issn 0075-5621 kentucky geological survey university of ... · issn 0075-5621 kentucky geological...

ISSN 0075-5621

Kentucky Geological SurveyUniversity of Kentucky, Lexington

Donald C. Haney, State Geologist and Director

ISSN 0075-5621Kentucky Geological Survey

University of Kentucky, LexingtonDonald C. Haney, State Geologist and Director

GROUND-WATER GEOCHEMISTRY AND ITSRELATIONSHIP TO THE FLOW SYSTEM AT ANUNMINED SITE IN THE EASTERN KENTUCKYCOAL FIELD

Thesis Series 5Series X1, 1993

David R. Wunsch

DISCLAIMER

The Kentucky Geological Survey provides online versions of its publications as a publicservice. Publications are provided as Adobe PDF (portable document format) files. Hard-copy versions are available for purchase by contacting the Survey at:

Kentucky Geological SurveyPublication Sales Office228 Mining and Mineral Resources BuildingUniversity of KentuckyLexington, Kentucky 40506-0107

Phone: 606-257-5500Fax: 606-257-1147

Selected KGS reports published before 1999 have been scanned and converted to PDFformat. Scanned documents may not retain the formatting of the original publication. Inaddition, color may have been added to some documents to clarify illustrations; in thesecases, the color does not appear in the original printed copy of the publication. Everyeffort has been made to ensure the integrity of the text. KGS maps and charts are suppliedeither whole or in part and some are too large to be printed on most plotters. Open-filereports are reproduced from the best available copy provided by the author, and have notundergone KGS technical or editorial review.

The Kentucky Geological Survey disclaims all warranties, representations, orendorsements, expressed or implied, with regard to the information accessed from, or via,this server or the Internet.

UNIVERSITY OF KENTUCKYCharles T. Wethington, Jr., PresidentLinda J. Magid, Vice President for Research and

Graduate StudiesJack Supplee, Director, Fiscal Affairs and Sponsored

Project Administration

KENTUCKY GEOLOGICAL SURVEY ADVISORYBOARDSteve Cawood, Chairman, PinevilleLarry R. Finley, HendersonHugh B. Gabbard, RichmondKenneth Gibson, MadisonvilleWallace W. Hagan, LexingtonPhil M. Miles, LexingtonW. A. Mossbarger, LexingtonRalph Palmer, WinchesterHenry A. Spalding, HazardRalph N. Thomas, OwensboroGeorge H. Warren, Jr., OwensboroDavid A. Zegeer, LexingtonKENTUCKY GEOLOGICAL SURVEYDonald C. Haney, State Geologist and DirectorJohn D. Kiefer, Assistant State Geologist for AdministrationJames C. Cobb, Assistant State Geologist for Research

ADMINISTRATIVE DIVISIONPersonnel and Finance Section:James L. Hamilton, Administrative Staff Officer IIRoger S. Banks, Account Clerk VClerical Section:Jody L. Fox, Staff Assistant VIRegina A. Gibson, Staff Assistant VIIShirley D. Dawson, Staff Assistant VEugenia E. Kelley, Staff Assistant VJuanita G. Smith, Staff Assistant V, Henderson OfficePublications Section:Donald W. Hutcheson, HeadMargaret Luther Smath, Geologic Editor IIITerry D. Hounshell, Chief Cartographic IllustratorRichard A. Smath, Geologist III, ESIC CoordinatorRobert C. Holladay, Principal Drafting TechnicianMichael L. Murphy, Drafting TechnicianWilliam A. Briscoe, III, Publication Sales SupervisorKenneth G. Otis, Stores WorkerGEOLOGICAL DIVISIONCoal and Minerals Section:James C. Cobb, HeadGarland R. Dever, Jr., Geologist VII

Donald R. Chesnut, Jr., Geologist VCortland F. Eble, Geologist VDavid A. Williams, Geologist V, Henderson OfficeWarren H. Anderson, Geologist IVGerald A. Weisenfluh, Geologist IVStephen F. Greb, Geologist IIIRobert Andrews, Geologist IDean D. Sheets, Assistant GeologistPetroleum and Stratigraphy Section:James A. Drahovzal, HeadMartin C. Noger, Geologist VIITerence Hamilton-Smith, Geologist VPatrick J. Gooding, Geologist IVDavid C. Harris, Geologist IVBrandon C. Nuttall, Geologist IVJoseph F. Meglen, Geologist IIMatthew Humphreys, Geologist IMara Chen, Post-Doctoral ScholarJames B. Harris, Post-Doctoral ScholarRobert R. Daniel, Laboratory Technician BDavid E. McFadden, Senior Laboratory AssistantRobyn A. Johnson, Staff Assistant IVAnna E. Watson, Staff Assistant IVFrances A. Benson, Staff Assistant IVLuanne Davis, Staff Assistant IVTheola L. Evans, Staff Assistant IV

Water Resources Section:James S. Dinger, HeadJames A. Kipp, Geologist VDaniel 1. Carey, Hydrologist IVJames C. Currens, Geologist IVDavid R. Wunsch, Geologist IVAlex W. Fogle, Hydrologist IIIPhilip G. Conrad, Geologist II0. Barton Davidson, Geologist IIDavid L. Harmon, Geologist IIDwayne M. Keagy, Geologist IIShelley A. Minns, Research AssistantEd Fortner, Jr., Geological TechnicianC. Douglas R. Graham, Geological TechnicianJoyce Belcher, Staff Assistant IV

Computer and Laboratory Services Section:Steven J. Cordiviola, HeadRichard E. Sergeant, Geologist VJoseph B. Dixon, Systems ProgrammerHenry E. Francis, Associate ScientistZhalet Baharestan, Senior Research AnalystXenia P. Culbertson, Research AnalystSteven R. Mock, Senior Laboratory TechnicianMark F. Thompson, Research Analyst

GROUND-WATER GEOCHEMISTRY AND ITSRELATIONSHIP TO THE FLOW SYSTEM AT AN

UNMINED SITE IN THE EASTERN KENTUCKY COAL FIELDABSTRACT

A comprehensive hydrogeologic and hydrogeochemical study was conducted at anunmined site in the Eastern Kentucky Coal Field. Sixteen piezometers were installed toapproximate a vertical grid in a ridge that is characteristic of the geologic andhydrologic conditions of the region. Piezometer placement was based on datacollected from geologic core description, geophysical logs, water-injection packertests, and downhole camera investigations. Water levels were measured onapproximately 10-day intervals. Water samples were collected monthly to evaluatetemporal variation in water chemistry. A one-time sampling for tritium analysis wasperformed to aid in the determination of recharge and discharge areas.

Piezometers monitoring shallow, fractured bedrock and coal beds showed thegreatest temporal water-level fluctuation and hydrochemical variation. These effectsdecreased with depth below the surface. Coal beds of the Hazard series probably actto dewater the upper portion of the ridge by laterally transmitting water that dischargesas springs or seeps. Interpretation of the tritium data indicates that a significantamount of recharge enters the ridge along the hillslope where fractures, along with abreak in the degree of slope, allow for the greatest infiltration of ground water.

Temporal variation in ground-water chemistry was minimal at the site except for afew specific cases. Ground water derived from coal seams contained the lowest pHand was predominantly a Ca-Mg-HCO3 water type. Water derived from fractured zonesvaried between CaHCO3 and Mg-SO4 water types. Ground water from the ridge interioris a Na-HCO3 type, which contained a high pH and characteristically high F-. Bariumwas found in significant concentrations in piezometers near the discharge area (valleybottom) where Na-CI water types were encountered, although ground water withelevated barium concentrations was shown to exist in other locations. Sulfate reductionand cation exchange appear to control the occurrence of barium.

Reaction path modeling of the geochemical evolution of ground water at the siteshowed excellent agreement with observed trends. A conceptual model is presentedthat shows the location of hydrochemical facies zones within the ridge and outlines aset of plausible water-rock reactions for their occurrence.

INTRODUCTIONThe bedrock of eastern Kentucky consists of relatively

flat-lying, alternating layers of coal, sandstone, shale,clay, and to a lesser extent, limestone, which varydramatically in thickness, as well as laterally. Thehydraulic properties of these rock types can varyconsiderably, but at shallow depths (generally less than100 feet) the movement of water is predominantlycontrolled by shallow fracture systems. it has beensuggested that these fractures have resulted from externalunloading, or stress relief (Wyrick and Borchers, 1981).

Conceptual models have been proposed to describethe ground-water flow that occurs in coal-bearing rocksin eastern Kentucky and neighboring states in theAppalachian Plateau Province. The primary emphasis ofthese models was to define the flow system, with little orno regard to the geochemistry of the ground water. Waterquality can vary greatly within an area, and particularground-water facies appear to correlate with the locationof water-producing zones or with respect to the elevationof the local surface drainage. For example, wellsbottoming below major drainage in eastern Kentuckytend to be Na-HCO3 or Na-Cl water types, whereas

ground water from wells or springs found along hillsidesabove drainage tend to be dominated by Ca-HCO3 or Ca-SO4.

Approximately 90 percent of rural Kentuckyhouseholds use ground water as their source of water.Barium concentrations in ground water greater than theU.S. EPA maximum contaminant level of 1.0 mg/L havebeen documented in areas of eastern Kentucky. It issuspected that other minor and trace elements are presentin concentrations exceeding U.S. EnvironmentalProtection Agency (EPA) drinking-water standards.Undesirable trace elements may have a considerableimpact on the health of many Kentucky residents.

Ground-water monitoring schemes for mining andother industries located in these areas have also beencomplicated by the fact that the background geochemicalconditions have not been accurately defined ordescribed.

This study documents the relationship betweengroundwater geochemistry and the flow system thatcontrols the movement of ground water. The flow systemis described by conceptual models and confirmed byfield investigation in a control area representative of theEastern Kentucky Coal Field.

2 Ground-Water Geochemistry and Its Relationship to the Flow System at an Unmined Site in the Eastern Kentucky Coal Field

This task was accomplished by installing piezometer/monitoring wells at a relatively undisturbed site typical ofthe area. The wells were drilled using air-rotary methodswhile maintaining tight environmental protocol duringwell installation to insure well integrity and minimize anypotential contamination. Methods employed to obtainstratigraphic control included the collection of drillcuttings and the examination of geologic core.Geophysical logging of the boreholes was performed toaid in stratigraphic positioning of piezometer screens. Theholes were examined with a downhole camera todocument specific features such as water-producing hori-zons and fractures.

Data collection included an intensive schedule forcollection of water-level measurements, detailed chemicalanalyses of water samples, and pressure testing of aborehole to determine the hydraulic characteristics of thewater-bearing rocks. Other facets of the study includewater sample collection for radioisotopic analyses(tritium) for relative age dating and the delineation ofrecharge zones, laboratory studies of water-rock reactions,and geochemical modeling.

A conceptual model describing the geochemicalevolution of ground water in eastern Kentucky ispresented, which will be useful in understanding therelationship of the groundwater flow system to thechemistry of ground water in other areas with a similarsetting.

Statement of PurposeThis investigation represents one of the most intensive

ground-water studies to be performed in the Eastern Ken-tucky Coal Field. Presently there exists a deficiency ofpost 1970's ground-water chemical data that includemeasurements of major cations and anions, stable isotopes,radionuclides, and trace metals. Also, the freshwater-saline water interface is not well defined in theeastern area of the state (Sprinkle and others, 1983). Thisstudy contributes to abating this lack of data by generatingchemical and isotopic data from ground water.

Mandatory monitoring of ground water in theAppalachian coal field has only been required since theSurface Mining Control and Reclamation Act (PL 95-87)was implemented in 1977. The results of this study shouldprove valuable to individuals, industry, and regulatoryagencies because it provides a major contribution to theunderstanding of the hydrogeology and chemistry ofground water in the region.

Location of Study AreaThe goal for site selection for this study was to find an

unmined area in the Eastern Kentucky Coal Field that wasreasonably representative of the geology andgeomorphology throughout the region. This proved to be adifficult task. Much of the land in eastern Kentucky thathas not been deemed unsuitable for mining usuallyexhibits some signs of mining-related disturbance.

The location of the control site that was chosen for thisstudy is on property owned or leased by the Star FireMining Company in Perry County, Kentucky (Figure 1).The site is contained on the Noble 7.5-minute quadrangle(Hinrichs, 1978).

Geography and ClimateThe Eastern Kentucky Coal Field lies in the maturely

dissected Appalachian Plateau Province and experiences ahumid continental climate. The southeastern region of thecoal field may receive up to 50 inches of precipitation ayear and has an average annual temperature ofapproximately 57oF. The dominant cation and anion foundin precipitation are typically H+ and S04

2-, respectively,with a pH range that is typically between 4.3 and 4.5(Coltharp and Brooks, 1991). The coldest month of theyear is January, where temperatures may fall below zero.The warmest temperatures occur in July, where they mayoccasionally exceed 100oF (Price and others, 1962).

Active coal mining occurs near the study area in theform of both surface and deep mines. Active oil and gasfields, along with isolated abandoned wells, are alsolocated throughout the region. The majority of the areathat has not been recently mined is covered by forestsconsisting of broadleaf deciduous hardwood trees andconifers.

Geology and HydrogeologyThe study area is located in the dissected Appalachian

Plateau Province, which is characterized by irregular,steeply sloping ridges separated by narrow valleys. Themost common lithologies are interbedded sandstones,siltstones, shales, underclays, and coals (Spengler, 1977).Some units are locally calcareous, or may containlenticular limestones or limy concretions. The major soilgroup in the area of study is Jefferson-Shelocta, which is awell-drained soil formed primarily from the siltstones,shales, and sandstones. The soil cover is generally thin onthe steep slopes common to the area, with the thickestaccumulations in the valley bottoms and flood plains(Quinones and others, 1981).

The majority of ground water in eastern Kentucky is obtainedfrom shallow bedrock wells generally less than 100 feet deep andfrom very shallow dug wells in alluvium or regolith. Salty waterand brines are commonly encountered at depths greater than 100feet below major drainage (Price and others, 1962). The largestyields are usually derived from wells in valley bottoms andgenerally decrease up slope (Price and others, 1962). Privateresidences in the area tend to be located in valley bottomsbecause of the greater availability of water and the preferencepeople have for flat land over mountain slopes for home sites.

Sandstone aquifers are responsible for most of the waterproduced, but in most cases water is also contributed fromthe various other lithologies penetrated by the well,especially coal seams (Kiesler, 1986). The relativelyimpermeable underclays and shales that typically underliecoal act as aquitards to the vertical movement of water,which may result in horizontal flow through coal seamsand other highly conductive layers. The vast majority ofwater wells in the area are shallow, but encounter severaldifferent lithologies.

The effects of fracture- related flow in the AppalachianPlateau (Wyrick and Borchers, 1981) and the EasternKentucky Coal Field (Kipp and others, 1983; Kipp andDinger, 1987) have been documented. Fracture flowusually occurs through joints, partings along beddingplanes, and near-surface frac-

turing. These forms of secondary permeability probablycontrol shallow ground-water flow, with intergranular andconfined flow gradually increasing in importance withdepth.

Previous InvestigationsThe importance of fractures, joints, and bedding planes

as avenues for enhanced ground-water flow wasmentioned in early published reports concerning coal-fieldhydrology and ground-water resource evaluation in theEastern Kentucky Coal Field (Price, 1962). Wyrick andBorchers (1981) postulated that ground-water flow in anarea in West Virginia was controlled by shallow fracturesystems. They concluded that vertical fracture systemscaused by unloading and slumping predominate along thevalley walls, whereas horizontal fractures along beddingplanes are more prevalent in the valley

bottoms. Their study did not discuss any geochemical as-pects of ground water; neither did they confirm theirpostulations concerning the presence of fractures alongvalley sides.

Kipp and others (1983) suggested that a complex flowsystem exists in the Eastern Kentucky Coal Field. Theirfindings suggest water is stored in intergranular spacewithin clastic rocks in the hill interiors, but that a relativelysmall amount of water is produced from the intergranularstorage. Shallow fracture systems (generally less than 50feet deep) control most of the flow within the smalldrainage basins characteristic of eastern Kentucky. Acascading type of flow may occur along hillsides, wherelocal confining layers prevent vertical movement of groundwater, causing its emergence as springs along the hillsides.No ground-water chemical data were reported in theirfindings, although it was ob-


served from a limited amount of samples that the waterderived from hillsides was a calcium-bicarbonate ormixedcation type, whereas sodium-bicarbonate-typewaters were found in water-producing zones below localdrainage (Jim Kipp, personal communication). Similartrends in water types have been observed in other areas ofeastern Kentucky (Ron Yost, personal communication).

Hopkins (1966) published a salt-water interface mapfor eastern Kentucky. This document shows that salt watercan be encountered at depths as little as 100 feet belowmajor drainage in many areas of eastern Kentucky. One ofthe earliest references to fracture-dominated flow ofground water in eastern Kentucky is found in the text thataccompanies the interface map.

Bienkowski (1990) delineated several aquifers ineastern Kentucky based on estimated and measuredproduction data from monitoring- and domestic-waterwells. Differences were noted in above- andbelow-drainage portions of aquifers, and decreasing yieldsin wells were most closely related to increasing depth. Thechemical data used in this study to correlate water qualitywith the aquifers were limited to measurements of iron,manganese, and sulfate concentration, along with pH andelectrical conductance data. The lack of data and otherconstituents precluded determination of the geochemicalmechanisms related to the generation of specific watertypes encountered in the study.

Harlow and LeCain (1991) performed packer injectiontests in several hundred coal exploration core holes inVirginia to determine the hydraulic characteristics ofvarious strata, including coal. They found a pronounceddecrease in hydraulic conductivity with depth in alllithologies except coal. This decrease is consistent with thefracture-dominated flow systems as described above, andthe relatively constant hydraulic conductivity of the coalseams was probably a result of the cleating in the coal,which may be more regionally controlled than thelocalized near-surface fracturing dictated by the surfacetopography. Originally, Harlow and LeCain intended totake water samples from many of the various horizons stu-died. Unfortunately, many of the core holes had been leftopen for significant lengths of time, causing uncertaintiesin the quality of the data that would have been collected(Gary LeCain, personal communication). Thiscircumstance, along with the factors of time and cost, didnot allow for the collection of chemical data.

Little research has been performed in hydrogeologicsystems to identify the cause of high barium concentrationsin ground water (Moore and Staubitz, 1984). Studiesperformed at the Illinois State Water Survey by Gilkesonand others (11981, 1983) showed that deepCambro-Ordovician and Pennsylvanian aquifers innorthern Illinois contained high barium concentrationswith anomalously low concentrations of sulfate. It wasconcluded that sulfur-reducing bacteria were responsiblefor the reduction of sulfate to hydrogen sulfide, thuslowering the dissolved sulfate content of the ground waterand allowing high barium concentrations to remain insolution. Caithamer (1983) performed an additional studyconcerning the mineralogical sources for barium in thisarea. The data collected were not conclusive as to thesource of

the barium, although accessory feldspar in sandstones wasmentioned as a possibility.

Arora and Rodenbeck (1983) noted the occurrence ofhigh barium concentrations in ground waters found in theGeorgia coastal plain aquifers, but did not speculate as tothe source or controls of barium in solution. Moore andStaubitz (1984) studied the occurrence of high bariumconcentrations in ground water on the Seneca Nation ofIndians Reservation in southwestern New York State.Their report showed barium concentrations as high as 21.5mg/L, and they claimed this number to be the highestreported concentration of barium from any naturalground-water system in the world. They concluded thesource of barium was from the dissolution of the mineralbarite. This conclusion would seem unlikely at the sulfateconcentrations reported in their study due to the lowsolubility of barite (Ksp = 10-10; Hem, 1985). Also, Mooreand Staubitz (1984) sampled potable water that typicallycontains low dissolved solids. This suggests that the ionicstrength effect imparted by the dissolved constituents onthe dissolution of barite would probably be minimal.

Wunsch (1988a) studied the cause and occurrence ofhigh barium concentrations in ground water in easternKentucky. This reconnaissance study of barium analysesfrom 130 wells in a two-county area showed thathigh-barium ground waters were restricted to drilled,bedrock wells bottoming in Pennsylvanian coal-bearingstrata. A more detailed study included the collection ofsamples for major cation-anion and barium analyses from46 of the 130 wells sampled in the reconnaissance stage.This sampling included wells known to have high bariumconcentrations as well as those with low bariumconcentrations. Analysis of these samples showed highcorrelations between barium and sodium, calcium, andchloride. Wells sampled in and around the town ofBuckhorn, Kentucky, had the highest number of wells withbarium concentrations greater than 1 mg/L, with the meanbarium concentration of 3.11 mg/L. The Buckhornsamples were also predominantly Na-Cl type waters.Stable Isotopic Ratio Analysis (SIRA) of hydrogen andoxygen in the water molecule showed slight enrichment ofboth 180 and deuterium, probably reflecting the influenceof mixing with brines. Brines typically are enriched in 180and deuterium because of water-rock interactions overlong periods of time (Hoefs, 1987).

Fleischer (1974) produced a map showing thedistribution of fluoride in ground water on a county-widebasis for each county in Kentucky. The data in this reportindicated that some counties in eastern Kentucky hadfluoride values as high as 1.4 mg/L, while little or no datawere available for others. Hopkins (1966) found fluorideas high as 2.6 mg/L and that concentrations frequentlyexceed 0.3 mg/L in ground water in eastern Kentucky. Theoccurrence of high fluoride content in ground water hasbeen noted in northeastern Ohio (Wunsch, 1982; Corbettand Manner, 1984). The occurrence of fluoride in this areawas related to wells penetrating the Pennsylvanian strata innortheastern Ohio. It was proposed that high fluoride wasthe result of anion exchange reaction where fluoride wasexchanged for aqueous hydroxyl ions. Zack and Roberts(1988) found high fluoride concentrations

Introduction 5

2CH2O + SO42-…….. 4HC03-+ HS-+ C02 + H2O (6)

Sulfur-reducing bacteria require less than 0. 1 mg/L oforganic material to support their activity (Gilkeson andothers, 1981). Ground water in which C02 has beengenerated by sulfate reduction typically has low sulfateconcentrations; thus, it can be clearly distinguished fromhigh-Na-HC03 water that has evolved only from thedissolution of calcite, oxidation of pyrite, and ionexchange processes (Freeze and Cherry, 1979). Otherevidence for bacterially mediated sulfate reduction is thepresence of hydrogen sulfide (H2S). The unappealingsmell of this gas is generally a detriment to water qualityand is common in the Eastern Interior coal basins(Banasczak, 1980).

Ground waters derived from coal layers may alsocontain metals such as iron and manganese (Stone andSnoeberger, 1978; Banasczak, 1980). The presence ofthese elements is most likely the result of the oxidation ofsulfide minerals. Secondary mineralization (e.g., barite,calcite) often exists in fractures and partings in coal, andmay also be a source of trace metals (Stach, 1982).Banasczak (1980) also noted the presence of microscopicparticles of coal suspended in ground waters from wellsreceiving their water from coal seams, often referred to as"black water."

Danilchik and Waldrop (1978) listed the followingsuite of minerals present in the stratigraphic section thatencompasses the study site: quartz, feldspar, mica, calcite.Their data are limited in that they did not conduct detailedpetrographic evaluations or determine the relativeabundances of each mineral.

Weinheimer (1983) performed sedimentological andmineralogical analyses of thin sections from four coresrepresenting the Breathitt Formation in eastern Kentucky.Two of the cores were taken from locations approximately4 miles from the site chosen for this study, and representthe best mineralogical control available for understandingthe geochemical reactions that may control the waterchemistry at the site. Analyses of the cores revealed thefollowing average percentages: quartz, 47 percent;feldspar, mainly K-spar, 29 percent; rock fragments, 11.9percent; mica, 5.4 percent; and heavy minerals (pyrite,siderite), 0.5 percent. The majority of

in ground water in the Atlantic Coastal Plain of SouthCarolina. The occurrence of fluoride was also attributedto anion exchange reactions. Robertson (1984) andRobertson and Garrett (1988) studied the occurrence offluoride in alluvial deposits in the southwestern UnitedStates and included extensive work on the solubilitycontrols of fluoride in ground water.

