104
HYDROGEOLOGIC STUDY EAST GRAY, MAINE for the Maine Department of Environmental Protection by Robert G. Gerber, Inc. Consulting Civil Engineer & Geologist South Harpswell, Maine 29 November 1982 N o

HYDROGEOLOGIC STUDY EAST GRAY, MAINE for the Maine ...and Mr Jame. Ks. Richard wh,o did th lineamene t analysi and s assisted i mappinn g surficia All geology memberl ofs th. e project

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

  • HYDROGEOLOGIC STUDY

    EAST GRAY, MAINE

    for the

    Maine Department of Environmental Protection

    by

    Robert G. Gerber, Inc.

    Consulting Civil Engineer & Geologist

    South Harpswell, Maine

    29 November 1982

    N

    o

  • ROBERT G. GERBER, INC.

    R. F. O. 1. BOX 483 ' • SOUTH HARPBWELL. MAINE O4O79

    2O7-833-6334

    29 November 1982

    Mr. John Krueger, Director

    Div. of Licensing and Enforcement

    Bureau of Hazardous Materials Control

    State House Station 17

    Augusta, Maine 04333

    Re: Transmittal of Final Report—Hydrogeologic Study, East Gray

    Dear Mr. Krueger:

    In accordance with my contract with the Department of

    Environmental Protection (DEP), I have completed my study of the

    migration of chemical contaminants in the ground water from the

    McKin Chemical Company site in East Gray, Maine. The attached

    report contains the summary of my findings.

    The project team consisted of the following: Robert G.

    Gerber who mapped surficial geology, conducted the computer

    analyses, and managed the project; John R. Rand who mapped

    bedrock, analyzed historical data, assisted in computer data

    preparation, and constructed the report figures; Weston

    Geophysical Corp., who performed the geophysical investigations;

    and Mr. James K. Richard, who did the lineament analysis and

    assisted in mapping surficial geology. All members of the

    project team are Certified Geologists in Maine. Mr. Gerber is

    also a Registered Professional Engineer in Maine.

    Through our field mapping, aerial photo interpretation, data

    analyses, borings, seismic refraction profiling, water sampling

    and analyses by DEP, and computer modelling, we find the following:

    1. The McKin Chemical Company processed oily wastes and other industrial wastes at the site in East Gray from at least

    1972 to 1977. As early as 1973, local residents began to

    complain of odor and peculiar tastes in their well water.

    Inilfial tests found organic chemicals such as trichloroethane

    (TCÊ ancT'̂ trTchloroethylene (TCEy) present in the domestic

    drilled water wells north of the chemical company. Studies took

    place during the period 1977 to 1980, which concluded that the

    ground water contamination had probably originated at the McKin

    Chemical Company site. In 1980 the DEP hired a contractor to

    remove 35,000 gallons of chemicals from the site, but more

    remain. In August 1982, my consulting firm was retained to do a

    more thorough investigation of the hydrogeology of the East Gray

    area.

    2. The area in which the ground water has been contaminated

    lies in the watershed of the Royal River. An important point of

  • Page 2 of 4, Krueger, 11/29/82

    Hydrogeologic Study, East Gray

    reference is the area of the "Boiling Springs" and the Maine

    Central Railroad bridge which lie about 2000' south of Gray

    Station and 4000' east of the McKin site. The drainage area of

    the Royal River above this site is 69.8 square miles. All ground

    water beneath the McKin site that flows in both the surficial and

    bedrock aquifers ultimately discharges into the Royal River. No

    surface or ground water moves southwesterly from the site toward

    Gray Meadow and the Meadow Brook drainage.

    3. There are three major soil units under and near the

    McKin site. Stratified sand and gravel covers an area north and

    west of the site. Glacial till underlies the sand almost

    everywhere throughout the study area and is exposed on upland

    areas to the west and south of the site and to the east of the

    Royal River. Glaciomarine fine sands, silts, and clay-silts

    overlie the sand and till along the steep slopes that lie

    adjacent to the Royal River and Collyer Brook. The clayey

    glaciomarine deposits have presented a barrier to ground water

    flow from the more permeable sand and gravel deposits. Springs

    and seeps occur where this ground water flowing from the sand and

    gravel escapes to the surface through the clayey deposits.

    4. The topography of the bedrock surface in the site area

    slopes generally down to the east, toward the Royal River. In

    detail, the bedrock surface forms a broad northeast-trending

    trough to the west of the site which passes easterly between an

    east-west ridge just south of Collyer Brook and the end of a

    narrow ridge which projects north-northeasterly from the site.

    The bedrock surface rises to the south from the site. This

    peculiar topography of the bedrock surface is a significant

    factor in controlling the initial movement of ground water toward

    the north, away from the site.

    5. The bedrock beneath the site is inferred from regional

    mapping to be composed of granite, pegmatite, or schistose

    migmatite, possibly transected locally by thin, tabular mafic

    dikes. Although these bedrock types are in themselves

    essentially impervious, they are capable of transmitting ground

    water along narrow partings where the rock has been broken by

    joints, cut by tabular dike intrusions, or, in the schistose

    migmatite, separated along micaceous layers. The preferred

    avenues for ground water movement in the bedrock appear to be

    toward the northwest along joint openings and toward the

    north-northeast along tabular dikes, with a relatively minor

    increment of ground water flow toward the east-northeast along

    micaceous partings in the migmatite.

    6. Ground water leaving the site in the surficial aquifer

    flows to the northwest initially, until encountering a coarse

    gravel deposit that trends east-west under the intersection of

    Depot and Mayall Roads. After merging with flow in this gravel

  • Page 3 of 4, Krueger, 11/29/82

    Hydrogeologic Study, East Gray

    deposit, ground water flows eastward and is discharged in springs

    and small streams that form part of a local surface drainage

    network that enters the Royal River adjacent to the Boiling

    Springs. The coarse gravel deposits lie in contact with or lie

    just beneath the Royal River between Boiling Springs and the

    Maine Central Railroad Bridge. Most of the ground water that

    originates at the site is discharged to the Royal River in the

    immediate vicinity of the Boiling Springs. The total ground

    water discharge rate in this area is about 800 gallons per

    minute. There is no evidence that any of the site ground water

    continues south of the Maine Central Railroad bridge, along the

    trend of the Royal River.

    7. Ground water that leaves the site in the bedrock aquifer

    flows toward the north initially, then spreads out in a diffuse

    manner to turn slowly eastward toward the Royal River. Most of

    the bedrock ground water that originates at the site is

    discharged to the overlying gravel deposits in the vicinity of

    Boiling Springs.

    8. Through computerized ground water simulations, we have

    been able to identify the approximate path and magnitude of

    concentration of the chemical TCE within both the surficial and

    bedrock aquifers. Our simulated distributions of the contaminant

    concentration are in good agreement with measured concentrations

    from water samples taken in the site area. The path of the

    contaminant plume within the surficial aquifer is relatively

    narrow and well-defined. Contamination entering the site soils

    would reach the Boiling Springs after about 5 years. Ten years

    after introduction of the contaminant at the site at an assumed

    average rate of about 26 gallons per year, the concentration of

    TCE would begin to reach a steady state condition at the Boiling

    Springs. Although the time history of chemical leakage into the

    aquifer beneath the site is not known, it may have been at rates

    as high as 125 gallons per year during the initial period of

    contamination. The remaining volume of chemicals contained in

    the soils under the site cannot be calculated from the available

    data.

    9. The distribution of contaminants in the bedrock aquifer

    is of greater extent than in the surficial aquifer. Contaminant

    is transferred to the bedrock aquifer from the surficial aquifer.

    The distribution of contaminants in the bedrock aquifer is

    controlled largely by the ground water flow pattern in the

    bedrock and the locations of the overlying contaminated surficial

    ground water. The presence of pumping wells in the bedrock

    aquifer will also affect the contaminant distribution and we

    infer that the domestic wells that were in use during the 1970's

    north of the site caused much higher concentrations of

    contaminants to enter the bedrock aquifer in the local area than

    would have otherwise been the case.

  • Page 4 of 4, Krueger, 11/29/82

    Hydrogeologic Study, East Gray

    10. Although concentrations of TCEy may be increasing in

    some wells, we believe this is due to a "retardation" effect

    peculiar to that chemical. Concentrations of TCE are generally

    stable or beginning to decrease at most sampling locations. We

    do not expect much further increase in contaminant levels in the

    surficial aquifer. Certain areas of the bedrock aquifer may show

    somewhat increased contaminant levels in the future— perhaps on

    the order of 10-20%.

    11. If the site is isolated in some way so that no

    additional chemicals can leak into the aquifer under the site,

    concentration levels will begin to decline in both aquifers after

    several years. After five years there will be major reductions

    in contaminant concentrations in the aquifers. After 10 to 12

    years most of both aquifers will have been reduced to less than

    30 parts per billion of TCE in most of the surficial aquifer, and

    less than 15 parts per billion in the bedrock aquifer. The rate

    of aquifer cleansing could be increased by a combination of

    ground water withdrawal wells and injection wells in the

    surficial aquifer. However, any contaminated ground water that

    is pumped from the aquifer in such a scheme would have to undergo

    costly treatment. It is our understanding that consultants hired

    by the Environmental Protection Agency will evaluate the costs

    and benefits of any site clean-up work.

