Proceedings of the

                     Third Conference
on  Hydrogeology, Ecology, Monitoring, and Management
            of Ground Water in Karst Terranes
                    December 4-6, 1991

                    Maxwell House/Clarion
                     Nashville, Tennessee
                          Presented by
        the U.S. EPA and the Association of Ground Water Scientists and Engineers,
                        a division of NGWA

-------
                      Ground Water Management
                         Book 10 of the Series

                          Proceedings of the

              Third Conference on Hydrogeology, Ecology,
              Monitoring and Management of Ground Water
                           in Karst Terranes

                         December 4- 6,1991
                        Maxwell House Clarion
                         Nashville, Tennessee
Abstract
The 1991 Karst Conference was comprised of 2 days of technical presentations,
and a day-long field trip.
A total of 46 papers were presented at the meeting.

Sessions were devoted to the following topics:

                 Ground Water Modeling and Hydrogeology
                          Site Characterization
        Geophysical and Other Techniques for Studying Karst Aquifers
           Hydrogeology and Processes Occurring in Karst Aquifers
                             Case Histories
                   Ecology of Caves and Karst Terranes
                         Ground Water Monitoring
           Emergency Response and Ground-Water Management
                       Ground Water Management


The meeting was co-sponsored by the Environmental Protection Agency, and the
Association of Ground Water Scientists and Engineers (a division of NGWA).

This bound volume is a compilation of papers that were presented at the
meeting.

Materials appearing in this publication are indexed to Ground Water On-Line, the
data base of the National Ground Water Information Center at (614) 761-1711.

-------
Ground Water Management
  A journal for rapid dissemination of ground water research

To minimize delay of new findings and insights, Ground Water Management
publishes papers exactly as submitted, without technical and grammatical editing
or peer review. It is our belief that these papers have technical merit. Complete
accuracy or technical viability cannot, however, be assured.  It is believed,
nevertheless, that early publication and rapid dissemination outweigh any
possible reduction in quality that may be encountered.

Papers published by Ground Water Management were presented at conferences
sponsored by the Association of Ground Water Scientists and Engineers, a
division of the National Ground Water Association.

Staff
Anita Stanley, Managing Editor
Jim Quintan, Special Editor (Karst Proceedings)
Chris Miller, Manuscript Coordinator
Pat Alcorn,  Business Manager
Robert I son, Circulation
Suzy Colley, Graphics

Subscription Rates
Subscriptions to Ground Water Management are for any six issues.  Special
coupon books are issued to each subscriber to redeem for issue topics of
interest. Each subscriber will receive notification of the printing of each issue of
Ground Water Management.

A subscription expires when  all six coupons have been redeemed.  A six-issue
subscription is $140 (192.50 to non-members), a savings of $70 when compared
to the single issue price. Subscribers who attend NGWA-sponsored conferences
have the option of using a subscription coupon for a second copy of the
conference's proceedings, or saving it for a future issue.

Responsibility
The Association of Ground Water Scientists and Engineers, a division of NGWA,
is not responsible for statements and opinions advanced in its meeting or printed
in publications.
 __,_, right
 1992 Water Well Journal Publishing Company
 All rights reserved.  ISSN: 1047-9023

 Claims
 Claims for missing issues must be made in writing within three months of
 publication, and will be subject to availability of back issues.

 Communications and Business Correspondence
 Ground Water Management, 6375 Riverside Drive, Dublin, Ohio  43017, (614)
 761-1711.
                                  i i

-------
Acknowledgements

The following people were co-leaders of the Conference Field trip:

                              Nick Crawford
                              Rocky Hannah
                               John Hoffelt
                               Jim Quinlan
                              Geary Schindel

A limited number of copies of the 86-page guidebook can be purchased from:
Friends of the Karst, Box 40922, Nashville, Tennessee 37204.

We wish to thank Geary Schindel for coordinating the field trip, and for
compilation of the guidebook.  We also wish to thank Eckenfelder Inc. of
Nashville for its support in preparation of the guidebook, and Grassmere Wildlife
Park for allowing the field trip participants to have access to the Park.

The following individuals served as session moderators:

                                Tom Aley
                               Dick Benson
                              Gareth Davies
                               Ralph Ewers
                               Jim Quinlan
Most of all, we want to thank the Environmental Protection Agency for partially
but very significantly underwriting the cost of the Conference.  The Agency's
financial support made it possible for more people to attend the meeting and to
vigorously participate in it.

-------
1 V

-------
                               Dedication

The National Groundwater Association's Third Conference on Hydrogeology,
Ecology, Monitoring and Management of Ground Water in Karst Terranes was
held in Nashville, Tennessee, in December 1991. Our field excursion on
December 5th took us to the shallow limestone plateaus around Murfreesboro,
30 miles to the southeast. A terrible, futile battle of the American Civil War was
fought there in December 1862 and January 1863; no doubt, some very elderly
people in Nashville today can pass on stories about it, told to them by oldsters
who, as children, heard the guns. It is just two generations away in recollection.
Returning to our hotels that evening, we learned that civil war flared again in
Croatia, with an artillery bombardment of Dubrovnik, one of the most beautiful
limestone cities in the world. This deeply saddened the many of us who have
made our pilgrimages to Yugoslavia and come to know and love it.

Yugoslavia (= "land of the southern Slavs") is the birthplace of karst studies in the
western tradition.  "Karst" itself is a Germanicization of Slavic and older words
meaning "stony ground" -- describing the rugged, denuded hillsides of (stria,
where much of the soil had been swallowed underground into karren pits and into
grikes. In Roman times, the regional name was Carsus. As early as 1150 AD
the city council of Trieste was promulgating statutes to limit karstic soil erosion --
officers of the modern EPA take their place in a long tradition! Between 1850
and 1910, the first intensive studies of karst geomorphology and hydrology
commenced in Slovenia and spread south  through Croatia, Serbia, Bosnia,
Herzegovina, and Monetenegro. These established the framework of academic
karst science, and many principles of karstland management as well, because
Yugoslav engineers have excelled in their careful, comprehending development
and regulation of the underground water resources.

Today the world karst specialists must travel there in order to measure their
particular karsts against the "classical" Dinaric model. When they do so, they
become enamored with the peoples and cultures of  these beautiful lands. In a
regional that is comparable to the Appalachians in length and breadth, there is an
astonishing variety of languages and faiths. Driving in the south with my family
one day, we encountered 3 different alphabets along one hundred  kilometers of
highway. Diversity is surely the greatest feature of our world cultural heritage.
We know that it can create local stresses, but grieve that these have led to
warfare in present-day Yugoslavia as they  did in the United States long ago.
This volume is respectfully dedicated to all the peoples and cultures of the
southern Slav lands. We urge them to resolve their difficulties without further
armed conflict.
                                                          Derek Ford
                                                          January 1992

-------
VI

-------
                                TABLE OF CONTENTS

                                                                       page
Keynote Address:
  Protecting Ground-Water in Karst Terranes: An EPA
  Priority for Action
  - LaJuana S. Wilcher	   3
Invited Lecture:
  Groundwater Modeling in Karst Terranes:
  Scale Effects, Data Acquisition and Field Validation
  - Georg Teutsch and Martin Sauter  	  17


                                     Session I:
                      Ground Water Modeling and Hydrogeology

Assessment of Hydraulic Conductivity in a Karst Aquifer at Local
  and Regional Scale
  - Martin Sauter 	  39
Approach for Delineating the Contributing Areas of a Well Field in
  a Carbonate-Valley Aquifer
  - Gary J. Barton and Dennis W. Risser	59
Effects of Quarry Dewatering on a Karstified Limestone Aquifer:
  A Case Study from the Mendip Hills, England
  - A.J. Edwards, S.L. Hobbs, and P.L. Smart	77


                                 Session II:
                            Site Characterization

Approaches to Hydrogeologic Assessment and Remediation of
  Hydrocarbon Contamination in Clay-Covered Karsts with
  Shallow Water Tables
  - Tony Cooley	95
Application of Dye-Tracing Techniques for Characterizing
  Groundwater Plow Regimes at the Fort Hartford Mine Superfund
  Site, Olaton, Ohio Co., Kentucky
  - Nicholas C. Crawford and Ginny L. Gray	113
Ground Water Monitoring in Unsaturated and Saturated Zones at
  a Site with Paleocollapse Structure
  - Richard Benson, Lynn Yuhr and Allen W. Hatheway	131


Luncheon Lecture:
  Chairman Mao;s Great Leap Forward  and the Deforestation
  Ecological Disaster in the South China Karst Belt
  - Peter W. Huntoon	149
                                  VI 1

-------
                                Session III:
                    Geophysical and Other Techniques for
                           Studying Karst Aquifers
                                                                         page

Electrochemistry of Natural Potential Processes in Karst
  - K.T. Kilty and A.L. Lange	163
Natural-Potential Responses of Karst Systems at the Ground
  Surface
  - Arthur L. Lange and Kevin T. Kilty	179
Water Temperature Variation in Springs in the Knox Group near
  Oak Ridge, Tennessee
  - Gareth J. Davies	
The Effect of Petroleum Hydrocarbons on the Sorption of
  Fluorescent Dyes by Activated Coconut Charcoal
  - Scott A. Recker, Michael J. Carey and Joe Meiman 	  213
Continuous-Plow Fluorometry in Ground Water Tracing
  - C.C. Smart and S-E. Lauretzen	231
Development of a Flow-Through Filter Fluorometer for use in
  Quantitative Dye Tracing at Mammoth Cave National Park
  - Martin Ryan	243


                                 Session IV:
                    Hydrogeology and Processes Occurring
                              in Karst Aquifers

Stable Isotope Separation of Spring Discharge in a Major Karst
  Spring, Mitchell Plain, Indiana, U.S.A.
  - Barbara L. Gruver and Noel C. Krothe	265
The Transmission of Light Hydrocarbon Contaminants in Limestone
  (Karst) Aquifers
  - Ralph 0. Ewers, Anthony J. Duda, Elizabeth K. Estes,
    Peter J. Idstein and Katherine M. Johnson	287
Velocities of Piezometric Waves Induced by Pumping in Karstic
  Aquifers
  - Claude Drogue	307
Solution Mining and Resultant Evaporite Karst Development in
  Tully Valley, New York
  - Paul A. Rubin, John C. Ayers and Kristin A. Grady	313


                                 Session V:
                               Case Histories

Hydrogeology and Ground Water Monitoring of an Ash Disposal Site at a
  Coal-fired Power Plant in a Karst System
  - Shelley A. Minns, Arsin M. Sahba, Lyle V.A. Sendlein, James C.
    Currens and James S. Dinger	331
Development of a Monitoring Program at a Superfund Site in a
  Karst Terrane near Bloomington, Indiana
  - Michael R. McCann and Noel C. Krothe	349
Heterogeneity in Carbonate Aquifers: Effects of Scale,
  Fissuration, Lithology and Karstification
  - P.L. Smart, A.J. Edwards and S.L. Hobbs	373
                                    VI 1 1

-------
                                    Session V:
                                  Case Histories
                                   (Continued)
                                                                         page

Caustic Waste Contamination of Karstic Limestone Aquifers in
  Two Areas of Jamaica
  - Basil P. Fernandez	389
The Interaction of Plow Mechanics and Aqueous Chemistry in a
  Texas Hill Country Grotto
  - Barbara Mahler and Phillip Bennett	405
The Oronoco Landfill Dye Trace III: Results from a Superfund
  Remedial Investigation in a Glaciated, Diffuse-flow Karat
  - E. Calvin Alexander Jr., Scott C. Alexander, Barbara J.
    Huberty, and James F- Quinlan	417
Deducing Karst Aquifer Recharge, Storage and Transfer Mechanisms
  though Continuous Electronic Monitoring: A Confirmation with
  Tracers
  - Peter J. Idstein and Ralph 0. Ewers	431
Petroleum Hydrocarbon Remediation of the Subcutaneous Zone of a
  Karst Aquifer, Lexington, Kentucky
  - Scott A. Recker	447
Correction of Background Interference and Cross-Fluorescence in
  Filter Fluorometric Analysis of Water-Tracer Dyes
  - C.C. Smart and P.L. Smart	475


                                 Session VI:
                     Ecology of Caves and Karst Terranes

Assessing Ground Water Quality in Caves using Indices of
  Biological Integrity
  - Thomas L. Poulson	495
The Use of Benthic Macroinvertebrates for Assessing the Impact
  of Class V Injection Wells on Carbonate Ground Waters
  - Albert E. Ogden, Ronald K. Redman and Teresa L. Brown 	 513

                                Session VII:
                           Ground Water Monitoring

The Response of Landfill Monitoring Wells in Limestone (Karst)
  Aquifers to Point Sources and Mon Point Sources of Contamination
  - Ralph O. Ewers	529
Development of an ASTM Standard Guide for the Design of Ground Water
  Monitoring Systems in Karst and other Fractured Rock Terranes
  - Michael R. McCann and James F. Quinlan	541
Ground Water Remediation May Be Achievable in some Karst Aquifers
  that are Contaminated, but it Ranges from Unlikely to Impossible
  in Most: I. Implications of Long-Term Tracer Tests for Universal
  Failure in Goal Attainment by Scientists, Consultants, and
  Regulators
  - James F. Quinlan and Joseph A. Ray	553

-------
                                Session VIII:
                           Emergency Response and
                           Ground-Water Management
                                                                       page
The Use of Groundwater-Level Measurements and Dye Tracing to
  Determine the Route of Groundwater Flow from a Hazardous Waste
  Site in an Area of Karst in Hardin County, Kentucky
  C.  Britton Dotson, Nicholas C. Crawford and Mark J. Rigatti.  .  .  .  561
Recommended Administrative/Regulatory Definition of Karst
  Aquifer, Principles for Classification of Carbonate Aquifers,
  Practical Evaluation of Vulnerability of Karst Aquifers,
  and Determination of Optimum Sampling Frequency at Springs
  - James P. Quinlan, Peter L. Smart, Geary M. Schindel,
    E. Calvin Alexander Jr., Alan J. Edwards and A. Richard Smith.  .  . 573
Legal Tools for the Protection of Ground Water in Karst Terranes
  - Gary A. Davis and James F. Quinlan	637

                                 Session IX:
                           Ground Water Management

Karst Geology and Ground Water Protection Law
  - Joseph A. Fischer, Robert J. Canace and Donald H. Monteverde  .  .  . 653
Analysis of DRASTIC and Wellhead Protection Methods Applied to
  a Karst Setting
  - Lyle V.A. Sendlein	669
Land Use Planning and Watershed Protection in Karst Terranes
  - Paul A. Rubin	769
Use of Dyes for Tracing Ground Water: Aspects of Regulation
  - James F. Quinlan	687
The Effects of Recharge Basin Land-Use Practices on Water
  Quality at Mammoth Cave National Park, Kentucky
  - Joe Meiman	697
The Sinkhole Collapse of the Lewiston, Minnesota, Waste Water
 Treatment Facility Lagoon
  - Nancy 0. Jannik, E. Calvin Alexander and Lawrence J. Landherr.  .  . 715
Erosion and Sedimentation Control Methodologies for Construction
  Activities over the Edwards Aquifer in Central Texas
  - Hank B. Smith	725
Case Histories of Several Approaches to Stormwater Management  in
  Urbanized Karst Terrain, Southwest Missouri
  - Wendell L. Earner and Patricia Miller	741
Contamination Investigation in a Karst Region
  - James E. Bentkowski	761

-------
                                 AUTHOR INDEX

                                                                      page

ALEXANDER, E. Calvin Jr.; Scott C. Alexander, Barbara J.  Huberty
  and James F. Quinlan; The Oronoco Landfill Dye Trace III: Results
  from a Superfund Remedial Investigation in a Glaciated, Diffuse-
  flow Karst	417
EARNER, Wendell L. and Patricia Miller; Case Histories of Several
  Approaches to Stormwater Management  in Urbanized Karst Terrain,
  Southwest Missouri 	 741
BARTON, Gary J. and Dennis W. Risser; Approach for Delineating
  the Contributing Areas of a Wellfield in a Carbonate-Valley
  Aquifer	59
BENSON, Richard; Lynn Yuhr and Allen W. Hatheway; Site Character-
  ization for a Landfill Located over a Mine with Superimposed
  Paleo-Karst Collapse: A Case History 	 131
BENTKOWSKI, James E.; Contamination Investigation in a Karst
  Region	761
COOLEY, Tony; Approaches to Hydrogeologic Assessment and
  Remediation of Hydrocarbon Contamination in Clay-Covered Karsts
  with Shallow Water Tables	95
CRAWFORD, Nicholas C. and Ginny L. Gray; Application of Dye-Tracing
  Techniques for Characterizing Groundwater Plow Regimes at the
  Fort Hartford Mine Superfund Site, Olaton, Ohio Co., Kentucky. . .  . 113
DAVIES, Gareth J.;Water Temperature Variation in Springs in the
  Knox Group near Oak Ridge, Tennessee	197
DAVIS, Gary A. and James F. Quinlan; Legal Tools for the Protection
  of Ground Water in Karst Terranes	637
DOTSON, C. Britton; Nicholas C. Crawford and Mark O.Rigatti;
  The Use of Groundwater-Level Measurements and Dye Tracing to
  Determine the Route of Groundwater Flow from a Hazardous Waste
  Site in an Area of Karst in Hardin County, Kentucky	561
DROGUE, C.; Velocities of Piezometric Waves Induced in Pumping
  of Karstic Aquifers	307
EDWARDS, A.J.; S.L. Hobbs and P.L. Smart; Effects of Quarry
  Dewatering on a Karstified Limestone Aquifer: A Case Study
  from the Mendip Hills, England	77
EWERS, Ralph 0.; The Response of Landfill Monitoring Wells in
  Limestone (Karst) Aquifers to Point Sources and Non Point Sources
  of Contamination	529
EWERS, Ralph 0.; Anthony J. Duda, Elizabeht K. Estes, Peter J.
  Idstein and Katherine M. Johnson; The Transmission of Light
  Hydrocarbon Contaminants in Limestone (Karst) Aquifers 	 287
FERNANDEZ, Basil P.; Caustic Waste Contamination of Karstic
  Limestone Aquifers in Two Areas of Jamaica 	 389

-------
                                                                      page


FISCHER/ Joseph A.; Robert J. Canace and Donald H. Moneverde;
  Karst Geology and Ground Water Protection Law	  653
GRUVER, Barbara L. and Noel C. Krothe; Stable Isotope Separation
  of Spring Discharge in a Major Karst Spring, Mitchell Plain,
  Indiana, U.S.A	265
HUNTOON, Peter; Chairman Mao's Great Leap Forward and the
  Deforestation Ecological Disaster in the South China Karst Belt.  .  .  149
IDSTEIN, Peter J.  and Ralph 0. Ewers; Deducing Karst Aquifer
  Recharge, Storage and Transfer Mechanisms though Continuous
  Electronic Monitoring: A Confirmation with Tracers 	  431
JANNIK, Nancy 0.;  E. Calvin Alexander and Lawrence J. Landherr;
  The Sinkhole Collapse of the Lewiston, Minnesota, Waste Water
  Treatment Facility Lagoon	715
KILTY, K.T. and A.L. Lange; Electrochemistry of Natural Potential
  Processes in Karst	163
LANGE, Arthur L. and Kevin T. Kilty; Natural-Potential Responses
  of Karst Systems at the Ground Surface 	  179
HAULER, Barbara and Phillip Bennett; The Interaction of Flow
  Mechanics and Aqueous Chemistry in a Texas Hill Country Grotto  .  .  .  405
McCAMN, Michael R. and Noel C. Krothe; Development of a Monitoring
  Program at a Superfund Site in a Karst Terrane near Bloomington,
  Indiana	541
McCAMN, Michael R. and James F. Quinlan; Development of an ASTM
  Standard Guide for the Design of Ground Water Monitoring Systems
  in Karst and other Fractured Rock Terranes 	  349
MEIMAN, Joe; The Effects of Recharge Basin Land-Use Practices
  on Water Quality at Mammoth Cave National Park, Kentucky 	  697
MINNS, Shelley A.; Arsin M. Sahba, Lyle V.A. Sendlein, James C.
  Currens and James S. Dinger; Hydrogeology and Ground-Water
  Monitoring of an Ash Disposal Site at a Coal-fired Power
  Plant in a Karst System	331
06DEN, Albert E.;  Ronald K. Redman and Teresa L. Brown; The Use
  of Benthic Macroinvertebrates for Assessing the Impact of Class V
  Injection Wells on Carbonate Ground Waters 	  513
POULSON, Thomas L.; Assessing Ground Water Quality in Caves using
  Indices of Biological Integrity	495
QUINLAN, James F.; Use of Dyes for Tracing Groundwater:
  Aspects of Regulation	687
QUINLAN, James F.  and Joseph A. Ray; Groundwater Remediation May
  Be Achievable in some Karst Aquifers that are Contaminated,
  but it Ranges from Unlikely to Impossible in Most: I. Implications
  of Long-Term Tracer Tests for Universal Failure in Goal
  Attainment by Scientists, Consultants, and Regulators	553
QUINLAN, James F.; Peter L. Smart, Geary M. Schindel, E. Calvin
  Alexander Jr., Alan J. Edwards and A. Richard Smith; Recommended
  Administrative/Regulatory Definition of Karst Aquifer, Principles
  for Classification of Carbonate Aquifers, Practical Evaluation
  of Vulnerability of Karst Aquifers, and Determination of Optimum
  Sampling Frequency at Springs	573
                                XI 1

-------
                                                                       page

RECKER, Scott A.; Petroleum Hydrocarbon Remediation of the
  Subcutaneous Zone of a Karst Aquifer, Lexington, Kentucky	 447
RECKER, Scott A.; Michael J. Carey and Joe Meiman; The Effect of
  Petroleum Hydrocarbons on the Sorption of Fluorescent Dyes
  by Activated Coconut Charcoal	213
RUBIN, Paul A.; Land Use Planning and Watershed Protection in
  Karst Terranes	769
RUBIN, Paul A.; John C. Ayers and Kristin A. Grady; Solution
  Mining and Resultant Evaporite Karst Development in Tully Valley,
  New York	313
RYAN, Martin; Development of a Flow-Through Filter Fluorometer
  for use in Quantitative Dye Tracing at Mammoth Cave National Park.  . 243
SAUTER, Martin; Assessment of Hydraulic Conductivity in a Karst
  Aquifer at Local and Regional Scale	39
SENDLEIN, Lyle V.A.; Analysis of DRASTIC and Wellhead Protection
  Methods Applied to a Karst Setting 	 669
SMART, C.C. and S-E. Lauretzen; Continuous-Flow Fluorometry
  in Groundwater Tracing 	 231
SMART, C.C. and P.L. Smart; Correction of Background Interference
  and Cross-Fluorescence in Filter Fluorometric Analysis of
  Water-Tracer Dyes	475
SMART, P.L.; A.J. Edwards and P.L. Smart; Heterogeneity in
  Carbonate Aquifers: Effects of Scale, Fissuration, Lithology
  and Karstification	373
SMITH, Hank B.; Erosion and Sedimentation Control Methodologies for
  Construction Activities over the Edwards Aquifer in Central Texas.  . 725
TEUTSCH, Georg and Martin Sauter; Groundwater Modeling in Karst
  Terranes: Scale Effects, Data Acquisition and Field Validation ...  17
WILCHBR, LaJuana S.; Protecting Ground-Water in Karst Terranes:
  An EPA Priority for Action	   3
                               XI 1 1

-------
             Keynote Address:
Protecting Ground-Water in  Karst
              Terranes:
   An  EPA  Priority for Action
             Lajuana Wilcher,
           Assistant Administrator,
          Office of Water, U.S. EPA

-------

-------
                       KEYNOTE ADDRESS
         PROTECTING GROUND WATER IN KARST TERRANESi
                 AN EPA PRIORITY FOR ACTION

                     LaJuana S. Wilcher
             Assistant Administrator for Water
            U.S. Environmental Protection Agency
                      Washington, D.C.
ABSTRACT
     Karst ground waters are unusually sensitive to
contamination because they have rapid flow rates, low
attenuation of contaminants, and large recharge areas that
could contribute contaminants to a receptor.  Caves and
karst areas contain unusual and unique organisms whose
habitats could be degraded before these organisms and their
environments can be fully characterized.  Each EPA program
is increasingly recognizing the importance of pollution
prevention as the prescription for protecting karst ground
water.  This conference is an expression of EPA's attention
to the uniqueness of karst ground water resources and to the
ecological consequences of failing to properly manage those
resources.  EPA has multiple ongoing programs and procedures
to protect karst ground water, including:  1) Prioritizing
karst as highly sensitive to contamination under EPA and
State Comprehensive Ground-Water Protection Programs, 2)
Detailed characterization, monitoring and analysis
requirements for siting RCRA Hazardous Waste Treatment
Storage and Disposal Facilities, 3) Recommended protocols
for field surveys, ground water monitoring, such as for
spring location, 4) Methods for and results of delineating
Wellhead and Springhead Protection areas in Missouri and
Kentucky, and 5) Health assessment of dyes used for
tracing.
TEXT OF ADDRESS

     It's a pleasure to join you for The Third Conference On
Hydrology, Ecology, Monitoring and Management Of Ground
Water In Karst Terranes.
     In fact, it is a dual pleasure, personal and

-------
professional.  On the personal side, my roots go deep into
the karst terrane of south-central Kentucky.  From the time
I was two years old, family outings to Mammoth Cave to see
its wonders were a special treat.  Add to that school field
trips, Girl Scout camping trips, my days as a National Park
Service naturalist and cave guide, and my ongoing
involvement with the Cave Research Foundation, and you can
understand my enthusiasm for the work you're doing.
     It's no wonder that I am pleased to be here, for I have
been lucky enough to grow up among limestone outcroppings,
to watch the River Styx flow from the hillside at Mammoth
Cave down to the Green River, and sometimes back again, to
rappel 60 feet into the earth, and then tramp, crawl and
wade for two hours through the cave's dark passageways, to
happen upon an undisturbed pool, and discover in it the
magic of an eyeless, translucent, troglobytic crayfish.
     Yet, too long we have abused these important parts of
our ecosystem, at real damage to the long-term
sustainability of the people and the "critters" that share
this planet.  So I am pleased to participate in a conference
that seeks to learn and share information about karst
terranes.
     Professionally, it's a pleasure to represent the U.S.
Environmental Protection Agency as one of the sponsors of
this conference, along with the National Ground Water
Association.  As you may know, the National Ground Water
Association, formerly the National Water Well Association,
is a well-established organization with a new name.  It's a
powerful thing, a group's name.  It's a beacon to all the
world of your goals and aims.  This name change reflects the
growing understanding in the U.S. of the importance of our
sub-surface water resources.  It suggests a logical and
holistic approach to the development and protection of our
ground water.  We look to the National Ground Water
Association to continue to provide leadership as we move
beyond a piecemeal, narrow view of the world to one in which
we recognize connections and ecosystems, and focus on
relationships-relationships between human activity and the
quality of our environment, between surface water and ground
water, and among all of the inhabitants of an ecosystem.  We
must work on behalf on those inhabitants whether they walk
on two legs or four; whether they fly, or swim, or slither
above or under the ground.

Ground Water and Karst Terranes-A Unique Link

     I'm probably preaching to the choir to tell you that
even though less than one per cent of the Earth's water is
ground water, it plays a disproportionately important role
in our lives.  In this country, about one-fourth of all
fresh water we use comes from the ground.  It supplies half
of our drinking water, enough for 125,000,000 thirsty
people.  In eight states, more than 90 per cent of the
population depends on ground water for potable and

-------
agricultural use.  Many Eastern states use nearly a third of
their ground water for industry.  In my home state of
Kentucky, industry uses ground water for nearly 60 per cent
of its needs.
     With that as background, it's worth asking why we as a
nation don't know more, understand more or care more about
ground water, and about the karst in which so much of it
lies.  After all, the characteristics of karst are easily
observed, and have been recognized for centuries.
     Early Europeans recognized the unique characteristics
of karst.  Their name for it meant "bare, rocky place."  The
Germanic word we use comes from the name for the Carso
region of Italy and Slovenia, which is blessed with a good
deal of soluble carbonate rock.
     Karst aquifers underlie about 20 per cent of the
continental United States.  Karst is found in almost two
fifths of the Eastern and Southern U.S., and in much of the
American Southwest.  And in the West, pseudo-karst in some
lava flows mimics many karst characteristics.
     The mechanisms by which karst terranes are sculpted has
been pretty well understood for well over 100 years in this
country.  In 1870, Ralph Seymour Thompson described his
visit to Mammoth Cave in a delightful account, The Sucker's
Visit To The Mammoth Cave ("sucker", by the way, was his
friendly term for a resident of rural Illinois.)  Thompson
wrote: "...a thousand, or a million, or a hundred million,
[years ago] as you please—a little stream of water
penetrated the soil somewhere in Kentucky, [and] wound its
way along through the crevices of the rock... Ages roll
away, and still this little stream follows its
course... through limestone...dissolving away the rock
through which it passes... Slow work it is true; but He to
whom a thousand years are but as yesterday when it is
passed, has plenty of time. ...[the spring] becomes a great
river, with branches in every direction, and draining a
great scope of country...."
     While some understanding of karst terranes has been
around a long time, our behavior in protecting it has lagged
far behind our knowledge.  We've made dramatic improvements
over the last 20 years in improving surface water quality-
Why haven't those advances been matched by protection and
improvements of our karst aquifers?  The reason, I suppose,
has to do with human nature.  The Middle-Ages philosopher
Thomas a1 Kempis said it best: "and when he is out of sight,
quickly also is he out of mind."
     It is like the terrible famines and natural disasters
on the other side of the world.  Unless we happen across a
news story or a TV special, we "tune them out."  Karst has
been "tuned out" for years.  But Nature has been producing
several "Specials" that are getting our attention.

The Consequences of Abuse

     When Nature puts on a drama, her stage is impressive.

-------
We have seen the land actively rebel against our uses and
our abuses of karst terrane.  For example, Bowling Green,
Kentucky won dubious fame in a 1921 article in Popular
Mechanics magazine as "the city with a million year old
sewer system."  The "system," of course, was an open drain
into sinkholes.  Out of sight, out of mind.  And it wasn't
just sewage.  Since 1969, chemical fumes from underground
were known to drift into homes there.  Ten years ago, five
houses had to be evacuated because of gasoline fumes rising
from caves into their basements.  The next year the Lost
River Cave beneath Bowling Green was filled with fumes.
Cave explorers found benzene and methylene chloride in the
underground river and traced them to a chemical mixing
facility with leaking underground storage tanks.  Bowling
Green has had two Superfund emergencies in the past six
years relating to explosive or corrosive cave fumes.
     And it's not just Bowling Green.  A sewage treatment
plant in West Plains, Missouri leaked 18 million gallons of
effluent into the karst aquifer in May, 1978.  The result-at
least 759 cases of viral gastroenteritis.
     Just three years ago, in October, 1988, four cavers
were preparing to descend a passage in Hicks Cave, near
Horse Cave, Kentucky.  Suddenly a ball of blue flame
appeared before them and exploded.  The cave ceiling boiled
with flame.  The cavers ran for their lives, and one was
seriously burned.  The cause—leakage from a propane gas
refueling station half a mile away!
     In Puerto Rico in the 1970s, three underground chemical
storage tanks ruptured, dumping more than 15,000 gallons of
very hazardous chemicals into the karst aquifer below.  And
that was just the "tip of the iceberg."  Pumping to recover
the chemicals sucked up over twice as much as had been lost
from the rupture.  The combined releases to the shallow
aquifer forced authorities to deny its use for drinking
water.
     A Puerto Rican landfill, in use for 20 years, has been
found to actually be in a large sinkhole.  Everything dumped
into it has been immediately transported to ground water.
Since the landfill is upgradient of most water wells in the
area, it is likely that residents have been drinking their
own waste for years.
     The caves, windows, sinkholes, fissures and fractures
of karst are among nature's most intriguing geological
creations.  They form a complex puzzle, challenging those of
us working to map it, or clean it up.  Discrete flow paths
and the changes in flow caused by storms makes it very
difficult to pinpoint ground water in karst, let alone a
slug of contamination in it.
     Because of these problems, it is especially important
that the Agency embrace a new way of doing business in karst
terranes; a new direction.  As United States Supreme Court
Justice Oliver Wendell Holmes said, "I find the great thing
in this world is not so much where we stand, as in what
direction we are moving:...we must sail sometimes with the

-------
wind and sometimes against  it—but we must sail, and not
drift, nor  lie at anchor."
     When I think of directions, I remember the story of an
old man, travelling by train.  The conductor came by,
collecting  tickets.  The old man rummaged through his
pockets but couldn't find his ticket, so the conductor
suggested that he mail it in later.  The man looked at him
-sternly and said, "Sir, the question is not where is my
ticket.  The question is, where am I going?"  The question
for us is, where is EPA going with regard to protecting
karst?

EPA's Role  in Karst Protection

     For a  number of years, EPA's direction sometimes
conflicted with those of the states concerning ground water
and karst protection.  Today I'm happy to say that EPA's
Office of Water, through its Ground Water Division, is
providing leadership in charting our course toward new
policies and more focussed  attention upon karst protection.
The help and the leadership of people like Phil Berger and
Ron Hoffer  in EPA's Office  of Ground Water and Drinking
Water, and  Ron Mikulak in our Regional Office in Atlanta, is
getting us moving in the right direction.
     Our efforts signal a new, comprehensive approach to
environmental protection.   Through various program offices,
EPA administers five different statutes containing ground
water protection language:  the Clean Water Act, the Safe
Drinking Water Act, the Resource Conservation and Recovery
Act, the Comprehensive Environmental Response, Compensation
and Liability Act  (Superfund), and the Federal Insecticide,
Fungicide and Rodenticide Act.  Every one of the Agency's
program offices, from pesticides to hazardous wastes, is
working together on a coordinated approach to ground water
protection.  The Office of  Water is pleased to be leading
this innovative effort.
     EPA Administrator Bill Reilly recognized the importance
of such efforts recently when he released EPA's new Ground
Water Strategy for the 1990s.  The Report re-states our
strong belief that states are in the best position to
protect their own ground water, and it sets a new policy
direction for the Agency to work closely with the states to
develop and implement Comprehensive Ground Water Protection
Programs.  We want to ensure that karst has the highest
priority in those state programs.  EPA will help states fill
any gaps they might find, in order to ensure comprehensive
ground water protection.  Our staff is working with every
state and national organization representing state managers
involved in ground water issues.  We intend to produce
national guidance next year on Comprehensive State Ground
Water Programs.
     Our Wellhead Protection Program includes technical
mapping and model protective ordinances.  We are actively
working on ways to help states designate wellhead protection

-------
areas in karst.  In fact, we have two case studies, one in
Missouri and one in Kentucky, that will serve as models.
     Our Safe Drinking Water Act Surface Water Treatment
Rule and forthcoming Ground Water Disinfection Rule will
have significant implications in karst as the public water
supply systems comply with setbacks and separation distances
between rural wells and septic tanks.  We need you to work
with EPA to help us understand what the rule should provide
for the special conditions in karst.  The Office of Water is
promoting evaluation of "Class Five" or, shallow,
Underground Injection Wells, to minimize effects on karst
aquifers.  Septic tanks, agricultural and storm water
drains, including "improved" sinkholes, should all be
evaluated.  EPA has been pleased to fund the work of Al
Ogden of the Tennessee Technological University and his
colleagues on the effects of these wells on ground and
surface water quality in karst areas.
     But we have a lot of catching up to do.  EPA's Office
of Solid Waste and Emergency Response has listed an
incredible 160 hazardous waste treatment, storage and
disposal facilities on karst terranes in the United States!
As you well know, hazardous waste spills into karst are a
remedial nightmare, and complete remediation is often not
possible at all, especially considering the effects of the
subterranean eco-system.  Contaminant flows of 25 miles in
12 days have been documented.  And the costs of attempted
clean-ups can be astronomical.  EPA estimates the cost of
remediating a hazardous waste spill in karst at between
$7,000,000 and $1.2 billion per incident, not including the
environmental costs of land lost or habitat damaged.  EPA is
now developing a proposed rule mandating strict siting
criteria for hazardous waste treatment, storage and disposal
facilities in karst.  That rule should be out sometime next
Spring.  In the meantime, as of 1987, nine states had acted
on their own to ban such facilities in karst, two states had
mandatory setback distances from karst areas and one state,
Kentucky, required a double-liner and leak-detection system
or a demonstration that karst features were sealed and
filled.  24 states have siting criteria that may restrict
such facilities.
     In addition, EPA's new municipal landfill rule, issued
just two months ago, creates the first meaningful national
solid waste landfill siting standards.  As of October, 1993,
developers must demonstrate engineering standards to
maintain the structural integrity of a facility sited in an
area of disruption.  The methods may include re-routing
surface water and a demonstration that karst features are
sealed and filled.
     But our regulations are only as good as our science.
EPA's Office of Research and Development has assessed the
risks and benefits of using ground water dyes as tracers so
that we can predict the fate and transport of hazardous
wastes leaked into karst ground water at Superfund sites.
It now appears that many dyes can be used safely if they are

-------
used properly.  Our final risk/benefit study may allay the
concerns some states have about dye-tracers.  The work of
EPA scientists Malcolm Field, Ron Wilhelm and Charles Auer
will be presented at this conference, to give you a more
detailed picture of what we're doing in this regard.
     And there's more.  EPA's Office of Pesticides and Toxic
Substances is promoting State Management Plans for
pesticides, tailored to differing local pesticide use and
ground water vulnerability.  Increased monitoring will be
reguired in areas such as karst which are particularly
susceptible to contamination.
     And Congress is getting the message that ground water
in karst areas needs special protection.  Late last year,
Congress passed the Food, Agriculture, Conservation and
Trade Act of 1990, the Farm Bill, which included a new
program to provide assistance to farmers to implement on-
farm water quality protection plans.  The areas of attention
include those "in shallow karst topography areas where
sinkholes convey runoff water directly into groundwater."
Next year, when Congress again debates re-authorization of
the Resource Conservation and Recovery Act, it will have on
the table a bill, introduced by Senator Max Baucus of
Montana, which includes proximity to karst terrane as a
criterion when siting landfills.
     While we are making great strides, we have much more to
do.

A Public Role in Karst Protection

     Government has an important role in ground water
protection, but government alone is not going to do the job.
We need to enlist everyone in this battle.  EPA has chosen
to do that by promoting preventive and corrective actions
before the environment is damaged.  Pollution prevention is
especially vital in karst terranes in order to prevent
contamination in the first place, rather than trying to
clean it up later.
     In the case of karst, pollution prevention means
wellhead protection programs and comprehensive ground water
protection plans.  It means identifying potential point and
nonpoint pollution sources, and it means setting priorities
for addressing those pollution sources.  It means geologic
and hydrologic mapping and water quality characterization.
     But to implement pollution prevention effectively, we
must reach out to people and businesses about causes and
effects—about environmental stewardship.  At the bottom
line, those who generate pollution—and that means every one
of us-must take action to keep it in check.  People need to
know about the connections between ground and surface water,
especially in karst terranes.  We have to get the word out
that people who dump into sinkholes might just as well dump
into the nearest river.
     As part of that process we must learn more and educate
the public more about the non-human subterranean

-------
environment.  We know that caves and karst areas contain
unusual and unique organisms-things like the Alabama
Cavefish, the Kentucky Cave Shrimp and the Comanche Springs
Pupfish.  Their habitats continue to be needlessly
contaminated, wiping out populations and whole species.

Our Ecological Responsibility

     We need to act on the advice of EPA's Science Advisory
Board last year, when it said that we should "...attach as
much importance to reducing ecological risk as ...[we do] to
reducing human health risk, because productive natural eco-
systems are essential to human health and to sustainable,
long-term economic growth, and because they are
intrinsically valuable in their own right..."
     While subterranean critters are valuable in their own
right, we also know that they can help us by showing us the
health of ground water.  Our holistic approach to improving
water quality has led us to begin developing biological
indicators of the health of surface water, and we must
develop similar indicators to measure the health of ground
water.  And, as we learn, we need to do our
characterizations and evaluations with the least possible
intrusion, so that other life-forms are as little affected
by human activities as possible.
     To promote understanding of the eco-system and
development of biological ground water indicators, we are
sponsoring the First International Conference on Ground
Water Ecology, April 27th through the 29th, 1992, in Tampa,
Florida.  We intend it to be the other must-do ground water
event on your calendar.

The Challenge

     Our job, and our challenge, is to learn, to teach, and
to protect.  As the African environmentalist Baba Dioum so
eloquently said, "For in the end, we will conserve only what
we love, we will love only what we understand.  We will
understand only what we are taught."
     Our task is not an easy one, but we must not fail!  I
assure you that we in EPA have a clear sense of the
importance of our goals, and that we will be working right
along with you to expand the volume of our knowledge.  I
wish you an exciting learning experience here at the
conference, and success in the important work you have ahead
of you.
BIOGRAPHICAL SUMMARY

     LaJuana S. Wilcher, the Environmental Protection
Agency's Assistant Administrator for the Office of Water, is
responsible for planning, policy and national program
management of EPA's water programs.  Under her leadership,


                              10

-------
the Office of Water has intensified its ecosystem protection
efforts, complementing its human health protection
activities.  Wilcher emphasizes wetlands and aquatic habitat
protection, a stronger scientific basis for regulatory
decision making, pollution prevention, and geographic
targeting.  Prior to her appointment as Assistant
Administrator, Wilcher was a partner in a 90-lawyer
Washington, B.C. law firm.  She specialized in environmental
litigation, legal counseling, and regulatory interpretation
of water, air, hazardous waste, and resource recovery
issues.  Her background includes extensive legal and federal
government experience and work as a naturalist and cave
guide at Mammoth Cave, Kentucky.  Wilcher holds a B.S. in
biology  (magna cum laude) and a Juris Doctorate.
                               11

-------
QUESTION:

1.   If we are protecting aquatic life in karst aquifers,
should we apply ambient water quality criteria (AWQC) as
ground water protection levels instead of maximum
contaminant levels (MCLs)?    {AWQC are to protect aquatic
life.  MCLs are to protect human life.}

ANSWER:

     EPA's aquatic life criteria methodology is based on
demonstrated adverse effects of particular toxicants on a
wide spectrum of aquatic plants and animals, including
sensitive fish species.  Therefore, in order to protect
aquatic life no matter where located, e.g., above ground or
below ground, EPA's recommended aquatic life criteria are
the criteria of choice.

     However, if both aquatic life beneficial uses and
public water supply uses are to be protected, the more
stringent of these criteria should be applied.  As a general
rule, the aquatic life criteria are more stringent than MCLs
for non-carcinogens.  For carcinogens, MCLs are more
stringent.

     Please address questions regarding which criteria apply
as a matter of State or Federal law to the appropriate State
Agency or to EPA's Office of Water.
QUESTION:

2.   In the light of unique problems associated with karst
areas and the poor performance record of most regulatory
agencies in recognizing them and dealing with many of them
as judiciously as might be desired, would it not be
desirable for EPA to set up a specific program to review
environmental problems in karst terranes?  Such a program
could include procedures for or assistance with
identification, evaluation, and remediation of karst-related
problems.  If not, why not?

ANSWER:

     While there is no question that the unique nature of
karst and the problems associated with it merit special
attention, it does not necessarily follow that new programs
must be established in order to attract that attention.
EPA's Office of Water, through its Ground Water Division, is
charting a course toward new policies and more focussed
                              12

-------
attention upon ground water protection, including that in
karst terranes.  Every one of the Agency's program offices,
from pesticides to hazardous wastes, is working together on
a coordinated approach to ground water protection.  The
Office of Water is pleased to be leading this innovative
effort.  It is true that we must do more to understand and
protect karst terranes, and EPA intends to be in the
forefront of such efforts.
                               13

-------
14

-------
                   Invited Lecture:


     Ground Water Modeling in  Karst

Terranes: Scale Effects, Data Acquisition

            and  Field  Validation

                   *Georg Teutsch,
     Institut fur Wasserbua, Universitat Stuttgart, Germany
                  and Martin Sauter,
                 Universitat Tubingen

-------
16

-------
                  GROUNDWATER MODELING IN KARST TERRANES:
           SCALE EFFECTS, DATA ACQUISITION AND FIELD VALIDATION
                                     Georg Teutsch
                                  Institut fur Wasserbau,
                              University of Stuttgart, F.R.G.

                                          and

                                      Martin Sauter
                                 Geological Department,
                             University of Tubingen, F.R.G.
ABSTRACT

       Karstified limestone aquifers are often a very prolific source of groundwater and have
therefore been studied in many countries to a great extent. In recent years the focus of many
studies has switched from 'water  resources development aspects' towards a  'water resources
protection' point of view. Consequently, the demand for quantitative contamination assessment
has increased. In  contrast, groundwater flow  in  karst terranes is  still considered  hardly
quantifiable. Predictions, based on classical hydrogeological investigation methods seem to  fail
quite frequently; similarly most attempts to employ standard mathematical models did not prove
satisfactory so far.

       This paper presents a recently developed concept for the consistent  classification of
groundwater flow problems in karst terranes which proved helpful in understanding some of the
'unexpected' hydraulic phenomena observed. This concept is based on a scale hierarchy approach
originally used to  describe  highly heterogeneous porous formations. Within  this  concept  the
spatial averaging properties of different investigation methods like pumping tests, tracer tests and
spring flow measurements are related to the size of the prevailing heterogeneities and the size of
the investigated domain. Similarly,  the applicability of various proposed modeling approaches is
analysed, taking into account the type and matureness of the respective karst system.

       To demonstrate its practical application a case study from the Swabian Alb limestone karst
aquifer  in Southern Germany is  briefly presented.  In this classical  karst  research  area a
considerable amount of hydrological, hydraulic and tracer test data has been gathered over the
past 10 years and analysed using single- and double-continuum flow and  transport  models of
different complexity.
INTRODUCTION

       In general,  the  effective primary  porosity of  carbonate  rocks  is considered  to  be
negligable. Consequently, in karst terranes the effective porosity is a result of a combination of
secondary porosity fissures and also a tertiary porosity in areas where the fissure system has been
enlarged to conduits by chemical dissolution. As a result, groundwater flow in karst aquifers may
show in some areas  a diffuse-type and in other areas a more conduit-type flow characteristic.
This is confirmed from field observations where hydraulic conductivity investigations conducted
at adjacent locations may produce very different results. Similarly, measurements from the same
                                             17

-------
location (e.g. one borehole) but taken at different scales (volumes of integration) may yield very
different hydraulic conductivities.

       In order to develop a more consistent understanding of karst terranes and their variability,
various attempts were made  employing classification schemes which  are based on  (mainly
qualitative) descriptions of the various flow phenomena observed in the field (e.g. White 1969,
1977; Thrailkill, 1986). In  a more recent study, Smart and Hobbs (1986) developed a scheme
where also the recharge distribution and the storage properties of the system are included.
Although very useful in providing a conceptual framework  for the description of various
observations made in karst areas, there still remains a need for a more quantitative analysis of the
field data commonly available. This quantitative flow analysis may be based either on physical
models (e.g. Snow, 1965; Irmay, 1964; Romm, 1966; Long et al. 1985; Smith et al. 1987, Teutsch,
1988; Kraemer and Haitjema, 1989) or a transfer function approach (e.g. Duffy and Harrison,
1987; Aiguang et al., 1988;  Simpson, 1988; Dreiss, 1989a, 1989b).

       In general the physical model is preferrable to the transfer functions (black box) approach
which commonly describes only  a single-parameter input-output relationship. Due to the non-
physical basis, these transfer functions have only very limited predictive properties. On the other
hand, physical models may imply considerable parameter identification problems when applied
to practical field conditions in karst terranes, where hydraulic measurements are usually scarce
and highly variable.

       It should be noted, that in karst areas the major parameter identification problem results
from the fact that data is  not available at the scale of the flow dominating heterogeneities, i.e. at
the scale of the main  conduits. Up til now, there is no standard investigation method available
which permits the reliable  detection of the major conduits within a given catchment area. The
chance to encounter a conduit in a borehole is extremly small if  no surface indication like a
sinkhole or a dry valley is available. Bearing in mind that in a natural karst system the hydraulic
conductivity contrast between a hardly karstified zone and a fully developed conduit system
could be as high as 5 to 7  orders of magnitude, the attempt to try and quantify groundwater flow
in a karst catchment appears almost impossible at first.

       On the other hand, due to the high risk of groundwater contamination in karst areas there
is a growing demand for  reliable protection measures. This has lead to a series of studies which
were  initiated in 1981 at the University of Tubingen (Behringer, 1988; Teutsch, 1988; Merkel,
1991; Sauter, 1990; Sauter, 1991 - this issue) and also at the  University of Stuttgart (Teutsch,
1989,  1990b; Lang et al.  199la, 1991b). The aim  of these  studies is the development of a
consistent methodology for the investigation and quantification of karst flow systems including
physical modeling. The area of investigation is the Upper Jurassic limestone aquifer of  the
Swabian Alb. The  aquifer is by far the most prolific source of groundwater in South Germany
and is extensively used for  the municipal water supply in that region.

       Figure 1 shows the Swabian Alb outcrop with the various regions selected for detailed
investigations. Region 2,  3, 4 and part of region 5 belong to the socalled  'Deep Karst' whereas
region 1 and part of region  5 belong to what is called the 'Shallow Karst'. The term 'Deep Karst'
refers to the area of the aquifer which drains to the river Danube.  There the piezometric levels
do not reach the aquifer bottom, the piezometric surface being controlled by the topographical
level  of the surface streams. In  the  'Shallow Karst'  the original drainage pattern to the river
Danube has been deviated during the Quaternary as a result of the erosional processes caused by
the river Rhine tributary system. There the aquifer piezometric  levels are controlled by  the
position of springs which developed at  the aquifer base just above the underlying aquiclude.
Because of different aquifer histories the observed flow characteristics are quite different in the
'Deep Karst' and in the 'Shallow Karst'. It is generally believed that in the 'Deep Karst' a diffuse-
type system dominates the  flow, whereas in the 'Shallow Karst' the original diffuse-type flow
system has matured to a conduit-type flow system during the  recent geological past.

Applying the scale hierarchy concept presented below, the range of  field findings obtained from
different investigation methods is discussed.
                                             18

-------
                                                                 Region 5
                                                                 Region 2
                                                                  legion 3
                                     0 10 20 30 40 50km
                     Figure 1:
Swabian Alb - Areas of Investigation
THE SCALE HIERARCHY CONCEPT

       The scale hierarchy concept as developed by Haldorsen (1986) proved to be extremly
useful  in the analysis of flow and transport problems in heterogeneous formations (Teutsch,
1990b; Teutsch et al., 1990). This paper describes its extention to the analysis of fractured and
karstified terranes where a scale dependent heterogeneity is commonly observed.

       Employing the terminology introduced by Dagan (1986),  three different scales are to be
considered:

       a) L, the length-scale of the flow or transport domain

       b) I, the length-scale of the flow dominating heterogeneities

       c) D, the length-scale (averaging scale) of the detection method used

       Figure 2 shows a schematic example of a heterogeneous system with the I, L and D length
scales.  The example might be viewed as a simplified cross-section through a fractured  or
karstified aquifer system, where L could be the size of the catchment, I the average length of the
fissures and  D for  example  the size of the  drawdown  cone during  a pumping test. The
heterogeneity of the system is extremly high at the scale of an individual fissure but certainly far
less visible at the larger scale D of the pumping test. Therefore, any movement of the detection
window D, i.e. conducting the pumping test at  different  locations, would lead to 'smoothly'
varying parameters but not to an abrupt change in the values. Consequently, due to the size of the
detection window one would not  expect to locate individual fissures or conduits. On the other
hand, any more selective, i.e. a smaller scale, detection method like for example a borehole
flowmeter-log or a point-source artificial tracer test would produce a much higher parameter
variability.
                                             19

-------
       For our example the scale hierarchy may be described as I „,_< L. Most conveniently,
it can be shown that many flow problems may be consistently classified on the basis of the scale
hierarchies only, i.e. I, L and D ratios without the need for absolute figures. From the above it
is quite clear,  that a laboratory analysis of micro-fissures at core scale may yield the same
variability of results as a catchment scale investigation dealing with major conduits, possibly at
the kilometer scale.
                                                                     I < D < L
              Figure 2:      Schematic example of length scale hierarchy (I: heterogeneity-, D:
                            detection-, L: domain-scale)
       The scale hierarchy concept as described above can be used to analyse various hydraulic
phenomena observed in heterogeneous systems like karst terranes. In the following paragraph a
few examples of commonly observed karst aquifer features are discussed applying the framework
of the scale hierarchy concept.
TYPICAL FIELD OBSERVATIONS

       The field examples presented below are grouped into hydraulic phenomena ranging from
regional to local scale and transport phenomena based on areal and point source input.


Hydraulic Phenomena

a) Large scale hydraulic tests

       In region 2 a large scale pumping test was conducted in 1982. Region 2 is located in the
'Deep Karst'  north of Heidenheim. Due to a limited shallow karstification, the vast majority of
the groundwater flow is believed to occur in a network of fairly narrow fissures, continuous at
a scale larger than the equivalent elementary volume, probably in the range of 100 meters. The
log-linear  distance-drawdown (steady-state)  relationship as observed  at  the end of  the
pumpingtest shows a straight line response of the karst aquifer wells. This reflects homogeneous
subsurface conditions at the given detection scale.  This finding  was further supported  in a
subsequent regional modeling study where ground water levels from 26 observation wells were
adequately simulated employing a single-continuum porous equivalent model. The scale hierarchy
for this problem would be I « D <= L.
b) Spring flow hydrographs

       Figure 3 shows the hydrograph from the Echazquelle in region 1 which is located in the
'Shallow Karst' where flow is believed to be dominated by large conduits. Typically for a karst
spring in that area, it shows a high Q^Q^ ratio and at the same time a sustained recession of
about 6 months during the summer.  This observation is inconsistent with a simple single-
continuum concept. A conduit-type of flow would explain the high Q^c/C^ ratio, but could not
explain the long recession and vice versa. It  is important to realize that spring hydrographs reflect
                                           20

-------
the response of the entire flow system to areal recharge events, i.e. they describe the spatially
averaged unsteady-state characteristics of the aquifer at catchment scale. The scale hierarchy for
this flow problem may be described as I <= D = L, i.e. the system heterogeneity is visible at the
detection scale. However, because the detection scale is equivalent to the catchment scale, the
distribution of the hydraulic conductivity within the catchment is not detectable.
                              ECHAZQUELLEN
o I M I o | J 1 F I M I A I tv/i I j I j  I A I
 19B2                13B3
                                                    Id IM iu
                                                               IF I rvi I A I N/I I
              Figure 3:      Echazquelle hydrograph for the period Nov.  1982 to May  1984
                             (Teutsch, 1988)
c) Groundwater level measurements

       The analysis of piezometric levels with respect to their spatial distribution and temporal
variation provides important information on the aquifer characteristics. This is especially true for
any heterogeneous system like a karst aquifer. Figure 4 shows the average piezometric levels as
measured in region 5 which is located partly in the 'Shallow Karst' and partly in the 'Deep Karst'.
The map appears fairly structured with the major (european) water divide extending in NE-SW
direction. Surprisingly, the location of the major spring outflow regions in the 'Shallow Karst' can
be clearly identified on the piezometric map. In these areas the groundwater flow is concentrated
(conduit-system), leading to a convex shape of the equipotential lines. This example shows that
groundwater level readings, even though being taken at single points (boreholes), may  represent
an area much  larger than the actual observation  point. In  other words, the piezometric surface
varies only gently between individual observation locations. Therefore, the hydraulic effect of
single conduits is far reaching and may be observable without having to tap them directly by
boreholes. The scale hierarchy for this problem is therefore I < D <  L, where I represents a
combination of small fissures and large conduits.

       Another way to look at groundwater levels is to analyse their temporal variation. Figure
5 shows the groundwater levels versus time as observed in  one of the boreholes within region 5.
The first part of the curve is based on weekly readings only. After the 4* of August 1988, an
automatic  recorder system  was  installed  in  the  borehole.  The  resulting  change  in  the
characteristics of the water level record is dramatic. The high time resolution of the automatic
                                              21

-------
recording  reveals  the high frequency components on  top of the smooth seasonal water level
variations. This 'dual-frequency' feature can be attributed to the 'dual-porosity' characteritsic of
the karst system in that area. This example demonstrates clearly  the importance of the time
domain resolution as compared to the spatial resolution discussed above.
           3D-Double-Porsity-Model
                  Magentalquelle
              Miihlbrunnenquelle
                                                                      104763
                                                                             Flow model
                                                                             boundary

                                                                             Measurement point

                                                                             Shallow karst spring

                                                                             Average piezometric
                                                                             head

                                                                             European water
                                                                             devide
                        °63 °°B< °°B5 °°BB °°67 °°BB °°6B °°70 °°71  °°7? °°73 °°74 °°75 °°76 °°77
                Figure 4:       Groundwater level map - 'Stubersheimer Alb' (Lang et al., 199la)
                            [ m )
                            SBO-D
                                             Installation of automatic
                                             datalogger (04.08.19SS)
                                  1987    198B
                Figure 5:       Groundwater levels at B33 with and without automatic recording
                               (Lang etal., 1991 a)
                                                   22

-------
d) Small scale hydraulic tests

       Figure 6 displays two types of plots of the recovery data of slug tests in the Gallusquelle
catchment (region 4). A log-log plot of test B8_3 (Sauter, 1991), which clearly reveals 'double-
porosity'  response  during early times, and  an analytical plot used for parameter evaluation
according to the procedure of Bouwer et al (1976). The analytical plot depicts three sets of
recovery data from the same borehole. The three tests only differ in the initial displacement and
the resulting hydraulic conductivity decreases with increasing initial displacement. This feature
demonstrates  very  clearly the particular  heterogeneous  characteristics of a double- porosity
system, which have also been described by Streltsova (1988). If the well is directly connected to
the fissure-conduit system, short tests will  only reveal the response of the highly permeable
conduits, whereas the longer the test (i.e. the  larger the displacement), the more will the test data
reflect the reaction of the entire system including the fissures. The radius of investigation, which
can be determined  according to Ramey et  al. (1975) and Sageev (1986) to range between 10 and
20 m, increases with increasing initial displacement. The scale hierarchy for this flow problem
can be described as I <= Dt « L at times t =  1,2, ..., n with D1 < D2 .... < Dn.

       Increasing the displacement depth even further leads to an assymptotic levelling off of the
hydraulic conductivity value. At this scale, the system responds almost homogeneously.
                      10-.

                       5\


                       2-

                       1
                    E°'5
                   10.2
                     0.1
                     0.05
                     0.02
                     0.01
Tesl3
Slope 0.0062
Hoi= 11.2m
                 TestBS 3
I,
i
the
                         Response of N atrix
                  At1 Response olttie Fracti res
                                                                  100000
              Skjgtest, BB. Bouwer & Rice
                              200
                                    400
                                          600    800
                                           Time [s]
                    1,000   1,200  1,400
                      Figure 6:
 Slugtest analysis at B8 (Sauter, 1991)
Transport Phenomena

e) Areal source tracer tests

       It has been recognised for some time that spring water quality variations provide valuable
information on  the  flow system within a  karst aquifer. They have  been frequently used to
evaluate aquifer characteristics and groundwater velocities and to distinguish between different
types of springs. They also allow the distinction between event and preevent water (Sklash and
Farvolden, 1979) and between the fast and the slow flow component (Dreiss, 1989a). The analysis
of the variation  of these parameters also provide information on the role of parts of the whole
system, e.g. the epikarstic zone (Williams, 1983).

       Figure 7 depicts time series of hydraulic and physico-chemical parameters during a four
monthly period, commencing with April 1989, observed in region  4 (Sauter, 1990). The spring
                                               23

-------
flow, groundwater hydrographs and the water quality variations suggest that two distinct flow
systems exist. Some of the parameters such as spring discharge integrate over the whole system,
others are more selective for the fast system (turbidity, electrical conductivity) and some provide
information on both systems (groundwater hydrographs, S^O, temperature).

       Spring water turbidity, frequently characterised by a bimodal breakthrough, is indicative
of the fast water component, because a  minimum hydraulic energy is required  for the clay
particles to remain  in suspension. The first peak is induced by high flow velocities due to the
initial pressure pulse, which also is responsible for the rapid increase in discharge, and the second
by the actual  fast water component arriving at the spring. Similarly, electrical  conductivity
changes can be used to identify fast water components, because the chemical reactions of the
carbonate system are  very  rapid, so  that slow water components cannot be distinguished from
preevent water.

       Spring water temperature variations are indicative of fast as well as slow water, because
the temperature can  be considered as  a more or less conservative parameter. Temperature
variations can be traced  back to a  particular event  even after three months.  The relative
abundance of oxygen isotopes can be employed as well as a regional tracer and displays a bimodal
breakthrough  after recharge  events. The  respective time series however  reveal very complex
patterns  due to the superposition  of several events  and the variations in input  6^O.  Details
concerning this particular time series are discussed in Sauter (1990).

       Obviously, the scale hierarchy for spring flow chemographs is usually I « D = L for the
fissure system and I <=> D = L for the conduit system.


f) Point-source tracer tests

       Point source tracer tests reveal  information on the hydraulic characteristics  of the
permeable rock, positioned between input and output location. Depending on the type of test and
the scale of investigation,  the test data  reflect the properties  of the  conduit  system and the
fissured  system (as the extremes) and a combination of both. If the input location is in the area
of a sinkhole (most of the tests), the hydraulic and transport characteristics of the  conduits can
be examined (neglecting the effect of the transport  through the unsaturated zone). With a forced
gradient test between two boreholes (probability to encounter a directly connecting fracture is
low), the breakthrough curve can be analysed for the parameters of the fissured system. Tracer
tests, with the input  in boreholes and where the breakthrough is  measured at a spring, are
influenced by the fissured and the conduit  system. The relative proportion of each system is
determined  by the  the distance the dye has to travel from the borehole to reach  the draining
conduit.

Below, different examples of tracer tests, belonging to the above three categories, are presented.
Conduit System

       In region 1 numerous dye-tracer tests were conducted during earlier studies (Villinger,
1969). Due to the intense and deep karstification, a considerable part of the flow is believed to
occur in conduits. Figure 8 shows the relationship between the spring discharge and the flow
velocities as measured for various artificial dye-tracer tests in the study area. Within the observed
range between 20 m/h and 200 m/h, the tracer velocities appear to  correlate linearly with the
spring discharge. However, the high flow velocities observed for most of the artificial tracer tests
are not consistent with the flow regime characterized by  the  steep slopes of the piezometric
surface.  As  described  above,  the piezometric levels  measured  in  observation  wells are
representative for the diffuse-flow as well as for the conduit-flow system, whereas the point-
source artificial tracer tests provide information on the tracer flow-path from the injection point
to the outlet spring or stream. The scale hierarchy for such  an investigation is I = D = L . There
is only one input-output information available for each traced flow-path n. As observed in many

                                             24

-------
  20 -
  10  -
-mm
Ji.
i .Apr.89 i .Ma'y.89
Groundwater Recharge
• 1 r ll
i.Juri.89 i.Jul.8'9
IT r
i. Aug. 89
  0.8




  0.5




  0.2



 686


 685


 684


 683



 580


 570


 560


 550





-10.0
_ /jS/cm
-10.5
        m3/s
                            Spring Discharge
                                                 B7
        Groundwater Hydrographs
          Temperature
                                 elect. Conductivity
    618O
    spring water
Recharge
mmm  > 20mm
^» 15 -20 mm
I--W---M 10-15 mm
i- • • i  5-10mm
1   i   < 5mm
                                                                   NTU  E
                                                                   B7
                                                                 m a.s.l.
                         618O
H                  rainfall input
 .  a sample not taken during abstraction
   + sample taken during abstraction
m  FW_ fast water
   SW~ slow water    n
   SCW subcutaneous d f
      watera
                                Oxygen Isotopes
                                                           1.0

                                                           0.5

                                                           0



                                                           661




                                                           660
                                                                          8.6
                                                           8.5
                                                                           -4
                                                                           -6
                                                                           -8
                                                                          -10
                                                                          -12
                                                                          -14
                                                                          -16
    Figure 7:      Time series of hydraulic and physico-chemical parameters (Sauter, 1990)
                                       25

-------
                     Va [m/h]

                       200
                                       ECHAZQUELLE
                        100 -
y =
0.21X
- 8.6
                                                r = 0.81
                                                         r> Q [1/S]
              Figure 8:
 .—i	1	1	1	1	r
300   400  500  600 700  800 900
Relationship between spring discharge and tracer velocity for the
Echazquelle  (Teutsch, 1988)
karst areas, a tracer test from another location within the same catchment may therefore lead to
a totally different parameter set.
Diffuse System

       Stober (1991) presented a  tracer breakthrough  of a test,  conducted between two
geothermal wells, completed in karstified limestones of the Upper Jurassic in the area of Saulgau,
situated approximately 50 km south of the Swabian Alb. The wells are 430 m apart. Correcting
for interferences from a third well, the breakthrough curve coincides with the theoretical curve
for porous media. Because of the low probability of a direct connection between input and output
well via a conduit, the test results clearly reflect the characteristics of the fissured system, which
obviously can be represented by a simple porous continuum approach. The average pore water
velocity was determined at less than 0.2 m/h. The scale hierarchy for this test is I « D <= L,
where I represents the fissure system only.
Mixed Diffuse-Conduit System

       Merkel (1991) performed a number of tracer tests with the input in some of the boreholes
of the Gallusquelle catchment (region 4). Samples were taken at the spring, which implies that the
breakthrough is a result of the influence of the fissured and the conduit system. The curves are
characterised by relatively high peak velocities (35 m/h - 70 m/h) and a long drawn-out tailing.
Because the dye was directly injected into the groundwater, the  unsaturated zone does have no
effect on the breakthrough. The shape of the curve is mainly determined by the properties of the
fissured system, whereas the variation in peak velocity depends on the distance between the input
location and the point where the  dye.enters the conduit system.
Some Conclusions

       From the field observations described above, it is evident that what is generally described
as the characteristics of a karst flow system may depend very much on the type and the scale (D)
of the investigation method used and also on the scale (L)  of the flow domain. A detection
method with a large averaging volume will produce parameter fields which appear to be almost
homogeneous, whereas from small  scale measurements  the  same aquifer may appear highly
heterogeneous. Similarly, an investigation method might be more selective in the analysis of the
high or the low hydraulic conductivity zones.

       Consequently, there  is not a singular method which  should be used  to investigate the
properties of a karst system. Only the combination of different methods at borehole, intermediate
and catchment scale may provide a somewhat complete view of the relevant flow processes.

                                             26

-------
MODELING APPROACHES

       As  described  above, it  may  prove useful to  analyse  and classify the various field
observations employing the scale hierarchy concept. However, for any quantitative interpretation
the development of an adequate model tool is a prerequisite. There is a whole variety of (physical)
model  concepts  which have been  proposed for the simulation of the hydraulic behaviour of
fractured rocks. They all employ either a discrete or continous representation of the fracture flow
system, together with some  algorithm describing the exchange between the fractures and  the
surrounding rock matrix. The key features of both concepts are briefly discussed below, with
respect to their applicability for the simulation of karstified aquifer systems. It can be shown that
the selection of the appropriate modeling approach has to be based on the  prevailing scale
hierarchy of the flow problem.

       Figure 9 shows a hypothetical karst aquifer block including  fissures  and  conduits of
different size, shape and orientation. This aquifer prototype is translated into five alternative
model  representations of different type and complexity. For any  practical problem, the major
model  selection criteria are the required investigation effort, the  capabilities to simulate karst
specific flow characteristics  and also the required accuracy of the  modeling.

       The simplest approach to karst flow is the single-continuum porous equivalent (SCPE)
representation shown in Figure 9. It is certainly a crude approximation to the complex variability
of the  prototype system. Obviously, using this approach individual fractures or conduits cannot
be adequately represented.  However, in  case  of  a large investigation domain  L, individual
heterogeneities may not be relevant for the overall system behaviour. To simulate the large scale
aquifer behaviour the spatially averaged system properties might be more relevant and have to be
investigated at a scale D much  larger than the size of the heterogeneities. Therefore, for  the
resulting scale hierarchy I « D < L, the  SCPE model could possibly be a valid approximation of
the real system. Of course, in a karst terrane these conditions apply only at very large scales and
only in cases  where the karstification process is  not too far advanced. Practical modeling studies
where  this  approach was used for large scale karst studies are described by Baoren and Xuming
(1988), Maclay and Land (1988), Teutsch (1988), Tibbals (1990).

       A much better representation of the prototype karst system would be achieved if discrete
flow-paths could be represented within the model. Two possible representations, the discrete
singular fracture set (DSFS) and the discrete multiple fracture set (DMFS) approach are shown in
Figure 9. In the DSFS model only one set of fissures and/or conduits is represented, whereas  the
DMFS model comprises multiple sets of discontinuities representing  the whole range of flow
systems with small scale fissures and also large scale conduits. The underlying theory of flow
through fracture networks has been developed e.g. by Louis (1967). Because of the improved
possibilities to represent complex reality, the DSFS and the DMFS models seem preferrable  to the
continuum models. However, it should be noted  that for  the discrete fracture approach  the
location and the  geometry  of  all individual  discontinuities  must be either exactly known
(deterministic model), or has to be described using statistical parameters (stochastic model). For
any regional  karst aquifer study this  leads to an almost unresolvable parameter identification
problem, where the scale hierarchy is I <= D « L. Bearing in mind  that the fracture permeability
is proportional to the  cube of the fracture width, there is very little chance that any of todays
geophysical techniques could provide the  required resolution. Therefore, the DSFS  and DMFS
concepts are applicable only for small scale investigations where adequate fracture statistics might
be obtained from  densely spaced measurement points. To our knowledge the DSFS or DMFS
concepts have not  been used so far for any real-world regional study in a karst terrane.

       Where appropriate, large scale information is available on the location and the extent of
the major conduits, a combination of the SCPE and the  DSFS approach might be most adequate.
As shown in Figure 9 the diffuse-flow system would be represented by a porous equivalent and
the conduit system through discrete fracture elements. The scale hierachy for such a system is I
« D < L for the SCPE representation and I >= D < L for the DSFS representation.  For such a
problem, the parameter identification problem is mostly reduced to the detection of the geometry
of the  large conduits system. Kiraly (1982) analysed some hypothetical configurations to

                                              27

-------
                               KARST AQUIFER
        SCPE
                         MODEL REPRESENTATIONS
DCPE
       I « D < L
         DETERMINISTIC
           APPROACH
SCPE-DSFS
DSFS
              I « D < L
              I >• D < L
                I <» D « L
DMFS
              I <= D « L
                                      STOCHASTIC
                                      APPROACH
        small
                              INVESTIGATION EFFORT
                                           •+•  high
        high  •*•
       PRACTICAL APPLICABILITY
                                  small
       limited
CAPABILITY TO SIMULATE HETEROGENEITY
                                                                    good
Figure 9:     Karst aquifer prototype and five possible model representations (SCPE: single-
            continuum porous equivalent, DCPE: double-continuum porous equivalent, DSFS:
            discrete singular fracture set, DMFS: discrete multiple fracture set)
                                       28

-------
demonstrate the resulting flow effects. Practical field studies were presented by Kiraly (1984) and
Yusun (1988).

       A compromise between the simple single-continuum and the complex discrete fracture
representation is the double-continuum porous equivalent (DCPE) approach shown in Figure 9.
Based on the available field evidence  from karst terranes, showing high fluctuations of the
piezometric  levels together with  very high groundwater flow velocities,  physical  reality is
approximated using the concept of two overlapping continua. One continuum is represented by
the more karstified areas (conduit-flow), yielding high groundwater flow velocities but hardly
any water level fluctuations. The second continuum is represented by the moderately karstified
aquifer  zones (diffuse-flow)  with lower hydraulic conductivity and higher storativity. The
advantage in using the DCPE  model instead of the SCPE is the  improved representation of the
karst aquifer heterogeneity. On  the other hand, the DCPE model does not require the detailed
geometrical information which is needed for the discrete fracture approaches. Typically, the scale
hierarchy where the DCPE model is  applicable would be  I « D < L  for the diffuse-flow
continuum and I <= D <= L for the conduit-flow continuum. The DCPE flow and transport model
approach was successfully used in the analysis of several karst systems of the Swabian Alb aquifer
(Teutsch 1988, 1989; Sauter 1990, 1991). To our knowledge, the only other attempt to use a DCPE
approach in  a Karst terrane is reported by Yilin et al. (1988) who simulated the flow in the karst
regions of northern China.

       Based on the above said, the authors believe that the double-continuum approach is a
favourable model concept for many practical karst aquifer studies where data is usually scarce but
detailed enough to show that a single-continuum model cannot be applied. One further advantage
of the DCPE approach is its simple mathematical formulation described below, which facilitates
the implementation into  existing program codes.
THE DOUBLE CONTINUUM POROUS EQUIVALENT (DCPE) MODEL

       Mathematically,  the double-continuum concept is similar to the well-known double-
porosity approach which is extensively used in the oil industry (Barenblatt et al. (1960), Warren
and Root, 1963; De Swaan 1976; Streltsova 1988). However, flow and transport may occur in both
systems with an exchange coefficient desribing the cross-flow between the two. The DCPE model
may be visualized as two porous blocks with a common outflow, which are connected through
hydraulic resistances at any of the block cells. Such a system is shown in Figure 10.

The equations describing the flow and transport of a non-reactive solute in a double-continuum
system are (Teutsch, 1988):

Flow (one-dimensional,  depth-averaged):

  ^ tVm \f ^m\   cm^* . /  ivm-l -/-Lut fc»4h       _, _ 1 O    (1\
  —l/Lr M 	1 =  A 	 + l-l)    Otl/l  —n  )       m = 1,Z,    (I)
  cbc       cbc        dt

Transport:


  etc     cbc        cbc                         dt

and

  _. *      V*  ™" •* /     f1\
  a.  =  a.  	——     (3)
                                             29

-------
where
       m = continuum index
       KX = hydraulic conductivity
       S = storage coefficient
       C = concentration
       ne = effective porosity
h = head potential
M = aquifer thickness
D = hydrodynamic dispersion coefficient
v = groundwater flow velocity
a = exchange coefficient
                                groundwater recharge
                                                               fissure-system
                                                                      Q
                                                             Q     = Q  +Qf
                                                                ges     m    i
                                                              conduit-system
                                                                      Q
              1 2  3  4  5  6 7  8  9 10 11 12 13 14 15 16 17 18
       Figure 10:     Double-continuum porous equivalent (DCPE) model (Teutsch, 1988)

       Two assumptions have to be made in order to apply this concept to a karstified aquifer.
Firstly, there is the assumption of a potential continuum in the highly permeable conduit-flow
system, at an intermediate scale smaller than the modelled domain. This is probably true for most
of the study regions presented, but not necessarily true  for those aquifer  zones comprising
extremly large active conduits. Secondly, it is assumed that the flow in both continua is laminar.
This is probably not always true for the fast conduit-system, especially  not following strong
recharge events. However, due to the lack of information  about the  discrete geometry  of
individual conduits, no correction for the nonlinearity of the flow can be introduced.

       Based  on  the  well  known MODFLOW groundwater flow program (McDonald and
Harbaugh,  1984) a three-dimensional double-continuum  model was recently developed at the
University of Stuttgart. This program is presently used for the analysis of the regional karst flow
in region 5, where a railway tunnel with a length of 13 km is planned to cut through the Swabian
Alb escarpment.
CASE STUDY

       The applicability of the DCPE model has been tested on the Gallusquelle groundwater
catchment (region 4), an area for which longterm (25 years) of flow and hydrograph data were
available, as well as numerous  tracer tests, small scale and large scale hydraulic tests and two
years of water quality data. These were good conditions to calibrate and test the model approach
on actual field measurements.
                                             30

-------
       Figure 11 and 12 demonstrate that the flow and the hydraulic characteristics could be
reproduced  very well,  considering  the assumptions in recharge  input  and  its respective
distribution with time and space (Sauter, in prep.)- Observed differences can be explained by the
absolute depth of recharge and by the fact that the hydraulic parameters vary with depth. In
Sauter (1991 - this issue) it is shown how the aquifer geometry, the hydraulic conductivity and
the storage coefficients have been determined for each part of the system and at the respective
scale, consistent with that used for model calibration.

       Without any further modifications, the regional transport of an inert areal tracer, such as
fi^O, was simulated with a double-continuum random-walk model (Teutsch, 1988) (Fig. 13). It
is interesting to note, that as opposed to the unimodal breakthrough of point source tracer tests,
areal source  tracers may produce a  bimodal breakthrough,  which,  as shown  in Figure 7, is
modified due to the interference of several other recharge events (dilution). The result is a highly
complex signal, which compares well with the isotope breakthrough observed in the field, which
would be very hard to understand without any model assistance.

       Further work will focus on the analysis of breakthrough signals resulting from point and
areal sources modified by the various physical and chemical processes. The model is a tool which
can  be  employed  to analyse specific scenarios concerning  resource development as  well as
contaminant migration. Furthermore, it allows the testing of various hypothesis such as autogenic
and allogenic recharge mechanisms and also the  simulation of chemical changes due  to e.g.
dissolution.
            710
            700-
            690-
            680 -
            670
                   m a.s.l.
                                   Groundwater Hydrograph (B14)
                                            Model Data
                                            Field Data
                                \\
                                       '"- .  ,•""**""
                AITOTSl D[J F M AM J JASON D|J F M A M J J AS D N DfJ 'F M'A'M'J'J 'A
                    1987          1988                 1989              1990
              Figure 11:    Measured and simulated groundwater hydrograph
              2 -
              1 -
                                     Spring Discharge
                             Model Data
                             Field Data
V S O N Dl J F M A M J J A S O N D J F
    1987           1988
                                                    A M J J A S ON DTJ 'F 'M 'A 'M' J 'J 'A
                                                        1989               1990
               Figure 12:    Measured and simulated spring flow hydrograph
                                              31

-------
                  Breakthrough of a conservative regional tracer (Event 1 April)
                                                     Event of 1 April
                                                --- Event of 17 April
                                                     Event of 23 June
                                                     Event of 10 July
             01.04.89
                         01.05.89
                                                 01.07.89
                                                             01.08.89
              Figure 13:     Simulated breakthrough of an areal-source tracer
SUMMARY AND CONCLUSIONS

The aim of this paper is to present a consistent framework for the classification of flow and
transport phenomena frequently observed in karst terranes, which in many cases appeared to be
incompatible in a single flow system. This approach is based on scale hierarchy concepts  which
are illustrated on various field observations. It is shown that the classification of the flow system
in a heterogeneous aquifer like the karst may depend strongly on the type and the scale  of the
investigation method used. Furthermore,  the relationship between the model conceptualisation
and the scale hierarchy is discussed. Various experiences from the Swabian  Alb karst aquifer
system show that the double-continuum porous equivalent (DCPE) model approach is superior to
other concepts for most karst aquifer situations, which require  an adequate representation of the
heterogeneities on a very limited data base.
REFERENCES

Aiguang, C., Hanchen, L., Juanming, X. and S. Jinli, 1988, Stochastic Modelling of the Karst
       Spring of Xin'an, Shanxi Province, in: Karst Hydrogeology and Karst Environment Pro-
       tection, Proc. of the 21st Congress  of the Int. Assoc. o.Hydrogeologists, Guilin, China,
       621-628.
Baoren, C.  and Z. Xuming, 1988, The mathematical simulation  of  hydrological regime of
       Chaohua spring  in Mixian County. Karst  Hydrogeology  and Karst Environment Pro-
       tection, Proc. of the 21st Congress of the Int. Assoc. o.Hydrogeologists, Guilin, China.
Barenblatt,  G.E.,  Zheltov, I.P.  and I.N.  Kochina,  1960, Basic concepts in the  theory of
       homogeneous liquids in fissured rocks. J. Appl. Math. Mech. (USSR), 24,  1286-1303.
Behringer, J., 1988, Hydrochemische Kurz- und Langzeitstudien im Malmaquifer der Mittleren
       Schwabischen Alb. Ph. D.  Thesis, Universitat Tubingen, 222p.
Dagan,  G.,  1986,  Statistical  theory  of  groundwater flow  and transport:  pore  to  laboratory,
       laboratory to formation and formation to regional scale. Water Resour. Res., 22(9)  120S-
       135S.
Dreiss, S.J.,  1989a, Regional scale transport in a karst aquifer, 1. Component separation of spring
       flow hydrographs, Water Resour. Res., 25, 117-125.
Dreiss, S.J.,  1989b, Regional scale transport in a karst aquifer, 2. Linear systems  and time
       moment analysis, Water Resour. Res., 25, 126-134.
Duffy, C.J.  and J. Harrison,  1987, The statistical  structure  and filter characteristics of tritium
       fluctuations in fractured basalts, Wat. Resour. Res., 23, 894-902.
                                             32

-------
Haldorsen, H., 1986,  Simulator parameter assignment and the problem of scale in reservoir
       engineering, in: Reservoir Characterization, Eds. L.W. Lake und H.B. Caroll, Academic
       Press.
Irmay, 1964, Theoretical models of flow through porous media. RILEM symp. on the transfer of
       water in porous media, Paris.
Kiraly, L., 1984, La regularisation de 1'Areuse (Jura Suisse), simulee par modele mathematique.
       in: A. Burger & L. Dubertret. eds, Hydrogeology of Karstic Terraines, International Con-
       tributions to Hydrogeology, Heise, Hannover.
Kraemer, S.R. and H.M. Haitjema, 1989, Regional Modelling of fractured rock aquifers, in: G.
       Jousma, J. Bear, Y.Y. Haimes & F. Walter, Groundwater Contamination: Use of models
       in decision making, Kluwer, Dordrecht, 467-476.
Lang, U., Teutsch, G. and Kobus, H.,  1991a, WeiBjura-Grundwasserbilanzmodell Stubersheimer
       Alb - Modellkonzeption und Programmtechnik. Bericht des Instituts fur Wasserbau Nr.
       91/1 - HG 139, Universitat Stuttgart, 87p.
Lang, U., Teutsch, G. and Kobus, H.,  1991b, WeiBjura-Grundwasserbilanzmodell Stubersheimer
       Alb - Stationare Modelleichung", Bericht des Instituts fur Wasserbau Nr. 91/20 - HG 150,
       Universitat Stuttgart, 81 p.
Long, J.C.S., Gilmour, P. and  P.A. Witherspoon, 1985, A model for steady fluid flow in random
       three-dimensional networks of disc-shaped fractures. Water Resour. Res., 21, 645-658.
Louis, C.,  1967,  StrOmungsvorgange in  kliiftigen  Medien  und ihre  Wirkung auf  die
       Standsicherheit von Bauwerken und Bdschungen  im  Pels, Ph.D.  dissertation, Univ. of
       Karlsruhe,  122p.
Maclay, R.W. L.F. Land,  1988, Simulation of flow in the Edwards aquifer, San Antonio region,
       Texas, and refinements of storage and flow concepts. U.S.G.S. Water Supply Papers, 2336.
McDonald, M.G. and A.W. Harbaugh, 1984, A Modular Three-Dimensional Finite-Difference
       Ground Water Flow Model. USGS, Reston, 527p.
Merkel, P., 1991, Karsthydrologische Untersuchungen im Lauchertgebiet (westl. Schwabische
       Alb).  Diplom thesis, Geologisches Institut, Universitat Tubingen, 108p.
Ramey, H.J. jr., Agarwal, R.G. and I. Martin, 1975,  Analysis of "slug" test or DST flow period
       data. J.Can.Petr.Technol., 37-47.
Romm, E.S.,  1966, Fluid flow in fractured rocks, Nedra, Moscow.
Sageev, A., 1986, Slug test analysis. Wat. Resour. Res., 22, 1323-1333.
Sauter, M., 1990, Double porosity models in karstified limestone aquifers, Proc. Int. Symp. Field
       Seminar on Hydrogeological Processes in Karst Terranes, 7. - 17.Oct.1990, Antalya, Tur-
       key.
Sauter, M., 1991, Assessment  of hydraulic conductivity in a  karst aquifer at local and regional
       scale. 3ri Conf. on Hydrogeology, Monitoring and Management of Ground Water in Karst
       Terranes, Nashville, Dec. 4-6  (this issue)
Simpson, E.S., 1988, The  discrete state compartment  model and its application to flow through
       karstic aquifers. Karst Hydrogeology and Karst Environment Protection, Proc. of the 21st
       Congress of the Int. Assoc. o.Hydrogeologists, Guilin, China.
Sklash, M.G.  and R.N. Farvolden,  1979, The role of groundwater in storm runoff, J.o.Hyd., 43,
       45-65.
Smart, P.L. and  S.L. Hobbs,  1986, Characterisation  of carbonate aquifers: A conceptual base.
       Proc. of the Environmental Problems in Karst Terranes and their Solutions Conference,
       National  Water Well Association, Bowling Green, Kentucky,  1-4.
Smith, L., Schwartz, F. and C. Mase, 1987, Applications of stochastic methods for the simulation
       of solute transport in discrete  and continuum models of fractured rock systems. Proc. of
       the Conference on Geostatistical, Sensitivity and Uncertainty Methods for Groundwater
       Flow and Radionuclide Transport Modelling,  S. Francisco, 425-440.
Snow,  D.T., 1965, A parallel plate model of fractured permeable media.  Ph.D. thesis, Univ. of
       California,  Berkeley, 330p.
Stober, I., 1991, StrOmungsvorgange  und Durchlassigkeitsverteilung innerhalb des WeiBjura -
       Aquifers  im  baden-wiirttembergischen  Anteil  des   Molassebeckens.   Laichinger
       HShlenfreund, 26, 29-42.
Streltsova, T.D.,  1988, Well testing in  heterogeneous formations. Wiley, N.Y., 412p.
de Swaan, A.O., 1976, Analytic solutions for determining naturally fractured reservoir properties
       by well testing. Soc. Petr. Eng. J., 16, pi 17.

                                              33

-------
Teutsch, G., 1988, Grundwassermodelle im Karst:  Praktische Ansatze am Beispiel  zweier
       Einzugsgebiete im Tiefen und Seichten Malmkarst der Schwabischen Alb. Ph.D. Thesis,
       Universitat Tubingen, 205p.
Teutsch, G., 1989, Groundwater models in karstified terraines - two practical examples from the
       Swabian Alb, S. Germany. Proc. of the 4th Conference - Solving Groundwater Problems
       with Models, Indianapolis, USA, Feb. 7-9 1989, 929-953.
Teutsch, G., 1990a, A variable scale approach for the delineation of groundwater protection zones
       using field and numerical modelling techniques. IAH Selected Papers of the 28th Inter-
       national Geological Congress, Vol. 1, 1990, 11  p.
Teutsch, G., 1990b, An extended double-porosity concept as a practical modeling approach for
       karstified terranes. Int.  Symp. Field Seminar on Hydrogeological Processes in  Karst
       Terranes, 7.  - 17.Oct.1990, Antalya, Turkey.
Teutsch, G., Hofmann, B., Ptak, T., 1990, Non-parametric stochastic simulation of groundwater
       transport processes  in highly  heterogeneous  porous formations. Proceedings  of the
       International Conference and Workshop on Transport and Mass Exchange Processes in
       Sand and Gravel Aquifers: Field and Modelling Studies, Ottawa, Canada, October 1-4,
       18p.
Thrailkill, J., 1986, Models and methods for shallow conduit-flow carbonate aquifers. Proc. of the
       Environmental Problems in  Karst Terranes and their Solutions Conference, National
       Water Well Association, Bowling Green, Kentucky, 17-31.
Tibbals, C.H., 1990, Hydrology of the Floridan aquifer system in east-central Florida. U.S.G.S.
       Professional Paper, 1043-E.
Warren, J.E. and P.J. Root, 1963, The behaviour of naturally fractured reservoirs, Soc. Petrol.
       Eng. J., 3, 245-255.
White, W.B., 1969, Conceptual models for limestone aquifers. Groundwater,  7, 15-21.
White, W.B., 1977, Conceptual Models for Karst Aquifers:  Revisited, in: Dilamarter, R.R. &
       Csallany, C.S. (eds), Hydrologic Problems in Karst Regions, Western Kentucky Univ.
       Press, Bowling Green, Kentucky, 76-87.
Williams, P.W., 1983, The role of the subcutaneous zone in karst hydrology. J. o. Hyd., 61,45-67.
Yilin, C., Hongtao, W. & X. Xinhui, 1988, Dual-media flow models of karst areas and their ap
       plication in north China. Karst Hydrogeology and Karst Environment Protection, Proc.
       of the 21st Congress of the Int. Assoc. of Hydrogeologists, Guilin, China.
Yusun, C. & B. Ji, 1988, The media and movement of karst water. Karst Hydrogeology and Karst
       Environment Protection, Proc. of the 21st Congress of the Int. Assoc. o.Hydrogeologists,
       Guilin, China.
BIOGRAPHICAL SKETCHES

Georg Teutsch:  He  received his M.Sc. in  Hydrogeology in  1980 from  the  University  of
Birmingham (UK) and his Ph.D. in 1988 from the University of Tubingen (FRG). He has been
recently appointed as an associate professor of geohydrogeology at the Institut fur Wasserbau of
the University of Stuttgart where he conducts a major part of the groundwater research since
1986. After various assignments in the field of groundwater resources evaluation and exploration,
his current interests focus on transport processes in heterogeneous media. He has authored more
than 30 scientific publications in the field of practical and theoretical  geohydrology with  an
emphasis on groundwater modeling.

Martin Sauter: He is  a research associate at the Chair of Applied Geology at the University of
Tubingen (FRG). In  1980 he obtained a diploma in Geology from Tubingen University and in
1981 an M.Sc. in Hydrogeology from Birmingham University (UK). Between 1982 and 1987 he
was  involved in various projects in  Germany and overseas, covering  areas of groundwater
contamination,  geophysical well logging and water resources  development. He is  presently
finalizing his Ph.D. project on regional flow and transport modeling and model validation in karst
terranes. His main interests lie in the area of groundwater flow and transport modeling in karst,
hydraulic testing methods and hydrochemistry.


                                            34

-------
                    Questions-and-Answers:

Ql:   You identified your model used as a modification of
     MODFLOW. Is this modified MODFLOW available in either draft
     form or final form ? If so, from whom ?

Al:   The double-continuum model presented in the paper is based
     on the well-known USGS MODFLOW package. However, the
     structure of the program is different because of two
     instead of one continuum and the cross-flow calculation.
     Furthermore, a routine has been developed which allows the
     reactivation of cells which fall dry during the simulation.
     A documented and tested version of the program will be
     available from the Univ. of Stuttgart, probably by mid of
     1992.
Q2:   You briefly mentioned that the northern aquifer system is
     dominated by diffuse flow [Remark: here the correct
     question should probably read "conduit flow" instead of
     "diffuse flow"] and the southern aquifer is dominated by
     diffuse flow. What are the geologic controls on [Remark:
     here the correct question should probably read "and"
     instead of "on"] which aquifer type is dominant ? How
     important are facies changes within the carbonate rocks.

A2:   In the northern part of the Swabian Alb karst aquifer the
     piezometric levels are controlled by a series of springs
     which developed at the aquifer base during the Quaternary
     as a result of  tectonic uplift and tilting of the limestone
     plateau. Due to the spatial concentration of flow and the
     increased hydraulic gradient, the original diffuse flow
     conditions therefore matured to a conduit flow system. In
     the southern part of the aquifer, the tilting of the
     limestone plateau caused a decrease in the hydraulic
     gradient and therefore the diffuse flow conditions were
     maintained. Within this regional flow pattern, the local
     flow conditions are of course highly dependent on the type
     of carbonate facies and therefore highly variable.
Q3:   The fit between the DCPE model and actual isotopic values
     is significantly worse than would be the case if a simple
     2-point moving average had been used. What then is the
     utility of the DCPE model ?

A3:   It should be noted that the isotope data was not used for
     the calibration of the DCPE model. The model was calibrated
     on the base of spring hydrographs, groundwater levels and
     a few point-input, artifical dye-tracer test data only,
     whereas the isotope data represent an areal-input,  natural
     tracer transport.  In fact, a recently conducted analysis of
     the isotope breakthrough curves has demonstrated the high
     sensitivity of the model outcome on the double-continuum
     cross-flow coefficent.  Referring to the proposed 2-point
     moving average approach, I don't see how this could help
     improve the model fit in a physically meaningful way.

                                35

-------
36

-------
               Session I:
Ground Water Modeling and Hydrogeology

-------
38

-------
        ASSESSMENT OF HYDRAULIC CONDUCTIVITY IN A KARST AQUIFER
                          AT LOCAL AND REGIONAL SCALE
                                      Martin Sauter
                                   Geology Department
                              University of Tubingen, FRG
ABSTRACT

The hydraulic parameters in a karstified limestone aquifer are usually only available for single
boreholes. There is a high density of information on wells in valleys, which are not necessarily
representative for the aquifer due to an increased degree in karstification close to the regional
outlet. As a result of these circumstances,  the values obtained cannot be directly used as input
parameters for a regional model. The selection of the appropriate hydraulic parameters requires
a careful analysis of the scale,  the values are applicable at, and the allocation of the parameters
to the respective part of the system, which consists of a fast transit (conduits) and a slow flow
(fissured) system. Slug tests, injection, packer and pumping tests were used to obtain parameter
estimates of the low and intermediate range of the hydraulic conductivity spectrum, which reflect
the response of the aquifer at local scale. Regional parameters could be evaluated using several
approaches. Applying Darcy and with the information on discharge and head gradient, average
transmissivities can be calculated, representing the slow regional system. Taking  the concept of
Rorabaugh (1964) for flow from bank storage, regional parameters were computed, using the
recession coefficients for the fast and slow flow system. The necessary storage coefficients for the
fast system could be derived from tracer tests and for the slow system from the quotient between
the discharged volume of water and the drained rock volume. The regional values were checked
for plausibility, using the local values and taking into account the different proportions of the
fast and slow system. The resulting calibrated model parameters compared reasonably well with
the observed regional values.
INTRODUCTION

During the last few years, double-continuum models have become a recognized tool for the
prediction of flow and transport in karstified limestone aquifers (Teutsch, 1988, Yilin et al, 1988,
Sauter, 1991, Sauter, 1992). This type of modeling approach manages to accomodate both, a
regional flow system, that is slowly depleted and at the same time a fast transit system to simulate
the rapid response after a rainfall event. It also simulates tracer breakthrough with the observed
high average velocities. Double-continuum models are also based on the physics of groundwater
flow and  capable of simulating spatial  variations in flow and transport  phenomena. These
advantages however require on the other  hand the knowledge of flow and transport parameters
of the fast and the slow system at the appropriate spatial resolution scale, which are frequently
difficult to obtain. Depending on the scale of the problem, different hydraulic conductivities will
apply as illustrated by Kiraly (1975) (figure 1). The hydraulic parameters of slow and fast system
will be different whether for example a borehole to borehole tracer test (several tens of meters)
or regional groundwater flow (several thousands of meters) is being analysed (Teutsch & Sauter,
1991, this issue).

In the present paper, the terms conduits and fractures are used interchangeably for the fast transit

                                           39

-------
 system and fissures and matrix are used to describe the slowly draining part of the aquifer, in
 order to show the  connection to double-porosity terminology and  its  applicability  to  karst
 aquifers.
                                                    River
                                                    Basins .
                 Laboratory
                               Boreholes
                                                                 Elfict of
                                                                 th* hirttic
                                                                 nitwcrk
                                                                 Erttct of
                                                                 pom ind
 Figure 1: Effect of scale on the hydraulic conductivity of karst aquifers (Kiraly, 1975)

 This paper attempts to demonstrate how the hydraulic parameters can be obtained from various
 tests applying different analytical methods and how the appropriate values can be found for the
 respective scale at which information is sought. The idea is to evaluate the hydraulic conductivity
 at the lower scales in order to be able to derive the resulting hydraulic  conductivity at the next
 larger scale, together with a knowledge of the various proportions of fast and slow flow system,
 which exist at every scale.  The parameters obtained can be compared with the calibrated values'
 of a regional double-continuum groundwater flow model.

 The results presented below are part of a karst research project, investigating flow and transport
 processes in a  spring catchment of  a karstified limestone aquifer in south Germany. The
 objectives are the numerical simulation of these processes  and the quantification of the input
 parameters and  groundwater recharge.


 SCALE PHENOMENA IN  KARSTIFIED LIMESTONE AQUIFERS

 In highly heterogeneous systems such  as karst aquifers, parameters like porosity and the related
 storage coefficient are highly dependant upon the scale of the investigation method (Teutsch &
 Sauter, 1991, this issue). Castany (1984) demonstrates how the porosity values obtained vary with
 the  volume of  rock  sampled, expressed as the radius of investigation. Starting at  a  micro
 (laboratory) scale, where the porosity can vary from  0 (tight rock) to  1  (void), homogeneous
 values can be measured if a sufficiently large number of heterogeneities (fissures, conduits) are
 integrated within the sample volume. Contrary  to the porosity, the  hydraulic conductivity
 increases  at a larger scale (Kiraly, 1975, figure 1, Department of Energy  1986)  which is
 plausible because of the added effect of the high permeability conduits. The porosity can also be
 assumed to increase but only very slightly, which however cannot be measured This  different
 behaviour could be explained by the cubic  law (Snow, 1965), which describes the relationship
 between permeability and the fracture aperture.


 RELEVANCE OF SCALE  EFFECTS IN GROUNDWATER MODELLING

 In a numerical model, simulating groundwater flow and transport, the hydraulic parameters
 dedicated to the respective discretised  unit (element/cell),  is  a  result  of the  sum
 (arithmetic/harmonic/geometric) of the parameters of the identified sub-systems  weighted by
 their respective proportions within  the particular model unit. The problem of ac'comodatine a
number of flow sub-systems in a  numerical model can be approached by  either employing
models, incorporating multiple  continua such as  MINC  (Narasimhan  et  al  1988) or by
                                           40

-------
determining the representative parameter at the modelled scale of a double-continuum model.
The  structure of the MINC model,  applied  to a  karst aquifer,  could be visualised as an
intertwined  system  of fast (fracture, conduit) and slow (matrix,  fissure)  flow, i.e.  double-
continuum systems,  whereby each slow system can again be split into a conduit and fissure flow
continuum.

The  fractured continuum of the lowest manageable double-continuum sub-system could be
visualised as the joints, spaced at decimeter intervals (10 cm - 30 cm) and frequently measured
in quarries. The complimentary parameters of the matrix (fissure) continuum can be measured in
the laboratory on cores with dimensions of several centimeters. The above  described  double-
continuum system could  be regarded as the  lowest in the hierarchy  of multiple  interacting
continua. It forms the matrix continuum of the next higher double-continuum system, where the
higher permeable set of fractures and the solution widened fissures are spaced apart between one
and  several meters. The  double-continuum system can be  investigated at this level in  the
hierarchy by double packer tests, where the packers enclose a test interval of less than one meter.
                                                             Project area
                                                             *   Boreholes
                                                             °   Spring
                                                             =— Tracer test
                                                             — Equipotentials
                                                                m above sea level
                                                             --- Catchment area
Figure 2: Gallusquelle karst aquifer catchment
                                          41

-------
Slug and injection tests with a test  interval and radius  of  investigation of several tens and
pumping tests with a scale of investigation of several hundreds of meters can identify the double-
continuum systems at the respective scales.

At the regional, i.e. catchment scale, where conduits and caves represent the fracture continuum,
the hydraulics of the matrix continuum could  possibly be derived from pumping tests, which are,
as outlined above, the result of the hydraulic characteristics and the interaction of all the lower
scale double-continuum flow systems.

For practical purposes, if the  problem in  question  asks for answers  at  regional scale, no
information  is required on e.g. the potential  change over a distance of several meters. It is also
difficult to obtain all the relevant input data at every scale. Therefore, the flow system can be
conveniently simplified to a single double-continuum model, assuming that the parameters can
be determined at the modeling scale.

Within the context of this study, the model was designed to simulate regional flow and transport
and the  breakthrough  of tracer tests with  travel distances  of several  thousands  of meters.
Therefore, the  nodes of the model grid were spaced  500 m apart, which was  a compromise
between the  lowest resolution of information on aquifer geometry, and the radius of investigation
of a large scale  pumping test.
PROJECT AREA

For the investigations, the spring catchment of the Gallusquelle was selected. It is situated in
south-west Germany on  the  Swabian  Alb,  a small mountain  range, that stretches in an
approximate south-west north-east direction for roughly 200 km (figure 2). Morphologically, the
project area dips gently from an escarpment (1000 m a.s.l.) in the north-west down to about 600
m in the region of the spring. The Gallusquelle groundwater basin forms a part of the catchment
(450 km2) of the Lauchert river, which is fed mainly by karst springs. The river Fehla represents
the boundary to the north-east.

The area is well suited for the intended measuring program. More than  25 years of continuous
records of discharge, waterlevel fluctuations (weekly readings), and climatic data were available,
as well as a number of about 20 wells that allow the construction of a water level map, even in
areas far removed from the discharge point. Fifteen tracer tests help to delineate fairly accurately
the catchment boundaries, which cover an area of approximately 45 km2.
GEOLOGY

Geologically, the area is composed exclusively of carbonate rocks of the Upper Jurassic (figure
3). At the surface, predominantly the massive limestones of the Kimmeridge 2/3 are exposed,
which reach a maximum thickness  of  between 90 and  140 m.  They are underlain by  the
Kimmeridge 1, a marly limestone sequence, with more or less expressed bedding and a thickness
of approximately 50 m. The lowest relevant geological unit consists of well bedded limestones, the
Oxford 2. The  whole stratigraphical succession dips south-east. The south-west border of the
catchment is formed by the Hohenzollern graben, which is tectonically still active. Another fault
zone strikes north-south and borders the project area in the south-east.
HYDROGEOLOGY

As shown in figure 3, the aquifer is formed by three geological units, the massive limestones
(ki2/3) in the south-east, the marly limestones (kil) in the centre and the well bedded limestones
(ox2) in the north-west. The aquifer base does not follow any stratigraphical boundary. Figure 3
shows highly transmissive and permeable zones of the Gallusquelle groundwater catchment. The
boreholes were all projected onto a profile line, assuming no variations vertical to the profile, and

                                           42

-------
the borehole logs were interpreted in terms of tight rock, fractured and slightly karstified and
highly karstified. Further evidence, regarding the differentiation  in hydraulic characteristics
could be provided by the analysis of long and short-term records of groundwater hydrographs
(Sauter, in prep.)-

Saturated thicknesses are estimated to reach approximately 30 m, in exceptional cases 50 m,
depending on the season, although  the  limestone  sequence  may be in excess of 150 m.  An
increased gradient of the piezometric surface can be observed in the centre of the catchment
(figure  4), where the water table  cuts  across the less  permeable  marly limestones. Further
upgradient, the aquifer is formed by the Oxford limestones. The water table constitutes the top
of the aquifer and unconfined conditions prevail in the entire catchment. The unsaturated zone
is highly karstified and reaches thicknesses of between 90 and 120 m.
   NW
tight rock (non-karstified)
fractured, slightly karstified
highly karstified

aquifer, highly permeable
                                                             waterlevels, max, average, min.
                                                             boundary W1/ki2
     Karst Aquifer Geometry
SE
                                                                                  m as.l
                                                                                  r 850
                      - 800
                                                                                  - 750
                                                                                  - 700
                                                                                  - 650
                     - 600
    km  14.0   13.0   12.0   11.0   10.0   9.0
                                       8.0
                                            7.0
                                                  6.0
                                                       5.0
                                                            4.0
                                                                 3.0
                                                                      2.0
                                                                            1.0
                                                                                 0.0
                                                                                    550
Figure 3: Geological and hydrogeological crossection of karst aquifer

Closer to  the spring, annual water level fluctuations  range  between 5 and  15 m and further
upgradient, between 10 m and  30  m, reflecting the  decrease in transmissivity and storage
coefficient away from the point of discharge.
DETERMINATION OF HYDRAULIC PARAMETERS AT THE RESPECTIVE SCALE

The evaluation of hydraulic parameters was subdivided into a regional assessment, taking into
account the existence of the fast (conduit) and the slow flow (fissure) system, local and sub-local
scale tests in boreholes and laboratory measurements derived from the literature. Particular
emphasis has been put on the separate evaluation of fast and slow system characteristics.
REGIONAL PARAMETERS

Common hydraulic testing methods such as slug tests, packer tests and pumping tests, have a
limited testing radius,  which is  usually in the order of several tens to hundreds  of meters,
depending on the type of aquifer  material. The only signal, exciting larger regions of the aquifer
system are natural pulses, such as rainfall events, which can be evaluated by applying recession
analysis to the output  signal,  i.e. the spring discharge.  This method  has  the  advantage that
information can be gained over a  large area, but because the registration of the output signal can
                                            43

-------
only be measured at the spring, a differentiation into regionally varying parameters is generally
not possible. If however the  regional hydraulic conductivity varies laterally and with depth
(figure 3), the value of the parameter obtained depends on the intensity of the signal (recharge
depth) and only reflects the hydraulic characteristics of that portion of the aquifer, dominating
the flow.

The approach taken for the evaluation of the regional hydraulic conductivity was first to measure
average  regional  gradients in the south-eastern, central and  north-western part of the
Gallusquelle catchment  during  three  different  flow conditions,  low  (0.12  - 0.36  m3/s),
intermediate (0.47 - 0.99 ms/s and high flow (1.25 - 2.66 ms/s). Assuming that there is no more
recharge to  the aquifer together with a knowledge of the respective discharge and the saturated
thickness, values can be calculated for T and K.

Secondly, the recession coefficients a, derived from spring discharge measurements were used to
calculate average transmissivities at different  fill  levels of the aquifer,  using  a relationship,
developed by Rorabaugh (1964) for groundwater discharge from bank storage.
Hydraulic Conductivity

Gradient (Darcy) approach

Figure 4 shows a block diagram of the aquifer, subdivided into three different areas, a highly
conductive one in the south-east, low conductive in the central and intermediately conductive in
the north-western part. The respective discharge is allocated as a fraction of the total, according
to the three different surface areas. The base of the aquifer was taken from figure 3 and  the
saturated thickness  varied according to the waterlevel in the boreholes, taken for evaluation of
the gradient.
                                                    Equipotentials
                                             ghly conductive zone
                North-West
                                 Central
Q1 (Qtot)
(5/5 of Q tot)
                                            South-East
Figure 4: Evaluation of regional hydraulic conductivity using throughput analysis

The slope of the watertable was determined using the waterlevels in boreholes B7 and the spring
for the south-east and that in B25 and B21 for the central region. Because of the lack of borehole
information, an average regional gradient was taken from Villinger (1977) and figure 2 for the
north-west.

The hydraulic conductivities varied between 2 and 13*10"4 m/s for the south-east and 2*10"5 and
3*10"4 m/s for  the central region. An average hydraulic conductivity of approximately  1.5*10"4
m/s was determined for the north-western region. The lower value  was generally  calculated
during low flow conditions, which is believed to represent "regional matrix", i.e. regional fissure
hydraulic conductivity. The  more conduits are included at higher  saturated thicknesses (high
hydraulic conductivity zone), the higher the resulting K. During intermediate flow conditions

                                            44

-------
the higher permeable zone, identified in figure 3 is partly saturated, derived from the water level
in B7  (above  660 m), and is therefore contributing to the flow. Above a level of 663 m,
permeability is reduced again. As could already be expected from the information on waterlevels
and aquifer geometry, the hydraulic conductivity is lowest  within the high gradient area and
highest within the south-eastern area close to the spring.



South-East
Central
North-west
Gradient:
low
Va
flow: 100
K regcond:
%frac
K reg 0.0002
Gradient
K regcond:
%frac
K reg 0.00002
Gradient
K regcond:
%frac
Kreg
Gradient
low
interm.
110
600
7.6
0.00007
0.0007
0.004
3.1
0.00003
0.0001
0.01
0.00015

high
250 m/h
1500 1/s
17.4 m/s
0.00006
0.0013 m/s
0.004
6.9 m/s
0.00004
0.0003 m/s
0.01
m/s
%
m/s
Gradient: high
low interm.
110
100 600
5.1
0.00010
0.0002 0.0007
0.006
1.3
0.00006
0.00002 0.0001
0.023
0.00015

high
250 m/h
1500 1/s
9.9 m/s
0.00011
0.0013 m/s
0.007
3.0 m/s
0.00009
0.0003 m/s
0.023
m/s
%
m/s
Table 1: Regional hydraulic parameters (Darcy approach)

The main source of error lies in the estimation of the saturated thickness. It was however not
possible to use transmissivities because of prevailing unconfined  conditions and because the
waterlevel fluctuations amount to a considerable  fraction of the  total saturated thickness at
average flow conditions.

The hydraulic  conductivities obtained are a result  of  the varying  contributions of fracture
(conduit) and regional matrix (fissure) system. At low flow conditions it can however be assumed
that the effect  of the regional conduit system is minimal, because  all the flow stems from the
matrix blocks and that the measured gradients are mainly determined by the regional matrix. At
higher flow  and flood conditions the hydraulic conductivities  are dominated by the conduits.
Evaluation of Fracture Contribution to the Hydraulic Conductivity

With a knowledge of the hydraulic conductivity of the regional fissured continuum (Kregmat),
together with an estimation of the hydraulic conductivity of the total system (K   ), a volume
                                                                           reg'
percentage of the conduits per unit volume permeable rock (%frac) can be determined (eq. 1, 2).
                                                           %frac                       (1)
with
                                                 aft,.,
                                     ngcoad
                                                 etc
                                                                                      (2)
This approach however requires information on the hydraulic conductivity of the conduit system,
which can be derived from the average velocity va of tracer breakthrough curves. The arithmetic
average of the relative contributions of the  two systems was used for the calculations (eq. 2).
Because the effective porosity (ne) is unity in fractures, the only unknown in the calculation of
K  ond is the hydraulic gradient in the fractures. It was attempted to to estimate the range of
gradients, likely to occur, using various approaches. The maximum gradients within the fractures
are those of the regional matrix and the minimum gradient is parallel to the aquifer base, derived
from figure 3. Lower gradients are unlikely to occur. With the approach described above, Kre cond
can be narrowed down  to one order of magnitude. Table  1 lists the results of the calculation and
it becomes apparant that the hydraulic conductivity of the  conduits is likely to range between a

                                             45

-------
few  and  approximately  10 m/s, depending on the gradient  and  the  flow  conditions.
Correspondingly, the volume percentage of the fracture contribution to the flow can be estimated
to range between 0.005% and 0.01%, which compares well with the conduit storage evaluated
below.

To further support the Kregcond values obtained, they were compared with those of Gale (1984).
The karstification in the project area however is much less developed, than in the catchments
discussed in Gale (1984). The author analysed the morphology of the conduit-bottom beds (e.g.
scallops),  composed of hydraulically  transported  sediments and  the solutionally developed
bedforms of accessible conduit walls. The author was able to derive mean values of flow velocity
and other hydraulic properties  for turbulent flow. Mean conduit flow  velocities,  which were
partly derived from literature varied between 0.43 m/s and 0.03 m/s. The flow velocities derived
from tracer tests and used for the evaluation of hydraulic conductivity (table 1) range at the lower
end of this spectrum, i.e. between 0.07  m/s and 0.03 m/s. These flow velocities are however an
underestimate of the actual mean velocity in the channels and conduits, because of the tortuous
nature of the flow paths. There  is also a certain bias towards higher flow velocities, because the
conduits have to have a certain  size, in  order to be accessible.
Baseflow Recession Method (Rorabaugh, 1964)

Rorabaugh (1964) showed that the slope of base flow recession curves of water released from
bank storage after flood events plotted on a log-lin diagram (discharge - log) is proportional to
aquifer diffusivity.  With a knowledge of the groundwater basin geometries ( L average distance
to the groundwater  devide) and  the storage coefficient, transmissivities  and hydraulic
conductivities can be evaluated according to the following eq. 3
                                           o  4 S L2
                                      T =
                                                               (3)
with a representing the base flow recession coefficient.
             10
           CD
            >
           O)
           c
           Q.
           C/D
             0.1
                      Recession Analysis
                                     a = 0.017
                                      •I
v\
                                                            I a = 0.25
                         29.8.88
                         Q = 0.27 m3/s
                         67 660.5   /
                         B14 683.5  /
                                                            =.0.0018
                             1988
                         1989
                                                                1990
Figure 5: Discharge recession of the Gallusquelle spring

Atkinson (1977) and Trainer et al (1974) successfully applied this method to evaluate regional
average transmissivities for limestone aquifers. As already discussed,  horizontal variations in
aquifer characteristics cannot be assumed to be detected with this method, because the dependant
variable - discharge - can only be measured at one single point. However, because of the decline
in water level,  a recession  curve reveals information on  lower zones of  the  aquifer  with
decreasing discharge.
                                           46

-------
Figure 5 displays the recession characteristics of the Gallusquelle groundwater catchment over
two and a half years. It can be observed that the recession coefficient decreases from ca. 0.25 d"1
to 0.0018 d"1. It is believed that the value of 0.0018 represents the "pure fissure" and the value of
0.25 the "pure conduit" recession, because in both cases,  the  other flow  component can be
considered negligible. The values of 0.017 and 0.006 can be assumed to be a combination of Kcond
and Krej-ls with varying proportions. The change in slope between a = 0.017 and a = 0.006 occurs
synchroneously with the water level dropping below the highly conductive horizon, identified in
figure 3, during  two  consecutive  years. It is not surprising that the  increase in K is more
prominent than in storage, considering the above explanation.
1

K regfis
"mixed"
"mixed"
K regcond
ilpha

0.0018
0.0060
0.0170
0.25
S

0.010
0.016
0.012
1
L
[m]
11000
11000
11000
11000
T sat.
[m2/dj
8.83E+02
4.71E+03
l.OOE+04
1.23E+07
thick
[m]
50
30
30
15
K
[m/.]
0.0002
0.0018
0.0039
9.0
Table 2: Regional hydraulic conductivities (Rorabaugh, 1964)

Table 2 summarises the results and lists the respective hydraulic conductivities. The matrix value
of 0.0002 m/s corresponds to the matrix value of the highly permeable south-east region, which
is plausible, considering that during low flow conditions that particular region is dominating the
flow due to its higher storage and transmissivity. Further upgradient it can be assumed that the
actual permeable zone of the aquifer is completely drained. The conduit hydraulic conductivity
also corresponds to values determined with the gradient  method. The value for the recession
coefficient was determined in February 1990 during a time period, when the contribution of the
fissures could be assumed to be negligible. This particular event was ideal for such an evaluation
because of its clean and sharp input function. The  mixed values are about twice as high as the
hydraulic conductivities, determined with the gradient method. This might be caused by the fact
that  a fixed  value of saturated thickness  is  used to  calculate  K. Transmissivities compare
somewhat better. Another source for differences could be  that with  the  gradient method,
hydraulic conductivities are determined for a fixed discharge, whereas the hydraulic parameters,
calculated using the recession method, apply to a range of discharges.
Storage

Storage at the regional scale is generally subdivided into conduit storage and diffuse storage.
Conduit storage could be evaluated with the approach suggested by Williams (1983). The values
obtained varied around 0.01% ± 0.01%. Another method employed tracer tests, whereby the dye
was injected into sinkholes, connected to preferential flow paths. The volume of the groundwater,
discharged from the time of injection until the arrival of the dye, devided by the volume of the
saturated rock produced similar values. Similarly, diffuse storage was determined by deviding the
water volume discharged by the aquifer volume drained (regional changes in  water levels).
Diffuse storage can be assumed to vary between 1% and 2%.
Summary

Combining the above results,  regional hydraulic conductivities of the fissured system ranges
between 2*10"5 m/s and 2*10"4 m/s and of the fractures between a few and approximately 10
m/s. The relative proportion of the conduits per unit rock volume most likely ranges somewhere
around 0.01% ± 0.01%. Total conduit and fissure storage ranges between 1% and 2%.
                                            47

-------
LOCAL PARAMETERS

Depending on the type of aquifer and the test duration, the range of investigation of pumping
tests can be expected to be in the order of several hundreds of meters, i.e. the local range. Only
a small number of full scale pumping tests are reported (Archive of the Geologisches Landesamt,
Baden-Wurttemberg), that had been  conducted in or close  to  the project area.  The tested
boreholes are usually located close to the river valleys, where most of the settlements are located,
the risk is lowest to drill a dry hole and where also the drilling costs are lowest because of the
thinner unsaturated zone.
Hydraulic Conductivity

The available pumping test data, which frequently consist of single readings were analysed, using
the iterative method of Walton (1970) for the evaluation of aquifer transmissivity and making
some  assumptions for storage and welloss. The resulting hydraulic conductivity (K,) ranged
between 4*10~5 m/s and 1*10~3 m/s, which is in the range of regional hydraulic conductivities,
determined for the fissures (Kre^18) of the central and the highly permeable south-eastern area
of the Gallusquelle catchment. Test results of a pumping test in compact non-karstif ied limestone
at MeBkirch yielded a hydraulic conductivity of approximately 1*10"6 m/s (Klmat).

Villinger (1988) determined hydraulic parameters for two wells, drilled in the Lauchert valley
near Veringenstadt and obtained hydraulic conductivities, that varied by more than an order of
magnitude although the wells were only 25 m apart. The hydraulic conductivity of the first well
corresponds to the values generally determined in valleys (1*10~4 m/s) and that of the second well
to the Kregfis  of the central low permeable area of the catchment (2*10~5 m/s). A  pumping test
was performed in borehole Bitz (north-west area of Gallusquelle catchment), and a hydraulic
conductivity  of 1*10~5 m/s could be evaluated, employing a double-porosity groundwater flow
model (TRAFRAP-WT, Huyacorn, 1983).


Storage

The pumping test data from Veringenstadt allowed the determination of a storage coefficient of
0.01%, which is  interpreted as the conduit storage  because the highly  permeable fractures
dominate  the flow towards the well. The storage of the fissured system at local scale could be
determined from tracer breakthrough data of a forced gradient borehole-borehole test (Stober,
1991) to range between 1%  and 3%, depending on the  method employed for evaluation.


Summary

The values determined by large scale pumping tests are believed  to  represent the hydraulic
characteristics of the "regional matrix" (Kregfls) (1*10~4 m/s - l*10"s m/s), because their radius
of influence, given a sufficiently long pumping period,  is estimated in the order of several
hundreds of meters using various steady state approximations. The borehole Bitz, situated in the
north-western area of the Gallusquelle catchment reflect the hydraulic characteristics (Krerfig) of
the central area (1*10~5 m/s) and  the results from the wells in the valleys, those  Kregfl"of the
south-eastern area. The storage values correspond to those, determined at regional scale, i.e.
0.01% for  the conduits and  1% - 3% for the fissure system.


SUBLOCAL PARAMETERS

The low conductivity end of the spectrum, not taking  into account laboratory measurements,
could  only be evaluated from boreholes further upgradient. Because the unsaturated thickness is
very high  (=>100m) and the 2.5" casing could not accomodate powerful pumps, it was impossible
to conduct full scale pumping tests. Considering the low hydraulic conductivities that had to be

                                           48

-------
expected within the lower part of the aquifer, tapped by the piezometers, and in view of the the
difficulties  involved  in  assessing high well losses, it was decided  to  conduct slug  tests and
injection tests.
Slug Tests

In the present study, slug tests were conducted by displacing the water within the borehole by
compressed air. The compressed air was supplied via a tank and after the well was put under a
predetermined pressure, no more air was added until equilibrium was reached between well and
formation  head. This could take up to 45 minutes, depending on the  initial change in water
column. Progress of the equilibration could be monitored with two pressure transducers, the first
one was placed below  the lowest expected water level, the second measured gas pressure.  The
compressed air within  the well was allowed to suddenly escape via a large valve, which did not
take longer than approximately 2 seconds,  and the recovery of the  water level was digitally
recorded. The tests were repeated at several different pressures in order to test the dependance
of the test  on the particular conditions of the well (Streltsova, 1988, p. 367).
Well
Test


Displacement
Ksl
Ksl
Skin


[m]



(Cooper, 1967)
(Ramey et al, 1975)
factor
B8
Test 1
1.67
1.2E-04
2.4E-05
4

Test 2
4.61
6.5E-05
1.9E-05
9

Test 3
11.2
bad fit
1.4E-05
17
B17
Test 1
1.1
6.0E-05
5.7E-05
45

Test 2
2.9
7.8E-05
4.6E-05
45

Test 3
6.95
3.2E-05
4.1E-05
45

Test 4
8.74
3.2E-05
4.1E-05
45
Unit

m
m/s
m/s
      K si (Bouwer et al, 1973)     2.3E-05 7.8E-06  2.9E-06 1.8E-05 4.2E-06 3.8E-06 3.7E-06   m/s
      Model Results (TRAFRAP-WT, Huyacorn, 1983)
      K slfis                   5.0E-06 3.0E-06  l.OE-09 l.OE-09 l.OE-09 l.OE-09 l.OE-09   m/s
      K si                     3.0E-05 l.OE-05  5.0E-06 l.OE-05 l.OE-05 7.0E-06 7.0E-06   m/s
Table 3: Hydraulic conductivities, derived from slug test data using different analytical methods

Examples are presented for the tests in borehole B8, because they display the effects of the
double-continuum  system. The results  of  some  typical  tests,  employing  various evaluation
methods, are compiled in table 3.

The data obtained from the tests were first analysed for homogeneous aquifer conditions using
the Cooper et al (1967) method. Reasonable fits could be obtained for all tests, apart from B7,
from about H/H0 = 0.6 down to H/H0 = 0.1. At the beginning, the measured curve systematically
stays below the theoretical one, towards  the end of the tests, the drawdown within the aquifer
might not be negligible anymore. The values obtained varied from 1*10"4 to 1.3*10"5 m/s; storage
coefficients were not determined because they were totally unrealistic. Barker et al (1983) also
point out that they  are usually a gross underestimate, even by a factor of 106.

The same data were also analysed with the method presented by Ramey et al (1975) and the fit
appeared to be somewhat better, with however the same problems, concerning early and late data,
as observed before. The hydraulic conductivity varied between 1*10~5 and 6*10"5 m/s, the skin
factors between 4 and 45. Dougherty et al (1984) presented a sensitivity analysis of the effect of
inner boundary conditions as  there are,  well storage, skin  effect and partial penetration. Well
storage  and skin effect are  accounted  for by  the  Ramey method.  The  neglect of  partial
penetration tends to overestimate hydraulic conductivity by about a factor of 6 - 7, assuming that
only about a quarter of the aquifer thickness is open to the well, which is probably the case in the
tested aquifer.

The data were therefore analysed with  the procedure described by Bouwer et al (1973) and
Bouwer (1989), figure 6a. The tests in  boreholes B8 and B7 were corrected for the initial rapid
increase, which  is  assumed to be caused by  the  fast drainage of the  highly  permeable area
immediately surrounding the  well (Bouwer, 1989).  By evaluating the  volume  of water, that

                                            49

-------
  suddenly entered the well, an equivalent corrected casing radius can be calculated for the initial
  head. The hydraulic conductivities, evaluated with the Bouwer and Rice method (table 3)  were
  as expected much lower and varied from 1*10~6 to 2*10~5 m/s.

  In order to be able to account for the geometry of the  well and its boundary condition, a
  numerical  finite element model  (double-porosity) was  run, to simulate the recovery  data
  (TRAFRAP-WT, Huyacorn, 1983). The fluid exchange between fractures and matrix can be
  modelled  as  transient flow (de  Swaan,  1976)  and  a skin surrounding the matrix blocks
  (interporosity skin) is integrated as well (Moench, 1984). Frequently, a pseudo steady state matrix
  to fissure flow (Barenblatt et al, 1960) represents the field data better, which can be achieved by
  setting the interporosity skin to a high value. The important calibrated model parameters are
  displayed in  table 3. The model results vary for  the tests with the highest initial displacement
  within a very narrow range, i.e. total hydraulic conductivities (Kgl, fissures + conduits) between
  0.3*10~B m/s (B14) and 0.7*10~5 m/s (B17). The fit between model and field data is generally
  good (e.g. figure 6b). Initial fluctuations of the test data could be either caused by the system
  itself that reacts underdamped or by the test procedure. The compressed air  requires a finite time
  to escape and a standing  wave might develop with compression and decompression cycles. Most
  of the tests could be simulated without any major matrix contribution (Kglfjs = 1*10~9 m/s,  sub-
  local fissure hydraulic conductivity). Only in tests B8  1 and B8_2, where the initial displacement
  was  lower, higher Kslfl8-values  (5*10~6 m/s -.  l*10"5ln/s) were used to improve the fit at early
  and  late times.
| 2',

  1

 0.5


 0.2

 0.1

0.05


0.02

0.01
                             TestBB 3
                0.01
                                   A13 Response o
                         A12 Response of f latrb
                    _At1 Response or the_Fract ires
                  0.1    10   1000  100000
                        Time (s)
 iTest 1
 iSlope 0.032
••|R6=T.6rrii	
 iK = 2.3E;5 m/s
Slugtest, B8, Bouwer & Rice
        200   400   600   800   1,000  1,200 1,400
                       Time [s]

  Figure 6: Estimation of hydraulic conductivity from slug test data
                                                           Model Simulation
                                                           (slug test data, B8 3)
                                                           Time [min]
  In trying to assess how representative the above tests are for the aquifer in question and to assist
  in allocating  an appropriate scale of investigation to slug tests, it is important to determine a
  radius of influence. Ramey et al (1975) and Sageev (1986) investigated the radius of influence of
  slug tests and developed appropriate diagrams. A critical parameter in this assessment is the scale
  of resolution of the recovery measurements, which however can be easily dealt with by using
  pressure transducer readings. For resolutions of H/H0 < 0.01 and a CD value (dimensionless well
  storage, Sageev, 1986, pi328) of 100, which the result is  not very sensitive to, the radius of
  investigation  is 1000 times that of the well radius, i.e. approximately 10 - 20  m.

  It was mentioned earlier on that the hydraulic conductivity  of  the  tests is  dependant  on the
  displacement depth, i.e. the duration  of the  test.  This feature,  which is attributed  to  the
  heterogeneous characteristics of the matrix/fracture system, has also been described by Streltsova
  (1988), who observed it whenever "the formation volume, influenced by the test is smaller than
  the representative formation volume  required  for fracture pattern replication" (p.366). The
  following conceptual model might explain this hydraulic behaviour. If the well is drilled within
                                               50

-------
a compact matrix block, the hydraulic response of an aquifer section is mainly determined by the
low matrix conductivity near the well. If however the well is connected to the fracture network,
short tests will only excite the highly conductive fractures and the longer the duration of the test,
the more the less conductive  matrix will contribute to the test result. Therefore a systematic
decrease in the hydraulic  conductivity with  increasing test  duration (displacement)  can be
observed.
Injection Tests

Two kinds of injection tests were used for parameter estimation. The first ones, conducted within
the context of the study consisted of the injection of water into cased boreholes (B7, B25, B14,
B8). It  was  attempted  to  obtain steady state  conditions  and  in  order to  circumvent the
inconsistencies in flow during the injection phase, only the recovery period was used for analysis.
The second kind  of test are  double  packer tests in  an uncased hole near the village Bitz
(unpublished data, Landesanstalt fur Umweltschutz, Karlsruhe), investigating vertical hydraulic
conductivity distribution. The profile within the saturated zone  was not complete because of
packer leakage and therefore did not sample the sections with high hydraulic conductivities. The
Deutsche Bundesbahn also kindly gave permission to use their data from double-packer tests,
performed in a series of boreholes in the eastern Swabian Alb, drilled for investigatory purposes
for a planned tunneling project.

The tests in boreholes  B7, B8, B14, B25  were  analysed  first using  the traditional  evaluation
method for Lugeon tests (e.g. Houlsby, 1976). Another standard method is the Theis recovery or
Horner plot, where the section of the time drawdown curve is used for the analysis when radial
flow conditions can be assumed. Barker (1981)  developed an analytical technique, specifically
designed for tests in fractured rocks. Transmissivity is evaluated by an iterative method using eq.
4
                                       2 * h
                                             In
                                                      - I)
                                                C r
                     (4)
with C as a constant (1.781), Kh and Kv the horizontal and the vertical hydraulic conductivity of
                                                        MJti-Rate Type Ctrve Andy sis Plot
                               Q = 1.5 l/s
                               As = 2.6 m
                               b = 6m
                               K= 1.76E-5m/s
                                                   V        tf        tt       tf
                                                  	Dimensioriess Time Group,
                    345         10
                    Time Function (t+At)/At

Figure 7: Estimation of hydraulic conductivity from injection test data
                                                                        k-B34/nd»—4.t«> OU30008bl/psi
                       K)'
At, hr
Figure 7a displays the Horner plot for borehole B8. The test B8 displays a delayed recovery,
which is caused by water running down the side of the steel casing. By the time the drop of the
water level in the borehole is slower than the speed of the water seeping down the casing, the rate
                                            51

-------
in head drop is inversed. The results are however not influenced to a major degree, because the
effect ceases before the onset of the infinite acting radial flow period. The S-shaped curve, a
typical characteristic of double porosity aquifer tests, also becomes apparant in test B8.

Test B8 was also analysed employing  the  well test analysis program STAR, developed by
Schlumberger. The field data could be fitted without using any skin, on the contrary, a negative
skin of -4 was obtained from the model fit (figure 7b). The peak observed in  the pressure
derivative at 0.1 hr is caused by the above described delayed seepage down the side of the casing.

In sum, hydraulic conductivities for the injection tests in B7, B8, B14 and B25 vary between
5*10~6 m/s and 5*10"5 m/s, depending on the borehole tested. The values are generally higher
than those obtained from slug tests. Assuming a regular fracture network in three dimensions and
a distance between fractures of 0.2 m, the matrix block shape factor a (Warren et al,  1963) can
be calculated. Together  with the knowledge of A.,  the interporosity  flow parameter and the
hydraulic conductivity of the fractures, which could be obtained from the  tests, the matrix
hydraulic conductivity (Kslflg) can be estimated to range between 5*10"7 m/s and 5*10~8 m/s.

The  radius of investigation can be calculated according to Streltsova (1988, p79, eq. 6), with R4
as the radius of investigation r\ as the diffusivity and t the flow period. A and c2 are constants,
that  depend on the definition of the radius and the system of units used. The radius of influence
calculated with the above method, with c2 being unity and A equal to 4.781, can be determined
between 10 and 20 m depending  on the diffusivity (m2/d) and the  time (d) used.
                                     Rt = A J c2 T] t                                  (5)
Test Results and Hydraulic Parameters of Double Packer Tests

The test sections  in borehole Bitz were preferentially located in horizons where fractures and
solution channels could be detected, using a selection of borehole logs and borehole television
inspection. However, highly productive zones could not be tested because of packer leakage and
the limited injection rate of the testing equipment. This explains hydraulic conductivities ranging
between 1*10"5 m/s and only 3*10"4 m/s.

Double packer tests in the eastern Swabian Alb yielded hydraulic conductivities between 5*10"3
m/s and 1*10~8 m/s with a maximum located between 1*10"6 m/s and 5*10"5 m/s.


Summary

At a sublocal scale, the system (conduits + fissures) hydraulic conductivity varies within a very
narrow range (1*10~5 m/s and 1*10"6 m/s). The slug and injection tests deliver fissure hydraulic
conductivities, ranging between 10"7 m/s and 10"9 m/s, which correspond to the low range Kglfis
of the double packer  tests. Because  of  the specific test setup,  no storage values could be
determined.
LABORATORY SCALE MEASUREMENTS

In order to complete the spectrum of hydraulic conductivities, measured at various scales, some
data could be found in WeiB (1987), which however only cover the bedded facies of the carbonate
series of the Upper Jurassic (Franconian Alb). A very definite maximum was measured between
10"8 m/s and 10"9 m/s for bedded limestone. For dolomite beds, the hydraulic conductivities
varied from 10"4 m/s to 10~8 m/s at the measured scale of 3 cm diameter cores. It is however
difficult to take samples of the massive karstified limestone. Hydraulic conductivities can be
assumed to vary by many orders of magnitudes, in the extreme  case no  porous or fractured
medium is encountered at all, if e.g. a "sample" is taken in the centre of a cavity. Storage values
were determined to range in the order of approximately 3%.

                                            52

-------
 CONCLUSIONS

 Figure 8 displays the karst system hydraulic conductivities of all the hydraulic tests and the
 regional evaluation. It is apparant  that the hydraulic conductivity increases with an increasing
 scale  of investigation. Lowest values are determined  in the laboratory and  the highest on a
 catchment scale. The values are categorised into a socalled "common  range", i.e.  hydraulic
 conductivities, that are most frequently measured and "low conductivities", which usually give an
 indication of the hydraulic conductivity of the fissure system at the respective scale. Among
 double packer tests, the higher values are usually underrepresented, which is generally a result of
 the testing equipment with its limited injection  rate.
w
5,
g
">
ta
3
T3
C
O
O
g
~B
2 1.0E-08
TD
><
   1.0E-10
                1.0E-12
                1.0E-02
                1.0E-04
                1.0E-06
                                           B
                                           B

                                 Low Conductivities (measured)
                                 common Range  Scale of Investigation [m]
                       0.01
                               0.1
                                       1
                                             10
                                                    100
                                                           1000
                                                                 10000
               Range of
               Investigation
                                 Laboratory Measurements
                                    Double Packer Tests
                                    - Slug/injection Tests
                                           Pumping Tests
                                                Regional Evaluation
 Figure 8: Relationship between hydraulic conductivity and scale of investigation

 It is interesting to note that measurements at the next lower scale seem to reflect the hydraulic
 characteristics of the fissures  at the next higher scale. Slug tests and injection tests produced
 fissure hydraulic  conductivities  of between  1(T6 m/s  and  1(T9 m/s,  which  correspond
 approximately to the total hydraulic conductivity  measured in the laboratory. The "common
 range" of the laboratory tests are probably an underestimate, because samples have only been
 taken from the bedded facies.  This range can probably be extended by one order of magnitude
 to ID'  m/s. The fissure system values, resulting from slug and injection tests (Kslflg) correspond
 to the  permeabilities, measured in less karstified, i.e. fissure zones of the double packer tests.
 Average hydraulic  conductivities measured  between 1*1(T5 m/s and 1*1(T6 m/s by slug tests
 correspond to fissure system values of pumping tests (Kmg), identified as hydraulic conductivities
 of unkarstified limestone. A similar relationship applies, if the pumping test results are compared
 with the regional fissure system values (Kregflg).

 Figure 9 graphically summarises the above observations, and shows that hydraulic conductivity
 (fissures + conduits) of the lower scale can serve as input for  fissure hydraulic conductivity at the
 next higher scale. Storage  values do not vary to a major degree and can be  estimated at 001%
±0.01% for the conduits and at 1% to 2%  for the fissures.

These findings have  been used as input for a regional double-continuum flow and transport
model,  which demonstrates  its usefulness  in  problems  concerning  groundwater resource
evaluation and contaminant risk assessment.
                                            53

-------
                                2000m
                                           Regional
                                           K reg: = 10E-3 m/s - 10E-4 m/s
                                           S reg: = 0.015
                                           % frac: = 0.0001 - 0.0003
                                           K regcond: « 3 m/s -10 m/s
                                           S regcond: = 1
                                           K regfls: 1*10E-4 m/s - 1*10E-5 m/s
                                           Local
                                           Pumping Test
                                           K I: = 1*10E-4 m/s - 1*10E-5 m/s (10E-3 m/s - 10E-6 m/s)
                                           S I: « 0.01 - 0.02
                                           % frac: = 0.0001
                                           K Icond: 0.01 m/s -10 m/s ?
                                           S Icond: 1
                                           Klfis:10E-6m/s?
                                           Sublocal
                                           Slug/Packer/Injection Test
                                           K si: = 1 '10E-5 m/s - 5*10E-6 m/s (10E-5 m/s - 10E-6 m/s)
                                           S si: 0.02 ?
                                           % frac: 0.0001 ?
                                           K slcond: = 0.03 m/s - 0.1 m/s
                                           S slcond: 1
                                           K slfis: 10E-7 m/s - 10E-9 m/s
                                           Laboratory
                                           K lab: «= 10E-8 m/s - 10E-9 m/s (< 10E-11 - > 1 m/s)
                                           S lab: «= 0.03 ( 0 - > 0.12)
Figure 9: Geometrical relationships and hydraulic conductivities at different scales

REFERENCES

Atkinson, T.C., 1977, Diffuse flow and conduit flow in a limestone terrain in the Mendip Hills,
       Somerset, J.o.Hyd., 19, 323-349.
Barenblatt, G.E., Zheltov, I.P. & I.N. Kochina, 1960, Basic concepts in the theory of homogene-
       ous liquids in fissured rocks. J. Appl. Math. Mech. (USSR), 24, 1286-1303.
Barker, J.A. & J.H. Black, 1983, Slug Tests in Fissured Aquifers, Water Res. Res., 19, 1558-1564.
Barker, J.A., 1981, A formula for estimating fissure transmissivities from steady-state injection
       test data.  J.o.Hyd., 52, 337-346.
Bourdet, D., Whittle, T.M. Douglas, A.A. & Y.M. Pirard, 1983b, A new set of type curves simpli-
       fies well test analysis. World Oil, May 1983.
Bourdet, D.  & A.C. Gringarten, 1980, Determination of fissure volume  and block size in
       fractured reservoirs by type curve analysis, presented at  the SPE-AIME 55th Annual
       Technical Conference and Exhibition, Dallas, Texas. Society of Petroleum Engineers,
       SPE-9293.
Bouwer,  H., 1989, The Bouwer & Rice slug test - An update. Ground Water, 27, 304-309.
Bouwer,  H.  & R.C. Rice,  1976,  A slug-test for determining  hydraulic  conductivity with
       completely or partially penetrating wells. Water Res. Res.,  12, 423-428.
Castany,  G., 1984,  Hydrogeological features of  carbonate rocks, in:  Unesco, Guide to the
       hydrology of carbonate rocks, Studies  and Reports in Hydrology, vol 41.
Cooper, H.H., Jr., Bredehoeft, J.D. & I.S. Papadopulos, 1967, Response of a finite diameter well
       to an istantaneous charge of water. Wat. Res. Res., 3, 263-269.
Department of Energy, 1986, Environmental assessment, reference repository location, Hanford
       site, Washington, DC, US. DOE/RW-0070.
Domenico, P.A. & F.W. Schwartz, 1990, Physical and chemical hydrogeology. Wiley, N.Y., 824p.
Dougherty, D.E.  & O.K. Babu, 1984, Flow  to a partially penetrating well  in a Double Porosity
       Reservoir, Water Res. Res., 20,  1116-1122.

-------
Gale, S.J., 1984, The hydraulics of conduit flow in carbonate aquifers. J.o.Hyd., 70, 309-327.
Houlsby, A.C., 1976, Routine interpretation of the Lugeon water test. Q.J.Eng.Geol., 9, 303-313.
Huyakorn,  P.S., Lester, B.H. & C.R. Faust,  1983, Finite Element techniques for  modelling
       groundwater flow in fractured aquifers. Water Res. Res., 19, 1019-1035.
Karasaki, K, Long, J.C.S. & P.A. Witherspoon, 1988, Analytical models of slug tests. Wat.Res.
       Res., 24, 115-126.
Kiraly, L., 1975,  Rapport sur 1'etat actuel des connaissances dans le domaine des characteres
       physiques des roches karstiques. In: A. Burger & L. Dubertret (eds.), Hydrogeology of
       karstic terrains. Int. Union Geol. Sci., Ser. B, 3.
Moench, A.F., 1984, Double-Porosity models for a fissured groundwater reservoir with fracture
       skin. Wat.Res.Res., 20, 831-846.
Narasimhan, T.N. & K. Pruess, 1988, MINC: an approach for  analysing transport in strongly
       heterogeneous  systems,  in: E. Custodio, A. Gurgui  & J.P. Lobo  Ferreira  (eds.),
       Groundwater flow and quality modeling, Reidel Dordrecht.
Ramey, H.J. jr., Agarwal, R.G. & I.  Martin,  1975, Analysis of "slug" test or DST flow period
       data. J.Can.Petr.Technol., 37-47.
Rorabough, M.I., 1964, Estimating changes in bank  storage as ground-water contribution to
       streamflow. Int. Ass. Sci. Hydrology, Pub. 63, 432-441.
Sageev, A., 1986, Slug test analysis. Wat. Res. Res., 22, 1323-1333.
Sauter, M., 1991, Double porosity models in karstified limestone  aquifers, Proc. Int. Symp. Field
       Seminar on Hydrogeological Processes in Karst Terranes,  7. - 17.Oct.1990, Antalya, Tur-
       key.
Sauter, M., (in  prep.), Quantification and forecasting of regional groundwater  flow and
       transport in the Gallusquelle karstified limestone aquifer, (SW. Germany): Ph.D. thesis,
       Universitat Tubingen.
Snow, D.T., 1965, A parallel plate model of fractured permeable media. Ph.D. thesis, Univ. of
       California, Berkeley, 330p.
Stober, I., 1991, Strdmungsvorgange und Durchlassigkeitsverteilung innerhalb  des WeiBjura -
       Aquifers  im baden-wiirttembergischen Anteil des  Molassebeckens. Laichinger
       H6hlenfreund, 26, 29-42.
Streltsova, T.D., 1988, Well testing in heterogeneous formations. Wiley, N.Y., 412p.
de Swaan, A.O., 1976, Analytic solutions for determining naturally fractured reservoir properties
       by well testing.  Soc. Petr. Eng. J., 16, pi 17.
Teutsch,  G., 1988, Grundwassermodelle  im  Karst:  Praktische Ansatze  am Beispiel  zweier
       Einzugsgebiete im Tiefen und Seichten Malmkarst der Schwabischen Alb. Ph.D. Thesis,
       Universitat Tubingen, 205p.
Teutsch, G., 1989, Groundwater models in karstified terraines - two practical examples from the
       Swabian Alb, S. Germany. Proc. of the  4th Conference - Solving Groundwater Problems
       with Models,  Indianapolis, USA, Feb. 7-9 1989, 929-953.
Teutsch,  G. & M. Sauter, 1991, Groundwater modeling in karst terranes: Scale effects,  data
       aquisition and field validation. 3rd Conf. on Hydrogeology, Monitoring and Managment
       of Ground Water in Karst Terranes, Nashville, Dec. 4-6 (this issue).
Trainer, F.W. & F.A. Watkins, jr., 1974, Use of base-runoff recession curves to determine areal
       transmissivities  in the Upper Potomac river basin. USGS J. Res., 2, 125-131.
Villinger,  E.,  1988, Hydrogeologisches Gutachten uber die  weiteren Ergebnisse der
       Wassererschlieflung im Gewann Stetten bei Veringenstadt, Lkr. Sigmaringen. Unpubl.
       report, Geolpgisches Landesamt Baden-Wurttemberg, 4763-6/88 SIG.
Villinger, E., 1977, Uber Potentialverteilung und StrOmungssysteme im Karstwasser der Schwabi-
       schen Alb (Ob. Jura, SW-Deutschland). Geolog. Jahrbuch, CIS, 92p.
Walton, W.C., 1970, Groundwater resource evaluation. Me. Graw Hill, Kogakusha, 664p.
Weifi, E.G., 1987, Porositaten, Permeabilitaten und Verkarsrungserscheinungen im mittleren und
       oberen Malm der Sudlichen Frankenalb. Ph.D. Thesis, Universitat Erlangen, 240p.
Yilin, C., Hongtao, W. &  X. Xinhui, 1988, Dual-media flow models  of karst areas and their ap
       plication in north  China. Karst Hydrogeology and Karst Environment Protection, Proc.
       of the 21st Congress of the Int. Assoc. o.Hydrogeologists, Guilin, China.
                                          55

-------
M. Sauter is research associate at the Chair of Applied Geology at the University of Tubingen. In
1980 he obtained a diploma in Geology from Tubingen University and in  1981  an M.Sc. in
Hydrogeology from Birmingham University. Between 1982 and 1987 he was involved in various
projects in Germany and overseas, covering areas of groundwater contamination, geophysical well
logging and water resources development. He is presently finalizing his Ph.D. project on regional
flow and transport modeling and model validation in karst terranes. His main interests lie in the
area of  groundwater flow and transport modeling  in  karst, hydraulic  testing methods and
hydrochemistry.
                                           56

-------
ASSESSMENT OF HYDRAULIC CONDUCTIVITY IN A KARST AQUIFER
AT LOCAL AND REGIONAL SCALE
Martin Sauter

Question: Rane Curl has determined that many caves (perhaps most) are fractal objects which have
a determinable fractal dimension. Caves are permeability (conductivity) paths of a scale. Your work
seems to be approaching the problem of scale-variant parameters from a broader perspective. Could
you comment on possible connections between Curl's work and yours?

Curl, R.L. 1986, Fractal dimensions and the geometry of caves: Mathematical Geology 18:765-784.


Answer: The fractal approach assumes that the same fractal dimension exists everywhere within the
investigated domain and that the principle of self-similarity applies at a small and at a large scale.
Fractals are typically scale-invariant. In order to achieve this self-similarity at all scales of permeable
pore space, the process, generating the hydraulic conductivity should not vary in space and time. As
far as caves are concerned, they usually form within a geologically short period and are the result of
a single process. The aquifer studied, formed over a period of millions of years since the Miocene
with the solution process modified by the glacial period,  valley erosion and backfill and global and
small  scale  tectonics, which  makes self-similarity in porespace geometry and therefore hydraulic
conductivity unlikely. The scale hierarchy approach  does not assume self-similarity  and scale-
invariant behaviour but attempts to recognize the different genetic  history of small and large scale
permeabilities (cf fissured basalt flows and highly permeable flow tops, in Dept. of Energy, 1986,
DOE/RW-0070)
                                             57

-------
58

-------
                       APPROACH FOR DELINEATING THE
                     CONTRIBUTING AREAS OF A WELL FIELD
                       IN A CARBONATE-VALLEY AQUIFER
                      By Gary Barton and Dennis Risser

              U.S. Geological Survey, Water Resources Division
                                Lemoyne,  Pa.
ABSTRACT

     The contributing area of a well that penetrates saturated sediments and
some sedimentary rocks can be generally estimated by use of principles of
flow in granular porous media. Unfortunately,  ground-water flow in bedrock
aquifers commonly occurs in fractures and conduits that poorly approximate
granular porous media at the well-field scale. The most common methods that
can be used to evaluate whether flow in a fractured-rock aquifer
approximates flow in a granular porous medium include (1) borehole
geophysical logging and mapping, (2) monitoring of ground-water levels, (3)
slug tests, (4) dye tracing,  (5) monitoring of ground-water-quality
fluctuations,  and (6) constant-discharge aquifer tests. If conduit flow
exists, the same methods can aid in estimating the contributing area to a
well or well field.

     The six methods were used to show that flow in a well field in a
fractured-rock aquifer in Nittany Valley, Pa., approximates flow in granular
porous media under nonpumping and pumping conditions. The well field
contains 12 wells in the Nittany Dolomite and Axemann Limestone of
Ordovician age. Fractures and conduits are approximately evenly distributed
in all directions throughout the well field and are interconnected.

