This status report has been prepared for the U.S. Environmental Protection Agency (EPA) in partial fulfillment of the statement of work for Response Action Contract (RAC) No. 68-W6-0037 for Region 6, Work Assignment No. 943-RSBD-0539. Tetra Tech EM Inc. (Tetra Tech) prepared this report to provide a summary of the current status of groundwater, surface water, sediment, and fish data for the Neal's Landfill site in Monroe County, Indiana. Tetra Tech reviewed available site data obtained from EPA Region 5 as well as data that became available to Tetra Tech from Viacom, Inc. (Viacom), the potentially responsible party for the site. Available data collected through October 2001 are included in this report.

This status report consists of five sections, including this introduction. Section 2.0 contains site background information for Neal's Landfill, Section 3.0 summarizes site hydrogeologic and spring flow data, Section 4.0 summarizes site chemical data, and Section 5.0 summarizes Tetra Tech's review findings. References used to prepare this report are listed after Section 5.0. Figures and tables cited in the text appear in Appendixes A and B respectively. Attachments 1 through 18 contain site information collected by various parties.

This section summarizes the site's history, the site geology and hydrogeology, and spring flow measurements.

2.1 SITE HISTORY

Numerous field investigations have been conducted at Neal's Landfill by EPA, Viacom, the Indiana Department of Environmental Management (IDEM), the U.S. Geological Survey (USGS), and the U.S. Fish and Wildlife Service (USFWS). Sampling efforts have focused on sediment, surface water, fish, and vegetation in Conard's Branch and Richland Creek; springs located near the landfill; on-site soil; monitoring wells on and off the site; residential wells in the vicinity of the landfill; and air upwind and downwind of the landfill. Data obtained by EPA and the Indiana State Board of Health indicate that high PCB concentrations were present in surface soil in the northeast portion of the landfill. PCBs were also detected in spring water near the site, in sediment in nearby springs and seeps, in fish in Richland Creek, and in fat from a calf that grazed on site. In addition, data for monitoring wells installed near the site indicate the presence of volatile organic compounds (VOC) in groundwater.

The site was placed on the National Priorities List in October 1981. In 1985, EPA, the State of Indiana, Monroe County, the City of Bloomington, and Westinghouse (now Viacom) signed a consent decree. Under the terms of the consent decree, Viacom is to remediate six sites in the Bloomington area containing PCBs. Neal's Landfill is one of the six sites covered by the consent decree. In 1987, a number of interim measures were completed at the site, including removal of capacitors from the ground surface; removal of a total of 4,500 linear feet of contaminated sediment from Conard's Branch (4,267 linear feet) and Richland Creek (233 linear feet); and installation of a soil cap over primary landfill areas. The interim measures also included construction of a spring treatment facility (STF), which began operation in 1990. The purpose of the STF is to collect and treat spring flow of up to I cubic foot per second (cfs). This flow is currently collected in a lined basin and travels by gravity to the STF. The STF consists of a clarifier, filters, and an activated carbon water treatment system that removes PCBs from the spring flow. The STF is designed to treat the water so as to achieve an effluent PCB concentration of I part per billion (ppb). (National Pollutant Discharge Elimination System [NPDES] monitoring data for the STF are presented in Section 4.5 of this report.) Spring flows exceeding 1 cfs overflow the lined basin and discharge directly to Conard's Branch.

Under the consent decree, Viacom was scheduled to excavate soil overlying the site and incinerate the soil in a municipal waste facility that Viacom was to build. However, remedial actions (RA) other than soil incineration had to be considered when the Indiana State Legislature passed a law in 1993 that was intended to block implementation of the incineration remedy.

In March and April 1998, EPA and Viacom conducted field investigations at the site. The objectives of these investigations were to characterize the landfill waste and identify PCB hot spots for removal. A total of 104 borings were drilled through landfill materials at approximately 100-foot spacings.

Figure A-2 is a geologic cross section (B-B') depicting the interim cap, underlying waste material, native lay, and bedrock as well as PCB data for several of the 104 borings. Cross section locations are shown in Figure A-3. Data from the investigations revealed the presence of PCB hot spots throughout the landfill at concentrations exceeding 500 parts per million (ppm). Data from the investigations were used to evaluate RA alternatives by Viacom and government parties. The selected alternative includes excavation of materials containing PCB concentrations exceeding 500 ppm and disposal of the materials at an off-site, Toxic Substances Control Act-permitted landfill. Materials containing PCB concentrations less than 500 ppm were to be consolidated on site under a Resource Conservation and Recovery Act (RCRA) Subtitle C cap.

In April 1999, Viacom and its RA contractors began implementing the selected source removal RA under the oversight of EPA, the State of Indiana, and Monroe County. Viacom completed material excavation, material disposal off site, waste consolidation, and construction of the RCRA Subtitle C cap in November 1999. Final waste limits and the extent of the RCRA Subtitle C cap are presented in Figure A-4.

2.2 SITE GEOLOGY AND HYDROGEOLOGY

Sections 2.2.1 through 2.2.4 discuss the Neal's Landfill site geology and hydrogeology, including the Northwest Spring System, the South Spring groundwater basin, and groundwater levels at the site.

2.2.1 Geologic and Physiographic Setting

The Neal's Landfill site is situated on a topographic saddle that overlies a watershed divide on an east- and west-oriented ridge. Natural surface drainage is generally to the north and west into tributaries of Richland Creek. The northern tributary is known as Conard's Branch and flows into Richland Creek 3,000 feet northwest of the site. The west-flowing tributary draining the southern portion of the landfill, referred to as Taylor Branch, joins Richland Creek about 6,000 feet west of the site. Headwater portions of Taylor Branch are clearly shown on 1957 topographic maps to have extended under the southeast portion of the landfill. During the early operation of Neal's Landfill, these valley areas were filled in to a depth of about 20 to 25 feet. The location of the buried portions of Taylor Branch are shown on early EPA reconnaissance mapping of the area (see Powell 1983, Figure 2 in Attachment 1). Surface drainage ponded behind the waste formed the cattail pond.

The site is located near the eastern margin of the Crawford Upland Physiographic Province, which is characterized by rugged topographic features that include steep slopes. The unconsolidated overburden consists mainly of clays and silts. At the site, waste material is separated from the underlying bedrock by a layer of clay (see Figure A-2). The bedrock is composed of thin limestones, sandstones, and shales assigned to the Mississippian-age West Baden Group that overlie limestone bedrock assigned to the St. Genevieve Limestone of the Blue River Group (Shaver and others 1986). Three different members make up the St. Genevieve Formation, including (in descending order) the Levias Member, the Spar Mountain Member, and the Fredonia Member. The St. Genevieve Formation is a relatively pure, well- bedded limestone that supports development of dissolution features and karst. Shale interbeds and chert beds are also present. Numerous sinkholes exist in the site area and are believed to exist beneath portions of the landfill outside the waste consolidation area. Valley bottoms immediately north and south of the saddle are eroded into lower parts of the St. Genevieve Formation, and the entire area is underlain by the St. Louis Limestone, the lowest unit of the Blue River Group. Attachment 1 provides a detailed description of the site geology and hydrology (Powell 1983).

Strata in the region generally dip or slope from the crest of the Cincinnati Arch to the west or southwest into the Illinois Basin at a rate of about 25 to 30 feet per mile. Local variations of this regional trend are not uncommon (Perry and Smith 1958). Correlation of chert marker beds in site boreholes suggests a northwesterly dip in the immediate area of the landfill (Powell 1983)

Subterranean flow routes from sinkholes, sinking streams, and wells to spring emergences in the site area were identified during fluorescent dye tracer tests conducted from 1990 to 1995. Subterranean drainage from the Neal's Landfill area is known to resurge at springs (see topographic map in Figure A-5) located at the head of Conard's Branch northwest of the landfill area and along this tributary to Richland Creek.

2.2.2 Northwest Spring System

The resurgences along Conard's Branch, which are collectively referred to as the "Northwest Spring System" by Viacom (2001 d), are known to discharge PCBs associated with Neal's Landfill. The springs include North Spring, South Spring, and several storm water overflow springs in the vicinity of South Spring (see Figure A-6). Most of the site is known to drain to the Northwest Spring System. The perennial discharge point for this flow system is South Spring. The southwestern portion of the site is known to drain to the southwest to Taylor Spring and Branam Spring, where low concentrations of PCBs are inconsistently detected.

North Spring is the lowest underflow spring associated with the Northwest Spring System. North Spring appears to be supplied in par, by the South Spring conduit system as well as by independent drainage from adjacent areas. The fact that lower PCB concentrations have been detected at North Spring than at South Spring may reflect this additional drainage contribution. Water discharging from North Spring enters a lined basin and is pumped to the STF. High flows from the North Spring overflow the lined basin and are discharged directly to Conard's Branch.

Five overflow springs are present in the vicinity of South Spring, and all are tributaries to Conard's Branch. These springs are normally dry but contribute most of their discharge to Conard's Branch during high-flow storm events. During such storm events, flow from the springs may be continuous for several days. The springs are referred to as Overflows 0, 1, 2, 3, and 4 (see Figure A-6).

Dye tracer experiments and groundwater quality data both suggest that all the springs comprising the Northwest Spring System are common discharge points for a single groundwater flow system and that they are fed by a single solutional conduit flow system. Such common-source overflow/underflow spring systems are common in the south-central Indiana karst area.

During a storm event, the overflow springs begin to flow in the order 0, 1, 2, 3, and 4. During the storm's recession, flow ceases at the springs in the reverse order. A small storm event may activate flow only at Overflows 0, 1, and 2. A larger storm event will also activate Overflow 3, and the largest storm events will activate Overflow 4.

Aggregate flows from North Spring and South Spring of up to about I cfs are routed directly to the STF. At present, some of the flow from Overflows 1, 2, and 3 is captured in small, riprap-lined collection basins and is routed to the lined basin at South Spring. If this flow exceeds the rated capacity of the STF, the excess flow overtops the lined basin and discharges directly to Conard's Branch above Gage CB.

2.2.3 South Spring Groundwater Basin

The land area within which infiltrating precipitation eventually reaches the Northwest Spring System may be referred to as the South Spring groundwater basin. The red line in Figure A-5 outlines the possible extent of the drainage area, or groundwater basin, that discharges to the Northwest Spring System emergence area under low-flow conditions. The Neal's Landfill site is located at the extreme downstream end of the Groundwater basin. Dye tracer tests performed from 1993 to 1995 helped to establish the size of the groundwater basin. The Sinks of Cave Creek, a large sinking stream, drain to Richland Springs under both low- and high-flow conditions (see Figure A-5). One off-site sinkhole located east of Cave Road (Taylor Sink) was found to drain about 6,000 feet to the northwest to South Spring. Based on this finding, the large karst valley extending north from the Sinks of Cave Creek is included in the South Spring groundwater basin. Harshman Sink is known to drain to the northwest to Rogers Spring, and therefore this sink area is excluded from the South Spring groundwater basin.

Groundwater flow beneath the site occurs in solutional conduits developed in the limestone. These solutional conduits develop along bedding plane surfaces and natural fractures. Bedrock samples collected during site investigations indicate that solution cavities are forming in the limestone below the groundwater surface. In addition, joints in the limestone bedrock were tectonically induced during Cincinnati Arch and minois Basin formation. Based on measurements of outcropping bedrock, east-west and north-south master joint sets both at the groundwater surface and at depth in the bedrock contribute to vertical and horizontal groundwater infiltration into the bedrock aquifer (Powell 1983).

