1.0 Introduction

The obligation for Viacom to remediate Lemon Lane (LL) Landfill was established in a 1985 Consent Decree (CD) (Reference 1). The remediation was performed in 2000 following the terms of the May 2000 statement of work (SOW) (Reference 2).

Details of the remedial approaches are described in a May 2000 work plan (WP) that covered excavation, removal and consolidation of waste (Reference 3). A final report was issued in June 2001 (Reference 4) that describes the work performed in the work plan and capping of the landfill. A RCRA Cap Inspection and Maintenance Plan (Reference 52) was agreed to by the CD parties in June 2001. In addition to the removal and closure actions, Viacom is required to submit a Long-term Groundwater Monitoring Plan for review and approval by the government parties to the CD; IDEM, USEPA, Monroe County and the City of Bloomington.

Groundwater monitoring is to occur during a five-year period starting with the year that construction began. After the five-year period, a review of all monitoring results will be conducted according to CERCLA Section 121(c) and Section 300.430(f)(4)(ii) of the National Contingency Plan. The review will occur five years after construction began at LL Landfill. The remedial action began in April 2000. The end of the five-year period will be April 2005. At that time, the CD parties will review the monitoring data and decide if the monitoring will be maintained, changed or eliminated. Also, during the monitoring period, changes to the plan may be proposed and the plan changed with the consent of all the parties.

Interim Groundwater monitoring required by the SOW was initiated before the remedial action began and will continue until the Long-term Groundwater Monitoring Plan is implemented.

A separate groundwater investigation plan to further evaluate PCB transport is also being developed and will be implemented by Viacom after review by the government parties.

1.1 Site Location

LL Landfill is located on the northwest side of the City of Bloomington in Monroe County, Indiana. The landfill location is shown on the General Site Location Map in Figure 1. Figure 2 shows a plan view of the site vicinity.

The original landfill area covered approximately 10 acres. The City of Bloomington owns a majority of the landfill property. Lemon Lane Road and a residential area along LL bounds the east side of the landfill. The CSX Railroad tracks (formerly Monon RR, as shown in Figure 2) border the southern edge of the Landfill. Directly south of the railroad tracks is Valhalla Cemetery. Jerry Pelfree owns the property directly north of the site and the northern part of the landfill occupies his property. Bordering the northeast corner of the site outside the fence is the Sexton property. On the east side of Lemon Lane Road are the Bender and Elliott properties. Viacom owns the undeveloped land to the west of the landfill. The Indiana State Route 37 by-pass is approximately 900 feet northwest of the landfill entrance. The Griffin property borders the southern portion of the east fence line of LL Landfill. This property was remediated during the LL remediation under a separate Administrative Order on Consent.

These additional investigations led to the fencing of the Illinois Central Spring (ICS) emergence area and Swallow Hole area in 1995, and the installation of the Illinois Central Spring Treatment Facility by the USEPA in 1999-2000, as discussed in Section 1.2.3., and the 2000 remedial action as described in Section 1.2.4.

An interim Groundwater monitoring program was initiated just before and during remediation at the site and continues since completion of the remedial action. A karst conduit study has also been ongoing for a number of years in an attempt to find major pathways of the PCB carried in the Groundwater from the site. This study includes additional monitoring well installations, pump tests, Groundwater elevation monitoring, including the installation of continuous monitoring instrumentation, and additional Groundwater and surface water sampling and analyses.

The water and sediment data generated during all these investigations from the 1980s to October 2002 are included in the tables of this plan and discussed in detail later.

1.2.2 Interim Remedial Measures (IRMs)

• In June 1983, an 8-foot high fence was installed by EPA around the entire perimeter of the landfill.
• In 1985, the EPA installed a fence around ICS. This spring is approximately 2000 feet southeast of the landfill and the spring waters and sediments were found to contain PCBs. It was suspected that the spring was the resurgence of Groundwater impacted by the site.
• In May 1987, the site was cleared of all trees and vegetation, and exposed capacitors were removed. An interim cap was placed on the site with the upper cap component being a 36-mil Hypalon geomembrane cover. This was installed to minimize rainwater infiltration into the landfill, prevent the erosion of soils from the site and minimize any potential air emissions. This work was done by Viacom under supervision of the other CD parties.
• In 1995, additional fencing was placed around the ICS area and Swallow Hole by Viacom.
• In 1996, Viacom modified drainage from the interim cap. A berm was installed on top of the cap to direct more of the runoff to Sargent's Pond and the cap was extended to the southwest of the landfill to prevent ponded water from infiltrating at the edge of the landfill.

Maintenance of all interim remedial structures was performed by Viacom.

1.2.3 Illinois Central Spring Treatment Facility

ù Remove materials in areas defined with greater than 100 ppm PCBs to an off site TSCA landfill so that the residual PCB average in the landfill will be less than 50 ppm.
• Reduce the final footprint area to be capped to approximately six acres from the original footprint of approximately ten acres.
• Cap all remaining waste with a RCRA Subtitle C cover.
• Manage surface runoff and potential run-on into lined storm water ditches and route this clean water into Sargent's Pond. This was to mir imize Groundwater recharge near the site, potentially reducing PCB loading to the springs.

1.2.5 Interim Groundwater Monitoring Program

• ICS, QS and Slaughterhouse Spring monthly for PCBs and total suspended solids (TSS). The QS samples represent the combined flow of Quarry A, B and C.
• Residential wells at 2320 West Beaumont Lane and 111 North Kimble Street with samples to be taken at the conclusion of the source control remedial action, assuming access to the properties was permitted.
• Water in Sargent's Pond. Samples were taken prior to excavation, monthly during excavation and after the completion of the RCRA cap.

Sampling of the springs continues monthly until replaced by the long-term Groundwater monitoring program. The PCB results from the interim program are discussed in section 1.5.

1.3 Topography and Surface Water Features

The site is located on the margins of the Mitchell Plain physiographic unit, which is characterized by moderate slopes and karst topography. The site was developed in two surface depressions that are part of a compound sinkhole. The compound sinkhole runs northeast to southwest and extends to the southwest of the LL site beyond the CSX railroad tracks.

Prior to fill operations at the site, the topography sloped from east to west into the two main sinkholes. After fill operations, the topography across most of the site sloped more moderately to the west with more severe slopes on the south and west edges of the site. The current surface of the landfill ranges in elevation from 860 to 887 feet above mean sea level (amsl).

Features identified in the vicinity of the landfill that are typical of karst terrains include sinkholes, caves, sinking streams and swallow holes (Reference 5) as shown on Figure 4. These karst features were located in and around the fill area by site surveys, inspection of topographic maps, interview with residents and historic aerial photography review.

Figure 5 is a schematic view of the drainage of the area. There are no surface water streams directly associated with LL Landfill that provide drainage. However, the site is located near topographic divides between the headwaters of three local surface streams, Clear Creek to the southeast, Stout's Creek to the northwest and Griffy Creek to the northeast. In lieu of surface streams, karst features such as sinkholes and swallow holes provide the surface water drainage in the immediate site area. These features route surface runoff and infiltration to the Groundwater system which in turn feed springs. The springs form the headwaters of the perennial streams.

A sinkhole pond, Sargent's Pond, is located northwest of the site. Sargent's Pond serves as the natural surface water drain for a majority of the site and the surrounding area. During the 2000 remedial action, a system of surface drainage ditches was installed around the landfill. These ditches take water from the capped area and from higher ground outside the capped area and direct it to Sargent's Pond. This pond has existed at the site for some time. In the eairly 1990s the pond reportedly drained overnight. The pond was repaired by digging out the bottom of the sink and plugging the bedrock opening with concrete.

Major perennial springs or spring systems, identified on Figure 5, are ICS, the QS, the Slaughterhouse Spring System (consisting of Slaughterhouse, Packinghouse Road, and Packinghouse Culvert Springs), Stony East Spring, and Stony West Springs. The other springs noted on Figure 5 are smaller and intermittent.

1.4 Geology and Hydrogeology Summary

A great deal of effort has been expended since 1982 to characterize and detail the geology, hydrogeology, and contaminant fate and transport for the LL Landfill and the immediate surrounding area. Section 1.4.1 briefly summarizes the extent of the investigations. The details of the various investigations are found in the reports cited.

Section 1.4.2 summarizes the results of the investigations including the stratigraphy and hydrogeology. A detailed review of the contaminant data in groundwater/surface water, including fate and transport, is in section 1.5.

1.4.1 Scope of Previous Investigations

Hydrogeologic investigations at the site began as early as 1982 and have continued to the present. A broad spectrum of investigation techniques have been used to study the site including:

• Borings and Gorings to delineate site stratigraphy
• Borehole and surface geophysics
• Monitoring well development, water level measuremet and sampling
• Aquifer hydraulics testing using slug tests and pump tests
ù
• Spring flow measurement and storm flow hydrograph analysis
• Spring flow contaminant sampling during non-storm and storm conditions
• Basin delineation and flow analysis using dye tests and modeling
• Site and area field surveys and historical photo reviews to identify key surface features related to karst development and structure

A chronological listing of all hydrogeological investigations is presented on Table 1. A brief summary of the investigations follows.

1.4.1.1 Borings, Corings and Wells

Bore holes and cores were drilled to determine the site stratigraphy. All of the holes were subsequently developed into monitoring wells of varying construction. While a few of the monitoring wells have been abandoned, most of them remain. A total of 72 wells have been installed over the years. A complete listing of all the wells and their characteristics is presented on Table 2. The locations of the wells are shown on Figure 6. The locations of wells which have been decommissioned are indicated on Figure 6 and on Table 2.

The EPA drilled four core-holes, which subsequently became monitoring wells MW-B1 through MW-B4 in July and August 1982. The wells were located by assuming that the regional dip (known to be to the southwest) controlled shallow Groundwater flow. Thus, one well, MW-B4, was located to be up gradient of the site. The other three wells were located in the assumed down gradient direction. Detailed core descriptions were prepared by Powell (Reference 5).

Thirteen monitoring wells, MW-1 thru MW-9, were installed by Viacom in December 1982 and January 1983 using air rotary techniques. Multiple wells were installed at the locations for MW-1, MW4, and MW-8. The multiple wells are indicated as deep (D), intermittent (I) or shallow (S). The basis for locating the wells included a fracture trace analyses performed by Krothe (Reference 6) and a desire to co-locate wells near the EPA wells previously installed. Caliper and gamma ray logging as well as slug tests were also conducted in the boreholes. The detailed data can be found in Reference 7.

In August and September 1987, Viacom installed five additional wells, MW-10 thru MW-14, around the perimeter of the landfill. Two of the wells were cored, C-1 (which later became MW-10) and C-2 (which later became MW-12). The others were drilled by air rotary equipment. All five wells were caliper logged and gamma-ray logged (Reference 8). The wells were located near existing wells to investigate different Groundwater flow zones discovered during the construction of the earlier wells. MW-10, MW-11 and MW-13 were installed to monitor the upper portion of the Groundwater system at elevations of 795 feet amsl or above. MW-12 and MW-14 were completed to monitor the lower portion of the Groundwater system at elevations of 785 feet amsl or below. MW-10 and MW-11 were used during the October 1987 low flow tracer test as an injection point and a sampling point, respectively. Detailed boring logs can be found in Reference 8.

Between July and October of 1998 seventeen (17) exploratory borings were drilled in the Valhalla Cemetery south of the site by Viacom (Reference 9). The Valhalla area was considered a likely location for conduits carrying venter the site because the cemetery is located between the ICS, which was shown to be the major discharge for site impacted waters, and the landfill. The wells were located based on a natural potential geophysical survey conducted in April 1998. These wells have designations such as "NN-300" which are derived from the grid line designation and distance from an arbitrary zero point along the line of the natural potential survey. All the borings were drilled by air rotary methods and, with the exception of NN-412, were completed to at least the 795 feet amsl elevation.

