Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
2 Exposure to Contaminants in Water Supplies at Camp Lejeune This chapter describes the scenarios of exposure to contaminants in the water supplies at Marine Corps Base Camp Lejeune and identifies gaps in understanding of the exposures of people who lived or worked on the base while the water supplies were contaminated. First, exposure assessment for epidemi- ologic studies is discussed to set forth concepts that will be used in other chapters that review epidemi- ologic evidence (see Chapters 5 and 6). Then, an overview of the water-supply contamination scenarios at Camp Lejeune and important considerations for characterizing them are presented, including hydro- geologic features of the site, the baseâs water-treatment plants and distribution systems, contaminated ar- eas, and water-quality measurements. Finally, information on the Tarawa Terrace and Hadnot Point water systems is evaluated. EXPOSURE ASSESSMENT FOR EPIDEMIOLOGIC STUDIES In public health, the term exposure refers to contact with an agent (such as environmental con- taminant) that occurs at the boundary between a person and the environment. Exposure assessment can be defined as the qualitative or quantitative determination or estimation of the magnitude, frequency, duration, and rate of exposure of a person or a population to a chemical (ILSI 2000). Often, the focus is on identifying one or more exposure pathways and, for each exposure pathway, the source, the environmental medium through which the contaminant is transported and possibly transformed, the receptor (individual or population), how contact occurs, and the route of exposure. The goal is to deter- mine how much of a contaminant is absorbed and at what rate (the dose) so that an assessment can be made as to whether the absorbed contaminant produced or might produce an adverse biologic effect (Lioy 1990). The possible routes of exposure are inhalation, if the contaminant is present in the air; ingestion, through food, drinking, or hand-to-mouth behavior; and dermal absorption, if the contaminant can be absorbed through the skin. In the field of exposure science, research has been focused on developing methods for quantifying the uncertainty and error in the exposure assessments and on correcting the assessments for such error or uncertainty when possible. New methods are being developed to account for cumulative exposure to multiple chemicals (ILSI 2000), as are probabilistic models for cumulative and aggregate exposure assessment (for example, Nieuwenhuijsen et al. 2006) and the application of exposure modeling based on geographic information systems (Nuckols et al. 2004; Mindell and Barrowcliffe 2005; Beale et al. 2008). A well-designed epidemiologic study should have the capability to evaluate exposure in relation to an appropriate latent period of a disease and to evaluate critical windows of exposure. In most epidemiologic studies, exposure cannot be measured directly or completely, and surrogate information is used to classify study subjects into exposure groups. Good surrogates for exposure elucidate the variation 28
Exposure to Contaminants in Water Supplies at Camp Lejeune 29 in exposure in the study population while minimizing exposure misclassification (error). Misclassification of exposure is of particular concern in environmental-epidemiology studies because the health effects of environmental exposures tend to be small, and it is usually difficult to accurately estimate exposure to environmental contaminants, which can occur by multiple pathways and in multiple locations. Further- more, environmental exposures are often at low concentrations, which make biases due to exposure mis- classification more likely to affect epidemiologic results. If misclassification of exposure is not differen- tial by health outcome, it commonly biases risk estimates toward the null (that is, toward finding no association) and can cause associations to be missed (Copeland et al. 1977; Flegal et al. 1986). To evalu- ate the degree of misclassification in an epidemiologic study, it is important to consider the ability of an exposure metric to correctly classify the magnitude of exposure in the study population and to differenti- ate between those who are exposed at magnitudes that could result in adverse health effects (sensitivity) and those who are exposed at lower magnitudes (specificity). It is important to maximize specificity when the prevalence of exposure in the study population is low and to maximize sensitivity when the preva- lence of exposure is high (Nuckols et al. 2004). Exposure assessment for epidemiologic studies of the effects of water-supply contamination in- cludes two components. The first is estimation of the magnitude, duration, and variability of contaminant concentrations in water supplied to consumers. An important consideration is hydrogeologic plausibility: an association between a contaminant source and exposure of an individual or population cannot exist unless there is a plausible hydrogeologic route of transport for the contaminant between the source and the receptor (Nuckols et al. 2004). The second component is information on individual water-use patterns and other water-related behaviors that affect the degree to which exposures occur, including drinking- water consumption (ingestion) and dermal contact and inhalation related to the duration and frequency of showering, bathing, and other water-use activities. Water use is an important determinant of variability of exposure to water-supply contaminants, particularly if it varies widely in the study population. Ideally, exposure-assessment strategies include both components, but in practice it may be difficult to obtain ei- ther adequately. A number of approaches have been used to assign exposures in studies of health effects of water- supply contamination. They have ranged from measures of exposure defined by geographic region or job classification (group-level or ecologic exposure) to more sophisticated measures that yield individual ex- posure estimates. Selecting an optimal approach for a given study is dictated in part by the epidemiologic- study design, the size and geographic extent of the affected population, and the quantity and quality of available exposure-related data. The approaches that have been used in epidemiologic studies of water- supply contamination are more fully described in Chapter 6. The following sections provide information on the water-supply contamination and exposure scenarios at Camp Lejeune. WATER-SUPPLY CONTAMINATION AT CAMP LEJEUNE In the early 1940s, the U.S. Marine Corps constructed a water-distribution piping system at Camp Lejeune. The source of water in the system was, and continues to be, groundwater wells. The water- treatment processes, distribution systems, and contributing wells have been modified to accommodate the additional demand due to population growth and to improve water quantity and quality. Four water sys- temsâHadnot Point, Tarawa Terrace, Marine Corp Air Station, and Holcomb Boulevardâhave supplied water to most of the residences and workplaces (see Figure 2-1). Other water-distribution systems on the base are Onslow Beach, Courthouse Bay, Rifle Range, and Camp Johnson. In late 1984 and early 1985, Marine Corps authorities removed a number of supply wells from service in the Tarawa Terrace and Hadnot Point systems after concluding that they were contaminated with solvents (GAO 2007). The sources of contamination of the two systems were different. Investigation into the source of perchloroethylene (PCE) contamination of the Tarawa Terrace water system concluded
30 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects FIGURE 2-1 Water-distribution systems serving U.S. Marine Corps Base, Camp Lejeune, North Carolina. Source: Maslia 2005. that it was due to waste-disposal practices at ABC One-Hour Cleaners, an off-base dry-cleaning facility (Shiver 1985). The dry-cleaning site was classified as a federal hazardous-waste site during March 1989 under the Comprehensive Environmental Response, Compensation, and Liability Act, commonly known as the Superfund Act, and remedial investigation began in 1990 (Faye and Green 2007). The Agency for Toxic Substances and Disease Registry (ATSDR) completed an extensive water-modeling study to pre- dict the extent of contamination (spatially and temporally) in the period January 1951âJanuary 1994 (Faye 2008; see discussion of the modeling later in this chapter). Quantitative estimates of contaminant concentrations in the water supply from that modeling effort will be used in current and planned ATSDR epidemiologic studies of the Camp Lejeune population. A report from the U.S. Government Accountability Office (GAO 2007) states that the sources of contamination at Hadnot Point are uncertain but are likely to include many on-base sites, including land- fills and base operations where solvents and other compounds were disposed of or used. ATSDR plans to do a historical reconstruction for the Hadnot Point water-distribution system to estimate the extent of groundwater contamination of wells and the extent to which water supplies of housing and public build- ings served by this system were contaminated (M. Maslia, ATSDR, personal commun., March 12, 2008). The committee is not aware of any extensive studies concerning potential contamination of wells serving other water-supply systems on the base. Those wells directly serve the Holcomb Boulevard, Ma- rine Corps Air Station, Courthouse Bay, Camp Johnson, Camp Geiger, and Rifle Range water-supply sys- tems and several smaller systems. Some water-supply systems are connected (for example, Holcomb Boulevard and Hadnot Point), and Bove and Ruckart (2008) documents some reports of intermittent de- livery of water from the Hadnot Point system to the Holcomb Boulevard system.
Exposure to Contaminants in Water Supplies at Camp Lejeune 31 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 con- fining 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 avail- able hydrogeologic data for the three units. Surficial Aquifer The thickness of the surficial aquifer at Camp Lejeune ranges from 0 to 73 ft and averages about 25 ft (Cardinell et al. 1993). The largest observed thickness occurs in the southeastern part of Camp Le- jeune. The aquifer consists of interfingered beds of sand, clay, sandy clay, and silt of both Quaternary and Tertiary age. The clay and silt beds that occur in the surficial aquifer are thin and discontinuous. The aqui- fer is often classified into several subunits; and the extent and depth of the subunits can vary among loca- tions. For example, in the vicinity of Tarawa Terrace, three minor units have been identified in the surfi- cial aquifer (the Brewster Boulevard unit, the Tarawa Terrace unit, and the Upper Castle Hayne River bend unit). Review of available cross-sectional hydrogeologic data does not indicate any distinct demar- cation between the subunits; hence, they were conceptualized as a single surficial unit in groundwater- flow models (Faye and Valenzuela 2007). According to Winner and Coble (1989), the surficial aquifer is composed of more than 90% sand in the eastern part of the base and about 70-90% sand in the western part. The aquifer is directly recharged by infiltration from rainfall that ranged from 28 to 70 in/year during 1952-1994. Tant et al. (1974) found that the soils in Camp Lejeune have good infiltration capacity. Effec- tive groundwater recharge is estimated to range from 6.6 to 19.3 in/year. The estimated average hydraulic conductivity of the surficial aquifer in the Camp Lejeune area is about 50 ft/day (Winner and Coble 1989). Conceptually, groundwater in the shallow surficial aquifer moves from areas of high hydraulic head in interstream divides toward areas of low hydraulic head at surface-water discharge areas (Harden et al. 2004). Castle Hayne Confining Unit The Castle Hayne confining unit lies beneath the surficial aquifer, and this clayey unit is concep- tualized as the top confining layer of the Castle Hayne aquifer. However, the lithostratigraphic top of Cas- tle Hayne aquifer is not continuous, and the thickness of the confining layer ranges from 0 to 26 ft, aver- aging about 9 ft where present. Harned et al. (1989) concluded that no continuous confining unit or clay bed appears to separate the surficial and Castle Hayne aquifers except in the easternmost side of the Had- not Point area. Furthermore, the thickness and distribution of the confining clay layers observed in vari- ous cross sections summarized by Harned et al. (1989) and Cardinell et al. (1993) are similar. The thin (5- 10 ft) and discontinuous clay layers observed in several cross sections indicate that the degree of hydro- logic connection between the aquifers could be substantial (Harned et al. 1989). The vertical hydraulic conductivity of the confining material, where present, is estimated to range from 0.0014 to 0.41 ft/day (Cardinell et al. 1993).