2.2.4 Groundwater Levels

Data collected from monitoring wells screened in the limestone bedrock indicate that subterranean passages transport contaminated groundwater. Groundwater elevation and dye tracer study data both indicate that groundwater flows to the northwest. A typical groundwater contour map is provided in Figure A-7. This map indicates a potentiometric surface graded to the north to the approximate 730-foot elevation of South Spring, the perennial spring discharge point for the groundwater flow system.

Seeps occurring throughout the landfill are discharge points for water that has infiltrated through or beneath the landfill. PCB concentrations detected in monitoring wells and at the seeps indicate that source materials may have been in contact with water within the landfill prior to source removal and consolidation activities (see Geologic Cross Section D-D' in Figure A-8). The low-lying waste above the highest recorded groundwater elevation of 773.5 feet above mean sea level was removed during the RA. Therefore, since the completion of source removal and consolidation activities in 1999, groundwater levels have been below the bottom of the waste (Viacom 2001d).

3.1 EARLY HYDROGEOLOGIC INVESTIGATIONS

Powell (1983) prepared a report detailing the geology and hydrology of the site. The report provided a detailed review of site geology, interpretation of site geologic structure and stratigraphy based on detailed rock core examination, and established the karst nature of the site. The report identifies 13 sinkholes or swallowholes, 5 possible sinkholes, 5 possible buried sinkholes or swallowholes, and 7 nonsinkhole depressions. The report also concludes that PCBs found in the North and South Springs were an indication of movement of contaminated water from beneath Neal's Landfill into Conard's Branch and Richland Creek.

Several site monitoring wells were installed in the early 1980s by both EPA and Westinghouse and became the focus of early groundwater monitoring efforts. USGS conducted an extensive surface water flow and groundwater level measurement program during this period. The USGS flow data are discussed later in this report.

3.2 ON-SITE GROUNDWATER LEVEL MONITORING

In August 1987, Westinghouse submitted a final on-site groundwater monitoring plan for the Neal's Landfill site (Blasland & Bouck Engineers, P.C. 1987). This monitoring plan relies heavily on use of monitoring wells completed in the limestone bedrock. Monitoring wells selected for Neal's Landfill in the plan include MW-2, MW4, MW-5, EPA-3A, EPA-5A, EPA-6A, and EPA-9A. The monitoring plan presents the results of various site surveys; monitoring well construction details; surface water, groundwater, and sediment sampling data collected during sampling events conducted from 1981 to 1984; a summary of groundwater elevation data collected during well installation from 1982 to 1985; boring logs for selected monitoring wells installed in 1982 and 1983; rain gauge monitoring data collected from 1982 to 1984; and calculated site water balances.

Table B-1 summarizes the groundwater monitoring well construction details for Neal's Landfill. Westinghouse performed semiannual groundwater sampling at the Neal's Landfill site from May 1989 to December 1998 in order to establish baseline PCB concentrations in on-site monitoring wells. Prior to each sampling event, water levels were measured in wells that were to be sampled as well as in adjacent observation wells. Water levels were measured in the following monitoring wells as part of the semiannual monitoring: MW-3, MW4, MW-5, EPA-1AA, EPA-2A, EPA-3A, EPA4A, EPA-5A, EPA-6A, EPA-8A, EPA-9A, and EPA-1OS (CBS 1999a). Water level data collected during the May 1998 sampling event were used to construct the groundwater contour map presented in Figure A-7 (CBS 1998c). Water level data collected during the December 1998 sampling event were consistent with the May 1998 data.

CBS began interim Groundwater pursuant to the Neal's Landfill Statement of Work (SOW) in April 1999. Monitoring has been conducted bimonthly since that time at a reduced number of wells (MW-3A, MW 4, MW-5, and MW-5A). The water levels that have been recorded at the wells being measured are consistent with historical records.

In the summer of 2001, several owners of private water supply wells in the general area of Neal's Landfill reported loss of well water supply. Under the provisions of the Indiana Emergency Water Rights statute (IC 14-25), the Indiana Department of Natural Resources (IDNR), Division of Water, conducted field investigations in the area to document the water supply problem. Residences of the area reported to the IDNR that their wells had failed to yield their normal water supply due to water withdrawal and blasting operations conducted at a crushed stone quarry located in the area. The IDNR investigations revealed that quarry operations of the Rogers Group, Inc., had substantially lowered Groundwater levels in the general area around the quarry (IDNR 2001). IDNR documented water level declines in the range of 14 to 29 feet in the areas south, west, and northwest of the quarry. Water level declines were noted in wells located ùover 1 mile from the quarry and including the general area of Neal's Landfill. In response to the water level declines, the quarry operators furnished temporary water supply tanks for several residences and initiated the installation of public water supply to affected residences.

Since the area where water level declines were noted by IDNR include Neal's Landfill, EPA expressed concern to Viacom that the declining water levels might adversely affect water levels, Groundwater flow direction, and spring flows at Neal's Landfill. Viacom initiated an investigation of water levels and spring flows at the landfill.

• Hourly water level records for monitoring well 5A
• Hourly water level records for landfill piezometer PZ-OI
• Semiannual hand measured water levels for monitoring well 1AA from May 1989 to December 1998 STF run records for the month of September 2001
• Hand measured water levels taken at monitoring well IAA on September 27 and October 2, 2001
• Hand measured water levels taken at monitoring wells EPA-2A, EPA-3A, MW-3, EPA dA, MW-4, MW-5, EPA-5A, EPA-8A, EPA-9A, and MW-1OS on October 22, 2001

3.3 TRACER TEST INVESTIGATIONS

In 1990 and 1992, dye tracer studies were conducted at the site to identify all low- and high-flow resurgences of groundwater originating at the site as well as any residential wells that might be in hydrologic connection with the landfill. Figure A-9 shows the locations of spring and stream monitoring stations used during the low- and high-flow tracer tests at Neal's Landfill.

The low-flow tracer test was conducted from September 23 through December 10, 1990. Spring and stream background samples were collected in July and August 1990 prior to the introduction of the tracer. The fluorescent dye Fluorescein was used in the low-flow tracer test. Fluorescein was injected sequentially at the following locations: (1) monitoring well 10SS; (2) the sink by the Southwest Seep pump station; (3) monitoring well EPA-3A; (4) monitoring well EPA-6A; (5) monitoring well MW-5; (6) monitoring well 11; and (7) monitoring well 1AA. The monitoring locations (and the concentration ranges of Fluorescein detected) included North Spring (below detection limit [BDL] to 1,450 ppb), South Spring (BDL to 1,800 ppb), Taylor Spring (BDL to 650 ppb), and Branam Spring (BDL to 160 ppb). The 1990 low-flow tracer test confirmed that most of the site drains to South Spring and Conard's Branch north of the landfill. Figure A-10 summarizes the 1990 low-flow test tracer results. As shown in this figure, a small area in the southwestern corner of the landfill drained to Taylor and Branam Springs. No other off-site springs or streams received tracer during the low-flow test to be used for the high-flow test (Westinghouse 1990). Spring and stream background samples were collected for dye analysis in 1991 and 1992 before th high-flow test. Low background concentrations of Fluorescein were present in the samples; a few samples had low background concentrations of Rhodamine WT, and one spring sample had a low background concentration of eosine.

The high-flow tracer test was conducted from April 18 through May 10, 1992. It did not begin until the two high-flow requirements for the tracer test were met: (1) the combined flow of the South Spring and Conard's Branch overflow springs exceeded 4,500 gpm, and (2) monitoring well EPA-5A had a water elevation of at least 743 feet above mean sea level.

In contrast to the low-flow tracer test, three fluorescent dyes (Fluorescein, Rhodamine WT, and eosine) were to be used for the high-flow test.

The tracers were injected in two rounds. Each dye was injected at two locations. Injections of a given dye were spaced 4 hours apart in the hope that separate dye breakthrough curves could be noted at dye recovery sites. The simultaneous injection of multiple dyes was intended to support a determination of where specific areas of the landfill drained. Because three different dyes were injected at two different times, individual well resurgence and landfill segment resurgence could be identified if the tracers went off site (CBS 1996). The first sequence of injections included the following tracers and locations: (1) Rhodamine WT in the sink by monitoring well EPA-3A (2) eosine in monitoring well 10SS, and (3) Fluorescein in monitoring well EPA-6A. The second sequence of injections commenced 4 hours after the first and included the following tracers and locations: (1) eosine by the Southwest Seep pump station, (2) Rhodamine WT in monitoring well MW-11, and (3) Fluorescein in monitoring well 1AA.

The 1992 high-flow tracer test had results similar to those of the 1990 low-flow test (CBS 1996). Figure A-11 depicts the results of the high-flow tracer test. This figure indicates that injection points on the southwestern side of the site drained to Taylor and Branam Springs. One injection point on the northeastern side of the site may have drained northwest to Pig Pen Spring, but this spring was not monitored during the test because of access considerations, and the nature of flow to the spring is conjectural. Most of the landfill drains to the Northwest Spring System at North and South Springs.

• North Spring Fluorescein - BDL to 0.771 ppb, cosine - BDL to 0.447 ppb, and i. Rhodamine WT - BDL to 1.566 ppb
• South Spring: Fluorescein - BDL to 6.599 ppb, cosine - BDL to 29.691 ppb, and Rhodamine WT - BDL to 2.793 ppb
• Conard's Branch: Fluorescein - BDL to 1.15 ppb, cosine - BDL to 7.32 ppb, and Rhodamine WT - BDL to 2.41 ppb
• Taylor Spring: Fluorescein - BDL to 1.81 ppb, cosine - BDL to 6.88 ppb, and Rhodamine WT - BDL to 115.18 ppb
• Branam Spring: Fluorescein - BDL to 1.39 ppb, cosine - BDL to 1.24 ppb, and Rhodamine WT - BDL to 2.56 ppb

Residential wells within 1 mile of the site were also monitored for tracer during the high-flow tracer test. One residential well (number 163) showed possible signs of tracer; however, during subsequent sampling events, no detectable levels of PCBs or other contaminants were found in this well. Attachment 2 shows the residential well locations near the landfill, and Table B-2 summarizes the residential well samples collected during the high-flow tracer test.

3.4 PIEZOMETRIC MONITORING OF WASTE DEPOSITS

Piezometers to monitor water levels within or beneath waste deposits were installed prior to the 1999 site RA. Additional piezometers were installed in 1999 to monitor the consolidated waste material after completion of the source control RA.

In March 1998, Tetra Tech installed eight polyvinyl chloride (PVC) piezometers to measure water (leachate) elevations in site areas where saturated waste materials were encountered. Piezometers NL-PZ14, NL-PZ20A, NL-PZ20B, and NL-PZ24 were completed in the southern portion of the landfill, and piezometers NL-PZ60, NL-PZ61, NL-PZ93, and NL-PZ94 were completed in the central portion of the landfill. Table B-3 summarizes the piezometer construction details. The eight piezometers were removed during the source removal RA and the construction of the RCRA Subtitle C cap.

The depth to water in the piezometers was measured on April 2, 1998, and again (as requested by EPA) on April 16, 1998, following a significant rainfall event in me Bloomington area (over 2 inches of rain in a 24-hour period). The depth-to-water data and corresponding water elevations are summarized in Table B-4. The data show that water levels did not change appreciably over the 2-week period from April 2 through 16 except in piezometers at borings NL-PZ60 and NL-PZ61, where water levels increased following the April 16 rainfall event. Additionally, the data indicate that piezometers NL-PZ14 and NL-PZ94 were dry on April 2 and 16, 1998.