In December of 1998, nine piezometers were drilled in the northeast corner of the landfill. These piezometers were installed within the shallow overburden to determine if saturated conditions existed along the perimeter of the capped area. The northeast corner of the site was chosen based on a shallow soil temperature survey conducted around the perimeter of the cap in 1998. A cold region in the northeast indicated the potential for lateral shallow inflow from offsite areas under the cap. Table 3 shows the survey coordinates, the elevations and the depths of the piezometers. An attempt was made to sample the piezometers on December 16, 1999. All but PZ1 were dry at the time as shown on Table 3. These piezometers were removed during the 2000 remediation.

In 1999 shallow offset borings were installed in Valhalla Cemetery within five feet of five deep (phreatic) borings and designated by the suffix of "A" at five locations, NN-12A, 00-125A, NN-300A, 00-300A and 00-587A. The driller's log and well construction logs for all the Valhalla wells are contained in Reference 10.

In May 1999, Viacom utilized resistivity imaging and spontaneous potential to identify possible karst conduits in the northern portion of Valhalla Cemetery (Reference 10). A possible subsurface fracture/conduit was identified and well 00-370 was installed.

In late 1999 resistivity imaging was performed along the southeast perimeter of the landfill. Based on this investigation, the LF series of wells were drilled in early 2000 to investigate shallow resistively & omalies. Water sampling and video logging was performed in these wells.

In late 1999, wells MS-1 and NN-700 were installed southwest of the landfill based on the fracture trace work of Parizek (Reference 11).

In the spring of 2000, shallow probing and sampling were done at the landfill to identify the top of rock. The probing was done with a direct push rod and refusal was interpreted as the top of rock. Sampling was done with a split spoon and was done both to identify the top of rock and to determine PCB levels in overburden at depth and to identify the soil/fill interface.

In late 2000, after remediation and consolidation, pairs of piezometers were installed at the landfill at the location of the deepest points of the north and south sinks. PZ-AS was installed at the deepest part of the fill-soil horizon in the southern sink and PZ-AD was installed in the deepest bedrock location in the southern sink. PZ-BS and PZ-BD were similarly installed at the corresponding depths in the northern sink. These piezometers are designed to monitor for the presence of water that may move beneath the landfill cover horizontally or rise from lower horizons during storms.

In late 2000, two additional shallow piezometers, PZ-C and PZ-D were installed on the west edge of the landfill, just outside the consolidated fill, in areas where free product was discovered during excavation. Pea gravel recovery galleries were installed at the depth of the free product. The piezometers were screened at this depth to be capable of functioning as recovery wells. These piezometers extend to approximately the top of rock at these locations.

In late 2000, additional shallow bedrock piezometers PZ-E, PZ-F, and PZ-G were drilled on the periphery of the landfill. They were located based on additional resistivity surveys in search of potential shallow conduits.

In late 2000 and early 2001, after the site remediation was completed, additional boreholes were drilled along the east side of the landfill. The specific location of each boring was determined by an additional resistivity investigation. During this period the east side of the site became more of a focus because careful water level measurements made in wells around the site indicated a consistent potentiometric low on the east side of the site. The borings became monitoring wells MW-15, MW-16, MW-17, MW-18, and MW-19.

In late 2001, more resistivity imaging and cross-hole seismic analysis were used to locate potential conduits near the MW-4 wells on the east side of the site. This specific area was of interest because the water levels in wells MW-41 ar,d MW-4S are consistently lower than the others around the site and high levels of PCBs were found in waters withdrawn from well 41 during short term pump tests. Wells MW-20 and MW-21 were drilled in September 2001 in the MW-41 area.

1.4.1.2 Geophysics

A number of geophysical techniques have been used at the site over the years including:

• Borehole geophysics
  • Caliper logging
  • Gamma ray logging
  • e-logging
  • Videologging
  • Cross hole seismic
• Surface Geophysics
  • Magnetometer
  • Electromagnetics
  • Resistivity Imaging
  • Natural Potential
  • Fracture Trace Analysis
  • Ground Penetrating Radar

During the initial installation of the MW-1 through MW-9 series of wells by Viacom, boreholes were caliper and gamma logged. The purpose was to aid in determination of borehole size changes and Ethology changes. The data is found in Reference 7.

In 1983, Viacom performed a surface magnetometer survey at the landfill to determine the limits of buried metallic material. This was done to aid in delineating the extent of buried landfill materials. (Reference 7).

Starting in 1993, videologging of select wells has been done. The purpose of the videologging was to help identify borehole stratigraphy and structure especially the presence of fractures and voids. Videologs have been completed on the following boreholes: MW-2, MW-3, MW-41, MW-7, MW-9, MW-11, MW-12, MW-13, MW-14, NN-370, NN-300, NN-625, NN-700, NN-12, 00-370, 00-387, 00-300, 00-125, 00-587, MS-1, LF-1, LF-2, LF-3, LF-5, LF-6, LF-7, MW-15, MW-17, MW-18, and MW-19.

The videos for each well are maintained by Viacom.

In 1995, EPA performed a surface electromagnetic survey on the site to determine the location of buried magnetic materials that may indicate the location of buried capacitors (Reference 12). The survey was done to identify potential areas of high metallic fill, which may indicate areas of buried capacitors, in support of sampling plan development.

Three fracture trace surveys have been conducted at the site. Fracture trace surveys are conducted by reviewing aerial photographs of the site area and identifying natural lineal features such as valleys, vegetation lines, and lines of sinkholes/springs that may indicate a region of underlying fractured rock. Fractured rock is more susceptible to conduit formation. The first study was conducted by Krothe in 1982 (Reference 6). BB&L also conducted a study as part of the Supplemental Hydrogeologic Investigation Plan (SHIP) (Reference 8). Parizek conducted another study in 1999 (Reference 11). The fracture trace work was used to locate boreholes with the hope that these locations would coincide with karst conduits.

Natural potential and surface resistivity are geophysical techniques that were used to identify likely locations of karst conduits. The natural potential method was tried first in April 1998 and then again in September 1998 (Reference 9). The natural potential method is the mapping of the potentials of the natural electric field across the ground surface. One of the major generators of that electric current is flowing groundwater.

Resistivity imaging method (RIM) is an electrical geophysical method fundamentally applied as an imaging or mapping tool of the sub-surface based upon a number of vertical electrical soundings (VES) performed along a prescribed line (profile) at certain prescribed spacing that depends on the lateral and vertical resolution desired for the image. Therefore, RIM is an imaging or mapping technique based on a number of VESs. RIM surveys in Valhalla Cemetery began in 1999 and they were applied at the site in 1999, 2000 and 2001. (References 10,13 and 14)

Ground penetrating radar was attempted from the surface to a borehole by EPA in 1994 (Reference 15). The thought was that anomalies in the rock would be identified. The EPA team concluded that there was too much clay and that radar would not work at the site.

A geophysics professor from Indiana University (IU) attempted surface ground penetrating radar in several areas around the site in 1998 and 1999. Again, there were problems penetrating the clay soils (Reference 16).

E-logging is performed by measuring the electrical resistivity in a borehole at different elevations. This was performed in 2000 in boreholes LF-6 and LF-7 (Reference 13).

1.4.1.3 Water Level Measurements

Water level measurements of well water have been conducted extensively at wells at and around the site. Spot check measurements have been made with hand water level probes and continuous water level measurements exist for many of the wells for various periods of time. The continuous measurements were made with downhole pressure transmitters and data loggers.

Tables 4 and 5 show all spot checks made by hand probe of water levels for all wells. Tables 6 and 7 show all periods for which continuous records of well level exist and the maximum/minimum water level recorded for each well. Groundwater contouring in karst must be done with care since the level at any particular well may be more dependent on the well's degree of connection to the flow system rather than ,-. tree reflection of the flow system potential near the well. Contours have been generated from the data and are discussed in section 1.4.2.

Well water level measurements made by hand are measured from a reference point on the top of the well casing (TOC). Some wells have only an outer casing while some also have an inner casing. A painted groove has been marked on each individual well to indicate the datum reference point used for the hand measurement. Figure 6 indicates when the current reference elevation was established and if the TOC reference point is on the outer casing or inner casing.

1.4.1.4 Spring Flow and Rain Measurement and Analysis

Flows have been measured at the springs around the site. Flows were measured by various methods ranging from visual spot estimates to continuous measurements using weirs or flumes. Table 8 shows periods of time for which continuous records exist and the maximum recorded flow for each time period.

For most of the periods of time when spring flows were monitored, rain fall was also continuously monitored at the site.

1.4.1.5 Dye Trace Testing

Dye trace tests are a useful technique in karst. Tracer tests are conducted by injecting a fluorescent dye and/or soluble salt into the grc ~ndwater (into wells or sinkholes) and then monitoring for the dye/salt at remote locations. In most cases, the aye was injected in a well or sinkhole and then monitored at springs that are near the site. In those cases the dye tests were used to determine where Groundwater that flows under or near the site emerges. In other cases, dye was injected in wells at the site and monitored in other wells around the site. In these cases the dye tests were structured to determine specific flow paths under and around the site.

In 1987, a low flow tracer test was conducted from wells around the site and springs around the site were monitored (Reference 8). The test was designed to determine which springs were connected to site Groundwater under low flow conditions.

In 1989 and 1990, high flow tracer tests were conducted again from wells around the site to nearby springs. The tests were designed to determine which springs were connected to site Groundwater under high flow conditions (Reference 17).

In May 1992, a moderate flow dye trace was conducted from five sinkholes near the landfill. The dye was monitored for at numerous springs in the area. The goal of the test was to determine which areas near the site drained to which springs and thus to estimate the drainage basin size (Reference 18).

In May 1996, high flow tracer tests were conducted by injecting dye into two sinkholes near the site. The tests were designed to determine the arrival time of the dyes relative to the arrival time of PCBs at ICS during a storm event (Reference 19). This information is used to determine the relative location of PCB source areas at the site.

In October and November 2001, a series of three dye trace tests were conducted from wells at or near the site (Reference 20). The goal of the tests was to determine specific flow paths for the dyes around the site from well to well and from wells to ICS.

In April and May 2002, a series of dye tests were conducted from the LF-6 well while monitoring at site and near site wells and ICS. The goals of the testing were to determine the arrival time of dye from LF-6 at ICS (during high and low flow), the direction of travel of the dye immediately after injection and if water flushed with the dye would induce a PCB peak at ICS (Reference 21).

From July 2002 to October 2002 a series of dye tests were conducted under non-storm conditions from several wells around LL Landfill. The goals of the dye testing were to determine travel times to and mass recoveries of the dyes at ICS, to determine travel paths around the landfill by sampling other wells after injection, and to see if a PCB response could be induced at ICS by injecting water in epikarst locations with known high PCB levels.

A general summary of the dye traces is given in Table 27. The first column shows the date and time of dye injection. The second column shows which dye, either Rhodamine WT (ROOT) or Fluorescein (FLR) was injected. First detection refers to the first sample location the dye was detected, and the hours from injection that the first sample was taken that had the dye detected. The fifth column is the date and time of first dye arrival at ICS Emergence and the next column shows that first arrival time in hours. The seventh column is the average flow rate at ICS between dye injection and first detection at the spring. The eighth column is the travel time predicted by the empirical equation derived from the PCB storm pulses. The last column is the percentage of the mass of dye recovered at the spring.

While a detailed report discussing the latest dye test results was not available as of the date of this plan, an investigation plan discussing all dye results and future aquifer test plans is being prepared. Based on an initial review of these latest dye results, it appears that dyes injected south of the railroad tracks in Valhalla Cemetery do not travel to the ICS in mass nor with the proper timing to match PCB responses during storms. On the other hand, dyes injected on the east side and especially near the four series wells yield strong mass recoveries and closely match expected travel times. Dye recovery from wells around the site was also stronger in the east side wells and especially near the four series wells. Wells around 00-370 in Valhalla also showed dye recovery for some injections.