    It has been a pleasure to have been of service to the State

    of Maine.

    Respectfully submitted,

    \ \%T.fV̂

    Robert G. Gerber ,

  • HYDROGEOLOGIC STUDY

    EAST GRAY, MAINE

    for the

    Maine Department of Environmental Protection

    by

    Robert G. Gerber, Inc.

    Consulting Civil Engineer & Geologist

    South Harpswell, Maine

    29 November 1982

  • TABLE OF CONTENTS

    Section Page

    INTRODUCTION 1

    HISTORICAL BACKGROUND 1

    METHODS 2

    SITE AREA GEOLOGY 5

    PHYSIOGRAPHY AND WATERSHEDS 5

    BEDROCK GEOLOGY 6

    Introduction . . . . . . . . . . . 6

    Bedrock Lithologies 7

    Bedrock Structure . . . . . 7

    Faults 8

    Joints, Dikes, and Foliation . . . . . . 8

    Lineament Analysis . . . . . . . . . 9

    SURFICIAL GEOLOGY 10

    Soil Thickness 11

    Soil Types and Distribution 11

    Significance of Surficial Geology to Contaminant Spread . 14

    HYDROGEOLOGIC ANALYSIS 14

    METHODS 15

    Delineation of Aquifer Boundaries 16

    Bottom of Surficial Aquifer 18

    Permeability and Transmissivity . . . . . . . 18

    Storativity and Specific Yield . . . . . . . 20

    Dispersivity . . . . . 2 0

    Pollutant Attenuation by Retardation and other Mechanisms 21

    Aquifer Inputs and Withdrawals of Water . . . . . 22

    Application of Contaminant to the Aquifer . . . . 23

    INTERPRETATION OF COMPUTER SIMULATION RESULTS . . . 25

    Ground Water Flow in the Surficial Aquifer . . . . 25

    Bedrock Aquifer Flow 26

    Contaminant Transport Modelling—Surficial Aquifer . . 27

    Contaminant Transport Modelling—Bedrock Aquifer . . . 28

    THE FUTURE OF THE CONTAMINANT PLUME 29

    MAINTAINING THE STATUS QUO 29

    PREVENTING CHEMICALS FROM LEAVING THE SITE . . . . 30

    GROUND WATER REMOVAL OR FLUSHING BY PUMPING . . . . 32

    SUMMARY AND CONCLUSIONS 32

    LIST OF REFERENCES 33

    Table 1—1982 Water Sample Locations and Test Results

  • TABLE-OF CONTENTS (continued)

    List of Figures

    Figure 1—Site Area Wells, Springs and Seismic Survey Locations

    Figure 2—Site Area—Geologic iMap

    Figure 3—Site Vicinity—Inferred Bedrock Elevation

    Figure 4—Site Vicinity—Inferred Elevation, Surficial Water

    Table

    Figure 5—Site Area Computer Simulation—Elevation of the

    Potentiometric Surface (Bedrock Aquifer)

    Figure 6—Site Vicinity—Computer Nets—Surficial

    Figure 7—Site Area Computer Nets—Bedrock

    Figure 8—Site Area—Historical Test Data: 1,1,1-Trichloroethane

    and Trichloroethylene

    Figure 9—Computer Simulation—Contaminant Distribution in

    Surficial Aquifer

    Figure 10—Computer Simulation—Contaminant Distribution in

    Bedrock Aquifer

    Figure 11—Computer Simulations—Contaminant Distribution in

    Surficial Aquifer after Isolating McKin Site:

    after 2, 5, and 10 years

    Figure 12—Computer Simulations—Contaminant Distributions in

    Bedrock Aquifer after Isolating McKin Site:

    after 2, 5, and 10 years

    List of Appendices

    Appendix A—Water Supply Questionnaire

    Appendix B—Well Data Summary and Analysis

    Appendix C—Weston Geophysical Corp. Seismic Refraction Survey

    Appendix D—Borings Logs and Sample Descriptions

    -11

  • HYDROGEOLOGIC STUDY

    EAST GRAY, MAINE

    INTRODUCTION

    Hydrogeology is the study of movement of water through soil

    and bedrock. This report contains a summary of the findings of a

    study conducted from August through November 1982 by Robert G.

    Gerber, Inc., on the hydrogeology of the East Gray area of Maine.

    The study was done under contract with the Maine Department of

    Environmental Protection (DEP) using funds appropriated by the

    Maine Legislature expressly for this purpose. The general objec

    tive of this study was to learn more about the extent of ground

    water contamination in the East Gray area that has been alleged

    to have been caused by chemicals leaking from the McKin Chemical

    Company site on Mayall Road.

    More specifically, the study objectives as defined in the

    contract with the DEP included:

    a) to place four new monitoring wells

    b) to undertake geophysical investigations to study the

    geologic profile in the area around the McKin site

    c) to map the bedrock and surficial geology

    d) to describe background surface and ground water

    quality

    e) to model the ground water aquifer using a specific

    computer model

    f) to meet with the DEP staff and public as the study

    progresses to keep them informed

    g) to prepare a final report and conclusions concerning

    the extent of the ground water contamination.

    HISTORICAL BACKGROUND

    The McKin Chemical Company processed oily waste and clean-up

    debris that was generated from a 100,000-gallon oil spill from

    the tanker "Tamano" in Casco Bay in 1972. The East Gray site was

    -1

  • also used for processing other industrial wastes and the McKin

    Chemical Company was in operation at the site for several years

    prior to 1972 (Gardner Hunt, pers. com.) and operated until 1977.

    Although the manner and locations in which chemicals entered the

    ground is not completely known, it is apparent that certain

    chlorinated hydrocarbons and other chemicals did enter the ground

    and pass into the ground water beneath the site.

    As early as 1973 local residents began to complain of odor

    and peculiar tastes in their well water. Initial tests found

    that dimethyl sulfide and chlorinated hydrocarbons such as

    trichloroethane (TCE) and trichloroethylene (TCEy) were present

    in domestic drilled water wells north of the site. Hart

    Associates was retained by the Environmental Protection Agency

    (EPA) and DEP to conduct a hydrogeologic study of the area to lo

    cate the source and extent of the contamination. Their report

    was submitted in November 1977. The DEP installed borings and

    monitoring wells on and near the McKin site in 1979 and 1980. A

    short interpretative report was prepared for the DEP in 1980 by

    BCI Geonetics. About 35,000 gallons of waste that had been

    stored at the site was removed by DEP-hired contractors in 1980.

    This was not the complete inventory of chemicals that had been de

    termined at the site, and some chemicals are still stored there.

    In June 1981, another report was prepared by Ecology and

    Environment, Inc., under contract to EPA. Except for the BCI

    Geonetics report, most of the other work was confined to sampling

    of local wells and surface waters and determining the magnitude

    and extent of the contamination. No detailed geologic studies

    were done in the area until this present study of 1982.

    METHODS

    Most of the methods that were used in this study were dictat

    ed by the DEP contract. The initial part of the study involved

    synthesis of the mass of data in the DEP files and collection of

    new data. John R. Rand went through the entire DEP file and sum

    -2

  • marized the important data pertaining to the hydrogeology, as

    presented on the Tables and Figures of this report. New data

    collection on water wells was accomplished by mailing a survey

    form (Appendix A) to all property owners of record in the East

    Gray area. About 33% of the questionnaires were returned,

    although many of the respondents indicated that there were no

    wells on their property. The results are tabulated in Appendix

    B. Well locations are shown on Figure 1. Local residents were

    very cooperative during this study and we are particularly

    appreciative for the cooperation of Portland Sand and Gravel,

    Blue Rock Industries, Mr. Frederick Farrell, Mr. Emerson

    Mitchell, and Ms. Lucymae Bowles.

    Geologic data collection included 8800 lineal feet of seis

    mic refraction profiling by Weston Geophysical Corp. at the loca

    tions shown on Figure 1 and described in Appendix C. During the

    geophysical field investigations, James Richard and Melissa

    Whitaker assisted Weston Geophysical. Five new borings were made

    (two borings made side-by-side at location B101) in which ground

    water monitoring wells were installed. The locations of these

    borings are shown on Figure 1 and the drillers logs and our inter

    pretive logs are given in Appendix D. Carol White provided the

    field inspection of the borings done by Henry Michaud & Son.

    Robert Gerber described the soil samples. Locations and eleva

    tions of all geologic data points were located as accurately as

    possible by reference to a photo-enlarged 1980 10-foot contour

    map of the area. Limited funds did not permit accurate

    surveying.

    Bedrock field mapping was conducted by John R. Rand. James

    K. Richard analyzed lineaments on topographic maps, aerial

    photos, and Landsat imagery. Robert G. Gerber, James K. Richard,

    Peter Garrett (DEP), and Carol White mapped the surficial geology

    of the area. The bedrock geology, as interpreted by John Rand is

    given on Figure 2. The generalized surficial geology, as inter

    preted by Robert Gerber is also given on Figure 2. John Rand pre

    -3

  • pared the bedrock contour map on Figure 3. Figure 7 contains a

    rose diagram that indicates the proportions of bedrock features

    determined from field mapping that trend in various directions.

