Hydrogeologic Features of Exposure at Camp Lejeune

On the basis of geophysical data and lithologic logs, several productive aquifers were found to exist beneath Camp Lejeune. The geologic cross-sectional details on the site, as reported in Harden et al. (2004), are summarized in Figure 2-2. The aquifers include the Castle Hayne aquifer and two other deep aquifers beneath the Beaufort confining unit, the Beaufort and Peedee aquifers. All the water-supply wells were installed within the Castle Hayne aquifer, so site characterization efforts focused on understanding the hydrostratigraphy of the upper three hydrogeologic units: the surficial aquifer, the Castle Hayne confining unit, and the Castle Hayne aquifer. Each unit is known to have multiple subunits that consist of seams of clay, silt, and sandy beds (as indicated in Figure 2-2). The sections below summarize the available hydrogeologic data for the three units.

FIGURE 2-2

Geologic cross section of Camp Lejeune. Source: Hardenet al. 2004.

Characteristics of Source Zones

Predicting the dynamics of contaminant transport from contaminant source zones requires the use of groundwater models that simulate a complex set of fate and transport processes. Results from these models should be interpreted in light of a conceptual framework that integrates the chemical and geologic complexities in sources and receptors to establish a relationship between the contaminant source and the groundwater wells. An example of such a source-receptor conceptual model for a waste site contaminated with volatile organic compounds (VOCs) like PCE or TCE is illustrated in Figure 2-3.

FIGURE 2-3

Conceptual model of DNAPL transport. The well is shown at an exaggerated scale. Source: Modified from Jackson 1998. Reprinted with permission; copyright 1998, Hydrogeology Journal.

At a typical waste site, spent VOCs are present in the unsaturated zone (a partially saturated soil layer above the water table) in the form of dense nonaqueous-phase liquids (DNAPLs). Pure-phase VOCs are DNAPLs that do not mix with water and have an “oily” texture. They can be trapped in soil pore spaces, and their dissolution (dissolving process) is limited by a complex set of mass-transfer processes (Miller et al. 1991; Jackson 1998; Clement et al. 2004b). Furthermore, considerable spatial variability in DNAPL mass distribution in a source region is almost inevitable; consequently, mass detection at DNAPL-contaminated field sites is extremely difficult and uncertain (Abriola 2005).

Laboratory-scale tank studies have indicated that under typical groundwater-flow conditions the DNAPL dissolving process will be limited by various mass-transfer processes, so concentrations of only about 10-20% of the maximum solubility level can be obtained (Clement et al. 2004a). Furthermore, waste DNAPLs, similar to the ones disposed of at Camp Lejeune, may mix with other chemicals that limit the mass-transfer kinetics further and lead to considerable reduction in solubility (Clement et al. 2002). Therefore, the presence of even a small volume of DNAPL can contaminate a large volume of groundwater for several decades as DNAPL continues to dissolve.

Figure 2-3 illustrates various possible pathways for groundwater contamination from a DNAPL source. If the quantity of the waste product (DNAPL) is high enough, the waste will migrate downward and penetrate the water table. The vertical migration will eventually cease, and the DNAPL will be trapped in the pore spaces or will pool over low permeable clay layers. The DNAPL phase will slowly dissolve into the water phase, and the dissolved plume will be transported toward the extraction wells. The migration patterns of DNAPL contaminants will also be highly influenced by local hydrogeologic conditions. The presence of low-permeability units (such as the Castle Hayne confining unit or any clay units) would limit vertical migration of both DNAPL and dissolved contaminants. At Camp Lejeune, all the groundwater-supply wells are beneath the surficial aquifer. Therefore, the ability of the contaminants to reach the receptor (well screen) at the site depends on local groundwater gradients, on the thickness (or existence) and geometry of the low-permeability clay or silt zones between the source and the well, and on the geometry of the hydrostratigraphic units. The presence of a thick clay unit between the source and the receptor retards transport; however, strong pumping could induce vertical gradients and enhance contaminant transport.

Water-Treatment Plants and Distribution System

A chronology of the water-supply systems providing water to the residential areas at Camp Lejeune from 1941 to 2000 is presented in Table 2-1. At various times, four systems have been the primary sources of water for residences other than barracks at Camp Lejeune since the first system was put into service: Hadnot Point, Tarawa Terrace, Marine Corps Air Station, and Holcomb Boulevard. Several smaller systems have supplied or still supply other areas of the base that have relatively low populations. For each system, a set of supply wells pumped water to a centralized water-treatment plant, where the water was mixed before distribution to housing areas, public buildings (such as schools), businesses, and workplaces.

TABLE 2-1

Water Supply of Housing Areas, Camp Lejeune, North Carolina (1941-2000).

Figure 2-4 provides an illustration of a conceptual model of a water-supply system at Camp Lejeune. Water-supply wells collected groundwater and pumped it to the water-treatment plant when the wells were turned on. Not all the wells operated at the same time. The wells were “cycled,” meaning that only a few wells pumped water to the treatment plant at any given time. Water from several wells was mixed at the treatment plant and processed before being distributed in the pipes that supplied water to the base. Limited historical information is available on the pumping schedules of the wells or the water-treatment techniques that were used.

FIGURE 2-4

Conceptual model of a Camp Lejeune water system. (1) The drinking water at Camp Lejeune is obtained from groundwater pumped from a freshwater aquifer located approximately 180 ft below the ground. (2) Groundwater is pumped through wells located near the (more...)