1.4.1.6 Slug and Packer Testing

Packer and slug testing was conducted on various boreholes to determine their hydraulic parameters. The value of slug and packer tests in a formation where the water is being transmitted by solution voids is for relative comparisons between wells. The absolute hydraulic conductivity values are not valid because the equations derived to produce them assume equal contribution along the borehole length being considered as if it were a porous medium.

Packer testing was performed at wells MW-7, MW-10, MW-11, MW-12 and MW-14 in August and September 1987 (Reference 8). The packer lasting was performed prior to installation of screen, riser, and sand pack. Double or single packer tests were performed depending on the zone tested within the borehole.

Slug tests were performed at wells MW-B1, MW-B2, MW-B3, MW-B4, MW-1 D, MW-2, MW- 41, MW-4D, MW-5, MW-6, MW-8S, MW-8D, MW-9, and MW-10 in October through December 1987. The slug tests were performed by dropping a solid cylinder of known volume beneath the static water level of the well (falling head test). The water level change and recovery back to the static level were measured by hand (Reference 8).

1.4.1.7 Pump Tests

Pump tests from various wells have been conducted for three purposes. The first is to determine the hydraulic parameters at the well and of the aquifer in general. Wells that are on or near a conduit may have higher transmissivity than other locations. The second is to see if highly contaminated water can be withdrawn while pumping, which may indicate the well is near a conduit carrying PCBs. The third is to determine the impact of pumping various locations on the PCB levels in ICS.

In October 1998, a pump test was conducted on well 00-387. The well was pumped at an average rate of 14.3 gpm for 4 hours (Reference 9).

In June 1999, a pump test was conducted on well 00-370. The well was pumped at an average rate of 12 gpm for 3 hours (Reference 10).

In April 2001, a series of short term pump tests were conducted on wells MW-16, MW-16 and MW-19. MW-19 was pumped for about 2.5 hours at a rate of about 10 gpm on April 4, 2001 (Reference 22). MW-16 was pumped at about 5 gpm on April 9, 2001, for 6 hours (Reference 23). MW-18 was pumped at about 14 gpm on April 17, 2001, for 2.5 hours (Reference 24). The goal of these tests was to determine if any of these wells were near a conduit carrying high levels of PCBs on the east side of the site. The theory was that if a well was near a conduit carrying high levels of PCBs, then pumping the well would preferentially tap the conduit water and eventually yield hich PCB levels in the water withdrawn during pumping.

In March 2001, PZ-C was pumped (Reference 14). The goal of the test was to determine the PCB content of the water and to see what the water recovery to the pea gravel gallery pit at this location would be.

In August 2001, PZ-D was pumped. The goal of the test was to determine the PCB content of the water and to see what the water recovery to the pea gravel gallery pit at this location would be.

In July and August 2001, two short term pump tests were conducted from MW-41. The goal of the tests was to determine if highly contaminated water could be withdrawn from the well.

In November 2001, two long term pump tests were conducted. In one test, MW-21 was pumped for 52 hours. In the second test, MW-16 was pumped for 140 hours (Reference 20). The main goal of the tests was to determine if highly contaminated water could be withdrawn from a well after long term pumping and to see what the impact on the ICS PCB levels would be with long term pumping.

In February 2002, short term pump tests were conducted at wells MW-41 and MW-4S (Reference 25). The goal of these tests was to determine if highly contaminated water could be withdrawn from the wells.

In June 2002, a short term pump test was again performed on well MW-41. For this test, packers were used to isolate the more transmissive part of the well. The goal of the test was to determine if more highly contaminated water could be withdrawn from this well with the lower zone isolated from the pump (Reference 26).

1.4.1.8 Injection Test

In November 2001, an injection test was conducted at Sargent's Pond. The goals of the injection test were to determine how fast the pond would release injected water and what impact this water would have on well levels near the site, ICS flow and ICS PCB levels. About 200,000 gallons of water were injected over several hours. This is water that had been just previously collected from the long term pump tests in the detention pond at the landfill. (Reference 20)

1.4.1.9 Sampling and Analysis

A large number of samples for PCB analysis have been taken over the years in all environmental media. Primary concerns for this plan are the samples of Groundwater and surface water. The water sampling events and results are discussed in detail in section 1.5. In addition to water samples, a large number of samples were taken from landfill materials and underlying soils at the site. Most of the soil samples were taken during the remediation in 2000 and the results are shown in Reference 4.

1.4.1.10 Isotope Analysis

Dr. Noel Krothe of Indiana University, along with his graduate students, has conducted a number of isotope studies between the site and ICS. Reference 27 is a typical paper generated from these studies and that paper references the other papers/studies conducted.

Isotopes of hydrogen, oxygen and carbon have been used in an attempt to determine the source of water flowing at the spring during storms. For example, rain water has a different isotopic composition than Groundwater. Therefore, by measuring the isotopic composition at the spring during a storm, the spring water can be separated into its constituent sources.

1.4.2 Results of Investigations

1.4.2.1 Site Stratigraphy

Prior to remediation the earthen cover in the landfill proper consisted of fill material and native soils and ranged in thickness from about 10.5 to 43 feet. The fill consisted primarily of ash from burning of municipal and industrial waste and construction debris. The fill material has been described as clayey, silty, sandy, and gravelly soils mixed with varying amounts of ash, cinders, glass, plastic, wood, wire, metal, asphalt, cement, paper, and brick.

Beneath the fill lies the indigenous soils which consist of Crider silt loams. Crider silt loams are silty and clayey soils that are red to red-brown in color and are usually referred to as "terra rosa". Soil depths were noted from 2 to 27 feet in borings.

Bedrock units encountered at the site are, in descending order, the St. Louis Limestone of the Blue River Group and the Salem Limestone of the Sanders Group. Both units are of Mississippian age and are susceptible to slow dissolution by water, which has created the karst landforms.

The St. Louis is composed of gray to yellow-brown limestone, dolostone, and shale. The limestones are generally fine to medium crystalline with scattered brachiopod, coral, and crinoid fossil fragments. Thin layered glauconitic limestones are noted in cores but were not abundant. Powell (Reference 5) states that the St. Louis - Salem contact is midway between the building stone unit of the Salem and what he calls a "stinkstein" unit (a micritic limestone with a sulfur odor). The contact is not defined iithologically, but by a foraminifer fossil (Globoendothyra baileyl).

Several lithologic units were identified as marker beds in the St. Louis. In descending order they include: a chart bed, a coral zone, a chalky or vuggy limestone bed, a lithographic limestone zone, the so-called "stinkstein" unit, a celestite band, and a bed of laminated limestone. None of the marker beds appeared in all drill holes.

The upper part of the Salem is lithologically identical to the St. Louis, being differentiated only by the aforementioned foraminifer. The building stone unit of the Salem is a characteristic light grayish yellow to gray, colitic, highly fossiliferous, finely crystalline, massive bedded limestone. The building stone was reliably identified in one core by Powell in MW-B1 at the 765 feet amsl elevation.

Bedrock in southern Indiana and the Bloomington area generally dips west to southwest at approximately 30 feet per mile. Powell shows structure contours for the coral zone and the lithographic limestone. The coral zone is shown dipping to the southwest and the lithographic limestone is shown trending south. BB&L also constructed a structure contour map, which is based on additional data points (Figure 7 and Reference 8). It shows the coral zone trending more westerly, and the lithographic limestone more to the southwest.

The exploratory borings along the east side of LL were turned into monitoring wells MW-15, MW-16, MW-17, MW-18, and MW-19 and show that there are definite zones of Syria,, (about 1-6") conduits that appear to be bedding plane anastomoses. These zones are at the following elevations (in feet amsl), in descending order: 850-852, 840-847, 834-836, 822- 825, 816-818, 803-808, and 795-800. This can be seen on Figure 8, Geologic Cross Section Eastern Edge of LL Landfill.

These zones seem to be stratigraphically related to brittle dolostones, often brecciated, or dense, brittle, lithographic limestone, that appear in the video logs with blocky, and sometimes conchoidal fractures from the air-hammer drilling. There may also be a relation to intercalations of limestone, dolostone and shales where the dissolution exploits the numerous bedding partings. Perhaps also, the more Brittle rocks are more densely fractured which provided more opportunistic openings for the dissolution process.

The widespread nature of the solutional zones at the site and their apparent correlation to brecciated zones has been noted. However, it has also been noted that elsewhere along the outcrop belt of the lower St. Louis Limestone, brecciated zones may be indicative of gypsum and anhydrite dissolution (Reference 53). Such dissolution might help explain the widespread nature of the solutional zones at the site and the apparent correlation of these zones to specific stratigraphic horizons.

There is also a water-producing zone of small conduits at approximately the 760' to 770' elevation amsl, in the Salem formation. This zone was recognized in the earlier reports, and is referred to as the deeper zone. Wells that monitor it are often appended with the 'D' identifier

Joints and fractures were noted in the rock cores. Some of these joints and fractures were associated with water-bearing zones, as noted above, and most were horizontal (Reference 8).

Fracture traces and lineaments have been delineated from interpretation of aerial photos and field reconnaissance. The general trends of the fracture traces are northeast to southwest and northwest to southeast. Two lineaments trending north to south were also observed in the area. The lineaments are shown in relation to the site, surface water features and karst features on Figure 9.

Fracture trace and photo-lineament analysis was updated in 1999 by Richard Parizek, and verified in the field by him (Reference 11). He identified a major lineament trending northeast-southwest through the site and noted an east-west lineament also. These are shown on Figure 10.

Many subsurface solution features were identified in St. Louis Limestone during the installation of the earlier monitoring wells (Reference 8). But it was not until the exploratory borings were installed post-remediation that all the zones were recognized.

A draft top of rock contour map has been prepared using the data from the boreholes as supplemented by top of rock probing. This is shown as Figure 11. The top of rock generally slopes from east to west with an elevation generally around 875 feet amsl on the eastern margin of the site. The sinkholes on the western portion of the site have bedrock lows generally around the low 830s feet amsl with some spots as low as 822 feet amsl.

1.4.2.2 Site Hydrogeology

A great deal of data on the aquifer has been acquired. At a macro scale, the aquifer can be described with precision. For example, the locations that serve as resurgences for site Groundwater and those that do not are well identified and characterized. The nature and characteristics of the Groundwater flow and spring resurgences have been well documented. The transport of PCBs has been very well characterized and a wealth of data exists against which future data can be compared.

However, the hydrogeology of the site is complex due to the karst nature of the aquifer. With regard to the specific karst pathways between PCB reservoirs in the rock and the main resurgence, there is much less known for certain. Additional study to delineate these pathways is underway and is not a subject for this plan.

For the purposes of this plan the hydrogeology will be described mostly from a macro standpoint. Ford and Williams (Chapter 6, Reference 28) lists a number of aquifer characteristics that can be evaluated. The macro aquifer characteristics that are important for a long-term monitoring plan include:

• Input and output sites
• The areal and vertical extent of the system
• Throughput rates and the response of the system to recharge

One of the key hydrogeologic parameters to describing a karstaquifer is to identify where the Groundwater exits the ground (the output sites). These locations are important as they are the most useful monitoring points and acts as transport/exposure points. The typical methods to identify the output sites are dye trace testing and contaminant sampling.

The low and high flow tracer testing conducted in 1987-1990 and PCB sampling of all springs near the site have identified ICS as the main resurgence of Groundwater impacted by the site. A small amount of site related water was also observed to flow to the Slaughterhouse/Packinghouse Spring System. The Quarry A and B Springs were shown to be a second resurgence of ICS waters. The ICS and QS waters form the headwaters of Clear Creek. The Slaughterhouse/ Packinghouse Springs form the headwaters of Stout's Creek.

