
On the basis of its review of data and experiences at dredging projects, the committee has reached the following conclusions:
The committee was generally unable to establish whether dredging alone is capable of achieving long-term risk reduction, because
Monitoring at most sites does not include the full array of measures necessary to evaluate risk.
Dredging may have occurred in conjunction with other remedies or natural processes, or insufficient time may have passed to evaluate long-term risk reduction.
A systematic compilation of site data necessary to track remedial effectiveness nationally is lacking.
Dredging remains one of the few options available for the remediation of contaminated sediments and should be considered, with other options, for managing the risks that they pose.
Dredging is effective for removal of mass, but mass removal alone may not achieve risk-based goals.
Dredging will likely have at least short-term adverse effects on the water column and biota.
Dredging effectiveness is limited by resuspension and release of contaminants during dredging and the generation or exposure of residual contamination by dredging. Those limitations are minimized if site conditions are favorable and the remedy is designed and implemented appropriately.
Favorable site conditions include
Little or no debris
A visual or physical texture difference or other rapid mechanism for differentiating clean and contaminated sediments.
Potential for overdredging into clean material.
Low-gradient bottom and side slopes.
Lack of piers and other obstacles.
Site conditions that promote rapid natural attenuation after dredging (for example, through natural deposition).
Absence of non-aqueous-phase liquid or readily desorbable contaminants.
Effective design and implementation factors include
Site characterization sufficient to develop a comprehensive conceptual site model and identify adverse site conditions.
Identification and control of sources on a watershed-wide basis.
Use of pilot studies, where appropriate, to identify adverse site conditions and appropriate management responses.
Application of best management practices to control residuals and resuspension (for example, operational controls at the dredge and on produced streams, appropriate equipment selection, and residual control measures).
Contracting and procurement mechanisms to encourage a focus on cleanup levels and remedial action objectives.
Engagement of experienced and innovative environmental-dredging contractors throughout the design and implementation phases of remediation.
Dredging alone is unlikely to be effective in reaching short-term or long-term goals where sites exhibit one or more unfavorable conditions. Where unfavorable conditions exist, increased contaminant resuspension, release, and residual will tend to limit ability to meet cleanup levels and delay the achievement of remedial action objectives unless managed through a combination of remedies or alternative remedies.
A remedy should be designed to meet long-term risk-reduction goals. The design should be tested by modeling and monitoring the achievement of long-term remedial action objectives.
Site conditions that influence dredging effectiveness should be recognized during selection, development, and implementation of the remedy. When conditions unfavorable for dredging exist:
Implementation of one or more pilot tests should be considered to identify optimal remedial approaches and assess their effectiveness.
Adverse effects of resuspension, release, and residuals should be forecast and explicitly considered in expectations of risk.
The ability of combination remedies to lessen the adverse effects of residuals should be considered when evaluating the potential effectiveness of dredging.
Best management practices should be implemented to minimize effects of adverse dredging conditions.
The possibility of adverse dredging conditions that are not anticipated should be recognized and planned for.
A good baseline assessment coupled with a well-designed long-term monitoring plan should be implemented to permit evaluation of dredging effectiveness.
Well-designed pre-dredging and post-dredging monitoring is necessary to establish effectiveness and indicate achievement of remedial action objectives.
Monitoring should be conducted to demonstrate achievement of cleanup levels and to confirm that the cleanup levels achieve remedial action objectives.
Data from monitoring should be managed and stored in electronic databases accessible for further analysis.
Further research, including during dredging pilots and full-scale operations, should be conducted to define mechanisms, rates, causes, and effects of dredging residuals and contaminant resuspension.
AGC (Advanced Geoservices Corporation). 2001. 2000 Annual Report Long-Term Monitoring Program: Marathon Remediation Site. Advanced Geoservices Corporation, Chaddsford, PA. April 2, 2001.
Alcoa Inc. 1995. The Grasse River: Non-Time Critical Removal Action Documentation Report, December 1995. Alcoa [online]. Available: http://www.thegrasseriver.com/ppt/Nov20CAP/NTCRA.htm [accessed October 11, 2006].
Alcoa Inc. 2000. Treatability Study for the ALCOA (Point Comfort)/Lavaca Bay Superfund Site, Draft. January 2000. 84pp.
Alcoa Inc. 2005. Remedial Options Pilot Study Work Plan: Grasse River Study Area Massena, New York. Alcoa Inc., Massena, New York. February 11, 2005.
Anderson, B.S., J.W. Hunt, B.M. Phillips, M. Stoelting, J. Becker, R. Fairey, H.M. Puckett, M. Stephenson, R.S. Tjeerdema and M. Martin. 2000. Ecotoxicologic change at a remediated Superfund Site in San Francisco Bay, California, USA. Environ. Toxicol. Chem. 19(4): 879-887.