32 FIGURE 2-2 Geologic cross section of Camp Lejeune. Source: Harden et al. 2004.
Exposure to Contaminants in Water Supplies at Camp Lejeune 33 Castle Hayne Aquifer The thickness of the Castle Hayne aquifer can range from about 200 to 400 ft. The aquifer is thinnest in the area of Camp Geiger in the northwest corner of the base and thickest in the eastern bound- ary. The bottom of the Castle Hayne aquifer is bounded by a regionally continuous clay unit, which is designated the Beaufort confining unit. All the groundwater-extraction wells in the base are in the Castle Hayne aquifer. The aquifer consists primarily of beds of sand, shell, and limestone (Winner and Coble 1989). The highly conductive material decreases from west to east across Camp Lejeune. The estimated hydraulic conductivity of the aquifer ranges from 14 to 91 ft/day (Cardinell et al. 1993). A portion of wa- ter from the surficial aquifer is able to infiltrate (move through or around) the upper confining unit, and this serves as the primary mechanism for recharging the Castle Hayne aquifer. Harned et al. (1989) also observed that in interstream areas the water level in the surficial aquifers can be 2-6 ft higher than the Castle Hayne aquifer and that the high vertical gradients can induce considerable vertical recharge. There is also some evidence of a potential for recharge of the Castle Hayne aquifer through the lower confining unit from the Beaufort aquifer (Cardinell et al. 1993). Finally, several paleostream channels have been identified within the Castle Hayne aquifer; these highly permeable, sandy channel beds can have consid- erable influence in local groundwater recharge, transport, and discharge patterns. 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. 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 âoilyâ 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 groundwa- ter 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
34 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects WELL UNSATURATED ZONE UNSATURAED ZONE SATURATED ZONE WELL SCREEN 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. 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 con- taminant transport. Water-Treatment Plants and Distribution System A chronology of the water-supply systems providing water to the residential areas at Camp Le- jeune 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. Figure 2-4 provides an illustration of a conceptual model of a water-supply system at Camp Le- jeune. 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 âcycled,â 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.
Exposure to Contaminants in Water Supplies at Camp Lejeune 35 In general, the water-treatment processes used by the Marine Corps generally included coagula- tion, sedimentation, filtration (with sand or anthracite), and lime softening (Marine Corps, personal com- mun., May 22, 2008). The American Water Works Association (AWWA) reported that efficiency of re- moval 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. TABLE 2-1 Water Supply of Housing Areas, Camp Lejeune, North Carolina (1941-2000) Housing Area Water-Treatment Plant Dates of Service Family housing areas Courthouse Bay Courthouse Bay 1942-2000 Berkeley Manor Hadnot Point 1961-1971 Holcomb Boulevard 1972-2000 Hospital Point Hadnot Point 1947-2000 Knox Trailer Park Tarawa Terrace 1952-1986 Holcomb Boulevard 1987-2000 Knox Trailer Park Expanded Holcomb Boulevard 1989-2000 Marine Corps Air Station Marine Corps Air Station 1958-2000 Midway Park Hadnot Point 1943-1971 Holcomb Boulevard 1972-2000 Paradise Point Cape Cod Hadnot Point 1948-1971 Holcomb Boulevard 1972-2000 Paradise Point Capehart Hadnot Point 1962-1971 Holcomb Boulevard 1972-2000 Paradise Point Cracker Box Hadnot Point 1947-1971 Holcomb Boulevard 1972-2000 Paradise Point general officer housing Hadnot Point 1943-1971 Holcomb Boulevard 1972-2000 Paradise Point two-story housing Hadnot Point 1943-1971 Holcomb Boulevard 1972-2000 Rifle Range housing Rifle Range 1942-1993 Onslow County 1994-2000 Tarawa Terrace I and II Tarawa Terrace 1952-1986 Holcomb Boulevard 1987-2000 Watkins Village Holcomb Boulevard 1978-2000 Barracks subcamps (not individual barracks) Camp Geiger Camp Geiger 1941-1976 Marine Corps Air Station 1977-2000 Camp Johnson Camp Johnson 1941-1986 Holcomb Boulevard 1987-2000 Courthouse Bay Courthouse Bay 1941-2000 French Creek Hadnot Point 1943-2000 Hadnot Point Hadnot Point 1943-2000 Rifle Range Rifle Range 1941-1993 Onslow County 1994-2000 Source: Marine Corps, personal commun., March 13, 2008.
36 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects FIGURE 2-4 Conceptual model of a Camp Lejeune water system. (1) The drinking water at Camp Lejeune is ob- tained from groundwater pumped from a freshwater aquifer located approximately 180 ft below the ground. (2) Groundwater is pumped through wells located near the water-treatment plant. (3) In the water-treatment plant, the untreated water is mixed and treated through several processes: removal of minerals to soften the water, filtration through layers of sand and carbon to remove particles, chlorination to protect against microbial contamination, and fluoride addition to help prevent tooth decay. (4) After the water is treated, it is stored in ground and elevated stor- age reservoirs. (5) When needed, treated water is pumped from the reservoirs and tanks to facilities, such as offices, schools, and houses on the base. Source: GAO 2007. 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). COMMITTEEâS WATER-SUPPLY EVALUATION APPROACH The committee focused its attention on the Tarawa Terrace and Hadnot Point water-supply sys- tems. The systems were evaluated differently because much more work had been done to characterize the contamination of the Tarawa Terrace system than that of the Hadnot Point system. For Tarawa Terrace, the committee relied exclusively on reports by ATSDR (Faye 2007; Lawrence 2007; Faye and Green 2007; Faye and Valenzuela 2007; Maslia et al. 2007; Faye 2008; Jang and Aral 2008; Wang and Aral 2008). The reports included analyses of the water-quality data conducted in conjunction with ATSDRâs
Exposure to Contaminants in Water Supplies at Camp Lejeune 37 TABLE 2-2 Contaminants Found in Soil or Groundwater at Hazardous Waste Sites Near Water-Supply Wells Approximate Number Identified Contaminants Detected in Soils (S) or Water System Hazardous-Waste Sites Monitoring Wells (M, D) Tarawa Terrace 2 Chlorinated solvents (S ,M, D) BTEX (S, M) Hadnot Point 13 Pesticides (S, M) Polychlorinated biphenyls (S) Metals (S, M) Chlorinated solvents (S, M, D) Fuel compounds (M) Benzene (M) Toluene (M) Ethylbenzene (M) Xylenes (M) BTEX (M) Petroleum products (S, M) Volatile compounds (S) Semivolatile compounds (S) Holcomb Boulevard 5 Pesticides (S, M) Volatile and semivolatile compounds (S, M) Metals (M) Marine Corps Air 6 Volatile and semivolatile compounds at two Station locations (S, M) Pesticides at one location (S, M) Rifle Range 2 VOCs (M) Camp Geiger 13 Chromium (M) Lead (M) VOCs (M) Camp Johnson 2 None Abbreviations: BTEX = benzene, toluene, ethylene, and xylene; D = deeper wells in Castle Hayne aquifer, source of water-supply wells; M = shallow wells, surficial aquifer, or soil vadose zone. Sources: Baker Environmental, Inc 1999; CH2M Hill and Baker Environmental, Inc 2005. water-quality modeling. For Hadnot Point, the committee conducted its own review of information that was in the public record. The committee used multiple sources, including the 2007 GAO report, remedial investigation reports (Baker Environmental, Inc 1993, 1994, 1995), data summarized in the âCamp Le- jeune waterâ(CLW) documents (CD accompanying Maslia et al. 2007), and planning documents from ATSDR (Maslia 2008). The goal was to get an understanding of the contamination of water supplies serv- ing Hadnot Point residents, including which VOCs were of potential concern and the degree to which contaminant concentrations in the water supply varied. In consulting the CLW documents, the committee focused on contaminant measurements taken while the contaminated wells were operating, including measurements of the water-supply wells and from the water-treatment plant and distribution system. As noted earlier, water from the supply wells was mixed at the water-treatment plant before distribution. Be- cause all water samples from the distribution system were taken after water from multiple supply wells was mixed, they were categorized as âmixedâ water samples. Sampling of mixed water occurred before and after water was treated or âfinished.â Samples taken from mixed water give a better indication of the concentrations of contaminants delivered to the tap than samples taken from supply wells. However, wa- ter-quality data on the individual supply wells shed light on the wells that were contaminated and permit preliminary documentation of the extent of contamination. In determining its approach to evaluating the water-quality data on Hadnot Point, the committee wrestled with reporting data that have not been collected by a process that involved standard quality- assurance procedures. The process that was used for abstraction of the water-quality data (see Appendix
38 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects C) did not consider multiple aspects of the data, including the sampling strategy, methods for sample col- lection and analysis, chain of custody of samples, recording and interpretation of detection or quantitation limits, and duplication of sampling results in source documents. Thus, the data cited are only for illustra- tive purposes, and references to the primary documents are provided to facilitate additional work. TARAWA TERRACE WATER SUPPLY 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). Fur- ther analysis confirmed the presence of PCE in finished water at 76-104 Âµg/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 Develop- ment (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 con- tamination. For example, PCE ranged from nondetectable to 132 Âµg/L and from 3.8 to 1,580 Âµg/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 Janu- ary 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 Âµg/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 Âµg/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 âstrong gasoline type odorâ 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 under- ground 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 gaso- line 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.