    Robert G. Gerber analyzed the ground water regimes and pre

    pared the computer models that simulated the ground water flow

    and contaminant spread. The methodology used for the computer

    analysis is described in more detail later in the report. The

    computer grids are given on Figure 6 (surficial aquifer) and

    Figure 7 (bedrock aquifer). John Rand contoured existing data

    from the surficial aquifer to produce the water table contour map

    in Figure 4. Figure 5 is the map of the potentiometric contours

    for the bedrock aquifer as generated by the computer analysis.

    Figures 9 and 10 show the computer-simulated distribution of TCE

    in the surficial and bedrock aquifers, respectively, for the 1982

    condition. Figures 11 and 12 show how the distribution of TCE in

    the surficial and bedrock aquifers will change with time if cer

    tain actions are taken to remove or isolate the chemicals under

    the McKin site.

    Water samples were taken for analysis of organic chemicals

    by the DEP laboratory and Energy Resources Company of Cambridge,

    Massachusetts. Field collection of water samples was done by

    Carol White, Peter Garrett (DEP), Edward Logue (DEP), Andrews

    Tolman (Maine Geol. Survey), Dorothy Tepper (US Geol. Survey),

    and John Williams (DEP). A graphic presentation of historical wa

    ter quality results through time at selected locations is present

    ed on Figure 8. Table 1 summarizes the results of the fall 1982

    water quality sampling. Sampling locations are shown on Figure

    1.

    Many discussions have been held with the DEP during the

    course of this study in order to keep them informed and to ex

    change data. The DEP has participated in several parts of the

    study and supplied information for this report. Several meetings

    were held in Gray to inform the Town of the progress of the

    study. After the contract award, an initial meeting was held to

    -4

  • discuss the scope of work, to exchange information, and to solic

    it help from the Town's people in the data collection effort.

    Richard Day, Gray Town Codes Enforcement Officer, and the Greater

    Portland Council of Governments were helpful in this regard and

    provided tax maps and local newspaper announcements. At the com

    pletion of the geologic field data collection, another meeting

    was held with the Town to discuss the interim findings. On 13

    December 1982 a final meeting will be held with the public to

    present the findings of this study.

    All graphics for the report and public presentations were

    prepared by John Rand.

    SITE AREA GEOLOGY

    PHYSIOGRAPHY AND WATERSHEDS

    The East Gray area has a very heterogeneous physiography

    (land surface form). Within the triangle of Depot Road, Mayall

    Road and Route 115 the ground surface is relatively flat or gent

    ly undulating on a sandy surface. Between Mayall Road and the

    Royal River and between Mayall Road and Collyer Brook, the

    gently-sloping sandy surface terrain gives way to closely-spaced

    gullies cut into silty and clayey soils. West of the site and

    west of Depot Road, the ground surface is deeply pitted with

    depressions formed on an irregular sand and gravel surface. Far

    to the south and to the west of the site the terrain rises onto

    moderately-sloping hillsides developed in a stony soil which has

    a moderately dense surface drainage pattern. This latter soil is

    known as glacial till and has a moderately low permeability (the

    measure of the rate at which water will pass through the soil).

    The deeply-pitted sand and gravel terrain is called "ice contact"

    terrain and was formed when swiftly-moving rivers flowed over and

    through glacial ice-covered terrain about 13,000 years ago. The

    flat sandy areas are ice contact deltas or glacial outwash that

    was deposited when the glacial meltwaters flowed into the sea

    which was about 300 feet higher then than it is now. The clayey,

    -5

  • highly dissected river banks are formed in what are called

    glaciomarine deposits. These soils consist of interbedded fine

    sands, silts and clay. Because the silts and clays have a

    relatively low permeability, most precipitation runs off the

    soils instead of infiltrating to become ground water. Therefore,

    the drainage network is much more dense in these soils than in

    more sandy soils.

    Although there is no direct surface runoff at the site, the

    site is included within the surface watershed of the Royal River.

    The watershed area of the Royal River above Brickyard Station,

    several miles southeast of the site, is 73.6 square miles. The

    watershed of Royal River at the railroad bridge, east of Boiling

    Springs, is 69.8 square miles. The watershed of Collyer Brook

    above its confluence with the Royal River, is 18.6 square miles.

    Neither surface water nor ground water from the site moves

    southwesterly toward Gray Meadow.

    BEDROCK GEOLOGY

    Introduction

    The study of the bedrock (ledge) geology has been made

    through a combination of literature search, field mapping, remote

    sensing techniques, analysis of water well data and by reference

    to analogous studies. The exact nature of the bedrock under the

    site is not known since it is covered by 36' to 65' of soil and

    no core borings have been made at the site. Conditions at the

    site and in other parts of the study area where no bedrock is ex

    posed are interpreted using standard geologic techniques of infer

    ence and extrapolation. Figure 2 defines the locations of

    bedrock outcrop that were examined as part of this study.

    Within one mile from the site, only two areas of bedrock out

    crop are known: granite and pegmatite are exposed in a stream

    bed about 1000' southeast of the site; and pegmatite and migma

    tite are exposed 4000 to 5000 feet southwest of the site. In ad

    dition, 10' of pegmatite bedrock was cored (see log in Appendix

    -6

  • D) from a depth of 104' to 114' in boring B103, 4000' east of the

    site.

    In the area from one to two miles from the site, bedrock is

    exposed at several widely-scattered locations to the southeast,

    east, north and northwest of the site. Rock types at these loca

    tions range from migmatite to pegmatite and granite. Thin north-

    to northeast-trending mafic dikes transect the country rocks at 5

    of the more distant outcrop areas to the east and southeast of

    the site. The observed dikes are thin, ranging from a few inches

    to a maximum of 30" in thickness.

    Bedrock Lithologies

    Regional geologic maps (Doyle, 1967; Hussey and Pankiwskyj,

    1975) place the site within the area of the Sebago Granite

    Pluton, in a location where the outcrop pattern of the pluton is

    markedly "necked" by the encroachment—from both the southwest

    and northeast—of phyllite and micaceous quartzite of the Eliot

    Formation. The indiscriminant occurrences of granite, pegmatite

    (coarse-grained granite) and migmatite (feldspar-rich or

    granitized schistose rock) that are exposed within two miles of

    the site suggest that the site bedrock lies in the roof zone of

    the Sebago Granite intrusive and may also consist variably of

    granite, pegmatite and foliated migmatite. These bedrock types

    inferred for the site may also have been intruded by one or more

    nearly-vertical, tabular mafic dikes (iron-rich igneous intru

    sives occurring in planar form).

    Bedrock Structure

    The features of bedrock structure that are particularly im

    portant in controlling the direction and rate of ground water

    flow within the bedrock include faults, joints (fractures), foli

    ation (micaceous layering or schistosity), mafic dikes, and sol

    uble lithologies (rocks that dissolve and develop open channels).

    -7

  • Faults

    We have found no evidence that suggests the presence of a po

    tentially-permeable fault at the site. Where fault zones in the

    crystalline rocks of Maine are sufficiently fractured and open to

    comprise important bedrock aquifers, they also commonly occur in

    troughs or valleys since the broken rock is less resistant to

    weathering and erosion than unbroken rock. Since test borings

    and water well records indicate that the bedrock surface forms a

    distinct ridge beneath and to the north of the site (Figure 3),

    there does not appear to be a significant likelihood of a

    permeable fault-zone aquifer there.

    With reference to Figure 3, the topography of the bedrock

    surface inferred from scattered borings, water wells and seismic

    refraction data suggests the presence of a distinctive bedrock

    trough passing about one-quarter mile northwest of the site.

    This trough runs on a trend of N60°E. If a bedrock trough truly

    does exist as our data suggest, it may reflect a fault zone, but

    seems more likely to us to reflect a band or roof pendant of rel

    atively soft, micaceous metamorphic rock. Numerous analyses of

    Collyer Brook water taken during the past 5 years have shown that

    no contaminant is entering the brook at the northeasterly projec

    tion of the zone.

    Joints, Dikes, and Foliation

    A rose diagram is presented on Figure 7 to summarize the ori

    entations of high-angle (steeply-dipping or nearly vertical) bed

    rock joints, mafic dikes and migmatitic foliation that were

    measured on outcrops throughout the area. While the 52 high-

    angle joints that were measured are seen to strike in all direc

    tions, a modest preference seems to be toward N50-60°W. The

    prominent rose trend toward N20-30°E is due principally to the

    strikes of 3 of the 5 mafic dikes measured in the area. These

    dikes inherently have higher bulk permeability than their enclos

    ing country rock, due to relatively close internal jointing.

    -8

  • Outcrops commonly show the dikes appearing in weathered troughs,

    suggesting that they are relatively soluble, and may contain open

    passageways at depth.

    Planes of potentially important bedrock partings are formed

    by foliation in micaceous migmatitic rocks. This foliation is a

    layering—like the pages of a book—that forms in certain rocks

    that are metamorphosed. As shown by the rose diagram on Figure

    7, foliation readings on 21 migmatitic outcrops show a close

    clustering of strikes toward the east-northeast. The dip or in

    clination of the foliation is commonly very steep. Computer sim

    ulations of bedrock ground water flow in steeply-dipping

    quartzitic schist at Wiscasset (Gerber and Rand, 1980) found that

    the permeability parallel to foliation was about 5 times greater,

    on the average, than parallel to the joints that ran perpendic

    ular to foliation. We do not believe that this is necessarily

    the case in East Gray, however. At none of the 21 observed migma

    tite outcrops in the East Gray area did the rock exhibit any nota

    ble tendency to weather and part along foliation planes, as it

    did in Wiscasset. It is our impression that ground water move

    ment in the rock in East Gray is controlled more by the joints

    and the mafic dikes than by foliation. The significance of these

    features is discussed in more detail later in the report.