Once the main output site was identified, extensive monitoring and sampling of that site was conducted. This took the form of water flow and conductivity measurements along with PCB sampling.

Karst aquifers can be classified as primarily dominated by diffuse or conduit flow and by the type of recharge that occurs (Reference 28). While different methods have been proposed to make the classification, spring hydrograph and chemograph analysis is typical. For conduit dominated flow systems with rapid recharge from sinkholes and/or sinking streams, the hydrographs and chemographs will show large variation during storm events.

For this aquifer system there has been extensive spring flow and chemical sampling at ICS. Refer to data for any of the larger storms in Appendix A, or the data presented in Figure 12. This data clearly shows a rapid and substantial response to precipitation at ICS for both the flow and chemical constituents (conductivity, TSS and PCBs). This shows that the flow feeding this spring is dominated by conduit flow and relatively concentrated input such as would occur at the numerous sinkholes throughout the basin. Therefore, the input sites for this basin are expected to be primarily fast flow sinkholes and swallets which can route surface runoff directly to the subsurface conduit network.

Prior to the installation of the landfill cap, Groundwater recharge to the flow system occurred through the sinkholes beneath the landfill and through the ones located immediately adjacent to the site, as well as by percolation through the overburden. The sinkhole locations adjacent to the landfill include areas to the southwest of the landfill, known as Martins Sink; to the west of the landfill, including Sargent's Pond; and possibly to the east- northeast of the landfill (Figure 5).

At present, after the installation of the landfill cap, no direct recharge to the flow system occurs through the landfill. The landfill runoff is now routed to lined ditches and through a lined detention pond to Sargent's Pond.

Sargent's Pond is a sinkhole pond that was sealed by the former landowner in the early 1990s. Sargent's Pond has been evaluated by level monitoring and a large scale injection test. The results of these tests show that the pond has a low leakage rate when the pond level is in its normal operating range. The in-line detention pond, constructed during remediation, functions to reduce the Sargent's Pond level during larger storm events.

In addition to rain infiltration at sinkholes and soil macropores, there are railroad tracks which abut the southern perimeter of the site. At the eastern portion of the site, the railroad tracks sit directly on bedrock since the entire soil column and some bedrock were removed when the tracks were constructed. This area of the tracks could also be a source of rapid infiltration of rain water to the bedrock.

An additional tracer test conducted in 1992 showed which sinkholes near the site contributed flow to ICS. This provided an approximate areal extent for the basin feeding the spring. This approximate basin area was evaluated in the basin flow models developed. Based on this dye trace testing, the basin feeding ICS, which includes the landfill and surrounding area, is estimated to be approximately 341 acres.

Extensive flow monitoring at ICS for many years (Table 8) has shown that flow at this spring ranges from less than 10 gpm during times of severe drought to over 4000 gpm during extreme periods of rain. Flow models for the basin can predict the storm response at ICS based on combinations of rain and pre-storm spring flow.

The relation between spring flow and rainfall was evaluated by multiple regression analysis. The statistically significant variables (p<0.05) were total rain and pre-storm flow. A pre- remediation storm set of 116 storms from 1994 to 1999 were evaluated. Those storms and the results are shown in Table 26. The best correlation established for peak flow was:

LN (peak flow) = 0.95246 LN (total rain) + 0.50688 LN (pre-storm flow) + 4.2462

The R2 coefficient for this equation was 0.84, which is considered excellent for hydrologic work.

The SEDIMOT II computer program was also used to model response to storm events at ICS. SEDIMOT II uses the SCS-TR-55 (Curve Number) method to model surface run-off and peak flow (Urban Hydrology for Small Watersheds, Technical Release 55, June 1986). The original purpose of the modeling was to evaluate the use of sinkholes as detention basins.

The initial watershed boundaries and program schematic used for the model were based on the IU graduate student dye traces of 1992 (Reference 29). Preliminary runs indicated that the model would predict excess runoff and peak discharge using the maximum probable drainage area of 341 acres as interpreted from the dye tracing. After consultation with EPA consultants and inspection of the watershed, an amended watershed of 234 acres was modeled. While this amended watershed gave good agreement with total runoff volumes measured, the peak flows were still much higher than what was measured at the ICS weir.

The large differences in peak discharge between the calculated and measured value with similar total runoff volumes indicates there is significant stormwater detention taking place in the karst watershed compared to a surface watershed. "System" detention basins were modeled to account for this temporary karstic storage. Table 1 from Reference 30 shows the results of the "system" model. Fairly good agreement is found with the model.

Viacom also contracted with Dr. Billy J. Barfield, the developer of the SEDIMOT II program, to model the iCS watershed, based partly on the work above, but using a 54 storm set to optimize the parameters. (Attachment 13 of Reference 10) He completed the model on an Excel spreadsheet and achieved good results, i.e. his predicted vs. measured R2 was ~0.8.

In addition to the ability to predict flow at ICS for a given rain event, the amount of PCBs expected to be discharged during a rain event and during non-storm periods is also predictable. The correlations are discussed in section 1.5.

The vertical extent of the system has been evaluated by several means. First, as detailed in section 1.4.1, numerous boreholes/wells have been drilled at the site and surrounding area. Monitoring well observations indicate that the phreatic system beneath LL Landfill appears to occur within two intervals.

The shallower phreatic Groundwater zone occurs within bedrock between 795 and 818 feet amsl (during non-storm conditions), and the deeper zone occurs within bedrock between 760 and 770 feet amsl.

The zone encountered between 795 and 818 feet amsl elevation is the main water bearing zone, and is perennially saturated. This zone can expand upward (backflooding) to 835 feet amsl during storms. A large number of wells that penetrate this zone have had their water levels continuously recorded for various time periods.

Based on hydraulic conductivity tests, the deeper zone appears to be much less transmissive than the upper zone. The deeper zone also has high conductivities and high sulfate concentrations, and appears to be the lower limit of meteoric circulation in the Groundwater basin. This apparent lack of circulation, and general absence of PCB detection, has directed most investigative efforts away from this zone.

Above the phreatic water levels, upper zones or epikarst were encountered from about 818 to 852 feet amsl elevation. These serve as water storage zones and flow on an intermittent basis mainly in response to storm events.

A horizon at 795-800 feet amsl elevation is part of the main phreatic zone and appears to be stratigraphically related to a brecciated dolostone. This horizon seems to be a persistent anastomotic maze that is well connected throughout the immediate landfill area. In fact, it is almost impossible to drill a well in the vicinity of LL and not encounter small bedding plane solution conduits that are part of this maze.

Quinlan (Reference 31) in his study of the sinkhole plain of central Kentucky delineated 28 Groundwater basins and sub-basins. He found an overall pattern of dendritic conduits with downstream convergence into major trunk conduit passages.

Palmer, first in his doctoral dissertation (Reference 32), and in two later studies, "Prediction of Contaminant Paths in Karst Aquifers" and "A Statistical Evaluation of the Structural Influence on Solution-Conduit Patterns", confirms that in low-dip carbonate terrains where sinkhole recharge predominates, dendritic convergence of conduits is the usual pattem.

This pattern, which he termed "branchwork", most often occurs with vadose flow going down dip and phreatic flow finding a major spring outlet along strike. He cites several Indiana examples, including Blue Springs Caverns in nearby Bedford.

Dye testing conducted from LF-6 has shown that at least some of the vadose flow goes down dip to the southwest. ICS and Slaughterhouse Spring are also oriented along strike from the site.

Groundwater flows primarily through solution enlarged fractures and joints. Based on the evidence from tracer tests, well levels and PCB sampling at the spring, there is a branchwork conduit system that ultimately converges and discharges at ICS. Superimposed on, or coincident with, this branchwork is a series of anastomosing channels that appear to be stratigraphically and perhaps lithologically related. The anastomotic network developed at the 795-800 feet amsl level is well interconnected, and well connected to the converging conduit system. This is shown from the water level monitoring that has taken place during storms and the pump testing between wells that penetrate to this zone. This zone is also the first place that significant water is encountered during drilling.

ICS elevation is 815.52 feet amsl at the emergence, and the elevation of the spring appears to be controlling the potentiometric levels in the shallower phreatic aquifer. The spring elevation then serves as the lower limit for phreatic Groundwater. The potentiometric surface near the site has been evaluated at both low and high flow.

In karst systems, the potentiometric surface can fluctuate over a wide range. For example, it is not uncommon for well levels to raise over ten feet within hours of a rain event. Wells that are connected to the conduit system feeding the main spring outlet will closely reflect changes in system flow.

When constructing these potentiometric surfaces, only wells that penetrate to the phreatic development zones are used. For example, a well such as 00-300A would not be used because the depth of the well is at 816.7 feet amsl. This elevation is not deep enough to ensure the well is in full communication with the zones of conduit development below the typical phreatic potentiometric surface

Figure 13 shows a typical potentiometric surface at low flow (non-storm conditions). This surface is a generalization of the water level data contained in the table on Figure 13. Average values for water levels in wells that are grouped together, such as MW-5S, MW-5S, MW-10, MW-20 and MW-21, were used for the site contour surface.

Note that even though the ICS is to the southeast of the site, the surface shows the potential for Groundwater flow around the site is to the easVnortheast. Note also that the head difference between the site and the spring is rather low (less than a foot). this indicates that there is a very well developed conduit system with low flow resistance.

There is also very low head differential indicated across the site. For example, the head difference between NN-625 and MW-17 (wells on opposite ends of the site, more than 700 feet apart) is only .1 feet. However, water level variation over small distances can be much larger than this. There is a head difference of 0.34 feet indicated between NN-625 and NN- 700, which are less than 100 feet apart. Among the group of five wells indicated above, MW-4S, MW-41, MW-10, MW-20 and MW-21, the variation in head is 0.55 feet.

While a good deal of care was taken in both measuring the water levels and performing the elevation surveys on the well reference points, there is likely some error in both measurements. At this point, Viacom assumes an error of approximately .1 feet in level measurements.

In all the water level surveys performed by Viacom, well MW-5S has consistently been the low point around the site. Again, this must be viewed with some caution. The well surface casing of this well is known to be bent, which will cause a water level to read low. The amount of level distortion caused by this has not been determined.

To construct a potentiometric surface during a storm event, data logger data is used. Care must be exercised to use data of known high quality and judgment exercised since Groundwater mounds will appear near areas of recharge to the phreatic zone. A typical high flow surface is shown in Figure 1 A The monitoring well elevation data indicated on Figure 14 is for a storm event on October 14, 2001. Note that the high is at well MOO-6 indicating recharge to the phreatic zone near this well. Well MW-41 is the low spot, again indicating overall flow around the site to be to the northeast.

Overall, the wells near the site that penetrate the 795-800 feet amsl horizon all respond identically to rain events (with differences only in total rise based on position in the basin) and mirror the flow of ICS. Figure 15 shows a typical well and spring response at ICS during a storm event to illustrate this point. This corresponds with pump test data that show all these wells to respond in kind with the spring during pumping and indicates a very well interconnected phreatic aquifer.

Well MW-6, which is south of the site in the northern fringe of Valhalla Cemetery, has historically been the well which shows the highest level rise during storms. This well also shows larger temperature and conductivity variations during storm events than most other wells. It is believed that these are indications that the well is on or close to a conduit that is carrying a mix of ground and storm water during storms.