Baumann, P.C. 2000. Health of Bullhead in an Urban Fishery after Remedial Dredging Final Report - January 31, 2000. Prepared for U.S. Environmental Protection Agency, Great Lakes National Program Office, Chicago IL, by
U.S. Geological Survey Field Research Station, Ohio State University, Columbus, OH [online]. Available: http://www.epa.gov/glnpo/sediment/Bullhead/index.html [accessed Jan. 9, 2006].
BBL (Blasland, Bouck & Lee, Inc.). 1994a. Engineering Evaluation/Cost Analysis. Volume I of II. Manistique Papers Inc. and Edison Sault Electric Company, Manistique, MI. Blasland, Bouck & Lee, Inc., Syracuse, New York. April 1994.
BBL (Blasland, Bouck & Lee, Inc.). 1994b. Response to Issues Raised by USEPA Region 5 in its Removal Action Recommendation and related documents. Prepared for Manistique Papers Inc., and Edison Sault Electric Company, Manistique, MI, by Blasland, Bouck & Lee, Inc., Syracuse, New York. October 1994.
BBL (Blasland, Bouck & Lee, Inc.). 1995. Non-Time-Critical Removal Action: Documentation Report, Vol. 1. Grasse River Study Area, Massena, New York, Aluminum Company of America, Massena, New York. Draft. Blasland, Bouck & Lee, Inc., Syracuse, New York. December 1995.
BBL (Blasland, Bouck & Lee). 1996. St. Lawrence River Sediment Removal Project Remedial Action Completion Report. Prepared for General Motors Power-train, Massena, New York, by Blasland, Bouck & Lee, Syracuse, New York. June 1996.
BBL (Blasland, Bouck & Lee, Inc). 2000. Feasibility Study Report - Phase I Allied Paper, Inc./Portage Creek/Kalamazoo River Superfund Site, Kalamazoo and Allegan Counties, Michigan, Appendix D. Site Profiles of Sediment Dredging Projects. Draft for State and Federal Review. Blasland, Bouck & Lee, Inc., Syracuse, New York. October 2000 [online]. Available: http://www.deq.state.mi.us/documents/deq-erd-kzoo-FS-apend-d.pdf [accessed Jan. 8, 2006].
Becker, G. 1983. Fishes of Wisconsin. Madison WI: University of Wisconsin Press.
Box, G.E.P., and D.R. Cox. 1964. An analysis of transformations. J. Roy. Stat. Soc. B 26(2):211-252.
Bremle, G., and P. Larsson. 1998. PCB concentration in fish in a river system after remediation of contaminated sediment. Environ. Sci. Technol. 32(22):3491-3495.
Bremle, G., L. Okla, and P. Larsson. 1998a. PCB in water and sediment of a lake after remediation of contaminated sediment. Ambio 27(5):398-403.
Bremle, G., P. Larsson, T. Hammar, A.Helgee, and B. Troedsson. 1998b. PCB in a river system during sediment remediation. Water Air Soil Poll. 107(1-4):237-250.
Bridges, T.S., Ells, S., Hayes, D., Mount, D., Nadeau, S., Palermo, M., Patmont, C., Schroeder, P. in press. The Four Rs of Environmental Dredging: Resuspension, Release, Residual, and Risk. ERDC TR-07-X. U.S. Army Engineer Research and Development Center, Vicksburg, MS.
Burton, G.A., B.L. Stemmer, K.L. Winks, P.E. Ross, and L.C. Burnett. 1989. A multitrophic level evaluation of sediment toxicity in Waukegan and Indiana Harbors. Environ. Toxicol. Chem. 8(11):1057-1066.
Canonie Environmental. 1996. Waukegan Harbor Remedial Action, Waukegan, IL. Construction Completion Report. Canonie Environmental. July 3, 1996.
CH2M Hill. 2001. Five-Year Review Report: Second Five-Year Review Report For Bayou Bonfouca Superfund Site, Slidell, St. Tammany Parish, Louisiana. Prepared by CH2M Hill, for U.S. Environmental Protection Agency, Region 6, Dallas, TX. June 2001 [online]. Available: http://epa.gov/earth1r6/6sf/pdffiles/bayou_bonfouca_5_year.pdf [accessed Jan. 10, 2007].
CH2M Hill. 2005. Lower Fox River Operable Unit 1 Pre-design—Basis of Design. Prepared for WTM I Company, by CH2MHill, Milwaukee, WI. March 2005.
CH2M Hill. 2006. Bayou Bonfouca - Hurricane Impacts Evaluation. Memorandum to Michael Herbert, EPA Region 6, from Renee Ryan and Scot McKinley, CH2M Hill. Project No. 334872.ET.02. January 25, 2006.