In general, the water-treatment processes used by the Marine Corps generally included coagulation, sedimentation, filtration (with sand or anthracite), and lime softening (Marine Corps, personal commun., May 22, 2008). The American Water Works Association (AWWA) reported that efficiency of removal of VOCs would be poor (0-20%) without lime softening and poor to fair (0-60%) with lime softening, of synthetic organic chemicals poor to good (0-80%), and of metals good to excellent (80-100%) except for chromium⁺⁶ (less than 20%) (AWWA 1995). Actual removal efficiencies are site-specific and depend on how each water-treatment plant is operated.

Review of Contaminated Areas

The committee evaluated data on hazardous-waste site locations and characteristics in the vicinity of the water-supply well and residential service locations for the water systems listed in Table 2-1 (Baker Environmental, Inc 1999, CH2M Hill and Baker Environmental, Inc 2005). Table 2-2 summarizes the contaminants found in soil or groundwater at waste sites near supply wells. Details of the contamination near supply wells serving Tarawa Terrance and Hadnot Point are presented later in this chapter. Waste sites in the vicinity of other water-supply areas are described briefly in Appendix C (Table C-1).

TABLE 2-2

Contaminants Found in Soil or Groundwater at Hazardous Waste Sites Near Water-Supply Wells.

Discovery and Investigation of the Contamination at Tarawa Terrace

The Tarawa Terrace water-supply system began operations in 1952. Seven wells initially supplied water to the system, and more wells were added over the years. A total of 16 wells served the system at some time between 1952 and 1987. The wells operated on a cycled schedule. Wells were taken offline or were closed for various reasons between 1962 and 1987 (Maslia et al. 2007).

During August 1982, a routine analysis with gas chromatography-mass spectrometry (to screen the water samples collected from the Tarawa Terrace water-treatment plant for chlorination byproducts) indicated high concentrations of halogenated hydrocarbons, a class of VOCs (Faye and Green 2007). Further analysis confirmed the presence of PCE in finished water at 76-104 μg/L (Faye and Green 2007). Sporadic sampling in 1982-1985 also indicated detectable concentrations of TCE, which is a degradation byproduct of PCE.

In January 1985, the North Carolina Department of Natural Resources and Community Development (NCDNRCD) began routine sampling of water from supply wells TT-23, TT-25, and TT-26 and finished water from the water-treatment plant (Faye 2008). The data indicated varied PCE and TCE contamination. For example, PCE ranged from nondetectable to 132 μg/L and from 3.8 to 1,580 μg/L in wells TT-23 and TT-26, respectively. Wells TT-23 and TT-26 were temporarily removed from service in February 1985. Later, well TT-26 was closed permanently, and well TT-23 was used intermittently for several days during March and April 1985 and finally shut down in April 1985 (GAO 2007). From January to September 1985, samples were taken from wells TT-30, TT-31, TT-52, TT-54, and TT-67, and PCE and its degradation products were not detected.

In April 1985, NCDNRCD conducted extensive field investigation to map the PCE plume and identify the contaminant source. On the basis of that investigation, the northwest edge of the plume was determined to be close to ABC One-Hour Cleaners. A shallow monitoring well installed close to the cleaners detected an extremely high PCE concentration of 12,000 μg/L (Faye and Green 2007). Such a high concentration is an indication of a source region that contains pure-phase PCE (the highest possible concentration of PCE in water is about 110,000 μg/L). Further investigations revealed that ABC One-Hour Cleaners had routinely used PCE in dry-cleaning operations since 1953. Shiver (1985) reported that PCE releases from various accidental spills entered the septic system through a floor drain. Furthermore, spent PCE was routinely put through a filtration-distillation process that produced dry still bottoms (sludge). Until about 1982, such waste products were used to fill potholes in a nearby alleyway. The exact date of the termination of those disposal practices is unknown; ATSDR estimates that they ceased in 1985 (Faye and Green 2007).

Several on-base sources and episodes were documented. Faye and Green (2007) report that a “strong gasoline type odor” was noted at water-supply well TT-53 during October 1986 while personnel from the U.S. Geological Survey (USGS) conducted a routine well reconnaissance. The well was not in service at the time. The gasoline contamination was traced to various spills and leaks from 12 underground storage tanks (USTs) associated with various buildings in the Tarawa Terrace shopping center. For example, on September 21, 1985, a catastrophic failure discharged about 4,400 gal of unleaded gasoline to the subsurface. A review of past releases indicated that small leaks of gasoline products probably occurred at the site beginning in the 1950s. As of May 4, 1987, more than 2 ft of floating gasoline was determined to be present above the water table in the vicinity of Building TT-2453.

Investigation of groundwater contamination due to sources other than the ABC One-Hour Cleaners began after 1990 (Faye and Green 2007). The investigations focused on above-ground petroleum-storage tanks, buildings that housed filling stations, and USTs. The above-ground tanks were between State Route 24 and the railroad tanks near water-supply wells TT-27 and TT-55. They were constructed in 1942 and stored petroleum until about 1980, when they were converted to waste-oil storage. Most of the remedial investigations of buildings and USTs focused on areas in or near the Tarawa Terrace shopping center. Information on the installation, use, and release histories of the USTs is sparse. At least some of the tanks may have been constructed as early as the 1950s. High concentrations of benzene and toluene were measured in samples taken from monitoring wells, and several benzene plumes were mapped as a result of those investigations (see Faye and Green 2007, Table E9 and Figures E7 and E9).