In general, data collected from the piezometers during the 1998 PCB hot spot delineation investigation indicated that the occurrence of perched saturated zones in the landfill is somewhat random and that no identifiable saturated zone extends throughout the landfill. More recent Piezometer water elevation data (Attachment 3 shows water elevation data for ters NL-PZ24, NL PZ61 and NL PZ93 monitoring well MW-5A from May through October 1998) provided by Viacom (CBS 1998d) indicate that the waste consolidation area at the southern end of the site is not prone to groundwater backflooding. Based on monitoring data, groundwater backflooding is apparently not occurring in other areas of the site. (Tetra Tech 1998c; Viacom 2001d).

As part of the source removal RA Viacom installed five Piezometers that are screened on top of bedrock w

ithin the RCRA Subtitle C cap area for future monitoring of water within the landfill. Piezometer construction details are presented in the Neal's Landfill construction completion report (CBS 2000a).

Data are being obtained from the five piezometers as part of the long-term monitoring activities for the site. Continuous monitoring data to groundwater elevations are presented in Table B-5 (Viacom2000a). In addition, Viacom conducts continuous monitoring of (1) rainfall; (2) the flow in Conard's Branch; and (3)water levels in site monitoring wells EPA-5A, EPA-10S and PZ-1 Data collected during storm events are reported by Viacom on a quarterly basis.

3.5 SPRING FLOW REGIME

3.5 EPA/USGS Flow Records (1982 to 1984)

Flow data were collected by USGS at two continuously recording flume installations at the Neal's Landfill site from October 1982 through November 1984. The data were collected usign digital tape punch equipment. Stage data were collected at 15-minute intervals. The South Flume (USGS Gage 391010086383301) was located just downstream of the present culvert crossing Conard's Branch and measured the discharge from South Spring and all the associated overflow springs (see Figure A-6). These measurements were made prior to the construction of the STF and therefore include all spring flow data. The North Flume (USGS Gage 391013086383401) measured discharge into Conard's Branch about 80 feet below the North Spring inflow (see Figure A-6).

Table B-6 provides an analysis of the South Flume provisional discharge data for a 754-day period from October 1982 through November 1984. The calculated total volume of flow for the 754-day period based on recorded mean daily discharge values was 438.8 million gallons, and the mean daily discharge was about 405 gpm. A plot showing the probability distribution is shown in Figure A-12. A summary tabulation of the flow probability data is presented in Coluruns 1 and 2 of Table B-6. Figure A-12 shows that there were 133 days (17 percent of the record) when the South Flume flow exceeded the 1 -cfs (448.8 gpm) design capacity of the STF.

Cumulative flow statistics were developed to determine the total volume of flow and the percentage of the flow volume discharged at South Flume at a mean daily discharge rate less than or equal to a given discharge. These data are plotted in Figure A-13 and a summary tabulation is provided in Columns 6 and 7 of Table B-6.

Based on Table B-6 and Figure A-13, about 75 percent (331.3 million gallons) of the discharge from South Spring (and the associated overflow springs) occurred at a rate exceeding 1.0 cfs.

3.5.2 Viacom Flow Monitoring at Conard's Branch (1993-1994 and 2000-2001)

After the construction of the STF, Viacom initiated Conard's Branch stream flow measurements at a location immediately downstream from South Spring and slightly upstream from the previous USGS South Flume location (see Figure A-6). Flow records are available for this gage from May 1993 until June 1994. The discharges from North and South Springs are not measured independently, but are totals from STF plant influent data. Neal's Landfill total flow was computed by adding together the Conard's Branch gage and plant influent flows. Because South Spring makes up most of the plant in fluent flow, the computed value is comparable to the 1982 to 1984 EPA/USGS data.

Continuous flow measurements were also conducted at the Conard's Branch gage in 2000. Low-flow monitoring results are of poor quality due to limitations in the monitoring system. High-flow measurements are based on an open channel rating curve. A v-notch weir flow gage was installed in Conard's Branch by Viacom in September 2001. This gage is located a short distance downstream of the EPA/USGS gage set up in 1982-1984.

3.5.3 EPA/USGS Measurements at Overflow Springs (2000-2001)

In May 2000, USGS under contract to EPA, installed four flow measurment devices to measure discharges from the Northwest Spring System. Flow measurement was conducted from June 2000 to November 2001. The combined discharge from Overflows 0, 1, and 2 and the combined discharge from Overflows 3 and 4 were measured by USGS using v-notch weirs (see Figure A-6). A third gage (Gage 0) was installed in the Conard's Branch channel immediately upstream of Overflows 1 and 2 to measure direct surface water flow off the landfill area. This gage, a 9-inch Parshall Flume, yielded marginal quality data due to submerged flow conditions in the flume throat. A fourth gage was established immediately below the North Spring STF collection sump. The gage was also a v-notch weir.

The EPA/USGS gages were installed to support storm-event sampling of the Overflow Springs and to provide better data related to flow bypassing the STF and directly entering Conard's Branch. The combined discharge of Gages 012 and Gage 34 is the total of Overflow Spring discharge entering Conard's Branch, and is roughly equivalent to the discharge measured at the Viacom Conard's Branch gage. The North Spring gage measured the quantity of North Spring flow that bypassed the STF collection sump and directly entaed Conard's Branch downstream of the STF.

Stage data from each gage were recorded using a float actuated shaft encoder and digital data recorda. Data wae recorded at 5- or 15-minute intervals. Raw field data were processed by the USGS to compute discharge for each 15-minute measurement. Data gaps in the flow record resulting from frozen float or equipment malfunction conditions were filled using flow correlation regression procedures. A hydrograph showing mean daily flow in gallons per minute measured at Gages 012, 34 and North Spring during the period June 2000 to November 2001 is shown in Figure A-14. Conard's Branch flow, also shown in Figure A-14, is estimated as the sum of Gage 012 and Gage 34. Total Conard's Branch flow volume during the 525 day monitoring period from June 7, 2000, through November 13, 2001, is estimated to be 124.4 million gallons. Of this, approximately 86.8 million gallons, or 70 percent of the total were discharged through Overflows Springs 0, 1 and 2. Generally, Overflow Springs 0, 1 and 2 flow for a longer duration and have a higher flow rate than Overflow Springs 3 and 4. This is related to the fact that Overflow Springs 0, 1, and 2 are located at a lowerr elevation than Overflow Springs 3 and 4 and therefore flow in response to a higher hydraulic head.

The flow volume estimates provided in the previous paragraph are conservative. Bypass of flow around Gages 012 and Gage 34 was noted on several occasions prior to March 21, 2001. Both gages were overtopped and damaged by a large storm event on October 5, 2000.

Table B-7 summarizes flow data for the ten storm flow events that produced the highest recorded instantaneous flow rates at Conard's Branch. The storms are ranked based on peak flow rate, as shown in Column 3. The mean daily discharge on the calendar date of the flow peak is shown in Column 4. The maximum 24-hour flow volume occurring over the peak of the storm is shown in Columns 6 and 7. Volumes are reported both in terms of millions of gallons (MG) and acre feet (Ac-Ft). The maximum flow volumes occurring for any 72-hour period during the storm are shown in Columns 6 to 9. Generally, the 72-hour data provide a good estimate of the total flow volume generated during a storm. The 72-hour data wae tabulated beginning at the even hour shown in Column 5.

The largest storm recorded during the June 7, 2000, to November 13, 2001, period occurred on October 5, 2000. The flow from this storm overtopped sandbags at both Gage 012 and Gage 34 and therefore these gages did not accurately record this storm. Flow rates and volumes for this storm are estimated from the Viacom Conard's Branch gage. The maximum recorded flow of 8,090 gpm through the USGS gages is about one-half of the total peak flow measured at the Conard's Branch gage. The peak flow recorded at the Viacom Conard's Branch gage for this event was estimated to be 17,700 gpm. The Octoba 5 storm also produced the largest 24- and 72-hour flow volumes during the monitoring period (see Table B-7). The October 5 storm was produced by large thunderstorm events occurring late afternoon on October 4 and continuing into the early morning hours on October 5. The Viacom rain gage at the STF recorded a rainfall amount of 4.02 inches occurring from 5:00 p.m. on October 4, 2000, until noon on October 5, 2000. The Monroe County Airport, located about 2 miles southeast of Neal's Landfill recorded 3.56 inches during this same time frame. The Indiana University Physical Plant in Bloomington recorded 4.15 inches on October 5. By comparison, 2-year 24-hour, 5-year 24-hour, and 10-year 24 hour rainfall amounts for Bloomington, Indiana, are 3.1, 4.0 and 4.5 inches, respectively. These data suggest that the October 5, 2000, storm was roughly equivalent to a 5-year recurrence interval event. The October 5, 2000, storm generated a 72-hour flow volume of about 90 acre feet, roughly 65 percent of which occurred during the 24-hour period around the peak of the storm.

The 24-hour and 72-hour flow volumes for the ten highest storms are generally related to the peak flow rate. This relation is controlled by the intensity of the storm and antecedent moisture conditions. Intense storms that occur during the growing season tend to produce a less sustained flow and lower flow volume relative to the peak flow than storms that occur during the non-growing season. The 72-hour flow volume is often impacted by back-to-back storm events that occur within a 72-hour period.

The USGS gages measuring overflows 012 and 34 did not capture all the flows into Conard's Branch; therefore, flows measured by Viacom at Conard's Branch were generally higher than reported by USGS. USGS gaging at Conard's Branch and Viacom gaging at the new Conard s Branch weir overlapped during the paiod October 9 to November 14, 2001. Figure A-15 shows a comparison of the hydrograph records during the October 24, 2001, storm event. Generally, the sum of USGS Gage 012 and Gage 34 flow data is very comparable with the Viacom measurement. The slight discrepancy at high flow rates for the four main storm peaks shown in Figure A-15 is most probably related to overflow from the South Spring STF sump during period of high flow. This overflow occurs downstream of the USGS gaging locations and upstream of the Conard's Branch weir. Figure A-15 further suggests that adding the flow from USGS Gage 012 to Gage 34 provides a reasonable estimate of total Conard's Branch flow.

The probability distribution of mean daily flow values measured by USGS during the periodd June 7, 2000, to November 13, 2001, is shown in Figure A-12. A summary tabulation of the flow probability data are provided in Table B-8. For all probability levels, the data plot below USGS South Flume data from 1982-1984. The 2000-2001 data do not include flow from South Spring. Flows from South Spring up to about 450 gpm are currently routed to the STF. However, the South Flume gage location (see Figure A-6) included South Spring flows in 1982-1984, and this is the principal reason for the difference in the curves at roughly the 5 percent probability level and higher. However, 2000 to 2001 flows at lower probability levels are also considerably lower. For example, the 1 percent probability flow in 1982-1984 was 6,300 gpm, but was only 2,500 gpm in 2000-2001. This differance is far greater than can be explained by the diversion of up to 450 gpm spring flow to the STF. This situation may be related to an apparent peak flow reduction that has occurred at the Northwest Spring System since the 1982 to 1984 paiod.

3.5.4 Conard's Branch Peak Flow Comparisons

In 2001, Viacom developed predictive equations that relate the peak flow at Conard's Branch to rainfall amounts and pre-storm flow rates. These equations have bean developed for various periods during which rainfall and comparative flow data wae collected. These periods include the 1982 through 1984 period when the USGS gages wae in place and the 1993 and 1994 period and 2000 period monitored by Viacom. Very good correlations wae obtained for the 1982 through 1984 data and the 1993 and 1994 data, with correlation coefficients of 0.8583 and 0.9071, respectively. The correlation for the 2000 data was not as good, with a correlation coefficient of only 0.6921.