    Lineament Analysis

    A lineament analysis of the study area was conducted using

    satellite imagery and standard panchromatic aerial photography at

    two scales to identify possible bedrock structural patterns. In

    the early phases of the study we had hoped that this would aid in

    identifying the preferred directions of bedrock ground water move

    ment, since so little outcrop was available for study near the

    site. Lineaments are visible as tonal, vegetal, or topographic

    alignments and may represent structural elements such as fracture

    zones, faults and foliation.

    -9

  • To provide an understanding of the regional setting, linea

    ments were first identified on 1:1,000,000 scale Landsat imagery.

    Two prominent lineament directions were detected. In the coast

    al area, lineaments parallel the regional strike (plane of foli

    ation) of the bedrock at approximately N20°E. A second set

    strikes northwest, roughly perpendicular to the first set of

    lineaments. A third, less prominent set, strikes approximately

    east-west and is particularly noticeable in the area surrounding

    Sebago Lake. This latter set appears to coincide approximately

    with the foliation developed within the Sebago Pluton and adjoin

    ing rocks.

    To obtain more detailed lineament data for the site area, it

    was necessary to use aerial photography capable of clearly depict

    ing cultural features without obscuring linear trends. Two

    scales were examined: 1:24,000 and 1:12,000. The 1:12,000 (1" =

    1000') scale photography was determined to be unsuitable as linea

    ments could not be precisely defined. The 1:24,000 (1" = 2000')

    scale photography provided an excellent overview of the area and

    better conditions for identifying linear features.

    Identifications of lineaments was completed using stereoscop

    ic and composited methods. Sixty-one lineaments were identified

    within the study area. Two general directions predominate. The

    strongest direction lies between N70°E and N90°E. The second

    direction is less well-defined and trends northwesterly, or ap

    proximately perpendicular to the first.

    By comparing the results of the lineament analysis and field

    mapping of the bedrock, it is apparent that the east-northeast

    trending set corresponds geometrically to local bedrock foli

    ation. The second set of local lineaments coincides with a minor

    orientation of field-mapped fractures.

    SURFICIAL GEOLOGY

    The surficial geology of the site area has been mapped by

    various governmental agencies. The pertinent literature refer

    -10

  • ences include Prescott (1980), Prescott (1977), Caswell (1979),

    Thompson (1976), and Prescott, Smith and Thompson (1976). All of

    the these studies were reconnaissance-level in nature and did not

    have the advantage of all of the data that were available for our

    interpretation. Therefore, the following description is based

    solely upon our detailed study of the local surficial geology.

    Soil Thickness

    A separate map of soil thickness has not been prepared for

    this report; however, one can determine the approximate soil

    thickness at any point in the study area by subtracting the bed

    rock surface elevation on Figure 3 from the surface elevation at

    the same point. With the exception of the southern edge of the

    area covered by the surficial finite element model and the vicin

    ity of node 72 shown on Figure 6, bedrock lies at a relatively

    great depth under the remainder of the surficial aquifer finite

    element model area. A glance at the seismic refraction profiles

    in Appendix C shows that soil averaged about 150' thick along

    most of the profiles (locations shown on Figure 1). Bedrock well

    #4 (Figure 1) encountered 200 feet of soil. According to the sta

    tistical analysis of the bedrock well data in Appendix B, the me

    dian soil depth is almost 100'. This large soil thickness caused

    great difficulty and expense in the drilling and geophysical

    field programs. The areas of greatest soil thickness appear

    along the northeasterly trending trough on Figure 3 and along the

    Royal River valley. Soil thickness is interpreted to be not near

    ly so great (30-50') along Collyer Brook, upstream of the Pumping

    Station that serves as the Pineland Center water supply.

    Soil Types and Distribution

    The surficial geology information shown on Figure 2 was de

    veloped from a combination of field mapping, aerial photo inter

    pretation, geophysical profile interpretation, and well and

    boring data interpretation. Field mapping was very detailed

    -11

  • around the site 'and the zone between Mayall Road and the Royal

    River and Mayall Road and Collyer Brook. The three separate grav

    el pit operations west of Depot Road (called "Pownal Road" on the

    topographic map) provided many exposures of glacial ice contact

    and outwash deposits in those areas.

    The Figure 2 depiction of the surficial geology fails to

    show the 3-dimensional nature of the deposits, which are very het

    erogeneous. Large areas of glacial till are shown to the far

    west and south of the site. These stony, silty soils were laid

    down under the glacial ice advance between 22,000 and 16,000

    years ago. They are compact and have a moderately low permeabil

    ity. Although glacial till was found in the bottom of boring

    B104, it was not distinguishable as a separate seismic refraction

    velocity (Appendix C). Along seismic line 4, however, a typical

    till velocity of 6000-6500 feet per second was measured, indicat

    ing that the majority of the soil profile in that area was till.

    Seismic lines 5 and 7 should have shown saturated till, but this

    could not be interpreted from the field data because of the great

    thickness of overlying unsaturated sand and gravel. There are

    large areas of Figure 2 that are indicated by the widely-spaced

    dotted pattern to be sand over till. In these areas, we infer

    that there is a significant thickness of glacial till overlying

    bedrock, which controls ground water movement in these areas. We

    did not find any evidence of any washboard or other moraines

    along the west side of the Royal River. Although there seems to

    be a parallel alignment of ridges and gullies trending east-

    northeast, similar to moraines farther to the south, the topo

    graphic pattern appears to be developed solely by erosion in

    glaciomarine deposits.

    Following the deposition of the glacial till, the continen

    tal glacier melted until its leading edge was resting in sea wa

    ter that was about 300' higher than present-day sea level about

    13,000 years ago. During summer melting periods, large rivers of

    glacial meltwaters flowed through the site area. Evidence for

    -12

  • this is seen in the exposed gravel banks of the Portland Sand and

    Gravel operation about one-half mile northwest of the McKin site.

    It appears that the major meltwater river entered the area of

    Portland Sand and Gravel's operation from the northwest, then bi

    furcated at the northwestern edge of the pit. The major branch

    flowed east toward the Royal River; the minor branch turned south

    and flowed toward Gray Meadow. Deltaic sand and gravel deposits

    formed in a large area, including under the McKin site, to the

    southeast of this meltwater stream bifurcation. These sediments

    spread out to cover areas of glacial till that had been deposited

    thousands of years earlier. Along the distinct courses of the

    two meltwater channels, the eastward-flowing channel apparently

    washed away all till along a narrow path to the Royal River and

    we infer it to consist of very coarse gravel and cobbles at the

    bottom. The southern meltwater branch appears to be perched on

    till or shallow bedrock to the west of sampling point 31 (Figure

    1) and there is no evidence to indicate that it now diverts

    ground water flow in that area to the south.

    The final major surficial unit of interest to the study is

    the soil unit identified as "GM" on Figure 2. This unit consists

    of a thin-bedded (typical beds are 1/16" to 1" in thickness) se

    quence of fine sands, silts and clay-silts. These deposits are

    the result of deposition in sea water of the finer particles of

    rock that had been pulverized by the continental glaciation.

    This deposition occurred concurrent with and immediately follow

    ing the deposition of the ice contact deposits. As the meltwater

    streams entered the water-filled valley, the coarse sands were de

    posited on the face of a delta and the fine sands, silts and

    clays settled to the valley bottom. A typical sequence of this

    "glaciomarine deposit" in the East Gray area consists of five to

    ten feet of fine to medium sand overlain by low permeability

    thin-bedded fine sands, silts and clay-silts, finally topped by

    none to 20' (locally) of fine sand. In boring B103, this

    sequence was found to be 103 feet thick. Although individual

    -13

  • clayey layers within this sequence are usually saturated below

    20" depth, thin beds of fine sand may not be saturated on the

    valley walls under the clayey layers. Therefore, seismic

    velocities that are typical of saturated glaciomarine

    deposits— 5000 feet per second—were not found except at the

    western end of line 2 (Appendix C).

    Significance of Surficial Geology to Contaminant Spread

    The observed spread of chemical contamination in the ground

    water appears to be directly related to the thickness, nature and

    permeability of the surficial soils. Bedrock and/or till ridges

    occur to the far west, far south, and on the north side of Mayall

    Road north of the Portland Sand and Gravel pit operations. The

    only major outlet for ground water flow from a relatively large

    watershed area is through the coarse gravels of the former gla

    cial meltwater channel that runs east from the Portland Sand and

    Gravel pit toward the Royal River. The low permeability clayey

    glaciomarine deposits on the east face of the delta act as a

    leaky dam. Ground water from the west leaks through the clay and

    erupts in springs in the slope east-northeast of the site as

    shown on Figure 2. Boring B102 suggests that the top of the ex

    tension of the sand aquifer that runs under the clay is about

    equal to the Royal River bottom elevation just east of the

    "Boiling Springs". At this major bend in the River there is ap

    parently a major discharge of the ground water that originates in

    a large area to the west, including the site.