Based on several lines of evidence, the 4-series wells appear to be near a major branch conduit. This evidence includes:

• Well 4s and 4i consistently being the potentiometric low for all the phreatic wells in and around the site.
• The apparent recharge deflection of MW-17 in distance-drawdown comparison between pumping MW-21 vs. MW-16 (Ref. 20, Figures 17 and 18).
• The greater degree of drawdown induced by pumping MW-16 vs. MW-21, indicating MW-21 to be nearer a recharge boundary (Ref 20, Figures 19 and 20).
• The recharge-indicated deflection of the 4-series wells on the distance-drawdown plot for the MW-16 pump test after equilibrium was reached (Ref. 20, Figure 22).
• Higher levels of tracers detected in MW-21 vs. MW-16 during pumping (Ref. 20).
• Convergence of tracers to 4-series wells during natural gradient tracer tests.
• Higher levels of PCBs, both during long-term pumping (Ref. 20) and other sampling (Ref. 25).

While the levels from phreatic wells can show the water level rise during storms from the saturated zone, they will not depict the water levels in any perched zones in the epikarst. Shallow wells at the site such as LF-1, PZ-E, and LF-6 have reflected water levels of approximately 855, 852 and 845 feet amsl, respectively. All these wells are completed much shallower than the phreatic zones and are reflecting localized conditions in the epikarst. These local epikarst zones have flow directions that are for the most part unproven, although dye testing from LF-6 shows that at least some of the water from that well flows southwest, which agrees with Palmers work that epikarst water flows down dip.

Dr. Noel Krothe and several of his graduate students at IU have taken ICS water samples over several storm events and analyzed the water for various natural isotopes. Their work indicates that most of the water discharged at ICS during a storm is stored Groundwater displaced by rain water.

Their work further indicates that most of the displaced water originates from the epikarst rather than the phreatic zone. These indications are somewhat in conflict with the interconnection and response of the phreatic zone seen at the site. But their results are supported by PCB levels at ICS which we generally higher than those found in phreatic wells near the site. As discussed in section 1.5, PCB levels in some epikarst wells at the site are much higher than those in the phreatic wells or ICS. Figure 16 is an example of the storm hydrograph separation results achieved with stable isotope sampling. Additional work to confirm Dr. Krothe's conclusions has not been conducted.

Several key points can be made about the hydrogeology:

• All lines of evidence; including dye tracing, pump testing, PCB sampling, flow and basin modeling, point to ICS as the primary resurgence of the ground water basin that includes LL Landfill in its drainage area.
• In terms of hydrologic analysis and response to transient events, i.e. storm events, injection tests, pump tests, etc., the basin has been well characterized from a "black box" perspective. That is to say, given input parameters, the flow response at the spring can be predicted with fair certainty.
• The basin appears to function as a conduit flow dominated karst system with input primarily coming from sinkholes.
• The ability to "take the black box apart" and understand the actual flow paths will require additional study. Two wells, MW-21 and MOO-6, may be on or very near conduits. But, additional study to verify this and understand the position of these potential conduits in the overall basin hydraulics and contaminant transport is required.

1.5 PCB in Water Data

Other PCB samples related to waterways are stream water and sediment samples and fish samples. Each is discussed separately below.

1.5.1 Groundwater Monitoring Wells

The use of Groundwater monitoring wells as an indicator of Groundwater quality in karst terrain is difficult to interpret making its use questionable. This is because most of the Groundwater in karst moves in discrete conduits. Unless the well happens to intercept one of these conduits, the contaminant data from the well cannot be used as an overall indicator of Groundwater quality at the site or the transport of contaminants from the site. Viacom and general industry experience is that attempting to locate conduits in karst and intercept them with monitoring wells is an extremely difficult and uncertain task. Because of this, the value of PCB data in the monitoring wells in assessing general Groundwater quality and contaminant transport is questionable.

The EPA has recognized this situation. EPA guidance for groundwater monitoring in karst is contained in "Groundwater Monitoring in Karst Terrain: Recommended Protocols and Implicit Assumptions" (Reference 33), and "RCRA Ground-Water Monitoring Technical Enforcement Guidance Document" (Reference 34). These documents state that the proper locations for monitoring the status of groundwater in karst areas are springs, cave streams and monitoring wells that have been proven by dye tracing to intercept conduits carrying water from the monitored location.

Nevertheless, Groundwater samples from the site monitoring, wells have been analyzed for PCBs since the first wells were installed in 1982. Table 9 is a complete listing of all well sample PCB analyses results. When reviewing the well data, there are a few things to note:

• A well with the designation "D" indicates a well that penetrates to the 760-770 feet amsl zone.
• Most other wells penetrate no deeper than the 790 feet amsi level.
• The majority of the wells at the site penetrate to the 795-800 feet amsl horizon.
• Refer to Table 2 for well details.
• Some wells were originally drilled penetrating both phreatic zones. These were later modified to be opened to only the upper zone.
• The early well samples, performed from 1982 through 1996, were taken by purging at least one well volume typically with a bailer and sampling after field parameters stabilized. Later, wells drilled after 1996 were sampled many times by simply lowering a bailer down the well without purging. in other cases, wells were aggressively pumped, sometimes at high rates, and samples taken periodically while pumping (a pump test). A few wells were also sampled during storm periods. These storm samples were typically taken without significant purging.

The EPA sampled the four monitoring wells they installed, MW-B1 through MW-B4, near the site for the first time in October 1982 and again in June 1983. All samples were found to be non-detect (ND) for PCBs.

Between December 1982 and Januarv ~ -~83, Groundwater samples were collected by Viacom at four locations on-site and nine locations off-< ite. The highest level of PCBs detected was in MW-4D at 2.4 ppb.

In February 1983, Groundwater samples were collected by Viacom from thirteen monitoring wells on or near the site. The highest sample result was at MW41 at 1.5 ppb.

In 1991, a program to sample most wells at or the near the site was initiated on a quarterly basis for one year. The highest results were found at well MW41 (99.65 ppb) and MOO-7 (11 ppb).

Viacom performed additional monitoring well sampling as part of the Field Sampling Plan for the Lemon Lane Groundwater Monitoring Investigation (Reference 35). This investigation included one year of monthly sampling at five wells on the periphery of the site. These results show that PCB levels in Groundwater on the immediate periphery of the site ranged from less than 0.1 ppb, up to 1.4 ppb. The results of this sampling effort demonstrated that these wells are not reliable monitors of PCB transport since sampling at ICS during the same period showed PCB levels at the spring higher than in any of the wells.

Starting in 1998, Viacom has been attempting to locate additional wells in conduits carrying PCBs from the site. The wells have been located using various geophysical techniques, as discussed in section 1.4.1. These wells typically were sampled with little to no purging. Between July and October of 1998, seventeen borings were drilled south of the landfill in and around the Valhalla Cemetery based on a natural potential geophysical survey. All the borings, except NN-412, were drilled to the 795 feet amsl phreatic elevation. Water was sampled from all the borings except 11-87,11-95 and 11-203, which were dry. Most of the borings were sampled from the upper epikarst zone and also from the lower phreatic zone as indicated in the table. The upper zone samples were taken during drilling before the drill had advanced to the phreatic zone. Some of the upper zone samples had high levels of PCBs (note results for wells NN-300 and NN-300A). Additional sampling of these wells was performed during late 1998 and early 1999, as shown on Table 9.

Well 00-387 was pump tested on October 28, 1998, and samples were taken while,pumping. The PCB levels ranged from .14 to .36 ppb. Well 00-370 was pump tested on June 17, 1999. PCB levels ranged from 2.5 to 18 ppb while pumping.

Several wells in Valhalla Cemetery were sampled during storms in early 1999. During a storm on February 11, 1999, wells MOO-6, NN-625, and NN-300A were sampled. PCB results in well NN-625 were all BDL. In well MOO-6, PCBs ranged from BDL to 17 and in well NN-300A ranged from 12 to 24 ppb. In addition to PCBs, temperature and conductivity were also measured during this event. Of note were the conductivity and temperature of well MOO-6 which had significantly lower conductivity and higher temperatures than the other wells. This indicates this well is near or on a conduit which routes a mix of storm and Groundwater to ICS. All these samples were taken by a dedicated bailer without purging.

On February 27, 1999, well NN-300A was sampled again during a storm. This time the samples were taken with a submersible pump at a continuous low flow rate of about 1 gpm. PCB results ranged from 3.1 to 11 ppb. The conductivity at the well also showed a dip indicating that some storm water flowed through the well.

On May 6, 2002, wells MW-5S, NN-300A and 00-370 were sampled for PCBs during a storm event. PCBs in well MW-5S ranged from 1.1 to 1.9 ppb, in well NN-300A from 3.9 to 18 ppb and well 00-370 from 7.3 to 11 ppb. All wells were sampled with dedicated bailers with no purging.

The LF series of wells were drilled in early 2000 along the southeast edge of the landfill to intercept suspected karst conduits. Significant water developed in LF-1, LF-2, LF-5, LF-6 and LF-7. LF-3 developed water very slowly and LF~ did not develop water. Water samples were taken soon after completion from each of the wells that developed water and a pump test was conducted from LF-1 and LF-5. PCB levels in LF-2 and LF-5 were less than 20 ppb. PCB levels in LF-3 were 100 ppb but the well developed very little water and was abandoned.

PCB levels in LF-1 were significant both with and without pumping. For example, during a pump test of the well on March 23, 2000, the PCBs varied from 230 to 16,000 ppb. Videologging of well LF-1 showed the well to be on the side of a small epikarst conduit at about the 858 feet amsl elevation. A pump test performed on LF-5 on March 22, 2000, showed PCB levels from 2.3 to 6.5 ppb.

PCB levels in LF-6 and LF-7 were also significant and a pool of PCB DNAPL was discovered in LF-6. These two wells were confirmed to be located in a substantial epikarst conduit that was mostly clay filled. This conduit was uncovered by excavating down through rock and some highly contaminated clay was removed during site remediation in the fall of 2000. Pea gravel was placed into the conduit area before covering the area with soil.

After remediation in 2000, exploratory borings were drilled along the east side of LL Landfill. These borings have become mor .oring wells MW-15, MW-16, MW-17, MW-18, and MW-19. Figure 6 shows the location of these wells. These wells were sampled in both the upper and lower zones in early 2001. Pump tests were performed on MW-16, MW-18 and MW-19 during this time. PCB sample results are listed in Table 9.

Based on additional resistivity imaging, MW-20 and MW-21 were drilled in April 2001 in the MW-41 area. These wells were part of a continuing attempt to find a major conduit that drains the PCB-contaminated Groundwater under the site. MW-20 was sampled on September 27, 2001 and April 4, 2002, with PCB results obtained at 25 and 4 ppb, respectively. MW-21 was sampled on September 27 and October 12, 2001 with PCB results at 4 and 5 ppb, respectively.

MW-21 was pumped for 52 straight hours from November 7 to 9, 2001. PCB levels in water withdrawn while pumping ranged from 5 to 7 ppb. MW-16 was pumped for over one hundred straight hours and PCB levels in water withd awn from the well ranged from .6 to 8.3 ppb.

Shallow offset borings were installed within five feet of five deep (phreatic) borings and designated by the suffix of "A" at five locations, NN-12A, 00-125A, NN-300A, 00-300A and 00-587A. These wells were sampled in the epikarst zone in February and March of 2002.

Pump tests were performed on MW-41 twice during 2001. PCB sample results exhibited high PCB content ranging from 40 to 4,000 ppb. The well was modified in the fall of 2001 to remove some of the bentonite sealant in the bottom of it. The pump test was repeated on MW41 in February 2002, and again in June 2002 (with a packer isolating the lower phreatic zone). PCB results for these tests were much lower, ranging from 1.7 to 15 ppb.

A short tern pump test was performed on MW-4S in February 2002. PCB sample results were generally anging from .81 to 9.6 ppb.

Piezometers PZ-EE, PZ-F, and PZ-G are located on the periphery of the final limits of consolidation of the landfill. They were located based on resistivity anomalies as part of the karst conduit study. PZ-G has remained dry and not been sampled.