Ground-Water GeochemistryIn Coal-Bearing Strata

The literature does not contain a voluminous amountof geochemical data on ground waters derived from coalseams or coal-bearing rocks. The majority of work on thistopic has been performed in the Western lignite coalfields. Table 1 shows the major cations and anions (indecreasing order), along with pH values listed by severalresearchers who conducted studies involving thegeochemistry of ground waters derived from coal orcoal-bearing strata. The studies shown represent analysesfrom the Western lignite coal fields (North Dakota andWyoming), the Paleozoic bituminous coals of theAppalachian Basin (northeastern Ohio), and the IllinoisBasin (southern Indiana). Although several water typescould be found in each of these areas, all of the aboveexamples, except for one, found that sodium andbicarbonate were the dominant cation and anion,respectively. It appears that geology and the setting of thecoal basins play a primary role in determining theground-water geochemistry in coal-bearing sediments.

Groenewold and others (1981) proposed a series ofprocesses that generate high-pH, Na-HC03 groundwaters. They can be summarized as follows:

1. C02 is produced in the organic horizons of the soilzone.CH2O + O2 � CO2 + H2O (1)2. The oxidation of pyrite liberates H+.4FeS2 + 14H2O + 14H2O � 4Fe(OH)3 + 16H+ + 8SO4

2- (2)3. Dissolution of carbonate minerals produces

bicarbonate.CaCO3 + H+ � Ca2+ + HC03

- (3)4. Precipitation and dissolution of gypsum may place

constraints on calcium and sulfate in the system.Ca2+ + S04

2- + 2H2O � 4CaS04+2H2O (4)5. Cation exchange reactions take calcium out of thesystem,Ca2+ + 2NaX(ads.) 2Na+ + CaX(ads.) (5)The shift in calcite equilibrium allows for more

dissolution of calcite (eqn. 3), which consumes H+ andresults in an increase in pH and bicarbonateconcentration. Ground waters derived from coals fromthe Western coal fields often contain significant amountsof sulfate because of gypsum dissolution.

Foster (1950) and Winograd and Farlekas (1974) havesuggested similar processes for the genesis of high-pH,NaHC03 waters in the Atlantic Coastal Plain, except thatthey proposed that C02 is produced at depth from eitherthe coalification (diagenesis) of lignite or by sulfatereduction requiring sulfur-reducing bacteria. Thereaction for the reduction of sulfate with the presence oforganic matter is:


the cement was determined to be ferroan calcite.Abundant authigenic kaolinite was found to fill porespaces and form reaction rims around feldspar grains.Opaque minerals and accessory components includedpyrite and detrital organic matter. Siderite and iron oxidewere also common. Dolomite was rare. Shales andclaystones in the Breathitt Formation contain illite,kaolinite, and chlorite (Papp, 1976). Diamond (1972)reported montmorillonite in the Magoffin Member.

Minerals that are the most abundant in rocks are notnecessarily those that have the greatest control on thedissolved constituents that occur in the water. Also, someminerals, especially aluminum silicate minerals, dissolveincongruently (Freeze and Cherry, 1979). Although thecarbonate minerals (e.g., calcite and siderite) are presentin minor amounts, often they can exert a pronouncedeffect on the geochemistry of ground water, especiallywhen the ground water is exposed to a high partialpressure of C02.

Powell and Larson (1985) related the chemistry ofground water to the mineralogy of the Norton Formationof Virginia. Their report describes in detail a suite ofreactions that may be responsible for the observed typesof water quality encountered in the coal-bearing strata ofsouthwestern Virginia. The Norton Formation of Virginiais very similar to the Breathitt Formation in Kentucky(Don Chesnut, personal communication). The study siteused by Powell and Larson is near the Kentucky-Virginiaborder, approximately 65 miles from the present studysite. The mineralogical determinations by Weinheimer(1983) also indicate a high degree of similarity betweenthe rocks of the Norton and Breathitt Formations. Becausethe geology and hydrogeology of the two areas aresimilar, the chemical reactions relating the reactions ofground water with minerals in the rocks described byPowell and Larson (1985) are considered valid to describethe reactions that control the ground-water geochernistryin this study. Therefore, a discussion based on their workis included in this study and can be found in Appendix 1.Their discussion presents an excellent description of theprobable weathering reactions taking place in or above thegroundwater system found in Pennsylvanian coal-bearingrocks in the Appalachian Basin.

Controls on Barium in Ground WaterBarium is an alkali earth element found in a number of

silicate minerals, mainly feldspars and micas. It oftensubstitutes for potassium because of the nearly identicalionic radii of 1.59 Angstrom units for potassium, and 1.50for barium. In sedimentary rocks, shales usually containthe greatest amounts of barium because it is commonlyadsorbed onto clay minerals or incorporated in theclay-mineral structure (Levinson, 1980). Wunsch (1 988a)found that barium content was the highest (1,180 ppm) inshales and lowest in coal (< 50 ppm) in cores fromcoal-bearing rocks of the Breathitt Formation in easternKentucky. These values are consistent with other valuesfound in the literature (Levinson, 1980).

Two barium compounds, barite (BaS04) and witherite(BaC03), can impose solubility constraints on barium innatural waters. Witherite has a solubility product of 10-8.64

at 25oC, which is similar to calcite (K = 10-8.48). Witheritesolu-

bility is largely dependent on the carbon dioxide partialpressure in equilibrium with the solution. The amount ofbarium in solution is most likely controlled by the solubilityof the mineral barite (BaS04) (Hem, 1985). The solubilityproduct of barite at 25oC has been given as 10-10 (Hem,1985) and 1.08 x 10-10 (Weast, 1975). Consistent with thelaw of mass action high activities of barium mustcorrespond to low activities of sulfate. For example, atsulfate molar activities near 10-4 (~10 mg/L) or 10-3 (~100mg/L) the corresponding equilibrium molar activities wouldbe 10-6 or 10-7 (~0.14 mg/L or ~0.014 mg/L), respectively(Hem, 1985).

Collins (1975) confirmed that the solubility of bariumsulfate increases with an increase in the ionic strength ofsolution. This important solubility effect is the probablereason for the high barium content in brines. Brines fromthe Appalachian Basin have been noted as containingsignificant amounts of barium (Heck, 1940). Brines high inbarium are characteristically low in sulfate content,probably due to sulfate reduction (Hem, 1985). Theresulting low sulfate concentrations allow for the bariumconcentration to increase, at least up to the point where thebarium carbonate solubility limit is reached (Hem, 1985).

Wunsch (1988b) showed that the low concentrations ofsulfate in ground water in areas of eastern Kentucky weredue to sulfate reduction mitigated by sulfur-reducingbacteria. Stable Isotope Ratio Analysis (SIRA) of sulfur inthe sulfate molecule showed that samples with low sulfatecontent had high S34, suggesting that the activity ofsulfur-reducing bacteria in the aquifer causes the reductionof sulfate to sulfide. Figure 2 shows the relationship ofsulfate content and the degree of sulfur-isotopefractionation to barium content.

Microbial culture analyses confirmed the presence ofsulfur-reducing bacteria in many of the wells from whichthe sulfur-isotope samples were drawn. The sulfur-reducingbacteria were found to be halophyllic, and probably thrivein the shallow mixing zone where brines and fresh waterproduce optimum salinities for their growth. The processaffects the chemical equilibria by keeping theconcentrations of sulfate low, thus allowing for greaterconcentrations of barium to exist in solution (Wunsch,1988b).

The effect of pH on barite solubility is thought to beminimal in natural waters. At pH values of less than about4, the HSO4- ions become important, which would tend toincrease the solubility of barite. However, most groundwaters in undisturbed settings in eastern Kentucky have pHvalues above 4, and the effect of this complex would not besignificant.

Fluoride in Ground WaterFluorine is the most electronegative (tendency of an ele-

ment to acquire a negative charge) member of the halogengroup of elements. In solutions, fluorine is found in theform of F- ions, and is generally found in natural waters inconcentrations less than 1.0 mg/L. Fluoride also may formcomplexes with aluminum and ferric iron.Aluminum-fluoride complexes are most common at low pHvalues (Hem, 1985). At a pH below about 3.5, the form HFo

could occur, although this value is below the pH range ofmost natural waters.

Fluoride ions have nearly the same ionic radius as thehydroxyl ion (ionic radius is 1.40 Angstroms; ionic radiusof F- is 1.36); therefore, these ions may replace each otherin mineral structures (Hem, 1985).

In sedimentary rocks, fluorine is found in fluorite(CaF2), amphiboles, and micas where it may alsosubstitute for hydroxyl ions. Fluoride ions may also beexchanged with hydroxyl ions at mineral surfaces inpH-dependent reactions:

XF + OH- = XOH + F- (17)Bower and Hatcher (1967) studied fluoride adsorption

on various soils and minerals. They found a high degreeof adsorption on gibbsite, kaolinite, and halloysite.

In natural waters with fluoride concentrations -5 mg/L,the solubility of fluoride is probably controlled byfluorite, such that high dissolved fluoride concentrationswould correspond to low calcium concentrations(Robertson, 1984):

CaF2 = Ca2+ + 2F- (18)

K = [Ca2+][F-]2 / [CaF2]At 250C the solubility product for fluorite has been

found to be near 10-11 (Hem, 1985). In ground waters thatare undersaturated with respect to fluorite,hydroxyl-fluoride exchange appears to be the moreimportant control on fluoride concentration (Robertson,1984).

METHODSSite Selection

The study site chosen for piezometer installation isshown on Figure 3. It is located on a ridge bounded on thenorthern and western sides by Lick Branch, a third-orderstream, and on the south by Balls Fork, a fourth-orderstream. Both of these streams drain large basins andmaintain their flow throughout the year. Contour-cutmining to remove the Hazard Nos. 7 and 8 coals occurredto the east of the study site during the 1970's. The locationschosen for piezometer placement were selected tominimize any effects of this mining on the occurrence andchemistry of ground water encountered during the study.The western edge of the bench resulting from mining isapproximately 500 feet from the nearest piezometer site,and it is also outside the apparent surface-water (andassumed ground-water) drainage divide as shown on Figure3.

Access for drilling and subsequent sampling was madepossible by the existence of a former coal-hauling road lo-cated on the northwestern side of the ridge. An access roadthat branches from the haul road was constructed to allowaccess to the top of the ridge. The effects, if any, on theshallow ground-water flow system due to the constructionof this access road are unknown, but are believed to beminimal. Its construction was completed 3 months beforethe commencement of well drilling for this study, and it isassumed that any disturbance in the flow system wouldhave declined. Kipp

Methods 9

Suggested Practices for the Design and Construction ofGround-Water Monitoring Wells" by Aller and others(1989).

Materials used in the installation of piezometers werefactory sealed. Detailed well construction diagrams alongwith specifications for the installation of piezometers canbe found in Appendix 2.

The units or water-producing zones to be monitoredwere chosen using data and observations from theexamination of geologic core, drill cuttings, water-injectionpacker test data, downhole camera investigations, andgeophysical logs.

Drill CuttingsSamples of the rock cuttings produced while drilling the

boreholes for piezometer emplacement were collected at5-foot intervals for each hole. The cuttings were placed inplastic bags and labeled for identification. Boreholelogging by field examination of the cuttings was performedin holes that bottomed at elevations below the level of CH1066 so that a continuous record of the geology of the ridgecould be obtained.

Geologic CoreA geologic core (Star Fire CH 1066) was drilled by the

Star Fire Mining Company for exploration purposes on Au-gust 8, 1990. The core hole was taken from the ridge topand was located between the locations for well 12 and 13(see Figure 4). The core was an NX hole (3-inch diameter)drilled to a depth of 210 feet. The depth was limited by theexploration needs of the mining company.

The core was described for lithology and features suchas fractures that may be indicative of water-producingzones. The geologic column adjacent to the profile shownon Figure 5 is based on the geologic data obtained from CH1066 and from the geologic quadrangle map (Danilchik andWaldrop, 1978) that encompasses the site, and from drillcuttings obtained from wells 11 and 41.

Water-Injection Packer TestsWater-injection packer tests were performed to

determine the hydraulic conductivities for the variouslithologies encountered in core hole 1066. The tests wereperformed by using a packer setup in which a 5-footperforated section of pipe was separated by two inflatablepackers. The perforated pipe permitted the injection ofwater into the zone of interest, which was sealed off by thepackers.

The method used for calculating hydraulicconductivities was taken from the U.S. Bureau ofReclamation (1974) technical publication, "The Design ofSmall Dams." The formula used is:

K = [Q / 2πLH] * [In / R] For L > 10r

Where:K = permeabilityQ = constant rate of flow into the holeL = length of the portion of the hole testedH = gravity head + pressure headr = radius of hole

and others (1983) showed that water levels in wellsdirectly affected by contour-cut mining returned topre-mining levels within 4 months. Additionally, thedisturbance to the landscape due to the installation of aone-lane access road is much less than a hillside contourcut created for mining purposes.

Ground-Water Monitoring ObjectivesThe piezometers were located so that comparisons

could be made between the head distribution and waterchemistry of ground water from the ridge interior,near-surface fracture zone and sites adjacent to and belowthe major surface-water drainage adjacent to the site. Inorder to make these comparisons, detailed geologic andhydrogeologic information was collected to provide for anaccurate monitoring scheme.

Monitoring Well LocationsFour locations along the access road on the

northwestern side of the ridge were selected for welldrilling and subsequent piezometer installation. Each ofthese locations were chosen on the following bases: (1) tomaximize the monitoring of all the prevalenthydrogeologic conditions at the site, (2) to provide a fairlyeven spacing from the ridge top to the adjacent valleybottom, (3) to provide easy access for the drillingequipment, and (4) to be as close as practicable to themain road to ensure access for sample collection duringwet weather conditions.

An 8-inch surface casing was set by drilling an 8-5/8-inch pilot hole to a depth that insured that the casingbottom was approximately 10 feet into the solid bedrock.The hole was then completed to the predetermined depthusing a 7-7/8-inch roller-cone drill bit. A minimumamount of water was added during drilling for dustcontrol.

The boreholes were drilled using a Schramm Rotodrillair-rotary drilling rig. The driller was required to keep thedrilling rig and equipment clean to avoid contaminationfrom synthetic byproducts used when drilling, such ashydraulic oils and fuel. All drilling rods and tools werekept on a flatbed truck or on plastic sheeting to avoidcontamination. Steel drilling rods were decontaminatedafter the completion of a hole by washing them withchlorinated water to avoid cross-contamination betweenboreholes.

The location of the wells can be seen on Figure 4. Aprofile of the ridge that passes through the wells along theline shown on Figure 4 is shown in Figure 5. Four wells,numbered 11, 12, 13, and 14, are located at the top of theridge, including a deep well (418 feet total depth) thatpenetrated the center of the ridge below the level of thesurrounding major-surface drainage. Wells 21, 22, 31, and41 are at descending locations on the access road alongthe slope of the ridge.

Piezometer InstallationSixteen piezometers were installed in the eight holes

during November and December of 1990. The installationof each piezometer was performed under the directsupervision of the author and staff from the KentuckyGeological Survey. The methods that were followed forthe construction of multiple-level monitoring wells aredescribed in the "Handbook of


Downhole Camera InvestigationsA PLS International 8750 Portable Television Inspection

System was used to view the open monitoring-well bore-holes. The viewing of each borehole was videotaped using aVCR and recorded on standard VHS-format videotape forlater viewing. The video recorder failed halfway through thetests, so recordings of the inspections are not available fortwo boreholes. In anticipation of this problem, written fieldnotes were taken as the examination of each hole took placeto insure the data would not be lost.

Geophysical LoggingGeophysical logging was performed by the staff of the

Kentucky Geological Survey on November 29,1990.Natural gamma, SP resistivity, and caliper logs werecollected.

Ground- Water-Level MeasurementsGround-water levels were measured by using a Slope In-

dicator Co. Model 51453 electric water-level meterequipped with an indicator horn and light. The indicatorprobe and lead wire were rinsed with distilled waterbetween each measurement and stored in a plastic bagduring transport to and from each site to avoidcontamination. Measurements were recorded to the nearesttenth of a foot.

Precipitation MeasurementsPrecipitation data were collected by a tipping bucket rain

gage equipped with a continuous digital data recorder. Thegage is located on the Star Fire Mine propertyapproximately 1.8 miles from the study site.

Water SamplingWater samples were collected on a monthly basis for the

period of 1 year beginning in February of 1991 and endingin January of 1992. Purging and sampling for the deepestpiezometer and other wells producing a limited amount ofwater was performed by using a stainless steel bailer with aTeflon ball and seat. The bailer was retrieved from thepiezometer using a pump-hoist truck equipped with asandline capable of lowering and retrieving the bailer at arate of up to 400 feet per minute. The sandline cable is3/8-inch nylon-coated steel wire, which allows forflexibility and easy decontamination. A polypropylene ropeleader was used to secure the bailer. The pump-hoist truck isoperated by the Water Resources Section of the KentuckyGeological Survey.

The remaining wells were routinely purged and sampledusing a Grundfos Redi-Flo 2-inch submersible pump. Wellswere purged of three well volumes before initiating samplecollection. The handling of the bailers and the pump wasperformed while wearing latex disposable gloves to preventcontamination. All objects that were lowered into the holewere placed on plastic sheeting and covered when not inuse.

Specific conductance, temperature, pH, and redox poten-tial (Eh) were recorded in the field. A YSI 3500flow-through meter was used to obtain the temperature, pH,and redox potential values. This device is an enclosed cellhosting the appropriate electrodes (pH andoxidation-reduction potential [ORP]). The cell wasconnected directly to the pump discharge hose or connectedto the bailer by way of an emptying

valve and hose to allow for simultaneous monitoring ofthe stabilization of the field parameters while purging thewell. The pH probe was standardized by calibrating to pHbuffers of 4.01, 7.0, and 10.0 (standard pH units) beforesampling each piezometer. The YSI meter is equippedwith automatic temperature compensating probes (ATC)so that the pH values are automatically corrected fortemperature. In order to check the integrity of the pHelectrode and determine the amount of drift throughout theperiod of this study, the probe was checked periodicallyby measuring the millivolts of pH buffers to determine theslope of the best-fit line describing the relationshipbetween pH and millivolts (mV). The ORP was checkedperiodically against a solution containing a known andstable redox couple (ZoBell's solution) to determine ifelectrode poisoning had occurred (Nordstrom, 1977).

Specific conductance was measured using a Cole-Parm-er 1481-55 conductivity meter. The meter is equippedwith an ATC probe that corrected all values to atemperature of 25oC. The conductance readings of watersamples were corrected to the values of conductivitystandard solutions by linear regression using the actualand observed reading from standards of 200, 2000, and20,000 microsiemens measured in the field each day

Water samples for chemical analyses were placed in250-ml acid-rinsed polyethylene bottles. Water samplescollected for dissolved metals were filtered by pumpingwater using a peristaltic pump through a 0.45-micronmembrane filter housed in a Fisher stainless-steel andTeflon filtering device. Two ml of a 1:1 mixture ofdeionized/distilled water and double-distilled HN03 wereadded to each sample for preservation and to prevent theprecipitation of solutes as hydroxides (Brown and others,1970). Samples collected for total metals analysis wereacidified but not filtered. Samples collected for dissolvedanionic species were neither filtered or acidified.

Samples were collected on a few occasions for total or-ganic carbon (TOC) and total metals. TOC samples werenot filtered and were placed in a 250-ml bottle andacidified and preserved by adding 1 ml of concentratedH2SO4- Chemical analyses of water samples wereperformed by the Laboratory Services Section of theKentucky Geological Survey. Analysis of metals wasperformed using a Thermal-Jarrell Ash InductivelyCoupled Plasma Emission Spectrometer (ICAP).Bicarbonate analysis was performed using a Mettlerautotitrator. Sulfate, chloride, and bromide analyses wereperformed using a Dionex ion chromatograph. Fluoridewas determined using a specific ion electrode. Totalorganic carbon was determined using a Dorhmann totalorganic carbon analyzer. A description of the specificmethod used for each analysis is given in Appendix 3.

Tritium AnalysisWater samples were collected on one occasion to deter-

mine the relative age of the ground water at the site and toadd insight to the rate of recharge. Tritium samples wereplaced in 125-ml boro-silicate glass bottles with conicalinset caps to displace any air bubbles created when fillingthe bottles. The caps were double-tightened and tightlywrapped with electrical tape and sent to the University ofMiami Tritium

Methods 13

Statistical AnalysesStatistical analyses of data were performed using the

STSC Statgraphics statistical software on a Zeos 386-SXmicrocomputer. In some cases univariate statistics were per-formed using a hand calculator or a spreadsheet program.

Geochemical ModelingGeochernical modeling of ground-water chemical data

allowed for the determination of the speciation of ions andthe degree of saturation with respect to several minerals. Insome cases, reaction-path modeling was used to validate aconceptual model covering the evolution of ground water atthe site. Ground-water mixing scenarios are also presented.

The geochernical model PHREEQE (Parkhurst and oth-ers, 1980) was used to perform the equilibrium calculationsin this report. BALANCE (Parkhurst and others, 1982) wasused for mass-balance calculations. Binary mixing of watersamples was modeled using HC-GRAM (McIntosh andMiller, 1989). The models were operated on a 386-SXmicrocomputer equipped with an 80387 math co-processorchip. The reader is referred to Appendix 3 for a generaldiscussion on the capabilities of each model.

RESULTSGeologic Core

A total of 210 feet of NX core was examined and loggedto evaluate potential monitoring zones. A lithologic descrip-tion of this core is shown in Figure 6. Four major coalseams, the Hazard Nos. 5, 6, 7, and 8, were encountered inthe cored section. Fractures were observed in the coreintervals from 28.6 to 33.6, 58.6 to 63.6, and 153.6 to 158.6feet.

Drill CuttingsSamples of drill cuttings were collected generally at

5-foot intervals from monitoring-well boreholes to obtaincontrol for the complete geologic column for the study siteand for laboratory experiments. In some locations cuttingswere collected from 10-foot intervals. A listing of thesampling interval and description of the cuttings is includedin Appendix 4.

Water-Injection Packer TestsFigure 6 shows the hydraulic conductivity values asso-

ciated with each 5-foot test interval in Star Fire core hole1066. A description of the core lithology of each intervalalong with any features of hydrogeologic significance is in-cluded. Eleven intervals out of a total of 39 took significantamounts of water; most of these intervals correlated toeither a coal seam or a fracture zone (as revealed from thecore removed from the borehole). Hydraulic conductivityranged from 3.5 x 10-7 to 9.3 x 10-4 cm/sec. The highesthydraulic conductivity values were obtained from theHazard Nos. 5, 7, and 8 coal seams.

Downhole Camera InvestigationsExcellent resolution of the boreholes was observed for

the portions of the holes that were not filled with water.Visibility was usually lost after the camera was loweredbelow the level of static water due to turbidity. The videoexamination provided great insight as to where water wasentering the hole.

Laboratory for analysis. Low level-proportional countingwas performed on each sample for 1,000 minutes. Thisproduced an accuracy and precision of ±2 tritium unitsTU) (Dr. Gote Ostlund, Tritium Laboratory, personalcommunication). Samples resulting in counts of less that 5TU were recounted using enrichment and low-level directcounting, which allowed for precision of less than 0.5 TU.A detailed description of sample collection andpreparation methods suggested by the lab can be found inAppendix 3.

Quality ControlA duplicate sample from a randomly chosen piezometer

was taken during most sampling events and sent to theKGS analytical laboratory for analysis for comparisonwith the results of the original sample. A field blank wasalso prepared in the field during each sampling event todetermine if any contamination was introduced to watersamples during collection. This involved passingdistilled/deionized water through the filtering device andacidifying the filtrate for total metals analysis. Theanalytical laboratory also performs detailed qualitycontrol/quality assurance procedures within the lab (HenryFrancis, Laboratory Manager, personal communication).