     The six methods show that delineating contributing areas to a well
field in fractured rock can be based on methods that assume a granular
porous medium. However, delineating the contributing area to the well field
is difficult because of the absence of observation wells located upgradient
of the field,  the heterogeneity of the aquifer, and poor definition of
ground-water and surface-water interactions. Because the methods used in
this study were based on fixed-radius assumptions, they do not provide
realistic contributing areas as do the uniform-flow and semianalytical
equations and ground-water-flow modeling which integrate hydraulic gradient
and aquifer heterogeneity into the solution.
                                    59

-------
INTRODUCTION

     As part of the Pennsylvania Department of Environmental Resources's
(PaDER) wellhead-protection program, the U.S. Geological Survey (USGS)
evaluated methods for delineating the contributing area of supply wells in
typical hydrogeologic settings throughout Pennsylvania. This report provides
an approach for delineating the contributing area of a well field in a
fractured carbonate-valley aquifer and evaluates the fixed-radius methods
that can be used to delineate contributing areas to a well field within the
valley. The aquifer is in the Spring Creek drainage basin within the Nittany
Valley, central Pennsylvania. Topography at the well field is hummocky with
vertical relief of approximately 70 ft (feet).
HYDROGEOLOGY

     The Nittany Valley is located in the Appalachian Mountain section of
the Valley and Ridge physiographic province.  Topography is characterized by
a succession of prominent northeast-trending ridges and valleys.
Hydrogeologic settings similar to that of the Nittany Valley are common
throughout the folded Appalachian Mountains.  Ten formations,  which range
from late Cambrian to late Ordovician age, crop out in the study area
(fig. 1). Upland areas are synclinal and consist primarily of resistant
shale, sandstone, and quartzite.  The Nittany Valley is anticlinal, and
soluble carbonate rock crops out in the valley floor (Parizek,  1971).

     The Houserville well field is located in the central part of the Spring
Creek basin. Basin area is approximately 175 mi2 (square miles).  Three
large-capacity supply wells completed in the Nittany Dolomite are owned and
operated by Pennsylvania State University and comprise the well field. These
wells are 1,250 to 1,950 ft apart (fig. 2). Nine wells, completed in the
Nittany Dolomite and the Axemann Limestone, are within 2,400 ft of the
supply wells (fig. 2). The Nittany Dolomite is 1,200 ft thick,  and the
Axemann Limestone is 400 ft thick (Wood, 1980, pi.  1). Both formations
strike approximately north 45° east and beds are tilted 15° to the
southeast. Transmissivity of the Nittany Dolomite is roughly one- to two-
orders of magnitude higher than that of adjacent formations (table 1). No
obvious relation exists between topographic position of the wells and
horizontal hydraulic conductivity. A cross fault extends through the center
of the well field and offsets the Nittany Dolomite and Axemann Limestone by
700 ft. Supply and observation wells are located on both sides of the fault.

     Most ground water in the well field flows to the northeast along
strike. The hydraulic gradient of the water table on May 16,  1990, was
0.001. In the broader area surrounding the well field, 4 mi (miles)
upgradient and 2 mi downgradient of the well field, the hydraulic gradient
was 0.003 in April-May. 1985 (Albert Becher,  U.S. Geological Survey, written
commun.,  1991). The water table at the well field does not reflect
topography. At the well field, Spring Creek and Slab Cabin Run are perched
above the water table (fig. 1). Average discharge of Spring Creek is
approximately 60 ft3 (cubic feet per second) at gaging station 01546400
located 1.0 mi downstream from the well field (U.S. Geological Survey,
1991).
                                     60

-------
          Pennsylvania  State
               University
                                                                            2 MILES
           GEOLOGIC UNITS

       Ob - B«llefont» Dolonltc
       Oa - Ax«ncn Lln«itonc
       Ot - Tr«ncon LinesCone
       On - Nitcany Colonies
       OB • Stonehenge Llmescone
                                         EXPLANATION
€w   Warrior Llmeitone
Oo -  Oswego Sanditone
€g -  Gacesburg Formation
OJ •  JuniiCa Formation
Or -  Reedsvllle Shale
 .PS35
                                                                              riLOMETEIS
        SUPPLY WELL AND LOCAL IDENTIFIER

	PERENNIAL STREAM

O^CE-2 SPRING AND IDENTIFIER

       • THRUST FAULT WITH TEETH ON SIDE
         OF UPPER PLATE.
         DASHED WHERE INFERRED
Figure 1.—Geology  and  the Houserville  well  field  in Spring  Creek basin.
              (From Parizek,  1971,  figure  6.)
                                                 61

-------
     77°50'32"
                                                                          77°49'53'
                                                    NITTANY   I
                                                    DOLOMITE  /
                                                          AXEMANN
                                                         LIMESTONE   II
40°49'11"
           1.05
          PS35
    EXPLANATION

SUPPLY WELL, LOCAL IDENTIFIER, AND
 DRAWDOWN AT END OF 72 HOURS IN FEET
                                                          500
                                                           I
1.000 FEET
                                                      100
                                                              200
                                                                     300 METERS
            1.34 OBSERVATION WELL,  LOCAL IDENTIFIER, AND
          • OB 6  DRAWDOWN AT END  OF 72 HOURS IN FEET

          .f)  Q _-DRAWDOWN IN FEET AT END OF 72 HOURS
                  COUNTOUR INTERVAL 0.2 FEET
                ORIENTATION OF PRIMARY FRACTURE SETS



                - CROSS FAULT


                 STRIKE AND DIP OF INCLINED BEDS
     Figure 2.—Well  network and geology at the  Pennsylvania State
         University Houserville well  field  and drawdown at  the end of
         a 72-hour aquifer  test pumping well PS33 at 630 gallons
         per minute.  (Geology by  Parizek, 1971, figure  6.)
                                        62

-------
                  Table 1.--Summary of hydraulic properties
                            [ft2/d, square feet per day; gal/min, gallons
                            per minute;  —,  no data;  NA,  not applicable]

Geologic Transmissivity
unit(s) (ft2/d)
Trenton 1 2 15
Limestone ^ * 560
Bellefonte 1 40
Dolomite 5 240
6 196 - 1,320 6
Trenton and
Bellefonte 7 1,000 7
Limestones,
and Reedsville
Shale, undivided
Axemann * 200
Limestone 8 3,700 and 7,100
7 5,000 7
Limestones 5 860
Ordovician Age
Nittany 9 3,800
Dolomite * 5,200
9 120,000
8 23,000 160,000
5 11,000
6 44,919 247,059 6
7 100,000 7
Stonehenge 1 80
Limestone 9 7,600
7 2.000 7
Gatesburg * 2,000
Formation 9 2,700
9 5,000
5 11,000
6 37,032 56,486 6
7 50,000 7

Storage
coefficient
3 .015
3 .015
3 .015
.00027 .00062
.003 .009



3 .015
.003 .009
--

9 .008
3 .015
3 .015
—
--
.003 .05
.003 .009
3 .015
9 .08
.003 .009
3 .015
9 .04
9 .04
—
.0003- .0006
.003 .009
Pump
discharge
(gal/min)
10
100
50
--
NA



100
NA
NA
—

200
500
1,000
NA
--
—
630 2400
50
500
NA
200
500
1,000
—
--
NA
1 Transmissivity based on specific capacity according to the method of Meyer, (1963) (Becher, U.S.
Geological Survey,  written commun.,  1991).
2 Data for the Coburn through Nealmont Formations (Wood, 1980, pi. 1), which are equivalent to the
Trenton Limestone (Parizek,  1971,  fig.  6).
3 Transmissivity based on specific yield (Becher, U.S. Geological Survey, written commun., 1991).
* Data for the Benner through Loysburg Formations (Wood, 1980, pi. 1), which are equivalent to the
Trenton Limestone (Parizek,  1971,  fig.  6).
5 Transmissivity based on specific capacity (Siddiqui and Parizek, 1971).
6 Transmissivity from Rauch and White (1971, table 2).
7 Data from transient simulation of three-day constant-discharge aquifer tests.
8 Transmissivity base'd on slug tests at the Houserville well field.
9 Transmissivity based on aquifer-test data (Becher, U.S. Geological Survey, written commun., 1991)
                                               63

-------
GROUND-WATER DISCHARGE AT AND NEAR THE HOUSERVILLE WELL FIELD

     Well PS35 (fig.2) is the only active well at the Houserville well field
and is pumped intermittently at 1,000 gal/min (gallons per minute); average
pumping rate during 1990-91 was 254 gal/min (John Gaudlip, Pennsylvania
State University, oral commun.,  1991). Wells PS33 and PS34 are a permitted
source of supply with a maximum withdrawal rate of 1,200 gal/min per well.
Springs CE-2 and CE-27 are 4,200 and 3,800 ft (fig. 1) south-southwest of
the well field in the Nittany Dolomite. Springs CE-2 and CE-27 discharge
about 540 and 100 gal/min, respectively, and are a permitted source of
supply at 175 and 35 gal/min. Spring CE-1 in the Axemann Limestone is
7,200 ft southwest of the well field and discharges about 4,000 gal/min.


METHODS FOR EVALUATING THE HYDROLOGIC RESPONSE OF THE AQUIFER TO NONPUMPING
AND PUMPING CONDITIONS

     One of the major issues in fractured-rock hydrology is whether the
ground-water flow through fractures in bedrock approximates flow in a
granular porous media. Hydrologic testing was conducted to address this
issue at the scale of the Houserville well field.  Results are used to
determine whether (1) flow occurs through multiple water-bearing fractures
and conduits that are sufficiently interconnected to approximate a granular
porous media (referred to as an interconnected fracture-conduit flow
system), or (2) flow occurs through a single or a few fracture(s) and
conduit(s) that are poorly connected and do not approximate a granular
porous media (referred to as a conduit-flow system).  Conduits are solution-
enlarged fractures that are capable of transmitting large volumes of water.


Borehole Geophysical Logging and Mapping

     Fluid flow in fractures and conduits is controlled by several factors,
including the density, length, orientation, aperture, surface roughness,
spatial distribution, and connectivity of the fractures and conduits (Barton
and Hsieh, 1989). The relation of these factors to fluid flow in a
fractured-bedrock aquifer is complex and difficult to quantify (Gale, 1982).
However, these factors are typically and most thoroughly measured by use of
borehole geophysical logging and mapping. Measurements in many cases are
limited to a fraction of the entire fracture and conduit network.
Information obtained about fractures and conduits by use of conventional
borehole geophysics (Paillet and Keys, 1984) is generally limited to less
than 5 ft beyond the borehole annulus. The highly experimental downhole
ground penetrating radar has been used to successfully map fractures and
their orientations at more than 300 ft from the borehole and tomographically
between boreholes (John Williams, U.S. Geological Survey, oral commun.,
1991). Information obtained by mapping is limited to the surface of a rock
exposure. Despite limitations of measurement and analysis, conventional
borehole geophysical logging and mapping are robust methods when used to
characterize the control that fractures and conduits impart on fluid flow.

     Numerous fractures and conduits detected in boreholes by caliper logs,
along with abundant measurements of fractures where bedrock crops out
indicates that the Houserville well field is a connected fracture-conduit
                                     64

-------
aquifer system. Borehole geophysical logging was conducted in all wells,
except in wells PS35, OB4, and OB9,  chiefly to identify fractures and
conduits and to estimate their apertures. Caliper, single-point-resistance,
gamma-ray, fluid-temperature, fluid-resistance and brine-trace logs, and
borehole video tapes were made (Paillet and Keys, 1984). Numerous fractures
and conduits were identified in all production and observation wells.
Caliper logs from all 9 wells show 7 to 22 fractures per 100 ft of borehole.
Fracture aperture ranged from 0.1 to greater than 1.5 ft. Fractures on
bedrock that crop out in the well field were mapped, chiefly to identify the
orientation of fractures. Two major sets of fractures were identified and
trend approximately 30° to the strike of bedding (fig. 2) along with some
minor sets of fractures with different orientations. The distance between
individual fractures of each major set is approximately 1 to 2 ft. Some of
these fractures extend below land surface and into the aquifer. Assuming two
major sets of fractures intersect each other at many points,  the
connectivity of the fractures and conduits is expected to be high.
Monitoring of Ground-Water Level

     A comparison of water-level fluctuations in the 11 wells monitored
could indicate the relative extent of connections between fractures and
conduits within the aquifer. An interconnected fracture-conduit network is
likely to produce water-level fluctuations similar to those that could be
expected in a granular porous media aquifer. Anomalous fluctuations in some
wells could indicate a fracture-conduit network in which parts of the
aquifer are poorly connected or isolated from each other. Seasonal water-
level fluctuations in the well field were nearly identical in all 11 wells.
Similarity in the phase, period, and amplitude of rises in water level
caused by recharge suggests that all wells intercept fractures and conduits
that are interconnected (Rorabaugh, 1960). The similarity in slope of water-
level recessions in all wells indicate that the aquifer's hydraulic
diffusivity is about the same throughout the well field.  The response time
of water levels to recharge can help evaluate whether the fracture-conduit
network acts as an interconnected fracture-conduit flow system or a conduit-
flow system (Shuster and White, 1972). In response to 2.8 in. (inches) of
precipitation, water-levels in all wells rose about 8 ft over a 20-day
period. This water-level fluctuation is not flashy as is typical of a
conduit-flow system, but gradual and more like an interconnected fracture-
conduit flow system. Some water-level fluctuations are caused by the cyclic
pumping of supply well PS35 in the Houserville well field (fig.  3). Water
levels in all wells respond rapidly to pumping and are most rapid and
greatest near the pumped well.
Slug Tests

     Slug tests provided a rapid and inexpensive method of estimating the
heterogeneity of aquifer hydraulic properties. The tests were analyzed by
methods that account for the inertia of the mass of water in the well and
aquifer (van der Kamp, 1976; Kipp,  1985). The median transmissivity is at
79,000 ft2/d (feet squared per day), and ranges from 23,000 to 160,000 ft2/d
wells completed in the Nittany Dolomite.
                                    65

-------
              931.4





            UJ 931.2

            UJ


            <
              931.0
            UJ
            "I 930.8

            t-
            LU
            UJ

            "• 930.6
              930.4
            tr
            UJ

            < 930.2
              930.0
                 28
                            29
   30

AUGUST 1990
                                                  31
                                    EXPLANATION

                                  PUMPING AT WELL PS35
Figure 3.—Water-level  fluctuation in  four observation wells caused

   by intermittent  pumping at well PS35.
            OS  3
          tot
          flit
           I a.
            o
          00
                     PUMPING RATE AT WELL PS33 - 830 GALLONS PER MINUTE
                      TIME 0 - RHODMINE-WT INJECTED INTO WELL OB8
                     TIME 930 - RHODAMINE-WT INJECTED INTO WELL OB9
                           500         1,000         1.500

                                 TIME, IN MINUTES
                           2.000
  Figure 4.—Dye-recovery  curve  for forced-gradient  dye-tracing tests.

-------
     Because the slug tests were conducted at wells having different open-
hole lengths, horizontal hydraulic conductivity is a better indicator of the
heterogeneity of the fracture system in the aquifer than transmissivity.
Horizontal hydraulic conductivities at five wells completed in the Nittany
Dolomite range from 800 to 4,500 ft/d (feet per day). The horizontal
hydraulic conductivities of the Axemann Limestone at wells OB2 and OB5 are
100 and 200 ft/d, respectively. The moderate to large hydraulic
conductivities at the seven wells tested indicate that these wells are
capable of yielding substantial amounts of water that can only be supplied
by water-bearing fractures intercepting the well bores. Therefore, slug
tests indicate that fractures are approximately evenly distributed
throughout the well field. The variability of hydraulic conductivity in the
well field indicates that the number and spacing of fractures and conduits
that intercept the wells are variable.
Dye Tracing

     Forced-gradient dye-tracing tests (Mull and others, 1988; Quinlan,
1989) were conducted during a 72-hour constant discharge aquifer test in
which well PS33 was pumped at 620 gal/min. Rhodamine-WT dye was injected in
observation wells OB8 and OB9 equidistant (120 ft) from the pumped well
(fig. 2).  The mean velocities for dye traveling from wells OB8 and OB9 are
1,030 and 720 ft/d, respectively. Dye break-through curves (fig. 4) show
about the same time-of-travel for dye during both tests, and the curves are
smooth, bell-shaped, and positively skewed.  Therefore, water-bearing
fractures and conduits appear to be well connected within 120 ft of the
pumped well.
Monitoring of Ground-Water-Quality Fluctuations

     Variations in ground-water quality can distinguish interconnected
fracture-conduit flow systems from conduit-flow systems in carbonate-valley
aquifers. Ground-water specific conductance during recharge events in
interconnected fracture-conduit flow systems is typically constant or
gradually increases and decreases with discharge, but in a conduit-flow
system, specific conductance fluctuates rapidly during and after recharge
(Shuster and White, 1972, p. 1069; Jacobson and Langmuir, 1974, p. 261).
Ground-water specific conductance was monitored continuously in well PS33
for 7 months. Specific conductance remained constant during recharge events
(fig. 5), providing additional evidence that the Nittany Dolomite is an
interconnected fracture-conduit flow system and behaves as a granular porous
medium at the well field.
Constant-Discharge Aquifer Tests

     Three 72-hour constant-discharge aquifer tests were conducted at 7 to
11 observations wells:  Well PS35 was pumped at 1,000 gal/min, well PS33 was
pumped at 630 gal/min, and wells PS33 and PS34 were pumped simultaneously at
2,400 gal/min (Todd Giddings and Associates, Inc., 1989). Water was
discharged into Spring Creek, a stream perched above the water table.
                                    67

-------
     BOO
    500
ALTITUDE OF GROUND-WATER TABLE


            \   XV   /\


         /V '
                                       SPECIFIC CONDUCTANCE

                                              \
                                                        sJ
                             8         10


                             AUGUST, 1990
                                                              L1.5
                                                              -\
                                        1
                                        2




                                        f
                                        P
                                    14
                                                                     r931
                                                                     -930
Z


J
                                                                     -929
                                                                     -92B
                                                                         LJ
                                                                         O
                                                                     -926
Figure 5.—Water-level  altitude and  specific conductance of water

   in observation well  OB8, and amount  of precipitation, August 4-14,1990.
                                     68

-------
Qualitative Analysis

     The shape of the drawdown curves shows that declines in water levels
did not stabilize during the aquifer tests (fig. 6a, 6b, and 6c).  Therefore,
the aquifer's water budget did not approach a new equilibrium during the
pumping. A new equilibrium can only be established if the volume pumped is
balanced by decreases in ground-water discharge to surface-water bodies,
such as Spring Creek and Slab Cabin Run and springs CE-2 and CE-27 (fig. 1) ,
or an increase in recharge from surface-water bodies such as Spring Creek
and Slab Cabin Run or both. Apparently, ground water withdrawn during these
tests was chiefly derived from storage.

     The aquifer test at well PS33 involved monitoring water levels in 11
wells that surround the pumped well. Drawdown decreased uniformly in all
directions with increasing distance between the pumped well and observation
wells (fig. 2). The cross fault extending through the well field did not
appear to be a barrier to, or a conduit for,  ground-water flow. The lagtime
between pumping and drawdown in well PS35, located on the opposite side of
the cross fault as pumped well PS33, is similar to that in wells located on
the same side of the cross fault as pumped well PS33.
Quantitative Analysis

     The shapes of the log-log (fig. 6) and semilog drawdown curves from all
three aquifer tests do not match any of the "classic" ideal drawdown curves
such as those for (1) an unconfined, confined, or leaky, unconsolidated,
homogeneous and isotropic aquifer; (2) a confined, fractured, double
porosity aquifer; or (3) a single vertical fracture in a confined aquifer.
The shape of drawdown curves commonly deviates from the shape of ideal
curves because of recharge and impermeable boundaries within the pumped
well's area of influence, well-bore storage, the partial penetration of
wells (Kruseman and deRidder, 1990, p. 48-53), and interference from other
pumped wells. The latter three conditions are not observed on the curves
from the three tests; however, the shape of the drawdown curves are probably
affected by boundary conditions. The nature of the boundary condition(s)
could not be identified from the drawdown curves. Therefore, in order to
simplify quantitative analysis of the aquifer-test data, a transient finite-
difference flow model (McDonald and Harbaugh, 1988), which integrated the
complex hydrogeology of the carbonate-valley aquifer into the solution, was
used. The transient simulations were used to (1) estimate an average
transmissivity and storage coefficient for the Nittany Dolomite and Axemann
Limestone because a wide range of values have been reported for both
formations (table 1), and (2) to estimate the areal extent of the area of
influence at the end of 72 hours of pumping, so that boundary conditions may
be better identified.

     For the sake of simplicity, the water-table aquifer was simulated as a
confined aquifer because drawdown is less than 10 percent of the aquifer's
saturated thickness (Franke and Reilly, 1987, p. 19). The model area is
72 mi2 (10.6 mi by 6.8 mi). Grid-cell dimensions are either rectangular or
square and range from 100 ft to 2,000 ft on a side. Grid cells along the
perimeter of the model were specified as no-flow cells because southeastern
and northwestern model boundaries extend to and parallel the base of Nittany
                                     69

-------
             o
             o
               o.oi
                    —i—i i i urn	1  i  i i inn	\—i  i i mi|    i  i i 111u
                     (•)         *                                    —
                     DISCHARGE - «30 GALLONS f£R MI HUT E



                     SIMULATED DftAl/DOUN:

                      	066

                      * * * OB2



                     MEASURED OO(Ut£CTEO DRAUDOUH:
                               10
                                          100
                                                                  10,0 0 0
10


^_
UJ
UI
U. 1

z
z
*
o
o
»
< 0.1
DC
Q

n n 4
- TTI nmi i i i mill i i i DIIIJ i i i um
DISCHARGE - 10OO CALUWS FOL MINUTE ~
XUOJIATED DBAUDOUK: A^""" —
— •— O17 i-* P*
•"" -^V°P'
KEAMfn^IKm, BU»KM<. ^. _
omi ^4 _.<> -
,'>* / :
-^A* P*'
S A .0
x a
" 'A .°' ~
^ o ~
-
IP
4
                                          100
                                                      1,000
                                             10,000
                 10
             I-
             Ul
             UI
             U.
             *
             o
             o
             *
             <
             a:
             a
O

2
                     (e)l  1 1 1 II If)    1  1  1 MIMt   1


                     COtfBUEO DISOUICB - 5400 CUUQHS PEK MINUTE
SUOHATEB

 _ _0«*

 • • • DBS


*f«»«™" oulmolaf

         •nd Aasoc.. lae.. 1919):
                               1 Illlj   1
                                                                 Iltt

                               10          100         1(0 0 0

                                    TIME. IN MINUTES
                                             1 0,0 0 0
Figure 6.  -- Measured and  simulated drawdown for  72-hour constant-discharge

               aquifer  tests pumping  (a)  well PS33,  (b)  well  PS35,  and  (c)

               wells PS33 and PS34  simultaneously
                                            70

-------
Mountain and Bald Eagle Mountain, and the southwestern and northeastern
boundaries are in the valley bottom (fig. 1). Internal boundaries were not
specified in the model because Spring Creek and Slab Cabin Run are perched
near the well field. Sensitivity analysis aided in assessing the effects of
no-flow boundaries in the valley bottom and also aided in estimating the
hydraulic parameters used in the system analysis. No-flow boundaries in the
valley bottom were evaluated by substituting constant-head boundaries. This
substitution did not affect drawdown near the pumped well. Transmissivities
used in the model are listed in table 1.

     The three aquifer tests were simulated by use of transmissivities of
100,000 and 5,000 ft2/d for the Nittany Dolomite and Axemann Limestone,
respectively. The storage coefficient was assumed to be uniform for all
formations. Pumping from wells PS35 and PS33 were simulated by use of a
storage coefficient of 0.003 and the simultaneous pumping of wells PS33 and
PS34 was simulated by use of a storage coefficient of 0.009. The need for
additional storage to simulate the simultaneous aquifer test may be caused
by heterogeneities in the aquifer. The storage coefficients used in all
simulations are considerably smaller than the specific yield of 1.5 percent
for the carbonate aquifer underlying Spring Creek basin (Giddings, 1974;
Albert Becher, U.S. Geological Survey, written commun.,  1991).  Storage
coefficients derived from short-term aquifer tests in unconfined aquifers
are commonly less than the specific yield determined from long-term aquifer
tests (Nwankwor and others,  1984).

     The hydrographs simulated for each aquifer test (fig. 6a,  6b, and 6c)
are generally in good agreement with measured aquifer-test hydrographs,
especially for days 2 and 3 of each aquifer test. Simulated water levels
show that pumping stress is propagated primarily within the Nittany Dolomite
and Axemann Limestone for a few miles upgradient and downgradient from the
well field (fig. 7). The area of influence encompasses approximately
3.5 mi2, including springs CE-2 and CE-27, a 0.7-mi stretch of Slab Cabin
Run, and 3 mi of Spring Creek.
APPROACH FOR DELINEATING AREAS OF CONTRIBUTION

     Delineating contributing areas of the Houserville well field is
difficult because of the absence of observation wells located upgradient
(southwest) of the well field, the large heterogeneity of the Nittany
Dolomite (table 1), poor definition of ground-water-flow directions, and
poor definition of ground-water and surface-water interactions. A
hydrogeologic model was developed and used to delineate the area of
contribution to supply wells. Results of simulation of the 72-hour aquifer
test pumping from wells PS33 and PS34 simultaneously at a combined rate of
2,400 gal/min, indicate that the area of influence (cone of depression) (1)
develops primarily where the Nittany Dolomite and Axemann Limestone crop out
because the adjacent Stonehenge Limestone and Beliefonte Dolomite act
largely as barriers to flow  (fig. 7). (2) is elliptical, and (3) extends
into areas not monitored during the aquifer test. Borehole geophysical
logging and mapping, monitoring of ground-water levels, slug test, constant-
discharge aquifer test, dye  tracing, monitoring ground-water-quality
fluctuations, and simulation of aquifer tests show that the aquifers
supplying water to the Houserville well field respond as granular porous
                                    71

-------


                                          Bald Eagle Mountain
     TRANSIENT FLOW MODEL BOUNDARY
         Pennsylvania State
            University
                  D

                    Sl»"
                                                        r->
                                                                   Jfv
                                              Nlttany Mountain
                                                                     2 HUES
                                                                     '
                                                                   3  KILOMETERS
                                                                                      •£

Figure  7
    wells
                                           EXPLANATION

                                1S33  SUPPLY WELL AND LOCAL IDENTIFIER

                           	 0.3	LINE OF EQUAL DRAWDOWN OF WATER TABLE

-—Simulated  drawdown after  72-hours  of  simultaneously pumping
 PS33 and PS34 at 2,400  gallons  per minute.
                                                        EXPLANATION

                                        PERCENTAGE OF COINCIDENT AREA

                                        K - HYDRAULIC CONDUCTIVITY,  FEET PER DAY
                                        1 - HYDRAULIC GRADIENT
                                        b - AQUIFER THICKNESS, FEET
                                        t - TRAVEL TIME, DAYS
                                        6 = AQUIFER POROSITY
                                        Q - PUMPING RATE, GALLONS PER MINUTE
                                            K-300, 1=0.001, b-300, t-365, e-0.015, Q-254
                                            K-300, 1=0.003, b-300, t-365, e-0.015. Q-254
                                            K-300, 1=0.001, b-300, t-365, a=0.015, Q-3,400
                                            K-300, 1=0.003, b-300, t-365, a=0.015, Q-3,400
                   10
               bt/(GQ)
                                                   »
                                                   •
                                                   O
                                                    ASSUMPTIONS:

                                                    (1) Flow Is two dimensional and steady;
                                                    (2) Aquifer is homogeneous, isotropic,
                                                    and extends infinitely in all directions;
                                                    (3) Hydraulic gradient is uniform
     Figure 8.—Percentage  of  time-of-travel  area coincident  between
         that delinated  using the  fixed-radius method  and  a method that
         includes the water-table  slope.
                                           72

-------
medium during nonpumping conditions and under short-term pumping stress.
Thus, delineation of contributing areas of the wells by methods that assume
a porous medium (Morrissey, 1987) can be made by estimating some or all of
the following hydrologic characteristics:  hydraulic gradient, direction of
ground-water flow, aquifer thickness, porosity and horizontal hydraulic
conductivity, ground-water and surface-water interactions, recharge areas,
rate of recharge from precipitation, pumping rate, and pumping duration. If
the hydrologic methods had indicated that flow in the fractured rock did not
approximate flow in a granular porous medium, the same methods could aid in
estimating a contributing area. Delineation of contributing areas by the
fixed-radius method are summarized below.
Delineation of Contributing Area

     The fixed-radius method is commonly used to estimate a time-of-travel
(TOT) diversion zone of a well. The method is based on estimating the
cylindrical aquifer volume through which ground water moves towards a well
for a given magnitude and duration of pumping (U.S. Environmental Protection
Agency, 1987, p.  4-6). The TOT diversion zone is the projection of this
cylinder boundary at the land surface. Aquifer characteristics used in the
calculations are listed in figure 8. In the case of supply wells PS33, PS34,
and PS35 pumped simultaneously, discharge from each well is summed to
represent an equivalent single well being pumped at a rate of 3,400 gal/min.
The radius of the 365-day TOT diversion zone for pumping 254 gal/min from
well PS35 is 1,125 ft and for pumping 3,400 gal/min is 3,400 ft. The
percentage of a 365-day TOT diversion zone delineated by the fixed-radius
method that coincides with methods that account for a sloping water table1
is small:  10 percent when pumping PS35 at 254 gal/min with a hydraulic
gradient of 0.003, and 18 percent when pumping 3,400 gal/min (fig. 8). The
fixed-radius method is poorly suited for application to this well field
because the method cannot account for the sloping water table. This fixed-
radius method delineates an overly large zone of diversion down-gradient
from the supply well(s) and too little upgradient.
1 Methods that delineate the contributing area(s) of a well and account for
a sloping water table include use of the uniform-flow equation (Bear, 1979,
p. 282), semianalytical equations such as used in the Wellhead Protection
Area model (Blandford and Hyukorn, 1989),  and ground-water-flow modeling.
Ground-water-flow modeling is the most powerful method because it is capable
of simulating complex hydrogeology; however, it also requires the greatest
amount of data.
                                    73

-------
CONCLUSIONS

     Borehole geophysical logging, water-level monitoring, slug tests, dye
tracing, monitoring of variations in ground-water quality, and constant-
discharge aquifer tests indicate that the aquifer supplying water to the
well field in a fractured carbonate-valley aquifer in Nittany Valley, Pa.,
respond as granular porous media during nonpumping and pumping conditions.
Fractures and conduits are more or less evenly distributed throughout the
well field and are interconnected. Because the aquifer responds behaves as
if it were a granular porous medium,  areas of contribution can be delineated
by techniques used for porous media.  Uniform-flow and semianalytical
equations and ground-water-flow modeling delineates contributing areas more
realistically than does the fixed-radius method because the equations and
flow modeling can account for hydraulic gradient and aquifer heterogeneity.
REFERENCES CITED

Barton, C.C., and Hsieh, P.A.,  1989,  Physical and hydrologic-flow properties
     of fractures:  Proceedings of 28th International Geologic Congress,
     36 p.
Bear, Jacob, 1979, Hydraulics  of groundwater:  New York,  McGraw-Hill, 569 p.
Blandford, T.N.,  and Huyakorn,  P.S.,  1989,  An integrated semi-analytical
     model for the delineation of wellhead protection areas:   U.S.
     Environmental Protection  Agency contract 68-08-003.
Franke, O.L., and Reilly, T.F., 1987,  The effects of boundary conditions on
     the steady-state response of three hypothetical ground water systems-
     results and implications  of numerical experiments:   U.S. Geological
     Survey Water Supply Paper 2315,  19 p.
Gale, J.E.,  1982, Assessing the permeability characteristics of fractured
     rocks,  in Narasimhan, T.N.,  ed.,  Recent trends in hydrology, Geological
     Society of America, Paper 189,  p. 163-181.
Giddings, T.M., 1974. Hydrologic budget of Spring Creek drainage basin,
     Pennsylvania:  unpublished doctoral thesis,  Pennsylvania State
     University,  76 p.
Jacobson, R.L. and Langmuir, Donald,  1974,  Controls on the quality
     variations of some carbonate spring waters:   Journal of Hydrology,
     v. 23, p. 247-265.
Kipp, K.L.,  1985, Type curve analysis  of inertial effects in response of
     well to a slug test:  Water Resources Research, v.  21, no. 9,
     p. 1397-1408.
Kruseman, G.P., and deRidder,  N.A.,  1990, Analysis and evaluation of pumping
     test data:  International Institute for Land Reclamation and
     Improvement, publ. 47,  376 p.
McDonald, M.G., and Harbaugh,  A.W.,  1988, A modular three-dimensional
     finite-difference ground-water flow model:   U.S. Geological Survey
     Techniques of Water-Resources Investigations, book 6, chap. Al, 586 p.
Meyer, R.R., 1963, A chart relating well diameter, specific capacity, and
     the coefficients of transmissibility and storage, in Bentall,  Ray,
     compiler, Methods of determining permeability, transmissibility, and
     drawdown, U.S. Geological Survey Water-Supply Paper 1536-1, p. 338-340.
                                    74

-------
Morrissey, D.J., 1987, Estimation of the recharge area contributing water to
     a pumped well in a glacial-drift aquifer:  U.S. Geological Survey Open-
     File Report 86-543, 60 p.
Mull, D.S., Liebermann, T.D.,  Smoot, J.L., and Woosley, L.H.,  Jr., 1988,
     Application of dye-tracing techniques for determining solute-transport
     characteristics of ground water in karst terranes:  U.S.  Environmental
     Protection Agency Report EPA 904 per 6-88-01, 103 p.
Nwankwor, J.A., Cherry, J.A.,  and Gillham, R.W.,  1984, A comparative study
     of specific yield determinations for shallow sand aquifer:  Ground
     Water, v. 22, no. 6, p. 764-772.
Paillet, F.L. and Keys, W.S.,  1984, Application of borehole geophysics in
     characterizing the hydrology of fractured rocks:  NWWA per USEPA
     Conference on surface and borehole geophysical methods in ground water
     investigations, San Antonio, Texas, Proc. p. 743-761.
Parizek, R.R., 1971, Hydrogeologic framework of folded and faulted
     carbonates, in Parizek, R.R. and others, ed., Hydrogeology and
     geochemistry of folded and faulted rocks of the central Appalachian
     type and related land use problems:  Pennsylvania State University,
     Cir. 82, 212 p.
Quinlan, J.F., 1989, Ground-water monitoring in karst terranes:  recommended
     protocols and implicit assumptions:  U.S. Environmental Protection
     Agency Report, EPA 600 per X-89 per 050, 79  p.
Rauch, H.W.,  and White, W.B.,  1971, Lithologic controls on permeability, in
     Parizek, R.R., and others, ed., Hydrogeology and geochemistry of folded
     and faulted rock of the central Appalachian type and related land use
     problems:  Pennsylvania State University, Cir.  82, 212 p.
Rorabaugh, M.I., 1960, Use of water levels in estimating aquifer constants
     in a finite aquifer:  International Association of Scientific
     Hydrology, publ. 52, p. 314-323.
Shuster, E.T. and White, W.B., 1972, Source areas and climatic effects in
     carbonate groundwaters determined by saturation indices and carbon
     dioxide pressures:  Water Resources Research, v. 8, no. 4,
     p. 1067-1073.
Siddiqui, S.H. and Parizek, R.R., 1971, Hydrogeologic factors  influencing
     well yields in folded and faulted carbonate rocks in central
     Pennsylvania: Water Resources Research, v. 7, no. 5, p. 1295-1312.
Todd Giddings and Associates,  Inc., 1989, The Pennsylvania State University
     wells 33 and 34:  unpublished consultants report, 70 p.
U.S. Environmental Protection Agency, 1987, Guidelines for delineation of
     wellhead protection areas:  EPA 440 per 6-87-010.
U.S. Geological Survey, 1991,  Water resources data for Pennsylvania, water
     year 1990--volume 2, Susquehanna and Potomac River basins: U.S.
     Geological Survey Water-Data Report, PA-90-2, 266 p.
van der Kamp, Garth, 1976, Determining aquifer transmissivity by means of
     well response tests:  the underdamped case,  Water Resources Research,
     v. 12, no. 1, p. 71-77.
Wood, C.R., 1980, Summary of ground water resources of Centre County.
     Pennsylvania:  Pennsylvania Geological Survey Water Resources Report
     48, 60 p.
                                    75

-------
       APPROACH FOR DELINEATING THE CONTRIBUTING AREAS
         OF A WELL FIELD IN A CARBONATE-VALLEY AQUIFER

               Gary Barton and Dennis Risser
What you  find with respect to aquifer properties depends upon how
you look.  Without tracer  tests from sinkholes  you  will  not
understand the conduit flow component.  Your  information is useful,
but only partly  describes the aquifer.  How can you justify  using
your data for wellhead protection when your study is so deficient
in analysis aided by tracers?

This project, as indicated in the background section of this
paper,  is charged with  evaluating methods  for  delineating
contributing areas  to supply  wells and  not with wellhead
protection.  The  data collected at the Houserville  well field can
be used to delineate a  generalized contributing area as long as
the accuracy of  the delineation is indicated. Tracers injected in
sinkholes (sinkholes have not been identified at the well field)
or observation wells upgradient of the  well  field and recovered in
the  active  supply well PS35  would increase  the  overall
understanding of the ground water flow  system and the contributing
area of the well field. However, capturing a tracer in supply well
PS35 during a forced-gradient test may  be difficult,  considering
that the  contributing area near and at the well  field is very
narrow (aquifer  is highly transmissive  and has  a large hydraulic
gradient)  and therefore the tracer could bypass the pumping well.

Borehole geophysical logging was used to evaluate  for fracture
concentration,  fracture orientation,  and fracture thickness. What
type of borehole geophysical log or logs were  run  to give such
information? Were the geophysical logs run in open holes,  water-
filled holes,  air-filled holes, and/or  mud-filled holes?

Borehole  geophysical  logging was used  to  estimate fracture
concentration and gross thickness (aperture) and was not used to
measure fracture orientation.  Characterization of fractures was
primarily based  on caliper logs.  Also, fluid resistance,  fluid
temperature, single-point resistance, and  gama-ray logs were
recorded.  The geophysical logs were collected in air-filled and
water-filled boreholes.

During the month of August,  the specific conductance measurements
did not vary.  You state  that this indicates interconnecting
fractures and conduits. Wouldn't the opposite  conclusion  be
expected  during times of recharge --  because of the influence of
lower-conductivity meteoric water?

Please  note that during the presentation I  indicated that this is
one of several possible interpretations. I believe this question
is ambigious and can not respond with a clear  answer.
                              76

-------
      EFFECTS OF QUARRY DEWATERING ON A KARSTIFIED LIMESTONE AQUIFER;

                A CASE STUDY FROM THE MENDIP HILLS, ENGLAND


                   A.J.  EDWARDS,  S.L.  HOBBS+ & P.L.  SMART,

               Department of Geography.  University of Bristol
                             Bristol, England

                   +Aspinwall and Co. Ltd, Walford Manor
                            Shrewsbury,  England
Abstract

The effects on conduit flow of dewatering, associated with development of  a
large  limestone  quarry  in  a  karstified  aquifer,  are  evaluated  using
Lycopodium spores and  fluorescent  dye tracing studies.  Point to point  and
quantitative tracer tests are used to define the conduit network, while  the
nature of the conduits is indicated by travel time/discharge  relations.  The
latter  is  strongly dependent  on structure,  strike  conduits  being vadose,
while  flow  up  dip  involves  hydraulically  inefficient  looping phreatic
routes. During quarry deepening intersection of conduits by the  quarry void
has  not occurred  (although  this  has occurred  elsewhere).   Dewatering is
shown to induce leakage  from  the conduit into the diffuse flow  zone. Where
conduits are  vadose  and/or distant  from the quarry, leakage is minor  and
the  conduit  function is  unaffected.  Where  conduits  are phreatic, leakage
prevents downstream propagation  of hydraulic head over  high  points in  the
looping  conduit,   and  the  conduit  becomes  non-functional.   Storm-related
increases in conduit flow both increase  leakage into the diffuse flow zone,
and  can cause reactivation of the  conduit,  as  shown by  increased tracer
recovery at springs
Introduction

The Carboniferous Limestone of the Mendip Hills is extensively  exploited to
supply the demand for aggregates in southern England.  Annual production is
currently  20  million  tonnes,  much  of which  is  abstracted from  a  small
number of  relatively  large quarries concentrated  in the East Mendip  area.
Until recently extraction  was wholly  from above the water-table, however 4
quarries are  now actively dewatering  the aquifer.  Water  is abstracted by
pumping from  sump  pools in the  lowest working levels,  it is then  ejected
into perched  surface  streams  which flow  out of the area.  Working  quarries
                                    77

-------
extending below  the water-table act  like  pumped large diameter  boreholes.
As the  level  of  operation deepens, there  is  a progressive increase  in the
associated drawdown and a widening of the  cone  of depression.  The resultant
lowering of the water-table can have  serious  effects upon  surface hydrology
and  public  water  supplies.  For  example,  mining  activity  in  Hungarian
karstic limestone  dried  up  boreholes, springs  and  streams over  an area  of
2500 km  (Ford &  Williams,  1989).   There is therefore a need  to  assess the
response of  the  aquifer to  this  abstraction,  and  an  extensive  network  of
observation boreholes and stream  gauges  has been established by  the  quarry
companies  and the National  Rivers  Authority  (which  is  responsible  for
protection  of  the  groundwater  resources)  to  monitor   the   effects  of
dewatering.   The  National  River Authority needs  to  know the   effects  of
proposed  dewatering  schemes   so   that   they  can  vet   suggested   quarry
extensions  and propose  remedial  measures  if  needed.   At present  it  is
extremely difficult  to  predict  the exact  nature  and effects of  dewatering
karstic  aquifers  due   to   the  heterogeneity  associated  with  flow.    In
homogeneous aquifers, the extent of the  cone  of depression and the required
pumping  rates can be  calculated  using  modified  Thiem-Dupuit  equations
(Driscoll,  1986).   These  equations  however assume homogeneity  in hydraulic
conductivity  and  laminar  flow  throughout  the  whole  aquifer.    Within
karstified Carboniferous Limestone, the  position  and function of a conduit
can  substantially alter the  nature of  the drawdown  surrounding a  quarry
(Atkinson et  al,  1973a)  and seriously affects calculations which are based
on homogeneous aquifers.  In  this  paper  we examine a conduit network which
is  adjacent  to   an operating  sub-water-table  quarry  and  elucidate  the
changes  in   the   function   of   this   network  which  have  resulted  from
dewatering, using  information obtained from tracer studies.
The Study Area

Torr Quarry, owned by  Foster  Yeoman Ltd,  is a large sub-water-table  quarry
on the southern limb of the Beacon Hill pericline at the eastern  end  of the
Mendip Hills,  England  (Figure 1).   The quarry has  an area  of 1.1 km2,  a
current working depth  of  50-60 m, and  an  annual  production of 6-8 million
tonnes.  The original  rest-water  level  in the area was approximately  150  m
ADD, but in the winter of  1987,  the quarry sump  was deepened to  130 m ADD,
and  subsequently  pumping  levels  have  been  between  135   and  140  m  ADD.
Average daily abstraction is  5.5 ML/d, but  this may vary between  0.7  and 31
ML/d depending upon the weather conditions.

The Carboniferous  Limestone  comprises massive, well-bedded,  pure, fine  to
coarse grained  limestones  of  low primary porosity.   In the study  area the
limestones dip  at  30°  to  40°   to  the  south, and are overlain unconformably
to  the  east by the  Jurassic  Inferior Oolite, a  massive,   fissured,  coarse
bioclastic and  ooidal  limestone  some 15 m  thick.   The two  limestone  units
are in  hydraulic  continuity,   and provide  the  major water supply for the
town  of  Frome  to  the  north-east.    At   the  base  of  the Carboniferous
Limestone  a  shaly  unit  (the  Lower Limestone  Shales) is  transitional  to
clastic sediments  of  the Devonian  Old  Red Sandstone,  which together with
Silurian volcanic  rocks  forms the  core  of the pericline  (Figure 1).   The
Old Red  Sandstone forms  the   highest  land,  rising  to 288 m  ADD.    Water
emerging from the  Old  Red Sandstone forms  small  streams,  which  sink  after
flowing onto  the  Carboniferous Limestone  across  the  Downhead  Fault.   The


                                     78

-------
             1/4
                     1/2    Mile
     0
      h-
     0         1/2         1 Kilometer

      Generalised Vertical Section
     o-
    50-


   100-

 in
 £ 150H
 CD


 S. 200 H
 OJ
 Q

   250



   300-


   350-
Jurassic Clays


Inferior Oolite Limestone
Unconformity
Carboniferous Limestone
Lower Limestone Shales


Old Red Sandstone

Unconformity
Andesite
      Key

£~rv>3 Quarry  —— — Fault         f Dip of limestone

—»— Conduit network (schematic)

  - *** Stream ----Piped stream 	Non-functional link

	9 Stream sink            • Spring

  'A   Borehole locations :T1 - pe - Torr Quarry boreholes
                        SF   - Shute Farm (Shute 1)
                        T     Tunscombe
                       Mf   - Manor Farm (Neilson  1)
                       WQ   - Westdown Quarry
                        A    - Ashley 1
                       Carboniferous
                       •  Limestone
                           Series
Figure 1.     Geology,  hydrology  and  hydrogeology  of  the  study area.
                                                79

-------
main stream sinks  are  at elevations between 181 and 223 m ADD and comprise
Downhead Swallet, Dairyhouse  Slocker, Bottlehead  Slocker  and Heale Slocker,
the latter being the nearest to Torr Quarry.   Atkinson et  al (1973b) have
previously traced  these  stream  sinks to Seven Springs  in  the Whatley Brook
(128 m  ADD).    Although  the  swallets  remain active  throughout  the  year,
Seven Springs  ceases to  flow in  the summer  months,  and  the  Whatley Brook
becomes influent at this point.
The Swallet/Rising Conduit Network

Tracing  studies  using Lycopodium  spores  (Atkinson et  al,  1973b)  and more
recently fluorescent  dyes  (Smart et al,  1991),  demonstrate that  flow from
the swallets  to  Seven Springs  is  via conduits.   Velocities are  high,  and
the  dye  breakthrough curves   show  a  characteristic  form  with  a  rapid
increase in concentration  to the peak and  a slower exponential  decline on
the falling limb  (Figure 2).   The minor  peaks are related  to  backflooding
of the spring due to  the intermittant release  of  abstracted  water from Torr
Quarry into the Whatley Brook.
            &
            I
            o
            0
            c
            a
            v
            B
            cc
                                                            -20
                            234567
                              Time (days after injection)
Figure 2.   Tracer breakthrough  curves  at  Seven Springs.   (Main Spring) for
            Rhodamine WT  injected at Downhead Swallet  and Bottlehead
            Slocker, and  Fluorescein injected at Dairyhouse  and Heale
            Slockers (Tests  2  and 4).
                                    80

-------
The pattern of tracer recovery  (Table  1)  can be  used to  suggest the conduit
network shown in Figure  1  (detailed  arguments  are presented in Smart et al,
1991).   Downhead Swallet,  Dairyhouse  Slocker and  Heale Slocker have  been
traced  to  both  the  Main  and  Pineroot  Spring at  Seven  Springs,  while
Bottlehead Slocker has been proven only to the main spring.   The  latter is
thus  inferred to  be a  completely  separate   link,  while  the other  three
swallets  are  tributary  to  a  single  conduit.   However,  while  Dairyhouse
Slocker shows complete  recovery of  the  tracer at  Seven Springs  (Table 1) ,
Downhead Swallet has much  lower recoveries (30 to 50%) , and  a distributory
upstream of the junction with the  Dairyhouse Slocker conduit  must  therefore
occur.   This distributory  is  shared  with the  Heale conduit,  which  like
Downhead  Swallet,   has  been  traced  to  Westdown  Quarry  borehole  using
Lycopodium  spores  (Atkinson et al,  1973b),  indicating underflow below the
Whatley Brook.   The final  outlet  for  this conduit  is  not proven,  but the
.consistantly  100%   recovery  of  tracer  from  Dairyhouse   Slocker  casts
considerable  doubt on  the  positive result  at Holwell  Rising  previously
reported by Atkinson et  al  (1973).   It  appears more  likely that this spring
drains the area of limestone to the  south east of the Whatley Brook.

Preliminary results  of  time of travel/spring discharge relationships  for
the individual swallets  have been used to infer if  the  conduits are water-
filled  (phreatic)  or  have  an  open water  surface  (vadose)   (Smart,  1981;
Smart et al, 1991).

The Downhead  Swallet conduit  is  essentially  vadose  (gradient log  time  of
travel/log  spring  discharge 0.5),  as  is  the  Dairyhouse  Slocker  conduit
(gradient 0.7).  The higher figure  for  Dairyhouse  infers that part of the
conduit  prior to  the  Downhead conduit  junction  is phreatic.    Evidence
presented below  will demonstrate  that the  Heale  Slocker  conduit  is  deep
below the water-table  and thus phreatic  at  least  adjacent to  Torr Quarry.
This accords  well  with expectations based  on  the  known structural control
of cave  passages (Ford  &  Ewers,  1978).   Strike directed flow may utilise
bedding planes which are laterally continuous  in the direction of  flow, and
allow development of solution conduits with relatively minor  loops  (state 3
or water-table cave  of Ford &  Ewers,  1978).   Conversely  flow down-dip,  or
in  the  case  of  Heale   and   Bottlehead  Slockers   up-dip,   must   utilise
infrequent and discontinuous joints  to pass through  the  dipping limestones.
This  results  in deep  looping  phreatic  passages (state  1 or  2  of Ford &
Ewers,  1978), which are  hydraulically  much   less  efficient  than  strike
conduits  because  of their  greater  length and the  high  probability  of
sediment constrictions  at the  bottom  of  the  loops.   This may explain why
the Lycopodium spore trace results show lower  velocities  for  Bottlehead and
Heale Slockers with  a high  up-dip  conduit component, compared to  Downhead
and  Dairyhouse   (Table   2A) .    Farrant  (1991)  illustrates  precisely  this
structural control in the newly discovered Cheddar Springs conduit.
Influence of Torr Quarry Dewatering on Conduit Function

In the preceeding  section  the results of our own tracing  studies  and those
of Atkinson et al  (1973b) have been used  jointly  to  elucidate  the  nature of
the conduit  network.   Our  tests  were carried out  in 1987 (Tests  1-4)  and
1990  (Test  5)  after sub-water-table  working  had  commenced in  Torr Quarry.

                                      81

-------
            Table  1A    Details  of Traces by Atkinson  et  al  (1973b)
Injection
Site
Downhead
Dairyhouse
Bottlehead
Heale
Tracer &
Quantity (g)
Lycopodium 1000
Pyranine 300
Lycopodium 500
Lycopodium 2000
Seven
Main
D
D
D
D
Springs
Pineroot
D
D
ND
D
Westdown Bh+
D
NS
ND
D
Holwell
ND
D?
ND
ND
Note: +Not accessible during recent tests

                 Table  IB    Details of  Dye Traces  (this  study)
Test    Injection  Dye +         Seven Springs
Number  Site       Quantity  (g)  Main  Pineroot Combined   Torr Quarry   Holwell
1

2

3

4

5


Key
ND
D
F
NS
Downhead RWT -47.6
Dairyhouse FL-200
Bottlehead RWT-47.6
Heale CBS-200
Downhead RWT-47.6
Dairyhouse FL-200
Downhead RWT -47.6
Dairyhouse FL-200
Downhead CBS -1000
Bottlehead FL-600
Heale RWT- 120

: not detected
: detected
: samplers frozen
: not sampled
D
D
D
D?
D
D
D
D
D
D
ND

RWT
FL
CBS

D
D
ND
ND
D
D
D
D
D
ND
ND

40
109
37
D
F
F
50
106
30
86
0

D
ND
ND
ND
D
ND
NS
NS
2
4
45

ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND

Rhodamine WT
Fluorescein
Tinopal

CBS-X





                                        82

-------
Table 2
A)
Flow velocities calculated using straight line distance with time of
first arrival and time to peak dye concentration for A)
swallet/Seven Springs conduits, B) boreholes and quarry sites.
Stream Sink
                        Groundwater Flow Velocity (km/d)

        First Arrival             Peak               First Arrival"1"
Downhead
Dairy House
Bottlehead
Heale
>6.7 - 4
4.0 - 3.
>2.40 - 1.
? - 1.
.0
35
90
65
6.7 -
2.25
2.40 -
?
2
-
0
1
.45
1.4
.92
.60
7.4< V <

2.4< V
3.K V
CIO
4
<2
<3
.4
.2
.8
.7
•'•From Atkinson et al (1973b) , Table 2.

B)
Test
Injection Site    Detection Site
                  Groundwater Flow Velocity (km/d)
                  First Arrival     Peak
Test 1
Downhead
            Dairyhouse
Neilson 1         10.2
Tunscombe         0.96
Shute 1           0.26
Pineroot Sp
Pineroot Sp       0.60
2.04
0.63
0.23
0.42
Test 5
Downhead
            Bottlehead

            Heale
Pineroot Sp       0.70
Ashley 1          2.37
Secondary Sump    0.24
Issues            0.14

Secondary Sump    0.45
                  Borehole P2
                  Sump
                  Issues
                  Tunscombe
                  >0.85
                  0.48
                  0.11
                  0.13
0.082 (0.022)
0.096
0.11

0.17

0.85
0.14
0.050
0.045
                                     83

-------
These can therefore be compared with the earlier  experiments  of Atkinson et
al (1973b) performed under natural (pre-dewatering) conditions.


1. The Heale Slocker/Seven Springs Conduit

A major  difference between  the  present and pre-dewatering results  is  our
failure  to  obtain clear  positive  results at   Seven  Springs  from  Heale
Slocker,  the closest sink to the  quarry.   As shown on Figure  2  only  a very
minor barely detectable dye  pulse reached  the springs.  The tracer was also
not detected at any of  the  monitored boreholes  (Figure 1) ,  nor  at the sump
outflow from Torr Quarry.  Unfortunately,  Tinopal  CBS-X (a  blue fluorescent
tracer) has  a  low  resistance to photo-degradation (Smart & Laidlaw,  1977),
and may  therefore  have  been  lost when  exposed  in the  quarry sump.   Thus,
for Test 5, Rhodamine WT which has a low and stable background fluorescence
and is resistant to photodecomposition, was injected at Heale  Slocker.   It
was not  detected  at Seven  Springs  but was first present  in hand  samples
from borehole  P2  in Torr Quarry  (Figure  1) some  25  hours  after  injection
(Figure  3a) .    The dye  breakthrough curve rose  rapidly  to  a peak  then
dropped  exponentially,  the   same  response   as the other  swallets  at  Seven
Springs and strongly suggestive of conduit  flow.   Borehole  P2  is cased to a
depth of 50  m,  but is  then  open to a depth of c.  200 m (c.  -50 m ADD;  all
other boreholes sampled are  uncased or have  slotted casing).   Therefore  the
Heale  conduit  apparently tapped  by the  unpumped borehole P2  is at  some
depth below  the quarry  floor, and has not yet  been intersected by removal
of limestone. Borehole P2 was not present while  Tests 1-4  were carried out.

Since  the  completion  of Tests  1-4,  significant  excavations  of up to 30  m
have occurred  in  the  northern end of the  quarry. At the time of Test  5  a
secondary  sump  had been excavated  on the  quarry floor north west  of  the
main sump, and when required to clear standing water on  the  quarry  floor,
water was  pumped  from this   via a  pipe  and  open  channel  to the main sump.
Traces of Rhodamine WT  from Heale  Slocker  were  detectable  in  the  secondary
sump 4 days after  injection  at very low but rising concentrations  which  are
indicative of  highly  dispersed diffuse  flow (Figure  3b) .   Nine days after
the  injection  an  intense 25 mm rainstorm affected the  East Mendip area,
substantial areas  of the north western  part of  Torr Quarry were flooded to
depths up  to 1  m.   Water was observed to discharge at specific  points from
rubble masking the north western  wall  of  the  quarry  (Figure 1)  .   Tracer
concentrations  in  one such  issue  ('Strong  Spring') are shown  in Figure  3c.
Initially, much higher  concentrations of  Rhodamine WT were present  than in
the secondary  sump,  but these  declined progressively with continued flow.
Tracer was  still  present in the  issues some 40  days  after  injection when
flow ceased, and was  also detected when flow  recommenced after heavy rain
52 days  after  injection.   The recovery of Rhodamine WT was  budgeted using
the rating  for the pump  discharging water  from  the  secondary  sump,  which
operated  from   days  10  to   19 after  injection.    The  dye  concentrations
mirrored those from the issues, peaking initially as  tracer was flushed out
by the  recharge event,  then declining  slowly,  as would  be expected for a
diffuse flow system.  Tracer recovery was  estimated as 30% on cessation of
pumping, but at this  time  hand samples from standing  water   on the  quarry
floor, combined with  the observed extent  and depth  of flooding suggest an
additional 2.5% to 3% of the tracer was  present, which  declined to about
0.5% on cessation  of sampling.  Significant tracer discharge  also  continued

                                     84

-------
                                   Figure 3.    Tracer  breakthrough curves
                                         for  Test 5:
                  Rhodamine WT (Heale Slocker)
                       Borehole P2
                       Tunscombe Borehole (x10)
                             —i—
                              20
                                   —i—
                                    30
                Time (days after injection)
b)
                        Fluoresceln (Bottlehead Slocker)
     1.5-
     1.0-
                   Rhodamine WT (Heale Slocker)
                — Tlnopal CBS-X (Downhead Swallet)
                                                   o
                                                   o
   2
   I
0.5-
a)  Borehole  P2  Torr
Quarry and.Tunscombe
borehole          for
injection          of
Rhodamine    WT    at
Heale  Slocker.

b)   Secondary   sump
outflow  Torr  Quarry
for       Fluorescein
(Bottlehead
Slocker),  Rhodamine
WT   (Heale   Slocker)
and   Tinopal   CBS-X
(Downhead Swallet).

c)    Strong    Spring
issues   Torr   Quarry
for       Fluorescein
(Bottlehead  Slocker)
and    Rhodamine    WT
(Heale  Slocker).
                  10          20
                Time (days after injection)
c)
   O)
   IS.
   c
   o
   c
   o
   o
     1.0-
0.5-
      0-
                             (Key as 'B')
                   w^
                  15          25
                Time (days after injection)
                                    35
                                   85

-------
for the  next  37 days  (Figure  3c) ,  but  could not be  formally budgeted  as
dilution in the main sump, which  has  a  large volume  (274 x 10  m  )  was  too
great.   A  maximum  additional recovery  of  12.5%  is  estimated  from  the
observed tracer  concentrations in the issues  and their estimated combined
discharge.    Thus the  total  estimated recovery  of  the  tracer injected  at
Heale  Slocker  is 45.5%  all  of which appears to have  reached the  quarry
sumps via diffuse flow routes.

One other  sample location Tunscombe  borehole,  showed the  presence of  the
tracer  from  Heale  Slocker (Figure  3a) ,  although concentrations  were very
low.   Tracer was first detectable following  the  heavy  rainstorm  on day  9.
Then  concentrations  increased  gradually until  a further  period of rain
which  commenced  with  15  mm  on day  23  and continued with smaller  totals
until  27 days  after  injection.  This recharge event caused a  second  pulse
of  tracer  to  move  into the  borehole, but concentrations  then fell   to
previous levels.   Dye  was  no  longer  detectable 42  days  after  injection.
Thus  in  addition to  that discharged  into  Torr Quarry,  some  of the  tracer
from Heale Slocker underflowed the Quarry  and was moving down  the regional
hydraulic gradient  towards  Seven Springs,  apparently as  a diffuse plume.
No  tracer was  however  detected at Seven Springs,  probably because  of high
dilutions in the swallet conduits  feeding the springs.
2. The Downhead Swallet and Bottlehead Slocker/Seven Springs Conduits

Fluorescein dye from  Bottlehead Slocker was detected  during Test 5  in  the
secondary  sump  in Torr  Quarry within  48  hours of  injection (Figure  3b),
indicating a very rapid transmission. Fluorescein was not present  in any of
the quarry floor  issues,  once  these  entered the secondary sump and  pumping
commenced  concentrations  therefore  decreased due  to dilution.   It was  no
longer detectable 16 days after injection.  Recovery was estimated at 4.3%,
compared to  86%  at Seven Springs,  giving a total  tracer  recovery  of  90%.
This  is  not  significantly different  from 100% given  the  probable  gauging
errors involved,  but  it is  noteable that  concentrations  rose again  in  the
secondary  sump  after  pumping  ceased,  suggesting  that  unrecovered  tracer
remained in  storage,  either in the aquifer,  or  more probably  the  quarry
sub-floor.    No fluorescein  was  detected  at  any  of  the  borehole  sampling
sites.  The essentially  complete  recovery in Test 5 compared with only  37%
recovery in Test 2 is significant and will be discussed further below.

In 1987, tracer injected  in  Test  1  at Downhead Swallet was  not detected at
the Torr Quarry  main sump,  but  was detected  in  several boreholes  (Figure
4).  At Manor Farm, a peak occurred some  12 hours after injection,  followed
by a plateau-like low concentration tail.  A similar tail was  also observed
at Tunscombe  and  Shute  Farm  boreholes,  1 and 3  days  respectively  after
injection.   The  clearance of  dye from  the observation boreholes  coincided
with a decline of tracer  concentrations at the Pineroot Spring (Figure  4),
which has been shown  above to  be  decoupled from the swallet conduit at  low
flow.    This  suggests that a pulse of  tracer  may have been moving  in  the
diffuse flow part of the aquifer towards  the spring, an observation similar
to that made  for the Heale Slocker (Test 5) test at Tunscombe borehole.
                                    86

-------
      3.CH
    O)
    o
    03
    o
    o
    o
      2.0-
      1.0-
       0--
                           Rhodamine WT (Downhead Swallet)
                                     Pineroot Spring
                            	Manor Farm Borehole
                            	 Tunscombe Borehole
                            	 Shute Farm Borehole
Figure 4.
       i       i        i       i       i        i       i
       1       234567
                Time (days after injection)

Tracer breakthrough curves for Rhodamine WT injected at
Downhead  Swallet  (Test  1)   at  Pineroot   Spring,  Manor  Farm,
Tunscombe and Shute Farm 1 boreholes.
Test 5 was undertaken under  similar groundwater level conditions at Test  1,
but tracer from  Downhead Swallet was not detected  in either the Manor Farm
or Tunscombe  Boreholes   (Shute  Farm was not  sampled) .   This  may be partly
because  of  the  high  and  more  variable  background  fluorescence  in the
Tinopal  CBS-X (blue) waveband.   In the recently  drilled  (1990)  Ashley  1
borehole  fluorescence  readings  in  the blue  waveband showed  a  double peak
before declining to  lower values 16 days after injection (Figure 5).   These
peaks  coincide with  and lag  recharge  events  associated  with   rainstorms.
There  is however  no comparable change  in  apparent  concentration  in the
Rhodamine WT  (orange)  waveband,  and the changes associated with the  third
rainy  period   (days  23  to 27  after  injection)  are much  smaller  (although
sample  frequency is much  lower) .   This  suggests  that  we are  not simply
monitoring changes  in background fluorescence  associated with organic-rich
soil  water arriving at  the water-table,  but  that  some  of the  labelled
Downhead  Swallet water  moved  from  the conduit  into  the  aquifer during
recharge  events.   Confirmation  of  this association  of tracer breakthrough
and recharge  is  also apparent  during Tests 1 and  3,  which were undertaken
during a  period  of  generally  falling groundwater-levels,  onto  which  minor
recharge  events  were superimposed.   No  tracer  was detected  at  either the
boreholes or  the quarry  sump for Test 3 (freezing of the samplers precluded
determination  of the tracers at the springs), which took place when swallet
flows were  in recession.   In  contrast  Test  1,  which  proved positive, was
associated with  a  minor recharge event and  high  swallet  flows  (50 L/s  at
Downhead compared to 24  L/s  during  Test  3).
                                     87

-------
              b)
                 o
                 15
                 o
                 c
                 o
                 o
                 s.
                 Q
                   0.5-
                              10       20      30
                              Time (days after injection)
                                                     40
                                    Tinopal CBS-X (Downhead Swallet)
            Rhodamine WT (Heale Slocker)*
                              10       20      30
                              Time (days after injection)
                                         40
Figure 5.
a)

b)
Rest water  level in Ashley  1  borehole and  daily rainfall
at Torr Quarry.
Apparent concentrations  of Tinopal CBS-X injected at
Downhead Swallet in Test 5 and Rhodamine WT (Background
fluorescence) in the Ashley  1  borehole
In  Test 5,  the  behaviour of  the Downhead  Swallet Tinopal  CBS-X tracer  in
Torr  Quarry  was  initially  similar  to  the  fluorescein   from  Bottlehead
Slocker (Figure 3b) ,  with rapid  initial arrival.   However,  Tinopal  CBS-X
was also present in the new issues, (Figure 3c),  thus concentrations  in the
sump  fell   less  rapidly  after  the  onset  of pumping.    Despite  this,  the
tracer  was  rapidly exhausted,  and recovery  was  only approximately 2%  of the
injected mass.   Concentrations only rose again  when pumping stopped    There
is  thus a  clear contrast between  the  style of breakthrough from Bottlehead
and Downhead swallets,  where diversion from the conduits  is minor  and  a
single  short  pulse reaches the  quarry,  and Heale  Slocker,  where capture  of
the conduit  flow is  complete and a sustained discharge  of tracer occurs via
the quarry.
                                     88

-------
Discussion

The tracing results  show an increasing degree  of functional change  in  the
conduit  as  a  result  of dewatering  with  distance  from  the  quarry,  and
coincidentally between phreatic  (near)  and  vadose  (far)  conduits.    The
Heale Slocker conduit  no  longer  discharges  to the Downhead Swallet  conduit
feeding Seven Springs,  but  there is no evidence  that it has been  directly
intercepted by quarrying.   Rather it appears  that  leakage  by diffuse  flow
from the  conduit under  the steepened hydraulic  gradient,  associated with
pumping from the quarry, prevents development of  sufficient head within  the
conduit to  enable  pressure-flow over the downstream loop thresholds. This
interpretation is  supported by the widespread  distribution  of tracer  from
the Heale conduit  within the  Torr  Quarry issues adjacent  to borehole  P2,
which is indicative of diffuse flow, and by the movement of tracer  down  the
natural  hydraulic  gradient north  east  of  the  quarry  to  the  Tunscombe
borehole.   Note  also  that  the  quarry  issues and  Tunscombe  borehole  show
'flushing' of the tracer in response to rainfall  events as swallet  recharge
pushes more dye tracer through the diffuse flow zone of the  aquifer.

The Bottlehead Slocker  conduit,  which lies  further  to the north than Heale
Slocker,  is  less   directly  influenced  by  dewatering.    During   Test  5
conditions were particularly wet  and swallet  flows  high.   A small pulse of
tracer  moved  initially to  the sump, suggesting  leakage into  the  diffuse
flow zone as  for Heale Slocker,  but  the  majority of  the  dye injected  was
discharged  through  the conduit  to  Seven Springs.   During  Test  2  swallet
recharge was much  less,  and the  hydraulic gradient  towards  Torr Quarry  was
higher because the sump was at a low level.   As a result, tracer recoveries
at Seven  Springs were much lower  (37%),  the  major part  of  the dye  moving
towards the  quarry by  diffuse  flow according to the  Heale  Slocker  model.
This  interpretation  is  supported  by  the  very  much  slower  travel  time
suggesting impairment of transmission along the conduit.  In fact no  tracer
was recovered in the quarry, possibly because of high dilutions in  the main
sump,   or   because  sampling  was  not   continued   for  sufficient   time.
Calculated  velocities  for  diffuse  groundwater  flow based  on  the  tracer
tests are typically an order of magnitude lower than  for conduit flow using
first arrival (Table  2A & B).    However,  given the  much greater dispersion
in diffuse flow, time  to  peak  (or ideally time to centroid) velocities  are
more representative.

The Downhead  Swallet conduit  is  least  affected by  quarry  dewatering both
because  it  is  furthest  away  from  the   quarry,   and  because   it   is
predominantly vadose.  Thus although some leakage from  the conduit  into  the
diffuse  flow zone  occurs,   as  indicated  by  detection of  tracer  in  the
observation boreholes  during Test 1,  this water  is  not drawn towards  Torr
Quarry, but follows  the  natural  hydraulic gradient  to discharge from Seven
Springs.    In  a  vadose   conduit  leakage   is   independent  of  the  head
differential in the  saturated  zone and essentially  constant.   In  fact  the
tracer  evidence from  Test 5 suggests that pulses of dye are expelled from
the  conduit during  high  flows, perhaps  when  pipe-full  flow conditions
occur.   The movement of these pulses in the diffuse  flow zone appears to be
complex,  and further  tests  are  necessary  to  determine  if  the   results
reported are  reproducible,  and  explainable  in terms  of differential  head

                                    89

-------
conditions imposed  on  the  general regional hydraulic gradient  towards both
quarry and natural outlet.

To our knowledge this  is the  first well-documented example  of change in the
function of a  conduit  network as a result of quarry dewatering.   Given the
increasing use  of  this option in extraction of karstified  limestones  it is
therefore of some interest.   Changes  in  the  conduit behaviour clearly occur
even  though direct  intersection by the  quarry  void has not  occured.   This
appears to be  the result of diffuse leakage  of  water from the conduit which
prevents  pressure  flow over phreatic  loop-tops.   Similar  leakage  from
vadose  conduits will  have much  less effect  on  conduit function  because
pressure  flow  is   not involved.   Despite  tracer tests  being  the  tool
preferred  by   karst  hydrologists   for   the   study  of  karst   conduits,
hydrogeologists  assessing  the  impact of  quarry  dewatering have  generally
adopted more conventional  approaches.  Perhaps  the information obtained in
.this  study  may  encourage  a  more   wider  utilisation  of  water  tracing
techniques for  improving our  understanding of both conduit  and  diffuse flow
in karst terrains.
ACKNOWLEDGEMENTS

The  authors  would  like  to thank  Foster  Yeoman Ltd  for access  to a  vast
quantity  of  hydrological data  aswell  as boreholes  in  and  around  Torr
Quarry, and  ARC for access to Seven  Springs.   A.J.  Edwards and  S.L.  Hobbs
were  supported  by NERC training awards in collaboration with ARC  Ltd,  and
Bristol Waterworks  Company and Wessex Water Authority respectively.   Thanks
to  Simon  Godden for drawing  the  figures,  and  Liz Owen  for processing  the
text.   This  paper is based on an  earlier  version published  in  Proceedings
of the University of Bristol Speleological Society (1991).
REFERENCES CITED

ATKINSON, T.C., BRADSHAW, R, & SMITH, D.I., 1973a. Quarrying in Somerset.
      Supplement No. 1, Hydrology and Rock Stability. Mendip Hills. A
      Review of Existing Knowledge. Somerset County Council.

ATKINSON, T.C., SMITH, D.I., LAVIS, J.J. & WHITAKER, R.J., 1973b.
      Experiments in tracing underground waters in limestones. Journal of
      Hydrology, 19, 323-349.

DRISCOLL, G., 1986. Groundwater and Wells, 2nd edition, Johnson Filtration
      Systems, Minnesota.

FARRANT, A.R., 1991.   The Cough's  Cave  System; exploration since 1985  and a
      reappraisal of the geomorphology.  Proceedings of the University of
      Bristol Speleological Society, 19(1), in press.

FORD, D.C. & EWERS, R.O., 1978.  The development of limestone cavern
      systems in the dimensions of length and depth. Canadian Journal of
      Earth Sciences,  15, 1783-1798.


                                    90

-------
FORD, D.C & WILLIAMS, P.W.,  1989. Karst Geomorphology and Hydrology. Unwin
      Hyman.  London.

SMART, P.L.,  1981. Variation of conduit flow velocities with discharge in
      the Longwood to Cheddar Rising Systems, Mendip Hills. Proceedings of
      the 8   International  Speleological Congress,  1, 333-337.

SMART, P.L.,  HOBBS, S.L. & EDWARDS, A.J., 1991.  Dye tracing in the Beacon
      Hill  Pericline,   East  Mendips.   Proceedings   of  the   University   of
      Bristol Speleological  Society, 19(1),  in press.

SMART, P.L. & LAIDLAW,  I.M.S., 1977.  An evaluation of some fluorescent
      dyes for water tracing. Water Resources Research, 13, 15-33.
BIOGRAPHICAL SKETCHES

Dr. P.L.  SMART is  a lecturer in  geography  at the  University of  Bristol,
      England, from  where he  also  received his first degree and  doctorate.
      He  has   extensive   research  experience   in   karst  hydrology   and
      geomorphology  in  many  different  areas  of  the world,  including  the
      Bahamas, China,  Indonesia,  Malaysia and several  European  countries.
      The  application  of  natural  and  artificial   tracer  techniques   to
      hydrological problems in karst areas has been a particular interest.

Dr. S.L.  HOBBS.  is a consulting hydrogeologist in a firm of Aspinwall  and
      Co. and  currently deals with  a variety of  projects on  contaminated
      land,  and site hydrological evaluation and instrumentation. His  first
      degree is in Geography  from Bristol University, where he  also  studied
      for his  doctorate,  which was on  the hydrology of the saturated zone
      in a karstified Carboniferous Limestone aquifer.

Mr.  A.J. EDWARDS  is  a  research  student  in  the   Geography Department,
      University of  Bristol,  from  where he also obtained his  first  degree.
      He  is  currently  completing his thesis  dealing with the  hydrogeology
      of quarries  and landfills  in the  karstified limestone aquifer of  the
      Mendip Hills
                                     91

-------
Q. How reliable are the lycopodium traces and dye traces from
   Heale Slocker to Seven Springs that were done before the
   quarry was excavated to below the water table?

A. Exceedingly.
Q. Could you sumnarize the hydrology anticipated after
   cessation of quarring operations?

A. Simple computer simulations have shown that the water
   table, within an abandoned sub-water-table quarry in the
   East Mendips, may take up to 24 years before it will
   recover to pre-dewatering levels.  This phenomenon is due
   to the immense volume the disused quarry void represents.
   If we assume excavations at Torr Quarry cease at 0 m AOD
   (Above Ordinance Datum, i.e., above sea level),
   approximately 165 million m3 of water would be required to
   fill the remaining quarry void up to the pre-dewatering
   level.  This immense lake (1.1 km2) will dampen the
   seasonality of the surrounding aquifer water table, and
   may even have a significant effect upon the whole
   catchment's storage potential.
                                        92

-------
     Session II:
Site  Characterization

-------
94

-------
  Approaches to Hydrogeologic Assessment and Remediation of

                 Hydrocarbon Contamination in

        Clay-Covered Karsts with Shallow Water Tables


                         Tony Cooley

                    University of Kentucky
                Lexington,  Kentucky 40506-0059
Abstract

A  conceptual  model  is  presented  of  water  and light  phase
hydrocarbon (LNAPL) movement through karsts  mantled by clayey
residual soils in which  a water table  is  present above top of
rock.   This  model is  then incorporated  into  approaches  to
hydrogeological assessment  in  such karsts.   These  points are
illustrated by examples from case histories in Alabama.

Water and hydrocarbon movement through the overlying soils and
upper portion of a karst  aquifer can be conceptualized in terms
•of the raacropore systems of the soil and the subcutaneous zone
of the rock.   Cracks and  other macropores would largely control
flow  of  water  and particularly of  hydrocarbons through the
soils.  An underlying  subcutaneous  zone at  and near top of rock
would then  converge  the downward  flow toward inlets  into  an
underlying, more  efficient  karst  underdrain system.   Where a
water table exists in the soils, cones  of depression would be
centered on these  inlets while the greater water flow in the
vicinity  of  these  inlets  would  result  in   locally  enhanced
permeability and porosity,  as  well as a  locally lower top of
rock.  A tributary system of solutionally enhanced routes along
the soil/bedrock  interface  and in  the upper  few feet of rock
feeds these inlets.

Assessment of contaminant movement  and design of  remediation in
such a system would best  focus  on identification  and character-
ization of the vicinity  of  such inlets.   An assessment scheme
could employ a combination of piezometers to define the effects
of the underdrain  system,  soil and water  sampling  to use the
existing hydrocarbon  contamination as  a  tracer, rotary wash
probe borings to locate top  of  rock by boring refusal and local
                              95

-------
areas of  greater permeability by  water losses,  and geologic
characterization of the site vicinity.

Remediation  could focus  on  source  removal  by  treatment  of
contaminated soils and interception of contaminated ground water
near the  points where it  is entering the  underdrain system.
Where fluid removal is appropriate,  location of the pumping well
in the vicinity of  a an affected  inlet with drawdown supple-
mented by vacuum is suggested to take advantage of the locally
greater permeabilities presumed  to be present  there, improve
connection to the system, and locally reverse natural gradients
to stop further releases.
Introduction

This  paper presents  a conceptual  model  of  the  flow  system
overlying a conduit underdrained karst with a subcutaneous zone
and  significant clay  cover,  discusses  hydrocarbon  movement
characteristics, and then presents approaches to characteriza-
tion of such a flow system and remediation design.  This paper
considers  the case  where a  water table  is  present  in  the
overlying  clay  a  significant portion  of the time and results
primarily  from authogenic recharge.

The advantage to working in an area with a water table above the
rock  is  that piezometers can  be  used  as indicators  of local
water flow and sampling of water is facilitated before it enters
the rock.   However, interpretation of the  resultant patterns
must be based on some conceptual model.
Conceptual model of shallow water movement

This paper addresses the situation in which a karst underdrain
system of solutionally enhanced conduits exists within an area.
Uniformly  distributed  recharge   from  precipitation  must  be
collected into the underlying system by some mechanism and this
collection occurs because the underlying  conduit system has a
relatively low head due to its connection by an efficient flow
path to  a distant discharge area.   Such a conduit  system is
considered  to  consist  of   a  tributary  system  of  conduits
collecting  water both  by direct inflow  from master  drains
connected to the top of rock  in the vicinity of the conduit and
by  diffuse  flow from  the   interconduit  areas  at a  greater
distance from a conduit.  This paper assumes the master drains
in direct connection to the conduit system are overlain by soil
which moderates the inflow and assumes direct swallet access to
the conduits is not present.

It is this collection system  that  is the subject of this paper.
The  basic conceptual  model  for  shallow water  flow  into an
underlying  karst  conduit was  outlined  in Gunn  (1981)  and
Williams (1983).   This collection system is assumed to involve
                             96

-------
discrete inlets  into  the conduit system  and these inlets are
considered to share the low head of the conduit system relative
to the surrounding soil and rock. The flow toward these inlets
may be either in soils  (called throughflow)  or in solutionally
enhanced pathways  at  the soil/rock interface or in the upper
portion of the rock (called subcutaneous  flow).

Rain water percolating downward through the soils is assumed to
be well distributed, but percolation may  be  concentrated along
abundant macropores through the soil that have higher permeabil-
ity to flow than the intact clay.  The most  important of these
are expected to be the  collapse  cracks resulting from repeated
collapse  of soil  cover as the top of  rock is lowered over
geologic time by solutional weathering processes. Such collapse
cracks are  subvertically oriented,  continuous from the top of
rock  to the  ground  surface,  dilated  at the   time  of their
formation,  and  very abundant.   Non-plastic  soils  and organic
matter may have been washed into these cracks as they heal.

Other  sources of  macropores  considered  significant  would be
cracks resulting from differential shrinking  and swelling due to
moisture  change,  including dessication  cracks,  and tree root
holes.  Observations of undisturbed soil samples and excavations
in  the  soils  of  Birmingham,   Alabama   indicates the  clays
overlying limestone are typically fissured,  often at very close
intervals, and subvertical slickensided surfaces are common.   A
conceptual  view  of  macropores  in  the  overlying soils  is
presented as  Figure 1.

The dense, underlying limestones are considered to be effective-
ly impervious except at discrete discontinuities such as joints
and  bedding planes.    The  downward percolating water  in  the
situation discussed here is assumed to accumulate at this top of
rock  interface  as either a perched or continuous water  table.
From this point it must either move laterally toward a  low head
inlet  into  the  conduit system,  percolate into the  diffuse  flow
system in the rock between conduits,  or  flow laterally  through
the soils toward some other discharge area.

Lateral  flow through  the  soils is  constrained  by  the  pre-
existing network of macropores,  which do  not readily change in
response to flow.   Flow within  bedrock discontinuities  and at
                  Figure 1  Common macropore types in soils
                               97

-------
the top of rock/soil interface can change, however,  due to
solutional removal  of  rock bounding these  flow routes.   Such
solutional removal  is  constrained by  the kinetics  of  calcite
dissolution as described by White (1977).   For slow moving flow
paths, most of the solutional enhancement  occurs near the entry
portions of such  flow  paths,  but can be linked by  mechanisms
described in the network models  of Ewers  (1989).

Solutionally  enhanced  flow paths would be  expected to  form a
network oriented to the low head drains,  called a  subcutaneous
zone.  A model of such a network is shown in  profile in Figure
2a, from Smart and Friederich (1986)  and the network connecting
mechanism of Ewers  (1989) would  imply  that in plan view such a
subcutaneous  system might look  like Figure  2b.   The  enhanced
flow  paths  would be  expected to have some  three-dimensional
aspect.  This is conceptualized to consist of discrete tributary
channels, while the portions of the  discontinuities between the
enhanced  tributary  flow  paths  would continue   to  have  low
permeability  and experience  little flow.

The flow paths would be progressively interconnected toward the
master drain, but,  if developed  like Ewers multitiered network
linking models  from the drain outward, would be progressively
separated by  divides of unenhanced permeability with distance
from  the drain.  This suggests that  relatively independent sub-
basins would  form within the area  of  influence of  the master
drain during  development  of  the  subcutaneous  system.

The head distribution in such a flow system would produce a cone
of depression centered over the master  drain into the conduit as
described by Williams (1983) .  Flow would be directed toward the
low  head of  the subcutaneous  network both  laterally  in the
unenhanced portions of the rock and  from the soil.  Whether flow
moves as  throughflow in the soil directly toward  the drain or
enters into the  subcutaneous system first,  the cone of depres-
sion  form would be expressed in the soil.  In the later case,
the  head in  the  soil  would reflect the  entry heads  into the
subcutaneous  system,  but these would  also decrease toward the
main  drain.   As  shown  on  Figure  2b,  the cone  of depression in
  Figure 2a)  Cross-section  of
          "subcutaneous zone
"2b)  Plan view of subcutaneous
    zone
                   (Smart and Friedarich. 1986)
                               98

-------
detail should also reflect the low permeability divides between
the proposed sub-basins  if  these are present.
In addition to  solutionally enhanced  flow routes developed  in
discontinuities within the rock, solutionally enhanced routes
may develop at the top of rock interface.   Patterns of solution-
al removal  at the soil/rock  interface, called  runnels, could
potentially develop  enhanced permeability-   In  general,  flow
along  this  interface  would be  constricted  as  the  weight  of
overlying soil presses the  clay  against the rock.  As shown  in
Figure  1 however, a  soil-roofed,  rock-floored  channel could
develop  at  this  interface due to solution along the floor and
arching of soil above the runnel. The  common observation during
rotary wash drilling or coring of occasional water losses at top
of rock  in  limestone areas  might be explained by such enhanced
runnels.
Aspects of hydrocarbon  movement in soils

Hydrocarbon movement through clayey soils is determined both by
the relative  saturation of the soils with  hydrocarbons and by
the pore  size of  the  soils.   Figure  3a  shows the  two phase
relative  permeability  relationship that  governs  movement of
hydrocarbons  in  saturated soils.   The three-phase relationship
through partially  air filled  soils is similar in that permeab-
ility to  a  fluid is related to the existing saturation of the
fluid  in  the medium.   For movement  into areas  not currently
containing hydrocarbons, the permeability to hydrocarbon flow at
the front of  the hydrocarbon plume is  substantially less than
that within the  plume itself.

In addition,  the hydrocarbon  must overcome the greater attrac-
tion that water  has for the soil particles.  To displace water
from a  pore,  the  hydrocarbon must overcome  the  meniscus that
expresses this greater  attraction.   The pressure required to
overcome  this meniscus  will be called the capillary resistance.
The strength  of the meniscus varies as the perimeter of the pore
or directly as the pore diameter.  The expelling force varies as
the cross-sectional area of the pore  or the  square of the pore
       100*  NAPL SATURATION
    CQ
    S5
    a
    UJ

    F

    uj 0.0
               S™
          WATER SATURATION
                                         Head relative to plume front-
                                         pressure
                                                  macropore
   a)  Effect of saturation on permeability

   Figure 3          Source: EPA (1989)
b) Pressure distribution within a plume
                              99

-------
diameter.  Thus, a smaller capillary resistance must be overcome
for hydrocarbons to expel water from a large pore than a small
one.  As a result, most hydrocarbon movement through soils will
be through the macropores, which have a larger  effective pore
size, with little movement into the saturated clay itself.

A  conceptualization of  the  movement of  a hydrocarbon  plume
through clayey soils is shown in Figure  3b.  A plume of mobile
separate phase hydrocarbons is assumed to have formed due to a
release  from  above.  The  shaded  area in  the figure 'contains
hydrocarbons and would thus have a permeability to hydrocarbon
flow many times greater than that at the  edge of the plume where
the hydrocarbons have not entered.  The rate of flow within the
plume would be constrained by the rate of advance of the front
of the plume.   In effect,  the hydrocarbons within the plume are
like in a bag created by the low permeability at the front.

As a result, nearly all of the available head would be lost at
the front of the plume,  producing a large head gradient there.
Because of its higher permeability, only  a small head loss would
be required within the plume to maintain  the flow rate permitted
by the advance of the front of the plume.   The hydrocarbon can
be considered  to be mounded inside the plume relative  to its
front.

This high  head gradient at the plume front,  produced  by the
confinement of hydrocarbons within the plume,  is a significant
factor in promoting advance of the plume and is  most effective
at the lowest part of the plume.  Where  the plume has advanced
farthest  in the vertical direction, as   in  the lower  right
portion of Figure 3b, the gradient at the plume front favoring
further advance of  the  plume  is also greatest and  thus would
tend to advance faster than the remainder of the plume.

To maintain itself  in  the pores,  the hydrocarbon must  have a
higher pressure at  this  interface than the water.  As a result,
the capillary  resistance pressure  portion of the head  at the
plume front is unavailable to promote flow. As noted above, the
capillary resistance is inversely related to pore size.   Thus,
where  flow  has followed a macropore  or encounters a  zone of
larger pores,  the available gradient for advancing the plume is
further increased.

The  above discussion  is  particularly  important  for a  sub-
vertically  oriented macropore.   Not  only will the  initial
permeability be greater along  such a feature, but the available
gradient at the plume  front in the macropore will  be greater
than elsewhere along the plume front  due  to smaller capillary
resistance, and as the plume preferentially advances along the
macropore  the available head  gradient  will  further increase
relative to that in the remainder of the plume.

This favoring of macropores and the gradient magnifying effect
of mounding within  the mobile  separate phase plume result in an
                             100

-------
irregular spread of the plume.  Figure 4 illustrates an example
from northern Alabama.   In this case, a release of Number 5 fuel
oil occurred in a clayey soil overlying a flat-lying limestone
(Tuscumbia formation) with a water table about 3 to 5 feet above
top of rock in the area of the release.   A zone of greater chert
content,  and  possibly   of  solutionally  enhanced runnels,  was
present at or near the  top of rock.

As seen in the cross-section of Figure 4b, the hydrocarbon plume
apparently did not spread when it encountered the water table.
At  the point  the plume  just reached  the  water  table,  the
difference in elevation between the base of the plume and that
of the release resulted in a mounded hydrocarbon height of as
much  as  13 feet.   This head  would  have  been available  for
lateral spreading along the water table as well as  for further
vertical penetration.   However, the capillary  resistance of the
small pores in the clay must be overcome for movement into clay
and this may potentially be greater than that of many feet of
water head.
It is hypothesized that subvertically oriented macropores, such
as collapse cracks or tree root holes,  allowed the hydrocarbon
access to the chertier or runnel-containing zone at top of rock.
The large pore  size  of  this top of rock  zone  then  provided a
route for spread of the plume at a smaller capillary resistance
than that of the clay-  In addition,  the height of the hydrocar-
bon in the plume would be greater relative  to  the top of rock
zone than at the water table, providing more available head and
thus a steeper gradient to promote expansion of  the plume at the
deeper level.

The resultant plume,  as defined by the borings shown on Figure
4a, has  a  very irregular  form,  apparently exploiting  local
differences in the pore  size and permeability along this top of
rock zone.  Hydrocarbon  analyses of the soils (CAL-DHS modified
8015,  TPH) did not detect hydrocarbons  in the soils outside the
immediate source area except within a few feet of  the  top of
rock.   It appears that the greater permeability and larger pore
size present near or at the top of rock  controlled the spread of
the hydrocarbons at this site.


As shown on the cross-section of Figure 4b,  this mechanism may
also  allow  mobile LNAPL's  to penetrate  into  a  solutionally
enhanced subcutaneous zone or conduit system  a moderate distance
beneath the water table.  In this case the surrogate for such a
system  is  a  man-made  brick  conduit  44  inches  in  diameter
constructed in  1903.  This  conduit was at  the  depth shown on
Figure 4b and  likely had  areas of deteriorated mortar that
allowed water into the conduit.  A finger  of the plume apparent-
ly extended along  the top  of  rock zone to  the  conduit  as the
plume advanced.
                              101

-------
                                                                   MONITORING
                                                                   WELL
                                                                   CONDUIT
                                                                   INTERVAL - ]'
                                                                   BEDROCK UELL
                                                           *Paired borings seven feet apart
                                                             One clean, one with separate phase
                                                           hydrocarbon
                        a)   plume distribution and water table
                               LOACHHO DOCK
HONTHWV9T
         Is
 oS
 II
LEGEND:

  ttO   »O». TFH ltv*l
  (•)   SAunj! LOCATrCN

  X    MH1UHIO WATIH LTVtU

  •    UCA9UHCO OCPTH TO IOHIH
                                                                       L IOH. tCHtBTY CLATl
                                                            rmm
                          b)  cross-section  B-B'
     Figure  4   Hydrocarbon  and water  table  distribution,  Huntsville
                                       102

-------
It is concluded that once the hydrocarbon encountered the open
conduit, it flowed into the conduit until it apparently drained
the  mounded mobile  hydrocarbons that  had  been driving  the
expansion of the plume.  The  absence of a capillary barrier for
flow into  an  open conduit and its extreme  permeability would
have favored such flow.  Once the mound had dissipated into the
conduit,  the  remaining heads  within  the   LNAPL  plume  were
relatively ineffective at further expanding the plume in other
directions against the capillary resistance and low permeabil-
ities of the plume front.
Approaches to assessment

The  special  goals  of   assessment  considered  here  are:  1)
determination  whether  the  site  is  underdrained  by a  karst
system, 2) location of the intakes to the conduits if present,
and 3) determination of where and how to monitor.   These goals
supplement the goals of a conventional exploration of outlining
the soil and water plumes and characterizing the hydrogeologic
properties of the soils.

Approaches  to achieving  the above  goals  would   include:  1)
detailed definition of water table contours  as  a  clue to flow
patterns, 2)  considering the site in the  context  of possible
discharge areas and geologic information, 3) use  of the observed
hydrocarbon distributions in soils and water as an "opportunis-
tic"  tracer,  4)   mapping the  top of  rock  configuration,  5)
pumping monitor wells before sampling to expand their radius of
influence and draw water from the subcutaneous  system,  and 6)
drilling  some  probe borings with rotary wash  techniques and
recording water losses  as an indication of the presence of more
permeable zones.   The  combination  of  these approaches  at  a
particular site would depend on  the judgement of the hydrogeol-
ogist and the particular conditions  of the site.

The detailed  delineation of the  water  table surface  in site
soils is to  check  for the cones of depression or other perterba-
tions that should be associated  with entries to the underground
conduits.  To delineate the detailed potentiometric surface,  a
network  of piezometers and monitor wells should be installed.
Most  of these should  be  screened at  the  top of  rock/soil
interface and not extend into rock to avoid possibly intercon-
necting  two areas  of  differing  head.   A limited  number of
piezometers into  rock  sealed a short  distance into  rock and
paired with piezometers in the soil at the soil/rock interface
are recommended to  check for relative heads between rock and
soil.

The definition of the water table in the soil is the target for
this exploration because  it is more economic,  avoids the risk of
spreading contamination  to greater  depths,  and provides good
information on the location of the entry points to the conduit
system.  If  the rock communicates with the soil infrequently so


                            103

-------
that water movement  is mostly  by throughflow to the drains,  a
cone  of  depression  defining  this  flow  would  be apparent.
Alternately,  if  flow is  mostly vertical  into a  subcutaneous
collector system, the decreasing heads in  such a  system toward
the drain should also be apparent in the soils near the  entrance
to the  subcutaneous  system and the  cone  of depression  in  the
subcutaneous  system  should thus be reflected  in  the overlying
soils.  Of course, if a significant difference in head  is noted
between piezometers  in  soil and those in  the  underlying rock,
this approach should be modified.
The greater economy of starting with a piezometer network at top
of rock with the greater density of data allowed by this economy
would recommend putting most of the initial phase of piezometers
at this location.  Most of these piezometers could be installed
in  unsampled probe  borings once sufficient  soil borings  and
monitor  wells are  installed to characterize  the plume.   The
maximum spacing  between piezometers will  depend on the size of
the  expected cone of depression in the soils,  but might be on
the  order of 40  to 70 feet at sites with 10 to 20 feet of soils
if  there are little topographic clues.   The borings  for  the
piezometers  may  also provide top  of rock  information.

Figure  5a  is an  example of such  a detailed water table defini-
tion.   This  shows the elevation  of the water level in soils at
the  site and indicates that water flow is to the southeast and
southwest with a groundwater divide within the site.  Gradients
are  fairly  steep,  being  about .05 to .07, while the overall
pattern suggests drain locations off-property to the southeast
and  southwest,  though the  divide  may be  simply  a sub-basin
divide  within a  larger general cone  of depression.

The  obtained water table and other potentiometric  data are then
considered in the context of topographic and geologic clues to
potential discharge areas and flow patterns.  As shown on Figure
5b,  the site is  located between three possible discharge areas
to local creeks.  The maximum available gradients to these areas
     Figure 5a)
Site water table and
BTEX contours
Site context, potential
discharge areas
                              104

-------
from the lowest water levels  recorded on site was calculated as
the remaining drop in elevation divided by the straight line
distance to the potential discharge area.   The actual gradient
of flow to these areas is  likely less both because the straight
line distance  is probably not the  flow path and  the on-site
gradients are probably not at the  lowest  possible  point.   The
maximum possible gradients were  .006  to two of  the areas and
zero to the  third,  ruling out the  later.   The  factor  of ten
difference between on-site gradients and the maximum available
gradients to  move  on-site water  the  rest  of  the way  to the
discharge area  confirms  underdrainage of  the  area by  a  more
efficient system than the soils.

The directions to the potential discharge areas are discrepant
to  those  in the  soils,   also  indicating  underdrainage.   The
geologic  information that the  site is  underlain  by dipping
limestone  striking  northeast parallel  to  a  major  fold  axis
provides an additional clue that the deeper flow is toward the
southwestern option, but this was not confirmed at this site.

The pattern of hydrocarbons  in the  soils  and  ground water can
also be used to indicate flow direction,  as well as travel times
if the time of release is definitely known.   A dissolved BTEX
(benzene,  toluene,  ethyl benzene,  and xylenes)  plume on Figure
5a indicates that the water table defined by the monitor wells
is compatible with the actual water flow.

Since  the hydrocarbon plume represents  true long-term  flow
patterns,  this will  generally be more reliable than the current
water  table  pattern at  some sites as  site work  or seasonal
effects may change water table recharge patterns controlling the
flow patterns shown by piezometric data.

The top of rock configuration can also provide clues for entries
to the rock system.   Where water flow into the rock is concen-
trated, it  is  common for the  increased flow to have locally
removed more rock, resulting  in a depression in the top of rock.
Top of rock data does not necessarily indicate current activity,
however,  and  should  be  interpreted  in  terms  of  associated
potentiometric data because  a low top of  rock  may  result from
other patterns of ground water flow or deposition of sediment in
conduit systems may  have  changed patterns of flow over geologic
time.

Figure  6,  from the  northern Alabama  site  discussed earlier,
shows the top of rock data gathered during piezometer installa-
tion together with one of  the obtained potentiometric surfaces.
One  clearly  defined low  area  extended  to  a  depth of  9  feet
deeper than nearby borings in the area just west of the eastern
building on-site.  Nevertheless, this is  not  reflected in the
water table surface even though a piezometer was placed inside
this low area.  In addition,  a Type III monitor well installed
in the rock 50 feet to the north showed a head in the rock 1.8
feet higher than that in the piezometer in  the depression.  The
                             105

-------
                                             op of rock contours
                                             BEDROCK

                                             HATER
                                             TABLE

                                             rtONITORINC
                                             UELLS

                                             BEDROCK
                                             WELL
                                             INTERVAL
iValer table contou
                                        pionitor well in rock with head
                                        1.8 feet above that of piezometer
                                        in depression
             Figure 6  Rock and water table contours, Huntsville.
top of screen of this rock well was one foot above the bottom of
the piezometer in the depression.  Considering that the site is
underlain by  flat-lying rock, the bedding planes connecting to
this depression were apparently not enhanced in permeability or
were clogged  with sediment as there  was no mound of upwelling
water  on the water  table as  would  be  expected  for such  a
gradient towards  the  depression.

Relative to monitoring a site, the above characterization of the
flow regime is important for  interpreting the data gathered and
checking that monitoring wells are correctly  located.  If much
of the movement is in the soils  as throughflow,  as indicated by
the plume outline,  then monitor wells screened  at  top of rock
may  be  sufficient.    However,  if the water  flow  is into  a
subcutaneous  zone  in the  rock,  even though piezometers may
correctly show a  cone of depression and  indicate the direction
to  the  master  drain,  conventional  monitor  wells  may  miss
downward moving dissolved contamination.  Monitor wells sealed
into the shallow  upper rock would be  an  appropriate supplement
to monitor wells  just in the soil.

However, in both the case of monitor wells in the  soil and those
in the  rock,  as  illustrated  by  Figure 2b,  a passive well may
miss the flow of contaminated water due to the convergent nature
of flow toward the favored  flow routes.  For this reason,  it is
recommended  that for  slow flowing wells,  which  thus indicate
                             106

-------
poor  communication  to  the  system, the  purging  operation be
continued until  on  the  order  of  10  to  20 gallons  have been
obtained.  This  may require use of a supplementary  vacuum to
assist removal of this much  water and  then a recovery period to
eliminate the effect of the vacuum on the volatiles.  The reason
for this is  to temporarily reverse  the local gradients to cause
flow  of  water toward the well from the  solutionally enhanced
flow systems.

Use of  drill water  losses  would primarily be useful  for the
location of fluid recovery wells for remediation to indicate a
tie into the  actual  permeable  subcutaneous  system.   For probe
borings where drilling with water would not interfere with other
goals of the drilling,  this is low cost data  to obtain and could
also be used in probe drilling for an actual  recovery well site.


Comments on remediation

Remediation of sites such as described would consist of source
removal from  soils  and recovery  of water  containing dissolved
contaminants.    Source removal  in the  soils  would  proceed
conventionally and will not be further addressed.   Suggestions
are made relative to  fluid  recovery,  however, with respect to
siting recovery wells, improving their connections to the flow
system,  and using  a  vacuum  assist  to  provide  an  increased
effective drawdown to better connect to the subcutaneous system
and "reach deeper into the  system".   The  following  ideas are
conceptual and have  been proposed but not yet tried in practice.

Siting of the recovery well should be as close to the main drain
as possible consistent with property  ownership  and other site
constraints.   This  will put the well   in the best position to
intercept flow from the site toward the main drain.  It should
also be placed where  the  top of  rock  is  locally deep to allow
for a greater cone  of depression.    If water  loss data was
obtained by  rotary  wash  boring or coring,  a  location  where
higher permeability  was observed should be strongly favored over
other considerations.   It may  even be desirable to do special
probing in an area of pinnacled rock where the recovery well is
to be sited to select the best location.

Head loss in communicating between subcutaneous  system and the
recovery well could  greatly limit the effectiveness of the well.
If the  well  is  in  soil, this connection  may be enhanced by
jetting  a cavity into  the  soil  at top of  rock  using  a high
pressure jet of water.   This would then be backfilled with sand
to stabilize  the hole.  In rock,  the bedding planes might be
separated to communicate to the enhanced routes in the rock by
hydrofrac at fairly low pressure.

The primary  suggestion made  for  recovery  is  to  used vacuum
assist  to  increase   the effective cone  of depression  of the
recovery well. A major limitation  on  recovery is  that drawdown
                             107

-------
of a  well is  limited by the  depth of enhanced permeability.
Regardless  of  the  depth  of  the  well,   the normal  cone  of
depression cannot be drawn down  deeper than the level at which
the water is  entering the well from  the enhanced  joints  or
bedding planes.  This means the cone of depression cannot reach
"downhill" in  the  subcutaneous system  toward the master drain
and if head losses are experienced due to poor communication to
the enhanced  flow  routes,  even interception  of  local flow may
not be possible.

Application of a vacuum to a well results in an increase in the
head available to water to enter the  well.   This has been used
for decades by the dewatering  industry to dewater fine-grained
soils.   An  effective drawdown of more then  20  feet  below the
bottom of the well  is often practical  if the well is constructed
with a vacuum in mind.   If a dual pump  system  is used where
water  is pumped  with  a downhole  pump  while  the  vacuum  is
maintained from the surface, there should also be no water lift
limitation on the drawdown obtained.

Figure 7  shows  the basic concept.  A vacuum  assisted recovery
well is shown that is near but not in  direct communication to a
subcutaneous enhanced pathway.  Part of the effective drawdown
       Capture of flow
    from adjacent branches
        Cone of depression
 vacuum assisted
.  r&coveryjt/ell
             Figure 7  Vacuum assisted recovery well
                             108

-------
is lost in the low permeability  connection to this flow path.
Once the enhanced  flow path is reached, however,  the cone of
depression would  expand preferentially along it.  If the cone of
depression is deep enough, this extension along the subcutaneous
enhanced flow  routes  will  reach  toward  the  main drain  and
reverse the gradient  of the  pathway below the junction point of
other branches.   Thus,  this  vacuum  assisted well can gather the
flow  from  parts of  the  subcutaneous   pathways  outside  its
immediate vicinity and can stop further discharge of water from
the site.
Conclusions

Movement  of water  and  hydrocarbons  through karsts  with  a
subcutaneous zone will  follow macropores  in the soils  and
solutionally enhanced channels  in the rock or top of rock toward
master drains into an underlying conduit system.   When a water
table is present in the soils,  this produces a cone of depres-
sion in  the soils.    The  subcutaneous network is  a  tributary
network  that  is  integrated  toward the  master drain, but  is
probably not integrated laterally.  Separate phase hydrocarbons
follow the macropores preferentially  and may  be  able  to enter
subcutaneous systems due to reduced capillary resistance and the
mounding effect of the plume.

Assessment in karst areas should  focus on determining whether
the site is underdrained, where the  intakes to the underdrain
system are, if present, and  how to monitor.   Close piezometer
networks and the  geologic and  topographic context  of  the site
often provide the  best clues, but use  of an existing hydrocarbon
plume  as a  tracer and  top of  rock  configuration  are  also
valuable.  Pumping poorly producing monitor wells  at least 10 to
20 gallons may improve detectability in the subcutaneous zone.
Fluid recovery for  remediation can be improved by  location of
the recovery well as  close  as practical  to the  drain intakes
defined  above,  improving the  connection to  the  subcutaneous
system by hydrofracing or jetting,  and using a vacuum assist to
improve  the effective  drawdown  of  the  well.   The  improved
effective drawdown is intended to reach deeper into the system
to  intercept  flow  of adjacent branches of  the  subcutaneous
system that connect  to the drain down-flow of the recovery well.
                             109

-------
References

Ewers, R.O., 1989. Cave development and patterns of ground water
     flow in karst: in course manual of Practical  karst
hydrogeology, NWWA short course, Dec, 1989, Volume 3, p.    D-l
to D-80

Environmental Protection  Agency,  1989, Transport  and fate of
     contaminants  in  the  subsurface,  Seminar   publication,
     EPA/625/4-89/019, 148 p.

Gunn, J. ,  1981,  Hydrological processes in karst  depressions,
     Zeit Geomorph, vol.25, no.3, pp. 313-331

Smart, P.L. and Friederich, H., 1986,  Water movement and storage
     in the unsaturated zone of a maturely karstified carbonate
     aquifer, Mendip Hills, England:  Environmental Problems in
     Karst Terranes and their Solutions Conference, Oct  28-30,
     1986, Bowling Green,  Ky, proceedings p. 17-31

White, W.B., 1977,  Role of solution kinetics in the development
     of karst aquifers:  in Karst Hydrogeology, Tolson and Doyle
     ed., Memoirs  of the  llth  Cong,  of the  IAH,  Huntsville,
     Ala., UAH Press

Williams, P.W.,  1983,  The role of the  subcutaneous zone in karst
     hydrology:  Journal of Hydrology, v. 61, p. 45-67


Biographical sketch

Tony  L.  Cooley  has  a B.S.  in  Geological Engineering from
Washington University in St. Louis and has done two years each
of  graduate work  in  engineering geology/soil  mechanics  at
Cornell University and rock mechanics at The Pennsylvania State
University-  He  has  worked as a consultant for  13 years as a
geological/geotechnical engineer, the last four years of which
were  mostly assessments  of hydrocarbon  releases.   He  is  a
registered engineer  in  Alabama and a  registered geologist in
Georgia and Tennessee.  At present he is a PhD graduate student
in hydrogeology at the University of Kentucky in Lexington.

Address;       Department of Geological Sciences
               University of Kentucky
               Lexington,  Kentucky 40506-0059

Phone:         606-233-9511
                             110

-------
    Approaches to Hydrogeologic Assessment and Remediation of
                  Hydrocarbon  Contamination  in
          Clay-Covered Karsts with Shallow Water Tables

                           Tony Cooley

1.   Please comment on methods of detection of hydrocarbon
contamination, especially with respect to economics.

The basic  analyses  used were  BTEX  (EPA Method 8020 or  602)  and
Total Petroleum Hydrocarbons  (TPH) by  either EPA  Method 418.1 or
California Department of Health Services  (CAL-DHS)  modification of
EPA Method 8015.  The CAL-DHS method is the more expensive of the
two TPH methods but has fewer false positives.  In clays, some of
the  non-polar  naturally  occurring  hydrocarbons  sometimes  get
polarized and get through the silica gel cleanup and show  on the IR
detector of the 418.1 method.  For specific cases, a fuel identifi-
cation using the GC portion of the CAL-DHS analysis  and a sample of
the free product may be done, but this is not common.

Economy  is achieved  by  limiting the  number  of  analyses  run.
Particularly for light  fuels, all soil samples are screened by soil
sample headspace  analysis in  the field  with a  PID or FID  and
profiles are  plotted.    This  information, together  with strati-
graphic and  water table information,  are used in  selecting  the
samples  for  analysis.    Boring   locations  and number   are  also
carefully  selected  and drilled  in  a certain  order so  that  the
results of each boring  are used  for  field decisions to influence
the implementation  of  the work plan.   This  results in  the best
economy.  As  noted in  the  paper,  once  the soils and ground water
are sufficiently delineated, further borings for  piezometers  are
logged by cuttings only but not sampled.
                                Ill

-------
112

-------
            APPLICATION OF DYE-TRACING TECHNIQUES FOR
         CHARACTERIZING GROUNDWATER FLOW REGIMES AT THE
                FORT HARTFORD MINE SUPERFUND SITE
                   OLATON, OHIO CO., KENTUCKY
                   Nicholas C.  Crawford,  Ph.D.
            Director,  Center for Cave & Karst Studies
                   Western Kentucky University
                        Bowling Green, KY
                       Ginny L.  Gray,  R.G.
              Site Manager, Ft. Hartford Mine Site
              Environmental & Safety Designs, Inc.
                           Memphis,  TN
This abstract is being submitted  for  the  Hydrogeology Session of
the conference.


This paper discusses  the application of dye-tracing techniques for
determining the  groundwater flow  paths  and  characterizing  flow
regimes above a  120 acre underground limestone mine.  The inactive
mine is  being used  to store  greater than  1.7  million   tons of
aluminum manufacturing by-product that, when wetted, emit toxic and
noxious gases.  Substantial quantities of  water are  entering the
mine and levels  of ammonia gas exceeding Immediate Danger to Life
and  Health  (IDLH)   concentrations,  require  that  all  personnel
entering the mine use supplied air  systems.  This investigation is
part of a Remedial Investigation/Feasibility  Study (RI/FS)  being
conducted  at  the   700  acre  site.    One  objective  of  this
investigation  is  to  identify   the  potential   and/or  actual
groundwater flow paths into the mine.   Greater than 25 sinkholes,
and 31  springs were discovered onsite  through aerial surveying and
field reconnaissance.  Dye  receptor packets were located in each
spring  and  at all water  intrusion points  within the mine.  The
results of 15 traces  suggest that  there are two aquifers located
above the mine,  which have the potential to "leak" into the mine:
a  turbulent  flow limestone aquifer,  which is perched above and
serves  as  recharge for,  a  laminar flow shale/sandstone  aquifer.
This evaluation is substantiated by residual  dye detected in the
largest spring up to  109 days after dye injection.  Sinkholes found
to contribute water  to  the mine  are  located in areas where the
faulting has displaced the low permeability shale/sandstone unit.
The  remaining  sinkholes  transmit  water  laterally  along  the
limestone shale/sandstone contact located above the mine roof and
discharge into surface streams.   A large construction project was
performed following the  dye trace to plug sinkholes and to divert
surface runoff away from the mine.
                                113

-------
              Application of Dye-Tracing Techniques
           for Characterizing Groundwater Flow Regimes
             at the Ft. Hartford Mine Superfund Site
                  Olaton, Ohio County, Kentucky

                       Nicholas  C.  Crawford
                   Western Kentucky University
                     Bowling Green, Kentucky

                          Ginny L.  Gray
               Environmental  & Safety  Designs,  Inc.
                       Memphis, Tennessee

INTRODUCTION

The Ft.  Hartford Mine Site is an  underground  limestone mine located
in rural western  Kentucky.   Since 1981  by-products of secondary
aluminum recovery, called salt cake fines have been stored within
the  interior  of  the  mine for  potential aluminum  and chlorides
recovery by Barmet Aluminum Corporation of Akron, Ohio.  Salt cake
fines are  a light  gray  powdery substance  regulated  as  a solid
waste.   In  a  dry  state the material  is  inert,  however,  when the
fines come  in  contact with water  several  gases  are generated:
acetylene,  methane, hydrogen sulfide,  and ammonia.

Investigation began at the site  due to environmental concerns that
ammonia gas, chlorides, and possible metals were posing a threat to
human health and  the  environment via numerous  pathways  from the
site.   In  June  1988,  the  U.S.  Environmental  Protection Agency
(USEPA)   proposed  that  the  Ft.  Hartford Mine be  added  to  the
Proposed National  Priorities List  (NPL) or "Superfund"  list for
investigation and  remediation under the Comprehensive Environmental
Response, compensation, and Liability Act (CERCLA).  Ft. Hartford
was placed on the final NPL on August 30, 1990.

On September  20,  1989 Barmet  Aluminum  signed  an  Administrative
Order by Consent  (AOC) with  USEPA  Region IV to perform Expedited
Response Actions  (ERAs)  and  a Remedial  Investigation/Feasibility
Study (RI/FS)  for the Ft. Hartford Mine  Site.  Two of  the primary
objectives of  the  ERAs were to:   1) identify the areas  where water
was entering the mine,  and 2) to prevent the  water intrusion and/or
relocate salt  cake fines away from the wet  areas.

When  ERAP  work began in  March  1990  there  were  five  apparent
locations where water  intrusion was occurring into the mine.  These
                                 114

-------
five areas were collapse locations which occurred  during limestone
mining  in areas of  limited roof  cover.   These collapses  allowed
direct  intrusion of  surface water into  the mine.   The mine had
accumulated  approximately  37 million gallons  of  impounded  water
since  mining began  in  the late  1950s.   Additionally an  aerial
survey had located numerous "closed depressions", commonly referred
to as sinkholes  above the mine.   All  the sinkholes were scheduled
for engineered  closure during the ERAs.

Since  the sinkholes  would be  closed  prior to  initiation of the
RI/FS,  the  karst investigation  was conducted concurrent with the
ERAs  to avoid losing  data pertinent  to  the characterizations of
groundwater   flow  regime(s)  at  the   site.     In  addition,  a
determination of  which sinkholes,  if  any, and the amount  of  water
that  they were  contributing to  the mine could be made.

PHYSICAL  SETTING

The Ft. Hartford Mine Site is located  in the east-central perimeter
of the  Western Kentucky Coal Field  physiographic province.   The
Coal  Field is located  within the Interior Low Plateaus  province of
the Interior Plains  Region as shown in Figure 1.
                                             FT. HARTFORD MINE SHE

                                             PENNSYLVANIAN  DUTCRDP
                                             EDGE  DF COASTAL PLAIN
          Mississippi
   ARK,   rEnbQ
   SOURCE: Hydrology and Geology of Deeo Sandstone Aquifer of Pennsvivanian
        Age in_Pgrit_sfLjthg_Wester.n_-Coalfleld Region. KenjycB^. 1974	
                                                       FIGURE 1

                                                  REGIONAL GEOLOGIC SETTING
                                                  DATE: 5 731/9Q
                                                          I DWG Njy^EL
Geologically,  the  area  is  in the  southeastern section of  the
Eastern Interior  (or Illinois)  Basin.  This region is  characterized
by  low   rolling  hills   formed  of  Pennsylvanian  age   shales,
                                 115

-------
siltstones, and sandstones with occurrences of coal and limestones.
However, the Ft.  Hartford  Mine Site is  located within the Rough
Creek Fault Zone  and Mississippian age sandstones and limestones
are exposed as  a result of normal and high angles reverse faulting.

The alluviated  valleys  comprise a small portion of  the area and
have  a  general elevation  of 380  to  420  feet  (msl) ;  the hills
surrounding the site rise to a maximum elevation of about 625 feet.
Surface  topography  in the  Ft.  Hartford Mine  area  is primarily
rolling  hills.   Slope grades  range from  5  to  20 percent.   The
highest and lowest points on the site form approximately 250 feet
of relief.

The majority of the  area above the  mine is also  heavily vegetated.
Two large  valleys traverse the site and trend northeast towards
Caney Creek and Rough River.  The most prominent surface features
found on the property are  the  numerous karstic features present.
Generally, these karstic features are identified by their surface
expression as a closed topographic  depression.   The "sinkholes" as
they are known,  have the potential of transmitting large quantities
of surface water  into the  underlying limestone aquifers.   At the
Ft. Hartford Site  the sinkholes appeared to be transmitting surface
water downward  into the formations  which  overly the  mine roof.
Three surface water streams, Caney Creek to the east, Rough River
to the north, and  Cane Run to the west provide the primary drainage
for the Ft. Hartford Site.

REGIONAL GEOLOGY

Mississippian age  sedimentary rocks are exposed at the Ft. Hartford
Site and throughout western  Kentucky along  the Rough Creek Fault
Zone,  a major  faulted anticlinal  feature  cutting  through  the
Western  Kentucky  Coal  Field (Figure 1  and Plate 2)  .   The Rough
Creek Fault Zone represents one segment of  a major lineament that
stretches  from  the Appalachian  region,  across  Kentucky,  and into
the  Fluorspar  Districts   of southern Illinois,  Kentucky,  and
Missouri.

The Rough  Creek Fault  System is a segment of  the 38th parallel
lineament, the designation given to a major basement  fault system
that stretches from  the Appalachians to the Ozarks and possibly to
the Rocky Mountains.  The New Madrid Fault System extends from Mark
Tree, Arkansas  to  Wabash, Indiana and crosses and connects with the
38th parallel fault  system.  At the confluence  of these two fault
systems is the  Hicks Dome  or anticline.  The Hicks Dome has been
described as a cryptovolcanic structure composed of mafic igneous
intrusive  rocks.    These   igneous  intrusions  are suggestive  of
incipient rifting along these major lineaments  (Hook, 1970).  The
fluorspar  districts of  Illinois,   Missouri,   and  Kentucky  are
developed within these intrusive rocks and their associated mineral
assemblages.  The other major faults that stretch to the east and
west from  the fluorspar district are the Rough Creek and Shawnee
Town fault segments  of the  38th  parallel system,  the St. Genevieve
and the Cottage Grove Fault Systems.
                                116

-------
The Rough  Creek Fault  System is  a positive  east-west trending
overturned anticlinal structural complex where uplift has shifted
Mississippian sedimentary rocks  in  the  Ft.  Hartford Mine area up
into a  primarily Pennsylvanian  sedimentary rock  terrain.   This
system is a compressional fold and is usually faulted as a single
high-angle overthrust from the south that develops into a complex
pattern  of  block  faults  that exhibit  thrusting,  dip-slip,  and
strike-slip movement  (Rehn,  1968) .   The  displacement along this
fault system  is divided along a series of step  faults and drag
faults.   These  step  faults  are predominantly gravity faults that
progress in a step-like manner from horsts to grabens.   One normal
step fault segment  crosses the Ft. Hartford Mine Site and the fault
cut is plainly evident in an outcrop seen  near the  entrance to the
Rough River side  of  the mine.   These  step faults  usually dip
between  60-90  degrees,  but  associated  reversals  of these faults
typically produce  dips within  5 degrees  of vertical.   Grabens
caught between the  major step fault systems can develop tension and
shear fractures  resulting  in the development  of  additional step
faults,  antithetic faults and cross  faults  in  the graben blocks.
Antithetic faults are generally parallel to the  strike of the fault
system,   have  dips  between  45-65  degrees  and  terminate  at  the
juncture of a major step fault.   Alteration  along the actual fault
zones usually consists of simple silicification in sandstones and
only slight alteration in limestones and dolomites.  Some strike-
slip movement has been noted along the rough Creek Fault  System due
to torsional forces with the north wall  of the fault moving to the
west (Heyl and Brock,  1960).  This movement is generally  limited to
the  western portions  of this   fault  systems  and is  in  direct
response to the  wrenching movement  related  to  the emplacement of
the Hicks Dome  mafic  anticlinal  feature in the Kentucky-Illinois
Flurospar Region (Hook, 1972; Trace, 1970).

RESEARCH PROCEDURE

To trace the groundwater flow patterns  in  the vicinity of the mine
and to determine which sinkholes, if  any,  were contributing to the
impounded water  in each lobe of  the mine  (Caney  Creek and Rough
River),   the  following research procedure was  recommended  by Dr.
Nick C.  Crawford of the Western Kentucky Center for Cave and Karst
Studies and implemented at the site.  This procedure is  consistent
with  that  in the  USEPA report  by  Mull,  Liebermann,  Smoot,  and
Woosley  (1988) .

The first step was  to conduct a comprehensive field reconnaissance
to inventory the karstic features within the  area of investigation.
The  field  reconnaissance involved  locating (field  mapping)  all
sinkholes,  springs,   caves,  karst   windows,   sinking  streams,
lineaments,  and areas  inside  the  mine where  significant water
intrusion was  occurring.  After  all karst  features and areas of
water intrusion  were  located,  a passive  dye receptor packet was
placed at  each location.   the  receptor packets used during the
study were small packets of  activated coconut  charcoal contained
within aluminum or fiberglass screen  mesh and small bundles of
surgical cotton. A total of 36 dye receptor packets were placed at


                                117

-------
springs and in streams above ground,  5 were placed inside the mine
at water  intrusion points  on  the Caney  Creek Side,  and  8 were
placed at  water intrusion points on the  Rough River  side for a
total of 49 dye receptor  locations.   Each receptor locations was
given a  letter and a  number that corresponded to  the group who
placed it and the order in which it was placed.

After these receptors  had been in place approximately one week, six
locations  were selected  to be  analyzed  in  the laboratory for
background fluorescence to create a baseline for comparison.  Water
samples were  also collected from major springs and streams and
analyzed for background fluorescence.   Results showed that there
were no problems with  background.  Each receptor location selected
for background was replaced  with a new packet.   Plates 1 and  3 show
all receptor locations aboveground and inside the mine.

Once the inventory of  the areas  was  completed,  the  dye receptors
were  in place,  and  background  fluorescence  was   analyzed,  dye
tracing was initiated.  The  four  dyes selected for  the study are
recommended by USEPA guidance documents  (Mull, Liebermann,  Smoot,
and Woosley, 1988) and are as follows:

a)   Fluorescein,  Color Index:   Acid Yellow 73
b)   Optical Brightener, Tinopal  5BM  GX, Fabric Brightening Agent:
     22
c)   Rhodamine WT, Color Index:  Acid Red 388
d)   Direct Yellow 96, Diphyenyl Brilliant Flavine 7GFF

These are  the  standard dyes most often used for dye  tracing in
karst  aquifers.    They  are  safe   for   this  purpose  in  the
concentrations used both  for human  consumption and aquatic life
(Smart and Laidlaw, 1977 and Smart,  1986).

Since the dye traces were  conducted during the summer months  (dry,
low flow  conditions),  all sinkholes were  "flushed"  with potable
water  in  lieu  of  waiting  for  them to be  flushed  naturally by
rainfall.  A minimum of 500 gallons of potable water was injected
into the openings to ensure that they drained sufficiently and to
wet the soil to minimize dye sorption onto  clays.  The dye was then
injected and flushed with  a  minimum of 1,000 gallons of additional
water.  Dye tracing was performed on a  maximum of  four sinkholes
per  test  before  the  receptors  were  collected,  analyzed,  and
replaced with new packets  for the following traces.

Following dye  injection and "flushing"  with water,  the receptors
were left in place a minimum of one week.   By allowing the dye to
flush through the groundwater system prior to the next trace, the
dye could be reused on subsequent sinks  without confusing the
results from previous  traces.  After the minimum waiting period,
all the  receptor packets  aboveground and  inside  the  mine were
collected for analysis.

The charcoal packets were  washed  and  elutriated with a  solution of
ninety-five percent isopropyl alcohol and a five percent potassium
                                118

-------
hydroxide  to  bring  the  Fluorescein  dye  to  the surface  of the
charcoal.   The packets were then visually observed for the presence
of dyes; if there was a question concerning the interpretation of
a  charcoal dye  receptor,  the  elutriant was  compared  with the
elutriant of the background  receptors on  a  Turner Fluorometer in
the Hydrology Research Laboratory at Western Kentucky University.
Rhodamine WT was usually visible on  the  charcoal  because of the
relatively  large  quantities used.    However,  the  elutriant was
always analyzed on  the Turner Fluorometer  for all Rhodamine traces
and  compared  with  background.    Rhodamine  WT,  if  present,  was
usually visible  when treated  with the  above elutriant  but the
"Smart" elutriant  consisting of  a  5:2:3 mixture  of 1-propanol,
concentrated NH40H  and  distilled water  was used  to analyze for
Rhodamine WT as needed (Smart and Laidlaw, 1977).

The surgical cotton dye  receptors were washed to  remove as much
dirt and debris as  possible and  then tested  for Optical Brightener
and Direct  Yellow  under  a long-wave  ultraviolet  lamp.   Positive
traces for Optical Brightener appear as a blue-white glow under the
lamp and Direct Yellow,  appears as a pale yellow glow, when it is
present.

DYE TRACING RESULTS

Dye tracing began with Sinkhole S10 on April  30, 1990  and continued
until September 30,  1990 with the final dye traces at  Sinkholes S9,
S3,  and S2.   Extremely  dry weather  and residual  dye  problems
account for the long duration of the karst  hydrology investigation.
Table  1 summarizes  trace results with the  sinkhole,  type of dye
used,  date injected, and  the  dye  receptor locations  that were
positive  for  each   trace.    Plates  1  and 3  show the  dye traces
 (straight-line)  from sinkhole injection point  to spring or water
intrusion into the  mine.

The  three most  important springs  in  the  vicinity of  the Ft.
Hartford  Site  are  II,  E2,  and E3.    Spring E3,  located  west of
Collapse locations  4 and 5,  is flowing from a small cave at the
based of the Glen Dean Limestone  (Mgd) (Plates 1,2,3, and 4).  It
is perched directly upon the Hardinsburg Sandstone Formation  (Mh)
which is primarily  shale  in  the  vicinity of  the Ft. Hartford Mine.
The surface stream which flows  from  Spring E3 has  a  very steep
gradient as it  breaches  the Hardinsburg  perching  layer  which is
about twenty feet thick at this  location.  As the stream flows off
of the Hardinsburg  (Mh) it sinks into the Haney Limestone  (Mgh) and
then flows to  Spring E2 for a final resurgence before flowing into
Cane Run Creek.   (All dye traces  which were  positive  for Spring E3
were also positive  for Spring E2.  However,  a dye trace from E3 to
E2 has  not been made to  confirm this assumption) .   During high
discharge when the  cave stream cannot accept  the entire flow of the
stream from E3,  some of the water flows on the  surface to Cane Run.
Spring  E3  is  located along  a  pair of closely-spaced,  east-west
faults and the geologic structure is extremely complicated  in this
area (Plates 2 and  4).
                                119

-------
                                                                 TABLE 1
                                                    SUMMARY OF DYE TRACE RESULTS
           Dye Trace
           Number
Sinkhole
Dye
 Date
Injected
Positive Dye Receptors
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
S10
SUB
SI
S11A
SI
S8B
S8A
SI
S6
S5B
S12
S7
S2
S4
S5
S2B
S9
S3
1 Ib. Fluorescein
2 liters Rhodamine
5 Ibs. O.B.
4 Ibs. D.Y.
4 Ibs. O.B.
1 gal. Rhodamine
2 Ibs. Fluorescein
2 Ibs. Fluorescein
5 Ibs. O.B.
1 gal. Rhodamine
4 Ibs. D.Y.
5 Ibs. O. B.
4 Ibs. D. Y.
Flush Pulse
1/2 gal. Rhodamine
4 Ibs. D. Y.
4 Ibs. O. B.
2 Ibs. Fluorescein
4-30-90
5-1-90
5-2-90
5-2-90
5-24-90
5-23-90
5-24-90
6-11-90
6-12-90
6-13-90
6-13-90
8-10-90
8-13-90
8-20-90
8-17-90
9-18-90
9-20-90
9-20-90
- (Not detected because Spring 11 was not found until 6-1 1-90)
Spring Kl above Breakthrough 2
- (Not detected because Spring 11 was not found until 6- 11 -90)
Spring Gl and into Cane Run
- (Not detected because Spring 11 was not found until 6-1 1-90)
Springs £3, E2, and into Cane Run
Springs E3, E2, and into Cane Run
Spring 11 and into Rough River
Inconclusive trace
Spring 11 and into Rough River
Mine Intrusion RR2 and Springs E3, E2 and into Cane Run
Mine Intrusion RR1 and Spring E3, E2 and into Cane Run
-Not detected (probably still in soil)
Mine Intrusion CC-R16
Mine Intrusion RR1 and RR2
Spring Cl
Mine Intrusion CC-R22
-Not detected (may have flowed to an intrusion in the flooded
sections of the mine).
NJ
O

-------
Dye traces from Sinkholes S8B, S8A, S12, and S7 were all positive
at Springs E3 and  E2.   All of these  sinkholes  are located on or
near the east-west  trending faults. The dye traces from Sinkholes
S12 and S7 were also positive at  water intrusions into the Rough
Creek lobe RR2 and RR1 respectively, indicated that the Hardinsburg
confining  layer  leaks  in  the vicinity of these  faults allowing
water to sink into the mine.   Dye traces  of  Sinkholes S4 and S9,
both located along or near these faults, flowed directly into the
Caney Creek lobe at  water intrusion  locations  CC-R16  and CC-R22
respectively and did not  flow to Springs E3 and  E2.  The dye trace
of Sinkhole S5  also located  along or  near these faults,  flowed
directly into the Rough Creek lobe at  water intrusions RR1 and RR2
 (Table 1, Plates 1 and 3).

Spring II is located along  the Rough River north of the Rough River
mine entrance.  It flows out of a massive pile of rocks displaced
by the  mining operation and  cascades about  three feet  into the
Rough River.   It appears that it probably resurges from a cave at
the  based of the  Glen  Dean Limestone  (Mgd) ,  perched  upon the
Hardinsburg  Sandstone   (Mh)  confining  layer,  which  has  been
completely buried by mine  debris.   The small  valley  north of the
Rough River mine entrance has been filled with rock from the mine
entrance to the  river.  The spring may  have originally been located
in  this valley  along the Glen  Dean  (Mgd)     Hardinsburg   (Mh)
contact.

Spring II was not found during the  initial  field  reconnaissance due
to high flow conditions in the Rough River.  Consequently, attempts
to dye  trace Sinkhole 1  on 5/2/90 and 5/24/90 and Sinkhole 10 on
4/30/90 were unsuccessful.  As summer approached,  the river level
dropped, and the spring  revealed  on 6/11/90.   Sinkholes SI,  S10,
and S5B were all traced  to Spring II.  Sinkhole S6 dye traced on
6/12/90 may also flow  to Spring II.   However,  there  is a chance
that the positive cotton dye  receptor at  II  may have been due to
residual Optical Brightener still  resurging at the spring from the
Sinkhole 1 traces of 5/2 and 5/24/90.   After performing traces of
sinkholes near S6 which all went into  the mine it  seems reasonable
that  the drainage from  S6  goes  there also.   However,  Optical
Brightener was not  detected on any dye receptors in the mine or on
the surface other than the one at Spring II,  but since there is a
chance  that it  may have been residual dye form  the  Sinkhole SI
traces,  the  results  of  the  Sinkhole  S6  trace  were  considered
inconclusive.

Plates  1  and  3  show  the  dye  tracing  results  at the  Ft. Hartford
Site.  Plate  1 shows the  approximate groundwater basins identified
by the dye traces.  Usually,  static water  levels are obtained from
water wells and monitoring wells which permit equipotential lines
to be drawn for the uppermost aquifer.  This information confirms
dye  tracing  results  and  permits  drainage basin divides to be
delineated.  Unfortunately there  was  only one  water  well in the
area  and monitoring  wells  have  not  yet been installed.   The
equipotential lines also provide   important  information,  such as
troughs in the water table,  which often indicate the approximate
                                121

-------
routes taken by cave streams from dye  injection points  to  springs.
Without the  water table data,  only straight  lines  were used  to
indicate the  flow from injection points  to discharge  locations.
With water table data this lines could have been curved to show the
approximate flow  routes of  cave  streams.   The highs in the water
table, indicative of drainage basin divides, would have permitted
more  accurate  delineations  of  the  groundwater basins.   Plate  1
shows  the  approximate drainage basins  based  on  the dye traces,
topographic divides, and geology.

CONCLUSIONS
The dye tracing investigation revealed that water sinking into  some
of the sinkholes  in  the Glen  Dean Limestone (Mgd) above  the  mine
flows along  solutionally enlarged conduits that are perched  upon
shale layers in the underlying Hardinsburg Sandstone  (Mh). In  most
areas the Hardinsburg (Mh) confining layer appears to prevent water
in the Glen  Dean  (Mgd)  from sinking into the mine located in the
Haney Limestone  (Mgh).  However, sinkholes located on  or  near the
east-west faults  which  cross  the  southern portion of the mine  do
permit water from nearby sinkholes to  flow through the  Hardinsburg
(Mh)  into the mine  (Plates 3 and 4).

The following dye traces of  sinkholes in the Glen Dean Limestone
(Mgd) located above  the mine revealed that water sinking into these
sinkholes flows to springs  and does not  flow into  the mine (Plates
3 and 4).

SINKHOLES                          DISCHARGE SPRINGS
SI                                 Spring  II and  into  Rough River
S10                                Spring  II and  into  Rough River
S5B                                Spring  II and  into  Rough River
S8A                                Spring  E3/E2 and  into  Cane Run
S8B                                Spring  E3/E2 and  into  Cane Run
Sll                                Spring  Kl and  into  Collapse  2
S11A                               Spring  Gl  and into Collapse  5
S2B                                Spring  Cl and  into  a tributary
                                   of Caney Creek

The following dye traces in the Glen Dean  Limestone  (Mgd) located
above  the  mine revealed that water sinking into these sinkholes
flows  through  the Hardinsburg  (Mh) confining  layer and  into the
mine.   All of  these  sinkholes  are on or  near the  two parallel,
east-west trending faults which cross the  southern portion of the
mine and it appears  that water draining  from these sinkholes flows
through the  Hardinsburg (Mh)  and  into  the Haney Limestone  (Mgh)
along these  faults  (Plates 3 and  4).

SINKHOLES                          DISCHARGE SPRINGS
S4                                 CC-R16     (Caney  Creek lobe)
S5                                 RRI  &  RR2  (Rough  River lobe)
S9                                 CC-R22     (Caney  Creek lobe)
S12                                RR2      (Rough River lobe) and
                                             also to E3/E2 and into
                                             Cane Run  Creek
                                122

-------
S7                                 RR1         (Rough River lobe)
                                             and also to E3/E2 and
                                             into Cane Run Creek

Primarily due to faulting,  the hydrogeology in  the vicinity of the
Ft. Hartford Mine is extremely complicated.  However, the results
of this investigation permit  the following tentative conclusions
about the hydrogeology.

The  Tar  Springs  Sandstone  (Mts)  does not  restrict water  from
percolating  through  it  into  the Glen   Dean Limestone  (Mgd).
Therefore, there does not  appear to be a  water  table in the Tar
Springs  (Mts).    Sinkholes in  the Tar  Springs  Sandstone   (Mts)
support this conclusion.  Water percolating through the Tar Springs
(Mts) has  dissolved  voids  in the  underlying Glen Dean Limestone
(Mgd).    The Tar Springs Sandstone (Mts)   has  slumped into these
voids and  the  erosion of sand  through  the solutionally enlarged
conduits in the Glen Dean  (Mgd) to springs is  the method by which
these  somewhat unusual  sandstone  sinkholes  have  been created
(Plates 3 and 4).

The Glen Dean Limestone  (Mgd)  has a karst-conduit, turbulent flow
aquifer perched upon the Hardinsburg Sandstone (Mh) which in this
area contains  numerous  shale  layers.   The aquifer  is  recharged
primarily  from surface  runoff  flowing  into sinkholes.    It  also
receives recharge from water percolating downward through the Tar
Springs Sandstone  (Mts).   Recharge  water  flows  rapidly through
solutionally enlarged conduits at the base of  the Glen Dean  (Mgd)
to springs  usually  located at the Glen Dean   (Mgd)  - Hardinsburg
Sandstone  (Mh)  contact (Plates 3 and 4).

The Hardinsburg Sandstone (Mh),  primarily  a shale  in  this area, is
responsible  for perching  the  karst  aquifer  in  the  Glen  Dean
Limestone  (Mgd).  In most areas it prevents the vertical movement
of water from the Glen Dean (Mgd) karst aquifer from sinking into
the  Haney  Limestone.   The dye  trace  investigation supports  this
conclusion as does the scarcity of  solutional features intersected
by the mine.   However,  in  some areas  faulting  permits water from
the Glen Dean (Mgd) perched aquifer to  leak through the Hardinsburg
(Mh)  directly  into  the  Haney  Limestone   (Mgh).    Dye  traces  of
sinkholes on or near the pair of east-west faults which cross the
southern portion of the mine support this  conclusion.

The  dye  study  also supports the conclusion that the Hardinsburg
Sandstone  (Mh)  contains  a porous-media, laminar-flow  aquifer.  The
top of this aquifer is the water table in  the  Glen Dean Limestone
(Mgd)  and  the  bottom   is perched  upon  shale  layers  in  the
Hardinsburg  (Mh).   It appears that very  little  water is leaking
through the  Hardinsburg (Mh)   except  along faults and  in places
where it is thin and weathered.  All five mine collapses occurred
in  places  where surface  streams  were  flowing  over   thin and
weathered Hardinsburg (Mh).  Most of the water  intrusions into the
mine (except those along faults) occur along the periphery of the
mine where mining was stopped because it was approaching a valley
                                123

-------
where the  overlying  Hardinsburg (Mh) is  thin and weathered  thus
permitting water intrusion into the mine.

The residual dye problems, at Springs II, E3,  and E2,  also indicate
two types of aquifers.   The  rapid  flow through from sinkholes  to
springs  revealed a karst-conduit  turbulent-flow  aquifer at the
bottom of the Glen Dean Limestone (Mgd) .  However,  residual dye was
still flowing  from Spring II  109  days  following dye  injection.
This  appears  to indicate  that the porous-media,  laminar-flow
aquifer in the Hardinsburg (Mh)  is  recharged  from  the conduit-flow
aquifer in the Glen Dean  (Mgd).  Therefore, the dye flows rapidly
through the conduit from the  sinkhole to the  spring but  the dye  in
the Hardinsburg  (Mh) aquifer is  released slowly over a period  of
several months.

The  Haney  Limestone  (Mgh),  protected  by  the Hardinsburg   (Mh)
confining layer, receivers very little recharge from above except
where  there are faults  or  the   Hardinsburg  (Mh)  is  thin and
weathered.  Very little water is entering into the mine  except  in
these areas.

There is  a shale confining  layer  at the  top of  the  Big Clifty
Sandstone (Mgbc). Johnson and Smith (1968)  in their description  of
the Big Clifty Sandstone  (Mgbc)  on the Olaton Geologic Quadrangle
refer to a  "clay  shale  common  in the  uppermost part."  Their
stratigraphic column for the  Big Clifty Sandstone  (Mgbc)  indicates
that this shale is  about 8 to 15  feet thick.   In other areas where
Crawford and Dotson (1990) have worked,  this shale confining layer
often perches springs which flow from a  karst  aquifer in the Haney
Limestone (Mgh) . Several springs at the Ft.  Hartford  Mine Site are
located at  the Haney  Limestone  (Mgh)-Big Clifty Sandstone (Mgbc)
contact.  They are springs E2,  A6,  C5,  Dl, and C4.  These springs
are  small  and  cease  to flow during dry periods.   This may  be
indicative  of  the  relatively  small amount  of water  which   gets
through  the  overlying  Hardinsburg   (Mh)   confining layer  into the
Haney Limestone  (Mgh).

The shale confining layer at the top of the Big Clifty  (mgbc) may
also  be responsible for the lakes  inside the  mine.    The   lake
elevations  appear  to  correspond  to the dip  of   the  Big Clifty
Sandstone (Mgbc) although probably  not the true dip.  From north  to
south the lake surface elevations on the map  of the mine  decrease
from 482 to 463 to  449  to 421 feet  (msl).  The lowest lake surface
elevation may  correspond to  the  surface elevation of Caney Creek
and therefore may represent the water table.  The  other  lakes are
probably perched upon the shales  at the  top of  the  Big Clifty
(Mgbc).

Since the  base-flow,  surface  elevation  of  the  Rough  River  is
approximately  395 feet  and  the lake closest  to  the  river is 482
feet, it certainly  appears that  this lake and the  463 feet and 449
feet lakes are  perched upon the shale confining layer at the top  of
the Big Clifty (Mgbc) .   If the true water table corresponds to the
base flow elevation of the Rough River  (and it may not because  of


                                124

-------
the faulting in the area)  then  the  water table in the Big Clifty
formation under the mine is probably located  at an elevation of
420-450 feet.  In  the  back part  of the mine  (south) the Big Clifty
aquifer may  be  confined by the same shale  confining layer which
perches the  lakes in the  Haney Limestone  (Mgh)  in  the northern
areas of the mine.

Figure 2  is  a geologic  cross section showing the hydrogeologic
profile across the Rough River side of the mine.  The cross section
illustrates the complexity of the hydrogeology in the vicinity of
the  Ft.  Hartford  Mine  Site.   The diagram shows  the  following
hydrogeologic features.

There is a perched, karst-conduit,  turbulent-flow aquifer at the
base of the  Glen  Dean Limestone (Mgd).   This  aquifer is perched
upon shale layers  in the underlying Hardinsburg Sandstone (Mh).

There is a perched, porous-media, laminar-flow aquifer in the upper
Hardinsburg Sandstone (Mh).  This aquifer is also perched upon the
same shale layers in  the Hardinsburg Sandstone  (Mh).   Therefore
this perched aquifer consists of two parts:  the basal part in the
upper Hardinsburg Sandstone (Mh) (porous-media, laminar-flow) and
an  upper  part  in the  lower  Glen  Dean  Limestone  (Mgd)  (karst-
conduit, turbulent-flow), both perched upon the shale layers in the
Hardinsburg  (Mh).

There are  cave streams at the base of the  Glen Dean Limestone (Mgd)
flowing to springs along the Glen Dean-Hardinsburg contact.  Some
of the cave  streams flow to  vertical openings  in the Hardinsburg
confining layer along  or near faults.  These openings permit water
from the upper  Hardinsburg - lower Glen  Dean  perched aquifer to
enter into the  Haney  Limestone (Mgh).   All of  the  sinkhole dye
traces which were positive at water intrusion  points in the mine
were located along or near this east-west fault indicated on the
cross section.

Figure 3 is  a cross section drawn along  the course of the stream
which flows  into  Sinkhole 1 and then  to Spring II  on  the Rough
River.   The  headwaters  is an ephemeral  stream which flows north
upon the Tar Springs Sandstone (Mts).   As it flows off of the Tar
Springs and  onto  the  Glen Dean Limestone (Mgd)  it flow directly
into Sinkhole 12.   Sinkhole 12 was dye traced to Spring E3 and also
to  mine  intrusion location  RR2.    The  stream  channel  continues
beyond Sinkhole 12 and it is obvious that at times it cannot take
all the discharge  and  some of the water continues north on dow the
channel toward Sinkhole  1.  It appears that  the headwaters of this
stream are being pirated by the cave stream which flows to Spring
E3  and also  to mine intrusion  location  RR2 probably due  to the
east-west faults in the area.  Water  flowing  past Sinkhole 12 flows
onto the Hardinsburg Sandstone  (Mh), joined by small seeps which
occur at the Glen  Dean-Hardinsburg contact.

The Hardinsburg is thin and  weathered and  there  is  an east-west
fault which crosses the  stream  at this location.   Collapse 1 has


                               125

-------
the faulting  in the area)  then the water table in the Big  Clifty
formation  under the mine  is  probably located  at  an elevation  of
420-450 feet.   In the back part of the mine  (south)  the Big  Clifty
aquifer may be confined by  the same shale  confining layer  which
perches the lakes  in  the  Haney  Limestone  (Mgh)  in the northern
areas of the mine.

Figure  2  is a  geologic cross  section showing the hydrogeologic
profile across  the Rough River side of the mine. The cross section
illustrates the complexity of the hydrogeology in the vicinity  of
the  Ft.  Hartford  Mine  Site.   The diagram shows  the  following
hydrogeologic  features.
 A'
SOUTH
FEET
  GEOLOGIC CROSS SECTION SHOWING HYDROGEOLOGY, ALONG LINE A - A'


ON PLATE 4, FORT HARTFORD MINE SUPERFUND SITE, OHIO COUNTY, KENTUCKY
 A
NORTH
                                          PERCHED. KARST - CONDUIT, TURBULENT - FLOW AQUIFER
                                                AQUIFER IN HARDINSBURQ SANDSTONE (Mb)

                                          irariATSrsnsito,,'— coNfiNiNG LAYER

                                               'ROUGH RIVER MINE ENTRANCE
There is a  perched,  karst-conduit, turbulent-flow aquifer at the
base of  the Glen Dean Limestone  (Mgd) .   This  aquifer is perched
upon shale  layers in the underlying Hardinsburg Sandstone  (Mh).

There is a perched, porous-media,  laminar-flow aquifer in  the upper
Hardinsburg Sandstone  (Mh).  This aquifer is also perched upon the
same shale  layers  in  the  Hardinsburg Sandstone  (Mh).    Therefore
this perched aquifer consists of two parts:  the basal part in the
upper Hardinsburg Sandstone  (Mh)  (porous-media, laminar-flow) and
an  upper part  in  the  lower Glen Dean  Limestone  (Mgd)  (karst-
conduit, turbulent-flow), both perched upon the  shale layers in the
Hardinsburg  (Mh).

There are cave streams  at  the base of the Glen Dean Limestone  (Mgd)
flowing to  springs along the Glen Dean-Hardinsburg contact.  Some
of the cave streams  flow to vertical  openings  in the Hardinsburg
confining layer along or near faults.   These openings permit water
from the upper Hardinsburg   lower Glen Dean  perched aquifer to
enter into  the Haney  Limestone  (Mgh) .   All  of  the sinkhole dye
traces which were positive at water intrusion  points in the mine
                                 126

-------
  were located  along  or near this east-west fault  indicated  on the
  cross section.

  Figure  3  is a cross section drawn along the course  of  the  stream
  which  flows  into Sinkhole  1  and then to  Spring  II on  the Rough
  River.   The  headwaters is  an ephemeral  stream which flows north
  upon the  Tar  Springs  Sandstone (Mts).  As  it flows off  of the Tar
  Springs  and  onto the Glen  Dean Limestone  (Mgd)  it  flow directly
  into Sinkhole 12.  Sinkhole 12 was dye traced to Spring E3 and also
  to  mine  intrusion  location  RR2.   The  stream channel  continues
  beyond  Sinkhole  12  and  it  is  obvious  that  at times it  cannot take
  all the discharge and some of the water continues north on dow the
  channel toward Sinkhole 1.  It appears that the headwaters of this
  stream  are being pirated by the cave  stream which flows to Spring
  E3  and also  to  mine  intrusion location RR2 probably due  to the
  east-west faults in the area.   Water flowing past Sinkhole 12 flows
  onto the  Hardinsburg  Sandstone (Mh),  joined by small  seeps which
  occur at  the  Glen Dean-Hardinsburg contact.

  The Hardinsburg  is thin and  weathered  and there  is an east-west
  fault which  crosses the stream at this location.   Collapse 1 has
  occurred  directly  under  this  stream.   Collapse 2 has  occurred
  directly  under a tributary to this stream located  about 200  feet to
  the west.  Before these collapses occurred both streams flowed over
  the east-west fault onto the Glen Dean Limestone  (Mgd) to a swallet
  at Sinkhole 1.  The dye trace  of Sinkhole 1 proved that this stream
  then  flows  through a cave, located  at  the base  of  the  Glen Dean
  Limestone (Mgd),  to Spring II on the  Rough River.