The correlations developed for the three monitoring periods are as follows:

• 1982 through 1984: Peak flow = 9,863.51 x rainfall + 1.88 x pre-storm flow - 3,382.91
• 1993 and 1994: Peak flow = 10,061.42 x rainfall + 17.64 x pre-storm flow - 7,929.03
• 2000: Peak flow = 4,088.47 x rainfall + 5.3 x pre-storm flow - 2,322.22

Based on an equivalent pre-storm flow of 200 gpm, the relationship between rainfall and peak flow at Conard's Branch during the three monitoring periods is shown in Figure A-16. It is noteworthy that storms occurring in 2000 appear to have had about one-half the peak flow for a given rainfall amount compared to storms occurring from 1983 through 1984 or in 1993 and 1994. These data are consistent with the flow probability plots shown in Figure A-12 and discussed in Section 3.5.3.

The reason for the apparent reduction in peak flow in 2000 is unclear. The most probable reason is the construction of several storm wata detention basins in the tributaries to the North Branch of Cave Creek (see Figure A-5). These detention basins, which have all been constructed since 1984, provide storm water storage for various urban developments in the North Branch Cave Creek watershed. The detention basins have been installed to reduce chronic flooding problems in the lower reaches of the Cave Creek valley at the Sinks of Cave Creek. Flooding in the lower reaches of the Cave Creek valley occurs to a depth of several feet because of either the limited capacity of a sales of swallets at the sinks or the limited capacity of the subterranean conduit systems that drain the swallets (see Figure A-5).

Recently constructed detention basins in the area include those at Park 48 Business Park and the Monroe County airport as well as a 70-acre-foot impoundment at the Fieldstone development. In particular, the large Fieldstone structure could have a significant impact on peak flows at the sinks. This structure went into operation in Fall 1998. The spillway for the Fieldstone structure is closed by a telemetric signal from a float switch located at the sinks. When flood water at the sinks reaches a critical level, the spillway gates are closed. The gates then remain closed, containing water in the 70-acre-foot impoundment until the water level at the sinks recedes below flood level.

The detention basin explanation for the peak flow reduction at Conard's Branch suffers from the fact that no one has established that the North Fork of Cave Creek is hydraulically connected to the Northwest Spring System discharging to Conard's Branch. In fact, during fluorescent dye trace studies conducted at the Sinks of Cave Creek under both low- and high-flow conditions, no dye had been detected at any of the Northwest Spring System emergences. All dye injected at the sinks has been reported to emerge at Richland Springs (see Figure A-5).

Efforts to locate additional sinks or swallowhole areas upstream from the Sinks of Cave Creek have not been successful. A detailed search of over 2 miles of streambed above the sinks by Viacom and EPA personnel in April 2001 revealed no significant sink areas. Therefore, the most likely type of hydraulic connection between the sinks and the Northwest Spring System (if one does indeed exist) is a subterranean overflow route. Such overflow routes are common in the Indiana karst geological area, and in this case, the overflow route would drain some water from the Cave Creek drainage to the Northwest Spring System during periods of high flow. Reduction of peak flow to the Sinks of Cave Creek would reduce the period of pending in the sink areas, resulting in less frequent use of the subterranean overflow route and reductions in peak flow at the Northwest Spring System emergences.

Another possible explanation for the peak flow reduction may be surficial drainage alterations that were made during the 1999 RA. During the RA, a large volume of fill was removed from the southeastern portion of the landfill and consolidated. In effect, this removal resulted in reexcavation and reestablishment of surface drainage to Taylor Branch south of the landfill. This drainage had formaly been blocked by landfill material placed across the threshold of the valley. The blockage resulted in the creation of a small ponded area upstream of the landfill, formaly referred to as the cattail pond (see Attachment 1, Figure 2). This pond had no natural overflow point, and swallowholes in the pond area diverted surface water flow into the subsurface. Flow of surface water into the underlying karst drainage system probably occurred. Removal of the landfill plug on the downstream side of the cattail pond in 1999 would have acted to reduce the rate of any such infiltration, and hence reduce spring flow rates in the Northwest Spring System.

Causes of the apparent peak flow reduction are currently being investigated by Viacom as part of a continuing Groundwater investigation.

This section describes the current status of chemical data obtained from groundwater,surface water, sediment, and fish sample analyses and from analyses of STF influent and effluent samples. In addition, this section presents estimates of the PCB mass discharged to Richland Creek during various periods.

4.1 GROUNDWATER

From July 1982 through December 1983, P.E. LaMoreaux & Associates, Inc. (PELA), and Ecology and Environment (E&E) performed Groundwater sampling at on-site monitoring wells and residential wells. Sample analytical data are presented in Table NL-9 of Westinghouse's final on-site Groundwater monitoring plan (Westinghouse 1987) and in PELA's water quality data (PELA 1983) for Neal's Landfill (see Attachment 4). The importance of the PELA and E&E sampling is that it was the first concerted effort to evaluate the flow and PCB discharges associated with the Northwest Spring System. PCB analytical results ranged from nondetect to 9.77 microgram per liter (ug/L) at South Spring and from 0.7 to 3.96 ug/L at North Spring. PCBs were also detected in South Spring overflow springs. PCBs were detected during one or more sampling events in monitoring wells 3A (4.4 ppb) and 6A (1.6 and 3.48 ppb). In addition, PCB concentrations exceeded the maximum contaminant level (MCL) of 0.5 ppb at all locations at which PCBs were detected. This PCB analytical information was the total data set that existed prior to the NPDES permit negotiations for the STF. The samples were also analyzed for volatile organic compounds (VOC), metals, and pesticides. During the sampling period, monitoring well MW-1AA contained methylene chloride concentrations as high as 22,000 ppb, and monitoring well MW-5A contained trichloroethene (TCE) concentrations as high as 56,000 ppb (PELA 1983, 1987). All the analytical results are summarized in Attachment 4.

In October 1983, EPA collected a groundwater sample from the Conard residential well. Analytical results for the groundwater sample indicated that the PCB concentration was below the laboratory detection limit of 1.0 ppb (EPA 1983).

In 1985, residential wells around Neal's Landfill were inventoried and sampled by the Indiana University School of Public Environmental Affairs (IU SPEA). Of the 37 residential wells sampled by IU SPEA, 5 wells had measurable PCB concentrations ranging from 3 to 7 parts per trillion (ppt). The remaining 32 wells had PCB concentrations below the laboratory detection limit of 1 ppt. The residential well locations and analytical data for the sampling event are presented in Attachment 5.

Westinghouse performed semiannual groundwater sampling at the site from May 1989 to December 1998 in order to establish baseline PCB concentrations in on-site monitoring wells. The following monitoring wells were sampled as part of this semiannual monitoring: EPA-3A, EPA-5A, EPA-6A, EPA-8A, EPA-9A, MW4, and MOO-5. Monitoring wells EPA-3A, EPA-5A, EPA-6A, EPA-8A, and EPA-9A are all screened in the Fredonia Member of the St. Genevieve Formation limestone. According to the groundwater monitoring report submitted by CBS for sampling conducted from December 16 through 18, 1998, PCBs were detected in samples from monitoring wells EPA-3A (25 ppb), MW4 (1.4 ppb), MOO-5 (0.34 ppb), and EPA-5A (2.1 ppb); PCBs were also detected in the duplicate sample from EPA-3A (33 ppb) (CBS 1999a). Analytical results for the other groundwater samples collected indicated that PCB concentrations were below the laboratory detection limit of 0.1 ppb. Table B-9 summarizes PCB analytical data for groundwater from 1982 through 2000.

In February 1999, CBS initiated interim groundwater monitoring at wells and springs in the site area. Sampling of monitoring wells, which was conducted biannually from 1989 to 1998, was reduced in favor of interim monitoring of groundwater discharges at wells and springs demonstrated by tracer testing to be hydraulically connected to the site. The interim spring sampling locations have included North, South, Taylor, Branam, and Pig Pen Springs. Four site monitoring wells (EPA-3A, MW4, MW-5, and EPA-5A) were included in the interim program, and these wells were sampled from April 1999 until April 2000. A summary of the monitoring well and private water supply well sampling results is provided below.

In February, March, and May 1999, CBS performed pre-excavation groundwater sampling at residential wells near Neal's Landfill (CBS l999b). CBS collected samples from the Branam well and Conard well to be analyzed for PCBs. The analytical results indicated that PCB concentrations were below the laboratory detection limit of 0.1 ppb. Table B-10 summarizes the 1999 pre-excavation groundwater analytical results for the two residential wells and the springs in the vicinity of the site.

On April 22, 1999, CBS collected six pre-excavation samples from four on-site monitoring wells (CBS l999b). PCBs were detected in five of the six samples: EPA-3A (1.0 ppb), EPA-3A duplicate (1.2 ppb), MW4 (0.95 ppb), MW-5 (0.57 ppb), and EPA-5A (2.2 ppb). The PCB concentration in the sixth (blank) sample was below the laboratory detection limit of 0.1 ppb. Table B-9 summarizes the groundwater analytical results.

On June 16, 1999, during landfill excavation, CBS collected six samples from four on-site monitoring wells (CBS l 999c). PCBs were detected in five of the six samples: EPA-3A (21 ppb), EPA-3A duplicate (19 ppb), MW4 (1.3 ppb), MOO-5 (0.36 ppb), and EPA-5A (1.9 ppb). The PCB concentration in the sixth (blank) sample was below the laboratory detection limit of 0.1 ppb. Table B-9 summarizes the groundwater analytical results.

On June 22, 1999, during landfill excavation, CBS collected two groundwater samples from two residential wells near Neal's Landfill (CBS l999c). The PCB concentrations in the Branam well and Conard well samples were below the laboratory detection limit of 0.1 ppb. Table B-l l summarizes the groundwater analytical results.

On July 26 and 27 and August 10, 1999, Tetra Tech collected samples from 14 residential wells during low-flow conditions; CBS had not previously collected samples from these wells. Tetra Tech's samples were analyzed for PCBs and VOCs. Analytical results indicated that no PCBs or VOCs were detected at concentrations above their detection limits. Table B-12 summarizes the residential wells sampled during the sampling event. Attachment 6 contains analytical results for the samples collected.

On August 10, 1999, during landfill excavation, CBS collected six samples from four on-site monitoring wells (CBS l 999c). PCBs were detected in five of the six samples: EPA-3A (32 ppb), EPA-3A duplicate (29 ppb), MW-4 (1.2 ppb), MOO-5 (0.19 ppb), and EPA-5A (1.8 ppb). The PCB concentration in the sixth (blank) sample was below the laboratory detection limit of 0.1 ppb. Table B-9 summarizes the groundwater analytical results.

On August 11, 1999, during landfill excavation, CBS collected one groundwater sample from the Conard well near Neal's Landfill (CBS l999c). The PCB concentration in the sample was below the laboratory detection limit of 0.1 ppb. Table B-ll summarizes the groundwater analytical results.

On October 29, 1999, CBS collected six post-excavation samples from four on-site monitoring wells (CBS l999d). PCBs were detected in five of the six samples: EPA-3A (36 ppb), MOO-4 (2.3 ppb), MOO-5 (0.46 ppb), EPA-5A (2.3 ppb), and EPA-5A duplicate (2.3 ppb). The PCB concentration in the sixth (blank) sample was below the laboratory detection limit of 0.1 ppb. Table B-9 summarizes the groundwater analytical results. In addition, Tetra Tech collected two (one split and one nonsplit) groundwater samples from monitoring well EPA-5A on October 29. The split groundwater sample contained 2.4 ppb PCBs (estimate), and the nonsplit groundwater sample contained 18 ppm TCE (estimate).

On November 1, 1999, CBS collected one post-excavation Groundwater sample from the Conard well near Neal's Landfill (CBS l999d). The PCB concentration in the sample was below the laboratory detection limit of 0.1 ppb.