Laboratory ExperimentsWater-Quality Effects From Well Construction Materials

A gravel pack consisting of fine silica sand was used tocreate a filter pack around the piezometer well screens.Visual examination of the sand revealed that it consistedof mainly quartz sand with minor amounts of feldspar. Inorder to determine if any significant contribution ofdissolved constituents found in water samples could beattributable to the gravel pack, an experiment wasconducted where 200.0 grams of the sand used in wellconstruction was placed in an acid-rinsed 1 -galloncontainer and filled with deionized/distilled water. A250-ml water sample was collected from the containerthree times throughout the duration of this study andanalyzed for total metals by the lab.

Barium Exchange ExperimentsExperiments were conducted to determine the amount

of barium that could be removed from rock samples fromdrill cuttings collected during monitoring-well drilling.The tests for barium extraction consisted of mixing equalamounts (by weight) of drilling cuttings from within thestratigraphic interval to be tested. The rock chips wereground up to create fresh reactive surfaces and air dried.Ten grams of each sample were placed in a 250-mlpolyethylene bottle and filled to volume. The solutionswere a 10 and 100 mmol NaCl solution along with adistilled/deionized water blank. One set of samples werereacted with a disodium-EDTA (ethylenediamine-tetraacetic acid) solution to solicit maximum extraction.Each sample set was run in triplicate to demonstratereproducibility. The samples were agitated for 1/2 hourand allowed to react for 72 hours. The solutions werefiltered using a 45-micron filter and analyzed.

14 Ground-Water Geochernistry and Its Relationship to the Flow System at an Unmined Site in the Eastern Kentucky Coal Field

where MEQc = cations in millequivalents and MEQa =anions in millequivalents (Freeze and Cherry, 1979). Manylaboratories consider charge-balance error of less than 10percent as acceptable, although advanced analyticalequipment is making it possible to obtain results with errorsless than this. Figure 12 is a bar graph with the distributionof charge-balance error and the corresponding percent oftotal analyses. The results achieved for this study are good,with 92 percent of the performed analyses with an error ofless than 10 percent. The average charge-balance error forthe 129 analyses was 5.36 percent.

Laboratory tests were performed to evaluate thechemical effects, if any, of the sand used as a filter packaround the piezometer screens. Samples were drawn from a4-liter con-

Results 15

Fractures and coal seams were easily identifiable, andoften water was observed flowing into the hole from thesefeatures. The camera viewing of each hole was recordedon videotape for close examination. Figure 7 showsphotographs reproduced from the videotape and shows afracture observed in well 14. The two photographs shownon each figure show either a close-up or an alternateviewing angle. Figure 8 shows a vertical fracture andcollapse of fractured rock in well 11. A fracture plane thatis infilled with clay or sediment is shown in Figure 9. Theplane defined by the fracture is inclined to the borehole, sothe outline of the fracture appears as an ellipse in thephotograph. Water flowing from the Hazard No. 7 coalseam is shown in Figure 10. Diagnostics from eachborehole were described and are included in Appendix 4.

Geophysical LoggingCaliper, natural gamma, and SP resistivity logs were

run in wells 11 and 41. These wells were chosen becausethe geologic sequence encountered in these wells overlap,allowing for a continuous logging sequence to be recordedfrom the ridge top to below drainage. Profiles of theselogs, including geologic contacts and units of interest, arepresented in Appendix 4. The geophysical data, inconjunction with the geologic data, aided in determiningthe location of coals and lithology changes, which wasimportant in accurately determining depths wherepiezometer screens were to be positioned.

Zones Monitored by PiezometersCareful evaluation of the results of geologic,

geophysical, and hydraulic data resulted in the selection ofthe discrete zones chosen for monitoring. The monitoringscheme resulted in a total of 16 piezometers beinginstalled in the eight drilled holes. The piezometer with thedeepest screened interval in each borehole was designatedthe "A" piezometer, followed by "B" and "C" for the otherpiezometers in order of decreasing depth from the surface.

Based on the data from the water-injection packer testsand downhole camera investigations, only coal seams andfractured bedrock, which includes vertical fractures,joints, and bedding-plane splits and partings, possessedsufficient hydraulic conductivities to transmit significantamounts of water. This observation is supported by thework of other researchers who studied the flow systems innearby or similar settings (Wyrick and Borchers, 1981;Kipp and Dinger, 1987; Harlow and LeCain, 1991).Therefore, water-producing zones monitored in this studycan be placed into three groups: (1) fractured bedrock,both near-surface and deep, (2) coal seams, and (3) solidbedrock zones monitored below the Magoffin ShaleMember at or below major drainage. Table 2 lists thepiezometer identification, the depth of each individualpiezometer, and the hydrogeologic zone to be monitored.

Precipitation DataFigure 11 shows the precipitation totals for each month

in 1991. The highest recorded precipitation occurred in thesummer months of June and July. The lowest monthlytotal occurred in October, when less than 1 inch of rainfell.

Examination of the daily precipitation data (Appendix5) shows that precipitation in the winter and early springmonths was relatively evenly distributed. The precipitationthat occurred in the late spring and summer was fromintense storm events. Because of the steep slope, andbecause maximum evapotranspiration occurs in thesummer months, it is presumed that most of the intenserain that fell during the spring and summer left the site assurface runoff or was transpired by vegetation, leaving asmaller amount for infiltration and recharge to theground-water system. The period of maximumground-water recharge is most likely in the winter andspring because of the relatively even rate of precipitationand low evapotranspiration.

Ground-Water LevelsStatic water levels were measured at roughly 10-day in-

tervals during the 376-day period from January 11, 1991,through January 23, 1992. During this period, 14 of the 16piezometers had water present in sufficient quantity tomeasure. Two of these piezometers, 13B and 14A, wentdry at times during the period. Continuous water levelscould thus be recorded from 12 piezometers.

Table 3 lists the water-level elevations of all thepiezometers. The maximum, minimum, and range ofwater-level fluctuation are also given for each piezometer.

CHEMICAL ANALYSESQuality of Analyses

Duplicate samples were collected to evaluate the preci-sion of the analytical laboratory. A different sample waschosen for duplication so that a sample from almost everypiezometer was duplicated throughout the sampling period.The results of analysis and the ID of the sample itduplicated are presented in Appendix 6. Overall, the resultsfrom the lab showed excellent precision.

An indication of the accuracy of water analyses data canbe obtained by calculating the charge-balance error, whichis the calculated difference between the sum of the cationiccharges and the sum of the anionic charges divided by thesum of the cation and anion charges; it is represented bythe expression


Figure 7. Photographs showing collapse of borehole along a fractured zone in well 14. To obtain color copies ofthese photographs, contact the Kentucky Geological Survey.

Chemical Analyses 17

Figure 8. Vertical fracture encountered in well 11. To obtain color copies of these photographs, contact theKentucky Geological Survey.


Figure 9. Infilled fracture encountered in well 13. Fracture plane is inclined to the vertical borehole, forming anelliptical pattern. Clay or sediment can be observed oozing from the fracture. To obtain color copies of thesephotographs, contact the Kentucky Geological Survey.

Figure 10. Water entering the borehole from the Hazard No. 8 coal in well 11. Bright areas on the left side of thephotographs are caused by the reflection of the camera light where water is flowing out of the coal. Areas whereFe-oxides have precipitated are outlined. To obtain color copies of these photographs, contact the KentuckyGeological Survey.


tainer holding the sand and distilled water and analyzed forsignificant metal content at roughly quarterly intervalsduring the period of study. The results of analyses areincluded in Appendix 6. The test results reveal that noappreciable metals concentrations result from leaching ordissolution of minerals contained in the sand. Theconcentrations of all metals were below 0.5 mg/L. Onlysix metals out of an analytical suite of 30 were present inconcentrations above the limit of detection of the ICAPinstrument. Therefore, it is highly unlikely that the resultsof chemical analyses from water samples would besignificantly affected by mineral-water reactions betweenformation water and the sand-pack material.

Tritium AnalysesThe results of the tritium analyses for water samples

collected in May of 1991 are listed in Table 4. Thirteenpiezometers had sufficient water from which a samplecould be collected for analysis. Water samples with tritiumvalues below 6 TU were resampled in June of 1991 andanalyzed using enrichment, which results in higherprecision and accuracy.

Water SamplesWater samples were collected on a monthly basis for the

period of 1 year beginning in February of 1991 and endingin January of 1992. All piezometers, with the exception of12B, 13B, and 21B, produced sufficient water for samplesto be drawn each month. Piezometers 13A and 14A hadmonths when the water level dropped below a level wheresatisfactory samples could be taken, and samples were notcollected during 8 of the 12 months for 13A; also, 14A islacking 1 month of data.

Compilation of chemical data from field and laboratoryanalyses is contained in Appendix 6. Univariate statisticsfor each ion or applicable parameter, including themaximum and minimum value, mean, standard deviation,coefficient of variation, and the number of samples areshown.

Examination of the tabulated data revealed that thechemical data for the first suite of water samples collectedfor this study in February of 1991 may not berepresentative of the water chemistry for each of thepiezometers from which samples were drawn. It can beseen from the tabulated data (Appendix 6) that theconcentrations of several parameters are



either considerably lower or higher when compared to thedata from all subsequent analyses. This disparity in thedata is most likely due to insufficient development afterthe piezometers were completed. It was noted that duringthe first sampling episode that water drawn from somepiezometers still exhibited considerable turbidity, mostlikely a remnant of well construction. Therefore, thechemical data resulting from samples taken on February 5and 6, 1991, will not be included in statisticalcalculations, graphical portrayal of data, or discussions inthis study. The samples used for discussion in this reportwill therefore be those collected from March 1991through January 1992.

Fluoride in Ground WaterThe occurrence of significant fluoride (in this study

concentrations above 1.0 mg/L are considered significant)in ground water at the site is restricted to piezometers 11Aand 22A. Both of these piezometers monitor the samezone, which is the ridge interior at an elevation that is justbelow drainage. The mean value for fluoride in these twopiezometers is 1.59 and 1.02 mg/L, respectively. All ofthe water samples from the other piezometers havemaximum and mean fluoride concentrations below 1mg/L.

Barium in Ground WaterTwo piezometers, 41A and 41B, show maximum and

mean values for barium exceeding 1.0 mg/L. This value isthe maximum contaminant level (MCL) currently suggestedby the U.S. Environmental Protection Agency (U.S. EPA,1975). The mean value for barium concentration inpiezometer 41A is 37.4 mg/L, and 1.72 mg/L in 41B.Piezometer 12A showed an increase in barium content witheach subsequent sampling, and the concentration exceeded1.0 mg/L during the period from September 1991 throughJanuary 1992.

Barium-Sodium Exchange ExperimentsThe results of laboratory experiments to determine the

potential exchangeabilty of sodium for barium on rockparticle surfaces from the study are shown in Table 5. Ahigh linear correlation (r = .99) exists between the amountof barium extracted into solution and the concentration ofthe Na-Cl solution used. A 100-millimolar Na-EDTAsolution was used to determine the maximum extractablebarium from the samples. The samples derived fromborehole 11 had a mean value of 1.43 mg/L of Ba (averagebased on triplicate samples) and 1.64 mg/L of Ba forsamples from borehole 41.


Total Organic CarbonSamples were collected for the analysis of total organic

carbon content (TOC) during September and October of1991 (see Appendix 6). The results of analysis show thatonly

three piezometers, 14A, 22B, and 22A, containedsignificant amounts of TOC. The instrument used todetermine TOC only has a resolution of 1 mg/L; therefore,these analyses do not preclude the occurrence of TOCconcentrations less than


1.0 mg/L. Piezometer 22B contained the highestconcentration with 24 mg/L, followed by 8.0 mg/L in 22A,each in September of 1991. Piezometer 14A showed 4.0mg/L in October of 1991.

DISCUSSIONThe Ground-Water Flow System

Figures 13 and 14 show a traverse of the study siteshown on Figure 1 with contours of potentiometric headconditions based on water-level readings from each of thepiezometers that contained measureable water. Figures 13and 14 show the head conditions based on the water levelsfor each piezometer recorded in February and November1991, respectively. The equi-potential lines are generatedby contouring the potentiometric head level correspondingto the elevation of the bottom of each piezometer. Theresulting contours of the head levels are shown on thetraverse along a line that passes through each cluster ofpiezometers.

This graphical representation of the potentiometric headconditions at the site is useful for examining the headgradient of ground water and defining the apparentground-water flow system. However, the traverse on whichthe contours are shown is a two-dimensionalrepresentation and lacks lateral control where othercomponents of flow could be defined. Therefore, flowlines or any other representation of actual ground-waterflow are not shown on this illustration.

Water-level data from February and November werechosen for representation in order to evaluate the changesbetween the periods of maximum and minimum waterlevels throughout the year. Overall, the minimum andmaximum water levels measured in each piezometeroccurred during the same month for all piezometers. Allbut two piezometers, 13A and 21A, achieved themaximum head reading in February of 1991. Piezometers13A and 21A reached their maximum lev-

els in April. All but three piezometers fell to theirminimum level in November 1991.

Both figures show a consistent drop in head from theridge top to an elevation below drainage (Lick Branch).The slightly closer spacing between contour lines in theupper area of the ridge indicates a higher vertical gradientthan the lower section. The equal head contours in theupper interior section of the ridge correspond closely totheir measured elevation (shown on the Y axis of bothfigures) within the ridge. This reveals that the total headlevels in the wells in the upper reaches of the site areapproximately equal to the elevation head, suggesting thatthe ground-water conditions near the ridge top areunconfined or semi-confined and are mostly a function ofthe elevation of where the piezometer is located. Thepresence of dry piezometers in this area indicates thatportions of the upper part of the ridge are unsaturated.The nearly horizontal attitude of the contours in this areaimplies a downward, vertical movement of ground water.

The contour lines at lower elevations (920 feet andbelow) exhibit a moderate amount of downwarddeflection in the area from approximately 200 feet abovedrainage to the elevations below drainage (i.e., LickBranch). The piezometric contours in this area are wellbelow their corresponding elevation level, indicating thatthe pressure head at this level provides a significantcontribution to the total head.

Figure 14 shows the contoured head elevations whenthe water levels were at their lowest conditions. There isvery little noticeable change in the shape or intensity ofthe contours between Figure 13 and Figure 14, stronglysuggesting a relatively consistent flow field remainsthroughout the year.

Piezometers Monitoring Fracture ZonesFour piezometers, 12A, 14A, 21B, and 31B, monitor f

ractured bedrock zones. Piezometer 21B did notaccumulate water during the study period. Piezometer12A is screened in

Discussion 25

a sandy-shale zone above the Hazard No. 5 coal. No frac-tures in the rock were observed at this elevation bydownhole camera observation because visibility was lost;however, examination of the geologic core indicatedfractures in shale at this elevation (the location of the corehole, CH 1066, is approximately 20 feet from the boreholecontaining piezometer 12A). The hydrograph forpiezometer 12A is shown on Figure 15. The hydrographshows that the water level remains fairly consistentthroughout the record period, with a subtle decline inwater level from its maximum of 1, 122.3 feet to its mini-mum of 1, 122.0 feet in December. The hydrograph doesnot seem to correlate well with the trends in precipitationmeasured at the site. Apparently, the fractures present inCH 1066 either do not intersect the immediate areamonitored by piezometer 12A, or fractures becomeineffective in transmitting recharging water to this depth.

Piezometer 14A is open to the shallow fracture zoneabove the Hazard No. 8 coal. The hydrograph for this pie-zometer is shown in Figure 16. The range of water-levelfluctuation is 7.7 feet. The hydrograph shows a fluctuatingwater level that remains between 1,248 and 1,252 feetfrom the beginning of the record period to August. Thewater level drops throughout the fall months, and thepiezometer becomes dry

in October. Water accumulates in 14A again in lateNovember and quickly rebounds up to near the level ithad attained at the beginning of the study. Thehydrograph exhibits a close correlation with the monthlyprecipitation data observed for the study period. Themaximum water level of nearly 1,252.2 feet is reached. inFebruary of 1991. The dry period for this piezometercorresponds well to the period of lowest precipitation(October) measured at the site. Water once again ac-cumulates in the piezometer following an increase inprecipitation in November. This pattern indicates that theshallow fracture monitored by piezometer 14A is greatlyinfluenced by atmospheric recharge and that the fracturesmaintain a good connection with the ground surface.

The hydrograph for piezometer 31B is shown in Figure17. This piezometer is screened at a zone where fractureswere observed at a depth of approximately 100 feet. Thehydrograph reveals an initial drop in water level by over 1foot. A slight rise in water level occurs from Februarythrough June, before falling to a minimum value inDecember. The hydrograph for this piezometer isgenerally smooth, more closely resembling the patternexhibited by 12A than the erratic hydrograph displayedby 14A. This observation suggests that the rapid andpronounced effect of recharge in fractured bed


rock decreases with depth below the surface.Additionally, the hydrograph for piezometer 31B may bean indicator of the effect of the degree of saturation of thehillside fracture system compared to the system at theridge top, which appears to exhibit saturated conditionsonly during wet periods. Piezometer 31B, representingthe hillside system, remained saturated throughout thestudy period. The water present in the hillside fracturesystem may act to buffer the incident recharge, whereaspiezometers monitoring fractures at the ridge top (e.g.,14A) would exhibit water-level pulses in response torapid recharge entry because the unsaturated fractureswould likely offer less resistance to rapid changes inground-water movement.

Piezometer 21B was screened in the shallow near-sur-face zone where fractures were observed with thedownhole camera. Piezometer 21B remained drythroughout the study period. The lack of wateraccumulation in this screened fracture zone indicatessome shallow fracture zones remain unsaturated. Theinfilled fractures observed by the downhole camera (seeFig. 9) support this observation.

Piezometers Monitoring Coal SeamsSix piezometers, 11B, 11C, 12B, 13A, 13B, and 21A,

were installed to monitor water in coal seams. Figure 18shows the hydrograph for piezometer 11C, whichmonitors the Hazard No. 7 coal. The maximum water levelwas recorded in February and the minimum level recordedin December. The greatest fluctuations occurred in thewinter months of February through March, with a steadydecline in water level taking place for most of theremainder of the year. This piezometer exhibited thelargest water-level fluctuation of all piezometers at the site(11.1 feet). A correlation of the water-level fluctuationwith monthly precipitation is observed. The steep drop inthe hydrograph that occurs during February and March issomewhat opposite the precipitation pattern observed atthe site during the same period. One possibility is that thespring awakening of vegetation may have resulted in thesharp water-level drop due to increased vegetative demandon the shallow ground-water supply.

A gradual decline in water level is observed for most ofthe remaining months, which may be due to increasedevapo-

Discussion 27

transpiration, along with the loss of recharge due to thehigh runoff usually associated with intense summerstorms, which composed most of the precipitation duringthe period of May through August. The minor peaks inwater level that seem to correspond to the increase inprecipitation observed in May through July indicate thatsome water from these storm events is able to enter thecoal seam during or after these events.

The hydrograph for piezometer 11B, which monitorsthe Hazard No. 5 coal, is shown on Figure 19. The waterlevel in this piezometer remained relatively unchanged,and ranged between 1,093 and 1,094 feet. Very littlecorrelation with the precipitation trends is observed, withthe exception of the sharp drop in water level inNovember, which may be a delayed response to the lowprecipitation measured in October. Piezometer 11B isscreened at a depth of 195 feet. A significant amount ofrock mass lies between the coal and the ground surface,which probably acts to buffer the movement of waterdownward to this level, resulting in the stable hydro-graph. Additionally, the occurrence and frequency offractures that transmit water at depth tend to decreasewith increasing depth from the surface (Wyrick andBorchers, 1981; Kipp and Dinger, 1987).

Piezometer 12B is open to an interval that includedthe Hazard No. 6 coal and accumulated no water duringthe re-

cord period. Water injection data (see Fig. 6) indicate thatthe interval that includes this coal had a higher hydraulicconductivity than the surrounding rock, suggesting that thiszone should contain water if it is stored in adjacent rocks.It is possible that the No. 6 coal was inadvertently sealedby bentonite or drill cuttings during drilling or wellconstruction, which would act to retard the production ofwater. This sealing could have occurred because the well islocated in the near-surface fracture zone where bentoniteand cuttings would be free to move into the fractures orcoal cleats, thereby sealing them.

The hydrograph for piezometer 13B is shown in Figure20. This piezometer exhibits a similar hydrograph patternto piezometer 11C, but within a smaller range. The rangeof water level fluctuation in 13B is 2.3 feet. Both of thesepiezometers monitor the Hazard No. 8 coal. A comparisonof the water-level data for each of these piezometersreveals that the difference in maximum water levelbetween these two piezometers is 12 feet. Also, piezometer13B became dry in August, whereas 11C did not.Apparently the screened interval for 11C may be in contactwith a fracture system in which rapid recharge can enterthat may not be present in 13B. Examination of thegeologic core showed that the Hazard No. 8 coal alsocontained several splits consisting of fire clay and shale. Adifference in the exact placement of the screens in


relation to the splits could also explain the difference inhead if the coal layers contributing water to eachpiezometer are not in direct hydraulic connection.Piezometer 13B is set at a depth of 50 feet and 11C is setto a depth of 48.5 feet. The difference in depth (1.5 feet)may have resulted in sufficient offset to limit thehydraulic connection between the two piezometers.Additionally, blockage to water entry resulting from drillcuttings being injected into the cleat may have occurred in13B, resulting in decreased flow. Piezometer 13B also be-came dry in August of 1991 and did not regain water forthe duration of the study.

Piezometer 13A is screened into the Hazard No. 7 coal.The hydrograph shown on Figure 21 indicates a net de-crease in water level for the record period beforerebounding in December. The maximum water level of 1,185.9 was reached in April, and the minimum of 1, 184.2was observed in October and November, indicating awater-level range of only 1.7 feet. The Hazard No. 7 coalhas approximately 100 feet of overburden above it in thelocation of piezometer 13A. The overlying bedrock, inconjunction with decreasing fractures with depth,probably acts to restrict the rapid influx of recharge intothis coal seam.

Piezometer 21A is situated in the Hazard No. 5 coal.Figure 22 shows that the water level remains relativelyconstant for the first half of the year near an elevation of1,090 feet,

until a gradual decrease begins in June and continuesthroughout the summer and fall months. Comparison ofthe hydrograph for this piezometer with the hydrograph ofpiezometer 11B, which is also screened in the Hazard No.5 coal but in the ridge interior, shows that the water levelin 21A is lower by an average of close to 4 feet. Thewater-level decrease observed in both 11B and 21Aduring the fall months occurs much sooner in 21A. Themore rapid hydrograph response in 21A, along with thedecreasing hydraulic gradient from 11B toward 21A,suggests that the water in the Hazard No. 5 coal isdischarging along the side of the hill in the direction of 21A where the coal seam crops out. Water was observeddischarging from the Hazard No. 5 coal where the accessroad intersects the exposed coal seam during the fieldreconnaissance for study-site selection. Because theHazard No. 5 is the lowest significant coal seam (>1 footthick is considered a significant coal seam) in the geologicsection, it probably acts as a significant control to theamount of ground water that can migrate down into theridge interior by laterally diverting the downward flow ofground water from the mass of rocks and coal above ft.

Figure 23 shows the piezometers that monitor coalseams plotted as a function of increasing depth below thesurface. A strong relationship is observed in thewater-level fluctuation in each piezometer as a function ofdepth below the surface.

Discussion 29

well's recovery after purging and sampling. Examination ofthe individual water-level data reading indicates that thesteep drop in water level was usually recorded on the firstwater-level measurement, which was usually several daysafter sampling (a minimum of three or more well volumeswas purged from the piezometers before collecting samplesfor chemical analysis). Subsequent water-level readingsshow a gradual increase in water level until the nextsampling event. Piezometer 22B could often be purged drywhen sampling.

The net drop in water level during the months ofmaximum precipitation indicates that little or no directrecharge was entering the water-producing zone screenedby piezometer 22B. The drop in water level, along with thedrawdown-recovery cycle exhibited by this piezometer,suggest that the water is released from rocks with a lowstorage capacity. Fractures probably do not penetrate tothis depth (200 feet); therefore, there is no route for directrecharge from the surface. Unfortunately, thedownhole-camera examination was limited in this hole dueto turbid water conditions, and the extent of fractures thatintersect the hole was not qualified.