Exposure to Contaminants in Water Supplies at Camp Lejeune 39 Investigation of groundwater contamination due to sources other than the ABC One-Hour Clean- ers 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 de- tected in supply wells, including TCE, 1,1-dichloroethylene (DCE), cis- and trans-1,2-DCE, benzene, toluene, and vinyl chloride. Many of these contaminantsâincluding TCE, DCE, and vinyl chlorideâmay 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 Âµg/L and total DCE at up to 1,200 Âµg/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 Âµg/L (estimated value on July 11, 1991) and 92 Âµg/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 sys- tem. 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 dateâat 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 contami- nated. The highest concentration (1,580 Î¼g/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 evi- dence of PCE contamination. Again, the highest concentration was found after a period of regular opera- tion in January 1985, and concentration was lower in later periods; notably, concentration was higher af- ter 24 h of continuous operation (on March 12, 1985) than at the beginning of that period of service. 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) sum- marized 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 â-â in the table). PCE ranged from undetected to 215 ïg/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,
40 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects TABLE 2-3 Observed Concentrations of PCE in Tarawa Terrace Water-Supply Wells Sample Date PCE, Âµg/L Detection Limit, Î¼g/L Supply well TT-23 Jan. 16, 1985 132 10 Feb. 12, 1985 37 10 Feb. 19, 1985 26.2 2 Feb. 19, 1985 ND 10 Mar. 11, 1985 14.9 10 Mar. 11, 1985 16.6 2 Mar. 12, 1985 40.6 10 Mar. 12, 1985 48.8 10 Apr. 9, 1985 ND 10 Sept. 25, 1985 4a 2 July 11, 1991 ND 10 Supply well TT-25 Feb. 5, 1985 ND 10 Apr. 9, 1985 ND 10 Sept. 25, 1985 0.43a 10 Oct. 29, 1985 ND 10 Nov. 4, 1985 ND 10 Nov. 12, 1985 ND 10 Dec. 3, 1985 ND 10 July 11, 1991 23 10 Supply well TT-26 July 16, 1985 1,580 10 Feb. 12, 1985 3.8 10 Feb. 19, 1985 64 10 Feb. 19, 1985 55.2 10 April 9, 1985 630 10 June 24, 1985 1,160 10 Sept. 25, 1985 1,100 10 July 11, 1991 350 10 Supply well TT-30 Feb. 6, 1985 ND 10 Supply well TT-31 Feb. 6, 1985 ND 10 Supply well TT-52 Feb. 6, 1985 ND 10 Supply well TT-54 Feb. 6, 1985 ND 10 July 11, 1991 ND 5 Supply well TT-67 Feb. 6, 1985 ND 10 Supply well RW1 July 12, 1991 ND 2 Supply well RW2 July 12, 1991 760 2 Supply well RW3 July 12, 1991 ND 2 a Estimated value. Abbreviation: ND = not detected. Source: Adapted from Maslia et al. 2007.
Exposure to Contaminants in Water Supplies at Camp Lejeune 41 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 Trans-1,2-DCE, Detection Sample Location or Event Date PCE, Âµg/L TCE, Âµg/L Âµg/L Limit, Âµg/L Tap water at Bldg. TT-2453 May 27, 1982 80 â â Unknown Tap water at Bldg. TT-2453 July 28, 1982 104 â â Unknown TTWTP Bldg. TT-38 July 28, 1982 76 â â Unknown TTWTP Bldg. TT-38 July 28, 1982 82 â â Unknown Tap water; address unknown Feb. 5, 1985 80 8.1 12 Unknown Well TT-26 shut down Feb. 8, 1985 a Well TT-23 initially shut down Feb. 8, 1985 TTWTP Tank STT-39 Feb. 11,1985 215 8 12 10 TTWTP Bldg. TT-38 Feb. 13, 1985 ND ND ND 10 TTWTP Bldg. TT-38 Feb. 19, 1985 ND ND ND 2 TTWTP Bldg. TT-38 Feb. 22, 1985 ND ND ND 10 a a a TTWTP Bldg. TT-38 Mar. 11, 1985 ND ND ND 2 b a a a TTWTP Bldg. TT-38 Mar. 12, 1985 6.6 ND ND 10 b a a a TTWTP Bldg. TT-38 Mar. 12, 1985 8.9 ND ND 2 TTWTP Bldg. TT-38 Mar. 12, 1985c 20a 1.1a 1.2a 2 c a a a TTWTP Bldg. TT-38 Mar. 12, 1985 21.3 ND ND 10 a a a TTWTP Bldg. TT-38 Apr. 22, 1985 1 4.1 ND 10 a a a TTWTP Bldg. TT-38 Apr. 23, 1985 ND 1.4 ND 10 TTWTP Bldg. TT-38 Apr. 29, 1985 3.7a NDa â 10 Well TT-23 ceases to operate May 1985 TTWTP Bldg. TT-38 May 15, 1985 ND ND ND 10 TTWTP Bldg. TT-38 July 1, 1985 ND ND ND 10 TTWTP Bldg. TT-38 July 8, 1985 ND ND ND 10 TTWTP Bldg. TT-38 July 23, 1985 ND ND ND 10 TTWTP Bldg. TT-38 July 31, 1985 ND ND ND 10 TTWTP Bldg. TT-38 Aug. 19, 1985 ND ND ND 10 TTWTP Bldg. TT-38 Sept.11, 1985 ND ND ND 10 TTWTP Bldg. TT-38 Sept. 17, 1985 ND ND ND 10 TTWTP Bldg. TT-38 Sept. 24, 1985 ND ND ND 10 TTWTP Bldg. TT-38 Oct. 29, 1985 ND ND ND 10 a Intermittent operation of well TT-23 after February 8, 1985, including at least March 11 and 12 and April 22, 23, and 29, 1985. b Samples collected downstream of TTWTP reservoir after well TT-23 operated for 24 h. c Samples collected upstream of TTWTP reservoir after well TT-23 operated for 24 h. Abbreviations: â = constituent not determined; ND = not detected; TTWTP = Tarawa Terrace water-treatment plant. Source: Adapted from Table E12 of Faye and Green 2007.
42 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects 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 ïg/L and trans-1,2-DCE at 12 ïg/L. Similar concentrations of TCE and trans-1,2-DCE (8 and 12 ïg/L, respectively) were reported in the water-storage tank sample (STT-39A, February 11, 1985). 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 ânot detectedâ to 2 Âµg/L and toluene from ânot detectedâ to 4 Âµg/L; all concentra- tions were below the stated laboratory detection limit of 10 ïg/L. (The accuracy of values below the de- tection 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 summa- rized. 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 ATSDRâs 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 condi- tions. 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
Exposure to Contaminants in Water Supplies at Camp Lejeune 43 TABLE 2-5 Benzene and Toluene Concentrations in Water Samples Collected at Tarawa Terrace Water- Treatment Planta Site Name Date Benzene, Âµg/L Toluene, Âµg/L TTWTP Bldg. TT-38 Feb. 13, 1985 ND ND Feb. 22, 1985 ND ND Mar. 11, 1985 1.6 â Apr. 22, 1985 ND ND Apr. 23, 1985 ND ND May 15, 1985 ND ND July 1, 1985 ND ND July 8, 1985 ND ND July 23, 1985 ND ND July 31, 1985 ND ND Aug. 19, 1985 ND ND Sept. 11, 1985 ND 4 Sept. 17, 1985 ND ND Sept. 24, 1985 ND ND Oct. 29, 1985 ND ND Dec. 2, 1985 2 â Dec. 18, 1985 1 â TTWTP tank SST-39A Feb. 11, 1985 ND ND a Detection limit for all analyses was 10 Âµg/L. Abbreviations: â = constituent not determined; ND = not detected; TTWTP = Tarawa Terrace water-treatment plant. Source: Faye and Green 2007. 2008). In addition, the team used a multiphase transport simulator, TechFlowMP, which has the capabil- ity 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 correspond- ing 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 in- jected continuously into the unconfined model layer-1 of the saturated zone at a constant rate of 1.2 kg/day (Faye 2008).