On December 17, 1999, CBS collected six post-excavation samples from four on-site monitoring wells (CBS 2000b). PCBs were detected in five of the six samples: EPA-3A (10 ppb), MW4 (1.1 ppb), MW-5 (0.37 ppb), EPA-5A (2.3 ppb), and EPA-5A duplicate (2.4 ppb). The PCB concentration in the sixth (blank) sample was below the laboratory detection limit of 0.1 ppb. Table B-9 summarizes the groundwater analytical results.

On December 18, 1999, CBS collected one post-excavation Groundwater sample from the Conard well near Neal's Landfill (CBS 2000b). The PCB concentration in the sample was below the laboratory detection limit of 0.1 ppb.

On February 22, 2000, CBS collected six post-excavation samples from four on-site monitoring wells (CBS 2000c). PCBs were detected in five of the six samples: EPA-3A (5.8 ppb), MOO-4 (0.16 ppb), MW- 5 (0.20 ppb), EPA-5A (1.8 ppb), and EPA-5A duplicate (1.6 ppb). The PCB concentration in the sixth (blank) sample was below the laboratory detection limit of 0.1 ppb. Table B-9 summarizes the Groundwater analytical results.

On February 22, 2000, CBS collected one post-excavation Groundwater sample from the Conard well near Neal's Landfill (CBS 2000c). The PCB concentration in the sample was below the laboratory detection limit of 0.1 ppb.

On June 13, 2000, Viacom collected one post-excavation Groundwater sample from the Conard well near Neal's Landfill (Viacom 2001a). The PCB concentration in the sample was below the laboratory detection limit of 0.1 ppb

4.2 SURFACE WATER

In November 1980, EPA collected seven surface water samples at and near Neal's Landfill. Based on the sample analytical results, North Spring and Northwest Spring were found to contain dichloropropane, dichloroethene (DCE), trichloroethane (TCA), and TCE. Attachment 7 presents the detailed analytical results and sampling locations.

From September 1982 through December 1983, PELA and E&E performed surface water sampling at and near Neal's Landfill. The samples were analyzed for VOCs, metals, pesticides, and PCBs (PELA 1983, 1987). PCBs were detected during one or more sampling events in runoff near monitoring well 6A (293 ppb), runoff between monitoring wells 8A and 9A (1.1 ppb), the on-site dug well (0.9 ppb), the South Flume (3.4 ppb), South Spring (1.73 to 7.0 ppb), North Spring (0.70 to 3.96 ppb), the North Flume (3.69 ppb), Southwest Seep (0.6 to 2.8 ppb), Overflow I (0.76 and 1.7 ppb), Overflow 2 (0.9 ppb), Overflow 3 (2.29 ppb), Deerlick Seep (1.41 ppb), Conard's Branch (1.7 to 2.7 ppb), Northwest Seep (1.17 and 2.8 ppb), and the on-site pond (0.2 to 3.0 ppb). PCB concentrations exceeded the ambient water quality criterion (AWQC) of 79 ppt (0.079 ppb) at all locations at which PCBs were detected. All VOC, metal, and pesticide analytical results are presented in Attachment 4.

In accordance with CBS's EPA-approved quality assurance project plan, pre-RA surface water samples were collected beginning in May 1998 from two locations along Richland Creek and one location along Conard's Branch. The sampling locations are shown in Figure A-17 (which is modified from McLaren/Hart Chemrisk 1998). The surface water samples were analyzed for PCBs. The analytical results for the two samples from Richland Creek indicate that PCB concentrations were below the laboratory detection limit of 0.1 ppb. The analytical results for the sample from Conard's Branch indicate a PCB concentration of 0.46 ppb (CBS 1998b). Table B-13 summarizes the surface water analytical results.

In February, March, and May 1999, CBS performed pre-excavation surface water sampling at springs around Neal's Landfill (CBS 1999b). CBS collected samples from Taylor Spring Branch, Branam Spring, South Spring, North Spring, and Pig Pen Spring. The samples were analyzed for PCBs. The February, March, and May water samples from South Spring contained detectable concentrations of PCBs ranging from 1.0 to 2.0 ppb. The samples collected from North Spring in February, March, and May contained detectable concentrations of PCBs ranging from 0.56 to 0.70 ppb. The analytical results for the samples from the other locations indicated that PCB concentrations were below the laboratory detection limit of 0.1 ppb. Table B-10 summarizes the surface water analytical results.

On June 22, 1999, during landfill excavation, CBS collected four surface water samples from South Spring, North Spring, Branam Spring, and Pig Pen Spring (CBS l 999c). Taylor Spring was not sampled by CBS, as the spring was dry. PCBs were detected in the samples from South Spring (3 ppb) and North Spring (1.5 ppb). The PCB concentrations in the Branam Spring and Pig Pen Spring samples were below the laboratory detection limit of 0.1 ppb. Table B-l l summarizes the surface water analytical results.

On July 27, 1999, Tetra Tech collected a sample from Anderson Spring; CBS had not previously collected samples from this spring. Tetra Tech's sample was analyzed for PCBs and VOCs. Sample analytical results indicated that no PCBs or VOCs were detected at concentrations above their detection limits. Attachment 6 contains the analytical results.

On August 11, 1999, during landfill excavation, CBS collected two surface water samples from North Spring and Pig Pen Spring (CBS l999c). Taylor and Branam Springs were not sampled by CBS, as they were dry; also, South Spring was not sampled by CBS, as the spring flow was inadequate for sample collection on August 11. PCBs were detected in the sample from North Spring (1.9 ppb). The PCB concentration in the Pig Pen Spring sample was below the laboratory detection limit of 0.1 ppb. Table B-l l summarizes the surface water analytical results. CBS sampled South Spring on August 16, 1999. PCBs were detected in the sample at a concentration of 2.8 ppb.

On November 1, 1999, CBS collected six post-excavation surface water samples from five springs around Neal's Landfill (CBS l999e). PCBs were detected in three of the six samples: the South Spring sample (1.0 ppb), the South Spring duplicate sample (1.1 ppb), and the North Spring sample (1.2 ppb). The PCB conceritrations in the Branam Spring, Pig Pen Spring, and Taylor Spring samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On December 18, 1999, CBS collected five post-excavation surface water samples from five springs around Neal's Landfill (CBS l999e). PCBs were detected in two of the five samples: the South Spring sample (2.4 ppb) and the North Spring sample (2.1 ppb). The PCB concentrations in the Branam Spring, Pig Pen Spring, and Taylor Spring samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On February 22, 2000, CBS collected seven post-excavation samples from five springs around Neal's Landfill (CBS 2000c). PCBs were detected in three of the seven samples: the South Spring sample (1.8 ppb), the North Spring sample (0.68 ppb), and the Taylor Spring sample (0.11 ppb). The PCB concentrations in the Branam Spring, Pig Pen Spring, Pig Pen Spring duplicate, and field blank samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On April 14, 2000, Viacom collected six post-excavation surface water samples from five springs around Neal's Landfill (Viacom 2001a). PCBs were detected in three of the six samples: the South Spring sample (0.81 ppb), the South Spring duplicate sample (0.73 ppb), and the North Spring sample (0.29 ppb). The PCB concentrations in the Branam Spring the April 14 sample container broke in the laboratory, so Branam Spring was resampled on April 25, 2000, Pig Pen Spring, and Taylor Spring samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On June 13, 2000, Viacom collected seven post-excavation samples from five springs around Neal's Landfill (Viacom 200 la). PCBs were detected in six of the seven samples: the South Spring sample (1.6 ppb), the South Spring duplicate sample (2.2 ppb), the North Spring sample (1.6 ppb), the Taylor Spring Branch sample (0.18 ppb), the Branam Spring sample (0.17 ppb), and the field blank (0.12 ppb). The Taylor Spring Branch, Branam Spring, and field blank sample analytical results were flagged "R" (unusable) by the laboratory. The South Spring, South Spring duplicate, and North Spring sample analytical results were flagged "J" (estimated) by the laboratory. The PCB concentration in the Pig Pen Spring sample was below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On August 15, 2000, Viacom collected seven post-excavation samples from five springs around Neal's Landfill (Viacom 2001a). PCBs were detected in three of the seven samples: the South Spring sample (2.3 ppb), the South Spring duplicate sample (2.4 ppb), and the North Spring sample (1.5 ppb). The PCB concentrations in the Pig Pen Spring, Taylor Spring Branch, Branam Spring, and field blank samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On October 19, 2000, Viacom collected seven post-excavation samples from five springs around Neal's Landfill (Viacom 2001d). PCBs were detected in three of the seven samples: the South Spring sample (1.1 ppb), the South Spring duplicate sample (1.0 ppb), and the North Spring sample (0.36 ppb). The PCB concentrations in the Pig Pen Spring, Taylor Spring Branch, Branam Spring, and field blank samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On December 21, 2000, Viacom collected five post-excavation surface water samples from four springs around Neal's Landfill (Viacom 2001b). PCBs were detected in three of the five samples: the South Spring sample (0.84 ppb), the South Spring duplicate sample (0.91 ppb), and the North Spring sample (0.33 ppb). The PCB concentrations in the Taylor Spring Branch and Branam Spring samples were below the laboratory detection limit of 0.1 ppb. Pig Pen Spring was frozen on December 21 and was not sampled. Table B-14 summarizes the surface water analytical results.

On February 27, 2001, Viacom collected seven post-excavation samples from five springs around Neal's Landfill (Viacom 2001c). PCBs were detected in three of the seven samples: the South Spring sample (0.83 ppb), the South Spring duplicate sample (0.78 ppb), and the North Spring sample (0.23 ppb). The PCB concentrations in the Pig Pen Spring, Taylor Spring Branch, Branam Spring, and field blank samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On March 21, 2001, Viacom collected seven post-excavation samples from five springs around Neal's Landfill (Viacom 2001c). PCBs were detected three of the seven samples: the South Spring sample (0.93 ppb), the South Spring duplicate sample (0.92 ppb), and the North Spring sample (0.31 ppb). The PCB concentrations in the Pig Pen Spring, Taylor Spring Branch, Branam Spring, and field blank samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On April 11, 2001, Viacom collected seven post-excavation samples from five springs around Neal's Landfill (Viacom 2001e). PCBs were detected in three of the seven samples: the South Spring sample (1.5 ppb), the South Spring duplicate sample (1.3 ppb), and the North Spring sample (0.63 ppb). The PCB concentrations in the Pig Pen Spring, Taylor Spring Branch, Branam Spring, and field blank samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On April 12, 2001, Tetra Tech collected surface water samples from two locations in Cave Creek and one location in Richland Springs. These sampling locations are hydraulically upgradient of Neal's Landfill and are therefore not affected by PCB discharges associated with the landfill. The PCB concentrations in all three samples were below the laboratory detection limit of 0.1 ppb.

On May 15, 2001, Viacom collected six post-excavation samples from four springs around Neal's Landfill (Viacom 2001e). PCBs were detected in three of the six samples: the South Spring sample (1.5 ppb), the South Spring duplicate sample (1.6 ppb), and the North Spring sample (1.1 ppb). The PCB concentrations in the Pig Pen Spring, Branam Spring, and field blank samples were below the laboratory detection limit of 0.1 ppb. Taylor Spring Branch was not sampled on May 15 because it was dry. Table B-14 summarizes the surface water analytical results.

On June 14, 2001, Viacom collected seven post-excavation samples from five springs around Neal's Landfill (Viacom 2001e). PCBs were detected in three of the seven samples: the South Spring sample (1.4 ppb), the South Spring duplicate sample (0.89 ppb), and the Taylor Spring Branch sample (0.57 ppb).