The screened interval for piezometer 11A penetrates tothe center of the ridge interior. The hydrogeologic zone ofinterest is interbedded sandstone and siltstones below theMagoffin Shale. The hydrograph for this piezometer isshown on Figure 25. A drop in water level of nearly 2.5feet occurred

Coals that are nearest the surface exhibit the greatestamount of water-level fluctuation. Fluctuation decreaseswith increasing thickness of overburden, and is most likelyattributable to the decrease in the movement of waterthrough fractures, and an increased granular flowcomponent with depth that would tend to buffer rapidground-water infiltration. The shift from fracture flow togranular flow probably is caused by the decrease in thefrequency and distribution of fractures, along with theinfilling of fractures by sediment or mineral precipitation.

Piezometers in the Below DrainagelRidge InteriorSix piezometers were installed to monitor ground water

in the ridge interior and valley bottom area where nonear-surface fractures or coal seams were observed. All ofthese piezometers, with the exception of 22B, were set atan elevation that is at or below the elevation of the nearestmajor drainage (Lick Branch). Each of these piezometerscontained measurable water throughout the period ofstudy.

The hydrograph for piezometer 22B (Fig. 24) shows aroughly cyclic rise and fall of its water level, which issuperimposed on a net decline that continues untilSeptember. The water level rebounds to above 990 feet inOctober and remains near this level for the remainder ofthe record period. The cyclicity exhibited by the graphmay be related to the


during February. This anomaly may be the result of waterinjected into the surrounding rock before construction ofthe piezometers had taken place, and subsequentlydischarging back into the hole. The water-level elevationmeasured in the open hole before piezometer 11A wasinstalled was approximately 1,049 feet, which is nearly150 feet higher than the observed water level aftercompletion. After piezometer installation anddevelopment, the water level dropped to a level thatseems more in line with subsequent measurements. Thewater level oscillated between 903 and 904 feetthroughout most of the spring and early summer, beforedropping steadily and reaching its lowest level inDecember. The oscillations in water level are all within arange of 1 foot and do not exhibit any obvious correlationwith specific precipitation events except for the gradualdecline in water level that began about August, andcontinued through the fall months.

The hydrograph for piezometer 22A is shown in Figure26. Piezometer 22A is screened in the same interval aspiezometer 11A, except that it is located downslope from11A. The hydrograph shows a similar pattern to thatexhibited by piezometer 11A. Piezometer 22A is secondin total depth at 290 feet. The initial rise, then fallobserved in 22A may also be a result of water stored inthe rock above the saturated zone before completion. Therange of water-level fluctuation during the study periodwas 6.6 feet.

The hydrograph for piezometer 31 A is shown in Figure 27.Piezometer 31A is also screened into the same interval aspiezometers 11A and 22A (see Fig. 1). The range ofwater-level fluctuation is 7.2 feet, which is largest of the fourpiezometers; set at this level. The hydrograph exhibits thesame water-level decline throughout the year, with theaddition of two significant drops in May and July. Thehydrograph shows a 4-foot water-level drop in May,followed by a rise in water level to within a foot of the levelbefore the drop occurred. The graph of the reboundresembles a pumping recovery curve, and this may representdrawdown induced by purging for water sampling on May 7,1991, which occurred 8 days before the large drop in waterlevel was recorded. A similar pattern was observed after themonthly sampling in July. However, this drop in water levelwas not observed in other months following monthlysampling, although approximately the same time interval hadpassed before the next water-level measurement wasrecorded. Another explanation for this hydrograph behaviorcan be related to the location and depth from the surface of31A compared to 11A and 22A. Piezometer 31A is situatednear the bottom of the ridge; thus, its depth (150 feet) fromthe surface is not as great as 11A and 22A (416 and 290 feet,respectively), which places 31A within the depth rangewhere near-surface fractures commonly occur. Fracturescould allow precipitation to recharge 31A more

Discussion 31

readily than the other piezometers screened at this level.Both of the large drops in water level in 31A occurred inthe spring and summer when less recharge was probablyentering the ground-water system, as evidenced by thegradual decline in hydrographs for all piezometers. Only0.3 inch of precipitation fell during the 10-day periodfollowing the sampling event that took place on May 8,indicating that the lack of precipitation probablycontributed to the rapid water-level declines, in addition toany effects induced by pumping.

Although the monthly precipitation total for May wasnearly 6 inches, the daily precipitation data show that mostof the rain fell as intense, short-term events (i.e.,thunderstorms). With high evapotranspiration and runoff,it is likely that only small amounts of precipitation fromintense events would move into the ground-water systemas recharge during the summer months unless theprecipitation is able to flow directly into highly conductivefractures that maintain a connection to the surface.

The hydrograph for piezometer 41B is shown in Figure28. Piezometer 41B also reveals an initial sharp drop inwater level similar to that observed in wells 11A, 22A, and31A. The consistent occurrence of this pattern in each ofthe four piezometers set at the level below the Magoff inShale suggests

that this response is probably real and not an artifact ofpre-construction conditions.

Following an initial drop to approximately 864 feet, thewater level in 41B remained consistent between 863 and864 feet, with the exception of a decline of nearly 1 foot onMay 15, 1991. This decline may correlate with the declineobserved in 31A at approximately the same time. Thegraph of monthly precipitation (see Fig. 11) shows that thetotal for April was approximately 3 inches less than theprecipitation total for the preceding and following months.Therefore, this decline in water level may represent adelayed response to precipitation shortfall.

Figure 29 shows the lateral placement of piezometers11A, 22A, 31A, and 41B with the contoured equi-potentiallines between the piezometers. The data indicate that thepiezometric surface between these piezometers slopes fromthe ridge interior to the southwest toward the direction ofwhere Lick Branch discharges into Balls Fork (see Fig. 1),indicating that water in the interior of the ridge flowstoward 41B. An average hydraulic gradient of 0.037 wascalculated along the inferred direction of flow, as indicatedby the arrow on the figure. The piezometric surface, asrepresented by the contours, indicates that the area ofhighest head lies beneath the


center of the ridge and that the potentiometric surface de-fined by the contour lines generally approximates thetopography.

The water levels for each of the four piezometers opento the interval below the Magoffin Shale are plotted onFigure 30. Piezometer 11A maintains its water level at thehighest elevation, followed by 22A, 31A, and 41B, whichis the order of encounter from the ridge center (11 A) tothe site adjacent to Lick Branch (41B). Each of thesepiezometers remained saturated throughout this study,indicating that saturated conditions exist deep within theridge interior.

In general, the water levels in piezometers 11A, 22A,and 31A indicate that an increase in the amount offluctuation occurs in the piezometers away from thecenter of the ridge (from 11 A toward 31 A). This mostlikely results from an increase in the influence of fracturesat the lower elevations because the thickness of bedrockoverlying the stratigraphic interval monitored by thesepiezometers decreases from the center of the ridge to thevalley bottom. The consistency observed in 41B probablyis related to its position in the discharge zone, where aconsistent ground-water flux is maintained from groundwater moving toward this area from higher elevations.

A strong correlation to specific precipitation events isnot observed in any of these piezometers, although agradual de-

cline in water level in each occurs for most of theobservation period, with a slight increase observed inJanuary. This trend probably reflects the lack of deeppenetrating recharge during the summer months. Thehigher water levels observed in all piezometers during thewinter months is probably related to cyclonic storms thatproduce soaking rains, which, in the absence ofevapotranspiration, provide recharge water that can reachthe deep bedrock zones below the fracture systems.

Piezometer 41A is screened to an interval at anelevation of 752.2 feet above MSL, which represents thelowest elevation monitored in this study. The hydrographfor 41A is shown on Figure 31. The borehole that containsboth piezometer 41A and 41B is adjacent to the majordrainage stream (Lick Branch). The water level in 41A(Fig. 2) rises to an elevation that ranges between 839.6and 843.4 feet, yielding a total fluctuation of 3.8 feet. Themaximum water-level elevation observed in 41A is belowthe level of the adjacent stream channel, which at itsclosest point lies west of the piezometer, approximately150 feet away at an elevation of 860.9 feet. The headlevels in 41B and 41A indicate that there is a verticalcomponent of ground-water flow between the two. Thehead level in 41B is higher than that of Lick Branch,indicating that ground water from the interval monitoredby 41B, and all intervals above this elevation, isdischarging into the stream.

Discussion 33

Field observations of Lick Branch during the study periodindicate that it is a perennial stream that appears to gainflow before discharging into Balls Fork. Lick Branchremained flowing throughout the observation period forthis study. Piezometer 41A apparently monitors a deeper,perhaps regional, flow system. Subsequent discussions onchemical characteristics of water samples derived frompiezometer 41A will support this conclusion.

Evaluation of Ground-WaterRecharge Using Tritium

Tritium is a radiogenic isotope of hydrogen with ahalf-life of 12.43 years (Ostlund, 1991). Tritium content iscommonly reported in tritium units (TU), which representone tritium atom in 1018 hydrogen atoms. The use oftritium in groundwater studies is a valuable tool fordetermining qualitative ages of ground water and in aidingin the delineation of ground-water flow paths and rechargerates (Davis, 1986; Hendry, 1988).

The amount of tritium present in the atmosphere has notbeen constant over time. It is estimated that tritium genera-tion from cosmic radiation reactions in the atmospherewould be responsible for average tritium values inprecipitation of approximately 10 TU (Davis, 1986).Atmospheric testing of thermonuclear devices in the early1950's created a large in-

flux of tritium into the atmosphere. As a result, tritiumconcentrations of 10,000 TU's were measured in thenorthern hemisphere during the 1960's (Hendry, 1988).This large spike of tritium into the atmosphere has allowedtritium to be used as a tracer in ground-water studies duringthe past 50 years.

In using tritium in ground-water studies, the effects ofunequal recharge rates and ground-water mixing must beacknowledged in order for the interpretations to be of highquality. The magnitude of these variables is not known forthis study; therefore, the following ages of ground waterbased on tritium concentrations are derived from thoselisted in Table 6, which only provides a qualified agedetermination based on the tritium content.

Figure 32 shows the site profile with contours of thetritium data. With the exception of piezometer 12B, all ofthe piezometers situated in the upper section of the ridge(Fig. 11) or at shallow depths near the slope containedtritium values above 10 TU. The piezometer with thehighest tritium content was 14A with a value of 23 TU andis interpreted as being recent recharge probably less than35 years old. This piezometer is situated in the shallowfracture zone near the top of the ridge, and exhibited alarge water-level fluctuation that correlated well withprecipitation trends, indicating that recharge rapidly entersthrough fractures. A one-time tritium sampling ofprecipitation from a thunderstorm in the summer


of 1991 near the site revealed that the tritium content forthat event was 6 TU. However, tritium data obtained fromthe U.S. Geological Survey indicate that in the late 1970'sthe tritium content in rain in eastern Kentucky was nearly60 TU. Most likely the water obtained from 14A is amixture of recent precipitation and relatively young groundwater stored in the rocks, which probably accumulatedwhen tritium content in precipitation was substantiallyhigher in the recent past.

Fractures could be observed in the steep, naturaloutcroppings of bedrock near the ridge top and in areasdisturbed during road construction. Fractures exposed atthe ground surface would allow for the easy entry ofrecharge water and probably result in the ground water inthis area having the highest tritium content. The highdegree of water-level fluctuation in the shallowpiezometers (e.g., 11C and 14A) in this area would supportthis conclusion. However, the fractures, in conjunctionwith the coal seams, could also allow for the rapiddischarge of ground water from the ridge top, which wouldact to "short circuit" the vertical migration of groundwaterrecharge into the ridge interior.

Evidence to support this interpretation is the depletionof tritium in the central area of the ridge. Figure 32 showsa zone of tritium-depleted (old) water beneath the core ofthe ridge, indicating very little recent penetration ofrecharge. For example, borehole 11 hosts threepiezometers that are set at

increasing depths into the center of the ridge. The shallowpiezometer, 11C, set at 48.5 feet below the surface, has atritium value of 13 TU. Piezometer 11B, set at 200 feet,measured 2.19 TU, followed by 11A with 0.32 TU at adepth of 418 feet. Piezometer 11B is set into the HazardNo. 5 coal. The contours of the tritium data show an abruptdecrease in the gradient of tritium content below theelevation of the Hazard No. 5 coal (elevation ofapproximately 1, 100 feet) in the ridge interior, which isreflected by the sharp break in the slope seen in the 4 TUcontour line directly below 11B. The two piezometers thatdid not accumulate water (1213 and 21 B) during the entirerecord period are situated above the Hazard No. 5 coalseam, as are the other piezometers that went dry at timesduring this investigation (13B and 14A). The drypiezometers, the occurrence of old water below the1,100-foot (No. 5 coal) elevation in the ridge interior, andthe occurrence of springs that crop out on the surface at ornear the elevation of the Hazard No. 5 and other coalseams indicate that the coal seams act to drain the groundwater that infiltrates into the bedrock above it. Springs andseeps were observed at several locations at the site, andnearly ail of them correspond to an elevation of one of themajor coal seams present in the ridge.

It was also noted that water that accumulated in corehole 1066 after the core was removed was at an elevationthat

corresponded to the elevation of the Hazard No. 5 coal. Itcould not be determined at that time if the water wasflowing into the hole from strata or fractures locatedabove the coal, or rising from below, but it appears thatthe coal imparted a control on the elevation of the water inthe hole. Each of the coal seams on the ridge top (theHazard Nos. 5, 6, 7, and 8) may contribute to draining theridge top, but the No. 5, being the lowest seam ofsignificant thickness (greater than 1 foot thick), ultimatelyconstrains the amount of ground water moving down intothe ridge interior.

Piezometer 21A is situated at a lower surface elevationthan 11B and also monitors the No. 5 coal. The coal seamis only 67 feet below the ground surface (see Figure 1 forpiezometer placements with respect to lithologic units) inthe location of 21A. The water obtained from 21Acontains 17 TU of tritium, indicating that much youngerwater is able to flow into the coal near the outslope,probably because of fractures that can intersect the coalseam near the outcrop. This observation is consistent withthe hydrographs for these two piezometers. Thehydrograph for 11B, located in the interior of the ridge,indicated very little water-level fluctuation throughout therecord period, whereas 21A exhibited some short-termfluctuations as well maintaining a water-level elevationthat averages 4 feet lower than that observed in 11B. Thishydraulic gradient within the coal indicates that water willflow

Discussion 35

along the No. 5 coal seam from the interior region andtoward 21A and probably discharge where the coal cropsout. The steady hydrograph of piezometer 11B suggests thatthe water present in the Hazard No. 5 coal may be retardedon Iithologies with low hydraulic conductivities that liebelow it. Water contained in the No. 5 coal is probably inequilibrium with the small but steady amount of rechargethat migrates from the leaky layers above.

The higher hydraulic conductivities for coal seams com-pared to the other lithologies present would suggest that themajority of water moving downward from the top of theridge would be diverted laterally along the coal seams. Nodata were found in the literature where vertical hydraulicconductivity values for coal-bearing strata in easternKentucky are given. However, a net vertical hydraulic valuecan be calculated from the tritium data that support a lateralflow mechanism.

The amount of tritium in piezometer 11A, which pene-trates the deepest into the ridge interior, is 0.32 TU, whichis conservatively interpreted as being at least 30 years old.Tritium values this low are commonly referred to as deadwater, and the water may be significantly older than this.Assuming recharge rates were relatively constant throughtime, this cor-


responds to an infiltration rate of 418 feet in 30 years, or1.3 X 10-5 cm/sec.

The vertical hydraulic conductivity value of 1.3 x 10-5

CM/ sec is an estimate of the average vertical hydraulicconductivity (K) of the bulk mass of rock and coal from theridge top to the deep interior. By means of a weightedaverage approach, a mean vertical K for the non-coal stratacan be estimated from the available data and by theacceptance of a set of assumptions concerning thepermeability of coal. The primary porosity of solid coal isquite low (Stach, 1982). The majority of permeability ofcoal is controlled by the cleat (Stone and Snoeberger, 1978;Banasczak, 1980; Shubert, 1980), which is a naturalfracture system that cuts across bedding planes. Because ofthe brittle nature of high-bituminous coal, well-definedcleat systems often form both laterally and verticallythroughout the seam. The coal seams examined in this studyare relatively thin compared to other stratigraphic intervalsin the section (generally less than 9 feet thick, includingpartings, with a mean of 5.25 feet); therefore, it isreasonable to assume that the cleat will penetrate the coalvertically as well as laterally. In this case, the verticalhydraulic conductivity can be assumed to be approximatelythe same as the horizontal. With this assumption accepted,an estimate of the vertical K of the non-coal sedimentary

where Kr = average vertical hydraulic conductivity of thenoncoal rock columnKa = estimated vertical K from the tritium data for lengthof the entire core holeKc = hydraulic conductivities of each coal seamcalculated from water-injection packer testsIc = thickness in feet of the interval tested to determinethe K of each coal seam (5 feet)n = number of intervals with coal seams present (n =10)T = total thickness (depth of core hole)Tc = sum of intervals containing coal seams (feet)

Solving for Kr yields an average vertical K for thenon-coal rock of 3.6 x 10-5cm/sec. The actual value isprobably lower, because the whole non-coal rock mass istreated as one interval in the calculation and because thewater is probably older than 30 years. This value isapproximately one order of magnitude less than theaverage value for the coal seams

rocks within the ridge can be estimated by the followingrelationship:

Discussion 37

(mean = 3.8 x 10-4 cm/sec). The difference in these esti-mates supports the conclusion that the higher hydraulicconductivity of the coals would cause coals to exert thegreater control in the transmission of ground water in thesection of the ridge where they are present. Theunsaturated or partially saturated conditions of the coalsnear the top of the ridge and their higher hydraulicconductivities would act to divert the downwardmovement of recharge water by promoting the horizontalmovement of water. The less conductive rock layersbeneath the coal seams would act to hinder the verticalmovement of water and cause ground water to dischargefrom the coals as springs or seeps at or near the outslope.

A downward deflection in the tritium contours in thearea between sites 2 and 4 indicates the penetration ofrelatively younger water to a greater depth than isobserved at the same elevation below the ridge top or thevalley bottom. This pattern probably is related to themovement of ground water through the near-surfacefracture zone. Also, there are no significant coal seamspresent at this lower level to divert the vertical migrationof ground water. Additionally, the break in the degree ofslope below an elevation of approximately 1,200 feet maycause a decrease in the velocity of flowing surface water,allowing for a higher percentage of surface runoff topercolate into the soil and bedrock to recharge theground-water system. This conclusion is supported bywater-

level data for the piezometers that monitor the intervalbelow the Magoff in Shale (11A, 22A, 31A, and 41B).These piezometers showed an increase in water-levelfluctuation downslope, which corresponds to the areawhere the tritium data indicate the penetration of youngground water. The increased fluctuation in water levels isattributed to the increased effects of fractures. This areaalong the hill slope where the penetration of recent groundwater is shown is interpreted as being the zone where thegreatest amount of recharge enters the ground-watersystem.

Conceptual Model of Ground-WaterFlow in the Ridge

Figure 33 presents a schematic view that summarizes therecharge and discharge relationships observed at the site,based on the previous discussion. The top of the ridge isshown as an area characterized as having a high surfacerunoff component because of the steepness of slope. Themajor coal seams of the Hazard group are concentrated inthis area. The coals and fractures act to divert a significantamount of the recharge that infiltrates into the bedrocklaterally to the outslope as seeps and springs. This watermay (1) remain on the surface and move to lowerelevations as runoff, (2) be transpired in theevapotranspirative process, or (3)


re-enter the ground-water system at or below the spring orseep.

The area between the location of site 2 and site 4 isshown as the area of maximum recharge. The break inslope and increased surface-area toward the base of theridge allows for a higher percentage of precipitation toenter the groundwater system. This is reflected by thedeeper penetration of relatively younger water, as indicatedby the tritium data. The area below site 4 down to thechannel of Lick Branch is shown as the discharge area.Water moving down-slope in the shallow fracture zonealong with ground water that is flowing from the interior ofthe ridge (see Fig. 31) would discharge in this area. Anarea of ground-water mixing would be expected in the areawhere the shallow, fracture-controlled flow and the groundwater from the ridge interior would converge. Thesignificance of this hypothesis will become more evident infollowing discussions on chemical considerations.

Several researchers have presented conceptual modelsof flow that are applicable to the Appalachian Coal Field(Wyrick and Borchers, 1981; Kipp and Dinger, 1987;Harlow and LeCain, 1991). Harlow and LeCain (1991)have presented the most recent model, which was based onthe previous investigations of others and a largeassemblage of hydraulic data from borehole testing. Figure34 shows their conceptualized flow system for the coalfield in Virginia. The model

presented in this study is in agreement with the locationand occurrence of the major features that controlground-water flow presented in their model.

The following discussion will address the observedgeochemical trends in this study as they relate to the flowsystem defined here.

Ground-Water Geochemistry Variation in Water Types

The use of Piper diagrams (Piper, 1944) is a common,practical means of graphically portraying watercomposition. To create a Piper plot, the concentrations ofmajor cations and anions from the chemical analysis ofwater are converted to millequivalents per liter (meq/L)and plotted as a percentage of the total cation or anionconcentration normalized to 100 percent on a triangularfield. Each apex of the triangle represents 100 percent ofan ion. Representative points from the cation and anionfields are then projected onto a diamond-shaped field. Thepoint of intersection of the two projections represents thesamples' water type. The following discussions will referto water types that are named according to the dominantcation and anion (based on being greater than 50 percentof the total cation or anion equivalents). If no ion makesup more than 50 percent, the water type will be defined byusing the two more abundant ions (in decreasing

Discussion3

order) that together represent more than 50 percent of thetotal.

The Piper diagrams show that several water types arerepresented at the site. However, similarities in watertypes are observed among piezometers that obtain theirwater from similar water-producing zones.

Figure 35 shows the 11 water samples taken from pie-zometer 12A plotted on a Piper diagram. All of thesamples plot as a Ca-HC03 water type. Water samplesfrom piezometer 14A are a Mg-S04 type and are shown onFigure 36. All of the samples from both of thesepiezometers plot in nearly the same location on theirrespective diagrams, indicating that very little variationoccurred within the percentages of major ions. Thecoefficient of variation (CV) for the total dissolved solidsfor piezometer 12A is 7.16 percent, indicating that theconcentrations of dissolved constituents remainedconstant throughout the sampling period. A coefficient ofvariation of 20.04 percent for 14A indicates considerablevariation in the concentration of dissolved constituentsthroughout the year and that it exhibited more variationthan 12A. Piezometer 14A is shallow (40 feet) comparedto 12A (167 feet), and the variation in piezometer 14A isprobably due to dilution from fresh recharge afterprecipitation events. This conclusion is consistent with thehydrograph and tritium data for 14A, which showed awater-level response that correlated well

with precipitation measurements and also that 14Acontained the highest tritium content (23 TU) of allpiezometers sampled. The stability exhibited by piezometer12A in both water quality and water level is probablyrelated to its depth (167 feet), where the immediate effectsof recharge events are minimized by the longer flow paththe ground water must travel.

Piezometer 31B had Na+ and HC03- as the dominant ions

in the beginning of the sampling period, but graduallyevolved to a Ca-Mg-HC03 water type (Fig. 37). Bicarbonateremained the major anion throughout the entire period. Thetotal concentration remained very consistent, as indicatedby the coefficient of variation (CV = 3.6 percent). A moredetailed discussion on the changes in water type in thispiezometer and one adjacent to it will be presented infollowing sections.

Three of the piezometers; monitoring the coal seams,11B, 13A, and 21A, are shown in Figures 38, 39, and 40,and reveal that each of these piezometers yield aCa-Mg-HC03 water, type. The coefficient of variation forthese three piezometers is 5.01, 5.4, and 6.0 percent,respectively, indicating that very little variation occurred inthe relative concentration of the total dissolvedconstituents. The Piper plot for piezometer 11C (Fig. 41)shows consistency in the total concentration of dissolvedsolids (CV = 10.01 percent), but exhibits a mi-


gration in the major cations from a calcium-magnesiumtype for most of the year to an Na-Ca-HC03 type inNovember of 1991. Piezometer 11C is set into the HazardNo. 8 coal seam, which is nearest to the surface at the topof the ridge. The fluctuating hydrograph in thispiezometer indicates that the influx of recharge followingprecipitation events may be responsible for the chemicalvariation. A tritium content of 13 TU indicates that it isslightly older than water derived from the fracture systemmonitored above it (e.g., 14A). Cation exchange or olderwater entering the coal from the surrounding bedrock mayalso account for the major ion variation during timeswhen recharge rates are low.