44 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects TABLE 2-6 Assumed Thickness and Layer of Castle Hayne Aquifer Units Geologic Unit Thickness, ft Layer No. Tarawa Terrace unit (surficial layer) 8-30 1 Tarawa Terrace confining unit (surficial layer) 8-20 1 Upper Castle Hayne aquifer-River Bend unit (surficial layer) 16-56 1 Local confining unit 7-17 2 Upper Castle Hayne aquifer-lower unit 8-30 3 Middle Castle Hayne aquifer confining unit 12-28 4 Middle Castle Hayne aquifer 32-90 5 Lower Castle Hayne aquifer confining unit 18-30 6 Lower Castle Hayne aquifer 41-64 7 Beaufort confining layer Bottom boundary N/A Source: Modified from Faye and Valenzuela 2007. ATSDR calibrated the MODFLOW and MT3DMS models for Tarawa Terrace by using a âhier- archical processâ that included the simulation of the following four successive scenarios: (1) predevelop- ment (before the 1950s) flow conditions without pumping, (2) transient flow conditions involving pump- ing, (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 in- dicated 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 EPAâs current maximum contami- nant level (MCL) for drinking water of 5 Âµg/L as early as January 1957 and that a corresponding break- through 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 con- centration 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 ATSDRâs model results, the estimated maximum concentration of PCE at the Tarawa Terrace water-treatment plant was 183 Âµg/L in March 1984. In the period November 1957- February 1987, the average concentration of PCE at the plant was 70 Âµg/L. The estimated PCE concentration range should, however, be interpreted with considerable cau- tion 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 monotoni- cally increasing trend, whereas the measured data are highly random. It is important to note that compari- son of monthly averaged model predictions with point measurements from various locations is problem- atic, 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 wellsâ water quality represent a single time and are relevant on a much shorter time scaleâhours instead of months. Similarly, spatial variations in concentration are averaged over a relatively large control vol- ume 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 2-7 Calibrated Model Parameter Concentrations Used to Simulate Groundwater Flow and Contaminant Fate and Transport in Tarawa Terrace and Vicinity Model Layerb a Model Parameter 1 2 3 4 5 6 7 Predevelopment groundwater-flow model (conditions before 1951) Horizontal hydraulic conductivity, KH (ft/day) 12.2-53.4 1.0 4.3-20.0 1.0 6.4-9.0 1.0 5.0 Ratio of vertical to horizontal hydraulic conductivity, KV/KHc 1:7.3 1:10 1:8.3 1:10 1:10 1:10 1:10 Infiltration (recharge), IR (in./year) 13.2 â â â â â â Transient groundwater-flow model(January 1951âDecember 1994) Specific yield, Sy 0.05 â â â â â â Storage coefficient, S â 4.0 Ã 10â4 4.0 Ã 10â4 4.0 Ã 10â4 4.0 Ã 10â4 4.0 Ã 10â4 4.0 Ã 10â4 Infiltration (recharge), IR (in./year) 6.6-19.3 â â â â â â d d Pumpage, Qk (ft3/day) â â 0 â 0 Fate and transport of PCE model (January 1951âDecember 1994) Distribution coefficient, Kd (ft3/g) 5.0 Ã 10â6 5.0 Ã 10â6 5.0 Ã 10â6 5.0 Ã 10â6 5.0 Ã 10â6 5.0 Ã 10â6 5.0 Ã 10â6 3 Bulk density, Ïb (g/ft ) 77,112 77,112 77,112 77,112 77,112 77,112 77,112 Effective porosity, nE 0.2 0.2 0.2 0.2 0.2 0.2 0.2 Reaction rate, r (dâ1) 5.0 Ã 10â4 5.0 Ã 10â4 5.0 Ã 10â4 5.0 Ã 10â4 5.0 Ã 10â4 5.0 Ã 10â4 5.0 Ã 10â4 Mass-loading ratee, qSCS (g/day) 1,200 â â â â â â Longitudinal dispersivity, Î±L (ft) 25 25 25 25 25 25 25 Traverse dispersivity, Î±T (ft) 2.5 2.5 2.5 2.5 2.5 2.5 2.5 Vertical dispersivity, Î±V (ft) 0.25 0.25 0.25 0.25 0.25 0.25 0.25 Molecular-diffusion coefficient, D* (ft2/day) 8.5 Ã 10â4 8.5 Ã 10â4 8.5 Ã 10â4 8.5 Ã 10â4 8.5 Ã 10â4 8.5 Ã 10â4 8.5 Ã 10â4 a Symbolic notation used to describe model parameters obtained from Chiang and Kinzelbach (2001). b Refer to Chapter B (Faye 2007) and Chapter C (Faye and Valenzuela 2007) reports for geohydrologic framework corresponding to appropriate model layers; aquifers are model layers 1, 3, 5, and 7; confining units are model layers 2, 4, and 6. c For model cells simulating water-supply wells, vertical hydraulic conductivity (Kv) equals 100 ft/day to approximate gravel pack around well. d Pumpage varies by month, year, and model layer; refer to Chapter K report (Maslia et al. in press) for specific pumpage data. e Introduction of contaminant mass began in January 1953 and ended in December 1984. Abbreviations: â = not applicable; dâ1 = 1/day. Source: Maslia et al. 2007. 45
46 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects FIGURE 2-5 Simulated (a) water level and direction of groundwater flow, and (b) distribution of tetrachloroethyl- ene (PCE), model layer 1, December 1984, Tarawa Terrace and vicinity, U.S. Marine Corps Base Camp Lejeune, North Carolina. Source: Maslia et al. 2007. 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 âcalibration target rangeâ 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 Âµg/L, and the upper limit was set at the analytic detection limit (Faye 2008).
Exposure to Contaminants in Water Supplies at Camp Lejeune 47 TABLE 2-8 Simulated and Observed PCE Concentrations at Water-Supply Wells and Calibration Target Range, Tarawa Terrace and Vicinity PCE Concentration, Âµg/L Site Date Observed Simulated Calibrated Target Range, Âµg/L RW1 July 12, 1991 ND 0.0 0.0-2.0 RW2 July 12, 1991 760 1,804 240-2,403 RW3 July 12, 1991 ND 0.0 0.0-2.0 TT-23 Jan. 16, 1985 132 254 41.7-417 Feb. 12, 1985 37.0 254 11.7-117 Feb. 19, 1985 26.2 253 8.3-82.8 Feb. 19, 1985 ND 253 0.0-10.0 Mar. 11, 1985 14.9 253 4.7-47.1 Mar. 11, 1985 16.0 253 5.2-52.5 Mar. 12, 1985 40.6 253 12.8-128 Mar. 12, 1985 48.0 253 15.4-154 Apr. 9, 1985 ND 265 0.0-2.0 Sept. 25, 1985 4.0 279 0.3-12.6 July 11, 1991 ND 193 0.0-5.0 TT-25 Feb. 5, 1985 ND 6.2 0.0-10.0 Apr. 9, 1985 ND 8.6 0.0-2.0 a Sept. 25, 1985 0.43 18.1 0.14-1.4 Oct. 29, 1985 ND 20.4 0.0-10.0 Nov. 4, 1985 ND 20.4 0.0-10.0 Nov. 13, 1985 ND 20.4 0.0-10.0 Dec. 3, 1985 ND 22.8 0.0-10.0 July 11, 1991 23.0 72.6 7.3-72.7 TT-26 Jan. 16, 1985 1,580 804 500-5,000 Feb. 12, 1985 3.8 804 1.2-12 Feb. 19, 1985 55.2 798 17.5-175 Feb. 19, 1985 64.0 798 20.2-202 Apr. 9, 1985 630 801 199-1,999 June 24, 1985 1,160 732 367-3,668 Sept. 25, 1985 1,100 788 348-3,478 July 11, 1991 340 670 111-1,107 TT-30 Feb. 6, 1985 ND 0.0 0.0-10.0 TT-31 Feb. 6, 1985 ND 0.15 0.0-10.0 TT-52 Feb. 6, 1985 ND 0.0 0.0-10.0 TT-54 Feb. 6, 1985 ND 5.8 0.0-10.0 July 11, 1991 ND 30.4 0.0-5.0 TT-67 Feb. 6, 1985 ND 3.9 0.0-10.0 a Estimated. Abbreviation: ND = not detected. Source: Faye 2008.
48 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects ATSDR stated concerns about uncertainties in the pumping-schedule data used in the PCE mod- eling 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 EPAâs MCL of 5.0 Î¼g/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. Be- cause 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 microorgan- isms 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 aqui- fer 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 characteris- tics. 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. Gaps in and Limitations of the Modeling The committee reviewed the Tarawa Terrace modeling reports and found that ATSDR applied the public-domain codes for MODFLOW and MT3DMS and two cutting-edge research codes PsOps and TechFlowMP to model the complex groundwater-contamination scenario at Tarawa Terrace. However, there are some important limitations in ATSDRâs modeling efforts because of the sparse set of water- quality measurements, the need to make unverifiable assumptions, and the complex nature of the PCE source contamination. The major gaps and limitations that the committee found with regard to the histori- cal reconstruction and modeling work are summarized below. Future modeling efforts for the Hadnot Point water system should be designed in light of these limitations. ï· The effects of the DNAPL in both unsaturated and saturated zones have not been included in the studies. As constructed, the DNAPL zone has no influence in any of the Tarawa Terrace groundwater models, because for each model ATSDR assumed that PCE was injected directly at a constant rate of 1.2 kg/day (that is, multiphase flow and dissolution reactions associated with DNAPL transport were ig- nored). PCE dissolution is a highly heterogeneous, rate-limited, mass-transfer process (Miller et al. 1990; Jackson 1998; Abriola 2005). Hence, the constant-mass injection approach used to model the complex PCE source zone may be prone to high uncertainty. Field data or other supporting evidence would be needed to justify the mass release rates. For example, Guilbeault et al. (2005) proposed some methods to characterize DNAPL source zones to estimate mass and contaminant release rates.