    HYDROGEOLOGIC ANALYSIS

    Most of the data collection and interpretation until this

    time has related to tracking the path of the chemical contamina

    tion in the ground water and surface waters. Only W. B. Caswell,

    in his short report to DEP (BCI Geonetics, 1980) began any anal

    ysis of the ground water regime of the site area. His analysis

    was based primarily on bedrock well data, most of which came from

    -14

  • domestic wells in the triangle between Depot and Mayall Roads,

    north of the site. Ground water levels, the contaminant plume,

    and the bedrock contours were all sloping downward to the north.

    No TCE or TCEy was detected in the Royal River north of Gray

    Station, and only a few low and questionable laboratory results

    suggested any contamination in Collyer Brook. The Boiling

    Springs on the west bank of the Royal River, east of the site

    were found to produce a fairly stable level of TCE contamination

    and the concentration of TCE in the Royal River at Brickyard

    Station has also been fairly constant, but inversely related to

    river flow rate. Prior to this study there had been no

    evaluation of the surficial aquifer flow nor of the level of con

    tamination, nor any evaluation of the mechanism of transfer be

    tween the surficial and bedrock aquifers.

    METHODS

    As required by contract, we have evaluated the ground water

    regime in both the surficial and bedrock aquifers through a combi

    nation of field investigations, water quality analysis, hand cal

    culations and computer simulation modelling.

    In order to understand a ground water regime in detail, it

    is necessary to be able to reproduce its response to stresses

    such as precipitation, pumping wells, and discharge to streams.

    The factors that control this response are the 3-dimensional

    boundaries of the aquifer, recharge and discharge rates, and the

    physical properties of the aquifer such as permeability and stor

    ativity. We have evaluated all of these factors to the best of

    our ability using the available information, and have developed

    simulation flow models that appear to simulate the behavior of

    the aquifers with reasonable accuracy.

    The methods that we have used to develop this model are de

    scribed in this section. One of the fundamental problems is to

    determine the general geology of the aquifer. This was done

    through a combination of literature search, aerial photo interpre

    ts

  • tation, field mapping, geophysical investigations, borings, anal

    ysis of water well data, and back-calculation of information from

    observed ground water levels and estimated permeabilities.

    Delineation of Aquifer Boundaries

    Four models were developed to study this problem: a) a re

    gional flow model of the surficial aquifer that either included

    natural aquifer boundaries, or extended far enough beyond the

    site area to have no effect on the results in the site area; b) a

    surficial aquifer model that covered an area near the site that

    was sufficient for contaminant transport studies; c) a regional

    flow model of the bedrock aquifer; and d) a site area contaminant

    transport model for the bedrock aquifer. The site area models

    use fixed water tables and water inputs and withdrawals for their

    boundary conditions that are established by the regional models.

    The transfer of ground water between the bedrock model and surfi

    cial model was simulated using water table elevations in the sur

    ficial model and leakage coefficients that described the amount

    of water transfer between the two aquifers as a function of surfi

    cial aquifer thickness, surficial water table position relative

    to that in the bedrock aquifer, and surficial aquifer vertical

    permeability.

    The regional models are the finite element models shown on

    Figures 6 and 7, for the surficial and bedrock models, respective

    ly. The computer program is a Galerkin-type finite element, 2

    dimensional saturated flow model—AQUIFEM—developed by Townley

    and Wilson (1980). The northern and eastern boundaries of the re

    gional flow models are interpreted to be no-flow boundaries along

    Collyer Brook and the Royal River. The western and southern

    boundaries of the bedrock regional flow model are also assumed to

    be no-flow boundaries (this assumption is probably incorrect at

    node 38, but did not appear to affect the model significantly).

    Many of the western and southern boundaries of the surficial

    regional flow model are treated as constant flux boundaries that

    -16

  • allow ground water flow either into or out of the aquifer perpen

    dicular to these boundaries. Inflow from upland till regions was

    estimated from recharge areas and typical recharge-per-unit-area

    values, then checked against estimated till transmissivities (per

    meabilities times aquifer thickness) and ground water gradients.

    Outflow between nodes 179 and 180 was estimated in a similar

    manner. Outflow between nodes 1 and 3 was a parameter that was

    varied during analysis to study its overall effect on the model.

    The many small drainageways within the surficial regional model

    were treated as intermittent or partially-penetrating streams.

    These streams are not treated as "line sinks" or constant head

    boundaries. They are treated in a manner permitted by AQUIFEM to

    accept ground water discharge at a rate proportional to the

    ground water gradient next to the stream, and a leakage parameter

    which is estimated from methods in Townley and Wilson (1980, p.

    68-70) and Rushton and Redshaw (1979, 207-209). When ground

    water levels are defined below the bed of the stream, no exchange

    of water takes place between the aquifer and the stream.

    A special condition occurs within the surficial aquifer

    along the string of nodes 166, 167, 185, and 168. Ground water

    discharge occurs in springs on the first three nodes, then re

    enters the ground as recharge at node 168. The springs were han

    dled with the standard leakage parameters, and the recharge was

    treated as injection at node 168. Several runs were necessary to

    simulate a balanced zero net flow.

    The areas covered by the solute (contaminant) transport mod

    els are shown on Figures 6 and 7, respectively, for the surficial

    and bedrock models. They occur in identical locations for both

    aquifers to simplify solute transfer calculations between aqui

    fers. The computer model is a block-centered, rectangular

    finite-difference, saturated flow model developed by Konikow and

    Bredehoeft (1978), often called the K&B model, which uses the

    "method of characteristics" to solve the solute transport equa

    tion. This model, which is referred to hereinafter as the "K&B

    -17

  • model", uses fixed water table elevations on the outside bound

    aries as determined from the regional flow model. Since this mod

    el does not have the same sophistication as AQUIFEM in handling

    partially-penetrating streams, stream discharge rates that were

    determined by AQUIFEM were placed in the same positions and at

    the same rates in the K&B model as if they were diffuse

    discharge.

    Bottom of Surficial Aquifer

    Using the AQUIFEM model, we input aquifer permeabilities in

    stead of transmissivities in the surficial aquifer, so that the

    position of the potentiometric surface would be solved as part of

    the model output. This was necessary since it was not known be

    forehand how much of the aquifer that lay beneath areas mapped as

    glaciomarine clay deposits would actually be artesian. This re

    quired that we know the approximate bottom of the aquifer. The

    bottom of the surficial aquifer was assumed to be the bedrock sur

    face. Using the well data (Appendix B) , the seismic refraction

    profile results, and the field mapping interpretation, we con

    structed the bedrock contour map of Figure 3.

    Aquifer thickness is important to the surficial contaminant

    transport model, since dispersion is inversely proportional to

    aquifer thickness, all other factors equal. For the bedrock mod

    el, thickness is not known since water may flow at depths as

    great as 700'. Generally, however, the majority of water flows

    within the top 300' of rock in granitic rock. Since no disper

    sion coefficients were applied in the bedrock contaminant trans

    port model, the thickness did not affect plume dispersion.

    Permeability and Transmissivity

    Aquifer permeability is a measure of the volumetric rate

    that water will flow through an aquifer if it is flowing on a

    slope of 1 on 1. Transmissivity is found by multiplying perme

    ability by aquifer thickness. There are no test data from either

    -18

  • aquifer that determined transmissivity, however localized

    permeabilities were estimated from borehole tests in B102 and

    B103 in the fine sand glaciomarine deposits. Initial permeabil

    ities were also estimated from field mapping and from correla

    tions of observed grain size distributions with

    permeability-versus-grain-size tables. For the bedrock model,

    rock transmissivity was initially estimated from the few (14)

    wells with reported yields (Appendix B—see statistics), but in

    the surficial and bedrock aquifers, permeability and transmissiv

    ity, respectively, were the major variables that needed to be de

    termined. Permeabilities in the surficial aquifer were assumed

    to be isotropic (equal in all directions at any point).

    Transmissivities in the bedrock aquifer were also treated isotrop

    ically although the model was designed to allow anisotropic treat

    ment if it appeared that it would be necessary to calibrate the

    model. The rose diagram (Figure 7) of the fracture orientations

    shows that the AQUIFEM model axes were aligned with the mafic

    dikes and major joint concentrations—the two major inferred av

    enues of ground water movement. Calibration of the bedrock flow

    model, although corroborating data are sparse, does not seem to

    indicate the need for anisotropic treatment.

    After many trial combinations of permeabilities and stream

    leakage coefficients, an acceptable agreement with field-observed

    or interpreted water table elevations was obtained. Since some

    deviations between anticipated and model-predicted water table el

    evations could not be eliminated with the original grid design,

    the final water table contour map presented with this report is

    our hand-drawn interpretation that we used to calibrate the surfi

    cial flow model (Figure 4). A scan of the final data set for the

    AQUIFEM surficial flow model indicates that typical permeabil

    ities of the sand and gravel deposits were in the range of 10 to

    100 feet per day, as we expected. The coarse-grained glacial ice

    contact deposits apparently have a very high (2000 feet per day)

    permeability, making them equal to some of the highest measured

    -19

  • permeabilities in Maine eskers. An inferred permeability for the

    lodgment till is as low as 0.01 feet per day (3 x 10~ cm/sec).