Other Contaminants of Concern at Tarawa Terrace

PCE is the primary contaminant at the Tarawa Terrace site, but other contaminants have been detected in supply wells, including TCE, 1,1-dichloroethylene (DCE), cis- and trans-1,2-DCE, benzene, toluene, and vinyl chloride. Many of these contaminants—including TCE, DCE, and vinyl chloride—may have resulted from degradation of PCE. Microorganisms in the subsurface degrade PCE to TCE under favorable anaerobic conditions. TCE later degrades to DCE (primarily cis-1,2-DCE [Bradley 2003]); similarly, DCE degrades to vinyl chloride and eventually to ethane, an innocuous degradation product (Bradley 2003; Clement et al. 2000; Clement et al. 2002). Some of the chlorinated compounds (including TCE, DCE, and vinyl chloride) can also be aerobically oxidized to yield carbon dioxide (Clement et al. 2000; Bradley 2003). At the ABC One-Hour Dry Cleaners site, water samples from monitoring wells in the waste-disposal zone contained TCE at concentrations up to 690 μg/L and total DCE at up to 1,200 μg/L on April 23, 1992 (Faye and Green 2007). The highest measured concentrations of TCE and total DCE in the Tarawa Terrace supply wells were 62 μg/L (estimated value on July 11, 1991) and 92 μg/L (measured value on January 16, 1985), respectively (Faye and Green 2007).

Water-Quality Data on the Tarawa Terrace System

ATSDR (Faye and Green 2007) lists 16 wells that served the Tarawa Terrace water-supply system. Two of them (TT-26 and TT-23 [also referred to as TT New Well]) were shut down on February 8, 1985, because of PCE contamination (GAO 2007). However, well TT-23 was used briefly after that date—at least on March 11-12, 1985, and on April 22, 23, and 29, 1985 (GAO 2007). ATSDR indicates that the well was removed from service in May 1985. Table 2-3 presents the PCE concentrations found in samples taken from various supply wells, including TT-23 and TT-26. Well TT-26 was highly contaminated. The highest concentration (1,580 μg/L) was obtained while the well was in service. Concentration decreased appreciably after the well was taken off line and then increased. Well TT-23 also showed evidence of PCE contamination. Again, the highest concentration was found after a period of regular operation in January 1985, and concentration was lower in later periods; notably, concentration was higher after 24 h of continuous operation (on March 12, 1985) than at the beginning of that period of service.

TABLE 2-3

Observed Concentrations of PCE in Tarawa Terrace Water-Supply Wells.

Measurements of mixed water samples suggest that supply wells TT-23 and TT-26 were major contributors to contamination of the Tarawa Terrace water supply. ATSDR (Faye and Green 2007) summarized results of analyses of PCE, TCE, and trans-1,2-DCE measured in water samples collected from May 1982 to October 1985 at the Tarawa Terrace water-treatment plant and locations (some unknown) throughout the water-distribution system (see Table 2-4). TCE and trans-1,2-DCE were not measured in all water samples (indicated by a “-” in the table). PCE ranged from undetected to 215 μg/L; the highest reported concentration was in a water sample collected from storage tank STT-39A on February 11, 1985, several days after wells TT-23 and TT-26 were removed from service. With the exception of that sample, quantified samples were collected on dates when TT-23 or TT-26 was contributing to the water supply. Most of the analytic results listed in Table 2-4 had nondetectable concentrations of TCE and trans-1,2-DCE, but not all samples were tested for these chemicals. Before February 8, 1985, those compounds were measured in only one water sample, which contained TCE at 8.1 μg/L and trans-1,2-DCE at 12 μg/L. Similar concentrations of TCE and trans-1,2-DCE (8 and 12 μg/L, respectively) were reported in the water-storage tank sample (STT-39A, February 11, 1985).

TABLE 2-4

Summary of Selected Analyses for PCE, TCE, and trans-1,2-DCE in Water Samples Collected at Tarawa Terrace Water-Treatment Plant and Tarawa Terrace Addresses.

Faye and Green (2007) also summarized analytic results for benzene and toluene in finished-water samples collected at the Tarawa Terrace water-treatment plant in 1985 (see Table 2-5). Benzene reportedly ranged from “not detected” to 2 μg/L and toluene from “not detected” to 4 μg/L; all concentrations were below the stated laboratory detection limit of 10 μg/L. (The accuracy of values below the detection limit is less certain.) It is notable that all samples in which benzene and toluene were detected were taken after February 8, 1985, the date when the two contaminated wells were closed, except for one sample with detection of benzene taken on March 11, 1985, during a period in which well TT-23 was temporarily back in service). The low concentrations (below the detection limit) of benzene and toluene in finished water and high measurements at a few monitoring wells (Faye and Green 2007) suggest that TT-23 and TT-26 may not have been the only source of VOC contamination in the Tarawa Terrace water-supply system. Analytic results on samples collected in 1986 from the Tarawa Terrace water-treatment plant are available (for example, on a CD accompanying Maslia et al. 2007) but have yet to be summarized.

TABLE 2-5

Benzene and Toluene Concentrations in Water Samples Collected at Tarawa Terrace Water-Treatment Plant.

Groundwater Fate and Transport Modeling

ATSDR performed a historical reconstruction and analysis of the contamination of the Tarawa Terrace water-supply system. It involved analyses of groundwater flow, contaminant fate and transport (of PCE and its decay products; benzene and other petroleum contaminants were not considered), and distribution in the water system. This section provides a brief review of the groundwater-modeling efforts reported in a series of ATSDR reports, including Chapters A, B, C, D, E, F, G, and H, that were made available to the committee (Faye 2007; Faye and Green 2007; Faye and Valenzuela 2007; Lawrence 2007; Maslia et al. 2007; Faye 2008; Jang and Aral 2008; Wang and Aral 2008).