Figure 42 shows that piezometer 22B has an Na-S04water type. The cluster of points on the Piper diagramindicates that the water quality for this piezometerremained consistent throughout the sampling period, butthe coefficient of variation (CV = 23.2 percent) indicatessignificant variation in the amount of dissolved solids.This trend also may be explained by an addition of groundwater to the zone monitored by piezometer 22B by moremineralized recharge. Comparing the piezometerhydrograph with the dissolved solids data (see Appendix6) indicates that the dissolved solids content is the highestwhen the water levels are high, and low values correspondto low water levels.

Fractures were not observed at the depth of piezometer22B by inspection, but the tritium content in ground waterfrom this piezometer (11 TU) is more on the order oftritium content found in piezometers at much shallowerdepths (see Fig. 32). This piezometer contained water thattypically contained dissolved solids concentrations greaterthan 1,000 mg/L, which is significantly higher than all otherpiezometers that monitor zones that are above the majordrainage.

Three piezometers, 11A, 22A, and 31A, are Na-HC03type waters and are shown in Figures 43, 44, and 45,respectively. These three piezometers monitor the samezone (below the Magoffin Shale) at an elevation ofapproximately 860 feet. The Piper plots for 11A and 22Aare nearly identical. Both of these piezometers are situatednear the core of the ridge, with 11A located in the center ofthe ridge, and 22A located downslope and slightly offcenter. The coefficient of variation for these piezometers is4.11 and 15 percent, respectively, which indicates thatslightly more variation in the dissolved constituentsoccurred in piezometer 22A. Piezometer 31A shows sodiumto be the dominant cation, but 31A contained a higherpercentage of calcium and sulfate compared to both 11Aand 22A. The coefficient of variation in dissolved solids is14.4 percent. Piezometers; 22A and 31A monitor locationsthat are progressively farther from the cen-

Discussion 41

ter of the ridge monitored by 11A. A gradual increase incalcium and sulfate content, along with the increasedvariation in dissolved constituents moving from 11Atoward 31A probably indicates the influx of ground waterinto the hillslope, piezometers (22A and 31A) through thenear-surface fractures where calcium and sulfate are moreprevalent, as shown previously. The zone monitored bythese piezometers is covered by progressively less bedrockas one traverses from 11A to 31A, and fractures probablyhave an increasing effect. The tritium data support thisconclusion by indicating the deeper penetration of youngerwater (recharge) in the area between the locations ofpiezometers 22A and 31A (see Fig. 32).

Piezometer 41B is also screened near the sameelevation as 11A, 22A, and 31A. The Piper diagramshowing data f rom this piezometer is shown on Figure 46.The water from this piezometer is an Na-Ca-HC03 type.The clustering of sample points and the small coefficientof variation (1.93 percent) indicate that the major ionchemistry remained consistent throughout the samplingperiod. The additional calcium content compared to theother piezometers screened to this level suggest a mixingof water from below the Magoffin zone with

water from the near-surface zone. This piezometer is nearthe presumed discharge zone, and the stable nature of thewater quality probably results from a homogenization ofvarious water types that blend as they move toward thislocation.

The water sampled from piezometer 41A plots as an Na-Cl water type high in dissolved solids (Fig. 47). The tightclustering of water samples indicates that the percentage ofmajor ions has remained relatively consistent throughoutthe sampling period. The coefficient of variation (CV = 14percent) indicates some variation has occurred in thequantity of dissolved constituents. This piezometercontained brine, and typically contained dissolved solidsgreater than 10,000 mg/ L.

Overall, the temporal variation in water quality ofground water at the site was quite low, with the exceptionof the shallower piezometers screened in coal and fracturedzones.

Geochemical Reactions ControllingGround-Water Types

Of the various ground-water flow paths that exist in theridge examined in this study, two can be defined with some


certainty. The contours of the piezometric head (Figs. 13and 14) and the tritium data indicate that (1) recharge thatenters the ridge at the top migrates vertically downwardinto the ridge core, and that (2) water that accumulates inthe ridge interior moves outward toward the discharge areaat Lick Branch.

Piezometer 14A is screened at the shallowest depth ofall the piezometers located on the ridge top and containsthe youngest water. The major water type is an Mg-S04.

Because there is no common source of sulfate in therocks in eastern Kentucky (i.e., gypsum is not commonlyfound in the strata nor is its presence observed in the coretaken from the ridge), the most likely source of sulfate inshallow ground water is pyrite. The sulfate content inpiezometer 14A most likely results from pyrite oxidation(Appendix 1, equation 14). Coal seams and rocks commonto the area often contain pyrite (Danilchik and Waldrop,1978), and pyrite was observed in the geologic coreexamined at the site. Powell and Larson (1985) found thatin a similar setting in Virginia the presence of sulfate isgenerally a function of pyrite oxidation. The iron producedby this reaction (eqn. 11) often precipitates along fracturesand in the cleat surfaces in coals. Iron-hydroxide, identifiedby its common orange coloration, was observed bydownhole camera investigation entering the borehole insome places where fractures and coal seams

were encountered. Fractures observed in the geologic orealso showed iron stains and solid Fe-hydroxide. A total ironcontent of 114 mg/L was determined in piezometer 14A,with the dissolved content less than detectable limits (seeAppendix 6). The high total iron content, and the relativelyhigh Eh value (mean of 170 mV) suggests that the ironwould occur as Fe+3, which is consistent with the presenceof precipitated iron-hydroxide that was often observed whenpurging piezometer 14A.

The source of the magnesium in 14A may be from reac-tions with chlorite (eqn. 9) and/or the solution ofmagnesian-calcite. Dolomite does not appear to be a likelysource of magnesium because it is not a ' common mineralin the rocks of the Breathitt Formation (Dafiilchik andWaldrop, 1978; Weinheimer, 1982).

Piezometer 12A maintained a Ca-HC03 water type. Thismost likely reflects the solution of calcium carbonate (eqn.12) and, because of the long ground-water migration timedue to depth, possibly plagioclase (eqn. 8). The major ionchemistry remained relatively constant throughout the sam-pling period with the exception of sulfate. The drop insulfate content corresponds to a drop in Eh and an increaseof barium concentration that exceeded 1.0 mg/L in the last 4months of sampling. This behavior will be discussed inmore detail in a later section.

Discussion 43

The water type observed in piezometer 31B began asNa-HC03, with calcium gradually replacing sodium as themajor cation. The relative percentages of anions remainedconstant for the sampling period, which suggests thatcation exchange has occurred where sodium is exchangedfor calcium (eqn. 5). Calcium is probably derived from thedissolution of calcite. Alternatively, the disruption of thebedrock by drilling and

purging of water on a monthly basis may have inducedflow from fractures or saturated zones in hydraulicconnection with the open interval, allowing for addition ofground water with a higher calcium content. The lowvariance in dissolved solids (CV = 3.6 percent) wouldsupport a cation exchange explanation. However, theintrusion of water with a higher percentage of calciumcould also explain this variation in the


major cations if the intruding water containedapproximately the same concentration of dissolvedconstituents. No discernible trend in dissolved solids withprecipitation events was observed that could eithersupport or refute this hypothesis.

All four piezometers that were situated in coal seams(11C, 11B, 13A, and 21A) are predominantlyCa-Mg-HC03 water types. Sulfate content was typicallyhigh in these piezometers, averaging greater than 25 mg/Lin each. The pH values were less than 7 for eachpiezometer (see Appendix 6). The lowest pH recorded atthe site (pH = 5.37) was measured in piezometer 11C. Thewater types in these piezometers are probably a functionof the solution of calcite (eqn. 12) with the concomitantoxidation of pyrite (eqn. 14), as evidenced by the highsulfate content and the acidic pH values. Iron probablyforms Fe-hydroxides, as was evidenced by thedownhole-camera investigations. The total iron contentwas high (generally three orders of magnitude greaterthan the dissolved content) (Appendix 6) in each of thesepiezometers, which is consistent with pyrite oxidation.The occurrence of magnesium may be related to reactionwith chlorite (eqn. 9), solution of magnesian-calcite, orpossibly ion exchange.

Piezometer 22B is classified as an Na-S04 water type.The alkalinity, as HC03

-, is also relatively high, with a meanvalue of 279 mg/L. This piezometer contains the highest av-erage dissolved solids observed in all of the piezometers;with the exception of 41A, which encounters brine. Theoxidation of pyrite along with the dissolution of calcite(eqns. 12 and 14) could explain the high sulfate andbicarbonate values, but the difference between the total anddissolved iron content ( 2.46 and 0.06 mg/L, respectively) isnot consistent with the iron content expected from theoxidation of iron sulfides. Piezometer 22B monitors a zonethat includes the Magoffin Member, which is a marineshale characterized as containing some minor limestone andsiderite nodules (Danilchik and Waldrop, 1978). The pH ofwater derived from this piezometer ranged between 7 and 8throughout the sampling period. Significant pyrite oxidationwould be expected to lower the pH by the addition of H+

ions. The dissolution of siderite (eqn. 13) could provide foran increase in iron into the ground water without the largeincrease in acidity that would be expected from theoxidation of pyrite (Powell and Larson, 1985). However,siderite dissolution in the vicinity of this piezometer shouldalso increase the iron content significantly if it is the


source of the sulfate observed in 22B (mean = 447 mg/Q.Iron hydroxides were not observed in the purge waterduring sampling.

One explanation for this water type is that the locationin the subsurface where the majority of the reactions thatinfluence the water chemistry found in this piezometer aretaking place somewhere other than in the immediatevicinity of piezometer 22B; thus, the iron may precipitateat some location along the flow path (i.e., fractures)before it enters 22B. The high sodium content of the wateris most likely a result of cation exchange (eqn. 5). Theonly other likely source of sodium

that could produce a high dissolved sodium content consis-tent with the observed concentration in piezometer 22B(mean Na+ is 218.3 mg/L) is by contamination by brine. So-dium contribution by brine is highly unlikely because (1)the piezometer is situated at an elevation well abovedrainage where brines are usually encountered (Price andothers, 1962) and (2) chloride concentrations were low(average 3.97 mg/L CI-) and are not consistent with thechloride concentrations that would be expected if brinecontamination was occurring. Evidently the redoxconditions near this piezometer are not conducive to sulfatereduction, as evidenced

Discussion 47

by the high sulfate content, and H2S was not noted duringsampling.

Piezometers 11A, 22A, and 31A are Na-HC03 watertypes. These piezometers are set at elevations slightlybelow drainage in the interior area of the ridge. The waterin these piezometers had to migrate through a considerablerock mass, which undoubtedly influenced the chemistry.The most likely evolutionary sequence for the generationof the Na-HC03 water type encountered in thesepiezometers involves cation exchange, where divalentcations are removed from water that has reacted withpyrite, calcite, plagioclase, and perhaps siderite in theshallow, fractured zones or coals near the ridge surface. Inthis case, water that does not move laterally and exit theground-water system by way of the coal seams andshallow fractures would migrate vertically into the ridgecore. This ground water would be in contact with the hostrocks for a significant amount of time because of the lowvertical hydraulic conductivities of the rocks within theridge. This would allow time for chemical reactions suchas cation exchange to proceed without interruption.Additionally, the high feldspar content of the rocks of theBreathitt Formation

(Danilchik and Waldrop, 1978) suggests that the hydrolysisof silicate minerals (eqn. 8) is likely to take place. The prod-ucts of this reaction include kaolinite, along with dissolvedsilica, metal ions, and bicarbonate ions. Increasedbicarbonate concentrations tend to occur during thehydrolysis reactions involving silicates in ground waterscontaining little carbon dioxide (Hem, 1985), and thisreaction may aid in the creation of ground waters with a highpH. The pH of each of these three piezometers wasconsistently above 7.0, and above 8.0 in piezometers 11Aand 22A. The highest pH measured at the site occurred in22A, with a reading of 8.92.

Figure 48 shows the stability fields of gibbsite, kaolinite,montmorillonite, and albite as a function of pH, Na+, andH4SiO4 content. Sodium was chosen as a variable on thediagram because it is the dominant cation in the deeper pie-zometers. The incongruent dissolution of feldspar (K-spar)results in the formation of kaolinite (eqn. 8). Water samplesfrom each piezometer (with the exception of 41A, whichplots in the montmorillonite field) plot within the kaolinitestability field, indicating this is the most stable phase formedby the silicate weathering reactions. The stability ofkaolinite with


ground water at this site is consistent with the findings ofWeinheimer (1982), who reported the high occurrence ofkaolinite in rocks of the Breathitt Formation.

Cation exchange (eqn. 5) most likely removes calciumand magnesium from the ground water in the ridgeinterior. The most likely host for the exchange reactions iskaolinite, because of its abundance. A low cationexchange capacity (CEC) (10 to 100 mmol/Kg) is usuallydetermined for kaolinite, with the majority of exchangebeing pH dependant and concentrated on the edges of theclay particles (Bohn and others, 1985). Despite itsrelatively low CEC, kaolinite would probably havesufficient exchange capacity to accommodate the calciumand magnesium concentrations found in the relativelyfresh ground waters (TDS generally less than 500 mg/ L)encountered at this site. Other clay minerals, such as illite,probably also contribute to this effect. As evidence, nodivalent cations are observed in high concentrations inground water derived from the ridge interior. This suggeststhat ample exchange capacity exist for the adsorption ofdivalent cations, which are preferred over monovalentcations on exchange sites (Bohn and others, 1985).Another possible host

for cation exchange reactions may be the carboxylic acidgroups found in organic matter. These radicals would mostlikely occur in coal. However, the stratigraphic data showthat the majority of coal seams at this site occur in theupper region of the ridge, and no major coal seams occur inthe ridge interior where the greatest effect of cationexchange is observed. However, carbonaceous material iscommon to the coal-bearing rocks in eastern Kentucky;therefore, this mechanism cannot be discounted ascontributing to the net divalent cation reduction.

The chemical analyses for water samples taken from pie-zometers 11A and 22A show very low calciumconcentrations (generally less than 5 mg/L), indicating thatthe cation exchange reaction (eqn. 5) has proceeded far tothe right. Freeze and Cherry (1979) indicated that thecation exchange reactions will proceed to adsorb calciumas long as (1) exchange sites are available, (2) the activityproduct for calcium and carbonate is less than theequilibrium constant for calcite (unsaturated), and (3) aslong as calcite is available for solution. Calcite has beenidentified as a cement in Breathitt sandstones, and isolatedlimestones occur within the forma

Discussion 49

tion (Danilchik and Waldrop, 1978; Weinheimer, 1982). Ascation exchange removes calcium from solution, it allowsthe water to remain undersaturated with calcite; therefore,calcite continues to dissolve, which has the net effect ofincreasing the alkalinity by the addition of HC03- (eqn. 3).Saturation indices calculated for calcite and plotted for eachwell are shown on Figure 49. All piezometers, with theexception of 41A, are undersaturated with respect tocalcite; thus, the condition for undersaturation with calciteis met for all of the piezometers that monitor the ridgeinterior. Calcite dissolution will add to HC03

- produced bysilicate hydrolysis, with the net result being an increase inthe total alkalinity and pH.

The sulfate concentration is generally low in bothpiezometers 11A and 22A. This is most likely due to sulfatereduction mitigated by sulfur-reducing bacteria (eqn. 6).These bacteria exist under anaerobic conditions and requiresulfur compounds and organic compounds such ascarbohydrates, organic acids, and alcohols (Alexander,1977) for metabolic activity. A minimum of 1.0 mg/L oforganic matter is needed for sulfate-reducing bacteria toexist in the ground-water system (Perry and others, 1980).In this study, several piezometers

(22B, 22A, 11A, and 31A) contained total organic carbon inconcentrations above the detection limit of 1.0 mg/L (seeAppendix 6). The exact organic constituents present in thisorganic matter are not known, but it is likely that thenecessary organic compounds to sustain bacterial activity arepresent. Most subsurface waters contain sufficient quantitiesof the necessary dissolved organic matter to sustain bacterialgrowth (McNabb and Dunlap, 1974).

Piezometers 11A and 22A are screened to depths at whichlow concentrations of dissolved oxygen would be expectedto be present. The Eh values obtained for these piezometersshow a mean value of 173 mV for piezometer 11A and 120mV for piezometer 22A, which are slightly above the Ehlevels where significant sulfate reduction normally occurs(Champ and others, 1979). However, the redox potential val-ues used in this study were not measured by the determina-tion of a redox couple at equilibrium and, therefore, may notbe indicative of the actual redox conditions in theground-water system. As indicated before, the measuredredox values for ground water in this study are used in aqualitative sense for comparison purposes. A slight odor ofhydrogen sulfide


was observed at times when sampling piezometers 11Aand 22A, indicating that sulfate reduction had occurred. Afield test (HACH kit) was performed on several occasionsto obtain a semi-quantitative analysis of H2S content, butthese efforts were unsuccessful. Apparently, the H2Sconcentration was below the detectable limits of themethod (0. 1 mg/L), although its presence was detectableby smell. The pH range of between 7 and 9 observed inboth piezometers 11A and 22A is near the favorable pHrange (6.5 to 8.5) where sulfate reduction occurs (Connelland Patrick, 1968).

On an average, the water from piezometer 31A is anNa-HC03 water type. Piezometer 31A is located downgradient of 11A and 22B and is also directly beneath 31B.The Piper plots for both 31A and 31B show that thepercentage of major cations varies throughout thesampling period (see Figs. 37 and 44). The majority ofvariation in ion chemistry in piezometer 31B occurred inthe first 5 months of the sampling period. Sodium was themajor cation during this period, but gradually gave way tocalcium. The relative percentages of anions remainedconsistent except for sulfate, which exhibited a small riseat the end of the period.

Figure 50 shows the trends in the percent of major ionsin both piezometer 31A and 31B along with the monthlyprecipitation data. It can be seen that the percentages ofions from August through the end of the study remain quiteconstant in 31B, yielding a Ca-Mg-HC03 type water. Thewater chemistry of samples taken from this piezometerduring this period is most likely controlled by thedissolution of calcite (eqn. 3) and the oxidation of pyrite(eqn. 14). This Ca-Mg-HC03 water type observed inpiezometer 31B during the latter sampling times is moreconsistent with the water types observed in otherpiezometers that are set in fractures. It has been shown thatNa-HC03-type water has been associated with the pie-zometers located in the ridge interior, which are generally ata greater depth than 31B (e.g., 11A and 22A). Also, thewater derived from the ridge interior tends to beconsiderably older water (based on the tritium data).

Some of the possible explanations for the cationic varia-tion and subsequent decrease in sodium are as follows: (1)As mentioned previously, the air-rotary drilling of theborehole containing the piezometer may have temporarilyaltered the flow path of water entering the well by blockingfractures by

Discussion 51

injecting drill cuttings. Subsequently, well development ofthe piezometers may have reopened the pathways forground water to enter the horizon. (2) Na-HC03-type watermay have moved upward into the zone monitored by 31Bwhen the borehole was left open between drilling andpiezometer construction. (3) Piezometer 31B is screenedinto a fracture zone, and water (i.e., water containinghigher calcium and magnesium) may be introduced duringstorm events and periods of high recharge.

Looking at simultaneous trends in the major ion concen-trations in both piezometer 31B and 31A may offer insightinto the variation in these piezometers. Figure 50 showsthat the considerable drop in sodium and bicarbonateoccurring during June 1991 in piezometer 31A correlateswith a simultaneous rise in calcium, magnesium, andsulfate. The same trend is observed in June in 31B,although it is not as pronounced. During June the relativepercentages of the major ions in water from 31A is verysimilar to the distribution observed in 31B. The bar graph(Fig. 50) showing the monthly precipitation illustrates thatthe highest amount of precipitation in a 1 -month periodoccurred during June. The June pre-

cipitation may have created a pulse of recharge water thatmigrated downward from the surface and affected thewater chemistry in 31A by adding calcium, magnesium,and sulfate while diluting the amount of sodium andbicarbonate. Eastern Kentucky experienced severethunderstorms that created flash flooding during June of1991. The intensity of these storms caused washouts insections of the access roads at the site. The intrusion ofground water related to summer storm events may providerecharge in the location of these two piezometers, therebyproviding water characterized by mineral reactions (e.g.,calcite dissolution, pyrite oxidation) that formCa-Mg-HC03 water types. This scenario may explain theprominence of calcium as the major cation for theremainder of the study period in 31B but not 31A. Subse-quent storms during the remainder of the year may nothave been of sufficient intensity to create the large pulseof recharge necessary to penetrate to the depth ofpiezometer 31A and affect its water chemistry, but weresufficient to reach 31B. This hypothesis is supported bythe daily precipitation data. During the 16-day period fromMay 18 to June 3 (which preceded the June watersampling), four storm events


quivalents) but gradually dropped to less than 7 percent atthe end of the study. Concurrently, the percent of anionsrepresented by bicarbonate remained approximately thesame, while chloride increased slightly. The pH also showsa steady rise from 6.78 in March to 7.73 in December of1991. The water chemistry toward the end of the periodmore closely resembles the water derived from 11A and22A, which are upgradient of 31A but screened in the samestratigraphic interval. The most likely cause for the watertype found in 31A during this period is ground-watermixing by the addition of ground water from the shallowfracture zone with ground water moving down gradientfrom the ridge interior. As precipitation (thus, recharge)decreases at times throughout the year, the water chemistryin 31A becomes more characteristic of water from theinterior (upgradient) area of the ridge. The increase inchloride content probably indicates a slight contribution ofolder water from depth.

Piezometer 41B is down gradient of 31A. The water typeencountered in this piezometer is an Na-Ca-HC03. The per-centage of major ions has remained stable throughout theperiod of study. The chloride content (mean = 27 mg/L) isfive

occurred where more than 1 inch of precipitation fell (seeAppendix 5). Two of these events produced more than 1.5inches. The site did not experience this many intenseevents in this brief a time period during the remainingtime frame of the study

Rapid infiltration in the vicinity of 31A and 31B isconsistent with the earlier findings that these piezometersare situated in the area where maximum penetration ofrecharge enters the ridge (see Fig. 32). Additionally,although the water chemistry and the distribution of majorions found in 31A is quite similar to that found in 11Aand 22A, the tritium content is significantly higher (0.32TU for 11A, 4.64 TU for 22A, and 10 TU for 31A,),showing an increasing trend from the ridge interioroutward. Although tritium is not a conservative tracer, thetritium content in 31A is more on the order of the 13 TUfound in 31B, which also supports the hypothesis thatground water is migrating downward from the shallowfracture zone above 31A.

At the beginning of the sampling period the watertaken from piezometer 31A showed that calcium plusmagnesium represented approximatelv 40 percent of thecations (mille-

Discussion 53

times higher than that found in the piezometers at the samelevel located in the ridge interior (mean = 4.48 mg/Lchloride for 11A and 4.80 mg/L chloride for 22A) andnearly twice that of 31A (mean = 17.7 mg/L chloride),which is nearest to and upgradient of 41B. Figure 51shows the iso-concentration lines for chlorideconcentration along the profile of the site. The data fromJuly 1991 are used and were chosen for the followingreasons: (1) July was one of the months when themaximum number of piezometers (13 of 16) had water insufficient quantity to sample, (2) the charge-balance errorfor each of the July analyses was less than 5 percent, withthe mean error being 2.27 percent, and (3) the hightemporal consistency in the chemical quality (asexemplified by the Piper diagrams) suggests that data fromone sampling event could reasonably represent thehydrochemical conditions at the site. Subsequently, thedata from July 1991 will be used in all iso-concentrationplots as well as for geochemical modeling in this study.

A definite trend of increasing chloride content is shownin Figure 51, which shows iso-chloride contours along theridge profile near Lick Branch. The higher chloridecontent in 41B

is probably indicative of ground-water mixing with brinesthat are present below in 41A. The area of high chlorideconcentration shown on Figure 51 is based on the availabledata, and this area may actually extend beneath all of theLick Branch valley bottom.