    The majority of the bedrock aquifer was found to produce a

    moderately high transmissivity of 100 square feet per day. The

    median yield of the 14 wells with data is 5 gallons per minute,

    an average yield for Maine wells. However, the low water level

    readings in many of the other wells intuitively suggest

    above-average bulk transmissivity for the area as a whole.

    The K&B model will only accept input as transmissivities, so

    for the long-term steady state simulation of contaminant trans

    port, the transmissivities calculated from the AQUIFEM output

    were input to the K&B model.

    Storativity and Specific Yield

    The specific yield of an aquifer is a measure of the amount

    of water released per unit of volume of aquifer when water is al

    lowed to drain from the aquifer by gravity. For most practical

    problems it can be equated with storativity when dealing with "wa

    ter table" or "unconf ined" aquifers, such as the case here.

    Storativity is the volume of water added to or removed from a

    unit surface area of aquifer under a unit change in water table.

    The only difficulty comes in the evaluation of rapid changes in

    water table in stratified soils where soil strata with relatively

    lower permeabilities restrict the rate that water can drain from

    coarser strata above them, producing an apparent storativity that

    is less than the specific yield.

    For this report, a knowledge of specific yield was not impor

    tant since only the steady state, average annual condition was

    simulated because long-term (10-year) simulations were necessary.

    Dispersivity

    The USGS solute transport model requires values of lon

    gitudinal (along the direction of flow) and transverse (perpendic

    -20

  • ular to direction of flow) dispersivities as input to the model.

    Dispersion is a molecular and physical phenomenon that causes a.

    contaminant plume to spread beyond the limits of what one would

    delineate by a simple convective, or flownet analysis.

    Dispersion is a scale-dependent phenomenon and a measure of the

    heterogeneity of an aquifer. For regional models involving sever

    al thousand feet of flow in stratified sand and gravel aquifers,

    the longitudinal and transverse dispersivities of 50 feet and 15

    feet, respectively, are appropriate according to the literature

    (e.g., Fried, 1975). Failure to account for dispersion can re

    sult in underestimating the apparent transport rate of a contam

    inant, and in underestimating the horizontal width of a

    contaminant plume.

    No dispersivity is assumed for the bedrock aquifer since

    this phenomenon is poorly understood in bedrock and there are no

    data available here to validate any particular value. Thus, the

    contaminant transport simulations done for the bedrock aquifer in

    this study assume convective flow without dispersion.

    Pollutant Attenuation by Retardation and other Mechanisms

    The 2-dimensional solute transport model does not take the

    sorptive capacity of the soil into account, nor the evaporative

    loss of the volatile chemicals that we deal with here. However,

    since these retardation factors are not well known, are different

    for each chemical, and greatly complicate modelling beyond the

    scope of this study, retardation was not taken into account. At

    this time, there are insufficient data to allow any calibration

    of retardation coefficients, even if they were incorporated.

    Thus it has to be kept in mind when viewing Figures 9 through 12

    that these treat the chemicals as "conservative" contaminants

    (i.e., contaminants whose concentration is only reduced by mixing

    with water) which move as tracers in the ground water.

    It is significant to note that TCEy apparently has a greater

    retardation coefficient than TCE. Recent test results for TCEy

    -21

  • are showing significantly increased concentrations over those mea

    sured at the same points in 1980, for example, whereas there has

    been no corresponding significant increase in TCE.

    Aquifer Inputs and Withdrawals of Water

    The major aquifer water input to the surficial aquifer is

    precipitation recharge. Using the results of USGS records and

    many other studies of this type that we have performed, we have

    assumed that on a yearly average, 60% of the average precipita

    tion enters the ground water to become recharge in the sand and

    gravel areas, 25% in the till areas, 5% in the glaciomarine clay

    ey areas, and intermediate values in the transition areas.

    Average annual precipitation was assumed to be 43" although the

    monthly and annual deviations from the average can be large (see

    precipitation graph on Figure 8). Since the long-term contam

    inant transport simulations were for 10 to 11 year periods, the

    average annual value is the most appropriate number, however.

    For the bedrock aquifer, water transfer to and from the aqui

    fer is a function of the water table position in the overlying

    surficial aquifer and the vertical permeability and thickness of

    the surficial aquifer. The rate of water transfer is proportion

    al to the difference between the potentiometric surfaces in the

    two aquifers, multiplied by the leakage coefficients. If the po

    tentiometric surface in the bedrock aquifer drops below the

    bottom of the surficial aquifer, the leakage rate becomes con

    stant and proportional to surficial aquifer thickness. The

    AQUIFEM bedrock flow model was developed using this principle of

    water transfer through leakage. Once the flow model was calibrat

    ed, the appropriate calculated leakage rates were input to the

    K&B model.

    Since water is necessarily added to the surficial model area

    by precipitation, it must be taken out at approximately the same

    rate, otherwise the water table would not cycle year after year

    around the same average position. In the surficial models, this

    -22

  • water is removed as stream flow or as a constant flux along the

    boundaries. The largest area of predicted discharge is the north

    ern branch of the major drainage leading west from the Royal

    River, just south of the Boiling Springs. In the zone between

    nodes 2 on the east and 89 on the west (Figure 6), about 800 gal

    lons per minute is discharged from ground water to surface water.

    Much of this probably occurs under the Royal River in this vicin

    ity. However the local tributary stream discharge is also signif

    icant. Field mapping found this area to be the largest

    concentration of springs in the entire study area, as suggested

    on Figure 2. This should be the case since the surficial model

    indicates that a large ground water flow in a high permeability

    zone occurs under this area. In the case of the bedrock aquifer,

    discharge occurs in a diffuse manner over most of the areas north

    and east of Mayall Road. The model also suggests, however, that

    there is a localized concentration of discharge to the surficial

    aquifer near the Boiling Springs.

    Application of Contaminant to the Aquifer

    No modelling was done of contaminant transport in the unsat

    urated zone of the aquifer. This is a difficult problem at best

    and there are no data to calibrate such a model at the site. The

    models developed for this study are 2-dimensional, saturated flow

    models that are incapable of accounting for fluid density differ

    ences or non-homogeneous aquifers. When contaminants are intro

    duced into the aquifer, simulations of contaminant concentrations

    with time reflect vertically-averaged contaminant concentrations

    for the entire grid element, which for this study is 340 feet by

    340 feet square.

    The time history of application of the organic chemicals at

    the site is not known, but since the concentrations of the chem

    ical TCE have been relatively constant with time at the Boiling

    Springs and in the Royal River at Brickyard Station, it appeared

    that the chemical was being added at a relatively constant rate

    -23

  • for at least several years in time. The probable source of the

    contaminant is a. leak or spill of the chemicals onto the ground

    at some time in the past. These chemicals are settling slowly

    through about 40' of unsaturated soil under the site. Whenever

    it rains, a little of the contaminant is transported downward to

    the water table. Over several years' time this can result in a

    fairly steady application rate of the contaminant to the aquifer.

    For the purposes of this study, the chemical was added as a con

    stant concentration of ground water recharge which was derived

    from precipitation during the time (ten years) of the surficial

    aquifer simulation. Starting with an arbitrary percentage input,

    simulated concentrations were back-calculated using a factor that

    was computed to adjust the predicted Boiling Springs concentra

    tion to about 200 parts per billion (ppb) of TCE, which is close

    to the historical observed value. TCE was used for modelling pur

    poses because there are more data on TCE concentrations around

    the area than for any other.

    Once the concentration with time was determined for each

    grid cell of the model, this was applied to the water that leaked

    into the corresponding recharge cells of the bedrock aquifer as a

    stepped time input until ten years was also simulated for the bed

    rock aquifer.

    For simulations of the effects of removing the contaminant

    or isolating the site from precipitation recharge, the initial

    concentrations of contaminant in each respective aquifer were as

    sumed to be the concentrations that developed after ten years of

    introduction of the contaminant. An additional ten years was sim

    ulated to observe the rate of cleansing of the aquifers. Again,

    for the bedrock aquifer, the concentration of contaminant that

    leaked into the aquifer from the surficial aquifer had to be in

    put in a stepped manner in time.

    -24

  • INTERPRETATION OF COMPUTER SIMULATION RESULTS

    Ground Water Flow in the Surficial Aquifer

    Our calibration of the surficial aquifer flow model suggests

    that .there is much more glacial till underlying the model area

    than we had originally anticipated. Therefore, the area of sat

    urated sand and gravel aquifer is much smaller than suggested,

    for example, on the map by Caswell (1979). We infer, for exam

    ple, that there is a ridge of till underlying the sand and gravel

    along Mayall Road, north of the Portland Sand and Gravel pit oper

    ation. The area under the Pownal Road, northeast of the intersec

    tion with Mayall Road, is also largely underlain by till. As

    previously mentioned, the narrow ridge of ice contact sand and

    gravel that lies west of node 166 on the surficial AQUIFEM model

    appears to be perched on till with very little saturated thick

    ness of sand and gravel over the till. The large spring at node

    166 is apparently at the contact of the gravel over the till.

    The recharge area that lies upgradient (southerly) of the

    site that is available for dilution of any contaminants leaving

    the site is small. Ground water flows away from the site to the

    northwest. Further from the site, the ground water that orig

    inates at the' site turns north, then east, and flows to the Royal

    River. This explains why no contamination was detected in the

    "McKin spring", just east of the site. South of the site, the

    ground water divide is in the vicinity of Route 115, with ground

    water to the south passing into a surface drainage that passes to

    the southeast of the site.