  Although the  Ft.  Hartford Mine Site  is  extremely  complicated,
  primarily because of faulting, this  investigation has  revealed a
  basic  understanding of the karst hydrogeology  as it exists in the
  Tar Springs Sandstone (Mts), Glen Dean Limestone (Mgd), Hardinsburg
  Sandstone (Mh) and Haney Limestone  (Mgh).   Exploratory wells will
  be  necessary to  investigate  the  hydrogeology  below  the Haney
   Limestone (Mgh).
SOUTHWEST
 FEET
            SURFACE - SUBSURFACE HYDROLOGY, FORT HARTFORD MINE SITE, OHIO COUNTY, KENTUCKY
NORTHEAST

  f EET
                                 NORTHWEST VIEW
                                   127

-------
                            REFERENCES

Crawford, N.C. and C.B. Dotson  (1990).  Groundwater  Investigation
for a Proposed Wetland Waste Treatment Facility at Logan Aluminum
Company, Logan County, Kentucky.  Prepared for Camp, Dresser,  and
McKee, 17 p. 2 plates.

Heyl, A.V.  and Brock,  M.R. ,  1961,  Structural Framework  of  the
Illinois-Kentucky  Mining District  and its  Relation  to Mineral
Deposits.  U.S. Geological Survey Professional Paper 424-D, pD3-D6.

Hook, J.W., 1972 Structure  of the Fault Systems in  the Illinois-
Kentucky  Fluorspar District.   In  Proceedings of  the Technical
Sessions, Kentucky Oil  and  Gas  Association,  34th and  35th Annual
Meetings,  1970  and  1971.    Special  Publication  21,  Kentucky
Geological Survey.

Johnson, W.D.Jr.  and A.E. Smith  (1968).  Geologic Map  of the Olaton
Quadrangle, Western Kentucky, U.S.  Geological Survey,  7.5 minute
quadrangle GQ-687.

Mull, D.S., Liebermann,  T.D., Smoot,  J.L.,  and L.H.  Woosley,  Jr.
(1988).   Application  of  Dye-Tracing Techniques  for  Determining
Solute-Transport Characteristics of Groundwater in  Karst Terranes,
U.S. Environmental  Protection Agency,  Region  IV, EPA  904/6-88-001,
103 p.

Rehn, E.E., 1968 Petroleum  Potential  of the Rough Creek Tectonic
Element in Kentucky.

Smart, P.L. (1986).  A review of  the toxicity of twelve  fluorescent
dyes  used  for  water  tracing.    National Speleological  Society
Bulletin.  v.46,  n.2,  p.21 33.

Smart,  P.L.  and I.M.S.  Laidlaw  (1977) .   An  evaluation  of some
fluorescent dyes  for water  tracers.   Water Resources Research.
v.13, p. 15-33.

Trace,  R.D.   (1972),   Illinois-Kentucky  Fluorspar  District  in
Proceedings  of  the   Technical  Sessions,  Kentucky  Oil and  Gas
Association, 34th and 35th Annual Meetings,  1970 and 1971, Series
X.
NOTE:  Copies of  Plates 1-4 will be available upon request from the
       authors.
                                128

-------
                      BIOGRAPHICAL SKETCHES
                   Nicholas C. Crawford, Ph.D.
            Director,  Center for Cave & Karst Studies
                  Western  Kentucky University
               Department  of  Geography  & Geology
            Environmental  Science &  Technology  Bldg.
                 Bowling Green,  Kentucky   42101
                           (502)745-5989


Nicholas C. Crawford,  Ph.D.  is a Professor  in  the  Department of
Geography and Geology and Director of the Center for Cave & Karst
Studies at Western Kentucky University.  He  has written over 140
articles and technical reports dealing primarily with groundwater
contamination of carbonate aquifers.  The  recipient of 25 grants
for hydrologic research on environmental problems of karst regions,
he was awarded Western's highest  award for Outstanding Achievement
in Research in 1985.   As a consultant  specializing in carbonate
aquifers for  the past fifteen years, Dr.  Crawford  has performed
over   600   dye   traces  and   worked  on  numerous  groundwater
contamination problems for private firms and for  federal, state and
local government agencies.
                       Ginny L. Gray, R.G.
              Environmental  &  Safety Designs,  Inc.
                     5724 Summer Trees Drive
                    Memphis,  Tennessee  38134
                          (901)372-7962


Ginny L. Gray, R.G.,  is a Geologist with Environmental  & Safety
Designs, Inc.  (EnSafe), Memphis, Tennessee.  She received  a B.A. in
Geology  (1987)  from Austin  Peay State  University,  Clarksville,
Tennessee.  At EnSafe she works as a consultant  to private industry
and government on groundwater contamination studies, environmental
compliance,  and  hazardous waste  management.   She  is  currently
managing two sites in rural  Western Kentucky slated for clean up
under CERCLA.
                                129

-------
              Application of Dye-Tracing Techniques
           for Characterizing Ground Water Flow Regimes
             at the Ft. Hartford Mine Superfund Site
                  Olaton, Ohio County, Kentucky

                   Nicholas C. Crawford, Ph.D.
                               and
                          Ginny L. Gray


1.   What  reduction  in  water  infiltration  into  the  mine was
     accomplished by the  $2 million sinkhole-filling and drainage
     diversion project you described?

As  part of  a  National  Pollution  Discharge  Elimination  System
(NPDES) permit  with the  state of  Kentucky, the  volume of  water
being pumped from the mine is currently being monitored for a four
month  period  to determine the water  infiltration reduction more
quantitatively- Qualitatively, the reduction of water infiltration
varies between the two lobes of the mine.  Water infiltration into
the Rough River lobe (which contained  the massive  collapses to the
surface) has been dramatically reduced.  When the  KPDES permit was
first  obtained in  November  of  1990,  there  were approximately
15,000,000 gallons  of impounded water  in the Rough  River  side.
This water  has since been  evacuated  with only small  amounts of
recharge.  The  Caney  Creek  lobe did  not contain collapses to the
surface;  however,  several  of  the   sinkholes  aboveground  were
contributing water to the impoundments.   This  lobe initially had
approximately 22,000,000 gallons of  impounded water which has also
been evacuated.  Recharge in this  lobe  has been more  substantial
due to several  partial  roof  collapses that continue to transport
groundwater into the mine.

2.   You stated the that  the dye  (Rhodamine WT?)  was  platykurtic
     (Broad) rather than the "normal curve" you expected for your
     main spring.  You  explained this as  possibly due to  diffuse
     recharge through the overlying  sandstone aquifers. Might the
     platykurtic  shape  of  the  breakthrough  curve  be  better
     explained by mixing and storage of the dye in the water mass
     impounded by the mine?

There were receptor packets placed within the impounded water and
all other water intrusion points inside  each lobe  of the mine that
were retrieved and  analyzed to  account  for this  possibility.
Secondly, the traces  which showed residual dye problems were in
units above the mine roof which did not penetrate the  Hardinsburg
Sandstone (primarily shale above the mine)  and  enter the mine, but
travelled laterally above the mine roof and discharged into the
Rough River just north of the mine.
                                130

-------
                 GROUND WATER MONITORING IN UNSATURATED AND
           SATURATED ZONES AT A SITE WITH PALEOCOLLAPSE STRUCTURE

                        Richard Benson and Lynn Yuhr
                                Technos,  Inc.
                            3333 N.W.  21st  Street
                              Miami, FL  33142

                              Allen W.  Hatheway
                       University  of Missouri   Rolla
                       School of Mines and Mineralogy
                     Department of Geologic Engineering
                              Rolla, MO  65401
ABSTRACT
Paleocollapse  structures   can  represent   major  obstacles   to  in   site
investigations for  large engineered  structures  and underground  facilities.
Recognition  of  paleocollapse  structures  become   a  vital part  of such  site
characterization.

Extensive underground mining of limestone has occurred  in  the  greater Kansas
City area.   A landfill  expansion  was proposed  over  an abandoned  limestone
mine located about 200 feet below the surface.   This case history deals  with
the development of a landfill  over an abandoned  underground mine,  the impact
of  paleocollapse  structures  and  a  subsequent  ground  water   monitoring
strategy.

The  site   characterization  work   identified  a   paleocollapse   structure,
probably  due  to   a  deep  seated   cavity   network  probably   within   the
Mississippian rock about  600  feet  below the  mine floor.   This collapse  had
induced  fractures   in   the  overlying   rock  and  soil.    When  the   mine
subsequently cut  through the  zone  of paleocollapse structure,  a major  14-
acre area of mine-roof  collapse  occurred.   The  paleocollapse  structure  lead
to  the  formation of  a  local  synclinal  basin with major  fractures.    These
features have a major impact upon  the mine stability and the  development of
a reliable ground water  system for  the site.

Ground water above the  mine  was found to  be perched and isolated  with  flow
through  isolated  joints,   fractures   and bedding  planes.     Ground  water
characterization must consider the perched  and  fracture  flow ground  water
conditions,   recharge   of  water  into  the  mine,  and  the   superimposed
paleocollapse  structural  aspects  in  order  to  develop  a   ground  water
monitoring strategy for  this  unique site.


INTRODUCTION

There   has   been   extensive  mining  of   the   Bethany   Falls   Limestone
(Pennsylvanian Age)  within the greater Kansas City area.   Kansas  City  area
mines are almost  always dry and stable  giving  the city the  distinction of
being Number One  in the  world  in  terms   of  human use  and  occupancy  of
underground  space.    More   than  200 businesses  are   located  underground
occupying more than 1,200 acres as  of 1983 (Hasan, et al.,  1988).

                                     131

-------
Due to  other  favorable (non-geologic)  conditions,  Waste Management  of North
America,  Inc.  chose  in  1986 to  expand  its  Forest  View  Landfill  over  an
abandoned limestone mine underlying much of the  site (Figure 1).

The loess soils  over the mine contained major fissures which  coincided with
the area  of mine-roof  collapse.   Preliminary site  characterization  by one
consultant identified vee-shaped features in the loess soil blanket  over the
collapsed portion of the mine  and suggested  that  the surface fissures were
due  to  the  mine-roof  collapse     (Figure   2).     This   was   a  reasonable
conclusion,   based  upon   limited   information   available   at   the   time.
Subsequently another consultant  suggested that surface subsidence,  and hence
fissures, were  not likely to occur based upon the general  conditions  within
the mine but did not address  the origin of the fissures.

There  were  two  key site  characterization  issues  to be  resolved  for this
project:
o      First,  characterization  of the  relation  (if  any)  between  mine-roof
       collapse   and the  surface fissures  along with  an  assessment of mine
       stability;
o      Second,  assessment of  the hydrogeologic conditions  so that  a  rational
       ground water monitoring strategy  could be  developed.

This paper  provides  a brief  summary  of the site characterization  (from both
the  engineering  geologic and  hydrologic  perspectives)  found necessary  to
proceed with  the landfill   development  over  an  abandoned mine  which  had
undergone  major  roof  collapse.    The  sequence  of  presentation within this
paper  follows  that of  the  actual site characterization process.
                Source: U.S.G.S. Shawnee, Kansas
                   Quadrangle, Photorevised, 1975.
                    Figure 1.     Site Location Map
                                      132

-------
GEOLOGIC SETTING

A thick  deposit of Pleistocene  loess overlays  the bedrock.   Bedrock is made
up of  an alternating sequence  of cyclothems consisting  of  relatively thin
beds of  limestone and shale,  which were  deposited in migrating shallow seas
(Figure  3).    The   Pennsylvanian  System  is  underlain   disconformably  by
Mississippian  limestone  and dolomite beds.

Regional  bedrock dips   gently  (10  to  20  feet/mile)   to  the  northwest.
However, since the site  is located  on  the northeastern  limb of  the Shawnee
Syncline extension,   the  local dip  is  to  the southwest^.   A  local synclinal
basin  occurs  over the collapse zone.   In the  early stage  of investigation,
it was not  clear  whether this local  synclinal basin was due to  paleocollapse
or  simply  a  local structural feature,  which  are commonly observed  in the
area with low  amplitudes of about 3 to 10 feet.

Much  of the   Pennsylvanian  strata  observed  in   outcrops   around the  site,
contain  near-vertical joints.   These joints  are  particularly evident in the
limestones, where fractures  of two  dominant directions  are  seen, generally
striking to the northeast and northwest directions.
                          VEE-SHAPED
                          SURFACE
                          FISSURES
                                        SHALE

                                        LIMESTONE
Figure 2.
               Previous (1986) Perception
               of Site Conditions:   Mine
               Collapse has Lead  to
               Surface Subsidence
                                                        ARGENTINE -
                                                        FRISBIE
                                                        LIMESTONE
                                                        LANE
                                                        SHALE
                                                        RAYTOWN -
                                                        PAOLA LIMESTONE
                                                        CHANUTE SHALE
                                                        DRUM LIMESTONE
                                                        QUIVIRA SHALE

                                                        WESTERVILLE
                                                        LIMESTONE
                                                         BLOCK LIMESTONE
                                                         FONTANA SHALE   If
                                                  Figure 3.     Geologic Section
                                       133

-------
CHARACTERIZATION OF MINE-ROOF  COLLAPSE AND SURFACE FISSURES

Much of  the data acquired  to  resolve  the mine-roof collapse  assessment and
its  relationship   to  the  surface   fissures  was   based   upon  geologic
observations, from within the  mine,  from  seven exploratory trenches cut  into
the loess  soil,  and at numerous rock  surfaces and cuts  exposed by trenching
or  by  excavation  during  construction.    An estimated  1,000  man-hours  have
been  spent  inside  the mine,   developing  an  initial  map  of  the mine-roof
collapse  area,  making  engineering  geologic  observations, monitoring mine-
roof collapse,  and  carrying out  a  QA  audit  of  the  mine  backfill program.
All observations within the mine and the  surface  were  extensively documented
by engineering geologic maps,  cross-sections and color photographs.


Initial Findings Regarding  Characterization of Mine-Roof Collapse  and
Surface Fissures

The  following  facts  evolved  as  part  of the  initial  mine-roof  collapse
characterization and  its  relationship  to the surface fissures  (Figure  4):
o     About  170 feet  of limestone  and  shale lie over the mine;
o     The  rock is covered by a thick loess soil up to 60 feet  thick;
o     The  fissures  appear  on  the  surface  as  large arc's  making up circular
      patterns  centered directly  over the  area  of major  mine-roof-collapse
      (Figure 5);
o     Observations  in  7  trenches revealed that  the fissures in  the loess
      soil were  much older (at least  many hundreds  to thousands  of years)
      than the  mine-roof  collapse which  occurred about  14 years before  this
      investigation;
                  EXPLORATORY
                  TRENCHES REVEALED
                  FISSURES ARE OLD
CURVILINEAR
FISSURES APPEAR
ON PRE-MINE AERIAL
PHOTOGRAPHY
                                       |   MASSIV
                                             E LIMESTONE
                  Figure 4.     Initial  Findings Regarding
                                Mine-Roof Collapse
                                      134

-------
o
o
            Figure  5.     Spatial Relation of
                          Surface Fissure and Mine-
                          Roof Collapse

Some  of  the   surface  fissures  were  observable  on  pre-mine  aerial
photography, indicating that the fissure existed before mining began;
The mine is a room-and-pillar mine, fourteen feet high;
A  random  pillar layout was  used before  the major  roof-fall  collapse
occurred;
A  regular  room-and-pillar layout  was  used  after  the major  roof-fall
collapse occurred;
The regular pillars were approximately  25 feet x  25  feet and  on  60-
foot centers;
The major  collapse  area (in  which major upward  collapse has  taken
place) covered an area of about 16 acres;
The major mine-roof  collapse  occurred  within the random pillar portion
of the mine;
Larger  (80  acres)  and much  older portions  of the  mine with  random
pillars  had very  little mine-roof collapse  and   no  major  mine-roof
collapse;
Typical mine-roof  collapse  extends upward 10  to  15 feet  through  the
Rubble Rock Zone, the Galesburg Shale, and the  Stark Shale  to  the base
of the Winterset Limestone;
A limited number of  local mine-roof collapse  extends  upward as much as
60 feet above the mine-roof to the base of the Westerville Limestone;
Upward  collapse  was  conical  in  shape becoming smaller  upward  (note
that the angle of draw is inward contrary  to  the outward  angle of draw
commonly used in coal mine subsidence);
The Westerville  Limestone immediately  above  60 feet was  quite massive
and thick (15 feet) and would tend to resist further upward collapse;
Roof rock  debris bulking factors  of  20  to  75% were  observed  in  the
mine and averaged 42%;
A bulking factor of  30% would choke off upward  collapse within 80 feet
from  the  mine-roof,  about  half-way  through  the   170  feet  of  rock
overlying the  mine;
Permit  conditions   required  backfilling   of  the  mine.     A  backfill
mixture of  fly-ash,  bottom ash  and kiln  dust was  initially used  to
backfill the mine.
                                    135

-------
                                          Source: Gentile, 1984.
                                        Figure 7.
Figure 6.     Conceptual Geologic Model
             to Account for Site
             Conditions
Conceptual Model of
Typical Paleocollapse
Structure in Kansas City
Area
  Conceptual Model of Mine Collapse and its Relationship to Surface Fissures

  The previous facts (the age of the fissures, and observations  of fissures in
  pre-mining  aerial  photography)   lead to  the  conclusion  that  the  mine-roof
  collapse had not caused the  surface  fissures.   A conceptual model  which can
  account  for  these  fissures  incorporates  the  presence  of  a  deep-seated
  paleocollapse structure deep below the mine (Figure 6) .   This  model has been
  used to  explain other  paleocollapse structures  (Figure 7)  in  the  greater
  Kansas   City  area  (Gentile,   1984).    Subsequent   investigations   of  rock
  exposures  resulting  from  site   development   has   strongly  supported  this
  conceptual model.
  Additional Engineering Geologic Work Included:

  o     Developing a detail map of the mine collapse;
  o     Monitoring mine-roof collapse by quarterly inspections;
  o     Observations of surface rock outcrops exposed by construction;
  o     Stabilizing the mine by backfilling.
  Developing A Detail Map Of The Mine Collapse

  A detailed  map  of  the  mine-roof collapse  area  and the  mine around  it was
  developed so that  the  collapse  areas  could be monitored.   This  map has been
  revised  periodically  to  reflect changes  in  mine collapse  as  well   as  to
  correct any errors in the map.   The mine map  also  provides a means of safety
  in mine operations.
                                       136

-------
Monitoring Mine-roof Collapse By Quarterly Inspections

Monitoring of mine  collapse  consisted of numbering and  photographing all of
the  collapse areas around  the  perimeter  of  the  central  collapse  area.
During each  subsequent  inspection,  photos were taken for  comparison.   After
five  mine   collapse   inspections  (1988-1991)  mine-roof  collapse  slowly
continues to expand laterally  around the perimeters of  the  central collapse
area (Figure 5).

Much of  the  mine-roof collapse is  thought to be  initiated  by deterioration
of  a  Rubble Rock which occurs at  the top  of the Bethany  Falls  Limestone.
The Rubble  Rock  Zone  is a  slake-susceptible horizon found  over  the greater
Kansas City area.   The Rubble  Rock  Zone  is  known to  deteriorate rapidly with
moisture.   Since  1987  mine  flooding has  increased substantially,  leading to
the  degradation  of the Rubble  Rock  Zone.    Generally,  mine-roof  collapse
extends  upward  (about 10 feet)  through  the  Bethany  Falls  Limestone (which
includes the Rubble Rock Zone) ,  the Galesburg and Stark  Shales to  the  base
of the Winterset Limestone.

At  a  few locations,  around the  perimeter of  the  collapsed  area,  collapse
continued to extend upward  to as much as 60  feet above the mine-roof.   At
this point,   the upward  collapse encounters  the massive  Westerville Limestone
which is  15 feet thick.   Only three  of these  areas were inspected  due to
access and  safety reasons.   In each  case,  these upward extensions  of mine-
roof collapse were  very  localized  (less  than 15-30 feet in  diameter)  and
became smaller as the collapse extended  upward.   The  three upward extensions
of  mine-roof  collapse   that  were  observed  were  all   located  along  a
curvilinear  line  of the paleofracture  and are  thought  to be  linked  to  the
paleo-structure.

Since the  angle  of draw is inward (contrary to  the outward  angle  of  draw
commonly used in coal mine subsidence),  and  the  bulking  in each of the three
areas inspected  was close  to  choking  off and  stopping any  further upward
collapse.    These  observations   support  early  assessments  that  the  mine
collapse would not lead to surface  subsidence.
Observations At Rock Outcrops Exposed By Construction

Further  evidence  for paleocollapse  activity  is  seen  in  rock  surfaces  and
cuts exposed by construction at the site.

Observations included:
o     A cross-section through one end of the  synclinal  basin which lies over
      the central collapse area with an amplitude of about 15 feet;
o     Numerous fractures  and shear zones  along  with sizeable  open cavities
      associated with fractures; and
o     The  fact  that  all  of  these  features  were  much  older  than  the
      relatively recent mine-roof collapse.
                                     137

-------
Based upon  these observations,  it is  reasonable to  assume  that  the local
synclinal basin  (which  is  spatially  coincident with  the Central Collapse
Area and the  area of  paleocollapse)  is in fact  related  to the paleocollapse
feature.

Since the  syncline has  up  to 15  and possibly 20  feet of  relief,   it  is
reasonable  to assume  that  the miners  had removed  rock  close  to, or even
above, the  top  of  the  Bethany  Falls  Limestone.   This  would have  clearly
weakened the mine-roof within this area.
Stabilizing The Mine By Backfilling

Surface  subsidence  is  not anticipated  based upon  observations within  the
mine.   However,  management has  made a  decision to  opt for  a conservative
approach and continue backfilling  the mine.   Backfilling of  the  mine is  now
being accomplished by a crushed-rock water slurry technique,  which  is a much
more efficient method of backfilling the mine.
CHARACTERIZATION OF HYDROGEOLOGIC CONDITIONS

The primary  source  of ground water in the  area  is from wells  in the Kansas
River Alluvium, which has a thickness of 40  to 70  feet,  yielding from 150 to
1,000 gpm  (O'Connor,  1971).   Only small  quantities of ground water  (1 to 10
gpm)  are  typically  available  from wells  placed  in  Kansas River  tributary
valleys   since   they  are   predominantly   silts   and   clays   with   low
transmissivities  (O'Connor,  1971).

Preliminary  assessment  of the  presence  of ground water, based  upon drilling
a number of  mine  fillholes,  indicated that most, if not all, of the rock at
the site  is  unsaturated.    Two open boreholes were monitored  with  borehole
television  over  a  few weeks  and  did  not  make  any  water.    Two  existing
monitoring wells had been dry  for more than  a year.   Localized  fracture flow
and  flow  along bedding  planes  were  occasionally  found  to   occur  in  the
uppermost highly weathered Argentine Limestone.

It had  been  proposed to use  the  Drum Limestone as  a zone  for ground water
monitoring since  it was  the   shallowest  strata  to  extend under  the entire
site.   Furthermore,  the literature indicated small amounts of ground water
were  extracted  from this  zone.   However,  the two  existing monitoring wells
in  this zone were  found  to be  dry and have  remained  dry for  a number of
years.

The mine was slowly filling with water  at a rate  of  about 1 million gallons
per month.   Since the  mine was  below the  level of the nearby Kansas River
(Figure 1) the river is a possible source of mine-water recharge.
                                    138

-------
Initial Findings Regarding Hydrogeologic Conditions

As a  result of  the initial  information on ground  water conditions,  based
upon  literature,   two  existing  monitoring  wells   and  numerous   borehole
television logs, the following conclusions were made:
o     The 200  feet or  more of  rock between  the  mine  and  the surface  was
      virtually unsaturated;
o     The one exception was  the  isolated encounters of perched water  in  the
      Argentine Limestone, the uppermost rock unit at the site;
o     The source(s) and rate(s) of mine-water recharge was unknown;
o     There  was   uncertainty  as   to  the  hydrologic   connection   of  the
      paleocollapse fracture from the ground surface down to the mine.


Additional Hydrogeologic Work Included:

o     Assessment of the hydrologic  connection of the  paleocollapse  fracture
      from the surface to the mine;
o     Further assessment of perched water zones;
o     Assessment of mine-water recharge;
o     An assessment of the nature of the mine-water quality;  and
o     Reassessment  of the ground water monitoring strategy.


Assessment Of The Existence Of The Paleocollapse Structure And Its
Hydrologic Connection To The Mine

One of  the  more significant fissures in the loess soil was  excavated to  top
of  the  Argentine  Limestone.   A major  solution-enlarged joint was  exposed
showing possible displacement of 12 to 18 inches (Figure 8).

Two angle  borings  were made  through the vertical  extension of the  exposed
joint  (paleocollapse   fracture)  and  observations were   made  with  downhole
television.    A number  of  fractures   and  voids  were  observed  near  the
projected  location of the  paleofracture (Figure  8).    In addition,   it  was
found that these boreholes would not hold water.

The paleocollapse  fracture  was also  observed within the mine  at a number of
locations.   The perimeter of the mine-roof  collapse was  accessible  in most
locations for visual  observations.   However,  the  core of the  collapse zone
was not accessible  due to choke  off  by  fallen  rock.   Around  the perimeter of
the collapse  zone,  vertical cracks were  observed  at three  locations.   Each
crack was relatively tight and would not  accept  a knife  blade.   Displacement
was not  observed at the crack but strata dipped downward at  an angle of 10
to 20 degrees on the  side of the collapse.   These cracks are  thought to be
due to paleofracturing by the collapse of a deep cavity system.

One  borehole   encountered  a  solution-enlarged cavity  in  the   Winterset
Limestone along the surface  projection of the  paleofracture.   This  cavity
was  found  to  be  about  2  to  3  feet  in  size  and  obviously enlarged  by
solutioning  because  of  the  smooth   rounded   rock  observed  on  downhole
television.
                                     139

-------
                                               VEE-SHAPED
                                               SURFACE FISSURE
               ARGENTINE /
               FRISBIE
               LIMESTONE
                                             Possible"i2"tol8 .Displacement"
                                             SOLUTION-ENLARGED JOINT
               LANE SHALE
                             APPROXIMATE PLANE-
                             OF PALEO FRACTURE
                                                        FEATURES
                                                        NOTED FROt.
                                                        BOREHOLE
                                                        TELEVISIOt
                 Figure 8.     Cross-Section Through
                               Surface Fissure and
                               Paleocollapse Fracture
The hydraulic connection  of the  paleocollapse  fracture extending from  the
surface downward to the  mine  (Figure  6)  was  obviously of  concern since  it
would provide  a  pathway for  leachate  migration.  This  was  assessed by  a  dye
tracer program in which  dye was  injected along  with  the  30,000  gallons  of
water at rates of up to  200 gpm.

Seven points  were  selected around  the perimeter  of  the collapse  zone  within
the mine  to  sample mine-water for the  presence  of dye.    The presence  of
background  dyes  in  the  mine-water  was assessed  by  obtaining water  samples
and placing  charcoal  bugs  at  each  of  the  seven  sample  locations  before
injecting the  dye.   After dye  injection,  these seven points were  sampled at
logarithmic intervals  of  approximately 1,  2, 5, 10,  20,  50  and 100 days.

Mine backfilling by the  rock slurry method began at about the  133  day  of the
dye tracer  test.  As  of this  time, no dye was found  within the mine.   The
process of  mine  backfilling  involves  withdrawal  of  mine-water which  is then
re-injected into the  mine  along with  rock as  a slurry at  pumping rates of
about 1,500 to 2,000  gpm.   Mine backfilling has now occurred in 12 fillholes
over more than 10 acres  of the mine.   More than 184 x 10^  gallons  of water
have been pumped  from the  mine and  re-injected.    Since  the mine  contains
about  60  to  70  million  gallons  of water,  this amount of  pumping and  re-
injection over an area  of  10 acres has  lead to  significant mixing  of mine-
water.    The  last  sampling  for dye at nearly  500   days  after  injection  was
made after  the significant mixing and  therefore can be expected to yield any
                                      140

-------
dye previously  missed.   Since  no  dye  was  found  within the  mine  after  a
period of  about  500 days, we  may conclude  that  the paleofracture  does  not
hydraulically interconnect  the surface  or  rock  strata  hydraulics  with  the
mine.

The lack of flow  through  the  paleocollapse  fracture can be  accounted for by
the presence of swelling shales  (probably the  Chanute  Shale)  which were seen
to completely close  off  three open boreholes  so  that they would  hold water
within a  10-12  month period.    The  closure by shales  was observed  in three
boreholes by  downhole  television and  subsequent  monitoring of water levels
above the swelled shale.
Further Assessment Of Perched Water Zones

The injection of 30,000 gallons of dyed water provided  a  windfall in that it
was found to have migrated  laterally within  the  Raytown-Paola Limestones via
fracture flow.   Water was observed on borehole  television to  flow  into one
borehole via  a  fracture.    This  borehole was  about  75  feet  from  the  dye
injection site.   Samples of  dye  water were  recovered in  the  Raytown-Paola
Limestones at  two  boreholes,  about 75  feet  and  500  feet  latterly  from the
injection site  1 day after  dye  injection.   The identification  of  fracture
flow  and  dye   identified  a  possible  alternate  zone   for  ground  water
monitoring within the Raytown-Paola Limestones.


Assessment Of Mine-Water Recharge

Long-term  monitoring  of  mine-water  levels  were  made  with  an  in-situ
electronic water level  instrument.   Water levels were found to  increase at
about   15   inches/year   (about   12   million  gallons/year   or  about   23
gallons/minute).

The  mine floor is  underlain by a  22-foot  thick  sequence  of shale  and
limestone strata.  The upper 10 feet of this  sequence  has been determined to
be  quite  permeable  due  to  a   very  high  secondary  porosity  related  to
dissolution-enlarged  joints.     This   permeable  zone   consists   of   the
Hushpuckney  Shale,   the  Middle   Creek  Limestone,   and  the  Ladore  Shale.
Recharge from the Kansas River (which  is significantly  higher than the mine)
flows through this  permeable  zone into the mine accounting for a portion of
the  recharge  of ground water  into the  mine.    Other  sources  of mine-water
recharge are also possible but were not assessed.


An Assessment Of The Nature Of The Mine-water Quality

A  mine-water  sampling  program   was  initiated  to  characterize  background
quality  of  the mine-water and  to  provide  geochemical  characterization.
Mine-water has  been  sampled  at   eight  points within  the  mine and  each of
these points have been  sampled three  times to provide  a  statistically valid
baseline  sampling.     Field  parameters   measured  included  pH,   specific
conductance and  temperature.   Laboratory  analyses  were carried  out  for the
expected  (not  yet  issued)  parameters  required  for  Subtitle  D     Phase  I
                                     141

-------
monitoring  (US-EPA).    In  addition,  the  samples were  analyzed  for several
pesticides and herbicides.

Geochemical  analyses   for  major   ions   indicate   that  the  mine-water  is
predominantly sodium  sulfate  in character.   Changes  in the  geochemical data
tend to correlate roughly with  distribution  of  the  original  fly-ash backfill
material within  the mine.    Mine-water composition  sampled  from  areas more
remote  from fly-ash backfilling  have higher calcium and magnesium ratios,
but are still dominantly sodium sulfate in character.

There  was  some  concern about  the long-term dissolution of  mine  rock  and
crushed-rock backfill  by the  mine-water.   However, at  one  location  where
surface water  entered the  mine under the mine  portal  road (through a rock
cover  of about 30  to 40 feet),  extensive precipitation of calcite occurred.
At  this  location,   water  dripping  from the  mine-roof  would  respond  to
rainfall events  within  24  hours.   The  rapid  response  to  rainfall  events
indicated  that  even  with  short  travel  time  and  distance  through  the
limestone,   the fresh rain-water had  become  saturated  and would  precipitate
calcite.  The observed  deposition  of  calcite by  water seeping  into the  mine,
the  high  pH  measured  in  the  mine-water  (7.6  to  12.4) and the  positive
saturation  index  (0.63)  calculated from geochemical  analyses,  indicate that
geochemical  conditions  are   such   that   dissolution  of  limestone  rock  or
crushed limestone used as the mine backfill should not occur.
Reassessment Of The Ground Water Monitoring Issue

Further drilling,  hydraulic  tests,  and hydrofracturing  were carried  out  in
the Raytown-Paola  Limestones  and Drum Limestone to assess  optimal locations
and conditions  for ground water monitoring wells.   It  was found  that  the
Raytown-Paola Limestones  contained  water  in 9  of  9 borings  and piezometers
and the Drum Limestone  did not contain water in 7  of 8  borings, piezometers
and wells.

Hydrofracturing  improved hydraulic  conductivity  within  the  Raytown-Paola
Limestones from 7.2 x 10"6 to 1.3 x 10"4  cm/sec.   In the Drum  Limestone  the
improvement in  hydraulic conductivity was  less.   As a  result  the  Raytown-
Paola Limestones have been selected as the optimal location for the shallow
monitoring wells.

The proposed  ground  water  monitoring system  is  shown  in Figure  8  with
objectives to:
o     Monitor the zone of perched water within the Raytown-Paola Limestones;
o     Monitor mine-water quality; and
o     Monitor ground water within the permeable zone beneath the mine.


CONCLUSIONS

Careful  site   characterization  consisted  in   large   part   of   extensive
observations and   deductions  to  provide  insight  into   mine collapse,  mine
stability,  the paleocollapse  structure, and  aid  in  developing a ground water
monitoring strategy at the site.
                                    142

-------
It is interesting  to  note that all four of the geologic  constraints cited in
"Geology of  Greater Kansas City, Missouri  and Kansas" (Hasan,  et  al. ,  1988)
were  found  to  exist at  this site.   They include:   detrimental  stability
effects of older unplanned,  random mine layouts; indications  of the presence
of paleocollapse   structure  such  as  described  by  Gentile (1984);  infilled
paleo-geomorphic  channels;  and  thinning  of  roof  strata  (due  to  mining
practices).   The  previous work by  Gentile (1984) proved to be a key  to the
insight  of  the paleocollapse issue.    Gentile has  identified  a  number  of
these paleocollapse structures in  the Kansas  City area.

The mine  stability issue  is  being resolved  by backfilling  the mine with  a
crushed-rock  slurry  (Figure  9) .     This  technique  has  proven  to  be  an
extremely  effective method and  has been  documented by  a Quality  Assurance
Audit  program  including  contour   maps  of the  backfill  (based  on  visual
observations), along with photos and video  documentation.

A ground water monitoring  strategy  has  evolved, based upon identification of
a perched  water zone  with fracture flow within the Raytown-Paola  Limestones
(Figure  9) .    Ground water  monitoring has been  enhanced by  improving  the
hydraulic conductivity  and the connection  of  wells  to the fractures in this
zone  by  at   least  two  orders  of magnitude through   hydrofracturing  the
limestone  rock  within the screened interval.   Monitoring of the  mine-water
quality will  provide  a back-up monitoring system and will also monitor  any
baseline changes due  to the  presence  of the  original  fly-ash backfill used
to fill  the  mine.    Monitoring of  the  permeable  zone below  the mine  floor
(which is  a  major  recharge zone for water flowing  into  or out of  the  mine)
will provide  a  continued means of  monitoring water  quality coming  into  (or
leaving) the mine.
                           -MINE
                           -PERMEABLE ZONE
                           -PERCHED ZONE
           ARGENTINE
           FRISBIE
           LIMESTONE
                                            ZONE OF PERCHED GROUND WATER
           DRUM LIMESTONE
           WINTERSET
           LIMESTONE
           BETHANY FALLS,
           LIMESTONE
                 Figure 9.     Conceptual Model of Site
                               Conditions and Proposed
                               Ground Water Monitoring
                               System
                                    143

-------
REFERENCES

Gentile,  Richard J.,  1984.    Paleocollapse  Structures:    Longview  Region,
        Kansas   City,   Missouri.     Bull.   of  the  Assoc.   of  Engineering
        Geologists, Vol. XXI, Number 2, p. 229-247.

Hasan,  Sidney,  R.  L. Moberly  and J.  A.  Caoile,  1988.   Geology  of  Greater
        Kansas  City,  Missouri  and Kansas, United  States  of America.   Bull.
        of  the  Assoc.  of  Engineering Geologists,  Vol .  XXV,  Number  3,  p.
        281-341.

Hatheway,  Allen  W. ,  N.  D'Andrea  and  R.  C. Benson,  1990.   Geologic  Factors
        Responsible  for Mine  Collapse under  a  Portion of  the Forest  View
        Landfill Site,  Kansas  City,  Kansas,  In:   33rd Annual Meeting  of the
        Association  of  Engineering  Geologists,   Pittsburgh,   Pennsylvania,
        October  1-5, 2p.

O'Connor,  H.  G.,  1971.   Geology  and  Ground Water  Resources  of  Johnson
        County,  Northeastern Kansas.    Kansas Geological  Survey  Bulletin,
        203, 68 p.
                                     144

-------
              GROUND WATER MONITORING IN UNSATURATED AND

     SATURATED ZONES AT A SITE WITH PALEOCOLLAPSE STRUCTURES

                        Richard Benson, Lynn Yuhr and Allen W. Hatheway



1.      Who was your client for this project?

       Waste Management of North America.


2.      Is the  increase in water  level in the mine, as related to the Kansas River, part of a regional/long
       term trend or is it a local phenomenon related to the mining activity? How will the filling of the
       mine with what is being  pumped into it affect water levels in the mine?  What  will be the effect of
       the rising water on the landfill?

       During active limestone mining, inflow of surface waters (portal run-in) and ground waters were
       pumped from the mine.  Since  abandonment  in 1965, the mine has gone unwatered.  The water
       level within the mine will now continue to rise until it completely fills the mine  or at least to such a
       height above the mine-floor that is in equilibrium with the sources of recharge.  One source of
       recharge is the Kansas River, some 1,200 ft distance from the mine.

       The crushed limestone backfill being placed within the mine is highly porous and will not affect the
       ultimate level of mine water at its final equilibrium. The rising water level should have no effect on
       the landfill, which is located upon more than 170 ft of overburden limestone and shale.


3.      I do not question the reality of there being a paleocollapse (or paleocollapse subsidence) structure
       present. Prove to me, please, that the faulting and fissures at the surface are not a consequence of
       re-activation  of  that paleocollapse by  the mining activity.  If such reactivation has occurred (is
       occurring), how  can you justify the integrity of the site until after such collapse is completed and is
       at some stage of equilibrium?

       A reactivation of the paleocollapse zone, from the mine workings, upward to the surface cannot
       occur due to mine collapse for the reasons cited within our paper  (primarily 170 feet of rock over
       the mine, support of overburden rock by 25' x 25' pillars on 60' centers, high bulking factors (an
       average of 42%), and bridging by the massive Winterset Limestone 60 feet above the mine-roof.
       Backfilling of the mine with crushed rock would also provide mitigation if significant subsidence
       were to be somehow occur.

       The rock, faults and soil fissures at the surface were initially caused by the paleocollapse.  They
       have been perpetuated and possibly accentuated by the fact that they are highly permeable zones
       which  readily accept surface water runoff. In  contrast,  the 20 to 60 feet of loess soils allows little
       percolation of surface water to the underlying rock.

       Re-activation of the paleocollapse zone above the mine-roof will not occur due  to  the on-going
       mine-roof collapse for the  reasons cited. If paleocollapse reactivation would occur, such would be
       due to deep-seated factors, hundreds of feet below the mine floor. Although rare, it is not unknown
       for paleo-sinkholes to reactivate leading to further subsidence.  We know of no such re-activation in
       the greater Kansas City area or  within  the tri-state (Kansas, Missouri and Oklahoma) lead mining
       area.  Therefore, we  feel that  the re-initiation of paleocollapse  by this cause is  essentially an
       improbability.

                                                145

-------
146

-------
                Luncheon Lecture:
Chairman Mao: The  Great Leap Forward,
 and the Deforestation Ecological Disaster
        in the  South China Karst
                 Peter Huntoon,
              University of Wyoming

-------
148

-------
    Chairman Mao's Great  Leap Forward and  the  Deforestation

       Ecological Disaster  in the  South  China  Karst  Belt


                       Peter W. Huntoon

             Department of Geology and Geophysics
                     University of Wyoming
                    Laramie,  WY  82071-3006



                           ABSTRACT

     Stone  forest  aquifers   comprise  an  important  class  of
shallow,  unconfined karstic  aquifers  in the south  China  karst
belt.   They occur  under flat  areas such as  floors of  karst
depressions, stream valleys, and  karst  plains.   The frameworks
for the aquifers  are the undissolved  carbonate  spires  and ribs
in epikarst  zones developed on carbonate  strata.    The ground
water  occurs   within   clastic  sediments  which   infill  the
dissolution voids.  The aquifers  are  thin, generally less than
100  meters  thick,   and  are  characterized  by  large  lateral
permeabilities  and small  storage.
     The magnitude and duration of  the  seasonal recharge  pulse
that replenishes  the stone  forest aquifers have been  severely
impacted by massive post-1958  deforestation in  the  south  China
karst  region.    The loss  of  seasonal  upland  storage  in  the
"green  reservoir" has  resulted  in  both  a   reduction  in  the
volume of  recharge  to  the  lowland stone  forest  aquifers  and a
shortening of  the seasonal  recharge event.    This  response is
compounded by  increased ground water  withdrawals as the people
attempt to offset the declining supply.


                         INTRODUCTION

     The  purposes  of   this  article  are   to  describe  the
hydrogeologic  properties   of   stone  forest   aquifers  which
constitute  an   important  class  of  thin,  shallow,  unconfined
aquifers in  the  vast  south  China karst belt,  Figure 1, and to
qualitatively assess the impacts  of  deforestation  on the stone
forest  aquifers.   This  article  is an  abridged   version  of
Huntoon (1992a,b).
                             149

-------
    SOUTH CHINA
                  r?
                  -' •} _ SGUIZIIOU
                       '  PROV.  3
                              V ^^  I
                          ..Dushan/-   S
                          -  /»x-*  n  '
    THAILAND
Figure 1.  Location of  Guangxi  Autonomous Region and  Guizhou
           and Yunnan provinces in south China.


              SETTING  OF THE STONE FOREST AQUIFERS

     Centered  around  the southern Chinese  provinces of Yunnan
and Guizhou,  and  the  Guangxi Autonomous Region are 500,000 km
of  the most  spectacular tower  karst  landscapes in  the world
(Zhao  1988).   A class of shallow aquifers,  herein called stone
forest aquifers,  occurs in  this terrane.   These aquifers owe
their  existence  to  intense but  shallow  dissolution  of  the
carbonate bedrock.
     The  stone forest  aquifers  occur under  flat areas within
the  karst  belt.   -They are most extensive under  the inland
plains  found within  a  few  hundred  meters  or so  of sea level
(Figure  2).   However,  as   shown  on Figure  3,   they  are  also
common  at much  higher  elevations including occurrences along
river  valleys in the mountainous  areas,  under large and small
karst  depressions  within  the elevated  areas, and  even under
flats  along, ground  water divides.
                             150

-------
Figure  2
Carbonate  peak  clusters  rising  from  a  karst  plain
near   Xiaopingyang,    Guangxi   Autonomous    Region,
China.  A  stone forest aquifer  underlies the  plain.
          PEAK FOREST PLAIN
                                    PEAK CLUSTER
                                    DEPRESSION
                                       PLAINS
                                       ALONG
                                       STREAM
                                       VALLEY
  PRIMARY
   RIVER
                          Cave draining peak
                          cluster depression
MOUNTAIN
 STREAM
        Top of clastic infill
        Buried stone forest
Figure  3.
Schematic cross  section,  vertical  scale  greatly
exaggerated, showing some  typical Bettings  for  stone
forest  aquifers  under  flat areas within  the  south
China  karst.   Epikarst  occurs on  all  the  carbonate
surfaces; however, only  the infilled epikarsts  under
the  flat areas which host  the stone forest  aquifers
are  shown.
                                151

-------
                 EPIKARST AND EPIKARST AQUIFERS

     Superimposed on  the  south China carbonate  terranes,  both
hills  and  plains,   is  an  epikarst  (Figure  4)  which  is  an
intensely dissolved  veneer  consisting of an  intricate network
of intersecting roofless dissolution-widened fissures, cavities
and  tubes  dissolved  in the carbonate  bedrock.   The  depths to
the  base of  the  epikarst zone  are  variable,  usually being less
than 100 m.   As revealed in quarries  and hand dug  wells,  the
crevassed, highly dissolved upper  part  ranges from  10  to  30 m
deep.    Widely  spaced  dissolution-widened  fractures  extend
another  30  to  70  m  below  the  crevassed   zone.    Externally
derived  sediments,  soils,   karst  breccias  and  residual  clays
infill the solution openings within the  epikarst zone.
Figure  4. The  famous  Kunming  Stone  Forest  110  km  from
           Kunming,  Yunnan  Province,   China.    This  type  of
           bedrock surface when infilled with clastic sediments
           serves as the framework for stone forest aquifers.

     The  term epikarst  aquifer  was  first  defined by  Mangin
(1975)  to   describe   saturated   zones   within  the  intensely
dissolved  veneers  found  on  carbonate  stratigraphic  sections.
Since  the  introduction of the concept  of  an epikarst  aquifer,
epikarst  has  been widely  adopted as  a  noun  to denote  the
morphology of  the  highly dissolved veneer  itself.   This usage
                             152

-------
is  employed  here.    The  epikarst  zone  owes   its  origin  to
dissolution  of the  carbonate  substrate beneath  a  mantle  of
soils, residual clays, and clastic materials (Song, 1986).
     The usage of epikarst aquifer by Mangin (1975),  as well as
by  most  subsequent  workers,  is  restricted  to the  saturated
parts  of  dissolution  veneers occurring  on elevated  outcrops
which are separated  from  underlying  areally extensive aquifers
by  a  vadose  zone.     Although   there   are  important  lateral
circulation  components  within the elevated  epikarst  aquifers,
they  are  conceptulized  as compartmentalized collector  systems
which ultimately funnel water  to  infiltration  conduits through
the  vadose  zone   (Williams,  1985).    Vertical flow  is  thus
emphasized over lateral circulation.
      In China, most of the exploitable saturated epikarst zones
occur on  broad plains and along stream  valleys.   The  water in
them  represents the  top of a fully  saturated substrate.   There
is   no  underlying   vadose   zone,   and   lateral   circulation
predominates.
                     STONE FOREST AQUIFERS

     Storage within stone forest aquifers occurs in dissolution
voids  in  the carbonates  and in intergranular  porosity within
the  infilling  sediments.    Permeability  through  the  aquifers
results from (1) interconnected dissolution cavities within the
carbonate  bedrock,   (2)  partings  between  the   carbonates  and
infilling sediments, and  (3) intergranular permeability within
the infilling sediments.  The permeabilities of the dissolution
cavities  and  partings are  often extremely large,  whereas  the
intergranular permeabilities of  the  arkoses and clays  are very
small.
     Figure   4   summarizes   the  porosity   and  permeability
distributions in a typical stone forest aquifer.  The fact that
the stone pillars have virtually no  porosity combined  with the
fact that the clastic  infills have porosities of about  10 to 20
percent implies that the porosity  of a stone forest aquifer is
considerably  less  than  that  of  a corresponding  volume  of
clastic rocks.  As  Figure  5  shows,  the reservoir capacities of
the stone forest  aquifers  are severely limited  by the  thinness
of the epikarst zone.
            HYDRAULIC RESPONSE TO SEASONAL RAINFALL

     Most  of  the  south  China  karst  belt lies  in  the  humid
subtropical monsoon climatic zone.  In a typical year, 70 to 75
percent  of  the  1  to  2  m of  precipitation  falls  during  the
monsoon season from April  to August.  This produces flooding on
the  karst  plains and  in  karst  depressions.   The  stone forest
aquifers become  fully  recharged to the point  that water levels
lie  at or  above the  ground  surface.   The water  levels  fall
                             153

-------
    Reservoir
  S 40
  B eo
   80
   100,
          Aquifer porosity
          assuming clastic
          ' infill has 20% /
          pore space
                 Or
   LATERAL
PERMEABILITY
Porosity of
epikarst zone
in carbonate
bedrock without
clastic infill
        POROSITY
       20  40   60  80  100

        POROSITY (Percent)
                                            Interclastic
                                            permeability

                                            Partings between
                                            clastic infills and
                                            carbonates and
                                            karstic conduits
                                            Partially filled
                                            dissolution widened
                                            fractures that are
                                            increasingly cemented
                                            with depth
                100
                 10"6 10'5 10""  10'3  10"2  10"'  1
               EST. HYDRAULIC CONDUCTIVITY (Meters/Second)
Figure  5.   Idealized porosity and permeability distributions in
            a typical stone  forest aquifer  based on observations
            from hand dug  wells and quarry  walls in the  vicinity
            of  Xiaopingyang   and   Laibin,   Guangxi  Autonomous
            Region, China.

rapidly with  cessation  of   the  monsoon  rains.    Early on,  the
water exits the karst plains both as surface runoff and lateral
discharge  through the stone forest aquifers.
     As  the   dry  season   progresses,    increasingly   greater
percentages of  the total  discharge from  the  plains  circulate
out through the stone forest aquifers.   By October or  November,
water  levels have  fallen below  the land surface to depths that
commonly   exceed  1/2  meter   or  more,  even   in   lowlands.  By
December,  the  bulk  of  the  annual  precipitation has flowed out
of the  region through the aquifer and  the  region is starting to
show signs of drought.
              QUALITATIVE IMPACTS  OF DEFORESTATION

     The  deforestation  of  south  China   took   place  in  three
phases  beginning  with  the   Great  Leap  Forward  campaign  in
1958.   Chairman Mao Zedong's  objective  in  instituting  the Great
Leap Forward was  to  build a modern infrastructure  in  China, an
effort  that required  steel and  cement  in  huge  quantities.   The
south  China forests  would  fuel this  effort.   This  first major
assault on  the  forests  began immediately and  was  particularly
effective.   The trees  were  cut and converted into  charcoal.  In
many  places  in  south  China,  as  shown  on  Figure  6,  thick
                                154

-------
subtropical forests  were  cut to  the  last  tree on  the  tops of
the most  remote  karst hills.   Once this phase  was completed,
cutters returned to the forests and uprooted the stumps by hand
which, in  turn, were  converted  into  charcoal.   The vastness of
this  program   is  unknown   to  me.   However,  the  forests  were
virtually  eliminated,  save  only  for  a  few  special  forest
preserves, for as  far  as  one  could  see  from  Guizhou and Yunnan
provinces, and the Guangxi Autonomous Region, which  I  visited
in south China in  1988 and 1990.
                        the  deforestation  is  revealed not  by
                       rather  by  the  following  account  related
                          Linxu  Commune  was  the center of steel
                        Guangxi  Autonomous  Region.   Tens  of
           of  people  were  divided into many  companies to mine,
transport  ore,  fell   trees,  make  charcoal,   smelt  steel  and
process timber.  Many blast furnaces were built and their fires
burned  very  brightly.    In   1958,  about  3,000   people  were
employed  to  cut  trees, each  responsible ofor supplying  a  kiln
                                                              i -^
     The  magnitude  of
grand statistics, but
to me by Chen Yao Yuan
smelting  near  Mashan,
thousands
load of wood per day --
of  wood  were  burned.
bare.
                        approximately 1 mj
                        The  nearby  karst
  Every day
hills  were
3,000 m-
stripped
Figure  6.    Contrast   between  deforested  karst  hills  and  a
           surviving vegetated  hillside  behind  a small village
           in  the  area south  of  Wuxuan,  Guangxi  Autonomous
           Region, China.
                             155

-------
     Deforestation  followed  in  two  more  phases.    Extensive
cutting took place during the Cultural Revolution between 1966
and 1976  when anarchy prevailed and  the peasants  did  as they
pleased with  nearby  resources.   The  last serious deforestation
phase occurred  in the period  following  the decollectivization
of  the  agricultural  communes  in  1979.    As  the  peasants
dispersed into  the  countryside,  they reentered  the forests to
harvest the remaining trees in order  to build new dwellings.
     The ensuing  impact  on ground  water  supplies was  the loss
of  what,   in  retrospect,   the  Chinese  call  their  green
reservoir.   The amplitude of the  flood-drought  cycle  has been
exacerbated.   This  impact  is compounded locally  by  climatic
changes  attending   the   loss  of  temperature   and  humidity
moderation  once provided  by the forests  giving  rise to drier,
hotter dry  seasons.   Desertification has begun  in  some of the
drier areas.
     The  impact  on  the   stone  forest   aquifers   is  directly
attributable  to  loss of  ground  water  storage in the  formerly
forested hills.  Small springs on the flanks of the  hills which
were  once  perennial  are  now  ephermeral.    Surface  flows,
essential for  late  dry  season irrigation and  recharge,  are now
characterized by  early and  rapid recessions.   These diminished
releases of  water  from  the  green reservoir  decrease dry season
recharge  to  the   stone   forest  aquifers which,  in turn,  is
manifested as reduced discharges from springs and greater water
level declines.   Consequently, dry wells and  dry karst windows
have become  more  common  as the dry season  progresses.   At the
same time,  the local population is attempting  to develop more
ground water  to mitigate  the losses.
     Reforestation   efforts   have   been  undertaken.     These
included  aerial reseeding  in  the  early  1960s.   Reforestation
programs have met with some success,  although two trends thwart
regrowth.     First,  the  area  is   experiencing  a   population
explosion,  and second,   this  population  relies  on   plants  for
fuel.  Everyday, without  exaggeration, armies of peasants climb
into the  hills to cut brush and weeds  to fuel  domestic stoves
and  various  cottage industries such as  brick  or lime  kiln
operations.   The  result  is  continuous  cutting at  a rate that
appears  to  exceed steady-state regrowth  in many  areas.  Where
before  a  small  forest  area would  produce  relatively  good
quality  fuel on a  reasonably  steady  basis, now  several hills
must  be  scavanged  to  produce  poor  quality  fuel having  an
equivalent BTU content.
     The   primary  attribute  of  the   ideal   tree  used  in
reforestation  is  that it grows  rapidly.   It  is considered a
great  species  if  it can  resprout  if   chopped  off at ground
level.   Chen Yao  Yuan  provided  the   following  list of species
that are in  common  use:   (1) Melia azedarach  Linn, (2) Rader
machera  sinica  (Hance)   Hemsl,  (3)  Toona  sinensis  (A.  Juss)
Roem, and  (4) Zenia insignis Chun.   The Chinese foresters and
hydrologists  are  continually  seeking  alternative species which
they can test on karst soils.
                             156

-------
                            SUMMARY

     Stone   forest   aquifers   have    very    large   lateral
permeabilities and very small storage capacities.  Consequently
they usually  have poor  development  potential.   Their  extreme
importance  in  the  region  results  from  the   fact  that  they
underlie most  of  the cultivated lowlands  floored  by carbonate
rocks in south China.
     The primary  shortcoming of  these  aquifers is  their  poor
ability  to  retain  in  storage  the  plentiful   waters  recharged
during the April to August monsoon season.  Within two to three
months  into  the ensuing dry  season, much of   the  water  in the
aquifers has  drained out of the region  owing  to large  lateral
permeabilities.
     The  poor  storage  characteristics   of  the  stone  forest
aquifers  have  been   exacerbated  since   1958  by   massive
deforestation  in   the  upland  areas.    The  "green  reservoir"
formerly retained  large  volumes of water  in the uplands which
was released  to the  down-gradient  stone  forest aquifers during
the  dry  season,   thus   partially   mitigating   water   level
declines.      Now   this   hydraulic  moderation  is   greatly
diminished.     The  result  is  a  water  supply  crisis that  has
become more  severe during  the past few  decades  as a result of
three mutually coupled trends.   (1)  The  surface  water  flood-
drought cycle has worsened through the cumulative hydraulic and
climatic  impacts   attending  deforestation.     (2)  Development
pressure  has  increased  on  the  stone  forest  aquifers   as  the
surface  water  resources  have  become  less  reliable.   (3)  A
population explosion in the  karst belt  has placed  increasing
demands  on  the total  water  supply  and  at the same time  the
people's  requirements  for  wood  fuels   have  contributed  to  a
serious lack of progress on reforestation.
                         ACKNOWLEDGMENT

     Financial  support  for this  research was provided  by the
Institute of  Karst  Geology,  Geological  Academy Sinica, Guilin;
Institute  of  Mountain  Resources,  Guizhou;  and  Comprehensive
Technical Development Center for Mountains of Guangxi, Nanning,
Peoples Republic of China.


                        REFERENCES CITED

Huntoon,  P.  W.    1992a.    Hydrogeologic characteristics  and
    deforestation of  the stone forest  karst  aquifers of south
    China.  Ground Water,  v. 30. no. 2.

Huntoon,  P.  W.  1992b.   Exploration  and  development of ground
                             157

-------
    water  from  the  stone  forest  karst  aquifers  of  south
    China.  Ground Water.  v. 30. no. 3.

Mangin, A.   1975.   Contribution a  1'etude  hydrodynamique des
    aquiferes  karstiques:   Annuals  Speleology,  v.  29.  no. 3.
    pp. 283-332; v.  29.  no.  4.  pp.  495-601; v.  30-  no.  1. pp.
    21-124.
Song Lin  Hua.   1986.   Origination of stone  forests
    International Journal of Speleology.   pp. 3-13.
                            in  China.
Williams,  P.  W.    1985.
    development  of  doline
    Geomorphologie N.F.  v
    Subcutaneous  hydrology  and  the
and  cockpit  karst.   Zeitschrift  fur
.  29.          pp.  463-482.
Zhao  Yannian.   1988.   Exploiting  the Rocky Mountains  for the
    prosperity of minority peoples.  Printed separate, IAH 21st
    Congress,   Karst   Hydrogeology   and   Karst   Environment
    Protection, Guilin, China.  5 p.


Peter  W.  Huntoon  (Ph.D.,  University  of Arizona,  1970)  is  a
professor at the University of Wyoming.   His research interests
include   ground   water   exploration,   karst   and   fracture
geohydrology  nuclear  waste  disposal,  structural  geology  and
salt tectonics, and aerial photographic^assessments.
                              158

-------
Chairman Mao's Great Leap Forward and the Deforestation Ecological
Disaster in the South China Karst Belt

Peter W. Huntoon

Question:  Has  a  comparison been  made  between  water  balance
           calculations before and after the deforestation?

Answer:     I was not privy to such information.
                                159

-------
160

-------
            Session III:

Geophysical and Other Techniques for
       Studying Karst Aquifers

-------
162

-------
        Electrochemistry of Natural Potential Processes in Karst
                     By K. T. Kilty1, A. L. Lange2

               1  Diamond-Anvil Company,  Cheyenne,  Wyoming
                2 Geophysics  Group,  Wheatridge,  Colorado
                                Abstract

   Natural potential (NP)  anomalies are observed over caverns lacking
streams as well as over those carrying karstwater.  NP surveys have
proved useful for tracing karstwater between endpoints of tracer tests
and for facilitating the siting of monitor wells that tap directly into
the stream course.
   Electrokinetic processes are responsible for the principal observed
anomalies over karst systems. A streaming phenomenon results in an
electric potential gradient being established along a hydrologic
flowpath that is proportional to the Darcian velocity. This gives rise
to a conduction current in the reverse direction. This conduction
current is measured as NP.
   NP anomalies arise from four different mechanisms: refraction of
regional current flow by the cave void, axial flow through the cave,
infiltration into a cave from the ground surface, and flow immediately
around a cave driven by evaporation and temperature.
                              Introduction

   Irreversible processes occurring in the subsurface may lead to the
generation of other,  coupled processes. The primary process is sometimes
referred to as a primary flow and may include heat-flow,  fluid
transport, or diffusion of chemical species. It is relatively
independent of other  flows that may occur and can be thought of as an
independent variable. The generated process (secondary flow) depends
entirely on the primary flow for its existence and constitutes a
dependent variable. Of principal interest in exploration work is
electrical current which is measured as natural electrical potential
(NP).  By making measurements of natural potential one may make
inferences about the  nature of the causative primary flow.
   Diffusion or flow  of groundwater through porous or fractured soils
and rock is a pervasive primary flow. It is determined principally by
the distribution of piezometric head and variation in hydraulic

                                  163

-------
conductivity, and is quite independent of electrical potential in
ordinary situations. Groundwater flow induces an electrical current and
related electrical potential. The phenomenon in varying situations is
called electrofiltration, streaming potential, or motoelectricity, but
always results from a common underlying mechanism.
                    Physics of streaming potentials

   A simple mathematical formalism is available to describe the
interaction of arbitrary primary and secondary flows. However, a
description of the physical mechanism underlying the streaming potential
may provide more understanding of the phenomenon and insight into its
variations.
   A solid surface immersed in a liquid typically disturbs the fluid
structure. A polar fluid like  water will attain a preferred orientation
with respect to the solid.  In  addition, material may dissolve from the
solid and diffuse into the  fluid; or material may become adsorbed onto
the solid surface from the  fluid.
   This molecular and ionic structure produces an electrified interface
displaying a potential difference between the interior of the solid and
the bulk fluid. When equilibrium is attained much of the potential
difference occurs across a  narrow pair of charged planes at the
interface. This parallel plate capacitor-like arrangement is termed the
Helmholtz pair. An additional  potential change occurs beyond the
Helmholtz region through a  diffused distribution of ions known as the
zeta potential. Electric field in such an arrangement may be large; but,
its direction and magnitude vary rapidly and when integrated in any
particular direction it amounts to nothing.
   If this electrified interface is now subjected to a flowing fluid,
ions in the diffused charge region, which are of one sign, are carried
preferentially by the flow. Near the interface, where fluid is
stationary with respect to  the solid, ions,  which are of opposite sign,
are stationary. The charge  distribution is  deformed and oriented in the
direction of the fluid flow.
   With an orienting fluid  flow the integrated electric field now
produces a nonzero electric potential along the flow known as the
streaming potential. An important observation is that a distributed
charge region  (or zeta potential) is required to produce a streaming
potential. Whatever eliminates the zeta potential, increased ionic
concentration for instance, will eliminate  the streaming potential. A
reversed zeta potential reverses the sense  of the streaming potential.
Ordinarily positive ions are located in the diffused charge region.
However, acidic groundwater against carbonate rock may result in a
predominantly negative diffuse charge. Thus, a particular groundwater
flow may produce a positive or negative NP  anomaly.
   Laboratory experiments have shown that NP acquired through flow of
water in porous material is linearly proportional to Darcian velocity
over a broad range of pressure gradient and fluid composition
(Bogoslovsky and Ogilvy, 1972). Presumably this is also true of flow
through fractures. Thus, one may justifiably treat streaming potential
with a linear model.
                                   164

-------
                           Mathematical Model

   In order to model NP one requires a mathematical theory based on the
underlying physical process. In the context of NP arising from
groundwater flow one need consider only the coupling between fluid flow
in a porous solid and electrical current. The linearized, coupled
equations describing this in 2x2 matrix form are:
 I.
J

v
GRAD
where;       j =  electric current density
             a =  electrical conductivity
             Cffk = Ck
-------
Time dependence

   Although not shown explicitly in the mathematical model, time
dependence may enter through two terms.
   DIV • j is equal to the negative time rate of charge accumulation. It
is not possible to accumulate charge in a conducting earth, so one may
justify setting this term to zero.
   Additional time dependence may enter through the primary flow. At
steady state in homogeneous material y2A is zero. However, this term is
not zero if the primary flow has a time dependence. Time dependence
often enters NP field observations through propagation of the diurnal
temperature wave into the soil. Since groundwater systems often have
seasonal components of flow, one might also expect a direct time
dependence of streaming potential.
Inhomogeneities

   In inhomogeneous material the taking of a divergence in the
linearized, coupled equations (Eqs. I)  will result in terms of the form
GRAD X  • GRAD H where X is any material property and H is the
corresponding potential. Thus, inhomoganeity,  even in the material
properties of the primary flow,  will provide a source of natural
potential.
   Inhomogeneity creates distributed current sources that act
proportional to the gradient of the primary or secondary potential. An
extreme example is a sharp interface between dissimilar materials in
which case there is a surface density of current sources on the material
boundary that equals the product of potential gradient normal to the
boundary and the difference in material property. Inhomogeneity may also
occur as a gradual variation of material property over a finite
distance. In this case there is a volume density of current source.
Boundaries

   In homogeneous material at steady-state, where the Laplacian of the
primary potential is zero, boundary conditions determine entirely the
form of the natural potential. This appears to lead to a belief that
sources of NP arise only on boundaries between dissimilar material. This
is true only under the restrictive assumptions of steady state and
homogeneous material. As we have shown, inhomogeneity and time
dependence each add sources of NP.
   Primary and secondary flows each must meet certain boundary
conditions which need not be the same for each flow. For instance,
electric current flow at the earth surface must be parallel to it;
whereas, fluid flow may originate at the surface or pass through it.
Fluid flow may not cross an aquiclude; whereas, electric current may
pass through it unimpeded.
   Boundaries may be replaced by a distribution of sources that force
each potential to meet the required conditions. This is not a
mathematical fiction. Because of changes that occur at boundaries in
hydraulic conductivity, electrical resistivity, or other material
properties, electric current sources actually accumulate there. If there

                                   166

-------
exists a means of specifying these boundary sources the problem of
determining the natural potential simply reduces to that of integrating
the source density.
   In the simplest case, where the earth is assumed to be homogeneous
for both primary and secondary flows, the most important boundaries are
the ground surface and cavern wall. Since electric current may not pass
across either of these the ground surface and cave wall must be electric
current streamlines.
                Models of natural potential near caverns

   There are several different effects that individually contribute to
NP observed over and within caverns. For example, the void itself may
refract regional currents to form a local anomaly; fluid flow within,
around, and above the cave may contribute a streaming potential; and it
is also possible that chemical diffusion or thermal convection play some
role.
Influence of the cave void

   Embedding a high resistivity cave (void) into a conductive material
will refract any regional electric current. The regional current might
be unrelated to the cave or even unrelated to anything geological. For
example, there may be cathodic protection on a nearby pipeline.
   Consider a model consisting of a horizontal cylinder (cave) of radius
a at depth h as shown in Figure 1. A regional electrical current flows
from right to left. An approximate electrical model of the cave is a
dipole current source centered at y = h with sufficient strength to make
the walls of the cave a current streamline. Introducing this dipole
disturbs the streamline on the ground surface. Addition of an image
dipole at y = -h will restore the surface streamline but disturb the
cave streamline. Thus, one would have to have an infinite series of
current dipoles and images to achieve an exact solution. However, a
model consisting of a single dipole and its image is sufficiently
correct to calculate the effect of the resistive cave.
   As Figure 1 shows, the regional potential gradient, which is uniform
in the absence of a cave, exhibits an anomalously steep decrease over a
cave. The residual NP anomaly consists of a paired high and low with the
crossover point directly above the cave. These total and residual
anomalies are unlike those actually observed over most caves. In
particular the theoretical effect of a void is to cause a deflection of
the regional potential gradient without producing a local maximum or
minimum. This is still true if one assumes that a region of altered
resistivity exists above the cave.
   One generally observes a local extremum of natural potential over a
cave. From this observation alone it is obvious that natural potential
observed over caves derives from mechanisms beyond simple current
refraction.
                                   167

-------
Fluid flow above a cave

   Presumably caves are loci of enhanced permeability. They occur along
joints, fractures, or faults. There is reason to believe that there will
be anomalous fluid flow above caves from this enhanced permeability.
Also having a void in the subsurface may cause large and strangely
directed piezometric gradients. It makes sense to assume that
infiltration in the soil above a cave will be directed toward the cave
in some way and enter it via the ceiling. This does not mean that water
will drip from the cave ceiling — it may simply evaporate. Here there
will be a distributed positive source not much wider than the cave
itself. Near the ground surface, at the point of infiltration,  there
will be a distributed negative source. This may be broad or narrow
depending on the width of the infiltration zone.
   Assume in the simplest case that the two distributions are of equal
width. They also have equal charge density since by charge conservation
their sum is zero.  This distribution of vertical bipoles, shown in
Figure 2, will exhibit a surface potential transverse to the cave that
has a deep central negative anomaly superimposed on a broader,  shallow
negative anomaly. The broad shallow anomaly results from an inverse
square decline of potential away from the sources; while, the additional
deep negative anomaly results from modulation by a cos(0) term. The
important observation here is that the anomaly has one sign. Under the
assumption of equal source density for positive and negative sources,
and a vertical orientation of bipole axes, NP measured on a surface
profile will always be of one sign.
   Equal current source density for recharge and discharge flow regions
seems an unlikely situation and restricts the variety of NP anomalies
that one expects to observe. More likely water inflow to a cavern is
collected over a wide region and funneled into a narrower region at the
cave ceiling (Lange and Quinlan, 1988) . Providing that the recharge
region is quite broad, that the discharge is quite narrow, and that the
vertical distance between the two is not large, the central portion of
the NP anomaly reverses polarity. What results is a sombrero-type
anomaly of one polarity or the other depending on groundwater chemistry.
Figure 3 shows this situation.
Flow within a cave

   Caves may contain flowing water or they may simply contain wet  •
sediments on their floor.
   A cave whose cross-section is completely filled with flowing water
ought to have an NP field  like that of a flowing stream. Fluid flow in
this case is predominantly along the cave axis and constitutes a three-
dimensional effect to be covered in the next section. An additional
approximately 2-D flow pattern contributes NP from two sources; a
diffusion front and a flow field.
   Krajew (1957) observed  that NP changes abruptly at the bank of a
surface stream. To explain this he idealized a stream as providing a
source of water to infiltrate the banks and bed producing a diffusion
front between waters of differing chemistry. The diffusion front is an
electrical double layer with the most mobile ions forming the layer
farthest from the stream axis. When an image of the surface stream is

                                   168

-------
added to make the ground surface an electrical streamline, the result is
a closed electrical double layer in the shape of a cylinder coaxial with
the stream. A theorem from potential theory states that the interior of
this surface (the stream) will have a constant potential equal to % the
potential difference across the double layer. The exterior of the
surface should have zero potential. This model explains the abrupt
change in potential at the stream bank; but, it does not square with
observations taken across caves. According to this model, a cave
completely filled with water should exhibit no NP anomaly since all
ground surface profiles are exterior to the cave. Yet, NP anomalies are
always observed on profiles transverse to caves.
   Contrary to Krajew's idealization water does not necessarily flow
radially outward from a stream or a cave. Caves, in particular those in
dense carbonate rock, allow virtually no penetration of the surrounding
rock through pores, but only along joints and fractures. This leads one
to conclude that a perfectly cylindrical electrical double layer from
.diffusion is not likely to obtain, and that some NP signal from
diffusion exists exterior to a cave. The resulting potential could take
a variety of forms and it make little sense to enumerate all
possibilities here.
   Caves that contain free water may, like streams, be gaining or losing
this water along their axes. Water is either entering the cave from the
surrounding ground or leaving the cave to its surroundings. In the first
case one expects cave walls to appear as a negative current source while
the surrounding earth acts as a distributed positive source. An opposite
situation holds in the second case. Our electrical model predicts a
concentrated potential anomaly in the midst of a distributed anomaly of
opposite polarity. The distributed source might be so diffused that it
may not be readily apparent and not observed against background NP
variations. In this case one observes what appears to be an isolated
positive or negative anomaly.
   Fluid flow immediately around a cave implies, again, generation of a
sombrero-type NP anomaly. Figure 4 shows one possible model and its
resulting NP profile. Note here that the anomaly has small amplitude
which is a reflection of having sources of one sign imbedded within
sources of opposite sign. The geometry is like a cylindrical double
layer which has a zero external potential.
Three dimensional effects

   Like surface streams, caves may be gaining flow in one segment and
loosing in another. Two-dimensional models presented above are therefore
abstractions. Only very unusual circumstances permit a purely two-
dimensional fluid and current flow around a cavern.
   One obvious three-dimensional effect is the obeservation that a
stream in any segment of the cave will flow in at one end and discharge
at another. This applies to the terminal openings of a cave at ground
surface, and also to any place within a cave where axial flow speeds up
(recharge) or slows down (discharge). Ions of one polarity are carried
by a stream from recharge to discharge and a return current is set up in
the surrounding rock or along the rock-water interface. A potential
gradient along the cave axis is thus created.


                                      169

-------
   A simple model for this situation is like one used to model NP over
sulfide ore bodies (Kilty, 1984). The model, as shown in Figure 5,
consists of a bipole current pair positive at the discharge end and
negative at the other. A profile transverse to the bipole axis at the
discharge end will exhibit a positive anomaly directly over the cave
discharge. An opposite anomaly occurs over the recharge end. Somewhere
between the two a transverse profile may have a zero anomaly, but this
is not likely to be observed in any real situation.
   Additional three-dimensional NP effects arise from influences
unrelated to the cave itself. For example, there is often a correlation
between ground surface elevation and NP. A streaming potential from
draining vadose water is the most likely cause. This anomaly can be
confused easily with that generated by sources directly related to the
cave.
   Cultural effects can be very important in populated areas. NP is
generated by metals corroding in soil, cathodic protection systems,
electrified railroads, and Faradayic rectification at ground points of
the power distribution system. We have observed TV and radio towers to
be the locus of NP anomalies. Grounded wire and chain-link fences have
an influence, as observed by us in a number of surveys.
NP anomalies in dry environments

   Certain caves over which Lange et al (1990)  have observed NP
anomalies are in very dry environments where very little moisture is
expected in the surrounding ground.  However, humidity within these caves
can be very high (100%) and the floor can be muddy or wet. That an NP
anomaly is observed at all suggests  that there  is at least enough soil
moisture in the surrounding rock to  provide electrical conduction; and
this moisture may still participate  in a fluid  flow.
   The floor of a cave that contains damp sediment acts as a source of
water which may drain under the influence of gravity into the cave
floor. In this case the cave floor will appear  as a negative anomaly and
the surrounding rock as a broadly distributed positive.
   Even if water is not drain it may move under variations in capillary
forces from evaporation or variations in temperature. It may be wicked
up along or near the cave walls to re-enter the cave along its upper
walls or ceiling. One would expect the cave floor to be a negative
current source and the ceiling to be positive.  Ceiling/floor
polarization is reported in several  caves by Lange and Quinlan (1988),
Lange et al (1990), and Lange and Wiles (1990).
   A ceiling/floor polarization constitutes a bipole current pair of
essentially equal size and density.  It should lead to a anomaly of a
single polarity, the polarity being  that of the current source nearest
the ground surface.
   Although there are no supporting  NP observations for this, a cavern
may act as an evaporative sink for capillary water in the surrounding
ground. The entire perimeter of the  cave is a discharge in this
situation, and should exhibit a single electrical polarity. It resides
within a broad and diffused current  source of opposite polarity, the two
sources constituting an electrical double layer. If the double layer
were cylindrically symmetric it would result in there being no NP
anomaly on a surface profile. More realistically, however, the double


                                   170

-------
layer would have some cylindrical asymmetry leading to a wide variety of
NP anomalies. In particular, if the line connecting centers of negative
and positive current sources is not vertical a positive/negative pair of
NP anomalies is generated; and, since one source is more compact the
resulting anomaly would appear as an asymmetric sombrero.
                               Conclusions

    We have examined several possible models of generating NP anomalies
near caves. Variations in NP occur both transverse to the cave axis and
longitudinally along it. Despite the apparent diversity of possible
mechanisms they all share one characteristic. Each generates a paired
positive and negative current source that are closely situated in space.
If the two sources are oriented with one vertically above the other, and
with the sources of equal current density,  the resulting NP anomaly is
of one sign. However, in reasonably realistic situations one current
source of the two is more compact. This produces sombrero-type anomalies
consisting of a sharp NP anomaly of one sign placed within a broader,
more subtle one of opposite sign. At times  the broader anomaly may be so
subtle that it goes unrecognized in normal  background noise.
                               References

Bockris, J. O'M., and A.K.N. Reddy, 1970. Modern electrochemistry.
Plenum Press. New York.

Bogolslovsky, V.A. and A.A. Ogilvy, 1972. The study of streaming
potentials on fissured media models. Geophysical Prospecting, v. 51 (1):
109-117.

Kilty, K.T., 1984, On the origin and interpretation of self-potential
anomalies. Geophysical Prospecting v. 32: 51-62.

Krajew, A.P-, 1957, Grundlagen der Geoelektrik. V.E.B. Verlag Technik,
Berlin, DDR. 357p. [In German, with separate Russian and non-Russian
bibliographies.]

Lange, A.L. and J. F- Quinlan, 1988. Mapping caves from the surface of
karst terranes by the natural potential method. In: Proceedings of the
2nd conference on environmental problems in karst terranes and their
solutions, National Water Well Association, Dublin, Ohio, p. 369-390.

Lange, A.L., Walen, P.A., and R. H. Buecher, 1990. Cave mapping from the
surface at Kartchner Caverns State Park, Arizona. American Society of
Photogrammetry and Remote Sensing: Third Forest Service Remote Sensing
Applications Conference, Proceedings: 163-174.

Lange, A.L. and M. Wiles, 1990. Mapping Jewel Cave from the surface!
Park Science, v. 11  (2): 6-7.
                                   171

-------
                                                meters
                               ,residual anomaly
                                               meters
Figure 1. Regional electrical current flowing from left to right is
refracted around the void representing a cave. This deflects the profile
of potential away from a constant gradient as shown. A void may produce
an anomalous potential gradient,  but it cannot produce extrema of
potential.
                             172

-------
                               Distance
                                                    Surface
                                                        p/v •-•
                                         Negative Source
          Positive Source-
                                Cave
Figure 2. An idealized electrical  model of infiltrative flow that
originates in the soil above a cave.  In this particular case the
negative and positive current sources have equal density.  Because  the
negative source is closest the ground surface, the resulting potential
is entirely negative.
                                 173

-------
 CO
 -*-J
 ~o
                               Distance
     Negative  Source
                                         ^Positive source
                                Cave
Figure 3.  Another possible model of infiltrative  flow.  In  this case the
negative current source is very broadly  distributed.  Groundwater flow is
funneled into  a narrow region directly above  the  cave.  This results in a
more compact positive current source at  the cave  ceiling.  An NP profile
at the surface above this cave would have a sombrero  shape.
                                 174

-------
 CO
 +->
 o
                               Distance
                                                     Surface
           Negative  Source
                            \   Cave
                +  -f +  +
-t- -f -j~i_:
+ + £TL
+++++++

f

%\ + -
:f;:
H
--" + + + * + +
+ + *
                                        ^Positive Source
Figure 4.  Water flowing or draining away  from a cave results in a
compact negative source at the cave walls and a distributed positive
source around the cave within a large  volume of rock. An NP profile at
the surface over this cave would show  an  inverted sombrero shape. Since
the negative current source is located within the positive source,  the
anomaly magnitude is greatly reduced over that of figures 2 and 3.
                                 175

-------
                                    Observed  NP
               Profile
       Observed NP
Figure 5.  Water  flowing axially through a cave  creates a longitudinal NP
gradient.  At  the recharge end of the cave NP  is predominantly negative-
at the discharge end it is predominantly positive.  This 3-dimensional
effect is  in  addition to 2-dimensional flow effects and results in
additional variety of observed anomalies.
                                  176

-------
Kevin T. Kilty
Response to question  "What  would be the most  appropriate way to
confirm the cause of a compound (s-shaped)  anomaly?"

   Presumably by s-shaped the  interrogator means sombrero-shaped
anomaly, since  this  type  of  anomaly  has  a  compound  source of
positive and negative currents. Two entirely different approaches
might provide confirmation of the cause of such an anomaly.
   In  very specific  cases,  a second  geophysical method could
provide  confirmation.  For  example,  in  the case  of  an   anomaly
resulting  from  an  underground  cavern,  a  gravity  survey  or
resistivity survey could confirm the presence  of a void space if
•the void were shallow enough for these methods  to  detect it. In
many cases  gravity or resistivity  surveys cannot detect  a void
because it is too deep or too small to resolve. Natural potential
might detect the same void  because of altered  ground water flow
between the void and ground surface.
   In  many cases  no  means  of  verification  exists   other  than
drilling. The shape of the NP  anomaly  provides information about
the maximum depth  of  the cause,  and if  drilling proceeds beyond
this depth without finding a cause, then one must assume that the
NP results from  broadly distributed, shallow sources.  In this case
the cause  is  difficult  to  confirm no  matter  what  approach  one
takes.
   The most appropriate method will depend on  which of the above
methods  is  least  expensive  and provides sure  confirmation.  Most
often drilling is the choice.
                                177

-------
178

-------
    NATURAL-POTENTIAL RESPONSES OF KARST SYSTEMS AT THE GROUND SURFACE


                  Arthur  L.  Lange* and  Kevin T.  Kilty**

               *The Geophysics Group, Wheat Ridge, Colorado

              **Diamond Anvil Corporation, Cheyenne, Wyoming
                                 ABSTRACT

   Caverns  and   underground   karst  streams  have  produced  characteristic
natural-potential (NP)  signatures  at  the surface  in a  variety of  karst
environments.  These effects have been  mapped  in  regions as diverse  as  the
Central  Lowlands, Black Hills,  Great  Basin  and the Edwards  Plateau.  Since
the  discovery  of  the  NP  phenomena  over  karst  features  in  1986,  the
technique has been  tested for  purposes  of   1)  Siting  monitor wells;   2)
Mapping   extensions  of  known   karst  systems;  3)  Avoiding  contamination
through   placement  of  roads  and  structures   over  such  systems;  and  4)
Minimizing the effects of petroleum drilling  and deteriorating  casings on
caverns  and groundwater.

   Several  different  mechanisms contribute to the  generation  of  the  NP
anomalies, involving,  for the most part,  the  streaming,  or eJectrokinetic,
effect.  In  this  process  an electric  potential  gradient is established by
water  flowing  in   the   ground, whether through  grains  of  soil,   rock
fractures, or open conduits.  Anomalies can  result from  water  infiltrating
the roof of a cavern;  from capillary action  of water moving upward from the
moist cave  environment;  and  from  streams flowing  bodily  through open or
tube-full conduits.  In addition, effects of a void can  be expected  in  the
electric field produced  by nearby,  strong artificial  sources.

   Examples from caverns  in the Edwards Plateau of  Texas and  Jewel  Cave
National Monument,  South  Dakota illustrate  the surface  effects  over caves
lacking  known streamflow.  Particularly sharp anomalies characterize flowing
underground  streams;  such  as,  Honey   Creek  Cave,  Texas;  Parker  Cave,
Kentucky; Cave Valley  Spring,  Nevada and Lost  River,  Indiana.
                                    179

-------
Introduction

   This  paper  provides field evidence  from diverse  regions  of the United
States  demonstrating the  existence and  nature of  natural-potential   (NP)
anomalies associated with cave systems and  karst conduits. Hopefully, these
examples will  help  to  validate  the electrochemical and physical mechanisms
hypothesized in  the previous paper by Kilty &  Lange (1991).  Actually, the
mechanisms  were   arrived   at   inductively,   based   on   field  evidences
accumulated  over  karst  systems  since  1986;   hence,  the  models  are  the
product  of cogitating  the  field  data,  rather than the cornerstone on which
the field tests were founded.
Acknowledgments

   The  data  that  we  view  here have been  gathered  in  part  through  the
research efforts  of the authors;  but also as  a result  of  field projects
funded by state and  federal  agencies.  These include investigations for the
development  of  Kartchner  Caverns  as  a  State  Park for  the  Arizona State
Parks  Department;  tests over  Kentucky caves sponsored  jointly by Mammoth
Cave  National  Park  and  the Environmental  Protection Agency;  and studies
over  Jewel   Cave  National  Monument,  carried out  under  a  grant  from  the
National Park Service.  These field projects were  reinforced  by a detailed
investigation of  the published literature on the  natural-potential method
funded  also by the  E.P.A.,  Las Vegas.  In  addition,  many  individuals  and
cave  managers have  played  a  role,  and  these  have been cited  in earlier
reports and  publications covering particular surveys.


Field procedures

   Natural  potentials,  either  on  the ground surface  or underground  in  a
cave,  are   in  effect   voltage distributions,   arising  from  natural  or
artificial  d.c.  currents  in  the ground.  These currents  occur everywhere
upon  the  earth's   surface—even  in  the  oceans—and  arise  from  redox
processes  around  buried metals  or  mineral  deposits;   localized thermal
sources  and  solar  heating;  biological  activity  and  chemical  gradients;
artificial   electrical   sources  such   as  cathodic  protection  devices  on
pipelines;   and  most importantly,  from water moving  into and  through  the
ground. We  can  liken  these  sources to  a collection  of batteries  of  all
different strengths  and  orientations,  buried in  the earth.  What we measure
between any two points on the ground surface is  the difference of potential
(voltage difference)  resulting from  the  combined  current  of  this  entire
assemblage of batteries.

   Voltage  differences  are  normally  measured  using  probes;  that  is,
electrodes,  such as  one would employ to troubleshoot a radio or TV set. In
field applications where measurements  in  the millivolt range are  required,
the usual metal  probes or stakes are  unsuitable  owing  to their tendency to
polarize or build  up  a charge;  we employ  instead non-polarizing  sealed
electrodes composed of a particular metal immersed  in one of its own salts;
most  commonly a  copper rod  immersed  in  a bath   of  copper  sulfate.  The
typical  NP  survey  utilizes  one stationary  electrode  partly  buried  at  a
central  location—the  reference  or base  electrode—and  a second  roving
                                   180

-------
electrode that samples the  ground  at  appropriate intervals along a grid of
lines,  oriented generally orthogonal  to the  axis of the target, such as an
underground stream. The two electrodes are  connected to a precision, high-
impedance multimeter  by  means of  a  long, color-calibrated  insulated wire
wound on  an aluminum reel.  In  karst work,  digital  meters  that  read to
millivolts or tenths  of millivolts  are required; and under dry conditions,
input impedances  of  1000MQ  (megohms) or greater may  be  necessary.  Under
unusually  noisy  conditions  it is  necessary to employ  signal  processing
equipment in order to obtain representative readings.

   Readings are made  by inserting  the  roving electrode  in  shallow  (10cm)
holes at  intervals of 3 to 10m or  less,  depending on  the target  size and
depth.  Stations  as close  as  one meter  may  be  required to  resolve small
conduits. We normally take two or more readings at each station to  insure
that  they  are  representative  of  the  location.  Furthermore,  multiple
readings  are  made  in  situ around  the  base electrode  before and  after
reading every line, in order to measure temporal drift. Drift occurs  in the
electrodes  due  to temperature  changes  and,  in  the soil, to temperature,
moisture and chemical  fluctuations.

   When electrode  separations  exceed  one  hundred meters or more, it may be
necessary  to  install  a stationary  array  of  electrodes and   a  recorder to
 Figure  1.  Natural-potential  profile  over  a section  of Jewel  Cave,  showing
 the typical infiltration negative associated with basic water. Positives on
 either  side may constitute fringe effects of the anomaly;  however,  that on
 the right  1s in the vicinity of the water well to which it may be related.
 The leftmost low is a  likely  fault zone that may contain solution  work.
                                   181

-------
monitor  temporal  variations  resulting from  natural  telluric  currents  or
urban electrical systems. This  is  particularly the case in areas of  highly
resistive ground during magnetic storms arising from solar flares.

   At the  end of  each  day field  data accumulated in  a log book  or data
logger are  entered into a portable  computer,  so as to  monitor the  survey
results.  Specially designed  programs correct  for drift,  and  adjust for
slope distances, topography and artificial potential gradients.  In the case
of a systematic,  gridded survey,  the  final output  is  a display of stacked
NP  profiles,   potential  contours  and  an  interpretive map,   illuminating
targets  of  interest,  be  they mineralized   bodies,   buried   channels   in
alluvium, or active karst conduits.

   Correlations of  NP anomalies with  subsurface  voids  are  best viewed  in
cross-section;  thus,  the  examples  that  follow  will  be  in  the  form   of
profiles and  their corresponding  underlying  structure  as deduced  by the
geophysicist.   In  some cases  these profiles  will  be drawn  from a  set  of
parallel  profiles;  in others, they  will  be individual  lines  measured for
testing  the  response  over  a  known  or  conjectured  cavern.  Figure  1
represents  one of  four  trial  lines  over  the  labyrinth of   Jewel  Cave
National  Monument, wherein the underlying  cave cross-section is  an accurate
portrayal of  the  mapped passages  crossed by the  surface  profile.  In the
following observations,  we  present at  least  one such  example  for  each  of
the generative mechanisms expounded  in the previous paper  (Kilty &  Lange,
1991).

   From  this  discussion,   it  should  be  evident  that meticulous  data-
acquisition procedures and processing are  required to resolve typical karst
features from  the  surrounding electrical  noise, even  in remote locations.
Understanding  of  the NP  record and  its  correct  interpretation is  an art
 Figure  2.  Potential  profile   between  Parker  Cave  (C),  Kentucky  and  a
 pipeline (P),  having  cathodic  protection,  1.5km distant.  Voltage at the
 pipeline measured  1.2V more negative than  that at the cave.
                                   182

-------
                     anomaly.
                                                meters
 Figure  3.  Effect  of a  cylindrical  void in  a horizontal  electric  field.
 Here,  current  lines  are solid, while  potentials are  dashed.  The  effect on
 the surface potential  is  a  sinuous  anomaly  superimposed on  the  electric
 field  gradient.
that is  not acquired  from  this paper  or from a  short course;  it  is  the
result of  long,  grueling days  under  adverse  field  conditions,  numerous
repeatings of  data,  and  acute  observations of  the immediate  environment,
with attention to  possible  current sources.  In  the  absence  of such  a
commitment on the part  of the operator, the instrumentation  becomes little
more than a dowsing rod employed for a  bit of Sunday exercize.
                                   183

-------
Electrical refraction around a void

   In the  environment of  an electric  field  gradient,  as  commonly  occurs
around a pipeline containing anti-corrosion devices  (Figure  2),  a  component
of the  NP response of  a void can  be  attributed to refraction of  current
lines around  the opening  (Figure  3).  While profiles  measured over  Parker
Cave,  Kentucky clearly  exhibited  the  effect  of  a pipeline  more  than a
kilometer  distant,   their  contribution   to   the   cave  anomaly  was   not
detectable (Figure 4)(Lange & Quinlan,   1988).

   During the course of that same field project, however, a  profile  was  run
over a railroad  tunnel  in  the town of  Park City,  where a strong  potential
gradient can  be  attributed to nearby  gas mains.  The  result  is a  sinuous
anomaly  over  the  tunnel,   made  more   evident  after  removing  the   linear
gradient  (Figure  5).  In this  case,  we evidently  are  dealing  also  with a
filtration  component  that  emphasizes  the  anomaly  over  the   void   and
displaces  the  crossover  to  the   right.  The   profuse  dripping  and   ice
formations in  the  tunnel walls testify to considerable filtration  through
the  thin  roof  of  the  structure.  In general,  however,   the  refractive
component of  cavern  anomalies is  small,  and  is  confined  to the  immediate
vicinity of pipelines or other urban disturbances.
                                           Brown
                                             River
                                 r~^\
                               w    V-M*
                             -<,Q^   VJu
                                *• ^"-i.     .«. i •
                                     ^4:1}
 Figure  4.  An  NP  profile  (dashed)  over  Brown  River  as affected  by the
 potential  gradient  of the  nearby pipeline (cf.,  Figure 2). The solid  curve
 is  the  residual anomaly  after  removing  the pipeline linear trend. Though
 noisy,  the  anomaly  is typical of that observed overastream cave,  wherein a
 peak  (here positive)  is flanked by lesser  excursions of  opposite  polarity
 (negative).  These  results were  obtained  during  January  1988  (Lange  &
 Quinlan,  1988). See also Figure 9.
                                   184

-------
               L  & N Tunnel
               Park  City' Kentucky
   ^natural potential
         40-
         20-
        -20'
                     10
                               20
                                         30
40
50 meters
         20H
                                   residual NP
        -20-P
                                                             50 meters
          0-
        
-------
the  pH  of  the electrolyte;  thus,  in  the  case of  basic  water,  negative
anomalies can  be  expected;  and  with acidic solutions,  positive anomalies.
Many field  measurements both  above and  below  ground will  be necessary  in
order to  test  this polarity  assertion  under   natural  conditions.  In any
event,  we  have observed both  positive as  well as negative anomalies over
cave passages   showing  evidence  of  active filtration  (V.  also  Ishido  &
Mizutani,  1981).

   Natural  Bridge  Caverns,  Texas,  presents  strong  negative  as well   as
positive  expressions  over  different  portions  of  the  system.  Negative
anomalies  appear  over  the  recently discovered  west region  of  the cave;
                 O
                  -400
                    Line D
                          -200
                                          :oo
                                                     mft
                                            Hall of the Mountain King
 Figure 6.  Filtration  anomalies over  different  regions of  Natural Bridge
 Caverns,   Texas,   a)   Sharp  lows   occur   in   profiles  over  the   recently
 discovered Jaremy  Room, where  downward vadose water  is  presently evident
 underground,  b) A  60mV positive anomaly  over the  largest  chamber of the
 caverns is believed to be  the  result  of  water moving from the cave depths
 towards the surface under  capillary flow.  A second,  undiscovered passage  is
 deduced from the  lesser peak.
                                   186

-------
     millivolts
                                                INNER SPACE CAVERN,
                                                              TEXAS
    INTERSTATE
  HIGHWAY 1-35
 Figure  7.  Positive NP anomalies  corresponding to  passages  in  Inner Space
 Cavern, Texas. The expressions are typical  of cave  anomalies to be expected
 from  downward filtration of  acidic  water.  (The extreme  negative anomaly
 alongside the highway is  due to a  buried pipeline.)

while the 60mV  positive  anomaly  in  the  tour portion of  the  cavern occurs
over a  relatively  dry cave  ceiling (Figure  6). One  logical  explanation  is
that the negative  expressions over the wet  Jaremy Room  are the  effect  of
filtration;  while  the positives over  the  Hall of the  Mountain  King arise
from upward  water movement (See the section on capillarity, below).

   Inner Space  Cavern  was  discovered during  test   drilling  for  highway
construction  at  Georgetown,  Texas.  Presently,  an  interstate  highway
junction,  a railroad  line and city  roads  pass over  the  tourist  cavern.  A
test line run over  four  typical cave  passages produced four corresponding
and definitive positive  anomalies  (Figure  7). Though  drippage underground
exhibited  a  pH of  7.0 at the time, the management reports that  drip water
is generally  acidic,  a  fact  that  agrees  with  the rule  that  infiltrating
acidic water produces positive anomalies.

   The preceding Texas examples illustrate  the effects of infiltration over
shallow, individual cave  passages.  When we  proceed  to deeper systems—50  to
100m or more—we can  expect to  lose  resolution of  particular passages, but
observe instead the  compound expression of  the "cave-as-a-whole"  (Figure
1). This is  the case  at  Jewel Cave in the Black Hills, whose  lower levels
occur  at depths  greater than 120m. The result  is that  93% of the labyrinth
crossed by  the  NP profiles  were  electrically low;   that  is,   they  were
negative relative  to the  surrounding country  rock.  Measurements of drippage
in the  cave yielded  pH's of  8.3  to 8.4,  definitely basic  (Bakalowicz  et
                                   187

-------
 Figure 8.  NP response over  the  main passages of  Kartchner  Caverns  State
 Park,  Arizona.  The anomaly  here is  a  compound  feature  consisting  of  a
 double positive  flanked  by lows, interpreted to  be the  result  of upward
 migration  of water from  the very  moist  cave environment.  (From  Arizona
 Conservation Projects,  Inc.,  funded by the Arizona  State  Parks Department).


al.,  1987),  and  are commensurate with the  idea  of downward moving  water,
though  flow  in places  can  be  directed horizontally by  local aquitards.


Upward  water  movement  by capillary action

   During a  dry  summer in the arid  environment  of Kartchner Caverns State
Park  of  southeast  Arizona,  one  might  have  difficulty  imagining  any
appreciable  downward  filtration  into the parched rock  surface;  in  fact,
even  during the  occasional  thundershower,  runoff flows  rapidly  from  the
rock  outcrop  towards  the  alluviated  arroyos.   Bulk resistivities  measured
in the  massive  carbonate  over the  cavern exceeded  1000Qm,   compared  with
about 400fim in the vicinity of Parker Cave,  Kentucky (Lange,  et al.,  1990).
Nevertheless,  25 to 50m  below the  surface,  the  cavern  is  very  wet.  Its
floors  are muddy, speleothems are actively forming  and portions of  the cave
flood,  following major storms.

   Natural-potential anomalies over  Kartchner  Caverns showed- both  positive
as  well  as  negative   expressions.   We  see  typical  negative  filtration
anomalies over  the arroyos  containing likely  underflow;  however,  on the
carbonate outcrop over the main portions  of  the cave system,  the principal
profile produced  a  sombrero-like, mainly  positive anomaly (Figure 8). The
NP  signature  was  considerably   more prominent  than that  obtained from
gravity measurements  along the same  line.  In the  Big  Room  of  the  cave,
ceilings  proved  to be consistently  more  positive  than  the floor.  These
observations  all  point  toward   upward  wick-like  movement   of  water   by
                                   188

-------
capillary  action   via  interstices   in  the   cave  walls   and   ceiling,
transporting water upward from some  likely underlying  reservoir towards the
surface  where evapotranspiration  likely  predominates  over  any  downward
filtration.  Capillary  effects  on natural  potential  have previously  been
suspected by  Hoover  (1976),  and  the  result  of evapotranspiration  observed
in German forests by Ernstson  &  Scherer (1986), though  not  over caves. The
phenomenon may likewise  account  for the large positive  anomalies  seen  over
the main chambers  of  Natural Bridge Caverns  (Figure 6)  and Sonora Caverns
in Texas.
Electrokinetics of cave streams

   In a stream of an  alluvial  environment,  we can expect  some  leakage into
the banks  and  the stream  bottom.  Likewise  in a  porous carbonate,  such  as
the Floridian  Ocala  limestone, we  foresee  partial  water movement  into  or
out of the  cave  walls in  addition  to the main flow within the tube.  In a
dense carbonate  medium,  on the  other hand,  flow is primarily confined  to
the conduit.  Thus,  if  we are to  see  an NP  anomaly  crosswise  to a  cave
stream, it must  come  about from flow  along  the  axis of the  passage,  where
the mobile  ions  give  rise to an  electric  convection  current.  The  return
current through  the  surrounding rock walls  comprise a rather distributed
conduction current. The result to  be expected at the  surface  is  an  axial
     Fig. 9
                          PARKER CAVE
                            KENTUCKY
      -Eio-koo
-400  -!i50  -5)0-250'  -200
1
I-f-
M~
^~
LrJ
S~^


.



i
*
i
i
i
i




i
i
i
i
i
i •
-i- J
i

[i—
i 	
i
i
i
'—^
p-

••M
1 	
1
1

P-
1
^^m
1
p—
1
i -
1
1
1
i
1
I
r^
r
i
i
i
i 	
1 i '
i
i
^
^^
                                                 q
                                                 d-
                                                 q
                                                CL d-
                                                CL
                                                LJ q
                                                   Fig. 10


                                 5   CAVE VALLEY SPRING ,

                                 I              NEVADA
                                                  '-20  -10
                                                               10
                                                 20   30
                                                  meters
 Figure 9. NP  profile along  a road,  crossing  over Brown  River of  Parker
 Cave,  Kentucky,  recorded  in June  1991.  This is a typical M-shaped  negative
 anomaly to be expected over underground streams, and contrasts  in  polarity
 with the  W-shaped  positive  observed  farther  downstream  in  January 1988
 (cf.,  Figure 4).

 Figure 10. Sharp  NP stream-type  anomaly  recorded  over a  small  limestone
 spring in  Cave Valley, Nevada.
                                   189

-------
                                          LOST RIVER
                                        BOWLING GREEN, KENTUCKY
                 response after removal

                   of cathodic-protection effect
 Figure 11. Positive  response  over  the cave passage near  the discharge of
 Lost  River,  Kentucky.  The  cave here  was almost  tube-full, so  that its
 cross-section is drawn  schematically rather than from survey data.
 peak  corresponding to the underlying stream,  and parallel limbs of opposite
 polarity  and lower amplitude on  either  side.  Here again,  we encounter the
 sombrero-type  anomaly;  though the hat may be  right  side up or upside down.
 Thus  under  certain conditions,  stream anomalies may become more positive in
 the downstream direction, so that the discharge  end,  or rise, will measure
 positive  relative  to  the upstream end,  or swallet.  On the other hand, under
 a  different chemistry, the  reverse  may  occur; that is,  the downstream end
 could  become increasingly negative.  Somewhere about halfway, no anomaly may
 be observed.

   Over natural  cave  streams we observe  both  positive and negative sombrero
 anomalies.   Brown  River  of  Parker  Cave,  Kentucky   exhibited  a  positive
 expression  (Figure 4)  during  January  1988,   yet in  June  1991  a negative
 expression  was measured  along  the  roadside a hundred meters or so farther
 upstream  (Figure 9).  While the main  cavern in Cave  Valley, Nevada produced
 a  6pmV positive sombrero anomaly (Lange & Quinlan,  1988),  a nearby karst
 spring yielded  a  sharp  negative  (Figure 10).  Other positive  expressions
 have  been recorded over  Lost River  Cave,  Kentucky and  Lost River, Indiana
 (Figures  11 &  12, resp.).  Meanwhile,  in the Texas  hill  country,  we see
remarkably  sharp positive M-shaped  anomalies  over  passages  of  Honey Creek
Water  Cave,  at locations separated  by  several kilometers  (Figures 13,14),
yet another part  of  the system  appeared negative,  while still  another
channel produced no measurable  response.
                                   190

-------
                                                     LOST RIVER
                                                       INDIANA
 Figure 12.  Natural  potential  along Highway 37 crossing the dry channel of
 Lost River,  Indiana. The cave depth and location are estimated.

   To properly test  the electrokinetic  stream  mechanism, we  will  require
one or more systematic surveys over a  discrete  mapped stream passage—with
measurements conducted at one  time of year. Only  in this  way,  will  we come
to fully confirm the surface potential distribution both  transverse to the
stream as well  as in the axial  direction.
Summary

   Evidences  of  four  different  mechanisms   postulated   for  generating
natural-potential  expressions  over  caves  and  karst  conduits  have  been
observed in the field.  The mechanisms are the following:

   1) Refraction of  ambient d.c.  currents around a void;
   2) Downward filtration through the cave roof and walls;
   3) Upward migration  of flow from a moist cave environment towards the
       surface by capillary action; and
   4) Electrokinetic effect of flowing water in tubes.

We   do   not   expect   these  mechanisms   necessarily   to  be   operating
independently; for example, a stream cave having air space can also receive
drippage from the surface.  The resulting surface anomaly is then a compound
effect.   Similar anomalies  can  be expected  from  artificial  tunnels  and
aqueducts.  Furthermore,  other  natural processes  may  be  acting  as  well  to
produce  NP  effects; so  that in  time, other  mechanisms  will  need  to  be
evaluated.
                                   191

-------
     HONEY CREEK WATER CAVE
     Fig. 13     _
                                           millivolts
Fig. 14
                                                                     100
  Figure  13.   Particularly  sharp  stream-type  anomalies  are  observed  over
  segments of  Honey  Creek  Water  Cave,  Texas.  Here each conduit is expressed
  as a sharp M-shaped positive.

  Figure  14.  Detail  of a stream-type anomaly over another  segment  of Honey
  Creek Water Cave,  remote  from those of the preceding figure.

   Finally,  in  addressing  an audience of practicing  karst hydrogeologists,
we need  not  emphasize the environmental  and engineering applications  of an
effective  tracking system  for  caves and  karst conduits.  We have  already
tested  the technique  for  the siting  of monitor  wells in karst (Lange  &
Quinlan, 1988); for the delineation of subsurface flowpaths and  the  mapping
of drawdown  and recovery  patterns  associated  with pumping  (Lange,  1991).
Tests have been made  for  the purpose of avoiding construction failures  and
environmental  misadventures  over  cave  systems,  and  for avoiding  the
possible contamination of a major cave resource  by  nearby  drilling.  Perhaps
the demonstration most needed at  this point is the mapping by the natural-
potential  method  of  a  karst  pathway  connecting   the   endpoints  of  a
successful dye trace. We can know where  the dye goes  in and  where it  comes
out,   but to  track the  intervening  flowpath,   we  must  rely on  a  remote
sensing  tool  such as  that of  natural  potential,  in effect, a rapid  and
cost-effective method for solving environmental  problems in karst.


References

BAKALOWICZ, M.J.,  D.C. FORD, T.E.MILLER,  A.N.  PALMER & M.V.  PALMER  (1987).
Thermal  genesis  of dissolution  caves  in the  Black   Hills,  South  Dakota.
Geological Society of America, Bulletin,  v. 99:  729-738.

ERNSTSON, K.  & H.U. SCHERER  (1986). Self-potential  variations with time and
their  relation  to  hydrogeologic  and meteorological parameters.  Geophysics,
V.51  (10): 1967-1977.
                                    192

-------
HOOVER,  D.B. (1976). Capillary pressure  potential—A  significant source of
error in  self-potential measurements  (Abstract).  Geophysics,  v.  41   (2):
356-357.

ISHIDO,  T.  & H.  MIZUTANI   (1981).  Experimental and  theoretical  basis of
electrokinetic  phenomena  in  rock-water systems  and  its  applications to
geophysics.  Journal of Geophysical Research, v. 86 (B3): 1763-1775.

KILTY, K.T.  & A.L.  LANGE  (1991).  The electrochemistry  of  karst systems at
the ground  surface.  National Water  Well Association,  Third Conference on
Hydrogeology, Ecology,  Monitoring and Management of  Ground Water  in Karst
Terranes.  Nashville, Tennessee,  December 4-6.

LANGE, A.L.  (1991). Hydrologic flownet mapping  and  karst-conduit detection
using the natural  electric  field. American Cave  Conservation Association,
Cave Management Symposium.  Bowling Green, Kentucky,  October 23-26.

LANGE, A.L.  & J.F. QUINLAN  (1988). Mapping  caves  from the surface of karst
terranes by the natural potential  method.  National  Water Well Association,
Second Conference on Environmental  Problems   in  Karst Terranes  and their
Solutions, Proceedings,  p.  369-390.

LANGE, A.L.,  P.A.  WALEN  & R.H.  BUECHER  (1990).  Cave  mapping  from  the
surface at  Kartchner  Caverns State Park, Arizona.   American  Society of
Photogrammetry  and Remote  Sensing:   Third  Forest Service  Remote Sensing
Applications Conference, Proceedings: 163-174.

LANGE, A.L.  &  M.  WILES (1991).  Mapping  Jewel  Cave  from the  surface!   Park
Science v. 11 (2): 6-7.

SCHERER,   H.U.   &  K.ERNSTSON   (1986).   Untersuchungen   zur  Lithologie-
Abhangigkeit geoelektrischer  Eigenpotentiale   [A  study of self-potentials
and  their  relation  to  lithology].   Neues   Jahrbuch  fur  Geologie  und
Palaontologie, Abhandlungen, v.  172 (1):  21-45.
                                    193

-------
Lange & Kilty/Natural-Potential Responses

1.     You summarized by saying that natural potential (NP) can detect both
      shallow caves and deep caves, ranging from small to large. Relative to
      this statement, I ask:
      A. What are the limits on cave size detectable as depth     increases?
      B. Under what geologic conditions might the NP method be
      inappropriate?

2.     Please discuss the following as related to the sensitivity of your
      instrumentation:
      A. What is the sensitivity and noise level of the     instrumentation?
      B. What are the electrode noises?
      C. What are the typical levels of field noise?

3.     Do you think that NP is the best geophysical technique for identifying the
      presence of a cave? What are the advantages and disadvantages of the
      technique?
                                    194

-------
Lange & Kilty/Natural-Potential  Responses:    Answers to questions

1a. We  have detected caves  to  depths as great  as 250m  (Western  Borehole of
Lechuguilla Cave);  however,  conduits  less  than  about  3m are unlikely  to be
resolvable  below about  50m.  We must  keep  in  mind  that,  in  the  case  of
infiltration into an empty void,  much of the response  may be coming  from the
region between  the  surface and  the  cave  ceiling.  Where  the  response derives
from  flowing  water,  its  amplitude  at  the  surface  will  depend on  several
factors;  particularly  flow  velocity,  velocity  gradient, and  the  coupling
coefficients   (electrochemistry)   between   the   carbonate   solid  and   the
electrolyte (water). Limits of  detectability, therefore,  will be dependent on
the overall environment and flow conditions.

1b. We  have not yet found  situations in which  the method  is  inappropriate;
however,  there  are  some  in which  resolving power  is considerably  reduced.
.Several  examples  are      a)   Sites  having  a  very  electrically-conductive
overburden  or  wet  shales  near  the   surface;  b)  Hot,  dry  and  rocky  desert
environments,  where  ground  contact  resistance is  high;    c)  Frozen  ground or
ice and  deep  snow  cover;   d) Paved  areas  of concrete and  asphalt. There are
ways to solve these difficulties; for example, the  drilling of  small  holes in
the pavement;  however,  they generally require more  time  in the  field,  and, of
course, increased cost.


2a. The instruments can  be  read to ±0.05 millivolts; although this sensitivity
is seldom needed. Instrumentation noise level is  less than that.

2b.  Using  copper-copper-sulfate  electrodes, the   reading changes  by  about
-360uV per  °C; however,  by maintaining both  the  base  and  roving  electrodes in
similar environments, in terms of shade and sun,  this effect can be minimized.
Inherent electrode  noise has been measured  in the lab as <3uV in  one hour.

2c. Field  noise  varies  greatly. Using lines  less  than  about 500m  in  length,
typical variations  of  1  to  3mV are  seen  in pastoral environments.  Telluric
noise during severe magnetic storms  can be as great as  ±10mV  for line  lengths
on the order of 500m in a granitic or carbonate  terrane—a condition  that may
require suspending operations for an  hour  or more until  the  noise  abates.  In
suburban settings,  local  artificial noise  is typically around  ±4mV;  whereas in
an urban  locale having  electric trains,   as in Staten  Island,  N.Y.,  sudden
jumps of 25mV  or more are encountered.  The  latter problem  can  be  surmounted by
utilizing an averaging,  signal-processing  receiver  which samples  as many as 16
readings at each position.


3. While NP anomalies can be  produced by a  variety  of processes;  it has  proved
to be a  definitive  indicator  of caves and  karst streams  in about 95%  of the
controlled cases studied. The  problem rests in  distinguishing cave  anomalies
from  those  produced  by  metal  conductors,  lithologic   changes,   localized
mineralization,  etc.—all  of which can produce   anomalies.  A  companion  survey
using an  EM device  can  help in  eliminating these.  As  a  rule  the NP cave-
resolving power seems to be  much higher than other geophysical  methods,  such
as gravity, resistivity,  and seismic;  furthermore,  it  is the  only  standard
                                195

-------
geophysical tool that  responds  to the movement  of  water rather than its mere
presence. The method is more economical  to employ than the aforementioned and
is  considerably  faster.  A  good  strategy   would  involve  running  an  NP
reconnaissance survey,  followed by NP  detailing  of  significant anomalies and,
finally,  localized  gravity traverses  over the  most  promising targets  as  a
confirming technique. The  main  disadvantages   lie in  the interpretation;  that
is, in distinguishing cave anomalies from  other  types and in evaluating these
in terms of depths  and  dimensions.
                             196

-------
      WATER TEMPERATURE VARIATION AT SPRINGS IN THE  KNOX

                GROUP  NEAR OAK RIDGE, TENNESSEE
                       GARETH J. DAVIES
                    Geraghty  & Miller,  Inc.
                     Oak Ridge,  Tennessee
                           ABSTRACT

         Water  temperatures  were  measured  during  1991  at
several springs  draining Cambrian-Ordovician  age,  Knox  Group
dolomites  near  Oak  Ridge,  Tennessee.    The  coefficients  of
variation  for temperature at each spring indicate,  there are two
different  groups  of springs, one group with essentially constant
temperatures,   and   a   second  group   with  more  variable
temperatures.  The maximum temperature occurrence  at  the springs
is observed to  lag  at least  50 days  behind the  reported maximum
mean air  temperature for  Oak Ridge  for 1991    (NOAA,  1991) .
Although the amplitude of temperature variation is different for
the two groups  of springs,  the lag-times  are  similar.   The lag-
time is related  to the  thermal  diffusivity of the  system,  and
the  amplitude  of  temperature  variation is  related  to  the
relaxation  length  or  temperature  damping factor  within  the
system.  The only model  that  best  explains the flow  system is
one where  there  are  different  proportions  of quick-flow  and
slow-flow components  at each spring.  This is because the groups
of  springs  have  similar  lag-times but  different  temperature
variability  and  all drain  the same aquifer.   Analysis  of
carefully  collected  temperature  data  can  yield  meaningful
insights into aquifer mechanics.
                         INTRODUCTION

         Ground-water flow  in  karst  aquifers  was  characterized
by Shuster  and White (1971)  as  a continuum,  the  two end members
of which are,  diffuse  flow or  conduit  flow.   Smart  and  Hobbs
(1986)  further  defined   the   characteristics   as   a  three-
dimensional continuum  where,  the  relations between  recharge,
                              197

-------
ranged from concentrated to  dispersed,  aquifer storage,  ranged
from low to high,  unsaturated to  permanently  saturated,  and  flow
ranged from conduit-flow to diffuse-flow.

          For  systems  dominated  by conduit flow,  hydrograph
response to rainfall is  relatively rapid  and  return  to base-flow
conditions  also  rapid.   For diffuse-flow  systems, hydrograph
response  to rainfall  is slow and  return to  baseflow likewise.
Chemograph response to precipitation for a mostly conduit-flow-
dominated  system  is flashy  and highly variable  and for  mostly
diffuse-flow  dominated  systems   is  much  less  so.   Ford  and
Williams  (1989, p.  204-210)  provide an  overview of how  spring
chemograph interpretation has been used to  study  karst aquifers.
Quinlan and Ewers  (1985)  and  Quinlan  e_L al. ,  (1991)  suggest  that
several measurements of  specific  conductance  at a  spring before,
during,  and after  several  storm events  is  a  simple  way to
determine  whether that  spring is fed by  mostly conduit-flow,
mostly diffuse-flow, or  is a  mixed type,  partly conduit-flow and
partly diffuse-flow.  Water  temperature  measurements have  also
been used as a method of assessing the flow characteristics  of a
spring by, for example,  Meiman et  al., (1988).

          When  karst aquifers  are  characterized using the  three-
dimensional cube model of Smart and Hobbs (1986)  and as  modified
by  Quinlan et al . , (1991,  Fig.  1)   involving  the  independent
affects of  recharge,  storage, and flow,  many different systems
can be conceptualized.   At one extreme a  predominantly fast-flow
(conduit)  system  would have discrete recharge via swallets, low
aquifer storage,  and  a  relatively straight connection  with the
spring,  a  diffuse-flow component  would  also  exist  in  such
systems  with    the  exception  of   systems   with   unique
characteristics.   An example of one  such unique system  would be
a conduit-flow system where the diffuse-flow component  would be
essentially  negligible  (e.g.,   Castleguard   Cave,   British
Columbia,  Canada)  where  during the winter months  and most  of the
year,  the  diffuse-flow component  could  be  considered  to be
effectively zero  (Smart  and  Ford, 1986).  At  the opposite end
member  of  the  continuum  a  purely  slow-flow  (diffuse-flow)
system, would have dispersed  recharge, high aquifer  storage, and
a complicated pathway to the  spring.

          It is known  that   the  pulse-through time (hydraulic
response)  is  often different from the  flow-through  time in
phreatic  systems  (Ford and Williams  1989, p. 227).  If  they are
similar then, as  Brown  (1972)  concluded, open channel  (vadose)
conditions  probably exist  within the  system.   When observing
water temperature  variation,  arrival  of a thermal  signature  at a
spring is  often associated with  a lag-time in response to  with
interaction  of the  recharge-water  thermal  signature  and the
aquifer's ambient signature.  If we consider the movement  of a
                              198

-------
slug  of  recharge  water  carrying  its  temperature  signature
through a  simple cave  system,  for heat  exchange to  occur by
mostly conduction,  the velocity of the  water  and the amount of
surface area  it  is  in  contact with  (the passage  size)  would
control the rate at which heat  is  lost  by the water and gained
by the rock.   So in a fast-flow system of  large conduits,  the
temperature signature would  probably  be  transmitted  as  far as
the spring.   In a slow-flow system, the  thermal signature of the
recharge water would  be damped or thermally  equilibrated with
the rock before reaching the  spring.

         Studies of temperature variation of  spring waters in
the Pennines of Yorkshire,  Great Britain, have used temperature
measurements to  estimate ground-water  contributions  to surface
flow  (Pitty, 1976,  1979) .  He  also noted temperature lag-times
at springs,  where the maximum temperature  at the  spring occurred
after the maximum mean  air temperature  or the maximum measured
air temperature.

         Roy  and Benderitter   (1986)   studied temperature  and
discharge  variation during  a  year at  a  spring in  the  Loire
Valley of  France;  there is  a  maximum  temperature  lag-time of
between 90 and 120  days.  Aley (1970)  measured the temperature
of a  small Ozark spring and  observed  a  temperature lag-time of
about 28 days.   Meiman et al..  (1988) collected temperature data
during a study of flood  pulse through a  karst aquifer at Mammoth
Cave, Kentucky, and  in this mostly  conduit-flow dominated system
short-term lag-times were measured  in hours.
                           STUDY AREA

          The  study area, a valley cutting through the Cambrian-
Ordovician Knox Group  dolomites near Oak Ridge, Tennessee, is in
the Valley and Ridge province  in the southern Appalachians.  The
geology of the valley  is summarized in Figure 1.   Older clastic
rocks of the  Rome Formation form  Pine Ridge,  to  the northwest,
with overlying  interbedded  clastic and carbonate  rocks  of the
Conasauga  Group outcropping  to  the  southeast,  with  the
Maynardville Limestone forming the floor of Bear Creek Valley.
Further to the  southeast,  Chestnut Ridge consists  of  the Knox
Group dolomites with a repetition of the sequence which lies to
the  northwest  cropping  out  to  the  southeast,  because  of
faulting.   Throughout  the  area,  beds dip  between 38  and 50
degrees southeast (McMaster, 1963).
                               199

-------
         EXPLANATION
      •C _ Conasauga Group

      OClr Knox Group

      ^C L, Chickamauga Group

      ^^^K Approximate Geologic Contact
            Figure 1.   Geologic Map of the Study Area
                       (after McMaster,  1963)

          Each carbonate sequence  is  karstified.  Dolines  occur
on the  Knox Group dolomites near  the  crests of the ridges,  and
springs emerge on  the  dip  slopes,  some of which sink  again into
swallets,  finally emerging  at  lower elevations.   Springs  also
emerge from along the  strike in valleys indented into  the  ridge.
The entire carbonate sequence is mantled  with  15 to  20 meters of
residual  soil   that   contains   numerous   chert   fragments.
Vegetation  on  the ridge consists  of  a  combination of  hardwoods
and pines with some deadfall and covering of leaf litter.

          Scarboro  Creek  is  a  perennial  stream  that   flows
southeastward  cutting  through Knox Group dolomites on  Chestnut
Ridge.   The  springs included in this  study  all emerge  near  the
channel  of Scarboro   Creek;  on Figure   1 they are  labelled  A
through  K and a  surface-water measuring point is  labelled X.
Although these are  not the only springs that have been  observed
to flow  near  Scarboro  Creek during  flood conditions, the  study
included  six  springs that  drain the  ridge  from the east,  A,  B,
C, D, F,  and H, and five  springs  drain the ridge from the  west,
E, G, I,  J,  and  K.  These  springs have discharges estimated to
be between 1 and 30 liters per second.  Springs  A, D,  and  H have
the  highest discharges and every  spring  except  Spring  F  is
perennial.
                               200

-------
                    TEMPERATURE MEASUREMENTS

         Each  temperature  measurement was made  with a digital
thermometer capable of measuring  to 0.1 degrees  Celsius.   The
accuracy of the thermometer  was checked with a mercury reference
thermometer every few weeks.  Spring temperatures were measured
generally every 2 or 3 days.   Each measurement  was always made
at the  same  location  in the  spring pool or flow.   Although a
coarser resolution has been obtained in this work when compared
to the  resolution that  would be  possible with  a  thermometer
measuring  continuously  or  to  0.01  degrees,  and  using  a data-
logger as  described by  Meiman et al. ,  (1988) a  large data set
for many different springs,  all of which drain the same aquifer
has been compiled.    This   allows  a comparison  of  temperature
variation  for  different springs  and a  relative  assessment of
likely flow geometries to be made.
                            RESULTS

          Table  1 shows  the  mean temperatures,  their standard
deviations,  and their  coefficients of  variation.   The table
represents data collected from January 1991 and  summarizes more
than 1,500 measurements.
r
Table 1. Sample
and
SPRING
A
C
D
E
H
J
B
I
G
K
F
BASED
^
Means, Sample Standard Deviations (SD),
Coefficients of Variation (CV)
MEAN
14.2
13.9
13.9
14.5
13.9
14.6
14.1
14.6
15.3
16.1
14.4
ON MORE THAN 1
SD
0.1
0.2
0.2
0.2
0.2
0.3
0.3
0.8
1.4
1.5
1.7
,500 DATA
CV
0.9
1.2
1.2
1.2
1.2
2.0
2.4
5.4
9.3
9.5
11.7
POINTS
          The  coefficients  of variation (CV = Std. deviation   x
 100/mean) show that the springs seem to fall naturally  into  two
 groups.   One  group  includes springs A, C, D, E,  H,  and J.   In
 this group the coefficients of variation of temperature  are less
 than 2%  and  in  most cases  close to  1%.   The mean temperatures
                              201

-------
range from  13.9  to 14.6 degrees  and  standard deviations  range
from 0.1 to 0.3.   It  is  interesting to note that Springs  C,  D,
and H have practically identical  statistics, are all  located  on
the east side  of Scarboro Creek,  but are  not located close  to  or
adjacent to each  other.   Spring  J  is  located  on the west  side
Scarboro Creek.   Although  the  eleven  springs seem to fall  into
two groups,  they are probably members of a  continuum  that  would
probably be more obvious if additional springs in the Oak  Ridge
area were to be sampled.

          Spring  A has the most constant temperature of all the
springs.    The  spring  outlet is a  large  pool  and  the   flow
discharges  through a  what appears to  be  a conduit partially
blocked  with  an alluvial  plug.   The spring outlet is located
beneath a large tree and flow  is  from  obliquely along  the strike
along  a  bedding  plane.    Only  in  the  absolutely  wettest
conditions has the flow from Spring  A become turbid.

          Spring C is  located  about 25 meters  down the valley
from Spring A.   This  spring emerges from along a bedding  plane
in  a  small  steephead  (steephead = a  spring-sapped,  steepened
area) near  the  spring  outlet.   Spring C has a mean temperature
of  13.9  degrees  and a standard deviation  of 0.2.   The flow  is
always clear,  even after the most  prolonged precipitation.

          Spring D is  located  about  250 meters downstream and  is
a  relatively  large spring.  The  perennial flow emerges from a
pool, the  alluviated  outlet,  but in wet  conditions  flow  also
emerges from a blow-hole nearby,  which itself then  discharges  to
the pool.   Spring  D shows  very little temperature variation and
becomes  slightly turbid  in  the  wettest conditions .    The
temperature of this spring responded in  a similar way  to Springs
A  and C  to  heavy, prolonged precipitation events,  and flow  at
the blowhole becomes slightly  turbid during floods.

          Spring E is  located about 40  meters  downstream  from
Spring D, emerging from a steephead on the  west bank of  Scarboro
Creek.   The variation  of temperature at  Spring E is also small,
a  standard  deviation of  0.2, but  the mean  temperature  is higher
at  14.5  degrees.   The  spring  water does  not  become turbid  in
even the wettest conditions.

          Spring H  is  located on  the  east  side of the creek,
about 150 meters  downstream.  Flow  emerges  from what  appears  to
be  the direction  of  strike,  from a small conduit.   This spring
has  a  constant temperature with  a  coefficient  of variation  of
1.2% and a mean temperature of 13.9 degrees.  The water  remained
clear  throughout  most of  the  year  and became  only  slightly
turbid during and after heavy  storms.
                               202

-------
         Spring  J  is located about  150  meters  downstream from
Spring H, on  the west side of  the channel.  Spring  J emerges
from a steephead very  close  to the  creek.   The coefficient of
variation of  this  spring is  2%  and the mean temperature 14 . 6
degrees.   The temperature at this spring became highly variable
during the summer  and  is thought to have been  affected by the
initial flow from an orifice near the channel of Scarboro creek
becoming plugged and thereafter  mixing  with flow from Scarboro
Creek, but  eventually flow from the  orifice resuming.

         Another   group  of   springs  with   more   variable
temperatures includes Springs B,  I,  G,  and  K.   Coefficients of
variation  are  greater  than  2%  but   less than   10%.    Mean
temperatures  of  this group ranges  from 14.1 to  16.2 degrees.
The standard deviations range  from 0.3 to 1.5 degrees.

         Spring  B  is located within a  few  meters  of  Spring C,
has  a very  small   discharge,  and  its  flow  is always  clear.
Spring  B  has a  coefficient  of  variation of  2.4  and  its
temperature variation  is quite dissimilar to nearby  Spring C.
Springs  B  and C  each emerge  from  steepheads  on  the  bank of
Scarboro Creek.

         Spring  I   is  located about 10 meters  downstream from
Spring J, also on  the  west  side of  the creek.   Flow  is from a
steephead along  the strike  at a  bedding  plane surface.   The
coefficient of variation  is  5.4%  and the mean temperature of the
spring  is  14.6  degrees.   Flow  from Spring  I  has never been
observed to become  turbid.

         Spring  G  is  on the west  side  of  the creek  about 60
meters downstream  of Spring F.   The flow  joins with the flow
from Scarboro Creek beneath  a  tree,  but  actually originates in a
collapsed  area  a  few  meters away.   Spring  G  has the second
highest mean temperature of all  the springs,  15.3  degrees, its
coefficient of variation  is  also high,  9.2 %.  The flow emerges
from a steephead  and was  clear throughout 1991.

         Spring  K  is a shallow pool  located on the west side of
the Scarboro Creek, about 20 meters  downstream of Spring H.  It
was not  originally  recognized as a spring but,  after observing
the pool for  several  days,  it was  realized  that it  could be a
spring as it was always clear and persisted as a pool even when
the tributary upstream of it  dried up.   The temperature of the
pool  was also  sometimes  similar to  the  springs  nearby and was
considerably different from the creek.   The mean temperature of
the pool is 16.0 degrees.  The coefficient of variation is high
at  9.9%.  The only flow  that  obviously enters  the  pool is from
the small tributary, in  flood conditions;   there is a  constant
amount of water  in  the pool  at  all times and when  the water in
                               203

-------
Scarboro Creek  is  turbid,  the pool  remains  clear.  During  the
wettest of conditions, overland flow occurs between  the  closest
section of Scarboro Creek and the  pool.

          Spring F  emerges from beneath  a tree about  50  meters
downstream of Spring E.   Spring F  has a mean  temperature  of 14.4
degrees, a sample  standard  deviation of 1.7, and  a  coefficient
of variation of 11.7%.  Water flows from beneath a  small  bedding
plane.  Flow ceases during dry conditions.

          Thermographs representing  measurements  at the  eleven
springs  from  January  1991  through early  December  1991  are
combined and shown  in Figures 2 and 3.

20—
0 1
o_^
fl,
3
"m 15-
0)
Q.
1 :
V

Spring J
SnrinaE ^J^Afi^tf ^ *
*^^ [^Spring D
Spring C
Spring H
JAN 1 FEB ' MAR ' APR 1 MAY 1 JUN ' JUL ' AUG ' SEP ' OCT ' NOV ' DEC
1991




     Figure 2.   Thermograph of Constant-Temperature  Springs
                           DISCUSSION

          The CVs of temperature in this study suggest  that  all
the springs are fed by a mostly diffuse-flow component.   This is
consistant with the CVs of specific  conductance at  four springs
A, D, E,  and H,  which also suggest that the springs are  fed by
mostly diffuse-flow  (CV less than 5%)  or  a mixture of conduit
and diffuse-flow  (5%
-------
        20-1
     U
     o>
     o
     Q.
     O
        15-
        10-
Spring G

 Spring I
     i
 Spring B
                                               Spring K
           JANlFEB I MAR I APR I MAY I JUN I JUL I AUG I SEP I OCT I NOV I DEC

                               1991
     Figure 3.   Thermograph of  Variable-Temperature Springs
  (Figures 2 and 3 are plotted with the same temperature scale)

However,  there are similarities and differences in the data  for
all the  springs.   The  similarities  are mostly  related to  the
lag-times,  and  the  differences  are  mostly  related  to   the
amplitude of  temperature  variation  through  the  year.   It is
important to point out that, from observations in the  study  and
the CVs of temperature,  there  is  no evidence  of  any relationship
(similarity or difference)  that is  related  to relative  locations
of  springs,  or  the  amount  of  their  flow,  or the  physical
appearance of the  spring outlets.

          The  temperature  lag-time,   when  it occurs,  provides
information about the thermal  diffusivity  of the system or  the
time required to displace the  resident thermal signature of  the
bedrock.   The thermal diffusivity of a substance can be  defined
as the thermal  conductivity  (cal/sec.  cm.  °C)  divided by  the
product of the heat  capacity (cal/g.  °C)  and  the  density  (g/cm3)
(Oke,   1978,  p.  39)  which  can be reduced  to and  expressed in
units  of  cm2/sec.    The  degree  of  thermal  diffusivity is
described as the depth of temperature penetration into the  rock
(or the system) of each pulse  of recharge  water temperature as
it  flows  through the  aquifer.   Another  way  of relating to
thermal diffusivity  would be to  recognize  that  it is  a  measure
of how  far each  molecule  of  recharge  water can  "carry"  its
temperature signature  into and  through  the   system.   A purely
conduit-flow system would  have a high  thermal  diffusivity  and
would  result in a short lag-time.  A purely  diffuse-flow system
                               205

-------
would have a low thermal diffusivity and would  result  in a long
lag-time.

          Wigley and Brown  (1976)  describe  a parameter  called
the relaxation length,  which is associated  with  the decay of,  or
damping, of outside air temperature to  an  asymptotic (constant)
value within a cave system.   The relaxation length is related to
the  size  of the passage  and the velocity  of the  fluid  moving
through it,  and its  roughness  factor.   More  simply put,  the
relaxation  length  is a  distance inside a cave  that  the  outside
air temperature effect can be  detected  (or  measured) .   Figure 4
shows  a representative thermograph  for a  variable-temperature
spring  (Spring  G) ,  and  a  curve representing the  daily  mean air
temperature for Oak Ridge in 1991  (NOAA, 1991) .   The daily mean
air  temperature curve  has  been smoothed by using a seven-point
moving-average  procedure  for  each  data point.   The thermograph
shows that maximum temperatures at  the  springs occurred about 50
days after  the  maximum daily mean  air  temperature for  Oak Ridge
 (NOAA,  1991) .   Fifty days is thought to be  more reasonable than
415  days  (50 + 365);   there  are  several  possible  temperature
oscillation cycles as explained by  Lange (1954) .

Temperature (°C)
-•• ro c
..?.,,.!. .,.?,.,,!, ...1


V

- •'"'/•" '*" '*
> ' ' " ^ - ~
*r" f • ^ Spring G
•
• *
,» ^^x-Mean Air Temperature ""
• ^ •
JAN 1 FEB 1 MAR ' APR I MAY ' JUN ' JUL 1 AUG I SEP ' OCT ' NOV ' DEC
1991
N




    Figure  4.  Thermograph for Spring G Compared with Mean Air
                 Temperature  for  Oak  Ridge,  1991

          From the relaxation length theory of Wigley and Brown,
 assuming  a  system  with   flow  pathways  of  roughly  equal
 dimensions,  a spring draining  a conduit-flow  system (a quick-
 flow  system)  would transmit a temperature  pulse  deeper through
                               206

-------
the  system,  possibly  to  the spring;  there  would  be a  long
relaxation length.   In  the opposite way, a  diffuse-flow system
(a slow-flow system)  would not transmit the  temperature pulse as
far,  or would  have a short  relaxation length; the  temperature
pulse would decay or be  damped within the system before reaching
the spring.

          As  previously  stated,  the  lag-times  for  all  the
springs are  essentially  the same  (with the  exception of  the
"noisy" data  for  Spring  J) ; as  Figure 5  shows,  the  constant
temperature  springs have lag-times  similar  to  the  variable-
temperature springs,  but,  the amplitude of temperature variation
is smaller.  Interpretation of various  conceptual  models of the
aquifer system suggests that  the differences  in the  data  could
be related to  the  systems being  either conduit flow  or diffuse
flow systems.
         16'
o
o

1
2
&
.4)
         13
                           Spring J
                                          Spring A
              Spring E
  Spring D
 Spring C
Spring H
         12~1 JAN I FEB I MAR' APR ' MAY ' JUN I JUL ' AUG ' SEP ' OCT I NOV I DEC

                                 1991
     Figure 5.   Thermagraph of Constant-Temperature Springs
         (Showing expansion,  for  clarity  of  temperature
                 scale  used in Figures  2  and 3)

          Relating thermal diffusivity to the  lag-times,  if the
systems are different  (i.e.,  conduit-flow or diffuse-flow) then
the thermal diffusivities of each system should be different, so
lag-times would be different  also.   The data do  not  fit  such a
model, the  lag-times are  similar,  so this hypothesis must be
rejected.   Also,  the  lack  of  significant turbidity in  the
springs after rain makes  it obvious that they are predominantly
diffuse-flow springs.
                               207

-------
         A  second hypothesis applies the  theory of relaxation
length.   If  it were valid  here,  then the  differences  in each
system could  be  explained  by  the different  ratios  of aquifer
length versus  relaxation  length.   With  such  a hypothesis, the
variable-temperature springs would be  closer to  the  recharge
area as  compared  to the constant-temperature  springs.   But as
previously stated,  thermal  diffusivity  is  a  time-and distance
related unit,  so  a  shorter  aquifer  length  would cause a change
in thermal diffusivity, and would probably result in different
lag-times  and  different  ratios  between   aquifer length and
relaxation length.  Again,  the  variation in the data cannot be
explained by this  hypothesis.

         A   third  hypothesis  involves  mixing  a  slow-flow
(diffuse-flow,  and at   a  constant temperature)  component with
different amounts  of  a quick-flow  (conduit-flow  -  variable-
temperature)  component.   This  mixture  would  always  include a
diffuse-flow  component with   a  variable  proportion  of the
conduit-flow component  present  within  each  system.  A mixture of
different proportions   of the  two  components could  produce a
certain temperature at  the output, and still maintain acceptable
thermal  diffusivity and relaxation  length conditions  for the
system.

         Multiple temperature measurements  at several springs
can  aid in   characterizing flow at  a spring,  or within  an
aquifer, but is  it  likely  that whatever  impression  is gained
about flow  at them, i.e.,  whether a  spring  or  an  aquifer is
dominated by mostly diffuse flow or mostly  conduit  flow, the
flow  is  probably  always a  mixture  of quick-flow and slow-flow
components.    The  response of a  spring or springs to rainfall or
temperature  change  is  only  the end-product of responses of the
whole system and  is only a  gross generalization of the sum of
those  responses;   they   include   individual  and  different
responses to  recharge,  storage, and flow.   It is highly  likely
that  no  system is  predominantly  conduit-flow or diffuse-flow.
Rather,  all systems are different  forms  of  mixed-flow systems.

          Temperature  variation should  be studied  at springs
because  it is  a physical parameter and would be expected to vary
somewhat differently   to   specific  conductance,  which  is  a
chemical parameter, but would provide equally useful  information
and would give insight  into aquifer mechanics.   Measurements of
specific conductance were  not  made  concurrently with the  1991
temperature  measurements and   in  retrospect  they  should have
been.  The temperature measurements  and  study of their  variation
should be supplemented  with concurrent  measurements  of  specific
conductivity and  study of  its variation.    Such  data are  now
being collected for all the springs to  allow  comparison between
the CV of temperature  with  that of  specific conductance.   These
                               208

-------
additional  data will be available  from  the  author in mid 1992.
However,  it  can be  seen  from the  (temperature) data described in
this paper that if  many  measurements are made at several springs
that  drain  the  same  or adjacent  geologic  unit(s)  and/or
aquifer(s),   such  information  can  yield  valuable  information
about the characteristics of that  aquifer.   Such  data that can
not be obtained from studying temperature variation at only one
spring, and  probably can not  be obtained  at  randomly located
wells, in part because of scale  differences between monitoring-
well  effectiveness  and  typical  karst aquifer  organization, as
suggested by Quinlan and Ewers  (1985)  and Ford  and Williams
(1989, p.  210-211).
                         ACKNOWLEDMENTS

          The  author  would like  to thank  Jim  Quinlan,  Ralph
Ewers, and Tom Aley for many useful  discussions  about the  study.
Particular thanks is given to Chris Smart for his  indispensable
help  with the  theoretical aspects of  interpretation  of the
results.  The drafting  skills  of  Brenda Altom are also greatly
appreciated.
                           REFERENCES

Aley, T.  J.,  1970.   Temperature  fluctuations  of a small Ozark
spring,  Caves and Karst,  12(4):25-30

Brown, M. C., 1972.   Karst hydrology  of the  lower Maligne basin,
Jasper,  Alberta,  Cave  Studies  13, Cave Research Association of
California.

Ford, D. C., and  P. W. Williams,  1989.  Karst  Geomorphology and
Hydrology, Unwin Hyman,  London,  601 p.

Lange,  A.,  1954.   Rock temperature  distribution underground,
Part II, Caves and Karst (7):  26-32.

McMaster. W., 1963.  Geologic map of  the Oak Ridge Reservation,
Tennessee.   Prepared  for Oak Ridge National Laboratory,  (ORNL-
TM-713)  .

Meiman,  J.  J.,   Ewers,   R.O,   and Quinlan,  J.F.  1988.
Investigation  of  flood pulse   movement   through a  maturely
karstified   aquifer  at  Mammoth  Cave  National  Park:  A new
approach, Environmental Problems in  Karst Terranes  and their
Solutions  Conference   (2nd,  Nashville  Tenn.)  Proceedings,
National Water Well Association,  Dublin, Ohio,  p.  227-263
                                209

-------
National Oceanic  and Atmospheric  Administration  (NOAA),  1991.
Local Climatological Data for  Oak  Ridge,  Tennessee

Oke, T. R.,  1978.   Boundary Layer  Climates,  Metheun,  London,  372
P-

Pitty, A. F., 1979.  Underground  contributions to  surface  flow,
as  estimated  by water temperature variability,  (in) Pitty, A.
F.,  (ed.)  Geographical  Approaches  to  Fluvial  Processes,  Geo
Abstracts,  Norwich,  p.  163-172

Pitty, A. F., 1976.   Water temperatures in the  limestone  areas
of  the  central  and Southern  Pennines.   Proceedings  of  the
Yorkshire Geological Society,  79:153-177

Quinlan, J.  F.,  and R.  0.  Ewers,  1985.   Ground water flow in
limestone  terranes :  Strategy,  rationale,  and procedure  for
reliable, efficient monitoring of ground-water quality  in  karst
areas.  National Symposium and Exposition on Aquifer  Restoration
and Ground Water Monitoring,  Proceedings,  197-234, Worthington,
Ohio, National Water Well Association

Quinlan, J.  F.,  Smart,  P. L.,  Schindel,  G.  M.,  Alexander, E.  C.,
Jr.,  Edwards,  A.  J.,  and Smith,  A.  R.,  1991.   Recommended
administrative/regulatory   definition  of  karst  aquifer,
principles   for  classification  of  carbonate  aquifers,   and
practical  evaluation  of  vulnerability  of  karst  aquifers.
Hydrology,  Ecology,  Monitoring, and  Management of Ground  water
in   Karst   Terranes  Conference   (3rd,   Nashville,   Tenn.)
Proceedings.  National  Ground Water Association,  Dublin,  Ohio.
[in this volume]

Roy,  B., and Benderitter,   Y.   1986.    Transferts  thermiques
naturels dans un systeme aquifere  carbonate fissure peu  profond,
Bulletin Societe geologique France,  (8)t. II, no. 4,  p.  661-666

Shuster, E.  T.,  and White, W.  B.  1971.   Seasonal  fluctuations in
the  chemistry  of  limestone springs:  a  possible means  of
characterizing carbonate  aquifers, Journal  of  Hydrology,  14:93-
128

Smart, C. C., and Ford,  D.  C., 1986.  Structure  and  function of
a  conduit  aquifer,  Canadian  Journal of Earth Sciences,  23(7):
919-929

Smart,  P.   L.,  and Hobbs,  S.L.   1986.    Characterization of
carbonate aquifers;  A conceptual  base.   Environmental  Problems
in Karst Terranes and Their Solutions Conference  (Bowling Green,
Ky.) Proceedings.  National Water Well Association,  Dublin,  Ohio.
p. 1-14
                               210

-------
Wigley,  T.  M. L., and Brown, S.L., 1976.   The  physics  of caves,
p. 329-358    (in) T. D. Ford and C. H. D.  Cullingford,  eds.,  The
Science  of  Speleology,  London,  Academic  Press.


                      BIOGRAPHICAL SKETCH

Gareth  J.  Davies  is a  hydrogeologist  and  quality  assurance
manager  with Geraghty &  Miller Inc.,  Oak Ridge, Tennessee.   He
received a  B.S. from  Millsaps College,  and  an M.S.  from  the
University  of  Southern  Mississippi.    He  is  a task group  and
sub-committee member of the  American Society for Testing  and
Materials and has  been interested in caves  and karst  research
for over 25 years,  with particular interest  in the  karst of  the
Upper Swansea Valley,  South Wales,  Great Britain.

Gareth J. Davies
Geraghty &  Miller Inc.,
97 Midway Lane,
Oak Ridge,  TN 37830
(615) 481-3000
                               211

-------
WATER TEMPERATURE  VARIATION AT SPRINGS  IN THE KNOX  GROUP NEAR
OAK RIDGE,  TENNESSEE

By Gareth J.  Davies


Do you think the time lags you observed are actual travel times?
If so, why?  If not,  why not?

The  lag times are  not  travel  times.  Rather,   they  provide
information  about  the  thermal diffusivity  of the system.   The
lag-time is an average  for the centroid  of  the  whole  basin that
drains to a particular spring.   A dye-trace provides information
about  travel  time  from  one  point  to  another,  and  is  not
necessarily  representative  of  the whole  basin.   If  a  lag-time
happens  to coincide with  a travel  time inferred  from a  dye-
trace, e.g.,  Pitty (1976,  1979)  the relationship is  just that,  a
coincidence.   Two  different processes are being monitored.
                               212

-------
     THE  EFFECT  OF PETROLEUM HYDROCARBONS ON THE SORPTION OF
         FLUORESCENT DYES BY ACTIVATED COCONUT CHARCOAL
                         Scott A.  Recker
              Delta Environmental  Consultants,  Inc.
                    Charlotte,  North Carolina

                        Michael J. Carey
              Delta Environmental  Consultants,  Inc.
                    Charlotte,  North Carolina

                            Joe Meiman
                      National Park Service
              Mammoth Cave National Park, Kentucky


                            ABSTRACT

The use of fluorescent dyes to delineate ground water flow-routes
and basin boundaries in karst aquifers has dramatically increased
in the  past  decade.   Dye  tracing techniques  have  been used in
conjunction with contaminant assessment investigations to determine
the extent of soil and ground water contamination caused by leaking
underground storage tanks.

Fluorescein and rhodamine  WT,  the two most common tracers,  are
passively collected  on activated  coconut charcoal  at  discharge
points or in wells  .   The presence of  petroleum  hydrocarbons in
ground water may retard the sorptive efficiency of the charcoal by
competing for carbon attachment sites or by reducing the actual dye
concentration due  to chemical alteration of the dye.

Solutions  were prepared  from ground  water  containing  known
concentrations of benzene, toluene, ethylbenzene and xylenes (BTEX)
and known concentrations of both fluorescein and rhodamine WT. Half
of the solutions were analyzed for  florescence intensity particular
to these  dyes, using a  scanning  spectrofluorophotometer.     The
second half of the solutions  were used  to immerse  a  sample of
activated  coconut  charcoal   (6-14  mesh),  mimicking  a  passive
detector.  After  a  contact  period of 48  hours, the  charcoal was
eluted with a mixture of 42% 1-propanol,  38%  ammonia hydroxide, and
20% distilled water for a period  of 24  hours.   The elutriant was
also analyzed for dye  concentration  intensity  characteristics of
the dyes using a scanning spectrofluorophotometer. Comparison of
the fluorescent intensity in the dye/hydrocarbon solutions and the
elutriants enabled  a determination  of sorption  retardation and
chemical  alteration caused  by the  presence  of the  petroleum
hydrocarbons.
                                213

-------
                           INTRODUCTION

Fluorescent dyes are commonly used tracing agents for hydrogeologic
investigations.  Heightened environmental awareness and  expanding
government  regulations have  increased the  need to  conduct  dye
tracing investigations in contaminated hydrologic systems.

Rhodamine WT  (RWT) and fluorescein,two of the most commonly used
dyes, are  often employed  in  conjunction with passive  activated
charcoal detectors.  Dye is introduced into a hydrologic  system at
an  input  point  and  is  collected  at discreet  discharge points
(springs)  or wells.  Historically, tracers have primarily been used
to  amass   information  concerning  the  flow  velocity  and basin
boundary   characteristics   of   free   flow   karst    aquifers.
Investigations  were rarely conducted in response to the loss of
organic contaminants.   The need to conduct tracer investigations in
the presence of petroleum hydrocarbon contamination has  increased
with the promulgation  of CFR 40, Part 280,  requiring hydrologic
investigations  in  response to releases  from underground storage
tanks (UST's).

Activated  charcoal is  useful in  tracer studies  as  it readily
adsorbs organic molecules.  However, the charcoal is  indifferent to
which organic  it will sorb in  the presence of multiple organic
compounds.  Rhodamine WT and fluorescein are organic  compounds that
adsorb readily to charcoal in a contaminant  free environment.  An
environment riddled with organic  contaminants may compromise the
sorption efficiency of the  charcoal  wi€h respect to  the dyes.  The
result  of  a tracer  study  conducted  in the presence  of organic
contaminants   may   be   inaccurate   because   the   charcoal  may
preferentially sorb petroleum hydrocarbons, rather than fluorescent
dyes.

The presence of organics, particularly petroleum hydrocarbons, may
also  chemically alter  RWT and  fluorescein  producing  erroneous
tracer  results versus  if  the organics were  not  present.   The
purpose of this  study  was  to  assess the sorption capabilities of
activated charcoal in the presence of  light petroleum hydrocarbons
(gasoline) and how hydrocarbons may possibly chemically  alter the
respective dyes.


                           METHODOLOGY

Representative  ground  water  and  liquid phase hydrocarbon (LPH)
samples were collected from a gasoline contamination site in  the
coastal plain of North Carolina.  Samples were immediately cooled
to  4°C  for shipment  to a  certified  laboratory.    Samples were
analyzed according to EPA SW-846 Method 602  modified for benzene,
toluene,  ethylbenzene and  total  xylenes (BTEX).    The  samples
contained varying concentrations of total BTEX as listed in Table
1.   Although  a sample was also collected from a monitoring well
which contained floating LPH,  its  total BTEX  concentration was  not
determined.
                                214

-------
Table 1.   Total BTEX concentrations (mg/L)

                   TOTAL  BTEX  CONCENTRATIONS

          BTEX Free                     0.0
          Low Dissolved                 3.17
          High Dissolved                9.95
          LPH                           132*

     *    Based on similar samples from monitoring wells containing
          LPH.