The potentiometric heads shown by the piezometersthroughout the site indicate that the valley bottom is thedischarge zone for water entering the ridge. Thepotentiometric head measured in 41B and 41A indicate adownward gradient between the elevations monitored bythese two piezometers. The brine in 41 rises to within 80feet from ground level, which is approximately 20 feetbelow the water level of 41B. The head levels measured inthese adjacent piezometers only represent one componentof the flow field (it was shown previously thatground-water flow between the piezometers at the level istoward the southwest; see Fig. 30). The head present in41A indicated that the brine has the potential to migrate towithin 40 feet of the stream channel of Lick Branch, andperhaps even higher when the fresh water above itexperiences decreased head from lack of recharge duringextended dry periods. The high chemical gradient be


tween the fresh water found at the surface and the concen-trated brine at shallow depths below may result in the diffu-sion of the salt water into the fresh zone if the salt-waterzone is not confined.

Examination of the conceptual models of ground-waterflow presented by Harlow and LeCain (1991) and Wyrickand Borchers (1981) show that the intensity of subsurfacefractures, both vertical joint patterns and bedding planes,increase in the valley bottoms. In this case, the greaterintensity and interconnection of fracture planes may breachaquitards that normally separate brine and fresh water nearthe valley bottoms and allow for the vertical migration ofbrines below drainage areas. The lower occurrence offractures in the ridge interior would probably maintain thecontinuity of any laterally extensive unit, such that itsproperties that retard the vertical movement of water wouldremain intact.

The brine found in piezometer 41A contains highconcentrations of Na-Cl with a measured average electricalconductance of over 26,000 microsiemens (see Appendix 6for chemical data). Piezometers 41B and 41A are the onlytwo piezometers in this study that yielded water that wascharac-

teristically high in barium content. Piezometer 41B had amean barium value of 1.72 mg/L, and 41A contained 37.4mg/L. These findings are consistent with the findings ofWunsch (1988a), where barium in ground water exhibiteda strong correlation with chloride and was restricted tobedrock wells open to intervals below major drainage invalley bottoms. Additional discussion concerning bariumin ground water will be covered in the following sections.

Both piezometers 41A and 41B contain ground waterthat is depleted in sulfate, considering the abundant sulfursources contained in the coal-bearing rocks. Piezometer41A has an average sulfate content of 10.9 mg/L, and 41Bcontains 9.1 mg/L. Often a strong odor of hydrogensulfide gas was detected in the water from both 41A and41B when purging took place. Wunsch (1 988a-b)identified the common sulfur-reducing bacteriaDesultovibrio in well water in eastern Kentucky. Thepresence of hydrogen sulfide is common in the groundwater in eastern Kentucky (Price, 1962), indicating theconditions for sulfate reduction are widespread in thearea.

Discussion 55

Low Eh values are characteristic of ground water in dis-charge areas (Champ and others, 1978; Freeze and Cherry,1978; Gilkeson and others, 1981). The Eh measurementsfound in 41A and 41B were the lowest encounteredthroughout the study period. At times the Eh measurementswere found in the negative range, with the lowest readingbeing -49 mV in piezometer 41A during the December1991 sampling. The mean Eh value for piezometer 41B(mean = 52 mV) and for 4 A (mean = 15 mV) are withinthe range (<100 mV) where active sulfate reductionmitigated by sulfur-reducing bacteria is likely (Connell andPatrick, 1968; Champ and others, 1978).

The total iron concentrations found in both 41A and41B were low when compared to total iron in water fromother piezometers at the site. Consistently low redoxconditions in the discharge area and the near-neutral pHreading observed in both piezometers; would predict thatthe majority of iron in ground water in these piezometerswould occur as ferrous iron (Hem, 1985). Continuoussulfate reduction in this zone results in the production ofhydrogen sulfide that probably

reacts with the dissolved iron and precipitates as iron sul-fides, which keeps dissolved iron concentrations low.

Geochemical Controls onFluoride in Ground Water

Figure 52 shows the saturation indices with respect tofluorite for water samples from each piezometer at the site.All samples are undersaturated with fluorite, and noapparent trend is observed between the degree ofsaturation and well depth, location, or water-producingzone, suggesting that the occurrence of fluoride is notcontrolled by fluorite (eqn. 18). Generally, fluorite (CaF2)is thought to be a control on fluoride concentrations inground water that contain > 5 mg/L fluoride (Robertson,1991). This argument is strengthened by the poorcorrelation (r = .25) of fluoride with calcium (Table 7).

Determination of fluoride concentrations in watersamples from all piezometers revealed that fluoride rangedfrom 0.02 to 2.20 mg/L. Figure 53 shows theiso-concentration lines for fluoride concentration along theprofile of the site.


High fluoride concentrations are restricted to the ridgeinterior, where the water type is predominantly Na-HC03and has a high pH. A strong correlation of fluoride with pH(r = 0.91) at the 0.01 percent level (p < 0.0001) suggeststhat pH-dependant exchange or adsorption reactions areimportant in the control of fluoride concentrations. In theexchange reaction, the elevated pH of the ground waterindicates that a high concentration of hydroxyl ions wouldbe present, which would be available to exchange withfluoride (eqn. 17), resulting in increased fluorideconcentrations in the ground water. Clay minerals are thelikely host for the exchange reactions to occur (Bower andHatcher, 1967); however, Bohn and others (1985) indicatedstrong adsorption of fluoride by goethite (FeOOH).Additional sources of fluoride may be from weatheringreactions of micas and hornblende. The stability diagrambased in sodium, silica, and pH indicate kaolinite is thestable clay mineral in almost all water samples; however,Weinheimer (1983) indicated the abundance of bothkaolinite and goethite in rocks of the Breathitt Formation.Therefore, both of these minerals may act as hosts foranion exchange reactions. As indicated earlier, kaolinitewas also suspected

as being the host for cation exchange reactions involvingcalcium, magnesium, and barium. Perhaps the fluoride andhydroxyl ions act to balance any net positive charge createdwhen positively charged cations are adsorbed on the edgesof clay particles. This study has not generated the types ofdata, nor was it a goal to determine the exact surface reac-tions that can account for this geochernical phenomenon.More study is needed to determine these relationships.

The decrease in fluoride content in ground water as itmoves from the ridge interior to the discharge area canprobably be attributed to dilution from ground-water mixingor by fluoride re-adsorption to clay minerals as the pHdecreases. Decreased pH indicates a lower concentration ofhydroxyl ions, thus less competition for exchange sites.

Geochemical Controls onBarium in Ground Water

Barium content in the piezometers at the site rangedfrom 0.024 mg/L to 48.6 mg/L. Table 7 lists Pearsonproduct-moment correlation coefficients for severalelements and pa-

Discussion 57

rameters for the chemical data from all piezometers. Theoccurrence of barium exhibited strong positivecorrelations with chloride, sodium, and calcium. Thesefindings are consistent with the findings of Wunsch(1998a), who attributed the occurrence of barium inground water in eastern Kentucky to the mixing of freshground water with brines.

Figure 54 shows the saturation indices for bariteplotted for each piezometer in the study. The watersamples for each piezometer, with the exception of 11A,22A, and 21A, are saturated with respect to barite. Thisshows piezometers with barium values less than 0.5 mg/Las well as those above 1 mg/L are saturated with respect tobarite. This finding is not unexpected because of therelatively high sulfate content common to ground water inPennsylvanian coal-bearing rocks. Sulfate can impose asolubility constraint on barium, especially under oxidizingconditions (Hem, 1985). Piezometer 21A shows asaturation index (SI) of -0.058, which is near equilibriumwith barite. Of more significance to the occurrence ofbarium in ground water is the trend in barite stabilityalong the flow path through piezometers 11A, 22A, 31A,and 41B. Piezometer 11A is highly undersaturated withbarite.

Piezometer 22A is also undersaturated, but to a lesser de-gree. Moving down gradient, one observes that both 31Aand 41B are supersaturated with barite (see Fig. 54).

Figure 55 shows a plot of the log value of the activityof select ions shown to exhibit strong correlations (Ca,Na, Cl; Table 7) with the occurrence of barium along withparameters involved with the solubility of barite (S04 andEh). Eh is represented as pe for the convenience of scaleon the diagram and is related to pe by the relationship Eh(volts) = pe/0.0592 (Hem, 1985). It can be seen in Figure55 that the activity of sodium and sulfate are fairlyconsistent along the flow path. The activity of chlorideshows a slight increase, and calcium and barium activitiesincrease significantly from the ridge interior area (11A) tothe discharge zone (41B). The pe remains consistentbefore exhibiting a dramatic drop in the discharge area(41B).

Wunsch (1988a) suggested that the sources of bariumcould be from its inclusion in brines that are encounteredat shallow depth, or from cation exchange. Cationexchange appears to be an important mechanism in theoccurrence of barium based on the data shown here, andis probably more


important than the solubility constraints imposed by barite,especially at low barium concentrations (<2.0 mg/L).

The water derived from piezometer 11A shows that thesulfate concentrations are low, with a mean sulfate valueof 8.34 mg/L and a mean barium concentration of 0.02mg/L. The sulfate concentration in 11A is comparable tothe sulfate found in 41B (mean = 9.1 mg/L), whichconsistently contains barium that exceeds 1.0 mg/L.Because the activity of sulfate is low in both 11A and 41B,and assuming the mineralogy along the stratigraphicinterval monitored by these two piezometers is consistent,it would be expected that barium concentrations in 11Awould be similar to those found in 41B if barite solubilitywas the only control on barium concentration. The groundwater derived from the ridge interior is noticeably devoidof divalent cations, which suggests that cation exchangereactions are removing barium, calcium, and magnesiumfrom solution. The lytropic series for cations indicates thatdivalent cations are preferentially adsorbed by clay min-erals over monovalent cations, with barium being the mostpreferred of these divalent ions (Bohn and others, 1985).

The chemical data show that as ground water movesfrom the ridge interior toward the discharge zone, anincrease in calcium and magnesium in ground water isobserved. As the concentration of these divalent cationsincreases, the competition for exchange sites increases,which may cause barium to be released into solution as thewater chemistry and the exchange sites move towardequilibrium with the cations contained in solution. Thisscenario can explain the observed trends in the ionactivities shown in Figure 55, which shows a goodcorrelation with the increase in the activity of barium insolution as the activity of calcium increases.

The data from the barium exchange experimentsindicate that calcium as well as magnesium was releasedfrom exchange sites with increased sodium solutionconcentration. As indicated earlier, the rock samples usedfor the lab experiments were taken from the samestratigraphic interval at different site locations (sites 11Aand 41B). The total calcium extracted by 100 mmolNa-EDTA shows that the average total calcium releasedfrom rock samples from borehole 41B contains nearlythree times the calcium compared to that de

Discussion 59

termined from the samples from 11A (mean = 90.6 mg/Lcalcium for 41B; 34.3 mg/L for 11A). Explanations forthis discrepancy are (1) the clay mineralogy varies laterallyfrom site to site, or (2) more of the exchange sites for the41B samples have been loaded with divalent cations(principally calcium) by ground water containing highercalcium concentrations, which would be consistent withthe hypothesis presented here and the observed trendshown in Figure 55. The increase in calcium from groundwater moving along the flow path from the ridge areatoward 41B would provide the "excess" calcium and isconsistent with this analysis. This relationship suggeststhat barium is released from exchange sites by competitionfrom calcium, and perhaps magnesium, and is somewhatindependent of sodium concentration if the assumption ismade that the total CEC for the rock samples from bothsites is approximately the same. The mineral that hosts theexchange reactions within the ridge interior would have ahigher percent of exchange sites containing monovalentcations, which would be available for exchange for dival-ent cations. This can explain why low concentrations ofdivalent cations are found in ground water derived fromthe ridge

interior. If the above scenario is accurate, ultimately thecontinuous loading of divalent cations on the claymineral exchange sites in the ridge interior may reach thepoint when barium and other divalent cationconcentration could increase in the ground water whenthe CEC is exceeded because of the continuousmovement of calcium-loaded water into the area.Exchange experiments with a calcium-chloride solutionmay provide additional insight into this mechanism.Unfortunately, most of the rock samples were consumedin the original experiments.

Additional evidence for this control on the occurrenceof barium in ground water is the water chemistry data forpiezometer 12A. Barium concentrations for the first fewmonths of sampling were below 1.0 mg/L. However,barium concentration increased during the study periodand exceeded 1.0 mg/L in October of 1991. This trendcontinued, with a maximum barium concentration of 1.29mg/L observed in December of 1991. Figure 56 show thetrends in several constituents related to the controls onbarium concentration found in piezometer 12A. It can beseen that the pe level drops signifi-


cantly during the period, which corresponds to a drop insulfate content. Sulfate content decreases approximatelyfivefold from a high of 51.0 mg/L at the beginning ofsampling to a low of 10.9 mg/L approximately a yearlater. The decrease in sulfate can be attributed to sulfatereduction by sulfur-reducing bacteria. The presence ofhydrogen sulfide was noted in the purge water from 12Aas the sampling schedule progressed. Also, thebicarbonate concentration increased throughout theperiod from 428 to 581 mg/L, which may be attributed tosulfate reduction (eqn. 6).

Piezometer 12A is the deepest piezometer thatmonitors a fracture zone and is at an elevation that is wellabove drainage. Although sulfur-reducing bacteria maynot have been present initially (i.e., no H2S odor detected,high sulfate), they may have been introduced bycontaminated sampling equipment. Althoughsulfur-reducing bacteria from wells near valley bottomswere found to be hallophyllic (Wunsch, 1988b), thesebacteria are known to adapt to variations in salt contentvery rapidly (Alexander, 1977), and could prob-

ably establish themselves in the above-drainageenvironment.

The divalent cation exchange mechanism presentedhere may also be the cause for the enrichment ofbarium in eastern Kentucky brines (Heck, 1940).Brines found in the Appalachian Basin usually containappreciable calcium (McGrain, 1953), which couldexchange for barium. In addition, high chlorideconcentrations have an increased solubility effect onbarite (Collins, 1975). The long residence time that isusually assigned to brines found in cratonic layeredsedimentary rocks (Freeze and Cherry, 1979) wouldprovide for the brine solutions to reach equilibriumwith the exchangeable sites on minerals in the hostrocks. The low redox conditions characteristic of oldconnate waters (i.e., brines) is also conducive to theactivity of sulfur-reducing bacteria, which keep sulfateconcentrations low (Heck, 1940) and allow forelevated barium concentrations to exist.

Discussion 61

Reaction-Path ModelingUsing Geochemical Models

Geochemical modeling can provide supporting evidencefor the validation of the proposed reactions used to explainthe chemical evolution of ground water at the site. In thisstudy the model BALANCE was used to determine thestate of solid phases (either dissolving or precipitating) inorder to satisfy the mass transfer of dissolved constituentsfrom one end member solution to another. Calibration ofthe model can be made by comparing the predicted state ofeach mineral phase used in the mass transfer model withthe saturation index calculated by the equilibrium model(PHREEQE). This method does not allow for a uniquesolution to each reaction problem (Parkhurst and others,1982), but it does demonstrate supporting arguments forconceptualization of groundwater evolution. Severalresearchers using this modeling approach alsoincorporated isotopic analyses of constituents into theirmodels, which allowed them to make semi-quantitativejudgements on which a particular mineral was the majorcontributor to the occurrence of a particular ion in solution(Powell and Larson, 1985; Wood and Low, 1988;Robertson, 1991). For example, carbon-13 determinationscan be useful in determining if the source of carbon inground water is derived from organic carbon or thedissolution of carbonates with a specific isotopic signature(Powell and Larson, 1985).

The data included in this study are limited in this respect;therefore, only supporting evidence for the plausiblemineral-water reactions discussed previously can be given.

It has been proposed that ground water entering theridge as recharge from precipitation events can travel byseveral paths before discharging in the area near LickBranch in the valley bottom. Only one flow path, however,can be predicted with a degree of certainty that precludesany likely effects from ground-water mixing. This flowpath is designated as precipitation entering theground-water system at the ridge top and migratingvertically into the center or interior area of the ridge. Afterground water accumulates in this area, the head gradientindicates that the ground water will move laterally towardthe discharge area. Mixing is most likely to occur in theseareas where recharge in the shallow fracture zone and coalseams migrates downward and mixes with ground waterflowing from the ridge interior toward the valley bottom ordischarges on the surface as springs and seeps.

Geochemical modeling is used to test the validity of thesets of plausible water/mineral reactions proposed in thisstudy as controls on the ground-water geochemistry. Theflow path chosen to model is one where ground water ismoving vertically from the ridge top to the interior down apath parallel to borehole 11, which contains piezometers11C, 11B, and 11A.

Discussion 63

These piezometers are chosen for the followingreasons: (1) Piezometer 11C is directly above 11A.Assuming a vertical f low path, the water at the level of11C would be expected to migrate to the level of 11A. (2)Piezometer 11C is screened to a shallow depth (50 feet) atthe ridge top. With few exceptions, the piezometers set inthe coal seams and those in the shallow fracture zoneswere similar in their major dissolved ion chemistry, as11A is to the other piezometers screened into the ridgeinterior. The reaction paths modeled here are shown onFigure 57.

The first step of the model is to simulate the reactionsencountered by precipitation as it travels through theshallow soil zone, moves through the bedrock byway offractures and coal, and arrives at the position ofpiezometer 11C. The second step is the simulation ofground-water/mineral reactions as ground water migratesdownward from 11C to the position of piezometer 11A(refer to Fig. 57).

The input solution is rain water, using chemicalanalyses from samples collected at the University ofKentucky Robinson Forest research station (Coltharp andBrooks, 1991). The mean values for constituentscontained in precipitation that fell in July 1991 are used inthe simulation. Chemical data for ground water are fromwater samples collected during July 1991.

In the first simulation, BALANCE is used to calculate themass transfer between the incident precipitation reactingwith minerals and the final water in this step, which is thewater found in piezometer 11C. The assemblage of mineralsand gases used were those that seem to be the most plausiblebased on the mineralogy contained in the rocks (Weinheimer,1982). Calcite is assumed to be the main source of calciumand bicarbonate. Magnesium is derived from chlorite. Ironand sulfate can form from the dissolution of siderite and theoxidation of pyrite. Potassium may be derived from potas-sium feldspar (K-spar). Ion exchange is used to supply so-dium by exchanging for calcium. Carbon dioxide gas is in-cluded because it is assumed that the system is enriched inCO2 due to its accumulation in the soil zone from the decayof organic material (Freeze and Cherry, 1979). The data usedin the models, as well as a summary of the computer output,are contained in Appendix 7.

Simulation 1 mass-balance calculations show that calcite,chlorite, pyrite, and CO2 gas dissolve, and the ion exchangereaction where calcium is removed and sodium is released isproceeding to the right (each denoted by a positive value),but siderite is precipitating (indicated by a negative value).The reader should note that the sign convention used byBALANCE is opposite that which suggests precipitation (su-persaturation) by PHREEOE. This output is inconsistent with


the mineral saturation data calculated using PHREEQE,which indicates that the water in 11 C is undersaturatedwith siderite. Because of the oxidizing conditions ofground water near the surface, and the relatively high Ehvalues recorded in 11C (mean = 190 mV), iron is mostlikely in the ferrous state, which would result in theformation of an iron hydroxide mineral. Therefore,goethite (FeOOH) is substituted in place of siderite insimulation 2. These results are consistent with the mineralsaturation data, which show that the reactions betweenground water and mineral phases listed above areplausible in the evolution of water as it enters the groundas precipitation and migrates to the level of piezometer11C. The positive delta values for the dissolvedconstituents used in the simulation indicate an increase inthe concentration of each ion species listed.

The second modeling step consisted of calculating themass balance resulting from the migration of the waterfrom step I (11C) and the final water, which is the waterfound in 11A. The phases considered in this step arecalcite, goethite, ion exchange (controlling calcium,barium, magnesium, and fluoride), siderite, goethite, ironsulfide, pyrite, hydrogen sulfide, and kaolinite. In order tomaintain the input values within the operationalconstraints of BALANCE (number of input species =number of output phases), some phases must be droppedfrom consideration or substitutions made. Carbon

dioxide gas is not included in this step because it isassumed that (1) its concentration will decrease as thewater moves from the soil zone, (2) the deeperground-water zone may be a closed system to CO2, and (3)CO2 is consumed in reactions along the flow path.

Simulation 3 shows that calcite and siderite dissolve,whereas goethite, pyrite, and kaolinite precipitate(Appendix 7). This simulation is inconsistent with thesaturation indices data, which indicate that 11A is vastlyundersaturated with pyrite. Simulation 4 substitutes ironsulfide (FeS) for pyrite. The assumption here is that ironmay combine with hydrogen sulfide resulting from sulfatereduction. The results for simulation 4 are in agreementwith the saturation indices calculated for each mineral,with the exception of the removal of iron (as indicated bythe precipitation of FeS as shown by BALANCE). ThePHREEQE data show that 11A is undersaturated withFeS. Two explanations may account for the lack of massbalance for iron. By one account, the dissolution offerroan calcite would increase the iron concentration in11A without requiring the oxidation of an iron-sulfidemineral. Because of the lower Eh, and the assumed lack ofoxygen in the deep interior area of the ridge, the oxidationof pyrite may be limited. Ferroan calcite is the dominantcementing agent and is common in Breathitt rocks(Weinheimer, 1982). The saturation index for calcitepredicts that calcite (and probably fer-

Discussion 65

Table 7.-Pearson Correlation Coefficients Calculated for Selected Constituents from all Piezometers Using July1991 Data. Significance Level Indicates the Probability of a Correlation Relationship Occurring by Chance isLess than the Value Given.

Pearson Correlation Coefficients

roan calcite) will dissolve in 11A, which wouldliberate iron and provide for the mass balance. Anotherpossibility is simply analytical error. The difference iniron concentration between the initial (11C) and finalsolutions (11A) is only .007 millimole (0.41 mg/L Fe).

The positive delta phases for calcite and sideriteindicate these phases dissolve. Negative delta valuesfor goethite and kaolinite indicate these phases wouldprecipitate. These predictions are consistent withpetrographic analysis that shows authigenic kaoliniteand iron-hydroxide coatings com-


mon in Breathitt Formation rocks (Wienheimer, 1982).Positive delta values for ion exchange for the divalentcations reveal that these cations are removed from solutionand sodium is released. A negative value for the ionexchange (reaction proceeds opposite to the way presentedin the model simulation) for fluoride is consistent withfluoride being released into solution. Examination of thedelta phases for the dissolved species from simulation 4shows that all of the species except sodium, carbon(bicarbonate), and fluoride have negative delta values.This indicates that as the water moves along the flow pathfrom piezometer 11C to 11A, the sodium, bicarbonate, andfluoride increase in concentration, resulting in thegeneration of an Na-HC03-type water with a high fluoridecontent. The findings of this modeling step are consistentwith the trends in water chemistry observed in piezometersthat monitor the ridge interior.

There are probably other mineral phase combinationsthat could produce results that are consistent with thetrends found along the flow path modeled here. Still, theseresults indicate that the controlling geochernical reactionspresented in this study represent a plausible explanationfor the geochemical evolution of ground water as it flowsthrough coal-bearing rocks in this hydrogeologic setting.

A binary mixing model using the water chemistry frompiezometers 11A and 31B was used to model the mixingthat would occur as water moves from the ridge interiortoward the valley bottom, where it would mix withground water stored in the shallow fracture zone. Themodeling results (Appendix 7) show that a mixing ratioof 71.15 percent of water from 11A and 28.85 percent ofwater from 31B would create a mixture that most closelyresembles the relative percent of major ions found in thewater from 31A. The largest error (-77.15 percent) in thecalculation occurred in the chloride content (seeAppendix 7). It should be noted that chlorideconcentrations in these samples are very low (less than 5mg/L), so a small variation in the concentrations resultsin a large percentage of error. Neither of the two endmembers contained sufficient chloride to account for thechloride concentration found in 31A. The higher chloridein this piezometer indicates additional mixing with otherground water containing a higher chloride content. Theincrease in chloride occurred as monthly samplingproceeded throughout the year, indicating that pumpingof the piezometer for purging and sampling purposes mayhave induced the migration of water containing higherchloride. Brines are frequently found within 200 feet ofthe surface in eastern Kentucky (Sprinkle and oth-

Discussion 67

ers, 1983). Brine is encountered in the study area inpiezometer 41A, which is located approximately 550 feetsouthwest and 100 feet below 31A. The mixing zonebetween the brine and fresh water is not well defined ineastern Kentucky (Sprinkle and others, 1983) or at thissite; therefore, pumping may draw in water containingsufficient chloride concentrations if the radius ofinfluence induced by pumping causes migration of higherchloride water from the salt-water interface zone.