Description of ATSDR’s Modeling Efforts for Tarawa Terrace

ATSDR personnel used the USGS model MODFLOW to simulate groundwater flow at the site (Faye and Valenzuela 2007) and the U.S. Environmental Protection Agency (EPA) model MT3DMS to simulate PCE transport (Faye 2008). MODFLOW is a three-dimensional finite-difference code that is capable of simulating groundwater head distribution under both steady-state and transient-flow conditions. MT3DMS is a three-dimensional transport model that is directly coupled to MODFLOW. MODFLOW and other MODFLOW-family transport codes are well-established public-domain codes that are routinely used in court cases to simulate the fate and transport of dissolved chemicals (Denton and Sklash 2006); however, they invoke several assumptions for simulating complex DNAPL contaminants, such as PCE. For example, MT3DMS can predict the transport only of dissolved contaminants, so a key approximation was made to represent the mass dissolved from the DNAPL source. To apply MT3DMS, ATSDR replaced the highly complex DNAPL contaminated source zone with a hypothetical model node where PCE was injected directly into the saturated aquifer formation at a constant rate (1.2 kg/day).

ATSDR in collaboration with personnel from the Georgia Institute of Technology also used a groundwater simulation and optimization tool, the Pumping Schedule Optimization System (PSOpS), to evaluate the effect of pumping-schedule variations on PCE arrival at water-supply wells (Wang and Aral 2008). In addition, the team used a multiphase transport simulator, TechFlowMP, which has the capability to use first-order biodegradation kinetics to simulate the fate and transport of PCE and its byproducts TCE, DCE, and vinyl chloride (Jang and Aral 2008). Unlike the MODFLOW and MT3DMS codes, the PSOpS and TechFlowMP codes lack validation by a broad spectrum of practicing geoscientists in an open-source environment.

ATSDR combined the hydrostratigraphic units above the Castle Hayne aquifer and modeled them as a single unconfined layer. The modelers assumed this layer to be underlain by a local confining layer. The permeable Castle Hayne aquifer formation, where all the water-supply wells are, is assumed to be below that confining layer. In the model, the Castle Hayne aquifer formation is divided into five distinct units. The details of all the modeled hydrogeologic units, their assumed thicknesses, and the corresponding model layer numbers that represent the units are summarized in Table 2-6. In both MODFLOW and MT3DMS, the subsurface was conceptualized as a fully saturated flow environment with seven layers that represented various hydrogeologic conditions. The model parameters used in the flow and transport models are summarized in Table 2-7. The boundary conditions of the models included generalized head boundary in the northern and northeastern edges of the model, no flow boundary in the western edge (which followed a natural divide), and constant head boundary conditions in the southern edge and part of the southeast direction. On the basis of rainfall data, an average recharge to the aquifer was estimated to be 13.2 in/year. The DNAPL source zone was represented by using a model node where PCE was injected continuously into the unconfined model layer-1 of the saturated zone at a constant rate of 1.2 kg/day (Faye 2008).

TABLE 2-6

Assumed Thickness and Layer of Castle Hayne Aquifer Units.

TABLE 2-7

Calibrated Model Parameter Concentrations Used to Simulate Groundwater Flow and Contaminant Fate and Transport in Tarawa Terrace and Vicinity.

ATSDR calibrated the MODFLOW and MT3DMS models for Tarawa Terrace by using a “hierarchical process” that included the simulation of the following four successive scenarios: (1) predevelopment (before the 1950s) flow conditions without pumping, (2) transient flow conditions involving pumping, (3) fate and transport of the PCE plume, and (4) concentration of PCE at the Tarawa Terrace water-treatment plant and water-distribution system. The first two steps involved flow modeling exclusively, and the latter two steps involved combined modeling of groundwater flow and PCE transport. The groundwater-flow patterns and PCE concentration contours predicated for the surficial layer (model layer 1) for December 1984 is shown in Figure 2-5. The results of the PCE modeling study with MT3DMS indicated that the vast majority of the PCE that reached Tarawa Terrace water-treatment plant came from well TT-26. The model results show that PCE at well TT-26 exceeded EPA’s current maximum contaminant level (MCL) for drinking water of 5 μg/L as early as January 1957 and that a corresponding breakthrough of PCE in well TT-23 occurred roughly in December 1974 (Faye 2008). The model-predicted groundwater concentrations and the simulated extraction rates were used in a mixing model to evaluate the flow-weighted PCE concentration at the water-treatment plant. Those estimates indicated that the concentration of PCE in the water-treatment plant output exceeded the MCL during October or November 1957 and that the concentrations remained above the MCL until the termination of pumping at well TT-26 in 1985. On the basis of ATSDR’s model results, the estimated maximum concentration of PCE at the Tarawa Terrace water-treatment plant was 183 μg/L in March 1984. In the period November 1957-February 1987, the average concentration of PCE at the plant was 70 μg/L.

FIGURE 2-5

Simulated (a) water level and direction of groundwater flow, and (b) distribution of tetrachloroethylene (PCE), model layer 1, December 1984, Tarawa Terrace and vicinity, U.S. Marine Corps Base Camp Lejeune, North Carolina. Source: Maslia et al. 2007. (more...)

The estimated PCE concentration range should, however, be interpreted with considerable caution because comparison of the model predictions with measured data at various locations, as summarized in Table 2-8 and by Faye (2008), shows that the model predictions systematically overpredicted the point measurements in samples from supply wells TT-23 and TT-25. Also, the model results show a monotonically increasing trend, whereas the measured data are highly random. It is important to note that comparison of monthly averaged model predictions with point measurements from various locations is problematic, although this practice is not uncommon in calibration of groundwater models like this application by ATSDR (Faye 2008). Clearly, the model predictions are influenced by temporal and spatial averaging effects. In the model, the temporal variations in pumping stresses are averaged over a month, and the temporal variations in the DNAPL source release rate are averaged over a year, whereas the data on the wells’ water quality represent a single time and are relevant on a much shorter time scale—hours instead of months. Similarly, spatial variations in concentration are averaged over a relatively large control volume represented by the model grid cells (the typical volume of a computational cell in layer 1 is about 100,000 ft³), whereas the water-quality data represent spatial variations on the scale of the control volume represented by the well (estimated at about 10-1,000 ft³).