Exposure to Contaminants in Water Supplies at Camp Lejeune 49 ï· Constant values of dispersivity (longitudinal dispersivity of 25 ft and transverse 2.5 ft) were used in the transport model. There is insufficient information available on the nature and amount of het- erogeneity to use these fixed values with a sufficient level of confidence in predictive simulations. ï· The basis used for setting the values of the âcalibration target rangeâ was unclear. The repeated samples collected at some of the wells (multiple samples in 1 year) may provide some important informa- tion about the variability of observations caused by subsurface variations and possibly pumping varia- tions. Perhaps these data could be used to determine observation variability that the computer model was not constructed to reproduce. ï· The numerical codes TechFLowMP and PSOpS used in the modeling are research tools and are not widely accepted public-domain codes, such as MODFLOW and MT3DMS, so their validation is im- portant. If data are not available, the results should be used with caution and should include appropriate uncertainty estimates. ï· The PSOpS modeling study is based on the premise that an optimization model can be used to evaluate pumping stresses. Without site-specific pumping and water-quality data, the results will be non- unique and uncertain. ï· Review of water-quality monitoring data indicates substantial temporal variability even at a single well. For example, the seven measurements taken on well TT-26 from January to September 1985 indicates that the concentrations at this well varied from 3.8 to 1,580 Î¼g/L (see Table 2-8). The model predictions for the same timeframe ranged from 732 to 804 Î¼g/L. The difference indicates that the real system is highly transient and that the model did not account for temporal and spatial averaging effects. ï· The TechFlowMP model predicted very high vapor concentrations. For example, TechFlowMP predicted that the PCE vapor concentration in the top 10 ft of soil beneath the Tarawa Terrace elementary school should be 137 Î¼g/L. Studies of PCE vapor concentrations in buildings that house or are near a dry- cleaning facility have reported measured concentrations around 55 Î¼g/L (McDermott et al. 2005). The PCE vapor concentrations predicted by TechFlowMP should be treated with caution because they are theoretical estimates that have not been validated against field data from Camp Lejeune or compared with any measured vapor concentrations at other similar dry-cleaner sites. ï· The biodegradation model used within the TechFlowMP code is based on an untested prelimi- nary research model. Biodegradation of chlorinated solvent compounds will be influenced by several types of complex biogeochemical processes. The simple first-order modeling framework that also used a single decay coefficient for the entire modeling domain may not capture those biologic complexities. Therefore, the predicted concentrations of TCE, DCE, and vinyl chloride should be considered âcrudeâ estimates, at best, unless validated with field measurements. In addition, biodegradation-model predic- tions are not supported by field data on biogeochemical indicators, which are commonly used to assess whether the assumed biodegradation pathways are active at a field site (EPA 1998a). ï· The TechFlowMP simulations assumed that the biodegradation byproduct of TCE is trans-1,2- DCE. However, the scientific literature indicates that cis-1,2-DCE is the predominant product of TCE reduction under in situ groundwater conditions (Bradley 2003). ï· Reporting absolute predicted concentrations of PCE and its biodegradation byproducts in fin- ished water delivered by the Tarawa Terrace water-supply system with a precision up to five significant figures without any error bounds (for example, Jang and Aral  report concentrations of PCE at 102.10 Âµg/L, TCE at 4.33 Âµg/L, DCE at 13.75 Âµg/L, and vinyl chloride at 7.50 Âµg/L) provides an unwar- ranted sense of certainty. Such reporting can contribute to misperceptions by the public and the epidemi- ology-research community that water-modeling efforts can produce a specific value for contaminant con- centration. Posting such precise point estimates for PCE, TCE, DCE, and vinyl chloride concentrations on public Web pages (www.atsdr.cdc.gov/sites/lejeune) and encouraging former Camp Lejeune marines and their families to find the estimated exposure concentrations of these contaminants leads to a misleading perception that reactive transport models can make accurate predictions. ï· In the absence of data, historical reconstruction efforts that use groundwater models can only provide a general conceptual framework for what happened at the site and why. At best, such models may
50 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects be used only to estimate a range of possible concentrations. Without historical geochemical data, the un- certainly associated with many of the input parameters (such as the biodegradation parameters) could be very high. In addition, current understanding of subsurface reactive transport processes is inadequate, so transport models cannot be expected to provide definitive concentration estimates especially for biodeg- radation byproducts. ï· The inherent and, in this case, profound limitations of historical modeling due to uncertainties in various model parameters and pumping stresses should be communicated along with modeling predic- tions. ATSDR has completed a detailed groundwater-modeling study and have used the best possible techniques and tools. Several of the gaps and limitations mentioned above are due to the difficulty of re- constructing accurate groundwater-contamination scenarios. Without historical data, the natural processes that occurred several decades ago simply cannot be reconstructed. The committee understands this limita- tion and acknowledges that ATSDR has done its best under these difficult circumstances. HADNOT POINT WATER SUPPLY Approximately 100 wells have supplied water to the Hadnot Point system since it began opera- tions in 1943, although not all were operational at the same time. ATSDR is currently determining the history of the individual well operations and capacities. Like the Tarawa Terrace system, water from the supply wells was pumped to the water-treatment plant and mixed and processed before distribution on the base. In July 1972, the Holcomb Boulevard water system took over supplying water to some areas origi- nally served by the Hadnot Point system. The two systems were connected, such that on several occasions the Hadnot Point system temporarily served or supplemented the Holcomb Boulevard system. Specifi- cally, water from the Hadnot Point system was used periodically during summer months and for 2 weeks in 1985 when the Holcomb Boulevard system was shutdown because of an emergency. A comprehensive water-modeling analysis has not yet been conducted for Hadnot Point, so the committee sought to identify documents that provided some quantitative information on the contamina- tion of the Hadnot Point water-distribution system. Relevant information was found in a 2007 GAO re- port, documents cited by ATSDR in its evaluation of Tarawa Terrace, remedial-investigation reports, and documents provided to the committee by the public. The selection of documents reviewed was not com- prehensive but was informed by discussion with the Marine Corps, ATSDR, and the public. Primary sources of information for the committeeâs review included contaminant measurements taken while the contaminated wells were operating and data collected from monitoring wells, which were installed to conduct testing and monitoring for remediation purposes after the supply wells were closed. The remedial investigations and the 2007 GAO report identify TCE and PCE as the primary con- taminants of concern at Hadnot Point. After reviewing additional preliminary information, the committee decided also to investigate eight other contaminants: trans-1,2-DCE, cis-1,2-DCE, 1,1-DCE, 1,1,1- trichloroethane (TCA), vinyl chloride, methylene chloride, benzene, and toluene. The most useful infor- mation on those contaminants was a set of CLW documents, available on the CD accompanying Maslia et al. (2007). The set has 1,110 files made up of over 8,700 pages of material and includes laboratory re- ports, memorandums, field notes, and other written documents. The CLW documents were not organized or cataloged, so it was not possible to search readily for documents that contained water-quality meas- urements. The committee asked the Marine Corps for guidance on which CLW documents contained wa- ter-sample values from any location on the base during 1980-1986 (see Appendix C, Table C-2, for the list provided by the Marine Corps). The committee abstracted data from the subset of CLW documents that contained water-quality results from Hadnot Point potable-well and mixed water samples before March 1985.
Exposure to Contaminants in Water Supplies at Camp Lejeune 51 It was beyond the scope of the committeeâs task to conduct an exposure assessment or even a full-scale data abstraction for Hadnot Point. Such an undertaking would have required a systematic re- view of standard laboratory practices for the sampling and analytic methods for collecting, analyzing, and reporting on water samples at the contributing laboratories during the 1980s; review of the source docu- ments for quality assurance; information on detection and quantitation limits; identification and elimina- tion of duplicate measurements recorded in multiple documents; and information on sampling location and conditions. The committeeâs review of the available documents presented below constitutes an illus- tration of the information that is available and should help to inform future efforts for evaluating contami- nation of water supplies at Hadnot Point. 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 contami- nation 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 under- standing 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 (in- stallation restoration [IR] site number), and they were grouped into operable units (OUs) to facilitate in- vestigation and data management. Most of the sites were active in the 1940s to 1970s, before implementa- tion 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. 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, refu- eling 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 investiga- tion 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 con- tamination of the upper aquifer with TCE and vinyl chloride was documented during the remedial inves- tigation process (Baker Environmental, Inc 1995).
52 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects TABLE 2-9 Installation Restoration Sites in the Hadnot Point Water-Supply Area OU and IR Site Site Description Contaminants Identified During Remedial Investigation OU 1, site 21 Transformer storage lot Soil contaminated with pesticides, polychlorinated biphenyls OU 1, site 24 Industrial area fly ash dump Soil contaminated with metals, pesticides; shallow groundwater contaminated with pesticides OU 1, site 78 Hadnot Point industrial area Groundwater (shallow and deep) contaminated with chlorinated solvents, fuel compounds (benzene, toluene, ethylbenzene, xylenes) Soils contaminated with pesticides, metals OU 2, site 6 Storage lots 201, 203; connected Soil contaminated with pesticides, petroleum products, metals to site 82 Groundwater (shallow and deep) contaminated with chlorinated solvents OU 2, site 9 Firefighting training pit at Piney Low concentrations of chlorinated solvents in shallow groundwater Green Road OU 2, site 82 Piney Green Road VOC area; Groundwater (shallow and deep) contaminated with chlorinated connected to site 6 solvents OU 7, site 1 French Creek liquids disposal Shallow groundwater contaminated with petroleum products, area chlorinated solvents OU 7, site 28 Hadnot Point burn dump Surface soils contaminated with volatile, semivolatile compounds Shallow groundwater contaminated with metals OU 7, site 30 Sneads Ferry Road fuel tank Soil contaminated with VOCs sludge area OU 15, site 88 Building 25, base dry cleaners Soil, shallow groundwater contaminated with solvents OU 18, site 94 Former PCX service station Groundwater contaminated with petroleum hydrocarbons, chlorinated solvents Pre-RI 10 Original base dump No significant contamination of soil or groundwater identified Pre-RI 12 Explosive ordnance disposal No significant contamination of soil or groundwater identified area Abbreviation: RI = remedial investigation. Site 28 is a former 23-acre burn dump, operated from 1946 to 1971, south of the Hadnot Point in- dustrial area (Baker Environmental, Inc 1995). Solid waste from industrial operationsâincluding con- struction debris, industrial waste, trash, and oil-based paintâwas 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 sam- ples from the same period revealed the presence of lead, which was detected sporadically in the deeper water (Baker Environmental, Inc 1995). Site 30, the Sneadâs 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).