Therefore, hydrochemical trends observed inpiezometer 31A can be explained by the location of thispiezometer in the recharge zone where ground-watermixing or the intrusion of ground water with a differentchemical signature takes place because of changes in thelocal flow field, which result from variation in the influxof recharge from the surface and/or changes imparted bywell development and pumping.

Conceptual Model for Hydrochemical FaciesHydrochemical facies are distinct zones that have

cation and anion concentrations describable withindefined composition categories (Freeze and Cherry,1979). The data presented in this study show that thepiezometers that are screened into discretewater-producing zones exhibit consistent, discerniblewater types that can be related to the flow system, theposition of the piezometer within the ridge, and to

the elevation with respect to the adjacent major drainagestream. These observations, in conjunction with hydraulic,isotopic, and chemical data, allow for the followingconceptual model to be presented, which shows areaswithin the ridge where the hydrochemical facies may bepredicted with a reasonable amount of confidence.

Description of Hydrochemical Model FieldsFigure 58 shows the outline of the ridge with shaded

fields that represent a particular hydrochemical facies. Eachfacies zone represents an area where a characteristic set ofwater types is observed and a plausible explanation can bemade for its occurrence. As presented here, these faciesfields represent areas where the probability is high that acombination of the cations or anions listed for that field willdefine the water type for a sample derived from a locationwithin the field.

The lines separating the zones are not to be interpretedas definite or absolute boundaries. Most likely theboundaries are somewhat gradational, but may be locallyabrupt where the boundary between fields coincides with alow or high permeability zone (e.g., the underclay beneath acoal or a highly fractured area). Coal seams have beenshown to impart a pronounced effect on the occurrence andmovement of ground water in this study area. The positionof coal seams within the stratigraphic sequence of a ridgemay act to expand or compress facies zones.

Figure 54. Saturation indices for barite for all piezometers. Chemical data are from July 1991.


The Ca+Mg, HC03+SO4 facies; zone represents thearea adjacent to the near-surface area of the ridge, andcorresponds closely to the area where a fracture zone ismost likely to occur (Wyrick and Borchers, 1983; Kippand Dinger, 1987). In this zone, the oxidation of pyriteand the dissolution of calcite in an oxidizing environmentimpart the greatest influence on the chemical character ofground water. The Na, HC03+S04 zone represents the areawhere the dissolution of calcite, cation exchange ofsodium for divalent cations, and possibly sulfatereduction control the occurrence of the major ions. TheNa, Cl zone represents old, connate water that mayconsist of very concentrated brines (McGrain, 1953). TheNa+Ca, HC03+SO4+Cl area represents a zone consistingof water formed when ground water from the Na,HC03+SO4 zone moved from the ridge interior and mixedwith water from the Ca+Mg, HC03+SO4 shallow zone aswater from both zones flowed toward the discharge area.Additionally, an increase in the chloride componentresults when additional mixing occurs with Na, Cl waternear the valley bottom. Therefore, any water derived fromthis zone will be a water type defined by varyingpercentages of Na, Ca, HC03, S04, and Cl. A decrease insulfate content due to sulfate reduction may be observedas the water approaches the discharge area in the valleybottom.

The Na, Cl zone is shown to be at its most shallow pointbeneath the valley bottom. This interpretation is based onthe chloride data, and is consistent with the observations ofothers (Price, 1962 ; Sprinkle and others, 1983). Conceptualflow models for similar settings to this study site have beenpresented by Wyrick and Borchers (1981), Kipp and Dinger(1987), and Harlow and LeCain (1991), and each suggestedthat a high occurrence of fractures exists in the valley bot-toms. These fractures may provide pathways for the migra-tion of salt water to shallow depths.

Probably the most probing question not answered by thisstudy is the exact location of the salt-water interface belowthe interior of the ridge. Funding and technologicallimitations precluded the drilling of a piezometer deep intothe ridge interior that may have resolved this question. InFigure 58, the boundary of the Na, Cl zone is shown asbeing depressed under the ridge interior, with a questionmark denoting this uncertainty. This interpretation is basedon the chloride data, which shows high chloride values tendto be found in water samples taken from piezometers at ornear the valley bottom. Very low chloride content is foundin water samples derived from beneath the ridge at the samestratigraphic level. Also, the potentiometric headmeasurements from within the ridge

Discussion 69

indicate that higher pressure heads are measured in theridge at the same elevation (stratigraphic interval) as pie-zometers near the valley bottom. The higher head impartedby the fresh ground water in this area should act to depressor displace the supposed underlying saline water. Osmoticeffects may also contribute to the pressure differential inthis area (Freeze and Cherry, 1979). Most likely there mayexist a layer or sequence of geologic stratum that may actas an aquitard to the downward movement of fresh water,or the upward movement of saline water in this area of theridge. In this case, the saline interface may not bedepressed significantly and may be encountered ratherabruptly. However, if this were the case, it would beexpected that over geologic time chemical diffusion wouldresult in the occurrence of chloride content greater than the5 mg/L or less typically found in the piezometers located inthe interior area of the ridge near the assumed salt-waterinterface. If the actual relationship between the salineinterface under the ridge presented here is accurate, itwould suggest that the saline interface throughout easternKentucky would occur as an undulatory surface where thedepressions in the interface surface would be an inverseapproximation of the surface topography or potentiometricsurface.

The pressure imparted upon the saline water "aquifer" bythe fresh water accumulating beneath ridges would be trans-ferred in the salt-water aquifer, causing the migration of saltwater away from the high-pressure area located in the ridgeinteriors toward areas of lower pressure (valley bottoms).These mechanisms may provide the hydraulic pressure thatdrives the brines upward (relative to other areas of the brineflow system), where breaches in the rock layers that confinethe salt water (presumably the highly fractured valley bot-toms) exist. Further study is needed to verify these relation-ships.

The conceptual model described here was created usingthe site-specific conditions detailed in this study. The hy-drochernical facies relationships represent a ridge in theEastern Kentucky Coal Field that has not been affected byman's activities to any significant extent. In addition, thissite is bordered by third- and fourth-order streams, and it iswidely held that a shallow salt-water interface exists belowdrainages with these ranks. The salt-water interface has notbeen well defined in areas below streams of a lower order orintermittent streams; therefore, this model, especially inreference to the occurrence of, and the geochernical effectsimparted


by, the salt-water zone may not apply to other areas ineastern Kentucky. Also, the effects of geology, such as thestratigraphic location of highly conductive coal seams andthe extent of subsurface fractures in a particular area, mustbe accounted for when using this model. Their presence orabsence in a stratigraphic sequence may act to expand orcompress the geochemical facies zones shown in thismodel. However, it is expected that this model will beapplicable to many areas in the maturely dissectedAppalachian Plateau that allow for these considerations.

CONCLUSIONSEight boreholes and one geologic core were drilled at

various locations at the study site. Examination anddescription of the core, along with packer-injection testsand downhole camera observations of the boreholesindicate that the majority of ground water moving atshallow depths (generally less than 200 feet) is occurringin fracture systems and coal seams. The deepestwater-producing fracture observed in this study was at 167feet. These observations are consistent with the conceptualmodels that describe ground-water flow in the region. Coalseams were the lithologic units with the highest hydraulicconductivities and probably act to dewater rocks that lieabove that are in hydraulic connection.

Fourteen of the total of sixteen piezometers that moni-tored the study site contained measurable water throughoutthe course of this study. Piezometers that remained dry(13B and 21A) were restricted to the upper area of theridge, indicating that the rocks in the ridge-top area are notcontinuously saturated. Piezometers screened into shallowfractures and coals near ground surface exhibited the mostrapid response to recharge events or periods of lowprecipitation. The amount of water-level fluctuation, whichcorrelated to precipitation events, decreased with depthbelow the surface. This is attributed to the increasedstorage volume available in the rock mass with depthbelow the surface, which tends to buffer the rapid influx ofrecharge, and to the decrease in the occurrence of fracturesthat transmit water. Piezometers deep within the ridgecontained the greatest amounts of water, and theseconditions were maintained throughout the year, indicatingthe rocks in the core interior were saturated.

All piezometers at the site showed a decline in waterlevel during most of the year, but most began to reboundtoward the end of the study period (winter), whichcorrelates with the return of long-term precipitation events.

Potentiometric head measurements from all piezometersat the site indicate a downward vertical gradient existswithin the ridge. Although temporal fluctuation in waterlevels was

Conclusions 71

Figure 58. Conceptualized model of hydrochemical-facies zones present in the unmined ridge. Hydrochemical-facies zonesindicate areas where there is a high probabilibyt that water present will have water type consisting of a combination of theions listed for that zone.

observed in all piezometers, contoured profiles of thehead data from periods when the maximum and minimumwater levels were recorded showed that no significantchange occurs in the flow field during the year. Wateraccumulates in the base of the ridge, which would form apiezometric surface that approximates the topography.Head measurements in piezometers screened at the samestratigraphic interval throughout the base of the ridgeindicate ground water flows from the ridge interiortoward the southwest at a gradient of 0.037.

Identification of recharge areas based on tritium datacollected at the site revealed that the ground water in thecrest of the ridge contained the highest tritium content,which is interpreted as having the youngest age. Theoldest water, based on tritium deficiency, was found inthe ridge interior and in brines in the discharge areabelow the elevation of the major drainage. A contouredprofile of the tritium data for the site indicates a deeppenetration of relatively younger water along the hillsideof the ridge where the slope decreases. This area isdirectly below the portion of the ridge where the rocks aresuspected of being de-watered by fractures and coalseams.

A conceptual model of ground-water flow at the siteshows that the coals may create a "short circuit" effect tothe

downward movement of ground water, which may reenterthe ground-water system along the hill slope after flowingout of the coal seams as diffuse flow or as hillside springs.

The break in the degree of slope along the midsection ofthe ridge may result in a decrease in the velocity of surfacerunoff during storm events, which would allow for greaterinfiltration into the subsurface. Fractures along the hillslope probably provide the initial mechanism for deeppenetration of young recharge water into the subsurfaceover extended periods of time.

The water chemistry was extremely variable at the site.The pH of water ranged from a low of 5.37 in water derivedfrom a coal seam to a high of 8.92 in the ridge interior.

Coal seams yielded Ca-Mg-HC03 water types and typi-cally contained high iron concentrations with an acidic pH.Variation in the amount of total dissolved solids was attrib-uted to dilution from fresh-water influx, especially inpiezometers set into shallow coals and fractures.

Piezometers; in fractures mainly yielded Ca-HC03 or Mg-S04 water types. One piezometer yielded water samples thatshowed temporal variation, mainly in the amount of calciumand sodium. This variability can be explained by intrusionby ground water associated with major storm events thatinfiltrated into the subsurface and mixed with the groundwater


The valley bottom/below drainage piezometers near ma-jor drainage consistently contained barium in excess of 1.0mg/L. Barium concentrations as high as 48.6 mg/L were en-countered in the piezometer that contained brine. Baritesolubility is probably important in controlling the bariumconcentrations at these higher levels. Low redox conditionsin the

present in the system. Apparently, the fracture system has athreshold for which an intense precipitation event is neces-sary to produce an intrusion of fresh water to depth.

The ground water derived from the interior of the ridgewas predominantly Na-HC03 water type with a high pH;one piezometer contained sulfate as the dominant anion.Ground water deep below the major drainage stream wasan Na-Cl type. Sulfate concentrations were low in the deepinterior area and in the discharge area near the valleybottom, most likely resulting from sulfate reduction.Na-HC03 ground water flowing down gradient from theridge interior mixes with ground water from thenear-surface fracture system (Ca- or Ca-Mg-HC03), whichresults in an increase in the calcium and magnesiumconcentrations along the flow path. Chloride content alsoincreases, resulting from the gradual influence of Na-Clbrines that are nearest the surface in the valley bottoms.

Reaction path chemical modeling using the PHREEQEand BALANCE computer codes produced results that wereconsistent with the observed trends at the site. The chemicalevolution of ground water begins with recharge entering thesystem as precipitation and reacting with minerals in theshallow subsurface or in fractured bedrock. The solution ofcalcite, the oxidation of pyrite, and the chemical weatheringof chlorite, feldspars, and siderite probably exert thegreatest influence on major ion chemistry in the shallowzone. Cation exchange coupled with sulfate reduction act tofurther alter the chemistry of the ground water as it movesdownward into the interior of the ridge. Cation exchangereactions where sodium is exchanged for the more preferreddivalent cations (Ca2+, Mg2+, and Ba2+) results in a decreasein the concentration of these ions in ground water and theformation of an Na-HC03 water type. Clay minerals, mostlyas kaolinite, probably act as the host for exchange reactions.Sulfate reduction also occurs when ground water moves toareas where redox conditions are favorable for the activityof sulfur-reducing bacteria. Evidence of sulfate reduction isincreased alkalinity, reduced sulfate concentration, and theproduction of hydrogen sulfide.

The solubility constraint exerted on bariumconcentrations by barite was not as important as the limitingeffect by cation exchange reactions. Piezometers in theridge interior contained very low concentrations of barium,although concentrations of sulfate were low. Ground watermoving from the ridge interior mixes with water from thenear-surface fracture zone, which has higher calcium andmagnesium concentrations. These cations can compete forexchange sites on clays as the ground water continues alongthe flow path and comes in contact with rocks that containbarium on exchange sites. Barium is released into solutiondue to competition for the exchange sites, resulting inincreased barium concentrations in the ground water.

valley bottom are conducive to the prolonged activity ofsulfur-reducing bacteria, which serve to keep sulfateconcentrations low. Sodium chloride ground water alsomay contribute to sustain high barium concentrationsbecause of its increased solubility effect on barite.

Elevated barium concentrations did occur in awater-producing zone that was above drainage. A drop inthe redox conditions, sulfate reduction, and an increase incalcium concentration correlated to the increase in bariumconcentration, which is consistent with the mechanism forthe occurrence of barium in ground water reported here.

The occurrence of elevated fluoride concentrations is re-stricted to the interior ridge areas that are characterized byan Na-HCO3 water type with a high pH. Fluoride mostlikely is released into solution during anion exchangereactions with hydroxyl ions on clay minerals (kaolinite) orgoethite. The decrease in fluoride concentrations as theground water moves away from the ridge interior isprobably a result of dilution and/or re-adsorption.

A conceptual model (Fig. 58) can be used to show thelikelihood for the occurrence of geochernical facies withinthe ridge. The model shows four zones where theprediction of the major cations and anions comprising thewater type for a particular water sample could be predictedwith a high degree of probability. The model shows adepressed salt-water interface below the ridge due todownward movement and accumulation of fresh water. Thehydrostatic pressure imparted on the salt water istransmitted in the salt-water zone, causing it to rise atlocations where fractures breech confining layers (valleybottoms).

The model is based on the site-specific data collectedand interpreted from this site. However, this model shouldbe applicable to other areas of the dissected AppalachianPlateau that possess a similar hydrogeologic setting.Site-specific conditions, such as the geology of the site andits proximity to surface drainage of the third order orhigher, must also be considered.

ACKNOWLEDGMENTSThe author would like to express his gratitude to his

graduate committee, Drs. Lyle V. A. Sendlein, JamesDinger, Sue Rimmer, V. P. "Bill" Evangelou, and JohnThrailkill, for their review of this report. I would like towish Dr. Thrailkill a happy retirement, and thank him onbehalf of all scientists for his many significantcontributions to hydrogeology and geochemistrythroughout his productive career. Special thanks go out tothe administration and staff of the Kentucky GeologicalSurvey for their help and support, especially the hydro-geologists in the Water Section. I would also like to extendmy sincere thanks to Cypress Southern Realty, especiallyMr. John Tate, for providing the suitable site for this study.This project could not have been performed without JimDinger's belief in the project and his patience andconfidence in me to get it done, and for that I will alwaysbe indebted. I thank Jim Kipp for his friendship andcamaraderie, and his thought-provoking discussions andcritical review during several stages of this study. PageTaylor, Douglas Graham, and Shelly Minns all helped withfield and computer work, and I am grateful to each of them.I sincerely hope they each found the

Acknowledgments 73

entire episode as great a learning experience as I have. MegSmath helped with the editing and completion of this finaldocument. Her patience and helpful comments throughoutmy tenure at KGS have helped me become a better writer. Ithank Nina Wunsch for her love and support when it wasneeded.

Finally, this dissertation is dedicated to TheodoreWunsch, my grandfather. His patience and humility havealways been an inspiration to his grandchildren. Althoughhis education was modest, hard work provided him themeans to allow his son to make a better life for himself,which ultimately allowed for his grandchildren to obtaincollege educations. May this dissertation serve as aninspiration to all of his descendents, present and future, towork hard to achieve their highest expectations in work,life, and love.

REFERENCESAlexander, Martin, 1977, Introduction to soil microbiology

[2d ed.]: New York, John Wiley, p. 362-367.Aller, L., Petty, R., Lehr, J. H., Nielson, D., and Denne, J.

A., 1989, Handbook for suggested practices for thedesign and construction of ground-water monitoringwells: National Water Well Association, 398 p.

Arora, R., and Rodenbeck, S. A., 1983, Barium andradioactivity levels within groundwater of Georgiacoastal plain aquifers: Geological Society of America,Southeastern Section, March 16-18, 1983.

Banasczak, K. J., 1980, Coals as aquifers in the UnitedStates: Symposium on Surface Mining Hydrology, Sedi-mentology, and Reclamation, University of Kentucky,December 1-5,1980.

Bienkowski, L. S., 1990, Delineation and characterizationof the aquifers of the Eastern Kentucky Coal Field:Ph.D. Dissertation, University of Kentucky, Lexington,319 p.

Bohn, H. L., McNeal, B. L., and O'Connor, G. A., 1985,Soil chemistry: New York, John Wiley, 341 p.

Bower, C. A., and Hatcher, J. T., 1967, Adsorption offluoride by soils and minerals: Soil Science, v. 103, p.151-154.

Brown, E., Skougstad, M. W., and Fishman, M. J., 1970,Methods for the collection and analysis of watersamples for dissolved minerals and gases: U.S.Geological Survey Techniques of Water ResourcesInvestigations, Book 5, 160 p.

Caithamer, C. E., 1983, Mineralogical sources for bariumin Cambro-Ordovician aquifers of northeastern Illinois:M.S. Thesis, Northern Illinois University, DeKalb,Illinois, 189 p.

Caruccio, F. T., Geidel, G., and Pelletier, 1980, The asses-sment of a stratum's capability to produce acid minedrainage: Proceedings of the Symposium on SurfaceMining Hydrology, Sedimentology, and Reclamation,1980, Lexington, Kentucky.

Champ, D. R., Gulens, J., and Jackson, R. E., 1979, Oxida-tion-reduction sequences in ground water flow systems:Canadian Journal of Science, v. 16, no. 1, p. 12-23.

Collins, G. A., 1975, Geochemistry of oilfield waters: NewYork, Elsevier Scientific Publishing Company, 496 p.

Coltharp, G. B., and Brooks, L. T., 1991, Incident andthroughfall precipitation chemistry on a mixedhardwood watershed in eastern Kentucky [abs.]: WaterResources Symposium, University of Kentucky,Lexington, Kentucky, March 14-15, 1991.

Connell, W. E., and Patrick, W. H., 1968, Sulfate reductionin soil: Effects of redox potential and pH: Science, v.159.

Corbett, R.G., and Manner, B. M., 1984, Fluoride in theground water of northeast Ohio: Ground Water, v. 22,no. 1.

Danilchik, W., and Waldrop, H. A., 1978, Geologic map ofthe Vest Quadrangle, eastern Kentucky: U.S. GeologicalSurvey Geologic Quadrangle Map GQ-1441.

Davis, S. N., 1986, Workbook for practical isotopehydrology: National Water Well Association, September8-9, 1986, San Diego, California.

Diamond, W. P., 1972, Differentiation of marine andnonmarine carboniferous environments of easternKentucky based on clay mineral composition: M.S.Thesis, University of Kentucky, 105 p.

Fleischer, M., 1974, Fluorine, in Geochernistry and theenvironment, the relationship of selected trace elementsto health and disease, v. 1: National Academy ofSciences.

Foster, M. D., 1950, The origin of high sodium bicarbonatewaters in the Atlantic and Gulf Coastal Plains:Geochemica. et Cosmochimica Acta, v. 1, p. 33-48.

Freeze, R. A., and Cherry, J. A., 1979, Groundwater:Englewood Cliffs, N.J., Prentice-Hall, Inc.

Gilkeson, R. H., Cartwright, Keros, Cowert, J. B., andHoltzman, R. B., 1983, Hydrogeologic and geochernicalstudies of selected natural radioisotopes and barium inground water in Illinois: Illinois State Water Survey,Champaign, Illinois, WFIC Report 180, 93 p.

Gilkeson, R. H., Perry, E.C., Jr., and Cartwright, Keros,1981, Isotopic and geologic studies to identify thesources of sulfate in ground water containing highbarium concentrations: Urbana-Champaign, Illinois,University of Illinois, Water Resources Center, GrantReport/Illinois State Geological Survey 1981-4, 39 p.

Groenewold, G. H., Rehm, B. W., and Cherry, J. C., 1981,Depositional setting and groundwater quality incoal-bearing sediments and spoils in western NorthDakota: SEPM Special Publication 31, p. 157-167.

Harlow, G. E., and LeCain, G. D., 1991, Hydrauliccharacteristics of, and ground-water flow in,coal-bearing rocks of southwestern Virginia: U. S.Geological Survey Open-File Report 91-250.

Heck, E. T., 1940, Barium in Appalachian salt brines:American Association of Petroleum Geologists Bulletin,v. 24, no.3,p.486-493.

Hem, J. D., 1985, Study and interpretation of the chemicalcharacteristics of natural water [3d ed.): U.S. GeologicalSurvey Water-Supply Paper 2254.


Hendry, J.M., 1988, Do isotopes have a place inground-water studies?: Groundwater, v. 26, no. 4.

Hinrichs, E. N., 1978, Geologic map of the Noble Quad-rangle, eastern Kentucky: U.S. Geological Survey Geo-logic Quadrangle Map GQ-1476.

Hoefs, Jochen, 1987, Stable isotope geochemistry [3ded.]: New York, Springer-Verlag, 341 p.

Hopkins, H. T., 1966, Fresh-saline interface map of Ken-tucky: U.S. Geological Survey, Scale 1 in. = 8 mi.

Kipp, J. A., and Dinger, J. S., 1987, Stress relief fracturecontrol of ground-water movement in the AppalachianPlateaus: Proceedings, National Water WellAssociation, Focus Conference on Eastern RegionalWater Issues, Burlington, Vermont, July 14-16, 1987,p. 423-438.

Kipp, J. A., Lawrence, F. N., and Dinger, J. S., 1983, Aconceptual model of ground water flow in the EasternKentucky Coal Field: Symposium on Surface Mining,Hydrology, and Reclamation, 1983, University ofKentucky, Lexington, Kentucky, p. 543-548.

Levinson, A. A., 1980, Introduction to explorationgeochemistry: Wilmeth, Illinois, Applied Publishing,Ltd., 923 p.

McGrain, Preston, 1953, Miscellaneous analyses of Ken-tucky brines: Kentucky Geological Survey, ser. 9,Report of Investigations 7, 16 p.

McIntosh, G. E., and Miler, K. P., 1989, HC-GRAM,technical support softwar0or the graphical portrayal ofhydrochemical data: U.S. Department of the Interior,Off ice of Surface Mining.

McNabb, J. F., and Dunlap, W. F., 1975, Subsurfacebiological activity in relation to ground water pollution:Ground Water, v. 13, no. 1 9 p. 33-44.