TABLE 2-8

Simulated and Observed PCE Concentrations at Water-Supply Wells and Calibration Target Range, Tarawa Terrace and Vicinity.

The modeling studies did not include any formal analysis to account for the temporal or spatial data-averaging effects. Instead, in the analysis presented by Faye (2008), the point measurements were used to set a “calibration target range” for constraining the model predictions; the range was arbitrarily set at about half the order of magnitude of the detected point measurements (Faye 2008); the actual target ranges used are shown in Table 2-8. For concentrations that are reported as nondetected, the lower target was set to 1 μg/L, and the upper limit was set at the analytic detection limit (Faye 2008).

ATSDR stated concerns about uncertainties in the pumping-schedule data used in the PCE modeling study discussed above and in the date when the MCL was predicted to be exceeded in water-supply wells and at the water-treatment plant (Faye 2008). ATSDR assumed that the major cause of uncertainty in the models was associated with pumping schedules. To address that issue, ATSDR applied PSOpS to evaluate the effect of variation in pumping schedules on the prediction of when the concentration of PCE would exceed EPA’s MCL of 5.0 μg/L (Wang and Aral 2008). Analysis of PSOpS results indicated that the change in pumping schedules would change the date when the MCL was exceeded in well TT-26 from May 1956 to August 1959 and at the water-treatment plant from December 1956 to June 1960. Because insufficient historical pumping data were available to constrain the model predictions from 1953 to 1980, the ability of the advanced optimization models to estimate the dates accurately is questionable.

Biodegradation is one of the major processes by which PCE can be removed from groundwater. Under favorable natural conditions, PCE can degrade to toxic substances. ATSDR used the multiphase research tool TechFlowMP to simulate the fate and transport of PCE with three decay products: TCE, trans-1,2-DCE, and vinyl chloride (Jang and Aral 2008). The TechFlowMP model also predicted PCE vapor concentrations. PCE biodegradation is mediated by a series of coupled reactive transport processes, primarily under highly anaerobic conditions (Bradley 2003), and little is understood about the underlying biodegradation mechanisms. There are several controversies about the types of subsurface microorganisms that could facilitate the decay reactions (Major et al. 2003; Nyer et al. 2003). Although it is not stated explicitly in the modeling reports, ATSDR made the following assumptions for the TechFlowMP simulations: (1) the entire aquifer is anaerobic (the only known biochemical condition in which PCE can degrade); (2) the aquifer has the necessary microorganisms, which are uniformly distributed; (3) the aquifer has a carbon source sufficient to support microbial growth; (4) trans-1,2-DCE is the only DCE species in the decay chain; and (5) there is no spatial variation in the microbiologic or geochemical characteristics. ATSDR indirectly invoked all those conditions by assuming, for example, a constant, first-order PCE biotransformation rate coefficient of 0.0005 day⁻¹ for all the layers in the aquifer. It is highly unlikely that that assumed biodegradation rate is applicable to the entire site. There are no microbiologic or geologic data are available to support the five assumptions. Therefore, the predicted concentrations of TCE, trans-1,2-DCE, and vinyl chloride in the Castle Haynes aquifer at the location of intake by Tarawa Terrace supply wells should be used with considerable caution.

Potential Sources of Contamination of Hadnot Point Water Supply

Descriptions of the sources of contamination and results of sampling of monitoring wells were obtained from remedial-investigation reports (Baker Environmental, Inc 1993, 1994, 1999). The reports summarize results of analyses of samples of groundwater collected during the late 1980s and early 1990s, after the contaminated wells supplying the Hadnot Point water-distribution system were closed. They also provide information on the timing and characteristics of waste-disposal practices that resulted in contamination of environmental media in the vicinity of water-supply wells. Those locations eventually required official remedial action under U.S. environmental laws, a process that continues today. In general, the water samples from the monitoring wells were analyzed for the presence of a suite of contaminants (EPA priority pollutants) and yielded insight into the fate and transport of the contaminants from the source to the groundwater. The committee used data from the remedial-investigation reports to gain a better understanding of the nature and extent of contamination and to refine the list of contaminants of concern.

The Navy initially identified 13 sites as potential sources of contaminants of the Castle Hayne aquifer in the Hadnot Point area (Baker Environmental, Inc 1999). Each site was assigned a number (installation restoration [IR] site number), and they were grouped into operable units (OUs) to facilitate investigation and data management. Most of the sites were active in the 1940s to 1970s, before implementation of more rigorous requirements governing waste tracking, handling, and disposal. The contaminants detected in soil or groundwater samples in the course of remedial investigations are summarized in Table 2-9. Figure 2-6 is a map of the sites in relation to housing areas and water-supply wells in the Hadnot Point area.

TABLE 2-9

Installation Restoration Sites in the Hadnot Point Water-Supply Area.

FIGURE 2-6

Designated hazardous-waste remedial investigation sites at Hadnot Point.

IR site 78, the Hadnot Point industrial area, has been a center of industrial activities since the 1940s. The site included as many as 75 buildings that housed such operations as maintenance shops, refueling stations, administrative offices, printing shops, warehouses, painting shops, storage yards, a steam-generation plant, and other light industry (Baker Environmental, Inc 1994, 1999). The remedial investigation for site 78 was preceded by several investigations that confirmed the presence of VOCs related to fuels and solvents in the groundwater. Those investigations were followed by ones that set the stage for the systematic sampling conducted for the remedial investigation in 1992.