Chemical Waste Management. 1997. Completion Report for Marine Remedial Action on the United Heckathorn Superfund Site, Richmond, CA. October 1997.
Connolly, J., V. Chang, and L. McShea. 2006. Grasse River Remedial Options Pilot Study (ROPS) Findings from Dredging Activities. Presentation at the Second Meeting on Sediment Dredging at Superfund Megasites, June 7, 2006, Irvine, CA.
Connolly, J., J. Quadrini, and L. McShea. 2007. Overview of the 2005 Grasse River Remedial Options Pilot Study. Presentation at the 4th International Conference on Remediation of Contaminated Sediments, January 22-25, 2007, Savannah, GA.
Cressie, N.A.C. 1991. Statistics for Spatial Data. New York: Wiley. 900 pp.
Dalton, Olmsted & Fuglevand, Inc. 2006. Remediation Action Construction Report, Part 1: Head of Hylebos Waterway Problem Area Commencement Bay Nearshore/Tideflats Superfund Site Tacoma, Washington, Review Draft, July 21, 2006. Prepared for Head of Hylebos Cleanup Group, Arkema, Inc, General Metals of Tacoma, Inc, by Dalton, Olmsted & Fuglevand, Inc., Kirkland, WA. July 21, 2006.
Dickerson, D., and J. Brown. 2006. Case Study on New Bedford Harbor. Presentation at the First Meeting on Sediment Dredging at Superfund Megasites, March 22, 2006, Washington DC.
Diggle, P.J., and P.J. Ribeiro. 2007. Model-Based Geostatistics. New York: Springer.
EcoChem, Inc. 2005. Duwamish/Diagonal CSO/SD Sediment Remediation Project Closure Report. Prepared for King County Department of Natural Resources and Parks, Elliott Bay/Duwamish Restoration Program Panel, Seat
tle, WA. Panel Publication 39. July 2005 [online]. Available: http://dnr.metrokc.gov/WTD/duwamish/diagonal.htm [accessed Jan. 3, 2006].
EPA (U.S. Environmental Protection Agency). 1984. Record of Decision: Outboard Marine Corp. EPA ID: ILD000802827 OU 01 Waukegan, IL. EPA/ROD/R05-84/007. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r0584007.pdf [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 1987. EPA Superfund Record of Decision: Bayou Bonfouca, EPA ID: LAD980745632, OU 01, Slidell, LA. EPA/ROD/R06-87/019. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r0687019.pdf [accessed Aug. 7, 2006].
EPA (U.S. Environmental Protection Agency). 1989a. Record of Decision: Marathon Battery Corp. EPA ID: NYD010959757. OU 02, Cold Springs, NY. EPA/ROD/R02-89/097. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r0289097.pdf [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 1989b. Risk Assessment Guidance for Superfund, Volume 1: Human Health Evaluation Manual (Part A, Interim Final). EPA/540/1-89/002. Office of Emergency and Remedial Response, U.S. Environmental Protection Agency, Washington, DC. December 1989 [online]. Available: http://rais.ornl.gov/homepage/HHEMA.pdf [accessed May 9, 2007].
EPA (U.S. Environmental Protection Agency). 1989c. Record of Decision: Commencement Bay, Near Shore/Tide Flats. EPA ID: WAD980726368 OU 01, 05, Pierce County, WA. EPA/ROD/R10-89/020. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r1089020.pdf. [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 1990a. Record of Decision: New Bedford. EPA ID: MAD980731335, OU 02, New Bedford, MA. EPA/ROD/R01-90/045. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r0190045.pdf [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 1990b. EPA Superfund Explanation of Significant Differences: Bayou Bonfouca. EPA ID: LAD980745632, OU 01, Slidell, LA. EPA/ESD/R06-90/900. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/e0690900.pdf [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 1991. Record of Decision: General Motors (Central Foundry Division). EPA ID: NYD091972554. OU 01 Massena, NY. EPA/ROD/R02-91/131. Superfund Information Systems, U.S. En-
vironmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r0291131.pdf [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 1993a. Record of Decision: E.I. Du Pont De Nemours & Co., Inc. (Newport Pigment Plant Landfill). EPA ID: DED980555122, OU 01, Newport, DE. EPA/ROD/R03-93/170. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r0393170.pdf [accessed Jan.9, 2007].
EPA (U.S. Environmental Protection Agency). 1993b. Record of Decision: Reynolds Metals Co. EPA ID: NYD002245967, OU 01, Massena, NY. EPA/ROD/R02-93/201. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r0293201.pdf [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 1994. PCB Action Level for Manistique River and Harbor. Letter to William E. Muno, Director, Waste Management Division, from Milton Clark, Senior Health and Science Advisor, U.S. Environmental Protection Agency, Chicago, IL. August 4, 1994.