Moore, R. B., and Staubitz, W. W., 1984, Distribution andsource of barium in ground water at Cattaraugus IndianReservation, southwestern New York: U.S. GeologicalSurvey Water Resources Investigations ReportWRI-84-4129.

Nordstrom, D. K., 1977, Thermochernical redox equilibriaof ZoBell's solution: Geochemica et CosmochimicaActa, v. 41, p. 1835-1841.

Papp, A. R., 1982, Extractable cations in claystones andshales of the Pennsylvanian Breathitt Formation in east-ern Kentucky as related to depositional environments:M.S. Thesis, Eastern Kentucky University, 153 p.

Parkhurst, D. L, Thorstenson, D. C., and Plummer, N. L.,1980, PHREEQE, a computer program for geochernicalcalculations: U.S. Geological Survey Water-ResourcesInvestigations Report 80-96, 210 p.

Parkhurst, D. L, Plummer, N. L., and Thorstenson, D. C.,1982, BALANCE-A computer program for calculatingmass transfer for geochernical reactions in groundwater: U.S. Geological Survey Water-ResourcesInvestigations Report 82-14.

Perry, E. C., Grundl, T., and Gilkeson, R. H., 1980,Hydrogen, oxygen and sulfur isotopic study of theground water in the

Cambro-Ordovician aquifer system of northern Illinois:Northern Illinois University Press, 43 p.

Piper, A. M., 1944, A graphic procedure in thegeochemical interpretation of water analyses:Transactions, American Geophysical Union, v. 25, p.914-923.

Powell, J. D., and Larson, J. D., 1985, Relation betweenground-water quality and mineralogy in the coal-produc-ing Norton Formation of Buchanan County, Virginia:U.S. Geological Survey Water-Supply Paper 2274.

Price, W. E., Jr., Mull, D. S., and Chabot, Kilburn, 1962,Reconnaissance of ground-water resources in theEastern Coal Field Region-Kentucky: U.S. GeologicalSurvey Water-Supply Paper 1607, 56 p.

Quinones, F., Mull, D. S., York, K., and Kendall, V., 1981,Hydrology of area 14, Eastern Coal Province, Kentucky:U.S. Geological Survey Water Resources InvestigationReport 81-137, 82 p.

Robertson, F. N., 1984, Solubility controls of fluorine,barium, and chromium in the ground water basins ofArizona, in Hitchon, Brian, and Wallick, E.I., eds.,Practical applications of ground-water geochemistry:Canadian/American Conference on Hydrogeology, 1st,Banff, Alberta, Canada, June 22-26, 1984, NationalWater Well Association.

Robertson, F. N., 1991, Geochemistry of ground water inalluvial basins of Arizona and adjacent parts of Nevada,New Mexico, and California: U.S. Geological SurveyProfessional Paper 1406-C, 89 p.

Robertson, F. N., and Garrett, W. B., 1988, Distribution offluoride in alluvial basins of Arizona and adjacent partsof California, Nevada, and New Mexico: U.S.Geological Survey Hydrologic Atlas HA-665, 3 sheets.

Schubert, J. M., 1980, Fracture flow of groundwater incoal-bearing strata: Symposium on Surface MiningHydrology, Sedimentation, and Reclamation, Universityof Kentucky, Lexington, Kentucky, December 1-5,1980.

Songer, N. L., 1987, Seepage velocity in stress-relief frac-tures in Eastern Kentucky Coal Field: KentuckyDivision of Water, Ground Water Section, Frankfort,Kentucky, MOA 006131, 87 p.

Spengler, R. W., 1977, Geologic map of the SalyersvilleSouth Quadrangle, Magoff in and Breathitt Counties,Kentucky: U.S. Geological Survey Geologic QuadrangleMap GQ-11373.

Sprinkle, C. L., Davis, R.W., and Mull, D.S., 1983,Evaluation of ground-water quality data from Kentucky:U.S. Geological Survey Water-Resources InvestigationsReport 83-4240.

Stach, E., 1982, Coal petrology: Gebruber Borntraeger,Berlin, Stuttgart, 535 p.

Stone, R., and Snoeberger, D. F., 1978, Cleat orientationand areal hydraulic anisotropy of a Wyoming coalaquifer: Groundwater, v. 15, no. 6.

U.S. Bureau of Reclamation, 1974, The design of smalldams: Technical Publication.

References 75

U.S. Environmental Protection Agency, 1984,Environmental radiation data report 39: U.S. EPA,July-September, EPA 520/5-85-009.

Weast, R. C., ed., 1975, Handbook of chemistry andphysics [51st ed.]: Cleveland, Ohio, Chemical RubberCompany, various pagination.

Weinheimer, R. L., 1983, Vertical sequence anddiagenesis in Breathitt sandstone of eastern Kentucky:M.S. Thesis, University of Cincinnati, Cincinnati,Ohio, 147 p.

Winograd, 1. J., and Farlekas, G. M., 1974, Problems in14C dating of waters of deltaic origin, isotopehydrology: International Atomic Energy Agency,Vienna, Austria, p. 69-93.

Wood, W. W., and Low, W. H., 1988, Solutegeochemistry of the Snake River Plain regionalaquifer system, Idaho and eastern Oregon: U.S.Geological Survey Professional Paper 1408-D, 79 p.

Wunsch, D. R., 1982, Fluoride distribution and relation tochemical character of ground water in northeast Ohio:M.S. Thesis, University of Akron, Akron, Ohio,101 p.

Wunsch, D. R., 1988a, High barium concentrations inground water in eastern Kentucky: Proceedings of theGround Water Geochemistry Conference, NationalWater Well Association, Denver, Colorado, February16-18, 1988.

Wunsch, D. R., 1988b, The effect of sulfur-reducingbacteria on high barium concentrations in ground waterin eastern Kentucky: EOS Transactions, AmericanGeophysical Union, v. 69, no. 44.

Wyrick, G. C., and Borchers, J. W., 1981, Hydrologiceffects of stress relief fracturing on an AppalachianValley: U.S. Geological Survey Water-Supply Paper2177, 51 p.

Zack, A., and Roberts, 1., 1988, The geochemical evolutionof aqueous sodium in the Black Creek aquifer, Horryand Georgetown Counties, South Carolina: U.S.Geological Survey Water-Supply Paper 2324,15 p.

Ground-Water Geochemistry and Its Relationship to the Flow System at an Unmined Site in the Eastern Kentucky Coal Field 76

APPENDIX 1: A Discussion on the Probable Weathering Reactions in Appalachian

Coal-Bearing Strata*

NOTE: Some of these reactions shown in this discussioninclude oxidation/reduction. For simplicity, only thestochiometric equations will be represented.

Water in contact with carbon dioxide in the atmosphereor concentrated in soil zones dissolves the gas, yieldingcarbonic acid by the following reaction:

CO2 + H2O -+ H2CO3 � H+ + HC03- (7)

Concentrations of C02 in the soil or shallow subsurfacecan reach several orders of magnitude higher thanatmospheric levels (Freeze and Cherry, 1979). Groundwater containing significant carbonic acid and dissolvedoxygen is important in the process of chemical weathering.Some of the weathering reactions that can be expectedwith the minerals typical to the coal-bearing rocks in thisstudy are as follows:

Silicate Hydrolysis

In the process of silicate hydrolysis, silicate mineralsreact with H+ and produce clays, dissolved silica, metalions, and bicarbonate ions. For example:

KAlSi3O8(s) + H2C03 + 9/2 H2O =(microcline)K+ + HCO3

- + 2H4SiO4 + A12Si205(OH)4(s) (8)(kaolinite)

The reaction consumes H+ and H2CO3, which increases thealkalinity and pH of the water. Hydrolysis reaction mayproduce ground waters with a pH as high as 10 (Freezeand Cherry, 1979).

Chlorite is also an abundant mineral in coal-field rocks.The weathering of chlorite can be represented as:

Mg3(OH)6(Mg2AI)(AlSi3)O10(OH)2+6H++HCO3- =

(chlorite) (9)

3Mg2+ + 6H2O+6HC03-+(Mg2A1)(AlSi3)O10(OH)2

(talc)Talc is not common to the mineral suite described by

Powell and Larson, and they suggest that talc reacts withadditional hydrogen ions, which may add Mg2+ to thesystem. If the chlorite is iron rich, ferrous iron will bereleased into the system, which may be oxidized to ferriciron and produce additional hydrogen ions, thus loweringthe pH. If the magnesium content is greater relative toiron, the tendency will be toward chlorite weathering toincrease alkalinity and pH.

4Fe2+ + O2 +4H+ = 4Fe3+ + 2H2O (10)Fe3+ + 3H2O = Fe(OH)3 + 3H- (11)

________________________________*Discussion largely taken from Powell and Larson(1985)

Carbonate Dissolution

The dissolution of carbonate minerals releases metalions (e.g., Ca2+, Mg2+, Fe2+), bicarbonate, and raises the pHof ground water. At least two carbonate minerals aresuspected of being present in the study area. Thedissolution of calcite with water containing carbonic acidis:

CaCO3 + H2CO3 � Ca2+ + 2HC03- (12)In this reaction, 2 moles of bicarbonate ions are formed

from the carbonate from the calcite plus the disassociationof the carbonic acid; therefore, this reaction cansignificantly increase alkalinity and pH.

The dissolution of carbonate minerals in the presenceof acid generated by the weathering of sulfide mineralscontributes less to alkalinity than does reaction withcarbonic acid.

Siderite was the most abundant carbonate mineralfound in the investigation by Powell and Larson. It wasobserved as nodules in Star Fire core 1066, and is noted asbeing present by Weinheimer (1983). The dissolution ofsiderite releases ferrous iron, which may oxidize as shownin reaction (4), releasing H+.

FeCO3 + H2C03 = Fe2+ + 2HCO3- (13)

(siderite)The H+ will be neutralized by bicarbonate derived from

the carbonic acid and siderite; therefore, the net effect isan increase of iron to the ground water without the largeincrease in acidity that the oxidation of pyrite wouldproduce.

A common reaction in coaI-field ground water systemsis the oxidation of sulfide minerals (pyrite and marcasite).The products of this reaction are sulfate, ferrous and ferriciron, hydrogen ions, and amorphous iron hydroxideprecipitate. The reactions are as follows:

Dissolution of Pyrite:

FeS2+3.5O2+H2O � Fe2++2SO42-+2H+

Oxidation of Ferrous to Ferric Iron:

4Fe2+ + O2 +4H+ = 4Fe3+ + 2H2O

Hydration of Ferric Iron:

Fe3++3H2O = Fe(OH)3(s)+3H+

The net reaction under oxidizing conditions is then:

77 Appendix I

FeS2 + 3.75O2 � Fe(OH)3(s) + 2SO42- + 4H+ (14)

With no mineral source of sulfate (i.e., gypsum is notpresent) the sulfate in the ground water is a goodindicator of the amount of pyrite oxidation takingplace. In the case of oxygen depletion, oxidation ofsulfide by ferric iron present in the water may takeplace. The net reaction:

FeS2 + 14Fe3+ + 8H2O � 15Fe3+ + 2SO42- + 16H+

In this reaction, the oxidation of Fe2+ to Fe3+ iscatalyzed by bacteria such as Thiobacillusferrooxidans. The activity of these bacteria isenhanced at pH less than 5.5 (Caruccio and others,1980).


APPENDIX 2: Geologic Core Description, Piezometer Construction Specifications,Piezometer Construction Diagrams

Appendix 2 79

1 . Drillers are invited to perform the installation ofapproximately 10 ground-water monitoring wells oruntil allocated grant funds are expended. Totalcumulative depth of completed holes is anticipatedto be approximately 1,300 feet.

2. Permanent casing will be 2 inch I.D., threaded, flush joint, PVC schedule 40. Casing shall beinstalled vertically in the ground such that a 6-footbailer will pass through it without obstruction (thismay necessitate the use of centralizers), extend 2feet above ground elevation, and have a threadedremovable cap. A bottom cap (plug) must beinstalled at the bottom of the screen to prevent sedi-ment from entering the well. All threaded joints willhave an O-ring seal. No grease or petroleum-basedlubricants will be allowed. Teflon-based or otherEPA-approved lubricants may be allowed at thediscretion of the geologistin-charge.

3. The monitoring wells will be completed in thefollowing steps (see Figure 2):(1) Placement of equipment over an existing

borehole of suitable diameter drilled to thetargeted depth,

(2) 1 foot of pure quartz (size FX20) sand emplacedto the bottom of the hole,

(3) The slotted screen (0.010 slot size) and pipe(2-inch) installed, centralizers may be installedscreen to center the pipe.

(4) Pure quartz sand (coarse sand, 0.033 to 0.46 inchdiameter) emplaced from bottom of the screen toa minimum of 3 feet above the screen,

(5) 3 feet of pure quartz sand (fine sand, 0.008 to0.12 inch diameter) emplaced,

(6) A minimum of 3 feet of bentonite seal(consisting of bentonite pellets) emplaced,

(7) Bentonite slurry tremied into the hole up to anelevation that is below the chosen elevation ofthe next piezometer,

(8) 3-foot bentonite seal (bentonite pellets)emplaced. A MINIMUM OF 1 HOUR MUSTELAPSE BEFORE STEP 9 IS STARTED. Thisis to allow for expansion of the bentonite pellets.

(9) Emplace minimum of 5 feet of sand,

(10) Slotted screen installed,(11) Start sequence over again with step(12) After step 6, the bentonite slurry will be tremied

or pumped into the top of the hole.

NOTE:1 . No bridging of bentonite, bentonite slurry, or sand

will be permitted in the hole.2. No materials such as cementthatwill alter the

ground-water quality may be introduced into the hole.3. All water used for slurry mixing, washing, and

decontamination shall be chlorinated, potable water.4. The delay between holes shall not be more than 1 day

unless approved by the University geologist-in-charge.

5. Locations shall be cleaned up by the contractor to thesatisfaction of the geologist-in-charge and cost of anydamages due to acts of the contractor shall be borneby the contractor.

6. All materials connected with well installation are to bemaintained in good condition. Plastic will be laid on theground surface so that all materials that will be installed inthe monitoring well such as, but not limited, to sand, wellproduction casing, and well screens can be stored on thisplastic, and plastic will cover the top of such materials tominimize contamination by foreign materials until they areinstalled in the well. PVC production pipe that is packagedin self-contained plastic bags or coverings will be pre-ferred. Buckets, water tanks, tremie pipes, and otherdevices used to carry and emplace materials in the hole willbe kept clean of foreign materials.

7. The contractor may partially complete a well bysetting the deeper of the two piezometers andcommence completion of an adjacent well while thebentonite and other materials are setting up in thefirst well.

8. All work is to be effected in a diligent manner usingstandard acceptable techniques. Equipment is to bemaintained in good working order.

9. Access rights to the property will be obtained by theKentucky Geological Survey. Suitable drilling padswill be the responsibility of the Kentucky GeologicalSurvey.

SPECIFICATIONS FOR INSTALLATION OF PRODUCTION PIPE AND SCREEN

Appendix 2 81

Appendix 2 83

Appendix 2 85

Appendix 2 87


APPENDIX 3: Analytical Laboratory Methods, Methods for Tritium Collection,Description of Geochemical ModelsKENTUCKY GEOLOGICAL SURVEY

Computer and Laboratory Services SectionAnalysis Parameters

INORGANIC-METALMethod: EPA 200.7a-Inductively Coupled Plasma (ICP)Element DML µg/LAluminum 18.0Antimony 42.0Arsenic 58.0Barium 0.40Beryllium 0.60Boron 27.0Cadmium 4.0Calcium 20.0Chromium 5.0Cobalt 10.0Copper 3.0Gold 8.0Iron 4.0Lead 36.0Lithium 75.0

Method: EPA 200.9 GFAA MethodsElement MDL µg/LArsenic 1.70Chromium 1.60Lead 1.36

INORGANIC-NONMETALParameterAcidityAlkalinityBicarbonateBromideCarbonateChlorideConductanceFluoridepHSulfateTotal HardnessOxygen (Dissolved)

Element MDL µg/LMagnesium 29.0Manganese 2.0Nickel 20.0Phosphorus 77.0Potassium 926Selenium 52.0Silicon 34.0Silver 4.0Sodium 21.0Strontium 0.30Sulfur 29.0Thallium 34.0Tin 99.0Vanadium 5.0Zinc 4.0

Element MDL µg/LCadmium 1.90Copper 1.45Nickel 2.11

Method MDL mg1LEPA 305.1 8.00EPA 310.1 3.00Calculated 3.00EPA-300.0 (IC) 1.00Calculated 1.00EPA-300.0 (IC) 1.00EPA 120.1 0.001EPA-340.2 0.020EPA-150.1 0.001EPA-300.0 (IC) 5.00Calculated 1.00EPA 360.1 0.010

Appendix 3 89

ADVICE ON SAMPLING REVISED 20 September 1985SAMPLING OF ENVIRONMENTAL WATER FOR LOW-LEVEL TRITIUM ANALYSIS

A. ExplanationTritium in environmental samples will be determinedwith a limit of detection of 0. 1 T-units JU) (0.0003pli/ml). Water vapor of the open air varies from 2 to100 TU. Indoors, the atmospheric humidity may reach10,000 TU from various luminescent dials. Exposureof the water to such air at any temperature might givebadly erroneous tritium results.

B. Sample bottlesFor lowest level of tritium samples we recommendusing I liter (1 quart) glass bottles with "PolySeal,"conical inset caps. The bottles should be clean anddry, preferably factory fresh. If transfer is to be madeindoors, the dry bottles should first be filled withargon gas. See below.If the very lowest detection level is not needed,heavy-wall plastic bottles may be acceptable. Musthave good caps. Hold a filled bottle upside down andsqueeze hard. No leakage is allowed. Remember thatthere are large pressure changes in air transport.

C. Sampling procedures1 . Sample transfer should be done outdoors, unless a

specially vented room is available with ban onwristwatches.

2. THE PERSON PERFORMING THE SAMPLETRANSFER IS NOT ALLOWED TO WEAR AWRISTWATCH, COMPASS, OR SIMILARWITH LUMINESCENT DIALS ORSO-CALLED "BETA" LIGHTS.

3. Fill the bottle close to the neck with sample. Donot rinse. Overflow is not desirable.

4. Replace and screw cap on tightly.5. Record bottle numbers on original field data

sheets, and fill in information on bottle label.6. If sampling indoors, never let the water be

exposed to the air. Pipe the sample water into themiddle of an argon-filled bottle (below the argonlevel). Do not pour the argon out before by tiltingan open bottle.

Mass Balance Modeling Using BALANCEThe model BALANCE (Parkhurst and others, 1982)

calculates the mass transfer (amount of materials orphases entering or leaving the liquid phase) necessary toaccount for the observed changes in ground-watercomposition collected at two points along a flow path(Deutsch and others, 1991). A phase can be a mineralsolid, gases, ion exchangers, and aqueous solutions. Themodel calculates the amount of each phase needed tosolve the following equation:

INITIAL SOLUTION(A) + REACTANT PHASES = FINAL SOLUTION + PRODUCT PHASES

A series of simultaneous equations using initial andfinal water compositions and selected mineral or gasphases define the changes in molality of each constituentalong a flow path by determining reaction coefficients forthe specified mineral stoichiometery or gas compositions(Robertson, 1991). The

mass transfer is defined as the number of millimoles perkilogram H2O of reactant or product phase that enter orleave the aqueous phase. The number of phases used in themodel must be equal to the number of elements in thechemical composition of the water used.

The model input requires the concentrations of the end--member solutions (in molar concentrations), the proposedmineral phases, and the stochiometric coefficient of theions contained with a mineral. The model output produces aset of values that indicate if a particular mineral phase isdissolving (positive) or precipitating (negative).

Geochemical Calculations Using PHREEQEPHREEQE (Parkhurst and others, 1980) is a computer

program designed to model geochernical reactions.PHREEQE stands for pH-Redox-Equilibria, and wascreated by merging parts of three previously developedmodels, WATEQF, WATEQ2, and MIX2. This modelcalculates the activities and concentrations of the solutionspecies entered into the model, along with calculating thepartial pressures of dissolved gases, and mineral saturationindices (Deutsch and others, 1991). Saturation indices aredefined in terms of the saturation index (SI):

SI = Log(IAP / K)where IAP is the ion activity product and K is theequilibrium constant. If the IAP is less than zero, thesolution is undersaturated with respect to a given mineral,and the mineral may dissolve. If the SI is equal to zero, thesolution is in equilibrium; if the SI is greater than zero, thesolution is saturated or supersaturated, and the mineralwould be expected to precipitate from the solution.However, the model does not take into account kineticfactors or stability effects of a particular mineral in thepresence of others. Therefore, it cannot be absolutely deter-mined if a species does indeed precipitate or dissolve aspredicted.

Reaction-Path ModelingWhen using a reaction-path model to describe the

changes in water chemistry along a flow path, the state ofthe phases (i.e., either dissolving or precipitating)determined by the mass-transfer model must be consistentwith the equilibrium model (Robertson, 1991). Most likelythere may be several possible scenarios that satisfy theconditions described by the BALANCE model andPHREEQE; thus, a unique solution may not be found.However, through a process of elimination, solutionscontaining suites of phases that satisfy the conditions setforth by the two models can most likely be determined,giving way to possible scenarios of geochernical reactionsbetween rocks and minerals in the system that may beuseful in understanding of the hydrogeochernical evolutionof water moving along a flow path.

The suite of minerals used in this study was postulatedfrom the following sources of information: (1) those usedby Powell and Larson (1985) because, as discussedpreviously, the geologic and geomorphic setting of the areaused in their study of

90 Ground-Water Geochemistry and Its Relationship to the How System at an Unmined Site in the Eastern Kentucky Coal Field

ground water in coal-bearing rocks in Virginia is verysimilar to this study site, (2) published mineralogical datafrom the literature (e.g., Weinheimer, 1983), and (3)minerals that are pertinent to this study that may imposesolubility controls on ions of interest (i.e., barite andfluorite). In a modeling scenario presented here, usuallythe number of possible phases or minerals exceeds thenumber of constituents (Plummer and others, 1983;Robertson, 1991). Therefore, the approach taken here willbe to form a "best fit" model, where reactions that aremost compatible with the water chemistry data andminerals present, along with solubility and kineticconsiderations, will be used.

Binary Mixing Using HC-GRAMHC-GRAM (McIntosh and Miller, 1989) is a program

designed to graphically display water-quality data usingeither Piper trilinear diagrams or Stiff diagrams. Theprogram allows for up to 400 water samples to be enteredat one time.

Binary mixing of two water samples is one of theoptions available in the program. The program willcalculate a straight-line mix between the two watersamples, with the user choosing the acceptable tolerancefrom a straight line used in calculations. The smaller thenumber, the less tolerance from a straight line will besearched for in the mix. The output shows the percent ofmajor ions (in millequivalents) of the input samples and acalculated mix of the two at the specified tolerance. Apercent error for the resultant mix water and the actualsample chemistry is also calculated.

Appendix 4

APPENDIX 4: Description of Drill Cuttings, Borehole Diagnostics from DownholeCamera Investigations, Geophysical Logs

91

Appendix 4 93


Downhole Camera DiagnosticsMonitoring Well 11

Appendix 4


95

Appendix 4


97

Appendix 4


99

Appendix 4


101

Appendix 4 103

Appendix 4

SF MONITORING WELL 11GAMMA LOG

11/29/90

105


SF MONITORING WELL 11SP RESISTIVITY LOGS

11/29/90

T. D. *172 FEET

DAILY PRECIPITATION DATA

0 0 0 0

0.197 0

0 0 0 0

0 0 0 0

0.459 0

0 0 0 0

0 0 0 0

0 0 0 0

0 0 0

0.656

APPENDIX 5: Daily Precipitation Data, Robinson Forest Precipitation Data

APPENDIX 6: QA/QC Chemical Data, Water-Quality Tests of Well Construction Materials, Water-SampleChemical Analyses

i I I I

Appendix 7 125

Appendix 7 127

issn 0075-5621 kentucky geological survey university of ... · issn 0075-5621 kentucky geological...

Documents