Exposure to Contaminants in Water Supplies at Camp Lejeune 53 FIGURE 2-6 Designated hazardous-waste remedial investigation sites at Hadnot Point.
54 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects 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 con- taminants 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 sam- ples (Baker Environmental, Inc 1999, 2001). Pre-RI site 12 is a 10-acre former explosive-ordnance disposal area. No substantial residual con- tamination 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 re- moved 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 Âµg/L, PCE at 386 Âµg/L, and trans-1,2-DCE at 3,400 Âµg/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 closedâFebruary 4, 1985; the sample contained TCE at 18,900 Âµg/L (Ensminger 2007; CLW 3269). Given that the water-quality data summarized in published reports were extremely sparse (for in- stance, 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 docu- ments that contained water-quality measurement data (see Appendix C, Tables C-3 and C-4, for data ab- stracted from the documents) for any samples connected to the Hadnot Point water supply that were col- lected 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 Octo- ber 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 recordedâsamples listed as ânon-detect,â âde- tect,â or âââ were excluded in these computationsâand percentiles were reported only if at least five samples contained a given compound. Sample concentrations that are listed as not quantified were re- corded in the source documents as D (detect) or ND (non-detect) or were not reported (shown as âââ in the data listing). Samples in which concentrations could not be quantified are summarized in Table 2-12. Of the nine analytes, the most prevalent compounds in mixed water samples collected from vari- ous locations in the Hadnot Point water-treatment plant and distribution system with measurable concen- trations were TCE (31 quantified samples had a mean of 399 Î¼g/L and a range of 1-1,400 Î¼g/L) and trans-1,2-DCE (21 quantified samples had a mean of 169 Î¼g/L and a range of 2-407 Î¼g/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 Î¼g/L and range of 5-18,900 Î¼g/L) and trans-1,2-DCE (14 quantified samples had a mean of 1,519 Î¼g/L and a range of 2-8,070 Î¼g/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 Î¼g/L), benzene (maximum, 720 Î¼g/L), methylene chloride (maximum, 270 Î¼g/L), and vinyl chloride (maximum, 655 Î¼g/L) in the potable well samples.
Exposure to Contaminants in Water Supplies at Camp Lejeune 55 TABLE 2-10 Contaminant Concentrations in Supply Wells of Hadnot Point Water System Concentration, Âµg/La Date Removed Trans- Methylene Vinyl Well from Service TCE PCE Benzene 1,2-DCE 1,1-DCE Chloride Toluene Chloride 602 Nov. 30, 1984 1,600 24 120 630 2.4 â 5.4 18 601 Dec. 6, 1984 210 5 ND 88 ND ND ND ND 608 Dec. 6, 1984 110 ND 3.7 5.4 ND ND ND ND 634 Dec. 14, 1984 ND ND ND 2.3 â 130 â ND 637 Dec. 14, 1984 ND ND ND ND â 270 â â 651 Feb. 4, 1985 3,200 386 â 3,400 187 â â 655 652 Feb. 8, 1985 9 ND â ND ND â â ND 653 Feb. 8, 1985 5.5 ND â ND ND â â ND a Detection limit for each contaminants was 10 Âµg/L. Abbreviation: ND = not detected. Source: GAO 2007. 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 con- tamination, with detectable concentrations of TCE in all the reported samples. Well 651 also had rela- tively 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 Î¼g/L (1,600 and 1,300 Î¼g/L, respectively). 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 âXâ 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
TABLE 2-11 Hadnot Point Water-Supply Quality Measurements (October 1980-February 1985) 56 No. Samplesa Summarized Data on Samples with Quantified Values, Âµg/Lb Water % Samples Individual 25th 75th Source Contaminant ND/NQc Quantified Quantified Samples, Âµg/L Mean Min Percentile Median Percentile Max Supply TCE 39 17 30 2,596 5 13 210 1,300 18,900 wells PCE 48 8 14 153 2 5 17 392 400 Benzene 50 6 11 180 2 4 62 230 720 1,1,1-TCA 56 0 0 1,1-DCE 54 2 4 2; 187 95 trans-1,2-DCE 42 14 25 1,519 2 9 165 700 8,070 MC 50 6 11 78 7 10 26 130 270 Toluene 54 2 4 5; 12 9 VC 51 5 9 205 7 18 168 179 655 d Mixed TCE 21 31 60 399 1 19 200 849 1,400 water d PCE 48 4 8 1; 4; 8; 15 7 Benzene 52 0 0 1,1,1-TCA 52 0 0 1,1-DCE 52 0 0 trans-1,2-DCE 31 21 40 169 2 9 150 321 407 MC 51 1 2 54 54 Toluene 52 0 0 VC 51 1 2 3 3 a Samples in this table listed separately in Appendix C, Tables C-3 and C-4. Samples treated as distinct if reported on separate laboratory reports; in some cases, multiple samples reported from same location on same date, but it is not known whether these were duplicate samples. b Sample concentrations presented as summary statistics if more than four samples were quantified. Quantified samples listed individually if four or fewer sam- ples quantified. c ND/NQ samples do not have reported concentrations for various possible reasons, including that they were not measured, were not detected, or were recorded merely as detected. See Table 2-12 for additional information about these samples. d Concentrations measured in seven of 11 samples collected before 1984 were assumed to be detected on basis of notes on laboratory reports and inferences from later laboratory reports. Abbreviations: DCE = dichloroethylene; MC = methylene chloride; ND = not detected; NQ = not quantified; TCA = trichloroethane; VC = vinyl chloride.
TABLE 2-12 Summary of Data on Water Samplesa from Hadnot Point Water System Recorded As Not Detected or Not Quantified in Table 2-11 Water Source ND/NQ Category Contaminant ND/NQ Reported as Detected <2.0 or <1 Âµg/L ND No data Supply wells TCE 39 39 PCE 48 48 Benzene 50 50 1,1,1-TCA 56 56 1,1-DCE 54 54 trans1-2,DCE 42 42 MC 50 50 Toluene 54 54 VC 51 51 b Mixed water TCE 21 7 6 7 1 PCE 48b 7 2 12 27 Benzene 52 14 38 1,1,1-TCA 52 14 38 1,1-DCE 52 14 38 trans1-2,DCE 31 7 10 14 MC 51 13 38 Toluene 52 14 38 VC 51 13 38 a Data listed separately in Appendix C (Tables C-3 and C-4). Samples treated as distinct if reported on separate laboratory reports; in some cases, multiple sam- ples reported from same location on same date, but it is not known whether these were duplicate samples. b Concentrations measured in seven of 11 samples taken before 1984 assumed to be detected on basis of notes on laboratory reports and inferences from later laboratory reports. Abbreviations: DCE = dichloroethylene; MC = methylene chloride; ND = not detected; NQ = not quantified; TCA = trichloroethane; VC = vinyl chloride. 57
58 TABLE 2-13 Characteristics of the Hadnot Point Supply Wells With At Least One Contaminated Sample Taken Between October 1980 and February 1985 Wella Well Contaminants, Âµg/Ld Depth, Screened Intervals, No. Water Trans-1,2- Year Well Date Well IR Siteb ftc Number (Range, ft) c Samplesd TCE PCE DCE 1,1-DCE MC VC Benzene Toluene Started Shut Downe 601 9, 78 195 4 (45-195) 3 26 ND 9 ND 1941 Dec. 6, 1984 210 4 88 ND 230 5 99 10 602 9, 78 160 5 (70-160) 4 38 ND 74 ND ND ND ND 1941 Nov. 30, 1984 340 ND 230 ND ND 120 ND 540 2 380 ND ND 230 5 1,600 24 630 2 18 720 12 603 78 195 5 (70-195) 3 ND ND 1941 ND ND 5 7 608 78 161.5 4 (61.5-161.5) 3 9 ND ND 2 1942 Dec. 6, 1984 13 2 ND 4 110 5 14 4 634 9, 78 225 10 (63-225) 3 ND ND ND ND ND 1959 Dec. 14, 1984 ND ND 2 ND ND 1,300 10 700 130 7 637 6, 9, 78, 172 5 (90-172) 3 ND 1969 Dec. 14, 1984 82 ND 270 642 9, 78 210 5 (112-196) 3 ND 1971 ND 38 651 9, 82 199 3 (125-194) 3f 3,200 386 3,400 ND 168 1971 Feb. 4, 1985 17,600 397 7,580 ND 179 18,900 400 8,070 187 655 652 1 9 1972 Feb. 8, 1985 653 6, 82 270 1 6 1978 Feb. 8, 1985 a Wells installed before March 1, 1985. Many other wells were operating before March 1, 1985, but are not included in list because contaminants not detected. b See Figure 2-6. c Data abstracted from Baker Environmental, Inc (1993a,b). d Samples listed in Appendix C, Table C-4. All readings are shown for individual compounds with at least one detection. e Well-closing dates reported in GAO (2007). f Includes two samples collected on same date and listed as âduplicatesâ on secondary source document. Abbreviations: DCE = dichloroethylene; IR = installation restoration; MC, methylene chloride; ND = not detected; PCE = perchloroethylene; TCE = trichloroethylene; VC = vinyl chloride.