EPA (U.S. Environmental Protection Agency). 1995. Superfund Site Close-out Report Marathon Battery Company Site Putnam County Cold Spring, New York. September 28, 1995.
EPA (U.S. Environmental Protection Agency). 1996a. Groundwater Remedial Action Five-Year Review: Bayou Bonfouca Superfund Site, Slidell, Louisiana. September 1996.
EPA (U.S. Environmental Protection Agency). 1996b. Record of Decision: United Heckathorn Co. EPA ID: CAD981436363, OU 01, Richmond, CA. EPA/ROD/R09-96/502. Superfund Information Systems, U.S. Enviromental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r0996502.pdf [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 1997a. Five-Year Review Type Ia Outboard Marine Corporation Site, Waukegan, Illinois. U.S. Environmental Protection Agency Region 5 [online]. Available: http://www.epa.gov/superfund/sites/fiveyear/f97-05002.pdf [accessed May 22, 2006].
EPA (U.S. Environmental Protection Agency). 1997b. Explanation of Significant Differences: Commencement Bay, Near Shore/Tide Flats. EPA ID: WAD980726368, OU 01, Pierce County, WA. EPA/ESD/R10-97/059. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/e1097059.pdf [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 1997c. Record of Decision: Harbor Island (Lead). EPA ID: WAD980722839, OU 07, Seattle, WA. EPA/ROD/R10-97/045. Superfund Information Systems, U.S. Environ-
mental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r1097045.pdf [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 1998a. Five-Year Review Report Marathon Battery Company Superfund Site Village of Cold Spring, Putnam County, New York. U.S. Environmental Protection Agency, Region 2, New York. June 1998 [online]. Available: http://www.epa.gov/superfund/sites/fiveyear/f98-02005.pdf [accessed May 4, 2007].
EPA (U.S. Environmental Protection Agency). 1998b. Record of Decision: New Bedford Harbor OU, 01 Upper and Lower Harbor, New Bedford, MA September 25, 1998. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/ne/superfund/sites/newbedford/38206.pdf [accessed Jan.9, 2007].
EPA (U.S. Environmental Protection Agency). 1999. Evaluation of Toxicity and Bioaccumulation of Contaminants in Sediments Samples from Waukegan Harbor, Illinois. EPA-905-R-99-009. Great Lakes National Program Office, U.S. Environmental Protection Agency, Chicago, IL. October 1999 [online]. Available: http://www.epa.gov/glnpo/sediment/waukegan/whrpt.pdf [accessed Jan. 9, 2007].
EPA (U.S. Environmental Protection Agency). 2000a. Realizing Remediation II: A Summary of Contaminated Sediment Remediation Activities at Great Lakes Areas of Concern. U.S. Environmental Protection Agency, Great Lakes National Program Office, Chicago, IL. July 2000 [online]. Available: http://www.epa.gov/glnpo/sediment/realizing2/index.html [accessed Aug. 16, 2006].
EPA (U.S. Environmental Protection Agency). 2000b. Record of Decision: Ketchikan Pulp Company. EPA ID: AKD009252230, OU02, Ketchikan, AK. EPA/ROD/R10-00/035. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r1000035.pdf [accessed Jan. 10, 2007].
EPA (U.S. Environmental Protection Agency). 2000c. Record of Decision: Puget Sound Naval Shipyard Complex,. EPA ID: WA2170023418, OU 02 Marine Bremerton, WA. EPA/ROD/R10-00/516..June 13, 2000. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r1000516.pdf [accessed May 4, 2007].
EPA. (U.S. Environmental Protection Agency). 2000d. Record of Decision: Newport Naval Education and Training Center. EPA ID: RI6170085470, OU4, Newport, RI. EPA/ROD/R01-00/155. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/r0100155.pdf [accessed Jan. 10, 2007].
EPA (U.S. Environmental Protection Agency). 2002. Second Five-Year Review Report for Outboard Marine Corporation Superfund Site, Waukegan, Lake
Country, IL. U.S. Environmental Protection Agency, Chicago, IL. September 2002 [online]. Available: http://www.epa.gov/R5Super/fiveyear/reviews_pdf/illinois/outboard_marine.pdf [accessed May 22, 2006].
EPA (U.S. Environmental Protection Agency). 2003a. Sediment Remedy ReEvaluation: Bayou Bonfouca Site, Slidell, St. Tammany Parish, Louisiana. U.S. Environmental Protection Agency, Region 6. February 2003.
EPA (U.S. Environmental Protection Agency). 2003b. Five-Year Review Report Marathon Battery Company Superfund Site, Village of Cold Spring, Putnam County, New York. U.S. Environmental Protection Agency, Region 2, New York, NY. June 2003 [online]. Available: http://www.epa.gov/superfund/sites/fiveyear/f03-02019.pdf [accessed May 4, 2007].
