The objective of this project was to demonstrate an integrated approach to fingerprint sediment polychlorinated biphenyl (PCB) contamination that combines sediment screening technologies on a large number of field samples followed by detailed PCB congener analysis with advanced chemical fingerprinting (ACF) data interpretation on a subset of selected laboratory samples to identify PCB sources to sediments. The current alternative approach without a forensics study merely assumes the most visible landholder closest to the sediment contamination is responsible. The performance assessment shows the PCB analytical measurements meet the required Data Quality Objectives (DQOs) and the investigation techniques generate reproducible results with simple artificial datasets as well as the real demonstration site datasets. Investigation techniques (including receptor models) were successful in reproducing original source signatures from artificial datasets that were constructed by mixing the original sources in varying proportions. Additionally, a real dataset from a demonstration site was used to show different investigation techniques produce comparable source results.
The demonstration integrated the following two technology components: (1) rapid sediment characterization (RSC) technologies that enable wide spatial and temporal coverage to delineate sediment contaminant gradients and semi-quantitative characterization in a cost-effective manner, and (2) ACF on a selected subset of samples to delineate sources. ACF includes both advanced laboratory chemical analysis of samples and the application of sophisticated data analysis and interpretation methods. The combined use of RSC and ACF, however, are only two steps in the overall Integrated Forensics Approach that is modified from earlier fingerprinting work. The overall sequence of steps employed in this project include evaluation of the site’s potential as a demonstration site, development of a conceptual site model, development and implementation of a defensible study design, demonstration of RSC screening, demonstration of ACF, and synthesis and presentation of the results.
The forensic study results from the two demonstration sites indicate both sites have multiple PCB sources to the sediments that were successfully discriminated by the demonstrated Integrated Forensics Approach. At Hunters Point Shipyard (HPS), there are three source patterns that appear to originate from two different areas. In the more recent surface sediment on the east side of the embayment near the former landfill, an Aroclor 1260 pattern is clearly seen, which appears to originate from the upland landfill area. In the more recent surface sediments on the west side of the embayment near Yosemite Creek, a mix of Aroclors 1260/1254/1248 is present and appears to be from combined sewer overfall outfalls. A third pattern appears more common in the deeper sediments (prior to 1970 when the highest PCB concentrations are present) from both sides of the embayment that is approximately a 50% mix of Aroclors 1260/1254.
At Ashtabula River, there are four PCB compositional patterns that appear to originate from two different areas. Most of the sediments in the dredge area contain an Aroclor 1248 pattern, some of which show varying amounts of a second dechlorination pattern. These two patterns probably represent the same source of PCBs and are reported to be from Fields Brook. A third deeper sediment pattern also believed to be from Fields Brook is seen in most cores (in sediments deposited prior to the last dredge event in the early 1960s) with a unique pattern enriched in highly chlorinated PCB congeners (e.g., PCB209). A fourth very recent pattern contributing Aroclor 1260 is observed in surface sediments and has been traced back to a drainage creek discharging from the west side of the Ashtabula River opposite from Fields Brook. When PCB mass is considered in addition to just PCB proportions, the Ashtabula dredge area shows greater than 99% of the PCB mass with compositional patterns that can be traced back to sources in Fields Brook.
The similarity in the three source patterns from HPS makes the apportionment less certain at that site compared to the four source patterns at Ashtabula, which show more unique characteristics that allow more precise apportionment.
The most important aspects to conducting a successful forensics study are to have high quality PCB data and experienced analysts interpreting the results. It is also important to have a technically defensible approach so that the Department of Defense’s fingerprinting case is fairly represented. It does not appear to matter which particular investigation technique (specific receptor model) is used, and in fact a weight of evidence approach that looks at a number of independent lines of evidence is probably best. Also important is to present the information in a number of ways to ensure it is communicated to an audience with varying levels of forensic expertise (such as receptor model details presented to forensic experts and spatial displays such as contour maps of PCB concentrations or source proportions presented to the general public). All of these techniques are commercially available from multiple sources. The analytical techniques are available from contractors that can provide high quality data. The number of individual PCB congeners that can be expected to be above detection limits will be related to the total concentration of the samples, and samples with total PCB concentrations below about 100 ppb will typically not have enough congeners above detection limits to be useful for forensic study. The data analysis and interpretation techniques are also widely available from multiple contractors. These contractors should also have experience with many types of visual displays that can be used to present the data in the most comprehensible fashion. Additional guidance on conducting a fingerprinting study can be found in a companion user’s guide on the Navy’s SPAWAR web site.