Exposure to Contaminants in Water Supplies at Camp Lejeune 59 TABLE 2-14 Concentrations of Contaminants in Mixed Water Samples Collected from Hadnot Point Water-Distribution System During Period of Documented Well Cyclinga Average Concentration, Âµg/Lb No. Water trans-1,2- No. Wells Date Samplesc TCE PCE DCE MC Pumping Wellsd Dec. 4, 1984 2 123 2 49 0 21 603, 608, 634, 637, 642, 652e Dec. 10, 1984 1 2 0 2 0 10 637, 652 Dec. 13, 1984 1 0 0 0 54 18 652, 653 Dec. 14, 1984 1 0 0 0 0 18 652 Dec. 15, 1984 1 0 0 0 0 15 642, 652 Dec. 16, 1984 1 0 0 0 0 13 642, 652 Dec. 17, 1984 1 0 0 0 0 13 603, 642, 652 Dec. 18, 1984 1 0 0 0 0 13 603 Dec. 19, 1984 2 1 0 0 0 13 603 f f f Jan. 29, 1985 3 463 18 603, 642, 651, 653 f f Jan. 31, 1985 14 618 225 19 603, 642, 651, 652, 653 a Dates estimated to be November 28, 1984, through February 4, 1985. b All nondetected values treated as having concentrations of 0. c The location from which the samples were taken are provided in Appendix C, Table C-3. d Wells with at least one detected analyte that were pumping on same day or up to 2 days before date specified. e Well 651 pumped 3 days before these samples were taken. f Contaminant not measured or reported for mixed water samples collected on this date. Abbreviations: DCE = dichloroethylene; MC = methylene chloride; PCE = perchloroethylene; TCE = trichloro- ethylene. period to ascertain the potential effect of well cycling on measured contaminant concentrations. To illus- trate the effect of well cycling on mixed-water contamination, the committee made the highly conserva- tive assumption that all ânon-detectâ samples had zero concentrations of the listed contaminants. The ta- ble 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. 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 chlorideâall of which were reportedly detected in the Hadnot Point supply-well samplesâeither 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 Âµg/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 Î¼g/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
60 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects 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 in- vestigation 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 re- ports 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 âtarget compound listâ (TCL) and the âtarget analyte listâ (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 Âµg/L. Similarly, detec- tions of trans-1,2-DCE ranged from 1 to 26,000 Âµg/L, of phenol from 2 to 22,000 Âµg/L, and of benzene from 6.7 to 35 Âµg/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 Âµg/L) and fuel con- stituents benzene (not detected to 9,200 Âµg/L), toluene (not detected to 18,000 Âµg/L), ethylbenzene (not detected to 3,000 Âµg/L), xylenes (not detected to 16,000 Âµg/L), and naphthalene (not detected to 260 Âµg/L) (Baker Environmental, Inc 1993, 1994). TAL metals that were commonly detected in shallow wa- ter, with some samples at exceedingly high concentrations relative to EPAâs current MCLs, were arsenic (405 Âµg/L), barium (1,200 Âµg/L), chromium (858 Âµg/L), lead (126 Âµg/L), and manganese (714 Âµg/L) (Baker Environmental, Inc 1993, 1994). Only five wells of intermediate depth (about 50-75 ft) were sam- pled as part of the remedial investigation, and detected chemicals were generally measured at concentra- tions below risk-based criteria. The results of groundwater sampling and analysis with monitoring wells provide additional in- formation 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 under- standing 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
Exposure to Contaminants in Water Supplies at Camp Lejeune 61 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 Âµg/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 recon- struct 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): ï· Discovery of new or updated site information after the second quarter of FY 2008 that substan- tially alters baseline information may add time to the current timeline estimate. ï· 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 com- pleted, a more refined estimate of required computational time will be made. Earlier in this chapter, the committee identified several limitations in the Tarawa Terrace histori- cal reconstruction and groundwater modeling. Because the contamination at Hadnot Point is more com- plex, the limitations and difficulties related to such modeling will be greater. WATER USE PATTERNS AND BEHAVIORS Place of residence is a key determinant of exposure to contaminants in water at Camp Lejeune, but individual behaviorâincluding water consumption, showering or bathing patterns, and other water- related behaviors (such as dishwashing)âalso would influence the degree of exposure. The committee is not aware of any historical information that documents individual water-use patterns and behaviors of residents of base housing. Some information on typical water use and other factors that affect individual exposure is available (EPA 1997, 2008). Some specific information on the Camp Lejeune population is being sought as part of ATSDRâs case-control study focused on birth defects and childhood cancer out- comes (see Chapter 8). However, as in all retrospective epidemiologic studies of water-supply contamina- tion, the validity of such information is open to question given that it requires retrospective recall of wa- ter-consumption habits and water-related behaviors that occurred decades earlier, increasing the like-
62 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects lihood that error due to recall bias could be substantial. The contaminated water systems also supplied nonresidential areas of the base, including schools, workplaces, recreational areas, and a hospital. Water- use patterns and behaviors in those setting are expected to differ substantially from practices in resi- dences. In addition, people could have been exposed to contaminated water at multiple locations, for in- stance, in both residential and nonresidential settings. EXPOSURE PATHWAYS Although most attention has focused on the ingestion of contaminated water, additional exposure pathways were possible, including the inhalation of chemicals that have volatilized from standing water in toilets or from faucet or shower water and dermal exposure from showering and washing. Although there are no contemporaneous data on the Camp Lejeune population, exposure via inhalation and dermal ab- sorption of VOCs from water used for household purposes has been shown experimentally to account for as much exposure as that from drinking the water (see Chapter 3). The intrusion of vapor from shallow contaminated groundwater into homes and offices is yet another possible inhalation-exposure pathway. ATSDRâs simulation efforts indicate a potential for vapors from plumes at Tarawa Terrace to have en- tered buildings, including an elementary school and family housing (Maslia et al. 2007). EPA recently examined the possibility of vapor intrusion at the Tarawa Terrace Elementary School and several housing units and did not find any current problems (EPA 2007a,b). Any estimates of past exposure to contami- nated groundwater should consider all exposure pathways. AFFECTED STUDY POPULATION Residential history in housing areas served by the contaminated water supplies during the period of contamination is an important determinant of exposure. There are two major categories of housing at Camp Lejeune: family housing for personnel on assignment to Camp Lejeune and barracks for enlisted personnel rotating through the base for training. The committee was provided with an estimated number of residential houses on Camp Lejeune by water-supply system in any given year from 1941 to 2000 by the Marine Corps (Appendix C, Table C-6). The first year with substantial residential water service was 1943, in which an estimated 919 units were served by the Hadnot Point water system, the first to serve a residential development on the base other than a barracks. Large increases in the total number of family- housing units on the base occurred in 1952, with the construction of Tarawa Terrace housing (3,065 units); in 1958, with the construction of Marine Corp Air Station housing (3,500); in 1961, with the con- struction of Berkeley Manor and Paradise Point Capehart housing (4,177); and in 1978, with the construc- tion of Watkins Village housing (4,550). Substantial shifts in the water-supply source for residential hous- ing occurred in 1972 when about 1,886 housing units were transferred from the Hadnot Point water system to the Holcomb Boulevard system and in 1987 when about 1,955 housing units were transferred from the Tarawa Terrace system to the Holcomb Boulevard system. Translating the number of housing units into the size of the population that may have been exposed would require knowledge of the number of residents per household or at least the number of residents by housing area in each year. To translate that into potential years of residential exposure for a given person or household, the duration of residence on the base would need to be ascertained. To assess potential exposure of that person or household to spe- cific contaminants in the water supply, more accurate information on the location and period of residence would need to be ascertained. Information on the population size or typical duration of residence of per- sonnel assigned to barracks was not available. Potential exposures in nonresidential settings should also be considered. Such exposures may oc- cur in schools and job locations on the base. Table 2-15 presents potential sites of nonresidential exposure to contaminants from the Tarawa Terrace and Hadnot Point water systems in 1943-1985. No information was available on the number of persons in occupational, school, or day-care settings with potential expo- sure to contaminated water.
Exposure to Contaminants in Water Supplies at Camp Lejeune 63 EXPOSURE ASSESSMENT IN STUDIES OF HEALTH EFFECTS OF WATER-SUPPLY CONTAMINATION AT CAMP LEJEUNE ATSDR has completed two epidemiologic studies of water-supply contamination at Camp Le- jeune (ATSDR 1998; Sonnenfeld et al. 2001). They focused on prenatal outcomes, including mean birth weight, small for gestational age, and preterm birth. The studies were limited to singleton live-born in- fants (with estimated gestational ages of 20 weeks or more) whose mothers resided in base housing for at least 1 week before giving birth in January 1, 1968-December 31, 1985. The earlier study (ATSDR 1998) also included stillborn infants. The results of those studies are presented in Chapter 8, and this section briefly summarizes the exposure assessments that were used in each. The 1998 ATSDR study evaluated residents of Tarawa Terrace and Hadnot Point, whereas the 2001 Sonnenfeld et al. study evaluated only residents of Tarawa Terrace. In both studies, exposure was defined by place of residence at delivery and ascertained by linking birth records to the baseâs family- housing records. In the ATSDR study, residents of trailer parks were excluded because of the incompleteness of housing information and the inability to identify their water source. Infants whose mothers resided at Ta- rawa Terrace for at least 1 week before giving birth were classified as exposed. Also included in the ex- posed group were infants whose mothers received water from the Hadnot Point water system in the Hos- pital Point housing areas or resided in the service area of the Holcomb Boulevard water system and were pregnant for at least 1 week in a 12-day period in January 27-February 7, 1985. During that period, Had- not Point water served or was present in the Holcomb Boulevard system for operational reasons. Infants whose mothers were residents in other base family housing (the Marine Corps Air Station, Rifle Range, and Courthouse Bay housing areas) were classified as unexposed, as were infants whose mothers lived in areas served by the Holcomb Boulevard water system (defined as Berkeley Manor, Midway Park, Para- dise Point, and Watkins Village housing areas) during the study period other than the 2-week period in winter 1985 when the Holcomb Boulevard system received contaminated water from the Hadnot Point TABLE 2-15 Potential Sites of Nonresidential Exposure to Contaminants in the Tarawa Terrace and Hadnot Point Water Systems, 1943-1985 Exposure Scenario Years Contaminated Employment at Hadnot industrial area or other workplace Unknown-1985 Employment location served by Tarawa Terrace water system 1957-1985 Tarawa Terrace Elementary School 1957-1985 Tarawa Terrace day care 1957-1985 Hadnot Point-Holcomb Boulevard area schools Until 1972; intermittent linkages with the Hadnot ï· Russell School, 1943-1987 Point system; and during a 2-week period in 1985 ï· Old high school/middle school, 1963-1987 ï· Berkeley Manor Elementary School, 1963-present ï· Stone Street Elementary School, 1959-present ï· Midway Park Elementary School, 1952-present Hadnot Point-Holcomb Boulevard area day-care services Until 1972; intermittent linkages with the Hadnot ï· New hospital, 1983-1987 Point system after 1972; and during a 2-week period ï· Building 712, 1966-1982 in 1985 ï· Building LCH4025, 1960-1987 ï· Building 799, 1953-1987 ï· Building 2600, unknown-1987 ï· Building 899, 1985-1987 ï· Building 1200, 1942-1987 Source: Marine Corps, personal commun., December 4, 2007.