EPA (U.S. Environmental Protection Agency). 2003c. Explanation of Significant Differences: Harbor Island (Lead). EPA ID: WAD980722839, OU 07, Seattle, WA. EPA/ESD/R10-03/011. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/e1003011.pdf [accessed Jan. 10, 2007].
EPA (U.S. Environmental Protection Agency). 2003d. Explanation of Significant Differences: Harbor Island (Lead). EPA ID: WAD980722839, OU 07, Seattle, WA. EPA/ESD/R10-03/010. Superfund Information Systems, U.S. Environmental Protection Agency [online]. Available: http://www.epa.gov/superfund/sites/rods/fulltext/e1003010.pdf [accessed Jan. 10, 2007].
EPA (U.S. Environmental Protection Agency). 2005a. Five-Year Review Report: General Motors (Central Foundry Division) Superfund Site, St. Lawrence County, Town of Massena, New York. U.S. Environmental Protection Agency, Region 2, New York, NY. July 2005 [online]. Available: http://www.epa.gov/superfund/sites/fiveyear/f05-02010.pdf [accessed Jan. 10, 2006].
EPA (U.S. Environmental Protection Agency). 2005b. Second Five-Year Review Report. E. I. du Pont de Nemours & Co., Inc. (Newport Pigment Plant Landfill) Superfund Site (a.k.a. DuPont-Newport Site), Newport, Delaware. U.S. Environmental Protection Agency, Region 3, Philadelphia, PA. March 31, 2005 [online]. Available: http://www.epa.gov/superfund/sites/fiveyear/f0503006.pdf [accessed Jan. 10, 2007].
EPA (U.S. Environmental Protection Agency). 2005c. Second Five-Year Review Report: Harbor Island Superfund Site, Seattle, King County, Washington. U.S. Environmental Protection Agency, Region 10, Seattle, WA. September 2005 [online]. Available: http://yosemite.epa.gov/R10/CLEANUP.NSF/9f3c21896330b4898825687b007a0f33/5a64831b6521f46b8825650200836f1c/$FILE/2nd%205%20Year%20Review%209-27-05.pdf [accessed Jan. 10, 2007].
EPA (U.S. Environmental Protection Agency). 2006a. Case Study Data and Information. U.S. Environmental Protection Agency.
EPA (U.S. Environmental Protection Agency). 2006b. Five-Year Review Report. Third Five-Year Review Report for Bayou Bonfouca Superfund Site, Slidell, St., Tammany Parish, Louisiana. U.S. Environmental Protection Agency, Region 6, Dallas, TX. May 2006 [online]. Available: http://epa.gov/earth1r6/6sf/pdffiles/bb_5yr_2006-0707.pdf [accessed Jan. 10, 2007].
EPA (U.S. Environmental Protection Agency). 2006c. Five-Year Review Report: Reynolds Metals Company Site, St. Lawrence County, Town of Massena, New York. U.S. Environmental Protection Agency, Region 2, New York, NY. April 2006 [online]. Available: http://www.epa.gov/superfund/sites/fiveyear/f06-02018.pdf [accessed Jan. 10, 2007].
EPA (U.S. Environmental Protection Agency). 2007a. Sediment Assessment and Remediation Report Assessment of Sediment Quality in the Black River Watershed. Final Report. U.S. Environmental Protection Agency, Great Lakes Contaminated Sediments [online]. Available: http://www.epa.gov/glnpo/sediment/BlackRiver/FinalReport.htm [accessed May 4, 2007].
EPA (U.S. Environmental Protection Agency). 2007b. Outboard Marine Corp. EPA ID: ILD000802827. NPL Fact Sheet. U.S. Environmental Protection Agency, Region 5 Superfund [online]. Available: http://www.epa.gov/region5/superfund/npl/illinois/ILD000802827.htm [accessed May 4, 2007].
Exponent. 2005. 2004 Monitoring Report for Sediment Remediation in Ward Cove, Alaska. Prepared for Ketchikan Pulp Company, Ketchikan, AK, by Exponent, Bellevue, WA. June 2005 [online]. Available: http://yosemite.epa.gov/R10/CLEANUP.NSF/9f3c21896330b4898825687b007a0f33/2dd5ab7462e4f004882567b30057eb7b/$FILE/KPC_monrpt_2004.pdf [accessed May 4, 2007].
Fort James Corporation, Foth &Van Dyke, and Hart Crowser Inc. 2001. Final Report 2000 Sediment Management Unit 56/57 Project, Lower Fox River, Green Bay, Wisconsin. Prepared for U.S. Environmental Protection Agency and Wisconsin Department of Natural Resources, by Fort James Corporation, Foth &Van Dyke, and Hart Crowser Inc. January 2001 [online]. Available: http://dnr.wi.gov/org/water/wm/foxriver/documents/finalreport/final_report.pdf [accessed Jan. 10, 2007].