The Taylor Spring Branch sample analytical result was flagged "R" (unusable) by the laboratory. The PCB concentrations in the Pig Pen Spring, Branam Spring, North Spring, and field blank samples were below the laboratory detection limit of 0.1 ppb. The North Spring sample analytical result was flagged "R" (unusable) by the laboratory. Table B-14 summarizes the surface water analytical results.

On July 27, 2001, Viacom collected seven post-excavation samples from five springs around Neal's Landfill (Viacom 2001e). PCBs were detected in five of the seven samples: the South Spring sample (1.6 ppb), the South Spring duplicate sample (1.5 ppb), the North Spring sample (0.89 ppb), the Branam Spring sample (0.13 ppb), and the Taylor Spring Branch sample (0.20 ppb). The Branam Spring sample analytical result was flagged "I" (estimated) by the laboratory. The PCB concentrations in the Pig Pen Spring and field blank samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On August 6 and 7, 2001, Tetra Tech collected surface water samples from Viacom historical sampling location 1 in Conard's Branch and four IDEM and Viacom historical sampling locations in Richland Creek downstream of the landfill. The PCB concentrations in the Conard's Branch samples ranged from BDL to 0.26 ppb. The PCB concentrations in all the samples collected from Richland Creek were below the laboratory detection limit of 0.1 ppb.

On August 16, 2001, Viacom collected seven post-excavation samples from five springs around Neal's Landfill (Viacom 2001e). PCBs were detected in three of the seven samples: the South Spring sample (1.4 ppb), the South Spring duplicate sample (1.5 ppb), and the North Spring sample (0.79 ppb). The PCB concentrations in the Pig Pen Spring and Branam Spring, and field blank samples were below the laboratory detection limit of 0.1 ppb. Taylor Spring Branch was not sampled on August 16 because it was dry. Table B-14 summarizes the surface water analytical results.

On September 14, 2001, Viacom collected seven post~xcavation samples from five springs around Neal's Landfill (Viacom 2001g). PCBs were detected in five of the seven samples: the South Spring sample (1.6 ppb), the South Spring duplicate sample (2.0 ppb), the North Spring sample (1.1 ppb), the Branam Spring sample (0.15 ppb), and the Taylor Spring Branch sample (0.20 ppb). The Branam Spring and Taylor Spring Branch sample analytical results were flagged "I" (estimated) by the laboratory. The PCB concentrations in the Pig Pen Spring and field blank samples were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

On October 19, 2001, Viacom collected seven post-excavation samples from five springs around Neal's Landfill (Viacom 2002). PCBs were detected in three of the seven samples: the South Spring sample (1.4 ppb), the South Spring duplicate sample (1.3 ppb), and the North Spring sample (0.57 ppb). The PCB concentrations in the Pig Pen Spring, Branam Spring, Taylor Spring Branch, and field blank were below the laboratory detection limit of 0.1 ppb. Table B-14 summarizes the surface water analytical results.

Tetra Tech has plotted Viacom's continuous monitoring data for surface water sampling at South and North Springs; data trends are shown in Figures A-18 and A-19, respectively. PCB concentrations in the springs have decreased somewhat since the completion of the RA at the site in December 1999; however, PCB concentrations continue to range from 0.7 to 2.4 ppb in South Spring and 0.2 to 1.5 ppb in North Spring.

Surface water monitoring during storm events is discussed in Section 4.6.

4.3 SEDIMENT

USFWS collected sediment samples from Richland Creek in September 1982. Of the 18 sediment samples collected, 13 contained detectable concentrations of PCBs ranging from 0.09 to 37.6 ppm. The sediment analytical results are presented in Attachment 8.

Westinghouse and O'Brien & Gere Engineers, Inc., performed stream sediment sampling in January and June 1983. Analytical results for sediment samples collected from the unnamed branch near Southwest Seep and Conard's Branch in January and June 1983 are presented in Attachment 9. Of the 45 sediment samples collected from the unnamed branch, 29 contained detectable concentrations of PCBs ranging from 1.6 to 20.6 ppm. Of the 231 sediment samples collected from Conard's Branch, 212 contained detectable concentrations of PCBs ranging from 2.3 to 2,535 ppm.

O'Brien & Gere Engineers, Inc., collected sediment samples for Westinghouse in April and May 1984. Of the 91 sediment samples collected from the Neal's Landfill area, 33 contained detectable concentrations of PCBs ranging from 1.2 to 411.5 ppm. The sediment analytical results are presented in Attachment 10.

In December 1984, EPA collected sediment samples from Richland Creek. Of the five samples collected, two contained detectable concentrations of PCBs. One of these two samples was collected at State Road (SR) 48 near Richland Church (0.71 ppm), and the other was collected at SR 43 south of Whitehall Cemetery (0.12 ppm). Attachment 11 contains the sediment sample analytical results.

On November 23, 1988, Westinghouse collected 20 sediment samples from two locations in Richland Creek following sediment and bank removal from Conard's Branch and Richland Creek in 1987. The sampling locations are presented in Attachment 12. All the PCB concentrations in the sediment samples were below the laboratory detection limit of 1 ppm (CBS l999d).

On October 26, 1992, IDEM and Westinghouse collected split sediment samples from South Spring, North Spring, Conard's Branch, and Richland Creek. Of the split samples that IDEM collected, 15 contained detectable concentrations of PCBs. The samples from South Spring and North Spring contained 7.7 and 0.23 ppm PCBs, respectively. Samples from Conard's Branch contained detectable concentrations of PCBs ranging from 0.51 to 1.8 ppm. Samples from Richland Creek contained detectable concentrations of PCBs ranging from 0.38 to 0.6 ppm. Of the split sediment samples that Westinghouse collected from South Spring, North Spring, Conard's Branch, and Richland Creek on October 26, 10 contained detectable concentrations of PCBs. The sample from South Spring contained 7.1 ppm PCBs; the PCB concentration in the sample from North Spring was below the laboratory detection limit of 1 ppm. PCB concentrations in three samples from Conard's Branch were below the laboratory detection limit, but five samples from Conard's Branch contained detectable concentrations of PCBs ranging from 1.2 to 21.0 ppm. The PCB concentration in one sample from Richland Creek was below the laboratory detection limit, but one sample from Richland Creek contained 2.0 ppm PCBs. Attachment 13 contains IDEM's and Westinghouse's split sediment sample analytical results (CBS l999d).

Westinghouse summarized analytical results for sediment samples collected in South and North Springs from September 1983 through May 1996. Samples from South Spring contained detectable concentrations of PCBs ranging from 2.8 to 40.6 ppm. A June 1990 sample from North Spring contained 2.3 ppm PCBs. The Westinghouse summary is provided in Attachment 14.

In May 1998, CBS collected sediment samples from 26 locations along Conard's Branch, 3 locations along Richland Creek and 1 location at the Southwest Seep of the unnamed branch. The sediment sampling locations are shown in Figure A-17. All the sediment samples contained detectable concentrations of PCBs ranging from 0.1 to 48 ppm (CBS 1998b). Table B- 13 summarizes the sediment sample analytical results.

On April 12, 2001, Tetra Tech collected sediment samples from two locations in Cave Creek and one location at Richland Springs. These samples were collected to evaluate potential PCB discharges to me Northwest Spring System drainage from suspected overflows of the Cave Creek system. The PCB concentrations in all three samples were below me laboratory detection limit of 0.05 ppm.

On August 6 and 7, 2001, Tetra Tech collected sediment samples from historical IDEM and Viacom sampling locations near the site. Tetra Tech sampled one location in Conard's Branch and four locations in Richland Creek downstream of the site. Composite samples were collected from depths up to 12 inches below te sediment surface at the center and both banks of Conard's Branch and Richland Creek. The PCB concentrations detected ranged from 0.10 to 0.26 ppm in Conard's Branch and 0.071 to 0.11 ppm in Richland Creek.

4.4 FISH

IDEM collected fish samples from various Richland Creek locations around Neal's Landfill in 1981, 1987, and 1993. Of te 26 fish samples collected, 24 contained detectable concentrations of PCBs ranging from 0.076 to 7.448 ppm. A summary of the fish analytical data is presented in Attachment 15.

USFWS collected fish samples in Richland Creek in September 1982. Of the numerous fish samples collected, many contained detectable concentrations of PCBs ranging from 0.015 to 279 ppm (see Attachment 16). In addition, USFWS collected two fish samples in Richland Creek downstream from Neal's Landfill in April 1991. The two samples contained PCB concentrations of 1.25 and 1.95 ppm (see Attachment 17).

In May 1998, CBS collected fish samples in Conard's Branch and Richland Creek for total PCB analysis. Table B-15 summarizes me fish analytical data. All the fish samples contained detectable concentrations of PCBs ranging from 0.03 to 25 ppm (CBS 1998a). Figure A-20 illustrates the historical PCB concentrations in fish samples collected at different distances (0.25, 1.0, and 5.5 miles) from south Spring, which lies northwest of the site. The fish samples collected 0.25 mile from South Spring were taken at the midpoint of Conard's Branch (location 1). The fish samples collected 1.0 mile (location 2) and 5.5 mile (location 3) from south Spring were taken at two points in Richland Creek. Concentrations of PCBs in fish samples collected from Conard's Branch correspond to the left ordinate of Figure A-20, and concentrations of PCBs in fish samples collected from Richland Creek correspond to the right ordinate. In addition, Figure A-21 plots the PCB analytical results for CBS's May 1998 fish samples by sampling location.

From August 7 through Lockheed Martin (a contractor for EPA's Environmental Response Team Center [ERTC] under me Response, Engineering, and Analytical Contract [REAC]) collected fish samples from six locations in Conard's Branch (location 1) and Richland Creek (locations 2 through 6) downstream of Neal's Landfill. The sampling locations included Tree historical CBS sampling locations (locations 1, 2, and 4) and Tree historical IDEM annual monitoring locations (locations 3, 5, and 6). A total of 87 fish (consisting of various species selected according to trophic-level feeding habits) were collected; 55 whole-body samples and 18 fillet composites were analyzed for PCBs. The remaining fish were archived at me REAC laboratory. Analytical data for me fish samples are presented in Lockheed Martin's technical memorandum for the sampling event (see Attachment 18). PCB Aroclor 1248 concentrations (wet weight) ranged from 5.9 to 9.0 ppm at location 1 in Conard's Branch, which was me sampling location nearest the site. In Richland Creek PCB Aroclor 1248 concentrations (wet weight) ranged from 0.32 to 3.6 ppm at location 2; 0.21 to 0.23 ppm at location 3; nondetect to 0.39 ppm at location 4; and nondetect to 0.24 ppm at location 5 (Lockheed Martin 2001). Figure A-22 plots the PCB analytical results for me August 2001 fish samples by sampling location.

4.5 STF INFLUENT AND EFFLUENT

Viacom has performed monthly influent and effluent surface water sampling at me STF since January 1989. NPDES data collected for me treatment facility influent from February 1990 to October 2001 show detectable PCB concentrations ranging from 0.2 to 8.7 ppb, and NPDES data collected for the treatment facility effluent during this period show PCB concentrations ranging from BDL (less than 0.1 ppb) to 0.85 ppb. Most of the effluent data are nondetects for PCBs. PCB data for me STF influent and effluent are presented in detail in Table B-16 (Viacom 2001f).