    Sensitivity analyses on the rate of constant flow leaving

    the aquifer under the Royal River between finite element nodes 1

    and 3 indicate that the flow is small there. Back-calculated per

    meabilities for this area suggest that any deposits under the

    clay in this area are no coarser than fine sand. This explains

    why almost all contaminants that originate at the site are dis

    charged to the Royal River in the small area between the Boiling

    -25

  • Springs and the railroad bridge where the sand and gravel aquifer

    terminates and is in contact with or close to the bottom of the

    River in this area. There is no suggestion that the coarse sand

    and gravel continues to the south of the Railroad Bridge. Water

    samples from Boring B103, near the Railroad bridge and south of

    Boiling Spring No. (Figures 1 and 2), showed no TCE or TCEy con

    tamination in surficial or bedrock materials at that location in

    October 1982.

    Bedrock Aquifer Flow

    Figure 5 represents the computer-simulated potentiometric

    surface contours in the bedrock aquifer. Bedrock ground water

    flow moves to the north from the site, then turns east and final

    ly southeast to discharge to the surficial aquifer in the

    vicinity of the Boiling Springs. Overall aquifer transmissivity

    is interpreted to be somewhat greater than average for Maine bed

    rock aquifers. However there are two apparent areas of low trans

    missivity: one area is 500 to 1000 feet due east of the site;

    the other area is in the vicinity of Philip Humphrey's well, bed

    rock well #4 on Figure 1. We interpret an area of relatively

    higher bedrock transmissivity to lie to the northwest of the

    site, in the vicinity of the bedrock trough shown on Figure 3.

    The bedrock aquifer model was simulated assuming that the

    Humphrey well, which is still being used, was being pumped at an

    average rate of about 50 cubic feet per day. Since the wells in

    the triangle between Depot Road and Mayall Road are not now being

    pumped, no pumping stress was applied on them in the model. Had

    we simulated the pumping of these wells, the potentiometric sur

    face would have obviously been lower in that area.

    The area of highest discharge rate to the surficial aquifer

    is in the area of the Boiling Springs. This would be expected be

    cause of the permeable nature of surficial aquifer there and the

    fact that the surficial aquifer is discharging to the River

    there.

    -26

  • Contaminant Transport Modelling—Surficial Aquifer

    As discussed earlier in this report, there are several diffi

    culties involved with simulating contaminant transport in the

    East Gray aquifers: the time history and rate of leakage of chem

    icals into the aquifer are not known; and the contaminant that

    was introduced was not a "conservative" contaminant. Despite the

    difficulties posed by these unknown factors, we believe that we

    have developed some useful information; however it must be eval

    uated with care.

    Figure 9 shows the results of 10 years of introduction of a

    constant rate of TCE into the surficial aquifer in a 340 foot by

    340 foot square area centered on the McKin site. The concentra

    tions under the site were back-calculated by applying a ratio

    that was necessary to adjust the Boiling Springs grid cell concen

    tration to about 200 ppb TCE. Treating the TCE as a conservative

    solute, the computer model predicts that about 26 gallons per

    year of TCE entered the aquifer at the site. Our hand-calculated

    estimates, however, based on our analysis of the concentrations

    observed over time at Boiling Springs and Brickyard Station,

    could place the amount of TCE discharged into the aquifer under

    the site at as high as 125 gallons per year. These discrepancies

    could be explained by the non-conservative nature of the TCE, in

    ability of sample results to develop correct "average" concentra

    tions in the Royal River (particularly at the Railroad Bridge,

    where the distribution of contaminant may not be uniform across

    the River channel), or by improper simulation of the time history

    of the chemical leakage at the site. With respect to the last

    point, there is some evidence to suggest, at least in the surfi

    cial monitoring wells at the site (e.g., well |8, Figure 8) that

    the amount of contaminant leaving the site has decreased with

    time since 1980. Since the simulation shows that it took 5 years

    for the first contaminant to reach Boiling Springs, then about 4

    more years to begin to stabilize near 200 ppb, the actual time

    -27

  • history of chemical leakage at the site could have been as a

    large slug initially, followed by lesser amounts decreasing with

    time thereafter. In a ten-year simulation this would not affect

    the results much at Boiling Springs, but could have resulted in

    greatly increased concentrations near the site in the early

    years.

    There are several other important implications of Figure 9.

    Notice that the path of the contaminant plume is first to the

    northwest, then turns east to run down the high-permeability zone

    to the Royal River. Notice that the plume does not turn to run

    to the south at the Royal River. This is in agreement with the

    water test results on Monitoring well BIOS, which found no contam

    ination in either the soil or bedrock well. Although the verti

    cally-averaged concentration of TCE is simulated to be between

    200 to 300 ppb between the Boiling Springs and sampling stations

    24 and 26 (Figure 1), west of Boiling Springs, only the spring at

    station 24 found any TCE contamination (25 ppb, Table 1—ERGO

    test results). The models used in this study are not three-

    dimensional and would not account for contaminant stratification

    within the aquifer. It appears that most of the contaminant is

    travelling along the bottom of the surficial aquifer.

    Contaminant Transport Modelling—Bedrock Aquifer

    Figure 10 shows our simulation of the contaminant distribu

    tion in the bedrock aquifer after 10 years of simulation. There

    are several interesting aspects of this simulation. First, no

    tice that even though no dispersion is accounted for in the mod

    el, the contaminant plume covers a much larger area than that of

    the surficial aquifer. Although the plume comes very close to

    Collyer Brook, there is no indication that it discharges to

    Collyer Brook in any significant amount, which is in agreement

    with test results on Collyer Brook. Notice that the center of

    the plume tracks toward Boiling Springs, although the initial

    -28

  • track from the site is more toward the north than the surficial

    plume.

    The simulation indicates that the bedrock plume first

    reached Boiling Springs about 2 years after initial contaminant

    introduction. However, the increase of concentration with time

    was slow and was still increasing at 10 years into the simula

    tion. This may explain why the concentration of TCEy at drilled

    wells 4 and 15 appear to have increased with time.

    While the simulation that is summarized on Figure 10 seems

    to agree nicely with recently measured concentrations of TCE in

    bedrock wells at sample locations 14 and 17 (Table 1), it does

    not explain the low readings at sampling locations 15 and 16.

    The well (#9) at sampling location 15 had very high TCE concentra

    tions in 1977. In fact the Figure 10 concentrations are about a

    factor of 10 lower in the triangle area between Mayall and Depot

    Roads than the values observed in the wells there in 1977, at the

    time the wells were in use. We have concluded that this is due

    to the fact that the wells were inducing a much greater rate of

    local recharge of contaminated surficial aquifer water into the

    aquifer at that time, than in the present condition with the

    wells not being pumped. At sample location 16, drilled well #6

    is near the edge of the plume and some adjustments of the bedrock

    flow model could move the simulated plume farther to the east,

    better to match the October 1982 sample results.

    THE FUTURE OF THE CONTAMINANT PLUME

    MAINTAINING THE STATUS QUO

    The time series of water quality tests for TCE and TCEy on

    Figure 8 do not seem to show any major downward trends in concen

    tration of the chemicals in most of the water sampling locations.

    Indeed, there is some indication that TCEy is increasing with

    time at some points. Although some of the surficial monitoring

    wells at the site (e.g. wells 8 and 16) may have begun to show a

    decline in concentrations, DEP site well #7 contained very high

    -29

  • levels of'TCEy (Table 1) during this fall's sampling. It is not

    known whether or not storage tanks are still leaking into the

    soil. If all of the chemicals were presumed to have entered the

    soil prior to 1977, the mechanics of unsaturated flow would sug

    gest that after some point in time, there should be a decrease in

    the absolute volume of chemical that would enter the aquifer.

    Even if the addition of chemicals to the soil stopped after 1977,

    our simulations show that no major change in concentration of the

    chemicals would be observed at Boiling Springs for another 5

    years. If the addition of chemicals to the soil is still going

    on, then the concentrations of the chemicals will likely remain

    the same, or even increase slightly throughout much of the bed

    rock and surficial aquifers. To the best of our knowledge, there

    is only one ground water user, Philip Humphrey, that is presently

    affected by the ground water contamination.

    PREVENTING CHEMICALS FROM LEAVING THE SITE

    There are several ways to prevent any further addition of

    chemicals to the ground water: a) creating a barrier around the

    site that would prevent further migration of ground water through

    or away from the site; b) sealing the top of the site so that no

    more precipitation could filter through the unsaturated soils and

    wash more chemicals into the ground; or c) removing the contam

    inated soils from under the site. This report is not intended to

    evaluate the cost effectiveness or feasibility of these alterna

    tives. This report was only intended to give a generalized eval

    uation of the hydrogeologic aspects of several of the major site

    clean-up approaches.

    Regardless of the particular method used to isolate the

    site, all of the methods can be treated basically the same from a

    hydrogeologic standpoint. The approach is to begin a new simula

    tion with the aquifer having initial concentrations of the chem

    icals as shown in Figures 9 and 10, cut off the addition of

    -30

  • further chemicals to the aquifer, then simulate the process of

    the aquifer cleansing itself.