PZ-E and PZ-F both have high levels of PCBs. PZ-E is located in the southeast corner of the landfill as shown on Figure 6. It was sampled twice in December 2000. PCB results of 10,000 and 3,100 ppb were obtained. In May 2001, it was sampled at 1,300 ppb. On February 14, 2002, the first water sampled from PZ-E was sampled at 2,000, and the sample taken after one well volume of water was removed was 2,700 ppb.

PZ-F is located in the middle of the southern perimeter of the landfill as shown on Figure 6. It was sampled on March 6, 2001, and February 15, 2002, at 42,000 and 150,000 ppb, respectively.

PZ-C is located just outside the western perimeter of the landfill so does not extend through the consolidated fill. It is screened into a gravel gallery that was constructed in a location where free product was discovered during excavation. It was sampled twice in 2001 and once in 2002, with analytical results ranging from 76 to 220 ppb. PZ-C was pumped on March 21, 2001. The water level was observed after pumping and the recovery found to be slow.

PZ-D is also located on the western perimeter of the landfill, outside of the limits of consolidation. It is also screened into pea gravel gallery that was constructed into another location where free product was observed during excavation. From May 2001 to February 2002, four sample results have ranged from 200 to 21,000 ppb. PZ-D was pumped in August 2001, and as with PZ-C, recovery was very slow.

Overall, much higher levels of PCB contamination (including DNPAL) have been found in shallow epikarst wells at and near the site rather than in phreatic wells. This indicates that the PCB oils originally dumped at the site did penetrate the soil/fill material into shallow bedrock and that some of this oil remains stored in the shallow rock.

Most of the shallow wells drilled at the site were concentrated in the southeast corner of the site. This area was investigated because the PCB levels in soil/fill in this area were found to be much higher and this contamination extended through the soil to the top of rock. Yet few capacitors were found in this area. This indicates that free oil must have been dumped repeatedly in this area, perhaps from scavenging operations. As noted in section 1.4.2, at least some of the epikarst water in this area has been determined, by dye testing, to flow to the southwest

1.5.2 Spring Emergence Water Sampling

PCB data in springs is important for several reasons. First, the PCB data shows which springs are impacted by the site. It backs up and confirms dye test data discussed earlier. Second, once the main springs impacted by the site were identified, a large data base of PCB results for those springs has been built up over the years to allow trends to be evaluated.

As discussed in Section 1.3, there are numerous springs in the vicinity of the site that serve as the headwaters of several perennial streams, as shown on Figure 5. The site is located near topographic divides between the headwaters of three local surface streams, Clear Creek to the southeast, Stout's Creek to the northwest and Griffy Creek to the northeast.

The main springs in the area include:

ù ICS and the related QS which form the headwaters of Clear Creek
• The Slaughterhouse Spring System (consisting of Slaughterhouse, Packinghouse Road, and Packinghouse Culvert Springs) which forms the headwaters of Stout's Creek
• Stony East and Stony West Springs which waters flow through Twin Lakes and eventually joins Clear Creek
• Jacks' Defeat East and West Springs which form the headwaters o.~ Jack's Defeat Creek
• Crestmont/Urban Springs which form the headwaters of Cascade Branch Creek
• The Detmer/Robertson Springs which form the headwaters of the West Branch of Stout's Creek.

As noted in section 1.4.2, dye testing has shown that the ICS and QS are the main resurgence of Groundwater impacted by the site, with a small amount of site related water also flowing to the Slaughterhouse System. This has been confirmed by PCB sampling of all the springs. Table 10 shows the PCB non-storm data taken at the ICS/QS and Slaughterhouse Springs. Table 11 shows the water samples taken from the springs in the area other than ICS/QS and Slaughterhouse Systems.

Note that the Detmer and Robertson Springs have indicated some low levels of PCBs. These springs did not receive dye during the LL traces and these springs have also shown low levels of other contaminants found at the ABBplant (the former Westinghouse capacitor plant). Therefore these springs are thought to be related to Groundwater that flows under the ABB plant and not associated with LL Landfill.

ICS is located approximately a half-mile southeast of the LL Landfill. Based on the evidence from tracer tests, well levels, and PCB sampling at the springs, it has been determined that there is a branch work conduit system underneath LL Landfill. The tracer tests and PCB sampling established ICS as the primary resurgence of Groundwater that flows under the LL Landfill, and indicated that the QS are the second resurgence of that same Groundwater.

In reviewing the PCB data, emphasis is placed on ICS with some discussion of the Slaughterhouse System. The data review for these springs is separated between low flow (non-storm) sampling and high flow (storm event) sampling since the PCBs present themselves differently for these conditions.

1.5.2.1 Non-storm Conditions

The first samples from ICS/QS System were taken by the City of Bloomington in August 1981. The QS sample contained 5.7 ppb PCB. Since that time, the system has been sampled many times. While most of the samples represent storm conditions, a good deal of data exist for ICS/QS System at non-storm condition. This data is shown on Table 10.

A graph of all the ICS data is shown on Figure 17. Note that the bulk of the data is between 5 and 20 ppb with no noticeable trend. Figure 18 shows a plot of the ICS data for only the last two years. In this plot it appears that there are periodic cycles to the data. These cycles are also found at the South Spring System at Neal's Landfill. It is believed that these cycles are related to flow conditions in the basin. When flows are generally higher at the spring, indicating ~ wet bar , PCBs are generally lower. When dryer conditions Act, PCBs are generally higher. Table 12 shows the ICS PCB and conductivity data. Figure 19 is a plot of all PCB/flow data pairs.

The same can be seen for PCBs versus conductivity of spring water. Figure 20 shows a plot of all ICS non-storm PCB data versus spring conductivity. Note that at high conductivity, PCBs are high and PCBs are similarly low at low conductivity. Therefore, because of the correlation between PCB concentration and flow/conductivity, to evaluate spring PCB concentrations and determine if there are true trends over time, a simple plot of PCBs versus time is not sufficient.

The implications of the correlations are that to properly trend PCBs in this spring system, you must also monitor either flow or conductivity, or both. The PCB relationships with flow and conductivity cause an apparent seasonal trend. In the dryer seasons of late summer and fall when flows are lower, PCBs and conductivities are relatively higher. Typically, in the wetter late winter and spring seasons, when flows are generally higher, PCBs and conductivities are generally lower.

The PCBs, if in a real descending trend, should show a family of curves, similar to those shown in Figure 21, that descend with respect to the same conductivity levels when plotted for successive time periods.

Inspection of Figure 21 shows that PCBs did appear to elevate during the remedy construction period, and that there does not yet appear to be a descending trend since remedy completion. Additional data should be collected to confirm and monitor this correlation and the trend at this spring.

Table 10 shows the historical non-storm PCB results for the Slaughterhouse and QS Systems. As can be seen, PCBs are generally less than .1 ppb for the Slaughterhouse System except for the period when the site was remediated. This again shows that the remediation at the site stirred up PCBs in the subsurface. The QS PCBs were generally about 40-60% of ICS PCBs prior to the construction of the treatment plant at ICS. After the treatment plant began operations, the QS PCBs are generally less than 2 ppb.

1.5.2.2 Storm Conditions

Since the inception of the project, Viacom has monitored a number of storms at the ICS/QS and Slaughterhouse Systems. Table 13 is a summary showing the characteristics of 27 storms that have been monitored since May 1995. The last seven storms occurred after the LL site was remediated and capped. Appendix A includes tabulated data for these events.

In May 1995, ICS and Slaughterhouse Springs were sampled during three storm events. ICS had a high concentration PCB pulse detected while Slaughterhouse was BDL for those same storms. A smaller storm was sampled in October 1995 and this time slight PCB detections were found at Slaughterhouse. In May 1996, sampling was conducted at QS and downstream Clear Creek, as well as ICS. This sampling found ICS to have the highest concentrations, and stations downstream were diluted compared to it. Subsequent storm event sampling has focused on the ICS only.

The best sampling coverage for a pre-remedy storm is the April 15, 1998 event. This event represents a large sized compound spring storm event for the Bloomington area. Figure 12 presents a summary of the relevant data taken at ICS for that event and shows the cumulative rainfall. As shown on Figure 12, the April 1998 storm had about a half inch of rain in the first three hours of the storm followed about eight hours later by an additional 1.67 inches over a ten hour period.

Because PCB concentrations can be impacted by storm water dilution, it is somewhat easier to view the PCB discharge history during an event by examining the cumulative PCB mass discharge history for the event, such as depicted in Figure 22.

By inspection of both Figures 12 and 22, it appears that PCBs begin a very slow rise during the initial half inch rain. As the intensity of the rain increases during the second part of the storm, the PCBs, flow and TSS rise and conductivity drops. The PCB mass discharge rate then drops off rapidly once the flow peak is reached and flow starts to recede. The calculated total PCB mass discharged at ICS during this event was estimated to be 719 grams.

Since completion of the remedy in the fall of 2000, Viacom has monitored seven additional storm events. To make comparisons and draw conclusions with storm data from one storm to another is a difficult task. Viacom's experience has shown that storms of varying magnitude will have different PCB discharge levels and that although the PCB response can be generalized, making specific conclusive comparisons is typically complicated by a lack of data from precisely comparable events.

In reviewing the post-remedial monitored storms listed in Table 13, Viacom believes that based on the limited data available to date, the most comparable event to the April 15, 1998 storm is the June 4, 2001 storm. Like the April 15, 1998 storm, it is also a large compound storm. The June 2001 event had a peak ICS flow rate of 2,739 gpm compared to 2,637 gpm for the April 1998 storm. The total rain for both storms was between two and three inches. Figure 23 shows the available relevant data for the June 2001 event.

Figure 24 shows PCB cumulative mass for both of the comparable storms on a time scale that begins with the first flow response during the storms. The chart shows the low level of PCBs during the initial half inch accumulation over the first 14 hours of the April 19,98 storm. PCBs started to increase as the ICS flow and the rainfall started to spike upward. The June 2001 storm shows a similar response. PCB samples were not taken for the initial part of this storm until flow and rainfall started to rise to their peak. The initial PCB sample shown on Figure 23 shows that sampling started just before the PCB levels staged to rise.

Both storms show a similar cumulative mass of PCBs emitted by the storm, 719 grams for April 1998, and 625 grams for June 2001. The June 2001 storm also seemed to exhibit a more rapid decline in PCB levels once the rain and ICS flow peaks were reached.

In further examining this data, there is another consistent finding. The receding limb PCBs appear to have initially increased from pre-remedial storms when viewed on a flow basis. The plots of PCB versus spring flow for receding limb data from pre-remedial storms is compared with the June 4 and June 6, 2001 post-remediation data. Figure 25 shows receding limb PCB versus flow data for the pre-remedial storms in April 1998, January 8, 1998 and June 2, 6 and 8, 1997. The post- remedial storms definitely tend to have higher receding limb PCB values than pre-remedial storms, for the same flow rate.

At Neal's Landfill, receding limb PCBs also initi. Iffy appeared higher during the April 2000 event which occurred about six months after the remedy at the Neal's Landfill site was completed. The low flow PCB data at the NW Spring System appeared to experience a high during and immediately after construction and then declined. It is theorized that because the interim cap was removed and the system was disturbed during the remedy the conduit system was loaded up with PCBs and it took a few large storm events to flush out the PCBs. This is discussed in Section 1.5.2.2 of the Neal's Landfill Long-term Groundwater Monitoring Plan (Reference 36)

Since the storm data at ICS in June of 2001 is only six months after the completion of the LL Landfill remediation, the conduit system under LL may not have recovered from disturbances caused by the remedy. Additional storms should be monitored to determine if receding limb PCB levels will eventually decline, as was the case for Neal's Landfill.