Sites 6 and 82 were used for open storage beginning in the 1940s. Many types of materials were stored on site, including pesticides and polychlorinated biphenyls. Groundwater in the vicinity of sites 6 and 82 was sampled as part of the Confirmation Study (1984-1988). Chlorinated solvents were detected in shallow and deep (Castle Hayne aquifer) monitoring wells during the remedial investigation study (Baker Environmental, Inc 1993).

OU 7 comprises sites 1, 28, and 30. The French Creek liquids disposal area (site 1) is 1 mile southeast of the Hadnot Point industrial area and was used by mechanized artillery units starting in the 1940s to dispose of waste petroleum, oil, and lubricants by ground spreading (dumping). Sporadic contamination of the upper aquifer with TCE and vinyl chloride was documented during the remedial investigation process (Baker Environmental, Inc 1995).

Site 28 is a former 23-acre burn dump, operated from 1946 to 1971, south of the Hadnot Point industrial area (Baker Environmental, Inc 1995). Solid waste from industrial operations—including construction debris, industrial waste, trash, and oil-based paint—was burned on site (Baker Environmental, Inc 1995). The remedial investigation found frequent detection of semivolatile and inorganic compounds and sporadic detection of VOCs in soil samples (Baker Environmental, Inc 1995). Shallow aquifer samples from the same period revealed the presence of lead, which was detected sporadically in the deeper water (Baker Environmental, Inc 1995).

Site 30, the Snead’s Ferry Road fuel-tank sludge area, was used by contractors to clean out fuel-storage tanks. A small amount of solvents was detected in soil samples collected in 1994, but there was no indication of groundwater contamination in samples from monitoring wells (Baker Environmental, Inc 1995).

Site 88 is the location of the former on-base dry cleaners. Underground storage tanks that were installed in the 1940s, which contained Varsol (a type of mineral spirits) and PCE, were removed in 1996. In 2005, it was determined that groundwater contamination extended 50 ft below ground surface, and the resulting plume of contaminants in the groundwater extended about 500 ft to the northwest. DNAPL was present in the groundwater, but aggressive treatment has reduced concentrations (CH2M Hill and Baker Environmental, Inc 2005).

Preremedial investigation (pre-RI) site 10, which was initially identified before the institution of the remedial investigation process, was the location of the original disposal area for Camp Lejeune waste. An investigation of the site in 1998 showed low concentrations of numerous organic and inorganic contaminants in the soil and in surface water and sediment from small ponds on site. Aluminum, arsenic, chromium, nickel, lead, and vanadium were detected at high concentrations in shallow groundwater samples (Baker Environmental, Inc 1999, 2001).

Pre-RI site 12 is a 10-acre former explosive-ordnance disposal area. No substantial residual contamination was detected during the remedial investigation process (Baker Environmental, Inc 1999, 2001).

Water-Quality Data on the Hadnot Point System

Published water-sampling data on Hadnot Point are sparse. One source (GAO 2007) reported on concentrations of contaminants detected in the Hadnot Point water-supply wells before they were removed from service in 1984 and 1985 (see Table 2-10). The highest concentrations of contaminants were reported for well 651, with TCE at 3,200 μg/L, PCE at 386 μg/L, and trans-1,2-DCE at 3,400 μg/L. The committee was also made aware of a water sample not included in the 2007 GAO report that was taken from well 651 on the day it was closed—February 4, 1985; the sample contained TCE at 18,900 μg/L (Ensminger 2007; CLW 3269).

TABLE 2-10

Contaminant Concentrations in Supply Wells of Hadnot Point Water System.

Given that the water-quality data summarized in published reports were extremely sparse (for instance, see Table 2-10), the committee expanded its evaluation to assess additional data collected in the 1980s that were summarized in CLW documents. The committee reviewed a subset of the CLW documents that contained water-quality measurement data (see Appendix C, Tables C-3 and C-4, for data abstracted from the documents) for any samples connected to the Hadnot Point water supply that were collected through February 7, 1985. The subset includes 56 samples of supply-well water collected during the period November 30, 1984-February 4, 1985, and 52 samples of mixed water collected during October 21, 1980-February 7, 1985. It also includes samples collected at locations ordinarily served by the Holcomb Boulevard water-distribution system but temporarily served by the Hadnot Point water-distribution system after a fuel spill on January 27, 1985. Appendix C contains additional information about the abstraction process.

In Table 2-11, the committee presents a summary of the analytic results for the nine contaminants of concern that it identified for the Hadnot Point water system. Summary statistics of concentrations were computed only for the samples that had specific values recorded—samples listed as “non-detect,” “detect,” or “—” were excluded in these computations—and percentiles were reported only if at least five samples contained a given compound. Sample concentrations that are listed as not quantified were recorded in the source documents as D (detect) or ND (non-detect) or were not reported (shown as “—” in the data listing). Samples in which concentrations could not be quantified are summarized in Table 2-12.

TABLE 2-11

Hadnot Point Water-Supply Quality Measurements (October 1980-February 1985).

TABLE 2-12

Summary of Data on Water Samples from Hadnot Point Water System Recorded As Not Detected or Not Quantified in Table 2-11.