Representative ground water samples were split during collection.
One was analyzed by a laboratory and the other was used to prepare
the dye mixtures.   Dye solutions of 50,000, 1,000, 1.0 and 0.5 ug/1
were prepared from ground water samples and distilled water.  The
various dye  solutions were  split into two bottles.   Half of the
solution was transferred into a bottle and kept in total darkness,
at a constant temperature  of approximately 4° C.   The other half of
the solution was transferred into a bottle with a  packet containing
approximately 3.0  grams  of  activated charcoal and kept in total
darkness, at a constant temperature of 4°  C.  After a period of 48
hours  the charcoal  packets were  removed and  shipped with  the
dye/ground water solutions for dye analysis.

Groundwater/dye solutions were  analyzed  on a Shimadzu scanning
spectrofluorophotometer with respect to RWT and  fluorescein.  The
50,000 ug/L solution was not  analyzed  because  its  fluorescent
intensity exceeded the capacity of the instrument.  Results of the
analyses are shown in Table 2.

Table 2.   Concentration of rhodamine WT and fluorescein elutriant.

                    RHODAMINE SOLUTION

                               (Concentration total BTEX (mg/L)

Dye            Distilled      0.0       3.17      9.95      LPPH
Concentration  Water

500            0.988          0.898     0.946     0.892     0.916
  0.5          0.006          0.005     0.005     0.004     0.007
  0.25         0.003          0.002     0.003     0.003     0.004

                    FLUORESCEIN SOLUTION

Dye            Distilled      0.0       3.17      9.95      LPPH
Concentration  Water

500            0.772          0.313     0.063     0.073     0.081
  0.5          0              0         0.001     0.001     0.001
  0.25         0              0         0.001     0.001     0.001
                               215

-------
Activated  charcoal  packets were  placed  in  a mixture  of 42%  1-
propanol,  38%  ammonia hydroxide  and  20% distilled  water for  a
period of  24 hours.   The resultant solution, known as  elutriant,
was analyzed on the  Shimadzu scanning spectrofluorophotometer with
respect  to RWT  and fluorescein.   Results  of the  analyses  are
included in Table 3.

Table 3.   Concentration of rhodamine WT and fluorescein solutions.

                    RHODAMINE ELUTRIANT

                              (Concentration total BTEX (mg/L)

                              0.0       3.17     9.95        LPPH
Dye            Distilled
Concentration  Water
500
  0.5
  0.25
               1.455
               0.004
               0.004
0.892
0.008
0.003
1.599
0.005
0.004
Dye            Distilled
Concentration  Water
                    FLUORESCEIN ELUTRIANT

                              0.0       3.17
500
  0.5
  0.25
               5.91
               0.009
               0.006
4.141
0.008
0.007
6.353
0.007
0.005
1.036
0.006
0.003
                    9.95
5.389
0.009
0.006
1.85
0.01
0.004
                    LPPH
3.473
0.1
0.007
                      RESULTS AND DISCUSSION
The  500 ug/L dye  concentration used  in this  study  is a  higher
concentration than commonly occurs  at aquifer discharge  points.
Dye  concentration  of initial  solutions versus dye concentration
measurements of resultant solutions were plotted as bar graphs for
each  BTEX value.   Graphs  are presented  in  Figures 1  through  4.
Review  of  these  figures  indicate  that  at  low dye  solution
concentrations  (0.5 and  1.0 ug/L), resultant dye  concentration is
insignificant for  both the dye  solutions and  for the  elutriant.
Measured  dye concentration  appears  to  be  within the range  of
instrument   fluctuation.     Dye   solutions   at  these   chosen
concentrations  were  not  useful  in   determining the  chemical
alteration of the  dye or the  sorption  capability  of the charcoal.
Concentration from elutriant though  cannot be considered purely
quantitative.   Although  care was taken to  standardize  charcoal
volume,   no   two  dye   packets   will   have  the   same  sorptive
characteristics.
                                216

-------
50.8

5 0.7
K
£ 0.6-j

| 0.5-
c
u 0.4

$ 0.3-
Q
h 0.2-



t°:
K
                   0.5              1             1000
                    DYE CONCENTRATION OF SOLUTIONS (ug/L)
      Distilled Water
      High Dissolved
                          Clean Ground Water
                          LPH
Low Dissolved
Figure 1. Resultant dye concentration versus dye concentration in
          prepared RWT solutions.
                                   217

-------
      0.8-
    E °-5^
    z
    w
    UQ.4
    o
    U0.3-
    O 0.2-

    5 0.1-



    I   °'
    K
0.5              1             1000
 DYE CONCENTRATION OF SOLUTIONS (u&/L)
        Distilled Water

        High Dissolved
          Clean Ground Water

          LPH
Low Dissolved
Figure 2.  Resultant  dye   concentration  in  prepared  fluorescein
           solutions.
                                     218

-------
                   0.5             1             1000
                   DYE CONCENTRATION OF SOLUTIONS (ug/L)
      Distilled Water
      High Dissolved
Clean Ground Water
LPH
Low Dissolved
Figure 3. Dye  concentration of elutriant versus  RWT concentration
          of prepared solutions.
                                   219

-------
  1-1
    7n
    6-
    5-
  w
    4-
  r
  K2-
   W

                 0.5              1              1000
                   DYE CONCENTRATION OF SOLUTIONS (ug/L)
      Distilled Water
      High Dissolved
Clean Ground Water
LPH
Low Dissolved
Figure  4.  Dye  concentration  of  elutriant   versus   fluorescein
           concentration in prepared solutions.
                                   220

-------
RWT Solution

Dye concentration versus  BTEX concentration was also plotted  for
the 500 ug/L RWT ground water solutions.  The plot is shown by  the
bar graph which is Figure  5.  Figure 5 illustrates the relationship
of chemical  alteration of RWT with increasing concentrations  of
BTEX.  Rhodamine WT concentrations  for  these  solutions  range from
0.988 mg/L in distilled water to 0.892  mg/L in the 9.95 mg/L BTEX
ground water solution.  The deviation in dye concentration between
these samples is not significant indicating little to no  chemical
alteration of the dye with increasing concentration  of dissolved
BTEX.  The highest concentration of BTEX shows an intermediate  RWT
concentration, also indicating that chemical  alteration of RWT is
not a function of BTEX concentration.
                        0.0      3.17     9.95
                       BTEX CONCENTRATION (mg/1)
132
Figure 5.  Dye concentration versus BTEX concentration in 1000 ug/L
          RWT solution.
                                221

-------
RWT Elutriant

Results relating to the concentration of 1000 ug/L of each dye with
differing concentrations  of BTEX were  quite visible.   Figure 6
illustrates  a  bar graph   of  dye  concentration  versus  BTEX
concentrations  for the  1000  ug/L   elutriant.    This  figure
illustrates that  dye concentration appears  to  be independent of
BTEX  concentration.  The sorption capability   of  the  activated
charcoal, therefore, does not  appear  to have been affected by an
increase in BTEX concentration.  This  is intuitively contradictory
with the theoretical situation where charcoal sites  should be used
up by the presence  of additional  organics other than RWT.   The
highest  dye concentration  is  evidenced by  the elutriant  from
charcoal  in the highest BTEX concentration and  the lowest in the
"clean" ground water sample.
      0
                        0.0      3.17     9.95
                        BTEX CONCENTRATION (mg/L)
132
          7
 Figure  6. Dye concentration versus BTEX concentration in 1000 ug/L
          RWT elutriant.
                                 222

-------
Fluorescein Solutions

Figure 7 is a plot of dye concentration versus BTEX concentrations
for  the 1000  ug/L  fluorescein  dye/ground  water solutions.   A
decrease in  dye concentration is  noted between  the solution of
distilled water and  dye versus the solutions of ground water and
dye.  The difference  in dye concentration may be  an indication that
naturally occurring organics in ground water may act upon the dye
and reduce its concentration through chemical alteration.

Dye concentration ranges widely with a maximum of 0.772  mg/L  in the
distilled water/dye solution to a minimum of 0.063 in the 3.17 mg/L
BTEX ground water/dye solution.  Notice the marked decrease  in dye
concentration in the  presence of detectable concentrations of BTEX.
This  drastic  decrease signifies that  the presence  of petroleum
hydrocarbons  may  have a   significant  effect  on  the  chemical
alteration of  fluorescein  in ground water.   In  other words the
concentration of fluorescein may be  decreased as a  result of the
chemical breakdown due to the  presence  of elevated  levels  of
petroleum hydrocarbons in the ground water.
                       0.0      3.17      9.95
                      BTEX CONCENTRATION (mg/L)
132
Figure 7-  Dye concentration versus BTEX concentration in 1000 ug/L
          fluorescein solution.
                                223

-------
Fluorescein Elutriant

Figure  8  shows  the  plot  of  dye  concentration  versus  BTEX
concentration for the elutriant  from charcoal soaked in the 1000
ug/L dye/ground water solutions.  Values for dye concentration in
elutriant range from a maximum of 6.535 mg/L in the 3.17 mg/L total
BTEX solution  to  a minimum  of  3.473 mg/L in  elutriant from the
solution  containing LPH.   Review of  Figure  7 also  indicates a
decrease in dye concentration of 2.880 mg/L from the highest to the
lowest  BTEX  solution  concentration.    This  decrease   in  dye
concentration with an increase  in BTEX concentration suggests that
the  efficiency of  charcoal  decreased with  an increase  in BTEX
concentration.  This is  contradictory to results  from RWT.  Each
dye, however, has a different molecular composition and structure
and may,  therefore, be sorped  to the charcoal at different rates
and/or different efficiencies in the presence of multiple organics.
In  other  words the  charcoal  may  "prefer"  RWT  over  petroleum
hydrocarbons.  Fluorescein, conversely, may take a "back seat" to
the  hydrocarbons   and   be  sorped   less   efficiently   had  the
hydrocarbons not been present.
                       0.0       3.17      9.95
                       BTEX CONCENTRATION (mg/L)
132
 Figure  8. Dye concentration versus BTEX concentration in 1000 ug/L
          fluorescein elutriant.
                                224

-------
                           CONCLUSIONS

Rhodamine WT

The 1000 ug/L RWT dye/groundwater solution shows little fluctuation
in dye concentration.  The presence of  increasing concentrations of
petroleum hydrocarbons  does not  appear  to affect  or chemically
alter RWT.  Elutriant from charcoal  immersed in the 1000 ug/L RWT
solutions shows no distinctive dye concentration pattern.   It is
unlikely,  therefore, that the  sorptive effectiveness of charcoal
has been compromised by the presence  of the petroleum hydrocarbons.
Even in  the  presence  of hydrocarbons  it appears that  carbon
effectively sorbs RWT.  Charcoal  sorption  sites  may "prefer" RWT
over constituent compounds of the petroleum hydrocarbons.

Use of  RWT as a ground water tracer  in an environment affected by
petroleum hydrocarbons is  recommended.  There appears to be little
if any chemical alteration of  the dye and  charcoal readily sorbs
the dye in the presence of hydrocarbons.

Fluorescein

Elutriant  from  charcoal  immersed in the  1000 ug/L fluorescein
solutions exhibited  a  distinctive decrease in dye concentration
with increasing  BTEX concentrations.    This  observation  may be
significant in evaluating the sorption capabilities of the charcoal
with respect to this dye; however, the\apparent chemical alteration
of the dye  must  first be considered.  " The  1000  ug/L fluorescein
dye/groundwater solutions  displayed a drastic reduction  in dye
concentration with  increasing  concentrations  of BTEX.   Based on
this information the dye seems to be  drastically compromised in the
presence of even low concentrations  of petroleum hydrocarbons.

Use  of  fluorescein  as a ground water tracer  in  environments
affected by petroleum hydrocarbons  is not  recommended.   Chemical
alteration of the  dye may result in a negative trace that may  have
been a positive  one had the contaminant not  been  present.   The
contaminant, therefore,  may affect an  area of the aquifer that was
previously traced  as "safe" using fluorescein.  A dye concentration
of 1000 ug/L  is perhaps a higher concentration  than is commonly
recovered at collection points.   Perhaps at lower concentrations
more common to real situations, petroleum hydrocarbons may have a
more detrimental alteration effect on the dyes.

                          FUTURE STUDY

As with any decent real world scientific  investigation, this study
seems to  raise more questions  than it has  answered.    Future
research using dyes  in  contaminated environments is necessary to
promote accurate,  dependable,  and  reproducible  tracer  tests to
ensure  protection of the environment  and human health.   Here are
several suggestions for further studies:
                                225

-------
A similar study should  be  conducted using a wider range and
number of dye concentrations.  A range from 100 ug/1 to 1000
ug/1 in increments of 100 ug/1 is recommended;

Studies should  be conducted using  a wide  range  of organic
contaminants such as heavier hydrocarbons (diesel fuel, fuel
oil, jet  fuel,  kerosene,  waste  oil) organic  solvents,  and
organic metals complexes.  The  fluorescent wavelengths of the
contaminants should be assessed prior to the study; and

An actual  ground water trace should be conducted in an aguifer
affected by  a  contaminant  that was  previously traced while
unaffected to assess the differences in dye interaction with
contaminants at various  stage.  Flow routes should be retraced
and confirmed using several different dye types.
                           226

-------
                        ACKNOWLEDGEMENTS

The authors would like to extend their thanks to Paula Recker  for
providing the clerical support to make this publication a reality.
We would also like to thank Hiram Rutledge who took time out of  his
already  overloaded  schedule  to collect  duplicate  ground water
samples necessary to perform this research.


                      BIOGRAPHICAL SKETCHES

Scott A.  Recker,  a native  of Cincinnati, completed a Bachelors
Degree  in geology at  the University of  Cincinnati  in  1985  and
received a Master of Science in hydrogeology  from Eastern Kentucky
University  in  1990.     Scott is   a   member  of  the  National
Speleological Society and his research centers around ground water
flow in the subcutaneous zone of karst aquifers.  He  is currently
residing  in  Charlotte,   North Carolina  and  is  employed  as a
consulting hydrogeologist for Delta Environmental Consultants, Inc.

Scott Recker
3300 Circles End Road
Charlotte, North Carolina  28226
(704) 541-8191

Michael   J.   Carey     is  a   Project   Hydrogeologist  at  Delta
Environmental Consultants, Inc.,  in Charlotte, North Carolina.  He
earned  his B.S.  in Geology  (1985)  at the  University  of Rhode
Island,  and  his M.S.  in  Geology  (1990)   at  Eastern  Kentucky
University.   Most  of his  experience  is related  to subsurface
assessment   and   remediation  of  petroleum  hydrocarbons.   His
specialties include the investigation of karst aquifers utilizing
dye tracing techniques.

Michael J. Carey
8107 Tifton Road
Charlotte, NC 28226
(704) 542-6776

Joe Meiman, a Kentuckian, is currently the hydrologist at  Mammoth
Cave  National Park.   His  B.S.  in  Geology  (1985)   and  M.S.  in
Geology/Hydrogeology   (1989)  were  earned   at  Eastern  Kentucky
University.  At the present, he is kept busy directing hydrologic
research at the park which includes:  water quality monitoring,  dye
tracing,  three-dimensional  schematic  karst  aquifer  modeling,
research and development of monitoring equipment, and the  general
quest of scientific knowledge.

Joe Meiman
P.I 0. Box 95
Mammoth Cave National Park, Kentucky 42259
(512) 758-2339
                                227

-------
     THE EFFECT OF PETROLEUM HYDROCARBONS ON THE SORPTION OF
         FLUORESCENT DYES BY ACTIVATED COCONUT  CHARCOAL

                         Scott A. Recker
              Delta Environmental Consultants, Inc.
                    Charlotte,  North Carolina

                        Michael  J.  Carey
              Delta Environmental Consultants, Inc.
                    Charlotte,  North Carolina

                            Joe Meiman
                      National Park Service
                     Mammoth Cave,  Kentucky


1.   You exposed the charcoal to dye and BTEX simultaneously.
     Would int not be more relevant to expose the charcoal to the
     BTEX for several days and then expose it to the dye?

Yes, we agree that BTEX exposure may be the preferred method as
it would better simulate a real world situation.  During every
dye trace it is necessary to place the detectors into
contaminated ground water prior to the dye drop.  The exposure of
the detectors to BTEX during the dye's travel time is certainly a
factor which should be considered.

2.   The fluorescence of fluorescein is widely known to be
     greatly suppressed at pH's below about 4.5 and maximized at
     pH's above 9.0.  These reactions are reversible, with no
     destruction of the dye.  What data,  if any, do you have on
     the pH of your mixtures of BTEX in water?  Without such data
     is it possible that at least some of your apparent loss of
     fluorescein is due to lowered pH?  Also, why not dilute
     samples with high concentrations of dye — as is standard
     analytical procedure, with high concentrations of any
     substance with any analytical technique?

Monthly pH measurements have been recorded in ground water
samples from this location at monthly intervals for a period of
three years.  The range of pH is small and generally remains
between 7.6 and 8.0.  It is unlikely that the prepared solutions
exhibited pH below this expected range.

Yes, the higher concentration samples should have been diluted to
allow for quantification.   This information may have provided
more concrete results on the reaction of the dyes in these
environments.
3.    I am distressed by the sloppiness of the design of the
     experiments described.  so many variables are unaccounted
     for that it is highly questionable whether there is any
     significance or reliability in any of your conclusions.  Why
                                228

-------
     did you present this paper NOW,  rather than after a properly
     designed series of tests?  By presenting possible erroneous
     results,  you may inspire someone to conduct the testes with
     adequate rigor, but you may also mislead both investigators
     and administrators to perpetuate your possibly false
     conclusions and innocently make  wrong decisions concerning
     spill investigations — how to conduct them and how to
     interpret them.  Rigorous experimental design requires more
     thought than money or time.

Any and all constructive criticism is welcome,  however,  this
question does not appear to fall into the constructive category.
This paper was submitted at this time because of our strong
belief that these questions need to be addressed in the dye
tracing community.   We have simply touched on the fundamental
issues using the resources we have available.  Scientific study
is evolutionary and many workers fail to present experimentation
which yield anything but presupposed  results.  We felt it
necessary to break this mold and present our information at this
time.  We would hope that our work would not mislead any workers
in that our conclusions are tentative and we have simply raised
questions for other workers to consider.
                               229

-------
230

-------
                 CONTINUOUS-FLOW FLUOROMETRY IN

                       GROUNDWATER TRACING

                           C.C.  Smart
                     Department  of Geography
                  University of  Western Ontario
                    London,  Ontario,  N6A 5C2
                             Canada

                         S-E. Lauritzen
                      Department of Geology
                      University of Bergen
                     Allegt 41, 5000 Bergen
                             Norway
ABSTRACT
     In dye tracing, continuous flow fluorometry limits contamina-
tion, significantly  improves  precision  and  accuracy  and  allows
immediate response to tracer  appearance.  However,  it is  logis-
tically more complex than simpler methods, and provides data rather
than actual water samples which may be critically important should
corrective  processing  be  required.     Disadvantages  outweigh
advantages for most tracer applications.  However,  an attempt to
investigate the detailed evolution of tracer dye through a defined
water-filled conduit lent itself to continuous flow fluorometry and
led to some developments  of the method.

     The Jordtulla system in Svartisen,  northern Norway drains a
27 km2 subarctic catchment through a 580 m largely phreatic conduit
averaging 5 m  in diameter.  The system drains directly from a large
lake, has clear well-buffered  water  and has been mapped by cave
divers.   It thus  provides  an ideal  site in which to determine the
relationship   between  conduit  hydraulics,  morphology  and  tracer
behaviour. Sampling occurred at the resurgence  (Fiskepole), and at
a window  (Mellomgrotta)  part  way through  the  system.   Portable
alternators  and automobile  batteries powered pumps  and standard
Turner Designs Model 10 Series Fluorometers.   Water temperature,
fluorescence   and  stage  were  recorded  on  Campbell CR21X  Micro-
loggers.  Thirty three traces were made  over 9  days,  providing a
total of 50 breakthrough  curves.

     Fluorometer  response was  inadequate on the unusually steep
                             231

-------
rising edge of the breakthrough curve,  demanding a complex logging
algorithm.   The  resulting  data  required substantial  processing
before  satisfactory  breakthrough  curves   could  be  defined.
Fiskepole breakthrough curves are consistent and reproducible,  but
the Mellomgrotta  data showed  a  distinctive structure  indicating
incomplete mixing.  Subtle detail in individual  curves  appears to
reflect the style of tracer injection.

     The  heavy power demand  of  pump  and  fluorometer  limit  the
portability of the present  system.    Sophisticated data  loggers
permit excellent control and measurement capability, but add yet a
further level of logistic complexity.   Continuous flow fluorometry
remains justified only by stringent experimental or legal  demands.


INTRODUCTION: SAMPLING STRATEGY AND PROJECT  DESIGN

     Strategies  adopted  in  groundwater  tracing  are  usually a
compromise between objectives and resources available.   The tracer
preferred by  most workers in karst terrain  is slug injected  dye
because it meets the fundamental requirements of safety,  availabil-
ity (including legality in most cases),  cost and detectability.  A
range of  sampling strategies  are available  for determining  dye
return  characteristics.   These  range  in  order  of  increasing
technology and information content from visual observation,  through
charcoal  detectors,  discrete  water sampling and continuous flow
fluorometry.   The  analytical system  similarly ranges from visual
detection, through filter fluorometers to spectrofluorometers.

     As a general  rule,  the more  advanced technology can  deliver
more precise and accurate data, but only at greater logistical  and
financial cost.    The  fundamental question to  be posed  at  the
inception of  any  tracer experiment is  the  information required.
Resource  availability  is a secondary  question,  although it may
prevent investigation of some questions.

     The  resources  available  should not  dictate the methodology
adopted.  On  the  one hand this may result  in data inadequate to
answer the question being posed.  On the other hand, the data may
be obtained at excessive expense.  In addition,  an  over-technical
trace may be  less reliable,    because  more  complex tests  present
more opportunity for error.

     The majority of important tracer tests will be simple point-
to-point detector  traces  with  the hydrogeographical objective of
determining tracer trajectories.   More  sophisticated traces demand
prior knowledge of the hydrogeography derived from  point-to-point
tracing or well-judged hydrogeological reconnaissance.

     However, there are  legal and  scientific  situations  where very
high quality  data may be required.     First,  as hydrogeological
litigation evolves, demonstrably  higher quality data will  become
more  important.    This   will  provide  a means  of  assuaging  the
increasingly  expensive  confrontation,  scrutiny and derogation of
                             232

-------
"expert" witness.   The  technology  required,  along  with simple
quality-assurance-quality-control requirements,  may make ground-
water tracing considerably more  expensive.   However,  it  seems more
rational to  expend resources  on a  high  calibre trace  than in
vacuous litigation.

     Scientific work may also demand high  quality data either to
verify  the  technique  or  the system under  scrutiny.    In  field
experiments,   the  difficulty of  exerting effective   scientific
control makes it critically important to minimise any error masking
the experimental  variance.  The remainder of this paper is a report
on the design and implementation of a continuous-flow fluorometric
system for field studies in  remote areas.


CONTINUOUS-FLOW FLUOROMETRY

     Modern field fluorometers are designed to allow water to pass
directly through the analytical cuvette, permitting a  continuous
reading of  water fluorescence.   Normally,  this water is  drawn
directly  from a  spring  or  stream  and  a  continuous  trace  of
fluorescence  is  printed  on  an  analogue  recorder  (e.g.  Turner
Designs 1978). There are problems and advantages  inherent in this
system.  These  will be addressed with special  reference  to  the
Turner Designs Model 10 Series Fluorometer which is the most common
and,  despite  its   age,  arguably the  most  suitable   instrument
currently available for  field fluorometry-

     Calibration of the continuous flow cell requires rather large
quantities of dye standard,  perhaps  a  litre  or  more.   Usually, a
short hose length is used to fill the  flow cell from the bottom.
Several rinsings  may be required until a stable reading is obtained
on  successive  flushes.   With high  temperature  coefficient  dyes
(e.g. the "Rhodamines")  it is essential to  record the temperature
of the calibration  standard.

     In remote areas without  A.C.  line voltage,  power consumption
is  a  serious  consideration, which  can   significantly increase
logistical load.     The Model  10 draws some 2 amps  of power  when
operating from a  12 volt source,  and  thus requires a rechargeable
battery source (usually one  or  more  automobile  batteries) and a
portable alternator with fuel (usually gasoline).  Such  instrumen-
tal energy consumption is appalling by contemporary standards,  but
is  dictated  by the "chopper"  and high voltage  photomultiplier
technology which gives  the  instrument its  impressive  analytical
capability.

     Instrument power  demands  are  often  overshadowed  where  it
becomes necessary to pump water up  from a stream at a suitably
rapid rate.   A heavy duty 12 v marine "bilge" pump will  often draw
3-5 amps.   A siphon can supply water at  a  suitable rate for  zero
power, but such systems are inherently  unreliable and their  flow
rate is strongly dependent on stream stage.  A  combination  pump-
assisted siphon is  very efficient, allowing  (once primed)  a  much
                             233

-------
lower power pump, providing a priming system is  incorporated,   it
is generally suggested  that  pumps should be placed  downstream of
the fluorometer  to  avoid over-pressurisation.   However,  we  have
found that in small diameter pipe  (e.g.  V  garden  hose),  this  can
result in degassing of water.  Instrument noise  increases and  dye
detectability suffers greatly from the stream of  bubbles produced.

     The lightest fluorometer-alternator package with a lightweight
motorcycle battery weighs at least 30 kg.   The system may operate
for only an  hour or so without recharging.  It is  imperative  in
these circumstances  to have prior knowledge  of likely travel times.
Where  A.C.  line  power is  available,  the   fluorometer will  run
indefinitely.  Line power instability may cause some interference,
e.g. low pressure mercury lamps (Alexander  pers. comm.  1991).   it
may be worthwhile to  run the fluorometer from a 12  volt  battery
continuously charged from an A.C. battery charger.   This  isolates
the fluorometer from line noise and improves precision.  Fouling of
the analytical  cuvette  and calibration  drift become the  primary
long-term problems.   These can be  corrected  on an itinerant basis.
Although it is possible to collect an excessive  quantity  of data,
over  long duration  tests,  or  those  requiring  high  frequency
sampling, continuous flow fluorometry may be much less expensive
and less  prone to error  than equivalent discrete sampling pro-
cedures .
RECORDING

     The simplest recording system is an analogue  chart  recorder,
matched to the instrument telemetry.  A chart record provides real-
time  "hard-copy"  of a  trace.   This  is valuable  as a  statutory
record.  It also allows  an immediate impression  of trace progress
and is invaluable in managing sampling  strategy,  e.g. chart speed,
regional  sampling  patterns,  sampling  frequency  and   sampling
termination.   However,   the  chart record includes  all instrument
noise, can be mis-matched to the fluorometer and is often unreli-
able.

     Modern digital data loggers are less expensive than analogue
chart recorders and considerably more reliable and  powerful. They
are no more technically demanding than chart recorders,  although
they  do  not  produce  any real-time  hard copy;  their memory and
recording media can  be  volatile.   Work with both  Campbell Scien-
tific and Aanderaa data loggers suggests that with  suitable memory
back-up facilities,  this is a mature  technology with any modern
field logger  system.

     The single greatest problem with continuous fluorometry occurs
where  the  direct  fluorescence  reading  is  dependent  on other
variables  than tracer concentration.   For example,  cross-fluor-
escence from other dyes,  background variations, temperature, pH and
turbidity  controls  on  fluorescence.    These  problems  normally
require  a discrete  sample  so that some  sample  preparation or
further  analysis  can  be undertaken  before  fluorescence  can be
                             234

-------
determined.     Many  data  loggers also  permit  data processing,
logical  process control  and  data management.  It  may  thus  be
possible to correct readings to  obtain a direct record of tracer
concentration.  For example, it is a simple matter to  record sample
temperature and correct to standard temperature.  In general,  it is
advisable to  obtain required  data on  site,  but  to avoid on-site
processing  because  errors  become  more  probable  as  processing
becomes more complex.   In the worst case, this may invalidate the
entire record.   Providing  sufficient  memory  is  available,  it is
better to  record  essential  parameters such  as temperature,  mass
injected etc.  and to  process the data  after downloading  to  a
suitable computer system or spreadsheet.

     A great  advantage  to logger  processing is the capacity to
smooth the data by averaging.   Measurements can be made with very
high frequency (perhaps every 0.1 s) and the average and standard
deviation recorded (perhaps every 5 minutes).   Each recorded value
in this example is thus the average of 3000 readings.  The signal
noise  which  obfuscates  analogue  chart  traces  is  removed  and,
providing  there  is  no  significant   trend   over  the  recording
interval,  a very smooth (precise)  record  is  generated.   This
facility is of such great value at low tracer concentrations that
it is now being applied to the analysis of discrete samples.

     Loggers can also be programmed to respond to  measured data.
Typically,  this will involve higher  frequency measurement during
rapid evolution of tracer concentration and lower sampling rates in
the tail of a breakthrough curve.

     Continuous flow fluorometry can  generate a very  large  data
base compared to that possible with discrete sample  analysis.  This
is advantageous, unless  there is  a  need for collection of physical
samples.  Such a need may arise for reasons of legal evidence, or
more commonly because some form of sample processing is necessary
to determine  correct dye concentrations.   Background,  turbidity,
pH,  temperature  and  cross-fluorescence  are examples  of  factors
which may require some pre-analytical  processing.

     The Model 10 combined with a data logger can provide excellent
field  data in  a typical  trace.   However,  the  Turner  Designs
fluorometer is unable to automatically switch over its full dynamic
range, a major constraint.   It  is also  possible  to use  only  a
single filter combination with each  fluorometer which limits its
value in multi-tracing.   Fountain  (1989 pers comm.) has developed
a  logger-controlled  hydraulic  switching  system which  can  route
water  from a number of  intakes  into  a  single fluorometer.   In
applications with very short rise times, the instrument is unable
to respond  sufficiently quickly to rapid increases in fluorescence.
The need to switch through all ranges on rising fluorescence while
only a single range change is necessary on falling fluorescence is
a fundamental design flaw,  albeit one  of limited impact.

     Logger  data  collected  from  continuous-flow  fluorometers
provides the best data currently available on tracer breakthrough


                             235

-------
curves.  In sites with high background or complex traces requiring
careful  interpretation,   spectrofluorometric  analysis  of  grab
samples is advisable.  Data acquired  from very rapid sampling and
very rapid  changes in  fluorescence can  be extremely  complex  to
process.  It has proven worthwhile  to  record at two levels during
critical  rapid  changes;  the  first  is  the  typical  averaging
approach, the other is a direct recording of readings which can be
invaluable in  interpreting the reliability  of  the  averaged data.
A visual  interactive editing  programme has been found  the  most
useful way to process fluorometric  data.


FIELD EXPERIMENTS

     Earlier work  on  continuous flow fluorometry in the Canadian
Karst  systems  at  Castleguard  and Maligne  are reported  in Smart
(a,b).   These  and other  (multi-tracer)  experiments  showed clear
advantages in continuous  flow fluorometry, especially for revealing
precise  detail  in the  breakthrough  curve  and  for  eradicating
contamination.

     More recent work on has taken place on the Jordtulla conduit,
Svartisen, Norway  (Lauritzen et al. 1985),  an exceptional natural
laboratory  for  experiments  on conduit flow.    It  is  a  5 m  in
diameter, 580 m  long, singular conduit, which has been completely
explored  and mapped by  divers and  (draining an extensive lake)
varies gradually in discharge  between  1 and 50 m3s~1.   A series  of
33 tracer experiments were made in May 1989,  a period of transition
from winter base flow to  spring floods  (Figure 1).   Injections  of
-20 g Rhodamine WT (Acid Red 338)  were made at the sink point or  at
one of two  karst windows, Mellomgrotta and  Waltergrotta.   Fluor-
escence was  measured  continuously on two Turner  Designs  Model  10
Series  Fluorometers  powered  from  12  v  automobile  batteries.
Samples  were  collected  at  Fiskepole,  the  resurgence  and  at
Mellomgrotta a karst window 150  m  from the sink point.   Fluor-
escence  and sample  temperature  data  were  recorded on  Campbell
Scientific CR21X Microloggers  as one minute averages or 60 readings
and as 10-second  direct  readings during  periods of high  fluor-
escence.  The data were segregated into discrete traces, corrected
to a  standard  injected mass and  standard temperature  and  edited
using  an interactive sorting routine.

     Figure  2  shows an example of tracer  breakthrough curves  from
Mellomgrotta and Fiskepole.  Note the  extremely rapid rise of the
Mellomgrotta curve and clear instability compared to the Fiskepole
trace.  When inspected in detail (Figure 3), the Mellomgrotta trace
shows  a rapid  decline in  fluorescence  background prior to arrival
of the dye.  The latter problem results from  desorption of dye from
the vinyl  hose used to  feed the  fluorometer.  It  is significant
because of the high dye  concentrations  involved and the cessation
of pumping  between traces.  Adequate  flushing time  is essential
before dye arrival.  The  impact of such adsorption can be minimised
by continuous  flushing  of the shortest possible  length of intake
hose.   Provided  each  subsequent  dye  impulse  is  comparable  in
                             236

-------
hO

U)

-J
      n (D rt
      o cn y
      3 o ro
      a ro
      f- 3 o
 ro
p-3 MI
N H"
D) ro

n o
           3-

           PI
      rt ^-^ PJ
      « 3 cn 3
        n 3
        rt •
     C
     O

5'« 2 ro


8g?"
•§8.eS
* * £ o
5.0)   ®
rr ^1
CD *" El D)
a [J ro H-
o H ^ w
n a ro x1

g|8-8
ro IT ro o
3 ^ cn M

r+rt  <">

Q, ^ H"*

•  «  cj:
        !-•• Hi M
        3 M ro
         C rt
        rt O
        3- f-l O
        ro i M)
                  JQ

                   Q.

                   Q.
             C
             O

            o


             0)


            Q
                      30~
                20-
                      10-
                 0
                                                                                             -4
                        138
                                 140
                                                142           144

                                             Julian  Date
                                                                                               12
                                                                                       -0
                                                                                              cn
                                                                                              L_
                                                                                              o
                                                                                              JZ
                                                                                              o
                                                                                              cn

                                                                                              Q
146

-------
   o n n>
   cr c:
to

LO

00
fl>

3
O

cn o
(D n
     o
aw g
H- J
'"'• 3 O
W ^Q h-h
t->
g 
1 — i
i~b i-j
8-S*
t?g s
H" 0) (B
3 ^ M
(D i u
\V ^-1
a o o
rt ^ iM
3- < n
t» ro o
3 ft
cn rt
rt 3* CD
3-0
(D £ CT
tn >-<
o> o DJ
rt O X
rt 3 rt
(D W 3*
• a o
(D C
I"X »/O
1 3"
Rn
DU
'S"
CL

v 	 x^O -

C
o
•43 30 H
^•T
D
L_
-4-»
C
(D 20-
o
C
O
0 10-

CD
^^N
Q
0



1 1 1
FP17 & MG17


*• • \

. •
•
;
.


.
•
/^~\

•
\
N»
^ •' '.
* '.
• '^



' I 'I ' I I |
                0
                         1            2           3

                      Time from  Injection  (h)
4

-------
K3
OJ
      rt o a to
      3* 0) fD
      (o n o ,.
        I-1- M
      Ifl M H- *
      ft 01 3
      ••< rt n> a
      (D H-
      OJ O
      0) H-
.rt
 0)
 H-
 M
 (D
 a
       Q.
         n
          H)
      o>
         a> 5*
         o- 1-
        .^< o
         (o g

       C rt
       3 O Qj
       rt C o
          rt
       H- 01 tJ1
       3 rfp fD
                         0.0            0.5            1.0           1.5
                            Time  from   Injection   (h)

-------
magnitude  to  its  predecessor,  there should  be  no  significant
impact.  However, it may be possible to mimic subtle variations in
the detailed form of the breakthrough curve by adsorptive exchange.

     The gross instability of the Mellomgrotta  curve was reproduc-
ible in frequency and amplitude, but not phase on  all  traces,   it
is  considered to be  an indication  of  incomplete  mixing in  the
system.   It  is thus a measure of the scale of turbulence in  the
stream.   Such  irregularity in a  continuous record can  be used to
indicate  inadequate mixing length in a tracer study.

     The  Fiskepole breakthrough curves were in general  well-mixed
and showed little deviation from  the anticipated  form.   Apart from
discharge, the single largest source of variation  in form results
from  the  injection  style.    Although   injection  technique   was
reasonably standardised, the  exact  pattern of  secondary flows  at
the injection site varied somewhat with discharge. A dye-trapping
eddy at the  sink thus resulted in slight shoulders on  the tracer
recession under higher flows.

     Our  technical   conclusions  from  this experiment  are  that
despite the excellent technical procedures,  there remain practical
limitations  to  the  precision of  field fluorometry,  although  the
error resulting from  imprecision is far from significant relative
to overall variance.   However, changes in  the breakthrough curve in
response to discharge are very subtle over  the range of  discharges
investigated.    For  this  system,  therefore,  extremely precise
procedures  are required.   Smaller  diameter conduits  or greater
changes in discharge  are necessary  for more satisfactory experi-
ments, but of course  this may prevent complete exploration.

     Further  work  will  investigate the  relationship  of tracer
dispersion and advection relative to injection site and  discharge.
Preliminary   studies  indicate   that the  limited  free-surface
associated with the karst windows have a detectable effect on  the
breakthrough curve.   Models developed for a pure conduit, will thus
require modification  to accurately represent the system.


CONCLUSIONS

     A continuous flow  fluorometer  coupled to  a  good data logger
can  provide  excellent   tracer  breakthrough  curves.    The  high
resolution  of the  record and  lack of  sample  handling provide
precise curve definition and  can  minimise labour.   However,   in
remote sites, or where tracer concentration is not  simply measured
the  technique  can  become  logistically   complex,   expensive   and
inaccurate.  However,  recent development of less precise, but quite
adequate  submersible  fluorometers (           1991) suggests that
multi-tracing  to multiple  springs  may  soon  become affordable,
although  still requiring considerable logistical effort.
                             240

-------
REFERENCES

Lauritzen, S.E.,  J. Abbott, R. Arnesen, G. Crossley, D. Grepperud,
          A,  Ive, and S. Johnson, 1985.  Morphology and hydraulics
          of  an active phreatic conduit.   Cave Science 12(4) 153-
          61.

C.C. Smart,  1988a.    Quantitative  tracing  of the Maligne Karst
          Aquifer, Alberta, Canada.   Journal of  Hydrology. V.98
          pp.185-204.

C.C. Smart, 1988b.  Artificial  tracer techniques for the determina-
          tion of the structure of conduit aquifers.  Groundwater.
          V.26 pp.445-453.
PERSONAL RESUMES

      Chris Smart obtained  a  BSc from the University  of Bristol
(UK) in 1974 where his interest in karst and hydrology was sparked.
The statistical properties of meandering cave passages provided the
focus of an MSc gained from the University of Alberta in 1977.  A
PhD. was awarded by McMaster University in 1983 following a study
on  the  hydrology of the  partially subglacial  Castleguard Karst
aquifer in the Canadian Rockies.  Present research interests focus
on instrumented field studies in alpine terrain, including karst,
till, glacier and fracture aquifers.  This work is complimented by
work  on  instrument  development,  data processing,  mapping  and
numerical modelling.

     An organic chemist by training, Stein-Eric Lauritzen converted
an amateur interest in caves and karst into  a  profession.   He is
currently  involved  in projects in  karst hydrology and  hydro-
chemistry,  quaternary geology  and   cave  archaeology,  glacier
hydrology and karst resource conservation.  He runs the Norwegian
Uranium-Thorium dating facility in the University of Bergen.
                             241

-------
Chris Smart - Continuous Flow Fluorometry


1.   Given that most applications of continuous flow fluorometry will
    probably be at sites where there is ready access to electric
    power from the mains,  I  think you have overemphasized the
    difficulties of the technique.   Please comment.

    Most springs and cave streams are remote from mains power.   Therefore,
    the arbitrary application of continuous flow fluorometry will  be
    constrained by the logistics of power supply.   At those few sites
    with safe mains power,  the technique is straightforward; however,
    other problems remain.   First are technical  problems of robust
    data recording.  Second  are irrecoverable analytical  problems  which
    may not be evident in the data  record.   Finally,  problems associated  with
    lack of physical  samples will  also remain unless a sampler  is  operated
    concurrently.  Therefore,  provision of mains power only overcomes
    the most superficial,  albeit immediate difficulties of continuous
    flow fluorometry.
                                      242

-------
  DEVELOPMENT OF A FLOW-THROUGH FILTER FLUOROMETER  FOR  USE  IN

     QUANTITATIVE DYE TRACING AT MAMMOTH CAVE NATIONAL  PARK


                           Martin Ryan

          Division of Science and Resources Management
                   Mammoth Cave National Park
                 Mammoth Cave, Kentucky  42259



                            ABSTRACT

A series of quantitative traces were completed in the Buffalo
Spring ground water basin in Mammoth Cave National Park as part
of the field testing of a newly developed filter fluorometer.
The RME flow-through filter fluorometer is an inexpensive, labor-
saving, battery-operated,  submersible device that, when
interfaced with a digital datalogger,  is  capable of precisely
measuring the travel time of two dye slugs (rhodamine WT and
fluorescein)  simultaneously.   It is also  able to measure the
approximate dye concentrations passing a  recovery point.
Interpretation of RME data yielded unprecedented information
concerning the hydrology of the Buffalo Spring basin—including
the unanticipated discovery of a major flow-route.


                          INTRODUCTION

Fluorescent tracer dyes are commonly used in the study of ground
water movement in karst terraries.   Qualitative dye tracing,  using
passive dye-detectors like cotton and activated charcoal to
recover the dye,  is frequently employed to approximate ground
water flow-routes and define ground water basin boundaries.
Quantitative dye tracing,  which requires  the measuring of
changing dye concentrations at a recovery point, is useful in the
determination of ground water velocities,  conduit condition
(phreatic or vadose), unexpected flow routes,  and water "budgets"
(for basins with multiple discharge points).   If the flow of dye
through an aquifer is closely documented  using quantitative
tracing,  models of soluble point source contamination events may
be generated.   Such models may be used by ground water managers
to aid in drafting contingency plans for  dealing with acute
ground water pollution.


                               243

-------
Quantitative dye tracing—much more expensive and labor  intensive
than qualitative tracing—is generally performed only as  a
supplement to qualitative tracing.  Typical methods used  to
recover dye include grab sampling, automatic sampling, and flow-
though fluorometry.  Each of these methods has inherent
drawbacks.  Grab sampling is enormously labor intensive.
Automatic sampling is moderately labor intensive and costly—
automatic samplers cost over $2000 each.  Samples obtained using
grab or automatic sampling must be quantitatively analyzed on a
fluorometer.  Fluorometer prices start at around $7000.   Flow-
through fluorometers are able to directly measure concentrations
of dye at a recovery point; however, in addition to being
expensive, they require a pump to generate flow through the
instrument.  The energy requirements of the pump and fluorometer
make this method impractical in remote areas where electrical
service is not available.  A need exists for the development of
cheaper and easier techniques capable of obtaining results of
similar precision.


                    THE  RME  FILTER FLUOROMETER

 An alternative to conventional dye recovery methods has  been
developed and is being used extensively at Mammoth Cave National
Park.  The RME flow-through filter fluorometer is an inexpensive,
battery-operated, submersible probe, supported by a digital
datalogger.  It may be deployed in the field for extensive
periods of time, and requires only occasional servicing.  The RME
uses 6 volts DC and draws less than 100 ma/hr.   Since it  is
submerged into the spring and is designed to slowly draw  water
through itself, no pump is required.  The RME is capable  of
continuously measuring small concentrations of two dyes
simultaneously—rhodamine WT (C.I. Acid Red 388)  down to  0.5 ppb
and fluorescein (C.I. Acid Yellow 73)  down to 5 ppb.  The
material cost of building an RME is approximately $175 per unit.
A datalogger with versatile programming is required to execute
and record the data measurements.  The datalogger and RME battery
power supply, attached to the RME through waterproof wire, must
be placed above the highest possible water level in a
weatherproof enclosure.

The RME fluorometer is two filter fluorometers in one package.
It has one light source, a 4-watt clear quartz mercury
ultraviolet lamp, sandwiched between two flow-through sample
tubes (Figure 1).  The flow-through tubes are made of 6 inch
sections of 1-inch ID aluminum box tubing.  Two elongate  windows
are milled into each tube at right angles to each other.  Clear
microscope slide glass is mounted across each window from the
inside using a silicone sealant.


LIGHT FILTER SETS

Situated between the lamp and the rhodamine sample tube is an
                                244

-------
                     PHOTORESISTOR ARRAY

                              PHOTORESISTOR
                        4 WATT Hg LAMP

        RHODAMINE TUBE

              EXCITATION FILTER SET
       FLUORESCEIN TUBE

EXCITATION FILTER SET
            I INCH
Figure 1.  Diagrammatic cross-section  of  the  RME  filter
           fluorometer.

excitation filter set, composed of a Kodak Wratten 61 gel filter
sealed between two Corning 1-60 colored glass filters
(recommended in Smart and Laidlaw, 1977).  This filter set is
designed to allow only the 546nm mercury  line light to illuminate
the inside of the sample tube.  The other major spectral lines
emitted by the mercury lamp  (578nm, 436nm, 405nm,  365nm, and
254nm) are absorbed by the filter set.  The 546nm light
illuminating the interior of the rhodamine tube is within the
excitation spectrum for rhodamine WT  (its excitation maximum is
about 555nm), so if that dye were present in  the  sample tube, it
would be induced to fluoresce.  An emission filter set,  composed
of a Corning 3-66 and a Corning 4-97,  is  located  between the
other window and the photodetective array. The secondary filter
set is designed to transmit a spectrum that has peak nearly
coinciding with the emission maximum of rhodamine WT (about
580nm); the filters are nearly opaque  to  wavelengths outside this
relatively narrow spectrum.

The excitation filter set for the fluorescein tube is a
combination Wratten 2A and a Wratten 47B.  It transmits the 436nm
mercury line, which is within the excitation  spectrum of
fluorescein (the excitation maximum is 490) and is nearly opaque
to the other lines.  The emission filter, located between the
emission window and a photodetective array, is composed of a
Wratten 2A, a Wratten 12, and a Corning 4-97  (recommended in
Turner Designs, 1983).  This filter set transmits a portion of
the excitation spectrum of fluorescein (the maximum is about
520nm) to the photodetective array.  All  Wratten  filters are
sealed inside clear glass to help preserve them.


THE PHOTODETECTORS

The RME uses cadmium sulfide photoresistors as photodetectors.
The electrical resistance of a CdS photoresistor  varies—in an
                                245

-------
inverse log relationship—with the intensity of light  striking
it.  The photoresistors are extremely sensitive even to tiny
changes in light intensity—especially in the 500nm to 600nm
range.  Both of the RME's sample tubes have an array of three
photoresistors, connected in parallel and located outside the
emission filter (Figure 1).   Another photoresistor, along with a
protective neutral density filter, is used to monitor  intensity
fluctuations in the mercury lamp.
PROGRAMMING THE RME

To conserve battery power and to extend the life of the heat
sensitive Wratten filters, the lamp, and the lamp circuitry, the
lamp is only operated periodically.  The datalogger (Campbell
Scientific 2IX microloggers were used by this investigator), via
a relay, switches the lamp on for one minute out of every ten.
At the end of that one minute the resistances of the rhodamine
array, fluorescein array, and the lamp reference are measured,
using a DC half bridge, and stored.  If dye above a pre-chosen
concentration is sensed, the sampling interval will change to
once per five minutes and then revert back to ten minutes when
that concentration is no longer exceeded.  This insures a better
probability of documenting short duration features.

Cadmium sulfide photoresistors have an undesirable inherent
behavior called a memory or light history.  If they are placed in
total darkness for even a brief period of time (as they are when
the lamp is off), they will become "stuck" in this very high
resistance dark state; small increases in illumination will not
cause any change in electrical resistance.  Minute increases in
dye concentration above background would therefore go unnoticed.
To counter this, an LED that keeps the rhodamine array slightly
illuminated is switched on when the lamp is switched off.  The
fluorescein array receives enough exciting light through its
secondary filter (an otherwise negative trait) that, when the
lamp is switched on, the array is quickly "snapped out" of the
memory state.
RME ENCLOSURE

The electronics and optics of the RME are enclosed in a
watertight 4-inch schedule 40 PVC pipe compartment.  The sample
tubes are connected to 3/4-inch PVC pipes which pass through the
endcaps of the compartment.  To prevent ambient light from
reaching the sample tubes, light baffles made of 45 and 90 degree
ells are inserted between the sample tubes and the outside
(Figure 2).  The front light baffles are removable to facilitate
cleaning the sample tubes.  All piping and the enclosure itself
are painted black as further protection against ambient light.
On the downstream end of the RME exterior is an inverted funnel-
shaped feature called a drag inducer.  When the RME is properly
oriented in a flowing stream, the drag inducer produces a vacuum
                                246

-------
                                              — --- L'lVK, INDUCEfl
                                               SAMPLE TUBES
                                               (SHOWING WINDOWS)
                                             WIRE CONDUIT
                                             FRONT LIGHT
                                           '-—' ~ BAFFLE
            DIAGRAMMATIC CUT-AWAY VIEWS
                   OF RME
        Figure 2.   Diagrammatic cut-away views of the RME.

effect which draws water through the  sample tubes.   This insures
that the RME is taking a sample representative of the water
around it at any given time.'
SUMMARY OF HOW THE RME MEASURES DYE  CONCENTRATION

The following is a very basic summary  of  the  physical
relationships employed by the datalogger  and  the RME to measure
dye concentration:
     1).  A change in dye concentration results  in a directly
     proportional change in fluorescence.
     2).  A change in fluorescence results  in  a  directly
     proportional change in the amount of  illumination striking
     the CdS photodetector.
      3).  A change in the amount of  illumination striking the
     photodetector results in an inverse  logarithmic change in
     the electrical resistance of the  photodetector.
     4).  A change in the electrical  resistance  of the array is
     measured as a proportional change in  the output voltage of a
     DC half bridge by the datalogger.
The datalogger records the output voltages of DC half bridge
measurements, which are downloaded from the logger onto a
cassette tape or into a data can and then  loaded into a PC
spreadsheet where the following transformations may be made to
it:
     1).  Using a conversion formula, output voltages are
     converted to resistances.
     2).  Resistances are converted into dye concentrations by
     interpolating from a calibration  curve—calibration curves
     are created prior to field deployment by plugging one end of
     an RME's flow-through tubes, pouring  in  a  series of
                                247

-------
     standards, and recording the resultant resistances.
     3).  Concentrations are temperature compensated using the
     formula provided by Smart and Laidlaw (1977) .
     4).  Instrumentational background is subtracted out by
     "zeroing" data immediately proceeding the leading edge of a
     dye slug.
     5).  Temperature compensated dye concentrations are
     multiplied by the discharge of the spring or stream to
     determine dye load.
     6).  The area under a dye load curve is calculated to
     determine the total amount of dye recovered.

Because of deficiencies in the present RME design (primarily in
the light filter sets), fluorescein concentrations may only be
roughly determined; therefore, fluorescein may only be reliably
used with the RME in ground water time of travel study.
     THE  DETERMINATION OF THE HYDROLOGY OF THE  BUFFALO  SPRING
            GROUND WATER BASIN USING RME FLUOROMETRY

INTRODUCTION

The Buffalo Spring ground water basin occupies about a 20Km2
portion of Mammoth Cave National Park,  Kentucky.  It is located
north of the Green River,  just west of its confluence with the
Nolin River within the Hilly Country of the Chester Upland
(George,  1989).   Buffalo Spring is stratigraphically located near
the middle of the Girkin Formation, the uppermost unit of a thick
section of highly karstifiable Mississippian limestone.  An
alternating sequence of relatively thin Mississippian sandstones
and limestones and the basal Pennsylvanian Caseyville Formation
are located above the Girkin Formation.   The regional dip is a
relatively gentle 5 to 15 m/Km to the west-northwest.  The Buffalo
Creek surface drainage splits into its two main tributaries, the
Wet Prong and the Dry Prong, about IKm from the Green River.
Surface flow is absent in both of these branches where the Girkin
Formation crops out,  except under high flow conditions.  The
surface streams are lost through a series of sequential ponors
downstream from the upper Girkin contact.   Many of the
tributaries to both the Wet and Dry Prongs also sink into the
upper Girkin.

Buffalo Spring is a rise pit type spring.   It has a highly
variable discharge that ranges between about 60 and 1800 1/s,
with an average discharge of about 500 1/s.   Qualitative dye
tracing by Meiman and Ryan  (1990) confirmed that sinking water
from the Wet Prong and Dry Prong sequential ponors resurges at
Buffalo Spring (Figure 3).   A large tributary to the Dry Prong,
Mill Branch, and numerous smaller tributaries to both Prongs were
also traced to Buffalo Spring.   Qualitative dye tracing showed
that Confluence Spring was an overflow spring for Buffalo Spring.
Fort's Funnel is a cave located on the flank of Collie Ridge just
northwest of the Dry Prong  (Figure 3)  and containing a large
                               248

-------
                                           DP332, DPIOO. AND
                                           DPI03 TRACES
                                           INJECTION POINT
         -X-   (^ L- WP33O TRACE
            \ <^ ^ INJECTION POINT     <-£,

                           \&
           A i   /         CP^
                                                 KELLVS CUT-OFF


                                               • MBI02 AND MBI03 TRACES
                                          ,.~	- INJECTION POINT
             ',  FF330 TRACE
               INJECTION POINT
«» BUFFALO SPRING V,. -


 % -  -'  ^
   ' ', ', BUFFALO CREEK
                                                      RHSEIOI AND
                                                      RHSEI05 TRACES
                                                      INJECTION POINT
                                                   DRY STREAM BED
                                                   TERMINAL SINKPOINT
                                                   QUANTITATIVE DYE TRACE
                                                      I KM

                                                  » SINGLE CHANNEL RME
                                                  »«OUAL CHANNEL RME
     Figure 3.  Map of the Buffalo Spring ground water basin.

stream.   The  discharge  of the cave stream is roughly  half that of
Buffalo  Spring.   Every  qualitative dye  trace performed in the
basin was recovered with  positive results at Fort's Funnel as
well as  at Buffalo Spring and Confluence  Spring  (if Confluence
Spring was flowing). Since the discharge  at Fort's Funnel was
considerably  less than  Buffalo Spring,  the  exact relationship
between  the Wet  and Dry Prongs and Fort's Funnel remained
problematic even after  recovering numerous  qualitative traces.


QUANTITATIVE  TRACES

Table 1  is a  summary of the  quantitative  tracer tests  performed
in the Buffalo Spring basin  using the RME filter fluorometer.   A
dual channel  RME,  capable  of recovering rhodamine and  fluorescein
simultaneously,  was placed at Buffalo Spring during all  the
traces. _   Single  channel RMEs,  rhodamine sensitive only,  were
placed in  Fort's  Funnel and  Confluence Spring only during the
April,  1991 traces.  Figure  3  shows the injection points
recovery points,  and straight line travel routes for each trace
and Table  2 summarizes the results of each  trace.
                                 249

-------
i-n
O
               Table  1.  Summary of quantitative dye traces
                         performed in the Buffalo Spring basin
                         using the RME.
Table 2.  RME-determined results  of  the
          Buffalo Spring basin quantitative
          traces.
Trace Injection Injection
number date site
DP100 4-10-91

DP103 4-13-91

DP332 11-28-90

MB102 4-12-91
MB103 4-13-91


RHSE101 4-11-91
RHSE105 4-15-91

WP330 11-26-90
FF330 11-26-90


Dry Prong
base flow
sink point
Dry Prong
base flow
sink point
Dry Prong
base flow
sink point
Mill Branch
quarry
ponor
Mill Branch
quarry
ponor
Raymer Hoi.
SE terminal
sink point
Raymer Hoi.
SE terminal
sink point
Wet Prong
base flow
sink point
Fort's Fun-
nel cave
stream
Recovery
point (s)
Buffalo Spring
Confluence Spring
Fort ' s Funnel
Buffalo Spring
Confluence Spring
Fort's Funnel

Buffalo Spring
Buffalo Spring
Confluence Spring
Fort's Funnel

Buffalo Spring

Buffalo Spring
Buffalo Spring
Confluence Spring
Fort's Funnel
Buffalo Spring

Buffalo Spring

Dye used/
Quantity
Rhod. WT/
150g
Rhod. WT/
150g
Rhod. WT/
119g and
Fluor./
150g
Rhod. WT/
150g
Fluor./
15 Og

Fluor./
150g
Rhod. WT/
150g
Rhod. WT/
119g
Fluor./
lOOg

Trace *Recovery Apparent
number point (s) travel
distance
Discharge Time to
(1/s) leading
edge
(meters)
DP100

DP103

DP332
MB102
MB103
RHSE101

RHSE105
WP330
FF330
BS
CS
FF
BS
CS
FF
BS
BS
CS
FF
BS
BS

BS
CS
FF
BS
BS
4320
4000
3120
4320
4000
3120
4320
3980
3630
2780
3980
4850

4850
4600
3750
2000
1200
875
150
395
950
460
770
280
940
360
700
960
930

1090
330
595
280
280
(hours)
12
11
8
10
8
5
28
11
9
7
9
14

12
11
7
11
18
.83
.13
.23
.00
.90
.75
.75
.27
.77
.67
.50
.72

.25
.85
.25
.92
.20
Time to
peak
cone.
(hours)
14.38
14.18
9.38
11.90
10.65
7.50
34.75
12.37
11.02
8.37
10.50
17.22

16.75
14.75
11.85
13.82
20.75
Approx.
peak
cone.
(mg/1)
.002
.005
.004
. 002
. 006
.010
.007
.004
.011
.024
-
_

.0005
.001
.0025
.100
-
*Buffalo Spring = BS
Confluence Spring = CS
Fort's Funnel = FF

























-------
                         November Traces

Three quantitative traces were recovered at Buffalo Spring in
November,  1990.   Flow conditions were low and relatively stable
during this period,  and Confluence Spring was not flowing.
Figures 4  and 5  show the recovery curves of traces initiated
simultaneously from Fort's Funnel and the Wet Prong terminal
sinkpoint.   The  dye slugs were recovered using both a dual
channel RME and  an automatic sampler.  Surprisingly, the
rhodamine  injected in the Wet Prong  (WPS30 trace) arrived at
Buffalo Spring more than six hours before the fluorescein from
the much closer  Fort's Funnel (FF330 trace) (Figures 4 and 5, and
Table 2).   This  shows that a primary flow-route exists between
the Wet Prong sink and Buffalo Spring with a gradient that is
significantly steeper than the flow-route between Fort's Funnel
and Buffalo Spring.   Conseguently, a difference in head must
exist between this newly discovered trunk conduit carrying Wet
Prong water and  the Dry Prong trunk visible at Fort's Funnel.
Enough of  the Wet Prong trunk is apparently pirated by the Dry
Prong trunk above Fort's Funnel to be detected using qualitative
dye tracing methods,  but not enough to cause a noticeable
secondary  dye slug to appear at Buffalo Spring while using
quantitative methods.

The RME results  (Figure 4)  compared favorably with the ISCO
sampler/Shimadzu spectrofluorophotometer results (Figure 5).   The
ISCO sampler, which was programmed to draw a sample hourly,
failed to  sample the peak rhodamine concentration.   The higher
resolution RME data shows that the peak rhodamine concentration
was considerably greater than what was determined by the
ISCO/Shimadzu methods.  Figures 4 and 5 prove that the RME is a
capable alternative to conventional dye recovery methods.  RME
and ISCO/Shimadzu results were also similar for a simultaneous
two-dye trace from the Dry Prong sinkpoint to Buffalo Spring
(DP332) .


                          April  Traces

Six quantitative traces—two using fluorescein and four using
rhodamine—were  performed in the Buffalo Spring basin in April,
1991.  Flow conditions were much higher than in November and
fluctuated due to several moderate rainfall events received
during the study period.  During the six day study period the
discharge  was measured, using a tape measure,  a survey stick, and
a Marsh-McBirney flow meter, five times at Buffalo Spring, four
times at Fort's  Funnel, and five times at Confluence Spring.
Discharges listed in Table 2 for each dye slug at each recovery
site were  interpolated from these measurements and are presumed
to be only moderately accurate.

If automatic sampling had been used as the dye recovery method at
all three  sites  for six days with a sampling interval of one
hour, it would have required changing 432 sample bottles and
                                251

-------
           0.00
                                                     SI
                                                     M
                                              200000  ^
                                              240000>
                                                   70

                                              280000 ^
                                                   3-

                                              320000 »
                                              360000
                                         Z
                                         O
                                         M
                                         O
                                         •*)
Figure  4.
                  5    10   15   20   25   30   35
                   TIME FROM INJECTION (hours)
Recovery curves  of  FF330 and WP330 simultaneous  traces
created from RME data.
x~x U.U /
X
gjo.06
£"J 2
| 2 0.05

C "^J
E *
Q £ 0.04
P> g
Q 5 0.03
< o
g | 0.02

o 3
to 0 0.01
i-^ O
T*
J-i
K 0.00
1 1
i i i i
RHODAMINE
-

_



-

_

-


. ~^^ WET PRONG
I "
i • -. FLUORESCEIN
; J FORT'S FUNNEL ~
*

.; .



\ M
\ • ^v
I ^ ^

\ ! V
J \ "•-
lNy •.

VJ. 1- ^i
O ~
ra S
o ^~
S a
z 5
£.
o fe
o a
0.2 O G
2 S
H 3
si
0.1 0 S
•y Z
p]
?g
(D
X
n n>-^
                  5    10   15   20   25  30
                   TIME FROM INJECTION (hours)
                               35
Figure  5
Recovery curves of  FF330  and WP330 simultaneous traces
created from automatic  sampler/spectrofluorometer
results.
analyzing  576  samples on the spectrofluorometer.   All this toil
would have produced only mediocre results because each dye slug
would have been sampled only a few times  (due  to  temporal
compactness  of the slugs)  and only a vague picture of a slug's
true shape would have resulted.  The ten  (or five)  minute
sampling freguency of the RME insures that a more realistic view
of the dye recovery curve will be recorded.


MB102 Trace

Rhodamine  trace MB102 was initiated at a discrete sinkpoint in
Mill Branch  and was recovered at Fort's Funnel, Confluence
                                 252

-------
Spring,  and Buffalo Spring.  The  resultant recovery curves are
shown in Figure 6.  It is apparent  in  that the peak
concentrations of Confluence Spring and Buffalo Spring are much
lower than Fort's Funnel.  The Wet  Prong trunk converges with the
Dry Prong trunk somewhere between Fort's Funnel and Buffalo
Spring and dilutes the dye-laden  Dry Prong waters.   Further
longitudinal dispersion was probably also a factor in this
reduction.  Hubbard et al. (1982) contains an excellent
description of dye dispersion in  streams during quantitative
traces.   Evidently, an input from the  Wet Prong trunk exists
between Fort's Funnel and Confluence Spring as well,  and it is
responsible for the sizable drop  in concentration at Confluence
Spring.   Helpful "black box" type models of conduit systems like
these were presented and discussed  by  Brown (1973).

The total mass of dye recovered at  Fort's Funnel during the MB102
trace was 122g, or 81% of the 150g  injected.   Only  45% of that
122g was recovered at the two terminal  springs;  the remaining 55%
was unaccounted for.  Since the total  discharge of  the Buffalo
Spring system was above average and increasing,  a plausible
explanation is that dye-laden conduit  water moved into diffuse
storage adjacent to the conduit—like  river bank storage.
Atkinson et al. (1973) describes  this  analogy in some detail.
Significantly, very strong positive results for rhodamine and
fluorescein were still being found  on  the passive dye detectors
at Buffalo Spring under low flow  conditions more than three
months later.   It is possible that  dye  in conduit adjacent
diffuse storage was slowly released as  the summertime base flow
condition was approached.  Other  factors that may have
contributed to the apparent dye loss may have been  adsorption or
the use of inaccurate (too small) discharge values.
g
h-
<
h-
Z
UI
O
Z
o
o
S
o:

I
             0.024
             0.020
             0.016
             0.012
             0.008
             0.004
             0.000
                               FORT'S TUNNEL

                               CONFLUENCE SPRING

                              - BUFFALO SPRING
                                    ': 700
                                        \ 360 l/i
                                            940 l/i
                             6      9      12
                            HOURS FROM INJECTION
                                               15
          Figure 6. Recovery  curves  from MB102 trace,
                                253

-------
                            MBI02 TRACE
                     WEI PRONG TRUNK
                                         I50g
                                 FORT'S FUNNEL 0=700 l/s
                                 CAVE     I22g
                             CONFLUENCE 0=360 l/s
                             SPRING    2Sg
                            BUFFALO 0=940 l/s
                            SPRING  29g
                     GREEN RIVER
                Figure 7.   Summary of MB102 trace.

Figure 7 summarizes the dye  recovery results for the MB102 trace.
The percentage of dye  (and flow)  going to each spring was
computed by considering the  total mass of dye recovered at both
terminal springs 100%, then  the mass recovered at either one of
the springs was divided by the total recovered at both springs
and multiplied by 100.  The  discharge of the Wet Prong trunk was
computed by subtracting the  discharge at Fort's Funnel (which was
assumed to be the entire Dry Prong trunk flow)  from the combined
Confluence Spring and  Buffalo Spring discharges.  Based on this,
approximately 47% of the discharge from Fort's Funnel resurged at
the overflow route—so about 329  l/s of Confluence Spring's 360
l/s discharge came from the  Dry Prong trunk.   The remainder of
the Dry Prong trunk flow and nearly all the Wet Prong trunk flow
resurged at Buffalo Spring.

DP103 Trace

Figure 8 shows the recovery  curves at the three recovery sites
for the DP103 dye trace.  When compared with the recovery curves
from the MB102 trace,  in which the same amount of rhodamine was
injected, several differences are discernable:  the peak
concentrations are all lower, the travel times are less,  and the
slugs are more dispersed longitudinally.  Because the discharge
was greater during this trace, the first two are believable—even
though the reduction in concentration was larger than such an
increase in discharge  would  warrant.   An increased longitudinal
dispersion, however, is the  exact opposite of what would
typically be expected  for a  trace initiated under higher flow
conditions.

The reason the DP103 slugs were more dispersed than the MB102
slugs may be related to the  fact  that DP103 dye entered the
subsurface through three widely spaced sequential ponors instead
of through one discrete ponor.  All three Dry Prong traces were
                                254

-------
          g
          5
          o
          o
          o
           g
           i
           CC
           a
           Ld
              0.010
              0.008
              0.006
0.004
          2   Q.Q02
           2
           en
              0.000
                            FONT'S FUNNEL

                            CONFLUENCE SPRING

                            BUFFALO SPRING
                            450 l/s
                                             950 l/s
                 0
                       3     6      9     12     15
                            HOURS FROM INJECTION

           Figure 8.   Recovery curves from DP103 trace.
initiated from the same point—just  above the Dry Prong base flow
terminal sinkpoint.  However, the  flow conditions were very
different for each trace: the terminal sinkpoint for the DP332
trace was the base flow sinkpoint, the terminal sinkpoint for the
DP100 trace was about 600 meters downstream at a ponor called
Kelly's Cut-off, and for the DP103 trace the terminus of surface
flow was a huge ponor 250 meters further downstream called Norain
Cave (Figure 3).  Dye from the DP103 injection entered the
subsurface at all three of these major ponors.   As a result of
this the injected slug was split into three separate slugs, each
with a slightly different route to follow at first.   Flow from
the three separate inputs eventually reunited and the three
slightly out of phase dye slugs were fused back together—
slightly more dispersed and with a lower amplitude than a single
input slug would have been.

Only 75g of the 150g of dye injected (50%)  was recovered at
Fort's Funnel.  The amount of that dye which was recovered at the
terminal springs was 73%.  The results,  summarized in Figure 9,
suggests that 55% of the total discharge passing Fort's Funnel
went to the Confluence Spring (about 423 l/s)  and 45% went to
Buffalo Spring (about 347 l/s).  The remainder of each was
supplied by the Wet Prong trunk.

When the DP103 trace results  (Figure 9)  are compared to the MB102
results (Figure 7) several important insights into the behavior
of this aquifer may be gleaned: the  Confluence Spring waters are
mostly derived from the Dry Prong trunk,  and the Wet Prong trunk
is not well connected to Confluence  Spring.   Thus the Confluence
Spring is predominately an overflow  spring for the Dry Prong
trunk.   If the discharge were increased in both the trunks
simultaneously, hydraulic damming by Wet Prong waters, which
basically have no place else to go but Buffalo Spring, would
cause a decrease in the percentage of Dry Prong water resurging
                                255

-------
                             DPI03 TRACE  I5°8
                     WET PRONG TRUNK
                                  DRY PRONG TRUNK
                               Xr
                               J FORTS FUNNEL 0=770 l/s
                                 CAVE      75g
                             CONFLUENCE  0=45O l/s
                             SPRING      30g
                            BUFFALO 0 = 950 1,'s
                            SPRING   25q
                     GREEN RIVER
                Figure 9.  Summary  of  DP103  trace.

at Buffalo  Spring and an increase in the percentage  overflowing
at Confluence  Spring.


        Repeated  Quantitative Traces as a Predictive Tool

The input-to-resurgence  travel time of a dye slug decreases  with
increasing  discharge.  Peak concentrations often decrease  with
increasing  discharge because dilution increases, and longitudinal
dispersion  decreases due to decreased dye slug travel time.
Figure 10 illustrates the results of a trace from Dry Prong  to
the terminal spring(s) repeated three times under different  flow
conditions.  The  aforementioned effects of increased discharge
are very clear.   Using results from these three traces,  reliable
predictions for almost any set of flow conditions could  be made
concerning  travel time,  peak concentration,  and dispersion for a
soluble contaminant  accidentally injected into the Dry Prong.
Mull et al. (1988) gives a detailed discussion of this important
topic.

                Determination of Conduit Condition

Conduit condition may be resolved eveji if a conduit  is
inaccessible by using quantitative tracing;  this was done  for
segments of the Dry  Prong trunk using the RME.  The  discharge of
a vadose conduit  is  increased by increasing the flow velocity
and/or the  cross-sectional area of the channel (by increasing
stage).  The only way to increase the discharge of a phreatic
conduit, since it is completely full and stage cannot be
increased,  is by  increasing the flow velocity.  So,  when log
discharge (X)  is  plotted versus the log travel time  (Y)  for  a
series of traces  through a phreatic conduit the result would be a
line with a slope of nearly -1 (Smart, 1981).   A plot of traces
through a vadose  conduit would be a line with a slope of less
                                256

-------
than -1.0  (but  probably greater than -0.3).   Figure 11 shows
first order  linear regressions  of log discharge versus log  travel
time for the three Dry Prong  traces recovered by the RME  for,  the
entire Dry Prong,  the segment of the Dry Prong trunk upstream
from Fort's  Funnel, and the segment of Dry  Prong trunk downstream
from Fort's  Funnel.  Judging  from their slopes,  which are
admittedly based on a paucity of data, the  segment downstream
from Fort's  Funnel is apparently mostly phreatic and the  segment
upstream is  mostly vadose.
                                              0.008
         BS=BUFFALO SPRING
         CS=CONFLUENCE SPRING
 Figure  10.   Recovery curves of Dry Prong  to the terminal
             spring(s) traces.
          LJ
          o
          Q
          U
          o 10
          UJ
          2
            10U
                       _0	  ENTIRE DRY PRONG CONDUIT

                       -T—  UPSTREAM FROM FORT'S FUNNEL

                        -V--  DOWNSTREAM FROM FORT'S FUNNEL
             1CT                10J                10*
                       BUFFALO SPRING DISCHARGE - (l/s)

 Figure 11.  Log-Log plots  of discharge vs.  time of travel for
            various segments of the Dry  Prong trunk.
                                 257

-------
              Hydrologic  Structure of Buffalo Creek

A pictorial summary  of  the  hydrologic structure of the Buffalo
Spring karst ground  water basin was generated by synthesizing all
the qualitative and  quantitative trace data and geomorphological
data collected  (Figure  12).   Smart (1988)  and Smart and Ford
(1986) presented a structural model of the Castleguard conduit
aquifer and laid the groundwork for this type of aquifer
representation.  Models like  these could be quite useful to
ground water managers charged with determining a course of action
during an accidental contamination event.


                     SUMMARY  AND CONCLUSIONS

The RME is an inexpensive alternative to conventional
quantitative dye recovery methods.   Extensive fieldwork in the
Buffalo Spring ground water basin,  including some in conjunction
with traditional dye recovery methods for comparison, proved that
the RME is a useful  dye quantification tool for field study.
Through use of the RME, subtle details concerning the hydrology
of Buffalo Spring basin were  recognized and described including
several previously unknown  ground water flow routes.   Also
generated was new information about the relationships of the
primary spring and the  over-flow spring to the two primary feeder
trunks and the response of  aquifer transmissivity to changes in
discharge.  Interpretation  of RME data helped to identify the
phreatic and the vadose portions of the Dry Prong trunk conduit.
A structural model of the Buffalo Spring basin was produced using
all the available dye tracing data.
                                        BASE FLOW
                                        SWtPOCHT
               WET PRONG TRUNK
                                    1—1"
                                        MILL BRANCH TRUNK
                                  DRY PRONG TRUNK
                                              /""RAYMER HOLLOW
                                                SE TRUNK

                                           :EU.Y'5 CUT-OFF
                   -BUFFALO SPRING
Figure 12.  Hydrologic structure  of  the  Buffalo Spring ground
            water basin.
                                258

-------
                        ACKNOWLEDGEMENTS

Dr.  Ralph Ewers of Eastern Kentucky University provided to me the
initial idea of developing a submersible fluorometer supported by
a digital datalogger.   Joe Meiman, my colleague at Mammoth Cave,
wrote the datalogger program used by the RME and contributed many
hours of assistance during development and testing.  Mammoth Cave
National Park provided financial support,  laboratory facilities,
and other resources.  I am extremely grateful to them all.
                           REFERENCES

Atkinson,  T.C.,  Smith,  D.I.,  Whitaker,  R.J.,  and Lavis J.J.,
     1973,  Experiments  in tracing underground waters in
     limestone:   Journal of Hydrology,  v.  19, p. 323-349.

Brown,  M.C.,  1973,  Mass balance and spectral  analysis applied to
     karst hydrologic networks:  Water Resources Research, v. 9,
     p. 749-752.

George, A.I.,  1989, Caves and drainage north  of the Green River:
     in W.B.  White  and  E.L. White (eds.),   Karst Hydrology:
     Concepts from  the  Mammoth Cave Area,  Van Nostrand Reinhold,
     New York,  p. 189-221.

Hubbard, E.F.,  Kilpatrick,  F.A.,  Martens,  L.A.,  and Wilson J.F.
     Jr.,  1982,   Measurement of time of travel and dispersion in
     streams  by dye tracing:   U.S. Geological Survey Techniques
     of Water-Resources Investigations, Book  3,  chap. A9, 44p.

Meiman, Joe and Ryan, M.T., 1990, Preliminary results of
     groundwater dye-tracer studies north of  the Green River,
     Mammoth  Cave National  Park:   Proceedings of Mammoth Cave
     National Park's First  Annual Science Conference: Karst
     Hydrology,  p.  137-142.

Mull, D.S., Liebermann, T.D., Smoot, J.L., and Woosley, L.H.,
     Jr.,  1988,  Application of dye-tracing techniques for
     determining solute-transport characteristics of ground water
     in karst terranes:  EPA 904/6-88-001, Atlanta, Ga., U.S.
     Environmental  Protection Agency, 103p.

Smart,  C.C.,  1988,  Artificial tracer techniques for the
     determination  of the structure of conduit aquifers:  Ground
     Water, v-  26,  No.  4, p.  445-453.

Smart,  C.C. and Ford, D.C., 1986, Structure and function of  a
     conduit  aquifer:  Canadian Journal of Earth Sciences, v. 23,
     p. 919-929.

Smart,  P.L.,  1981,  Variations of conduit flow velocities with
     discharge in the Longwood to Cheddar Rising system, Mendip
     Hills: in Beck, B.F. (ed.),  Proceedings  of the VIII
                               259

-------
     International Congress for Speleology, Bowling Green,
     Kentucky, p. 333-335.

Smart, P.L. and Laidlaw, I.M.S., 1977, An evaluation of some
     fluorescent dyes for water tracing:  Water Resources
     Research, v. 13, no. 1, p. 15-33.

Turner Designs, 1983, Fluorometric Facts—Fluorescein, Bulletin
     No. 103, 3 p.
                        BIOGRAPHIC SKETCH

Martin Ryan is a Hydrologic Technician at Mammoth Cave National
Park.  Current research includes developing and implementing new
quantitative fluorometric techniques to model flow in karst
aquifers, delineating ground water basins north of the Green
River in the Mammoth Cave area, and performing water quality
monitoring.  He received his BS in geology from Eastern Illinois
University in 1987 and he is currently pursuing a MS in geology
at Eastern Kentucky University-
                               260

-------
   Development of a Flow-through Filter Fluorometer for Use in
     Quantitative Dye-tracing at Mammoth Cave National  Park

                        by Martin T.  Ryan

I understand that inclusion of a turbidity measurement was
contemplated for an earlier model of your flow-through
fluorometer.  Why did you decide not to monitor turbidity?

The nephelometer/turbidity meter prototype we were experimenting
with had severe problems with water leakage.   Additionally,
laboratory experiments had showed that the design of the
instrument,  which used a jumbo LED with a Fresnel lens and two
photoresistors,  was adequate only for approximation of turbidity
(which is itself only an approximation of suspended sediment
concentration—the maligner of dye fluorescence.)  Even if a
successful turbidity meter had been deployed  alongside the
fluorometer, I do not believe that a turbidity-compensating
mathematical formula could have been generated.   Rather,
turbidity data may have been useful in helping to explain
anomalies in the fluorometric record by intuitive comparison—for
example,  if the mass of dye recovered during  a quantitative trace
is unusually low,  a partial explanation may be available if the
turbidity data is examined and found to be exceptionally high
(eg. because a brief storm event)  during the  time period
including the recovery of the dye slug.

We are still trying to develop a nephelometer/turbidity meter
that is interfaceable with a digital  datalogger,  but it is not
urgently required to validate data acquired through RME
fluorometry — rather it may, at times, be a  helpful supplement.
                               261

-------
262

-------
              Session IV:

Hydrogeology and Processes Occurring in
            Karst Aquifers

-------
264

-------
     Stable Isotope Separation  of  Spring Discharge in a Major
          Karst Spring, Mitchell Plain,  Indiana,  U.S.A.

               Barbara L. Gruver and Boel C.  Krothe
                   Dept. of Geological  Sciences
                         Indiana University

ABSTRACT
     Major-ion chemistry and isotopes of oxygen and deuterium
where used to identify components  of water resurging at
Orangeville Rise, a large perennial  spring.  The Orangeville R^se
basin IB recharged by precipitation  that infiltrates a 125 km'
sinkhole plain and upland area  developed on Mississippian
limestone and clastic sequences.
     In October,  1990, 53 mm of rain fell on the basin in a 40
minute period. Isotopic composition  of  the rain was -8.2%0and
-56%-for 6 I8O and^D respectively.  Prior to this,  at baseflow,/
 TD and 6D  of Orangeville Rise  was -6.27oc.and -407oorespectively,
and specific conductance was 770 uS.  Discharge increased rapidly
from 0.3 nrs   to  3.4 m s   ,  however major ion concentrations  did
not show a rapid drop which would  have  been the case had an
influx of rainwater been responsible for the increase in
discharge. Rather, the most dilute waters were found 24 hours
after peak discharge. Isotopic  composition of waters also
reflected this trend.  Isotopic  separation of storm hydrograph
Into pre-storm and storm water  components revealed that 75% of
discharging water consisted of  pre-storm water.
     Similar monitoring was conducted during April,  1991,  when 4
rains, totaling 38 mm produced  an  increase in discharge from 1.6
     to 4.9 ills'1.  £T) of these  rains ranged from -7 %o to -45%,,
and 5 TD ranged from -2.67oo to —6.8%0. Again,  majoi—ion and
isotope chemistry did not reflect  a  rapid influx of rainwater,
and isotopic hydrograph separation showed that rain water
comprised only 20% of the total volume  of discharge. These
results seem to indicate that water  stored in the epikarst may
discharge at Orangeville Rise prior  to  the arrival of the mass of
rain water.

UTTRODUCTIOH
     Discharge from springs in  karst terrains responds to
influxes of precipitation in many  ways.   The response is governed
by the type of recharge  (concentrated or dispersed), as well as
the routes of subsurface flow within the basin (conduit or
diffuse).   It is also highly dependent  upon the volumes of water
stored in the unsaturated and saturated zones.
     Historically, individuals  have  studied karst systems by
                              265

-------
mapping cave passageways, and  by conducting dye-tracing studies.
Analysis of storm hydrographs  from karst springs has also been
performed, concentrating on  recession limb behavior CMangin,
1974, 1975; Torbarov, 1976;  Milanovic,  1976;  and Atkinson, 1977).
     Major—ion analyses have proven useful in discerning
properties of a karst system,  and have  been used by numerous
investigators, (e.g., Shuster  and White,  1971;  Ford and Ewers,
1978; Friedrich and Smart, 1982;  Gunn,  1983;  and Scanlon and
Thrailkill, 1989).  Changes  in water qiaality associated with
storm hydrographs of karst springs have been helpful in
interpreting sources of water  which contribute to discharge
(Williams, 1983).
     Stable isotopes have been used to  show that new rainwater
can displace old water in storage in karst systems (Bakalowicz et
al., 1974).  The two were found to mix  in different proportions
to create fluctuations in  A  O  across the storm hydrograph.
Dreiss  (1989) separated storm  discharge from karst springs in
Missouri, into new event water and pre-storm water on the basis
of calcium and magnesium ion concentrations.
     This study looks at water quality  changes associated with
storm hydrographs at Orangeville Rise,  a perennial spring in a
karst region of Indiana.  Goals of this study include quantifying
the proportion of rainwater  contributing to storm discharge
through the use of oxygen and  deuterium isotopes,  accounting for
observed water chemistry changes across the hydrograph, and
attempting to identify subsurface storage zones which contribute
to discharge at Orangeville  Rise.

Geologic and Hydrologic Setting of the  Study Area
     Orangeville Rise is a perennial spring located in south-
central Indiana,  approximately 150 km south of Indianapolis
(Figure 1).  The ground water  basin supplying water to
Orangeville Rise,  as defined through dye-tracing studies  (Murdock
and Powell, 1968;  Bassett, 1974),  has an areal extent of 115 W
and lies within two physiographic provinces,  the Mitchell Plain
and the Crawford Upland.
     The Mitchell Plain is a flat-to-gently-rolling lowland
surface developed on middle  Mississippian limestones of the Blue
River and Sanders Groups (Figure 2).  These formations consist of
dense, thinly bedded, highly fractured  limestones which have
experienced profound solutional modification to produce features
typical of karstic terrains.   Indeed,  the Mitchell Plain  is known
for its extensive karst topography expressed In a myriad of
sinkholes, caverns and blind valleys.
     The Crawford Upland occupies a region where alternating
formations of sandstone, shale and limestone of the Mississippian
Chester Series crop out at the surface.   These formations are
more resistant to erosion than underlying Blue River and Sanders
limestones and hence form an upland surface that stands 45 to
60 m above the Mitchell Plain.   The Crawford Upland is a rugged,
deeply dissected surface.  Karst features (i.e., sinkholes and
sinking streams) have developed where stream erosion has out
through Chesterian Strata to expose Blue River limestones.  As a
result, the hydrologic behavior of Orangeville Rise is dictated
                               266

-------
                                       I
                                048 kilometers


                                 Mitchell (Sinkhole) Plain



                                 Crawford Upland
Figure 1. Location of  the Orangeville  Rise,  and  physiographic

           provinces within  the basin.
                    West Baden/ Slephensport Groups



                    Blue River Group



                    Sanders Group
0          5 miles
i i  i   i  i
5 milt
                                        048 kilometers
    Figure 2. Geologic Map of  the Orangeville  Rise basin.
                                    267

-------
by the karstic nature of  the  Blue River bedrock.
                                                      % - 1
     Discharge at  Orangeville Rise ranges from  0.06  nt s   (2 cfs)
to 5.1 m3s"'  (180 cfs).   The Rise is situated within  a  bedrock
alcove forming a pool that  is approximately 15  m in  diameter, and
at low-flow is 6 ra deep.  The Rise is fed by two vertical
conduits which are 6 m  long.   Scuba divers descending  these
shafts, found them to connect to narrow horizontal conduits at
their base.  Discharge  from Orangeville Rise forms a perennial
surface stream that occupies  an alluviated channel 5 m deep.
     At low flow,  when  suspended sediment loads are  low,  waters
in the discharge pool take  on a greenish cast.   After
precipitation events, discharge increases rapidly.    Depending on
the magnitude of the storm, the pool surface may rise 3 m or
more.  The waters  become  heavily laden with suspended  sediment,
and a bailing of the pool surface can be seen as water is forced
up through the vertical conduits.
     Ho perennial  surface streams discharge at Orangeville Rise.
.Rather, discharge  is derived  predominantly from precipitation
that infiltrates soils  of the sinkhole plain,  with a secondary
component coming from sinking intermittent streams draining the
Crawford Upland.
     Soils of the  Orangeville Rise basin consist of clay and clay
loam derived as residuum  from underlying limestones,  as well as
residual material  produced  during erosional retreat of the
Crawford Upland, and thin loess deposits (Ruhe,  1975). These
soils are susceptible to  fracturing due to their high clay
content, creating  macropores  for rapid infiltration  of
precipitation to the subsurface (Wells and Krothe,  1989).   Soils
are up to 5 m thick, but  are  locally much thinner    In many
places, particularly around sinkholes,  soil material has been
completely removed by erosion,  creating  another avenue for rapid
subsurface recharge.
     Southern Indiana has a humid-temperate climate,  with a mean
annual rainfall of 1130 mm.   Precipitation Is distributed through
the year,  such that there is  a pronounced dry time in the fall
when monthly average precipitation is 70 mm,  and a wetter time in
late winter through spring, when average monthly precipitation is
115 mm.  July is the warmest  month on average,  with  mean daily
maximum and minimum temperatures of 90°and 64°  F respectively.
In contrast, January is the coldest month,  when mean daily
maximum and minimum temperatures are 43°and 25 ° F respectively.

Previous Investigations at  Orangeville Rise
     Bassett  (1974) has identified the waters from Orangeville
Rise to be calcium-bicarbonate in character,  and he saw an
inverse relationship between  discharge and total dissolved
solids.  Subsequent studies have confirmed these observations
(Libra, 1981; Tweddale, 1987).   The carbonate aquifer(s)
supplying water to Orangeville Rise have a strong conduit
component,  however, it is  more complicated than a simple system
of conduits.  Based on  major-ion chemistry and sulfur  isotopic
composition, Krothe and Libra (1983) differentiated  two flow
systems within the ground water basin.   One is a shallow conduit
system, dominated  by surface  flow entering the subsurface through
                               268

-------
 large fractures.   Discharge at Orangeville Rise responds rapidly
 to precipitation  by  water entering the aquifer via this route.  A
 second system consists  of a deeper diffuse network of
 interconnected  fine  fractures and joints which respond more
 slowly to inputs  from precipitation.

 BACKGROUND
 Conceptual Model  for Ground Water Storage
     A simplified model illustrating water storage compartments
 for Orangeville Rise is shown in Figure 3.   The model is simple
 in that it includes  no  impermeable bedrock or caprock, and
 recharge occurs only from precipitation that falls within the
 basin. These identified compartments could exist in many karst
 systems, and have important ramifications for supplying water as
 discharge to any  fcarst  spring.
     Recharging precipitation may be temporarily stored in the
 vadose zone  (unsaturated zone)  as soil moisture, epikarst
 storage, diffuse  vadose storage,  or vadose conduit storage.
     The volume of water stored as soil moisture at any one
 moment is highly  variable and dependent on soil properties of
 depth, texture  and composition,  as well as on the timing and
 volume of precipitation events.   After extensive drought
 conditions, soil  moisture will drop to a minimum.   When
 precipitation events are large,  or follow in rapid succession,
 soils may become  completely saturated.  Recharging water from
 precipitation moves  through soil horizons as an advancing front,
 acting like a piston to displace and replace previous soil
 moisture.
                                                    01
Figure 3.  Model of water storage compartments  of  the vadose and
          phreatic zones in the vicinity of  Orangeville Rise.
                                269

-------
     Water which passes through  the soil zone may reside for some
time in epikarst storage.   Mangin (1974, 1975) defined epikarst
as a region in the upper weathered  layers of rock at the base of
soil horizons, lying above  the permanently saturated (phreatic)
zone.  Williams  (1983) described the formation of epikarst as
follows: Water flowing through soil dissolves COj from  the  soil
atmosphere, and becomes aggressive  toward underlying limestones.
As infiltrating water encounters bedrock,  its ability to dissolve
the rock is at a maximum.   These aggressive waters preferentially
flow along prominent fractures,  dissolving bedrock along the
fracture, making thus a zone  of  maximum dissolution, and
producing substantial secondary  porosity.   As the water flows
deeper  into  the fracture,  its capacity to dissolve is
diminished, and the fracture  narrows with depth,  limiting the
rate of water transmission  to deeper zones.   Following
precipitation events of high  intensity and/or large volume, a
substantial body of water can accumulate within the epifcarst, and
form a region where water saturation exists,  resulting in a
perched water table.
     Water will drain from  the epikarst along solutionally
modified fractures, but will  also move into the rock mass through
the system of interconnected  fine joints and fractures.
Depending on the amount of  primary  porosity and the extent of
fracturing and jointing within the  bedrock,  the volume of water
within  diffuse vadose storage may be minimal or qtiite large.  The
degree  of interconnectedness  of  the Joints and fractures will
also govern the rate at which water can move into and out of this
compartment.
     There exists a transition zone between diffuse vadose and
diffuse phreatic storage, as  well as between vadose conduit and
phreatic conduit storage.   It is well documented that changes in
water table elevation in karst terrains can be quite dramatic,
and can occur on a seasonal cycle,  or following precipitation
events.  The magnitude of change in conduits can be on the order
of tens of meters or more,  and can  occur at an alarming rate
after heavy rainfall.  The  response is slower and more subdued
within  the diffuse bedrock  compartments.   Because conduits flood
quickly and drain more rapidly than the diffuse zones,  the two
areas do not appear to be in  equilibrium,  and it is therefore
difficult to speak of a water table in karst terrains.  •
nonetheless,  it is helpful  to visualize water storage in terms of
the vadose and phreatic zones since it lends an element of
elevation or position within  the subsurface.

Stable  Isotopes  in Ground Water  Stxidies
     Stable isotopes of oxygen and  hydrogen bound in a water
molecule, under  low temperature  conditions,  are altered only by
physical processes such as  diffusion,  dispersion, mixing and
evaporation, and may therefore behave in a conservative manner.
As a result,  these isotopes can  act as tracers and their analysis
may aid in determining the  geochemical history of a water mass.
Stable  isotopic analysis of waters  may also provide information
on movement and mixing of water  bodies, provided that the
isotopic composition of each  water  body differs by at least the
                               270

-------
analytical error of Isotoplc  ratio identification.
     In this study, stable  isotopes of  oxygen and hydrogen are
used in an attempt to identify  two time-source components of
discharge at Orangeville Rise:  1)  water that existed within the
basin prior to a storm event  and 2) new rainwater that fell
within the basin.
     If it is assumed that  discharge at Orangeville Rise is
derived by simple mixing between two components:  pre-storm water
(Qpr) and rainwater  (Q >,  then two  mass  balance  equations can be
written that describe the water  flux and the isotope flux at the
Rise:
where Q. le the total measured discharge at any  instant  in time
and is defined by the sum of the two  components.   Delta notation
represents the  °O or deuterium (D)  composition of the
instantaneously measured discharge  G>^>, the rainwater  ($p ), or
pre-storm water (Spe) •  By combining these  two equations, a third
equation can be written which  identifies the  rainwater
contribution to discharge at any  instant  in time, in terms of the
isotopic composition of the measured  discharge,  pre-storm water
and rainwater.
     To use this equation as  a  means of separating discharge into
rainwater and pre-storm water components,  the isotopic
composition of rainwater must be  significantly different than
pre-storm water,  and the pre-storm water should have an
identifiable and uniform isotopic composition throughout the
basin.   This technique has been applied to surface streams by
Fritz et al. (1976) and Sklash  and Farvolden (1979),  where the
pre-storm water was interpreted to be ground water.
     In this report, stable isotopic composition of water is
reported relative to standard mean ocean water (SMOW) ,  in parts
per thousand (or premil) notation,  such that:
                   "
                     O.D]
                                 p
where R is the ratio of D/H or  ' O/ 0,  depending on which ratio
is being identified.
                               271

-------
FIELD IITVESTIGATION
     The field investigation was conducted in two stages-   During
the first stage, water was sampled on a bi-weekly schedule to
establish baseflow chemical characteristics.   It was critical
during this stage to determine  the baseflow or pre-storm isotopic
composition to be used in the mase-balance equations   Water fro.
domestic wells was also  sampled during this stage of the
investigation to establish isotopic composition of the ground
water within the basin.
     The second stage of sampling  concentrated on short-term
changes in water chemistry associated with changes in discharge
brought on by two storm  events  in  the Orangeville Rise basin: in
October, 1990, corresponding  to the driest time of the year; and
in April, 1991, during the wettest time of the year  ^en
recharge to the carbonate aquifer  is a maximum  (Bassett  1974).
Sampling frequency during this  stage of monitoring was keyed to
thePrat! of discharge  increase  and decrease:  every two to four
hours during the time of rapid  discharge increase, peak flow and
initial discharge  recession,  and  less frequently during later
stages when discharge decreased less rapidly.

RESULTS
     Results from  the  bi-weekly sampling confirm the results o±
previous  investigations.  Discharge at Orangeville Rise is
calcium-bicarbonate  in character,  and there  is an inverse
relationship between discharge and total dissolved solids.   It
was  also  found that  as discharge approached  baseflow, the water s
 isotopic  composition repeatedly approached -40 7<*>and -6.2&o for
4'D and b 18O respectively.  These values were  used as the isotopic
signature of pre-storm water in hydrograph separation of spring
discharge.
     Ground water  isotopic composition was found to be quite
 uniform,  even though there was variable major-ion chemistry in
the  basin (Table  1).   The average 6'D and S I8O of ground water was
 found  to  be -44 %0 and -6.6%0 respectively, and  is similar to that
 of the  average isotopic composition of rainwater  in southern
 Indiana CSheppard et al.,  1969).

 October,  1990 Storm
      On October 4,  1990, 53 mm of rain fell  on  the Orangeville
 Rise basin in a 40-minuter period.   Rain  was sampled  from two
 locations in the basin to check for areal variability  in  the
 amount and isotopic composition of  the rain.   The amounts of
 rainfall  at each station were  identical,  and isotopic
 compositions were within analytical  error, and were  averaged.

 Table  1.   Selected chemical data  from domestic  wells  located in
           the Orangeville Rise basin.  < n =  26  )
PARAMETER
Specific Conductance (uS)
Calcium (mg/l)
Sulfate (tng/l)
£°smow <*•>
* *0»mow «J
RANGE
408 (o 1985
70 to 388
13.9 to 976
-43to-46
-6.4to -63
AVERAGE
668
113
IZ5
-44
-6.6
                             272

-------
   Discharge
    (m3/s)
                 0   12   24  36  48  60   72   84   96  108

                               Time - Hours

      Figure  4.  Stnrm hydrograph  for Orangeville  Rise,
                October 4, 1990.
leotoplc composition of the rainwater  was -56%oand -8.2?c,ofor
SD and  &I6O respectively.  Isotopic  composition of the rain may
have shifted during the course  of  the  40 minute storm, however
since the duration of the storm was  short in comparison to
variable travel time for water  in  the  entire basin, the potential
shift was considered insignificant for the purpose of this study,
and a bulk isotopic composition was  determined.
     Prior to this storm, baseflow discharge at Orangeville Rise
was 0.3 nTs1  and specific conductance  of  discharging waters was
770 uS.  Calcium, magnesium,  sulfate,  and bicarbonate
concentrations were at the highest levels measured at any time
during the study.                                      . .
     Discharge  increased rapidly to  a maximum of 3.4  itfs  within
6 hours after the storm  (Figure 4).   There was no precipitation
during the next four days, and  discharge underwent a steady
exponential decrease to  0.3 nrs ' ,  with on exception between 26
and 36 hours, where there was a small  perturbation in the
otherwise smooth recession.
     Results of chemical analyses  for  waters collected across
rising and receding limbs of  the hydrograph are shown  in Figure
5.  It can be seen that  major-ion  concentrations slowly decreased
during the first 8 hours, even  though discharge increased
dramatically.  Concentrations were not at their minimum until 24
hours after peak discharge,  when the HCOj concentration had
decreased to 75% of its  baseflow value and SO^ dropped to only
21% of its baseflow value.   These  minimums were preceded by a
small yet significant  increase  in  ionic concentration  for several
species, which  corresponds  to the small  irregularity  in
recession.   Ion concentrations  remained  low during much of the
recession, and  only started to  recover in the  last  24 hours of
monitoring.
     Isotopic ratios remained fairly constant during  the rapid
rise in discharge, and showed a small decrease during peak
                                273

-------
        Discharge (m3/s)
                   3.5
                                          72
                                                 96
                                           BOO

                                           750

                                          - 700

                                           650

                                          -600

                                          ^550

                                           500

                                           450
                                                           Specific
                                                          Conductance
                                 Time - Hours
      B
Discharge
 (rrvVs)
            Discharge
-20

-16

-12

-8

- 4
                                             72
                                   Time - Hours
                                                     96
                                                     HCO3
                                                     (mg/T)
                                                   20


                                                   16


                                                   •12


                                                   -8
                                                                  g21, Na*
                                                                  (mg/l)
                                                                     ,cr
                                                                  (mg/l)
Figure  5.  Chemical data  from Orangeville  Rise,  October 4-8,  1990:
            A)  specific conductance,  B) cation concentrations,  and
            C)  anion concentrations.
                                       274

-------
          ^ OS
           Discharge 2 0.
           (tn'/s)
                       24     48    72     96
                           Time - Hours
Figure 6. Isotopie  composition of discharge at Orangeville  Rise,
          October 4-8,  1990.
discharge and  Initial  discharge recession  (Figure 6).   Oxygen and
deuterium shifts mirrored one another during a period of rapid
isotopic fluctuations  between 18 and 40 hours,  when two pulses of
Isotopically light water  discharged at the Rise.  During much of
the discharge recession,  isotopic composition remained steady,
and approached that of baseflow values.

April 1991 Storm
     Chemical variability of  discharge was monitored during a
second storm event in  the spring,  1991 (Figure 7).  Because of
frequent rainfall in the  winter and spring, discharge at
Orangeville Rise does  not achieve baseflow conditions.
Therefore,  to establish pre-storm conditions for this sampling,
Orangeville Rise was monitored for 6 days prior to the storm
event.   Major-Ion concentrations and isotopic composition of
discharge remained fairly steady at the values illustrated  in
Figures 8 and 9 for time  equal to zero-hours.
     As in the October storm,  precipitation was collected at  two
locations within the basin.   Again,  the amount and isotopio
composition of the two rainwater samples were similar,  and  were
averaged.   Rainfall history during this event was much more
complicated than the October  storm,  and is outlined in  Table  2.
Because of the complex rainfall history,  discharge response was
significantly different from  that observed in October  (Figure 7).
                                275

-------
 Table 2.
Isotoplc  composition of rainfall,  April 12-15,  1991,
Orangevllie  Rise basin.

Date Time
4-12 / 1930
4-1 2/2330
4-13/0430
4-14/0645
4-15 /0500
Amount
(mm)
3
7
19
3
6
•ID
(X.)
-12
-18
-7
-9
-45
i'«0
<%.)
-2.6
-3.4
-2.3
-2.7
-6.8
Volume Averaged
iD
«.)
-12
-16
-12
-II
-16
SO
(X)
-2.6
-3.4
-2.7
-2.7
-3.3
                                            _3 "1         3 -I
      Discharge increased  rapidly from 1.6 nrs  ^o 42 ms   In 12
 hours,  in response to 29  mm of  precipitation. During the next 24
 hours discharge leveled off,  then rose again following  a short
 rainfall,  to a maximum of 4.9 ms".   After  the peak, discharge
 slowly decreased, following an  irregular linear recession.
      Behavior of major-ion  concentrations during this storm were
 markedly different from that observed during the October
 monitoring (Figure 8).  Concentration of several ions increased
 during an increase in discharge.   Calcium and magnesium increased
 by 5 and 10% respectively during the first 6 hours, while sodium
 fluctuated between 8 and  12 mg/1 during the first 25 hours.
 Bicarbonate remained constant during the first 12 hours of
 monitoring when discharge increased from 1.6 ms   to 3.2 ms  ,
 and after an initial decrease,  sulfate concentration increased to
 a maximum at 18 hours.  There was also a Jump in ionic
 concentrations at 25 hours  when Ca , Ha* and SO,~ increased by 13
 to 31% from the previous sample.   Minimum specific  conductance
 and ionic concentrations occurred 4 hours after peak discharge.
 Concentrations increased  from this minimum, however the trend was
 marked by erratic fluctuations.
                5.0-

                4.0-

         Dlscharge   3.0 -
          (nrVs)
                2.0-

                1.0-
11
1
i

B


._ ~0<5o_
• 0
• 5
• 10
• 15
I- 20
                                             Precipitation
                  ^   i  I	1	1	1   I	1	r—i	1	1—
                   0  12  24  36 48  60  72 84  96  108 120 132
                                Tims - Hours
                                                  144
Figure 7- Storm hydrograph and rainfall  history for Orangeville
          Rise,  April  12-18,  1991.
                                276

-------
            Discharge
                     2.0-
                                     48     72      96
                                         Time - Hours
                                                          120
                                                                    r 4-10
                                                                    •••too
                                 •360

                                        Specific
                                 ^320  Conductance
                                         (uS)

                                 -200
                                                                  T-L240
                                                                  144
                    5.0-
           Discharge
            (m3/s)
                                        Gallons
                             24
48     72     96
    Time - Hours
                                                                 40
                                                                     20
                                                                    -10
                                     Mg2', Na1
                                       (tnoyl)
                                                              144
                                                              HCOi
            (m3/s)
                                         Anions
                                   48     72     96
                                       Time - Hours
                   120    144
                                 -39
                                 -33
                                 -27
                                 ^21
                                 -15
                                 r9
                                 -3
SO4=
Oa", ci
(tng/l)
Figure  8.  Chemlonl  data  fTom Orangoville  RlRe,  April  12  18,   1991;
              A>  Bpoclfic  conductance,  B)  cation  eoncentrationR,   and
              C)  anion  concentrations.
                                             277

-------
        Discharge
         (m3/s)
5.0-

4.0-

3.0-

2.0-

10
                                 Average 518ORa|n- -3.3%o
                                      Discharge
                    —I	1	1	r
                    24    48
                              72

                          Tims • Hours
                                    96
                          —I—
                           120
                                144
 Figure 9.  Isotopic composition  of  discharge at Orangeville Rise,
           April 12-18,  1991.
     Isotopic composition of discharge Just prior  to the onset of
rainfall was -44 foe, and -6.57oofor ^D and i °O   respectively
(Figure 9).  During the first 12 hours, when  discharge increased
dramatically, & D  increased by only 2 7oo and  &   O increased by
only Q.2%o.  During the next 24 hours when discharge increased
only slightly,  the  isotopic shift was more rapid.   The heaviest
discharge occurred  from 62 to 65 hours, 24 hours after peak
discharge, when  & D reached ~307o0 and  & O climbed to -4.5%o.
Isotopic composition of discharge became lighter during the
remainder of recession but did not obtain values seen in pre-
storm discharge.
Hydrograph Separation
     As a means  of  investigating components  of  discharge at
Orangeville Rise, storm hydrographs were separated into pre-storm
and rainwater components.   Separations were  performed using both
deuterium and oxygen isotopic data as a means of checking for
consistency in the  data.
     Hydrograph  separation of the October storm is shown In
Figure 10 along  with the equation used to solve for instantaneous
rainwater discharge.   At peak discharge, rainwater comprised 12
to 15% of the total discharge.  The greatest percentage of
rainwater at any time on the hydrograph occurred 18 hours after
peak discharge,  where it made up 39 to 46%  of the total.
Separation revealed that a secondary pulse  of pre-storm water
                                278

-------
               Hydrograph Separation based on Deuterium Isotope Data
   Discharge
     (rrrVs)
            3.0-
            2.0'
            1.0'
                                    Q,-^ ^-40)
                                              16
                       24        48       72

                             Time - Hours
96
   Discharge
     (m3/s)
            3.0 H
            2.0 A
            I.CH
                Hydrograph Separation based on Oxygen Isotope Data
                                                 6.2)
                       24      48       72

                            Time - Hours
96
Figure  10.  Hydrograph separation of  discharge
              from Orangeville  Rise into pre—
              storm  and  rainwater components,
              October 4-8,  1990.
                                                                                  Hydrograph Separation based on Deuterium Isotopic Data
                     Discharge
                       (m3/s)
                     Discharge
                       (m3/s)
                                                                                                                   (5DM - 44)
                                     24
                                                                     144
                                                                                              Time - Hours
                                                                                   Hydrograph Separation based on Oxygen Isotope Data
                                     24
                                                                     144
                                                                                              Time - Hours
                    Figure  11. Hydrograph separation of  discharge
                                 from Orangeville  Rise into pre-
                                 storm and rainwater  components,
                                 April 12-18,  1991.

-------
arrived  just  after the time of maximum rainwater  contribution.
For the  entire five  days  of monitoring,  rainwater made up 20 to
25% of the total  discharge at Orangeville Rise.
     Separation of the April,  1991 storm hydrograph into pie-
storm and rainwater  components is shown in Figure 11.  The
volume-averaged isotopic  signatures of rainwater shown in Table 2
were used to  calculate the rainwater component of discharge at
any one  time  (McDonnell,  et al.,  1990).
     Hydrograph separation shows that at peak discharge, the
rainwater component  made  up 27 to 33% of the total discharge.
Twenty-four hours after peak discharge,  the rainwater
contribution  to total  discharge was at a maximum and equaled 47
to 52%.  The  proportion of rainwater decreased from this point to
a very small  to non-identifiable component,  110 hours into
monitoring.   As in the October storm, rainwater made up 20% of
the total discharge  during the six days of monitoring.  This
separation also showed pulses of pre—storm water, but the pulses
.were more subdued than that observed in the October hydrograph.

DISCUSS I OF and COUCLUSIOJTS
     Vhile discharge increased rapidly in both storms, ionic
concentrations decreased  only moderately or even increased as in
the April storm,  indicating that rainwater could not have
contributed to a  majority of discharge at peak flow.  Isotopic
data also indicates  that  there was not a large rainwater pulse
arriving at peak  discharge.   Hydrograph separations reveal that
rainwater made up only a  small proportion of discharge at peak
flow, and the largest  proportion of rain water contribution to
instantaneous discharge occurred 18 to 24 hours after peak flow.
     If  the observed isotopic composition of discharge at
Orangeville Rise  were  a product of simple mixing between two end
members, namely rainwater,  and pre-storm water identified by the
isotopic composition of discharge at baseflow, then all data
should fall on a  line  connecting the two extremes.  Figure 12
shows the global  meteoric water line (Craig,  1961) and the
isotopic composition of rain samples collected in the Orangeville
Rise basin.   The  average  ground water and baseflow also fall on
the meteoric  water line,  and hence if two component mixing were
responsible for shifts in the isotopic composition of discharge,
then all data would  lie along the line.   Data from the October
and April storms  do  not fall on the meteoric water line,  but
rather define another  unique line.   These results identify an
element  of complexity  not accounted for in the simple two
component mixing  model.   Since the isotopic composition of all
phreatic waters in the basin lies on the meteoric water line, one
may be compelled  to  look  to the vadose zone as a source of water
which could cause discharge to deviate from the meteoric water
line.
     The rapidly  increasing discharge indicates that there must
be substantial concentrated recharge through sinkholes, and rapid
transmission  of water  through soils via macropore flow.  Large
volumes  of water  must  flow quickly through conduits to discharge
at Orangeville Rise, however,  this water cannot be rainwater.
Rather "old"  water must be displaced by new water.  The storage
                               280

-------
          -50
          -80
                             Rain : April, 1991
                   5D- 85"O + 10
                                     5D-55"0-9
               ' Rain : October. 1990
• Rain
i Orangeville Rise: October. 1990

o Orangevilla Rise: Apnl. 1991
+ Basetlow
* Average Ground Water
                                       •2
 Figure  12.  Isotopie data from Orangeville  Rise and rainwater
            collected within the basin,  relative to the meteoric
            water line of Craig, 1961.
compartments which  could respond most quickly  would be
which contain water stored in conduits,  or contain  water that is
in direct contact with conduits,  including epikarst storage,
vadose conduit storage and phreatic conduit storage.   An influx
of rainwater through concentrated recharge could be quickly fed
to the epikarst, increasing hydraulic heads there,  and causing a
rapid discharge of  water stared within the epikarst to
Orangeville Else.
     The Increased  hydratilic heads associated  with  recharge will
also result in an increased flow of water through more diffuse
compartments of storage.   However,  transmissivity of the diffuse
stores are orders of magnitude lower than that associated with
conduit flow.   It is difficult,  therefore, to  see how diffuse
water could arrive  at the Rise ahead of the water transmitted via
conduit flow.   As a result,  we believe that epikarst and conduit
water contributed to the bulk of discharge at  peak  flow,  and
diffuse stores of water  may be responsible for the  pulses of pre-
storm water seen during  discharge recession.
     The rain of October 4  broke drought conditions in the basin.
Because it had been so dry,  vadose—s?one water  storage was at a
minimum,  which can  be seen In the rapid rate of discharge
recession.   The April monitoring occurred at a time of high
recharge to the ground water systems.   Water storage compartments
contained large volumes  of water,  as can be seen in the slow
discharge recession.  Ion concentrations remained constant,  or
increased with initiation of storm flow, and may represent the
flushing of waters from  epikarst and/or conduit storage.
     It is interesting to note that in both storms,  the tjme of
                               281

-------
minimum epecific conductance  did not correspond to the time of
maximum rain water discharge  as identified by the isotopic data.
Discharge at any moment of time is a product of all the
contributing sources.   It  is  possible that at the time of maximum
rainwater arrival, water was  also arriving from sources with
higher than average  total  dissolved solids, the resultant
discharge having a greater conductivity than might be anticipated
for the time of peak rainwater discharge.

Aknowledements
     This study was  funded by the American Geophysical Union,
Horton Research Fellowship.

References Cited
Atkinson, T.C., 1977, Diffuse flow and conduit flow in limestone
     terrain in the  Mendip Hills,  Somerset (Great Britain); Jour.
     of Hydrology, v.35, p.93-110.
Bakalowicz, M. ,B. Blavoux, and A.  Mangin,  1974, Apports du
     tracage isotopique nature1 a la connaissance du
     fonctionnement  d'un de trois systemes des Pyrenees France*
     Jour, of Hydrology, v.23,  p.141-158.
Bassett, J.L., 1974, Hydrology and geochemistry of karst terrain,
     upper Lost River drainage basin,  Indiana;  M.A.  Thesis,
     Dept. of Geology,  Indiana University, 102p..
Craig, H., 1961, Isotopic  variations in meteoric waters;  Science,
     v.133, p.1702-1703.
Dreiss, S.J., 1989,  Regional  scale transport in a karst aquifer
     1. Component separation  of spring flow hydrographs;  Water
     Resources Research, v.25,  n.l,  p. 117-125.
Fritz, P., J.A. Cherry, K.U.  Veyer , and H. Sklash, 1976,  Storm
     run-off analysis using environmental  isotopes and major
     ions; Interpretation  of  environmental Isotope and
     Hydrochemical Data in Groundwater Hydrology,  International
     Atomic Energy Agency, Vienna,  p.111-130.
Gunn, J., 1983, Point recharge of limestone aquifers - a model
     from Hew Zealand karst;  Jour,  of Hydrology,  v.61, p. 19-29.
Krothe, H.C., and R.D. Libra,  1983,  Sulfur isotopes and
     hydrochemical variations of spring waters of southern
     Indiana, USA; Jour, of Hydrology,  v.81,  p.267-283.
Libra, R. D. ,  1981, Hydrogeology and sulfur isotope variation of
     spring systems; south-central Indiana; M.A.  Thesis,  Dept. of
     Geology,  Indiana University,  109p.
Mangin, A., 1974, 1975, Contribution a 1*etude hydrodynamique des
     aquiferes karstiques. DES thesis,  Univ.  Dijon,  France;  (Ann.
     Speleology, 1974, v.29,  n.3,  p.283-332;  v.29, n.4, p.495-
     601; 1975, v.30, n.l, p. 21-124.
McDonnell, J.J., M.  Bonell, M.K.  Stewart,  and A.J. Pearce, 1990,
     Deuterium variations  in  storm rainfall:  Implications for
     stream hydrograph separation;  Water Resources Research,
     v.26, n.3, p.455-458.
Milanovic, P.T., 1976, Water  regime in deep karst. Case study of
     the Ombla spring drainage area; in Karst Hydrology and Water
     Resources, v.l, Karst Hydrology,  V.YevJevich (ed.),  Fort
     Collins, Colorado, Water Resources Publication, p. 165-191.
                                282

-------
Murdock,  S. H. ,  and R.L. Powell,  1968,  Subterranean drainage
     routes of Lost River, Orange County,  Indiana;  Indiana
     Academy of Science Proceedings  n.77,  p.250-255.
Rune,  R.V.,  1975,  Geohydrology of karst  terrain,  Lost River
     watershed, southern  Indiana;  Indiana  University Water
     Resources Research Center,  Report of  Investigation,
     n.7, 91p.
Sheppard, S.F.M.,  R.L.  Nielsen,  and  H.P. Taylor,  1969,  Oxygen and
     hydrogen isotope ratios of  clay minerals from porphyry
     copper deposits; Economic Geology,  v.64,  p.755-777.
Shuster,  E.B.,  and V.B. White, 1971,  Seasonal fluctuations in the
     chemistry of limestone springs:  A possible means of
     characterizing carbonate aquifers;  Jour-  of Hydrology, v.14,
     p.93-128.
Sklash.M. G. ,  and R.IT. Farvolden, 1979, The role of groundwater in
     storm runoff; Jour,  of Hydrology, v.43,  p.45-65.
Torbarov, K. ,  1976, Estimation of permeability and effective
     porosity in karst on the basis  of recession curve analysis;
     in Karst Hydrology and Water Resources,  v.l,  Karst
     Hydrology, V.YevJevich  (ed.), Fort       Collins, Colorado,
     Water Resources Publication, p.121-136.
Tweddale, J.B. , 1987, The relationship of  discharge to
     hydrochemical and sulfur isotope variations in spring waters
     of south—central  Indiana; M.S.  Thesis,  Dept.  of Geology,
     Indiana University,  165p.
Wells, E.R. ,  and N.C. Krothe, 1989,  Seasonal fluctuations in ^15IT
     of ground water nitrate in  a mantled  karst aquifer due to
     macropore transport  of fertilizer-derived nitrate; Jour, of
     Hydrology, v.112,  p.191-201.
Williams, P. W. , 1983, the role of  the subcutaneous zone in karst
     hydrology; Jour, of  Hydrology,  v.61,  p.45-67.
                               283

-------
Gruver - Stable Isotope Separation

1.     I want to congratulate the speaker for her particularly thoughtful
      discussion, and her stimulating suggestions.  The stable isotope records
      presented are suggestive of piston-flow (at least in part).  Piston-flow
   -  response is difficult to reconcile with the suggestion that the source of the
      expelled (pre-storm) water deviating from the meteoric water line in the
      epikarst. Please comment.

2.     Can you suggest a method for spiking clean injection water with heavy
      oxygen or deuterium in order to perform a tracer test with stable isotopes?
      This could be useful in areas where, for political reasons, it was not
      practical to use dyes in a karst aquifer.  Do you know of such studies?
      Would light water be more easily obtained? What is the cost per liter (or
      kg) for heavy water and light water? Having bought same, how much of it
      is ordinary water? (This is asked in order to get an idea of how much
      tracer might be heeded for ah "average" trace and what the cost would
      be.)

3.     The delta 1 &O  - delta D relations of the spring response suggests a third
      hydrological component, presumably in the vadose zone. However, the
      end member occupied by the two storms are different, although both lie  on
      the "evaporite line."  Why are they different, and why are they displaced to
      the left and right of the "meteoric water line" for the two storms?
                                     284

-------
Answer to  Questiontfl:
       Water  collected  during baseflow at  Orangeville  Rise, and  ground waters
sampled from domestic  water wells lie  on the meteoric  water line  (MWL).   As  all
phreatic and  rain  water samples  lie on  or near the MWL,  one may need  to  look to the
vadose zone  for a source of water that could cause discharge to shift from  the  MWL.
       The first few  samples collected  during each  storm  event  lie  very  close  to  the
MWL.  I  believe  these  samples represent displaced  phreatic  conduit water.   Six to
twelve hours  into storm  flow,  the isotopic  signature of  discharge begins  to deviate
from  the  MWL, as well as  in  the direction  of the  rainwater isotopic signature.   It
seems reasonable   to interpret these waters as  rain  and  epikarst  water  that has  been
rapidly   transmitted  via  shaft, and conduit flow.   Very little work has been done on
chemical  processes  within   waters  of  the  epikarst,  or what fractionation  processes
might  occur  there.    This  study points  to  the  need  for research  which  addresses  these
questions.

Answer to Question #2:
       I  am unfamiliar  with any  investigation  which has used "light"  or  "heavy"
water  in  place  of traditional dyes in karst  aquifer tracing.   If one  feels  compelled to
use this  technique, however, they  would be  advised  to use  a water  tracer  that has the
greatest isotopic difference  from that of the natural  system.     The  amount  of  "spiked"
water  used would be dependent upon the volume of  water in the  natural system.   I
could not  begin to  guess  the price of "spiked"  water.    I can say  that a  large  expense
in a stable isotope study is the  cost of  analysis.   Count  on at least $35  per sample  for
deuterium  analyses,  and  $45  for  oxygen  analyses.

Answer to Question #3:
       The delta  ISO-delta  D relationships  suggest  at  least  a  third  component
contributing to  total discharge.   I  would suggest  that, while the data  seem  to  lie  along
a  common line with  a  slope of 5, upon  closer examination,  the October  data lie along a
line  with  slope 4, while the  April  storm lies  along a line  with slope  6.   I believe  that
the cause  of the  shifts  for  each  storm is  different.   Fractionation  processes acting in
the dry  fall-time,  compared  with  those of the spring  may  be different  and  may be
responsible for  shifts in  the  isotopic  signature  of  discharge in  opposite  directions
from  the  meteoric water  line.
                                         285

-------
286

-------
      The Transmission of Light Hydrocarbon Contaminants

                 In Limestone (Karst)  Aquifers
       Ralph  0.  Ewers  Ph.D./  Eastern Kentucky University
               and Ewers Water Consultants Inc.
                Anthony J. Duda, Dames & Moore
       Elizabeth K. Estes M.S., University of Minnesota
      Peter J. Idstein M.S., Eastern Kentucky University
          Katherine M.  Johnson,  U.  S.  Army (USATHAMA)
                           ABSTRACT
  Hydrocarbon contaminants  with densities less than water  may
enter karst aquifers through soil percolation, directly by  way
of surface runoff into open  sinks,  or by  a combination  of  these
two  routes.    Once  in  the aquifer,  the  movement of   these
hydrocarbons is  governed by  four factors:  (1)  the  level of  the
principal  dissolution  conduits   relative  to  the   zone   of
saturation, (2)  the flow regime (turbulent or non-turbulent)  in
the conduits which  contain  the contaminant,  (3)  the  vertical
complexity of  the  framework  of  dissolution  porosity in  the
aquifer,  and (4) the nature  of  the recharge to the aquifer.
  Field investigations indicate that  light hydrocarbons can move
in these  aquifers  as gross  concentrations  of  free product  at
speeds of the  order  of  kilometers  per  hour.    Under   other
circumstances, the contaminants may be trapped in  the aquifer so
completely that  spring water originating from the  point  of  gross
hydrocarbon release shows none  of the contaminant.
  Trapping of the hydrocarbons occurs so  long as two conditions
exist.   First,  the  active  conduits  must remain,  to  a  large
extent,  in a completely water-filled  (phreatic) condition.   If
a free surface  (an  air-water  interface)  develops throughout  a
large percentage of the active conduits the  floating fluid  may
move as "free product".  Second, the groundwater velocity in  the
vicinity   of  the  floating product  must   remain  slow.    Gross
turbulence entrains  the "free product" in the swift  flow of  the
active conduits.
                         INTRODUCTION

  Contaminant movement in karst aquifers often appears  strange
and erratic  to  hydrogeologists  more  familiar  with granular
                             287

-------
aquifers.  Light hydrocarbon contaminants, those with densities
lower than water,  appear especially capricious.  The behavior of
these materials in karst aquifers is related to the dissolution
porosity,  often referred  to  as conduit  porosity, which they
possess.

   Dissolution  porosity  in paleozoic  carbonate rock  normally
occurs  along  partings  such  as bedding  planes,  joints,  and
faults.   In  areas of flat lying or gently  dipping carbonates,
conduits formed along bedding planes are commonly most important
in  forming  the master  conduits  which  permit  long  distance
transmission of groundwater.  Joint and fault related conduits
provide routes for admitting meteoric water  to the aquifer.  The
total length of explorable conduits primarily guided by bedding
planes in such areas commonly exceeds the  total length of those
primarily  guided  by  joints by  10 to  1  (Ewers,  1972).    On  a
worldwide basis, which includes karst aquifers in dipping rocks,
the  ratio of   bedding  plane to  joint  and  fault  controlled
conduits is of the order of 1.3 to  1 (Palmer, 1991)  This is the
case,  not only because bedding  planes  commonly  extend  over
greater horizontal distances than joints,  but also because they
can transmit water in any  direction, unlike joints and faults.

   Unlike the  diffuse,  dispersive,  laminar  flow  of  granular
aquifers,   groundwater   movement   in   karst   aquifers   is
concentrated,  convergent  toward  the  conduits,  and  at  times
turbulent  (Quinlan  and  Ewers,   1985).    Surface  water  and
contaminants may  have direct  access  to the conduits by  way of
sinkholes with  swallet openings and with  sinking streams.   The
presence of conduits may permit rapid  movement  of  liquid phase
hydrocarbons.   Because  of the turbulent  flow regime  in these
conduits, contaminant movement may  be in the form of globules of
entrained  "free  product"  as  well   as  a  dissolved  phase.
Groundwater  and contaminant velocities in  these  conduits  are
commonly  as  high  as  several  kilometers  per  day  (Quinlan  and
Ewers, 1983).
    FACTORS AFFECTING LIGHT HYDROCARBON MOVEMENT  IN KARST

   The  movement of  immiscible floating  hydrocarbons  in  karst
aquifers   is   controlled   by   factors   relating   to   the
characteristics of the framework of dissolution porosity and to
the type of flow regime found in these conduits.  These factors
can be conveniently grouped into the following four categories.

1.- The level of the principal conduits relative to the zone of
saturation
   Conduits in karst aquifers occur deep within the phreatic zone
(the zone of saturation beneath the potentiometric surface) and
well above  normal  levels  of groundwater  circulation.   Their
range in a specific area is related to the thickness of soluble
carbonate  rock,  the  local  geomorphic history,  and  climatic
factors.   Light hydrocarbon  contaminant transmission  is less
                             288

-------
likely in master conduits  located deep in the phreatic zone than
in those which are along the upper part of this  zone.

Phreatic Conduits
  Where   the  master   conduits  responsible  for   horizontal
transmission of groundwater lie permanently within the phreatic
zone, floating hydrocarbons should not move  easily.   The entry
point for the contaminants should form at least a temporary trap
for this free product at the top of  the phreatic zone (Fig. 1,
point "A").  Likewise,  any other portion  of  the  master conduit

  Hydrocarbon Traps
                   Water Table

                      i    ^  Jo
                       Master
Conduit
                            T
   Free Hydrocarbons
  FIGURE 1  Light hydrocarbons can be trapped above phreatic
  master  conduits  in  ("A")  aquifer  entry  points,  ("B")
  solution widened  joints,  ("C") horizontal  conduits above
  bedding  planes,   and  ("D")  the  soil   filling  of  grikes
  (solution widened  joints)  and  in  the  soil around nearby
  clints (bedrock pinnacles).

which extends above  the  potentiometric surface should  form an
additional trap for any free product which becomes entrained in
the flow of the conduit,  only that  portion  of  the hydrocarbon
which  dissolves  in  the  groundwater  should  move  easily  in
phreatic conduits.
                             289

-------
Epiphreatic Conduits
  Where  the master conduits  are  in the epiphreatic  zone (the
zone in the immediate vicinity  of  the potentiometric surface),
floating hydrocarbons tend to  move  more freely.  Under base-flow
conditions,  some  master  conduits   will   have  an  air-water
interface  along  their   entire  length.    This   provides   a
continuously sloped surface for the  movement of floating "free
product", identical to a surface stream.   The hydrocarbons will
move  at  a  speed  nearly  the  same  as  the  water  in  this
circumstance.   More commonly,  epiphreatic conduits  have air-
water  interface conditions spaced intermittently  along their
length.  To the degree that this interface diminishes, floating
hydrocarbons will move with increased difficulty, their mobility
becoming more  and more dependant  upon the nature  of  the flow
regime in the conduit.


2.-  The  flow  regime  in the  conduits   which  contain the
contaminant

Turbulent Flow
  During periods of turbulent flow, floating hydrocarbons would
be readily entrained in conduit water.   Groundwater velocities
in typical karst  aquifers  range from 30  ft/hr  at  base  flow  to
1300 ft/hr  in  periods  of  high flow  (Quinlan and Ewers,  1985).
The  hydrocarbons,  normally  excluded   from  phreatic  master
conduits, could be mobilized by the turbulence and would travel
at nearly the velocity of the water.

Laminar Flow
  When laminar flow prevails,  floating  free  hydrocarbons are
unlikely to be  entrained in the groundwater.  The "free product"
will separate from the water  and collect at the highest points
along the conduit ceiling.  Diffusion of the hydrocarbons into
the water  will occur, but the  concentrations  of  contaminants
carried in this way may be very low.  The  large quantities  of
groundwater  moving   in   the   conduits   may   limit   these
concentrations  as well as the relatively high  octanol-water
partition  coefficients  for many   of  these  substances.    Both
trapping of  the  floating  free product  and  diffusion  of the
hydrocarbons into the water  are   dependant upon  the  vertical
complexity of the conduits.


3.- The  vertical complexity  of the  framework of dissolution
porosity in the aquifer

  Vertical conduits in carbonate aquifers have two roles in the
transmission of light hydrocarbons.   They  provide entry routes
for these contaminants and they also provide traps for the free
product.   These  conduits  together  with abandoned horizontal
conduits are collectively  referred to as the epikarst (Mangin,
1974-75).  This term  is used  here in a  broad  sense to include
any solution  porosity  lying above  the  level  of the  active
                             290

-------
drainage trunks, including that which may be  above  or below the
normal potentiometric surface.  Karst aquifers with an extensive
system  of  interconnected  vertical  and  horizontal  phreatic
conduits above deep phreatic  master conduits have  an increased
ability to trap light hydrocarbons and a reduced likelihood of
transmitting  these  contaminants.    Those  with  simple  shallow
conduit systems are more prone to  contaminant movement.

  Hydrocarbon Entry
             Sinkhole
      With Swallet Opening
                     Leaking Underground
                        Storage Tank
With Cover of Residuum
       Water Table

     	t	
  FIGURE 2  Light  hydrocarbons can  enter karst aquifers in
  several ways.   They may  enter  directly with  concentrated
  surface recharge entering swallets (point  "A"),  by seepage
  through sinkhole bottoms  covered by residuum  (point "B"),
  or through the soil cover  and minor solution widened joints
  in the carbonate rock  (point "C").

Hydrocarbon Entry
  Hydrocarbons  may  gain  entry  to  a  karst aquifer through
sinkholes with  open  swallets  which  give direct access  to  the
conduit system or by  seepage through the residuum  in  a sinkhole
bottom (Fig.  2,  points "A" and "B").   In  either case,  a vertical
conduit, typically developed  along a  joint,  provides  both  the
means  for  the sinkhole  to  form  and  the  entry route for  the
contaminants.
                              291

-------
  A  much  larger  number  of vertical  epikarst  conduits  are
associated  with  features  called  grikes.    These  are  linear
depressions in the bedrock surface with narrow bottoms and wide
tops which develop along joints.   They  can conduct hydrocarbon
contaminants  from the  bottom of  the soil  horizon  to  master
conduits in the  phreatic or epiphreatic  zone.   These conduits
are often common  in  areas  where no depressions  are visible at
the surface of the overlying soil.


Hydrocarbon Trapping
  Master conduits are  frequently intersected by joints.   Where
these  joints  are enlarged by dissolution,  they form traps in
their phreatic portions for floating hydrocarbons (Fig. 1, point
"B").   Vertical  conduits  of  this  type  intersecting  master
conduits which are accessible to direct observation frequently
have volumes  in  excess of 1000  gallons.   Where the  carbonate
rock is heavily jointed, these potential trapping situations are
seen to occur at intervals of a few meters or less.   Thus,  they
may  provide  a  considerable  storage  capacity  for  floating
hydrocarbons.

  Vertical  conduits  may connect the master conduit  with  higher
horizontal  conduits  in the  phreatic  portion of the epikarst
(Fig.  1,  point   "C").   These  conduits  are  often  laterally
discontinuous due to  truncation  by soil filled grikes.  Conduits
of  this  type   may   provide  additional  storage  for   light
hydrocarbons.

  Vertical  conduits  may extend from the master conduit  to  the
upper bedrock surface and  the base of the  soil.   In such cases
the hydrocarbons  may collect at the soil bedrock interface if
the  potentiometric  surface  reaches  above  this  level  either
permanently or intermittently (Fig. 1, point  "D").

  Where the interval between the level of the master conduit and
the level of the uppermost  phreatic part of the vertical conduit
elements is great, the groundwater in  the master conduit  may be
in poor communication with  the floating  contaminants.    Both
entrainment of the hydrocarbons  by master conduit turbulence and
their diffusion into this flow will be very limited.


4.- The nature of the recharge to the aquifer

  Two  basic  types of  recharge occur  in carbonate  aquifers,
diffuse recharge and concentrated recharge.  The nature of this
recharge has  a significant effect upon  the movement of light
hydrocarbon contaminants.   Diffuse recharge which passes through
a soil  cover  and  enters the  aquifer through  a  large  number of
joints is less likely to produce gross turbulence in the aquifer
conduits than concentrated recharge.  This  latter  includes  the
concentrated runoff which enters swallets in sinkholes and both
                             292

-------
allogenic and  autogenic sinking  streams.   Concentrated  storm
runoff from  these  sources regularly  produce  stage changes  in
conduit systems  of 30  feet,  and  changes  of 75  feet are  not
unusual.   Flows of  these magnitudes produce gross  turbulence in
the conduit  systems  which propel  coarse  gravel  from  spring
orifices.  A modest  amount of turbulent recharge  entering  the
top  of  a   free   product  trap  may  entrain  the   collected
contaminant.    Similarly,  hydrocarbons  trapped close  to  the
master conduit may be mobilized  by master conduit turbulence
induced by the storm flow of a sinking stream.

   Karst  aquifers  with thick  soil  covers,  few open sinks,  and
without  sinking   streams  have   a   reduced  likelihood   of
experiencing movement of trapped light hydrocarbon contaminants.
Those with the opposite characteristics  are at  higher risk  for
contaminant movement.

     CASE STUDY lr  AN AQUIFER  WITHOUT CONTAMINANT  MOVEMENT
               Quarles Spring Groundwater Basin

   Quarles Spring, located in Christian County, Kentucky,  is used
as a potable and irrigation water  supply.  Dye tracing by Ewers
and his students (Ewers  Water Consultants, 1989;   Carey,  1990;
Greene,  1990)  has  shown  that  this  spring  is  the  principal
discharge point for a groundwater  basin which  includes Campbell
Army Airfield (Fig. 3).  Jet fuel  spillage  at the  airfield  has
appeared in several wells in the karst aquifer at the  airfield.
As  much  as  16  feet  of  "free  product"  has  been  measured  in
monitoring well  MCI-2  (Point  "C", Fig.  3;  Fig.   4)  by  Ewers
(Ewers Water  Consultants, 1991).    Dames & Moore  report free
product  in   four   other  wells   (Dames  &  Moore,   1991).
Investigations by  Duda  (Dames  & Moore, 1991)  suggest that  the
fuel entered  the aquifer through a sinkhole which takes drainage
from the airfield  (Point "B", Fig.  3).   Repeated sampling  of
Quarles Spring by Duda (Dames  & Moore, 1991) has failed  to show
jet fuel, dissolved  or free.    Thus,  it  appears that floating
hydrocarbons are not mobile  in this aquifer.  The known  and
inferred  aquifer   characteristics  fit  the  criteria  for  low
mobility outlined above.

The principal conduit system-
   Quarles  Spring  possesses   a deep  rise-pool,  suggesting  a
phreatic condition in the master conduit which  is  tributary to
it.  Information from Well MCI-2 (Fig. 4)  at the airfield gives
further evidence that the conduit system  is well within  the
phreatic zone.  This well intersects a solution conduit at about
459 feet msl.  This conduit has a  minimum vertical  extent of 16
feet  (MCI,  1987),  and  is a possible source  of  the  jet fuel
contaminant which is  seen in the annulus of this well.  The well
bore entry point into  the conduit is approximately 29 feet lower
than  the normal  level  of Quarles  Spring,  well  within  the
phreatic zone (Table 1; Fig.  4).
                             293

-------
Quarles Spring Groundwater Basin
   Quarles Spring
                      ~r - '.n
                         *
                      55
    FIGURE  3   Quarles Spring is the principal discharge point for  a  groundwater
    basin which includes Campbell Army Airfield.  Jet fuel spillage at the airfield
    has appeared in  several wells in the karst aquifer.  Point "C"  is  well MCI-2.
    Points  "B", "C", and "H" are the location of dye inputs.  The tracer dyes were
    recovered  at Quarles Spring, number 55.  No jet fuel has been  recovered from
    this spring.

-------
         Vertical Section - Quarles  Spring Basin
    Elevation MSL
    57 0\-
                                       Well I
              Airfield Elevation 559 msl.
    520
    470
N5
VO
    420
Quarles Spring
   488 \rnsl.
                               /Product reported by
                              / .  Dames & Moore
Product Level 6/90
 494' msl. (approx)
                                                   Water Interface 6/90
                                                            478' msl.
                                                   Cavity
                                                  j  Top 459'- Bot. 443'
                                                      (width inferred)
                  Surface Elev

                  Open Hole
                                Bedrock Interface

                                Conduit
   Casing
   Free JP-4
                                    0000 Epikarst (position inferred)
            FIGURE 4  This schematic vertical section between well MCI-2 (point  "C", Fig.
            3) and Quarles Spring is based upon information from 37 wells and numerous soil
            investigations.   The  conduits shown are  inferred,  except for  that  in  the
            vicinity of well  MCI-2.  The typical and extreme  range of the  soil bedrock
            interface is shown.

-------
  Table  1  ELEVATION OF FEATURES RELATED TO MONITORING WELL
   MCI-2,  THE  BEDROCK,  AND QUARLES SPRING

  Well Features      Elevation (ft.)  Bedrock & Spring Features

  Top of Casing	    560.16
  Ground Level	    559.65
                                      Bedrock/Soil Interface-
                            534  	 Shallowest (MCI, 1987)
  Product  Level  6/90	    494
                            491  	Floodplain at Quarles Spr.
  Water  Level  Without-
         Floating JP-4	    490.25
  Bottom of  Casing	    489.74
                            488  	Water level in Quarles Spr.
  Water  Interface 6/90—    478
  Top of Cavity	    459
  Bottom of  Cavity	    443
                                      Bedrock/Soil Interface-
                            433  	Deepest  (MCI, 1987)


  The conduit  intersected by this well was shown by dye tracing
to be connected to Quarles Spring  (EWC,  1989).   It  is  reported
to have significant amounts of fine sediment moving through it.
The driller  recorded that  a five  foot  layer of mud was  swept
into this conduit during a  severe rainfall event which occurred
during the installation of  the well (MCI, 1987).  We may assume
that this  conduit  is   a part  of  the  active conduit  system,
tributary to Quarles spring.  If this conduit were significantly
lower than the active conduit  system and if  it were  not a part
of that system, then it would probably be sediment filled.  The
groundwater gradient to Quarles Spring is very  low.   The head
loss from the well to the spring is approximately 2 feet at base
flow (Table 1)  over a distance of  11,800 feet,  a typical value
for a phreatic conduit.  These observations suggest  that  the
conduit system beneath  Campbell Army Airfield and tributary to
Quarles Spring is phreatic  in nature, even under extreme low-
flow circumstances.

The flow regime and the aquifer recharge-
  Flow in  the master conduits of the Quarles Spring groundwater
basin are  almost certainly turbulent much of  the  time.   The
spring  discharges   water   even  when   heavy  withdrawal  for
irrigation occurs during very dry periods.  Base flow discharge
is estimated at  0.5  ft3/sec.   However,  it discharges primarily
fine sand-sized sediment and is typically quite clear.   It does
not have the flashy response and wide range  of flow velocities
exhibited by many karst aquifers.  Flow velocities in the region
of most  of its  concentrated  recharge  points is probably  not
turbulent.  The reason  for the lack of  turbulence can be found
in the nature of the aquifer recharge.
                             296

-------
  Recharge  to the  Quarles Spring groundwater  basin does  not
include sinking streams.  All of  its  recharge is autogenic, it
occurs directly over  the  groundwater basin where  only  soluble
bedrock with chert bands and cherty residuum overlie the bedrock
aquifer.  Sinkholes are common, but they  rarely  contain direct
openings to the bedrock.  A layer of  cherty residuum, 25 to 127
feet in thickness, mantles the bedrock and sinkholes restricting
most of the  recharge to  soil percolation (Dames & Moore,  1991).

  Low-flow  and high-flow  time of travel  studies utilizing  dye
tracers were  undertaken between the  drywell  at point  "H"  and
Quarles Spring  (Fig.  3)(EWC,  1991).   The velocities obtained
were 95 feet per hour  and 226 feet per hour respectively.  These
values are quite different from those obtained from the  Mammoth
cave  region where  the  aquifer is recharged  directly  through
swallets and by sinking streams.   Values of 9 meters per hour
and 390 meters per hour  are quoted by Quinlan & Ewers (1985)  for
that area.

The vertical complexity of the conduits-
  If  the  conduit  encountered  in the well  at point  "C"  (Fig. 3)
is indicative  of  the  level of  the master conduit,  31  feet of
space in the vertical dimension are available for  the trapping
of contaminants  (Table  1; Fig,  4).    Fracture  trace analysis
(Dames and Moore,  1991)  and the extremely variable  elevation of
the bedrock soil  interface (Dames  & Moore,  1991;  MCI,  1987)
strongly  suggest  that  fractures  enlarged by  dissolution  are
common  in  the  bedrock  aquifer.     Therefore,  vertical  and
horizontal  conduits should be  common  in  the  epikarstic  zone.
These   should   be  quite   effective  at  trapping  floating
contaminants in the Campbell  Army Airfield setting.  The wide
range  of  the  soil  bedrock interface,  reported  by drillers,
should  insure  that there  are several horizons  of  horizontal
epikarstic  solution  porosity  which  are  poorly  integrated
laterally (Fig. 4, Table 1).

  The 16  foot layer of  floating jet fuel in the well reported
by Ewers  is associated with  the  conduit porosity  encountered
when this well was drilled.  The free product reported  by Duda
was discovered at the soil bedrock interface,  and  is probably
fuel which has migrated upward through solution widened joints
from the master conduit  below.  These horizons are  likely to be
poorly connected to sources of concentrated recharge because of
the relative  lack of  open sinkholes.   Without this turbulent
recharge,  accumulated product is unlikely to  be flushed  from
these reservoirs.   Because these  reservoirs extend  well  above
the active  flow  system  there  would  be  little opportunity  for
measurable quantities  of the floating immiscible contaminants to
dissolve in  the groundwater circulating in the main conduits.
                              297

-------
   CASE  STUDY  2,  AN AQUIFER WITH GROSS CONTAMINANT MOVEMENT
            Little Sinking Creek - Big Sinking Creek

   This site is located near the Big Sinking  Creek oil field in
Lee County, Kentucky.  Both Little Sinking Creek and Big Sinking
Creek  are,  in  part,  subsurface streams.    Portions of  their
surface  valleys are perennially streamless (Fig.  5).   Conduits
of explorable  dimensions with  multiple  entrances pass  beneath
the dry  portions  of these  valleys.

   Little  Sinking Creek  - Big Sinking Creek, KY
                                                 Legend
                                             -<•• Spring

                                             -—^ Sinking Stream

                                             A  Oil Tanks
                                  1000   0    1000   ?000  3000   4000
  FIGURE  5   Little Sinking Creek and Big Sinking Creek are,
  in  part,  subsurface  streams.    Several  smaller  sinking
  streams are tributary  to the master conduits beneath these
  valleys.     oil  spills  at   several  points   have  been
  transmitted quickly  to the  springs at points "A" and  "B".

  This  karst  aquifer  is  developed  in the  Mississippian  age
Newman Limestone,  oil  production is from units 700 feet beneath
the karst aquifer.  These horizons are isolated from the aquifer
                              298

-------
by several hundred feet of shale.  Pipeline breaks, storage tank
cleaning,  and  brine discharges have  introduced  chlorides  and
crude oil into  these sinking  streams  and into the  conduit
entrances.  The  interior surfaces of  the conduits are  coated
with petroleum and  tar balls  are  common along the  surface  and
subsurface portions of the  streams.   A spring, perched  upon  a
shale layer beneath the  Newman Limestone discharges from  the
lower end of each of these valleys near their  confluence (Fig.
5).  Floating  free  product  is discharged with the  groundwater
from these spring points after spills have  occurred.  Thus,  it
appears that floating hydrocarbons are  extremely mobile  in this
aquifer.  The known  and inferred aquifer characteristics  fit the
criteria for high mobility outlined above.

The principal conduit systems-
   The  master  conduits   in  both  valleys are  located  in  the
epiphreatic zone.  Deeper phreatic circulation is prevented in
this case by the  presence of insoluble  shale beds  in the  Renfro
Member of the Borden Formation.  Virtually the  entire  length of
conduit beneath  the valley of  Little  Sinking  Creek  shown  in
figure 6  is  explorable and occupied by  a stream  with an  air-
water interface  above.    The  conduit  which forms  the   spring
orifice has been observed to maintain  an air-water  interface
under very high flow conditions.  The master conduit beneath the
valley  of Big Sinking   Creek  does not  possess  an  air-water
interface  over  its  entire  length,   but  the crude oil   is
transmitted along the total conduit system (Fig. 7).

The flow regime and the aquifer recharge-
   The  aquifer recharge  at  this site  is  almost  entirely  from
allogenic sources, by way of sinking streams. This concentrated
recharge is clearly turbulent,  and during heavy rainfall  events
it is grossly  turbulent.   The conduit flow in the aquifer  is
visible at several points along each valley  and turbulent flow,
similar  to  that  in  the  sinking  streams   and  the   spring
resurgences, can  be directly observed.

   The portion of  the Big  sinking Creek  conduit  above mist cave,
shown in Figure 7, may not possess an air-water interface even
under base flow conditions.  Water ponds over the swallet of  Big
Sinking Creek during high-flow conditions and  the crude  oil  is
temporarily trapped at the surface.  When the flow  subsides,  the
ponded  water  lowers  and  turbulence  entrains   the  floating
contaminant, and  it  is conducted through several spring and sink
points to its final resurgence.

The vertical complexity of the conduits-
   Only  a  few  feet of  limestone overlie the master  conduits  at
this  site.    Therefore,   the  opportunities  for   trapping  of
hydrocarbons  along the  few  sections  of  phreatic  flow  are
severely limited.
                             299

-------
Vertical Section - Little Sinking Creek
  Swallet of Little Sinking Creek
  c
  CO
  E

  0)
  z
                   Oil Tanks
  c
  0)
  13

  O
  CQ
                                        Arrow Cave
                                                      Flood Cave
   Resurgence
      of
Little Sinking Creek
FIGURE 6  Vertical  section  through  Little  Sinking Creek
The  Swallet  of Little  Sinking Creek  and  Flood  Cave  are
points "A"  and "C"  in Figure 5, respectively.
Vertical Section - Big Sinking Creek
                                                     Dune Buggy Cave
  Swallet of Big Sinking Creek
                              Mist Cave
    Resurgence
       of
  Big Sinking Creek
  V)
  2
  _i
  c
  (0
  E
FIGURE 7  Vertical  section  through  Big Sinking Creek
The  Swallet  of Big Sinking Creek and Dune Buggy Cave  are
points "B" and "D"  in Figure  5, respectively.
                             300

-------
   CASE STUDY 3, AN AQUIFER WITH MIXED CONTAMINANT MOVEMENT
      Lancaster Road Gasoline Station - Richmond Kentucky

  An  underground storage tank  at  this site released  gasoline
into a karst aquifer consisting of  Ordovician Limestones.   Free
product was discharged into Tennis Court Spring, an intermittent
seep 150 ft from the  leak site (Fig. 8).  Fluorescent tracer dye
injected  into  the  storage  tank pit showed that groundwater
flowed to Tennis Court Spring and also to  Little  Caesar Spring
a perennial spring 900 ft distant.   Gas chromatograph  analysis
of  samples from these  springs showed  that  the  free  product
discharging from Tennis Court Spring was nearly identical to the
tank contents.   No gasoline constituents could be discerned  in
the samples  from Little  Caesar Spring  either in high-flow  or
low-flow conditions.