64 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects system. ATSDR also computed the number of weeks that a mother lived in the residence specified on the birth certificate on the basis of information about occupancy dates from the housing records, which were then categorized and used in analyses to explore the effects of duration of exposure on the adverse preg- nancy outcomes that were under investigation. However, ATSDR discovered after the study was com- pleted that the Holcomb Boulevard water-treatment plant had been in operation since 1968 (rather than 1972), so pregnant mothers receiving water from that system in 1968-1972 were incorrectly classified as âunexposed.â A reanalysis to correct that error is planned; exposure estimates from the water-modeling study (http://www.atsdr.cdc.gov/HS/lejeune/erratum.htm) will be used. In the Sonnenfeld et al. study, infants born to mothers living at Tarawa Terrace for at least 1 week before delivery were classified as exposed. With the exception of people who were excluded because they lived in base trailer parks or in areas served by distribution systems outside Tarawa Terrace that were also contaminated with TCE, all other infants whose mothers resided in base family housing were classified as unexposed. Misclassification of women as unexposed if they resided in areas served by the Holcomb Boulevard water system and were pregnant in 1968-1972 also affected this study. For each birth, length of maternal residence at Tarawa Terrace before delivery was computed by using dates of occupancy from the housing records and then categorized and used as another surrogate of exposure to explore effects on prenatal outcomes. Given the nature of the contamination at Camp Lejeune, the committee found the application of broad classifications of exposure based on place and duration of residence to be an appropriate approach for assessing exposure in the studies described above. Historical reconstruction and groundwater model- ing at Tarawa Terrace have provided additional characterization of potential exposure to PCE and an es- timated timeframe for the contamination, but it is questionable whether the additional information im- proves the exposure assessment for epidemiologic studies. Retrospective data on time-activity patterns of water use and water-related behaviors could improve exposure assessment but will be of questionable accuracy because the assessment is for periods that extend 20 years or more into the past. CONCLUSIONS The Tarawa Terrace and Hadnot Point water supply systems were contaminated with VOCsâ particularly TCE, PCE, and DCEâfor decades ending in the middle 1980s. Most of the organic contami- nants originated from DNAPLs, which have the potential to contaminate large volumes of groundwater over long periods. The hydrogeologic data indicate a high potential for contaminants from surface sources to migrate to water-supply wells in some areas of the base. The absence of a continuous impermeable bar- rier between the surface (source area) and the Castle Hayne aquifer (primary aquifer) supports the field observations that show contaminants in deep monitoring wells at the same depth as the water-supply wells. The exact extent of the contamination at Camp Lejeune cannot be documented with certainty, but it is known that a few highly contaminated wells supplied water to the Tarawa Terrace and Hadnot Point systems and that the contaminated wells were in operation for multiple years. The contaminant concentra- tions in the water-supply system varied because well pumping was cycled (the contaminated wells were not operated continuously, so there were fluctuations in contaminant concentrations). The qualitative evi- dence suggests that the magnitude of groundwater contamination was much higher in the Hadnot Point system than in the Tarawa Terrace system. It is also known that the Hadnot Point system supplied water to the Holcomb Boulevard water-supply area before 1972 and periodically after 1972. Widespread water- supply contamination in other water systems on the base was not evident from available documentation, but the committeeâs review was too limited to be conclusive in this regard. The fundamental problem in estimating exposure to contaminants in the water-supply systems of Tarawa Terrace and Hadnot Point quantitatively is the lack of information on water quality and treatment- system operation during the period of contamination. There are no water-quality data for the period before the 1980s, and this leaves a 40-year period for which the extent of water-supply contamination is un-
Exposure to Contaminants in Water Supplies at Camp Lejeune 65 documented. In addition, little documentation is available on water-treatment techniques, which would shed light on the efficiency of contaminant removal during treatment. Also lacking is information on well cycling, which is important for documenting when contaminated wells were pumping raw water into the system. For those reasons, any estimates of water-supply contamination must rely on unverifiable as- sumptions. ATSDR applied best practices and cutting-edge modeling approaches to predict the complex groundwater-contamination scenario at Tarawa Terrace. The ultimate outcome of the modeling was aver- aged monthly predictions of the concentrations of contaminants in the water supply to which people could have been exposed. Although ATSDR recognized and tried to account for the limitations and uncertain- ties associated with developing its models, it is extremely difficult to obtain quantitative estimates of his- torical levels of exposure to PCE and its degradation products reliably on a monthly basis. Reporting such model predictions without clear error bounds gives the impression that the exposure of former residents and workers at Tarawa Terrace during specific periods within a given year can be accurately defined. It is the committeeâs judgment that ATSDRâs model is suitable only for estimating long-term exposure quali- tatively. From that perspective, a single exposure category of âexposedâ appears to be applicable for per- sons residing or working at Tarawa Terrace at any time during 1957-1985. Efforts at historical reconstruction of exposures at Hadnot Point will be even more problematic. The contamination scenario at Hadnot Point is so complex that the committee judges that only crude es- timates of contaminant concentrations in the water supply can be obtained. RECOMMENDATIONS The history of water-supply contamination at Hadnot Point is much more complex than the his- tory of that at Tarawa Terrace because of the multiplicity of sources and contaminants and the ill-defined period of contamination. Therefore, the committee recommends the use of simpler approaches (such as analytic models, average estimates based on monitoring data, mass-balance calculations, and conceptually simpler MODFLOW/MT3DMS models) that use available data to rapidly reconstruct and characterize the historical contamination of the Hadnot Point water-supply system. Simpler approaches may yield the same kind of uncertain results as complex models but are a better alternative because they can be per- formed more quickly and with relatively less resources, which would help to speed-up the decision- making process. As needed, and if available, better field characterization and details (such as active supply wells and cycling schedules, geologic data, and source characteristics) may be added to the conceptual models to improve understanding of transport at selected locations where potential exposure was high. Detailed MT3DMS modeling studies should be performed only for selected sites (using locally-refined grids) and only after establishing a priori the clear need, objectives, and expected outcomes for such studies. On the basis of the results of the Tarawa Terrace models, application of cutting-edge research codes for ground- water modeling (such as PSOpS and TechFlow) is unlikely to reduce uncertainty in the Hadnot Point ex- posure scenarios, which are expected to be much more complex than at Tarawa Terrace. Future modeling efforts should also be aided by additional field information about the physical and chemical characteristics of the sources and receptors (aquifers). Specifically, the hydrogeologic char- acterization of the recharge zones of the primary aquifer that was and is the source of water for the water- supply systems at Camp Lejeune should be determined. For example, the extent and characterization of the Castle Hayne confining unit are critical for understanding the potential for hydraulic connectivity be- tween the waste sites identified and the source aquifer for the water-supply wells over the period of poten- tial exposure (1943-present). It is well documented that the confining layer is neither continuous nor con- fining in all areas beneath the Camp Lejeune geographic boundary. The committeeâs effort to evaluate potential exposures to contaminants in the Tarawa Terrace and Hadnot Point water systems was hampered by the fact that the available data on water quality of those systems was found in hundreds of documents. Most of the documents are publicly available on line, but
66 Contaminated Water Supplies at Camp LejeuneâAssessing Potential Health Effects they were not readily searchable or cataloged in an organized fashion for research. To facilitate future exposure-assessment efforts, the committee strongly recommends that a comprehensive, accessible data- base of water-quality measurements (including data from remedial investigations) be created and main- tained. Such a database should include information on sample location, date, analytes measured, labora- tory quality-control information (including limits of detection), and other information relevant to exposure assessment that relies on environmental samples collected in the course of investigating water, soil, and air quality at Camp Lejeune. Because of the sparseness of water-quality data and the insufficient ability of water-quality mod- eling to make up for the absence of information, most exposure estimates in epidemiologic studies at Camp Lejeune will rely heavily on unverifiable assumptions and projections. Therefore, the most useful exposure assessment will likely be relatively crude and based for the most part on ascertaining the most likely time period and location (water supply system) of contamination, typical locations the study par- ticipant spent time on the base (for example, residence, school, daycare, workplace), and crude categori- zation of personal water-use activities during the exposure period.