Of the nine analytes, the most prevalent compounds in mixed water samples collected from various locations in the Hadnot Point water-treatment plant and distribution system with measurable concentrations were TCE (31 quantified samples had a mean of 399 μg/L and a range of 1-1,400 μg/L) and trans-1,2-DCE (21 quantified samples had a mean of 169 μg/L and a range of 2-407 μg/L). PCE was quantified in four (8%) of the 52 samples. Benzene, 1,1,1-TCA, 1,1-DCE, and toluene were not detected or quantified; methylene chloride and vinyl chloride were each detected in one sample. As in the mixed water, the most prevalent compounds in potable well-water samples were TCE (17 quantified samples had a mean of 2,596 μg/L and range of 5-18,900 μg/L) and trans-1,2-DCE (14 quantified samples had a mean of 1,519 μg/L and a range of 2-8,070 μg/L). There was at least one detection of all contaminants except 1,1,1-TCA. In particular, there were a few high concentrations of PCE (maximum, 400 μg/L), benzene (maximum, 720 μg/L), methylene chloride (maximum, 270 μg/L), and vinyl chloride (maximum, 655 μg/L) in the potable well samples.

In Table 2-13, the committee provides a detailed summary of Hadnot Point area supply wells that had at least one nonzero value for at least one of the nine analytes. It shows the well number, IR sites near the well, well depth, screen interval, a summary of measured VOC concentrations, and dates of operation. Some of the water-supply samples were collected after individual wells were closed, and it is important to note that pumping can affect the degree of contamination in wells. Of the 10 wells summarized in Table 2-13, eight were closed from late 1984 through early 1985 (GAO 2007). Well 651 had the highest contamination, with detectable concentrations of TCE in all the reported samples. Well 651 also had relatively high readings of trans-1,2-DCE, PCE, and vinyl chloride. Wells 602 and 634 each had one sample with a TCE concentration above 1,000 μg/L (1,600 and 1,300 μg/L, respectively).

TABLE 2-13

Characteristics of the Hadnot Point Supply Wells With At Least One Contaminated Sample Taken Between October 1980 and February 1985.

Hadnot Point Supply-Well Operation and Implications

The supply wells for the Camp Lejeune water system were on a cycled pumping schedule; that is, generally only some of the wells were pumping raw water to the water-treatment plant at any given time (GAO 2007). Typically, pumps at various wells are scheduled to cycle on or off at different times during the day, so a dynamic mixture of water from different wells flows into the water-treatment plant and into the distribution system serving residences and other facilities. Well cycling is important to consider if one wants to understand the presence of contaminants in the distribution system inasmuch as concentrations of contaminants might vary greatly from day to day or even over the course of a single day, depending on whether contaminated wells were pumping.

The committee is aware of one document (CLW 6950) that summarizes well-cycling information during a period assumed to be November 28, 1984-February 4, 1985 (Marine Corps, personal commun., February 26, 2008). The document lists Hadnot Point well numbers and some date information (calendar days without accompanying months or years) with an “X” whenever a well pumped on a given date. If the inferred dates are correct, the document shows that individual wells operated on the average for 38% of the days over the 69-day period, with a large range of operation frequency (individual wells pumped on 0-96% of the days). On the average, 16 wells pumped each day; the range was 9-27. In Table 2-14, the committee presents the well-cycling information in combination with water-sampling data from the same period to ascertain the potential effect of well cycling on measured contaminant concentrations. To illustrate the effect of well cycling on mixed-water contamination, the committee made the highly conservative assumption that all “non-detect” samples had zero concentrations of the listed contaminants. The table indicates that 10-21 wells delivered raw water to the water-treatment plant on days when at least one mixed-water sample was analyzed. At least one well with demonstrated contamination pumped on the same day or previous 2 days from the dates when water samples were collected, but contamination in the mixed water was not detected on all dates on which a sample was collected.

TABLE 2-14

Concentrations of Contaminants in Mixed Water Samples Collected from Hadnot Point Water-Distribution System During Period of Documented Well Cycling.

TCE, PCE, trans-1,2-DCE, and methylene chloride were detected in mixed-water samples taken during November 28, 1984-February 4, 1985. Benzene, 1,1-DCE, toluene, and vinyl chloride—all of which were reportedly detected in the Hadnot Point supply-well samples—either were not included in the laboratory analysis or were not detected in measurable concentrations in mixed-water samples during that period. The two dates with the highest average TCE concentrations (463 and 618 μg/L) were the dates when well 651 was supplying water to the system on the current and/or previous 2 days; this suggests that well 651 was an important source of contamination of the Hadnot Point water-supply system. In addition, the 14 mixed-water TCE measurements in samples from one of those days (January 31, 1985) had a range of 24-1,148 μg/L.

Hadnot Point Area Monitoring Wells

The committee focused its review on some of the earliest deep-groundwater monitoring data available from the remedial-investigation reports for waste sites 6, 9, 78, and 82 in the Hadnot Point area (Baker Environmental, Inc 1994). Monitoring wells were used to collect water samples from depths of about 148-153 ft below ground surface. Screens (elevations of water-intake portals in the well pipe) in most of the wells that supplied water to the Hadnot Point water system were installed at depths of 60-190 ft below ground surface. Each supply well had three to five screens. Thus, the analytic results on water samples taken from deep monitoring wells should be representative of contamination of the Castle Hayne aquifer at a depth consistent with water withdrawal from the water supply, albeit at least 7 years after the discovery of contaminants in the Hadnot Point supply wells.

The remedial investigation of site 78 was preceded by several investigations, including an initial assessment study (1983) that identified the groundwater contamination and a confirmation study (1984-1988) that documented the presence of VOCs related to fuels and solvents in the groundwater. A later supplemental characterization step study (1990-1991) and pre-investigation study (1992) set the stage for the systematic sampling effort for the remedial investigation in 1992 (Baker Environmental, Inc 1994).