  There are  no  data  from wells  or other sources which  can  give
direct insight  into  the  nature of  the  conduits  which  connect
these two  springs to  the leak  site.   However,  the two  springs
clearly indicate the presence in the aquifer of a  natural light
non-aqueous  phase  liquid   separation  system.   This system
operates in  an  environment  where water flow clearly occurs  in
discrete conduits.   The tracer introduced  into  the tank pit
leaves no  doubt about this assertion.   It  appeared  at  both
springs  overnight.    As expected,  the  level  of  the spring
discharging product was significantly higher than  the one which
discharged water only (Table 2).   This karst aquifer has  no
sinking streams and  few  sinkholes,  none with  swallet  openings
which could admit water  directly to the aquifer.
  Table  2  ELEVATION OP FEATURES RELATED TO THE LANCASTER ROAD
  GASOLINE STATION

  Elevation (ft.)                             Spring Features

  995 	  Top of gasoline storage tank pit  (dye injection
                point)
  985 	  Tennis Court Spring - Groundwater, gasoline,  and
                dye were discharged at this point.
  935 	  Little Caesar Spring - Only groundwater was
                discharged at this point.
                          CONCLUSIONS

  The  empirical  studies  suggest that the following conditions
decrease the probability  of  movement of light hydrocarbons  in
karst aquifers:

1-The  limestone  formations  are thick  and  conduits  are  well
within the phreatic zone.
                            301

-------
   Lancaster Road Site, Richmond,  KY
                Gasoline Leak  Site
                  Tennis Court Spring
                                   /
      0    100   200   300

      I     I     I     I
             Feet
       Little Caesar Spring
                   935' msl
     Eastern Bypass
FIGURE 8 An underground storage tank at this site released
gasoline into a karst aquifer.   Free  product was discharged
at Tennis Court Spring, an intermittent seep.  Fluorescent
tracer dye injected into the storage tank pit showed that
groundwater flowed  to Tennis  Court Spring  and  also  to
Little  Caesar  Spring,   although  no  gasoline  could  be
detected at this  latter spring.  The numbers indicate the
elevation of the  leak site  and the springs.
                          302

-------
2-The limestones are covered by  a  thick mantle of soil causing
recharge to the aquifer to be slow and  non-turbulent.

3-The  aquifer is  not  recharged by  sinking  streams  so  that
turbulence in the master conduits is minimized.

4-An extensive system of epikarstic porosity overlies the master
conduit providing cavities for trapping of the hydrocarbons.

Given  all  of these  characteristics,  hydrocarbon  mobilization
could occur if drought  conditions  can  lower  the potentiometric
surface  sufficiently to alter  the phreatic  condition in  the
master conduit.

   Human  activities or natural events which would  permit  storm
water  to flush  the epikarstic  zone  could  possibly  mobilize
contaminants stored there.  These would include  intense  highly
localized rainfall events,  developing  sinkholes,  excavations
near bedrock, and dry-wells.   Attempts  to retrieve contaminants
through recovery wells  by creating depression cones  extending
near  or  below the  level  of  the spring outlets could  create
conditions for their transport, particularly if this occurred at
a time when gross turbulence existed in  the trunk conduits.
                       REFERENCES CITED

Carey, M,J.,  1990,  The delineation of Karst drainage to identify
possible  contaminant migration  routs from  the Campbell  Army
Airfield,  Fort Campbell,  Kentucky.   Master of  Science  Thesis,
Eastern  Kentucky University,  Department  of Geology,  Richmond
Kentucky, 6Op.

Dames & Moore, 1991, Remedial Investigation Report, Contract
DAAA 15-88-D-0008 to USATHAMA

Ewers, R.O.,  1972,  A model  for  the development of  subsurface
drainage routes along bedding planes, M.S. Thesis, University of
Cincinnati, Cincinnati, Ohio, 84p.

Ewers Water Consultants (EWC),  1991, Groundwater Time of Travel
Study, Campbell  Army  Airfield,  Implications  for  Contaminant
Movement, R.O. Ewers. Contract to Ft. Campbell and USATHAMA.

Ewers Water Consultants (EWC), 1989,  Fort Campbell  Groundwater
Study, Final Report,  R.O.  Ewers, M.J. Carey, and D.L.  Greene.
Contract to Ft. Campbell and USATHAMA.

Green, D.L.,  1990,  Hydrogeology of the  potable water  supply
spring   (Boiling  Spring)   on   the  Fort  Campbell   Military
Reservation,   southwestern  Kentucky  - northwestern  Tennessee.
Master   of  Science  Thesis,   Eastern  Kentucky   University,
Department of Geology, Richmond Kentucky, 43p.
                             303

-------
Mangin, A.,  1974-75, Contribution  a 1'etude  hydrodynamic des
aquiferes karstiques, Annals de Speleologie,  29(3).

MCI, 1987, Subsurface investigation for JP-4 contamination, POL
facility 7226, Fort Campbell Army Airfield,  Fort Campbell KY, to
US Army Corps of Engineers, Nashville  Dist.  by  C.S.  Higgins,
M.J. Levy, and C.B. Huggins, MCI Consulting Engineers.

Quinlan,  J.F.  and  R.o.  Ewers,   1985,  Groundwater  flow  in
limestone  terranes:   strategy  rational   and  procedure  for
reliable,  efficient monitoring of groundwater quality in karst
areas.  Proceedings   of  the   Fifth  National  Symposium  anrj
Exposition on Aquifer Restoration and  Groundwater Monitoringr
National Water Well Association.

Quinlan,  J.F.  and  R.o.  Ewers,  1985,  Groundwater  Hydrology
geomorphology of the  Mammoth Cave region,  Kentucky,  and of the
Mitchell  plain,  Indiana  (Field Trip 7);  with J.A. Ray,  N.C.
Krothe, and  R.L.  Powell; in Shaver,  R.H.  and Sunderman J.A.,
eds.,  Field  Trips  in Midwestern Geology:  Bloomington.  Ind.f
Geological Society of America  and  Indiana Geological Survey, v.
2, p. 1-85.
                     BIOGRAPHICAL SKETCHES

   Ralph o. Ewers is professor  of geology and director  of  the
Groundwater Research Laboratory at Eastern Kentucky University,
and a principal  in  Ewers Water Consultants,  a consulting firm
specializing in carbonate aquifers.  His  B.S.  and M.S.  degrees
in geology were earned at the University  of  Cincinnati  and his
Ph.D. was  earned  at McMaster  University  (1982).    Professor
Ewers' special interests include the applications of tracer and
electronic monitoring techniques  to the  solution  of  practical
environmental  problems   in  karst  groundwaters.     He  was  co-
recipient of  the  1986  E.B. Burwell  Award from  the Geological
Society  of  America  for a  "work of distinction  in engineering
geology."

Ralph O. Ewers Ph.D.
Ewers Water Consultants Inc.
160 Redwood Drive
Richmond, KY 40475
(606) 623-8464

   Anthony  J.  Duda is senior hydrogeologist with  Dames & Moore,
where he has worked since 1980.   He received his  B.A.  degree in
geology  from Rutgers University.   Mr Duda has been involved in
planning and  implementing groundwater  remediation strategies,
soil  gas  sampling  programs,   and  geophysical  surveys.    His
experience includes investigations at more than 30 governmental
and private  sites  in more  than 10 states.   He  has  been lead
hydrogeologist at the Ft. Campbell, Campbell Army Airfield site
where he was involved in a major UST investigation.
                             304

-------
  Elizabeth K. Estes is a doctoral  candidate  at the  Department
of Geology and Geophysics at the university of  Minnesota.   She
received her  M.S.  degree  from Eastern  Kentucky University.
Among her many accomplishments is a five year  study of the  area
in the vicinity of Big  Sinking Creek in Lee  County,  Kentucky,
and several studies with the Cave Research Foundation at Mammoth
Cave  Kentucky-     Her  research   interests    include,  karst
geomorphology, speleogenesis, and groundwater geochemistry.

  Peter  Idstein received  his Bachelor  of Science  degree  in
Geology at Eastern Illinois  University.   He  is  completing his
Master  of Science  degree   in  Geology  at   Eastern  Kentucky
University. Mr.  Idstein  has  spent one  year  working at  the
Florida  Sinkhole  Research   Institute  conducting studies  on
conduit dominated groundwater  flow.   He has also spent a  year
working  for  Ewers  Water  Consultants conducting  dye  tracing
studies and continuous  electronic  monitoring studies in  many
karst dominated and  non-karst terranes.

   Catherine  A. Johnson is  a  project officer  for the United
States Army Toxic  and Hazardous  Materials Agency (USATHAMA).
She  graduated with  General  Honors  from  the  University  of
Maryland with  a B.S. degree in Chemical Engineering.  Her duties
at  USATHAMA   involve  managing  Preliminary   Assessments/Site
Inspections,  Remedial Investigations/Feasibility Studies,   and
RCRA Facility Assessments/Investigation studies  at Army sites.
Ms.  Johnson has been the project officer at the  Fort  Campbell,
Campbell Army  Airfield site  for the  last two years. During  that
time she has managed extensive field work including; geophysical
investigations,   sediment,   Water,   and  soil   sampling,   and
monitoring well  installation.
                             305

-------
The Transmission  of Light Hydrocarbon  Contaminants  In  Limestone
(Karst) Aquifers

Ralph O. Ewers,  Anthony J Duda, Elizabeth K. Estes, Peter J Idstein
and Ratherine M. Johnson.

1.   What  are  the major  difference  between  the  physical  and
chemical properties of the dyes and the  light hydrocarbons used in
you study?  Why not try to use a tracer that  has   properties
similar  to  the   hydrocarbons?     Has   there  been   a   chemical
transformation of the hydrocarbons in the  limestone which affects
their properties?

ANSWER- The dyes are water soluble substances that are non-toxic.
The light  hydrocarbons are toxic  immiscible organic  substances.
I know  of no non-toxic  immiscible hydrocarbons  that regulatory
agencies would  allow me to place in an  aquifer in the quantities
that would be required.  In any case, tracers with the properties
of the  light  hydrocarbons would not give  us  the answers  that  we
need in the time required.  They would be  poor tracers.

     I  suspect that   there  are  many  chemical  and   biochemical
transformations which  have occurred.   However,  I do  not  believe
that  such  changes  are the principal reason for their  apparent
retention in many karst situations.  I believe that immiscibility
and density are  the primary reasons.  The demonstrated  travel  times
are too fast for these processes to be effective.

2.   What were the  dissolved BTEX levels  associated with the Little
Caesar Spring?  Was dilution sufficient to limit  benzene  level  to
below 5 micrograms  per liter.  What were  dissolved levels at low
flow conditions?

ANSWER- I do not know the BTEX levels because they were below the
level of detection at the time.

     Demonstrably,  the separation  system   in  the  aquifer and the
dilution were sufficient.   Hundreds of gallons  of gasoline were
probably involved,  and the spring flow was relatively  small, a  2-
3 cubic feet per  second  at maximum.  Separation  of the  floaters
seems to be very important.
                               306

-------
                   VELOCITIES OF PIEZOMETRIC WAVES CAUSED BY

                           PUMPING IN KARSTIC AQUIFERS


                                       C. Drogue

Laboratoire d'Hydrogeologie, Universite de Montpellier II, Place E. Bataillon, 34095 Montpellier
                                    Cedex 05, France



Abstract

The velocities induced by pumping in a karstic area were studied at a test site in 20 boreholes at
5m intervals. The site was 4.6 km from a spring. Pumping of 1 to 2 m^ s ~1. was performed at the
spring. The time for the effect of pumping at the spring to reach the boreholes varied from 6 to 8
minutes, corresponding to an apparent velocity  of 9.6 to 12.8 m s~V Wave propagation probably
occurs in  drainage "conduits"  (connected to  the  spring) at a velocity related to physical
parameters such as the water (modulus of  elasticity), rock (Young's modulus) and  the
characteristics of the conduits (equivalent diameter) and the type of flow (under head or open
surface). The  recorded velocities show that flow was not entirely under head since they would
have been much higher in  this case  because  of the  rigidity of the  limestone.  At the other
boreholes (which boreholes) impact times varied from 26 minutes to 5 hours, corresponding to an
apparent velocity of 2.9 to 0.2 m s  . These observations are interpreted using a conceptual model
with double diffusivity taking into account  low diffusivity fissured blocks and high diffusivity
cracks. The boreholes directly  affected by pumping at the spring penetrated a fracture in the
cracks with blocks in the model. The boreholes with low velocity are in the blocks themselves and
far from the drain. One of the  results of this difference in piezometric  behaviour is that certain
apparent piezometric gradients are reversed between pumping and periods of stoppage.


Introduction

Measurements of the velocity of piezometric waves caused by pumping in a karstic aquifer can
contribute to  better surveying  of the heterogeneity of  the medium. This is particularly fruitful
when numerous piezometers are installed in an aquiferous zone and  the fissure structure can be
surveyed.


General considerations

In porous  media, pressure transfers can be interpreted  by solving the diffusivity equation (Bear,
1972). There are two possible procedures for karstic aquifers:
                                         307

-------
a) The karst is identified as a continuous medium equivalent, as has been proposed for certain
fissured rocks (Maini & Ockins, 1977). The problems are handled like those of porous media. This
is possible if the area considered is very large in relation to a  basic grid representative of the
high conductivity network. However, even more than in fractured rock, heterogeneity is found at
all scales in karst (Drogue, 1988) and it is difficult to consider it as a porous medium.
b) The second procedure takes  into account the geometrical and hydraulic characteristics of
aquiferous karstic networks. However, schematization  is necessary because  of the extreme
complexity. This is the case of the conceptual models proposed by fissured  aquifers. In these
models, matrix porosity is combined with  fracture porosity (Barenblatt et al, 1960; Wilson et al,
1974; Shapiro & Anderson, 1983; Huyakorn et al, 1983; Moench, 1984).
A double porosity fissure model is used for karstic aquifers (Drogue, 1980, 1988). Little-karstified
fissure blocks with low hydraulic conductivity (K' = 10"7 to 10~9 m s'1) are separated by cracks or
karstic channels with high hydraulic conductivity (K' = 10~3 to 10'1 m s'1) (Fig. 1). The model can
be used  in many  karstic aquifers to  interpret  the  phenomena observed:  piezometry,
hydrochemistry and temperature patterns (Drogue,  1985, Drogue,  forthcoming).
The channels between the blocks form the aquifer drainage network which  converges  on the
channel feeding the spring. If the channel  is under head it can be considered as a conduit. The
velocity of a pressure wave can then be expressed as follows:
      c =
where: p = voluminal mass of water, K = modulus of elasticity of the water, D = diameter of the
conduit, E = Young's modulus of the conduit material, e = thickness of the wall

In the extreme case of an indeformable conduit, the ratio D/Ee is zero and C = 1425 m s~* (at 15°C),
where p = 10'2 and K = 2 10~8 kg f mr2.
This velocity  that of sound under water  - is certainly not attained in karstic aquifers. Indeed,
these rocks undergo deformation by the piezometric variations caused by rain water alone. This
was demonstrated by extensometric measurements performed  in caverns (Crochet et al., 1983). In
addition, the channels are often combined with cracks with open surfaces corresponding to the
blocks in the model. These cracks contain damping volumes which slow the pressure  waves.
Aquiferous non-karstic soil cover may also play the same role through the drainage effect.


Experimental conditions

The experimental site consists of 20 boreholes 60 m  deep laid  out in a 500 m2 area (Fig.  2). The
boreholes are 5 to 7 m apart. They run through much-fractured, karstified Upper Jurassic and
Lower Cretaceous limestone forming a free aquifer. All the boreholes reached the saturated zone.
Piezometric levels vary from 20 to 50 m according to the season. Various work  at the site - test
pumping, tracing, logging,  piezometric monitoring over  a  period  of  several  years and
hydrochemical investigations  have verified that there is  hydraulic continuity  between all the
boreholes. It has also been possible to site each borehole in relation to the drains and blocks of the
conceptual model (Drogue & Grillot, 1976).
The spring at which pumping is carried out is 4.6 km from the experimental site. The natural flow
ranges from 0.4 m3 s'1 at low water to 8 m3 s'1 during flood periods; average flow is 2 m3 s"1. It
drains a 400 km2 aquifer within which the  experimental site is  located. Numerous faults between
the spring and the experimental site place  karst and non-karstic formations in contact with each
other.
Exploitation of the spring with discontinuous pumping started  in 1968 at 0.8 m3 s"1.  Pumping
discharge has been up to 1.5 m3 s'1 in recent years.
                                         308

-------
Fig.  1.  Conceptual  model of a  double
fissural porosity karst. Blocks with low
permeability (B) are bounded by  open
cracks with high hydraulic conductivity
(C). When the  karst outcrops, the upper
part of the aquifer is extremely fissured (A)
because  of decompression and  surface
weathering.
                                                      Meteorological  station
                           Laborato
Fig.  2. The  experimental site
(Terrieu, karst formation in
the South of France) equipped
with 20 piezometric boreholes
60 m deep. Only the boreholes
mentioned  in the  text are
numbered. They are repres-
entative  of  the  various
piezometers monitored at the
site.
                                                                              Phreolic high
                                             Phreatlc low
                                Ground water flow to the spring
Fig. 3. Piezometric evolution in
several  boreholes  during the
first pumping test: A, pumping
at the   spring;   B,  pumps
stopped.
                                    I9;5 .
I
N
e
5.
   20.0
                                     hours
                                       309

-------
    48-
              2h                               3h                               4h
                                                      Time (hours)
Fig. 4. Piezometric evolution during the second pumping test. 1, piezometry at the spring; 2,
piezometry at borehole P20; 3, piezometry at borehole P9; a, 2-minute response time; b, 24-minute
response time.
                                                Drain
                      PIO
                       X
                                      Block
                   P20,
                             1
I I I I I1
LJ _L'._i
rt+4
                Spring
                                                        Z*- (pumping
Fig. 5. Diagram (based on piezometer velocity) of the spring and boreholes P20, P9 and PIO in a
double porosity fissure model.
                                     block
                                            drain
                                                     Time
Fig. 6. Piezometric data for a drain and an adjacent block during pumping. Dephasing of the
piezometric movement accounts for the inversion of the piezometric gradient (a).
                                        310

-------
Experimental results

Preliminary experiment

This preliminary  experiment was performed with natural flow of 0.5  m^ s~* at the spring.
Pumping discharge was 0.8 m^ s~l with 0.5 m maximum drawdown at the spring.
Piezometric variations at the experimental site displayed the following features (Fig. 3):
- drawdown was considerable at some boreholes: 0.3 to 0.4 m. In contrast, drawdown was slight or
non-existent at other boreholes only 5 or 10 m from the former;
- the piezometric levels in the boreholes little-affected by pumping were higher than those of the
boreholes  in which there was considerable drawdown.  When the pumps were stopped, the
apparent piezometric gradients between these boreholes were reversed.
In order to confirm and complete these observations, pumping was performed at high discharges to
cause greater drawdown at the sight and enable better observation of the phenomena.

High discharge pumping

After pump stoppages limited to  30  min because of water requirements, piezometric variations
were triggered by instantaneous  pump starting at discharges of  1.5 and then 2 m^ s~l. Three
piezometers representative of the different domains of the double porosity model were observed:
P20: on a drain,
P9: in a block and near a drain,
P10: in a block.
The results are shown in the table below.

Response time (min)
Apparent velocity
(ms'1)
P20
2
38
P9
24
3.2
P10
approx. 120
approx. 0.6
Drawdown was too slight for accurate measurement at borehole P10.
Discussion

It should first be noted that the hydraulic distance between the spring and experimental site is
doubtless greater than the geographical distance of 4.6 km because of twists and turns of the
network. Real wave velocities were thus greater than the apparent velocities in the table above.
Borehole P20 responded very rapidly. It cut through  a drain which must be connected to the
karstic network which feeds the spring. P9, which is on a block but near a drain reacted more
slowly. P10, located in a block, reacted much later. These results thus confirm the suitability of
the model for the interpretation of the velocities of piezometric  waves. The respective positions
of the spring and the piezometer boreholes can be schematized in  the model (Fig. 5).
Dephasing  times are  related to reductions in the amplitude of  piezometric fluctuations. Thus,
stopping pumping between discharges of 1.5 m3 s-1 and 2 m3 s-1 caused piezometric rises of 0.65 m
in borehole P20, 0.25 m in P9 and  approximately 0.03 m in P10. There were  thus apparent
piezometric gradients between boreholes. This is explained by the dephasing of the piezometric
variations between a  drain and an adjacent block during short duration pumping (Fig. 6). This
phenomenon might be a  characteristic of the karst. It  is  interesting to observe that these
piezometric inversions were similar to those observed in the same boreholes during a flood caused
by infiltration (Drogue, 1980).  In the latter case,  rapid piezometric recharging  caused a
                                        311

-------
temporary rise in level in the drains in relation to the piezometric levels in the blocks.  This
piezometric inversion phenomenon also occurred between the spring and borehole P10 and was
caused by the low apparent piezometric gradient between the site and the spring (8.7 10~2) when
pumping was stopped.
Conclusion

Piezometric wave velocities thus provide extremely accurate information about the aquiferous
structure of the karst and its degree of heterogeneity. The representativeness of a piezometer in
aquifers of this kind should thus be examined attentively. The hydrodynamic calculations made
using data from such a piezometer only apply to a certain domain of the aquifer.
References

BARENBLATT, G.I., IV.P. ZHELTOV and I.N. KOCINA, 1960   Basic concepts in the theory of
seepage of homogeneous liquids in fissured rocks.
      J. Appl. Math. Mech. Engl. Transl., v. 24, p.1286-1303.
BEAR, J., 1972 - Dynamics of fluids in porous media.
      Elsevier, New-York, 764 p.
CROCHET Ph., LESAGE Ph., BLUM P.A., VADELL M., 1983 - Extensometric deformations linked
with  rainfalls.
      Annales Geophysical, 1, 4-5, p.329-334.
DROGUE  C. and  GRILLOT  J.C., 1976   Structure geologique  et  premieres observations
piezometriques & la limite du sous-systeme karstique de Terrieu (Perimetre experimental).
      2e coll. Hydrologie en pays calcaire. Besangon, France. Ann. Sc. Universite de Besangon,
      fasc. 25, 3e serie, p.195-209.
DROGUE C., 1980 - Essai d'identification d'un type de structure de magasins carbonates, fissures.
Application £ 1'interpretation de certains aspects du fonctionnement hydrogeologique.
      Mem. Soc. Geol. France, No.ll, p.101-108.
DROGUE C., 1985  Geothermal gradients  and ground-water circulation in fissured and karstic
rocks : the role played by the structure of the permeable network.
      J. of Geodynamics, 4, p.219-231.
DROGUE C., 1988 - Scale effect on rock fissuration porosity.
      Environ. Geol. Water Sci.  vol.11, No.2, p.135-140.
DROGUE C., forthcoming, 1992  Hydrodynamics of karstic aquifer, results from experimental
sites (Mediterranean karsts in the South of France)
      "Hydrogeology of Selected Karst Regions, I.A.H., vol.13.
HUYA KORN P.S., LESTER B.H. and FAUST C.R., 1983  Finite element techniques for modeling
groundwater flow in fractured aquifers.
      Water Resour. Res., v.19, No.4, p.1019-1035.
MAIM  and OCKINS, 1977  An examination of the feasibility of Hydrologie isolation of a high
level waste  repository  in  crystalline  rocks.  Invited paper,  geological   disposal of  high
radioactive waste session, annual meeting of Geological Society of America, Seattle.
MOENCH, A.F., 1984 - Double porosity models for a fissured groundwater reservoir with fracture
skin.
      Water Resour. Res., v. 20, No.7, p.831-846.
SHAPIRO  A.M. and ANDERSON J., 1983   Steady  state fluid  response in  fractured rock: a
boundary element solution for a compled discrete fracture continuum model.
      Water Resour. Res., v.10, No.2, p.328-335.
                                          312

-------
                  Solution Mining And Resultant Evaporite
                Karst Development In Tully Valley,  New York
           Paul A. Rubin1, John C. Ayers2,  and  Kristin A.  Grady3
           'Oak Ridge National Laboratory, Oak Ridge, Tennessee
               2Vanderbilt University, Nashville, Tennessee
                3Law Environmental, Inc., Albany, New York

             The views and conclusions put forth in this paper
               are expressly those of the authors and do not
           necessarily reflect those of the State of New York,
                  its employees,  experts and/or agencies.

                                 Abstract

     A solution mining operation was conducted  in Tully Valley, New York
from 1889 to  1988. In excess of 37 million m3 of halite was  removed  from
335 to 518 meters  below the ground surface. An  interbedded sequence of
gypsum,  shales,  limestones,  and sandstone overlie the halite beds. This
sequence is capped by thick,  unconsolidated deposits of till,  sand and
gravel,  and lacustrine clay.

    As a result of this mining, large void cavities were created, followed
by numerous fractures extending upward to the ground surface.  The resulting
settlement area is in excess of 550 hectares. Within this area sinkholes
formed,  gaping fractures developed and streams  were pirated into the
subsurface. Interformational mixing of groundwater now occurs between
formerly separate  flow systems, providing substantial recharge to deep
formations.

    Some 2 kms downvalley of the brine fields,  in a smaller settlement
area, mud "volcanos"  effuse weakly saline groundwater that flows into
Onondaga Creek.  The clay fraction of the effluent gives Onondaga Creek the
appearance of chocolate milk for the =26 kms it takes to reach Onondaga
Lake. The location of the mud volcanos appears  to coincide with an up-
valley moving salt front.

    The number of  mud boils and their areal extent has substantially
increased since the onset of brining operations.  By characterizing the
chemistry of  groundwaters in local formations and performing mixing
calculations  based on mass balance, the volcano effluents were shown to
represent a mixture of groundwaters from 3-4 formations.  Several working
hypotheses are advanced and critically evaluated in an effort to define the
dynamics necessary for rapid mud volcano growth in a karst setting.

                           Location  and Geology

    Tully Valley is a glacially scoured Finger  Lake valley located 24 km
south of Syracuse, New York (Fig. 1). The valley floor is blanketed by
Pleistocene ice marginal deposits 18 to 134 m thick consisting of
gravel/clayey gravel, a middle thick rhythmic clay sequence and a basal
sandy gravel  (Getchell,  1983; Mullins et al., 1991). Bedrock consists of an
interbedded sequence of shales, limestones, and sandstone overlying four
evaporite beds ranging up to 21 ± 8 m in thickness (Fig.  2). All beds dip
approximately 2° to the south.
                                   313

-------
      SURFACE HYDROLOGY AND  LOCATION
             MAP OF TULLY VALLEY,  NY
                                   .  BRINE WELL OR
                                     CORE HOLE
       VESPER
         FIELD
                           WEST FIELD
                                             Sand Volcano U Volcano Area Ram Pum
            0   500  1000 meters
Figure 1:
      General location map of the features in Tully Valley, including
      brining fields and associated collapse.

-------
                               SCHEMATIC    CROSS-SECTION
SOUTH
                             Showing  Collapse   Structures  and  Flow
                                 Regime  of  Tully  Valley,  New  York
                                                                                                      NORTH
                                                                                                         CARDIFF
                                                                                                          WELL
50


100
                                                      1000
                                                              2000  meters
 D
POST GLACIAL LACUSTRINE AND GLACIAL OVTWASH
   Clayn and Sllte, Sand and Grovel
     DEVONIAN SHALES - Hamilton Croup
     Skeneatellea and Marcalliu »h*lfi
     ONQNDAGA LIMESTONE
     ORISKANY SANDSTONE
     DEVONIAN/SILURIAN LfUESTONE AND DOWIOTSS
         Uanliua Lm./Readout and
       Coblesldll Dolomites and Bertie Fm.
SILURJAN-CAHILWS SHALE
 with lowvr gypsum bed
                                         GYPSUM BED
                                   SALT- *
                                            * B*"1", 3  ««1 4"1
                                            Beda plotUd tc6elher
                                                         - Syracuse Pm.
                                        SUurimn Veroon Shale
Disrupted beds =
shale, gypsum
                                                                            Infiltrating meteoric
                                                                            and ground water
                                                                           flow direction
                                                                           of brine waters
                                                                          Regional Groundwater flow
Brine Potentiometric
    Surface
                                                                                                      1 •*. *~ FRACTURE
                                                                                                     Original
                                                                                                   land surface
                                                                                                    Brine Cavities
                                                                                                    and brine /tow path
    Figure 2:      Generalized cross-section  of Tully  Valley showing  stratigraphy,  collapse
                    structures and  flow regime.  See text for further discussion.

-------
             Solution Mining Operation and Resulting Subsidence

    A solution mining operation was conducted  in Tully Valley  from  1889
until 1988. Until  the late 1950's salt was usually mined by injecting  fresh
water down one well to the formerly dry salt horizons and then pumping or
air lifting brine  out another well. Brine was  piped downvalley to Syracuse
for the production of soda ash. Some  150 wells were drilled, many in close
proximity  (=  40  m)  to each other. Continuing brine extraction  caused the
solution cavities  to expand and interconnect.  It was not uncommon to drill
new wells  and find that uncontrolled  solutioning had already removed some
or all of  the upper halite beds (Fig. 3).

    Seven  large  sinkholes formed catastrophically in the Tully brine fields
between 1949  and 1980. A monument network was  installed and periodically
surveyed from 1959 to monitor subsidence. Although the extent  of early
subsidence will  never be known, a steady growth in the subsidence area has
been  documented. Today, this settlement area  is in excess of 550 hectares,
has subsided  up  to 14 meters vertically, and  is progressively  expanding
outward from  the cavities. A comparison between the 37 million m  of halite
extracted  and the total settlement volume of  5 million m3 calculated as of
1982  (Tully,  1983) indicates that only 14 percent of the total possible
settlement volume has occurred.

    Such extensive settlement resulted from fracturing and subsequent
collapse of beds overlying the brine  cavities. Evidence supporting  this
includes the  continued growth of the  large  surficial settlement area,
individual sinkhole formation, the presence of large gaping fractures  open
to the  surface,  sheared well casings, and the piracy of streams  (e.g.,
Emerson Gulf) into the subsurface. In essence, extensive brine cavity  roof
collapse  and  bedrock disruption has resulted  in a fracture network  of  high
permeability.
        E WELL GROUPING OF  THE EAST FIELD
               YEAR'DRILLING COMPLETED
         1889 1889 1B90 1890 1B82 1930 1934 1937 1940  1941  1958
   3
   1
w
g-iao
2
  -ISO

                                        Aver, top of
                                     -- 1st Salt Bed

                                        Aver, top of
                                        2nd Salt Bed
                                       Aver,  bottom of
                                        2nd Salt Bed
                                        Aver, top of
                                        3rd Salt Bed
                                          Aver, bottom of
                                          Salt Beds
                                                            WELL LOCATIONS
                                                            RE-1
                                                                   S-2
                                                                    o
                                                                      S-1
                                                                 NE-2
                                                                    o 4 E-2
                                                           t
    NE-3  '



~^J4E




  0™^""ToOmel
         E-l E-2 E-3 £-4 E-5  _j NE-2 NE-3 NE~» NE-1 RE-1
•
B
KEY
Depth to first SALT
encountered
SALT
   Figure 3:
      Time versus depth plot of the E well grouping.  By 1937 when NE-3 was
      drilled, the first salt bed was already uncontrollably dissolutioned by
      brining activities.
                                     316

-------
    Hydrologic evidence supports the presence of at least one major
confining bed near the top of the Manlius Limestone (Fig. 2). Large
artesian pressures have been encountered when penetrating this horizon
during drilling.  Recent subsidence and fracturing has led to
interformational  mixing of formerly separate hydrologic flow systems and
greatly increased the natural recharge to all subsurface formations.
Increased recharge meant that by the late 1950's it was no longer necessary
to pump fresh water down into the salt horizons to obtain brine during
mining operations.

    Further evidence of the extent of subsidence-induced fracturing and
interformational  mixing is found by comparing potentiometric surfaces
between pumping brine wells and glacial aquifer wells. For example, brine
wells H-7 and MV-5 are 432 and 428 meters deep respectively, while CATO-4
penetrates deep glacial deposits some 1.8 kilometers from well H-7.
Measurement of the potentiometric surfaces of these wells over a period of
four years by Allied-Signal and the U.S.G.S. reveals similar mimicking
responses to brine pumping, although they are some 1.2 kilometers apart and
differ in depth by approximately 300 meters. During a period of high brine
field pumping (August 1985) the water table in CATO-4 was drawn down 15.7
meters below its  "normal" flowing elevation. At the same time the
potentiometric surface in H-7 was drawn down approximately 38 meters.
Continued observations revealed that the magnitude of fluid level
fluctuations in these wells also significantly decreased when pumping rates
decreased.

    The similar hydrologic response of bedrock and soil ag_uifers to pumping
suggests that the ==300 m between them is now hydraulically connected by the
increased permeability resulting from subsidence-induced fracturing. The
damage to the structural integrity of beds beneath and adjacent to the
brine fields and  the resultant interformational mixing may now be so
extensive that they are irreparable. Similarly,  the piracy of surface water
from Emerson Gulf and the increased recharge through formerly competent
lacustrine clays  and bedrock must increase the flow through subsurface
formations.

                 Appearance of Mud Boils and Sand Volcanos

    Some 2 kms downvalley of the brine fields, in a smaller settlement
area, sand "volcanos" effuse weakly saline groundwater that flows into the
Onondaga Creek. Projection of geologic units described in distant wells
into this area and observation of the growth pattern (N to S, and E and w)
of these mud boils suggest that they may mark the present location of the
subsurface salt front. Figure 4, time sections 1-3, depict the natural
southward migration of the salt front.

     An early newspaper documents the presence of a sand volcano at Otisco
Road (Fig. 1) in  1899. These early sand volcanos were areally limited.
Perhaps they were a reflection of the earliest down valley effects of the
solution mining operation. The number of mud boils and their areal extent
have substantially increased since the onset of brining operations.
Individual effusion features vary in size from small mud boils to broad,
flat cones (sand  volcanos) measuring 12 m in diameter and 1 m in height.

    Historic accounts and the interpretation of historic aerial photography
reveal that the mud boil areas expanded rapidly during solution mining
operations. Sometime between July 27, 1936 and October 15, 1951, the clay
fraction of the mud boil effluent caused Onondaga Creek to change from
clear to turbid.  Onondaga Creek now has the appearance of chocolate milk
for the =26 kms it takes to reach Onondaga Lake.
                                  317

-------
     Time  Section  1  --  (>125,000  years ago)

     d       Tully  Valley pre-Wisconsinan glaciation                  d
     a:                                                               a:
SOUTH |                                 |g               NORTH  SOUTH <

     8      BRINE FIELD                   u3  *
                                       sS  R
                                             Approx. 15 km
                                 BENJAMIN    £
                                 ARM WELL C  o
                                              Down valley
                                                                     Time  Section  2  —  (<10,000  years  ago)
                                                                                 Tully Valley Postglacially
                                                                                      (soil deposits in place)
                                                                                                                       NORTH
SOUTH <
            Time  Section  3--(circa  1900)
                  The beginning of salt mining
                    operations  in  Tully  Valley
          Shdles
               10
               Limestones
 Lacustrine Clays — Otisco Clay
 Sand and Gravel or Till

- inferred contact between clay deposits
   and sand and gravel deposits

 Oriskany Sandstone
                                     V*

                                     \
                                          Subsidence Fracture with
                                          infiltrating groundwaters
                                         Confined groundwater

                                         Infiltrating surface "waters
                                         Mixed groundwaters
                                         Salt Front - headward
                                          retreat due to dissolution

                                         Brine
                                                                          Time  Section  4—(circa  1950)
                                                                              Collapse  of  the brine  field area.
                                                                          New circulation flow path(s)  established
                                                        NORTH   SOUTH <
Cm

60

100
                                                                                    SCALE
                                                                                    1000
                                                                                            2000 meters
                  Figure 4:     Time sequence diagrams showing the flow regimes and sequence of events
                                that led to the present day.  See text for further discussion.
                                                                                                                        NORTH
                                                                                                                      CARDIFF
                                                                                                                       WELL

-------
    Massive subsidence in the recharge area coincides in time with the
greatest sand volcano activity.  The largest and most prolifically effusing
sand volcano field,  approximately 400 meters south of Otisco Road, was not
present as of April  28,  1967. By April 29, 1972 it measured an irregular =
60 m by 25 m; by March 28, 1981, = 160 m by 73 m; and on November 29, 1989,
191 by 175 m. Continued settlement, to a depth of 6 m, has further enlarged
this area in the last two years. Much of the growth of the mud boil areas
appears to develop along the strike of the inferred location of the salt
front.

    The physical setting of the  mud boils is inconsistent with natural
springs.  Springs typically occur in one or a few closely spaced discharge
points and are stable in size and discharge location over time. The
artesian Tully Valley mud boils  have migrated rapidly southward along
Onondaga Creek, and  southwestward along a tributary of Onondaga Creek in
the last 80 years. This is not in keeping with known short term geologic
phenomenon.

            Chemistry  of Groundwaters  and  Sand  Volcano Effluents

    The chemical relationships among uncontaminated groundwaters in Tully
Valley and sand volcano effluents (SVE) can be used to identify the source
of SVE. Stiff diagrams were found to have limitations when trying to define
the chemical relationships between groundwaters and SVE. Because there are
many groundwaters that are potential sources, and most have distinctive
chemistries, multivariate models were necessary to differentiate among
them. The multivariate models give quantitative results in the form of
significance levels  of chemical  discrimination and calculated mixing
proportions of groundwaters.

      The chemical analyses were taken from three main sources; the New
York State Department of Law; the New York State Department of
Environmental Conservation; and  a thesis by Noble (1990). Many samples from
Noble (1990) were taken just outside Tully Valley, but from the same
geological formations. Chemical  analyses with > 5% error in ionic balance
and > 10% error in mass balance  were deleted from the sample database.

    Water samples were classified into hydrostratigraphic units based on
geologic interpretation and well logs. The five units present in Tully
Valley in order of increasing depth (Fig.  2) are Glacial Till, Devonian
Shale, Silurian-Devonian (SD) Carbonate (Onondaga-Devonian/Silurian
carbonates), Evaporite (Silurian shale-Syracuse Fm.), and Vernon Shale
(Noble, 1990). The compositions  of dilute groundwaters in the Glacial Till
and Devonian Shale units are similar,  therefore samples from these units
are grouped together with the name Near-Surface waters.  Average
compositions of water from these four units and the SVE are given in Table
1. The SVE contain significant halite and gypsum components.

Statistical Tests

    If SVE derive from a single  hydrostratigraphic unit, their compositions
must be the same within error as samples from that unit. Because samples
were analyzed for >  8 chemical species, a multivariate technique called
multiple discriminant analysis was used to test for statistical differences
in composition between each unit and SVE.  Multivariate statistics give at
least as good a discrimination as any univariate technique. Although some
variables (species concentrations) have non-normal distributions for some
hydrostratigraphic units (log transformations do little to correct this),
this should not affect the results of the discriminant analysis, although
the significance levels obtained may be questionable  (Le Maitre, 1982).
                                  319

-------
                                  TABLE 1

          Average Chemical Compositions of Waters in Tully Valley*
Element
Near-
Surface

 n=48
SD Car-
 bonate

  n=5
Evap-
orite

 n=5
Vernon
Shale

 n=6
                                                 SVE
                                                 n=6
Na


Ca


Cl


S04


Sr*


Mg'


HCO3*
1.05
(0.66)
37.4
(64.2)
73.5
(34.3)
54.4
(80.9)
35.1
(35.7)
0.69
(0.96)
17.3
(9.15)
271
(85.3)
0.94
(0.46)
18.2
(12.2)
207
(180)
51.6
(30.2)
320
(405)
3.0
(1.3)
32.6
(17.2)
314
(35.1)
232
(20.2)
110,000
(10,500)
1880
(198)
171,000
(19,900)
2300
(1150)
76
(11)
187
(106)
75.7
(50.9)
11.1
(12.4)
51.0
(51.6)
392
(144)
117
(130)
829
(390)
5.4
(2.5)
32.3
(16.0)
274
(73.4)
6.47
(2.00)
1800
(963)
241
(64.1)
3500
(1610)
329
(136)
84
(16)
126
(29.5)
90.0
(51.4)
Model
 #15
                                                          6.47
                                                          1900
                                                  241
                                                 2980
                                                  329
                                                  4.0
                                                 37.1
                                                  401
Model
 #2*
                                                         6.48
                                                         2650
                                               239
                                              4160
                                                350
                                                4.6
                                               35.4
                                               306
    Concentrations in mg/1;  standard deviations in parentheses.
    Element not included in model calculations (not conserved).
    Model #1 proportions of end-members (calculated using non-linear least
    squares):  SD  Carbonate = 0.91, Near-Surface =  0.062,  Evaporite  =
    0.024,  SSR (sum of  squares of residuals)  = 2.3 E6.
    Model #2 proportions of end-members (calculated using linear
    programming and multiple linear regression): SD Carbonate =  0.38, Near-
    Surface =  0.53, Evaporite =  0.017, Vernon Shale =  0.11,  SSR  = 3.9 E5.
     The objective of discriminant analysis  is to  find  functions that
maximize the differences between chemical groups.  Specifically, the
functions  are  linear combinations of variables  that  maximize the ratio  of
between-group  to within-group variances  (Le Maitre,  1982).  Here the groups
are  the four hydrostratigraphic units  and SVE and the  variables are the
concentrations in mg/1 of the ions Na, K, Ca, Mg,  Cl,  and SO4.  Use of
element ratios did not improve the discrimination between groups.
Calculations were made using the program DISCRIM  in  the SAS software
package  (SAS Institute, © 1988) using  equal weighting  for each group.

     Over 99% of the variability in the sample data set is accounted for by
the  two discriminant functions CAN1 and  CAN2:

     CAN 1  = 0.0011*Na + 0.045*K - 0.00033*Ca -  0.0063*Mg -
            0.00043*C1 - 0.0016*SO4

     CAN 2  = 0.000025*Na - 0.0075*K + 0.011*Ca + 0.0083*Mg
            -0.00014*C1 + 0.00070*SO4
                                  320

-------
Species  concentrations are in mg/1. The angle between these two vectors  is
90°,  which  allows plotting of the values of CAN1 and CAN2 for each sample
on an x-y diagram (Fig. 5). Waters from Near-Surface and SD Carbonate are
not significantly different. However, compositions of SVE and waters  from
the four hydrostratigraphic units are distinctive. The F-statistic shows
that SVE are  significantly different from all units at the 99% significance
level. The  SVE therefore cannot form by direct sampling of a single unit.

    If waters in SD Carbonate are more saline than our analyses indicate,
SVE may  be  derived directly from this unit. Samples taken from the SD
Carbonate unit during drilling of the M-series wells (Phalen, 1928) were in
fact more saline than modern samples taken from wells penetrating SD
Carbonate.  However, multiple discriminant analysis shows that the SVE and
M-series samples are compositionally different at the 99% significance
level.

    Since SVE cannot be derived from a single hydrostratigraphic unit, they
must represent a mixture of waters from the various units (chemical
end-members). In Fig. 5, lines connecting the end-members define a mixing
polygon. Any  mixtures formed from the end-members must lie within the
polygon. The  location of average SVE suggests mixing of Vernon Shale  + SD
Carbonate + Near-Surface ± Evaporite. The next section discusses
quantitative  tests of these general conclusions.
            8
            7
            6
       CN   5
       O   3
            2
            1
             Sand Volcanos
             Surface  Brines
SD Carb
                                          Near—Surface
             -3      -2      -1       0
                                 CAN1
    Figure 5:  Chemical variation diagram relating compositions of the sand
volcano effluents  and surface brines to the end-member groundwaters in
Tully Valley (Evaporite  is off-scale). CAN1 and CAN2 are the functions
obtained from  the  multiple discriminant analysis (see text). Mean values
for SVE and surface brines (filled circles) and end-members (open circles)
are plotted with ± 1-sigma error bars. Lines connecting the end-members
define the mixing  polygon.
                                   321

-------
Mass-Balance Calculations

    The process of groundwater mixing can be modelled by mass-balance  if
species behave conservatively. In the mass balance model, the  average
composition of SVE represents the mixture and the average compositions of
the four units represent the end-members (Table 1). Species  are  conserved
if no dissolution or precipitation of minerals occurs between  mixing and
sampling. On short time scales precipitation is more likely  to occur,  but
only if the mixture is saturated in a mineral, i.e., the saturation index
(SI) is > 1. Calculation of SI for various minerals (Truesdell and Jones,
1974) shows that SVE are saturated only in calcite, dolomite,  and
strontianite. Species other than Ca, Mg, Sr and HCO3 should behave
conservatively. Because there must be as many species as end-members in the
mixing calculations, we were forced to use Ca as a variable, but the fact
that the Ca data fit the model suggests that Ca behaves conservatively.
Magnesium and Sr have anomalously high concentrations in SVE,  and
comparison of analyses of filtered and unfiltered samples suggest this
results from clay particles that remain suspended even after filtering to
0.45 urn.

    Two different models were used to calculate the mixing proportions, Xjf
of the end-members. The required constraints are that all Xj  >  0  and ZXj =
1. Model #1 is a non-linear model based on the Marquardt-Levinson algorithm
(Press et al., 1985). The objective function minimized was S(M; - SX^)2,
where Mi is the concentration of element i in the mixture and  Cy  is the
concentration of element i in end-member j. Model #2 is a linear model that
uses multiple linear regression to minimize the same objective function.
However, since there is no way to constrain all Xj > 0 using  multiple
linear regression, the first step is to choose those end-members that  can
mix in positive proportions to form the mixture (Wright and  Doherty, 1970).
This is accomplished using linear programming based on the simplex
algorithm (Press et al., 1985). The optimal proportions of the chosen
end-members are then calculated using multiple linear regression. The
disadvantage of this approach is that the objective function for the linear
programming step (minimize E(M; -SXjCy)) is different from that  for the
multiple linear regression step. However, this discrepancy is  unlikely to
have any effect when choosing the best mixing model.

    Table 1 shows the results from Models #1 and #2. Although  the
calculated proportions of end-members are different, both models fit the
data within error. Model #1 gives the proportions 0.91 SD Carbonate, 0.062
Near-Surface, and 0.024 Evaporite with SSR (sum of squares of  residuals) =
2.3 E6. Model #2 gives the proportions 0.38 SD Carbonate, 0.53
Near-Surface, 0.017 Evaporite, and 0.11 Vernon Shale with SSR  =  3.9 E5.
Note that the relative proportions of SD Carbonate and Near-Surface have no
meaning, since the compositions of these two end-members are not
significantly different (Fig.5). Also, the stated proportions  of
end-members were calculated for average SVE, but the same calculations for
each sand volcano effluent show highly variable proportions  of end-members.
This is to be expected in such a complex mixing scenario.

    The fact that the solution to the mass balance problem is  non-unique
(i.e., averaged SVE chemistry can be explained by several different models)
is not surprising, since there are too few high quality constraints on the
mixing problem. This results from having too many end-members  and not
enough elements. Also, the variance of each element is too large,
distributions of some variable are non-normal, some elements are
correlated, and some end-members have similar chemistries. More  data are
required before a choice can be made between mixing models.  However, we can
conclude that SVE waters form from a mixture of groundwaters that are
present beneath the sand volcanos, apparently to a depth of  at least 335 m.
This figure represents the depth to the Vernon Shale, as projected into
this area from distant wells.
                                  322

-------
    Many wells down valley of the sand volcanos at Otisco Road have
encountered dilute brine waters after penetrating the confining Otisco
Clay.  Similar waters effuse from a large spring called the Sulfur Hole  (=
3500 m north of Otisco Rd. and = 500 m west of Tully Farms Rd.), which has
been present near the valley flank for at least 130 years. We assume that
its waters probably rise along the clay/valley flank contact or through
fractures in the clay where it thins out. Discriminant analysis of these
surface brines reveals that they are chemically similar to the Vernon Shale
(Fig.  5), yet distinct from SVE at the 99% significance level. The average
chemistry of these waters is modelled as a mixture of 97.4% Vernon Shale
and 2.6% Evaporite. Thus, the natural brines appear to rise undiluted from
the Vernon Shale and up-valley Evaporite units because these units are
closer to the surface than at Otisco Road and because the uppermost
confining beds have been erosionally removed.

                       Valley Axial Groundwater Flow

    Discharge from the sand volcanos responds dynamically with significant
infiltration in Tully Valley. This may result from rapid flow through
highly transmissive bedrock conduits or by exertion of a large pressure
head on confined porous media.

    A more complete geologic column to the south limits natural recharge to
brine-rich bedrock formations. Similarly, the thick lacustrine clays near
the surface of the valley act as a confining bed, severely limiting natural
surface infiltration. Several possibilities for the saline discharge flow
routes exist; all of which require substantial system recharge to highly
transmissive formations (Fig. 2).

    Groundwater in deep regional flow systems moves slowly (Fetter, 1988).
Rapid flow through deep-seated formations is rare, unless there is
substantial fracturing at both the recharge and discharge ends of the flow
system. Naturally-occurring fractures are present at the up-dip end of the
Tully system, but the mining has provided access routes at the down-dip
(brine field) end (Fig. 2). To generate rapid flow through intervening
tight bedrock formations requires pathways such as fractures and/or the
dissolution of soluble geologic units (e.g., carbonates or evaporites).

    A surcharged pressure head acting on highly fractured limestones is one
flow route option (e.g., brine field to volcanos). However, the rapid
increase in volcano discharge following system recharge would, if
transmitted from the brine field to the volcanos, require solution
conduits. The natural transmissivity of fractured limestone would not be
expected to communicate a hydraulic pressure wave over any appreciable
distance due to pulse damping under laminar/non-open conduit flow
conditions. Neither should the increased discharge from the sand volcanos
originate from valley flank sources, because recharge waters would have to
breach density stratified waters and one or more confining beds (e.g., clay
and/or bedrock). Fracturing of the thick Otisco clay along the valley flank
contact (e.g., Sulfur Hole area) would,  if present, probably serve to
discharge artesian valley flank and axial pressures rather than serve as a
recharge site.

    In evaluating whether the limestones would be preferred flow routes one
would have to assess the likelihood that solutional conduits were present
pre-glacially between the brine field and the volcanos. Solutional conduits
develop in limestone only where a pre-existing network of integrated
openings connect the recharge and discharge areas (Palmer, 1991).
Furthermore, carbonate dissolution requires that undersaturated water
remain in contact with soluble fracture or conduit walls. A minimum of
10,000 years is required to widen joints and partings enough to produce
significant permeability (Palmer, 1984;  Dreybrodt, 1990; Palmer, 1991). A
typical water sample obtained from the Devonian shales contained 51.9 mg/1
calcium, or 129.6 mg/1 CaCO3  equivalents.  For this concentration,  the rate
                                  323

-------
of wall retreat (e.g., joint enlargement) in limestone is calculated  to be
0.05 cm'/liter  in  10,000  years  (Palmer,  pers.  comm.)•  Based on an
evaluation of these considerations, it is our opinion that geologic
conditions have not been favorable pre- or post-glacially for the
development of significant solution conduits in the limestones.

    The authors evaluated several other lines of reasoning in assessing the
likelihood that the main flowpath between the brine fields and the sand
volcanos occurs in the limestones. This included an evaluation of natural
infiltration pathways for calcite undersaturated waters. Geologic logs
reveal that little or no exposed limestone was present up valley of the
sand volcanos pre-glacially. Furthermore, much of the mud boil subsidence
has occurred along bedrock strike, which is inconsistent with typical
downdip vadose cavern development in this type of geologic setting (Palmer,
1991). Not only are the effusion areas migratory in nature, they exceed the
largest natural sinkhole size in New York State by well over an order of
magnitude. Conditions often attendant to sinkhole formation, such as
aquifer dewatering, were also not present. The variable mud boil discharge
in recent time would not be expected from a naturally deep-fractured karst
aquifer system if it was physically separated from large scale recharge.

     The hypothesis might be advanced that the artesian pressures and
volcano effluents may derive from infiltration entering alluvial fans at
the outlets of either Rattlesnake Gulf or Rainbow Creek (Fig. 1). The
chemistry of the SVE would require this water to breach density stratified
brines and discharge up-valley against the regional hydraulic gradient. In
addition, significant recharge would have to completely penetrate the thick
confining Otisco clays. It is the combination of these unlikely events and
the areally wide and migrating distribution of the mud boils that virtually
removes all potential sources of system recharge other than the brine field
settlement area. It is more likely that any infiltration into the
Rattlesnake Gulf alluvial fan flows a short distance below the ground
surface, daylighting in the Ram Pump Spring area (Fig. 1).

    The variable discharge rate and the chemistry of SVE are critical to
our understanding of likely source areas. The presence of gypsum and halite
components in the effluent is a clear signature telling us that at least
some volcano effluent must derive from the deep evaporite unit. The salt
content of SVE is diluted by shallow groundwater in sediments and bedrock
in the valley. Geochemically, the SVE represent mixtures of waters from all
of the hydrostratigraphic units beneath the sand volcanos to a depth of 335
m. SVE have moderately variable salinities, but groundwaters from the
hydrostratigraphic units show much less variation. The chemistry of SVE
suggests fluctuating amounts of dilution of deep saline and gypsiferous
waters by near-surface fresh waters. The brine must originate at a location
having a higher hydrostatic head than at the sand volcanos. Mixing occurs
during ascent along vertical pipes (joint or fracture flow paths) with
discordant hydraulic efficiencies.

    Two separate flow regimes are probably responsible for the response of
volcano discharge following recharge events; deep evaporite and shallow
confined sediments. The high salinity of SVE observed during storm events
may indicate the development of a subsurface flow route, first through
fractures in the Vernon shale and then through salt dissolution and conduit
formation along the shale/salt contact (Fig. 2). Density-driven circulation
initiating at the brine fields may now rise along the fractured U-shaped
(areally) salt front beneath the volcanos. Jenyon (1986), discusses the
natural circulation systems attendant to salt fronts. Formation of
fractures at the salt front  (Fig. 4, sections 1-3 and Fig. 2) normally
results in infiltration and downvalley flow. Instead, water rises at the
mud boils, almost as if the flow system was artificially short-circuited
(Fig. 4, section 4). The large sub-vertical arrow in Time Section 4 of
Figure 4 indicates upward flow (not against the regional flow gradients)
along salt front fractures resulting from this short-circuited circulation.
                                  324

-------
Alternately,  the rise of saline waters south of Otisco Road may directly
relate to the erosional removal of critical confining beds near the sand
volcanos. Whatever the flow path of the saline volcano component, the storm
discharge of  the volcanos requires conduits or a confined porous media
capable of rapidly transmitting a recharge pressure pulse.

        Several flow routes have been hypothesized to explain the rapid
increase in volcano effluent following recharge events, but the actual
pathway is unimportant. Visual observation of increased volcano discharge
during periods of high infiltration indicate that the source area responds
in tandem with surface recharge, operating through and beneath the thick
Otisco clay confining bed.  It is unlikely that significant storm
infiltration  through this clay can naturally occur to cause the increased
discharge observed at the mud boils. It is more likely that the sudden
surcharge on  the hydrologic system from extensive infiltration in the brine
field settlement basin acts as a pressure pulse. Rubin has proposed several
means of verifying these relationships.

    The sudden increase in pressure head, exerted through one or more
saturated formations, is partially released at the mud boils. While some of
the volcano discharge originates in the evaporite units,  much of the
observed discharge may reflect a pressure wave being transmitted through
the sandy gravel aquifer confined beneath the Otisco clay.  As sand is
entrained and removed from below the clay, a void develops until the
mechanical strength of the roof is exceeded. Ring fractures develop, the
subsidence area grows, and the efficiency of the exit pathways continues to
increase. The SVE are piped upward to an elevation well in excess of the
shallow water table common to valley centers. The only viable explanation
for the storm discharge observed at the sand volcanos (Fig. 6) is from
recharge through severely fractured brine field clay and/or bedrock. Here,
interformational mixing of groundwaters from massively fractured clay and
bedrock formations has resulted in a vast pressure head on a saturated flow
network.

    Fluids from the deep saline beds are successfully diluted at this time.
It is the quantity of discharge that is of concern. This turbid water
degrades Onondaga Creek, Onondaga Lake, the aquatic ecosystem, and the
aesthetic quality of the valley.


                                Conclusion

    Solution  mining operations resulted in large-scale subsidence.
Subsidence-induced fractures serve to significantly recharge formerly
confined flow systems. This increased recharge has resulted in the
formation of  sand volcanos downgradient. The bulk of the sand volcanos
appear to have formed by a sequential combination of: 1)  increased recharge
to formations above solutioned out evaporite beds that has unnaturally
surcharged the flux through geologic units below the brine fields. This
flow through  has created an unnatural circulation pattern that vents upward
at the sand volcanos. Pumping of brine increased recharge through formerly
competent and confining beds leading to the 2) solution of salt and the
transmission  of a recharge pressure pulse through one or more geologic
horizons. This water 3) rises under artesian pressure through pre-existing
fractures underneath the sand volcanos, 4) mixing with other groundwaters
during ascent. Rising waters 5) suspend unconsolidated sediments  (e.g.,
clay, silt, and sand) and discharge to Onondaga Creek via the sand
volcanos.
                                  325

-------
            lit
     Figure 6: Typical  Tully Valley sand volcano.
                Diameter is approximately 2 meters.
                                Acknowledgement

    Preparation of this manuscript  occurred while the senior author  was
employed  in the Environmental Sciences Division at the Oak Ridge National
Laboratory. ORNL is  managed by Martin Marietta Systems,  Inc., for  the U.S.
Department of Energy under contract DE-AC05-84OR21400.
                              "The  submitted  manuscript has been
                              authored by a contractor of the U.S.
                              Government under contract  No.  DE-
                              AC05-84OR21400. Accordingly, the U.S.
                              Government  retains  a  nonexclusive.
                              royalty-free license to publish or reproduce
                              the published form of this contribution, or
                              allow others to do so. for U.S. Government
                              purposes "
                                References Cited

Dreybrodt,  W., 1990,  The role of  dissolution kinetics  in the development of
       karst aquifers  in limestone:  A model  simulation  of karst  evolution:
       Journal of Geology, v. 98,  no. 5, p.  639-655.
Fetter,  C.W., (1988),  Applied Hydrogeology,  Columbus,  Merrill Publishing
       Company.
Getchell,  F.A., 1983,  Subsidence  in the Tully Valley,  New York,  M.S.
       Thesis, Syracuse University,  140p.
Jenyon,  1986, Salt  Tectonics, Elsevier Applied Science Publishers Ltd.,
       London, 191 pp.
                                     326

-------
Le Maitre,  R.W.,  1982, Numerical Petrology: Statistical Interpretation of
      Geochemical Data. Elsevier Publ. Co., Amsterdam.
Mullins,  H.T.,  Wellner, R.W., Petruccione, J.L., Hinchey, E.J., and Wanzer,
      S.,  1991,  Subsurface geology of the Finger Lakes region, Ebert, J.R.
      (ed.),  New York State Geological Association 63rd Annual Meeting
      Field Trip  Guidebook, p. 1-54.
Noble, J.M.,  1990, A reconnaissance of the natural groundwater
      geochemistry of Onondaga County, New York, Unpublished MS Thesis,
      Syracuse University, 129 pp.
Palmer,  A.N., 1984, Geomorphic interpretation of karst features in
      Groundwater as a geomorphic agent, LaFleur, R.G. (ed.), Allen and
      Unwin.  Inc., London, p. 173-209.
Palmer,  A.N., 1991, Origin and morphology of limestone caves: Geological
      Society of America Bulletin; v. 103, p. 1-21.
Phalen,  W.C., 1928, Memo MK-87, Underground water conditions in new wells
      at Tully, The Solvay Process Company.
Press, W.H.,  Flannery, B.P., Teukolsky, S.A., and Vetterling, W.T., 1985,
      Numerical Recipes: The Art of Scientific Computing, Cambridge
      University Press, New York.
SAS, 1988,  SAS Institute Inc., Gary, N.C., 27512, USA.
Truesdell,  A.H.,  Jones, B.F., 1974, WATEQ, a computer program calculating
      chemical equilibria of natural waters, U.S. Geological Survey Journal
      of Research, v. 2, p. 233-248.
Tully, W.P.,  1983, Relationship of brining operations in the Tully  Valley
      to the behavior of groundwater and geologic resources, Summary
      Consultants Report, prepared for The Solvay Process Allied Chemical
      Corporation, Camillus, NY.
Wright,  T.L., Doherty, P.C., 1970, A linear programming and least  squares
      computer method for solving petrologic mixing problems, Geol. Soc.
      Amer. Bull., v. 81, p. 1995-2008.
                                   327

-------
                  Solution Mining And Resultant Evaporite
                Karst Development In Tully Valley, New York

            Paul A.  Rubin, John  C. Ayers,  and  Kristin A.  Grady


1.      What is the likelihood that the sand boils are a result of the
       expulsion of water due to progressive collapse of cavities created
       by ongoing solution along the upper portion of the salt by water
       entering at cracks?

       The process of cavity collapse, exemplified by sinkhole formation in
       limestone karst settings,  occurs at either an extremely slow rate as
       soil is sapped downward and removed or as a catastrophic event.  The
       downward sapping of sediments and rock debris from above the salt
       horizons would occur so slowly that it could not be expected to
       continuously displace quantities of water.  If catastrophic cavity
       collapse were the driving force behind sand volcano effusion, then
       sand boils would form instantaneously, resulting in a high effusion
       rate immediately following cavity collapse.  This effusion rate
       would rapidly drop to zero.  Without a large and continuous fresh
       supply of rock or sediment to displace groundwater, the sand boils
       would soon be removed by erosion.  Therefore, the likelihood that
       the sand boils are a result of the expulsion of water due to
       progressive collapse of cavities is unlikely.  Instead, we see a
       slow, steady growth of the sand boils, and although the effusion
       rate may be variable,  there is always significant flow.


2.      There is an incredibly vast,  international literature on the effects
       of both natural and anthropogenic solution of salt in the
       subsurface.   I know of at least 100 references and could probably
       find 1000.  Some of the most useful that come to mind are:

          Calvert,  A.F., 1915. Salt in Cheshire, Spon, London, 1206 p.
       (includes many photos).
          Evans, W.B. et. al., 1988.  Geology of the country around
       Macclesfield — Geological Survey of Great Britain, Memoir (plus a
       more recent memoir on Cheshire published 1987±).
          Reuter, F. and Tolmanacer,  V.V. 1990. Bauen and Bergbau in
       Senkungs und Erdfallgebieten.  Schriftenreihe fur Geogische
       Wissenschaften, Akademie Verlag, Berlin, No. 28., 176 p.
          Weber, H. 1952. Pliozan und Auslaugung im Gebiet der Oberen
       Werra,  Geologica Akademie Verlag, Berlin, v.8, 136 p.
                                  328

-------
  Session V:
Case Histories

-------
330

-------
 Hydrogeology and Ground-Water Monitoring of an Ash-Disposal

      Site at a Coal-Fired Power Plant in a Karst System


    Shelley A. Minns1, Arsin M. Sahba1,  Lyle V.A.  Sendlein1,
             James C. Currens2 and James  S.  Dinger2

            Department  of  Geological  Sciences  and
                 2Kentucky Geological Survey
          University of  Kentucky,  Lexington,  Kentucky
Abstract

     A coal-fired power plant located along a lake in south-
central Kentucky is situated on karstified Mississippian-age
carbonates.  Ground-water flow and chemistry are evaluated to
assess impacts of current and past coal-ash-disposal practices.
Ash disposal in an unlined, active slurry pond is allowing ash
and pond water to enter the ground-water system through
underlying sinkholes and fractures.  Water infiltrating through
20-year-old ash in a closed-out slurry pond enters the ground-
water system similarly-  Ground water discharges at springs
near lake pool elevation.  Sinkholes, swallets, and springs
were field mapped, and a dye-tracing plan was implemented.  Dye
traces indicate separate flow systems for each disposal site.
Dry fly ash was leached to determine constituents present in
fresh ash.  Water-quality analyses were obtained from
upgradient and downgradient springs and surface-water sources.
Wells installed in the closed-out slurry pond were sampled.
Monitoring data indicate that major ion composition in the
active slurry pond is more similar to the lake water used for
the ash slurry than natural ground water.  Active-pond water is
elevated in sulfate, barium, boron, copper, iron, manganese,
and zinc compared to background waters.  Spring discharge
associated with the active slurry pond shows major ion
composition intermediate between pond water and natural ground
water.  Boron, iron, manganese, and zinc are elevated in these
springs.  Spring discharge associated with the closed-out
slurry pond is similar to background springs except for
increased concentrations of boron and iron.  Water samples
collected from wells drilled into the ash are similar in major
ion composition to background springs but show elevated
concentrations of barium, boron, iron, manganese, and vanadium.
                             331

-------
Introduction

     The purpose of this research is to study the ground-water
flow system and chemistry in a karst aquifer underlying a coal-
fired power plant.  The results of this research will be used
by the power company to develop a ground-water-monitoring
program which meets the requirements of the state regulatory
authority.
     Coal ash is an aluminosilicate glass which forms from the
burning of coal at temperatures of approximately 1500 degrees
celsius (Elseewi, 1980).  Ash is composed primarily of silicon,
aluminum, iron, and calcium oxides.  Lesser amounts of sodium,
magnesium, potassium, and sulfur are typically present
(Cherkauer, 1980; Elseewi, 1980).  The major constituents are
derived from thermal breakdown of minerals, primarily illite
and kaolinite.  As a result, the composition of coal ash is
highly dependent on the composition of the pre-combusted coal.
     Coal ash consists of two products.  Fly ash is fine
grained and often forms concentrically layered spheres.  This
fraction accumulates in electrostatic precipitators and
comprises approximately 20 percent of the total ash fraction at
the study site (Rai and Zachara, 1990, personal commumication).
Bottom ash is coarser grained material which falls to the
bottom of the combustion chamber.
     Elements associated with ash may be incorporated into the
aluminosilicate matrix of individual particles or may
accumulate on the outer surface of the spheres.  Many elements
volatilize during combustion and subsequently condense on the
outer surface of the cooling particles.  Antimony, boron,
fluoride, selenium, arsenic, cadmium, lead, zinc, barium,
chromium, copper, chlorine, sulfur, and vanadium are often
preferentially enriched on ash particles in this manner
(Ainsworth and Rai, 1987; Rai, 1987).  Aluminum, cobalt, iron,
manganese, nickel, strontium, and vanadium are also associated
with coal ash, though not generally enriched on the surface of
the particles (Wu and Chen, 1987; Rai, 1987).


Setting

     The power plant was constructed in 1964 and houses two
generating units which produce a total of 320 megawatts of
electricity.  Maximum coal usage is 5,000 tons per 24-hour
period.
     The site has two ash-disposal facilities (Figure 1).  From
1964 to 1978, coal ash was sluiced to a 40-acre slurry pond.
This disposal facility was retired in 1978 when its storage
capacity was reached.  The maximum thickness of ash in this
facility is 35 feet.
     Since 1978, ash has been disposed of in an unlined pond
that has an area of approximately 50 acres.  Fly ash and bottom
ash are transported to this facility as an ash-water slurry via
a 3,500-foot-long pipeline.
                              332

-------
                               ACTIVE
                               ASH PDND
4
                                CLDSED-DUT
                               SLURRY  PDND
Figure 1.  Map showing location of power station and ash-disposal
facilities.
     The plant is located in  the Mississippian Plateau
carbonate region of  Kentucky,  adjacent to a major impounded
river.  The topography is a rolling, upland, karst plain which
exhibits subsurface  drainage.   In the immediate vicinity of the
plant, the land surface is dissected by small v-shaped valleys.
Surface flow in these  channels is ephemeral and often
disappears into swallets in the stream bed.  The land surface
drops steeply, often vertically, from the upland karst area on
which the plant is situated to the  surface of the lake.
     Rocks underlying  the site are  primarily limestone and
dolomite (Figure 2).  Structures associated with the plant are
constructed on the Ste. Genevieve Limestone and St. Louis
Limestone.  The Salem  and Warsaw Formations (undifferentiated)
crop out a few feet  above permanent lake pool elevation.
     The Salem and Warsaw Formations, which are considered
together because of  their lithologic similarity, are the
stratigraphically lowest formations that crop out near the
plant.  Total thickness ranges from 30 feet to 75 feet (Lewis,
1974; Taylor, 1975).  The two formations represent an
intertonguing sequence of dolomite, limestone, shale,
siltstone, and sandstone, which shows significant lateral
variation.  The exposed section of  Salem and Warsaw near the
                              333

-------
                                    Mbha
                                     N
(PbO   Breathltt Fn.

(Mp)  Pennlngton Fn,

(Mbha)  Bangor Ls. &
       Hartselle Fn,

(Mnk)  Kldder Ls, Mbr,


-------
                           ACTIVE
                           ASH  PDND
                                           LEGEND
                                                  DYE INPUT
        o •
ACTIVE-POND •
SPRINGS  •
                                                  WELLS

                                                  SPRINGS
                            CLUSED-DUT
                           SLURRY PDND
                                                   SAMPLED
                                                   SPRING
                                                O SINKHOLE
                    POWER
                    STATION
Figure 3.   Hydrogeologic features  map.
     A survey was conducted to locate springs  potentially
related to subsurface drainage from the site.   The survey began
in late winter when lake levels were highest.   As summer
approached,  lake levels declined and additional springs became
visible.  Mapped springs are shown on Figure 3.
     A submerged spring situated about 22  feet below permanent
pool elevation was noted by plant employees during the time the
closed-out slurry pond was an active facility- It was thought
that this spring may represent a major discharge point from the
conduit system associated with the closed-out  slurry pond and
that stratigraphically higher springs may  be wet-weather
overflows.  During early spring of 1990, the spring was
documented when a conductivity plume was delineated relative to
ambient lake conductivity at the same depth. In July 1990, an
attempt was made to locate this spring using SCUBA but the
attempt was not successful.
     By October 1990, the lake level had fallen and exposed a
small seep at the top of the agillaceous dolomite zone in the
Salem and Warsaw Formations.  It was concluded that this spring
maintains a very low flow in the summer and fall and is not a
major discharge point.
     A dye-tracing program was initiated in the spring of 1990
to determine hydraulic connections between swallets and springs
                             335

-------
located around the periphery of the site.  The program was
developed with the assistance of Dr. John Thrailkill, Professor
of Geology at the University of Kentucky, and James Currens, a
hydrogeologist at the Kentucky Geological Survey.
      The fluorescent tracers Fluorescein LT (Acid yellow 73,
CI 45350) and optical brightener (Tinopal CBS-X) were chosen
for this study.  Fluorescein imparts a fluorescent green color
to water, and optical brightener fluoresces blue-white under
ultraviolet light.  These dyes have been used in the karst
regions of Kentucky by both Thrailkill and Currens.
Fluorescein and optical brightener do not interfere with one
another and can be used concurrently.
     Passive detectors were used to adsorb the tracers from the
water.  Laboratory-grade activated charcoal in nylon screen
bags was used for fluorescein adsorption.  Cotton broadcloth
test fabric stretched over a wire frame was used for adsorbing
optical brightener.
     Prior to the addition of dye into the conduit system, a
background test of fluorescence was conducted using both types
of detectors.  No evidence of potential interference with
either dye was detected.
     Several measures were taken to prevent accidental
contamination of detectors with dye.  Dye and detector
materials were not stored or handled in the same room, nor were
dye and detectors handled on the same day.  To minimize
handling of dye during the course of the project, all bulk dye
products were prepackaged into 50,  100, and 200 gram bottles
before any dye tracing was initiated.  As an extra precaution,
bottles of dye were sealed inside plastic bags.  Once the
amount of dye needed for a trace was calculated, the
appropriate amount was chosen from the prepackaged selection.
Detector placement and dye injection were always performed on
different days.
     The amount of fluorescein introduced into each point was
calculated from the following formula (Aley and Fletcher,
1976):

     Wd = 1.478 D x (Q/V)   where:

     Wd = Weight of dye in kg
     D = Distance travelled in km
     Q = Approximate discharge in mVsec
     V = Approximate velocity in m/hr

The calculated amount was rounded to the next even 50 grams.
     For optical brightener, Quinlan (1986) suggests that
doubling the amount calculated from the above formula works
well for the average Kentucky spring.  This rule of thumb was
used for determining the amount of optical brightener to
inject.
     Eight sites were selected for dye injection, four near
each disposal facility.  One detector site was located at the
submerged spring.  Detectors were attached to a rope anchored
to a 30-pound weight on the lake bottom and could be raised and
                              336

-------
lowered using small pulleys.  A float was tied to the anchor in
order to mark the site.  Use of this method allowed easy access
from a small boat.
     Seven of the eight dye traces were successful.  One
fluorescein dye trace was not detected at any of the detectors.
Since laboratory tests showed that fluorescein is strongly
adsorbed onto coal ash, it is possible that this trace was lost
as a result of adsorption onto ash which may be present in the
conduit system.  It is also possible that not enough water was
used to flush the dye through the system.
     Three ground-water-monitoring wells were installed in the
closed-out slurry pond.  Well locations are shown in Figure 3.
These wells were used to monitor water levels and to collect
water-quality data in the dewatered, closed-out facility-  Each
well was completed in ash using 10-foot-long screens, the
bottoms of which were placed at the top of bedrock.
     Water samples were obtained from each monitoring well
during March and May to analyze for non-metals.  One set of
field-filtered samples was collected for dissolved-metals
analyses.  Each well was purged prior to sampling.  When
possible, three well volumes were removed.  Low-yield wells
were purged until dry, then allowed to recover enough to permit
sampling.  A bladder pump was used for sampling when there was
sufficient water volume.  Wells with little water were sampled
using a stainless steel bailer with a bottom-emptying device.
Springs were sampled by filling a sample bottle with water
obtained from near the spring mouth.
     Water samples for dissolved-metals analyses were preserved
immediately upon collection using nitric acid.  All samples
were placed on ice immediately after collection and delivered
to the laboratory the next working day.  Water-quality analyses
were performed by the Kentucky Geological Survey laboratory.
Analyses were done according to EPA procedures described in
"Methods for Chemical Analyses of Water and Wastes."  Metals
analyses were performed using the Inductively Coupled Argon
Plasma method (ICAP).
     Fresh, dry ash samples were collected directly from the
hoppers within the plant.  Numerous leaching tests of varying
duration were conducted by Battelle-Pacific Northwest
Laboratory on fly ash and bottom ash through a range of
solid/solution ratios.  Data were obtained on elements
associated with the ash fractions and their leachability.


Results

     Dye-trace results show that the active disposal pond and
the closed-out pond are located in different ground-water-flow
systems.  Figure 4 shows the dye input points and the resulting
positive traces.
     The lower boundary of karst development in the closed-out
slurry pond flow system is located approximately 30 feet below
the top of the Salem and Warsaw Formations.  The lower boundary
is marked by an argillaceous dolomite layer (Figure 5).  Spring
                               337

-------
                                  CTIVE
                                '• ASH PDND
                                   CLDSED-DUT
                                   SLURRY PDND
Figure 4.   Locations of positive dye traces.
ft
ELEV.
(Ft.)
1000 .
950 .
900 .
850 -
800
750 _
700
650
\ BANGDR LS. &.
-V^RARTSELLE FM, 
-------
flow in the closed-out pond system varies seasonally.  The
greatest flows generally occur in the winter and spring and the
lowest flows are in the summer and fall.
     The lower boundary of karst development in the active-ash
pond flow system appears to be argillaceous zones at the top of
the Salem and Warsaw Formations.  All springs associated with
the active pond occur in the lower section of the St. Louis
Limestone (Figure 6).  Spring discharge from this system
remains constant throughout the year.
I
ELEV

-------
Table 1.  Maximum and minimum concentrations for dissolved
metals (mg/L).
Constituent

aluminum
antimony
arsenic
barium
beryllium
boron
cadmium
calcium
chromium
cobalt
copper
iron
gold
lead
lithium
magnesium
manganese
nickel
phosphorus
potassium
selenium
silicon
silver
sodium
strontium
sulfur
thallium
tin
vanadium
zinc
Background
springs
Max
.139
<.017
<.028
.032
•C.OOl
.020
<.003
123
.010
<.004
.008
.116
•e.004
<.03B
•e.006
8.62
.010
<.012
•c.068
<1.32
•e.052
3.01
<.003
6.67
.028
14.7
<-026
<.026
.006
.009
Min
<.027
•e.017
<.028
.028
<.001
<.006
<.003
85.1
<.006
•c.004
.005
<-004
-C.004
•c.038
<.006
5.55
•c.OOl
•c.012
<.068
.400
<.052
.074
<-003
1.77
.252
11.0
<.026
<-026
<.003
.006
Wells in
closed-out
. ,2
ash pond
Max
.057
<.017
.071
.125
.001
1.06
<.003
70.2
.006
.004
.005
.050
.007
•C.038
.118
8.35
.654
<.012
.463
5.79
<.052
10.5
<.003
13.2
2.29
27.6
<.026
<.026
.031
.010
Min
.030
<.017
<.028
.082
<.001
.657
<.003
53.3
<.006
<.004
<.004
.028
.004
<.038
.110
6.95
.160
•C.012
<.060
5.67
<.052
4.62
•c.003
8.55
1.71
23.5
<.026
<.026
<-003
.008
Springs down-
gradient of
closed-out
ash pond
Max
.118
<.017
•c.028
.044
•C.OOl
.175
.003
73.9
.009
•e.004
.011
.157
•c.004
<.038
.028
7.88
.040
•e.012
<.068
2.80
•e.052
1.44
<.003
11.84
.399
20.9
<.026
.066
<.003
.088
Min
.101
<.017
<.028
•c.OOl
•c.OOl
.175
<-003
71.7
.008
•c.004
.008
.142
<.004
•C.038
.024
7.74
.015
<.012
•c.006
2.71
<.052
1.27
<.003
10.1
.398
20.1
•C.026
<.026
•c.003
.011
Springs down-
gradient of
active pond
Max
.117
.020
.029
•e.OOl
<.001
.081
<.003
82.9
.010
<.004
.010
.198
•c.004
<.038
.013
6.12
.005
.013
•e.068
2.14
<.052
1.38
•C.003
5.63
.404
11.5
•c.026
<.026
<.003
.014
Min
.099
•e.017
<.028
<.001
<.001
.061
•c.003
82.2
.010
<.004
•C.004
.154
<.004
<.038
.011
6.02
.001
•c.012
<.068
1.70
•c.052
.992
<.003
2.98
.391
9.30
•c.026
<.026
<.003
.010
Surface water
samples of
active pond
Max
.047
•e.017
•c.028
<.001
<.001
.231
.006
28.7
<.006
.005
.095
.730
<.004
•c.038
.138
8.42
.100
.035
•c.068
4.64
<.052
2.55
<.003
14.08
.246
25.6
.029
.116
.005
.038
Min
•e.027
•c.017
•C.028
<.001
<-001
.125
<.003
26.1
•e.006
<.004
.009
.006
•C.004
<.038
.119
7.20
.043
.012
•c.068
4.20
<.052
.581
•e.003
6.94
.233
25.3
<.026
•c.026
<.003
.007
Lake
wa-
ter
1
value
.086
•c.017
•C.028
<.001
<.001
.011
•e.003
16.9
•c.006
.004
<.004
.008
•c.004
<.038
<.006
7.40
.006
•e.012
<.060
<1.32
<.052
.073
•e.003
12.2
.082
15.1
<.026
.062
•e.003
.008
1 Three analyses except for Sb, Be, Cd, Co, Au, Pb, Se, Ag, S, and Sn which have two analyses.
2. Two analyses.
3. Three analyses except for Sb, Be, Cd, Co, Au, Pb, Se, Ag, S, and Sn which have two analyses.
                              340

-------
Table 2.  Maximum and minimum concentrations  for non-metals
(mg/L).
Constituent



pH
alkalinity
as CaCO?
bicarbonate
total
dissolved
solids
chloride
fluoride
aulfate
Background
springe


Max
8.08
249

304

314

2.85
.15
32.6
Min
7.47
225

274

258

2.31
.10
21.6
Wells in
closed-out
ash pond


Max
7.18
208

254

294

2.10
.37
22.8
Min
6.55
122

149

146

1.43
.16
15.4
Springs
down-
gradient
of closed-
out ash
pond
Max
7.61
216

263

310

4.03
.20
95.6
Min
7.34
122

149

224

3.03
.17
19.8
Springs down-
gradient of
j*
active pond


Max
7.42
221

270

290

2.84
.17
33.7
Min
6.93
210

256

274

2.18
.14
20.9
Surface water
samples of
active pond


Max
7.79
10

12

202

3.99
.20
116
Min
5.92
8

10

128

2.76
.10
80.7
Lake
water


1 value
6.94
39

48

128

2.21
.06
45.4
1 Five analyses except for alkalinity, bicarbonate, and TDS which have four analyses.
2 Five analyses.
3 Nine analyses.
4 Five analyses.                                                >
5 Six analyses except for alkalinity, bicarbonate, and TDS which have three analyses.

     Table 3 contains a partial summary  of  the ash-leaching
results using deionized water.  After 15 minutes of contact
time,  high concentrations of  some elements  were present  in the
leachate.  Leaching continued for most elements after  one week
of contact time.   In general,  the most enriched leachate
resulted from the  fine-grained fly-ash fraction.

Table  3.  Concentrations  (mg/L)  of dry ash  leachate leached
with deionized water at different solid  solution ratios  for
different time increments.
Constituent
(mg/L)


aluminum
arsenic
barium
boron
cadmium
chromium
chloride
cobalt
1:1
Fly ash;
15 min.

700
40
.84
62
.18
14.2
1.9
1.6
1:1
Bottom
ash; 15
min.
239
7.9
.240
14.1
<.010
2.86
8.07
.44
1:10
Fly ash;
15 min.

65
8.0
.262
5.5
•e.010
1.4
.89
.16
1:10
Bottom
ash;15
min.
20.3
.86
.454
1.38
<.010
.22
1.53
.04
1:1
Fly
ash; 1
week
619
19
.08
66
.24
1.2
2.8
1.7
1:1
Bottom
ash; 1
week
287
<.050
.15
21
.12
.3
22.37
.6
1:10
Fly
ash; 1
week
81
2.7
.072
7.82
.023
.139
.10
.206
1:10
Bottom
ash; 1
week
69.2
<.050
.033
2.34
.014
.020
3.50
.088
                                 341

-------
Table 3 continued.
Constituent
(mg/L)
copper
fluoride
iron
lithium
manganese
nickel
selenium
strontium
vanadium
zinc
1:1
Fly ash;
15 min.
29.1
12.61
19.6
30.5
4.06
4.65
.78
20.4
33.6
7.86
1:1
Bottom
ash; 15
min.
6.44
4.61
25.5
7.1
4.11
1.27
<.010
7.69
1.10
2.15
1:10
Fly ash;
15 min.
2.81
2.33
2.53
2.6
.379
.434
.39
2.77
3.34
.761
1:10
Bottom
ash; 15
min.
.718
.54
2.32
1.04
.308
.115
.018
.913
.15
.186
1:1
Fly
ash; 1
week
23.5
8.80
5.69
24.2
4.77
4.9
	
16.4
.72
7.95
1:1
Bottom
ash; 1
week
4.54
6.39
6.48
10.2
	
2.10
	
7.19
	
2.90
1:10
Fly
ash; 1
week
3.02
3.89
1.38
3.0
.540
.597
	
4.40
.219
1.00
1:10
Bottom
ash; 1
week
.602
1.18
3.01
1.11
2.18
.253
	
1.78
	
.324
 Summary

      Dry-ash samples collected from the site indicate that coal
 ash generated at the plant is enriched in most elements tested.
 Dry-ash leach tests show that significant leaching of the ash
 particles occurs after a 15-minute contact time in deionized
 water and that continued leaching occurs.  The system of ash
 disposal at the plant which incorporates large volumes of water
 with a 9 to 15 minute transit time, effectively leaches many
 trace elements from the surface of the ash particles during
 sluicing.
      Dye traces in the vicinity of the closed-out slurry pond
 define a flow system which is approximately 2,000 feet wide and
 4,000 feet long.  The closed-out pond lies within the boundary
 of the flow system.  Anecdotal evidence suggests that spring
 flow in this system was much greater when it was an active
 slurry pond.  Numerous sinkholes reportedly opened while water
 was impounded, allowing water to enter the ground-water system.
 Ash was reportedly observed at some spring outlets when the
 pond was active.
      Major-ion chemistry of the springs connected to the
 closed-out pond is very similar to background spring chemistry
 (Table 4).  Comparison of mean values for background and down-
 gradient springs for trace elements associated with coal ash
 shows that leaching of the ash has already occurred (Table 5).
 On the average, constituent values for springs below the
 closed-out pond are similar to background values.  Only boron
 and iron are more than two times greater than background for
 these springs.
      Wells completed directly in the ash show that boron and
 manganese exceed background by more than one order of
 magnitude.  Barium and fluoride are two to three times
                              342

-------
Table 4.  Mean concentrations of major ions (mg/L)
Constituent




bicarbonate
Bulfate
sodium
potassium
magnesium
calcium
Back-
ground
springs


286
27.8
4.00
<1.32
6.97
99.4
Lake
water



48
45.4
12.2
<1.32
7.40
16.9
Hells in
closed-
out ash
pond

196
18.4
10.8
5.73
7.65
61.8
Springs
down-
gradient of
closed-out
ash pond
264
28.5
4.31
1.92
6.07
82.6
Springs
down-
gradient of
active pond

173
80.14
10.93
2.76
7.81
72.8
Surface
water
samples
of active
pond
11
95.9
9.49
4.38
8.13
27.4
Table 5.  Mean concentrations of trace constituents (mg/L)
Constituent
aluminum
antimony
arsenic
barium
boron
cadmium
chromium
chloride
cobalt
copper
fluoride
iron
lead
lithium
manganese
nickel
selenium
strontium
sulfur
vanadium
zinc
Back-
ground
springs
.079
<.017
<.028
.030
.015
<.003
.008
2.60
<.004
.006
.13
.057
<.038
<.006
.005
<.012
<.052
.276
12.9
<.003
.008
Lake
water
1 value
.086
<.017
<.028
.021
.011
•e.003
<.006
2.21
.004
<.004
.06
.008
<.038
<.006
.006
•e.012
<.052
.082
15.1
•e.003
.008
Wells in
closed-
out ash
pond
.044
<.017
.049
.104
.859
<.003
.006
1.66
.004
.005
.28
.039
•e.038
.114
.407
<.012
<-052
2.00
25.6
.017
.009
Springs
down-
gradient of
closed-out
ash pond
.108
.019
.029
.036
.071
<.003
.010
2.44
<.004
.007
.16
.176
<.038
.012
.003
.013
<-052
.398
10.4
<.003
.012
Springs
down-
gradient of
active pond
.110
<.017
<.028
.044
.175
.003
.009
3.53
•c.004
.010
.19
.150
•c.038
.026
.028
<.012
•e.052
.399
20.5
•C.003
.050
Surface
water
samples
of active
pond
.035
<.017
<.028
.203
.168
.005
<.006
3.43
.005
.031
.15
.204
<.038
.128
.071
.024
<.052
.237
25.5
.003
.024
                              343

-------
background levels.  Vanadium and strontium values are
approximately six times greater than background concentrations.
Other constituents are similar to background measurements.
     Dye traces define a wedge-shaped flow system which
encompasses the active pond and is at least 2,900 feet wide and
6,200 feet long.  Total ground-water spring flow in this system
is approximately two orders of magnitude greater than spring
flow observed in the closed-out pond system.  The source of
much of this flow is apparently the active disposal pond.
Water from the pond enters the ground-water system through
fractures and sinkholes.
     Comparison of water quality from springs and the disposal
pond with background quality shows that leaching has occurred
in this system.  Most trace constituents are similar to
background spring and lake water quality for the elements
listed (Table 5).
     Barium, boron, iron, and manganese are more than 10 times
greater than raw lake water in the surface-water samples taken
from the active pond.  Copper and zinc are eight and three
times greater, respectively.  The remaining constituents are
similar to raw lake water concentrations.  Water quality in the
active pond reflects the major-ion composition of the raw lake
water used for sluicing (Table 4).  The pH, alkalinity,
bicarbonate, total dissolved solids, and calcium are depressed
compared to background springs.
     Boron values in the springs below the active pond are
approximately 10 times greater than background spring
measurements.  Manganese and zinc concentrations are
approximately six times the background values in springs
associated with the active pond.  Iron is three to four times
greater than background spring values.  Major-ion composition
for these springs is intermediate between background and active
pond water.


Conclusions

     1.  The ground-water flow system in which the disposal
sites occur have been defined relatively well.  Even though
deep wells have not been installed at the site, the base of the
flow system can be placed near the contact between the St.
Louis Limestone and Salem and Warsaw Formations.
     2.  Total dissolved solids are relatively low with the
maximum value occurring in the background springs.
     3. Mean values for all constituents analyzed are within
U.S. EPA standards, even though some elements are 10 times
background spring concentrations.


References Cited

Ainsworth, C. C., and Rai, p., 1987, Chemical characterization
     of fossil fuel combustion wastes: Palo Alto, California,
     Electric Power Research Institute, EA-5321.
                              344

-------
Aley, T., and Fletcher, M. W., 1976, The water tracer's
     cookbook: Missouri Speleology, v-16, no. 3.
Cherkauer, D. A., 1980, The effect of fly ash disposal on a
     shallow ground water system: Ground Water, v. 18, no. 6,
     p. 544-550.
Elseewi, A. A., Page, A. L., and Grimm, S. R., 1980, Chemical
     characterization of fly ash aqueous systems: Journal of
     Environmental Quality, v. 9, no. 3, p. 424-428.
Lewis, R. Q., Sr., 1974, Geologic map of the Somerset
     Quadrangle, Pulaski County, Kentucky: U.S. Geological
     Survey Geologic Quadrangle Map GQ-1196, scale 1:24,000.
Quinlan, J. F., 1986,  Practical karst hydrology, with emphasis
     on groundwater monitoring (course manual): National Water
     Well Association, Dublin, Ohio, 898 p.
Rai, D., 1987, Inorganic and organic constituents in fossil
     fuel combustion residues, volume 1: a critical review:
     Palo Alto, California, Electric Power Research Institute,
     EA-5176.
Taylor, A. A., Lewis, L. Q., Sr., and Smith, J. H., 1975,
     Geologic map of the Burnside Quadrangle, south-central
     Kentucky: U.S. Geological Survey Geologic Quadrangle Map
     GQ-1253, scale 1:24,000.
Wu, E. J., and Chen, K. Y., 1987, Chemical form and
     leachability of inorganic trace elements in coal ash: Palo
     Alto, California, Electric Power Research Institute,
     EA-5115.
Biographical Sketch

     Shelley A. Minns obtained the B.S. degree in geology from
West Virginia University in 1977.  She worked for 10 years
throughout the southern Appalachian coal field region,
primarily for coal-related consulting firms.  She obtained the
M.S. degree in environmental studies from Ohio University in
1989 and is currently a Ph.D. candidate in geology,
specializing in hydrogeology, at the University of Kentucky,
Lexington, Kentucky-  Her research interests are coal-ash
disposal and coal hydrology.  She is currently employed with
the Kentucky Geological Survey as a research assistant.
                             345

-------
   Hydrogeology and Ground-Water Monitoring of an Ash-Disposal

       Site at  a Coal-Fired  Power  Plant  in  a  Karst  System


     Shelley A.  Minns1, Arsin M. Sahba1,  Lyle V.  A. Sendlein1,
              James C. Currens2  and James S. Dinger2

              Department of Geological Sciences and
                   2Kentucky Geological  Survey
           University of Kentucky,  Lexington,  Kentucky


Question 1:  Did you define two flow systems or two drainage
basins?

     The two disposal ponds are located in different surface
water drainage basins.  Dye traces were conducted on all
available sinkholes considered upgradient of each pond.  The dye
traces defined the flow systems and as it turned out, they also
reflected the surface water basins.


Question 2:  What was the rationale for selecting  5 times
background concentration as the criterion for determining whether
the levels were elevated above background?

     Concentrations of metals for water associated with the two
disposal sites range from nearly identical to background spring
concentrations to about two orders of magnitude greater than
background concentrations.  The ranges of five times greater than
background and ten times greater than background were selected to
use during the presentation of the paper at the conference as a
way to get a feel for how metals were distributed when compared
to background spring concentrations.  These numbers to do not
imply any significance relative to background concentrations and
are not used in the paper.  The paper contains detailed data of
constituents in the water associated with the two disposal sites.


Question 3:  What leaching tests were found to be most
representative of actual processes presumably operative within
the  ponds?  Was there a significant difference in the trace
element nature of the coals burned at the site between the old
and the active ponds?  Were there other  significant waste streams
contributing to the ponds?

     Leaching tests were not conducted as part of our research
effort.  Battelle-Pacific Northwest Laboratory conducted the
leaching tests as part of a separate investigation.  We have not
received the complete results of these tests.  We do know that
the solid/solution ratio, which is generally representative for
the sluicing system used at the plant, is ten parts ash to one
part lake water.
                                346

-------
     Trace element analyses of coal burned at the plant when the
closed-out pond was an active facility are not reported.

     No other waste streams impact the disposal ponds.  The
contributing watershed is mainly undisturbed forest.  The coal
stockpile area is located downgradient of both ponds and does not
affect the ash disposal site.
                                347

-------
348

-------
     DEVELOPMENT  OF A MONITORING PROGRAM AT A SUPERFUND SITE

          IN A KARST TERRANE NEAR BLOOMINGTON, INDIANA
         Michael R. McCann, Westinghouse Electric Corp.
                Noel C.  Krothe,  Indiana University
                      Bloomington, Indiana
Abstract
     Lemon Lane, a  former municipal dump  from 1933 to  1964,  is
located  entirely  within  two  sinkholes   on   a   ridgetop  over
Mississippian limestones.  Materials containing PCBs were discarded
there from 1957 to  1964.

     Twenty-two monitoring wells have been installed in stages into
two separate ground-water zones.   Results from three tracer tests
conducted from 1987 through 1990, and  interpretations  of limited
potentiometric  data,  indicate  that the site  is  drained by  a
convergent conduit system that, during low to moderate flow, drains
through a  main conduit  which exits at  a  series of  springs.  A
quantitative tracer test showed that 97%  of the tracer recovered
flowed to  one  spring.  During high  flows,  when  the  conduit  is
flooded,  some water is temporarily stored in the epikarst and the
fractured bedrock adjacent to the conduit. Small amounts of water
in storage  may spillover into  adjacent  spring drainage basins,
suggesting the site is near the ground-water  drainage  divides of
several subterranean basins.

     Fluorescein injected as  a tracer in an  on-site well, in 1989,
was still detectable at the main  spring  outlets 19 months later.
Monitoring at selected on-site wells has  shown tracers at levels
near  the  detection  limit in  water  (0.005 ppb)   of  a  scanning
spectrofluorophotometer.

     The springs, indicated by tracing to be directly connected to
the  site,   will  provide  the most  time-sensitive and  relevant
monitoring   locations   for   potential   off-site   migration   of
contaminants. Existing monitoring wells should effectively monitor
for  contaminants   in  temporary   storage.   As   a  result,   the
installation of  additional  monitoring wells  is  not recommended
because the probability  of   any  of them  intersecting  conduits
draining the site is minimal.
                                '349

-------
Description of Lemon Lane Landfill

     Lemon Lane  Landfill  is located  on  the west-central edge of
Bloomington in Monroe County, Indiana.   Bloomington is  located in
the Mitchell Plain physiographic unit.  The Mitchell Plain is a low
plateau underlain with carbonates of Mississippian  age  on which a
characteristic  karst  terrane  has  developed.    The  Lemon  Lane
landfill, comprising about 9 acres, was located entirely within two
sinkholes  on a  ridgetop  at  or  near several  surface  water or
topographic divides.  However, all of the natural  surface drainage
at the site  is  into several  sinkholes.   The 1908 topographic map
for the area (Figure 1) , shows the two sinkholes that were used for
the disposal  of  solid waste,  which are  actually  part of a  large
compound sink extending  to the  northwest  and southwest of  Lemon
Lane.

     Lemon Lane was operated as  a municipal dump from 1933 to 1964.
A  nearby plant  manufactured  capacitors  containing PCBs  as an
insulating fluid beginning in  1957.  Materials containing PCBs were
discarded at  Lemon  Lane from 1957 to  1964 by local  contract  waste
haulers.
Geology of Lemon Lane Landfill

     The  site  is  underlain  by  Mississippian  age  limestones,
dolostones,  and shales  of  the  St.  Louis Formation.   The  Salem
Limestone, which has strata similar to the St.  Louis in  its  upper
part underlies  the St.  Louis  Formation.   The  lower part of the
Salem  is  a thick,  cross-bedded  calcarenite  which is famous as a
building  stone  known as the "Indiana  Limestone" (Shaver et al.,
1986) .  The contact between the St. Louis and the Salem is not well
defined lithologically in south-central Indiana.

     The  strata generally dip  westward from their outcrop on the
flank  of  the Cincinnati Arch  into the Illinois  Basin.   The dip
averages about 30 feet to the mile in the  Bloomington area (Gates,
1962) .


Monitoring Well Installation

     Twenty-two monitoring wells were installed in  two  phases at
Lemon  Lane.  Seventeen wells were installed,  from 1982 to 1983, in
the  first  phase and five additional wells were  installed in 1987
(Figure  2) .   Seventeen  of the  wells were  installed to  monitor
ground water in shallow bedrock and  five of the wells  monitor a
deeper zone.   Figure 3  shows  the  stratigraphy encountered.  The
shallow zone occurs between  elevations 798 ft. and 820 ft. amsl and
the deeper zone occurs between  elevations  760 ft. and 770 ft.  amsl.
Monitoring wells MW-4D, MW-5, MW-8D, MW-12, and MW-14 are installed
in the deeper zone and the rest are installed in the shallow zone.
                                 350

-------
uo
Ul
                               R2W  R1W
                                                    0      2000 FEET
                                                     i    i     i
Figure  1.
                                          Location of Lemon Lane  Landfill relative to
                                          sinkholes prior to use  as  a  landfill.
                                          (Bloomington, Indiana,  Quadrangle 15 minute
                                          series, 1908)

-------
                                            N
                          MW-12
                            O
                      MW-1D
                        °0
                       MW-1S
MW-8S
    MW-8D
                                 MW-B4
                                   O
                               (not to acale)
                              (well la off mop)

MW-B3
O
/\
\fc
\°»
\\
X^
Vfe
r
I
i


i
1
MW-9
Oi
^rf 1
^ MW-2
X,g
~^JMW-13
MW-3
or 	
MW-14O |
MW-4S
O I
MW-41 OMW-10
O O
MW-40 1
                 SCALE 1 inch - 200  FEET
                                                                  MW-B2
                                                       MW-11    O
                                                             O O
                                                                MW-5
                E
_t
         Figure 2  - Monitoring Well and Cross-Section  Location
                                    352

-------
          ELEVATION
          ABOVE M8L
oo
Ul
CO
                                                               Figure 3  -  Geologic  Cross-Section
                                                                                                                                                   BLACK SLACEOUS CORAL
                                                                                                                                                   LS
                                                                                                                                                   DOLOSTONE
                                                                                                                                                   LS
                                                                                                                                                    IpLOSTONE

                                                                                                                                                       TONE
                                                                                                                                                   SILT STONE WITH PYRITE

                                                                                                                                                   POROUS ZONE
                                                                                                                                                   BLACK SHALE IN UW-40
                                                                                                                                                L.J

-------
The two zones  appear to be separated by  a shale unit that  is  at
elevation 782 ft. to the east and 778 ft.  to  the west (see Cross-
section, Figure 3).

     Packer  tests and  slug tests  conducted  during  installation
indicated the wells  in the  deeper zone encountered fractures much
less  transmissive than  those  encountered by  the wells  in the
shallow zone.


Water Level and Water Quality Measurements

     Potentiometric  surface maps for water levels taken  on 2/2/88
and  6/14/88  are shown  for  the  upper and  lower zones in  Figures
4,5,6,and 7.  The interpretation of ground water flow directions  in
the lower zone appears to be relatively simple.   During the higher
flow period (2/2/88)  recharge is from the west and flow directions
are to the east.  During the lower  flow period  (6/14/88) recharge
is from the west and east and flows  are  generally to the  north and
northwest.  Interpretations of  flow directions  in the upper zone,
however, are  problematic.   During the higher-flow  period,  ground
water appears to be  mounded in  the vicinity of  wells  MW-7,  MW-B3,
MW-4I,  MW-8S,  and  MW-11.   Water  levels  are   depressed  in the
vicinity of MW-9, giving the appearance of  a sink.  During the low-
flow period, mounding is apparent near MW-B3, MW-3,  and  MW-11.  In
neither  flow  regime  is  there  an  unambiguous  flow  direction
indicating the probable direction of ground-water movement.

     Table 1 shows the analytical results of PCB sampling at the  17
monitoring wells installed in  1982 and some  springs in the vicinity
of Lemon Lane.   Wells MW-10,  MW-11, MW-12, MW-13,  and MW-14  were
installed in  1987 and have  not  yet been sampled.   The results are
difficult to  interpret.  For example, although  PCBs were found  in
wells  MW-8S  and  MW-8D,  the  potentiometric  maps  for  both  flow
regimes  seem to  indicate  these wells are upgradient.  Illinois
Central and  Quarry  springs, located about  2000  feet  southeast  of
Lemon  Lane,  have  concentrations of  PCBs  one  to  two orders  of
magnitude higher than the on-site wells.

     The original monitoring plan for Lemon Lane required  on-site
monitoring wells to  be  selected on  the basis  of ground-water flow
directions and the results of initial PCB sampling.  Locations for
off-site wells were  to be selected  on the  basis of preferred flow
directions  as  interpreted  from on-site   wells.    The  original
monitoring  plan was  based  on the assumption  that an  aquifer was
present that behaved as  a porous-media-equivalent and that  ground-
water  flow  direction could be  predicted  by  idealized  flow  nets
constructed  from  water  level  measurements.   Contaminant movement
could  thus  be predicted  from  aquifer  properties  deduced  from
pumping tests which were analyzed by standard  equations of  ground-
water flow.  In an attempt to resolve the ambiguities, tracer tests
were conducted.
                                  354

-------
                         I	  Site Boundary
                                                      MW-5
      Figure 4  -  Lower Zone Potentiometric Map  6/14/88
                             en	° Site Boundary
                                                      MW-5
Figure 5 -  Lower Zone Potentiometric Map  2/Z/88
                                 355

-------
                   MW-1S
  'Big
                                                  MW-4S
                                                    MW-11
   Figure 6  -  Upper Zone Potentiometric Map  6/14/88
                MW-1S
MW-8S
                                                 N
                                                   MW-11
   Figure  7  - Upper Zone  Potentiometric Map  2/2/88
                             356

-------
 Location

Springs/Surface Water

Sargent Pond #1
Sargent Pond #2
Stout Creek
Robertson Spring
Snoddy Spring
Stoney Spring-East
Stoney Spring-West
P.H. Road Spring
Detmer Spring
Packing Plant Spring
III. Central Spring
Quarry Spring
Slaughter House Spring
Hinkle Wet Weather Rise
                                     Table  1
          PCB ANALYTICAL DATA FOR GROUND WATER AND SURFACE WATER
                                LEMON LANE LANDFILL
                               BLOOMINGTON, INDIANA

        6/25-29/81     7/1-2/81    8/17/81    10/7-8/81   7/28-29/82   10/19/82
   ND
   ND
0.9*

 ND
                       ND
                       ND
                     5.7***
                        ND
                         ND
                         ND
                         ND
                         ND
                         ND
                         ND
                        6.8*
ND
ND
                                                                  12/9,15/82
                                             ND
                      ND
1.85**
  ND
  ND
  ND
  ND(ND)
12.2**
2.7***

  ND
ND
ND
ND
ND
                               10.**
                               5.5***
   Location

 Monitoring Wells

   MW-B1
   MW-B2
   MW-B3
   MW-B4
   MW-1S
   MW-1D
   MW-2
   MW-3
   MW-4S
   MW-4I
   MW-4D
   MW-5
   MW-6
   MW-7
   MW-8S
   MW-8D
   MW-9
ND
ND
ND
ND
              12/17/82-1/5/83    2/2/83-2/14/83    6/8/83
                                       ND
                                       ND
                                       ND
                                       ND
                   0.4
                   0.3
                  <0.01
                   0.5
                   0.8

                   2.4
                   0.2
                  <0.01
                   0.6
                   0.6
                   1.4
                   1.1
                        0.11
                     0.13(.05 dup)
                        0.02*
                        0.04*
                        0.07
                        1.5*
                        0.09
                        0.02*
                     0.02(.05 dup)
                        0.10
                        0.11*
                        0.07
                        0.40*
    NOTES:  All analyses for mixed Aroclors unless noted by * or **
            All analyses in ug/l or parts per billion (ppb)
            * Aroclor 1016/1242   _
            ** Aroclor 1248          Arcolor 1
-------
1987 Low Flow Tracer Test

     The first tracer test at Lemon Lane was  conducted under dry,
fall  conditions.    Injection occurred  on  November  10, 1987  and
sampling continued until November 27, 1987.  Monitoring wells MW-7,
MW-1D, and MW-10 were selected as injection locations.  Springs to
be sampled were:  Crestmont A and B,  Detmer A, Illinois  Central,
Quarry, Packinghouse Road (PH Road),  Pumping Station, Snoddy A and
B,  Stony East,  Stony  West  A  and B,  and Urban  (Figure  8) .
Monitoring wells to be sampled were: MW-5, MW-11, MW-8S, and MW-8D.
Lithium bromide (Br-) was chosen as the tracing agent and two sets
of background samples were  taken prior to injection.   Background
concentrations of Br- were in the range of 0.01 to 0.39 ppm with a
detection limit of  0.01 ppm.

     Sampling frequency was  every 6  hours for the first  72  hours
and then every 24 hours  for  the  remainder  of  the  sampling period.
Only  Illinois  Central  Spring  and Quarry  Springs had  definitive
breakthrough curves for Br-  that had a classic rising limb, a peak,
and a receding tail.  The peak at  Illinois Central was 5.11 ppm and
the peak at Quarry  was 4.83 ppm.   The time  to  peak  from  first
detection at both  springs was  about 250 hours.   Illinois  Central
rises  and flows along the  surface for 600 feet to where  it  sinks
about  200 feet above where Quarry Springs resurges.   Based on this
and the tracer detection times,  Quarry  Springs was presumed  to be
second resurgence of the same water.  This was later confirmed by
a  dye trace run from where the  Illinois  Central waters  sink to
their  outlet  at Quarry Springs.   Crestmont  A had one  Br- sample
within 15 hours of injection that was 0.50 ppm and Urban spring had
a  similar one time  occurrence of Br- at 0.66  ppm  within 1.5  hours
after  injection.   Well  MW-8D had 1.49  ppm Br- detected in  it on
November 25, 1987.   It is possible that these Br- detections may be
the tracer injected at Lemon Lane, but it is  not considered likely
because they are isolated occurrences.


1989 High Flow Tracer Test

     A tracer  test during high  flow periods was scheduled  to be
performed  in  the  spring of 1988.  However,  drought  conditions
prevailed and the test  was rescheduled  for the spring of  1989.  A
different fluorescent dye was injected  in  each well  along with an
aliquot  of Br-  in  each  well.    Color  Index (C.I.)  Fluorescent
Brightner  28  (FB28)  was  injected  in  MW-7.   MW-10  received
Fluorescein (C.I. Acid  Yellow  73; AY73).  Instead of MW-1D,  well
MW-1S  was injected  with Direct Yellow 96   (DY96)   (Figure  2).
Tracers  were  distributed throughout the  bedrock column  in  these
wells  because  injection occurred during  rising water  levels  and
because there was interconnection between nested wells.  The trace
was to be  qualitative with detection to be  accomplished  visually
for the fluorescent dyes.  FB28 and DY96 are adsorbed onto sterile
cotton fabric which is suspended  in the resurgence.   The cotton is
washed and then observed under a hand-held ultraviolet lamp.   FB28
will  fluoresce blue, and  DY96  will  fluoresce yellow.    If  both
                                358

-------
         V
RESIDENCE WITH
WELL IN USE
SPRING
BACKUP STATION
STREAM
DYE RECOVERED
LESS THAN 0.01%
2000 Feet
                                                4WEIMER ROAD
                     Figure 8  - 1987  Low Flow Tracer Test

-------
tracers are  present,  the cotton  will fluoresce a  characteristic
blue-white.   AY73 is  adsorbed onto  activated  charcoal which  is
suspended in screen packets in the resurgences.   The dye is eluted
from the charcoal with a hydroxide solution and will  appear  as a
yellow-green layer in  the elutant above the  charcoal.   Background
samples taken before injection indicated minor  amounts of  FB28  at
some stations.   This  is expected  since  FB28  is  a component  of
laundry  detergents.  However,  no  background samples  were taken
during storm flows.

     Injection of tracers took place  on May  26,  1989 after a  1.35
inch rainfall.  Several  storms of 1"  or greater rainfall occurred
during sampling, including a 2.36"  rain on July 12,  1989.   Figure
9  shows  the location  of springs and streams  that were  sampled
during the test until  July  22,  1989.   Monitoring wells  that  were
sampled were MW-8S, MW-6, MW-5, and MW-11.   Sampling began within
6 hours after  injection, and Illinois Central  and Quarry  springs
had distinct positives for FB28 on the first samples.  Breakthrough
curves for Br- were  also detected at Illinois  Central  and Quarry
springs with peak concentrations at Illinois  Central  of 8.7  ppm and
10.5 ppm  for Quarry.    Distinct  visual  positives  for AY73  were
detected at Illinois Central and Quarry springs  beginning 55 hours
after  injection  and   continued throughout  the 57  day sampling
period.  The tracer DY96 was not recovered  visually at any  station
during the sampling period.   Apparent  detections  of AY73  were
recorded  for all  stations  on  samples collected  following  rain
storms.   However, the coloration  was not  the distinctive  AY73
yellow-green and it was  suspected other organic constituents  were
being  adsorbed   by   the   charcoal  and  making    the  definite
identification of AY73 impossible by visual  means.  There was no
visual detection  of  dye at  any of the monitoring wells, although
Br-  was  detected at   MW-8D.     AY73  was  detected  visually at
residential  well  #83  (Figure  9) .  There were  Br-  detections at
Clear Creek of 2.96 ppm  at 839 hrs  (35 days)  and 1.45  ppm  at  1147
hrs  (47  days)   after   injection,  but  since  they  are isolated
occurrences they are not considered conclusive.

     Since the visual  detection of the dyes at many stations was
inconclusive,  random   samples  were   submitted to  Quinlan  and
Associates  for  analysis on  a scanning  spectrofluorophotometer
(SSFP).   These analyses indicated  AY73 and  DY96 were  present  in
some samples, even though they were not recognizable visually.  A
SSFP was  purchased  in February 1990,  and  most  of the  grab water
samples  taken for Br- analysis  were  analyzed  for  fluorescence.
Table  2  shows the  results  of  that analysis.    These  results
confirmed the suspicion that Lemon Lane was at or near the  divides
of  several subterranean  drainage basins,  and,  that upper level
conduits  exist  that  would divert ground  water, during  high  flow
storm events, into those adjacent spring drainage  basins.   Actual
dye concentrations were  not calculated for these samples,  but the
relative  intensity  of  fluorescence  as  measured  by  the  SSFP
indicated  that  Illinois Central  and  Quarry  Springs received the
majority  of  the  tracers and all  the other stations  received  only
minor amounts.  It was obvious that the proportional distribution
                                 360

-------
                                            •SSSVVSSS*
                                                          *ILLINOl'sYTHaADAMS
                                                          ^CENTRAL :
RESIDENCE WITH
WELL IN USE
SPRING
BACKUP STATION
STREAM
DYE RECOVERED
LESS THAN 0.01%
2000 Feet
                                                ^WEIMER ROAD  '
                    Figure  9 - 1989 High  Flow Tracer Test

-------
Table  2 -  Results  of  Spectrofluorophotometer Analysis
                 1989  High Flow  Tracer Test

Station
Detmer A
Detmer B
Defeat E.
Defeat W.
K1rby Rd.
Stony E.
Stony M.
Sinking Cr.
WN-1
WS-2
ICG-1
ICG-Z
ICG-3
ICG- 6
Pump St.
Urban
Bypass 37
S. House
PHRoad
Hinkle
Snoddy ASB
17th St.
IL Centra]
8 Allen
Weiner Rd.
Stouts U.
Stouts E.
HW-8S
MU-6
MM- 11
MW-5

5/26
ND
FB28
DY96
FB28
ND
ND
—
.FB28/
AY73
iV7T
HI / J
FB28/
DY96
ND
ND
FB28
FB28
DY96
ND
FB28/
AY73
FB28
FB28
FB28
FB28
FB28
FB28
DY96
FB28
ND
NO
FB28
SS
S§
5/27
ND
FB28
DY96
DY96
DY96
DY96
DY96
iY7T
ft I / 0
ND
ND

FB28
FB28
DY96
ND
DY96
ND
ND
FB28
ND
DY96
ND
ND
FB28/
AY73
ND
ND
ND
ND
o o
So
o
5/27
DY96
DY96
—
ND
ND
-
ND
FB28/
AY73
ND
ND
ND
FB28/
AY73
AY73
ND
ND
DY96
ND
DY96
ND
ND
ND
ND
DY96
ND
FB28
ND
ND
FB28
ND
ND
0 o
o o
0 tNJ
o o
5/28
DY96
ND
DY96
DY96
ND
FB28
ND
AY73
AY73
ND
DY96

ND
ND
ND
ND
ND
ND
DY96
ND
ND
DY96/
AY73
ND
ND
ND
ND
ND
ND
o o
o o
<£> CO
0 0
5/28
DY96
ND
DY96
ND
FB28
ND
DY96
AY73
un
nu
ND

ND
NO
ND
NO
AY73
ND
ND
FB28
--
AY73
ND
ND
ND
NO
FB28
ND
0 O
0 O
5/28
ND
ND
DY96
DY96
ND
DY96
AY73
ND
NO
--

ND
ND
ND
ND
AY73
AY73
ND
DY96
ND
DY96/
AY73
ND
ND
ND
ND
ND
ND
o o
§§
i— i CO
5/28
DY96
DY96
ND
ND
DY96
ND
DY96
AY73
wn
nu
ND
AY73

ND
ND
ND
ND
AY73
AY73
ND
FB28
DY96
ND
ND
NO
NO
ND
ND

5/29
ND
NO
DY96
DY96
DY96
ND
ND -
DY96/
AY73
un
nu
ND
AY73

ND
--
DY96
NO
AY73
AY 73
NO
ND
DY96/
AY73
ND
ND
ND
ND
ND
FB28
ND

5/30
ND
NO
DY96
DY96/
AY73
DY96
ND
ND
FB28/
AY73
un
nu
ND
AY73

ND
ND
ND
ND
AY73
AY73
DY96
ND
DY96/
AY73
DY96/
AY73
ND
ND
NO
ND
NO
ND

5/31
ND
OY96
DY96
ND
DY96
AY73
AY73
DY96/
AY73
ND
ND
DY96/
AY73

ND
ND
ND
AY73
AY73

DY96
AY73
ND
NO
ND
ND
AY73
ND
ND

6/1
ND
DY96
OY96
ND
DY96
AY73
ND
AY73
MD
nu
AY73

—
ND
ND
ND
AY73
AY73

DY96
AY73
AY73
ND
ND
ND
ND
NO
ND

6/2 6/9 6/16
OY96 DY96 ND
ND DY96 DY96
DY96 - DY96
DY96 ND
ND DY96 ND
DY96 ND ND
AY73 DY96 ND
AY73 ND DY96/
AY73
ND
AY73 DY96/ AY73
AY73

AY73 ND ND
ND ND ND
DY96/ DY96/ AY73
AY73 AY73
ND ND ND
AY73
AY73
DY96 —
AY73 ND ND
DY96/ —
AY73
ND ND OY96
ND DY96 ND
ND NO OY96
ND ND ND
ND FB28/ ND
AY73
NO FB28 ND
ND ND ND

6/23
ND
--
ND
OY96
ND
DY96
DY96
—
AY73

ND
ND
AY73
ND

—
ND
AY73
ND
ND
ND
ND
ND
ND
ND

6/30
DY96
-
OY96
ND
ND
DY96
--
—
AY73

AY73
ND
DY96/
AY73
ND

-.
ND
AY73
DY96
AY73
ND
ND
ND
ND
ND

7/7
DY96
DY96
DY96
NO
DY96
--
--
DY96
AY73
--

AY73
DY96
DY96/
AY73
DY96

—
DY96
AY73
NO
DY96
ND
ND
ND
ND

7/13
DY96
DY96
DY96
ND
ND
--
--
AY73
ND
—

ND
ND
DY96
ND

ND
DY96
AY73
ND
ND
ND
ND
ND
ND

7/22
NO
ND
ND
ND
DY96
--
-
AY73
—
--

ND
ND
ND
ND

—
ND
AY73
ND
ND
ND
ND
ND
ND
 ND - Not Detected
 	 Not sampled or not analysed
 FB28 Optical brightners detected
 AY73 Fluorescein detected
 OY96 Direct yellow 96 detected
                                      362

-------
of groundwater would have to be ascertained.  A quantitative high-
flow tracer test was scheduled for the spring of 1990.


1990 High Flow Tracer Test

     Background sampling was conducted at all monitoring stations
to ascertain which fluorescent tracer would have the least amount
of  background   interference   and  be  most   suitable  for  the
quantitative trace.   The  sampling  indicated that  Rhodamine WT,
(C.I. Acid  Red 388;  RWT)  would  be  appropriate.    Some stations
however, did have background concentrations  of RWT in grab samples
of water analysed on the SSFP.  The detection limit of RWT on the
SSFP was 0.01 ppb (Table 3).

     RWT was detected at  7th & Adams on 4/19/90  in the elutant from
a charcoal packet.  In addition AY73 was detected in Quarry Springs
at  0.51  ppb.    Eighty-two residential wells in the  vicinity of
Lemon Lane  were monitored during the tracer test  by  teams from
federal, state, and  local agencies (Figure 10).  The one residence,
#83, that had AY73 detected in its well during the 1989 test, was
subsequently  connected  to city  water  and this   well was  not
monitored during  1990.   Residential wells  #82 and  #212 had AY73
detected in them during background monitoring for the 1990 test.

     Spring and  stream stations  were  gaged, where  possible, and
Table  4  lists the  stations  and  the  flow  monitoring  method.
Illinois Central spring,  per se,  was not monitored for this test,
rather a gaging station was installed below Quarry Springs.
              Table 3 - Background Determination of
                   Rhodamine  WT  in  Water  Samples
                    in parts  per billion  (ppb)
STATION
Quarry Spring
Urban Spring
Bypass 37
Stony East
Clear Cr @ 1st St.
ICG-2 Spring
Cascade Br.
Weimer Rd.
7th & Adams
3/20/90

0.075
0.095
0.075
0.058
0.31
0.15
0.095
0.058
  ND
4/2/90

  ND
  ND
  ND
  ND
0.125
  ND
  ND
  ND
  ND
4/19/90   5/7/90
  ND
  ND
  ND
  ND
  ND
  ND
  ND
  ND
  ND
 ND
 ND
 ND
 ND
0.11
 ND
 ND
 ND
 ND
                                 363

-------
                                                 STOUTS^STOUTS
                                                 WEST rl FAST
                                                                         *
-------
  TABLE 4 - MONITORING  STATIONS FOR 1990 TRACER TEST
                Station is downstream
                   of a Spring	
                13. Bypass 37  (re)
                26. Defeat Cr. E.  (re)
                27. Defeat Cr. W.  (re)
                28. Kirby Road  (re)
                29. Sinking Creek  (re)
Station is near
   a Spring
l. Quarry Spring (w)
3. ICG-1 (re)
4. ICG-2 (w)
5. ICG-3 (w)
6. ICG-6 (w)
7- Fell Iron (nm)
9. Crestmont (w)
10.Pumping Stn.  (nm)
11.17th Street  (nm)
12.Urban (w)
14.S.House  (w)
15.PH Culvert  (w)
16.PH Road  (w)
17.Snoddy A  (w)
IS.Snoddy B  (w)
19.Hinkle (w)
20.Abrams (w)
21.Walcott A (w)
22.Walcott B (w)
23.Robertson (re)
24.Detmer A  (nm)
25.Detmer B  (w)
30.Stony West  (re)
31.Stony East  (re)
32.WN-1 (w)
33.WS-2 (w)
(w)  -  90°  v - notch weir
(re)  - rating curve with staff gage
(nm)  - not  measured
Station is a backup
point downstream
of several primary
of several Springs
                 2. 7th & Adams  (re)      Cascade Br. (re)
                 8. Clear Cr. @  1st  (re)  Stouts Cr. W.  (re)
Stouts Cr. E.  (re)
111 Cen.@ Allen (re)
Weimer Road  (nm)
Clear Cr. @ 7th (nm)
                           365

-------
     Injection of tracers began on May 12,  1990 after a 1.45"  rain
fell.  RWT was injected as a 20% solution and the  injected amounts
are corrected for this dilution.  Well MW-1S  received 9,545 grams
of RWT,  MW-10 received  9,545g, and MW-7 received 8,183g for a total
of  27,273  grams  of RWT  injected.   Large  amounts  of  RWT  were
injected because RWT is known to suffer large adsorption losses and
high suspended clay and organic matter loads  were observed during
storm flows at Quarry Springs.

     RWT was  first detected  at Quarry Springs  3.75 hours after
injection and  reached  a peak concentration  of at least  4300 ppb
(distinctly visible) at  5.25  hours.   By 10 hours after  injection
the concentration had  receded to 40  ppb and remained at  or below
that level through the last sample taken on August  24,  1990.
Five other stations had repeated detections of RWT.  They were  ICG-
1, Slaughterhouse,  PH  Road,  PH Culvert,  and Clear Creek  at First
St.  However, Clear Creek did not have a breakthrough curve during
the first 72 hours of sampling.   Only two samples  from Clear Creek
were above the background level of 0.31 ppb.   They were  a 2.6 ppb
detection at 584 hours after injection and a  0.7 ppb detection at
593 hours.  The Clear Creek watershed is urbanized and the specific
spring heads could not  be  located or monitored.  A determination of
Clear Creek's connection with Lemon Lane above its confluence  with
the Illinois Central-Quarry Spring Branch is  not conclusive.

     Ten  stations  had sporadic and  low  (near background level)
detections of RWT  in the  grab water  samples.  They  are  listed in
Table 5,  along with the  date,  time,  concentration  detected,  and
amount detected in background samples.  In addition  the  following
stations on  the following dates had  RWT detected in the  back-up
activated charcoal samples:  Fell Iron Spring on 6/8/90 and 6/15/90,
Snoddy A on  7/6/90,  Snoddy  B on 7/6/90,  Detmer A on 7/13/90,  and
Stony East  on 7/14/90.   Three residential  wells,  #82,   #86,  and
#212, had  RWT detected in  them.   Residential  well #86 had  been
abandoned when the owner  connected to city water in  1986,  but was
elected to be sampled anyway by state agency representatives.   Well
#212  also had  AY73 detected  in it.   Wells #82  and  #212   were
analyzed for PCBs and none were found to a detection limit of 0.1
ppb.

     Calculations for mass balance were performed to determine the
dye  recovery  at  the  six  stations  that   had  repeatable  RWT
detections.  Those results are  listed in Table 6.   Clear Creek at
First St. was included even though the connection is problematic.
Only 18% of  the  total  amount of RWT  injected was recovered.  The
amount  not  recovered  is attributed  to  losses by  adsorption,
underestimation   of   peak   flows,    underestimation   of   peak
concentrations due to quenching effects of high concentrations on
the SSFP, and tracer in ground-water  storage.   Sampling  frequency
was  sufficient  to catch  all  peaks of dye  detections and is not
considered  to be  a significant contributor to the lack of dye
recovery.  Slightly less than 97% of  the amount recovered occurred
at Quarry Springs.  ICG-1 appears to  be related to Quarry, perhaps
                                 366

-------
TABLE 5 - RHODAMINE WT APPEARANCES AS OF  6/29/90
Station
Kirby Rd.
Stony West
WN-1
Stouts W.

ICG-6

Clear Cr.
@ 7th



Cascade Br.
WS-2
Bypass 37
Defeat E.
Date
5/19/90
5/12/90
5/12/90
5/12/90

5/18/90
5/29/90
5/13/90


5/14/90
5/15/90
5/25/90
5/15/90
5/28/90
6/15/90
Time
0148
2230
2250
2033
2233
1130
1130
0520
0720
0930
0525
0520
1215
0641
0920
0930
Concentration
360 ppt
260 ppt
290 ppt
115 ppt
280 ppt
58 ppt
58 ppt
85 ppt
75 ppt
120 ppt
42 ppt
44 ppt
275 ppt
54 ppt
42 ppt
65 ppt
Background
ND
ND
85 ppt
ND

ND

ND




95 ppt
ND
75 ppt
ND
                        367

-------
as  a subsidiary  overflow  route  during  storm  events  or  as a
distributary resurgence.   Based on their morphology,  elevations,
and  comparison  of dye breakthrough curves,  it is concluded  that
Slaughterhouse, PH Road,  and PH Culvert are distributary outlets of
a main conduit  draining in that direction.  The conclusion then, is
that about 98%  of the ground water from Lemon Lane resurges in the
vicinity of Quarry  Springs, about 1%  or  less flows to the three
distributaries to the northwest,  and another  1% or less may flow to
other headwaters of Clear Creek  above  the Illinois Central-Quarry
Spring Branch confluence.   Some  of ground water from the vicinity
of  Lemon  Lane  may  be diverted  to numerous  other spring basins
during storm flows, but the amount of  ground water is  probably no
more than 0.01% of  the total  from Lemon Lane based on the tracer
recovery ratios (Figure 10).


Hydrogeology of Lemon Lane  Landfill

     An examination of the Cross-section in Figure  3 shows  that
shallow strata on the east dip to the  south-west and strata on the
west slope trend to the south-east.  This is interpreted to be an
effect of sinkhole subsidence  and collapse, as well as  development
of the conduit along a structural sag  and  trough.   Strata adjacent
to the conduit  that  collapsed also subsided in the  direction of the
collapse.  A zone of interbedded limestone,  dolomites, and shales
at approximately the 798'-805' elevation is  where the  major water
producing fractures occur.   The different  strain properties of the
different interbedded  lithologies  caused  bedding plane openings,
vertical jointing,  and general  fracturing  of the  rock units in
response to the subsidence stresses.    This  produced  the zone of
enhanced permeability noted on the cross-section.  A higher density
of  fracturing  also occurs  within  the first  10'  of  the bedrock.
This is  in accordance with Williams (1983), Ford and William (1989,
p.  206) ,  and  others,  who  report that  the   uppermost layers of
bedrock constitute a suspended aquifer termed the subcutaneous  zone
 (Williams,  1983) or the epikarst (Mangin 1975).
                 TABLE 6 - RHODAMINE WT RECOVERY
Station

Quarry
ICG-1
Clear Cr. @ 1st
Slaughterhouse
PH Road
Ph Culvert
Mass (grams)
Recovered

4744.243
  64.135
  60.192
  22.032
   5.589
   2.562
4898.753
% of
Injected

 17.40%
  0.24%
  0.22%
  0.08%
  0.02%
  0.01%
 17.97%
% of
Recovered

  96.85%
   1.31%
   1.23%
   0.45%
   0.11%
   0.05%
     100%
                                  368

-------
            MW-1S
                                                 MW-11
Figure  11  - Hypothetical  Base Flow  Map
             MW-1S
                                                MW-10
                                 *&'
vMW/7^--!ite.!!.unlary
                                -838
\
£
                                                  MW-11
 Figure 12  -  Hypothetical High Flow Map
                         369

-------
ground-water flow and contaminant, movement based on that assumption
erroneous.  The original monitoring plan, which was predicated on
the basis of  intercepting a  contaminant  plume,  with wells, is
incapable of yielding reliable or accurate monitoring - except by
improbably good luck.

     (2)  The  aquifer has  been  shown  to be  a karst  aquifer, one
dominated  by  conduits  in which  flow  is  convergent;  therefore,
efficient,  reliable, and  time-sensitive monitoring of  off-site
movement of contaminants is best accomplished through the sampling
of springs  shown to be  connected  to the  landfill by  tracing as
described by  Quinlan and  Ewers  (1985)  and  Quinlan  (1989).   The
tracer tests have clearly shown that most of the ground water  flows
rapidly via conduits to Illinois Central-Quarry springs during high
flows.  Off-site releases of contaminants in directions other than
the  Illinois  Central-Quarry  system  would more  efficiently and
reliably be detected at some,  or all,  of those 32  springs shown by
tracing to have intermittent connection with Lemon Lane rather than
off-site wells.

     (3)  Since  movement  of  ground  water  occurs  in  different
directions during different flow regimes, monitoring schedules must
be adjusted on the  basis  of hydrologic criteria.    For  example,
during pre-excavation baseline sampling, it is proposed to collect
samples during  storm events as well  as during base  flow periods.
When material is being excavated from Lemon Lane, it is proposed to
sample Illinois Central Spring and Quarry Springs  weekly and also
during every  storm  event of sufficient magnitude and duration to
elevate the discharge of those springs above their previous week's
base flow discharge.

     (4) On-site monitoring wells will provide water level data and
samples to  accomplish monitoring of  ground water in  storage near
the site.  This is a necessary component of  a  complete monitoring
plan but, at Lemon Lane,  the on-site wells were shown by tracing to
be unreliable monitors for ground water leaving the site.  This is
because wells tend to intersect ground water in the diffuse bedrock
aquifer and miss the conduits that  convey most of  the water.   Part
of that  ground water  is  recharge  from the  backflooding of the
conduits.  However,  the  probability of wells  intersecting conduits
conveying  the  majority  of  the ground water,  and,  hence, the
contaminants leaving the site is low.  For the same reason, the use
of proposed off-site wells to intercept contaminants that have left
the site is considered to be inefficient and unreliable, especially
since  tracing  has  shown  the  springs  to  be  time-sensitive and
accurate monitoring stations for the potential off-site release of
contaminants.
                                 370

-------
     DEVELOPMENT OP A MONITORING PROGRAM AT A SUPERFUND SITE

          IN A KARST TERRANE NEAR BLOOMINGTON,  INDIANA

         Michael R. McCann, Westinghouse Electric Corp.
               Noel C. Krothe,  Indiana University
                      Bloomington, Indiana
Q: Do you have any reason to believe that all the dye is out of the
system?

A: On  the  contrary,  we have evidence  the dyes are still  in the
system. As late as September 23, 1991 Illinois Central Spring had
8.7 ppb Fluorescein and 12.3 ppb Rhodamine WT and Quarry Spring had
1.8 ppb Fluorescein and 8.0 ppb Rhodamine WT detected in grab water
samples.    We  believe this  is  due  to two  reasons.   One  is  that
injection  wells MW-1S and  MW-10  were  relatively "tight" and  a
portion of  the dye simply remains in the fractures near the wells,
draining slowly  into  the conduit system.  This is particulary true
of well MW-1S; during purging of this well for PCB  sampling a great
deal of concentrated  Rhodamine  WT was evacuated.  Apparently,  very
little of  the dye  from this  well  entered the  system at  all.
Secondly,  a portion of the dye is thought to be in storage in the
epikarst and adjacent  fractured bedrock.  This storage is either an
artificial  result of injection  into rock borings under rising water
levels, or  a  result  of  the backflooding  of the conduit  system
during the  storm event.
Q: When is it best to  inject dye?  Why didn't you catch the dye at
its peak concentration?

A: If time and money will  allow, it is best to inject dye first on
a moderate flow event,  followed by repeat injections on a low flow
and  then  high  flow event.    At a  minimum,  one  low  flow event
followed by one high flow  event should be conducted.  The low flow
event is necessary  to  formulate your base-flow baseline sampling
schedule.   The high flow event is necessary, not only to establish
your high-flow baseline sampling schedule, but  also  to make sure
all off-site flow routes have been established, because upper level
overflow routes may exist that operate  only at the higher flows.
     The  dye pulse came  through  quickly,  and  even  though  the
sampling  frequency  was every 15  minutes,  it  may not  have been
enough to  catch the  peak.  In retrospect, when dye begins to appear
visually,   it is probably necessary to  increase  one's  sampling
frequency to every 5 minutes or less.
                               371

-------
372

-------
                   HETEROGENEITY IN CARBONATE AQUIFERS;

     EFFECTS OF SCALE, FISSURATION, LITHOLOGY AND KARSTIFICATION


                   SMART, P.L., A.J. EDWARDS AND S.L. HOBBS+

                    Department of Geography, University of Bristol,
                             BRISTOL, BS8 1SS, England.

                        +Aspinwall and Co.  Ltd, Walford Manor,
                          SHREWSBURY, SY4 2HH, England.
Abstract
The  flow transmission  properties of carbonate  aquifers may be  determined  and inter-
compared using  cumulative  probability  plots of log hydraulic  conductivity obtained from
slug  and bailer tests in  observation boreholes.  Such tests are simple and inexpensive, and
allow determination of both  average hydraulic conductivity,  and an  indication of the extent
of aquifer heterogeneity.  Karstified aquifers, such as the Carboniferous Limestone of the
Mendip Hills,  exhibit a  large range of hydraulic conductivities, while fissure flow aquifers,
such as  the Great Oolite  of the adjacent  Cotswold Hills,  have   both  higher hydraulic
conductivities  and are  less  heterogeneous.   The  importance  of fissure  density  and
karstification can  be illustrated  using  caliper logs  and aquifer  tests of different scales.
Spatial variations  of  aquifer  properties also  occur,  and can  be related to  variations  in
lithology and structure,  and  to the effects of quarrying.
Introduction
It is generally recognised that the behaviour of carbonate aquifers  is dependent on the type
of groundwater flow  occurring, although the recharge and storage properties are also of
importance (Smart and Hobbs,  1986).  Various authors suggest a continuum of flow types
ranged between  two  end members  (conduit  and diffuse  flow;  White (1977), Smart  and
Hobbs (1986)),  three  end members  (conduit,  diffuse  and  fissure network  flow; Atkinson
(1985)) and even  four end-members (granular, fracture, diffuse and conduit flow; Quinlan
(1988)).  Hydrodynamic (Aley, 1975; Atkinson, 1977) hydrochemical (Schuster and  White,
1971) methods have been  used to characterise flow type for particular aquifers,  and a list
of possible characteristic  criteria  was presented  by  Smart and Hobbs (1986).   Another
approach is to consider the spatial  variation of aquifer transmission properties which can
be obtained  from borehole  aquifer  tests  (specific  yield,  transmissivity  or  hydraulic
conductivity).   Diffuse flow  aquifers would  be expected  to exhibit only limited  spatial
                                       373

-------
variations while  karstified  aquifers  are  characteristically  heterogeneous and would  show
much greater  variation (Daoxian, 1986;  Hobbs  and Smart, 1988), in this paper we present
data derived from simple slug and bailer tests, which can conveniently be displayed  using
cumulative probability  plots,  a  tool often used by  hydrogeologists to display well specific
capacity  data  (Csallany and Walton,  1963).  The data  may also assist in  understanding the
factors controlling the spatial distribution of aquifer hydraulic conductivity,  and  provide a
data-base suitable for parametarisation of numerical aquifer models.
Methods
Slug and bailer tests (Ferris and Knowles, 1963) provide a simple, quick and cheap method
for  estimating   aquifer  hydraulic  conductivity.    Unlike  the   pumping  tests,   more
conventionally employed  to determine  aquifer parameters, they do  not require installation
of submersible pumps, and can thus be undertaken in small diameter observation boreholes.
The logistic support required to install, power and discharge water  from the pump is thus
eliminated.  Typically slug  and bailer  tests  can be completed  in a  few hours and require
only one or at  the most two  operators they are therefore  ideal for  geotechnical and other
low costs investigations  in  which  development  of a  pumping supply borehole  is not the
prime objective.  In these tests, a slug of water is either  introduced into or removed  from
the borehole, and the return of the water  level to pre-existing conditions is monitored,
preferably using  a sensitive pressure transducer and chart recorder or data logger.  The
initial change in  water level should theoretically be instantaneous, and it is often easier to
approximate  this  condition  by  either  inserting or  removing  a long  weighted  float  to
generate the disequilibrium in head  conditions.   A number of theoretical solutions to this
problem have been presented  which permit direct calculation of hydraulic conductivity in a
variety  of different situations (see  review  by Karasaki et al,  1988).   In our  study of water-
table aquifers the solutions  of Bouwer and Rice (1976) and  Van der Kamp (1976) for the
underdamped case were employed.   It should be noted that these solutions are based on the
assumption of  laminar flow  within the aquifer,  wheras  in  aquifers with dissolutionally
enlarged voids turbulent flow may  develop under the test conditions.

Hydraulic conductivity data follow a log-normal distribution and thus can be linearised  by
plotting log hydraulic conductivity on  cumulative frequency probability paper.  The mean
and standard deviation are  determined for  the  normally  distributed log  data, and can  be
used  to calculate the  coefficient  of  variability.   The  standard   deviation can  also  be
converted back  to non-logarithmic  form to  provide upper and lower  limits  which are
asymmetric  about  the  mean,  but  are  expressed  directly in  the units  of hydraulic
conductivity (here m/d).
Comparison of Hydraulic Conductivities from Four Contrasting Carbonate Aquifers


Hydraulic  conductivities  have been  obtained  from  slug  and bailer  tests (henceforth  slug
tests) in four contrasting aquifers.   The Carboniferous  Limestone aquifer of the Mendip
Hills  is a maturely  karstified  aquifer  developed  in  massively   bedded,  well  jointed
limestones  of low primary porosity (< 0.1 %) (Green  and Welch, 1965).  Flow  in the aquifer
is predominantly via  conduits,  but  fractures  and  fissures provide  the majority of the
saturated zone  storage in the diffuse  flow  zone adjacent  to  the conduits (Atkinson, 1977).
Abstraction is  wholly  from  spring sources. The slug tests  were  conducted  in the eastern
Mendips (Hobbs and Smart, 1988; Atkinson et al, 1973),  where surficial karst features are
less  well developed than in the  better known and  more  mature  central  Mendips  area
                                        374

-------
                                  Fractured Quarry Sub-floor
                                 • (Carboniferous Limestone)
: Pleistocene
   Lucayan
  •\ Limestone
                                     Fissured
                                      Great
                                       Oolite
                                    Karstified
                                    Carboniferous
                                     Limestone
                                                                           EB73          Caliper extension (cm)
                                                                              10       15        20       25       30
                                                                                        35
                                                                      2-
                                                                      4-
                                   6-
                                                                   Q.
                                                                   0
                                                                   Q
                                                                      8-
    10-2
                            50   70     90

                    Cumulative probability (%)

Figure   1     Cumulative  probability  plot  of  log   hydraulic
conductivity  determined from slug  tests for  quarry  sub-floor,
Lucayan Limestone,  Great Oolite and Carboniferous  Limestone
aquifers.
                                                                     10-
                                                                     12-
                                   14J
                                                                                                    'Intact' sections - Black

                                                                                                    'Fissured' sections - White
                                                                           -
                                                           Total saturated thickness 13.70m

                                                           Total'fissured'sections   11.55m
                                                           Total Intact' sections      2.15m

                                                           Flssuration =    '5S x 100 = 84.3%
                               Figure 2   Caliper  log for Grand Bahama borehole EB 73, and
                               calculation of fissuration index.

-------
described by Atkinson (1977).  Within the  East Mendip  area, the Carboniferous Limestone
has been extensively  extracted  by hard-rock quarries,  which  extend below  the natural
water table (Edwards et al; this volume).  Modern blasting techniques, which are  designed
to comminute  the  rock,  also cause extensive fracturing of the  Carboniferous Limestone
both below the quarry floor and laterally away from the walls (Holmberg and Maki,  1981;
Gunn and Gagen, 1987).  Where it is  below the water table the sub-floor blast zone can be
considered a separate aquifer characterised by high  fracture density, low  primary porosity
and negligible conduit development.  Our data are derived from shallow (< 10m) boreholes,
drilled into the sub-watertable quarry  floor.

The Great Oolite aquifer of the southern Cotswolds comprises massive oolitic and bioclastic
limestones (Kellaway  and  Welch, 1980),  and has  been  extensively  developed by  large
abstraction  boreholes  (> 10  ML/d).    Transmissivities  are high  and  are  dominated  by
development of dissolutional fissures upon  the bedding planes and extensive joint systems.
These  fissures provide  the majority of the storage  (specific yield c 3%),  but there is also
significant intergrannular porosity (up to 15%). Test data has been derived from a network
of observation  boreholes.  Unlike the Great Oolite and  Carboniferous  Limestone aquifers,
the Pleistocene Lucayan Limestone aquifer of the Bahamas has  never suffered  deep burial
(Beach and  Ginsberg, 1980), and has a high intergrannular porosity  (typically 40%) and
specific  yield (5-6  %).   It has however  been karstified,  with  vadose meteoric  leaching
during times of low Quaternary sea-level,  and dissolution  in the  freshwater phreatic and
mixing zones at times of high  sea-stands (Smart and  Whitaker,  1988), resulting in  vuggy
porosity  and development of  coastal (dune-flank)  caves  (Mylroie and Carrew, 1990).  The
data presented  here have been  derived from  observation boreholes and drainage  wells on
the island of Grand Bahama.

Hydraulic conductivity data from these four contrasting carbonate aquifers  is presented in
Figure 1, while summary statistics are given in Table  1.  The densely  fractured blast zone
aquifer  has  the highest  mean hydraulic  conductivity, and  is also  spatially  the  most
homogeneous of the aquifers.  The  Lucayan  Limestone and Great Oolite  aquifers attain
similar maximum hydraulic conductivities but exhibit a  much greater range of test values,
indicating greater  heterogeneity.   In the  case of the fissured  Great Oolite this greater
heterogeneity results from  a relatively small  proportion of sites  having  either unusually
high, or  very  low  hydraulic conductivities,  the  lower  quartile spanning  two orders of
magnitude.  The former are associated with fissure network zones which are  responsible for
the rapid transmission  observed in  the aquifer  from  tracer tests (Smart  1976), while the
latter represent  tight  zones  where   bedding  plane  and  fracture  density  are low,  and
extensive dissolution to  generate fissures has  not  occurred.  A  similar pattern is  seen for
the karstified  Carboniferous  Limestone, the  data  from which  also span four orders of
magnitude.   None  of the boreholes tested  intersects on  open conduit.   Had this been the
case, much higher  maximum  apparent hydraulic conductivities would  have  been obtained,
significantly  increasing the range.  The mean hydraulic  conductivity of the Carboniferous
Limestone is much  less  than  that of  the Great Oolite,  both because of the more massive
bedding, and because  concentration of  flow  into conduits at an early  state in the aquifer
development limits the extent of dissolution in the diffuse zone  (Ewers 1978).  In the more
densely fractured  Great Oolite, competition between flow routes of similar size ensures a
more uniform distribution of  dissolution (Palmer, 1984).

Thus conduit  flow aquifers  are characterised  by  high heterogeneity and a rather low
transmissivity  dominated  by the diffuse  flow  zone.   Fissured  aquifers  have a  higher
transmissivity and  significantly lower  heterogeneity,  as indicated by the coefficient of
variation (Table  1), and the gradient  of  the log probibility plot.  Both  the Pleistocene
Lucayan Limestone, in  which  intergrannular flow  is  significant,  and the  fracture flow
quarry sub-floor have high transmissivity and low heterogeneity.  The mean transmissivity


                                        376

-------
                        Hydraulic Conductivity (m/d)
                               Quarry Sub-Floor
                               Lucayan Limestone
                       65    70    75    80   85    90    95   100
                                      % Fissuration
Figure 3  Relation between fissuration and log hydraulic conductivity for Lucayan Limestone
and quarry sub-floor aquifers.

                         Hydraulic Conductivity (m/d)
                      10l==
                     10'= =
                      10;
                      10==
                            n   Quarry Sub-Floor

                            0   Carboniferous Lstn.
                            A   Lucayan Limestone
njf]
                                                           <\A
                                                              D
                                                             A
                                                             A
                                                       A
                        EE     0   0
                                    0 0
                      IOH	1	1	1	1	1	1	1	h—
                        10   20   30  40   50  60   70   80  90  100
                                      %  Fissuration

 Figure 4   Comparison  of relation  between fissuration and  log  hydraulic conductivity for
 Carboniferous Limestone and Lucayan Limestone  and quarry sub-floor aquifers.
                                           377

-------
is thus an indication of the  extent of dissolution enlargement of voids  within the aquifer,
while the coefficient of variation (or slope of the log probability plot) indicates the  degree
of flow concentration occuring on the continuum, from diffuse to conduit flow (Smart and
Hobbs,  1986).
Role of Dissolution and Void Integration
If the  aquifer hydraulic conductivity at  a  particular site is controlled by the development
of dissolutional fissures,  it should  be possible  to  predict  hydraulic  conductivity  directly
from the aperture  and number  of  fissures  penetrated by  the  borehole.   In order  to
investigate this relationship, selected boreholes were caliper logged using a Gerhart T-450
borehole logging system  (Keys and Maccary, 1971).   An index  of  fissuration was then
derived from the caliper  logs by  summing the  length  of  borehole segments which were
greater than  the  nominal drilled diameter,  and expressing this sum as a percentage of the
total saturated borehole depth  (Figure  2).   The  index should  thus  depend both on the
number and  size of fissure openings, although in practice the latter will be overestimated
by 'break-out' of  the  drill  bit into  the fissure.   It  is also  important  to  remember  that
widening  of  the hole may not simply be  due to the presence  of open  fissures, but can
occur  in  softer  rock  units,  such  as  poorly cemented  limestones in  the  diagenetically
inmature Lucayan Limestone, or shales in the Carboniferous Limestone.   Furthermore, the
amount of 'break-out' which  occurs into  a  fissure  is  strongly dependent on  method of
drilling, drilling rate and rock matrix strength, which are not controlled in our data set.

Figure  3  presents data from the  Lucayan  Limestone  aquifer  of Grand Bahama, and the
quarry  sub-floor  aquifer.   For  these  there  is  a  remarkably  good   relation  between
percentage fissuration and the measured hydraulic conductivity (which is not dependent on
the coincidence of the quarry and Lucayan Limestone relationships).   Thus even in young
limestones with high intergrannular porosity, the transmission properties  of the aquifer are
governed  by development of dissolutional  fissures.  In  the Bahamas  these form along the
frequent exposure  surfaces which terminate  early carbonate sub-unit and  are developed
during  low  sea-stands; such surfaces are laterally continuous and  thus generate very high
horizontal hydraulic conductivities.

In Figure 4,  the Bahamas  and quarry sub-floor  data are supplemented by a data from the
karstified Carboniferous  Limestone  aquifer.   Because  the  intergrannular  porosity of this
limestone is  very low,  secondary  voids would be  expected  to  dominate  the  transmissivity.
In fact there is  not  a  significant relationship between the fissuration index and  borehole
hydraulic conductivity  (it should however be noted that  unfortunately the logged boreholes
did  not include  any sites with above average  hydraulic  conductivities).    The  probable
explanation  for this unexpected finding is  related to the connectivity of  individual fissure
openings.    Packer  testing  has  shown  previously  that  some  fissures  yield  hydraulic
conductivities similar to the bulk  rock,  indicating that  they  do not  interconnect laterally
with other voids in the aquifer (Hobbs and Smart, 1988).  They thus contribute to storage
within the aquifer  by delayed leakage, but  are not important for flow. Thus as emphasized
by Sendlein and  Palmquist (1977)  void intergeration  is critical in controlling transmission in
karstified rocks. Whilst in the blast zone and Bahamas  fissure voids  integrate laterally,  in
the  Carboniferous  Limestone this is not the case  and the resulting  aquifer is both less
transmissive  and more heterogeneous (Figure 1).