The peak ICS flow experienced during the May 2002 storms was 2,832 gpm, which is slightly higher than the peak flows measured during the June 2001 and April 1998 storms. However, the peak PCB concentration obtained during the May 2002 storm was 100 ppb, compared to 510 ppb for the June 2001 storm and 190 ppb for the April 1998 storm. Also, even though the PCB sampling was not extended out over the entire receding limb of the May 2002 storm, the cumulative PCB mass released over all three of the May 2002 storms is 558 grams. This is less than the PCB mass released during the June 2001 and April 1998 storms. This may be indicating that the conduit system under LL is recovering from the disturbances caused by the remedy.

There is a fair amount of uncertainty in comparing these storm events. First, there is uncertainty in the measurements made during the storm. Storm flows were monitored by using a water level recorder at the ICS weir up until mid-1999. Since then flow measurements have been made at the inlet of the treatment plant built by EPA. This data may be inaccurate at times due to variations in flow conditions upstream of the culvert, as well as depth sensor calibration. The ICS flow is recorded by using a Doppler area/velocity flow meter and also by calculating the flow based on the rate at which the spring receiving sump (SRS) building fills with water. The Doppler data are biased low at all flow rates, and the errors appear to be related primarily to the velocity measurements. The principal errors in the SRS building data calculations involve resolution of the sump level measurements and calibration of four magnometers used to determine pump discharge rates out,of the SRS building and return flow rates from the storage tanks.

The PCB measurements are also a source of uncertainty. All the storm data was collected at the Sam? locations, by the same methods and analyzed by the same lab However, Viacom has observed up to a 100% difference in duplicate samples taken during storm events over the years. These differences are attributable to both normal lab variations and actual heterogeneity in conditions over very short periods of time in the stream. There is also uncertainty in the comparability of the storms. None of the monitored storms are exactly the same and even two storms that are nearly the same can have somewhat different responses. There is also some uncertainty over the best way to analyze contaminant loading and storm data from karst systems. There is little guidance available on the subject.

The uncertainties can best be addressed by sampling multiple storms over an extended period of time and continuing to improve the quality of the measurements made. With this approach, enough of a database will be developed to lend more weight to the conclusions.

1.5.2.3 Illinois Central Spring PCB Loading Analysis

The substantial data collected at ICS over the years has allowed detailed analysis of the PCB release. The analysis has focused on three issues: 1 ) how much PCB is released at the spring in a year, 2) how much of the release is associated with non-storm versus storm conditions, and 3) what is the relationship between storm flows and the timing of the peak PCB period during a storm event.

Correlations between flow and PCB concentrations at ICS were developed, as discussed in Section 1.5.2.1, during non-storm conditions (Reference 54). Additionally, the amount of PCBs released during storm periods has been shown to be related to the size of the storm, as discussed in Section 1.5.2.2. These correlations, along with the substantial historical flow records available, allow the estimation of the total loading to the Clear Creek system from ICS.

In order to perform the analysis, a complete record for hourly flow at ICS was first constructed. As noted in Section 1.5.2.2, Viacom maintained a weir just downstream of ICS from 1995 to mid-2000. Viacom selected 1996 as the year to construct total PCB discharge because this was a fairly wet year and flow records were mostly complete. Any holes in the flow records were filled by using a correlation developed between MOO-6 water level and ICS flow (Reference 54).

After constructing the complete flow record, the correlations between PCB concentration and flow developed for the non-storm and receding limb flow conditions was applied against the flow data. This yielded PCB concentrations at the spring for all but the peak periods during storms. To complete the analysis, a correlation between peak storm flow and PCB mass released during the peak period was developed.

The analysis resulted in an estimated discharge of 6.99 kilograms of PCBs for 1996. Of this 7 kg, the breakdown by flow condition was as follows:

As can be seen, most of the estimated PCB annual discharge is associated with periods of high flow, specifically the receding limb periods of storms that generate and sustain high flow for longer periods of time. The peak storm periods, while generating high relative concentrations of PCBs, are short lived and not the highest mass contributor on an annual basis. These relationships can be updated based on later data and the mass calculated for any time period for which flow and/or MW-6 level records exist.

To determine the relationship between the peak PCB periods during storm events and the flow, the data from all the storms monitored for PCBs and flow were consolidated into a set of corresponding travel times and flows. The travel times were taken as the time from the first flow response at ICS during a storm until the time the first significant rise in PCBs occurs at the spring during that storm. The flow is the average flow during the time period from the first flow response to the arrival of the rise in PCBs. The travel time is then plotted against average flow. The data is shown in Table 14 and the plot is shown in Figure 26. As can be seen, there is a good correlation between average flow and travel time. The literature ~ reference 37) suggests that this indicates the PCBs are traveling from the site to the spring mostly in a phreatic pathway.

1.5.3 Stream Water Sampling

The ICS/QS System forms the headwaters of Clear Creek. This system serves as the major discharge of site related Groundwater. The Slaughterhouse Spring System forms the headwaters of Stout's Creek. This system has shown much smaller amounts of PCBs. Stout's Creek is also adjacent to the Bennett's Dump PCB site and PCB data from Stout's Creek is discussed in the monitoring plan for that site (Reference 38). This plan will discuss the PCB data from Clear Creek.

Clear Creek flows through Bloomington past the Winston Thomas Wastewater Treatment Plant and ultimately flows into Salt Creek. Figures 27-29 show Clear Creek from its headwaters to Salt Creek. Table 15 lists all the known stream surface water sample results from the upper ICS/QS branch. The purpose of Table 15 is to list comparable PCB data along the ICS/QS stream. For comparison purposes, Table 15 includes the ICS emergence data at Location 1 for sampling events that also obtained PCB data further downstream. Figure 27 shows Location 1 as the ICS emergence or as the stream immediately downstream of the emergence. Conversely, Table 10 lists all the spring emergence data itself, for ICS, Quarry B. and Slaughterhouse Springs. The February 26, 1982 data for Location 1 along the ICS is shown in Table 15 without any downstream surface water data. The PCB result of 18 ppb was obtained from CBU and the exact location of the sample (i.e. emergence or immediately downstream) is not known.

Table 15 also lists Quarry A, B and C spring emergence PCB data (locations 6, 7 and 10, respectively, on Figure 27) for comparison to downstream surface water PCB data. The presence of PCB in the Quarry C branch and emergence is unexplained. Tracer tests have shown that Quarry C is not connected to the underground drainage system from Lemon Lane Landfill or to the ICS swallow hole. A possible explanation for PCBs in Quarry C is that the PCB arise from back flooding of Quarry A and B during heavy storms when the whole Quarry Springs area floods.

Table 16 shows all the PCB results for samples taken from lower ICS/QS and Upper Clear Creek. Table 17 shows PCB results further down in Clear Creek and in Salt Creek. Figures 27, 28 and 29 show the locations of samples listed in Tables 15, 16 and 17, respectively.

The earliest stream sampling was performed by the state of Indiana (performed concurrent with fish sampling) and IU in Clear Creek. Results from Clear Creek at Country Club Road, upstream of the Winston Thomas (WT) Wastewater Treatment Plant and at Gordon Pike, downstream of WT, showed a significant increase in PCB levels as Clear Creek flowed past the WT plant. PCB levels in Clear Creek at Gordon Pike were up to 69 ppb in 1977 and 1978, while the Westinghouse plant was emitting PCBs to WT (Reference 39). PCB levels in surface water of Clear Creek at Gordon Pike were measured at 6.4 ug/L in December 1980, while WT was still operating, but the PCB emissions from the Westinghouse plant had been stopped (Reference 40).

Sampling of surface water by IU on August 2,1978, showed PCB levels in ICS/QS (Spanker's Branch) at 7 ug/L. PCB levels in the confluence of ICS/QS and Clear Creek were 3 ug/L (Reference 41).

PCB surface water sampling detected PCBs at 14 of 26 locations sampled TV the USEPA in November 1991 (Reference 42). PCB concentrations were predominantly ,~ citified as Aroclor 1248, with a maximum concentration of 10.0 ppb from a sample collected 150 feet upstream of ICS swallow hole.

In November 1994, Normandeau Associates sampled surface water along ICS/QS as part of an ecological assessment for Viacom. Four samples were taken along ICS/QS and three samples were taken in Clear Creek. Sampling locations are shown in Figures 27 and 28. Only two of the samples had detectable levels of PCB above the detection limit of 2 ug/L. Sample ICS/QS-1 was taken immediately downstream of the ICS emergence and was found to have a PCB content of 28.5 ug/L. Sample ICS/QS-2 was taken 100 feet upstream of the swallow hole, directly downstream of the culvert under the railroad tracks leading from the ICS emergence. This sample was analyzed at 18.0 ug/L. The remaining two ICS/QS samples and all three Clear Creek samples were <2 ug/L.

In November 1996, Normandeau Associates again sampled surface water along ICS/QS as part of an ecological assessment / fish sampling event for Viacom. Surface water was sampled from six locations shown on Figures 28 and 29. The sample locations were in the ICS/QS and Clear Creek, ranging from 1.3 to 20 miles downstream of the LL site. The water sample taken at Location 1 in ICS/QS north of Allen Street was analyzed at 0.21 ppb, total PCBs. The samples taken at the other five locations, further downstream in Clear Creek, were all <0.1 ppb.

In November 1999 and again in Novvember 2000 Viaccm performed surface water sampling at three locations on Clear Creek. Samples were collected at the bridges that cross Clear Creek at Strain Ridge Road, Fluckmill Road and Country Club Road. All these results were less than 0.1 ppb.

Overall, the historical surface water data shows that while the WT plant was operating, it was the largest source of PCBs to Clear Creek. After the facility was shutdown in 1982, the PCB levels in Clear Creek downstream of the WT plant are an order of magnitude or more lower. Upstream of the WT plant, the most data exists for the Country Club Road area. The data at that station also shows a substantial decline from the late 1970s to the late 1990s.

In order to facilitate the trending of data, it would be helpful to have flow, conductivity and TSS data concurrent with the PCBs. It is likely that PCBs in this stream are also a function of flow conditions, just as it is known that PCBs out of ICS are very dependent on flow conditions.

1.5.3.1 PCB Mass Flow in Streams

An investigation of the PCB mass flow in ICSIQS/CC stream waters was performed in November 1997 to investigate the possibility of additional PCB sources contributing to Clear Creek along its course through Bloomington and beyond. The mass flows of PCBs in the water at several points along ICS/QS/CC were calculated.

To determine the mass flow of PCBs at any point in the stream the water flow was measured in gallons per minute and the PCB concentration of the water was determined in ppb. Multiplying these two values and converting units, results in the mass of PCBs passing that point during a set period of time. Table 18 shows the flow measurements and PCB concentrations for eight locations from the ICS emergence to 16 miles downstream at Ketchum Road. The calculated mass of PCB at each location is also shown. Figure 30 plots the calculated values of PCB mass flow in Clear Creek from the ICS emergence to Ketchum Road.

Ideally, if only one source of PCBs exists, as additional tributaries of "clean water" enter along the creek path, the concentration and mass of PCB should decrease both because of dilution and PCB mass loss processes such as sedimentation and volatilization. Figure 30 shows that the mass flow of PCBs decreases significantly from the ICS emergence down to the confluence of ICS/QS with Clear Creek. This suggests that most of the PCBs volatilize or deposit out of the water along ICS/QS Branch. The measured flow in ICS/QS before the confluence with CC was 99 gpm and the PCB concentration was only 0.035 ppb. In Clear Creek, upstream of the confluence with ICS/QS, a mass flow of PCBs was measured at the same magnitude as the PCBs entering Clear Creek from ICS. This suggests that an additional source of PCBs upstream of ICS contributes as much PCB to Clear Creek as ICS/QS does under these flow conditions.

No reduction in PCB mass flow is seen from the ICS/QS/CC confluence to Country Club Road. This could mean that no PCBs were lost along this length or it may mean that additional PCB is added to Clear Creek to make up for the PCB that leave the water column.