Groundwater in the vicinity of sites 6 and 82 was also sampled as part of the confirmation study (1984-1988). The remedial investigation of sites 6 and 82 included three rounds of groundwater sampling, conducted in two phases: phase 1 in 1992 and phase 2 in 1993 (Baker Environmental, Inc 1993). The investigation at each site, including groundwater sampling and analyses, continued after the publication of the remedial-investigation reports. The committee judged that the focus on the remedial-investigation reports for Hadnot Point sites was justified because they provided a reasonable snapshot of contamination closest to the period of interest.

For the remedial investigation, groundwater samples were generally analyzed for two suites of common chemical contaminants known as the “target compound list” (TCL) and the “target analyte list” (TAL). The results of the detections are summarized below; a more complete discussion is presented in Appendix C (Table C-5).

The monitoring-well data identify TCE, phenol, benzene, cis- and trans-1,2-DCE, and 1,1-DCE as the most prevalent contaminants in groundwater at the locations and screened depths of the wells. Other contaminants with multiple detections were arsenic, cadmium, 1,2-dichloroethane, and PCE. TCE, phenol, and cis- and trans-1,2-DCE had the highest prevalence of concentrations measured above their limits of detection.

Concentrations reported in the remedial-investigation reports varied widely among the well sites. For example, the concentrations of TCE in 11 samples ranged from 1.3 to 58,000 μg/L. Similarly, detections of trans-1,2-DCE ranged from 1 to 26,000 μg/L, of phenol from 2 to 22,000 μg/L, and of benzene from 6.7 to 35 μg/L. The most contaminated locations were near supply well 651, next to sites 6 and 82.

At most locations, shallow groundwater (sampled at a depth of less than 25 ft) had the greatest number of contaminant detections, including such TCL chemicals as TCE (0.5-2,100 μg/L) and fuel constituents benzene (not detected to 9,200 μg/L), toluene (not detected to 18,000 μg/L), ethylbenzene (not detected to 3,000 μg/L), xylenes (not detected to 16,000 μg/L), and naphthalene (not detected to 260 μg/L) (Baker Environmental, Inc 1993, 1994). TAL metals that were commonly detected in shallow water, with some samples at exceedingly high concentrations relative to EPA’s current MCLs, were arsenic (405 μg/L), barium (1,200 μg/L), chromium (858 μg/L), lead (126 μg/L), and manganese (714 μg/L) (Baker Environmental, Inc 1993, 1994). Only five wells of intermediate depth (about 50-75 ft) were sampled as part of the remedial investigation, and detected chemicals were generally measured at concentrations below risk-based criteria.

The results of groundwater sampling and analysis with monitoring wells provide additional information regarding the presence of contaminants in the aquifer. In many ways, the data are secondary to the analytic results on samples taken from the supply wells or the tap, at least for the purposes of understanding historical exposures. However, because the available information on such samples is sparse, it is important to consider all available data, including those from monitoring wells.

Contaminants of Concern in the Hadnot Point Water Supply

The paucity of water-quality measurements of the Hadnot Point water supply, both temporally and spatially, makes it difficult to characterize the contaminants of concern accurately. Multiple waste and operational sites have contributed to the groundwater contamination since the 1940s, so the nature of the contamination has probably varied. The few available measurements were taken during the 1980s and 1990s, decades after the contamination could have begun. The principal contaminants discovered in the wells that supplied Hadnot Point in the early 1980s were TCE and PCE. TCE, phenol, benzene, cis- and trans-1,2-DCE, and 1,1-DCE were the most prevalent contaminants in samples collected in 1992 and 1993 from deep monitoring wells. Other contaminants with multiple detections in monitoring wells were arsenic, cadmium, 1,2-dichloroethane, and PCE . The chemical 1,1,1-TCA, which was on the preliminary list of contaminants of concern compiled by the committee, is given only cursory attention in this report because it was not observed in any Hadnot Point water-quality samples collected before February 8, 1985. However, 1,1,2-trichloroethane was detected in one sample from a monitoring near well 651 at 5.8 μg/L (see Appendix C, Table C-5).

Groundwater Fate and Transport Model for Hadnot Point

ATSDR has proposed that the methods that were used for Tarawa Terrace be applied to reconstruct the historical contamination of water supplied by the Hadnot Point water-treatment plant (Maslia 2008). The proposed reconstruction will simulate the groundwater concentrations of TCE, PCE, and BTEX (benzene, toluene, ethylbenzene, and xylene). The preliminary data-screening efforts started in January 2008, and work is expected to be completed on October 2009. The study includes 10 technical tasks: analysis of data from16 sites; computation of mass of PCE, TCE, and BTEX at about six major sites; review of capacity histories of about 100 wells; statistical analysis of existing data; fate analysis; fate and transport model selection; grid design and data input; fate and transport analysis; water-distribution system analysis; and uncertainty analysis. ATSDR is also committed to providing updates on its progress by participating in external progress meetings and Community Assistance Panel meetings and by preparing and disseminating data analyses and model simulations. On the basis of work already carried out, ATSDR also indicated the following (Maslia 2008):

• Discovery of new or updated site information after the second quarter of FY 2008 that substantially alters baseline information may add time to the current timeline estimate.
• Because of the expanse of the area being modeled, computational time for fate and transport analyses may be longer than previously estimated. When model selection and grid design have been completed, a more refined estimate of required computational time will be made.

Earlier in this chapter, the committee identified several limitations in the Tarawa Terrace historical reconstruction and groundwater modeling. Because the contamination at Hadnot Point is more complex, the limitations and difficulties related to such modeling will be greater.