Careful examination of the data for the individual  aquifers (Figure  1) suggests that there
may be sub-populations present in the  data,  as  indicated by  linear segments of differing
gradient on  the  cummulative  probability plot.   This is illustrated for  the Carboniferous
                                        378

-------
               10'
            o
            •§  10°-
            o
            o
            o
            1
            t
               10-'
               10-
                              'Well Connected Fissures'
                                     'Fissures'
Fractures
                                                      Unconnected
                                                    ^ Fractures'
                            10       30    50    70       90
                                Cumulative probability (%)
                      99
Figure 5  Slug test data for the karstified Carboniferous Limestone sub-divided into linear
components.
 Table 1        Comparison of hydraulic conductivity data for the four aquifers
Carbonate
Aquifer Type
Fractured Quarry
Sub-Floor
Pleistocene Lucayan
Limestone
Fissured Great Oolite
Karstified Carboniferous
Mean
(m/d)
305
96.2
31.8
0.214
Coefficient
of Variation
(%)
10.4
29.5
70.1
179
Total Number
of Boreholes
11
44
24
46




 Limestone
                                        379

-------
Limestone  aquifer (for  which  we have  most data) in  Figure 5, which shows  four sub-
populations represented  by straight line segments.  These sub-populations may simply be a
statistical artefact, but could also represent  the  combined  effects of void  integration  and
dissolution.   Fractures  (width mm)  are unmodified  joints  and beddings developed  by
unloading  an  re-exposure of  the limestone,  while  fissures  (width cm)  have suffered
dissolutional modification, and in some cases  may  have  turbulent flow.   They have,
however, not expanded to the size of  conduits  (width tens of cm to m).  Thus while
fractures give lower hydraulic conductivities  than the larger aperture fissures, the degree of
void  integration  is  also  critical.    Unconnected  fractures  give  very  low  hydraulic
conductivities, which increase significantly once  integration  of the net occurs, for instance
by  continued unloading (or  in the quarry  sub-floor case by blasting).    The enhanced
circulation  permits dissolution generating fissures, but initially these  are  interconnected
only by fractures  which effectively limit the  increase in hydraulic conductivity. Continued
circulation expands these fracture links giving a net of highly connected fissures with high
hydraulic conductivity.   In fact  it is  precisely  such  a net which characterise  the Great
Oolite aquifer.   These  suggestions could be more  rigourously  tested  by  application  of
simulation models, such as those  developed by  Foster and  Milton (1974)  for the  Chalk
aquifer of  southern England.
Controls on the Spatial Variation of Aquifer Hydraulic Conductivity


Because slug tests are relatively rapid and simple they can  be used to  generate sufficiently
large data sets that controls  on the  spatial variations of  aquifer hydraulic conductivity can
be investigated (provided sufficient suitable boreholes are available).  For the East Mendip
Carboniferous Limestone data set, an initial sub-division was made into boreholes on the
steeply dipping (c 70°) northern limb of the Beacon  Hill  pericline,  and  those from the more
gentle southern  limb (dip  c  30°)  (Figure   6).   These differ significantly  in hydraulic
conductivity (at the  95%  confidence  interval),  the northern limb  having  an  average
hydraulic conductivity on order of magnitude less  than the southern  limb.  The northern
limb also has markedly less sites in  the 'fissure' range (10° to 10 ^  m/d).  These  differences
are paralleled by a much greater  degree of  fissuration on the southern (average fissuration
index  78.6%)  than  the northern limb (34.6%).  As  both areas have a similar geomorphic
history, this difference  may  be  attributed  to  the  difference  in  dip which controls the
number of  bedding planes intersected by a  vertical borehole.  This is much lower in the
steeply dipping beds, of  the  northern  limb, than  for  the  more  gently dipping southern
limb.   Because  it is the  bedding planes  which  provide  laterally continuous and  thus
interconnected openings, the steep dips yield lower hydraulic conductivities on the northern
limb.

The Carboniferous Limestone data was  also sub-divided with respect  to the  lithological
group within  the  Carboniferous Limestone  Series (Green  and  Welch,  1965).   Again
significant differences are apparent, both with regard to the average hydraulic conductivity
and the heterogeneity  indicated  by the  coefficient of  variation (Figure  7,  Table  2).
Although there are lithological differences  between the groups,   (the  Vallis and Hotwells
Limestones  are  coarse,  pure  bioclastic  limestone,   the Clifton  Down  contains a major
mudstone unit,  and the  coarse bioclastic Blackrock  Limestone has both  significant shales
and chert beds), these lithological factors appear to be less important than disposition in
the  aquifer.  Thus  the impure  Blackrock  Limestone  would be  expected on  lithological
grounds to be less transmissive than the pure  Vallis  Limestone.  The converse is in fact the
case, because the Blackrock  Limestone has been exposed to  much  greater dissolution as it is
at the base  of the  limestones, and receives  aggressive allogenic runoff and diffuse leakage
from the topographically  higher  siliclastic  rocks  forming the core  of the  anticline.   The


                                        380

-------
CO
00
              101
           o
o
o
o
              10°-
             10-'
 Northern Limb
Dip 70° (34.6%)\
                                             (Percentage fissuration
                                                  in brackets)
VSouthern L/nrt>
     30° (78.6%)
                             10     30   50    70     90
                             Cumulative probability (%)
                                                  99
        Figure  6    Comparison  o)  hydraulic  conductivity  data for
        boreholes from  the northern and southern limbs of the Beacon
        Hill pericline. Carboniferous Limestone aquifer.
                                                                            10
                                                                                                     •  Black Rock
                                                                                                     o  Clifton Down
                                                                                                     •  Hotwells
                                                                                                     •  Vallis
                                                                            10-2
                                                                          10      30   50   70     90
                                                                         Cumulative  probability (%)
                                                             Figure  7    Slug  test  data  for  the Carboniferous  Limestone
                                                             aquifer sub-divided into lithologic groups.

-------
Vallis  Limestone  in  contrast  has received  only autogenic  recharge,  and dissolutional
integration of the initial fractures has  been limited.

Interestingly,  the Blackrock Limestone  has also  been more extensively  quarried  than  the
other limestone groups, and  some of the  tested  boreholes are  in  fact  drilled  within  the
quarries.  If we only consider sites which  have  water-levels well below  (< 10 m)  the sub-
floor blast zone, these have hydraulic  conductivities much higher than boreholes which  are
remote from quarries (>150 m;  Figure 8).  Whilst  this may be explained  by the  lithological
controls discussed above, the implications for geotechnical aquifer parameterisation studies
which draw data only from the  quarry sites are potentially serious.
Table 2       Hydraulic conductivity data for the Carboniferous Limestone sub-divided
              into different lithological groups
Group
Blackrock
Clifton Down
Hotwells
Vallis
Mean
(m/d)
0.97
0.123
0.102
0.025
Coefficient
of variation
(%)
9330
120
88.4
43.8
Total Number
of Boreholes
22
7
7
9




Effect of Scale
Kiraly (1975) has  suggested  that in karstified aquifers there is  a systematic  relationship
between  the  effective transmissivity of the aquifer and the scale considered.   Thus at the
catchment scale  conduits dominate transmission, which at the other extreme in core samples
only intergrannular voids are present, and measured hydraulic conductivities are very low.
Slug tests involve displacement of only a limited volume of water,  and thus 'sample' only a
relatively small zone of the aquifer  within 1 to 2 m of the  test borehole (depending on slug
size and  aquifer hydraulic conductivity). In contrast pump tests may cause more extensive
drawdown over  distances from tens  to hundreds of metres from the test borehole.  Figure 9
contrasts our slug test results with pump test data for Grand  Bahama obtained by Little et
al (1975). Over 60% of the pump test  results have higher hydraulic conductivities than the
maximum recorded in  the slug tests, but the two data sets give similar minimum values.
This suggests that  in pump  tests  the cone  of depression expands  to intersect dissolutional
conduits, whose spacing is  sufficiently wide that  they  are not  penetrated  directly  by
random boreholes.   It  should also  be  born in mind that  at  high  hydraulic  conductivities
such  as  these  turbulent  flow may  develop  and the transmissivities  determined  may be
erroneous.
                                        382

-------
                100%
                 75%
                 50%
                 25%
                   0%
                       10-3   10-2    10-1   10 0   10 1   10 2   10 3
                            Hydraulic Conductivity (m/d)
                                      Borehole Location
                       •• Outside Quarry      I   I Within Quarry
                       I   I Quarry Sub-Floor
Figure 8  Comparison of hydraulic conductivities for boreholes within quarries, in the quarry
sub-floor, and greater than 150 m from quarries,Carboniferous Limestone aquifer.
                           Slug and
                          Bailer Tests
                  10'
                                                                99
                       1        10      30    50    70      90
                                Cumulative probability (%)
Figure 9  Comparison of  hydraulic conductivity  data  for  the Lucayan Limestone aquifer,
Grand Bahama derived from pump and slug tests.
                                        383

-------
Conclusions
In our study we have made use of small diameter observation boreholes which do not have
pumps installed, and are therefore readily tested by slug and bailer methods.  In other areas
where large  numbers of  domestic abstraction wells fitted with  pumps are available,  use
could be made of borehole specific capacity data (or transmissivity derived from this data)
see for instance (Sallany  and Walton, 1963).   In both  cases log probability plots offer an
easy and effective method for quantification of  the position of  the aquifer on  the flow
spectrum from difuse to  conduit  within the conceptual model of  Smart and hobbs (1986).
Different carbonate aquifers  can be readily compared and contrasted, and insight gained on
the flow properties.  Hydraulic conductivity in carbonate aquifers is controlled by void size
and void integration.  In the  more densely  fractured aquifers, void integration  is effectively
complete, giving  a  relatively  homogeneous  aquifer, and  the density  of  fissuration  and
degree  of   dissolution  enlargement  are  the main factors  controlling   mean  hydraulic
conductivity.  In  less densely  fractured aquifers,  void  integration becomes  much more
important, and where dissolutional development is not extensive, the aquifer is much more
heterogeneous, especially  where flow concentration  into conduits  has  occured.  Lithology
and structure may be important in controlling aquifer transmissivity, but disposition of  the
unit within  the aquifer may be equally  important  where dissolutional  void enlargement is
important.   Thus, the approach to aquifer characterisation exemplified in this paper has
two  important applications.   On  a practical  basis it provides a means of describing  the
transmission  properties  of carbonate aquifers, providing  data  in a  format  suitable  for
stochastic parameterisation of aquifer models. It also provides a simple design tool useful
for prediction of  borehole yields, and the  preferred mode of aquifer  development (Smart
and Hobbs,  1986).  On  a  more academic level it provides data to elucidate the controls on
transmission  and structure of  carbonate aquifers.   This also  has  implications for aquifer
management,  for  instance transmission of pollutants  from waste disposal sites via  the
diffuse flow zone in karstified limestones (Edwards and Smart, 1989)  and the validity of
geotechnical measurements of aquifer parameters.
Acknowledgements
The authors  would like  to  thank the  following  for access to,  and in some  cases drilling,
boreholes: ARC Southern Ltd, Foster Yoeman Ltd, Wimpey Hobbs Ltd, ECC Quarries Ltd,
National River Authority;  Wessex  Region, Wessex  Water  Pic., the  Government of  the
Bahamas, the Bahamian Ministry of Works  and  Utilities and the Freeport Water Company.
The authors  would also  like to  thank the  British Geological Survey  for  the  loan of  the
geophysical calliper log  equipment.   A.J.  Edwards and  S.L.  Hobbs were  supported  by
NERC training awards in collaboration with ARC Southern Ltd, and Bristol Waterworks
Company and Wessex Water Authority respectively.  The Bahamas fieldwork  was supported
by  Amoco UK Ltd.
References Cited


Aley, T. 1975.  A predictive hydrological model for evaluating the effects of land use and
       management  on the quantity  and  quality of  water  from  Ozark  Springs.  Ozark
       underground Laboratory, Springfield Missourri.
                                       384

-------
Atkinson,  T.C.  1977.   Diffuse flow and conduit flow in limestone terrain in the Mendip
       Hills, Somerset.  Journal of Hydrology,  35,93-110.

Atkinson,  T.C.  1985.  Present and future directions in karst hydrogeology.  Annales de la
       Societe Geologique de Belgique, 108, 293-296.

Atkinson,  T.C., Smith,  D.I.,  Lavis, J.J. and  Whitaker, R.J. 1973.  Experiments in tracing
       underground water in limestones. Journal of Hydrology, 19, 323-349.

Beach, D.K.  and  Ginsberg, R.N.  1980.   Facies  and  succession  of Pliocene-Pleistocene
       carbonates, Northwestern Great Bahama Bank.  American Association of Petroleum
       Geologists Bulletin, 64, 1634-1641.

Bouwer, H.F. and  Rice, R.C.  1976.  A slug test for determining hydraulic conductivity of
       unconfined aquifers with completely or  partially penetrating wells.  Water Resources
       Research, 12, 423-428.

Csallany, S. and Walton, W.C. 1963.  Yields  of  Shallow Dolomite Wells in Northern Illinois.
       Illinois State Water Survey, 1-43.

Dao Xian, Y.  1986.   On the heterogeneity  of karst water.   International Association of
       Hydrological Sciences Publication,  161, 281-292.

Edwards,  A.J., Hobbs, S.L. and Smart,  P.L.  1991.   Effect of quarry dewatering on a
       limestone aquifer: case study from the  Mendip Hills, England.   Proceedings of the
       Third Conference on  Hydrology, Ecology,  Monitoring and  Management of  Ground
       Water in Karst Terraines, (in press).

Edwards,  A.J. and Smart, P.L. 1989.  Waste  Disposal on karstified Carboniferous Limestone
       aquifers  of  England  and  Wales.    Proceedings  of  the  3rd  Multidisciplinary
       Conference on  Sinkholes  and  Engineering  and  Environmental  Impacts of  Karsts.
       Balkema, Rotterdam,  165-182.

Ewers, R.O. 1978.  A  model for  the  development of broad scale networks of groundwater
       flow in steeply  dipping  carbonate  aquifers.   Transactions of the  British Cave
       Research Association, 5, 121-125.

Ferris, J.G. and Knowles, D.B. 1963.   The slug-injection  test for estimating the coefficient
       of Transmissivity  of an aquifer.   United  States  Geological  Survey  Water  Supply
       Paper,  1536-1,  299-304.

Foster, S.S. and Milton, V.A.  1974.  The  permeability and  storage  of an unconfined chalk
       aquifer. Hydrological Sciences Bulletin, XIX, 485-500.

Green, G.W. and  Welch, F.B.A. 1965.  Geology of the Country  around  Wells  and Cheddar.
       H.M.S.O.,  London, pp. 225.

Gunn, J.  and Gagen,  P.,  1987.   Limestone quarrying and sinkhole development in the
       English  Peak  district.  Proceedings  of the 2nd  Multidisciplinary Conference on
       Sinkholes   and   Engineering   and  Environmental  Impacts  of   Karst.    Balkema,
       Rotterdam, 121-125.
                                       385

-------
Hobbs, S.L. and Smart, P.L., 1988.  Heterogeneity in carbonate aquifers: a case study from
       the  Mendip  Hills,  England.    Proceedings  of  the  Internation Association  of
       Hydrogeologist 21st Conference.  Geological Publishing House, Beijing, 724-300.

Holmberg, R. and Maki, K., 1981.  Case examples of blasting damage and its influence on
       slope stability.   Proceedings of the 3rd  International  Conference on  Stability in
       Surface Mining.  Vancouver, British Columbia, 773-793.

Karasaki, K., Long,  J.C.S. and Witherspoon, P.A. 1988.  Analytical models of slug tests.
       Water Resources Research,  24, 115-126.

Kellaway, G.A. and Welch, F.B.A. 1980.  British Regional Geology, Bristol and Gloucester
       District (Second Edition).  H.M.S.O., London.

Keys,  W.S. and Maccary,  L.M.  1971.   Applications  of borehole  geophysics  to water
       resources investigations.  Techniques of Water-Resources Investigations of the United
       States Geological Survey, chapter El, 1-26.

Kiraly,  L.  1975.    Rapport  sur  e'etat  actuel  des connaissanies dans  le  domaine  des
       characteres physiques des roches karstique.  In  Bwrger, A. and Duberet, L. (eds)
       Hydrology of Karstic  Terrains, International Union of Geological Sciences, Series B,
       3, 53-67.

Little, B.C., Buckley, O.K., Cant,  R., Jefferies, A.,  Stark, J. and Young, R.N. 1975. Land
       Resources  of the  Commonwealth   of  the  Bahamas,  Grand  Bahama.   Overseas
       Development Authority, Surbiton, Surrey, pp. 87.

Mylroie,  J.E.  and Carew,  J.L.   1990.   The  flank margin  model  for  dissolution cave
       development in  carbonate  platforms.  Earth  Surface Processes and Land forms, 15,
       413-424.

Palmer, A.N. 1984.  Geomorphic interpretation of karst features.   In  La Fleur, R.G. (ed).
       Groundwater as  a Geomorphic Agent.   George Allen and Unwin,  New York, 173-
       209.

Quinlan,  J.F.  1988.    Special  Problems of  groundwater  monitoring  in karst  terraines.
       Symposium  on  Standards  Development  for   Groundwater   and  Vadose  Zone
       Monitoring   Investigations.     American  Society  for  Testing   and  Materials,
       Philadelphia, 1-37.

Sendlein, L.V.A. and Palmquist, R.C.  1977. Strategic placement of waste disposal sites in
       karst regions.   In   Tolson, J.S.  and  Doyle, F.L.,  (ed).   Karst Hydrogeology.
       University of Alabama, Huntsville, 323-336.

Schuster,  E.T.  and White, W.B. 1971.   Seasonal fluctuations in the chemistry of limestone
       springs:  a  possible  means  for characterising  carbonate   aquifers.    Journal  of
       Hydrology, 14, 93-128.

Smart, P.L.  and Hobbs,  S.L. 1986.  Characterisation of  carbonate aquifers:  a conceptual
       base.   Proceedings  of the  Environmental  Problems  in Karst  Terrains  and their
       Solution.  National Water Well Association, Dublin, Ohio, 1-14.
                                        386

-------
Smart,  P.L. and Whitaker,  F.F.  1988.  Controls on the rate and distribution  of  carbonate
       bed-rock dissolution in the Bahamas.   Proceedings of the 4th Symposium on  the
       Geology of the Bahamas.  Bahamian Field Station, Fort Lauderdale, 313-321.

Smart,  P.L. 1976.  Catchment  delimination in  karst areas  by the use of quantitative tracer
       method.  Proceedings  of the 3rd  International Symposium of Underground Water
       Tracing.  Institute of Karst Research, Postojna, 291-298.

Van der  Kamp, G. 1976.  Determining aquifer transmissivity by means  of well response
       tests.  Water Resources  Research, 12, 71-77.

White,  W.B. 1977.  Conceptual models for carbonate aquifers: Revisited.   In  Dilamarter,
       R.R. and Csallany, S.C., Hydrologic Problems  in Karst Regions.  Western  Kentucky
       University Press, Kentucky, 176-187.
                                      387

-------
HETEROGENEITY IN CARBONATE AQUIFERS: EFFECTS OF SCALE, FISSURATION,
LITHOLOGY, AND KARSTIFICATION

By: Peter L. Smart, Alan J. Edwards, and Steve L. Hobbs


Q.  Relative to the use of  caliper logs  to characterize the degree
    of fissuration,  how was fracture aperture  determined?   What
    about single solutionally enlarged fractures? The caliper log
    would seem to be an indicator of borehole  instability and be a
    measure of factors other than single fracture density.

A.  Borehole caliper logs are controlled by the degree of aquifer
    fissuration  as  well  as  the  drilling method  and  borehole
    instability.   The  percentage fissuration  index described in
    this  paper  is a crude method  of  measurement.    However,  it
    offers  a pragmatic  and  simple technique  which we believe
    generates interesting and useful results.


Q.  Did you use  different methods of slug-testing? Did you observe
    any differences between the hydraulic conductivities determined
    with various slug-test methods and a bailer?  If  so, how do you
    account for any observed differences?

A.  In all slug  and bailer tests, weighted floats  were inserted and
    removed  from the  water within  the tested  borehole.   This
    technique was utilized in all the tests.  Hydraulic conductivi-
    ties determined from the slug rather than the bailer  test were
    generally higher.  This was  because the rise in water level,
    due to float insertion,  resulted in  an immediate water loss to
    the surrounding  unsaturated  rock.  In order  to  overcome this
    disparity, the average hydraulic conductivity from 6 slug and
    bailer tests was calculated for each tested borehole.
                                388

-------
  CAUSTIC WASTE CONTAMINATION  OF KARSTIC LIMESTONE  AQUIFERS

                    IN TWO AREAS OF JAMAICA


                       BASIL P. FERNANDEZ

                  UNDERGROUND WATER AUTHORITY
               HOPE GARDENS, KINGSTON,  JAMAICA
Abstract

Jamaica's bauxite  deposits occur in association with the
karstified White Limestone Formation of  Tertiary Age.  The
refining of bauxite  to  alumina results  in  a  waste known
locally as "red mud", which is stored  in unsealed mined out
karstic depressions.  The red mud, which is  more than 70% water
is highly caustic  and infiltrates to the groundwater table.
Groundwater contaminated by "red mud",  shows increased sodium
concentration and  increased pH.  Monitoring  of  groundwater
quality in two areas  of the island, where  bauxite/alumina plants
are located, indicate significant contamination over the last
20 years.  In these  two areas, caustic  contamination has resul-
ted in over 20 km  of  aquifer being unsuited  for groundwater
development.  The  mud disposal ponds are located in the direct
path of groundwater  flow and still poses a  serious threat to
groundwater reservoirs  and consequently  the  groundwater reserves
of the island.  Remedial measures to reduce  contamination have
included sealing of  disposal ponds, thickening  of mud and solar
drying and recycling  of liquid fraction.


Introduction

A highly caustic waste  known locally as  "red mud" is a by-product
of the extraction  of  alumina from Jamaican  bauxite, which occurs
as an interfingered  blanket deposit in  association with the
limestone deposits forming the exploitable  aquifer of the island.
Four plants located  on  the south coast  of  the island, produce
approximately 3Mm  of red mud waste annually and an increase to
4.5Mm  is expected with expansion of the present plants and
construction of one  new plant.  The red  mud  waste is stored  in
mined out or topographic depressions in  the  limestone.
                                389

-------
                      TABLE 1   CROUNDWATER RESOURCES AVAILABILITY
                                     M»3/Yr     Hied     Percent of
                                     Mm /Yr     MIgd       Total


           GROUNDWATER SAFE YIELD          3418      2062
             FROM LIMESTONE AQUIFERS       3294      1987        96

             FROM ALLUVIUM AQUIFERS        124        75         4
           PRODUCTION (WELLS & SPRINGS)      839       506

              FROM LIMESTONE AQUIFERS       '06       426        84
              FROM ALLUVIUM AQUIFERS        133        80        16
           UNUTILIZED RESOURCES
                                      2579      1555        75
Limestone  rocks  in  Jamaica have  a high risk  factor  for  pollution
because of  the development of  relatively  large channels  for
groundwater  infiltration  and percolation  (karstification).  The
limestone  aquifers  in Jamaica  are a major  source  of water supply
because of  the high transmissivities  and  storage  especially
along fault  and  fracture  lines  and in  areas  where  karstification
has  produced a network  of solution openings.   Limestone  rocks
underlie 67% of  the island (figure 1)  and  important groundwater
reservoirs  are to be found in the limestone  close  to major urban
and  agricultural  areas.   There  are more than  400  wells  tapping
the  limestone aquifer and abstracting  approximately 706  Mm/ yr
of water or  8470  of  the  total water produced  (Table  1).

Th&  disposal sites  (figure 1)  are directly  in the  flow  path of
groundwater  to major reservoir  areas.   No  structural or  litho-
logic barriers exist in  the limestone  and  changes  in groundwater
abstraction  patterns could induce a more  wider.spread of  the
contaminant.  At  sites  1  and 4  (figure 1)  caustic  contamination
of groundwater has  been  detected.  Several  springs  and  wells
have  been  abandoned and  a significant  portion of  the limestone
aquifer adjoining  these  sites  is unsuitable  for further  ground-
water development.


Geology/Hydrogeology of  the Limestone

The  White  Limestone Formation  is a thick  series of  pure  calcium
carbonate  ranging  in age  from  middle  Eocene  to Miocene  (47M.'yrs -
10M  yrs).   The entire sequence  is estimated  to be  more  than
3,000 meters thick  and  has been  subdivided  into ten (10)  bio-
stratigraphic and  four  (4) hydro-stratigraphic units.
Hydrogeologic properties  of the  entire series are  broadly similar
and  there  is hydraulic  continuity between  all the  units.
Extensive  tectonism has  resulted in folding,  faulting and frac-
turing of  the rocks.
                                   390

-------
    Figure
                            SIMPLIFIED  GEOLOGY  OF  JAMAICA
        QroundwQtar Flow Direction
        Faults
        FTOURE  2
                       HYDRO-STRATIGRAPHIC   MAP  OF   JAMAICA
           E II 0
     HTtmO-STHATOOIWHIC WITS
  l::-:l Alluviim AqiHftr/AquteMt
  E3 Cental AquioMl/Ai|
-------
In the lowlands,  solution of the  limestone has resulted  in  an
integrated network  of  solution openings,  imparting to  the
aquifer almost homogenous and isotopic  properties.  In the
lowlands conduits which originate  in  the  uplands may continue
as cave systems  or  subterranean streams.   Most of the  recharge
of the limestone  aquifer takes place  in  the uplands,  where  soil
cover is absent  and  rainfall is the highest,  while storage  is
in that portion  of  the  aquifer underlying  the coastal  plains.
Lateral flow of  groundwater takes  place  both  in the vadose  and
phreatic zones.   In  the vadose zone,  lateral  flow is  rapid  and
confined to narrow  channels which  may continue into the
phreatic zone.   Most of the water  in  the phreatic zone moves
through the interconnected network of solution openings and
fracturejoints.


.Limestone Hydrostratigraphic Units

The limestone  is  subdivided in four (4) hydrostratigraphic
units (figure  2)  viz:

         Limestone Aquifer
         Limestone Aquiclude
         Coastal  Aquifer
         Coastal  Aquiclude
The limestone  aquifer  is comprised of members of  the  White
Limestone Group.  They  form a sequence of  moderately  compacted,
well bedded, partially  crystallised bioclastic and micritic
limestones with  thickness in excess of 2,000  metres.  The
aquifer exhibits  mature karstic features,  typified by  a very
high infiltration capacity,  predominant sub-surface drainage
and highly compartmentalised sub-surface conduit  flow.
Its karstic nature makes the limestone aquifer very suscep-
tible to contamination.

The limestone  aquiclude is a soft, fine grained chalk  of  low
permeability which  fringes the coast ponding  groundwater
within the juxta  posiitoned limestone aquifer.   Thickness
approaches 1000  metres.

The coastal aquifers are raised reefs patchily deposited  along
the north coast  of  the  island.   These reefs form  highly karsti-
fied limestone aquifers of high permeability  and  low  groundwater
storage potential.   Thickness is  usually less than 50  meters.

The coastal aquicludes  are soft marls fringing the coast  of the
island.  It dams  groundwater flow which occurs along  faults
through the limestone  aquiclude such that  springs issue along
the limestone/coastal  aquicludes  boundary.  Thickness  approaches
300 me t e r s.
                               392

-------
   TABLE 2
             RED KUD INSOLUBLE SOLIDS
                                      TABLE3	    RED MUD SOLUBLE SOLIDS
CONSTITUENT
L10

SlOj
AI2°3
F'2°3
P2°5
CiO
N,20
Tt°2
MnOj
Miscellaneous
PERCENT

11.0
5. 5
12.0
49. 5
2.0
8.0
3 . 5
5.0
1 .0
1 .5
                                   CONSTITUENT
                                                            -AMOUNT
                                   A12°3
                                   NaOH
                                   Nacl
                                   So ec . Gravity
                                   BOD

                                   COD
2.5g/kg liquid

3.'g/kg

1 -6g/kg



0. 7e/ko

0.1c/kc

1 .008

10.5 units

  6 pptn

 1<*8 ppm
                       ANALYSES Of HID HUD LIQUID riACTIOH AND
                         •tCOKHENDCD COKCEHTR AT10H LIMITS

Cenit ltu«nt
C.Utu.
Hl|i«llu.
Sodium
tot • I • 1 ua
Cirbonctf
Sulch.t.
Chlor ld<
Ktrdncil •• C«CO.
AU.Ilnll, .1 C.COj
°H
Turbidity
Colour
a s.,,u.

•* V«t«T Quality Crlt«rlA
Fro* W.d50
10600 " 500
- 500 1300 " 700 2100 •• 100
12.6- 69.] 6.59 3.39.1
unit* 30 unit!
3 30" 1200 until
601

by Hek.« 
-------
Its potential  for  contamination  of  groundwater lies  in  the
concentration  of  sodium and hydroxide  ions; the presence  of
iron oxide;  and  the  organic substances  which on decomposition
impart an unpleasant smell to the waste.   Tables 2 and  3  show
respectively  the  chemical analyses  of  insoluble and  soluble
solids of Jamaica  red mud.


•For  domestic, industrial  and  agricultural water,  colour,  taste,-
 smell,  high soda content  and  high  pH makes  the "red  mud"  a
 potential  agent for degrading groundwater quality.   Table 4
 shows  an  analysis of the  red  mud liquid fraction  plus  the
 recommended concentration  limits.
 Contamination Criteria

 Analyses  to detect above average  concentrations of the  chemical
 constituents present plus esthetic  indices such as colour,
 taste  and smell determine the  degree  of contamination.

 Five  indices were specifically  used  to detect contamination.

 (1)   Sodium to chloride concentration ratio exceeding the
      maximum  ratio encountered in  uncontaminated groundwater
      in  Jamaica of 1.5
 (2)   High sodium content.  This alone is not a precise
      indicator as sodium chloride waters are found in the
      limestone aquifers as a result  of saline intrusion.
      However,  in this form  of  contamination high sodium
      concentrations are associated  with high chloride concen-
      trations, not the case with caustic contamination.

 (3)   Sodium to Calcium concentration  ratio in excess of the
      ratios generally encountered in  uncontaminated ground-
      wat e r  of  1.0.

 (4)   High pH values in excess  of  8.0  units, the maximum
      encountered in uncontaminated  groundwater.

 (5)   The  presence of suspended  solids,  red discoloration,
      poor smell and unpleasant  taste.
 Disposal  Methods

 The  two  plants,  around which contamination has been detected,
 have  similar disposal methods.  The  red  mud waste is piped
 to  and ponded in a natural karstic depression from where the
 bauxite  had been previously mined.   The  two sites are located
 atop  highly karstified limestone with  numerous solution
 openings.   No special precautions were  taken to prevent or
 reduce infiltration before storage of  waste began.  Infiltra-
 tion  was  seen  as one way to extend  the  life of the disposal
 pond.  No  proper geological, hydrogeologica1 and engineering
 studies  were done before the selection  of  disposal site was
 made .
                               394

-------
        Figure 3:  Diagrammatic Section through a Red Mud Pond1
           OUTFAU.   UUO — DELTA  EVAPORATION  PRECIPITATION  RUNOFF
                              i^CAUSTC  LIQUID v^V;:;^^/;
                  FRACTURED LIMESTONE
Movement of Contaminant into Aquifer

A diagramatic section through  a  red mud pond  is  shown as
Figure 3.  The pond  is  formed  by  a  sloping, elongated, steep
sided karst depression  with a  narrow outlet at  the downstream
end.   Storage is  improved by the  construction  of  a rock fill
dam across the outlet with increases in dam height as necessary.
Limestone is exposed  along the surface and at  the  base of  the
pond.  The depression  is naturally drained by  sinkholes or
other solution openings.   Lateral  and vertical  percolation takes
place through the  network of fractures and solution openings
typical of the karstified limestone.  Leakage  also occurs  through
the rockfill dams.  With  time, the  accumulation  of solid material
will  form a seal  over the limestone, reducing  the  percolation
rate.  Calculations  indicate that  only 30 - 40%  of the waste is
retained in the pond.

The formation of  vertices, rapid  fall of mud  level and overnight
disappearance of  mud  and  water,  indicate rapid  and high infiltra-
tion  losses.  The  formation of mud  deltas at  the  outfall pushes
the liquid faction towards the rockfill dam where  the capacity
for leakage is high.   Evaporation  from the pond  surface concen-
trates the pollutants in  the waste.   The ponds  act as collectors
of rainfall runoff, mixing of  this  water with  the  waste and then
infiltrating into  the limestone.
                                 395

-------
Contaminated Areas

Of  the  four  sites shown on  figure  1,  three have  been constructed
atop  karstic limestone and  one  atop alluvium.  Of  the three
sites  atop  karstic limestone,  contamination has  been detected
at  two  sites - site 1 and 4.   Site 1  is associated  with the
Ailcan Jamaica  Company (ALJAM) Ewarton  Plant while site  4  is
associated with  Alumina Partners (ALPART)  Nain Plant.
Site 1 - ALJAM  Mt.  Rosser Mud Pond

This site has the  largest mud disposal  pond in the island with
13.17 Mm  of waste  in storage up  to  September 1991, and a
surface area of 40  hectares.  The pond  occupies a large elonga-
ted karst depression approximately 6.5  km north of the plant
(figure 4).  This  pond known as Mt.  Rosser or Schwa 1 lenburgh,
had no special  preparation before mud  storage began. At the
lower eastern end  a rockfill dam  of  limestone blocks was
constructed.  Massive material losses  were recorded during the
early disposal  period.  However,  it  was  felt that the impact on
the surrounding environment would not  be  significant.  There was
also the expectation that with increasing submergence of the
pond floor, the mud would form a  sufficiently good seal against
further effluent loss.  This has  not been the case and effluent
losses have continued unabated.   At  first, attempts were made
to locate and seal  all the sinkholes and  fractures with clay to
stop the leakages.   Attempts have have  also been made to create
beaches of  tailings around the face  of  the dam to keep the
liquid faction  from seeping through  but  this was also unsuccess-
ful in reducing effluent loss.

The Mt . Rosser  Pond is located on a  faulted anticline, the
northern limb of which continues  into  the Moneague Sub-basin
(a  syncline) and  the northern limb  continues into the Linstead
Sub-basin (also a  syncline).  The latter  merges with a faulted
anticline at Bog Walk, where impervious  strata are brought to
the surface and forms a barrier to groundwater flow.

The Moneague syncline forms a large  depression into which ground-
water discharges through springs.  This  depression has no surface
water outlet and the spring discharge  enters the aquifer via
sinkholes and flows northwards to the  headwaters of the rivers
on the north coast  of the island.  Caustic contamination of
groundwater has been detected at  a number of springs and wells
to the north and south of Mt. Rosser Pond.  To the north the
Rio Hoe Spring  and  the Walkers Wood  well  show evidence of caustic
contamination.

The Rio ^Hoe ".Spring  rises at 'the' sou-thern  edge of the- Moneague
Sub-basin,  flows for 2km and enters  the  Moneague Lake.  Water
from the Moneague  Lake enters the aquifer via the Walton
Sink and flows  to  the Walkers Wood well.    Figure 5 shows the
                               396

-------
                                        plot  of sodium  concentration
                                        for  the period  1973-1991 .for
                                        Rio  Hoe.  The contamination
                                        was  first detected  in  1971
                                        and  rose progressively to
                                        peak  at over 350  mg/1  sodium
                                        in  1983, and the  spring which
                                        was  a source of domestic
                                        water has been  declared unfit
                                        for  human consumption.   The
                                        Moneague Lake fed by  the Rio
                                        Hoe  Spring is also  contamina-
                                        ted  with maximum  sodium
                                        concentration of  65 mg/1
                                        (Mar. 1990).

                                        The  contamination in  the
                                        Walkers Wood well was  detec-
                                        ted  in 1977.  The well  was
                                        drilled in 1976 and the
                                        sodium concentration  during
                                        the  pump test was 5mg/l.
                                        The  contamination has
                                        increased and monthly  monito-
                                        ring, which began in  1990,
                                        shows a worsening situation,
                                        (figure 6).

                                        The  elevation of  the  surface
                                        of  the red mud  waste  is at
                                        an  ;altitude  of 466.3  metres
                                        while the spring  rises  at
                                        305  meters and  there  seems
                                        t o .i>e continuity between the
                                        pond  and the spring by  a
                                        system of conduits  above
                                        the  regional water  table.
                                        The  spring is above the
                                        regional water  table  and
                                        during high rainfall  the
                                        discharge becomes brown and
                                        turbid, while the water table
                                        does   not exhibit this  change.

The area is now being  assessed and  so  far  5 monitoring  wells  vary-
ing in depth from  200-350 metres have  been  completed.   To date,
only 3 of the monitoring wells have  tapped  contaminated ground-
water.  The probable area of contamination  is shown on  figure 4.
To the south the Weatherly Spring and  Alcan's Deepwell  2  have
shown evidence of  caustic contamination.   The plots of  sodium
concentration for  Weatherly Spring  and  Deepwell 2 for 1973-1991
are shown as figures 7  and 8 respectively.
                              397

-------
       Figure  5 :  SODIUM   CONCENTRATION   RIO   HOE  SPRING   1973 - 1991
I
                                                     I9B2   1963   196*   1985   1985
        1973   197+  1975  1976  1977  1978  1979
                                                                               1937  1963  1969   1990  1931
               Figure  6'  SODIUM   CONCENTRATION   WALKER'S  WOOD  WELL  1976-1991
                1976  1977  197B  1979  1980  1931  1982  1933  19B<
                                                     19E5   1966  1987  1983  1939  1990   1991
 In  1989/90  the  Underground  Water  Authority  and ALJAM  collaborated
 on  a project  to  assess  the  level  of  contamination.    Six  monitoring
 wells  were  drilled  and  average  sodium  concentration  varied  from
 9-30 mg/1.    The  results  of  the  project  indicate:

 (i)  the  area  of  contamination  has  been  better  defined
       and  has  been  found  to  be  smaller  than  originally
       env i s ioned.

 (2)  two  zones  of  contaminated  groundwater  separated  by
       clean  water  were  intercepted  and  confirmed  the more
       significant  role  that  compartmentalised flow  plays  in
       the  movement  of  the  pollutant  than  diffuse  flow.
  (3) The  upper  contaminated  zone  has  been  interpreted  as  being
       the  direct   result  of  the  plant,  its  attendant  dumps  and
       other  adjacent  effluent  generating  facilities.
                                                 398

-------
    Figure  7  :  SODIUM  CONCENTRATION  WEATHERLY   SPRING   1973 - 1991
3000

2700

2«OC

21OQ

1800

1500.

1200.

 900.

 600

 300.
                                  1980  1961  1982  1983  1964  1985  1986
    1973  1974  1975  1976  1977  1978  1979
                                                                1967  1988  1969  1990  1991
    Figure 8 : SODIUM  CONCENTRATION   DEEPWELL
1973 - 1991
 10QQO

 9000

 BOOO

 7000

 6000

 5000

 «00

 33a)
 2000

 1003
    1973  1974  1975  1976  1977  1978  1979
                                 19BD  19B1   1932  1933  1934  1935  1935
                                               Monfe
                                                               1937  I SEE  1933  1993  1991
   To  the  north  and  south  the  total area  of  the  aquifers  contaminated
   is  approximately  18  km^  (figure  4).


   Site  4  Alpart -Nain  Mud  Pond

   There.are two  (2)  disposal ponds  at  this  site,  a  south mud  pond  and
   a north  clear effluent  pond  (figure  9)  .   The ponds are  located  on
   a thick  sequence  (>1500m)  of faulted  and  fractured brecciated
   limestone.   The presence  of  numerous sinkholes  and  the  absence of
   surface  drainage  indicate the potential  for  rapid  infiltration.
   Thesites  are  located close  to the groundwater divide which separates
   northerly  flow to  the Upper  Morass  from  southerly  flow  to  the  sea.
   The surface area  of  the  ponds are 90 hectares and  the  volume  of
   mud in  storage is  estimated  at  12.8  Mm3.    Red  mud  disposal was
   being  restricted  to  the  south lake  but  caustic  enriched  clear
   effluent  is being  disposed  of in the north  lake.    The  formation  of
   tailings  "deltas"  were being  restricted  to  the  northern  and eastern
   perimeters  of the  south  lake  with the  resultant  ponding  of superna-
   tant  liquor directly against  the limestone  along  the southern  and
                                         399

-------
flgur. «i  AREA CONTAMINATED
      BY CAUSTIC WASTE FROM
      ALPART ALUHIHA
      PLANT
western  lake perimeter.

The plant  began  operations  in  1969  and
five wells were  drilled in  the  vicinity
of the plant to  supply industrial  and
domestic  water.  As  early as March  1970
high sodium concentrations  were  noted
in the well water.   This was followed
by a decrease in well yields and the
drilling  of a replacement well  (#6)  to
satisfy  demands.   The sodium concentra-
tions  increased  until all the  wells
around the plant are now contaminated.
Domestic  water now comes from  a  new well
     field established 6.4  km  north  of
     the  plant at  Pepper.   The  Pepper
     well  field began operations  in
     1974  and has  since been showing a
     trend  to increasing  sodium  concen-
     tration (figures  10,  11, 12, 13).

     To the south  a  1990/91  investiga-
     tion  has indicated a low level
     of contamination  of  domestic wells
     at  New  Forest.   This contamination
     is masked  by  saline  intrusion into
     the  aquifer but  once the molar
     ratio  between  sodium and chloride
     have  been  reconciled, excess
     sodium remains  in  the groundwater
     and  its source  can only be  the mud
     pond.   The area  of the  aquifer
     contaminated  is  approximately 20 km2
     and  the domestic  wells  that  supply
     the  large  urban centre  of Mandeville
     are  now threatened.
                                  Reduction  of Contamination
                                  The Underground Water  Authority has
                                  been working with  both Alpart and
                                  ALJAM  to  reduce groundwater conta-
                                  mination.
system called "mud stacking
consists  of  thickening  the
thinly on  a  sloping,  sealed
sun.  All  run-off is  collec
to a sealed  holding pond,  f
Once the  mud has dried  on  t
can be cleaned and returned
Company  to dispose of  less
in the Mt.  Rosser red mud  p
                                  At ALJAM  a new mud
                             and drying"  has been  imp
                            red  mud to 22%  solids,  sp
                             drying bed,  where it  is
                            ted  at the toe  of the  bed
                            rom  where it  is recycled
                            he  bed (it does not reslu
                             to  use. This has allowed
                            than 5% of the  annual  was
                            ond, and only in the very
                          disposal
                          1emented.  This
                          raying  it
                          dried  by  the
                           and  transferred
                          into  the  plant.
                          rry)  the  bed
                           the  Bauxite
                          te  generated
                           high rainfall
                                 400

-------
     Figure   10  :   SODIUM    CONCENTRATION    ALPART    I    WELL    1975 -  1991
1000
 900
 eco
 TOO

 sen

 ra
 200
 ICO
     1975
            1976
                  1977
                          1978
                                 1979
                                        1330
                                               199
                                                      1EEB
                                                             19B3
                                                                     196*
                                                                            19B5
                                                                                   1S6B
                                                                                          1987
                                                                                                  19BB
                                                                                                         1969
                                                                                                                1990
                                                                                                                       1991
       Figure   II:   SODIUM    CONCENTRATION     PEPPER   I   WELL    1975-1991
    9Q
    «
    40
    s
    a
    z
    XI
    IS
        Illlllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll
       1975   1976   1977    1978    1979    I960    199
                                                        1S62
                                                               1S83
                                                                      1934-
                                                                             1985
                                                                                     1985
                                                                                            1987
                                                                                                   1983
                                                                                                                 1980
                                                                                                                        1931
              Figure 12 :  ANNUAL  MAXIMUM  Na/CI
                           RATIO  ALPART I  1975-1991
          2000
          1500
          1000
           500
           003
                 i  f   I  I   I  I   I  I  I  I  I  I   I  I  I  I  I
               1975  1977  1979' 193  19S5 19E6  1967  1989  1991
                  1976  1978  I960 1932  19B4  1SBB  19B3 1SGO
                                   Years
        Figure 13:  ANNUAL  MAXIMUM  Na/CI
                     RATIO  PEPPER  I  1975-1991
I
                                                                        250
                                                                        200
                                                                        150
                                                                        oso
     oco
           I  l   l  l  1   l  l   l  1  r i  I   r r  i  I   I
         1975  1977  1979  1981  1EG3 1SB5  19B7  1S6B  1991
            1976  1978 'Sffl  1962  19B*  19B6  1SGB  1980
                             ran
                                                        401

-------
periods when  the  drying cycle becomes  longer.   Alpart is now
investigating  the  suitability of this  method  of disposal for
implementation  at  its  plant site.
Summary and Conclusions

The red mud wastes  produced from .the bauxite  alumina operations
contain sufficiently  high concentrations  of  pollutants to
contaminate the  groundwater,  making it unusable.   The method
of disposal of the  waste utilizing natural  or  mined out
depressions in the  limestone  creates a potential  hazard in view
of the high degree  of karstification of the  aquifer.
Contamination of  the  groundwater in the vicinity  of.:'the two
sites have been  proven.  The  solutions to reducing the contami-
nation lie in the  sealing of  the disposal pond and possible
total impoundment  of  the effluent.  Pond  management needs to
be improved to minimize the possibilities and  effects of further
contaminat ion.
Acknowledgements

The Author wishes  to  thank the following persons/agencies who
assisted in the preparation of this paper.
                             (UWA)
                             (ALJAM)
                             (ALPART)
                             ( JBI)
Mr- Dexter Lewis  who  assisted   with the preparation of figures
Miss Michelle Wint  who  compiled data   and
Mrs. Nerma Wynter who typed the paper-
Underground Water Authority
Alcan Jamaica Company
Alumina Partners Limited
Jamaica Bauxite  Institute
References
1 .
2.
3 .
                                                          : ion
    ALJAM/UWA -  Preliminary report on Groundwater  Pollute..
    by caustic effluent  from the Ewarton Plant  and associated
    facilities -  Dec.  1989.  A joint project  report.

    FERNANDEZ, BASIL:   Caustic waste pollution  of  groundwater
    in Jamaica.   Two  solutions - May 1983.  An  unpublished
    Underground  Water  Authority Report.

    Underground  Water  Authority:  Water Resources  Development
    Master Plan  Report 1  - Water Resources  Inventory  -
    Dec. 1985.   An  unpublished Underground  Water  Authority
    Report .
                               402

-------
Question

Are there any springs near the plants that are used
as water supplies?  If so, are they hydraulically
connected to the plant?
Answer

Yes,  there are 4 springs located near to the plant that
can be used for water supplies.  One spring is located
to the north,  and 3 are located to the south of the
pi ant (see fi gure 4).

Of the 4 springs, 3 are hydraulically linked to the mud
disposal pond, and show signs of contamination.

These are:

(a) North - Rio Hoe Spring
(b) South - Weatherly Spring
            Cashew Tree Spring

The fourth spring is  not hydraulically linked, and is
not contaminated.  It is used as a source of domestic
water for a small village in the hills above the plant.
                         403

-------
404

-------
              The Interaction of Flow Mechanics and Aqueous Chemistry

                           in a Texas Hill Country Grotto

                         Barbara Mahler and Phillip Bennett

           Department of Geological Sciences, University of Texas at Austin
                                  Austin, Texas
Abstract
   The physical characteristics of flow, or flow mechanics, through a carbonate terrane
affect the chemical signatures of groundwater inputs to Hamilton Pool, a grotto in the
Central Texas Hill Country portion of the Trinity Aquifer. Because flow mechanics
influence geochemical processes, their identification is necessary to protect water quality at
this unique site, which is used for both recreation and habitat preservation.
   During baseflow, two groundwater sources supply all the water in the pool:  1)  small
perennial springs half a kilometer upstream from the pool are the headwaters for Hamilton
Creek,  which enters the pool via a 25-meter waterfall (0.1-0.3 ftVs), and 2) stalactites
hanging from the ceiling of a large overhanging limestone ledge drip water directly into
Hamilton Pool (0.05 ft-Vs). The two sources exhibit differences in pH, calcium
concentration, Ca/Mg ratio, and temporal variation in chemistry. The chemistry of the drip
water is affected by degasing of carbon dioxide and equilibration with ambient temperature
upon contact with the atmosphere. Backmodeling of these environmental processes with
PHREEQE resulted in a subsurface chemistry of the drip water very similar to that of the
spring water. Analysis of conductivity variation in response to precipitation revealed that
the chemistry of the upstream springs is controlled by conduit flow, while the ceiling water
is controlled by diffuse flow.
   These results suggest that a) both groundwater inputs to Hamilton Pool are from the
same recharge source, and b) the conduit-controlled upper springs render Hamilton Pool
vulnerable to contamination should the catchment area be urbanized.
                                        405

-------
Introduction

   Carbonate aquifers are distinguished from other groundwater media by their physical
flow characteristics. Flow rates at different points within a single carbonate aquifer can
vary many orders of magnitude, from centimeters per day to kilometers per day. The
mechanics controlling flow rate influence both the chemistry of the water and the
susceptibility of the water to contamination. This paper highlights the findings of a more
extensive study of geochemical response to flow mechanism in a Texas Hill Country grotto
(Mahler, 1991).
   Diffuse flow and conduit flow represent opposite extremes of carbonate flow.  Diffuse
flow is similar to flow through a granular, porous media;  water flows through extremely
tiny joints, partings, bedding planes, and fractures, the diameters of which are on the scale
of millimeters or less. Conduit flow, in contrast, involves flow through integrated,
solutionally-developed conduit systems with diameters on the scale of centimeters to meters
(e.g., Shuster and White, 1971).
   The risk of contamination of conduit-controlled water sources is high.  Conduits
behave as pipes rather than as filters; water moves through them with high velocities and
short residence  times, and without undergoing dilution. The processes that normally
cleanse groundwater (e.g. filtration by soils, adsorption of contaminants by mineral grains,
dilution and dispersion through large volumes of water, and decay and die-off during long
residence times) often do not affect conduit flow.
   Springs controlled by the two different flow mechanisms can be distinguished by their
chemistry. Shuster and White (1971) found that diffuse flow springs showed much less
variation in hardness than conduit flow springs. Similarly, Jacobson and Langmuir (1974)
found that diffuse flow springs showed less variation in solute amounts than conduit flow
springs. They also found that after a storm, conductance remained the same or increased
slightly for diffuse flow  springs due to salts and solutes swept in from soil, while
conductance decreased in conduit flow springs due to dilution with low conductivity
meteoric water. They therefore suggested using specific conductance as an indicator
variable of subsurface flow  mechanism.
   The objective of the investigation presented here is to characterize the aqueous
geochemistry of inflows to Hamilton Pool,  Travis County, Texas, to determine the flow
mechanism controlling each source, and to investigate each inflow's genetic relationship to
the other baseflow sources.
   The Study Site. Hamilton Pool is a pristine waterfall-fed grotto in rural western
Travis County, Texas, approximately 20 miles southwest  of the city of Austin. The pool is
the focal point of Hamilton Pool Preserve, a county-operated facility. Water quality at the
pool is utmost important both for recreational use and species habitat.  Although current
                                        406

-------
land-use in the Hamilton Creek watershed is primarily agricultural, commercial and
residential development are rapidly expanding westward from the Austin area.
   The Hamilton Creek watershed is located in the rocks of the Lower Cretaceous Trinity
Division. Five outcropping formations of the Trinity Division can be viewed at Hamilton
Pool Preserve. Sycamore Sand outcrops in the bed of Hamilton Creek from the Pedernales
River to approximately one kilometer upstream, where it slopes into Hammett Shale.  The
shale forms the eroded walls of the cliff surrounding the east side of Hamilton Pool, and
Cow Creek Limestone forms the overhanging ledge. The rolling grasslands surrounding
Hamilton Pool are on Hensel Sand, which outcrops along the creek bed and smaller
drainages. Glen Rose Limestone outcrops in the hills east of the pool, and is the most
areally extensive formation cropping out within the watershed (Figures  1 and 2).
   There are two principal inflows to Hamilton Pool.  The main source is Hamilton Creek,
which empties into the pool via a 25-meter waterfall. During baseflow, the sole source of
this water is a group of small springs a few hundred meters upstream from the waterfall. A
second inflow consists of groundwater dripping from a large travertine formation ("Drippy
Rock") and stalactites which line the ceiling of a cave-like structure which overhangs the
pool. Both these inflows issue from the Cow Creek Limestone. During baseflow, Drippy
Rock and stalactite flow total approximately 0.035 ft3/s, while flow from the upper springs
varies from 0.1 to 0.5 ft^/s.
Approach

    The aqueous chemistry within the Hamilton Pool Preserve was examined to determine
the water source, flow mechanism, and subsurface processes. Water samples were
collected weekly for six months from the four springs comprising the Upper Springs site,
from Drippy Rock, and from Hamilton Pool itself. Field measurements of pH,
temperature, alkalinity, and conductivity were recorded, and analyses of major anions and
cations performed in the laboratory. The chemistries of the four upper springs, ah1 located
within three meters of one another, were compared to see if one spring could be chosen as
representative of all four. Because all four sites displayed similar solute concentrations and
temporal variation, it was decided to use analyses of East Spring as indicative of the
chemistries of all the upper springs.
     Mineralogy of solid phase samples was analyzed to estimate the contribution from
each formation to aqueous chemistry.  Fresh samples of the Glen Rose Limestone, the
Hensell Sand, the Cow Creek Limestone, and the Hammett Shale were collected in the field
and analyzed for major cations.
                                       407

-------
                                     •XV:':VJ Glen Rose Limestone
                                          Hensell Sand
                                          Cow Creek Limestone
                                     q^HHil Hammett Shale
                                          Sycamore Sand
                        50 200     600ft
Figure 1.  Plan view of Hamilton Pool  Preserve.
                   408

-------
^i Glen Rose
):£j Limestone
   Hensell
   Sand
    Cow Creek
    Limestone
    Hammett
m Shale
Sycamore
Sand
                                      Hamilton
                                      Pool
                                                             900'
                                                             850'
                                                   S5^saaasHE; ,800'
                                                             750'
                                                            L700'
                                                             600'
                            VERTICAL EXAGGERATION X 10
               500ft.
           Figure 2. Cross Section of through Hamilton Pool Preserve.

-------
Results

   Environmental processes and flow mechanics cause differences in aqueous chemistry at
Drippy Rock and East Spring.  The formation of travertine where Drippy Rock water
issued from the subsurface indicated that the drip water became supersaturated with respect
to aragonite upon contact with the atmosphere. Additional carbon dioxide may have been
lost to the atmosphere from the collection vessel in the amount of time that it took to
accumulate enough drip water for a sample, as suggested by White (1988).  Saturation
indices for aragonite, calculated with WATEQF (Plummer et al., 1976), an equilibrium
speciation program, showed that Drippy Rock water was indeed supersaturated with
respect to aragonite, while East Spring was slightly undersaturated. Analyses of Drippy
Rock revealed that Drippy rock was depleted in calcium ion and had increased pH relative
to East Spring, as expected. In order to compare the subsurface chemistries of the two
sources, PHREEQE (Parkhurst et al., 1980) was used to back-model the environmental
processes affecting the drip water.  Loss of carbon dioxide was reversed by reacting
Drippy Rock water with a fixed pCO2 equal to that of baseflow at East Spring (-1.70), the
effects of ambient temperature were reversed by holding temperature fixed at the baseflow
temperature of East Spring (23°C), and the water was held in equilibrium with calcite.
Figure 3 compares the chemical analyses for Drippy Rock and East Spring with the results
of the simulated subsurface Drippy Rock water. Modeling Drippy Rock water in the
subsurface eliminated most of the difference in pH and  calcium concentration for the two
waters, but had no effect on magnesium concentrations.
   Drippy Rock and East Spring had contrasting responses to rainfall. Conductivity and
rainfall for a three-month period are shown in Figure 4. Conductivity at East Spring drops
after a major rain, while the conductivity at Drippy Rock remains the same or increases
slightly.
   In general, the back-modeled Drippy Rock water and East Spring water had similar
overall chemistries. Both were both calcium/magnesium bicarbonate waters. They
contained similar levels of magnesium, chlorine, sulfate,  and silica, and had very low
levels of iron, phosphate, boron, and potassium. These similarities suggest that the source
of these waters was similar. To investigate possible rock-water interactions, solid samples
of Glen Rose Limestone, Hensell Sand, Cow  Creek Limestone, and Hammett Shale were
analyzed for major cations. The results are displayed in Table 1, which shows that the
Cow Creek Limestone did not have any detectable magnesium, in contrast to the other
formations, which contained magnesium levels similar within an order of magnitude. All
four formations had similar proportions of calcium and strontium. These results indicate
that 1) the Cow Creek Limestone has been entirely recrystallized into a pure calcite
limestone, and 2) that the magnesium in water samples issuing from the Cow Creek
                                       410

-------
                                    PH
        |
                      Drippy Rock

                      East Spring

                      Simulated Dttppy Rock
                            Calcium Concentration
                           Magnesium Concentration
Figure 3. Temporal variation in pH, calcium, and magnesium versus rainfall.
Simulated water is fixed at pCO2 = -1.70, T = 23.0° C, and is in equilibrium
with calcite.

-------

                       Drippy Rock
             L
                            \

                           §
                                              500
                                             U400
                                                   E




                                              300  f


                                                   t>


                                                   1
                                                   o
                                                   o
                                             -200
                                              100
                                    S   a
                                               500
     I  2



     1
     a:
                                               100
                    3)   TO
Figure 4. Conductivity versus rainfall at Drippy Rock and East Spring.
                            412

-------
         Table 1.  Cation analyses for solid phase samples.
   Hammett Shale 1
   Hammett Shale 2
   Hammett Shale 3
   Hammett Shale 4
   Cow Creek Limestone 1
   Cow Creek Limestone 2
   Glen Rose Limestone
   Hensell  Sand
                                Ca
Mg
Sr
0.324
0.442
0.284
0.476
0.443
0.536
0.393
0.245
0.043
0.008
0.012
0.003
0.000
0.000
0.028
0.075
0.00038
0.00033
0.00051
0.00056
0.00063
0.00028
0.00047
0.00019
   Units given in gram constituent per gram sample.
Limestone must be derived from other formations, most likely the Glen Rose Limestone,
which covers the majority of the catchment area. These results suggest that water at both
East Spring and Drippy Rock originate from recharge water falling on the Glen Rose
Limestone.
Discussion

   The results of the chemical analyses suggest that water at Drippy Rock is chemically
similar to water from the upper springs, but that each is controlled by contrasting flow
mechanics.
   The chemical behavior of East Spring suggests that it is controlled by conduit flow.
The decrease in conductivity and calcium concentrations after rainfall at this site indicates
dilution of baseflow by low-conductance meteoric water. Once baseflow water has been
diluted by rainfall in a conduit system, it moves swiftly through the aquifer and emerges
from a spring before it has reached equilibrium with the surrounding rock matrix
(Jacobson and Langmuir, 1974).  An interesting physical manifestation of conduit flow at
this site was the build-up of small mounds of sand and gravel at the mouths of the upper
springs after a particularly heavy rain; the sediment had evidently moved through the
conduits as bedload.  The conduit nature of flow at this site greatly increases its
susceptibility to contamination, as water which enters the conduits undergoes minimal
filtration and adsorption of contaminants.
                                      413

-------
   In contrast, diffuse flow mechanics control the water chemistry at Drippy Rock, as
evidenced by the lack of variation in conductivity. The slow rate of diffuse flow allows
water to approach equilibrium with the rock matrix, and water at this site remains
unaffected by dilution from rainfall. In diffuse flow systems, rainfall may also flush soil
water with high carbon dioxide concentrations into the rock matrix. Such an increase in
carbon dioxide could cause the increase in dissolved calcium concentrations observed at
Drippy Rock after rainfall.
Conclusion

   The two groundwater inflows to Hamilton Pool are genetically very similar. The
contrasting chemistries of these inflows to Hamilton Pool are caused by physical
differences in flow as the water comes into contact with the atmosphere.  Drip water from
Drippy Rock undergoes much more degasing upon atmospheric contact than spring water
at East Spring; it also responds more quickly to ambient temperature.  As a result, Drippy
Rock water undergoes an increase in pH accompanied by a reduction in calcium and
bicarbonate concentrations as it emerges from the subsurface. After removing the effect of
these processes through geochemical modeling, the simulated Drippy Rock subsurface
water and the East Spring water both appear to originate as recharge through the overlying
Glen Rose Limestone.
   Contrasting response to rainfall indicates that the two sources are controlled by
contrasting flow mechanics. East Spring is controlled by conduit flow. As a result, after
rainfall East Spring water shows an increase in conductivity and a decrease in
concentrations of several ionic constituents including calcium.  Because conduit-controlled
springs are susceptible to contamination, Hamilton Pool is vulnerable to water quality
degradation from this source. In contrast, the second source (Drippy Rock) is controlled
by diffuse flow. After rainfall, water at Drippy Rock undergoes a slight increase in
conductivity and an increase in calcium concentration due to flushing of soil carbon dioxide
into the water.

Acknowledgements
   Very special thanks  go to Travis County Parks and Recreation staff at Hamilton Pool
Preserve, particularly Terri Siegenthaler and Dan Chapman.  Additional thanks to Roger
Lee, Raymond Slade, and Colleen Stapleton.
                                       414

-------
References

Jacobson, Roger L. and Donald Langmuir, 1974, Controls on the quality variations of
       some carbonate spring waters, /. of Hydrology, 23, 247-265.

Mahler, Barbara J., 1991, A Geormorphic, Hydrologic, and Geochemical Study of
       the Hamilton Creek Watershed, Travis County, Texas, Masters Thesis,
       University of Texas at Austin.

Parkhurst, D.L., D.C. Thorstenson and L.N. Plummer, 1980, PHREEQE  A
       Computer Program for Geochemical Calculations, U.S. Geol. Surv. Water-
       Resour. Invest. 80-96, NTIS Tech. Report PB 81-167801, Springfield, VA.

Plummer, L. N.,  B.F. Jones, and A.H. Truesdall, 1976, WATEQF - a FORTRAN IV
       version of WATEQ, a computer program for calculating chemical equilibrium in
       natural waters, U.S. Geol. Survey Water Resour. Invest. 76-13,, 61 pp.

Shuster, Evan T. and William B. White, 1971, Seasonal Fluctuations in the chemistry
       of limestone springs: a possible means for characterizing carbonate aquifers,
       Journal of Hydrology, 14, 93-128.

White, William B., 1988, Geomorphology and Hydrology ofKarst terraines, Oxford
       University Press, N.Y.
Biographical Sketch

Barbara Mahler is a Master's Candidate in the Department of Geological Sciences at the
University of Texas at Austin. She received her Bachelor's Degree from Boston
University, and will begin working toward a Ph.D. in hydrogeology at U.T. in the
spring of 1991. Dr. Philip Bennett is an Assistant Professor in the Department of
Geological Sciences at the University of Texas at Austin.
                                   415

-------
The Interaction of Row Mechanics and Aqueous chemistry in a Texas Hill Country Grotto.

By Barbara Mahler and Philip Bennett

Q: Do either or both of the chemistries of the diffuse and conduit-dominated flow regimes vary
seasonally?

A: Water samples were collected from June through October, 1990, a sampling period insufficient
to verify existence or lack of seasonal variation in chemistry.  During this period no seasonal
variation was detected in the conduit-controlled regime. In contrast, the samples collected at the
diffuse-controlled site had calcium concentrations which were inversely related to ambient
temperature, due to rapid equilibration of the drip water with surface conditions.  However, we
assume that within the subsurface the diffuse-controlled regime would display the same sensitivity
to seasonal variation as the conduit-controlled regime.
       The sampling period was necessarily limited due to  the scope of the project (field work for
a M. A. thesis). However,  regular sampling over a two year period may reveal differences in
seasonal variation between the two regimes, which would provide more information about the
relation between the two water sources.
                                           416

-------
THE ORONOCO LANDFILL DYE TRACE III:  RESULTS FROM A SUPERFUND

 REMEDIAL  INVESTIGATION  IN A  GLACIATED,  DIFFUSE-FLOW  KARST.


       E.  Calvin Alexander,  Jr. and Scott C. Alexander

            Department of Geology  and Geophysics
       University of Minnesota, Minneapolis, MN 55455


                     Barbara  J. Huberty

            Olmsted  County Public Works Department.
                     Rochester, MN 55904


                      James F. Quinlan

       Quinlan  & Associates,  Inc., Nashville,  TN  37222
                          ABSTRACT

The Oronoco  Landfill in  Olmsted County, Minn.,  a Superfund
site 10 miles north  of  Rochester,  is built on the Ordovician
Prairie  du  Chien  Formation,  and  important regional  karst
aquifer.   As part  of the site's R.I.,  a dye trace  was con-
ducted to determine:   1) the local directions(s) and speed of
ground water flow, 2) if  an existing monitoring well network
effectively  intercepted ground water flow  beneath the site,
and 3)  which local  private  wells were most likely to be af-
fected  by  potential  ground water  contamination attributable
to the landfill.

Samples were collected  from 11  on-site  monitoring wells, 12
springs, and 201 private  wells.   Continuous pulses  of dye
were detected  in 6  on-site  monitoring  wells, and 2 private
wells located 2.2 and 2.9 km  northeast  of the landfill.  The
continuous pulses of dye detected in the wells have persisted
for up  to  two  years.  The concentration  of dye  in the wells
periodically  increases  after  major  recharge  events.   This
karst is a diffuse-flow system with ground water flow to the
northeast  at velocities of  the  km/yr.   The monitoring wells
                             417

-------
appear  to  effectively  intercept  ground water  flowing under
the site.   Conventional  quarterly sampling  will detect the
largest, chronic hypothetical  landfill releases but may miss
short-term, acute pulses  of  pollutants.  The continued pres-
ence of  dye in the  system has proven to be  a useful guide
during the installation of additional monitoring wells at the
site.
                         Introduction

The Olmsted County Sanitary Landfill is located approximately
10 miles  north of Rochester  ,  Minnesota  (Figure  1)  and was
purchased by the City of Rochester in 1969.  A permit to con-
struct a  landfill on the  site  was  granted  by the Minnesota
Pollution Control  Agency  (MPCA)  in 1970.   Site development
began in 1972.  The City of Rochester operated the site until
1983, when the City and Olmsted County agreed to transfer the
permit to Olmsted County.

The  landfill  received  waste  from  local refuse  collection
haulers serving  residences throughout  Olmsted  and adjacent
counties.   Local commercial establishments, institutions, and
industries also used the site.  In addition to the usual ref-
use from these generators, the site also received incinerator
ash,  utility  ash,  wastewater   treatment plant  sludge  and
sludge  ash,   miscellaneous  industrial  process sludges,  and
debris resulting from a 1978 flood.  In March 1987, the land-
fill was  closed  to municipal solid  waste,  but continues re-
ceiving demolition, coal ash,  and asbestos waste.

In 1983,  analysis of water samples  from  monitoring wells at
the site  indicated the  presence of VOCs.  As  a result, MPCA
and EPA began Superfund  actions  at the  site  in 1986.  Subse-
quently MPCA and  EPA agreed  to  establish MPCA as  the lead
agency for remedial  action.   The  MPCA  negotiated a Response
Order  by  Consent  with  the   City  of  Rochester and  Olmsted
County which became effective in December 1989.

As part of the site's Remedial Investigation (R.I.), the Olm-
sted County  Landfill Dye Trace  Study was conducted to deter-
mine:   1)  the  local  directions(s)  and  velocity  of ground
water flow, 2) if an existing monitoring well network effect-
ively intercepted  ground water  flow beneath the site, and 3)
which local  private  wells  were  most likely to be affected by
potential  ground  water  contamination  attributable  to  the
landfill.   The Study was  initiated in April,  1989 and con-
tinues at  the present  time (October, 1991).   This report is
based  primarily  on  data  gathered  between  April,  1989  and
September,  1990  and  is  excerpted  from Alexander  et  al.
(1991).   The  Study was  divided into  two phases.   Phase 1
extended from the  initiation of the project through March 31,
1990.   Phase  2 extended from April  1, 1990 to the present.
                             418

-------
                                                            N
Figure 1.   Map of  Olmsted County  Minnesota  with the  town-
     ships,  cities,  major  highways   and  County  Sanitary
     Landfill  near Oronoco shown.
                           419

-------
                  Site  Geology and Hydrology

In the vicinity  of  the site,  glacial overburden generally is
less than 100 feet thick.  Within the site, the overburden is
0 to  60  feet thick and  includes  sands  and gravels, clay and
silt.  Clay  and silt units predominate in the southern part
of  the  site.    Sand  and  gravel underlies the  central and
northern part of the site, including the waste cells.

The first bedrock encountered under the  site is the Prairie
du Chien Group.   It is up to  250  feet  thick and is predomi-
nantly dolomitic  limestone.    It is fractured,  jointed, and
contains numerous karst  features.   Underlying the Prairie du
Chien Group is the Jordan Sandstone (Olson, 1988a, 1988b).

The water table  is  encountered in the  Prairie du Chien Group
(Kanivetsky,   1988).    In the  vicinity  of the  waste  cells,
groundwater  monitoring wells  completed  at the  water  table
have  static  water levels at  depths  of 80  to  115 feet below
ground surface.   Ground  water  flow in  the vicinity is gener-
ally to  the  east  or northeast.   Residential wells in the vi-
cinity draw  water  from  both  the Prairie  du Chien and the
Jordan aquifers.
                      Dye Trace Methods

Groundwater flow velocity and direction in limestone is aqui-
fers  difficult  to  predict  because the  exact nature  of  the
fractures, joints  and  solution cavities in  the  limestone is
unknown.  Groundwater flow in these conditions can be studied
by introducing dye at one or more locations and then monitor-
ing the groundwater  at  many locations to determine the speed
and direction of groundwater and dye movement.

Rhodamine WT  dye  was selected for use  in  this study princi-
pally because of  its ease and economy of  analysis.   Dye was
introduced into  carbonate  bedrock  in a  quarry west  of  the
landfill in May 1989 (Phase 1).  A second introduction of dye
was made in March 1990 in a sinkhole located east of the site
(Phase 2).

A study area  approximately  1.5 miles square was established.
Groundwater samples  were taken daily  from approximately 100
residential wells and  from the  landfill  monitoring wells.
Springs and  sites along  the  Zumbro River  were  sampled less
frequently.

The residential  wells  were distributed  throughout the study
area.    They  included  up-gradient wells,  generally  west and
south of  the  landfill  site, and wells north and east of the
Zumbro  River, a  hydrogeologic boundary.    Residential well
samples were taken by the residents.
                             420

-------
Springs  were sampled  because  they  were  considered to  be
potential discharge  points  for  dye.   The Zumbro  River was
sampled in an effort to  detect  dye which might seep directly
into the river.   These samples were taken by County staff.
                   Results and Conclusions

Rhodamine WT dye  reached  the first monitoring well less than
24 hours after  introduction (see Figure  2).   Dye reaching a
total  of  six  monitoring  wells  in  the following  weeks and
months  documented  a  complex pattern of  groundwater  flow
roughly west  to east under  the  waste  cells.   The continued
presence of dye in the monitoring  wells  documents that they
are adequately  sampling the local  groundwater  flow.   All of
the positive  monitoring  wells  are  in the Prairie  du Chien
aquifer.

Rhodamine WT dye  reached  private well 151 north-northeast of
the landfill 77 days after introduction (see Figure 3).  That
well  is  1.2  miles northeast  of  the  dye  introduction point.
The minimum  speed the  leading  edge  of the dye  traveled to
reach  well  151  is  5.7 miles per  year.   Dye continues  to
emerge from that  well in October 1991.   Dye reached private
well  108, 1.5  miles  northeast of the introduction point,  in
207 days.  Dye also continues to emerge from that well.  Both
wells  are older Prairie du Chien wells.   These two positive
wells  document  a  groundwater flow  path from the  landfill to
the north-northeast in the Prairie du Chien aquifer.

Rhodamine WT  from the  first dye  introduction has  not been
confirmed at any of the other original monitoring wells, res-
idential wells, springs,  or river  stations.   However, trace
levels of Rhodamine  have  appeared  at spring 412  (see Figure
4)-   Rhodamine WT from the first  dye introduction has been
detected in several  of  the  new monitoring wells drilled into
the Phase 1  dye plume.   Rhodamine  WT from the second intro-
duction of dye  was detected during  the installation of a new
monitoring well nest in June 1991.   No other detections from
Phase  2 dye have occurred.

The dye  pulses  in five of  the six  positive monitoring wells
respond rapidly to recharge events  producing  large fluctua-
tions  in the  dye  concentrations.   The monitoring well break-
through curves  illustrate the complex  nature  of  the ground-
water  system  beneath  the  landfill.    The Prairie  du Chien
aquifer under  and surrounding the  landfill  has flow charac-
teristics  that are   intermediate  between  those   of  mature,
conduit-flow karst aquifers and porous-media aquifers.

In addition to  the continuous dye  pulses discussed above, a
low level contamination problem complicated and increased the
cost of the Study-  During the summer  of 1989, dye detections
                             421

-------
 PHASE  I  DYE  MOVEMENT   PHASE I DYE  MOVEMENT
         1 DAY
 PHASE  I DYE  MOVEMENT
         4 DAYS
PHASE  I DYE MOVEMENT
         7  DAYS
        45  DAYS
               LEGEND
          DYE INTRODUCTION LOCATIONS
           I • - PHASE I
          II* - PHASE II
             DYE PLUME
          SAMPLING LOCATIONS
           »  - PRAIRIE DU CHIEN WELLS WITH DYE
           '  - PRAIRIE DU CHIEN WELLS WITH NO DYE
             - JORDAN WELLS WITH NO DYE
     SOLE!
Figure  2.  Phase 1 dye movement during the first 45 days,
                           422

-------
PHASE I DYE MOVEMENT    PHASE  I DYE  MOVEMENT
        4 MONTHS
 PHASE  I DYE MOVEMENT
        10.5 MONTHS
                   LEGEND
       8 MONTHS
PHASE  I DYE  MOVEMENT
       11  MONTHS
              DYE INTRODUCTION LOCATIONS
               I • - PHASE I
              II* - PHASE II

                 DYE PLUME

              SAMPLING LOCATIONS
               A  - PRAIRIE DU CHIEN WELLS WITH DYE
                 - PRAIRIE DU CHIEN WELLS WITH NO DYE
                 - JORDAN WELLS WITH NO DYE
         SCALE I
Figure 3.  Phase 1 dye movement, four  to eleven months
                           423

-------
•t-

K5
             c
             n
             o
             i-(
             o
             en
             en

             en
             o>
             o
             rt
             H-
             o
             O
             M)
             CD



             O
             <
             (D
             3
             (D
             3
             rt
                         LOCATION

                             MAP
                                 3000
                                           2500
  \

2000      1500

DISTANCE (FEET)
                                                                                                                                       THE HORIZONTAL AND VERTICAL

                                                                                                                                       SCALES ARE DIFFERENT.
                                                                         1000

-------
began to occur  erratically at isolated,  random locations in
the residential  wells.   The  number of  residential  wells in
the study was expanded.  Ultimately the number exceeded 200.
These scattered, short-term detections were eventually shown
to be a random  phenomenon  unrelated to the Rhodamine WT used
in the study.

The results of an anion survey conducted as part of the Study
indicates that the Jordan aquifer in the Study Area is hydro-
geologically  isolated from  the overlying Prairie  du Chien
aquifer.   In many parts of southeastern Minnesota these two
aquifers behave  as single aquifer.    In the  Study Area,  the
Prairie du  Chien aquifer contains  elevated  chloride and ni-
trate contents presumably  from a variety of surface sources.
The Jordan  aquifer,  in contrast, shows  very little evidence
of surface  impact.   The pattern of dye  movement revealed by
the positive  wells is consistent with the chemical data and
with Study Area scale potentiometric mapping.


                       Recommendations

Based on dye  trace study results through September 1990, ad-
ditional geotechnical investigations were recommended for the
site.   Specific  recommendations included:   1)  slug and pump
tests on the existing monitoring wells,  2)  investigation of
the  source  of the  vertical  head gradients at  the  site, 3)
gamma  logging of  the monitoring wells  and  selected nearby
private wells, and 4) investigation of the existence and size
of rapid pulses of contaminants in the monitoring wells.
It was  also recommended that  all of the existing monitoring
wells should  be maintained as part of the expanded Environ-
mental Monitoring System and private wells 151 and 108 should
be  added  to  the  array.    There  is  a  danger that  access to
samples from  wells  151 and 108 could be lost as their owners
seek alternative water supplies.  The  ability to sample wells
151 and 108 should be maintained.

Several  new  monitoring wells  were  recommended.    Several
shallow water-table,  mid-level Prairie du Chien and at least
one  Jordan  monitoring well  were needed  in the northeastern
part of the landfill  site.   Mid-level Prairie du Chien moni-
toring wells  and another Jordan monitoring  well were needed
in the area between the landfill site  and private well 151.

Finally,  if the  anion evidence for leachate  pulses  in the
monitoring  wells  is confirmed, the  ability  of routine quar-
terly sampling  to  adequately  define the contamination poten-
tial will need to be examined.
                             425

-------
                    Remedial  Investigation

During the summer of  1991, work was conducted on the site as
part of  the  Remedial Investigation  (RI)  phase of the  Super-
fund process.  The second stage of field work  for the R.I. is
currently underway-    Seven  new  monitoring wells  have been
installed and  geophysical  logging was  conducted  on a  subset
of the new and old monitoring wells and residential wells 151
and 108.  A  "maintenance mode" of  sampling for dye has been
maintained throughout the  R.I.   The presence  of  dye in the
aquifer was  useful  during  the  installation of the new moni-
toring wells  to confirm correct  siting of  the wells and to
assist in selecting the depth for screen placement.
                          References

Alexander, E.  Calvin,  Jr., Barbara  J.  Huberty  and Keith J.
     Anderson  (1991)    Final  Report  for  Olmsted  County Dye
     Trace Investigation  of  the  Oronoco  Sanitary Landfill.
     Prepared by Donohue and Associates, Inc. April 1991 in 4
     volumes.

Kanivetsky, Roman  (1988)   Bedrock Hydrogeology-   Sheet 5 in
     (N.H.  Balaban,  ed.)  Geologic  Atlas  Olmsted  County,
     Minnesota.   County  Atlas Series  Atlas  C-3, Minnesota
     Geological Survey, St. Paul, MN.

Olson,  B.  (1988a)    Bedrock Geology.    Sheet  2 in   (N.H.
     Balaban, ed.)  Geologic Atlas  Olmsted County, Minnesota.
     County  Atlas  Series Atlas  C-3,  Minnesota  Geological
     Survey,  St. Paul, MN.

Olson, B.  (1988b)   Depth to  Bedrock and  Bedrock  Topography.
     Sheet 4  in  (N.H.  Balaban,  ed.) Geologic Atlas Olmsted
     County,   Minnesota.    County  Atlas   Series  Atlas  C-3,
     Minnesota Geological Survey, St. Paul, MN.
             BIOGRAPHICAL SKETCHES OF THE AUTHORS
E. Calvin Alexander. Jr.r is a Professor in the Department of
Geology  and  Geophysics at the  University of  Minnesota.   He
has a B.S. in chemistry (1966) from Oklahoma State University
and a  Ph.D.  in chemistry (1970)  from  the  University of Mis-
souri at Rolla.  The central theme of his current research is
the rate  of  movement of fluids in hydrogeology.   He and his
research group  are  utilizing a  variety of methods to measure
flow and  residence  times,  which  can range  from hours to ten
of thousands of years.  His full  address is as  follows.
                             426

-------
            Department of Geology  and  Geophysics
                   University of Minnesota
                 Minneapolis, MN 55455-0219
                       (612)  624-3517
Scott C. Alexander  is  a Senior Laboratory  Technician in the
Department of  Geology  and  Geophysics  at  the  University of
Minnesota.    He  has a  B.S.  in  geophysics  (1986)  from the
University of Minnesota.   His current  research involves the
application  of   chemical,   isotopic,  and  computer  modeling
techniques to  the  solution of ground  water  pollution  pro-
blems.   He  is  acting as a  technical  consultant for federal,
state  and  local  research  and water  planning  groups  while
supporting the  day-to-day  operation  of a  hydrogeologic re-
search facility.  His full address is as follows.

             Department  of Geology  and Geophysics
                   University of Minnesota
                 Minneapolis,  MN 55455-0219
                        (612)  624-3517
Barbara J. Huberty is an Environmental Analyst in the Olmsted
County Public  Works Department,  Solid Waste Division.   She
has a B.S. in biology (1977) and a B.S. in broad-area science
education (1979), both  degrees  from the University of Minne-
sota.   Her  current  responsibilities  include  environmental
monitoring of  the Olmsted  County's  landfill operations and
oversight of a Superfund  Investigation.  Her full address is
as follows.

           Olmsted County Public  Works Department
                    2122 Campus  Drive S.E.
                  Rochester, MN 55904-4744
                        (507) 285-8231
James  F.  Ouinlan  is  an independent  consultant  and  former
research geologist  for the  National Park  Service at Mammoth
Cave Kentucky.  He has a B.S. in geology (1959) from Virginia
Polytechnic Institute  and a  Ph.D.  in geology (1978) from the
University  of Texas  at Austin.    His  field  experience in-
cludes:  32 years of  research and observations in karst ter-
ranes of more than  20  countries  and 25  states; more than 500
dye-traces  in the  Mammoth   Cave  region of  Kentucky  and in
various  states;  the  environmental  applications  of  dye-
tracing;  evaluation  of waste-disposal  sites  in  limestone
terranes;  design  of  ground   water  monitoring  networks; and
analysis and  remediation of sinkhole development.    He and
                             427

-------
Ralph  Ewers  received the  1986 E.B.  Burwell Award  from the
Geological Society  of America for their  paper  on monitoring
ground  water in  karst  terranes.    His  full  address  is  as
follows.

                  Quinlan & Associates,  Inc.
                       P.O.  Box 110539
                  Nashville,  TN 37222-0539
                        (615)  833-4324
                            428

-------
                          Questions

1.    There is an explicit vertical path followed by the first
     trace.    What  is  the  possibility  of there  being  an
     additional   vertical  "step"   and  a  deeper,  undetected
     (untested)  flow system?

     Response:   The possibility exists.  The monitoring wells
at the landfill  do  not sample the lower  part  of the Prairie
du Chien aquifer and there was no way to monitor that part of
the aquifer in this study.   The  major  horizontal flow seems,
however,  to be in the  middle of the aquifer,  and we did not
see any  dye  exiting at  springs  that appears  to have missed
the wells.  Finally, the  flow system does not  extend locally
into the  underlying Jordan  (sandstone) aquifer  as  evidenced
by the lack  of  dye detection in any of  the numerous Jordan
wells monitored.

2.    In  retrospect,  how  could  (should?)  your investigation
     have been  organized  to maximize  efficiency and  lower
     costs without sacrificing reliability?

     Response:   Given the initial  state  of hydrogeologic and
dye-tracing  knowledge   for  this   site,   coupled  with  the
administrative  requirements  and  the   residents'  concerns,
there  is  little   that   we  could  have   done  to   maximize
efficiency and lower costs.   Armed with  three  years  and half
a  million dollars  worth  of experience,  requirements  for
future   traces   can  be   refined.     We   would  remove  the
requirements   for  immediate  public  response   and   pre-set,
automatic sampling plan expansion.   Rather, necessary changes
in the project plan should be technically driven.
     The  people  who pay  for dye  traces and  those  who  are
affected by the  results  find it dificult to accept  unknowns
(costs,  time-spans, out-comes,  etc.)  However,  unknowns can
not  be  completely  removed  and  will  continue  to  make  dye
traces complicated and expensive.
                            429

-------
430

-------
    Deducing Karst Aquifer Recharge, Storage, and Transfer

     Mechanisms Through Continuous Electronic Monitoring -

                  A Confirmation With Tracers


            Peter  J.  Idstein,   Ralph O.  Ewers Ph.D.

 Eastern Kentucky University and Ewers Water Consultants  Inc.
                      Richmond, Kentucky



                           ABSTRACT

       Rainfall events produce  significant changes in the stage,
conductivity,  and  temperature of the groundwater mass in a karst
aquifer.    In   theory,  the arrival  time of  these  thermal  and
conductivity pulses,  relative to the stage pulses,  should reveal
the  nature  of  the  aquifer  recharge,  storage,  and  transfer
mechanisms above the point of  measurement.
     An analysis  of  these fundamental  aquifer  properties  was
made  utilizing continuous  monitoring  records  from  a  karst
aquifer in Virginia.  These records suggested that there  is no
allogenic water entering this portion of the  aquifer.   This  was
evidenced  by   the  subdued  response   in  temperature   during
precipitation  events.  The monitoring record also suggested that
the basin contributing  flow  to the part  of  the aquifer  under
consideration  does  not contain a  significant  conduit  storage
component.   This  was  shown  by  the simultaneous  response  of
stage, conductivity,  and  temperature to  even  the smallest  of
storm  events.   The  rapid response to  storm events  and  the
limited rise in stage indicates that the basin is of  very small
dimensions.
     Dye  tracing of the two sources of  allogenic water,  in  the
area, confirmed that they do not  contribute  to the portion of
the aquifer that  was continuously monitored.   The  basin that
contributes  to  groundwater flow at the monitoring point has been
shown by dye tracing to be an area elongated along  the strike of
the local  rocks, collecting only autogenic waters.
                              431

-------
                             THEORY

     In  theory,  the  nature  of  the  flow path  or  paths  that
groundwater in a karst aquifer traverses, to reach a particular
site will be reflected in the physical and chemical adjustments
each path  imprints  upon the water passing through it.   Ashton
(1966) proposed  that  the flow at  any one point  in  the system
will be  the sum of the  response of the  portions tributary to
that location.  Given  the most simple configuration the response
of  a  measured parameter would  be a  deviation from  base  flow
conditions  to a peak and  then  a recession  to   the  base  flow
condition.  Response  for a system that  consists  of  two simple
inputs that combine flow up stream from  the  monitored location
will result in  a record that reflects  this  as the  sum of  the
characteristics  of  the  two  individual  components  (Figure  1).
The  resultant record will  thus  have  peaks  that reveal  the
existence  and nature  of  each of  the  inputs, including  their
phase relationships.   Ashton suggests that in addition to  the
flow rate  other parameters  of  utility  may  be turbidity,  pH,
hardness,and  temperature.     He  also  suggests  that  specific
conductance as a measure of the  ionic character of the water is
a practical alternative for hardness.  The response for each of
these parameters during the passage of  a flood  pulse  will  be
modified  by the various components  of   the  aquifer that  are
encountered.  The changes  that the original  pulses undergo  and
the relative timing of these pulses should reveal  the nature of
the  storage,  transfer  mechanisms and  the  magnitude  of each
component of the aquifer system.
                          STUDY SITE

     The Cathedral  Hall Passage  in Unthanks  Cave  has a  free
surface cave stream that flows through  it.   The Cathedral Hall
Cave Stream contains an  extraordinary assemblage of cave adapted
aquatic invertebrates.  One of these troglobitic organisms is an
endangered  species  of  Isopod  Crustacean Caecidotea  recurvata
that the Virginia Chapter of The Nature Conservancy is committed
to preserving.   This study was conducted to determine the extent
of the drainage basin  and the transfer mechanisms for the water
which contributes to the Cathedral Hall Cave Stream.
     Karst by  it's very nature presents certain  problems  that
must be overcome by the  groundwater scientist.  The methods that
have been used  in this study have been implemented to understand
the special nature of  karst groundwater flow.


LOCATION OF STUDY AREA

     The study  area is  located in Lee  County  Virginia (Figure
2), within the  Powell Valley.  The  Powell Valley is one of the
most  western  valleys  of  the  Appalachian  Valley  and  Ridge
physiographic provence.   The  valley is bounded on the northwest
by Cumberland  Mountain  and to the  southeast by  Wallen Ridge.


                             432

-------
                            me

Figure 1.  Flood wave complexing from multiple  inputs
          (modified from Ashton 1966).
r.. v
                                     '
                                           V
                                  T VIRGINIA

                               TENNESSEE
Figure 2.  Location  of  the  study area.
                             433

-------
The limit of  the  investigation was controlled by the location of
the Powell River and Wallen Ridge.
REGIONAL GEOLOGY

     The rocks in the  region are Ordovician carbonates with some
shales  in  the  valley and  Silurian siltstones  on  the  ridge.
Unthanks Cave is formed in  the upper portion of the Martin Creek
Limestone near the base of the Hurricane Bridge Limestone.  The
Sandy Ridge  Anticline  follows  the regional  structure  and  is
located between the cave and the Powell River.  Dip of the rocks
in  the  area  ranges  from  horizontal  to over  45 degrees,  but
within  the  cave,  dip never exceeds  20  degrees.  There  are  no
major mapped  faults  with  any  surface  expression in  the study
area.
                     CONTINUOUS MONITORING

     Continuous monitoring of the Cathedral Hall Cave Stream was
undertaken  using a  Campbell  Scientific  Incorporated  digital
micrologger model  21X.   This  is  the central  core of  a  Karst
Water Instrumentation System (KWIS).  The micrologger is a very
versatile  device that can  be used  to  monitor many  different
kinds of  sensors and physical parameters.   The  KWIS  used  in
Unthanks  Cave  had  three  sensors  that  were monitored by  the
micrologger for  the  time  that the  system was  installed in  the
cave.   These  sensors  were  used  to monitor  the  stage  (water
level),  temperature and conductivity (specific conductance)  of
the water in the Cathedral Hall Cave Stream.
     The stage of  the  cave stream was monitored  using a  Druck
pressure transducer. This  device possesses a silicon wafer that
varies in  resistance as  the pressure around it changes.   This
change is proportional to the height of  water that is above  the
sensor.   This transducer can detect changes in  stage as small as
.001 feet.
     Water  temperature   was  monitored   using  an   Platinum
Resistance  Thermometer  (PRT)  supplied  by  Omega  Engineering
Incorporated.  The  resistance of  a platinum wire changes with
the water temperature.   This resistance  is measured relative to
a   100   ohm precision   resistor   that  is  attached   to  the
micrologger.  Temperature is measured in degrees Celsius with a
resolution of  -01 degree.
     The  specific  conductance (conductivity)  sensor  that  was
used at this site was designed and  constructed  specifically for
this site.  The  new design  was developed  in  order to avoid the
large current demand of earlier instruments.  The new design has
greatly reduced  current demands and  is  thus  able  to be powered
from the  micrologger directly.  Another  advantage to  the  new
design  is  the  elimination  of  the  "ground  looping"  problems
associated with most other conductivity measurement systems.
     The datalogger  and  sensors  were located  in  the Cathedral
Hall stream passage of Unthanks Cave. This  passage is oriented


                             434

-------
along the  strike  of the  rocks  in the  area  and is  apparently
controlled by the  fractures in the Martin Creek Limestone.   The
cave stream is about one  inch deep by one foot across.   There
are numerous pools  in  the bottom of the  accessible  potions  of
the cave  stream  that  are up to  18  inches  deep.    The  sensor
cluster used in this study was located in one of these pools  to
ensure  that  they  always remain  covered  with  water.    The
datalogger was located five  feet  above  the  cave  stream on  a
ledge.


TYPICAL RESPONSE  OF THE SYSTEM

     A typical response for the  aquifer system at Unthanks Cave
can be seen in Figure 3.   This record is related to the day 193
rain a single  short uncomplicated event.  The  system responds
with a rapid change  in stage as the pulse of  water  progresses
through the  aquifer.   Temperature of  the water experiences  a
rapid rise  at the  same  time as  the stage  changes  are  seen.
Water  temperature  for  this  event  responds   in  a  increasing
manner, events during  colder  seasons  are  represented  by  a
decrease in temperature.   There is also an immediate decrease in
the specific conductance of the water from the base  flow values.
     A short  term  reversal  of  the  trend of  temperature  and
conductivity occurs within a few minutes  of  the initial system
response.    The detail of  this occurrence  can be   seen  most
clearly in Figure  4.
     The  peak  (the  maximum  deviation  from  base  flow  values
independent  of  direction)  is  reached  for   each   of  these
parameters at  nearly  the  same  time.   Recession  of stage  is
typically  a  simple  return  to  base  level without any  abrupt
changes in slope.  The return of temperature to base flow values
usually experiences  a  noticeable  change in  slope during  its
recession.  Conductivity may also experiences a change of slope
during the recession to base flow values.
     All  of  the  responses to  storm events  take  this  general
form.    Deviations from this form are  believed to  be due  to
variations in antecedent conditions.  These changes  may be  due
to the system being  in a state of recovery from a previous storm
event and  thus its response  is that of  the  combined  pulses.
Another factor connected  with  these  deviations is the  present
state of  saturation of   the  soil and  rock.   This  may  vary
depending  upon the length of time between precipitation events
and the  seasonal  variations  of evapotranspiration.    Further
variation  may also arise from the stage increase associated with
large storm events which bring other flow paths into operation
that were  not active during smaller events.
     The simultaneous response of temperature,  conductivity and
stage is believed  to be an indication that fresh storm water is
arriving  at  the  monitoring  site  from  inputs  very nearby  and
without significant restriction of  flow.    Hess  and  White
(1973,1988) and Meiman  et. al. (1988) observed a lag between the
arrival of the  stage response and a response  in temperature  and
conductivity.   This has been attributed to the propagation  of
                             435

-------
   Figure 3.
Figure  4.
 300
                                                            22:00  24:00
   193
        194
             195   196
              JUUANDAYS
                                                    18:00  20:00  22:00  24,00
                        197
                                     300
                                      12:00
                                           MOO
        16:00  16:00  20:00  22.00  24:00
         BOMB (MIAN MY 113)
Figure  3.  Aquifer  response to  rain event  of Julian  day 193.
           Representative of typical aquifer response.


Figure  4.  Detailed inspection  of the fluctuation  in specific
           conductance and temperature  shortly after first
           response to the Julian day 193  rain event.
                                436

-------
the stage pulse  as  a  pressure wave  through the  phreatic  portion
of the aquifer in advance of the fresh input waters.  This  lag
effect has not been observed at this site, thus  it  is asserted
that there is no significant phreatic contribution  to the flow
that passes the  monitoring site.
     The  reversal  of  trend  in temperature  and conductivity
(Figure 4) a short  time after the initial  response to the storm
event is presumed to be the  arrival of higher  conductive water
with  a  temperature  signature  resembling that  of  base flow
values.   This water is assumed to be the flushing of  stored
water  as the fresh  input  waters  affect  other  parts  of  the
system.   This may  be  the  arrival of  a  flow component from
further up stream or displacing water from nearby storage that
exhibits more restricted  flow  of  extended residence time than
the initial response component.  The cessation  of this reversal
of trend is interpreted as  the  flushing  of the stores  of this
second component or  the  overwhelming  of   the  influence by  an
increasing volume of fresh input water.
     Meiman et.   al.  (1988)  saw multiple  slope changes  in  the
conductivity  response   that  was attributed  to  the  input   of
separate concentrated  allogenic sinking  streams in the  basin.
There is normally only  one slope change seen in  the conductivity
response with the present study.  The arrival  of  this pulse is
seen as  a  more  conductive water than  the fresh  input  waters.
This high conductive signature  is  not  likely to  be  associated
with an allogenic sinking stream.
     Continued  departure   from  base  flow   values  for   all
parameters is seen  as the  fresh  input waters dominate the flow.
Maximum stage occurrence is associated with the cessation of  the
source of fresh  input waters to  the system.  The  return  to base
flow conditions   occurs as the excess water  moves through  the
aquifer.    Conductivity  and  temperature peaks are   reached
slightly after the stage  crest  is  reached.  Recession  to base
flow conditions  for conductivity and temperature is marked by an
initially rapid  change that corresponds to the swift  recession
of  stage.    The  slope  of the  recession  of  conductivity  and
temperature changes as  the stage response  approaches base flow.
This change in slope  is due to reduced significance of the quick
flow component of the fresh input water and an  expansion of  the
influence  of the   restricted  flow  extended   residence time
component.


SYSTEM RECOVERY  AFTER STORM EVENT

     The response to  the day 259 precipitation  event is  similar
to the record for all  single storm events.  This  event  clearly
displays a temperature phenomenon that exists  in  the record of
most events that have  been  observed  in this  study  (Figure  5).
After the peak response of the event has been  attained  for  all
parameters and the recession  to base flow  conditions  begins,  the
return of temperature  to  base  flow values lags behind  that of
stage,  and to a  lesser extent it lags  behind  conductivity.
                              437

-------
    Figure 5.
 Figure 6.
  2.00
    252   254  258  258  260  262   264   266
2.401
2.20
2.00-
1.80-
1.60 "

0 80 •

IS




^

5 20C




L , ^ ^

205 .



U\,"''i ... •••[•••
TA....I 	 J..1

10 215 220 225 230 235



\J \


24
1 /.U





13.5-
1.1 0 •






s^- 	














\


                                        195  200  205  210  215  220  225  230  235  240
                                      500
                                      I 450
                                      I 400
                                      •150-

                                      i

                                      | 300
  300
    252  254  256  258  260  262   264   266
               JUUANDU3
                                      230
195  200  205  210 215 220 225 230 235  240
            JUUiNUYS
Figure 5. System response  to storm  events of  Julian days  259
           and  255.   The response to the day 259 event  exhibits
           the  delay in recovery of  temperature, conductivity
           and  stage relative to each other.   The initial
           response  to the  day 255 event show  an initial
           increase  in conductivity.


Figure 6. System response  to drought conditions.
                                 438

-------
     In contrast to this  study Meiman et. al.  (1988)  observed
that the return  to  base flow  values  for temperature  was  more
rapid than for conductivity.   Storage of  thermal  energy  in the
rock mass  in  the vicinity  of the  active flow  routes is  the
likely reason  for any lag effect in temperature.  The release of
this thermal energy  from the rock  back into  the water slows the
return of  temperature  to base level  until  the thermal  stores
have been depleted.
     This  difference  in response  in  the two  studies may  be
related to the presence or absence of phreatic passages  or the
distance between the monitoring site  and the  source  of  input.
The diffuse stores in the subcutaneous zone  and saturated  rock
mass with restricted flow conditions will have a relatively  high
conductance due to an extended residence  time.  The  aquifer at
Mammoth  Cave   has  been  shown  to  receive   flow   from   both
concentrated and diffuse sources under both  flood  flow and  base
flow conditions.  The  combining of  multiple sources  will  slow
the return  to base  flow values of  the  conductance  at a  down
stream monitoring point.  Additionally, the  aquifer  at Mammoth
Cave has a  significant  phreatic  zone.   As  the storm  pulse  is
transmitted through the aquifer  a significant portion of  the
fresh  input waters  may  be  pushed from  the  conduit  into  the
conduit adjacent stores.  These short residence time waters  will
be released from the conduit adjacent  stores  as the  flood event
wanes.  The low  conductivity waters  from the  conduit adjacent
storage will  slow the  recession  to  base flow values.  If  an
aquifer receives the majority of  its  base  flow  from extended
residence  time  water   the  response  will  be  such  that   the
conductivity may  return to  a  base  flow  condition before  the
thermal stores in the rock from the storm event can be depleted.
     Meiman et.  al.  (1988)  saw  that  as a  storm  pulse   was
transmitted through  the system, there  is a  lag in the initial
response of temperature relative  to conductivity  and  stage.
They  also  note  that  the  amount  of  lag is  affected by   the
antecedent conditions.    The absence  of  a thermal  lag at  the
onset of the  storm event,  in the  present study,  suggests  that
this initial response is dominated by a quick  flow  component.
This  may  reflect  concentrated   autogenic   contribution   into
enlarged fractures and  vadose shafts.   Through this  route  of
transit the water would  not have enough time  to react chemically
with the  rock to significantly change its  conductance.    The
leading  edge  of  the  thermal  response  would  not  be delayed
relative to the other values if the input is  near the  monitoring
site.   Additionally there is  sufficient  contact with  the  rock
that a  reduction of the  temperature  of  the  water  without  a
significant delay in  the  arrival of  the initial  temperature
change is  likely.   If  flow  at  the monitoring site  received
contribution from the allogenic sinking streams in the area  this
water would not have sufficient contact with  the rock to  subdue
the thermal response.
                             439

-------
CONDUCTIVITY RISE AT THE ONSET OF AN EVENT

     Meiman et. al. (1988) saw an  initial  rise in conductivity
at the onset of storm events.  This is explained as the flushing
of high conductive stores from portions of  the aquifer that are
not typically active during base flow.  The  storm event of day
255 (Figure  5)  does show a small initial rise  at the onset of
the storm event.  This is the most noticeable occurrence of this
effect  seen during the time  this  site  was  monitored.    The
typical response has been, as stated before,  an immediate drop
in the  conductance at the  same time that the  other  parameters
respond.  This suggests  that the  flow route for the fresh input
water must be very direct and possessing very limited stores in
order   to   dominate  the  initial   response,   and  that   only
periodically are  there  sources of  high conductive water  to be
flushed ahead of this  storm pulse.
LOW FLOW CONDITIONS

     The flow of the cave stream during the period from day 195
to day  235 was  seriously reduced  due to  the dry  conditions
(Figure 6).  Starting about day 201  the stream flow was reduced
enough that the level of  water in the pool, where the probes had
been located,  was no longer sufficient to maintain a normal flow
out  of   the pool.    The  reduced  pool   level  appears  to  be
controlled by fractures at a location lower than the normal pool
level.
     In  response to  several  small  storm  events,  that  occur
during this period of time, the pool level is temporarily raised
to the overflow point that signifies the  return to stream flow.
This pool height is  only maintained for a short period of time
until the flow is cut off and the surface is returned to a level
controlled  by  the  fractures.   Water  temperature during  this
period is very stable and there is only a  limited  response that
can be  associated with  the small  storm  events that have  been
recorded.  This  thermal  stability indicates  the limited extent
of these  storm  events and/or  reflects the  efficiency  of  the
thermal stores.  Conductance of the water fluctuates with each
storm event but is stable between storm events.  When the stream
ceases  to  flow, the pool  is  once  again  isolated  and  so
measurement of the  conductance represents  the  character of the
last water to be flowing in the stream before flow stopped.
     During  a portion  of  the  time  that  the stream  in  the
Cathedral  Hall  Cave  Passage had  ceased  to  flow, the  sinking
streams that flow from the ridge still maintain a low  level of
flow.  If  these  streams  had contributed flow to the  portion of
the cave that  was continuously monitored during this time period
the flow of the cave stream would have been maintained.
                              440

-------
                       DYE TRACE RESULTS

ALLOGENIC SINKING STREAMS

     Two allogenic sinking streams  flow  off  of  Wallen Ridge in
the area of the Cathedral Hall Passage.  The "Scott Farm Sinking
Stream" (SFSS) was found to sink at  the contact of the Hurricane
Bridge  Ls.  and the  Martin  Creek  Ls.    Water  flowing in  this
stream includes runoff from a few farms and a winding road that
cross the  ridge,  State Route 758.   The sinking  stream  "SFSS"
(Figure  7)  was traced  at three  different  times  due to  it's
proximity to the cave stream of interest.  At no time during the
study was this sinking stream found to connect to the Cathedral
Hall Cave Stream.   Dye from "SFSS" has been detected in the cave
at  monitoring  points  down  stream from  the  Cathedral  Hall
Passage.
     The water  flowing  in  the  Horn Farm Sinking  Stream  (HFSS)
was shown to flow  away from Unthanks Cave parallel to strike but
in the opposite direction  of all other traces.  Water Sinking at
this  location on  the  Horn  Farm was  never  shown,  under  the
conditions  investigated,   to  contribute  to  any  portion  of
Unthanks Cave.
AUTOGENIC SOURCES

     During one of  the  research trips into the  Cathedral  Hall
Passage a  strong  odor  of  "diesel" fuel  was  detected.  After
exiting the  cave  an  investigation  of  local  sources  of  this
product was initiated.  It was  determined that the  most  likely
source was due  to dumping of  excess  product  from  large metal
drums along side a sinkhole located behind "Bacons Store"  (BSD).
This  is  the   site of a small  country  store  that  also sells
gasoline.   This sinkhole  was  later  used  for  dye injection  in
order to further understand its connection  to  the cave system.
Water  entering  at  this point flowed  to  the Cathedral  Hall
Passage and was  detected at the monitoring site.  Dye introduced
at this site  was  also detected at dye monitoring sites  in the
cave down stream from the Cathedral Hall Cave  Stream.
     "JWD1" is  a  sinkhole  located a short distance  south  of
State Route 758  in a wooded rocky portion  of a farm. Dye input
at this location  was detected  in  the Cathedral Hall  Passage.
This dye was also detected at  dye  monitoring  sites  down  stream
from  the  Cathedral  Hall  Cave Stream.    This  site  was  only
investigated  during high flow.
     Sinkhole "JWD2" is also located  a short  distance  south of
State Route 758 on  the  same  farm.   This  sinkhole is located a
few hundred feet East of JWD1.   Dye introduced at this  site was
detected at dye  monitoring  sites down stream from the Cathedral
Hall  Cave  Stream.    This   site was  tested   under  high  flow
conditions  but  no  dye  from   this  site  was   detected in  the
Cathedral Hall Passage.
     "RSD"  is  a  sinkhole that  is located very close  to  and  just
south of State  Route 758.   It is about half way between "BSD"
                              441

-------
                                                          4000 FEET
                                   DRAINAGE BASIN CONTRIBUTING TO
                                   CATHEDRAL HALL PASSAGE CAVE STREAM
LOCATION OF CAVE
Figure 7. Map  showing the location  of  dye insertion sites  and
          the  basin that contributes to the Cathedral Hall
          Cave Stream.

-------
and the Cathedral  Hall Passage.   Dye introduced  at  this point
was detected in the Cathedral Hall Passage and at dye monitoring
sites further down stream.
                          CONCLUSIONS

     The simultaneous response of all parameters at the onset of
a  storm  event  establishes  the existence  of  a  quick  flow
component to  the  flow of the Cathedral Hall Cave  Stream.   The
lack  of  delay in  response  of   conductivity  and  temperature
relative to stage response confirms  the absence of a significant
phreatic flow path for this quick flow contribution.
     Variations of  conductivity  and temperature shortly  after
the  initial  response from  base  level reflect  the arrival  of
fresh  input  quick flow  waters with flushed  stores from  more
distant or restricted flow sources in the aquifer.   Flushing of
these stores or overwhelming of their signature by the increased
volume  of  fresh input is considered to  be the reason  for  the
cessation of  these  occurrences.   The ordinary absence  of  high
conductance waters  to be flushed from storage at  the onset  of
storm events suggests that the routes of quick flow are normally
drained of extended residence time waters.
     Return of the waters conductance to  base flow values  more
rapidly than temperature, reflects  the absence of  concentrated
allogenic inputs and/or conduit adjacent  storage  of fresh  input
waters.   Concentrated allogenic  sources  and conduit adjacent
storage of  fresh input  water would have low conductance  and
compete with  the  high conductance restricted flow  waters  that
maintain the  base flow.   The  absence  of  these sources in  the
flow at Unthanks Cave allows conductance  to return  to base  flow
conditions before temperature.
     The cave  stream  stops  flowing  during dry periods  but  the
sinking streams flowing from Wallen  Ridge, for a portion  of  this
dry period, continued to flow and sink.   There flow must not  be
related to the cave stream under these conditions.
     Tracing of both allogenic sinking streams in the area  shows
that "HFSS" does not contribute to any  portion of Unthanks  Cave
and that "SPSS" contributes to portions of the cave down stream
from the Cathedral  Hall  Passage.   Tracing of sinkholes in  the
area has shown the location of the drainage basin and confirmed
the source of the water is exclusively autogenic.
     The dye tracing has  supported the conclusions of the nature
of the drainage basin  and the continuous monitoring has revealed
more detail about the transfer mechanisms than the qualitative
dye tracing could do.
                       REFERENCES CITED

Ashton,  Ken,  1966, The analysis of flow data from karst drainage
     systems:   Transactions  Cave  Research  Group of    Great
     Britain, v.  7,  p.  161-203.
                              443

-------
Hess, J.W., and White, W.B., 1974, Hydrograph analysis of karst
     aquifers:     Pennsylvania  State   University,   Institute
     Research Land Water Resource Publication no.  83,  63 p.

Hess, J.W-, and White,  W.B.,  1988, Storm response of the karstic
     aquifer  of  Southcentral Kentucky:   Journal  of  Hydrology,
     no. 99, p. 235-252.

Meiman,   J.,   Ewers,   R.O.,   and   Quinlan,    J.F.,   1988,
     Investigation of  flood pulse movement through  a maturely
     karstified  aquifer  at Mammoth  Cave  National Park: a  new
     approach: in Proceedings of Environmental problems in Karst
     Terranes and there Solutions Conference  p. 227-263
                     BIOGRAPHICAL SKETCHES

     Peter Idstein  received  his Bachelor of  Science  degree  in
Geology at Eastern  Illinois  University.   He  is  completing his
Master  of  Science  degree  in  Geology  at  Eastern  Kentucky
University.  Mr.  Idstein  has  spent  one  year working at the
Florida  Sinkhole  Research  Institute  conducting  studies  on
conduit dominated groundwater  flow.   He has  also  spent  a year
working  for  Ewers  Water  Consultants conducting  dye  tracing
studies and  continuous electronic monitoring studies in many
karst dominated and non-karst terranes.

Peter Idstein
Dept. of Geology, Rm. 9 Roark
Eastern Kentucky University
Richmond, Ky. 40475
606-622-1273

     Ralph O. Ewers is professor of geology and director of the
Groundwater Research Laboratory at Eastern Kentucky University,
and  a principal  in  Ewers Water Consultants,  a consulting firm
specializing in carbonate aquifers.   His  B.S. and  M.S. degrees
in geology were earned at the  University  of Cincinnati and his
Ph.D. was earned  at McMaster  University  (1982).   Professor
Ewers' special interests include the applications of tracer and
electronic monitoring  techniques  to the  solution  of  practical
environmental  problems  in  karst  groundwaters.     He was co-
recipient of  the  1986 E.B. Burwell  Award  from  the Geological
Society of America  for a  "work of distinction  in engineering
geology."

Ralph 0. Ewers
Ewers Water Consultants Inc.
160 Redwood Drive
Richmond, KY 40475
606-623-8464
                             444

-------
    Deducing Karst Aquifer Recharge, Storage, and Transfer

    Mechanisms Through Continuous Electronic Monitoring -

                 A Confirmation With Tracers


           Peter J. Idstein,  Ralph O. Ewers Ph.D.

 Eastern Kentucky University and Ewers Water Consultants Inc.
                      Richmond,  Kentucky
Question:
     Are there temperature  lag-times,  and  are  they  the  same  for
low flow and high flow conditions?

Response:
     The study at Unthanks Cave showed no temperature  lag-times,
at the  onset  of  the  storm  events during the period that  the
system was monitored.   Work at Mammoth Cave  National Park  by
Meiman et.al.  1988 did show significant temperature  lag-times at
the onset of storm events.
     There were temperature  lag-times in the return to base flow
values.   Both stage  and conductivity  returned to  base flow
conditions before temperature.  The difference between  low flow
and high flow  conditions does  not seem  to be  as  significant as
the variability in temperature of the fresh input waters.
                              445

-------
446

-------
            PETROLEUM HYDROCARBON REMEDIATION OF THE
              SUBCUTANEOUS ZONE OF A KARST AQUIFER,
                       LEXINGTON, KENTUCKY

                         Scott A. Recker
              Delta Environmental Consultants, Inc.
                    Charlotte, North Carolina

Investigation of karst aquifers  using dye  tracing techniques has
reached widespread  acceptance  in  the  hydrogeologic  community.
These techniques have  only  recently been applied  to  tracing the
release of  contamination  from  a point source.  These studies fail,
however to provide  suggestions  for the remediation of  the karst
aquifer affected by the release.

Following   the  release  of  1300 gallons  of  gasoline  from  an
Underground Storage Tank an interceptor trench  was installed and
recovered   approximately  800  gallons of  the  lost product.    A
subsequent  tracer  study  from the  loss  location  indicated  flow
through the subcutaneous  zone at 400 ft/day  and discharge to the
ultimate resurgence, 1.75 miles  away,  in 67 days.  Ground water
samples collected at the spring resurgence indicated non-detectable
levels of petroleum hydrocarbons.  Residual soil and ground water
contamination still  remained  in the  subcutaneous zone  at  the
release  location.     This  residual  contamination  acted  as  a
continuing  source  of  ground  water  contamination into  the karst
conduit drainage network  and resulted in gasoline vapors in local
businesses.

A remediation system was  designed to remove the residual gasoline
contamination through  the use of soil vapor  extraction,  in-situ
soil washing and standard pump &  treat technology.  The system was
put on line during  April  1991.  The recovery and treatment system
continues  to remove gasoline  contamination  from the subcutaneous
zone  and  reintroduce  treated ground  water—mimicking  constant
precipitation event.

                          INTRODUCTION

Tracer investigation in karst  regimes has become an exact and well
defined technique to determine conduit drainage network boundaries
and conduit flow paths.   Tracer techniques have recently been used
to trace the  transport fate of contaminants from  a point source
release location within the  conduit drainage network.   Studies of
this type have  been used to estimate the residence times of the
contaminants in the aquifer but  fail to  address remedial options
for the affected portions of the aquifer.  Releases of contaminants
are often  times to the  overburden  and  subcutaneous  zone of the
aquifer rather   than   directly   to  the  well  developed  conduit
drainage.

Remedial options for karst aquifers are often not considered due to
the fact that they are applied to the conduit system  and fail to
address contamination of  the subcutaneous zone.   This study
                              447

-------
presents   information  collected  using   classical   contaminant
investigative techniques  for unconsolidated overburden  materials
and using dye tracing techniques and how is is applied applied to
design and implement  a  remediation plan of the subcutaneous  zone
for the case study presented.

The subject area is located at the corner of South Limestone Street
and Gazette Avenue in  Lexington, Kentucky (Figure  1 & 2) .   The  site
lies in  the Inner Blue Grass Physiographic  Province of  central
Kentucky.  North of the site are office buildings and  houses, the
University  of  Kentucky is  to the  east,  office  buildings and a
residence to the west, apartment buildings  to  the southwest, and a
bank, convenience store, and restaurant to the south.

SITE HISTORY

The site has been a gasoline station and convenience store for at
least 20 years.  A map showing the  former, Shop-N-Go  (Super America
station),  surrounding businesses   and  residences is  included as
Figure 2.  On April  12,  1986,  the  Shop-N-Go reported a  sudden  loss
of 1,300 gallons of  gasoline from  an underground  storage  tank.  On
April 13, 1986 gasoline  was detected in the  basement  of Mr.  Gatti's
Pizza Restaurant,  located approximately 400  feet to the  south.
Because the detection of gasoline occurred approximately 24 hours
after the loss,  the gasoline apparently originated at the  Shop-N-Go
site.

Several remedial  activities were  undertaken  in  response to  this
loss.   Activities  included  excavation and  removal  of  leaking
underground  storage  tanks,  removal of  contaminated  soils,  and
removal of liquid phase hydrocarbons (LPH)  from the tank basin at
the  Shop-N-Go   site.    A recovery trench  system was  installed
adjacent to Mr.  Gatti's Pizza Restaurant and has  been  in  operation
since April  1986.   Initial  remedial  activities  resulted in the
recovery of nearly 800  gallons  of gasoline and the excavation of
350  cubic  yards of  contaminated soil from the site.   Additional
gasoline was reported to have discharged to the storm water sewers
at  the  site.   Approximately 800  gallons of  the gasoline  was
recovered during the first few weeks after  the reported spill.  No
LPH  has  been detected in the recovery  sump  since December 1986;
however,  contaminated  ground  water  has  been   recovered since
remediation was initiated.

                    GEOLOGY AND HYDROGEOLOGY

GEOLOGY

The  site  is situated within the Inner Bluegrass  Physiographic
Province of central Kentucky.  This area is characterized by  flat
lying to very gently dipping limestones, dolostones and  shales of
Ordovician Age.   Strata  beneath the site consists of  limestones and
thin shales of the Lexington Limestone.  Members  of this  formation
show an intertonguing relationship throughout the Lexington area.
The members underlying the site are, in descending order,  the
                                448

-------
                                                           SITE
HE
                             KENTUCKY
                            0
                            fc

                            SCALE (MILES)
                  Figure 1.   Site Location

-------
                                                                    °YE INJECTI°N P°INT
                                                                                                 SOUTH LIMESTONE AVENUE
Ul
o
             .LEGEND;

             —X—X- FENCE
                                     ABANDONED INJECTION
                                         I BORINGS
/
CHECKERS
FOOD
7

MR. GATTIS
PIZZA





OFFICES






•^
TREATMENT .^
YSTEM LOCATION V
1


Kl
o
If
0












                                                                                                                                 60
                                                                                                                        SCALE FEET
                                                                          Figure  2.    Site  Map

-------
Tanglewood,  Brannon and Grier Members of the Lexington Limestone.
The Tanglewood and Grier Members are characterized by thinly bedded
argillaceous  limestones  with  very  minor  and  isolated  shale
partings.  These limestones are subject to solutional modification
by ground water flow as evidenced by solutional features along road
cuts as well as abundant closed  surface  depressions  in the area.
The Brannon  Member  is  characterized  as argillaceous  limestone
containing  shale and chert beds  and solutional modification is
uncommon (Miller,  1967).

A total of  36 test borings were  drilled  during  November 1987 and
twenty three additional  borings were advanced by Delta in February
1988 and May  1990.  The boring locations are  shown in Figures 3 and
4.    Material  beneath  the   site  consists  of  lean   to  fat
unconsolidated overburden with minor amounts of chert and limestone
rock fragments increasing with depth.  The depth to the Tanglewood
Member ranges from 8.5 to  16  feet below grade.

Soil was screened  with a PID  and soil samples were collected from
the 14 borings advanced in May 1990.

Soil screening and analytical results indicate that elevated levels
of  petroleum  hydrocarbons are  concentrated at  the  soil/bedrock
interface over a large area.  Levels appear to be higher in a small
area west of  Mr. Gatti's Pizza Restaurant.  The approximate extent
of elevated  hydrocarbon levels in soil is shown in Figure 5.

HYDROGEOLOGY

Ground  water was  not  encountered  during  the  boring  programs,
therefore, there have been no static measurements of  ground water
at  this  location.    Flow  mainly  occurs along  the soil  bedrock
interface through tiny  channels  and solutionally enhanced joints
and bedding  planes known as the subcutaneous or epikarstic zone.

                    Ground Water Tracer Study

In order to determine the ultimate pat for ground water and, hence,
contaminant  movement,  it was necessary to conduct a dye trace from
the areA of  the  gasoline loss.

After  initial  reconnaissance on  April  15,  1989,  the  following
observations were  made:

     1)  Directly underlying the site is the Tanglewood Member of
         the Lexington  Limestone.  The Tanglewood is karstif iable,
         as evident by many  area sinkholes.

     2)  Most area springs are discharged from the Grier Member of
         the Lexington Limestone.

     3)  Between  the Tanglewood and Grier  is the  Brannon Member
         which,  although  contains  some  shale beds and chert, is
         considered a semi-permeable unit.
                               451

-------
                                               , DYE INJECTION POINT
                                                                             SOUTH LIMESTONE AVENUE
                            PUMP
                           ISLANDS
                      ABANDON INJECTION
                          BORINGS
         ^\^

         ^
I
                           •^-^ <,8-6
                                                     g_2    B_7(,    B-25,
                                                           B-Bn
'/
   «?
  «^
                            ^
UGEND;

—X—X- FENCE
    (1   TCT SOIL BORING LOCATIONS (2/BB)
«G.
                                                     8-4
                                                                           26
                                                 B-27
                                                 «
1


CHECKERS
FOOD
MART







MR. GATT1S
PIZZA

R
*
H

                                                                                                    -32
                                                                                                                     OFFICES
                                                                                              B-30<»
                                                     .B-M  B-1
                                                                                           B-29
                                                                   B-13-    B-17-
                                                           B-10
                                                                                      B-23a
                                                                        B-19
                                                                              LIMESTONE
                                                                              APARTMENTS
                                                                           APARTMENTS
                                                                                               B-21,
                                                                                                  B-
                                                                                                              E
                                                                                                              o
                                                                                                            60
                                                                                                   SCALE FEET
                                    Figure  3.    Borings  advanced  during  1987

-------
                                                                      SOUTH LIMESTONE AVENUE
1ECQJQ;

—X—X-  FENCE
 A-1 •   SOIL BORING LOCATIONS (SEPTEMBER 198B)
 C-l»   SOIL BORING LOCATIONS (MAY 1990)
                                           LJ


                                           1
                                           A-3
                                                            C-H
SECOND
NATIONAL
 BANK
                                                           C-4
'
CHECKERS
FOOD
MART
7
/
C-9»
MR. GATTIS
PIZZA

                                                                           RECOVERY^
                                                                            TRENCH
                                                       >C-2
                                                 A-51
                                          A-4
                                                  A-6
                                                                    C-7 I
                                                                                                           omcES
                                                                                           A-B
>v C-1 C-6 /^



X

UJ
'k
A-


^
LIMESTONE
APARTMENTS

•
\S \S •*'
K-lf~

s\ s\
«.«.

                                                                                            C-10
                                                                                                                           [3
                                                                                                                           o
                                                                                                                           E
                                                                                                                           o
                                                                                                 60
                                                                                          SCALE FEET
                         Figure  4.   Borings  advanced during  1988  and  1990.

-------
                                                                 SOUTH LJMESTONE AVENUE
1EGEND;

—X—X- FENCE

—— > 20 mg/kg TOTAL BTEX

- —— - 1-20 mg/kg TOTAL BTEX

—.—.._ < 1 mg/kg TOTAL BTEX
                                                         SECOND
                                                        I NATIONAL
                                                          BANK
/
CHECKERS
FOOD
MART
7
/
/

MR. GATTIS
PIZZA

                                                                                                   OFFICES
                                                                      TRENCH
                                                                                    ^SEPARATOR
                                                                                       TANKS
                                                                 LIMESTONE
                                                                 APARTMENTS
APARTMENTS
                                                                                          60
                                                                                   SCALE FEET
Figure  5.   Estimated  extent of soil  contamination  based  on  soil boring  information.

-------
                    Table 1
Detector Locations                      Location Codes

Recovery Trench                              RT
Mr. Gatti's Sump                             MG
Storm Sewer                                  SS
Storm Sewer at Railroad Culvert              RR
Big Elm Country Club                         BE
Wolf Run                                     WR
Vaughn's Branch                              VB
Town Branch                                  TB
Hickman Creek                                HC
McConnell Spring                             MS
Preston's Cave Spring                        PS
                       455

-------
-fc-
Ul
             LEXINGTON WEST. KY.
             SE/4 GEORGETOWN 15' QUADRANGLE
                  38084- A5-TF-024

                      1965

                  PHOTOREMSED 1887
             DMA 4060 n SB-SERIES VB53
                                                                                                IDEALIZED FLOW ROUTE
    SCALE  1:24000
                           1 MILE
1000    20OO   3000   4000    5000
 CONTOUR INTERVAL 10 FEET
                                        Figure 6.   Dye  detector locations  and  tracer results.

-------
     4)    Local  dip is to the northwest  at  approximately 20 feet
          per mile.  There are no mapped faults  in the immediate
          area of the site.

     5)    The storm sewer (later used as a dye detector site) which
          passes the site and  receives water from the separator
          tank,  ultimately discharges in Vaughn Branch (also later
          used as a detector site).

     6)    Dye detector locations chosen are included in Table 1 and
          shown  in Figure 6.

Dye Background

Prior to  dye tracer  studies,  an  analysis for  the presence  of
background levels of fluorescent dyes is  necessary.  Dye detector
sites were chosen  (Table 1  and Figure 6).  Background detectors
were deployed and  analyzed  on two  occasions.    These  detectors,
collected on July 14, 1989 and July 21,  1989 indicated the presence
of extremely  weak concentrations of fluorescein  dye in the storm
sewer and McConnell Spring sites.  The dye was determined to be of
insignificant concentrations to negate  the use  of fluorescein for
the tracer study.

Dye Injection

On July 21, 1989 at approximately 11:00  a.m., fluorescein dye (acid
yellow 73) was injected  into a  fresh 10 inch diameter soil boring
at the Shop-N-Go  site.   The location  of the  boring is  shown  in
Figure 2.  The boring encountered bedrock  at approximately 12 feet.
Prior to  the  boring of the  injection well, two  additional borings
were drilled  approximately 3 and 6  feet  from the  injection well
(Figure  2) .   These wells have  a total depth of 9 and  10 feet,
respectively,  and would  not effectively take a slug of water.  The
injection well readily  accepted  over  100 gallons  in a  period  of
less than 10 minutes.  After the initial slug of water, 3 pounds of
fluorescein mixed with  2 gallons of water was injected  into the
well. The dye was  then  followed with approximately 200 gallons of
water.

Dye Recovery

Dye detectors were  retrieved at weekly  intervals  throughout the
course of the study.  Dye detectors were eluted  in a solution of
potassium hydroxide  and  isopropyl  alcohol and  analyzed on  a
spectrofluorophotometer. Detector analysis results can  be found in
Table 2.

Dye Trace Results

The first sampling point to yield a confirmed positive was the sump
in the basement of  Mr. Gatti's.   The  dye was present  on the
detector  collected July  27,  1989.  The second  round of detector
                               457

-------
 SITE




 RT




 MG




 SS




 RR




 BE




 WR




 VB




 TB




 HC




 MS




 PC
                                                              Table  2
7/14    7/21     7/27    8/3      8/9      8/16    8/24    2/1
 VW
+V     +VB    +VB




+VB    +VB    +VB









                W
        VW
DETECTOR RESULTS




        =      dye not detected




VW     =      very weak




W      =      possible dye detected, weak




+      =      dye detected




+VB    =      dye detected, very strong




X      =      detector not collected as past + confirmed




LS      =      detector lost or stolen




DATE   =      detector collection dale
 X




 X




LS
                                                LS
                                                LS
                                                        X




                                                        X
2/9     9/15     9/26     10/10    10/19    10/27    11/S




 XXX       X      X       X       X




 XXX       X      X       X       X
                                                                                W
- - -
LS
LS LS
-
LS
LS
LS LS
h LS LS
h
                                                                                                         W
                                                                                                                W
NOTE:  Test was conducted 7/14/89 to 11/15/89

-------
collection on August 3,  1989 showed positive at Mr. Gatti's and in
the recovery trench.  The rapid movement  of the  dye to these two
nearby sites demonstrates the relatively high ground water flow
velocity associated with the  solutionally enhanced upper bedrock
(the epikarstic zone),  which was the horizon of injection.

In  the  weeks  following the  initial confirmed  positive  at Mr.
Gatti's  and the  recovery   trench,   there  were  a  few  "false
positives".  These  false  positives  of  fluorescein  dye  can  be
expected when  tracing   in urban areas  since  fluorescein  in the
coloring agent  in many popular anti-freezes.  These false positives
are not a problem in a tracer  study of this design.  Since a large
amount of  fluorescein  was  injected  upon  the  epikarst,  one may
expect a  rather  extended dispersal  period over a  large lateral
area.   Additionally,  the  detectors were retrieved at  weekly
intervals and a positive was not assumed  until dye was detected for
at least four continuous weeks at any  particular site.  Thus, false
positives can be detected and noted accordingly.

There were several false positives during the course of this study.
For  example, detectors at  Big Elm,   Town Branch, and  McConnell
Spring showed  false positives on several occasions.    All false
positives in the above  sites were followed with negative readings
in  subsequent  weeks.    The  storm  sewer also  displayed  false
positives  throughout much  of  the  study  period.   This  site was
adjacent to a road with  heavy traffic, and undoubtedly encountered
many urban-induced fluorescein spills. These spills were not great
enough to  cause positive readings  at the railroad  storm sewer a
quarter mile directly downstream.

One focus of  this study was to determine  the ultimate resurgence of
the injected dye.   The  detectors collected on  September 26,  1989
marked the arrival of the first confirmed positive  of the dye at
McConnell  Spring  and Preston's  Cave Spring sites.   This first
confirmed  positive   occurred   some   67  days  following  the  dye
injection.  The connection of  McConnell Spring to Preston's Cave
Spring was  first  documented by James Rebmann of  the Lexington-
Fay ette Urban County Government in January 1988.

Dye  introduced at the  soil-bedrock  interface moved very rapidly
through the epikarst toward Mr. Gatti's and the  recovery trench.
The epikarst is noted for its  rapid ground water flow velocity and
wide lateral dispersion of  recharge.   After passing  through the
epikarst,  it appears that the dye  moved very slowly  through a
poorly developed conduit system toward  McConnell Spring.   After
passing through McConnell Spring,  the  dye resurged  at Preston's
Cave Spring.

                    Ground Water Contamination

Ground water contamination,  as detected  at Mr.  Gatti's and in the
recovery sump has decreased from the presence  of  free product in
April 1986 to dissolved  contamination in the low parts per million
range in December 1989.   Fluctuations in contaminant levels have
                               459

-------
been  documented  during this  time  period.    Increases  can  be
attributed to heavy periods of recharge where ground water moving
along  poorly developed conduits  in the  epikarstic  zone leach
residual soil contamination.  Decreases are marked by dry periods
where precipitation  has little effect in  leaching residual soil
contamination.

              REMEDIATION DESIGN AND IMPLEMENTATION

In  order to  properly  remediate  the subcutaneous  zone  of  this
location, an approach different from classical pump and treat was
needed.   The design had to  be specifically   matched  with  the
geological and hydrogeological conditions of the subcutaneous zone,
satisfy  state  requirements of  remeditaion  to  background,  and
satisfy  the  third party landowner that every  possible avenue to
remediation  was  being  attempted  using  the  fastest  technology
available.

When designing this recovery system we had to account for:

          contaminated soil above the soil/bedrock interface that
          was contributing to contamination;

          inaccessible contaminated soil in solutionally enlarged
          pore space of the subcutaneous zone;

          quickflow  of  contaminated  ground   water  in  poorly
          developed conduits of the subcutaneous  zone;

          observed increases in BTEX  concentration in ground water
          with an increase in recharge;

          available space and property boundaries; and,

          the possibility of volatile organic vapors in buildings.

SELECTED TECHNOLOGIES AND APPLICATION

                         Soil Excavation

Elevated levels of gasoline constituents were identified in an area
of soils west of Mr. Gatti's Pizza Restaurant as outlined in Figure
5.   Approximately 1960  tons of this soil  was  excavated from the
area  and transported to  the Lexington-Fayette  County Municipal
Landfill.

                          Soil Flushing

Soil flushing was selected for this site for several reasons.  The
nature of  flow  along the soil bedrock interface is likely guick
flow  through poorly developed  open  and  soil  filled solution
channels in  the subcutaneous  zone.   Levels of dissolved gasoline
components  in ground  water have  shown  a sharp rise  following
periods  of heavy recharge from precipitation.  Reinfiltration of
                               460

-------
treated ground water was designed to mimic a constant recharge to
this zone, thereby continually flushing contaminants from the soil
in the  epikarstic zone.  Soil flushing is also very cost effective,
provides an option for the fate of the treated ground water and can
increase and  enhance ground water flow to the interceptor trench.

An infiltration gallery was designed to accept and reinfiltrate the
volume  of discharge  generated by  the ground water recovery and
treatment  system.   At this site discharge will be the  result of
ground   water  recovered  from  a  maximum  of  two  continuously
operational recovery sumps.

The  purpose  of the  infiltration  gallery  is  two  fold.    First
reinfiltrated  water  serves  to  recharge the  perched  surficial
epikarst and move any ground water contamination to the interceptor
trench  more quickly-  Second,  reinfiltrated ground water  flushes
residual soil contamination at the soil bedrock interface causing
the leached material to  become mobile in ground water for removal
at the  interceptor  trench.

During  the tracer study conducted  July 1989, a  small  conduit was
discovered at the soil  bedrock  interface at a  location shown on
Figure  2.   This  conduit was easily  able  to accept at  least 100
gallons of water instantaneously.   This conduit was also traced by
fluorescent dye to  the location of the present recovery trench in
less than  24  hours.    The infiltration  gallery intersects  the
conduit in the epikarst therefore  providing constant  recharge to
the  surficial flow  system  and  constant  flushing  of  residual
contamination from  soil.

The gallery is 40  feet  long,  3 feet  wide and 7 feet deep.   The
discharge  water is  introduced by means of 4"  diameter perforated
pipe to allow  drainage  across  the entire trench.  The  trench is
backfilled with washed gravel to allow for even percolation of the
treated ground water.  Two  clean-out ports have been placed on the
perforated pipe to  remedy any sediment,  clogs  or biofouling.  Two
piezometers  were installed at  opposite ends of  the  infiltration
gallery to monitor  its  efficiency  in  reintroducing  ground water.
Emergency  high level floats keep the infiltration gallery from over
flowing if   its  capacity   is  exceeded.    A  schematic  of  the
infiltration   gallery  is  shown  in Figure  7.    The  gallery  was
installed  during  the second week of January 1991.   To ensure the
gallery was   hydraulically "connected"   to water  flow  in  the
epikarstic zone approximately 1000  gallons of water was introduced
to the  open hole after it was excavated.  The water quickly flowed
from the gallery  and into the subcutaneous zone.
                               461

-------
            ,— TREATMENT SYSTEM EFFLUENT IN

              /— PIEZOMETER
      PIEZOMETER
• CLEANOUT
t
3'
t
X / i
C) i CO • > •• • < ' • ' ' • • ' i '(

/ X
II 1 1 I 1 1 1 I t 1 1 1 1 1 l( }

\
• o

                              PLAN  VIEW
                  TREATMENT SYSTEM EFRUENT IN
                             CROSS SECTION
Figure 7. Infiltration gallery design.

                           Soil Venting

Soil venting was also a  selected technology for implementation at
this site for this site  for  several reasons.  Soil venting allows
for remediation of  the  residual  soil  contaminants  with minimal
disturbance of  surface structures  and  roads.  Soil  venting is a
cost-effective  and field proven technology  for  remediating soil
contaminated with petroleum hydrocarbons.  Application of a vacuum
to the interceptor trench should also increase the hydraulic yield
of the  trench and enhance  the  rate of  ground  water  remediation.
The primary reason,  through,  was to remove hydrocarbons  from soil
is the epikarst that could  not  be removed by excavation.
                                462

-------
The soil vapor extraction system consists of a regenerative blower
connected  via 2-inch schedule  40  PVC piping to  the ground water
recovery/vapor  extraction  trench.   The trench is screened through
the vadose zone to accommodate the vapor removal.   Depression of
the potentiometric surface due to pumping of the ground water also
increases  the volume of contaminated soil exposed for remediation
by venting.

Soil Remediation System Equipment

Figure 8 presents the layout for the equipment of the soil venting
system which consists  of  a  regenerative blower  connected to the
recovery  well via  the  previously  described PVC  piping,  an inlet
filter to remove particulates  from the air stream  entering the
blower, and  a  coalescing unit  to remove  moisture from  the air
stream entering the blower.   The effluent from the blower will be
discharged to the atmosphere via a 2-inch schedule 40 PVC pipe.
As the liquid level in  the coalescing unit reservoir increases, it
activates  a  sensor which turns off the blower and opens a solenoid
valve in the coalescer,  thereby allowing the condensate to gravity
drain into the diffused aeration  tanks.  A timer  in the control
panel reactivates the blower and closes the  solenoid valve after
all condensate  has drained.

GROUND WATER RECOVERY AND  TREATMENT

                 Ground Water Recovery Equipment
                                     \
The ground water  recovery system installed in  1986 at the Lexington
site  consisted   of  a  ground  water  interceptor  trench  and  one
recovery  sump.   The former interceptor trench was extended and an
additional recovery sump  be added  as  shown in  Figure 9.    The
ground water recovery system consists of two total fluids recovery
pumps installed in the  recovery sumps of the enlarged interceptor
trench.  The pumps intercept and pull  contaminated ground water
towards the  interceptor trench.   A  cross  section  of  the trench
construction is shown in Figure 10.

The static depth to ground water in the  recovery sump installed in
1986 was approximately  8 feet below grade.   Both recovery pumps are
positioned at  a depth  of  11  feet.   The depression  pump on/off
floats are   positioned at  11  feet  and  10  feet  below  grade
respectively in  order to  establish  a maintained drawdown  of
approximately 2 feet in the interceptor trench.   The operation of
the total fluids pump  is controlled by  high  and low level on/off
sensors  in  order to  facilitate  the   efficient  removal  of
contaminated ground water  from  the soil bedrock interface.

Ground Water Treatment  Equipment

The ground water treatment technology implemented at this site is
a three step process:
                                463

-------
                                  VACUUM GAUGE-
MANHOLE COVER
                           VACUUM ENHANCED RECOVERY
                           WELL
                                                                        COALESCER
                                                                                             -VENT STACK
                                                                                                 REGENERATIVE BLOWER
                                                                           AIR F10W
                                                                                                       BLOWER
                                                                                                       SUPPORT
                                                                 DRAIN LINE TO
                                                                 DIFFUSED AERATION TANK
                                                                               NOTE:

                                                                               THIS DRAWING IS NOT TO SCALE
                             Figure 8.    Soil venting system  schematic.

-------
               SOUTH LIMESTONE AVENUE
   SECOND

  NATIONAL

    BANK
1
CHECKERS
FOOD
MART


/ L
/
/
v/
i /

MR. GATTIS
PIZZA





                                                     Ld
                                                     <

                                                     G.
PARKING
                                                                  PARKING
                                                      TREATMENT SYSTEM
                                                         BUILDING
                                                   RECOVERY
                                                   SUMP
                                                                 APARTMENTS
                    PARKING
                 LIMESTONE
                 APARTMENTS
                                                      LU
                                                      3
                                                      Q.
                                                                 APARTMENTS
                                                                RESIDENCE
            FENCE
Figure  9.   Recovery trench  and  treatment  system  location.
                                    465

-------
APPROX.
 4'-5'
 APPROX. 1'-2'
           •|1 RISER
                            NATIVE BACKFILL OR
                         GRAVEL UNDER PARKING AREA
 5'
 \- 2" SLOTTED PVC
	 30'-0" •	
30'-0"
 d
         I
    APPROX. 1'-2'

                                                                    COARSE GRAVEL
                                                                                           -#3 RISER

                                                                                               RECOVERY WELL
                                                                                                  8" PVC
                                                                            r POLY-PLASTIC UNER
                                                                            \ (IMPENETRABLE)
8' —
                                                                                                 SCREENED
                                                                                                              5'-0"
                                                                                                                          S'-O"
                                         BOTTOM OF TRENCH
                                                     4" PERFORATED DRAIN PIPE     ''TiT'''TIT1
             NOTE:   DRAWING IS NOT TO SCALE.

                      ALL PVC IS SCHEDULE 40
                    Figure  10.   Recovery trench  construction  details.

-------
1)    particulate  filtration  as  a  pretreatment  step  to  remove
     suspended solids;

2)    LPH/water gravity separation as a pretreatment step to provide
     for the flotation separation of LPH and flow equalization; and

3)    diffused aeration to  remove the dissolved volatile organic
     compounds.

Pretreatment  Process

The  ground  water  pretreatment  system  consists   of  an  in-line
particulate filter followed by an LPH/water separator tank.  Figure
11 presents the treatment system lay out.

The ground water recovery pumps transfers ground water through an
in-line filter to a  1000  gallon LPH/water separator tank to allow
for  flotation  separation of the  LPH from the ground water and
particulate  settling.  A totalizing flow  meter installed  at the
intake of the  separator  tank continuously records  the  amount of
ground water  and LPH that has been pumped.

Effluent from the separator tank flows by gravity through a contact
chlorine chamber  containing  solid calcium hypochlorite  tablets.
This process  reduces biological build up in the treatment system.
 The  chlorination   step  inhibits xthe   growth   of   biological
microorganisms  that could potentially  cause  the  lines  in the
aeration tanks to become  clogged.

Diffused Aeration Equipment

Effluent from  the contact  chlorine chamber will  flow  by gravity
into a channeled  diffused  aeration tank.  This  tank will be four
feet wide  by  six feet long by two feet deep.  A  2 hp, 230V, single
phase pressure blower will introduce  air into  the aeration tanks
through perforated   PVC  pipes  to  facilitate   stripping  of  the
dissolved volatile organics.

Effluent from the diffused aeration tanks flows by gravity into a
120 gallon transfer tank.  This  transfer tank  is  equipped with a
submersible  electric sump pump with a nominal operating flow rate
of 10 gpm. The sump pump moves the treated ground water from the
transfer tank to an  infiltration gallery located  upgradient from
the interceptor trench.

    Soil Venting and Ground Water Remediation System Security

The soil venting and ground water remediation equipment are housed
inside a pre-fabricated  building located  at a  location  shown in
Figure 9.
                                467

-------
AERATION TANK-
                                  I I TREATED EFFLUENT TO
                                  I I INFILTRATION GALLERY
                                  I I
                                  I I
                                  I I
                                                         JL
                                                               EXHAUST VAPORS
CHIOR1NATOR
                                   EMERGENCY
                                   HIGH LEVEL
                                   FLOAT
                             TRANSFER
                             TANK WTO
                             SUUP PUMP
                                                             LEVEL SHUT-OFF PROBE
                                                                    D
                                                           CCTBEBVATON VENT
                                                               1000 GALLON
                                                               SEPARATOR TANK
                                                                    o
                                                              OBSERVATION PORT
                                                       SA14PUNC -
                                                       PORT
                                                       BALL VALVE-
                                                                        ^-FLOWUCTER

                                                                     I INLINE FILTER
                                                                                          AIR INLET
                                                                                        FROU RECOVERY
                                                                                           WELL
                                    Locxm
                                    SATE
                                                                     r—	——•———~—-\ FROU PRODUCT
                                                                     I	«==_.» RECOVERY WEU.
                  Figure  11.    Treatment  system  layout.
                                                 468

-------
                         RESULTS TO DATE

Analytical results from water samples  collected  from the former
emergency sump and the current ground  water  recovery system are
plotted against time in Figure 12, illustrating a wide variation in
contaminant concentrations  since  recovery operations  have begun.
Maximum dissolved concentrations of approximately 15,000 ug/L total
benzene, toluene,  ethylbenzene and total xylenes  (BTEX)  have been
recorded.   Although  flow readings  do  not exist for the  early
recovery system, a general correlation between recharge events and
increasing contaminant concentratons has been  made.  The emergency
recovery sump was  taken off line  from July 1989 through February
1990 to conduct  the dye tracer study-
       TOTAL BTEX CONCENTRATIONS  (ug/L)
                     SHOP & GO FOOD MART
    PQ
       05/07/90          11/23/90         06/11/91          12/28/91
               08/15/90          03/03/91          09/19/91
                                DATE
Figure 12.
Total  BTEX  concentrations   in  treatment  system
influent since May,  1990.
The current ground water remediation system was monitored monthly
to verify that the system is operating properly.   Sample ports are
provided at several  locations  along  the  treatment  system.   Water
from the treatment system influent and effluent sampling ports are
collected monthly, and analyzed for volatile organic constituents.
                               469

-------
from the treatment system influent and effluent sampling ports are
collected monthly, and analyzed for volatile organic constituents.
A totalizing flow meter is installed to document the total volume
of ground water being pumped from each recovery sump.

Plots  of flow  volume over  time from  both  recovery sumps  are
presented as Figures  13  and 14.   These  illustrations  verify the
flashy nature of  flow  in  the  subcutaneous  zone.   Flow rates vary
from recovery well 1 from almost negligible to approximately 5500
gallons per day.   Flow variations  from recovery well  2  are even
more dramatic ranging from 3000  to  almost 25,000 gallons  per day.
This information  also  correlates with periods of  observed  heavy
rainfall events through that time period.
                     RECOVERY  WELL  1
                   FLOW RATE (GALLONS PER DAY)
      6000
        03/03/91
       05/17/91
07/31/91
10/14/91
                                 DATE
Figure 13.
Flow in gallons per day from Recovery Well  1  since
system start-up.
A plot of BTEX concentration in the treatment system influent since
the currently system was put on line  included as Figure 15.  Figure
15 shows that the concentration of total BTEX varies widely but the
maximum measured levels have deceased since remediation began.  The
peak concentrations are also roughly similar to maximum  times of
flow  or  maximum   recharge.     This  may  suggests  that   more
contamination is being leached from  soils, mobilized and removed
during times of  heavy recharge and that the constant recharge from
the  infiltration  gallery  may  aid   in  expediting  this  removal
process.
                                470

-------
                 RECOVERY  WELL 2
               FLOW RATE (GALLONS PER DAY)
      03/03/91
      05/17/91
                          07/31/91
10/14/91
                            DATE
Figure 14.
Flow in gallons per day from recovery well 2 since
system start-up.
      TOTAL  BTEX CONCENTRATIONS (ug/L)
                   SHOP & GO FOOD MART
     10000

   U
   CJ
1000
 03/03/91
                        06/11/91
                            09/19/91
               04/22/91
                   07/31/91
                                            11/08/91
                              DATE
Figure 15.
Total  BTEX  concentration  in  treatment  system
influeratLnce system start-up.
                            471

-------
The performance of the soil venting system is illustrated in Figure
16 as a plot of soil vapor concentration since  the  system was put
on line.   Soil vapor concentrations were  at  a maximum when the
vacuum  blower  was  first  started  and  have  decreased  to non-
detectable since that time.   The reason  for the quick  decrease is
believed to be removal of BTEX laden soil vapor from the immediate
area of the vapor extraction equipment.  The system has since been
turned off to allow equilibration of soil vapor with BTEX on soil
particles. The system will then be put back on line and  a decreased
but similar pattern of soil vapor removal is expected.


           SOIL VAPOR  EHRACTION SYSTEM
               PHOTOIOMZATION DETECTOR READINGS
      250
      03/23/91
      05/22/91
07/21/91
09/19/91
                                DATE
Figure 16.
Photoionization detector readings of the soil vapor
extraction emissions since  initiation.

            CONCLUSIONS
The  likelihood  of a  release  of contaminants  directly into the
conduit network of a karst aquifers is low.  Most releases occur
within the overburden and move into the subcutaneous zone of the
aquifer thereby decreasing the quickflow components and  increasing
residence times in the intergranular overburden porosity and the
poorly developed  conduit  system of the subcutaneous  zone.   The
results of  this effort have  been  presented as  a site specific
approach to remediate the  subcutaneous  zone of a karst aquifer.
The remediation design was not arrived at using classical pump and
treat, intergranular flow manner approach  to the  problem and set
                                472

-------
aside the notion that this  karst  aquifer was not "remediateable"
as if it were conduit network contamination.  A remediation system
was designed around the specific characteristics of the surficial
subcutaneous karst aquifer to be effective as a remediation tool in
this setting.  The use of excavation, soil flushing, soil venting
and pump and treat has been shown effective at removing soil and
ground  water contamination from  the s  ubcutaneous  zone  as  a
decrease in  maximum contaminant  concentration  over  time.   The
system  also  appears to  be working  as  designed by  mimicking  a
constant   recharge  event   thereby  leaching   residual   soil
contamination  from  the   subcutaneous  zone  for  removal  by the
interceptor trench.  The  system is currently operating as designed
and data being collected for  a future  submittal  concerning its
efficiency and performance.