Foth & Van Dyke. 2000. Summary Report: Fox River Deposit N. Wisconsin Department of Administration, Wisconsin Department of Natural Resources. April 2000 [online]. Available: http://dnr.wi.gov/org/water/wm/foxriver/documents/sediment/depositn_report.pdf [accessed Jan. 10, 2007].
Fox, R., M. Jury, and J. Kern. 2006. Overview of Lower Fox River OU1 Dredge Residuals. Presentation at the Second Meeting on Sediment Dredging at Superfund Megasites, June 7, 2006, Irvine, CA.
Fox River Group. 1999. Effectiveness of Sediment Removal. September 27, 1999 (Appendix C in GE 2000).
Goovaerts, P. 1997. Geostatistics for Natural Resources Evaluation. New York: Oxford University Press.
Green, G., V. Buhr, and M. Binsfeld. 2007. Controlling Dredge Cut Accuracy, Sediment Resuspension, and Sediment Dewatering: Fox River OU1 Contaminated Sediment Project. Presentation at the 4th International Conference on Remediation of Contaminated Sediments, January 22-25, 2007, Savannah, GA.
Helsel, D.R. 2005. Nondetects and Data Analysis: Statistics for Censored Environmental Data. Hoboken, NJ: Wiley. 250pp.
Herrenkohl, M., L. Jacobs, J. Lally, G. Hartman, B. Hogarty, K. Keeley, J. Sexton, and S. Becker. 2006. Ward Cove sediment remediation project revisited: Long-term success of thin-layer placement remedy. Pp. 421-431 in Proceedings of the Western Dredging Association 26th Technical Conference and 38th Annual Texas A&M Dredging Seminar, June 25-26, 2006, San Diego, CA, R.E. Randell, ed. Center for Dredging Studies, Ocean Engineering Program, Civil Engineering Department, Texas A&M Univer-sity, College Station, TX.
Hollander, M., and D. Wolfe. 1973. Nonparametric Statistical Inference. New York: Wiley.
ILDPH/ATSDR (Illinois Department of Public Health and the Agency for Toxic Substances and Disease Registry). 2004. Health Consultation. Outboard Marine Corporation/Waukegan Harbor Waukegan, Lake County, Illinois. EPA Facility ID: ILD000802827. Illinois Department of Public Health and the Agency for Toxic Substances and Disease Registry. April 20, 2004 [online]. Available: http://www.atsdr.cdc.gov/HAC/PHA/outboard2/omc_toc.html [accessed Jan. 10, 2006].
Kemble, N.E., D.G. Hardesty, C.G. Ingersoll, B.T. Johnson, F.J. Dwyer, and D.D. MacDonald. 2000. An evaluation of the toxicity of contaminated sediments from Waukegan Harbor, Illinois, following remediation. Arch. Environ. Contam. Toxicol. 39(4):452-461.
King County. 1999. King County Combined Sewer Overflow Water Quality Assessment for the Duwamish River and Elliott Bay. King County Department of Natural Resources, Seattle WA; and Parametrix, Inc. Kirkland, WA. February 1999 [online]. Available: http://dnr.metrokc.gov/wlr/waterres/wqa/wqrep.htm [accessed April 25, 2007].
Levinton, J.S., S.T. Pochron, and M.W. Kane. 2006. Superfund dredging restoration results in widespread regional reduction in cadmium in blue crabs. Environ. Sci. Technol. 40(24):7597-7601.
Mackie, J.A., S.M. Natali, J.S. Levinton, and S.A. Sanudo-Wilhelmy. 2007. Declining metal levels at Foundry Cove (Hudson River, New York): Response to localized dredging of contaminated sediments. Environ. Pollut. 149(2):141-148.
Malcolm Pirnie, Inc. and TAMS Consultants, Inc. 2004. Final Engineering Performance Standards Hudson River PCBs Superfund Site, Appendix: Case Studies of Environmental Dredging Projects. Prepared for U.S. Army Corps of Engineers, Kansas City District, by Malcolm Pirnie, Inc., White Plains, NY, and TAMS Consultants, Inc., Bloomfield, New Jersey. April 2004 [online]. Available: http://www.epa.gov/hudson/eng_perf/FP5001.pdf [accessed Aug. 16, 2006].
MIDEQ (Michigan Department of Environmental Qualilty). 1996. Manistique River Area of Concern Manistique, Michigan Remedial Action Plan Update. February 20, 1996 [online] Available: http://www.epa.gov/glnpo/aoc/mantique/1996_Manistique_RAP.pdf. [accessed May 4, 2007].