Tetra Tech collected two STF influent samples and two STF effluent samples on November 19, 1998, during low-flow conditions. Table B-17 summarizes me high-resolution gas chromatograph/high- resolution mass spectrometer (HRGC/HRMS) analytical results for the samples, which were analyzed by Wright State University in Dayton, Ohio (Wright State University 1998). The two influent samples had detectable PCB concentrations of 1,362 and 1,397 ppt. The two effluent samples had detectable PCB concentrations of 14.4 and 16.2 ppt. In addition, two Laboratory blanks contained PCB concentrations of 1.91 and 1.93 ppt. Based on mese results, the STF reduced me PCB concentrations in the influent to the ppt range before me water was discharged to Conard's Branch.

4.6.1 Storm Event Flow and PCB Discharge Relationships

4.6.2 1995 and 1998 Storm Events

Prior to me site RA, Viacom collected samples for PCB analysis during two storm events, one in May 1995 and me second in April 1998. The May 1995 sampling occurred across two closely-spaced storm events. Sampling was conducted at scum Spring. The first storm pulse on May 17 produced a peak flow of about 1,900 8pm. A peak PCB concentration of 10 ug/L occurred at South Spring 1 hour prior to me peak discharge. The second storm pulse on May 18 produced a much larger peak flow of 21,500 gpm, and a peak PCB concentration of 4.5 ug/L occurred 2 hours prior to me storm peak. A total of 18 samples were analyzed for PCBs.

The April 1998 storm event produced a peak discharge of about 9,000 8pm, and was a comprehensive storm sampling event at Conard's Branch. A total of 48 samples were collected for PCB, TSS, and specific conductance determination over a 94 hour period. A peak PCB concentration of 5.2 1lg/L occurred 5 hours prior to me flow peak. A graph showing me storm flow and the PCB mass discharged during me April 1998 storm is presented in Figure A-23.

4.6.3 2000 Storm Events

Five storm events were monitored in April, May, June, October, and November 2000. A hydrograph and PCB chemograph for me April 2000 storm is presented in Figure A-24. The flow peaked at about 10,500 gpm, and me peak PCB concentration was over 30 ug/L. The peak PCB concentration occurred on the rising limb of the storm hydrograph, about 2.5 hours prior to the storm peak. No flow data are available for the May storm; therefore, it is not included in Table B-18. A total of 12 samples were collected from the Conard's Branch gage site, and the peak PCB concentration was 3.4 ug/L.

The June 2000 event was a small storm. Peak flow a Conard's Branch during His storm was about 520 gpm. The peak PCB concentration at Conard's Branch was 4.0 ug/L, occurring about 1 hour before the storm peak.

The October 5, 2000, storm was the largest storm event measured since the source removal RA was completed. USGS Gages 012 and 34 were bypassed by floodwaters during this storm, and therefore peak flow measurements are not available from me USGS record. However, me Conard's Branch gage recorded a peak flow of approximately 17,000 gpm. Sampling during this storm was not initiated until 9 hours after the storm peak. Thus, the highest PCB levels were probably not measured. The initial PCB concentration of me first Conard's Branch sample was 13 ug/L. All succeeding recession limb samples were in the range of 0.83 to 2.1 ug/L, suggesting that the first sample result may be anamolous. The November 2000 storm was smaller, with a peak flow of 2,200 gpm and a peak PCB concentration of 2.1 ug/L. The impact of the site RA in reducing PCB discharges from the site was demonstrated during the November storm.

Figure A-25 shows data gathered during a storm event on February 25, 2001. The figure shows the rainfall amount, the PCB concentrations and specific conductance measured at Gage CB, the discharge at Gage CB, and the discharge at USGS Gages 012 and 34. Also shown is the total discharge from the two USGS gages, which should approximately equal the flow reported at Gage CB. PCB sampling occurred at hourly intervals across the storm peak, thereby providing good sample coverage for this storm event. This storm event had reasonable flow estimates by both Viacom and the EPA, and hourly PCB samples over a 48-hour period. It is one of the best post-remediation storm events monitored at Conard's Branch. This storm was chosen for examination because a large amount PCB data were gathered, the rising limb of the storm hydrograph was sampled, and a particularly prominent, and unusually high, PCB peak of 12 1lg/l occurred on the rising limb. The PCB peak, represented by a single hourly sample at Conard's Branch occurs in conjunction with a conductivity peak. The conductivity peak occurs after surface water runoff in the Conard's Branch channel ceases, and prior to the arrival of the surge of storm flow through the conduit system. This situation has been observed during other sampling events, and may represent long-residence-time groundwater containing elevated PCB levels that is flushed from the karst drainage system by storm water very early in a storm event. PCBs appear at maximum concentration very early in a storm event due to the extreme downstream position of the landfill within the ground water basin.

Figure A-25 illustrates the following features of the Northwest Spring System storm-flow hydrology and PCB discharge:

• Overflow springs respond rapidly to rainfall and reach peak flows within 3 to 4 hours of the onset of a precipitation event. The small flow peak and corresponding large conductivity dip are indicative of surface water runoff from a small area of the landfill that discharges directly to Conard's Branch upstream from Gage 012.
• Overflows 0, 1, and 2 flow for a longa period than Overflows 3 and 4.
• Most of the storm flow emages at Overflows 0, 1, and 2 during a storm flow on the order of 5,000 gpm.
• Pronounced spikes in PCB concentrations occur early in the rising limb of the hydrograph during a storm, perhaps even with the first discharges from the overflow springs. During the February 25, 2001, storm event, the spike coincided with an apparant peak in specific conductance. The apparent conductivity peak occurs after surface wata runoff in the Conard's Branch channel ceases, and prior to the arrival of the storm-flow surge through the conduit system. This situation has been observed during subsequent sampling events and may represent long-residence time groundwater containing elevated PCB concentrations being flushed from the karst drainage system by storm wata very early in a storm event. The PCB discharge achieves its maximum concentration early in a storm event because of the extreme downstream position of the landfill within the Groundwater basin.
• PCB concentrations drop dramatically during the recession of a storm as low-specific conductance, short-residence time Groundwater arrives at the spring emergences from the outa portions of the Groundwater basin. The spring discharge at this time may include large volumes of point-source recharge from sinking stream drainage areas.
• As specific conductance values begin to increase several hours after a peak flow, a second spike in PCB concentrations occurs. This second spike is smaller than the first. PCB concentrations than decline to a consistent level during the latter part of a storm.

From June 4 through 6, 2001, storm event sampling was conducted by Tetra Tech, USGS, and Viacom at the overflow spring gages and Gage CB. On June 4, 2001, a storm event with a peak flow of 7,640 gpm was monitored by collecting samples for PCB analysis at 1-hour intervals using autosampla equipment. On June 5, 2001, a second peak flow of 3,060 gpm occurred and was also monitored. Figure A-26 plots the storm flow, PCB concentration, and specific conductance values.

Figure A-26 illustrates additional features of the Northwest Spring Systam storm-flow hydrology and PCB discharge:

• Specific conductance values at Gages 0, 012 and 34 are consistent, indicating a common conduit flow source.
• PCBs are discharged from all overflow springs unda all flow conditions. The highest values are ganaally associated with the rising limb of a storm flow hydrograph, but PCB concentrations do not track as consistently as specific conductance values.

On July 4, 2001, storm event sampling was conducted by Tetra Tech at the overflow spring gages. A peak flow of about 7,640 gpm was monitored by collecting samples for PCB analysis at 30-minute intervals using autosampler equipment. Figure A-27 plots the storm flow and PCB concentrations measured. A specific conductance chemograph is also shown in this figure. Peak PCB concentrations of 3.2 to 4.2 1lg/L wae observed on the rising limb of the storm hydrograph. Vay similar PCB and specific conductance bands wae observed at Gage 0, 012, and 34.

A storm event occurring from October 23 through 25, 2001, was monitored by Viacom. Figure A-15 plots the storm flow and PCB concentrations measured. Two storm peaks occurred on October 24, 2001. The first peak produced a peak runoff of 11,000 gpm, but was not sampled. Hourly PCB samples were collected over the peak of the second storm that occurred lata in the day. This storm produced a smaller peak flow of approximately 7,200 gpm. The peak PCB concentration for the second stolen was 3.1 1lg/L, which occurred about 1 hour before the storm peak.

4.6.5 Estimates PCB Mass Discharged

This section discusses three analyses of the relationship of the Groundwater discharge rate of the Northwest Spring System to PCB mass discharged. The first two analyses were performed using data obtained prior to the 1999 site RA, and are presented to provide an historical background. The last analysis compares pre-RA and post-AA storm flow data, as outlined above.

4.6.5.1 1987 Westinghouse Analysis

The report titled "Analysis of PCB-Flow Data, Neal's Landfill Spring Water" was prepared for Westinghouse in 1987 (Westinghouse 1987). The data and analysis presented in that report were used as justification for a 1 cfs/1 ppb treatmant standard for PCB-contaminated Groundwater Emerging from the Northwest Spring Systam during NPDES permit negotiations for the STF. On page 4 of the report, a regression model is presented that relates the South Spring PCB concentrations to the mean daily discharge at Conard's Branch. This model was developed using 15 PCB discharge data points from December 1982 through Septamba 1983 (see Section 4.1). The maximum recorded Conard's Branch discharge flow rate was 2.3 cfs or about 1,000 gpm. However, because PCBs ware not detected in either of two samples collected at the discharge point, both sample results were excluded from the regression analysis. The regression equation used is as follows:

where
    PCB = Total PCB concentration in ug/L      Q = Discharge (cts)

The regression equation was used to estimate a PCB concentration of 0.169 ug/L at a flow rate of 1.0 cfs (see Table B- 19, column 3). At slightly higher flow rates, the estimated PCB concentration is less than the analytical reporting limit of 0.1 ug/L. These data wae used to support the contention that a 1cfs /lppb PCB treatment standard was appropriate and that further storm event sampling proposed in the draft NPDES permit was unnecessary.

PELA, an EPA contractor, performed monthly sampling at the Northwest Spring Systam emergences from December 1982 to October 1983. Sampling locations included South Spring, North Spring, and during periods of high flow, some of the overflow springs. Sampling was conducted during one high-flow (greatathan1 cfs) event on April 11 and 12, 1983. The USGS flow data being recorded at this time indicate that the Northwest Spring System was in the recession limb of a large storm event that peaked at over 30 cfs on April 9. Flows remained high on April 11, whan the mean daily discharge from the USGS South Flume was 2.30 cfs. PCBs wae not detected in the South Spring sample collected on April 11,1983. However, because of the high-flow conditions, PELA collected additional samples directly from Overflow Springs 1 and 3 containing PCB concentrations of 0.76 and 2.29 ug/L, respectively. In addition, a sample collected from North Spring on April 11 contained 1.25 ug/L PCBs. On April 12, an additional sample was collected at Overflow 2, and its PCB concentration was 0.9 ug/L. The mean daily discharge at the South Flume on April 12 was 1.7 cfs.

The analytical results for the four samples collected from the Northwest Spring System on April 11 and 12, 1983, contradict the South Spring sample result and indicate that during high-flow periods, PCBs were discharged from the Northwest Spring Systam. Yet these results were not considered in the 1987 Westinghouse analysis. One explanation given in Section 3.0 of the 1987 report is as follows: "As well as can be determined, only South Spring PCB data represents PCB concentrations in the actual spring wata at the time the flow measurements wae made. The South Spring overflow data represent the PCB concentrations in the pool behind the sediment fence, not necessarily of the same concentration as that in the spring at a given time." The 1987 report goes on to state that "while the data from North Spring may be representative of the actual spring data, the corresponding North flume flow data includes flow from the South Spring. Consequantly, PCB-to-flow correlations are not possible at North Spring." Subsequant sampling of the Northwest Spring Systam to the present day has indicated that storm water flows carry the highest PCB concentrations, and that the 1987 assumption that the PCB content is negligible at flow rates above 1 cfs is erroneous.