                            REFERENCES

Miller, R.D., 1967, Geologic Map of the Lexington West Quadrangle,
Fayette  and  Scott Counties, Kentucky;  United  States Geological
Survey Geological Quadrangle Map - GQ - 600.

                       BIOGRAPHICAL SKETCH

Scott A. Recker,  a  native  of  Cincinnati, completed  a Bachelors
Degree  in  geology at the University of  Cincinnati in  1985 and
received a Master of Science in hydrogeology from Eastern Kentudy
University  in  1990.     Scott is  'a  member   of  the  National
Speleological Society and his research^ centers around ground  water
flow in the subcutaneous zone of karst aquifers.  He is currently
residing  in  Charlotte,   North Carolina  and  is  employed   as  a
consulting hydrogeollogist  for Delta  Environmental Consultants,
Inc.

Scott Recker
3300 Circles End Road
Charlotte,  North Carolina  28226
(704) 541-8191
                                473

-------
            PETROLEUM HYDROCARBON REMEDIATION OF THE
              SUBCUTANEOUS ZONE OF A KARST AQUIFER,
                       LEXINGTON, KENTUCKY

                         Scott A. Recker
              Delta Environmental Consultants, Inc.
                    Charlotte, North Carolina


1.   Has the remediation effort been evaluated at the spring?

No, the spring showed non-detctable amounts of petroleum
hydrocarbons during two sampling events.  The spring has also
been documented as the ulimate resurgence for a ground water
basin which encompasses half of the City of Lexington.  Remedial
evaluations at the spring would therefore be unsuccessful due to
the large amount of dilution and the presence of other potential
petroleum hydrocarbon constributors.
                                474

-------
  CORRECTION OF BACKGROUND INTERFERENCE AND CROSS-FLUORESCENCE

      IN FILTER FLUOROMETRIC ANALYSIS OF WATER-TRACER  DYES

                          C.C. Smart,

                    Department of Geography
                  University  of Western Ontario
                             London
                        Ontario, L8S 4K1
                             Canada

                           P.L. Smart
                     Department of Geography
                      University of Bristol
                        Bristol, BS8 1SS
                            England.


ABSTRACT
     Filter fluorometry is a widespread analytical technique for
the  quantitative   analysis   of   fluorescent   tracer   dyes  in
hydrogeology.    However,  mixtures  of  tracers  and  fluorescent
background materials may result in ambiguous results unless steps
are taken  to isolate the tracer materials>  Analytically  it may be
possible to chemically suppress or mask interfering substances, or
to select  filter  combinations capable  of  excluding those  wave-
lengths  frustrating analysis.

     Post-analytically,  an  algebraic   solution   is   possible,
providing   care has  been  taken  to  run  calibration  standards
({y}=l,2,3) through  all filter  arrays ({x}=A,B,C)  to  obtain a
matrix of  fluorescence readings F*   The matrix  of cross- fluor-
escence  calibration coefficients  {k*} = F*/k* can  be  determined,
where z= A, B,  C  for  y=l,2,3  respectively.   By  definition, the
positive diagonal  kx values will always equal  one.   Off-diagonal
k* may be zero  (indicating absence of cross-fluorescence), or non-
zero (indicating cross-fluorescence and the need for correction).
Note that  exact  values  of k*  require empirical  determination,
because  they depend on characteristics of dye  stock,  individual
filters, and light sources.
                              475

-------
Table of Cross-Fluor-
escence Calibration Coef-
ficients {k*}
Filter
Set
(x)
A
B
C
Dye (y)
1
kA
K1
kB
K1
kc
K1
2
VA
K.7
kB
K?
kc
K?
3
kA
K3
k8
*h
kc
K^
                                                      If  no  cross-
                                                      fluorescence
     If tracer dyes 1, 2 and 3 are matched to filter sets A,  B and
C respectively and calibration allows  k*=k|=kij=l,  then  an approxi-
mate but general solution to the  cross-fluorescence problem can be
derived.  The actual fluorescence of dye 1 in a sample (FA)  can be
expressed in terms of measured fluorescence  of  that sample  in all
three filter sets (FA,FB,FC), and the coefficients  {k*} :
                       - FB(k* -
                            _L

                                      C-j jCii
                                                           (10a)
The  relationship  can be  further  simplified  for  two-component
mixtures, to produce the common correction equation  given  by  Kass
(1967) :
                                                             (11)
     Where  cross-fluorescence  coefficients
further simplification is possible:
are  small  numbers,
                                                             (12)
     The model will be robust for all fluorescent materials with a
consistent  spectral signature.   This  may allow correction  for
background  interference  from organic  materials,  provided  that
background variations are due to fluctuations in spectral amplitude
rather than  shifts  in  spectral character.  The  background "stan-
dard" used to determine k£ must be entirely free of tracer dye, but
can be of arbitrary concentration.
     The technique will not allow cross-fluorescence correction in
waters containing fluorescent contaminants closely mimicking dyes,
nor accommodate  dramatic shifts  in  background signature.   These
problems  may  be particularly  acute  in  elutant from  activated
charcoal, which  must remain  a semi-quantitative  technique.
     Where precise interpretation is required, a spectrofluorometer
remains the  instrument  of choice.  The  same method may be applied
to correct for cross-fluorescence.
                               476

-------
INTRODUCTION
     Fluorescent dyes are an  invaluable tracing agent, widely used
in surface and groundwater studies.  They are inexpensive, benign
(Smart 1982)  and  readily detectable at  concentrations  below the
parts per billion (ppb)  level.   It  is also possible to use several
dyes simultaneously,  allowing several components  of a hydrological
cycle to  be  monitored  under  a given  set  of  flow  conditions.
However,  in practice, multi-tracing can be complex,  expensive and
possibly ambiguous,  especially  if  tracers can not  be completely
separated from  one  another  (cross-fluorescence)   or  from natural
background  materials.      In practical  applications,  this  will
sometimes be seen as exact,  but highly  attenuated mimicking of a
primary breakthrough curve in a neighbouring spectral trace.  Where
the two tracers  have been injected simultaneously,  it may  be of
considerable  importance  to determine if the  attenuated  trace is
real,or simply the result of cross-fluorescence.

     The problem  is illustrated in  Figure 2 where  cross fluor-
escence from Dye I has  caused an increase  in the apparent fluor-
escence of Dye 2  (broken  line) .  The exact magnitude of the problem
will  depend  on  the  overlap of the  spectra and  the  relative
concentrations of the  two tracers.  Also  note that cross-fluor-
escence is  "non-commutative"; despite the interference  of  dye 1
with  measurements  of  dye 2, measurements  of dye  1 are little
influenced by concentrations of dye 2.

     The problem of  two-component  cross-fluorescence  in spectro-
fluorometry has received some attention in the  European literature.
The present paper first reviews a correction methodology and then
provides an  initial  attempt  to   formalise the problem  for multi-
component cross-fluorescence, with  particular emphasis  on filter
fluorometers.

     There  are  two  strategies  to  solving  cross-fluorescence
problems.   Pre-analytical  treatments are  designed  to   mask  or
suppress  fluorescence  of  interfering tracers  in mixed samples
without altering the fluorescence  of  a  selected  tracer.   Behrens
(e.g.  1982)  has  derived  an array  of  such  treatments   such  as
filtration and pH control allowing  spectacular segregation of dyes
prior  to analysis.    Further   segregation  may  be  gained  using
chromatographic separation techniques  (e.g.  Rochat  et  al. 1975).
In  general,   these  techniques  require  relatively  sophisticated
processing and although automated techniques may be adopted, remain
prohibitively expensive for most environmental investigations.

     The alternative  strategy where  moderate cross-fluorescence
occurs in mixtures of two or  three  tracers is  post analytical.  It
involves application of an algebraic correction  (Kass 1967)  based
on  the  theory  of  mixtures.    This  requires  very  simple cross-
calibration of the fluorometer using single dye standards.  Cross-
calibration  is  best performed  at  the time  of   analysis, but in
emergencies can be  done some time subsequent to the  analysis if
necessary.    The method  is straightforward and  inexpensive,  and
should be adopted in any multi-tracer test.


                               477

-------
n M) P) M
O M
MOW
M g ti w
fD fD O
Q cn o y
rt H- rt fD
  o 0)
  I rt
  H> H-

  H1 O
      O
      M
  O 0)
  g IQ
      O (D
      W
      cn en
      I ft (D
      HI o rt p)
      M 3 hj g
      c a H-
      O 0) O O
     (0 O- 0>
     cn w
         Hi
-fc-

^J

OO
        0) s;
 (D ro o
  P- O
 rt en g
 »•• -a
 (0  O
   3
 n 1-3 n>
 B) £T 3
 I- (0 rt
 H-
   o
        o
        cn
  •8
  o
      ft
      O
                 80
                 70-
  60H
CD

^

0)
O
C/)
0)
                 30-
                 20-
                 10-
                      CALIBRATION
                                         Dye  2
                         h
                                                                     B
                                    A
                                           B    Wavelength

-------
rt < i-3 rt rt cn N)
M Q> D- 3" *; T) .
C I-1 (D (D O (D
 C      O CO
  

?&3

°8S

83s
    n
   M 0)
   C rt
   O H-
c H n o
>-( 3" O
< to 3 a
CD  ft> h-
cn tr rt 0)
  If t-t iQ
rt o H-
(D 0)
  3

3 O
a c
(D l-(
^ <
M fD
       o p>
         3
       0)
       3 O
       0) H,
       en o
    1 3 3 O (D
    *Q a MI en
      H-  o
    en o o CD
    Ti D)  3
    ID rt *: o
T3 O
o rt

(D en '
3-a'
rt (D
en o '
•  rt JT n fo OJ m
  •1 2 rt m rt
  » * ^  fD en
  M ^ 0) rt i-t T3
   —  y  a>
  p" ^ o ro en o
  ro ™ HI  a> rt
  H- £'  3 3 i-i
 iQ ti rt fD >a C
  3- 2 31 0) H- 3
  rt .v fD en fD
  en m  e  o
   "  rt i-( O tT
  rt  « fD O rt
  O 2 O a 3 0>
        rt H-
  O "> a rt 0) 3
  cr —^ n M-fo
  rtj-31 fD 0) 3 a
  cu  en n M-
  H-  • fD 3 H)
  3   - iQ t-!
         o
         3
                          SAMPLE
                                              A
                                                                 B       Wavelength

-------
CROSS-FLUORESCENCE: SPECTROFLUOROMETERS

     A fluorescent  material absorbs  light  over a  characteristic
waveband and re-emits  light at a characteristically lower waveband.
In  fluorometric  analysis a  sample is  irradiated  (excited) with
light of wavelength within the absorption spectrum of interest, and
emittance measured at an exclusive longer wavelength.   The  key to
effective  measurement  is  provision  of  suitable,  but mutually
exclusive  excitation  and  emission  wavelengths.    In spectro-
fluorometers,  this  is  achieved by very  precise  control  of the
wavelengths using diffraction gratings.

     A  spectrofluorometric  analysis  is  usually  undertaken  by
scanning with excitation and emission wavelengths a fixed distance
apart (typically "  ^m) .  This produces a typical spectral trace of
overlapping  peaks  where  peak  position   is   characteristic  of
particular fluorescent materials and peak height its concentration.
Unfortunately, fluorescence is additive so that if the tail  of one
spectral peak underlie the peak of another,  the  peak  height  is no
longer a  simple measure of  concentration  of  a single  material.
This is  the  problem of cross-fluorescence.    In worse cases, the
overlap may  result in  complete masking of  smaller peaks  in the
flanks of more dominant ones.   The problem  becomes more acute at
shorter wavelengths when the very broad fluorescence  of  naturally
occurring organic materials may underlie tracer  peaks.

     In spectrofluorometry,  the problem of  cross-fluorescence is
usually readily apparent in a compound form  to  the spectral  trace.
For two component  mixtures,  it may be corrected using the method of
Kass (1967).  As there is no  formal statement readily  available on
the correction procedure,  it is summarised as follows.     First,
single tracer standards are run, noting the  height of  the peak and
shoulder at the characteristic peak wavelengths for both dyes.   A
matrix of trace height readings  is thus obtained for each dye (1,2)
at each diagnostic wavelength (A,B).  For example,  Figure 1  gives
an array as follows:
Calibration Measurements
Wavelength
A
B
Dye
1
h*=39.9
1^=5.40
2
h$=0.00
h^=53.2
The  convention  introduced here is  that  the superscript  (A,B...)
indicates the wavelength(s) used, and the subscript  (1,2,...) the
dye  of interest.

     Cross-fluorescence  correction  factors  are  the  ratios  of
spectral readings  to peak height.   They are thus independent of
concentration,  and are obtained  by dividing each  column by the
measurement obtained for  the  matched wavelength-dye pair  (Dye 1-
                                480

-------
wavelength A; dye 2-wavelength B).  Thus,
Cross Fluorescence
Calibration Coefficients
r vx j
4Kvl
Wavelength
(x)
A
B
Dye(y)
1
k*=i.oo
ki|=0.135
2
k*=0.00
kjHl.OO
The cross-fluorescence calibration factors are generalised as
{k*} where  x  indicates the wavelength (normally the spectral peak
for a dye), and y indicates the dye of interest.  Note that  0 <
k* <  1,   the  positive  diagonal  members will always be unity.
Off-diagonal values will be zero where there is no cross-fluor-
escence  (eg.  k*  above),  and non-zero  (usually small numbers)  when
cross-fluorescence occurs.   Ka'ss (1967) and others  (e.g Behrens
18982) employ a and 6 to indicate these correction factors,  but
their derivation may not be clear and their meaning can be
ambiguous.

     Figure 2 provides an example of a sample analysis, where hA
and hB are  measured  peak heights from a sample and other terms
are defined above.  The following estimates of actual peak
heights h1 ^ may be obtained by
                    i-kfkf
20.0 - 0.0 x 13.4
 1 - 0.00 x 0.102
                                              =20.0
(1)
i.e there is no cross-fluorescent interference with dye 1,
dye 2
                               For
                 hB -
                             13 .4  - 0. 135 x 20.0
                              1 - 0.00  x 0.135
                  = 10.7
(2)
In this case a 20% correction has been applied to the raw data.
A formal derivation of this simple case is developed below.

CROSS FLUORESCENCE: FILTER FLUOROMETERS

In contrast to spectrofluorometers,  filter fluorometers employ
glass and gelatine colour filters and light sources  (collectively
referred to as a "filter set") which are carefully selected to
optimise fluorescent sensitivity, although the spectral width of
both absorption and emission wavelengths are much wider and less
discriminating.   The problem of cross fluorescence  is thus more
likely to occur, and will be far less evident to the casual
analyst.   For this reason, it will always be worthwhile obtain-
ing cross-fluorescence calibration coefficients if more than one
                               481

-------
fluorescent tracer is likely to be encountered in a tracer  test.


     Measurement of fluorescence using a filter fluorometer is
shown in Figure 3 which may be contrasted to Figure 1.  The
filter cut-off is shown to be unrealistically sharp, this simpli-
fication in no way invalidates the analysis which follows.
A filter fluorometer integrates across a broad spectral band to
obtain fluorescence.  It can be seen that Dye 2 now has a small
overlap with spectral band A, in contrast to the complete exclu-
sion possible with a spectrofluorometer.
     Despite the increase in cross-fluorescence, the calibration
and analytical procedure is identical to that developed above.
First, calibration standards (of arbitrary concentration) are run
through both filter sets to get a matrix of fluorescence readings
Calibration Measurements
{F*}
Filter Set
A
B
Dye
1
FA=241.
FB=45.9
2
FA=1.1
FB=260
The calibration measurements are divided by the column matched
pair (FA and FB) to generate the correction factor matrix  {k*}.
Cross Fluorescence
Calibration Coefficients
te\
Filter Set
(x)
A
B
Dye(y)
1
kA=1.00
kB=0.191
2
kA=0.004
kB=1.00
These factors are then employed on sample analyses to correct for
any cross-fluorescence.  Figure 4 provides an example where FA
and FB are 313  and 196 respectively.   Substituting FA and  FB for
hA and hB  in eguations  4 and 5 above gives estimated sample  FA and
FB of 312  and  136 respectively,  with -0.2% and -30% change from
the measured fluorescence.  Fluorescence measurements may be
converted to concentrations in the usual manner.

     Colour filters, light sources and electronics are specific
to each filter fluorometer.  The same idiosyncracy applies  to dye
stocks, even from a single batch if poorly homogenised.   It is
not possible or desirable to transfer calibration data between
machines, components, dyes or over time.  A final caution con-
cerns the high temperature coefficient of rhodamine dyes  (Smart
and Laidlaw 1977).  Where sample temperatures differ from the
temperature of the standard used in calibration, temperature
measurements will be necessary on both A and B analyses, and the
                                482

-------
OO

LO
o> o g w cj
3 Q) 0)
Q- rt X" i-t)
   (D
  c o 1 g
w o i-< i a
ro M
  (D CO
  W rt c
ft O i-l o
fl> fl> O
   3
     D
X 3 3 6
ft rtiQ g
   fD (D iQ
HI ^1 i-J rt n
O (D  MO)
i 01 o p- a
 ~  o
  •O  0
    0> H)

    0) rt

    M§
    en
    P- O
    CO O


    'I
    0) W
    H-
    M O
    W 3
o a (-(
HIS; n>
 n> cn
rt en o
      a>
a>

O 0)
Q) 3
M a
H"
CT NJ
    ft 3 (D
      3 ff (D
      O (D 3
       0 O
       O ^
       X O
         M
       O W
       M) I
     f-h 0) 01 M
     H-H -a c
     I-1 (D (t> O
     rt o> o n
     fl> W (-•• fl>
     !-( HI cn
      TJ (-• O
     en*" o a>
     rt
     cn
    • 3
    • n
    (D

    H-
    3
                 80




                 70-




                 60-
                      0
                          CALIBRATION
Dye  2
                                     Dye   1
                                                                Waveband

-------
00
ft 0) w *>.
H- 3 0) •
O Q. 3
3 n in
 *i M o
rt " n> sr
o '  ID
  n 3
rt ^ rt H-
P) ID £U O
H- ,- H-
3 § 3 a
 2 P- H-
rt ^3 0)
        a o>
        *< 3
        (D
        tn o
        Ml
(D
cn
s»
ai
 rt
0) rr
3 (D
a
 Mi
     £•
        O
        3
     >Q 2
        H-
       (D O


       n> o>
       3 3
       ft 0)
      e O^
      r 3 0)
      H' |-J H-
      S^"
      m n o

      n ^ ""
      2 n> PI
      2 a
  m Q)
   rt
CD 20^
U

C

CD

U
00

CD
L_

O
                       SAMPLE
                                  Dye  1
                                               Dye  2
                                                       Waveband

-------
correction factor applied to the estimated fluorescence of the
rhodamine dye.
THEORY OP CROSS FLUORESCENT CORRECTION

     The methods above have a theoretical basis which can be
expanded to deal with a more general case of three dyes  (1,2,3)
and filter sets (A,B,C).

     Calibration of three dyes through three filter sets will
give a 3X3 matrix of fluorescence readings, which divided by the
matched dye-filter pair for the column will yield the following
general matrix of correction factors {k*}:
Table of Cross-Fluor-
escence Calibration Coef-
ficients {k*}
Filter
Set
(x)
A
B
C
Dye (y)
1
kA
K1
kB
K1
kc
K1
2
kA
K?
k8
K?
kc
x?
3
kA
K7
kB
KT
kc
KT
                                                     If no cross-
                                                     fluorescence
The total (i.e.  measurable)  fluorescence of a mixture of three
cross-fluorescent dyes in a particular filter set is the sum of
the matched dye  fluorescence,  plus the other dye fluorescences in
their matched filter set, corrected with the appropriate cross-
fluorescence calibration factor.   Thus,
                       FA =

(3)
                                                             (4)
                                                             (5)
     The quantity of interest is the positive diagonal F*,  the
actual fluorescence due to a particular dye in its matched filter
set.   Re-expressing 3-5 gives

                                                             (6)
                               485

-------
                           =F* - JtA - *BFC                    (7)
                           = Fc - JfcT - *CFB                    (8)
     It is necessary  to  remove unknown terms from these  express-
ions by substitution.  Taking (6)  as an example, substituting  (7)
for F| and (8)  for F3, collecting terms and then  resubstituting
gives the following general  expression
A     -       -  32   -     *
                                   - Fc(k
              FT =
                                ABAC
                           1 ~~ JCp/C- ~ /C-7/C*
     The product  of  three off -diagonal k*  will be negligible.
Ignoring such terms  (collected in the lower line of  (9)) not only
simplifies the expression,  but leaves the required fluorescence
expressed in measurable  and known terms.
                                 .*}FLit  ~~ jV-jjC*
The complementary  equations to solve for the true fluorescence of
the other two  dyes are  as follows:
                     F* -
                             J-  VC « -«^p   "^ ?
                                                           (lOb)
                              _ ,C B  _  B  C _ ^C, A

                                 2 ^'     V 2    1 2^        (IOC)
                             -L   /C i -K-z   "^?  "^
In two component  mixtures (e.g.  dyes 1 and 2) all terms  with
superscript  C  or  subscript 3 will equal zero, giving
In many cases  the product of two k£ will  be negligible,  thus
allowing a very simple cross-fluorescence correction.
                                 486

-------
     It is unlikely that a full three-component cross-fluor-
escence correction will be necessary where widely different dyes
and filter sets are employed (e.g. the Red, Green, Blue suite
outlined by Smart and Laidlaw,  1977).   It may be possible to
separate dyes previously rejected because of excessive cross-
fluorescence, although no experiments have been tried as yet.
The most common application will be for Rhodamine WT interference
with fluorescein sodium (i.e.  Uranine, CI 45350, Acid Yellow 73).
Very high rhodamine concentrations may mask a weak fluorescein
peak.   Alexander (1991, pers comm.) reports just such an occur-
rence which was only detected following correction for cross-
fluorescence.
BACKGROUND CORRECTION

     The single greatest problem in contemporary fluorometric
tracing is that of natural background fluorescence.  Although
seldom a serious problem with "red" dyes,  it is a common diffi-
culty with green dyes,  and especially acute with blue dyes.  This
consistent wavelength dependent sensitivity to background inter-
ference indicates that background has a characteristic spectral
signature (e.g. Smart et al.  1976).  Smart and Laidlaw (1977) and
Smart and Friederich (1982)  have employed this property to
establish the ratio of background fluorescence in dye-free water.
Significant departure from this line is an indication of a
positive sample.  A critical displacement of two standard devi-
ations from the line was suggested, although a statistical
confidence band might provide a more objective criterion.  The
technique will fail if the tracer dyes are present in a similar
ratio to the background fluorescences.

     The viability of the ratio technique suggests that it is
possible to correct for background interference with red and
green dyes by treating the background as a third dye component.
A third filter set, sensitive to background (an ultra-violet:blue
combination will be most suitable)  constitutes the C waveband
(Figure 5) .   A dye-free background sample is essential to cali-
brate the method, but the concentration for all cross-fluor-
escence calibrations is arbitrary,  so composition is less import-
ant than stability and representativity-  Fluorescence readings
are taken using the three standards in all three filter sets to
give a 3x3 matrix {F*}.
                               487

-------
00
00
        H> Q, (Jl
           3
           0)
         Hi ft
C (D
O 01 W
l-i O
(D 0) 3"
cn 3
o
(D

O
(D C O
 O

tn o> H-
 01 0>
M O *Q
0) (D l-l
cr 3 01
fo rt g

MCT o
(D (M Hi
d O
• ^i H>
W O ft
m c a>
n> 3 M
 a
ft • Hi

i> 
 O
a i-! o
(D O H)
rt cn
Q) 0) rt
           *:
           O
                     CALIBRATION
                     Background
                 o
                                                           B    Waveband

-------
     An example based on idealised data can be drawn from figure
5 to give
Table of Calibration
Measurements {F*}
Filter
Set
(x)
A
B
C
Dye (y)

243
44.1
4.75
2
1.50
269
0.00
3
85.2
6.53
238
     Division of each row by its positive diagonal element will
give the {k*} matrix.
Table of Cross-Fluor-
escence Calibration Coef-
ficients {k*}
Filter
Set
(x)
A
B
C
Dye (y)
1
1
0.182
0.0195
2
0.0056
1
0.0
3
0.358
0.0274
1
     Taking an imaginary example (Figure 6) of FA=292,  FB=117 and
Fc=340 and  applying these values with  {k*}  into equations lOa,
lOb, and lOc gives F*=169,  F*=76.9, and F^=337,  corrections of
-42, -34 and -0.9 percent respectively.   It is clear that the
magnitude of the background problem can be severe and probably
accounts for many cases of excess dye recovery estimations.

     The cross-fluorescence method of background correction is
the best developed to date, although it reguires considerable
analytical commitment to provide all the necessary data.  The
method is currently only theoretical and will require consider-
able testing and verification before its reliability can be
assured.  The simplest empirical indication of an error in
application is the estimation of negative fluorescence.  The key
assumption is that the spectral signature of the background is
steady and changes only in amplitude over time.  The influence of
spectral shift is unknown,  but may not be too significant if a
sufficiently broad wave band is employed in characterising the
background signature.  There is very little information available
on the pattern of background variation between springs and with
time at a single site.  Quinlan (1991, pers. comm.) suggests that
variations between springs are greater than variations at a
single spring.   This means it would be necessary to establish a
{k*} matrix  for  each  spring if  local stability  were assured.

     Tracer samples eluted from activated charcoal detectors
often exhibit the most serious background problem.  Unfortunate-
ly,  the concentration and exchange of organic molecules on tracer
                               489

-------
cr e rt oi
D) CO 3" D)
n (o (D 3
o
    O 3
    O 0)
    3 rt
    rt H-
 d ^
H- 01 2 o> o
3 01 5 H-
rt I 2 3 &
(D H> £ I-1- H-

t~h C ^ ^ ^Q
(D O rt  ^
(D (D
3 01
O O
(D (D
• 3
 rt
  5 « S
  3 o
     o
  S D- ">
  (D
    o>
     rt
     tr
 V^ A "-"
 O, (I) O. H)
 CD 2  *-•
 M g B) C

 ctgtr?
  m
    a
 n
 o
 o
 rt
 o
 n

 01

 o
 ff
   vQ
 01
  C O

  ' O> D)

  O O
  ' O M
  3 ><
    en
    3 01
    (D
    3 O
  -rtH,
 01
 (D

 (!) ... O
 hj ft '•
 (D
  •< (D
    i-(
            50
            40-
         (D
         U
         C 30
         0)
         O
         (f]
         (U

         020-
            10-
                 SAMPLE
                Background
                      C
                                            FA  \
                                                  \    Dye   2
                                                           B  Waveband

-------
samples eluted from charcoal detectors creates dramatic vari-
ations in background composition and resulting signature  (Quinlan
pers comm.),  so that the cross-fluorescence method will not be
applicable to this particular problem.

     Experiments are currently under way to test and evaluate the
cross-fluorescence methodology.  The results  and conclusions
will be reported following this completion of this work.

CONCLUSIONS

     Cross-fluorescence is a potentially serious problem in the
analysis of fluorescent tracers, especially with filter
fluorometers.  Cross-calibration should always be undertaken
during analysis to determine the extent of the problem and to
allow post-analytical correction if necessary.
     The three-component general theory of cross-fluorescence as
presented can be extended to more complex mixes with considerable
spectral overlap, providing care is taken to validate the
approach.  The method promises to provide a useful operational
correction for background interference, but substantial field
experiments will be necessary to assess the reliability of the
method.
REFERENCES

Behrens,  H.,  1982.  Verfahren zum qualitativen Nachweis von neben-
     einander vorleigenden Fluoreszenztracern. Beitrage zur
     Geologie der Schweiz- Hydrologie 28 I 39-50.

Kass, W. ,  1967.   Erfahrungen mit Uranin bei Farberversuchen.
     Steirische Beitrage zur Hydrogeologie, Jg 1966-67. 123-135.

Rochat,  J. J. Alary,  J. Molinari, R. Chariere, 1975. Separations
     physicochemique de colorants xantheniques utilises comme
     traceurs en hydrologie. Journal of Hydrology 26 277-293.

Smart, P.L.,  1982.   A review of the toxicity of 12 fluorescent
     dyes.  National Speleological Society Bulletin.

Smart, P.L.,  B.L. Finlayson and C.M. Ball, 1976.  The relation of
     fluorescence to dissolved organic carbon in surface waters.
     Water Research 10 805-811.

Smart, P.L.,  and H. Friederich, 1982.  An assessment of the
     methods and results of water tracing experiments in the
     Gunung Mulu National Park, Sarawak.  Cave Science 9 100-112.

Smart, P.L.,  and I.M.S. Laidlaw, 1977.  An evaluation of some
     fluorescent dye for water tracing.  Water Resources Research
     13  p.15-33.
                               491

-------
492

-------
            Session VI:
Ecology of Caves and Karst Terranes

-------
494

-------
                 ASSESSING GROUNDWATER QUALITY IN CAVES

                   USING INDICES OF BIOLOGICAL INTEGRITY


                             Thomas L. Poulson

                            University of Illinois
                              Chicago, Illinois



                                ABSTRACT

The states of populations and communities give an integrated record of past and
ongoing pollution  whereas toxicity Dioassays and chemical analyses may detect
pollution occurring  only  at  the moment  of  sampling.   Of the  population  and
community  biocriteria that  are  derived from field  data,  indicator species,
diversity indices,  and community similarity metrics are either totally flawed or
have more disadvantages than advantages.  The preferred biocriterion  is an Index
of Biological Integrity (IBI) which  combines  habitat data with population and
community data.   In caves there are  few species and so most must be  considered.
For fish, salamanders, and crayfish all  individuals are located on a map, measured,
and examined for reproductive  condition and lesions.  The smaller species and
habitat characteristics are sampled at  random  stations.  Since  surface waters
enter caves  by  different  routes,  I  sample  communities  from diffuse  input
formation areas,  from mixed  input areas like  terminal breakdowns, and  from
direct inputs associated with sinking  streams, vertical shafts, and backflooding
from surface rivers.  Through 30 years  of study, mainly in the Mammoth  Cave
Region, I  have documented or inferred the non-lethal effects on  populations and
communities  due  to homogenizing  of  habitat  structure  by si Ration,  due to
favoring of facultative over obligate cave species by excess food Input, and due to
compromising  reproduction  and   growth  of  very   long-lived  species  by
biomagnification of toxins along food chains.

                                    495

-------
  BIOLOGICAL MONITORING IS CHEAPER THAN CHEMICAL MONITORING AND GIVES A
          MORE INTEGRATED RECORD OF PAST AND ONGOING POLLUTION

      Chemical toxicity tests have now been flawed as a single approach because
they have been  done on a minority of  suspected toxicants, can not assess effects
of bioaccumulation or resulting indirect ecological  effects, use only a few test
species, and toxicity of one chemical changes as the overall chemistry of water
changes.   Increasing use of whole effluent  ("out of  the pipe") toxicity tests has
only partly  mitigated these problems.  Neither  chemical-specific  criteria nor
whole effluent tests assess impacts of habitat modification, such as water level,
or non-toxic  pollutants, such as of sediments and  nutrients.   Furthermore both
depend on sampling at the moment of pollution input,  unlike the case for  field
biological criteria.

      The composition  of  ecological communities  integrates  ongoing  and  past
pollution over space and time whereas toxicity tests usually only detect pollution
occurring at  the instant  of water sampling.   Yearly  biological surveys  often
suffice   to  demonstrate  ongoing sublethal  pollution  or past  pulses of lethal
pollution (as  with "midnight dumping") whereas these impacts are usually missed
by monthly or even weekly sampling of water  for  chemical and toxicity tests.
Thus biological  surveys should be used as an inexpensive early warning system  to
detect the presence  of a  problem.   Expensive chemical  analysis  and  toxicity
testing should be reserved to pinpoint the exact nature of a problem and to develop
discharge limits for particular pollutants. This combination of biological surveys
and  chemical tests provides a  complementary balance in the  protective and
regulatory process.

      In  karst regions monitoring water quality is more difficult  than in surface
waters because of problems of  inaccessibility and complications of  diffuse flow.
Quinlan and  Alexander (  1987) and Quinlan (1990)  show that  nearly continuous
sampling is needed to detect organic  pollutant pulses (e.g. the herbicide atrazine)
coincident with hydrograph peaks during conduit flow in streams.  It may be that
headwater point inputs  of organic  pollutants via sinking streams and  open
sinkholes on the Sinkhole  Plain  will be diluted before they reach  Green River
Springs  but  diffuse  input of   pollutants  can  linger   and  cause  insidious
bioaccumulation in cave organisms  (Barr  1976).     Diffuse input of pollutants
through the soil and epikarst into wells  and eventually into streams  is  slow and
complex  as Quinlan and Alexander (1987) demonstrate for an in organic fertilizer
(nitrate).
                                     496

-------
THE NEED FOR A COMMUNITY AND HABITAT BASED INDEX OF BIOLOGICAL INTEGRITY

      Of the biocriteria available toxicity bioassays, Indicator species, diversity
indices, and community similarity metrics are either totally flawed or have many
more  disadvantages than advantages.  The preferred biocriterion  is an index of
biological  integrity (IBI)  which  combines  habitat data  with  population  and
community data (Karr 1989).
Single Species Toxicity Bioassays

      EPA's acute and chronic  bioassays on Ceriodaphnia  and fathead minnows
have the advantage that they have been standardized and are rigorously analyzed
statistically but they are fatally flawed as single  best biocriteria for assessing
impacts of pollution (Cairns  1986,  Karr  et al.  1986).  Cairns points  out that the
central flaw in toxicity testing is the "myth of the most sensitive species".  One
corollary of this myth is the unwarranted assumption that  the responses of the
test species correspond to those of the many species not tested.  Another flawed
assumption is that there are no responses that are more sensitive than the 50%
lethal dose usually measured. For example  even the standard test organism may
have  sublethal responses more sensitive  than LC50,  such as  depression of
respiratory rate.   Such sublethal  responses in nature could change behavior of
predators  or competitors and  so  have  reverberating  efffects  throughout the
community. Still  another flawed assumption is that the cost savings using single
species tests more than compensates for the cost of failing to act when the single
species test does  not predict negative effects in nature.  In addition, Karr points
out (1988) that even in situ field toxicity tests are not sensitive to environmental
degradation due to non-toxic pollutants  like sediments  and nutrients.    In situ
toxicity  tests are   more  reliable at  detecting intermittent pollution  than
laboratory tests using water collected at one instant in time but the costs  of  in
situ tests and their practical difficulties are great. The most important practical
difficulty is inacessibility of the caged test organisms during flood peaks  when
we expect the  greatest  concentrations  of  many organic pollutants  (e.g. Quinlan
 1990).

      I  believe that there is little hope of  developing a "white mouse"  toxicity
test  using a cave species because  of the complexity of culturing cave organisms
and  the  conflicting  toxicological  results  using  field  collected cavernicoles.
Reliable culturing of a troglophilic amphipod has  been difficult (Dan Fong personal
communication) and nobody has successfully gotten an aquatic trogobitic species
through even one generation in the laboratory.  The one  toxicity test using EPA
protocols  on  field-collected  animals   (Bosnak  and  Morgan  1981)  had the
counterintuitive result that the cadmium LC50  for a troglobitic isopod was higher

                                     497

-------
than for a troglophilic isopod.  My explanation for this result is that the trogloblte
would be less sensitive to acute exposure to a toxin if  it had a lower metabolic
rate than the troglophile, as is consistent with the troglobite having a  16 fold
lower body concentration of cadmium with a  15 fold higher LC50 than  for the
troglophile.   Barn's acute toxicity studies at Mammoth Cave  (1976) are not very
sensitive because he used only  1000 ppM of  several  toxins  and  his  results are
equivocal.  Given the complexities in his own data and the conservation problems
of using  so many individuals for tests,  Barr recommended to NPS that no more
toxicity tests be done.
Problems of Using a Single Indicator Species in the Field

      Use of a single indicator species or even a single genus as a field indicator
of adverse impacts has fewer advantages and some of the same disadvantages as
just discussed for single species  toxicity testing.  Presence  or absence of an
Indicator is a simple  criterion but  there are many problems of  interpretation. In
addition to  the same problems as  with  the  "myth of a most sensitive  species"
there is the problem that a species may be absent due either to  pollution or for
natural reasons.  One natural constraint is inability to  live in some habitats. For
example shrimp do not live at upper levels or even at the headwaters  of  shaft
master drains ( Leitheuser et al. 1982)  for reasons that  are not clear.  Another
natural constraint is of predation or predation mediated competition. For example
the isopod  Caecidotea stygia  does not occur in lower level  habitats  possibly
because of  predation by  crayfish and/or fish but  this does not appear to be a
problem for Caecidoteabicrenata (Lewis and  Lewis 1980).  In other situations
food and predation may interact to affect the distribution and  abundance of a
species. An example is of isopods  in three unimpacted master shaft drains under
Flint  Ridge (Poulson   1968).   Isopods  are  most  abundant where  there  is  a
combination of  gravel  and  rock cover,  that mitigates the  impact of fish and
crayfish predation, and a lot of fine particulate organic matter, that serves as
food.
Interpretation of Species Diversity is Difficult Because Not All Species are Equal

      The number of species present, species richness, is easy to understand and
the  the  diversity of species,  H, is easy to calculate  but  both  have critical
problems with measurement and/or interpretation.

      As a  single biocrlterion  species  richness is especially  sensitive to the
recurring problem of standardizing sampling effort and efficiency.  Just how long
                                     498

-------
should one sample over how large aan area to be sure that one has found all the
species present?

     The information theory index of diversity H (see Poulson and Culver  1969
for a detailed description)  is  not  so  sensitive  to sampling effort  but its
interpretation is made difficult  because it  confounds the separate effects of
numbers of species and their relative abundances.  Because of this problem one
should report both numbers of species and their relative abundances and not use
the single index H.

     Quantitatively both species richness and H  are flawed by the  implicit
assumption that all species are equal in their effects in a community.  In  fact the
potential impact of a  species  is measured  by its  absolute  importance value
(Poulson and Kane 1981). Importance Value of a species is the  composite of its
frequency, density, and impact per individual. A very  frequent species, found  in
all habitats and at all times, is more  likely to have an  impact  than a rare  species.
A species which is locally very  dense could have a high local  impact even if it is
not frequent in space and time. And, a species which is large relative to others in
its community could have a great impact even  if it is infrequent  and  has a  low
density.

     Qualitatively  both species  richness and H are  flawed  by  the  implicit
assumption that communities with natural and exotic species are equally
indicative of biological integrity.  In the case of caves this is equivalent to saying
that  two communities are the same if one has many accidentals and troglophiles
and the other has only troglobites, as long as  both communities have the same
number of species and the same H.  An example shows that this is patent nonsense.
Clearly the pollution stressed Hidden River Cave community dominated by tubificid
sewage  worms and  several  large protozoan  species with    high numbers  of
troglophilic crayfish and rare unspecialized troglobitic amphipods (Crangonyx} is
very  different than the unimpacted Eyeless Fish Trail under Flint  Ridge with a
regular troglobitic cave fish and crayfish, one or two common copepods,  uncommon
Isopods, and rare but specialized troglobitic amphipods (Stygobromus}.
Measuring  Statistical  Similarity  of  Community  Species  Composition  Has
Advantages but it is Not A Good Single Biocriterion

      The simplest measure of community similarity uses only presence-absence
data and and is calculated by dividing the species in common by the sum of species
in the two communities to get a percent similarity.   This measure easily detects
the differences between a badly impacted community such as in Hidden River Cave
and an unimpacted one such as in Eyeless Fish Trail.  However it does not give an
                                    499

-------
early warning sign if there are slight changes In  community composition and or
changes in abundance of species without changes in species composition.

      Multivariate statistical techniques, such as Principal Components Analysis,
use both species identity and abundance data to group species along two or more
axes. The axes often do not  correspond to any simple environmental variable and
so the species clusters can be difficult to  interpret ecologically.   None-the-less
statistical changes  in species position in a cluster or in location of a cluster,
compared to  the  natural  variability  seen  from  time  to time  in  baseline
communities,  could signal  subtle changes and provide a good early warning sign
before negative impacts proceed too far.
An  Index  of  Biological   Integrity  Keeps  the  Advantages  and  Avoids  the
Disadvantages of Other Biocriteria and An IBI Has Unique Advantages

      A  maximum  value for  an IBI  Indicates  that  a sampled  area has  the
"capability  of   supporting  and  maintaining a  balanced,   integrated,  adaptive
community having a species composition and functional organization comparable
to that of the natural (unimpaired, original) habitat in the area"  (Karr et al. 1986).
The EPA now mandates use of both a fish and a benthic macroinvertebrate IBI in all
surface stream surveys.  An IBI  includes many different kinds of criteria that are
ecologically meaningful and so it is sensitive to a wide array of differing insults
to the natural environment.  An IBI does a particularly good job of integrating, over
space and time, the effects of past and ongoing negative impacts. An IBIs detailed
interpretation depends on the professional judgement of an ecologist because the
use of a variety of  qualitative and quantitative measures does not  allow for
statistical treatment. However the methodology for measuring each  metric used
is clearly enough explained that anyone can be trained to use the IBI.

      The  multiplicity  of  criteria  used  to  calculate an IBI  include  major
categories of habitat heterogeneity, energy source, numbers of species and guilds,
and well-being of each species.   Available IBIs and my proposed cave IBI give
positive  points  to  sensitive natural species like cave fish and   give negative
points to troglophiles, exotic species, and pollution tolerant species not normally
found in  the community.  The two scales of habitat heterogeneity measured are
also applicable  to caves. At a large scale one measures characteristics  of the
stream itself such  as size metrics, meander length, presence of undercut banks,
and the proportions of pool, raceway, and riffle habitats.   At  a finer scale one
measures the sediment size distribution on the bottom from bedrock and cobble,
through gravels and sand,  to silt  and clay.    In  a    cave  there   is  no  plant
photosynthesis and  so the  relevant  measure of energy input  is  the amount  of
coarse and fine  particulate  organic matter and the amount of non-toxic dissolved

                                     500

-------
organic matter. Most of the energy will be on or in the substrate and so I have
used weight loss on  ignition at 550C as an index of total organic matter in  the
sediments (Poulson 1968).  Simple indices of well-being can be applied to most
cave species. A species is doing well if it is abundant and has a size distribution
that includes the small individuals indicative of recent reproductive success.   In
fish a  rounded  belly and subcutaneous  fat deposits  are good signs  and  in
translucent crayfish the presence of yolked eggs in the ovary is a good sign.
           A CAVE IBI WILL INCLUDE MOST KINDS OF AQUATIC SPECIES
                     AND SHORELINE TERRESTRIAL SPECIES

      Cave ecosystems have few species and so currently EPA approved I Bis for
fish (Karr et al. 1986) and for benthic macro invertebrates (Ohio EPA 1987) cannot
be used or even simply modified  for use in caves.  The major  reason for low
species diversity in caves is that lack  of photosynthesis  means there is a small
energy base for food webs.  The problems of low food supply mean that most cave
species must be generalists in both feeding and habitat use and this  in turn makes
it difficult for new species to invade unimpacted cave communities.  For example
cave crayfish feed as detritivores,  scavengers, and predators and cave isopods
occur in fast and quiet water and on all  kinds of substrates.
All Kinds of Species Will Be Sampled

      Because there are few species, a cave IBI must include all kinds of easily
sampled  organisms.   Field census without collection  is  easily done for fish,
crayfish, shrimp, and flatworms and can be done for isopods and amphipods if
species identification is not  critical (Poulson 1968).   Censuses  of  the larger
shoreline terrestrial  species  is also easy in the field.    Baiting and trapping
techniques  will be used where organisms are rare, where much of  the habitat is
inaccessible, and where  turbidity  and high  water  preclude  visual observation
(Cooper and Poulson  1978,  Cooper 1975, Poulson  and Culver  1969).   A high
intensity diving light will  greatly increase efficiency of census any time that
waters are  slightly turbid and/or deep.
Detailed Data Taken On All Individuals Of Large Species

      For extremely  long-lived  species  with  slow   growth  and  infrequent
reproduction (fish Poulson  1963, crayfish Cooper  1975,  and shrimp Leitheuser et
al. 1982) the entire accessible stream will be searched and all individuals will be
measured. Size  distributions are a good index of current and past reproductive
                                     501

-------
success, as explained earlier.   To check for the presence of fin erosion, sores or
tumors, and ectoparasites the fish must be examined in a translucent  lucite box
for a few minutes.   Even more detailed data will be taken on crayfish  because
their hard exoskeleton protects them during handling.  It is easy to sex them and
assess male reproductive condition (form I vs form II).  In addition the translucent
exoskeleton of the  trogobitic species  allows detailed  assessment of female
reproductive  status  even  by observing  them  in  the stream  (Jegla and  Poulson
1970). Unlike crayfish, shrimp are too delicate to handle but they are transparent
and  so  status  of  ovarian  or hatched  eggs  is  observable  in  the  stream  with
appropriate snorkel or SCUBA gear ( Leitheuser et al.  1982).

Stratified  Random   Sampling  To  Census   Smaller   Organisms  and  Habitat
Characteristics

      Smaller organisms have shorter generation times than fish, salamanders, or
crayfish and so are better  indicators of short term changes in water quality.  Their
more cryptic  habits and higher densities require  that random station be sampled
[Permanent stations are useful for short  term studies of life cycles (Lewis  1982)
but can  be misleading when there are shifts of organism microdistribution  as
locations of substrates or food shifts after periods of high stream flow.]  At each
random  station  aquatic   macro invertebrates  are  censused   stream  habitat
characteristics  measured,  abundance  of  surface  particulate organic matter
scored, large shoreline terrestrial species  censused, and  tiny terrestrial species
trapped. Both  large  aquatic  and terrestrial  species  will be  censused  using a
combination of timed and areal searches to allow comparison between areas with
abundant and  sparse populations.   All  microhabitats will be included  in the
censuses.  If there is a stream current  then mini-surber samples will be  taken
using a small silk aquarium net (in my experience this  picks up copepods as well
as some large protozoa and some bacteria ).   Habitat characteristics will  be
assessed just as discussed earlier for Karr's fish IBI  but with special attention to
refuges from floods such as undercut banks, breakdown, and rocks and gravel  in the
streambed.  Cores of the stream and/or shoreline  sediments may be taken for
weight loss on  ignition and to assess the presence of lenses of particulate organic
matter that could  be exposed by flooding. Often there is no input of new leached
organic matter for years and so redistribution of the low  quality old particulate
matter may be  important.
  BASELINE IBIS MUST BE DEVELOPED AT ECOREGION, SUBREGION, DRAINAGE BASIN,
                        AND LOCAL WATERSHED SCALES

      For  karst, contrasting ecoregions are  the Valley-Ridge  of  the central
Appalachians vs the Interior Low Plateau in the Mammoth Cave area. Cave streams

                                     502

-------
have different characters In these two ecoreglons  and so there are different
baseline faunas (e.g. Culver 1982).  Within the Mammoth Cave ecoregion southeast
of Green River there are two subregions, the Sinkhole  Plain and the Mammoth Cave
Plateau.   These  subregions  differ  greatly  in  geology,  hydrology,  aand cave
development (as reviewed by many authors  in White  and White 1989).    Each of
these areas has its characteristic communities (Barr and Kuehne 1971, Lewis  and
Lewis 1980, Lewis  1981, Poulson  1968, Poulson  and Culver  1969).   In addition
there is a community that includes shrimp.  It appears to be restricted to  the
interface of downstream flows and backflow, during times of flood from the Green
River, through alluviated springs (Lisowski and Poulson 1979, Duchon and Lisowski
1981, Leitheuser et al. 1982).  The still different Hilly Country to the northwest
of Green River is now being studied  hydrologically but has scarcely begun to be
studied biologically.
           DIFFERENT COMMUNITY RESPONSES TO NATURAL FOOD INPUTS AND
                   DIFFERENT CATEGORIES OF POLLUTANTS:
                      A SERIES OF NATURAL EXPERIMENTS

      In this final section I review what baseline data we  have so far to use in
developing IBIs and discuss what natural experiments can tell us about community
change with categories of pollutants that include  sediments, human  and  animal
wastes, pesticides, heavy metals, and a potpourri  associated  with oil and  gas
exploration.  More detail is provided  In the unpublished Proceedings of the 1990
Mammoth Cave National Park Karst Conference.

            General  kinds of  responses  of   cave communities  to different
categories of disturbance have been considered elsewhere.  Thus Poulson and Kane
(1977) discuss the general  impacts of removing single  species, simplifying
habitat structure  by siltation, modification of  the  timing  and  extent of  floods,
favoring  facultative  cave  species (troglophiles)  by excess  nutrient input,  and
stress on long-lived  obligate cave species (trogiobites) associated with toxins
that are biomagnif fed along food chains.
Natural Variation Among Vertical Shaft Master Drain
Streams Under Flint Ridge

      Vertical shaft drains and master drains under the Mammoth Cave Plateau are
almost certainly in a near natural state and base level streams under Flint and
Mammoth Ridges are probably also natural since some receive little input from the
Sinkhole Plain.  Thus data  on these streams (especially Poulson 1968) are a  solid
                                     503

-------
start for developing I Bis to use In assessing impacts on more impacted baselevel
streams under Joppa Ridge.

      The diversity  of fauna among the streams is  correlated with diversity of
habitat and kinds of food supply. The stream with the highest diversity of aquatic
species and the least census to census variation  is the  longest stream with both
the greatest diversity of the food base and the greatest diversity of habitats. Thus
the right habitat combination for each species is to be found somewhere  at all
times. In contrast the stream  that is smallest in extent has a low diversity of
habitats.  Even though  it  has  by  far the greatest  particulate organic matter it
also has the greatest variation in aquatic organism density as the stream varies in
extent among years.  On the other hand the variation in the terrestrial fauna is the
least of the four streams.  The veneer of coarse particulate organic matter on the
ceiling provides a virtually unvarying food supply that supports by far the richest
terrestrial fauna of the four streams and  the  highest  densities of troglobitic
millipede  (Scoterpes copei) and harvestman {Phalangodes armata} that I have
censused anywhere In the Mammoth Cave Area.
Tradeoff of High Food  Supply  and Low Predation  Pressure  with  Low Habitat
Stability in Upper Level Seeps, Drip Pools, and Shaft Streamlets

      In the caves under the Mammoth Cave  Plateau the upper  level aquatic
habitats do not  have  a  diverse  fauna  but  an  IBI should be developed  for  them
because they are likely to be the first areas to be impacted by any local pollution
on the plateau and their waters often reach the shaft stream complexes at lower
levels.   In general, whether impacted or not,  these upper  level  habitats are
dominated by amphipods and f latworms that have adaptations for surviving in the
mud when pools  or seeps dry up annually and can take advantage of high seasonal
inputs of bacteria and fine particulaate organic matter. When there is permanent
water isopods are added (Caec/cfotea stygia, Lewis  1982)   and  a  troglophilic
amphipod may occur if there is unnaturally high organic input (Crangonyxpackardi,
Lewis 1987).
Cave Base-Level Community Compromised By Historic Alterations Of Green River
Discharge Patterns And Depth

      The  combined effects of  three  dams  have compromised  the base-level
communities,  of  the  Styx-Echo River  areas  of  Mammoth  Cave,  through a
combination  of  increased  siltation  and decreased  food  availability.    This
conclusion is based on historical records pre and post installation or Lock and Dam
*6 downstream on Green River and on biosurveys of  fish, crayfish, and  shrimp
                                     504

-------
during development of negative Impacts of the upstream dam  In the 60s and 70s
(Barrand Kuehne 1971, Llsowskl and Poulson 1979, Duchon and Llsowskl 1981) and
continuing to present.  A main  focus in the 70s was the apparent decline of the
Kentucky cave shrimp, PaJaemoniasganterl', in the Styx-Echo River area, a pattern
that has now been borne out by detailed  study  of  the  shrimp's distribution and
abundance (Leitheuser et a).  1982).
Short-lived Species May Be More Sensitive To Acute Exposure To Toxins
And Long-Lived Species More Sensitive To Chronic Exposure

      Cave communities will respond differently to acute and chronic exposure to
heavy metal  or organic toxins depending on whether the exposure is acute and/or
chronic.  Single or repeated acute exposures are  likely to kill  small, short-lived
species with relatively high metabolic rate.  This will decrease or eliminate the
food supply for the larger, longer-lived and lower metabolic rate species of upper
trophic levels. The early warning signs of this scenario are easy  to detect from
biosurveys and IBIs. More subtle and Insidious are the effects of sublethal chronic
exposure to  toxins because the early warning  signs are first seen in a further
decrease in  the  normally infrequent  reproductive success  of  fish  and  with no
elimination  of species.   With chronic toxin pollution   the  cave  fish  will  have
increasing body loads of  toxins due both to bioaccumulation associated with long
life and to biomagnification along the food chain.  If we are lucky  there will also
be obvious  signs  of  problems  such as  with  the hemorrhaging  and  impaired
swimming of "broken-back syndrome" in cave fish (Keith and Poulson 1979). Cave
crayfish are less likely  than  cave fish to be  affected by chronic organic toxins
because they store less fat and less likely to be killed by severely acute exposures
because they can leave the stream and so decrease exposure as was observed to
occur in summer 1979 with an acute hydrocarbon exposure in Hawkins River.

      If there are changes in IBI or individual species well-being due to suspected
toxin exposure then  identification of  the culprit(s)  will  depend on chemical
testing since all  toxins  can  kill  both  by  acute  exposure and by  chronic
bioccumulatlon and biomagnification during prolonged sublethal  exposures.
Inorganic Fertilizers  Affect  Cave Communities Only If  They Stimulate Primary
Productivity Of Surface Waters Which Are Subsequently Diverted Into Caves

      There is no photosynthesis in caves so the stimulating effect of inorganic
fertilizers, especially phosphates and nitrates, on aquatic plant growth can only
affect cave communities  indirectly.   Impacts  will  occur when algal  choked
sinkhole  ponds  overflow  into  caves  or  when  the  sinkhole  pond  disappears
                                    505

-------
underground as Its drain suddenly opens. A natural experiment that simulated this
scenario was of a catastrophic leakage of the Job Corp sewage  treatment lagoons
into vertical shafts that feed  Eyeless Fish Trail under Flint Ridge in 1967.  This
was a one time event and did not cause a detectable change in the distribution,
abundance, or well-being of cave fish or crayfish (Poulson 1968 and unpublished)
but it had a sudden depressing effect on the species diversity and abundance of the
shoreline terrestrial  community (Poulson and  Culver  1969).   This depressing
effect was presumably due to toxic effects of the  blue-green  bacteria Spfru/ma
that was left as a band along the shore as the pollution laden crest of water
passed and receded.  This influx of sewage lagoon  waters caused  the  substrate
organic  content to increase from 1.05-1.23% to 2.44-3.06%.
Human And Animal Wastes Impact Cave Stream Communities Mainly By Favoring
Species That Tolerate Low Oxygen And Are Reproductively Stimulated By Increased
Food Supply

      The destruction of the cave  community in Hidden River  Cave was due
initially to a combination of past toxic effects of heavy metals  pollution  (Barr
1976)  and  the high biological oxygen demand  of  decomposing cheese whey and
human wastes (Austin personal  communication) but  the current  slow recovery
seems due to residual effects of human wastes.  My conclusion  is based on natural
experiments in other caves where the only pollution is by human or  animal wastes.
Holsinger (1966)  and Lewis et  al. (1982)  provide qualitative  observations,
Weingartner (1977) provides detailed quantitative data,   and  I have preliminary
quantitative data  on different degrees of  impact.   At  high  impact  levels, as
directly under a  feed  lot or a large sewage treatment plant the high BOD kills all
macroscopic organisms and  leaves only strands of colonial  sewage bacteria and
associated protozoa.  If the BOD does not completely remove  oxygen then tubificid
sewage worms become part of the community as has recently been  the case in the
south branch of Hidden River Cave (Lewis et al. 1982, Poulson unpublished).  If the
amount of wastes is  not too great, as with septic  field diffuse  input or with
dilution by less polluted water as in east branch of Hidden River, the sewage  fauna
drops out and the higher than normal food  supply favors survival and reproduction
of  shorter-lived  macroscopic  cave  fauna,   especially  troglophilic  isopods,
amphipods, and crayfish  which may replace troglobitic species.  As the input of
waste becomes less  accidental chironomid midges and troglophiles survive, but do
not reproduce, and reproduction of short-lived troglobitic isopods and f latworms
is stimulated. At still lower impact levels the reproduction of larger troglobites
may also be stimulated.  Far downstream from a presumed   feed  lot input into
Black River of Roppel Cave there are incredibly dense populations of cave crayfish
(Leitheuser personal  communication)  and by  far  the highest densities of
meiofaunal  nematodes  (280/liter  of sediment) and   harpactacoid  copepods

                                     506

-------
(858/llter of sediment)  measured  anywhere by  Whitman  (in Leitheuser et  a).
1986).   High meiofauna) densities  may be an  early warning  sign of increasing
human  and  animal waste pollution because  normally   the meiofaunal  species
diversity and density decreases downstream from vertical shaft inputs.
                            ACKNOWLEDGEMENTS

      I am indebted to a host of colleagues with whom I have discussed aquatic
cave ecology over the past 30 years.  My interpretations  do not always match
theirs but I could not have generated my ideas without their verbal input and their
data.  I cannot mention everyone who has helped to form the ideas expressed herein
and/or has helped me in the streams  in  the Mammoth  Cave Area  but  a few  bear
special mention. In alphabetical order they are Tom Aley, Bill Austin, Tom Barr,
Nick Crawford, Dave Culver, John Cooper, Dave Griffith, Jack Hess, Tom Kane, Jim
Keith, Terry Leitheuser, Jerry Lewis,  Ed Lisowski, Eric Morgan, Rick Olson, Jim
Quinlan, Stan Sides, and both Bette and Will White. Finally I  could not have learned
what I know without the continued cooperation of landowners and the help  and
encouragement of the staff at Mammoth Cave  National Park  and  of the JVs  and
members of The Cave Research Foundation. I thank you one and all!
                             CITED LITERATURE

Barr, T. C. Jr. 1976. Ecological effects of water pollutants in Mammoth Cave.
      Final Technical Report to NP5.  Contract *CX-5000-50204
Barr, T. C. Jr. and R. A. Kuehne 1971. Ecological studies in the Mammoth Cave
      System of Kentucky. II. The ecosystem. Annales de Speleologie, 26(1): 47-96.
Bosnak, A. D. and E. L Morgan 1981. Acute toxicity of cadmium, zinc, and total
      residual chlorine to epigean and hypogean isopods (Asellidae).
      National Speleological Society Bulletin, 43: 12-18.
Cairns, J. Jr. 1986. The myth of the most sensitive species: Multispecies
      testing can provide valuable evidence for protecting the environment.
      Bioscience, 36(10): 670-672.
Cooper, J. E. 1975. Ecological and behavioral studies in Shelta Cave, Alabama,
      with special emphasis on decapod crustaceans.
      Ph.D. in Zoology, University of Kentucky. 364 pp.
Cooper, J. E. and T. L. Poulson 1978. A guide for biological collecting in caves.
      Caving Information Series of the National Speleological   Society.  14 pp.
Culver, D. C. 1982. Cave Life: Evolution and Ecology.
      Harvard University Press, Cambridge MA, USA. 189 pp.
                                    507

-------
Duchon, K. and E. A. Llsowski  1980. Draft environmental assessment of Lock and
      Dam Six, Green River Navigation Project on Mammoth Cave National Park.
      Cave Research Foundation, Dallas TX
EPA, Ohio 1987. Biological criteria for the protection of aquatic life. Vol. 2.
      Users manual for biological  assessment of Ohio surface waters.
      Division of Water Quality Management and Assessment, Columbus OH
Holsinger, J. R. 1966. A preliminary study on the effects of organic pollution
      of  Banners Corner Cave, Virginia.
      International Journal of Speleology, 2: 75-89.
Karr, J. R., K. D. Fausch, P  L. Angermeier,  P R. Yant, and I. J. Schlosser 1986.
      Assessing biological integrity in running waters: A method and its rationale.
      Special Publication 5. Illinois Natural History Survey, Champaign, IL
Keith, J. H. and T. L. Poulson 1979.  Broken-back syndrome in Amblyopsis
       spelaea, Donaldson-Twin Cave, Indiana.
      pp 45-48 in Cave Research Foundation Annual Report
Leitheuser, A. T. and various authors. 1982-1986.  Phases I-VI.  of Ecological
      analysis of  the Kentucky cave shrimp Palaemoniasganteri Hay,
      Mammoth Cave National Park. Old Dominion University Research
      Foundation, Norfolk VA. NP5 Contract * CX-5000-1 -1037.
Lewis, J. L. 1981. Observations on  aquatic communities in the Historic Section of
      Mammoth Cave,  pp  17-19 in Cave Research Foundation Annual  Report
	1982. The life cycle and distribution of two troglobitic Caecidotea in
      Mammoth Cave National Park.
      pp 10-11  in Cave Research Foundation Annual Report
	1987. Aquatic communities  in the Cathedral Domes section of the
      Mammoth Cave,  pp 35-38 in Cave Research Foundation Annual  Report
	and T. M. Lewis  1980. The distribution and ecology of two  species
      of subterranean Caecidotea in Mammoth Cave National Park.
      pp 23-27 in Cave Research Foundation Annual Report
	T. M. Lewis, and J. Eckstein  1982.  A biological reconnaissance of
      a polluted cave stream: The Hidden River groundwater basin.
       pp 9-10 in Cave Research Foundation Annual Report
Lisowski, E. A. and T. L. Poulson 1979. Impacts of Lock and Dam Six on baselevel
      ecosystems in Mammoth Cave.
      pp 48-54 in Cave Research Foundation Annual Report
Poulson, T. L. 1963. Cave adaptation in Amblyopsid fishes.
      The American Midland Naturalist,  41: 263-290.
	 1957 Comparisons of cave  stream communities.
       pp 33-34 in Cave Research  Foundation Annual Report
	 1985. Evolutionary reduction by neutral mutations: Plausibility
      arguments and data from Amblyopsid fish and Linyphiid spiders.
      National Speleological Society Bulletin, 47(2): 109-117.
                                    508

-------
	and D. C. Culver 1969. Diversity in terrestrial cave communities.
      Ecology, 50(1): 153-157.
	and T. C. Kane 1977. Ecological diversity and stability:   Principles
      and management.
      pp 18-21 in T. Aley and D. Rhodes eds.,Proceedings of the 2nd Cave
      Management Symposium, Speleobooks, Albuquerque, NM, USA
	and W. B. White 1969. The cave environment: Limestone caves
      provide unique natural laboratories for studying biological and
      geological processes.  Science, 165:971-981.
Quinlan, J. F. 1990. Special problems of ground-water monitoring in karst
      terranes. pp 275-304 in D. M. Nielsen and A. I. Johnson eds.,
      Groundwater and vadose zone monitoring. ASTM Special Technical
      Paper 1053, American Society for Testing and Materials,
      Philadelphia, PA, USA
	and E.  C. Alexander Jr.  1987. How often should samples be taken
      at relevant  locations for reliable monitoring of pollutants from an
      agricultural, waste disposal, or spill site in a karst terrane?
      A first approximation,  pp 277-286 in B.  Beck and W. L. Wilson eds.,
      Proceedings from the 2nd Multidisciplinary Conference on Sinkholes and the
      Environmental Impacts of Karst, Florida  Sinkhole Research
      Institute, University of Central  Florida, Orlando, FL, USA
White, W. B. and E. L. White eds. 1989.  Karst hydrology: Concepts from the
      Mammoth Cave Area. Van Nostrand-Reinhold, NY, USA
                            BIOGRAPHICAL SKETCH

Tom Poulson got his B.S. at Cornell University and his  Ph.D. at the University of
Michigan. After a postdoctorate at U.C.L.A. he taught 10 years at Yale and 3 years
at Notre Dame before coming to the University of Illinois at Chicago where he is
now Professor of Biological Sciences. He is Chief Scientist of The Cave Research
Foundation.  He  Studies morphology, physiology,  life  history, and  behavior in
ecological and evolutionary contexts.  In caves he studies fish, crayfish, spiders,
millipedes,  springtails,  flies,  and  beetles.    Outside  he  has  studied  many
communities including tropical forest dung beetles, salt marsh and desert  birds,
mountain and desert chipmunks, floating plants, dunes grasses, and forest trees.
                                      509

-------
U5E OF BIOLOGICAL INDICATORS OF GROUNDWATER QUALITY
Thomas L. Poulson

Q.  When using an Index of Biological Integrity (IBI) to assess groundwater
contamination in a cave, how necessary is it to have a pre-contamination
assessment? Will  nearby cave streams be sufficient to provide a baseline?

A.   It is sufficient to define maximum values for an IBI using the variation seen
for the most pristine cave stream(s) in the local region.  Only slight adjustments
to expectations for non-impacted streams are made due to abiotic factors such as
stream order and above or below average refuges from  flooding and predation.
Q. How useful and reliable might the IBI approach be for karst springs where caves
are not accessible?

A. The principles behind an IBI will hold for any aquatic community including
karst springs. I would develop an IBI using spring basin faunas and floras for the
region in question and, if possible, use a net to collect the organisms washed out
of the spring over a week long period both at times of low and high flow.  The
problem with spring outwash data alone is that one cannot assess the underground
habitat to determine what it can support and so it might be important to have pre
contamination data on the outwash fauna for a particular spring system.
Q. Are the top species of the food chain sufficiently rapid in their responses to be
useful bioindicators of pollution?

A. Fish and crayfish at the end of the food chain react quickly to some and slowly
to other contaminants. Crayfish crawl  out of the water when there is an acute
input of a toxin at high concentrations.  Increased proportions of females becoming
fecund may indicate chronic input of human and animal waste at low
concentrations.  But neither long-lived  fish nor crayfish may respond quickly to
chronic input of toxins at low concentrations that take many years to
bioaccumulate enough to cause visible tumors or obviously compromise
reproduction.  If toxin pollution is suspected then I would find out what is used in
the watershed (which pesticides and herbicides) and monitor the most dangerous
toxin in the water at peak flow when runoff will be greatest.
                                    510

-------
Q.  Have you seen any significant changes in the aquatic fauna of Mammoth Cave
since the 1950s and if so why have the changes occurred?

A.  No species has been lost or gained but the diversity and abundance of the Echo-
Styx River communities apparently declined since the early 1900s and has clearly
declined since the 1950s when I first censused the area (Lisowski and Poulson
1979).  At present cavefish, cave crayfish, and cave shrimp are either absent or
very difficult to find and the upper level species of  isopod may have replaced the
usual river species.  These changes were associated with changes in flow regime
and water level starting with the construction of Lock and Dam *6 just
downstream on Green River around 190  and then continued with the construction
of  the Nolin River Dam just downstream and the Green River Dam far upstream in
the 1950s.  I do not know  the mechanism but have hypothesized that it is increased
silt deposition in the cave. Siltation does decrease  substrate heterogeneity and so
may have decreased refuges of copepods, isopods, amphipods, and shrimp from
floods and/or predation by fish and crayfish.  Siltation may also cover or dilute
particulate organic matter and so cause declines of  the bacteria and protozoa
which are eaten by copepods, isopods, amphipods, shrimp, and small crayfish.
                                    511

-------
512

-------
   THE USE OF BENTHIC MACROINVERTEBRATES  FOR ASSESSING THE

 IMPACT OF CLASS V INJECTION WELLS ON CARBONATE GROUND WATERS

   Albert E.  Ogden1, Ronald K. Redman2 and Teresa L. Brown3
     Center for  the Management, Utilization and Protection
      of Water Resources1  and  The  Department of Biology2,
   Tennessee Technological University,  Cookeville,  TN  38505
                              and
   First Tennessee Development District3,  207 N Boone Street,
              Suite 800, Johnson City, TN  37604

                           ABSTRACT

     The  U.S. Environmental  Protection Agency contracted  the
authors to investigate the impact  of  Class V injection  wells on
ground water  in karst  terranes  to  demonstrate the need  for
regulations over these shallow injection methods.   Any  sinkhole
that  has  been modified to better accept drainage, including
storm water runoff, was considered a Class V injection well.  In
addition,  service station  bay  drains  that  lead to septic  tanks,
pits, or  dry  wells were  a  focus of the study.

     The  study  was  conducted  in  Cookeville and Johnson  City,
Tennessee, to compare contaminant transport  in  both flat-lying
and folded rock.   A major cave system underlies  both cities.
Ground water flow paths and velocities were determined using dye
tracing methods.  Water and  benthic macroinvertebrate samples
were  taken where water  entered Class V injection wells  and at
the springs influenced by these waters.   As  a  control, samples
were   also gathered  from   a  sinking  stream  and   spring  in  a
forested area.   Concentrations of MTBE,  BTEX, TPHC,  ethylene
glycol, and metals were below detection limit at most sites and,
therefore, were  not  useful in demonstrating an impact.    This
emphasizes that  grab samples  in karst seldom  indicate overall
water quality conditions.   In contrast, the  Shannon Diversity,
Shannon Evenness,  and  Family  Biotic indices,  coupled with  a
cluster analysis of the   benthic  data,   were  very useful  in
classifying sampling sites as to their degree of pollution where
water chemical data did  not indicate degradation.

                          INTRODUCTION

     Part C of the Safe Drinking Water Act (Public Law 93-523)
authorized the  U.S.  Environmental Protection Agency  (EPA)  to
establish regulations to assure that potable ground water is not
                              513

-------
endangered by  the  underground injection of waste.   Guidelines
for underground  injection and the classification  of injection
wells  come  under  Part  146.04  of  the  Federal  Underground
Injection  Control   program  (U.S.  EPA,  1981).    This  program
created five classes of underground injection wells.  Classes I
through IV include such categories as radioactive and hazardous
waste  injection  and disposal  of brines from  the oil  and  gas
industry.

     Class V wells  are generally defined as  those which inject
only  nonhazardous   fluids  into  or  above  strata that  contain
underground sources  of  drinking water (USDW).   USDWs  not only
include aquifers which are currently  serving  as drinking water
supply sources but  also aquifers which are of acceptable quality
for possible  future use.   Class V wells  include any  type  of
injection well not  covered in  the UIC definition of Classes I,
II, III, or IV.  EPA has classified Class V injection wells into
six groups based in part  on the expected quality of the injected
fluid.  The following is  a listing of these  groups as  found in
regulations  for  the  State  of Florida  which is  particularly
sensitive to karst aquifers:

Group 1 - wells  associated   with   thermal   energy   exchange
          processes, which include air conditioning return flow
          wells  and cooling  water return flow wells.   Cooling
          water  return flow  wells may be part of a closed-loop
          system, with no hazardous additives;
Group 2 - recharge  wells,  saltwater  intrusion  barrier  wells,
          connector   wells,    and  subsidence   control   wells
           (associated with aquifer overpumping);
Group 3 - wells  which are part  of  domestic  waste  treatment
          systems,  swimming pool drainage wells, injection wells
          used  in   experimental  technologies,   wells  used  to
          inject spent brine into the same formation from which
          it was withdrawn after extraction of halogens or their
          salts;
Group 4 - nonhazardous industrial and commercial disposal wells,
          which  include laundry waste, dry wells, sand backfill
          wells,  and nuclear   disposal  wells  used to  inject
          radioactive wastes, provided the concentrations of the
          waste do not exceed drinking water standards contained
          in Chapters 17-22, FAC, and injection wells  used for
          in-situ recovery of phosphate,  uraniferous sandstone,
          clay,  sand,  and  other minerals  extracted  by  the
          borehole slurry mining method (lignite, tar sands, oil
          shale, coal);
Group  5 -  lake level drainage and stormwater drainage wells; and
Group  6 - geothermal wells and "other" wells.

     The U.S.  EPA  funded twenty-four projects nationwide under
its Shallow Injection Well Initiatives Program to help evaluate
the  impact of Class V  injection wells on ground  water and to
establish best management practices.  The authors of this paper
                             514

-------
were chosen to help evaluate Class V injection well practices in
the karst of Tennessee.  Any sinkhole that had been modified to
accept waste,  including  stormwater  runoff,  was considered  a
Class V injection well in the study.

                          OBJECTIVES

     The primary goal of the proposed study was to determine if
Class V  injection  wells  have created a  ground water pollution
problem.  Evaluating the  effects of  service station bay drains
(Group 6) that lead to septic tanks, pits, or dry  wells was a
primary target of the investigation.   Other high priority Class
V  injection   wells  that  were  investigated  include:     1.
Agricultural and municipal drainage into  improved and unimproved
sinkholes  (Group  1),  2.  Industrial drainage into  sinkholes
(Group 4),  and 3.  Domestic wastewater drainage into sinkholes
(Group 5) .

     To  fulfill  these objectives,  samples  were collected from
waters   entering   sinkholes  and  the   springs  hydrologically
connected to them as determined  by ground water tracing methods.
A benthic macroinvertebrate study was then conducted at most of
the  sites  to  evaluate the impact  of  contaminants  on stream
biota.   Since  ground water  chemistry in  karst terranes changes
rapidly during storm events (Ogden,  1988 and Quinlan, 1989), it
was  felt that the  benthic  survey might prove a  more reliable
indicator of long-term water  quality.

                HYDROGEOLOGY OF THE STUDY SITES

     Two geologically different karst terranes in Tennessee were
chosen so that the results  would  have  applicability throughout
much  of  the  karst  in  the  United  States (Figure 1).  Around
                Western Villi
                                         Cumberland Plitfi
   Mnmslppi River Vtllcy
 Figure  1.  Location of study areas  (adapted  from Miller,  1974).
                              515

-------
Cookeville (Putnam County), the Mississippian-aged carbonates of
the Eastern Highland Rim Province are flat-lying allowing ground
water to move along a wide range of orientations corresponding
to  joint  and  photo-lineament  trends  (Ogden et  al.,  1989).
Streams originating on the Cumberland Plateau east of Cookeville
flow over shales and sandstones until underlying limestone beds
are intersected.  Some streams sink when they reach the Bangor
Limestone  while  others sink  into the  Monteagle  or  St.  Louis
limestones.  Subterranean water moves  through caves,  pits,  and
solution-enlarged  fractures  until emerging  as  spring  flow,
commonly at the St.  Louis-Upper Warsaw  boundary.  Within much of
the city limits of Cookeville, surface runoff is perched on the
upper Warsaw Formation which  is a cross-bedded, reddish-brown,
sandy  limestone.    This  water  then   sinks  into  the  middle
limestone  member.   A  four-mile-long cave  that drains much of
Cookeville's storm  water runoff  occurs within this unit.   The
lower  Warsaw member  is a  relatively impermeable  calcareous
siltstone and shale with some argillaceous limestone beds.  The
largest springs  around  Cookeville  are  found perched above this
lower member.

     Ground  water conditions in  Putnam County  have not  been
extensively studied.  An early regional analysis of the aquifers
and a listing of water wells was made by Smith (1962) .  A ground
water resource analysis around Center Hill Lake,  which included
the Cookevile  area, was made by Moore and Wilson (1972).   A
water  quality  survey  of  some  water  wells  around  town  was
performed  by Collar and Ogden (1990).   Faulkerson  and  Mills
(1981),  Pride  et  al.   (1988), and Hannah  et al.   (1989)  have
performed  some  dye traces  around Cookeville.   Physiochemical
degradation of the water within the four-mile-long cave beneath
Cookeville  has  been documented by Smithson (1975),  Faulkerson
and Mills  (1981), Wilson (1985),  and  Pride et al.  (1988).   A
recent assessment of sinkhole flooding in Cookeville from storm
water drainage has  been conducted by Mills  et al.  (1991).

     The  second  study area used  for the  investigation  occurs
within  the  city limits of  Johnson City   (Washington  County)
located  within the  Appalachian  Valley and Ridge  Province of
eastern Tennessee  (Figure 1).  In the Valley and Ridge Province
around Johnson City, the Ordovician-aged Knox Group carbonates
are  complexly  folded  and   faulted,   and   ground  water  moves
predominantly  along   stratigraphic  strike  within  solution-
enlarged  bedding planes.    The  Knox Group is over  3,000 feet
thick  and  has  not been subdivided in  the  study area.   Much of
Johnson  City is  developed  on a sinkhole  plain.   Many  of  the
sinkholes  were  designated  as Class  V  injection wells  by this
study  due  to storm water routing  into  them and/or the presence
of  discharging  pipes  (often of  unknown  origin)  within  the
sinkhole basin.

     Early reconnaissance  ground water studies of the Johnson
City area  were  performed by Maclay (1956)  and DeBuchananne and

-------
Richardson  (1956).   Many  of  the  public water  supplies  near
Johnson  City  utilize  springs  that  emerge  from  cavernous
limestones and dolomites.   The vulnerability of these municipal
water supplies prompted the First Tennessee Development District
(Stanley Consultants,  Inc.,  1983; Matthews,  1986;  and  Brown,
1987)  and  TVA  (Foxx,   1981)   to  make a survey  of  potential
pollution sources within close proximity  of the springs.   Other
springs  in  the  area have  shown probable  contamination  from
septic  tanks   (Wilson  et  al., 1988).    Many  of  the  eastern
Tennessee springs  become turbid  after  storm  events and  show
elevated levels of fecal coliform (Brown, 1987).   This reflects
the  impact  that natural  and  anthropogenic  activities in  the
recharge area have on ground water quality.  Ogden et al.  (1991)
recently  completed  a   "wellhead"  protection  study  of  nine
municipal-used springs  in eastern Tennessee that involved ground
water tracing and a time-series analysis  of water quality.   The
springs within the city limits  of Johnson City were not included
in any of these studies.

     No ground water tracing had been conducted in Johnson City
until  initiation  of  this  project.     Field  investigations
performed for this study showed that most of the  storm drainage
in  north Johnson  City flows  through cave systems and  then
emerges at springs  along  Knob and Cobb creeks which flow  into
the Watauga River.   These creeks are being  studied by the  U.S.
Geological Survey  as part  of the nationwide effort  to monitor
urban   storm   water   runoff   quality    (Johnson,    personal
communication).

                            METHODS

     Samples were collected of  waters entering Class V injection
wells  (sinkholes) and at the springs influenced by these waters
during the 1991 wet and dry seasons.  As a control, samples were
also  gathered  from a  sinking  stream  and a  spring  with  a
predominantly forested recharge area.   Dissolved  oxygen (DO) ,
pH,  conductivity,  and  temperature were  measured in  the  field
with  a Hydrolab field  monitor.  Laboratory analyses used as
indicators  of contamination from  agricultural activities  and
septic   tanks  included   nitrate,   chloride,  fecal  coliform
bacteria, and fecal streptococcus  bacteria.   Zinc,  chromium,
lead, methyl  tertiary butyl ether  (MTBE,  a  gasoline additive),
benzene-toluene-ethylene-xylene    (BTEX),   total    petroleum
hydrocarbon (TPHC), and ethylene glycol levels  were measured as
indicators of waste products and leaks from service stations and
parking  lot  runoff.   Seven  springs  and   seven sinking streams
were sampled  around  Cookeville; whereas,  six springs and three
sinking streams were sampled around Johnson City.

     To  enhance the interpretation  of the  impact of  Class V
injection wells on  spring waters,  benthic macroinvertebrate
samples were collected at riffle areas and pools at most of the
injection points and the springs.   Samples  were  collected with
                             517

-------
a  modified kick  net  and a  Surber  sampler  (0.09  m2) .    All
organisms were preserved in a 10% formalin solution, enumerated,
and identified to  the genus  level when possible.   Statistical
analysis  of  the  benthic macroinvertebrate  data  were  then
conducted.

     Qualitative  ground  water  tracing  was  conducted  using
fluorescein  dye  and  activated  charcoal  detectors.    Optical
brighteners and cotton detectors were also used for tracing.  A
control  packet  was  placed  at  a  spring  known  not  to  be
hydrologically connected  to  the tracer input  site  during each
test.

                           RESULTS

Water Quality Sampling

     Both  wet  and dry  season  samples from  nearly all  of  the
twenty-three sites in the two study areas showed below detection
limits  for BTEX,  MTBE,  ethylene glycol,  chromium,  lead,  and
zinc.   One site,  the  storm  drain for Tennessee  Technological
University, showed below detection for BTEX in one  sample  and
high levels in another  (benzene  -  1200 ppb,  toluene - 220 ppb,
xylene  -  780 ppb) .   Fumes  were reported  in this  drain which
initiated  its  sampling.  The  drain  receives  runoff  from  the
university's   garage   and  lies   next  to  several   gasoline
underground storage tanks.   In addition,  a  nearby underground
storage tank used to provide fuel for a background power supply
in the  basketball  arena was  recently pulled,  and  soil samples
showed TPHC levels as high as 2088 ppm. No remediation or site
characterization was performed at the site,  and a  new building
has since  been constructed over  it.   The  storm drainage system
beneath the university  has been  diverted  into  a  large sinkhole
which  also receives  storm  water  runoff  from other  areas  of
Cookeville.   Raw  sewage  from  the  city  lines is  occasionally
bypassed into the same sinkhole.  Ground water tracing showed a
hydrologic connection between the sinkhole and Big  Spring.

     MTBE  was  found  in only two  samples,  and it was  at  low
levels  (<  5  ppb) .   TPHC was above the detection  limit at only
one  site  during  the  wet season with  a  level   of   309  ppb.
Sampling  during  the  dry season showed detectable  TPHC at only
the  Cookeville city  garage drain  site with a value  of  6 ppm.
Zinc was measured  at  levels  above  detection at five sites with
concentrations being up to 73 ppm.

     Chloride  levels  were  significantly  higher  at  springs
receiving  recharge from Class V injection wells compared to the
control  springs.   The control  springs  had  chloride  levels
between 1  and  3  ppm;  whereas,  the other sites  ranged from 3 to
64 ppm  with  an average of 14 ppm.   Nitrate concentrations were
below  2 ppm at all but two sites.   The highest  measured value
was  5  ppm.   Fecal coliform and fecal streptococcus bacteria
                              518

-------
counts were  dramatically  different between  the wet  and  dry
sampling periods.  The wet season samples were  relatively free
of  these  bacteria.    Only  five  of  the  samples  had  fecal
streptococcus bacteria  with the  highest count  being  only  21
colonies/100 ml.  All but  two  of the sites had  fecal  coliform
bacteria  counts  over  fifty    with  the  highest  being  350
colonies/100 ml.  In  contrast,  most of the dry  season samples
had fecal streptococcus levels that were too  numerous  to  count
(TNTC).   Fecal  coliform  bacteria counts were greater  than 500
colonies/100 ml  at  thirteen of  the sites  for  the dry season
samples and only one site had less than 25 colonies/100  ml.  The
storm drain at Tennessee Technological University and Big Spring
had the highest  fecal coliform bacteria counts in the Cookeville
samples  related  to  leaky  sewers and bypassing  of raw sewage
(2900  colonies/100  ml  and  TNTC,   respectively).    The  fecal
coliform/fecal streptococcus ratios suggest  that  most of  the
other sites are being  polluted  from  animal versus human wastes.

Benthic Macroinvertebrate Sampling

     Benthic  macroinvertebrate  samples  were  collected   from
twelve sites in the two study areas.  To  date,  identifications
have  been  completed  for six samples  collected  during the  wet
season.  As  a control,  a spring with a forested recharge area
was chosen  for  comparison to the springs with urban  recharge
areas.   Macedonia Spring  in Cookeville and  Taylor Spring  in
Johnson City served as the control springs.

     Shannon   Diversity   indices    (Washington,   1984)   were
calculated on each spring and sinking stream  (Table 1).   Wilhm
(1970) found values between  3  and 4 for the  Shannon Diversity
Index (d) in unpolluted waters,  but  in badly polluted waters it
was generally less than  1.   All of  the sites shown on Table 1
exhibited low Shannon Diversity  (Base 2) values  with Macedonia
Spring having the highest (2.77)  indicating  only slight organic
pollution.  However,  EPA biologists  for the  southeastern region
have  found  d  not  to  be  sensitive  enough  and  have   used
equitability  e   (Shannon   Evenness)  to   better   demonstrate
degradation of stream waters (Weber,  1973).   E  is  sensitive to
slight  levels  of degradation.    Polluted streams usually have
values  between  0.0 to  0.5;  whereas,  unpolluted streams fall
between  0.5 to  1.0.     In  the  Cookeville  area,  the  Shannon
Evenness  Index   (Table  1)  clearly  distinguishes the  polluted
urban  springs   (Big and Pigeon  Roost)  from  the rural  spring
(Macedonia).  In the Johnson City area,  this distinction is not
as  apparent.    This  may be  due to  Taylor  Spring's  immediate
confluence with  a surface  stream which may not  allow  time for
the  benthic  environment  to change   from  heterotrophic  to
autotrophic.  According to  Vannote et al.  (1980), this change is
necessary for increasing species diversity.  Also, the substrate
at Taylor  Spring is  almost completely composed  of  sand,  silt,
and  clay  with  nearly   no  gravel  or   cobble-size material.
                             519

-------
Table l.   Shannon Diversity, Shannon Evenness, and Family Biotic
          indices calculated for the wet season samples.

Macedonia Spring
Big Spring
Pigeon Roost Spring
Taylor Spring
Shoe Spring
Car Wash Insurgence
Shannon
Diversity
(Base 2)
2.77
0.88
0.15
2.14
1.64
2.15

SP
P
P
SP
SP
SP
Shannon
Evenness
0.69
0.31
0.08
0.58
0.55
0.68

UP
P
P
UP
UP
UP
Family
Biotic
Index
3.97
5.84
6.01
6.08
7.05
5.50

VG
FP
FP
FP
P
F
P = polluted; SP = slightly polluted; UP= unpolluted; VG = very
good; F = fair; FP = fairly poor; P = poor

Macedonia  Spring,  Big  Spring,   and  Pigeon  Roost Spring  also
emerge from caves but their substrates are composed primarily of
gravel, cobble,  and  boulder sizes allowing  for  more diversity
and greater  numbers  of  species.   Car Wash Insurgence and Shoe
Spring have substrate compositions composed of gravel and cobble
mixed with sand, silt,  and clay.

     Hilsenhoff's  (1988)  Family-Level  Biotic  Index  (FBI)  was
also calculated for each site  to see  how  it compared  to the
Shannon Base 2  values and equitability  e values  (Table 1).  In
general,  the higher the FBI,  the  greater  is  the  amount  of
organic pollution with  10 being the highest number possible.  In
the Cookeville  area,  the rural  spring  was  rated as  very good
while the urban springs were ranked as fairly polluted.  In the
Johnson City area,  a distinction between the rural  and urban
sites  was   not  seen,   probably for   the   reasons  discussed
previously.   Visual and olfactory  evidence of  pollution was
found  at  the  two urban springs in Johnson  City,   and water
quality differences  between the two groups  were seen based on
chloride and conductivity results (Table 2).  This water quality
data   helps   support  the  substrate   and   heterotrophic  vs.
autotrophic  hypotheses just presented.

     A  cluster  analysis of the  benthic macroinvertebrate data
was also performed  for  the  study sites.  The cluster analysis,
together with  the Shannon Diversity, Evenness,  and the Family
Biotic  indices,  gave   further   insight  to  which  sites  have
threatened   benthic  communities and   possible  ground  water
degradation.  Big,  Shoe,  and Pigeon Roost springs are polluted
by  city runoff into Class  V injection  wells  and thus  cluster
together  (Figure 2).  Low  number of genera at  Taylor  Spring,
related to substrate composition, caused it  to be grouped with
                             520

-------
Table 2.   Wet and dry season  chloride and  conductivity values.


Macedonia Spring
Big Spring
Pigeon Roost Spring
Taylor Spring
Shoe Spring
Car Wash Insurgence
Chloride (Mg/L)
Wet season
1.70
4.00
18.20
2.82
7.92
10.20
Dry season
2.40
15.60
56.30
2.30
7.39
11.80
Conductivity (/unhos/cm)
Wet season
364
158
286
281
478
475
Dry season
142
236
262
572
442
863
the  polluted  urban  springs.    Macedonia  Spring,  which  has
relatively  pristine   water,  was  grouped  alone.    Car  Wash
Insurgence, a small sinking stream, was clustered separately due
to  large  annual  water  temperature fluctuations  which  have  a
significant effect  on the biologic community structure not seen
at the springs.
                       Cluster Analyses
                                             •  Macedonia

                                             1  Big Spring

                                               Shoe Spring
                                               Pigeon Roost
                                               Taylor Spring
                                              Car Wash Insr.
     Linkage Clusters
       1

       2

       3

       4

       5
Big Spring

Big Spring

Big Spring

Macedonia

Macedonia
2     .4



  Linked


 Shoe Spring

 Pigeon Roost

 Taylor Spring

 Big Spring

 Car Wash Ins.
  .6



Similarity


 .50446

 .38494

 .28472

 .04615

 .00098
                                         .8
 Prob


.29000

.15000

.10000

.01000

.01000
 Figure 2.  Cluster analysis results.
                               521

-------
                          CONCLUSIONS

     The water quality tests performed  for  this study were not
definitive in  assessing  the impact  on  karst ground  waters  of
volatile chemicals entering Class V  injection wells.   The open
conduit flow  in the karst  aquifers  allows contaminants  to  be
rapidly  flushed  through  the  ground  water  system.    Nearly
continuous water quality  monitoring would be necessary to detect
contaminants dumped or spilled  into Class V injection wells such
as from service station bay drains.   In addition, turbid flow in
the  caves  and sinking streams causes  rapid  degassing of  the
volatile  components,   thus  explaining  why some sample  sites
showing the presence of oil  and grease had no measurable amounts
of volatiles.   Also,  if volatiles were detected, it  would  be
nearly  impossible  to  determine if they were derived  from  bay
drains or from leaky underground storage tanks.   Fecal coliform
and fecal streptococcus bacteria levels, on the other hand, were
very high  during the  dry  season demonstrating the  impact  on
ground  water  from leaky sewers,  sewage bypassing, and  animal
wastes  entering  Class  V  injection   wells.     Chloride  and
conductivity values  were also useful  in demonstrating  ground
water degradation.

     Benthic  macroinvertebrate  analyses  may  be  better  for
accessing the impact of Class V injection wells  on ground water
since  these  organisms can be very  sensitive  to daily  water
changes not detected by  collecting occasional grab samples for
chemical  analysis.    In  this study,   the  Shannon  Diversity,
Shannon Evenness, and Family Biotic indices, in conjunction with
the  cluster  analysis,  proved  useful in distinguishing  spring
waters  that have been degraded by pollutants entering Class V
injection wells.
                       ACKNOWLEDGEMENTS

     This  project   was   funded  by  the   U.S.   Environmental
Protection Agency with matching funds provided by the Center for
the Management, Utilization and Protection of Water Resources—
Tennessee  Technological  University and the  First  Tennessee
Development District.  This paper  does  not  necessarily reflect
the views of the agencies.
                           REFERENCES

Brown, T., 1987, Survey of public ground water supplies:  First
     Tennessee Development District, Johnson  City,  TN,  217  p.

Collar,  P.O.  and A.E. Ogden,  1990,  A hydrochemical  survey of
     carbonate   groundwaters,   Putnam   and  Jackson  counties,
     Tennessee:  Proc. 3rd TN Hydro. Sym., U.S.G.S. Nashville,
     TN, pp.  30-37.
                              522

-------
DeBuchananne,   G.D.  and  R.M.  Richardson,  1956,  Ground  water
     resources of east Tennessee:   Tenn.  Div. of Geol.,  Bull.
     58,  Nashville,  TN,  393 p.

Faulkerson,  J. and H.  Mills,  1981,  Karst hydrology,  morphology,
     and water quality in the vicinity of Cookeville, Tennessee:
     Report to the City of Cookeville,  Tennessee  Technological
     University,  Cookeville,  TN, 67 p.

Foxx, Suzanne  A.,  1981, A summary of the geohydrology and of the
     spring recharge zones within the Tennessee portion  of the
     208 study area  of the  First Tennessee-Virginia  Development
     District:  TVA Office of  Natural  Resources,  Division  of
     Water Resources,  Report No. WR28-1-520-116.

Hannah,  E.D.,  T.E.  Pride,  A.E. Ogden,  and  R.  Paylor,  1989,
     Assessing ground water flow paths from pollution sources in
     the karst of Putnam  County,  Tennessee:   Proc. of  Third
     Multidisciplinary Conference on Sinkholes:  Their Geology,
     Engineering,  and Environmental  Impact,  Florida  Sinkhole
     Res. Inst.,  Orlando,  FL, pp.  183-188.

Hilsenhoff,  W.L.,  1988,   Rapid field  assessment  of  organic
     pollution with a  family-level biotic  index.    J.N.  Am.
     Benthol.  Soc.,  7(1):  65-68.

Johnson, G.,  1991,  Personal  communication regarding  U.S.G.S.
     activities in Johnson City, Knoxville Field Office.

Maclay,  R.W.,  1956,  The  ground  water  hydrology   of  the
     Elizabethton and Johnson City, Tennessee area:  Unpubl. M.S
     Thesis, University of Tennessee, 54 p.

Matthews, M.R., 1986, Groundwater  situation assessment  of the
     Tennessee Valley Region:   TVA  Office  of  Natural Resources
     and  Economic  Development,  Division  of  Air   and  Water
     Resources, Chattanooga,  TN.

Miller,  R.A.,  1974,  The  geologic  history of  Tennessee:   TN
     Division of Geology,  Bull. 74, Nashville, TN,  63 p.

Mills, H.H., H.N. Taylor,  A.E.  Ogden, Y.  Robinet  Clark,  and R.
     Forde, 1991, Predicting sinkhole flooding in  Cookeville,
     TN, using SWM model and  CIS:  Proc. Appalachian Karst Sym.,
     Natl. Spel. Soc., Huntsville, AL, pp. 159-167.

Moore, G.K.  and J.M. Wilson,  1972, Water resources of the Center
     Hill Lake region, TN, State of  TN  Dept.  of Cons.,  Div- of
     Water Res., Series No. 9,  77 p.

Ogden, A.E.,  1988,  Distinguishing  flow regimes of  springs in
     carbonate rock terranes:   Proc. Natl. Water Well Assoc.,
     Ground Water Chemistry Conf., Dublin, Ohio, pp.  53-73.
                              523

-------
Ogden, A.E., W.A. Curry, and  J.L.  Cummings,  1989,  Morphometric
     analysis of sinkholes and caves in Tennessee comparing the
     eastern Highland  Rim and  Valley and Ridge physiographic
     provinces:  Proc.  of Third Conference on Sinkholes:  Their
     Geology,  Engineering,  and  Environmental Impact,  Florida
     Sinkhole Res. Inst., Orlando, FL, pp. 135-142.

Ogden, A.E., T.L.  Brown, and K.G. Hamilton, 1991, Delineation of
     wellhead  protection  areas  for  municipal-used  springs  in
     eastern Tennessee: UT -  Tennessee Water Resources Research
     Center, Report #124, Knoxville, TN, 134 p.

Pride, T.E., A.E.  Ogden,  M.J.  Harvey,  D.B.  George,  1988,  The
     effect  of urban  development  on spring  water  quality  in
     Cookeville, TN:   Proc.  Natl.  Water Well Assoc.  2nd Karst
     Conference, Dublin, Ohio, pp. 97-120.

Quinlan, J.F.,  1989, Ground water monitoring in karst terranes:
     Recommended protocols and  implicit  assumptions:  U.S.  EPA
     600/X-89/050, Las Vegas, Nevada, 79 p.

Smith,  O.,  1962, Ground  water  resources  and municipal  water
     supplies  of the  Highland  Rim  in  Tennessee:    Tennessee
     Division of Water Resources, Water Resources Series Number
     3, 257 p.

Smithson, K.D.,  1975,  Some effects  of  sewage effluent  on  the
     ecology  of  a  stream  ecosystem:   Unpubl.  M.S.  Thesis,
     Tennessee Technological University, Cookeville, Tennessee,
     74 p.

Stanley Consultants, Inc., 1983, Ground water protection plan -
     final   report:     First   Tennessee-Virginia   Development
     District  208 Water  Quality  Management  Program,  Johnson
     City,  TN.

Vannote,  R.L.,  G.W.  Minshall,  K.W.  Cummins,  J.R.  Sedall,  and
     C.E. Gushing,  1980,  The river contiuum concept:   Can.  J.
     Fish.  Aquat. Sci. 37:  130-137.

Washington,  H.G.,  1984,  Diversity,   biotic  and  similarity
     indices:    A review  with  special relevance   to  aquatic
     ecosystems:  Water Resources  Res.,  19(6):653-694.

Weber,  C.,  1973, Biological  field and  laboratory  methods  for
     measuring quality of surface  waters and  effluents:  USEPA
     670/4-73-001.  Cincinnati,  OH.

Wilhm,  J.L.,   1970,   Range   of  diversity  index   in  benthic
     macroinvertebrate populations:   JWCF., 42(5):   R221-R224.

Wilson,  D.,   1985,  The  physical,  chemical,  and  biological
     recovery  of  Pigeon Roost Creek following the closing of  a
                             524

-------
     sewage treatment  plant:   Unpubl.  M.S.  Thesis,  Tennessee
     Technological University, Cookeville, Tennessee, 73 p.

Wilson, T.M.,  J.A. Gordon,  A.E.  Ogden,  and J.D.  Reinhard,  1990,
     Detection  of  failing  septic  tanks  in  east  Tennessee
     utilizing infrared  color aerial photographs:   Proc.  NWWA
     Conf.  on  Eastern Regional  Ground  Water Issues,  Dublin,
     Ohio, pp.  177-188.

U.S. Environmental Protection  Agency, 1981, 40 CFR Parts 122 and
     146  Underground Injection  Control  Program  Criteria  and
     Standards:   Federal Register,  V.  46, no. 190,  pp.  48243-
     48255.

                     BIOGRAPHICAL SKETCHES

Albert E.  Ogden  received his B.S.  degree in geology  from the
Pennsylvania State University  and his Ph.D. in hydrogeology from
West Virginia University.  He  was the hydrogeology professor for
the  University  of  Arkansas  from   1976-1981.    He  then  was
assistant director  and hydrogeologist  for the Edwards  Aquifer
Research  and Data Center in  Texas  from 1981-1985.   Following
this, Dr. Ogden was  the senior hydrogeologist for the RCRA and
Superfund programs within the  Idaho Division of Environment from
1985-1987.   He  presently  is  an associate  professor for  the
Center for the Management, Utilization  and Protection of Water
Resources,  Box   5033,   Tennessee   Technological   University,
Cookeville,  Tennessee    38505,   (615)  372-3353.   Dr.  Ogden's
research  interests  include  site characterization  of hazardous
waste  sites  and  solving  water  quality  problems  in  karst
terranes.

Ronald  K. Redman received his  B.S.  degree in  biology  from
Arkansas  Tech  University.   He then  worked  as a manager  for  a
fish   hatchery  where  his   duties  included  water   quality
monitoring.    He  now  is  working  on his  M.S.  in biology  at
Tennessee Technological University, and this paper is a portion
of his thesis research.

Teresa L. Brown received her  B.S. degree in  geology from James
Madison University  followed by graduate work in  the Hydrology
and Environmental Science Program at Indiana University.  After
college,  she  worked  as a   geologist  and  pollution  control
specialist for the Virginia Division of Mined Land Reclamation.
Since   1986,   she   has  been  the  environmental  management
coordinator for  the First  Tennessee Development  District,  207
North  Boone Street,  Suite  800,  Johnson City, TN   37604,  (615)
928-0224.    Ms.   Brown has   experience  in  siting  landfills,
assaying  the  impact of land  application  of  sludge,  conducting
water  quality  surveys,  and  delineating  wellhead  protection
areas.
                              525

-------
The Use of Benthic Macroinvertebrates for Assessing the Impact of Class V Injection
Wells on Carbonate Groundwaters
Albert E. Ogden, Ronald K. Redman, and Teresa L Brown
Question 1.  The Shannon-Wiener diversity indices are problematic because they:
               a)  confound the numbers of species and evenness of numbers of
                   individuals among species; and
               b)  count tolerant  species, e.g., surface chironomids, the same as
                   obligate cave species (e.g., Caecidotea isopods).  So,  what were
                   the species and  their  degrees  of  restriction  to caves  and
                   groundwater in springs with outflows from control and contaminated
                   basins?

Response:   No samples were taken in caves.
Question 2.  Most, if not all, of the insect orders of your control surface streams are not
            normally found in caves.  What orders of Crustacea and what kinds of
            species (Facultative vs. obligate cave species) did you find in outflows from
            your control and contaminated basins?

Response:   In the control  springs, we found decapods, amphipods, and isopods. We
            also found them  in some contaminated springs, but amphipods and/or
            isopods were usually absent.
                                     526

-------
      Session VII:
Ground Water Monitoring

-------
528

-------
            The Response of Landfill Monitoring Wells

                  In Limestone (Karst) Aquifers

     To Point Sources and Non Point Sources of Contamination


                      Ralph O. Ewers  Ph.D.

            Director, Groundwater Research Laboratory
       Department of Geology  Eastern  Kentucky University,
                  Ewers Hater Consultants Inc.
                       Richmond, Kentucky


Abstract
     The terms "point source" and "non-point source"  in the context
of groundwater contaminants are relative terms.   Landfills can be
considered non-point sources,  relative to a monitoring well at the
boundary of a  typical  landfill,  if contaminants  are  released at
hundreds of points,   Conversely,  the landfill is a point source
relative to these wells  if  contaminants are released at one or two
points.
     This author and several of his  colleagues  have maintained that
the convergent flow  in karst aguifers  should make monitoring wells
unreliable.  Implicit in these arguments was the assumption  that
the landfill was a  point source of  contamination.
     In  the  non-point  source case,  virtually  every  joint  and
bedding plane would be exposed to contamination.   Thus, any down-
gradient monitoring will which intersects the groundwater should
show  contaminants,   although  they may  be  unrepresentative  of
leachate from the entire landfill.
     In  the   point  source  case,  the monitoring  well  may  not
intersect  the  conduit  which carries  the  contaminant.    This
contention was  confirmed  at  a western  Kentucky landfill  where
rhodamine-WT was injected into an  up-gradient well, simulating a
point source  of contamination.  Two down-gradient monitoring wells
did not show  the dye.  although the dye appeared at three off site
springs approximately three miles distant.
     Therefore, neither contaminated  nor uncontaminated  landfill
monitoring wells in  karst  aguifers should give  us  confidence in
their reliability.
                                529

-------
                           Introduction

Landfills As Non Point Sources Of Contamination

     When a monitoring well is placed down-gradient from a landfill
an  assumption  is  made  that  it  will  intercept  a   plume  of
contamination  if  the landfill  leaks.   If the  leachate escapes
uniformly  over the entire  liner or  at hundreds  of  points, the
traditional system  of one monitoring well  up gradient and  three
down-gradient  will  signal  that  this  has  occurred   (Fig.  1).
Relative to the monitoring wells,  such a landfill is a  non-point
source of contamination.

     Although the leakage in this case is homogenous the  leachate
may not be. Should  the most hazardous waste be located  in a single
cell,  and  if  that  cell  is  located  along  a  flow vector midway
between two of the monitoring wells, dispersion may be insufficient
to  bring  this  substance  to  these  monitoring  points.    The
contamination  detected  by  the  monitoring system would not  be
representative of the chemical nature of the  leachate as  a whole.
Traditional monitoring  wells  may  not  provide  a representative
sample of leachate unless the landfill waste  is homogenous.


Landfills As Point Sources Of Contamination

     If the leachate passes through the landfill liner at a  small
number of discrete points, the traditional monitoring well system
may not signal that this has  occurred  or  it may give an incomplete
understanding  of  the  problem  (Fig.  2).   Only favorably located
points of  leachate  escape may  develop plumes which intersect the
wells.   Relative  to  the traditional monitoring wells, such  a
landfill is a point source of contamination.

     This   scenario  also  presents  problems  of  obtaining  a
representative sample.   Should the landfill  waste be  less than
homogenous, the most troublesome leachate may go undetected.  This
would  be  the  case  unless each  point of release  is sufficiently
distant from  the  well so  that dispersion has time to  spread the
contaminants,   or  unless  each  point  of  release is  directly up-
gradient from a monitoring well.


Landfills In Karst

     An additional  complication arises when  the  aquifer beneath
the landfill is composed of carbonate  rocks.   These aquifers  often
possess  very   little  primary  (intergranular) porosity  but they
exhibit  significant fracture  porosity  (secondary porosity) and
tertiary conduit porosity.   While contaminant plumes may develop
in the primary and secondary  porosity, the plumes  may be  truncated
and their contaminants entrained by the  conduit porosity  (Fig. 3).
                                530

-------
                         Background Monitoring Well
                     Landfill Boundary
Figure 1.  The contaminant plume  at  a hypothetical  landfill which
releases Iqachate  uniformly  over its entire liner.   The stippled
area indicates the  plume.
                         Background Monitoring Well
                     Landfill Boundary
                            Monitoring Wells

Figure 2.  The contaminant plumes at a hypothetical landfill which
releases  leachate  at  two discrete  points.   The stippled  area
indicates the plume.
                                  531

-------
                     Landfill Boundary
 Figure  3.   The contaminant plume at a  hypothetical  landfill in a
 karst aquifer with a high density of solution conduits.  Leachate
 is  released  at  two  discrete points.   The  presence  of  conduit
 porosity is  indicated by the  black branching line.
Figure 4.  The  contaminant  plume at a hypothetical landfill  in  a
karst aquifer with  a  low density of solution conduits.   Leachate
is  released  at  two discrete  points.    The presence  of  conduit
porosity is indicated by the black branching line.
                                532

-------
     If the landfill  leaks  at discrete points,  the point source
case, there  is a  considerable  likelihood that  the traditional
monitoring wells will  not  function  as they  were intended.  Quinlan
and  Ewers  (1985)  made  this   point,   although  they  did  not
specifically describe  the development of plumes and their spatial
relationship to the conduits.  The probability of  the wells working
as intended in  carbonate  aquifers  will depend not  only upon the
geometric relationship of the points of  leachate release and the
monitoring wells,  but also  upon the density of  conduits  in the
aquifer.   If the  conduits are widely spaced,  the probability of
their  being located   between  the contaminant   source  and  the
monitoring wells will  be  less  and the monitoring wells are more
likely to be efficacious (Fig.  4).   Conversely, if the density of
conduits  in the aquifer  is  great,  it would  be unlikely that the
wells would be  useful.

     If the landfill is a non-point source, virtually every joint
and  bedding  plane  beneath  the  landfill  and immediately  down-
gradient  would  be  affected by leachate.   Although some portion of
the  leachate may  be  intercepted by conduits,  any  nearby  down-
gradient  monitoring well will intercept some leachate.   However,
samples from such  a well are unlikely to be representative of the
landfill  leachate  unless the waste is homogenous.


A Kentucky Landfill Example

     A recent study at a Western Kentucky  landfill  confirmed the
contention that traditional  monitoring wells may be ineffective in
limestone aquifers.   Two  down-gradient monitoring  wells  at this
site were tested for efficacy by injection of a tracer into the up-
gradient  background monitoring well.  The  landfill  is  located in
a karst region.  The  aquifer beneath the  site is a  Mississippian
age limestone,  the upper member of  the  St.  Louis  Formation.

     Four pounds of 20 percent Rhodamine-WT dye was  injected into
the up gradient well.   The dye was  diluted in 50  gallons of water
before injection.   The well was first tested by  an injection of
several hundred gallons of  potable water.   This  insured that the
well would accept  the  dye and also  served to wet the pathways that
the dye  could  follow.  After  the  dye injection  several  hundred
gallons of potable water was injected into the well to flush the
dye into  the aquifer.

     The   results  of  the  dye  study are  depicted  in   Figure  5.
Rhodamine-WT from  the well  injection appeared in the Town Spring
at Cadiz, Kentucky and an  adjacent spring.   These  two  sites are
probably   part  of   a   local  distributary system.    This was  the
principle  location  of  dye  recovery  from  the  well  injection.
Rhodamine-WT was also  found  in very low concentrations at two other
springs,  one north and the  other southeast of the landfill.   The
                               533

-------
            N
                                           Muddy Fork, Little River
                                                     LEGEND


                                                   • Spring


                                                   O Swallet


                                                   Q Monitoring Well


                                                   A Dye Injection Well
            Miles
Figure  5.   Results of  a  dye study at  a Western  Kentucky landfill.
The  long  arrows  indicate  the  schematic  path  of the dye.    The
principal  dye recovery point  was at  the Cadiz  town  spring.   The
local gradient at the  landfill  is indicated by  the short arrows.
                                  534

-------
dye reached all of these sites nearly three miles distant in less
than four days.

     No dye was recovered from charcoal  dye detectors in the two
proposed down-gradient monitoring wells located 1100 and 1300 feet
from the dye injection well.   Both of these  wells were equipped
with airlifts which continuously circulated the water in the well
over the dye detectors.   Dye detectors  were  exchanged regularly
over a period of 38 days.   Apparently the tracer was intercepted
by conduit porosity before reaching the monitoring wells.


Appropriate Methods Of Monitoring In Karst

     Quinlan and Ewers (1985)  suggest that appropriate monitoring
methods  in  karst  should  utilize  springs.    When  it  can  be
demonstrated by proper dye-tracing  procedures  that  a particular
spring drains a landfill site it  may  be  the most appropriate and
dependable  monitoring  site.    However,  springs  are,  in  some
instances,  demonstrably inappropriate monitoring sites, even though
they  are  shown  to  be  connected  to  the  area  of  potential
contamination release.   This  is  the  case at  the  cited Kentucky
example for two reasons.   First, the spring  is quite  large and
therefore the dilution of landfill  leachate would  be very great.
Only  a  near catastrophic release  would  likely be  detectable.
Second,  the probable recharge basin for  the spring contains many
potential sources  of contamination.  It could be very difficult to
assure that a given contaminant was  the result  of landfill leakage
or due to some  other problem.

     A alternative monitoring  procedure which is more  likely to be
specific to the landfill  would  be to employ  a pumped monitoring
well at the site.   During dye tracing  programs  in karst, wells are
frequently employed as  dye  recovery sites.  Those that are even
modestly pumped during the tracing period often show the dye, those
which are not pumped rarely  do.  An  excellent example can be found
in the  work of Aley  at a proposed landfill   site near Pindall,
Arkansas (Aley, 1986).   In this study, dye was  recovered from five
wells, ranging  up  to a mile from both the dye injection point and
the spring resurgences where the dye appeared.  Apparently, local
gradients serve to carry dye,  and presumably contaminants, to the
conduits, establishing the convergent  flow which is characteristic
of these aquifers.   Pumping stress apparently reverses these local
gradients,  extracting  the  entrained  dye from the conduit (Fig. 6).

     If tracer  dyes can be recovered  from  pumped wells in karst,
by inference, contaminants could also be recovered from such wells.
A protocol for establishing a pumped  monitoring well  would be as
follows.

1- Locate the  lowest  point on the  potentiometric  surface,  down-
gradient from the site to be monitored.   This  could  be done with
                               535

-------
   Pumped Monitoring Well
Figure  6.    A pumped  monitoring well  lowers  the potentiometric
surface   (dotted  line)   near  a   conduit  which  is   carrying
contaminants.    This  pumping  stress  draws  contaminants  to the
monitoring  well  that would  otherwise  remain  entrained  in the
conduit.
                                536

-------
several small-bore  temporary  piezometers.   Low  points  of  the
potentiometric surface correlate well with the location of conduits
(Quinlan and Ewers,  1981).

2- Install a well at this point and  outfit it with a submersible
domestic well pump having a capacity  sufficient to create drawdown
in nearby piezometers in a short time.   A site so located should
be near conduits which may drain from the landfill.

3- Qualify the well with a dye test  from an appropriate up gradient
site or sites.   This would help to define the capture zone of the
proposed pumped monitoring well.  If the capture zone is sufficient
to encompass the site it could be used as a monitoring point.

4- Sample  the well for contaminants on the required schedule after
an appropriate  amount of pumping  has occurred.   The  amount of
pumping required could be determined  from "3" above.

Two monitoring wells of  this  type are ready to be installed at two
sites,  one in Kentucky and one in Puerto Rico.


Conclusions

     Clearly, the traditional monitoring wells in the cited example
cannot be  relied upon to accurately indicate the performance of the
landfill's clay  liner.    If  one  of  the wells  should show  the
presence of  leachate,  one cannot be confident that it monitors the
entire landfill or that it fairly represents the character of all
of  the leachate  that  is  released.    Neither  contaminated  nor
uncontaminated traditional  monitoring  wells  in  karst  aquifers
should give us confidence in  this method of  monitoring.   Springs
or pumped monitoring wells offer a  more reliable alternative.


REFERENCES CITED

Aley,  T.,  1986/1989.  Assessing presumptions  of negligibly  slow
downward water movement through deep residuum beneath a proposed
landfill site,  Pindall Arkansas. National Water  Well Association
Karst Hydrology Course Manual, 1989.

Quinlan, J.F. and R.O.  Ewers, 1985. Groundwater flow in limestone
terranes:  strategy rational and procedure for reliable efficient
monitoring of ground water quality  in karst areas.  Proceedings of
the Fifth  National Symposium and Exposition on Aquifer Restoration
and Ground Water Monitoring Proceedings,  pp 197-234.

Quinlan J.F.  and R.O. Ewers, 1981. Hydrogeology of the Mammoth Cave
Region, Kentucky,  in Roberts, T,G.,  ed. ,  1981  GSA Cincinnati '81
Field Trip Guidebooks:  Washington D.C., Am. Geol.  Inst., v. 3, p.
496-501.
                               537

-------
Biographical Sketch

     Ralph O. Ewers   is  professor of geology and director of the
Groundwater Research Laboratory at Eastern Kentucky University, and
a  principal  in  Ewers  Water  Consultants,  a  consulting  firm
specializing in carbonate aquifers.  His B.S.  and  M.S. degrees in
geology were earned at the University of Cincinnati and his Ph.D.
was  earned at  McMaster  University  (1982)   .  Professor  Ewers's
special interests include the applications of tracer and electronic
monitoring techniques to the  solution of practical environmental
problems  in  karst groundwater.    He  has more  than 30 years  of
research and consulting experience gained throughout much of North
America and  Europe  and  he serves  on several  state  groundwater
advisory boards.  He  is author or co-author of more than 100 papers
and reports dealing  with groundwater in karst.    For one  of the
papers concerning procedures for monitoring ground water in karst
terraces,  he and co-instructor James Quinlan received the 1986 E.B.
Burwell Award from the Geological  Society  of America  for a "work
of distinction which  advances  knowledge concerning principles of
practice of engineering geology."
                                538

-------
The Response  of Landfill  Monitoring Wells  In  Limestone  (Karst)
Aquifers To  Point Sources and Non-Point Sources of Contamination

Ralph O. Ewers

1.   In England and Wales, 56 domestic (municipal?)  waste sites are
located  on the karstified  carboniferous  limestone.    Unpumped
monitoring wells have been successfully used to monitor contaminant
plumes leaving the sites,  yet  only  4 springs have been  affected.
Why do you think so few springs have been adversely affected?
Why do you think the monitoring wells have been so successful?

ANSWER-   I suspect  that the  reason so few springs seem to be
contaminated is that  dilution  of the contaminants is  too great.
This  is one  of the  reasons  I  believe that springs  are often
inappropriate monitoring points.

     A successful  monitoring well can be defined several ways.  If
you define success as having no contamination, you  may be deluding
yourself.  The aquifer may  be contaminated  you are  just  unaware of
that fact.  If  you define  success as showing some contamination,
you may be unaware of the real  scope of the problem.


2.   Once you have determined  the most  appropriate well location
(at the  lowest  point on the potentiometric  surface),  how do you
determine the "appropriate" pumping rate?

ANSWER- I would judge the pumping rate and time on the basis of the
tracer tests that are used to  qualify the  well.   If pumping at a
given  rate gives  a  strong  dye  concentration  which appears  to
plateau after pumping 2000 gallons  of water, I would assume that
would be an appropriate rate and quantity-


3.   In many  county  landfills  in Tennessee, no municipal sewer
system is  available  to  receive water pumped  as you propose.  The
volume of water you imply would be pumped would be a problem.  Do
you have any suggestions for disposal of such pumped water?

ANSWER-  I  would  store  the  water until after  the analysis  was
complete and then  release it, if it is clean, to some appropriate
receiving stream,  assuming the  appropriate discharge permits were
obtained.   Alternatively,  it   could be  trucked  to  a disposal
facility.


4.   How often do  you propose to continue pumping to determine if
there is leakage?  What do you propose to do with the volume of
water?  Once the initial pumping test has been conducted and, for
example, the results  indicated that the landfill  is  not  leaking
are your proposing to cease pumping at the monitoring point until
the next scheduled monitoring date,  at which time you would begin
the entire pumping and sampling procedure again?  Would additional

                                539

-------
wells be installed to enhance interception of the  leakage?

ANSWER- I would pump as often as required by law or  good  science.
Perhaps  quarterly,  I  would prefer  to  pump  at  during  low-flow
conditions.  I am not sure that high-flow samplings  are necessary
as they may be for spring monitoring.

     See 3 above regarding disposal of the water.

     I  see  no reason  at this time  to  continue  pumping between
sampling periods.

     Additional pumped  monitoring wells  may  be required if the
landfill is large or if  the  dye  test results indicate that it is
infeasible to extend the capture zone over the required area.


5.   I  suspect  that  the  cost  of   continuing  to  analyze  for
pollutants in your well pumped to draw water from the  conduit, plus
the cost of water handling and disposal, would make  it cheaper to
combine a natural potential survey with your potentiometric map and
drill several "pilot" holes (let's say  5) as potential monitoring
wells.   If  one of the 5  holes  hits the conduit  (as proposed by
tracing) you have maximum reliability and no  problems with water
disposal.  Convince me that I am wrong!

ANSWER-  I  doubt  that the  natural  potential method  is developed
sufficiently at this time to find  the smaller  conduits which are
likely  to  be  associated with the  average landfill.   If  a large
conduit is suspected, the type investigated by A. Lang and others,
I would certainly advise using this technique.
                               540

-------
              DEVELOPMENT  OF  AN ASTM STANDARD GUIDE

        FOR  THE  DESIGN OF  GROUND-WATER MONITORING SYSTEMS

              IN KARST AND FRACTURED ROCK TERRANES
         Michael R. McCann, Westinghouse Electric Corp.
                      Bloomington, Indiana

             James F. Quinlan, Quinlan & Associates
                      Nashville/ Tennessee
ABSTRACT

     Permeability  in  karst  terranes  and  many  fractured  rock
terranes is developed in joints, fractures,  and bedding planes.  In
karst terranes  some  of these  three  discontinuities  have  been
solutionally  enlarged, sometimes to form conduit-like  caves.   In
both terranes the assumptions underlying the description of ground-
water flow as  taking  place  in a homogeneous,   isotropic,  porous
medium are usually  invalid.   As a result, ground-water  flow and
pollutant flow  in most  karst  terranes and  many fractured  rock
terranes  is  not  described by  the  radially dispersive  charac-
teristics of  flow  in  a granular  medium.   Monitoring  systems
designed with the assumption of  dispersive flow will not produce
accurate, reliable,  or efficient monitoring of either ground-water
flow or  pollutant flow.

     A Standard Guide, developed under the auspices  of ASTM Sub-
committee D18.21.09 on  Special Problems  of  Monitoring  Karst and
Fractured Rock Terranes, has been deemed necessary to  assist in the
design  and  implementation of  accurate and  reliable  monitoring
systems  in those hydrogeologic  settings which depart significantly
from the characteristics of a porous medium.  This Guide is based
on  recognized  methods  of monitoring  system design and  imple-
mentation for the purpose of collecting  representative ground water
data. The design guidelines  are applicable both to the detection
of  contaminant  transport  from  existing  facilities  and to  the
assessment of proposed facilities.   The recommended procedures of
the Guide are designed to obtain representative aquifer and ground-
water information.  Use of the Guide will assist  in the development
of accurate conceptual hydrogeological models which are integral to
the design of a time-sensitive, relevant,  and reliable monitoring
program.
                                541

-------
INTRODUCTION

     The American Society for Testing and Materials, better known
as ASTM,  was organized  in 1898  and has grown into  one  of the
largest voluntary standards development systems in the world.  It
is a non-profit organization that provides a forum for producers,
users,  and  all others  having a  general interest  in materials,
products, systems,  and services.  Its 134 committees and additional
subcommittees  write   standard  test   methods,  specifications,
practices, terminology,  guides,  and classifications for metals,
paints, plastics,  textiles, petroleum,  construction, energy, the
environment,  consumer  products,  medical  devices,  computerized
systems, electronics, and many other areas.   More than 8500 ASTM
standards are published each year in the 68 volumes  of the Annual
Book of ASTM Standards.

     One of the 134 ASTM Committees, D18,  is concerned with soil
and  rock.    One of  its  subcommittees, .21,  is concerned with
groundwater  monitoring,  and  it  has  been divided   into  10 Task
Groups, one of which,  21.09, is concerned with monitoring in karst
and fractured rock terranes.  This  paper is  written to tell what
that task group is doing and  why.   But first,  we wish to briefly
establish that the problems of karst and fractured  rock aquifers
are different from those of other aquifers.


HOW KARST AND FRACTURED ROCK AQUIFERS PHYSICALLY DIFFER FROM OTHERS

     The main  physical  difference between aquifers  in karst and
fractured rock terranes and aquifers in other terranes  is the type
of permeability  (effective porosity).   In aquifers  consisting of
unconsolidated material,  and  in many  consolidated  aquifers, the
permeability  is a  primary feature of composition and  deposition.
In  fractured and  karst aquifers,  however,  the  permeability is
secondary,  a result  of  fractures,  bedding  planes,  and commonly
dissolution along them.   Underlying the description and analysis of
ground water and contaminent  flow is the implicit assumption that
the  aquifer  is behaving as a porous medium.   Even  though it is
widely recognized that ground-water flow in many rock  aquifers is
through secondary openings, the assumption that it is  behaving as
a  porous  medium   is  still   made   at  some  working  scale  of
representation.  The problem  in many fractured  rock  aquifers, and
most karst aquifers, is that this  assumption  is  made  uncritically.


HYDRAULIC AND HYDROLOGIC DIFFERENCES

     A  result  of  the  porous media assumption  is  the further
assumption that the aquifer is homogeneous and  isotropic,  and that
flow  is  laminar.   The range in aquifer characterstics within the
spatial  limits of the  aquifer  is  frequently  large in  fractured
rocks,  and  most  karst  aquifers  are  strictly non-homogeneous.
Fractured rock and karst aquifers  are typically  highly anisotropic


                               542

-------
in  three  directions.  Hydraulic  conductivity  and  ground-water
velocity can range over several orders of magnitude, depending upon
the direction of ground-water movement. For  example,  in extreme
cases in some karst  aquifers, ground water appears  to be moving
parallel to the  equipotential  lines of a  mapped potentiometric
surface (Arnow,  1963) .   This  is,  of  course,  due to an inadequate
amount of  data points used to  describe the surface.  Nevertheless,
until another means  is utilized  to  check the  ground-water flow
direction,  a potentiometric  map  can be misinterpreted to  be a
correct representation  under the porous media assumption.

     Flow in  porous media  is  assumed to  be  laminar  over the
macroscopic scale of investigation.   Although this assumption is
reasonable for the  average of  hydraulic  properties  in  a porous
medium,  water  movement  in  fractures,  especially  solutionaly
enlarged fractures,  is  often rapid and turbulent.  This is due not
only to size of the  apertures in the rock, but also to the many
direct inputs from  the surface to the  aquifer.  This rapid and
turbulent flow in both recharge and  movement through the aquifer
often  results   in  large  variations  in  head  in  response  to
precipitation,  and large spatial and temporal variations in water
chemistry.   Monitoring wells  installed  in  these terranes vary in
the number and type of  fractures  encountered from well  to well.
 Some wells encounter  larger  fractures with good  connection to
surface inputs or major  conveyances; others encounter relatively
"tight" and  non-transmissive  fractures.    The  result  is poten-
tiometric maps  that can be difficult to interpret or are entirely
misleading.   For example, Figure 1 shows a potentiometric map for
a site in  an Indiana  karst terrane during the  February wet season.
In  fact, ground water  from  this site drains to the southwest, as
demonstrated by tracer tests.   Such  a conclusion, however, could
not be interpreted from this map.  The tracer  tests were needed to
test and establish flow directions and destinations.
DIFFERENCES IN DESCRIPTION OF GROUND-WATER AND CONTAMINANT FLOW

     Standard equations describing ground-water flow are based on
Darcy's Law which assumes laminar flow in a porous medium, and is
invalid in nonlaminar flow and other media.  The equations are not
valid for karst and fractured rock aquifers because flow in them is
neither  laminar nor  in  a porous  medium.    This  means  that the
parameters  for  transmissivity  and  storativity  which  may  be
calculated from pumping tests are not representative of  the aquifer
as a whole.    Consequently, predictions  of contaminant movement
based on assumed radial  dispersion and advection, and calculated
using these aquifer parameters, will be erroneous and misleading in
both direction and magnitude.  Monitoring systems designed on  false
assumptions will  not  produce  accurate,  reliable,  efficient,  or
time-sensitive  monitoring.    Examples from  case  studies  of the
inadvisability  of  applying  the porous-media  assumption  to the
design  of ground-water  monitoring  systems are  given  by:  Aley
(1988),  Bradbury  et al.  (1990),  Crawford  (1988),  Field  (1988),
McCann and  Krothe  (1991),  Quinlan and Ewers  (1985),  and Quinlan
                                543

-------
                                                                 N
                                                                    1W-4S
                                            0	200

                                            Scale ir\ feet
Figure 1 - Lemon Lane Landfill  Potentiometric Map, February 1988
           (after McCann and Krothe, 1991)
                                  544

-------
(1989).   Table 1 gives a comparision of characteristics of porous
media,  fractured rock aquifers, and karst aquifers, and highlights
their significant differences.


WHY A STANDARD GUIDE IS NECESSARY

     Quinlan (1989), in the foreword  of  his document written for
the EPA, "Ground-Water Monitoring in  Karst Terranes: Recommended
Protocols and Implicit Assumptions", stated four major reasons why
that document was necessary.   They are also four good reasons why
an ASTM Standard Guide is necessary:   1)  The  hydrology of karst
terranes  is  significantly   different   from  that  of  terranes
characterized by granular  and fractured rocks — flow velocities in
karst may  be several  orders  of magnitude higher than  in other
ground-water  settings; Darcy's  Law  describing  flow  is  rarely
applicable; 2)  For monitoring  to  be relevant and reliable in karst
terranes, monitoring procedures  must  be  radically different from
those in non-karst  terranes;  3)  There is  a need  for a practical
guide  that   tells  engineers,  geologists,   hydrologists,   and
regulators what the monitoring problems are in karst terranes and
how to solve them; and  4)  Create  awareness  of the state-of-the-art
in  monitoring  in  karst  terranes  — and  provoke  thought  and
discussion about the subject and  its implications for ground-water
protection strategy.  Although Quinlan (1989) focused specifically
on karst terranes, ASTM Subcommittee D18.21.09 recognized that many
fractured rock aquifers share similar problems in monitoring design
and decided  to include them  so  the  Guide  would  have the widest
practical application possible.

     Almost 20% of the United States and 40% of  the country east of
Tulsa,  Oklahoma,  is  underlain by carbonate rocks.  The  extreme
vulnerabilty of  carbonate aquifers  in these terranes to contam-
ination make it imperative that a reliable guide to monitoring be
made available.  Very few  state agencies have guidelines that take
into account aquifers that do not behave as  porous  media.  However,
some  states such  as Alabama, Indiana,  Kentucky, Missouri,  and
Tennessee have begun  to require  water tracing  studies to confirm
ground-water flow  directions  in  their karst terranes.   Although
some EPA documents  do address the problem of  monitoring in non-
porous-media aquifers, most do not.  In particular, the Technical
Enforcement Guidance  Document  (TEGD)  which governs monitoring at
RCRA sites assumes a  porous media exists in most applications of
its monitoring guidance.   Its one brief discussion of  karst, p. 66-
68, is ambiguous and probably  erroneous;  no data are given for the
conclusions  reached.    Few universities,  in  their  hydrogeology
programs, specifically address how to characterize aquifers that do
not behave as porous media.

     A Standard Guide, developed  through  the extensive peer review
and consensus  process of  ASTM,  will provide  a much  needed and
nationally  disseminated  standard  on which  to  base  design  of
monitoring systems  in karst  and fractured rock  terranes.  Such  a
Standard Guide does not yet exist.  The thorough and complete ASTM
                                545

-------
AQUIFER
CHARACTER-
ISTICS
Permeability


Flow

Isotropy

Homogeneity

Flow
predictions
Storage




Head
variation

Water
chemistry
variation
AQUIFER TYPE
POROUS MEDIA FRACTURED ROCK
Mostly primary


Slow, laminar

Most
isotropic
Most
homogeneous
Darcy's law
usually applies
Within
saturated zone



Minimal
variation

Minimal
variation

Mostly secondary


Possibly fast
and turbulent
Less
isotropic
Less
homogeneous
Darcy's law
may not apply
Within
saturated zone



More
variation

More
variation

KARST
Almost
entirely
secondary
Likely fast
and turbulent
Highly
anisotropic
Non-
homogeneous
Darcy's law
rarely applies
Both in
saturated &
un-saturated
(epikarstic)
zone
Can have
extreme
variation
Can have
extreme
variation
Table  l  - Comparison  of porous,  fractured,  and karst  aquifers
(adapted from Bradbury et al., 1990)
                                546

-------
process would best produce  the necessary document which would have
the widest practical application and acceptance.  Such a document
would be most useful to three groups of people:  1) Administrators
who must  evaluate existing or  proposed networks  for monitoring
water  quality in  karst  or  similar terranes,  but  have  minimal
experience in non-porous-media hydrogeology;   2)  Consultants and
others who must  design monitoring networks but may or may not have
extensive experience  in  karst or  similar  terranes; and   3)  The
well-experienced water tracer who  is  already  familiar with the
hydrology and geomorphology of karst or  similar terranes  but has
minimal familiarity with monitoring problems.


HOW  A  STANDARD  GUIDE   WILL  ASSIST  IN  DEVELOPING A  RELIABLE
MONITORING SYSTEM

      As to be explained within the Standard Guide, there are three
main objectives  to be accomplished in assisting  the development of
a  reliable monitoring system:    1)  The  Guide will aid  in the
accurate  characterization  of  karst- and  fractured-rock aquifers
because the development of  a conceptual hydrogeological model that
identifies and defines the various components of the  flow system is
a necessary first  step to  the design of  a  monitoring system;  2)
The Guide will be based on  recognized methods of monitoring system
design   and   implementation  for   the  purpose   of  collecting
representative  ground-water  data.  The  design  guidelines  are
applicable both to  the  determination of  ground-water flow and
contaminant  transport  from  existing  sites  and  assessment  of
proposed  sites;    3)  The  objectives of  the  Guide are to recommend
procedures for obtaining  information on aquifer characteristics and
representative water quality.

     In  order  to  accomplish these objectives,  the Guide  will
discuss   and   compare   qualitative   differences    in   aquifer
characteristics  between  porous,  fractured,  and karst media. The
special characteristics of karst and fractured-rock  aquifers will
be  pointed out,  and explanations  of  why  typical   investigative
procedures  often  fail   to give  accurate  information  will  be
provided.  The  Guide will recommend procedures to  obtain repre-
sentative aquifer and ground-water information. These recommended
procedures include:  1) Criteria for determining whether or not an
aquifer under investigation  is  behaving as a porous medium;  2)
Procedures  for   potentiometric   mapping;     3)  Procedures  for
interpreting  aquifer tests;   4)  Criteria  for  determining  when
tracing is necessary;  5) Procedures for conducting  tracer tests;
6)  Criteria   for  determining monitoring station  location;    7)
Criteria  for  establishing  sampling protocol;   8)  Procedures for
interpreting  chemical  and hydrologic  data;   and    9)  Meeting
regulatory requirements.

     It is not the intent of the Guide to provide a rulebook on how
monitoring must  be conducted in non-porous-media aquifers.  Rather,
the members  of  ASTM Sub-committee D18.21.09 wish to provide the
users of the Guide with the criteria and procedures  to conduct an
                               547

-------
investigation  so  that they  may  determine  the  most  accurate,
reliable,  efficient,   and  time-sensitive   monitoring  program
possible.

     People interested  in  assisting in formulating and  reviewing
the Guide  described  herein are invited to  contact either of  its
Task Group co-chairman,  the authors of this  paper.   Such parti-
cipation involves attendance at meetings,  review of documents,  and
voting  on drafts  proposed  for  adoption.    Membership  in  sub-
committees developing Standards or Standard Guides is  open to all,
but only members of ASTM can vote on them.
REFERENCES CITED

Aley, T., 1988.  Complex radial flow of ground water on  flat-lying
     residuum  -  mantled limestone  in the Arkansas  Ozarks.   In:
     Environmental Problems in Karst  Terranes and Their Solutions
     Conference  (2nd,  Nashville,  Tenn.),  Proceedings.   National
     Water Well Association, Dublin,  Ohio.  pp. 159-170.

Arnow, T., 1963.  Ground  water  geology of Bexar County,  Texas.
     U.S. Geological Survey, Water Supply  Paper 1588.   36 pp.

Bradbury, K.R., Muldoon, M.A., Zaporozec,  A., and Levy, J..  1991.
     Delineation of wellhead protection areas in fractured rocks.
     Guidelines for Delineation of Wellhead Protection Areas, U.S.
     Environmental  Protection  Agency,  Office  of   Ground-Water
     Protection, Washington, D.C..  EPA 570/9-91-009.   144 pp.

Crawford, N.C.,  1988.   Karst  ground  water  contamination  from
     leaking  underground  storage  tanks:  Prevention,  monitoring
     techniques,  emergency   response  procedures,    and  aquifer
     restoration.  In:  Environmental Problems in Karst Terranes and
     Their Solutions Conference (2nd,  Nashville, Tenn.), Proceed-
     ings.  National Water Well Association.   Dublin,  Ohio.  pp.
     213-226.

Field, M.R., 1988.  U.S. Environmental Protection Agency's strategy
     for  ground-water  quality  monitoring at hazardous  waste land
     disposal  facilities located  in karst terranes.  In:  Inter-
     national  Association  of  Hydrogeologists,   Congress   (21st,
     Guilin, China), Proceedings, vol.  2,  pp. 1006-1011.

McCann, M.R.,  and Krothe, N.C., 1991.  Development of a  monitoring
     program   at  a  Superfund  site   in   a   karst   terrane  near
     Bloomington, Indiana.  In: Hydrogeology,  Ecology, Monitoring,
     and  Management  of Ground Water  in Karst Terranes  Conference
      (3rd, Nashville,  Tenn.),  Proceedings.  National Ground  Water
     Association.  Dublin, Ohio.   (in press).
                                548

-------
Quinlan, J.F., 1989.  Ground-water monitoring  in karst  terranes:
     Recommended  protocols,  regulatory  problems,  and  implicit
     assumptions.      U.S.   Environmental   Protection   Agency,
     Environmental  Monitoring  Systems  Laboratory,  Las  Vegas,
     Nevada.    EPA/600/X-89/050.    [Draft;  final  version to  be
     released 1991]   79  pp.

Quinlan, J.F.  and  Ewers, R.O.,  1985.  Ground water  flow in lime-
     stone terranes:    Strategy,  rationale,   and  procedure  for
     reliable,  efficient  monitoring of  ground-water quality  in
     karst areas.    In:   National Symposium  and Exposition  on
     Aquifer  Restoration  and   Ground   Water  Monitoring  (5th,
     Columbus,  Ohio), Proceedings.   National  Water  Well Associ-
     ation.   Worthington,  Ohio.  pp. 197-234.
BIOGRAPHICAL SKETCHES

     Michael R. McCann is Senior Project Geologist for Westinghouse
Electric Corporation's Bloomington Project. He supervises geologic
and hydrologic investigations and ground water monitoring for the
six PCB-contaminated Superfund sites near Bloomington,  Indiana.
Previously he was  a  consultant with Skelly and Loy Engineers in
Lexington, Kentucky,  and prior  to that he was  hydrogeologist for
the Kentucky  Division  of Water.  Mr.  McCann was  also assistant
geologist at Mammoth Cave National  Park.   He holds a B.S. degree
from Indiana  University, a  M.S.  degree  from  the  University of
Kentucky, and is  a  Certified  Professional  Geologist with the state
of Indiana.  Mr. McCann is currently  serving as co-chairman of ASTM
subcommittee D18.21.09  on Special Problems of Monitoring Karst and
Other Fractured Rock Terranes.

     Michael R.  McCann
     Westinghouse Electric Corp.
     P.O. Box 997
     Bloomington, IN  47402
     812-334-0030
     Dr. James  F.  Quinlan,  P.G.,  is president  of  Quinlan and
Associates, a consulting firm specializing in problems  of carbonate
terranes.  He was  Research Geologist  for the National  Park Service
at Mammoth  Cave, Kentucky, for 16 years and has been an independent
consultant  on karst for more than 10 years.  He earned a Ph.D. in
geology  at  the   University  of  Texas  at  Austin  (1978) .    His
experience  includes 36 years of  research and observations in karst
terranes of 26  states,  2 territories,  and  23 countries and work as
a  consultant in  many of  them.   This  experience has  included
environmental applications  of  dye-tracing, evaluation  of waste-
disposal sites,  design of ground-water monitoring networks, peer-
review of  reports and both analysis  and  remediation  of sinkhole
development.  He  has  written  or co-written more  than 160 publi-
cations on  karst-related topics.   For one  of the papers concerning
                                549

-------
procedures for monitoring ground  water in karst terranes, he and
co-author Ralph Ewers received the 1986 E.B. Burwell Award from the
Geological Society of America.  For  the past 6 years, he and two
others  have  annually  taught a  NWWA  course on  Practical  Karst
Hydrogeology, with Emphasis  on Ground-Water Monitoring.   He served
for  4  years as  a Director  of  the  Association of  Ground  Water
Scientists and Engineers and is  co-chairman of ASTM Subcommittee
D18.21.09 on Special Problems  of Monitoring Karst and  Fractured
Rock Terranes.

     Dr. James F. Quinlan
     Quinlan & Associates,  Inc.
     P.O. Box 110539
     Nashville, TN  37222
     615-833-4324
                                550

-------
DEVELOPMENT OF AN ASTM STANDARD  GUIDE  FOR THE  DESIGN OF GROUND
WATER MONITORING SYSTEMS IN KARST AND OTHER FRACTURED ROCK TERRANES

by Michael R.  McCann and James F. Quinlan


It is appropriate  to mention that funding of the entire  ASTM ground
water monitoring  standards effort — all  10 committees  — is being
paid  for  by  the  U.S.  Navy,  the  U.S.   Environmental  Protection
Agency, and the U.S.  Geological Survey.

We gladly acknowledge this support.
                             551

-------
552

-------
GROUND-WATER REMEDIATION MAY BE ACHIEVABLE IN SOME KARST AQUIFERS

THAT ARE CONTAMINATED, BUT IT RANGES FROM UNLIKELY TO IMPOSSIBLE

       IN MOST:   I.  IMPLICATIONS OF LONG-TERM TRACER TESTS

            FOR UNIVERSAL FAILURE IN GOAL ATTAINMENT

            BY SCIENTISTS,  CONSULTANTS,  AND REGULATORS


               JAMES F.  QUINLAN and JOSEPH A.  RAY*
                    Quinlan & Associates,  Inc.
                       Nashville,  Tennessee

                             *Now with
                    Kentucky Division of Water
                       Frankfort, Kentucky



                        EXTENDED ABSTRACT

     Not  a  single contaminated aquifer in  the United States has
been confirmed to have been  successfully restored  (remediated) by
conventional  "pump and treat" technology!  At  present,  "No matter
how much money the federal government is willing to spend,  contami-
nated aquifers can not be restored to a condition compatible with
health-based  standards." (Travis & Doty, 1990).  Numerous distin-
guished authorities  agree  (Feenstra and Cherry, 1988; Freeze and
Cherry, 1989; Mackay and Cherry, 1989; Mackay,  1990; Rowe, 1991).

     Many  karst  hydrogeologists,   including  the authors,  have
believed  that aquifer  remediation  in  karst terranes  might be
quicker  and more easily achieved  as a consequence  of  the rapid
flow-through times and resultant short residence-times characteris-
tic of  most karst  aquifers.   This might  be  so.   This might be
especially  so in  highly  integrated,  maturely  karsted  aquifers
typified by much of  the Mammoth Cave area, much of western Kentucky
and southern  Indiana, and  much of the midwest and southeast.  We
now, however, doubt the probability of quick  and  easy remediation
and self-cleansing  —  except where  contaminants might enter the
aquifer  through  swallets  and  sinkholes  draining  directly  to
underground streams and to springs.  Recent multi-tracer  investiga-
tions, begun  in the epikarst zone of karst aquifers  characterized
                                553

-------
chiefly by diffuse flow, strongly suggest that the probability of
remediation in most karst aquifers, diffuse- or conduit-flow, may
range  from unlikely  to  nil.    (The  tracer dyes  were  used as
surrogate pollutants.)

     Interpretation of judicious extrapolations of tracer-recovery
data from  a total of  9 randomly-located dye-injection wells at
sites in the folded Appalachians of Pennsylvania and Tennessee —
as  contrasted  with discrete  inputs at  sinking streams  or  cave
streams  — data  from  more  than  a thousand  water  samples and
hundreds of activated-charcoal-detectors, all regularly collected
over periods ranging from 6 to 20 months, strongly indicates  that
dye-residency  times  in  the  sub-water-table part  of  many karst
aguifers  can   range  from several  years  to many  tens  of years
(Quinlan,  1992;  Quinlan and Ray,  1991;  Quinlan, et  al. , 1990a,
1990b,  1991) .

     Application of "pump and treat" technology at an Appalachian
superfund site in Beekmantown (Knox) dolomites has prevented  some
migration  of dye from  it, but  has not  diminished  the long-term
concentration of dye in the aquifer.  Similar long-term residency
of  dye  (and,  therefore,  of  pollutants)  in  parts  of epikarst
aquifers propinquitous to those parts with rapid flow-through and
short-term residency has been recognized  by  others also  (Smart and
Friedrich, 1986; Alexander et al., 1991).

     An assumption that rapid remediation might be characteristic
of karst and epikarst  aquifers  in,  for example,  the Mammoth Cave
area,  ignores  the  following  facts:   1)  More than  99%  of the
approximately 600 dye-tests performed there  have been from  sinking
streams and cave  streams  rather than from wells randomly  drilled
into the  diffuse-flow part of the  aquifer  (between conduits and
their tributaries), and 2) No tracer tests have been performed in
the epikarst there.  The time necessary for possible remediation of
an aquifer in  the Mammoth Cave area  would be highly variable and a
function  of where  pollutant input occurs  above  the  aquifers.
Remediation there could be practically impossible.

     These conclusions are applicable to most karst areas  and are
relevant to interpretation of  traces performed as  part of spill
response in most karst terranes.


                         REFERENCES  CITED

Alexander, E.G., Jr.,  Alexander,  S.C.,  Huberty, B.J., and Quinlan,
     J.F.   The Oronoco  Landfill dye  trace III: Results  from a
     Superfund remedial investigation in  a glaciated, diffuse-flow
     karst.  Hydrogeology, Ecology, Monitoring, and Management of
     Ground Water in Karst Terranes Conference  (3rd, Nashville,
     Tenn.),  Proceedings.   National  Ground Water Association,
     Dublin, Ohio.  [in this volume]

Feenstra, S.,  and Cherry, J.R.   1988.   Subsurface contamination by
                               554

-------
     dense non-aqueous phase liquid  (DNAPL)  chemicals.   Interna-
     tional Association of Hydrogeologists, International Groundwa-
     ter Symposium (Halifax,  N.S.),  Proceedings, p. 16-69.

Freeze,  R.A.,  and  Cherry, J.A.  1989.  What has gone wrong?  Ground
     Water.  27 (4) :458-464.

Mackay,  D.M.   1990.   Characterization of the distribution and be-
     havior of contaminants in the  subsurface,  in Water Sciences
     and Technology Board.   Ground Water and Soil Contamination
     Remediation:  Toward  Compatible  Science,  Policy and Public
     Perception.   National Academy Press, Washington, B.C.  p. 70-
     90.

Mackay,  D.M.,  and Cherry,  J.A.   1989.  Groundwater contamination:
     Pump-and-treat remediation.   Environmental  Science and Tech-
     nology.   23:630-636.

Quinlan, J.F.   1992.  Interpretation of dye-concentration and dye-
     recovery curves from four simultaneous interwell traces in a
     karst aquifer.   Privately published Christmas  card, [Nash-
     ville, Tenn.] 1 p.

Quinlan, J.F., and Ray, J.A.  1991.  Application of dye-tracing to
     evaluation of a landfill site  in a  karst terrane:  Rationale
     and an Appalachian case study (abs.).   Geological Society of
     America,  Abstracts with Programs.  23(3):54.

Quinlan, J.F., Ray,  J.A.,  and Elliott, W.G.   1990a.  Flowpath de-
     lineation, interpretation,  and control in Pennsylvania karst
     terrane:  Utilization of dye-tracing to successfully accomplish
     and monitor these objectives  (abs.).   Geological  Society of
     America,  Abstracts with Programs.  22 (7):371.

Quinlan, J.F., Ray,  J.A.,  Elliott, W.G., Smith, A.R., and Behrens,
     H., 1990b.   Interwell tracing  with  simultaneous use of four
     different fluorescent dyes  in a Pennsylvania karst terrane
     (abs.) Geo ,  17(2-3):82.

Quinlan, J.F-, Ray, J.A., and Schindel, G.M.   1991.  Application of
     dye tracing to  evaluation of a  landfill site  (abs.),  in Kast-
     ning, E.H.,  and Kastning, K.M.,  eds.  Appalachian Karst.  Na-
     tional Speleological Society,  Huntsville,  Ala.  p.  168.

Rowe, W.D., Jr.   1991.   Superfund and  groundwater remediation:
     Another perspective.  Environmental Science and Technology.
     25(3):370-371.