    The decreases with time of the TCE contaminant concentra

    tions in the aquifers are shown on Figures 11 and 12 for the sur

    ficial and bedrock aquifers, respectively. Dealing with the

    surficial aquifer first, notice that after two years, little

    change takes place in the aquifer, except in the immediate site

    area. After 5 years, the western half of the contaminated

    surficial aquifer has shown significantly decreased

    concentrations, however, the area of the Boiling Springs still

    has not changed much. After 10 years the aquifer is much

    improved in quality overall, with only the low-permeability soils

    northwest of the Boiling Springs still not having changed much.

    In the series of Figures 12A, B, and C, the change in concen

    tration of the chemical TCE in the bedrock aquifer is shown over

    time, after the site is isolated. Notice that after 2 years, the

    only area in the bedrock aquifer that has decreased is immediate

    ly under the site. All other areas have actually increased be

    cause the contaminants are still being added to the bedrock

    aquifer from the surficial aquifer and the bedrock aquifer had

    not reached chemical equilibrium after the initial 10 years' sim

    ulation. After 5 years of site isolation, however, concentra

    tions are decreasing throughout the entire aquifer. At ten years

    TCE concentrations are even more improved. Although not shown in

    this report, the simulation was carried out to 11.4 years, by

    which time the western half of the contaminated portion of the

    bedrock aquifer had decreased everywhere to 3 ppb or less, but

    the area under Boiling Springs was still in the range of 10 to 15

    ppb. Because of the mechanisms of contaminant transfer between

    the surficial and bedrock aquifers, it is apparent that the rate

    of cleanup of the bedrock aquifer would be greatly increased if

    the surficial aquifer could be cleansed at a faster rate, partic

    ularly north of the site, in the triangular area between Mayall

    and Depot Roads.

    -31

  • GROUND WATER REMOVAL OR FLUSHING BY PUMPING

    Some ground water contamination in other parts of the coun

    try has been handled by pumping the contaminated water out of the

    ground and treating the water, then injecting it into the ground

    or discharging it to surface waters. Another method is to inject

    water to control the direction of flow or the rate of dilution of

    the contaminant. For the bedrock aquifer and low permeability

    soils within our study area, manipulation of the ground water by

    pumping or injection would probably not be effective or feasible.

    In the high-permeability surficial zone, these techniques could

    be used to accelerate the clean-up of the surficial aquifer,

    which would also accelerate the clean-up of the bedrock aquifer.

    The optimum location for a withdrawal well in the high-

    permeability zone would be about 1000' south of the intersection

    of Mayall and Depot Roads. Clean ground water injection wells

    could prove useful in moving contaminated water out of the area

    under the site, which has a moderately low permeability. We have

    not evaluated the technical aspects of these options in detail,

    since additional field data are necessary and modifications to

    the existing ground water simulation models are necessary before

    a meaningful evaluation of these alternatives can be performed.

    With a combination of withdrawal and injection wells, the contam

    inant distributions as shown for 10 years after site isolation in

    Figures 11 and 12C might be achieved in one to three years. This

    assumes that the site was isolated at the start of the pumping.

    SUMMARY AND CONCLUSIONS

    The summary and conclusions of this report are given on the

    letter of transmittal at the beginning of this report.

    -32

  • LIST OF REFERENCES

    BCI Geonetics, Inc., 1980. A consultant report by W. B. Caswell

    to the Maine Dept. of Environmental Protection concerning

    the ground water contamination in East Gray, Maine

    Caswell, W.B., compiler, 1979, Sand and gravel aquifers, maps 11

    and 12, Cumberland and Androscoggin Counties, Maine. Maine

    Geological Survey, Dept. of Conservation, Augusta, ME 04333

    Caswell, W.B., and E.M. Lanctot, 1976 (rev. 1978), Ground water

    resource maps of Cumberland Co. Maine Geological Survey,

    Dept. of Conservation, Augusta, ME 04333

    Doyle, R.G., 1967, Preliminary geologic map of Maine. Maine

    Geological Survey, Dept. of Conservation, Augusta, ME 04333

    Ecology and Environment, Inc., 1981, Preliminary site assessment

    and emergency action plan, McKin site, Gray, Maine. FIT

    project prepared for U.S. EPA, Contract No. 68-01-6056

    Fried, J.J., 1975, Groundwater pollution. Elsevier Scientific

    Publ. Co., Amsterdam, 330 p.

    Gerber, R.G. and J.R. Rand, 1980, Geology and hydrology—Mason

    Station ash disposal facility, Wiscasset, Maine. A

    consultant report to Central Maine Power Co., Edison Drive,

    Augusta, ME 04336

    Hart, Fred C. Associates, 1978, Analysis of water contamination

    incident in Gray, Maine. Prepared for U.S. EPA, Contract

    No. 68-01-3897

    Hussey, A.M. II, and K. Pankiwskyj, 1975, Preliminary geologic

    map of southwestern Maine. Open-file map 1976-1, Maine

    Geological Survey, Dept. of Conservation, Augusta, ME 04333

    Konikow, L.F., and J.D. Bredehoeft, 1978, Computer model of

    two-dimensional solute transport and dispersion in ground

    water: Book 7, Chapter C2, Techniques of Water-Resources

    Investigations of the U.S. Geol. Survey, Washington, D.C.

    20242, 90 p.

    Prescott, G.C., Jr., 1980, Ground water availability and surfi

    cial geology of the Royal, Upper Presumpscot, and Upper Saco

    -33

  • River Basins, Maine. Water-Resources Investigations

    79-1287, U.S. Geol. Survey, Washington, D.C. 20242

    Prescott, G.C., Jr., 1979, Maine hydrologic-data report no. 10,

    ground-water series, Royal, Upper Presumpscot, and Upper

    Saco River Basins, Maine. U.S. Geol. Survey, Washington,

    D.C. 20242

    Prescott, G.C., Jr., 1977, Ground water favorability and surfi

    cial geology of the Windhara-Freeport area, Maine.

    Hydrologic Investigations Atlas HA-564, U.S. Geol. Survey,

    Washington, D.C. 20242

    Prescott, G.C., Jr., 1976, Maine basic-data report no. 9,

    ground-water series, Windham-Freeport-Portland area. U.S.

    Geol. Survey, Washington, D.C. 20242

    Prescott, G.C., Jr., G.W. Smith, and W. B. Thompson, 1976,

    Reconnaissance surficial geology of the Cumberland Center

    Quadrangle, Maine. Maine Geological Survey, Dept. of

    Conservation, Augusta, ME 04333

    Rushton, K.R., and S.C. Redshaw, 1979, Seepage and groundwater

    flow. John Wiley & Sons, Chichester, England, 339 p.

    Thompson, W.B., 1976, Reconnaissance surficial geology of the

    Gray Quadrangle, Maine. Maine Geological Survey, Dept. of

    Conservation, Augusta, ME 04333

    Townley, L.R., and J.L. Wilson, 1980, Description of and user's

    manual for a finite element aquifer flow model, AQUIFEM-1.

    Technology Adaptation Program, Mass. Inst. of Technology,

    Cambridge, MA 02139, 294 p.

    -34

  • TABLE 1--1982 WATER SAMPLE LOCATIONS AND TEST RESULTS

    (Sample locations are shown by numbered arrows on Figure 1)

    • No. Map Designation Land Owner Tax Lot Maine Coordinates DEP Sample TCE TCEy Notes/Other

    (Fig 1) (Figure 1) Map No. North East Number µg/l µg/l

    1 Gerber--BlOlA Blue Rock Indust. 33 30 383,930 465,030 Could not sample well 2 Gerber--Bl02 Lucymae Bowles 39 10 383,610 469, 720 104558 J71 Jl20 Methyl ethyl ket·:me Jl3 ,000 3A Gerber--Bl03 soil Fred. Farrell 38 13 383,080 470,220 105111 Kl Kl 3B Gerber--Bl03 rock Fred. Farrell 38 13 383,080 470,220 105137 Kl Kl 4A Gerber--Bl04 upper Emerson Mitchell 39 8 386,020 467,020 105017 Kl Kl Electron capture detection 4A Gerber--Bl04 upper Emerson Mitchell 39 8 386,020 467,020 KO.l 0.5 ERCO test results 4B Gerber--Bl04 deep Emerson Mitchell 39 8 386,020 467,020 105019 Kl Kl Electron capture detection 5 DEP Well 8 (site) Richard Dingwell 38 20 382,990 466,140 Could not sample 6 DEP Well 13 (site) Richard Dingwell 38 20 382,660 466,370 l05074 Kl Kl 6 DEP Well 13 (site) Richard Dingwell 38 20 382,660 466,370 105073 16 0.7 ERCO test results 7 DEP Well 14 (site) Michael Valente 38 35 382,690 466,070 105076 1(C Kl 8 DEP Well 15 (site) Michael Valente 38 35 382 '890 465 '890 105075 Kl Kl 9 DEP Well 7 (site) Michael Valente 38 35 383,310 466,130 10419 2 53 130,000 Methylene chloride JlOO

    10 Test Well 17 Blue Rock Indust. 33 30 382,910 464' 770 L04577 Kl -~Kl

    11 Drilled well 1 Ralph Wink 39 20 385,880 470,100 Could not sample