4.6.5.2 1998 CBS Analysis

Based on this equation, predicted PCB concentrations at various flow rates up to I cfs are presented in Table B-19, column 4.

The second correlation was developed using data gathered during an April 1998 storm event. The regression equation used, which is based on 28 data points, is as follows:

Based on this equation, predicted PCB concentrations at various flow rates up to 1 cfs are presented in Table B-19, column 5.

Figure A-28 shows the predicted PCB concentration and flow relationships for Conard's Branch based on the various CBS regression models discussed above. Based on CBS's 1998 analysis, the 1987 analysis conducted by Westinghouse underestimated the PCB concentrations in Groundwater Emerging from the Northwest Spring System.

4.6.5.3 Postremediation Analysis

Sampling has been conducted during several storm events since the site RA was completed in 1999. Figure A-29 shows data from the storm hydrograph recession limb for six postremediation storm events that occurred from June 2000 to October 2001, and for which flow data records are available. Comparison of recession limb PCB data from storm to storm is one useful measure of identifying bands in PCB discharges from the site. Historically, the recession limb data show more consistent PCB levels than the rising limb data and are indicative of the largest proportion of the spring flow. This does not state that the rising limb data are not important. Genaally, the highest PCB concentrations are found on the rising limb, but there are few storms for which good rising limb data sets have bean gathered. For comparison, data for the praemediation, April 1998 storm event are also shown in Figure A-29.

Table B-18 presents a tabulation of all of these data. The April 1998 data are shown in blue; storm data from 2000 are shown in black and 2001 data are shown in red.

Based on the best available flow data, the relationship betwean PCB concentrations and flow for the 208 postramediation recession limb data points shown in Figure A-29 can be approximated using the following equation:

Most postramediation storm event PCB values range from 0.5 to 2 ug/L. The PCB concentrations are remarkably uniform over a wide range of storm flow rates. PCB concentration spikes of over 6 ug/L occur in the recession-limb data.

Figure A-29 suggests that the PCB ramoval and landfill capping completed at the site in 1999 have impacted PCB concentrations in Conard's Branch. The recession-limb data for storms in 2000 and 2001 ganaally exhibit Iowa PCB concentrations than the April 1998 recession-limb data. This band is particularly evidant for flow rates below about 2,000 gpm (see Figure A-29); however, there is no apparent difference between the 2000 and 2001 storm data. More storm event data are needed to evaluate long-tam bands in PCB data.

In 1998, Tetra Tech estimated the PCB mass discharged from Neal's Landfill via Groundwater to North Spring, South Spring, and Conard's Branch in 1993 and 1994 and in April 1998. The mass PCB loading rate was estimated by multiplying the spring flow rates by the corresponding PCB concentrations. The peak PCB mass estimates for all peak spring flows following the rainfall events in 1993 and 1994 and an April 1998 storm event wae calculated using correlations developed by Viacom.

Tetra Tech also estimated the PCB mass that bypassed the STF in 1993 and 1994 and in April 1998. According to Viacom, the STF operates at a design flow rate of 450 gpm. To calculate the PCB mass captured by the STF, Tetra Tech estimated the PCB concentration corresponding to the STF's design flow rate of 450 gpm using a correlation provided by Viacom. To estimate the PCB mass that bypassed the STF, the PCB mass captured by the STF was subtracted from the total PCB mass discharged. Tetra Tech's estimates of the total PCB mass discharged, the PCB mass that bypassed the STF, and the PCB mass removed by the STF during the two study periods are summarized in earlier reports (Tetra Tech 1998a, 1998b). Table B-20 shows the estimated PCB mass discharged to Conard's Branch in 1993 and 1994 and in April 1998.

In February 2001, Tetra Tech used hydrograph information collected at Neal's Landfill by USGS from June through December 2000 to calculate the mass of PCBs released that bypassed the STF during storm events. Fourteen monitored storm events occurred during this paiod. Tetra Tech used an equation based on data collected during storm events in 1998 and 2000 to calculate the concentration of PCBs released during each hour of the storm events. Using this concentration and the volume of flow during the time period of each storm event, the PCB mass released was calculated using the following two equations:

PCB concentration (ug/L) =-2.22213 x 10~5 x flow rate (gpm) + 1.502

PCB mass released (gram [g] pa hour) = PCB concentration (~g/L) x [1 g / 1 x 106 micrograms ù(,ug)] x [1 liter (L) / 0.2642 gallon] x [60 minutes / hour] x flow rate (gpm), or in shortened form without units, (PCB concentration / 264,200) x 60 x flow rate

p>

The total PCB mass released during the storm events was estimated to be approximately 261.18 g (Tetra Tech 2001a). This amount of PCBs was distributed throughout Conard's Branch and its confluence with Richland Creek downstream of the Neal's Landfill site. This amount is comparable to the 170 g of PCBs that Viacom estimates was released during the April 7, 2000, storm event (Viacom 2001d).

In November 2001, Tetra Tech evaluated data for storm flows that occurred from February through July 2001 (see Figures A-25, A-26, and A-27). These data and the two equations presented above were used to calculate the PCB mass released during this paiod (Tetra Tech 2001b). Tetra Tech obtained hydrograph information collected at the Neal's Landfill site by USGS from February through July 2001 and calculated the mass of PCBs released during storm events. To make the USGS data manageable, Tetra Tech analyzed the hydrograph information to identify six storm events generally producing flow rates greata than 50 gpm. Tetra Tech used USGS's 60-minute flow data increment as the flow volume for each hour of the storm events. The total PCB mass released during all the storm events was calculated to be 110.92g (Tetra Tech 2001b).

Viacom (2001f)presented cumulative PCB mass curves for a few monitored storm events, including the February 25, 2001, storm event. EPA believes this type of analysis has mait in assessing long-term trends in PCB storm discharges. Viacom (Figure 28 of Viacom 2001f) indicates that the total PCB mass released during the February 25 storm event was about 19 grams. EPA indepandantly performed a calculation using the same flow and PCB data and daived a mass release estimate of 18.1 grams. This analysis used the average values from the flow hydrograph and PCB chemograph for the storm event.

Using the recession limb PCB flow regression, Tetra Tech calculated the PCB mass release for the February 25, 2001, storm at 11.7 grams, or about 65 percent of the value calculated using the averaging procedure. The difference is attributable to the flow regression model not considering the prominent, though short-term PCB spike on the rising limb of the storm. Tetra Tech regards the mass release estimates using PCB-flow regression as conservative, but not unreasonable, givan the relative lack of more definitive data at this point in time to make a more informed PCB mass release model. Only a few post-ramediation storm monitoring events have bean conducted at Neal's Landfill to date, and rising limb storm data are not available for some of these storms. However, Tetra Tech believes the current simple flow regression analysis provides a reasonable, conservative estimate of PCB mass discharged from the Neal's Landfill site.

Table B-20 shows the estimated PCB mass discharged from June 2000 through July 2001. This amount of PCBs was distributed throughout Conard's Branch and its confluence with Richland Creek downstream of the Neal's Landfill site.

5.0 SUMMARY OF FINDINGS

The Neal's Landfill site lies in karst tarain. Groundwater movement is characterized by rapid flow through open conduit systems. Dye track testing has demonstrated that these conduit systams discharge at springs northwest and southwest of the site. The southwestern springs, known as Taylor and Branam Springs, drain into an unnamed tributary of Richland Creek. The tracer test results suggest that a small portion of the southeastern part of the site drains to these springs. The Northwest Spring System at the headwater of Conard's Branch consists of the perennial South Spring, North Spring (a related underflow spring), and several overflow springs. The overflow springs are normally dry but carry large volumes of storm wata flow. Wata quality and track test data suggest that South Spring and the various overflow springs discharge from a common conduit system. The Northwest Spring System discharges to Conard's Branch, a tributary of Richland Creek. Historical Groundwater surface water analytical data indicate that PCBs have bean detected at concentrations above MCLs and AWQC, respectively, at and near the site. According to sample analytical data from May 1998, groundwater in four monitoring wells screened in fractured limestone contained PCB concentrations above the detection limit. Surface water sample analytical data from May 1998 show that PCB concentrations above the detection limit was presant in Conard's Branch. Sample analytical data from May 1998 also show that sedimant and fish in Conard's Branch and Richland Creek contained PCB concentrations above detection limits. However, July 1999 analytical data for Groundwater samples collected from residential wells in the vicinity of Neal's Landfill during low-flow conditions indicate that no PCBs were present in the wells sampled.

Current hydrogeologic and chemical data indicate that Groundwater has come in contact with PCBs and discharges to the surface through various springs. Detections of PCBs in Taylor and Branam Springs are sporadic. PCBs are consistently detected in the discharges from the Northwest Spring System. Since 1989, low-flow discharge from the Northwest Spring System has been routed to the STF, where PCB concentrations are treated to meet the 1-ppb discharge criterion. Flows exceeding the STF's design capacity of 1 cfs (449 gpm) are discharged directly to Conard's Branch from South Spring. Data collected by USGS in 2000 and 2001 indicate that during storm events, flows of several thousand gpm bypass the STF and discharge directly to Conard's Branch. During the period of June 7, 2000, to November 13, 2000, at least a dozen storm events with peak flows in excess of 5,000 gpm discharged directly to Conard's Branch. Total flow bypassing the plant during this period is estimated to be in excess of 124 million gallons.

In 1999, Viacom completed excavation of PCB hot spots, consolidation of non-hot spot wastes in an area of the site that is not prone to backflooding, and placement of a RCRA Subtitle C cap over the consolidation area. In addition, a storm water diversion system was constructed to prevent water infiltration and to control runon and runoff at the site. These remedial activities are expected to gradually reduce the mass of PCBs discharged to the Northwest Spring System, Conard's Branch, and Richland Creek; however, low concentrations of PCBs persist in these areas. Groundwater monitoring well and spring surface water analytical data for samples collected after the source control RA was completed in November 1999 indicate the presence of PCBs above MCLs and AWQCs, respectively. The discharge of PCB-contaminated groundwater from the Northwest Spring System to Conard's Branch is the principal avenue by which PCBs are released to the aquatic environment.

Storm event sampling subsequent to the site RA in 1999 has shown that the Northwest Spring Systam flows typically contain 0.5 to 2 1lg/L PCBs. Much higha PCB levels may occur for a short period of time early on the rising limb of a storm. An instantaneous peak of about 12 ppb was observed early on the rising limb of a storm in February 2001. The appearance of the maximum PCB concentrations very early in a storm event is related to the location of the landfill in the extreme downstream portion of the karst drainage network.

Although storm wata flows appear to contain lower PCB levels than they did in 1998, a comparison of recession limb storm data from six storms does not indicate a significant difference in PCB levels from 2000 to 2001.

The springs in the vicinity of the site, Conard's Branch, and Richland Creek must be the focus of long- term Groundwater monitoring of the site. Monitoring will need to be conducted until PCB concentrations in surface water, sedimant, and fish consistently meet applicable regulatory criteria.

The mass of PCBs released to Conard's Branch during postramediation storm events from April 2000 to July 2001, is estimated to be 372.1 grams. This PCB mass is deposited in sediment in both Conard's Branch and Richland Creek.

The sample analytical data for the August 2001 sediment sampling event conducted by Tetra Tech indicate that PCB releases have occurred in Conard's Branch since the site RA was completed in 1999. In addition, fish sampling was conducted by EPA's ERTC contractor, Lockheed Martin, in August 2001. The sample analytical results indicate the presence of PCBs in all fish species sampled.