Montgomery Watson. 2001. Final Summary Report: Sediment Management Unit 56/57 Demonstration Project, Fox River, Green Bay, Wisconsin. Prepared for Fox River Group and Wisconsin Department of Natural Resources, By Montgomery Watson. September 2001 [online]. Available: http://dnr.wi.gov/org/water/wm/foxriver/documents/finalreport/final_summary_report.pdf [accessed Jan. 10, 2007].
Nadeau, S. 2006. SMWG Review and Analysis of Selected Sediment Dredging Projects. Presentation at the Second Meeting on Sediment Dredging at Superfund Megasites, June 7, 2006, Irvine, CA.
Nadeau, S.C., and M.M. Skaggs. 2007. Analysis of Recontamination of Completed Sediment Remedial Projects. Paper D-050 in Remediation of Contaminated Sediments-2007: Proceedings of the 4th International Conference on Remediation of Contaminated Sediments, January 2007, Savannah, GA. Columbia, OH: Battelle Press.
NRC (National Research Council). 2001. A Risk-Management Strategy for PCB-Contaminated Sediments. Washington, DC: National Academy Press.
NYSDEC (New York State Department of Environmental Conservation). 1997. Record of Decision: Cumberland Bay Sludge Bed-Wilcox Dock Site (OU-1), Plattsburgh, Clinton County Site Number 5-10-017. New York State Department of Environmental Conservation, Division of Environmental Remediation. December 1997.
NYSDEC (New York State Department of Environmental Conservation). 2001. Cumberland Bay Sludge Bed Removal Project. New York State Department of Environmental Conservation. April 2001.
Ortiz, E., R.G. Luthy, D.A. Dzombak, and J.R. Smith. 2004. Release of polychlorinated biphenyls from river sediment to water under low-flow conditions: Laboratory assessment. J. Environ. Engin. 130(2):126-135.
Patmont, C. 2006. Contaminated Sediment Dredging Residuals: Recent Monitoring Data and Management Implications. Presentation at the Second Meeting on Sediment Dredging at Superfund Megasites, June 7, 2006, Irvine, CA.
Patmont, C., and M. Palermo. 2007. Case Studies Environmental Dredging Residuals and Management Implications. Paper D-066 in Remediation of Contaminated Sediments-2007: Proceedings of the 4th International Conference on Remediation of Contaminated Sediments, January 2007, Savannah, GA. Columbia, OH: Battelle Press.
Qian, P.Y, J.W. Qiu, R. Kennish, and C.A. Reid. 2003. Recolonization of benthic infauna subsequent to capping of contaminated dredged material in East Sha Chau, Hong Kong. Estuar. Coast. Shelf S. 56(3-4):819-831.
Reible, D.D, , D. Hayes, C. Lie-Hing , J. Patterson, N. Bhowmik, M. Johnson, and J. Teal. 2003. Comparison of the long-term risks of removal and in-situ management of contaminated sediments in the Fox River. Soil Sediment Contam. 12(3):325-344.
Stern, J., and C.R. Patmont. 2006. Evaluation of Post-Dredge Monitoring Results to Assess Net Risk Reduction of Different Sediment Cleanup Options. Presentation at the SETAC North America 27th Annual Meeting, Nov. 5-9, 2006, Montreal, Canada.
Stern, J., C.R. Patmont, and D. Hennessy. 2007. Evaluation of Post-dredge Monitoring Results to Assess Net Risk Reduction of Different Sediment Cleanup Options. Presentation at 2007 Georgia Basin Puget Sound Research Conference, March 26–29, 2007, Vancouver, BC.
Steuer, J.J. 2000. A Mass-Balance Approach for Assessing PCB Movement During Remediation of a PCB-Contaminated Deposit on the Fox River, Wisconsin. U.S. Geological Survey Water Resources Investigation Report 00-4245. U.S. Department of the Interior, U.S. Geological Survey. December 2000 [online]. Available: http://www.dnr.state.wi.us/ORG/water/wm/foxriver/cuments/sediment/usgs_5657_report.pdf [accessed Jan. 11, 2007].
Stow, C.A., S.R. Carpenter, L.A. Eby, J.F. Amrhein, and R.J. Hesselberg. 1995. Evidence that PCBs are approaching stable concentrations in Lake Michigan fishes. Ecol. Appl. 5(1):248-260.
TetraTech. 2004. Final Interim Remedial Action Report for Site 1—McAllister Point Landfill, Operable Unit 4—Marine Sediment/Management of Migration, Naval Station Newport, Middletown, Rhode Island. Prepared for Department of the Navy, Engineering Field Activity, Northeast Naval Facilities Engineering Command, Lester, PA, by Tetra Tech. FW, ICN, Langhorne, PA. September 28, 2004.
TetraTech NUS. 2006. Round 1: December 2004 Long-Term Monitoring Report for McAllister Point Landfill Naval Station Newport Newport, Rhode Island. Prepared for Engineering Field Activity Northeast Naval Facilities Engineering Command, Lester, PA. Contract Number N62472-03-D-0057. March 2006.