From Country Club Road, which is just upstream of the WT plant to Gordon Pike, which is just downstream of WT, an increase in PCB mass is shown. This would suggest that PCBs were being added to Clear Creek from the WT site in 1997. Furthermore, the mass flow of PCB continues to increase as Clear Creek travels down to Fluckmill Road and Ketchum Road.

As can be seen by comparing the magnitude of the mass flow of PCB at ICS/QS at the confluence with CC, to the mass flow beyond the WT site, even if the PCB emitted by the ICS emergence were stopped completely, PCB levels downstream of WT would not be significantly reduced.

This data is only one snap shot in time. This data should also be considered a gross estimate only since the PCB lab analyses were done with a non-approved method, and stream samples were done by hand dipping a bottle in the flow centroid. Additionally, flow measurements at two of the stations could not be reliably measured and were estirriated based on visual observations and stream characteristics, as stated in Table 18. For these conclusions to be strengthened, additional sampling under varying flow conditions with more accurate techniques would be required.

1.5.4 Stream Sediment Sampling

1.5.4.1 Sediment Sampling Events

Table 19 lists all the sediment sample results that have been associated with the LL Landfill, including sampling that was done in ICS/QS Branch, Clear Creek, Salt Creek and other area springs. Sediment sampling that was performed in Clear Creek and Salt Creek downstream of the WT site is also included. The WT plant site in Bloomington, contributed PCBs to Clear Creek prior to it's shutdown in 1982. In 1987, sediments were removed from Clear Creek from 25 feet upstream of WT to a point 500 feet downstream of WT, in accordance with the Consent Decree

The earliest sediment sampling alias done in 1976 by the Indiana State Board of Health (ISBH) in Clear Creek, directly upstream of WT and downstream to the Pleasant Run Creek confluence with Salt Creek, as shown in Figures 28 and 29. These samples showed the considerable contribution of PCBs to Clear Creek from WT. Additional sampling was done by the state in 1977 and 1980 downstream in Clear Creek and Salt Creek.

In October 1981, offsite sediment samples were collected by EPA at four locations: 1) the headwaters of Stout's Creek at the intersection of Vernal Pike and Woodyard Road, 2) Quarry Spring, 3) Stoney Springs East, and 4) Stoney Springs West. Quarry Spring was at 0.5 ppm. All other samples were analyzed for total PCBs and were below the detection limit of 0.01 ppm.

In February 1982, the City of Bloomington took the first known sediment sarr.ples from ICS. Five sediment samples from ICS area were analyzed. The highest PCB levels were 360 ppm from north of the railroad tracks and 65 ppm south of the railroad tracks. In June of 1983, E&E took sediment samples for the EPA from ICS, north and south of the railroad tracks. Sample results were 5.8 ppm of Aroclor 1248 north of the tracks and 2.5 ppm of Aroclor 1248 south of the tracks.

Sediment sampling of two indicator areas occurred in June and July of 1988, after the sediment removal in 1987. Ten places downstream of Gordon Pike and ten places downstream of Old Route 37 at Harrodsburg were sampled to determine if contaminated sediments had washed downstream during the sediment removal. All ten samples at Gordon Pike and nine out of ten samples at Harrodsburg were non-detect at <1.0 ppm. One sample at Harrodsburg was 1.2 ppm.

Sediment samples from the springs associated with LL Landfill were collected on three occasions in 1991 for PCB analyses. Thirteen sediment samples and one duplicate sample were collected by the IDEM for PCB analysis in June 1991. PCBs were detected in ten locations, with a maximum concentration of 22 mg/kg in a sample from ICS. In October 1991, sediment samples were collected by Viacom from seven locations and analyzed for PCBs. PCBs were detected in five of the samples, with a maximum concentration of 4.1 mg/kg reported from ICS.

In November 1991, 26 sediment samples and two duplicate samples were collected and analyzed for PCBs from the ICS emergence, downstream to Clear Creek at Gordon Pike, downstream of the WT plant. Sediments from the QS and ICS tributaries were included in this sampling. PCBs were detected in 18 of the samples, with a maximum concentration of 58 mg/kg reported from ICS. PCB levels had dropped to <1 ppm at the confluence of ICS/QS with Clear Creek. The sampling locations generally corresponded with the EPA's water samples taken during the same time period.

In November 1994, Normandeau Associates sampled sediments along ICS/QS for Viacom. PCBs were detected at 37.4 ppm near the emergence and at 165 ppm near the swallowhole. PCB levels dropped to <0.6 ppm at the confluence of ICS/QS with Clear Creek.

Based on these results additional sediment sampling was performed by Viacom on May 8, 1995, at ten feet increments from ICS swallow hole. PCB concentrations in the samples were found to be up to 1,100 ppm on a dry basis near the swallowhole.

Eight sediment samples were also taken by Viacom on June 15,1995, along QS. Results ranged form 5.1 to 10 ppm dry.

PCBs were detected at Slaughterhouse and Packinghouse Road Springs, as well as at ICS and QS during May of 1995. Packinghouse Culvert Spring was found to be BDL three times in 1995.

Viacom collected 45 sediment samples at nine peripheral springs at LL on August 31,1995. One sediment sample was collected at the spring emergence and four others were collected at selected locations farther downstream of each spring where sediment was observed to collect. Only one of 45 samples had a detectable level of PCBs over the detection limit of 0.1 ppm. The furthest sample downstream of Stoney West was 0.67 ppm. A duplicate sample for this location was BDL. A similar low level detection was also obtained on an EarthTech split sample at Urban Spring.

In October 1995, Viacom collected sediments at six connected springs as part of the storm event sampling under the Lemon Lane Field Sampling Plan. PCBs ranged from BDL to 500 ppm at ICS.

On November 28,1995, Viacom sampled a sediment core in an area of sediment deposition, about ten feet from the swallowhole. The core was taken from 0 to 16 inches deep. PCB content was found to be up to 130 ppm at 6-7 inches deep, and 11 ppm at the bottom of the core at 15-16 inches deep. Sediment samples were also taken in a karst window about 50 feet northwest of the emergence. PCB content was 3.8 and 8.8 ppm.

In November 1996, as part of a Clear Creek ecological investigation, Viacom collected sediment samples as well as fish samples at six locations along ICS/QS and Clear Creek. Sample locations ranged from 1.3 miles from te Landfill in ICS/QS at Allen Street and progressed downstream to 20 miles from the site in Clear Creek at Strain Ridge Road. Multiple samples were taken at each sample station. Total Organic Carbon (TOC) was also determined for the samples to allow the PCB content to be normalized for TOC content. The highest PCB content was at the upstream most sampling station in ICS/QS at Allen Street where PCB samples averaged 1.63 ppm. Samples in Clear Creek had a fairly consistent PCB content from 0.48 ppm at the confluence with ICS/QS, 2.5 miles from the site, to 0.35 ppm at Strain Ridge Road, 20 miles from LL.

In November 1999, Viacom again performed fish and sediment sampling in Clear Creek. As part of this study, Viacom collected sediment samples at three sites in Clear Creek. The sites were at Strain Ridge Road, Fluckmill Road, and Country Club Road. In November 2000, Viacom repeated this sediment sampling. All results were <1.0 ppm.

1.5.4.2 Sediment Data Analysis

The earliest sediment sampling by ISBH in 1976 showed considerable PCB content in Clear Creek sediments largely because of Winston Thomas rather than LL Landfill. Sediments sampled in Clear Creek, 180 feet upstream of WT were analyzed at 1.13 ppm PCBs. However, at 50 feet downstream of WT sediments were found to contain 111 ppm of PCBs. Sediment PCB levels remained at up to 27.3 ppm in Clear Creek for more than six miles downstream of the WT plant. After WT was shutdown in 1982, PCB content in sediments declined drastically. By 1984, before the 1987 sediment removal, sediment PCB contents were less than 1 ppm directly downstream of the WT site, at Gordon Pike, and all along Clear Creek to Hobart Road. Comparing the PCB content in Clear Creek from 1984 to the latest sampling in 2000 shows similar PCB levels at less than 1 ppm.

PCB sediment samples were first taken in the ICS/QS stream in 1982 by the City of Bloomington. PCB levels up to 360 ppm were obtained from sediments in the emergence area at that time. Viacom obtained sediment PCB contents up to 1,400 ppm at the ICS emergence in May of 1995.

Sediment sampling by the state in November 1991, and Viacom in 1994, showed that even though sediments in ICS/QS had high PCB values near the emergence and swallowhole (up to 165 ppm), levels decreased rapidly at the downstream end of ICS/QS (2.05, 0.6 ppm) where it flows into Clear Creek. During both sampling events, PCB levels in sediments were less than 0.6 ppm in Clear Creek itself.

Also, both the 1991 and 1994 sampling results show that ICS/QS may not contribute much to sediment PCB levels in Clear Creek. In 1991 a sediment sample upstream of the confluence with ICS/QS had a PCB content of 0.2 ppm. Directly downstream of the confluence with ICS/QS the PCB content only rose to 0.29 ppm. Similar results were obtained in November 1994, where sediments had a PCB content of 0.51 in Clear Creek upstream of the confluence with ICS/QS and only 0.52 directly downstream of the confluence.

In ICS/QS, PCB content drastically decreases with distance from the site. However, for Clear Creek sediment PCB content stays fairly constant at <1 ppm from upstream of the confluence with ICS/QS at 2.2 miles from LL down to Strain Ridge Road, 20 miles from LL.

Since 1994, all sediment samples by Viacom have also been analyzed for total organic carbon (TOC). PCBs would typically be sorbed on the organic carbon fraction of sediments (Reference 43). Therefore it is helpful for interpreting trends to have concurrent TOC values and to normalize the PCB content of sediments with TOC. This can remove some of the inherent variability caused by extreme differences in organic content at a particular location in a stream.

For example, if at a particular location in a stream there is both fine organic rich sediment and sandy/gravelly sediments, it would be expected that the fine organic materials would have a higher PCB content than the sandy material. Obviously, if in one year samples of the sandy material were taken and another year the fine organic materials were taken, then the large differences seen in PCBs may suggest a trend when in fact none may exist.

Table 20 shows the PCB and TOC data and normalized PCBs. The normalizing of PCBs does not change the interpretation of the sediment data discussed above.

1.5.5 Pond Sampling

In June 1981, offsite surface water samples were collected by the City of Bloomington Utilities in the LL Landfill area, and were analyzed for PCBs (Reference 44). Two surface water samples were collected from private ponds on the north and northwest side of the site, plus a surface water sample from Hensenberg Creek. The sampling event was conducted in relation to the residential well samples collected in June 1981. All sample results showed non- detectable levels of PCB. The analytical detection limit is unknown.

On July 1, 1981, an offsite surface water sample was collected by EPA in the LL Landfill area, and was analyzed for PCBs. As part of an EPA soil sampling event occurring on June 30 and July 1,1981, one surface water sample (Reference 45) was collected from a private pond (Sargent's Pond), 25 yards west of the northwest corner of the site, along with one blank sample. PCBs were detected at 0.9 ppb in the pond water.

Sargent's pond was sampled as part of the Interim Monitoring Program during the 2000 remediation. PCB levels varied from BDL on February 23, 2000, before remediation began, to 2.9 ppb on October 19, 2000, near the end of the remediation. PCB levels have declined since the completion of the cap, as shown in Table 21.

1.5.6 Fish Sampling

There has been a considerable amount of fish samples taken in creeks and streams downstream of the LL Landfill and analyzed for PCBs. The fish samples taken were done by several organizations with varying purposes and methods. Some of the data was taken to document PCB in biota downstream of the site to support litigation. Others were taken to guide health advisories for fish consumption. Still others were taken to support potential risk assessment activities. Organizations that have conducted fish sampling include the USEPA, USFWS, Viacom, ISBH and IDEM.

1.5.6.1 Factors Affecting Fish Sample Results

For a long term monitoring plan, data should be collected to determin