Thibodeaux, L.J., and K.T. Duckworth. 1999. The effectiveness of environmental dredging: A study of three sites. Remediat. J. 11(3):5-33.
URS (URS Woodward Clyde Group Consultants Inc.). 1999. Construction Monitoring Report Christina River Remediation: Newport Superfund Site, Newport, DE. Prepared for DuPont Corporate Remediation Group, Wilmington, Delaware, by URS Woodward Clyde Group Consultants Inc., Buffalo, New York. December 1999.
URS (URS Group Inc.). 2006. Final 2003 Marine Monitoring Report, OU B Marine, Bremerton Naval Complex, Bremerton, Washington. Prepared for Naval Facilities Engineering Command Northwest, Poulsbo, WA, by URS Group, Inc., Seattle, WA. February, 2, 2006.
USACE (U.S. Army Corps of Engineers). 1992. Remediation of Marathon Battery Superfund Project: Cold Spring, New York, Supplemental Information. IFB No. DACW41-92-B-0030. U.S. Army Corps of Engineers, Kansas City District. May 1992.
USACE (U.S. Army Corps of Engineers). 1995. New Bedford Hot Spot after Dredge Sediment Sampling. Memorandum to Dave Dickerson, U.S. Environmental Protection Agency, from Mark Otis, U.S. Army Corps of Engineers. CENED-PD-E. April 19, 1995.
USACE (U.S. Army Corps of Engineers). 2005. Benthic Recovery Review. Draft Technical Note ERDC TN-DOER-E. August 2005.
Weston (Weston Solutions Inc.). 2002. Final Comprehensive Post-Removal Summary Report Manistique Harbor and River Site U.S. Route 2 Highway Bridge Manistique, Michigan. Prepared for the U.S. Environmental Protection Agency, Region 5. November 12, 2002.
Weston (Weston Solutions Inc.). 2005a. Final Manistique Harbor and River Site, Manistique, Michigan, Data Evaluation Report: Revision 1-19 May 2005. Document No. RWW236-2A-ATGN. Prepared for U.S. Environmental Protection Agency, Chicago, IL.
Weston (Weston Solutions Inc.). 2005b. Human Health Risk Assessment GE/Housatonic River Site, Rest of River, Vol. 1. DCN:GE-021105-ACMT. Environmental Remediation Contract, GE/Housatonic River Project, Pittsfield, MA. Prepared for U.S. Army Corps of Engineers, New England District, Concord, MA, and U.S. Environmental Protection Agency, Boston, MA, by Weston Solutions Inc., West Chester, PA. February 2005 [online]. Available: http://www.epa.gov/NE/ge/thesite/restofriver/reports/hhra_219190/219190_HHRA_Vol1.pdf [accessed May 7, 2007].
WI DNR/EPA (Wisconsin Department of Natural Resources and the U.S. Environmental Protection Agency). 2002. Record of Decision, Operable Unit 1 and Operable Unit 2, Lower Fox River and Green Bay. Wisconsin Department of Natural Resources, Madison, WI, and the U.S. Environmental Protection Agency, Region 5, Chicago, IL. December 2002 [online]. Available: http://dnr.wi.gov/org/water/wm/foxriver/documents/whitepapers/Final_Record_of_Decision_OUs_1-2.pdf [accessed Jan. 11, 2007].
Windward. 2005. Lower Duwamish Waterway Remedial Investigation. Data Report: Fish and Crab Tissue Collection and Chemical Analyses. Final Report. Prepared for U.S. Environmental Protection Agency, Region 10, Seattle WA, and Washington State Department of Ecology, Bellevue, WA, by Windward Environmental LLC, Seattle, WA. July 27, 2005 [online]. Available: http://www.ldwg.org/assets/fish_crab_tissue/final_fish%2bcrab_tissue_dr.pdf [accessed April 25, 2007].
Windward. 2006. Lower Duwamish Waterway Remedial Investigation. Data Report: Chemical Analyses of Fish and Crab Tissue Samples Collected in 2005. Final Report. Prepared for U.S. Environmental Protection Agency, Region 10, Seattle WA, and Washington State Department of Ecology, Bellevue, WA, by Windward Environmental LLC, Seattle, WA. April 19, 2006 [online]. Available: http://www.ldwg.org/assets/fish_crab_tissue/2005dr/2005_fish%2bcrab_dr.pdf [accessed April 25, 2007].
Zarull, M.A., J.H. Hartig, and L. Maynard. 1999. Ecological Benefits of Contaminated Sediment Remediation in the Great Lakes Basin. Windsor, Ontario: International Joint Commission [online]. Available: http://www.ijc.org/php/publications/html/ecolsed/ [accessed Sept. 12, 2006].