Smart,  P.L.,  and Friedrich,  H.  1986.  Water movement in the un-
     saturated zone of  a maturely  karstified carbonate aquifer,
     Mendip Hills, England.   Environmental  Problems in Karst Ter-
     ranes and Their  Solutions Conference  (Bowling  Green,  Ky.),
     Proceedings.   National  Water Well Association, Dublin, Ohio.
     p.  59-87.
                               555

-------
Travis, C.C., and Doty, C.B.  1990.  Can contaminated  aquifers at
     Superfund  sites  be  remediated?   Environmental  Science and
     Technology.  24 (10):1464-1466.
                      BIOGRAPHICAL SKETCHES

Dr. James F- Quinlan, P.G., is president of Quinlan & Associates,
Inc.,  a  consulting  firm specializing  in problems  of carbonate
terranes.  He was Research Geologist  for the National Park Service
at Mammoth Cave,  Kentucky,  for 16 years and has been an independent
consultant on karst  for  more than  10 years.   He earned a Ph.D in
geology at the University of Texas at Austin  (1978).  His experi-
ence  includes 36  years of  research and  observations  in karst
terranes of 26 states and 25 countries, and work as a consultant in
many of them.  He has written or co-written more than 170 publica-
tions on karst-related topics.   He is co-chairman  of ASTM Subcom-
mittee D18.21.09 on Special Problems  of  Monitoring Karst and Other
Fractured Rock Terranes.

Dr. James F. Quinlan
Quinlan & Associates, Inc.
Box 110539
Nashville, TN 37222-0539
(615) 833-4324


Joseph A. Ray is now Senior Hydrologist  for the Groundwater Branch
of the Kentucky Division of Water where  he  is  involved in evaluat-
ing vulnerability of aquifers throughout the state, but especially
karst aquifers.   Previously,  he was  a  karst  hydrogeologist with
Quinlan  and  Associates,  Inc.,  and the  National  Park  Service at
Mammoth Cave, Kentucky.   His graduate work  on  environmental change
in Columbia, South America, earned a Master's  degree in geography
at Western Kentucky University.  He is extensively experienced in
karst field  investigations and dye tracing.   His other research
interests  include  the  origin of  prairies and savannas  in humid
regions, and geomorphology.

Joseph A. Ray
Groundwater Branch
Kentucky Division of Water
18 Reilly Road
Frankfort, KY  40601
(502) 564-3410
                               556

-------
Ground-water remediation may be achievable in some karst aquifers
that are contaminated, but it ranges from unlikely to impossible in
most: I. Implications of long-term tracer tests for universal fail-
ure in goal attainment by scientists, consultants,  and regulators

By: James  F. Quinlan and Joseph A. Ray

Q.  Have you  seriously  questioned how  many of your  fluorescein
    recoveries are your dye and how  many  are from  other sources?
    Also,   there  can  occur  what  some believe  to  be  "breakdown
    products" of Rhodamine  WT that  have  special  characteristics
    similar to those of fluorescein.  We  have seen this in long-
    term tests in Missouri.  Also, we sometimes find  a long-term
    background for fluorescein.

A.  Yes.   Paranoia  is a healthy  state  of mind.   For procedural
    reasons too lengthy to summarize here,  we are satisfied that no
    tests  we have run have  had recovery  of  fluorescein from the
    tests  of others, from industrial  or household  background,  or
    from Rhodamine WT breakdown products.   In brief,  however, the
    possibility of such false positives can be minimized by repeat-
    edly monitoring for  them in background before the tests and by
    sampling often  enough  so  that one  can  closely and reliably
    monitor the breakthrough,  peaking, and  recession  of all dyes
    used,  and by  recognizing that landfills and  gasoline stations
    can be  a  source of fluorescein  (from the  colorant  in anti-
    freeze) .

    The breakdown of Rhodamine WT to something that might be con-
    fused  with fluorescein was first reported by Jim Duley in the
    proceedings of the first NWWA karst  meeting, in 1986 (p. 396-
    397) ,  but  we  are unaware  of  any further study of  such com-
    pounds.  Dave Sabatini has  studied the properties of Rhodamine
    WT (Ground Water, 1991.  29(2):341-349) and shown that it actu-
    ally consists of two structural isomers with different sorption
    tendencies (EOS,  1991, 72(44):154).  These isomers explain why
    Rhodamine WT  breakthrough curves commonly have a plateau at a
    C/C0 value  of about  0.4.

    As for the long-term fluorescein background in Missouri, all
    one can  do is  repeatedly monitor with a  scanning spectro-
    fluorophotometer, not trace  indiscriminately,  and (seemingly
    contradictorily)  use enough dye to unambiguously exceed back-
    ground  and the known  fluctuations  of  it.    The problem  of
    accumulation  of fluorescein background within a karst aquifer
    has also been recognized in Switzerland and  is the subject of
    a recent paper by Aurele Parriaux and two  colleagues (Hydrogeo-
    logie,  1990.   (3):183-194).

Q.  When will regulatory agencies  recognize  the futility of com-
    plete  remediation in Appalachian karsts?  How do we make them
    realize this?  They insist  on pumping and treating  for up  to 10
    to 20  years.   Costs skyrocket and frustration rules.
                               557

-------
We wish  we knew.  We  are fully sympathetic  to the physical
impossibility of remediation  of  some aquifers and agree with
the authors  we  cite.   Nevertheless,  regulatory agencies are
required  to  develop,   implement,  and enforce  regulations as
enabled by state or federal  legislation.   Many states have a
non-degradation policy or legislation that requires remediation
to background or detection limits.  Such legislation is created
by politicians,  many of which do not necessarily understand the
complex and technical  issues  involving hazardous waste remedi-
ation problems.   They  do understand that the voting public has
zero tolerance for any level of contamination — irrespective
of risks, potential exposure,  or cost of remediation.   Some-
times, the fault is with well-meaning  regulators, legislators,
and legislative  staff  who, for various reasons, have written
rules or statutes more  stringent than is necessary or intended.
We do not  mean  to  seem naive, but we see the first  steps in
improvement of  communication  between regulatory agencies and
both the regulated and  their consultants are to be education of
all parties  as  to  the nature  of reality,  establishment of
mutual respect for the technical competence and the integrity
of both  consultants  and regulators,  and creation of  a non-
adversarial climate.   If mutual good faith is established, and
if there are  alternative  interpretations  of the regulations,
the more  reasonable one  can be  sanctioned.   These desirable
goals are more easily stated than obtained.
                           558

-------
            Session VIII:

Emergency Response and Ground-Water
            Management

-------
560

-------
The Use of Groundwater-Level Measurements and Dye Tracing to Determine the Route of Groundwater

                      Flow from a Hazardous Waste Site in an Area of Karst

                                   in Hardin County, Kentucky


                   C. Britton Dotson1, Nicholas C. Crawford2, and Mark J. Rigatti3

                                       'Roy F. Weston, Inc.
                                       Florence, Kentucky
                                  2Western Kentucky University
                                     Bowling Green, Kentucky
                              3O.H. Remediation Services Corporation
                                         Atlanta, Georgia
ABSTRACT

The nature of karst groundwater flow dictates that groundwater investigations in karst areas be approached
in a different manner than those in more traditional hydrogeologic settings. Water-level measurements over
relatively large areas and dye tracing have proven to be suitable techniques for determining characteristics
of groundwater flow in karst areas.  This paper presents a case study of a two-phase investigation consisting
of, first, water-level measurement and  identification of hydrogeologic features and, second, dye tracing to
determine the groundwater flow route from the Middleton Hazardous Waste  Site in Hardin  County,
Kentucky.

Groundwater measurements were obtained over an area of approximately 70 square miles.  From these
measurements a water-table contour map was constructed. Based on the slope of the water-table gradient
and the form of the water-table contours, a hypothesis was formed: groundwater flowing from the Middleton
Hazardous Waste Site would resurge  at a spring 4.5 miles to the west.  This hypothesis was tested by
performing dye traces.

Passive dye receptors were placed at all groundwater resurgences, interface locations and, where possible,
cave streams in the study area. An automatic water sampler was placed at the hypothesized  resurgence.
Dyes  were injected into a sinkhole collapse adjacent to the site and at a nearby sinking stream.  After
allowing sufficient time for the  dye to flow through the system, the passive dye receptors and water samples
were analyzed for presence of  the dyes. The hypothesis was proven correct by the dye traces.

Based on interpretation of the  water-table contour map and results of the dye traces (both qualitative and
quantitative) the direction, destination, velocity, and general route of groundwater flow from the Middleton
Hazardous Waste Site were determined.
                                              561

-------
INTRODUCTION

Groundwater  investigations typically involve  the  use of monitor wells to measure the  movement of
groundwater and/or  contaminants (Fetter, 1988)  (Driscoll, 1986).   In karst aquifers  the  majority of
groundwater flow is turbulent (White, 1988) compared to flow through porous-media aquifers which is
generally laminar. Relatively rapid transmittance of groundwater through discrete conduits in karst aquifers
precludes the use of monitor wells as a means of accurately measuring the movement of groundwater and/or
contaminants. In karst settings water and/or contaminants are typically on the ground surface for only a
short tune before sinking into the subsurface through features such as sinkholes, sulking streams, and soil
macropores. These features provide direct access for liquids and solids to migrate from the surface into the
subsurface with little or no filtration by soils.  Once in the subsurface, the turbulently flowing groundwater
can rapidly transport liquids and solids considerable distances. Most water wells and springs in karst areas
become muddy after hard rains because solid material transported by surface water flows from the ground
surface into the  aquifer through conduits large enough to transmit the particles. The subsurface conduit
system will then  transport solids (sediments) much like surface flow systems, the more turbulent the water,
the greater its capacity to transport solid material.  Due to the unique characteristics of groundwater flow
in karst areas, investigations should be designed to  encompass a larger geographical area of study than the
site itself and incorporate water-table measurement data to identify general flow directions and dye tracing
to determine general flow routes.

This paper  presents a case study of a two-phase  investigation designed to determine characteristics of
groundwater flow in  the vicinity of a hazardous waste site  in a karst area.   The initial  phase of this
investigation involved the identification of all applicable hydrogeologic features in the study area and the
preparation of a water-table contour map based on  water-level measurements from private water wells and
elevations of surface and subsurface streams. From the information acquired through these activities, the
second phase of the investigation, dye tracing, was designed and performed.  The results of both phases of
the investigation determined the direction, destination, velocity and general route of groundwater flow from
the hazardous waste site.
SITE BACKGROUND

Several hundred electric transformers were burned over a period of years at a rural "junkyard" located 10
miles south of Elizabethtown in Hardin County, Kentucky (Figure 1). The "junkyard" is referred to as the
Middleton Hazardous Waste Site (MHWS).  The transformers  were burned  in  order  to facilitate the
extraction of copper wiring housed inside the transformers. As a result of the activities at the site, oil laden
with polychlorinated biphenyls (PCBs) and associated metals were released into the soil. PCBs have a strong
affinity for organic materials in soils and therefore constitute a threat to groundwater in karst areas due to
the direct surface-subsurface link provided by karst features.  The uppermost aquifer in the vicinity of the
MHWS is a karst aquifer. At the time of this investigation the aquifer served  as  a water supply for the
majority of residences in the area.  Due to the potential impact of the "junkyard"  activities on the quality of
the area's groundwater, a karst groundwater  investigation was performed (Crawford  et al., 1989) as a
component to the overall  site remediation activities of the Region IV U.S. EPA.

Physical Setting

Hardin County, Kentucky  is located in the Mississippian Plateaus Physiographic Province. The area of study
encompasses approximately 70 square miles of mature karst topography which is bounded on the east, north
and west by the Noun River as shown  in Figure 1.   This river serves as the base-level stream for
groundwater flow in the area.  Another surface  stream, Dorsey Run, flows toward the west and sinks 2 miles
south of the MHWS.

The geology of the  study area is comprised of two  slightly westward-dipping  (<1 degree) limestone
formations of Mississippian Age. The oldest unit is the St. Louis Limestone, which is bedrock in the eastern
                                              562

-------
        HARDIN COUNTY
                                           1-69
STUDY AREA
                                            MIDDLETON HAZARDOUS
                                            WASTE SITE
FIGURE 1 - Location map for the Middleton Hazardous Waste Site
          Groundwater Investigation
                           563

-------
two-tMrds of the study area.   Bedrock in the western one-third of the study area is  the  younger Ste.
Genevieve Limestone (Moore, 1964,1965).  The study area exhibits features common to areas underlain by
relatively homogeneous carbonate rock.
METHODOLOGY

The groundwater investigation consisted of two phases.  The initial phase involved determining the water-
table elevation in as many water wells as possible in the area and collecting information concerning locations
of applicable hydrogeologic features from landowners, area residents, and members of the Fort Knox Grotto
of the National Speleological Society. The second phase involved dye tracing from a sinkhole adjacent to
the MHWS and the sinking stream located south of the MHWS.

Phase One  -  Identification of Hydrogeologic Features and Determination of the Water-Table  Gradient

A door-to-door survey was conducted to  identify residences with accessible water wells and to  acquire
permission to obtain water-level measurements. Measurements were made with water-level indicators.  Only
data representing  the water-table aquifer were collected.  Information from well owners indicated the
majority of the wells were cased only to bedrock.  The wells were uncased from bedrock to the total depth
of the well.  Due to the size of the study area and the distribution of wells, it was impractical to determine
the precise  elevation of the wells.  Surface elevations were estimated from the Sonora and Summit 7.5
minute  topographic quadrangle maps (USGS, 1967 and 1972).  The contour interval of the topographic
quadrangles in the vicinity of the site was 20 feet.  Estimates of surface elevations from these large contour
interval maps decreased the accuracy of the water-table elevations in the area but did not preclude the use
of the technique in providing a generalized map of the water table.

During  the acquisition of water-level measurements from wells, inquiries were made of landowners and area
residents concerning locations of various hydrogeologic features such as springs, surface streams, sulking
streams, sinkholes with water flow, and caves, in particular those caves  with streams.  These features, as well
as those identified by local cavers, were investigated and located on topographic maps. In anticipation of
the second phase of the investigation, passive dye receptors were placed at the aforementioned hydrogeologic
features.  These receptors  (strips of untreated surgical cotton and packets of activated coconut charcoal)
once analyzed, served as an indicator of the presence or absence of fluorescent dyes in the groundwater flow
system.  Placement of receptors at the identified hydrogeologic features throughout the study area prior to
dye tracing provided a measure of background fluorescence to which the post dye-trace receptors would be
compared.  Before initiating the dye traces, all background dye receptors were removed and replaced with
fresh receptors. Locations of measured water wells and passive dye receptors are identified in Figure 2.

The information acquired from water-level measurements and the estimated elevations of surface and cave
streams were  used to prepare a  water-table contour map  (Figure 3). Interpretation of the water-table
contour map influenced the second phase of the investigation.  Due to its proximity to the MHWS, a spring
located 1.3 miles east of the site (Location 1) was initially hypothesized as the primary resurgence of water
from the site.  However, the slope of the water-table gradient  and the form of the water-table contours
suggested the most likely resurgence was at Waddell Spring, 4.5 miles to the west (Location 8). Anticipating
the resurgence of dye at this spring, an ISCO automatic water sampler was installed and programmed to
collect a sample every four hours.  Collection and analysis of these samples provided information necessary
to make a quantitative interpretation of groundwater flow from the MHWS.

Phase Two -  Dye Tracing

Six pounds of fluorescein dye (C.I. Acid Yellow 73) was injected into a collapsed sinkhole approximately 200
feet east of the site (Figure 4). Fifteen-hundred gallons of water was injected to insure the drainage capacity
of the collapse while at the same time wetting the soil to reduce dye sorption. Nine thousand-five hundred
gallons  of water was then used to flush the dye into the system.
                                              564

-------
                                                                             MIDDLETON HAZARDOUS
                                                                             WASTE SITE
                                                         432
MIDDLETON HAZARDOUS WASTE SITE
   GROUNDWATER INVESTIGATION
                                     1    0.5    0
                                                       1 MILE
                                                                        LEGEND
•  MEASURED WATER WELL
A  PASSIVE DYE-RECEPTOR LOCATION
                                            Figure 2

-------
                                                 NOUN RIVER
    i\
    II
                                                                                 MIDDLETON HAZARDOUS
                                                                                 WASTE SITE
                                                                                                  sao
MIDDLETON  HAZARDOUS WASTE SITE
   GROUNDWATER INVESTIGATION
    WATER-TABLE CONTOUR MAP
                                      1     0.5
                                                          1 MILE
                                                                            LEGEND
 .  MEASURED WATER WELL
 A  PASSIVE DYE-RECEPTOR LOCATION

«•«- GENERAL GROUNDWATER FLOW DIRECTION
                                                                    WATER-TABLE CONTOURS ARE IN
                                                                    FEET ABOVE SEA LEVEL
                                              Figure 3

-------
                                                NOLIN RIVER
                                                                                MIDDLETON HAZARDOUS
                                                                                WASTE SITE
MIDDLETON HAZARDOUS WASTE SITE
   GROUNDWATER INVESTIGATION
    WATER-TABLE CONTOUR MAP
                AND
            DYE TRACES
1    0.5
                                                                           LEGEND
MEASURED WATER WELL
PASSIVE DYE-RECEPTOR LOCATION

DYE INJECTION LOCATION
DYE TRACE AND DIRECTION OF FLOW
WATER-TABLE CONTOURS ARE IN
FEET ABOVE SEA LEVEL
                                             Figure 4

-------
To further characterize the groundwater flow of the area, 15 pounds of optical brightener (Tinopal 5BM GX,
F.B A. 22) were injected into Dorsey Run Swallet (Figure 4).  This sinking stream provided an ideal location
to inject dye, and receptors capable of capturing the optical brightener  had been placed along with those
being used to intercept fluorescein dye.

Every four days water samples being taken by the automatic sampler were removed and fluorometrically
analyzed hi the laboratory for presence of fluorescein dye. After allowing sufficient time for the dyes to flow
through the system, all passive dye receptors were replaced with fresh receptors. The dye receptors removed
from the locations were analyzed for the  presence of both fluorescein and optical brightener.  Analysis for
the presence of fluorescein involved rinsing  the coconut charcoal with high-pressure water to remove
sediment and elutriating the charcoal with a solution of 9.5:1 isopropyl  alcohol and potassium hydroxide.
Both visual and fluorometric analysis of the elutriant and comparison to  the background receptor elutriant
indicated which hydrogeologic features the dye flowed through.   Analysis for optical brightener involved
observing the spray-cleaned strips of untreated cotton under a long-wave  ultraviolet light. Receptors which
fluoresced bright blue-white, as compared to background receptors from the same locations, were identified
as locations which were hi the flow path  of the optical brightener introduced into the system.
RESULTS

During the inventory of the hydrogeologic features a number of springs, cave streams, and karst windows
were located (Figure 2, Locations 1-9).  Dye receptors were placed at each of these features.  Because the
Nolin River is the base-level stream hi the area, it was unnecessary to include hydrogeologic features north
of the river in the investigation. Passive dye receptors were positioned at several features south of the area
depicted hi Figure 2. However, these locations did not prove to be a part of the groundwater flow system
in the vicinity of the MHWS and, therefore, are not indicated hi this figure.

Groundwater elevations obtained by measuring water wells were combined with estimated elevations of the
surface and cave streams to produce a water-table contour map (Figure 3). Data from forty-three wells were
used hi preparation of the water-table contour map.  From this  map it is apparent that groundwater in the
vicinity of the MHWS flows west,  contrary to the original hypothesis  of flow to the east. Based on the
relative size of Waddell Spring and the direction of groundwater flow, it was selected as the most probable
location for the resurgence of groundwater from the site.  The water-table contour map indicates that water
sulking at Dorsey Run Swallet is part of the groundwater flow found at Locations 2, 3, 4, and 5. This water
also flows hi a westerly direction toward Waddell Spring.  The flow of water from the spring at Location 1
originates south and southeast of the spring rather than from the west  as originally hypothesized.

Qualitative Dye Trace Results

Dye receptors were collected and analyzed for the presence of fluorescein and optical brightener.  The
results  of the dye traces  and the water-table contours are shown in  Figure 4.  Activated charcoal dye
receptors positioned at Locations 6, 7, and  8 were positive for the presence  of fluorescein.  All other
receptors were  negative for the presence of fluorescein.  Surgical cotton  dye receptors positioned at
Locations 2, 3, 4, 5, 6, 7, and 8 tested positive for the presence of optical brightener, and all other receptors
tested negative.

Quantitative Dye Trace Results

Water samples collected at Waddell Spring by the automatic sampler  were fluorometrically  analyzed for
fluorescein.  Results of this analysis are illustrated in Figure 5.  The first arrival of dye at Waddell Spring
occurred approximately 86 hours after  injection.   The concentration  of dye at  the  spring  peaked
approximately 12 hours after first appearance and was no longer  detectable 80 hours later.
                                              568

-------
FLOW-THROUGH CURVE FOR FLUORESCEIN DYE AT WADDELL SPRING
                          Dye Injected on September 7, 1989 at the MHWS
    c
    g
    '•*-•
    O 0_
    §^
    O
    0)
    >.
    Q
         0.08
          0.06-
0.04
                                      Peak Dye Concentration
         First Arrival of Dye

        Dye Injected
                                            Dye No Longer
                                            Detectable
         0.02-
              0
                      T     i    r
                     80  100  120  140 160  180  200
20  40   60
        Time (Hours From Injection)

Figure 5 - Dye Flow-Through Curve
                                                                 Sample Data Points
                                             Distance from Waddell
                                             Spring to the MHWS is 4.5
                                             miles. First arrival of dye
                                             was approximately 86 hours
                                             after injection. Velocity of
                                             flow was 0.052 mph.

-------
DISCUSSION OF RESULTS

Groundwater from the MHWS flows toward the west through karst windows at Locations 6 and 7 before
resurging at Waddell Spring (Figure 4).  Dorsey Run, which sinks 2 miles south of the site, flows through
Locations 2, 3, 4, and 5 before joining the flow from the MHWS.  The convergence of these  two cave
streams occurs between Locations 5 and 6.   The rate of flow from the MHWS to Waddell Spring is
approximately 0.052 miles per hour. This flow rate is based on linear distances between positive fluorescein
detection points. The actual flow rate is higher given the tendency for cave streams to meander in a manner
similar to surface streams.
CONCLUSIONS

A two-phase approach to the Middleton Hazardous Waste Site groundwater investigation determined the
direction, destination, velocity, and general route of groundwater flow from, and in the vicinity of the site.
The use of water-table measurement data produced a generalized water-table map of the study area and was
beneficial in directing  the second  phase of the investigation.  Dye traces identified the destination and
general flow routes of groundwater from the site and in the vicinity of the site.

Error introduced to the water-table measurements by estimating surface elevations from 20-foot contour
interval topographic maps would not typically be acceptable in hydrogeologic investigations.  However, in
this investigation the generalized water-table map proved useful in identifying the most probable resurgence
of groundwater from the site.  Based on this information, a water sampler was  placed at that resurgence.
Information gained from the fluorometric analysis of the samples was used to prepare a dye flow-through
curve.  The  added certainty of the dye flow-through curve and the associated information it provides were
necessary given the nature of the investigation.  A dye flow-through  curve more positively identifies the
connection between injection point and sampling point than visual or fluorometric analysis of elutriant from
passive dye  receptors. From the dye flow-through curve it was possible to estimate the flow rate from the
site to the spring. The water-table contour map and dye-trace information obtained from this study will also
serve to focus any groundwater sampling activities associated with the site investigation.

If the water-table map had not been  prepared prior  to dye tracing, the water  sampler  would have been
incorrectly located at the spring east of the site. Analysis of samples collected at that spring would not have
identified the resurgence of groundwater from the site. Since passive dye receptors had been placed at all
hydrogeologic features  in the  area, Waddell Spring  would have been identified as the resurgence of
groundwater from the site. However, it would have been necessary to perform another dye trace with the
water  sampler housed  at Waddell Spring to provide information  for a quantitative  interpretation of
groundwater flow from the site. An additional dye trace would have required more time and resources, and
an accurate interpretation of the second dye trace may have been hindered by dye from the first trace
remaining in the system.

This case study reinforces the need for hydrogeological investigations  at hazardous waste sites in karst
terrains to include regional groundwater data to determine general flow directions and dye tracing to identify
the destination and general route of groundwater flow. The more traditional approach at hazardous waste
sites using site monitor well data may result in a completely false characterization of groundwater flow.
 REFERENCES

 Crawford, Nicholas C. and Dotson, C. Britton. 1989. Groundwater Flow in the Vicinity of the Middleton
 Toxic  Waste  Site, Hardin County,  Kentucky.  Prepared  for  O.H.  Materials  Corporation  and  U.S.
 Environmental Protection Agency, Region IV. 15 pp.
                                             570

-------
Driscoll, Fletcher G. 1986. Groundwater and Wells. Second Edition. Johnson Filtration Systems Inc., St.
Paul, Minnesota. 1089 pp.

Fetter, C.W. 1988. Applied Hydrogeology. Second Edition. Merrill Publishing Company, Columbus, Ohio.
592pp.

Moore, F.B. 1964. Geology of the Summit Quadrangle, Kentucky. Kentucky Geological Survey and USGS
Geologic Quadrangle Map.

Moore,  F.B.  1965. Geologic Map of the Sonora  Quadrangle,  Hardin and Larue Counties, Kentucky.
Kentucky Geological Survey and USGS Geologic Quadrangle Map.

USGS. 1967. Sonora Quadrangle, Kentucky, 7.5 minute series (topographic).

USGS. 1972. Summit Quadrangle, Kentucky, 7.5 minute series (topographic).

White, William B. 1988. Geomorphology and Hydrology of Karst Terrains. Oxford University Press, New
York. 464 pp.


C. Britton Dotson

Britton Dotson has a B.S. in Geology and a M.S. in Geography from Western Kentucky University. At the
time of this research he was employed by the Center for Cave and Karst Studies at Western as a research
hydrogeologist.  He is now employed as  a geologist by Roy F.  Weston, Inc., under  the Response,
Engineering, and Analytical Contract (REAC)  for the Environmental Protection Agency's Environmental
Response Team, in Florence, Kentucky.
Dr. Nicholas Crawford

Nicholas C. Crawford, Ph.D. is a Professor in the Department of Geography and Geology and Director of
the Center for Cave and Karst Studies at Western Kentucky University.  He has written over 140 articles
and technical reports dealing primarily with groundwater contamination of carbonate aquifers. The recipient
of 25 grants for hydrologic research on environmental problems of karst regions, he was awarded Western's
highest award for Outstanding Achievement in Research in 1985. As a consultant specializing in carbonate
aquifers for the past fifteen years, Dr. Crawford has performed over 600 dye traces and worked on numerous
groundwater contamination problems for private firms and for federal, state and local government agencies.
Mark J. Rigatti

Mark J. Rigatti is the program manager for the U.S. EPA Region IV Emergency Response Cleanup Services
(ERGS) contract for O.H. Remediation Services Corporation.  Mr. Rigatti holds a B.S. in aquatic biology
and has been associated with several studies dealing with groundwater contamination in carbonate aquifers
in Kentucky and Tennessee  as a component to hazardous waste site remediation activities.
                                            571

-------
               The Use of Groundwater-level Measurements and Dye Tracing to Determine
                      the Route of Groundwater Flow from a Hazardous Waste Site
                             in an Area of Karst in Hardin County, Kentucky

                         C. Britton Dotson, Nicholas C. Crawford, Mark J. Rigatti

Question 1  Was there any indication of site contaminants in the local wells or springs?

Question 2 - What assurances do you have that dye-tracing would intersect the same conduit system with the
sinks at the waste site?

Answer 1   The primary objective of Crawford and Associates was the identification of flow direction and
destination of groundwater in the vicinity of the site. Crawford and Associates was not involved in any sampling
activities.  To date there has been no extensive sampling of water wells and springs in the  study area.  A
domestic water well located on the site property was sampled during the initial EPA response.  There were no
site contaminants detected in this well.

Answer 2 - The uppermost aquifer in the vicinity of the hazardous waste site is a karst aquifer.  Sinkholes on
the site and in the vicinity drain into this aquifer. Since individual groundwater drainage basins exist within the
aquifer, as evidenced by multiple groundwater resurgences, all sinkholes in the study  area do not contribute to
the same  drainage basin.  It is possible that water flowing into one sinkhole may  contribute to a different
drainage basin than water flowing into another  sinkhole only 200 feet away.  Based on the hydrogeologic
inventory only two springs (Locations 1 and 8) were likely candidates for resurgence of dye injected at the site
or in the vicinity.  Dye that was injected into the sinkhole 200 feet east of the site resurged 4.5 miles west of the
site at Location 8 (Waddell Spring). This fact, accompanied by information provided by the water-table contour
map, indicated that the sinkhole where dye was injected contributed to the same drainage basin as the sinkholes
on site. If dye had been injected into  a sinkhole on the site it is unlikely that it would have flowed east to
Location 1.  To do this, flow would have been against the gradient, as indicated by the  water-table contour map,
and would be opposite the general flow direction identified by the dye trace.
                                                 572

-------
RECOMMENDED ADMINISTRATIVE/REGULATORY DEFINITION OF KARST AQUIFER,

       PRINCIPLES FOR CLASSIFICATION OF CARBONATE AQUIFERS,

   PRACTICAL EVALUATION OF VULNERABILITY OF KARST AQUIFERS, AND

      DETERMINATION OF OPTIMUM SAMPLING FREQUENCY AT SPRINGS


                         JAMES F. QUINLAN
                    Quinlan & Associates, Inc.
                       Nashville,  Tennessee

                          PETER L. SMART
                       University of  Bristol
                         Bristol, England

                         GEARY M.  SCHINDEL
                         Eckenfelder Inc.
                       Nashville,  Tennessee

                     E. CALVIN ALEXANDER, JR.
                      University of Minnesota
                      Minneapolis, Minnesota

                         ALAN J. EDWARDS
                       University of  Bristol
                         Bristol, England

                         A. RICHARD SMITH
                  Municipal Solid Waste Division
                      Texas Water Commission
                          Austin, Texas
                              SUMMARY

      A major problem of  karst hydrology is an  inadequate under-
 standing by  many  regulators, planners,  attorneys,  geologists,
 hydrologists,  engineers,  and other non-karst specialists of what a
 karst aquifer  is and how  it differs significantly from other aqui-
 fers.   As  a first  step  in solving this problem,  we offer the fol-
 lowing definition.  A karst aquifer is an aquifer in which flow of
 water is or can be appreciable through one or more of the follow-
                                573

-------
ing:  joints,  faults, bedding planes, and cavities — any or all of
which have been enlarged by dissolution of bedrock.

     The above definition is in terms of hydraulics,  not landforms
(such as  sinkholes),  grikes  (soil-filled  joints),  or  hydrologic
features  (such as swallets, sinking streams, springs, and  caves).
The presence of sinkholes and any or all of these other features is
diagnostic of a subjacent  karst aquifer,  but the absence  (or un-
recognized presence) of  any or all of them  does not mean that a
karst aquifer  is  not present.  As  a generalization, however, if
carbonate rocks such as  limestone, marble,  or dolomite, or even
more soluble rocks such as  salt or gypsum,  are present, assume that
the terrane is a karst  — until or  unless the opposite can  be con-
vincingly proven.   Almost  all carbonate terranes are underlain by
one or more karst  aquifers, but some terranes and aquifers are more
karstic than others.

     Having defined karst  aquifers, we classify them according to
three fundamental, independent attributes governing their behavior
— recharge, storage, and  flow — each of which can be  visualized
as forming orthogonal  edges of  a cube.   Each of the three attri-
butes is  part of  a  continuum  between two end-members,  yielding a
classification based on a total of 6 karst end-members, many inter-
mediate members, and 2 non-karst end-members.  The cube, modified
after one proposed by Smart and Hobbs (1986), can be visualized as
a  3-dimensional conceptual model within the boundaries of which
various aquifers can be plotted  (mapped) and their relations to one
another can be seen, as shown in Figure 1.

     All of the many types  of  karst aquifers  can be grouped into a
fourth  continuum  based on the  sensitivity of  their response to
variations  in  recharge,  storage,  and  flow,  and their  attendant
relative vulnerability to the  adverse effects of spills  and impro-
per  disposal  of  waste:  hypersensitive aquifers,  very  sensitive
aquifers, moderately  sensitive aquifers,  and  slightly  sensitive
non-karst aquifers.  These  new terms are plotted on the  same cube.

     Hypersensitive aquifers are characterized  by concentrated re-
charge  at discrete  points, low to  moderate  storage, and  conduit
flow.  Consequently, they  are extremely vulnerable to groundwater
contamination.  Very sensitive  aquifers  are  characterized by the
occurrence of either conduit flow,  point recharge, or low storage,
or a  combination  of two  of these,   as  shown  in Figure  1.  Again,
they  are  sensitive  to  pollution, but less so than hypersensitive
aquifers.   Moderately  sensitive  aquifers have  dispersed,  slow
recharge,  high  storage,  and  diffuse  flow,  as  also   shown  in
Figure  1;  such  aquifers  are  therefore less vulnerable to rapid
transmission of pollutants  than are very sensitive karst aquifers.

     The  continuum from hypersensitive  to  moderately  sensitive
karst aquifers can be discussed in terms of  two  end-members which
have  traditionally,  but  misleadingly, been  called  "conduit-flow
aquifers" and "diffuse-flow aquifers".  Recognition of the position
of an individual aquifer in this continuum  is essential  to  provid-


                               574

-------
ing adequate aquifer protection and appropriate management of waste
disposal  and also to the design of reliable groundwater monitoring
schemes.   In  general,  movement  of pollutants  in  hypersensitive
karst aquifers is very rapid, commonly tens  to  thousands  of feet
per hour, depending upon stage.  Reaction  time,  the time  between
the start of a recharge event  (or when a spill occurs) and when it
is first  sensed by the aquifer, is short.  Duration of the adverse
effects of a pollution incident may be relatively brief (depending
upon the  properties of the  contaminant  and how and  where  in the
system the contamination occurs).  In contrast, movement of pollu-
tants in moderately sensitive aquifers is usually several orders of
magnitude less rapid than  in hypersensitive aquifers, but  it is
usually orders of magnitude  faster than in most granular aquifers.
Reaction  time  is much longer than in hypersensitive aquifers.  The
results of a pollution  incident can persist for tens of years.

     The  terms conduit flow and diffuse flow,  as  descriptors of
aquifer type used for classification,  should be abandoned  because
they prevent  necessary  consideration of the important roles of
recharge  and  storage.   Conduit flow  and  diffuse flow should be
retained, however,  as descriptors of flow  type.

     The  distinctions between  the two vulnerability end-members in
karst aquifers can be made  in several ways;  they are most easily
determined by interpreting  variations  in specific  conductance
(hereafter called conductivity), as characterized  by its  coeffi-
cient of  variation (CV).   The variations in conductivity are more
important than the absolute values.   The distinctions between aqui-
fer vulnerability can also be determined from fluctuations in one
or more  of the  following alternative water quantity  or  quality
parameters:  discharge,  turbidity, temperature, and carbonate hard-
ness.  These distinctions  can only be determined by site-specific
fieldwork. The necessary data for these conductivity and alterna-
tive vulnerability-type discrimination measurements are easily and
inexpensively  obtained,  but they are not included  in geologic or
hydrologic publications  or reports.

     The  spring is the pulse  of a karst aquifer.   Monitor  it in
order to  understand the  nature of an aquifer.  While springs pro-
vide the  most  accessible and perhaps the most representative moni-
toring points,  alternative monitoring  points  may  include  cave
streams,  seeps,  and wells  that may intercept fracture zones, dis-
solution  zones,  or  conduits.   If monitoring  is  done merely to
attempt to characterize an aquifer — as contrasted with attempting
to detect pollutants from a specific site  — the wells and seeps
used as alternative monitoring points are  likely to undergo much
smaller fluctuations in chemical and physical characteristics than
are springs.

     Interpretation of variations in conductivity of a spring as an
indicator of the vulnerability of a karst aquifer is probably reli-
able for  the vast majority of North American karst aquifers and, we
believe,  essentially all those with a drainage-basin area smaller
than about 500 square miles.   Although the controls on conductivity
                                575

-------
are many and complex,  it  is  a reliable measure and integrator of
hydrologic and water-quality response to the dynamics of an aqui-
fer.   Conductivity senses and  integrates the  covariance  of the
complex response and allows a simple measure of it.

     Six important caveats must be remembered:

     1. An aquifer can not be classified as  being "not a karst"
        simply because  one can  successfully model it numerically
        over so  large an area  that  it seems to  behave with the
        properties of  a porous medium.   The  local perturbations
        that  so dramatically  affect  site characterization  and
        remediation are beyond the ability of  analysis by present-
        day  computer  models.   Such model  codes  are  most often
        applicable only to granular aquifers where aquifer parame-
        ters do  not  vary as dramatically  and  over  as  short  a
        distance as they can in karst aquifers.

     2. Data from wells alone might not show karstification, even
        if it exists,  because the scale of organization  of flow in
        a karst aquifer is not  likely  to  be adequately sensed by
        just a few wells.   If wells alone  are used for monitoring,
        they must  be  shown  by  rigorously designed and executed
        tracer tests to have hydrologic connection to the site to
        be monitored.

     3. A karst aquifer within  a single groundwater basin may be
        locally characterized by either of the two end-member types
        of vulnerability, or by an intermediate member.  In part,
        this is  because contaminants introduced  into  a sinkhole
        connected  to  a conduit  will move  more  rapidly  than if
        spilled on the ground or  introduced  in the area between
        swallets or open sinkholes, or  introduced in an  area where
        dispersed recharge and  diffuse flow  may predominate and
        conduits may be rare or relatively inaccessible.

     4. Just because  a spring flows  from a conduit,  it does not
        follow that it is a hypersensitive spring  or that it is a
        discharge point of an aquifer  dominated by conduit flow.
        It could be a  moderately sensitive spring  draining a simi-
        lar aquifer,  as determined by interpretation of  variations
        in its water quality and discharge.  For  nomenclature deci-
        sions, aquifer response to recharge events takes priority
        over physical appearance.

     5. Site-specific fieldwork may require  examination  and evalu-
        ation of geology, hydrology, and  karst features within a
        radius of a mile from a site to perhaps as much  as 40 miles
        away, depending upon the structural setting and probable
        size of groundwater basins.

     6. The conductivity CV is not an infallible  indicator of the
        nature and vulnerability of  a  spring  and the  aquifer it
        drains.   Nevertheless,  the  CV  is  extremely  useful and
                               576

-------
        probably reliable so often that it can be employed with a
        high degree of confidence.  Its reliability can be tested
        by tracing.

     An alternative to using conductivity at discharge points such
as springs to  characterize the nature  of  a karst  aquifer is to
obtain hydraulic conductivity or transmissivity data from aquifer
tests (slug,  bailer, or pump tests).   Such data are characterized
by a low mean and a high CV in aquifers dominated by conduit flow
and by a  moderate  to high mean but  a much lower  CV in aquifers
dominated by diffuse flow.  Non-karstic carbonate aquifers gener-
ally have a low hydraulic  conductivity  mean and low CV (Smart et
al., 1991).  Interpretation  of data from aquifer  tests in karst
terranes, however,  should be  done cautiously,  for the  reasons
discussed.

     Aquifer tests may yield valuable information (albeit inaccu-
rate) about the flow and quantity  of water in the portion of karst
aquifer tested  but  say nothing about its  recharge or storage beyond
the immediate vicinity of a well.   Such tests  are complemented by
interpretation of conductivity CV's which say nothing about aquifer
parameters but do imply much about recharge,  storage, and flow.

     An appendix outlines  two procedures for  using conductivity
data to prevent aliasing  (loss of recognition  of  the signal from
rapidly changing parameters,  a consequence of  insufficient frequ-
ency of  sampling)   and  to determine  an appropriate,  practical,
aliasing-free sampling procedure  for using water  quality data to
characterize an aquifer,  its spring, or a conduit-fed well, and for
the establishment of sampling frequency during tracer tests.


                            CONTENTS

Summary   	573
Introduction  	 578
Importance of aquifer vulnerability to design  of
     groundwater monitoring plans for karst aquifers  . . . 583
Recognition of karst terranes   	 584
Definition of karst aquifer   	 586
Recognition of karst aquifers 	 587
Principles for classification of carbonate aquifers .... 588
Relation between types of karst aquifer and
     sensitivity to groundwater contamination    	593
Characterizing carbonate aquifer sensitivity	594
     Introduction 	 595
     Background	596
     Geochemical variation  	 597
     Hydrological behavior	598
     Limitations	599
Practical evaluation of vulnerability of karst aquifers   .601
     Procedure	601
     Difficulties and limitations of vulnerability
          evaluation	602

                               577

-------
Use of aquifer test data to characterize flow in and
     vulnerability of karst aquifers   	 604
A cautionary tale	607
Caveats   	608
Discussion	608
Solicitation  	 615
Acknowledgements  	 615
References cited  	 615
Appendix:  Procedure for prevention of aliasing and for
     determination of appropriate sampling frequency  for
     a spring or a conduit-fed well    	629
Biographical sketches   	 632
Questions and answers   	 634
                           INTRODUCTION

     This paper is written primarily for geologists, hydrologists,
and engineers who are not  specialists in karst but who work in car-
bonate terranes and deal with regulators, planners, and attorneys
who must make important decisions concerning such terranes.  It is
also written for regulators and planners who are  not  likely to be
trained in karst hydrology but who must educate some of the consul-
tants they  deal with.  The  discourse is written as a practical
guide for "greater utility" (Franklin, 1781)  .  Accordingly, we have
tried to avoid the Feynman Effect (directing the content toward our
colleagues, karst specialists,  when we are supposed to be  focusing
on communication with non-karst specialists; Bartlett, 1992), but
we believe  that karst specialists will find  it to be a  valuable
synthesis which revolutionizes concepts of karst aquifer classifi-
cation and  of what constitutes necessary sampling frequency.   A
less technical  summary of  this paper will be published elsewhere.

     The Summary, the sections titled Definition of karst  aquifer,
Principles for classification of carbonate aquifers,  and Practical
evaluation of vulnerability of  karst aquifers — as well as the Ap-
pendix and the penultimate indented paragraph of the Discussion —
are the most important parts of this paper; the  rest  comprise the
foundation for  it  and the  necessary rationale.   The Discussion is
a potpourri of important points that,  although  extremely relevant,
might have detracted from  the clarity and continuity  of the paper
itself.

     About 20% of the conterminous U.S. and 40% of the U.S. east of
Tulsa, Oklahoma, is underlain by carbonate rock,  mostly limestone.
We maintain  that almost all of these  areas are, therefore, some
type of karst and probably include some type of karst aquifer.  It
is far more  judicious  to  operate on the  assumption that  they are
karst aquifers  —  until and unless  conclusively  proven otherwise.
An understanding  of  the extraordinary hydrology of such  areas is
necessary if their waters  are to be protected and used most wisely.

     Groundwater movement in  karst aquifers differs  from that in
granular media.  It  is commonly orders of magnitude faster, seem-

                               578

-------
ingly (but not  actually)  less predictable,  and commonly through
conduits or  dissolutionally enlarged joints  and bedding planes.
Water movement in karst aquifers can be described by known physical
laws but generally not the same ones that successfully describe and
predict flow in granular media.

     Karst terranes can be usefully classified according to eight
major attributes:  cover, rock-type, climate, geologic structure,
physiography, hydrology, modification during or after karstifica-
tion, and dominant landforms, each of which forms the basis for a
separate classification that can be combined with the other seven
(Quinlan,  1978,  as partially summarized by White,  1988, p. 117-118;
1990b).   Simple as  the above classifications are,  and useful as
each of them and a combination  of them may be for geomorphological
and regional analysis,  they are obviously much more than is needed
for day-to-day use by professionals concerned with environmental,
regulatory, and  administrative problems of terranes potentially af-
fected by hazardous materials.

     Karst is not rare.   Its hydrology  is not unpredictable.  When
the appropriate methods of investigation are used, the properties
and flow within  a karst  are more predictable than most profession-
als realize  (Quinlan and Ray, 1989).

     Although various sections of  the  long-awaited,  newly issued
Subtitle D regulations  for RCRA (Resources Conservation and Recov-
ery Act) and the CWA  (Clean Water Act)  can be easily interpreted to
sanction special monitoring protocols that may  be  necessary for
karst aquifers  —  as they should —  there  is  only  one section
(Section 258.15) that specifically  mentions karst, in referring to
unstable areas.   Karst  is mentioned in the appendices that discuss
this section and others.   Unfortunately,  in Appendix C (Supplemen-
tal Information), the discussion repeatedly and mistakenly refers
to  karst  terraces where,  from context,  it  clearly means  karst
terranes (U.S. EPA, 1991, p. 51,020, 51,047-48).  Sinkhole develop-
ment can be extremely important at  waste disposal  sites, but it is
our considered opinion that the potential for groundwater contami-
nation at such sites is far more common and nearly ubiquitous.

     One of the  more practical recent applications of academically-
inspired karst research has been the recognition that springs and
cave streams, shown by tracing to drain from a site, may be optimal
points for monitoring groundwater quality and that only wells shown
by tracing to drain  from a site can be considered  to be monitoring
wells (Quinlan and Ewers,  1985; Quinlan,  1989, 1990a).  These con-
cepts have been  well  received by the hydrogeologic community (Beck
et al.,  1987) and the regulatory community  (Quinlan, 1989) and are
implicit in this paper.

     There is a  need  for recognition of degrees of vulnerability of
the water  quality that can occur in different types of karst aqui-
fers.  Such recognition enables the user to think in terms of why
one is concerned with the karstic nature of a terrane (Field, 1988,
1990).  The "why" is  answered in terms of sensitivity to effects of
                                579

-------
groundwater contamination and in terms of how to reliably monitor
the various  types of karst  so  that the monitoring is relevant,
reliable, and time-sensitive.

     Although much of this paper is concerned with the potentially
severe consequences of spills of hazardous materials, no attempt is
made here to discuss procedures  for response to  them.  See Quinlan
(1986) and Crawford  (1991).

     How, then,  can karst aquifers be adequately  protected?  We
will discuss why degree of vulnerability of karst aquifers is rele-
vant to groundwater monitoring plans and suggest how to recognize
a karst terrane and/or a karst aquifer.  Then we  will state princi-
ples for classification of karst aquifers.  This will be followed
by a discussion of flow, recharge, and storage within karst aqui-
fers,  relations  between karst  aquifer  types and  sensitivity to
groundwater contamination, and characterization  of carbonate aqui-
fer sensitivity-  We then give a simple technique for reliably pre-
dicting aquifer vulnerability insofar as it is relevant to problems
of groundwater protection and monitoring, and close with a review
of the use of aquifer tests to characterize flow  in and vulnerabil-
ity of karst aquifers, some caveats, a concluding discussion, and
an appendix that describes a  new procedure for reliably establish-
ing the optimal frequency for sampling a spring.

     A cautionary quotation is appropriate:  "Of all the words in
the hydrologic vocabulary, there are probably none with more shades
of meaning  than  the term aquifer.  It means different things to
different people, and perhaps different things to the same person
at different times.   It is used to refer  to individual geologic
layers, to complete geologic formations, and even to groups of geo-
logic formations.  The term must always be viewed in terms of the
scale and context of its usage." (Freeze  and Cherry, 1979, p. 47).

     An aquifer  can  be  defined  as a saturated  permeable geologic
unit that can transmit significant quantities of water under ordi-
nary hydraulic gradients (Freeze and Cherry/  1979, p. 47).  Stated
another way, an  aquifer  is a  body of  rock or sediment capable of
yielding usable quantities of water to wells and springs.  If one
uses  or  could  use the water, its host is an aquifer.   With this
less precise definition,  the rock adjacent to a cave stream that is
perched above the water  table could be classified as and regulated
as  an aquifer,  even though  most  of  the rock  is  not saturated.
[Some state regulations  would consider this  cave stream  as surface
water because it is in a defined channel; there  is case  law to sup-
port  such interpretation  (Davis and Quinlan, 1991).]

     The term  aquifer is  purposely imprecise with respect to hy-
draulic  conductivity; it refers only to the ability  to  yield  a
usable quantity  of  water.  A rock or  sediment  mass yielding less
than one gallon per minute to a well and supplying one household is
just  as much an  aquifer as one  yielding thousands  of gallons per
minute to a  well for a  municipality or industry-   Obviously, the
latter is a better aquifer (if one needs  a high yield),  but the
                                580

-------
former is no less an aquifer.

     By water we mean both phreatic water  (water below the water
table — groundwater  in  the  traditional sense)  and vadose water
(water in the  unsaturated zone between  the water  table  and the
ground surface),  as discussed by Domenico and Schwartz (1990, p. 9-
11) .   Indeed, for the past  15 years,  the nature and movement of
water and contaminants in the vadose zone has been a major frontier
in research of both groundwater hydrology  in general (Devinny et
al..  1990,  p.  42-43)  and karst  hydrology  (Smart and Friedrich,
1986; Bonacci,  1987,  p.  28-35;  Chevalier,  1988).   In karst ter-
ranes,  it is especially important to monitor and characterize the
flow of vadose water because the vadose zone  in karsts provide sig-
nificant storage; flow from it to the  saturated zone can be quite
rapid.  Monitoring of groundwater in the unsaturated  zone, particu-
larly in karst terranes,  is  an as yet unexplored frontier,  both
practically and administratively.

     Several relevant  definitions are given in  recently issued
Federal regulations concerning landfills (EPA, 1991,  p. 51,017-18).
"Aquifer" means  a geological formation,  group  of formations,  or
port[i]on of a formation capable of yielding significant quantities
of ground water to wells  or springs."  [The problem words here are
use of ground water rather than water and the definition of signi-
ficant.  What is  the difference between significant and apprecia-
ble?   We would argue that significant,  as  defined by Webster's
Ninth New Collegiate Dictionary, means "of a noticeable or measur-
ably large  amount", and  that large is a relative  term.   In the
southwestern U.S.,  a  spring  may  flow a  few quarts  per hour,  an
insignificant quantity only by the standards of the humid east, but
enough to sustain an  entire  local ecosystem and  to be  used as a
water supply.]  "Ground water means water below the land surface in
a zone of saturation.  ...  Saturated zone means  that part of the
earth's crust in  which all voids are filled with water."

     The emphasis of monitoring sections of these new regulations
is on water below a water table  or in  a confined  aquifer.   EPA
proposed in  1988 to require monitoring elsewhere,  including unsatu-
rated zones.  The agency is currently  evaluating comments on that
proposal and is preparing a final rule.   The current regulations,
however,  do not preclude  states from  requiring  monitoring in the
unsaturated zone  (U.S. EPA,  1991,  p.  51,067).

     An encyclopedic compilation of aquifer and oroundwater defini-
tions could  be obtained from hydrology  texts and  case law, but such
a compilation is  beyond the scope of this  paper.   We  are not obses-
sed with definitions,  but we believe  it  is  necessary to agree on
them in order to  achieve clear,  unambiguous communication.

     Four excellent  texts in  English provide thorough expositions
on the hydrology of karst aquifers (Milanovic,  1981;  Bonacci, 1987;
White,  1988;  Ford and  Williams,  1989),  each  with  a  different
emphasis.
                               581

-------
     Even though there are numerous excellent texts on groundwater
hydrology (Freeze and Cherry, 1979; Driscoll,  1986;  Fetter,  1988;
Domenico and Schwartz,  1990), applied chemical and isotopic ground-
water hydrology  (Mazor, 1991), and on hydrology  in general (e.g.,
Gupta, 1989; Viessman et al..  1989),  none of these texts or others
describe the hydrology of  karst aquifers in more than a superficial
fashion, if at all.   Most  hydrology texts  completely ignore karst.
Otherwise excellent, state-of-the-art texts  on environmental sci-
ence and engineering and on watershed management don't even mention
karst (Henry and Heinke, 1989;  Brooks et al., 1991).   Recent defi-
nitive books on groundwater development, on groundwater monitoring,
on landfill design, and on hazardous-waste site  management or re-
mediation mention karst only in passing (Roscoe Moss Company, 1990,
p. 14, 167;  Nielsen, 1991, p.  64,  407; O'Brien and Gere Engineers,
Inc.,  1988,  p. 51,  53,  56, 78),  ignore  it  (Ward et al., 1990;
Bagchi, 1990; Pfeffer,  1992; Goldman et  al.,  1986;  Christiansen et
al.,  1989;  O'Leary  et al., 1986), or erroneously refer to karst
terranes as being "rare" (Devinny et al..  1990, p.  145).  A recent
environmental geology text says nothing about  the  special  charac-
teristics of groundwater pollution in karst  terranes  (Montgomery,
1992) .

     Although the second edition of an authoritative water encyclo-
pedia includes data on velocities characteristic of  conduit flow
and diffuse flow (van der  Leeden et al.,  1990, p.  286), little else
of its content suggests the importance of  karst aquifers in Ameri-
ca.  An otherwise excellent text on environmental geology includes
a brief discussion  of sinkhole development but says nothing about
groundwater movement in karst  aquifers,  especially as  related to
groundwater contamination (Lundgren,  1986).  A highly readable com-
pendium  on  groundwater contamination  in the  United States says
nothing  about  karst (Patrick et  al.,  1987)  and neither does an
otherwise  excellent  introduction  to groundwater contamination
written for executives,  plant managers, and politicians (Bailey and
Ward,  1990).   A fascinating synthesis  of geochemistry  and fluid
mechanics of permeable rocks ignores karst (Phillips, 1991) as does
a reference  handbook by the National Fire Protection Association
written as guidance for response to hazardous  materials incidents
(Henry, 1989).  Even the National  Research Council totally ignores
karst  (Committee  on Ground Water  Quality Protection,  1986; Water
Science  and  Technology Board,  1988, 1990a)  or gives it only one
paragraph in a book on groundwater models  (Water  Science and Tech-
nology  Board,  I990b,  p. 99-100) .   And  the National  Ground Water
Information Center, operated by the National Ground Water Associ-
ation, does not include karst as a Descriptor/Keyword on its index
sheet for Proceedings volumes.  The situation  with respect to re-
cognition of karst-related problems by  non-karst specialists in
Europe  is  little better,  but  there are  some notable  exceptions
(Erdelyi and Galfi, 1988;  Fookes and Vaughan,  1986;  Lallemand-
Barres and Roux, 1989; Nonveiller, 1989; Wynne,  1987).

     The books cited above are some of the more  recent on various
topics.  The same could be said about the  inadequacy of exposition
of karst-related topics in earlier books.  Accordingly,  most con-
                                582

-------
sultants,  whether hydrogeologists or engineers,  and most regula-
tors,  whatever  their education,  lack formal  training  in  karst
hydrology  and rarely have the opportunity to learn about it.  Many
are unaware of the significance of karst.  Too many others deny its
existence  or  importance.

     Only  six North American universities and the National Ground
Water Association regularly offer formal  courses devoted entirely
to various aspects  of karst hydrogeology.   Training  in  karst is
also available at several  additional universities where the topic
is a research interest of a faculty member.  All of these educa-
tional efforts are taught by well-experienced staff,  but they reach
an inadequate number of potential users of the concepts espoused.


             IMPORTANCE  OF AQUIFER VULNERABILITY TO
    DESIGN OF GROUNDWATER MONITORING PLANS FOR KARST AQUIFERS

     In an ideal world, the  concepts advocated in this paper would
be  regularly  applied to planning  and management.   In  the  real
world, however,  the  concepts can and should be so applied, but most
applications  will be to groundwater monitoring  and responding to
contamination situations that already exist.

     All aquifers are sensitive to groundwater contamination, hence
the need for aquifer protection policies and groundwater monitoring
to ensure  compliance with them.  Knowledge of whether a spring (and
the aquifer  it  drains)  is hypersensitive rather  than moderately
sensitive  to  contamination will affect the design of  its monitoring
plan and can  affect the  design and operation of tracer tests used
to establish  that plan.   Hypersensitive  springs typically respond
rapidly and possibly greatly to recharge events; they will require
frequent  sampling,  beginning early  in  an  event.   In  contrast,
moderately sensitive springs  typically respond  slowly and  in a
relatively subdued manner;  sampling frequency can be less and can
start later  in  an  event  but sampling duration will be longer.
These sampling matters are discussed in more detail by Quinlan and
Alexander   (1987)  and Quinlan  (1989,  1990a),  where we  used the
traditional  (but not equivalent)  terms  conduit  flow  and diffuse
flow,  respectively,  and  by Blavoux and Mudry (1988).

     In this  paper,  we offer a simple technique for aquifer charac-
terization on the basis of its relative vulnerability and the vari-
ability of its specific  conductance because there is a great need
for reliable, rapid, and inexpensive methods for site characteriza-
tion and  monitoring during  remediation  (Mackay,  1990)  and,  more
important, before remediation  is  ever necessary.  We suggest three
other techniques that yield complementary  results  and may or may
not give a higher reliability in assessment and characterization,
but they involve aquifer testing  and a significantly greater cost.

     We recognize that evaluation of aquifer vulnerability is just
one aspect of  land-use  planning  and watershed  protection  often
necessary  for karst terranes (Rubin,  1991).
                               583

-------
                  RECOGNITION OF KARST TERRANES

     When environmental professionals think of karst,  many of them
think of the Mammoth Cave area of Kentucky with its tens  of thou-
sands of sinkholes, hundreds  of  sinking streams and  springs, and
hundreds of miles of accessible cave passages, some so large that
two garbage trucks could be driven side-by-side through them.  Most
of these same people are unwilling to recognize, for  example, the
Nashville, Tennessee,  area as a karst.  To judge a carbonate ter-
rane not to  be  a karst because it does  not resemble the  Mammoth
Cave area is to  judge the National Enquirer not to be English prose
because  it lacks the  style,  grace,  and wit  of  a Shakespearean
comedy.  Indeed, the classic examples of  karst in the  Mammoth Cave
area and The Taming of  the Shrew, however delightful and fascinat-
ing they may be, are trivially small non-representative examples of
the spectrum of karst types and the spectrum of English prose.

     Others have defined a  karst  terrane in terms of the number of
sinkholes  shown on  a topographic map — "X" sinkholes per square
mile or per  ninth of a 7.5-minute  quadrangle.  This  is objective
but naive, and not recommended.  Aside from the fact that sinkholes
are rare or only subtly present in some karst terranes, in others
most  sinkholes  are not shown on  the  topographic map  for  many
reasons, ranging from non-intersection of them by contours, to non-
recognition of them by  photo-interpreters making topographic maps,
to farmers having partially filled them.   Indeed,  85% of  the 535
sinkholes in Winona Co., Minnesota, are not on the 7.5-minute topo-
graphic  maps and were  found  only by  field work  (Dalgleish and
Alexander, 1984).

     A karst is not a landscape.   Rather, it is a variable aggre-
gate of diagnostic surface and subsurface features that are known
to be genetically related to  its  development.  [Admittedly,  this
definition is a bit circular.]  We use karst terrane  interchange-
ably with  karst to  refer to both  the surface and the subsurface,
the lithologic expanse considered as a whole.  Strictly speaking,
however, karst terrain  refers only to the surface.   [Consult Hansen
(1991, p.  177) for discussion of the distinctions between terrane
and terrain.]

     The necessary  (essential) elements  of a definition  of karst
are that it  be  composed of  both  the  landscape and the subsurface
features formed as a result of dissolution of bedrock  (usually, but
not necessarily limestone and/or dolomite), that it be character-
ized by a distinctive subsurface hydrology, and that it include any
of a suite of distinctive surface and subsurface features  that are
sufficient  (non-essential)  for fulfillment of the definition.

     The presence of any one of the following types of distinctive
surface  and subsurface features  is  sufficient for  diagnosing a
terrane as a karst.  No specific  one of them is necessary.  Lack
(or apparent lack) of nearly all of them does not mean a terrane is
not a karst:
                                584

-------
     1.  Sinkholes (any closed depressions,  with or without a dis-
        crete opening at their bottom,  formed by dissolution and/or
        collapse of bedrock, with flushing and/or collapse of soil
        into a subjacent cavity)  and internal  drainage to them;
     2.  Dry valleys  (in humid climates);
     3.  Springs (draining carbonate, sulfate,  or halide rocks);
     4.  Caves (open to the surface  or accidentally encountered by
        drilling);
     5.  Sinking streams (that sink  at a hole known as a swallet);
     6.  Dissolutionally enlarged joints and/or bedding planes (as
        seen in cores  or outcrops);
     7.  Grikes  (soil-filled,  dissolutionally  enlarged  joints  or
        grooves; also  known as cutters or soil karren);
     8.  Karren (dissolutionally, subaerially, water-carved grooves
        on rock, commonly subparallel).

     A few examples  of types 1 through  5 may  be  shown on a topo-
graphic map, but most  of  them and  types 6 through 8  can only be
found by field work.  If one looks in the field, and does so dili-
gently,  some of these eight features, perhaps all of them, will be
found.  We are so confident of this  that we say "If there is car-
bonate rock, there is  almost certainly some type of karst."

     The apparent lack of some or all of the above eight features
in a  carbonate  terrane,  or their non-recognition, does not mean
that a karst terrane  is not present.  It probably means that more
field work or drilling  is  needed to  find them.   Similarly, we con-
sider the presence of just one of the features to be diagnostic of
a karst  because that one is usually representative of a much larger
population.

     The above definition of karst  in  terms of what  is necessary
and what is sufficient is totally  (and independently) consistent
with current European  thought  on the subject, thought  which has
evolved during the past 150 years.   To paraphrase  and abbreviate
Gams (1991), karst is a territory formed by dissolution of bedrock
and in which underground drainage of  precipitation prevails.  These
he identifies as the  absolute (necessary)  elements of the defini-
tion. The  so-called karst landforms  and other surface features are
the possible (sufficient)  elements.  Gam's definition of karst and
ours were independently conceived but are remarkably similar.

     Some people confuse  a  sinkhole with the karst  itself.   The
senior author has reviewed reports that describe proposed landfill
sites in flat-lying limestones such as the Mississippian-age St.
Louis Limestone.  These reports state,  for example:  "There is no
karst on the  property except for one,  in the southeast corner,
where there is  an open hole."  The report writers ignored the other
closed contours (indicators of sinkholes)  without an open hole at
their bottom,  and they truly confused an open sinkhole with the
karst itself.   This is like confusing  Cyrano de Bergerac's nose
with the man himself.   All will agree that there was a lot more to
Cyrano than just his  nose.  So also, there is  far more to a karst
than just a sinkhole or  a cave.   In the example cited here, the
                                585

-------
whole property is part of a karst terrane.

     Like sinkholes,  springs  are commonly omitted on topographic
maps.  Fewer than  5%  of  springs relevant to regional movement of
groundwater in the Mammoth Cave area of Kentucky are  shown  on 7.5-
minute (1:24,000) topographic  maps.   The remaining 95% had to be
found by field work (Quinlan,  1989, 1990a), but the presence of a
few  large  springs  not shown  on the maps  could  be inferred from
diagnostic contours suggestive of their alcove.  Non-indication of
springs is characteristic of  topographic maps of the eastern and
midwestern U.S.  MORAL:  In  consideration of the unreliability of
topographic map  depiction of  sinkholes and  springs,  absence of
recognizable karst  landforms  or  karst hydrologic features  on a
topographic map must not  be  interpreted to mean  a karst terrane is
absent.  Topographic maps are useful — but not reliable —  indica-
tors of the karstic nature of  a terrane.

     Obviously, there are degrees of karstification,  just as there
is a range in the quality of English prose.  One can quibble over
the subjective judgement  as  to degree of karstification or  quality
of writing, but nearly all carbonate terranes are a karst  and the
National Enquirer is English prose.

     The best map showing the  distribution of American karst ter-
ranes, albeit incomplete, is the hitherto almost  unknown useful map
(in color)  by Davies et al.  (1984).   Discussion  of our strong dis-
agreement with many of their classification criteria, however, is
beyond the scope of this paper.


                   DEFINITION  OF KARST AQUIFER

     A karst aquifer is an aquifer in which flow of water is or can
be  appreciable through  one  or more  of the  following:    joints,
faults, bedding planes,  and cavities  — any or  all of which have
been enlarged  by  dissolution  of bedrock.   This definition is in
terms  of  hydraulics,  not landforms (such  as  sinkholes),  grikes
(soil-filled joints),  or hydrologic features  (such  as swallets,
sinking streams, springs, and  caves).   The presence of sinkholes
and  any  or  all of  these other  features  is diagnostic  of  the
presence of  a karst  aquifer,  but  the  absence  (or  unrecognized
presence) of any or all of them — on the ground  or on a topograph-
ic map — does not mean  that a karst aquifer is absent.

     The definition of appreciable  (as  related to the quantity of
flow), like the definition of  aquifer itself, may vary as  the use
varies.  Appreciable is  a relative term and does not refer to the
same quantity of water in all  aquifers.   According to the standard
American law dictionary,  appreciable means "capable of being esti-
mated, weighed, judged of, or  recognized by the mind.  Capable of
being perceived or recognized by the senses.   Perceptible,  but not
a synonym of substantial" (Black, 1989,  p. 92).

     Dissolutional enlargement of joints and cavities in limestone
                                586

-------
is initially extremely slow.  Dissolution rates and their kinetics
are described by Palmer  (1991) and  Dreybrodt (1988,  p. 229-240).
Routing of much of the initial flow may be like that described by
Ewers (1982; partially summarized by  Ford  and Williams,  1989, p.
249-261,  and  Dreybrodt,  1988,  p.   223-229,  240-245), Johns  and
Roberts (1991), and Tsang et al.  (1991).

     It is probable that most flow in most karst aquifers is turbu-
lent (Ford and Williams,  1989,  p.  142-148).  We believe,  however,
that it would be a mistake to use such a criterion to distinguish
a karst aquifer from  one that  is not.  Aside  from the fact that
some flow in some karsts is probably  laminar (Hickey, 1984), accur-
ate measurement of flow velocities, fissure  widths, and hydrostatic
heads at inaccessible  parts of an aquifer is difficult and ambigu-
ous.  To routinely spend  considerable time and resources trying to
make such problematic,  unnecessary measurements distracts from the
likely reasons for assessing an aquifer.

     The presence of  non-carbonate  beds such as sand,  clay,  or
sandstone  at the surface of a terrane does not mean that the under-
lying carbonate beds do not comprise a  karst aquifer.  They usually
do.  Such mantled, covered, or caprock-protected aquifers, be they
shallow or deep and possibly confined,  can be regionally important.


                  RECOGNITION OF KARST AQUIFERS

     As a generalization,  if carbonate rocks such as limestone,
marble,  or dolomite are present,  or  if more soluble rocks such as
salt or gypsum are present,  assume  that the water  moving through
these rocks is in a karst aquifer — until  or unless  convincingly
proved otherwise.   This statement is  correct probably 95%  of the
time. Assume also an Orwellian nature  to the definition of a karst
aquifer:   All carbonate terranes are karstic and underlain by one
or more karst aquifers, but some are more  karstic than others.

     If there  is evidence for a karst terrane, assume that there is
also a karst aquifer.   The  aquifer, however, may be several orders
of magnitude less active than it once  was,  because  it is  a relict
karst and  no longer has its original climate or hydrographic set-
ting, or  because it  is  a  buried  karst which  is  completely  or
largely de-coupled from the present  hydrogeochemical  system (Ford
and Williams, 1989, p.  507-512;  Quinlan,  1978, 1990b; Samama, 1986,
p. 200-242;  Bosak et al., 1989;  Wright et  al.,  1991).

     The karstic nature of an aquifer can  often  be  inferred from
pumping tests in which the drawdown is  stepped, even though pumping
is continuous.  Such stepped  drawdown curves, however, can also be
a consequence of flow through fractures that have not been dissolu-
tionally enlarged.   All  one  can  be  confident of is that flow is
non-Darcian, so one still must be cautious.  Conversely,  however,
a smooth drawdown curve does not mean  an aquifer is not karstic.

     Other  indicators  of the karstic  nature of an aquifer — or
                               587

-------
that it is fractured, or both — are:

     •  Irregular cone of depression or no cone of depression, as
        defined by multiple observation wells around a pumped well;

     •  Non-linearity of the drawdown vs. discharge plot;

     •  Stair-stepped, irregular  configuration of potentiometric
        surface (where there are enough  data  to depict the surface
        with reasonable accuracy);

     •  Strongly bimodal  distribution of the logarithm of hydraulic
        conductivity of a suite  of wells completed in the same for-
        mation (Smart et  al..  1991).  Caution:  Karst  aquifers may
        also have a unimodal log-normal distribution  of hydraulic
        conductivity  (Moore, 1988a, 1988b);

     •  Non-coincidence of water levels  in closely adjacent wells;

     •  Bimodal or polymodal  distribution of daily or continuous
        measurements  of  specific  conductance (Ford and Williams,
        1989, p. 213-214; Bakalowicz and Mangin, 1980);

     •  Significant variation in specific conductance and hydraulic
        conductivity, as  interpreted  from wellbore fluid logs (Ped-
        ler et al.,  1990);

     •  Significant variations in the distribution of discharge, as
        measured by an impeller meter moved vertically during pump-
        ing at a constant rate  (Molz et a!., 1989);

     •  Significant variations  in the  distribution  of  flow in a
        pumped or unpumped well, as measured  by a movable electro-
        magnetic or thermal flowmeter in the borehole  (Nyquist et
        al.. 1991; Kerfoot et al.. 1991; Hess and Paillet, 1990);

     •  Significant differences in the breakthrough  curves in a
        well for different dyes  injected into several wells and re-
        covered in that well  (Quinlan,  1992).

     The first four  of these indicators are illustrated and briefly
discussed  by  Wisconsin   Geological  and  Natural History Survey
(1991).  Proof of appreciable flow through joints, faults, bedding
planes, and cavities can  also be obtained  by  inference from cores,
video  logs,  caliper  logs,  and temperature  logs.    Lacking such
proofs, if the bedrock is  limestone or dolomite, one would  be well-
advised to believe in the probable presence  of a karst aquifer.


       PRINCIPLES FOR CLASSIFICATION OF CARBONATE AQUIFERS

     The behavior of  an aquifer is dependent on three essentially
independent properties:   the mode of recharge, extent of  storage,
and  type of  groundwater transmission  (flow)   (Smart and Hobbs,
                               588

-------
1986).  In karst terranes, recharge may range between predominately
point recharge  (for  instance,  via sinking streams)  to dispersed
recharge  (for instance,  where water enters the karst aquifer over
a broad area and through  an  overlying permeable sandstone, thick
soil,  or  perhaps densely fractured rock with a thin soil).  Simi-
larly,  flow can be considered as a continuum between conduit flow
in large-diameter,  pipe-like caves and diffuse flow via fractures
and intergranular pores.  Storage is largely controlled by porosity
and geological disposition of the  aquifer, but ranges  from high,
where there is a deep saturated  zone, to low, where the aquifer is
perched and has a limited extent and thickness.  In order to char-
acterize  an aquifer,  it is necessary to define  its position on each
of  the three  continua,   effectively  mapping  it  into a  three-
dimensional space defined by recharge,  storage,  and flow (Figure
1) .  Also shown in this  figure are the general types of recharge,
storage,  and  flow which  characterize  each continuum.   These are
discussed further below  and the rationale  for them is  given by
Smart and Hobbs (1986) , but it is important to  stress here that the
boundaries are  fuzzy.   Each type  grades  into and  overlaps with
those adjacent.

     Conduit flow, in the strict sense,  has been and should be used
to refer to  flow through dissolution passages with  diameters of
centimeters to meters,  following  the  usage of Shuster and White
(1971), Atkinson and Smart  (1981), and Smart and  Hobbs  (1986).
Velocities are commonly high; flow  is commonly turbulent.  Diffuse
flow,  in  the strict sense, is confined to tight  fractures and pores
with small,  interconnected openings, with diameters of centimeters
or less.   Velocities are low; flow is  commonly laminar and may be
Darcian.  These two terms, conduit flow and diffuse flow, have been
subsequently  (and  erroneously)  broadened by  Quinlan  and Ewers,
1985;  Quinlan 1989, 1990a) and by numerous  other writers to include
implicit  characteristics  of water-quality and discharge variabili-
ty.   Between the two end-members,  we can intuitively recognize a
continuum of intermediate flow types.  Pipe flow in open cave pass-
ages represents a flow type close  to the  conduit  end-member.  In
other karst aquifers, dissolution has resulted  in dense networks of
open fissures which are  intermediate in character between conduit
and diffuse flow.  Finally,  flow through  intergranular pores and
fractures essentially unmodified by dissolution  approximates the
diffuse-flow end-member.

     Diffuse flow, as used in describing  karst  aquifers, should not
be construed to be the laminar, dispersed flow characteristic of
granular  aquifers and described  by  Darcy's Law.  The term diffuse.
as used here,  is intended to mean slow,  both laminar and slightly
turbulent flow of water  through a system of  small,  discrete path-
ways that  are being dissolutionally  enlarged, albeit extremely
slowly.  These pathways  range from fractures and near-microscopic
conduits  to intergranular pores.  Hence,  a diffuse-flow karst aqui-
fer may be described as having  flow generally restricted to dis-
crete pathways, with minimal turbulence rather  than the extremes of
turbulence characteristic of  conduit-flow aquifers.  The principle
difference between a diffuse-flow part of a  karst aquifer and a
                                589

-------
                       CLASSIFICATION  OF  CARBONATE  AQUIFERS
                                             VULNERABILITY
                                                   HYPERSENSITIVE
                                                    KARST AQUIFERS
                                                   VERY SENSITIVE
                                                    KARST AQUIFERS
                                                   MODERATELY SENSITIVE
                                                   KARST AQUIFERS

                                                   SLIGHTLY SENSITIVE
                                                   NON-KARST AQUIFERS
                                                   (In Fractured Rock)
                                                    NON-AQUIFERS
Figure 1.  Conceptual model for carbonate aquifers which recognizes
independence of recharge, storage,  and flow.  Volumes #1 through #4
depict the relative vulnerability of karst and non-karst carbonate
aquifers, ranging from hypersensitive to slightly sensitive.  Each
of the three axes represents a different  non-arithmetic continuum
that ranges from  0% to 100% for each end-member.   Boundaries be-
tween vulnerability fields are approximate, intuitive, and transi-
tional.  The relationships shown and implied are depicted within a
cube because there are three independent  variables, each of which
may have a value up to 100%.  The coefficient of variation  (CV) of
hydraulic  conductivity,  as  discussed by Smart  et  al.  (1991), and
the CV of specific conductance  (conductivity), as discussed  in this
paper, are each at  a  maximum at the upper far  vertex of the cube
and a minimum at the lower  near  vertex; the  CV of both parameters
decreases  along the diagonal connecting them.  Not all the boun-
daries of  volume  #5 (non-aquifers)  are shown.   This solid extends
been used in two senses:   the  strict sense,   referring to water
transmission through  variously sized apertures, and  in the broad
along part of  the back edge of  the cube but its  edges have been
omitted,  for  clarity.    (Greatly modified after Smart and Hobbs,
1986)
                                 590

-------
Darcian granular aquifer is that  the  flow pathways in the former
have been and are being significantly enlarged by dissolution.

     To partially summarize,   conduit flow and diffuse flow have
been used  in two senses:  the strict sense,  referring  to water
transmission through variously sized  apertures,  and in the broad
sense,  referring to aquifer type as represented by the sum of all
responses to recharge,  storage, and flow.   The sense intended by a
writer must be interpreted from context.  We henceforth use them in
the strict sense, to describe transmission.

     Recharge ranges from point to dispersed.  Point is used in the
same sense that Smart and Hobbs (1986) used concentrated. but the
former term was  selected to be in conformity  with its widespread
usage  in  America.    However,  we  have deliberately not  used its
false,  apparent antonym, non-point,  because it does not mean aeri-
ally distributed.   Although large  quantities  of "high-strength"
agricultural runoff  may enter  an aquifer at the swallet of a sink-
ing stream, i.e., at a  point,  it is  technically classified by most
state agencies as non-point pollution.   Accordingly,  we have re-
tained dispersed in the same sense as Smart and Hobbs (1986).

     Point recharge  is  characterized by sinking and losing streams.
However, even if these are  absent,  concentration of recharge may
still occur in karst terranes with closed depressions where drain-
age is often via a shaft system (Smart and  Friedrich, 1986).  This
may also occur in limestone aquifers overlain by a thick regolith
and lacking obvious  depressions.   Aquifers with recharge through
shaft systems occupy the center of the recharge continuum, grading
into recharge through  fissures toward the dispersed  end-member.
The  latter is  characterized  by  recharge  through  fractures  and
intergranular voids.

     Storage ranges between high  and  low end-members  (defined in
terms of the  ratio  of storage volume to  annual  recharge) .   Two
factors control  storage, the effective porosity of the aquifer and
its geological disposition in relation to regional base-level and
any confining beds (White, 1977).   In practice, the aquifer config-
uration dominates.   Hence we divide  the continuum into three tran-
sitional groups.  The first comprises aquifers with limited stor-
age, predominantly in the unsaturated  zone,  primarily in soils and
the subcutaneous (epikarst)  zone  (Smart and Friedrich, 1986).  The
second is similar to the first but  also  has a seasonally drained
saturated zone that  may be part of a shallow water-table aquifer.
Finally, the high-storage end-member comprises aquifers with an un-
saturated zone,  a seasonally inundated saturated zone, and either
perennial storage in a confined aquifer  or a  water-table aquifer
which extends well  below spring  level.   Examples  of  such high-
storage karst aquifers  are the Madison aquifer, chiefly in Montana
and Wyoming  (Downey,  1984; Plummer et al.,  1990), that  in the
carbonate-rock province of the Great Basin, chiefly in Nevada and
Utah (Burbey and Prudic, 1991), and in the Floridan aquifer (Sun
and Weeks,  1991, p.   5-9; see  also U.S. Geological Survey Profes-
sional  Papers 1403-A through I).

                               591

-------
     Storage in conduits (pipes),  as a fraction of the  total stor-
age within an entire aquifer,  is always very small.  Thus aquifers
having storage only  in  conduits would represent the extreme low-
storage end-member.

     Each of  the three independent  variables  that determine the
nature of an aquifer — recharge,  storage,  and  flow —  ranges from
the extremes  shown in  Figure  1  by  upper-case  letters,  POINT to
DISPERSED, LOW to HIGH, and DIFFUSE to CONDUIT.  The  three major
types  of  each of  the three  independent variables are  shown in
lower-case letters  indented between their two  extremes.   These
three sets of three major types are  each part of a  non-arithmetic
continuum that extends  from 0  to  100% fracture and intergranular
recharge, 0 to 100%  soil  +  subcutaneous + seasonally  saturated +
perennial storage,  and 0 to  100% dominance by conduit-flow,  as
shown.

     The eight corners of the cube correspond to the following end-
members, most of which  probably do not exist as pure  end-members
and two of which  are not karst. As with any continuum,  however, it
is easier to think in terms of the  end-members while recalling that
most of reality  lies  between them.   The pure end-member vertices
are:
     1. Point recharge, high storage, conduit-flow  karst aquifer;
     2. Point recharge, low storage, conduit-flow karst aquifer;
     3. Point recharge, high storage, diffuse-flow  karst aquifer;
     4. Point recharge, low storage, diffuse-flow karst aquifer;
     5. Dispersed recharge,  high storage, conduit-flow  karst aqui-
        fer;
     6. Dispersed recharge,  low storage, conduit-flow  karst aqui-
        fer;
     7. Dispersed recharge, high  storage,  diffuse-flow non-karst
        aquifer;
     8. Dispersed recharge,  low storage, diffuse-flow non-aquifer.

     Obviously,  there are numerous  intermediate types  of karst
aquifers, but we  do not propose to describe them or the  end-members
in detail here or to  give examples of each of  them.   Instead, we
have stated and  illustrated the principles for classification of
carbonate aquifers.

     Remember, Figure 1 is a classification of carbonate aquifers,
not karst terranes.   As mentioned in the Introduction, there are
seven other criteria by  which karst can be usefully classified.  To
be practical, however, we  use here only the criterion that  is most
relevant to the movement of and monitoring of pollutants —  aquifer
hydrology.

     Moderately  sensitive karst  aquifers  may  behave  in  a less
flashy manner than hypersensitive  karst aquifers because of either
diffuse flow or high storage or dispersed recharge or any combina-
tion of them, not simply because of  diffuse flow.   In  fact, chem-
ograph/hydrograph criteria do  not separate which of these is the
case, only that aquifer responses are slower and more  damped.

                               592

-------
           RELATION BETWEEN TYPES OF KARST AQUIFER AND
            SENSITIVITY TO GROUNDWATER CONTAMINATION

     The particular susceptibility of karst aquifers to groundwater
pollution  is related to the rapid transmission of pollutants with
little attenuation,  dilution,  or dispersion  via dissolutionally
enlarged fractures  and cave conduits (Aley, 1986;  Hoenstine et al. ,
1987; Hannah et al.,  1989; Murray et al.,  1981; Quinlan and Ewers,
1985; Edwards and Smart,  1989;  Quinlan and Rowe,  1977).  However,
recharge and storage  processes  are also important.  If recharge is
via dispersed routes such as a  porous sandstone caprock overlying
the limestone,  any pollutant entering from the surface will be con-
siderably  attenuated and delayed before reaching the cave conduit
in the underlying limestone.  The risks of severe contamination of
a spring or other discharge point are thus somewhat reduced,  al-
though they  could  still be  high if  the contaminant  had  direct
access to  the subsurface, for instance, via a drainage well (Craw-
ford, 1984).  If there were a substantial body of water stored in
phreatic fissures and pores adjacent to a  conduit, further dilution
and attenuation are possible before the pollutant is discharged at
the spring or other discharge point.  Thus, moderate to high stor-
age and dispersed recharge will reduce susceptibility of discharge
points to  contamination.

     Intuitively, we can infer  the  susceptibility of carbonate
aquifers to groundwater contamination by considering  the type of
aquifer defined with reference  to the recharge, flow,  and storage
continua described above.   Accordingly,  in Figure 1, we introduce
and define four new terms for zones representing relative sensitiv-
ity to groundwater pollution: hypersensitive karst aquifer (at the
upper far  vertex, volume #1), very sensitive karst aquifers (volume
#2), moderately sensitive karst aquifers  (volume #3), and slightly
sensitive  non-karst aquifers  (at the lower near vertex, volume #4) .
There is also a non-aquifer class (volume #5)  in which the lack of
concentrated flow and low saturation preclude practical abstraction
of water.  Effectively, we have added a fourth continuum to Figure
1 which ranges from hypersensitive to groundwater contamination to
slightly sensitive thereto.  Note  that  all  aquifers  of whatever
type are sensitive to groundwater contamination  and thus require
proper management.    For  this reason, we  rejected  the end-member
terms non-sensitive and robust.

     Hypersensitive  aquifers are the most  karstic of  all  karst
aquifers and are characterized by conduit flow in  pipes (caves),
concentrated recharge (via swallets, stream beds,  or shaft drains),
and low storage  (limited saturated zone).   There is a high prob-
ability that pollutants introduced at the  surface will pass rapidly
downward,  without attenuation,  directly  into conduits where they
will be transmitted rapidly via pipe-flow  to springs  (generally the
only easily found  high-yield source of  water in such aquifers).
Even if there is limited  saturated  storage, little  dilution or
attenuation of pollutants will occur  as  point recharge generally
bypasses the storage sites  (Atkinson, 1977;  Smart  and Friedrich,
1986) .
                                593

-------
     Very sensitive aquifers are characterized by the presence of
either well-developed  flow in pipe-like  caves  (conduits), point
recharge via swallets,  or storage ranging from very low (unsaturat-
ed)  to very high (saturated), or a combination of two of  these.

     Moderately sensitive  aquifers do not  have  pipe-flow, point
recharge  through  swallets  and  losing  streams,  or  low  storage.
However, there is  still a  significant risk of  groundwater  contami-
nation,  with the possibility  of  rapid recharge through closed
depressions, rapid  transmission  through dissolutionally-enlarged
fissure networks,  and limited dilution when  seasonal  storage is at
a minimum.

     As  emphasized  in  our  earlier  definition, karst aquifers are
characterized by appreciable flow through dissolutionally developed
voids.  It must be recognized,  however,  that  for several  reasons,
some carbonate aquifers (extremely few, in our experience)  have not
undergone significant dissolution.   These reasons include limited
exposure time, presence of overlying carbonate-rich regolith, and
low solubility of bedrock.  These non-karst carbonate  aquifers are,
therefore, identical in character to clastic aquifers in  which flow
is via intergranular pores  or fractures developed during unloading
or exposure  in  the  near-surface environment.  They  are shown as
volume #4  in Figure 1.  For these  non-karst carbonate aquifers,
special attention  is not needed;  conventional techniques  of aquifer
development, protection, and monitoring are probably applicable,
but may have to be modified because of fracturing.


          CHARACTERIZING CARBONATE AQUIFER SENSITIVITY

Introduction
     The scheme described above uses observable criteria  and pro-
vides a  simple, intuitive  basis  for assessing the sensitivity of
carbonate aquifers to groundwater contamination.   However, we are
not yet able to adequately fix the position of individual  aquifers
within the  continua.   Clearly,  for administrative and  regulatory
purposes, we need a much simpler scheme, one which,  although less
rigorous, is capable of differentiating  aquifers which have a high
pollution risk from those which are less susceptible.

     A potential solution to this problem is to use the geochemical
or hydrological behavior  of a  major spring  to  characterize the
nature of the aquifer it drains.  We treat the aquifer as  a black-
box  which transforms  the   "signal"  from  natural storm  recharge
(which has  a distinctive  chemical  signature)  into an output res-
ponse  (hydrograph or chemograph) at the spring,  or possibly at a
seep or well shown to intersect fracture zones, dissolution zones,
or conduits.   If  this  signal  is substantially smoothed,  delayed,
and attenuated by passage through the aquifer, then the same will
happen to a pulse of pollutant; the aquifer will be only  moderately
sensitive  to contamination.  If it  is  transmitted  rapidly, with
little or  no attenuation  or dilution, the aquifer is hypersensi-
tive.
                                594

-------
Background
     Fortunately,  characterization and interpretation of geochemi-
cal and hydrological variation of  springs  has  been the basis for
many studies  of  karst aquifers   (see  the summary  in Ford  and
Williams,  1989,  p.  193-214,  and the analysis by Kiraly and Muller,
1979).   In a classic paper,  Shuster and White (1971) discussed the
variations of  water hardness, degree of saturation with respect to
calcite and dolomite,  and  Ca/Mg  ratios of springs  in  the Appal-
achians of Pennsylvania.   They  recognized two types  of springs:
(1) those  with highly variable water quality, wide fluctuations in
discharge, and rapid response to recharge events (which they called
conduit-flow springs)  and  (2) those with much  smaller changes in
water quality  parameters,  higher saturation state,  small fluctua-
tions  in  discharge,  and  lagged,  prolonged hydrograph  response
(which  they  called diffuse-flow  springs).   Similar  studies  by
Newson (1971)  in the Mendip  Hills,  England, supported the observed
continuum in  geochemical  response but suggested that  the highly
variable springs there were dominated by point recharge at swal-
lets,  while the  less variable springs were recharged predominantly
by diffuse percolation.   The proportion of diffuse to point re-
charge was similarly used by Atkinson (1977)  in explaining Mendip
spring hydrograph response.  In contrast,  Aley (1975,  1977) dis-
cussed the same  hydrograph  features in terms of the proportions of
transit and storage water.

     Despite the disparity  in these explanations of an essentially
identical range  of spring behavior, all of them are  correct.  As we
have discussed previously,  the most karstified aquifers (hypersen-
sitive aquifers) are characterized by low storage,  pipe flow, and
point recharge.  There is a strong tendency  for these three charac-
teristics to be  genetically and functionally linked.  Thus maximal
concentrated  dissolution of the  bedrock occurs  where aggressive
surface water  drains into limestone as recharge from adjacent im-
permeable rocks.  Caves develop (Palmer, 1990).  Conduits eventu-
ally link the  swallets  directly to  the springs which discharge the
sometimes large flow entering at the swallets (Ford and Williams,
1989,  p. 249-271).  Where conduits are developed, rapid discharge
of groundwater can occur,  and the  amount  entering  into long-term
storage is relatively limited.  The flashy, highly variable behav-
ior of Shuster and White's conduit-flow springs  is a consequence of
the fact that  much of their recharge is rapid, at discrete points,
and into conduits, and the fact that the conduit flow reduces the
effective aquifer storage.   However,  the more subdued response of
their diffuse-flow springs could equally well be explained either
by lack of point recharge,  or by a  geological disposition favoring
substantial storage,  or by the dominance of diffuse flow.

     In Figure 1, variability in the chemistry and flow of a spring
(or in  the  chemical  composition and water level of  a well that
intersects a fracture zone, dissolution zone, or conduit) decreases
systematically along all axes of the cube  as  one moves away from
the most karstified end-member, at the upper far vertex.  The most
damped  response  is expected  at  the  farthest  distance  from the
hypersensitive vertex.  As one might expect, this  is the diagonally
                               595

-------
opposite lower  near  vertex — aquifers  that are not appreciably
karstified and  are thus  only slightly  sensitive  to groundwater
contamination.

     Although the above generalizations by Shuster and White were
based primarily on water hardness and related to discharge and lag-
time, they are equally valid for numerous other water quality and
quantity parameters of a spring or an aquifer.  The easiest para-
meters to  monitor,  however, are  specific conductance  (hereafter
referred to as  conductivity),  temperature,  turbidity, discharge,
and lag-time between onset of recharge and the time of hydrograph
peak.  The conductivity,  an easily determined electrical property
of water,  correlates with the sum of dissolved major-ion concentra-
tions in water and often with  a single dissolved-ion concentration
(Hem, 1985; Miller et al.. 1988).  It is a quick, extremely inex-
pensive surrogate for a chemical  analysis when  it is not necessary
to know what specific major ions are present.

     The nature of a spring or other suitable monitoring point (and
the  aquifer  that  feeds  them)  can be  adequately  determined  by
systematically measuring its conductivity and by observing and es-
timating its  turbidity,  discharge, and  lag-time.    These latter
three parameters,  as well as temperature,  can be quantified, but it
is not always  necessary to do so.   Indeed,  variations in turbidity,
discharge, and  lag-time can often be  estimated accurately enough
for reliable characterization by interviewing local residents.


Geochemical Variation
     The direct relationship between water hardness and conductivi-
ty allows us to use  the same  numerical  values for discriminating
spring type  (or aquifer type) as did Shuster  and  White (1971).
Diffuse-flow aquifers  (as defined by them) are characterized by a
conductivity in which  the  coefficient of variation (CV; standard
deviation x 100 •*•  mean) is less than 5%.   Conduit-flow aquifers (as
also defined by them)  are  characterized by a  range of conductivity
in which the  CV is  greater than 10%.   Aquifers  that  have a con-
ductivity CV between 5 to  10%  are interpreted  to be a mixture of
diffuse flow and conduit flow.  [These empirical class boundaries
are arbitrary but  utilitarian.  The general validity of using hard-
ness at all,  however,  has  been challenged, as discussed in the two
paragraphs following the next.  Nevertheless, conductivity CV has
seemingly  worked  well to  date.]   We have  assumed the  same  CV
boundaries for  conductivity  as an  index  of aquifer sensitivity.
For  moderately  sensitive  aquifers  the  conductivity CV  is  5%  or
less.  For very sensitive aquifers it is between 5 and  10%, and for
hypersensitive  aquifers the CV is greater than  10%.   Work is in
progress to verify and, if necessary,  adjust these boundaries.

     We estimate  that  most of the other chemical parameters will
probably have similar CV's.

     Shuster and White's use of carbonate hardness for defining the
nature of an aquifer was strongly criticized by Bakalowicz  (1976)
                               596

-------
who emphatically asserted that conductivity is a regionalized vari-
able — in the sense  of Matheron  (1965) and as the concept is dis-
cussed by David (1977, p.  91-114)  and by Henley (1981, p. 9-34) —
and that it provides a more accurate and sensitive indicator of the
characteristics of an aquifer.  Regular measurements by Bakalowicz
(1979), cited by Bakalowicz and Mangin (1980)  and Ford and Williams
(1989, p. 213-214, Fig.  6.24),  have shown the following about the
frequency distribution of  conductivity  at springs in various types
of aquifers:
     Granular aquifers:   unimodal, relatively high conductivity
     Fractured aquifers:   unimodal, relatively low conductivity
     Karst aquifers:   polymodal,  wide range of conductivity

     Bakalowicz and Mangin  (1980) and Mangin (1984) stated that the
CV of  hardness  is not a valid measure of the complexity  of the
polymodal distribution of  hardness.   This is partly  because the
accuracy of the  standard deviation used to calculate CV is based on
the validity of the assumption of a normal distribution (Hamburg,
1977, p. 42-44).   They  are technically correct, but  we disagree
with the necessity for their rejection of  its use.   The standard
deviation can still be calculated for a polymodal distribution, and
it is still a measure of dispersion about the mean.   However, its
meaning is different.  For  example, the mean + two standard devia-
tions might include only 80% of the conductivity values rather than
95.5% of them.  We justify use of  the CV of conductivity, imprecise
as it is for a polymodal distribution,  with a quote from the emi-
nent statistician  John W. Tukey:  "Far better an approximate answer
to the right  question which is often vague,  than an exact answer to
the wrong question, which can always be  made precise" (Tukey, 1962,
p. 11) .  [This same rationale could be employed to justify cautious
use of standard techniques for interpreting  slug, bail,  and pump
tests in karst aquifers.]

     Bakalowicz (1979) also showed that the frequency distribution
for daily measurements of conductivity at  a particular karst spring
in the Pyrenees varied significantly from year to year over a 4-
year period and was trimodal; the amplitude  and  frequency of the
maxima changed  from year  to  year and,  as one might  expect,  from
storm to storm.  We have used the daily  data in his dissertation to
calculate that the CV of conductivity at this spring for the 4-year
period was 6.4%, 7.9%, 6.4%,  and  5.7% (mean =6.6%) — in spite of
its high  variability  — suggesting that  it  is a very sensitive
spring but not greatly different  from a hypersensitive spring.  We
interpret the slightly more than  low value of  its CV to be a func-
tion of fluctuations  of recharge,  storage, and flow in the aquifer
studied.  We  do not have comparable long-term data for American or
British karst springs, but a review of existing data and a study of
other springs are  underway.

     The use  of  conductivity as a measure of the nature of an aqui-
fer is not an issue here.  The issues are  whether its CV is a reli-
able  measure  thereof and  what arbitrary  CV boundaries  can (or
should) be drawn between spring and aquifer types.   These matters
are being evaluated by us  and will be described in future publica-
                                597

-------
tions.  In the meantime,  we consider CV < 5%, CV = 5 to 10%, and
CV > 10% to  be practical  boundaries  for  discrimination between
aquifer vulnerability types.

     Conductivity CV's have been used to identify and  quantify the
difference between two groundwater basins (Quinlan and Ewers,  1981,
p. 488, 490).  A  hypersensitive spring  (mean = 326 micromhos per
cm; CV = 14%) was clearly differentiated from a moderately sensi-
tive spring (mean = 488  micromhos per cm; CV = 2%) with about one-
thirtieth the drainage area of the hypersensitive spring and dis-
charging in the middle of a distributary.  The latter is used for
water supply; the former is not.

     Two generalizations,  based on  the field experience  of the
senior author, can be made about the relations between conductiv-
ity, spring and aquifer type, and groundwater basin size.  First,
for equivalent basin size,  a  moderately sensitive spring  has a
significantly higher conductivity than a hypersensitive spring; 20
to 30% higher is  not unusual.   Second,  for both spring types, con-
ductivity  is directly  proportional to basin size,  which  is  a
control on residency time (time for reaction) of water.


Hydrological Behavior
     For hypersensitive aquifers, the ratio  between flood-flow and
base-flow discharge may well be 10:1 to perhaps 100:1.  Lag-times
between storm events  and  hydrograph maxima  will  range from  a few
hours to a  few days and be directly  proportional to basin  size.
Duration of response may be a few days to a few weeks.  In moder-
ately sensitive aquifers, the  ratio between flood-flow and  base-
flow discharge may be 3:1 or less; lag-time will range  from days to
weeks or even months, again in  direct proportion  to basin size, as
discussed by White (1988, p. 186), but as conduit-flow  and diffuse-
flow aquifers. Duration of response may be several weeks to sever-
al months.

     One could use just the ratio of flood-flow discharge to  base-
flow discharge to determine flow-type (and,  by inference, probable
spring type or aquifer type), as discussed by White  (1988, p. 184-
187) concerning aquifer type,  but  such data are essentially non-
existent in either published or open-file form.

     One might also  analyze the exponentially decaying recession
limb of several spring hydrographs and calculate the response  time,
tR, that is  characteristic of  the  spring and its  aquifer  type.
Physically, tR is the time it takes for the hydrograph to fall to
about 37% (= 1/e)  of  its maximum.  This time ranges from a few days
to a few thousand days and depends upon many factors not yet  fully
understood.  More accurately, tR is the reciprocal of epsilon  (what
has been called the  exhaustion coefficient), the constant in the
variable exponent of e in the  empirically  derived equation that
describes the  relationship  between  spring discharge and time, as
discussed by  White (1988,  p. 186).   Epsilon is  the slope of the
straight line describing that relationship when the data are  plot-
                                598

-------
ted on  semi-log paper,  as explained  by  Krumbein  and  PettiJohn
(1938,  p.  209-210).    [This graphical procedure is mentioned here
because a  similar technique will be proposed, in the Appendix, for
interpreting the recession  limb of storm-related conductivity data
and design of an optimal plan for reliably  sampling  springs and
conduit-fed wells.]  Such an analysis of discharge  may be useful if
one has the luxury  of  time  and money for instrumentation  of  a
spring and  if  it is  physically possible  to gauge it  reliably.
There are, however, other problems,  as discussed by White (1988, p.
186) :

     "Precipitation  events  are variable in intensity, spacing
     and duration.   If tR is  much less  than the mean spacing
     of precipitation events, hydrographs  will be flashy and
     the yearly  record will  consist of  a series  of  sharp
     peaks.   If tR  is on the  same order of magnitude  as the
     mean spacing of precipitation  events, individual  storms
     will tend to overlap but seasonal  changes in precipita-
     tion will appear.   If tR is  much longer  than  the  mean
     spacing of  precipitation events, the hydrographs will be
     broad and relatively featureless.  Large systems tend to
     have longer response times than small systems simply be-
     cause of the longer times needed  to  transmit the storm
     impulse from input to output.  Thus, short response times
     and high  discharge ratios  are indicators   of  conduit
     systems only for small basins."

     White's analysis, above,  is generally valid and we agree with
it, but  he assumes  basin  homogeneity  and ignores the  possible
influence of local,  near-spring input  of  storm runoff  in  a large
basin.  By inference, his conclusion about  short response times and
high discharge ratios would be also applicable to interpretation of
conductivity variations.  The validity of this inference  depends
upon the distinction between small  basins  and large basins.

     Within our experience in the  eastern, midwestern,  and south-
western U.S., we estimate  that damping of  spring hydrograph and
chemograph response  caused by basin size is not  significant at
springs draining less than about  500 square miles.   We conclude
that,  for whatever the several reasons  that control a spring's be-
havior, monitoring of its  conductivity and  determination  of its
conductivity CV are reliable measures of spring  (aquifer) variabil-
ity and response to recharge which may  include pollutants.   Clear-
ly, research is  needed on basin hydraulics and response as a func-
tion of numerous variables.


Limitations
     Using only  the variation in conductivity CV and discharge for
spring (aquifer)  characterization is not infallible.  For example,
the fourth author of  this paper, E.G. Alexander, Jr., and his stu-
dents have been  studying the  hydrology of Lanesboro Spring, near
Lanesboro, Minnesota.  Its hydrographs,  chemographs, and conductiv-
ity plots are those of  classic diffuse-flow springs — more than
                                599

-------
99% of the  time.   Yet dye injected  at  either of two wet-weather
swallets, 2.5 miles away, during base-flow conditions was recovered
in only 8.5  hours and with a breakthrough curve having a  very steep
rising limb.   The  recession limb is  complex,  has multiple peaks
that are reproducible in repeated tests, and takes 2 weeks to re-
turn to background levels.  The occasional muddying  of  the Lanes-
boro spring waters (3 times  in  2  years)  and the results of these
repeated tracer tests* during base flow can be explained in terms
of infrequent storm-related point-recharge into an otherwise dis-
persed recharge, moderate to high  storage,  conduit-flow aquifer
that is probably very sensitive  to pollutants  for more than 99% of
the year and hypersensitive for  less than 1% of the year.  The be-
havior of this spring is totally consistent with the  ideas expres-
sed here and by Smart and Hobbs  (1986), but at  variance  with those
of Shuster  and White  (1971)  if  one  thinks of  the aquifer just in
terms of diffuse flow (as suggested by the very low CV for its con-
ductivity)  rather  than  recharge,  storage, and flow.   The spring
behavior also brings  into question the reliability of conductivity
CV in assessing aquifer vulnerability.

     We do not have enough data to know how common such inconsis-
tency is between low  CV  of conductivity and occasional hypersensi-
tivity of a spring  and its aquifer,  or whether  it occurs as a gen-
eral phenomena.  We  do  believe, however, that such  storm-related
behavior is probably uncommon and contributory to a distinctive-
mode  on  a  conductivity-frequency plot.   Nevertheless, we  also
believe that use of the  conductivity CV,  as we  advocate, will usu-
ally be  reliable for an overwhelming majority of  springs.   This
reliability justifies its use.

     The valuable concepts of conduit-flow aquifer and diffuse-flow
aquifer  (as defined by Shuster and White,  1971) have  been usefully
applied  for more than  30  years,  even though  Bakalowicz  (1976),
Bakalowicz  and Mangin  (1980), Mangin (1984),  and Smart and Hobbs
(1986) questioned them and called  attention to  their deficiencies.
We recommend here that these two terms be retired as  the basis for
classification of aquifer type  —  even  though there is usually a
* Briefly stated, most tracer tests involve injection of a recog-
  nizable non-toxic  substance at  one  point and recovery of it at
  another, yielding information  about  flow-direction, -dispersion,
  -destination,  and  -velocity   (Quinlan  and  Alexander,  1990;
  Atkinson and Smart, 1981; Lepiller  and Mondain, 1986).  A well-
  designed, properly  performed,  and correctly interpreted tracer
  test, most  commonly employing one or more fluorescent dyes, is
  the most powerful  technique for studying karst aquifers  (Aley,
  1986; Ford and Williams,  1989, p. 219-241; Quinlan,  1989; Quinlan
  and Ray, 1989) but, until recently, they have rarely been used,
  except by karst specialists.  The value of tracing in all types
  of aquifers, however,  is  now widely recognized (Moltyaner,  199Oa,
  1990b; Maloszewski and Zuber,  1990;  Tsang et al. , 1991; Adams and
  Davis, 1991).
                               600

-------
partial overlap of so-called conduit-flow aquifers with hypersensi-
tive aquifers and usually a partial overlap of so-called diffuse-
flow aquifers  with  moderately sensitive  aquifers.   The  terms
conduit-flow and  diffuse-flow should be retained, however, as des-
criptors of flow type  and to refer to aquifers dominated by one
type of flow or the other.


     PRACTICAL EVALUATION OF VULNERABILITY OF KARST AQUIFERS

Procedure
     A spring is the pulse of the aquifer it drains.  Understanding
the variations in  its  discharge and  water quality  (or  those of
other suitable monitoring points)  is the key to understanding the
hydraulics of its  (their)  aquifer and assessing  aquifer vulner-
ability -

     We believe that interpretation of water quality and quantity
parameters,  as described herein, even though less  meticulous than
proposed by Smart and Hobbs  (1986),  is  practical  and justifiable
for the purposes we propose — to identify aquifer vulnerability —
and adequate for use by non-specialists.   The classification pro-
cedure we advocate  is  less  subjective  than what  they proposed.
Smart  and Hobbs  are more rigorous,  but we are more  practical.
Quantitative characterization of recharge,  storage, and flow pro-
portion are difficult,  time-consuming tasks best left to special-
ists.

     The conductivity CV of a spring  (and, therefore, of a ground-
water basin  presumed to be representative of the aquifer it drains)
should be determined  from at least 20 to  40 daily, weekly, or other
regular or random measurements of its conductivity.  The measure-
ments should be taken at  least  a day  apart  and  must include both
base flow and flood flow during the estimated peak discharge.  We
can not say  how many  samples are required; we can only advise that
caution must be used  to make the samples statistically representa-
tive of the  various conditions within the aquifer.  This is easier
said than done.

     It is critically important to include  conductivity measure-
ments  collected  under  both base-flow and  storm-flow conditions.
Twenty conductivity measurements taken only under base-flow condi-
tions,  perhaps under  the pressure of an arbitrary 30-day deadline,
will probably have a CV of less that 5% and will "prove" that the
spring  is characterized by diffuse  flow.   Such a conclusion is
wrong because it is based  on  data  that  do  not represent the true
variability of the system.  Reliable interpretation  of conductivity
is aided by also having daily measurements of rainfall,  the maximum
daily rainfall,  and the maximum rainfall per storm.

     We reiterate:   If  the  conductivity CV  is  5% or less, the
spring (aquifer)  is moderately sensitive  to pollution and is prob-
ably characterized by mostly diffuse  flow.  If the  conductivity CV
is greater than 10%,  the spring  (aquifer) is hypersensitive and is
                                601

-------
characterized by a much greater proportion of conduit flow.  If the
conductivity CV is between 5 and  10%,  the spring  (aquifer)  is very
sensitive and is probably a mixture of both diffuse flow  and con-
duit flow.  Also, reliable inferences  can be made about whether an
aquifer is dominated by moderate  sensitivity or is hypersensitive,
both by observing the turbidity and discharge during, immediately
after, and shortly  after  storms  of different intensities, and by
interviewing landowners who are familiar with the springs and their
behavior.

     Not a single published American report known to us has enough
data that would enable an  investigator to run a few simple conduc-
tivity CV calculations and determine the vulnerability (sensitivi-
ty) of the groundwater basins in a given area.  Therefore, if one
needs such data for a specific area, it is highly likely that they
can be acquired only by field work.


Difficulties and Limitations of Vulnerability Evaluation
     The conductivity CV is a useful method for characterizing the
vulnerability of aquifers;  it must be remembered,  however,  that the
hydrograph or chemograph (a plot  of chemical concentration or con-
ductivity vs. time)  of a  spring  reflects the recharge,  flow, and
storage characteristics of  the groundwater basin that it drains.
For site-specific  investigations in karst aquifers,  the springs
used to characterize an aquifer will probably not be on a site to
be evaluated and may even  be several  miles  away.  Their  possible
relevance to the hydrology of the site must be evaluated  with the
aid of geologic maps of the area  adjacent to the  site; dye tracing
may also  be necessary  to prove or disprove their  relevance  of
particular springs to a given site  (Quinlan et al., 1988).

     The conductivity CV method has some limitations in areas where
there are few springs  and  where  springs are  only located several
miles from a site under investigation  and tracing has demonstrated
that flow velocities are,  for example, 1 mile per year or slower.
An alternative approach in these  areas is to interview well owners
and determine if their well water turns turbid or  muddy after heavy
rains.   Be cautious.   Approximately  90 to  95%  of the  wells  in
karsts dominated by  conduit  flow are  in the  diffuse-flow part of
the aquifer that feeds the conduits.   [These are usually hypersen-
sitive aquifers.]   Accordingly,  many  well owners,  perhaps 150 to
200, need to be  interviewed before one  can be confident about using
wells to ascertain whether an aquifer is hypersensitive rather than
moderately sensitive.  Some  of these  5 to 10% of wells that turn
turbid or muddy are the ones most  likely to  be  positive for dye
during a tracer test.

     We hear the argument that "My client can't afford to interview
150 to 200 well-owners."  We  reply that "If you need the truth, you
can't afford not to  do so."

     Evaluation of conductivity CV  has severe limitations  for some
karst terranes, such as those in  much  of Florida, that are charac-
                               602

-------
terized primarily by  recharge  and flow rather than  also by dis-
charge.  Discharge at springs may be  several  tens of miles away,
perhaps beneath the ocean; flow velocities may be no more than a
few hundred feet  per  year.  The  springs,  if accessible,  can be
monitored for conductivity, but they will probably be inaccessible.
Well-to-well tracer tests  may  be practical for determination of
flow velocities for either  the design  of monitoring systems or for
determination  of wellhead and springhead protection areas based on
time of travel,  but tracer  tests would be unsuccessful in delinea-
tion of  wellhead  or  springhead protection  areas  based  on basin
boundaries because it  would take too long to do so.  Conductivity
CVs of wells in such terranes are not likely to be meaningful.

     Currently, some  of  the  serious  limitations  in  applying the
conductivity CV at springs as a method of assessing aquifer vulner-
ability are:  1) the  false assumption (by some agencies,  clients,
and consultants),  that monitoring wells alone are  sufficient to
characterize a karst aquifer, and 2)  an unwillingness by some of
the same to extend monitoring efforts beyond  the  boundaries of a
site under investigation, to all possibly relevant springs, wells,
and base-level stream  segments.   [If such short-sighted views and
restrictions are allowed to prevail,  and they should not be, the
quality  and credibility  of a  tracing investigation  is  severely
compromised.  So also  is  the integrity of the investigator compro-
mised. ]

     We argue  that a limited number of boreholes, for example five
at a given site, do not provide a statistically valid confirmation
of the existence or probable non-existence of conduits in the sub-
surface.  Benson and La  Fountain  (1984) have  shown that drilling
1,000  3-cm  holes  per  acre is  necessary in order to have  a 90%
chance of intercepting a  cavity about  2.3 m in diameter.  A lesser
number would be needed for  finding a conduit of the same diameter,
but many of the conduits  of interest at a given site are likely to
be only a tenth as wide (or smaller) .   A thorough survey of springs
and existing wells (see above discussion),  followed by  a well-
designed tracer test,  properly performed, and correctly interpret-
ed, will provide much more  cost-effective and accurate information
about the site.

     We also hear  the  argument that "My client (and his attorneys)
does not wish to perform any investigations off-site." In the real
world,  water-flows  (and  contaminants)  do  not respect  property
boundaries; they  travel  beyond them.   A consultant  may  have to
convince clients and their attorneys that judicious investigation
that minimizes  potential  liability  requires  following  the flow
wherever it may go.  Off-site field work  is probably always cheaper
than the off-site courtroom work it can prevent.

     Well-meaning supervisors, clients, regulators, and consultants
with whom  the  reader  interacts may  exert pressure  to  assess an
aquifer from the office or  without the amount  of field work we are
recommending as minimal.   In response, ask whether he  or she would
respect or trust a C.P.A.  who was willing  to  audit accounts over
                               603

-------
the telephone.  A field problem can only be resolved by field work.

     We wish to emphasize that:

     1. If a  terrane is underlain  by limestone  or dolomite, as
        indicated by an accurate  geologic  map and/or site-specific
        observations, the site is extremely likely to  be part  of a
        karst and have a karst aquifer;

     2. With one exception, there is no way without fieldwork, as
        described, to determine whether the aquifers in a site  area
        are hypersensitive  rather than moderately sensitive to  con-
        tamination and probably  dominated by  conduit flow rather
        than diffuse  flow.   The  sole exception  is when  so  much
        field work has already been done in the area that one can
        confidently  interpolate  or extrapolate  on the basis of
        experience or pre-existing state-of-the-art literature.


            USE OF AQUIFER TEST DATA TO CHARACTERIZE
           FLOW IN AND VULNERABILITY OF KARST AQUIFERS

     In some karst areas,  where domestic  wells  are sufficiently
abundant and accessible, it may  be  possible to characterize  flow
within an  aquifer by using aquifer tests  and  to make inferences
about  its  vulnerability from such  data.   One advantage  of  this
approach  is  that  it directly measures  the transmission (flow)
properties of an  aquifer,  whereas the conductivity CV and hydro-
graph analysis methods (discussed above) integrate flow, recharge,
and storage properties of the aquifer.  In addition, slug or bailer
tests and single-well pump tests are simple, rapid, and cost-effec-
tive methods for determining average of the vertical distribution
of aquifer horizontal properties.

     Briefly, the aquifer-test method  of assessing aquifer vulner-
ability involves  determination of hydraulic conductivity* from  slug
or bailer tests,  single-well pumping tests, or multi-well pumping
tests, calculation of CV of log hydraulic  conductivity, and inter-
pretation of cumulative probability  plots  of log hydraulic conduc-
ivity.  The distribution of hydraulic conductivity tends to be  log-
normal.  Therefore,  the  logarithm of  hydraulic conductivity  data
should be  used for calculation of  statistics,  and the geometric
mean of hydraulic conductivity should be used to best represent the
average of log-normally distributed values  (Domenico and Schwartz,
* Hydraulic conductivity, the volume of fluid (usually water) that
  will move through  a medium in a unit of  time,  T,  under a unit
  hydraulic gradient through a unit area measured perpendicular to
  the flow^ has the dimensions L/T because the L of volume divided
  by the L  of area cancels  to L, a  length.   Consequently, hydrau-
  lic conductivity, a volume, is often erroneously thought to be a
  velocity.
                               604

-------
1990,  p.  66-67; Bradbury and Rothschild, 1985).  The geometric mean
and CV of the hydraulic conductivity distribution are then used to
make inferences about aquifer vulnerability.  As a rule of thumb,
we suggest that at least 20 data points should be used to calculate
these statistics.  Additional  data points, however, will provide a
more representative  estimate of the distribution of hydraulic con-
ductivity within an  aquifer.

     For hypersensitive aquifers (those probably dominated by con-
duit flow),  the CV will be very high (>500%), with wells penetrat-
ing both highly transmissive fissures and conduits plus non-conduit
blocks with low permeability-   [The probability  of  a well inter-
cepting a conduit is  extremely small, but  doing  so  would simply
increase the CV of the sample]  .  However, the geometric mean of the
hydraulic conductivity will be low.   For moderately sensitive aqui-
fers (those  probably dominated by diffuse flow), the mean hydraulic
conductivity will be moderate to high,  reflecting the dissolutional
widening  of openings  in  the  limestone, but the CV will  be much
lower, reflecting the more homogenous  distribution of flow within
the aquifer (Smart  et al. ,  1991) .   In both cases,  the log-normal
probability plot of  hydraulic conductivity may well be polymodal.

     This hydraulic  conductivity approach has some advantages over
the indirect analysis of  specific conductance CV and hydrographs,
discussed above, because  it allows direct  determination  of flow
characteristics of  carbonate  aquifers.  The specific conductance
and the hydrograph are a response to the processing by the aquifer
of a signal  that is  not uniquely controlled by its flow character-
istics and  is  also  dependent on the  nature of recharge  to  and
storage in the aquifer.   Fortunately,  many of the most karstified
(and therefore most troublesome) carbonate  aquifers  have conduit
flow,  concentrated  recharge,  and low  storage, thus  providing an
unambiguous signal.   Other aquifers show a more damped signal,  im-
plying diffuse  flow, but this may result from an absence of concen-
trated  recharge  (for instance,  where  recharge  is  through  a
permeable caprock or thick soil mantle not breached by sinkholes)
or high  storage  (for instance,  in  a  deep,  saturated zone  of an
aquifer).

     There may be a  methodological  problem with the use of hydra-
ulic  conductivity for characterizing  a carbonate aquifer.   The
thickness of the open interval of  a borehole tested may  bias an
evaluation of  its hydraulic  conductivity-  More specifically, long
intervals may tend  to average and dampen  its true extremely high
and low values.  The measured high hydraulic conductivity of a well
is always less than  the actual value for a  fracture or other cavity
intercepted by  it  (Maureen  Muldoon, Wisconsin  Geological  and
Natural History Survey, written communication, January 1992).  The
possible relationship between the thickness of developed interval
in a borehole and the  hydraulic conductivity in different types of
carbonate aquifers needs  to be evaluated.

     As appealing, objective, useful, and diagnostic as interpreta-
tion of,  say,  30 hydraulic conductivity measurements may be, in
                                605

-------
actual practice obtaining  them may be difficult, time consuming,
and infeasible in  some areas.  There is little doubt, however, that
evaluation of the CV of  hydraulic conductivity from slug or bail
tests is superior in many ways (but not all ways) to evaluation of
the  CV  of  specific conductance.   The  latter technique  can be
cheaper but it takes longer and may have to be delayed because of
climatic seasonality.

     Slug and bailer tests  are reviewed by Fetter (1988, p. 196-
200) , Domenico  and  Schwartz  (1990,  p. 164-168),  and  Dawson and
Istok (1991).  There are many limitations to applying such tests,
both theoretical  and  logistical.   When  an  investigator uses the
standard techniques for interpretation  of  bailer or  slug tests
(Hvorslev,  Papadopulos,  and many others) or for single-well pumping
tests, he must be familiar with and willing to accept the assump-
tions inherent in  those analytical solutions.  All of these methods
assume Darcian flow  and assume also that the hydraulic conductivity
is homogeneous over the interval tested.  Since these assumptions
are rarely,  if ever, valid for karst aquifers,  the resulting hydra-
ulic conductivities  should not be used to calculate  flow velocities
or times of travel or to calibrate computer models.  The variation
in the hydraulic conductivity distribution, however, may be used to
draw conclusions about relative vulnerability of an aquifer.

     Logistically,  slug  tests are  probably the  easiest aquifer
tests to perform.  The ideal method of characterizing the variation
of  hydraulic  conductivity of  a  karst aquifer would be  to drill
wells specifically  for that purpose, then perform packer tests at
a set interval in them (Milanovic, 1981, p. 237-251; Domenico and
Schwartz,  1990,  p.   163-164).    This  is  not practical  for most
studies  and,  as a  result,  slug tests are commonly  performed in
monitoring wells and in existing domestic wells where the pump has
either not yet been installed,  or it has been pulled in order to
perform the tests.

     If  an investigator  is using  domestic wells as  points to
measure hydraulic conductivity,  single-well  pumping tests  (those
without an observation well) can not be used to reliably determine
hydraulic conductivity,  but they do provide an  advantage over slug
tests in that the pump does not have to be pulled.  The drawdown
and recovery data (easily recorded with a  downhole  pressure trans-
ducer) from a single-well pump test can, however,  be used to calcu-
late residual drawdown and transmissivity (Domenico and Schwartz,
1990, p. 162-163;  Kruseman and de Ridder,  1990, p.  193-197; Dawson
and  Istok,  1991,  p. 59-61).  A major problem,  however,  with any
attempt  to  reliably determine transmissivity or  even hydraulic
conductivity from any test  but a  packer test,  is that in a karst
aquifer it is often difficult, if not  impossible, to do so from  a
well that partially  penetrates  an aquifer in which the thickness is
unknown or  (in some cases)  indeterminable.

     The hydraulic conductivity measured by a slug  test, a single-
well pump test, or a multi-well pump test is the average hydraulic
conductivity measured over the entire test interval.   Using  this
                                606

-------
"average" hydraulic  conductivity to  predict flow  velocities or
travel times will result  in  significant  errors,  however, because
karst aquifers are characterized  by highly heterogeneous hydraulic
conductivity distribution  (i.e., low-permeability blocks,  and high-
conductivity conduits or dissolution zones) and most of the flow is
through the high-permeability zones.

     Hydraulic conductivity can also be measured using multi-well
pump tests.  The use of these tests  is limited to those projects
with budgets  that  are  sufficient to  install a  pumping  well and
several observation wells. The standard techniques  for the inter-
pretation of data from any type of pump test (Theis, Hantush-Jacob,
etc.) require  the  erroneous  assumption of  isotropic conditions;
other analysis techniques, however,  can be used to calculate both
the horizontal anisotropy  ratio (Papadopulos, 1965), as well as the
horizontal to  vertical anisotropy ratio (Weeks,  1969).  The assump-
tions of  the  different pump  test  interpretation  techniques are
clearly discussed by  Kruseman and de Ridder (1990) and Dawson and
Istok (1991) .  They also discuss techniques developed by Streltsova
and Boulton and by Streltsova that  may be more suitable for karst
aquifers.  Thiery et  al.  (1983) have also tried to develop a reli-
able technique for interpreting pump-test data from karst aquifers,
but they too have been unsuccessful.   The  discharge of  a well, par-
ticularly one  in a karst aquifer, is the integral of the hydraulic
character  (hydraulic conductivity,   dispersivity,  storage,  and
tortuosity) of all the different dissolutionally-enlarged fissure-
systems that feed it  (Quinlan,  1992).  The hydraulic conductivity
calculated from pump test data still  represents  a value averaged
over the interval tested and  should be not used in calculations of
Darcian or non-Darcian  flow.

     In spite  of the  potential errors in hydraulic conductivities
calculated from slug and  pump tests  in karst aquifers,  and their
possible misuse, we  do believe in  their value for evaluation of
aquifer hydraulic properties and probable vulnerability.   The  CV of
specific conductance  represents a response to the combined effects
of  recharge,  storage,  and flow, and gives a  direct, integrated
measure of the vulnerability  of a karst aquifer to pollutants.  In
addition, the  data are relatively inexpensive to collect.  However,
it may take longer to  collect or be  delayed because of climatic
seasonality.   Both  types of information  are complementary and
needed.  Neither type is intrinsically superior to the other.
                        A CAUTIONARY TALE

     A detailed study, using geophysical logs, borehole video logs,
lithologic core analyses, and unspecified aquifer tests, was made
of the  dissolution porosity and  permeability  at a  Florida site
(Robinson and Hutchinson, 1991).   [The thoroughness of their well-
designed and carefully executed study was  much more  than we have
recommended here.]   Tracer tests  were  performed by  injecting a
fluorescent dye into the open-hole interval of a well and measuring
the movement induced by pumping 1000 gallons per minute at a well
                               607

-------
200 feet away.  Dye concentration at the pumped well was measured
continuously with a filter fluorometer.   Based on standard assump-
tions about flow in porous media and a porosity of 20%, the theo-
retical arrival time for the dye should have been approximately 40
days, with a  4-day persistence time for the  dye.   Actual break-
through time was 5  hours.  The peak was at 22 hours  and the persis-
tence time was 28 days.  A second breakthrough of dye occurred at
36 days after injection and its persistence time was  8 days.  This
second breakthrough  is not  attributable to   response to  a storm
(James Robinson, U.S. Geological Survey, Tampa, Fla.,  oral communi-
cation, December 1991).  The bimodal distribution of tracer arrival
time was interpreted to indicate dual porosity in the aquifer, but
we suggest that bifurcation of a flow route, followed by rejoining
of the trunk route, is  a more plausible explanation (Tsang et al.,
1991; Ford and Williams, 1989, p. 226-228).

     This Florida study illustrates the heterogeneity of the aqui-
fer  and  the  impossibility  of using carefully  (and  expensively)
acquired  aquifer parameters  to make  correct predictions  about
travel times in karst aquifers.*  Most important, it shows that the
assumptions of uniform distribution of porosity and attendant dif-
fuse flow, as has been assumed and proposed in current strategies
of wellhead protection for the Floridan aquifer (DeHan, 1988), are
probably  not  valid.   Again,  the  conceptual  model  inadequately
corresponds to reality.
                             CAVEATS

     The summary of this paper lists six important caveats.  They
require no additional  explanation.  Please go back and reread them.
                            DISCUSSION

     Experience has shown that,  in the midwest  and south, there is
sometimes a rough  correlation between spring morphology and flow
type.  Springs dominated by conduit flow tend to have high-energy
discharge from a large  alcove that is eroding headward, commonly
have dry  high-level orifices that function only in response to
storms, and may have a deeply eroded channel. In contrast, springs
dominated by diffuse flow commonly have low-energy discharge from
a very small  alcove,  shallow channels,  and fontaphilic (spring-
loving) vegetation.  These correlations  are guides,  for  first  ap-
* The discrepancy between actual tracer velocities and those based
  on  predictions from  analysis of  cores and  logs  is  partly a
  consequence of extrapolation from a borehole scale of  study  to a
  field  scale,  as discussed for karst  aquifers by Kiraly  (1975;
  summarized by Ford and Williams,  1989, p.  134,  Fig.  5.4)   and
  Smart  et al.  (1991) ,  and as  recognized in  other aquifers by
  Domenico and  Schwartz  (1990, p. 84-87,  371-84)  and  many others.


                                608

-------
proximations during the first few days of field work.

     A review of post-1960 classifications  of karst aquifers was
made by Ford and Williams  (1989, p.  166-170).  No attempt has been
made here to repeat it.   The  aquifer classification we have pro-
posed  is  consistent with  and complementary  to those  of  Mangin
(1984)  and Ford and Williams (1989, p. 169).  It is an update and
revision of one first proposed by Smart and Hobbs (1986).

     Figure 1 depicts a generalized conceptual hydrogeologic model
of karst  aquifers.   As such,  it incorporates  all  the essential
features of their physical  system.  It is  not,  however,  a site-
specific,  accurate  representation  of the conditions  beneath and
adjacent to  a  particular  site.    Although  the degree  of  detail
necessary  for  various  sites differs,  presentation  of a  site-
specific conceptual model  should be accompanied by maps and cross
sections.    Conceptualization  is a  means  of achieving  a graphic
idealization of the actual geologic/hydrologic conditions,  and it
deliberately ignores minor features that are not important to the
overall picture.

     The arbitrary but convenient dividing points in the continuum
between conduit flow and diffuse  flow are 20% and 80%.  The inter-
mediate members of the continuum, having 20  to  80% conduit flow and
80 to  20%  diffuse flow,  have been known as mixed  flow aquifers
(Atkinson and Smart, 1981; Quinlan and  Ewers,  1985) .   This term,
too, should be  retired.  [Strictly  speaking, all aquifers that are
not 100% one flow-type or the other are mixed flow, but it has been
convenient to draw intermediate boundaries  at 20% and 80%.]

     Diffuse flow can be  better understood by  making an analogy
with the flow of water through a stack of bricks; flow is chiefly
between the  bricks  rather than through them.   We  recognize,  of
course, that water  also moves within the rock  matrix of a karst
aquifer, just as it would within the bricks,  but we consider the
intergranular Darcian component of  the  flow to be unimportant to
this discussion. The importance of such dual porosity, especially
during tracer tests, has  been analyzed in  a  series  of papers by
Maloszewski and Zuber, the most recent of which  is Maloszewski and
Zuber (1990).  A recent mathematical analysis of dual porosity has
been made  by Douglas  and Arbogast  (1990)  and  its role in water
quality distribution within fractures and intergranular pore space
has been briefly described by Edmunds (1981).

     The complex relationships  between conduit  flow  and diffuse
flow, particularly as related to the role of dual porosity during
recharge events, have been mathematically analyzed by Onder (1986).
He showed that base-flow recession through a conduit may be related
to aquifer properties,  as  might be expected.

     A substantial amount  of the water in an aquifer dominated by
conduit flow moves as  diffuse  flow,  as  explained  by Atkinson
(1977).  In the British aquifer he  analyzed, 60  to 80% of the dis-
charge is via conduits; active storage in the diffuse-flow part of
                               609

-------
the aquifer that feeds  them is  30 times greater than in the con-
duits.  Approximately 50% of the recharge is from sinking streams
and sinkholes; the  remaining 50%  is via slow percolation through
the soil.

     There is a perception  among  some hydrologists and engineers
that, because  conduits  comprise  such  a  small percentage  of an
aquifer  and  because  most of the rock  surrounding them  may be
characterized by diffuse flow,  overall, there  is no significant
problem of aquifer vulnerability.  Wrong!  This is like saying that
the hole blown into  the  bottom of a battleship by a torpedo is only
10 feet in diameter but, because the rest of the steel is imperme-
able, there is no problem of sinking.  Wrong again!

     There are two further complications.  First, almost  all  of the
discharge from all aquifers dominated by conduit flow occurs at a
point or group of points along  a  line.   A small component of the
discharge  is  as  diffuse  flow,  as described by Atkinson (1977).
Second,  most karst  aquifers dominated by diffuse  flow  also dis-
charge at a point or group of points,  but some may have almost all
of their discharge  over a strip or a broad  area,  as discussed by
Quinlan (1989)  as an exception to  his Implicit Assumption #1.  Our
experience suggests  that point  discharge is predominant  in most
carbonate terranes.   Accordingly, no consideration  to  such non-
point discharge has been given  in this paper.

     In lieu of daily or twice- or thrice-weekly measurements of
specific conductivity, it is likely to be far more cost-efficient,
and could be more reliable, to continuously record it with a trans-
ducer and a data-logger.  Also,  having a continuous record of con-
ductivity facilitates making an  analysis that enables optimization
of sampling frequency,  as discussed below,  and if  such would be
useful for a given project.   The utility of the conductivity data
could be inexpensively  enhanced by using  the same data-logger to
also  continuously  record temperature with  a platinum-resistance
thermometer and  turbidity.   Interpretation  of  temperature data,
although  it can  be  quite  subtle,  can give  significant  useful
insights into  aquifer hydraulics   (Meiman, et  al. ,  1988;  Davies,
1991).

     Continuous recording of water-quality parameters also elimi-
nates the possibility of aliasing, a sampling error that can intro-
duce significant bias into estimates of the magnitude and frequency
variation of a parameter and reliable characterization of it  — all
as a result of sampling with insufficient frequency.  Aliasing is
a phenomenon in which a  high-frequency component of  a signal takes
on the identity of one with a lower frequency.   [The  familiar illu-
sion  of  stagecoach wheels  in a movie  first  going  forward, then
apparently backward, then forward  again is an example of aliasing.]
Signal processing  theory, information  theory,  spectral analysis
theory, and time-series  analysis theory all include a theorem which
states that a  continuous, periodic signal,  with frequency  compo-
nents in the range / = 0 to  f =  fmax per unit  of time,  can be  recon-
structed from  a  series  of equally spaced samples if the sampling


                               610

-------
frequency exceeds  2/max  samples per unit  of time.   /^ = 1/2At,
where At = the sampling  interval and fmax = the Nyquist number, the
fastest frequency detectable.   The slowest frequency detectable is
zero; it never repeats.   Trend  lines have zero frequency and infi-
nite period.   These concepts and many others relevant to interpre-
tation of  water-quality data  are discussed by Meade  and Dillon
(1991, p.  16-19), Karl (1989, p. 4-8),  Bandat and Piersol (1986, p.
335-339),  Pierce (1980,  p. 35-39),  Bras and Iturbe (1984, p. 172-
177), Davis  (1986,  p.   257-258,  349),  Gardner  (1988,  p.  49-52),
Chatfield (1975, p. 155-158),  Gottman (1981, p.  15-16).  Aliasing
can be prevented by continuously recording a parameter or by taking
samples extremely  frequently,  perhaps  every 30 seconds.   A pro-
cedure for using  conductivity  data  to prevent  aliasing of water
quality data and for determination of an appropriate aliasing-free
sampling protocol for using water-quality data to characterize an
aquifer or its spring is given in the accompanying appendix.  The
procedure is similar to that already discussed as used  to analyze
exponential  decay  of the recession limb  of a hydrograph.   We
believe the procedure is also applicable to the design  and inter-
pretation of tracer tests.

     Aliasing of water-quality  data can be  a  serious problem be-
cause it can  inadvertently and seriously bias their interpretation,
but it is not mentioned by Gilbert  (1987)  or  in any other water-
quality publications we have reviewed.   Only Davis (1986), of the
several geostatistical  books  known to us,  discuss aliasing  of
spectral data, but there are several signal-processing  texts such
as  that  by Robinson (1980) which  recognize aliasing  of seismic
data.

     It is highly likely that most water-quality studies and moni-
toring  of  springs  and  wells  characterized by  conduit flow are
flawed by aliasing unless they have been sampled frequently  (per-
haps daily) and/or as related to storm events.   Unless  they have,
such effects are incapable of accurately characterizing the dynamic
changes in  water quality except by improbably good luck.   Most
assuredly, this  is a fatal defect  of the monitoring  studies at
Mammoth Cave National Park described by Meiman  (1991); as designed,
such studies are intrinsically incapable  of achieving  their pur-
pose.   Probable aliasing of water  quality parameters  is  also a
deficiency of the recent synthesis of data on the largest and most
famous spring in France, Vaucluse (Puig,  1990).

     Time-series analysis and spectral  analysis  of data  from well-
designed  studies  of regularly  sampled and continuously sampled
spring waters was first  applied in various papers by Mangin and by
Bakalowicz, as partially summarized  by  Ford and  Williams (1989, p.
210-214).   Application of these data-analytical techniques to other
studies, like  those described by Mazor  et al.  (1990)  and those
cited in the  following paragraph, will lead to significant improve-
ments in understanding  of the  processes  operating in karst aqui-
fers.

     The complexity of  flow within an aquifer dominated  by conduit
                               611

-------
flow, plus  interpretation of  the  lack of  coincidence among the
pulses of stage,  conductivity,  and temperature, are discussed by
Meiman et al. (1988)  and,  to a lesser extent,  by Ford and Williams
(1989, p. 204-214).  Interpretation of coincidence among the pulses
associated with a different  type of conduit flow is discussed by
Idstein and Ewers (1991).

     Further discussion of the interpretation  of hydrographs of
springs and streams is given by Bonacci (1987, p.  75-81), Ford and
Williams  (1989, p. 193-203),  Domenico and Schwartz (1990,  p. 15-
17),  and Hall (1968).  Spectral analysis and time-series analysis
of spring discharge have been used by Bakalowicz  and Mangin  (1980),
and  Mangin  (1982,  1984)  as  a basis  for  aquifer classification.
Chemograph separation is discussed by Ford and Williams (1989, p.
204-214).  Detailed summary of these topics  is beyond the scope of
this paper, but some are discussed in the accompanying appendix.

     Spills in karst terranes frequently occur on surfaces with a
soil profile 5 to perhaps 50  feet thick and are  sometimes confined
to the soil.  With luck, the  entire spill  is confined to the soil.
Many workers would assume that Darcian flow equations adequately
describe the movement of liquids through the soil.  Unfortunately,.
non-Darcian macropore flow (Quinlan and Aley,  1987; Everts  et al.,
1989; Wells and Krothe,  1989) commonly occurs in  the soil; contami-
nants can move rapidly through the soil to, and perhaps into, the
bedrock.  Fluid movement  through soil with macropores can not be
analyzed  as  though  it  were  through a granular  medium.    Indeed,
Watson and Luxmore (1986)  found that macropores and mesopores  in an
Appalachian karst area, which together constitute only  0.3% of the
soil volume, account  for  96%  of all infiltration.  This topic is
discussed in great detail  by  Wilson et al.  (1991).  Further, the
pollutant may be perched at the soil-bedrock contact and may move
hundreds  of  feet  along it before  intercepting  a dissolutionally
enlarged  joint which  allows access  into bedrock of  the karst
aquifer below.

     One  of  the more  challenging  frontiers in contemporary karst
hydrogeology is the epikarst  (also known as the subcutaneous zone).
This is the dissolutionally well-developed zone immediately  beneath
the soil and above the relatively dry zone above  the phreatic zone.
In a sense,  it is much like a perched aquifer,  but usually there is
no discrete perching  bed.  As discussed by Ford and Williams (1989,
p.  120-121  and  in  many  other places  throughout their  book),
Williams  (1983,  1985), Bonacci  (1987,  p. 28-35),  Friedrich and
Smart (1981), Smart and Friedrich (1986), Field (1990),  Quinlan and
Ray  (1991),  and  McCann  and  Krothe  (1991), there is  significant
water-storage capacity  and sufficient interconnection to  diffuse
the  movement of pollutants and tracer dyes widely.  Drainage from
the  epikarst  is not  uniform  but is down preferential pathways to
the  subsurface.  Residence time for pollutants  or tracers in the
same epikarst can range from hours to tens of years (Quinlan and
Ray, 1991).  Much is being learned about flow and storage  in epi-
karst, both  in the U.S. and in Europe  (Smart  and  Friedrich,  1986;
Chevalier, 1988), but  most of it has not yet been published.  We
                                612

-------
believe that epikarst hydrology  is the most enigmatic  in all of
karst.   Assume that an epikarst exists beneath any soil mantle —
until you prove otherwise.  Hydrologists, engineers, and regulators
should  be prepared to encounter epikarst.  They should be prepared
also to be  confused and frustrated by it.

     Some  experienced hydrologists believe that  slug  and bailer
tests measure  little  more than  skin-effects  and a  small,  local
volume  adjacent to the well bore.  The results of  such tests can be
non-representative and misleading.  They may be  in  vogue chiefly
because they are quick, inexpensive, and readily create a favorable
response in some regulators.  All this is more true than not.  But,
as such tests are orders of magnitude superior to laboratory deter-
minations of permeability of a  core sample and more representative
of aquifer  hydraulic properties in a karst or any other aquifer, so
also, pump  tests are significantly superior to slug and bail tests.
Nevertheless,  tens to perhaps hundreds  of pump tests are necessary
to accurately  characterize an aquifer.  All  this is to be expected.

     As pointed out by Kiraly  (1975) and  quoted by Ford and Will-
iams (1989,  p.  134, Fig. 5.4),  karst aquifers become both more het-
erogeneous  and more anisotropic with passing time.  Their average
total porosity is  directly proportional to the reference volume of
rock considered and may  range  over three orders  of  magnitude —
depending upon the scale  at which investigation is performed.  Ac-
cordingly,  it could be convincingly argued  that  a well-designed,
carefully  interpreted tracer  test, because it  may sample  a far
larger  volume of  an aquifer,  can be superior to  a pump test for
characterizing some properties of  an aquifer.   The  two types of
test are complementary, however.  Neither is intrinsically superior
to the  other.

     Many  topics  have  been discussed  in this paper.    Its most
important points are that:

     •   Clear, accurate definitions of karst and karst aquifer —
        as  well as practical techniques for  recognizing them  — are
        needed.  We have fulfilled this need.

     •   Regulations concerning karst should focus on  protecting
        waters of  karst aquifers.  Sinkhole  collapse is a relative-
        ly  minor problem.

     •   Karst aquifers should be classified according to the three
        highly variable major controls  on their nature:  recharge,
        storage,  and flow, as shown on  the cube depicted in Figure
        1,  and grouped into three classes of vulnerability:  hyper-
        sensitive, very sensitive,  and moderately sensitive.

     •   Karst aquifers should no longer be classified as conduit-
        flow aquifers or diffuse-flow aquifers.  These terms should
        be  retained,  however,  as useful descriptors of flow type.

     •   Systematic measurement of the specific conductance  of
                               613

-------
        springs during base flow and during and after flood flow,
        with evaluation  of  the coefficient of  variation (CV) of
        such data,  is a valuable and probably reliable predictor of
        not only the net effect of an aquifer's recharge, storage,
        and flow characteristics but also its relative vulnerabili-
        ty to the  adverse effects  of  contaminants and mismanage-
        ment.

     •  A major deficiency of most attempts  to monitor water qual-
        ity of springs and wells in karst terranes  is  aliasing, the
        failure to  detect and recognize high-frequency fluctuations
        (which can  also be  high-magnitude fluctuations)  in water
        quality because the  water is not sampled often enough.  The
        highest frequency signal that can be  resolved  has a wave
        length that is twice the distance (time) between successive
        regular observations.

     •  Two objective, sensitive, flexible procedures  for determin-
        ing a cost-efficient,  aliasing-free  sampling  frequency for
        reliable  characterization  of   springs,   aquifers,   and
        conduit-fed wells now exist.   They are described in the
        accompanying appendix.   We believe these procedures are
        applicable to:   1) Reliable assessment  of water quality, 2)
        Sampling for and interpretation  of  tracer tests in which
        dye-concentration is  not  monitored continuously,  and 3)
        Design, operation, and interpretation of pulse tests (Ford
        and Williams, 1989,  p. 226-228).

     •  The empirically established values  for  the CV  of conduc-
        tivity, recommended for use in separating the three types
        of vulnerability of  karst aquifers, are < 5%,  5 to 10%, and
        > 10%.  The reliability of these values is being reviewed
        by  the authors,  but we  do not  expect  any substantial
        changes in them.

     •  Imperfect as their results in karst aquifers usually are,
        slug, bail, and  pump tests should be  performed in an effort
        to characterize the  aquifer.  An individual test has little
        significance, but evaluation of  the  geometric mean and the
        CV of a representative sample (at least tens) is a useful
        technique for approximating the hydraulics of karst aqui-
        fers and their differences.

     Finally, a few words should be said about how this paper was
written.   Although a manuscript  was completed  in  time  for the
meeting, it was constructive dialogue with participants there that
forced all of us to rethink our assumptions, interpretations, and
conclusions, and to reformulate our ideas.  The three  authors of
the original manuscript believe that  this final version has been
significantly enhanced by the  input and active participation of the
others and the reviewers we have acknowledged.  We  know other meet-
ing participants have had similar experiences in  completing their
manuscript.  This  is a  good,  satisfying way to do better science
and is one reason why we believe that conference proceedings should
                                614

-------
be distributed after a meeting, not before or at it.


                          SOLICITATION

     Readers of  this paper  are invited  to correspond  with the
authors and suggest improvements in the definition of karst aquifer
and ways  to classify such  aquifers meaningfully,  usefully,  and
practically.


                        ACKNOWLEDGEMENTS

     Tom Aley,  Richard C.  Benson, Gareth J. Davies,  Gary A. Davis,
Malcolm S.  Field, Michael R. McCann, Maureen Muldoon,  Joseph A.
Ray, Paul A. Rubin, and Joe P.  Sparks made numerous useful sugges-
tions.  We think that the  content and clarity of expression of the
ideas discussed  herein has  been significantly enhanced by their
critical review.  William  B.  White graciously clarified mathemati-
cal interpretation of discharge and conductivity data during hydro-
graph recession.  We  are indebted to them.  The typesetting talents
of Marcia Williams  in lettering Figure 1 are greatly appreciated.
We thank them all.
                        REFERENCES CITED

Adams, M.C., and Davis, J.   1991.  Kinetics of fluorescein decay
     and its  application  as a  geothermal tracer.   Geothermics.
     20(1/2):53-66.

Aley, T.  1975.  A predictive hydrologic model for evaluating the
     effects of land use and management on the quantity and quality
     of water  from Ozark  springs.  Ozark Underground Laboratory,
     Protem, Mo.  [Republished in Missouri Speleology, 1978.  18:1-
     185. ]

Aley, T.   1977.   A model for  relating land  use  and groundwater
     quality  in southern Missouri,   in  Dilamarter,  R.R.,  and
     Csallany, S.C.,  eds.   Hydrologic Problems in  Karst Areas.
     Western Kentucky University, Bowling Green,  p. 323-332.

Aley, T. 1986.  Complex radial  flow of ground water in flat-lying,
     residuum-mantled limestone in the Arkansas Ozarks.  Environ-
     mental Problems  in Karst Terranes and Their Solutions Confer-
     ence (1st, Bowling Green,  Ky.),  Proceedings.   National Water
     Well Association, Dublin,  Ohio.   p. 159-170.

Atkinson, T.C.  1977.   Diffuse  flow and conduit flow in limestone
     terrain  in the   Mendip Hills,   Somerset  (Great  Britain).
     Journal of Hydrology.  35:93-110

Atkinson, T.C., and Smart, P.L.   1981.  Artificial tracers  in karst
     hydrogeology,  in A Survey of British Hydrogeology.  The Royal
                                615

-------
     Society, London,  p. 173-190.

Bagchi, A.   1990.   Design, Construction  and Monitoring of Sanitary
     Landfill.  Wiley-Interscience, New York.  284 p.

Bailey, P-,  and Ward, W.D., eds.  1990.  Understanding Ground-Water
     Contamination:  An Orientation Manual.  Executive Enterprises
     Publications, New York.  223 p.

Bakalowicz,  M. 1976.  Geochemie des eaux karstiques.  Une methode
     d1etude de 1"organization des ecoulements souterrains.  Col-
     loque d'Hydrologie en Pays Calcaire  (2nd, Besancon-Neuchatel) ,
     Proceedings.  Annales Scientifique de 1'Universite de Besan-
     con.  Series 3, 25:49-59.

Bakalowicz,  M. 1979.   Contribution  de la  geochemie des eaux a la
     connaissance de 1'aquifere karstique  et de la karstification.
     Doctoral thesis, L'Universite Pierre et Marie Curie  - Paris 6.
     269 p.

Bakalowicz,  M., and  Mangin,  A. 1980.   L'aquifere  karstique.   Sa
     definition,   ses characteristiques  et son identification,  in
     Hydrogeologie:  Interactions  entre 1'eau souterrain  et  son
     milieu.   Societe  Geologique de France, Memoire  hors-serie
     no. 11.  p.  71-79.

Bartlett, A.A.    1992.   The  Feynman Effect  and the  boon docs.
     Physics Today.  45(1):67,69.

Beck, B.F.,  Quinlan,  J.F.,  and Ewers, R.O.  1987.  Presentation of
     the E.B. Burwell,  Jr.,  Memorial Award to  James F. Quinlan and
     Ralph  O.  Ewers.   Geological Society of America,  Bulletin.
     97:141-143.

Bendat, J.S., and Piersol, A.G.   1986.   Random Data: Analysis and
     Measurement  Procedures.    2nd  ed.    Wiley-Interscience,  New
     York.   566 p.

Benson, R.C.,  and La Fountain,  L.J.  1984.   Evaluation of subsi-
     dence  or collapse potential  due to  subsurface  cavities.
     Multidisciplinary  Conference on Sinkholes  Conference (1st,
     Orlando, Fla.), Proceedings.   Balkema, Rotterdam, The Nether-
     lands,   p. 201-215.

Black, H.C.   1979.   Black's Law  Dictionary.  5th  ed.   West, St.
     Paul,  Minn.   1511 p.

Blavoux, B.,  and Mudry,  J.    1988.   Importance of  the sampling
     rhythm in the hydrochemical study and hydrocinematical knowl-
     edge of karstic aquifers.  Karst Hydrology and Karst Environ-
     mental Protection (Guilin, China),  Proceedings.  (Internation-
     al Association of  Hydrogeologists, 21st Congress).   Geological
     Publishing House,  Beijing.   Memoir 21.   2:864-869.
                                616

-------
Bonacci,  O.   1987.  Karst Hydrology, With Special Reference to the
     Dinaric Karst.  Springer, Berlin.  184 p.

Bosak,  P-,  Ford,  D.,  Glazek,  J.,  and  Horafiek,  I., eds.   1989.
     Paleokarst:    A  Systematic and  Regional Review.   Elsevier,
     Amsterdam.   725  p.

Bradbury, K.R.,  and Rothschild, E.R.   1985.  A computerized tech-
     nique for estimating the  hydraulic conductivity of aquifers
     from specific capacity data.   Ground Water.   23(2):240-246.

Bras, R.L.,  and Rodriguez-Iturbe, I.   1985.   Random Functions in
     Hydrology.   Addison-Wesley, Reading, Mass.   559 p.

Brooks,  K.N., Ffolliott, P.F.,  Gregersen,  H.M.,  and Thames,  J.L.
     1991.   Hydrology and the Management of Watersheds.  Iowa State
     University,  Ames.   392 p.

Burbey,  T.J., and  Prudic,  D.E.  1991.   Conceptual  evaluation of
     regional ground-water flow in the carbonate-province of the
     Great Basin, Nevada, Utah, and adjacent states.   U.S. Geologi-
     cal Survey,  Professional Paper 1409-D.  84 p.

Chatfield,  C.  1984.   The Analysis of Time Series:   An Introduc-
     tion.   3rd ed.  Chapman and Hall, London.  286 p.

Chevalier,  J.  1988.   Hydrodynamique  de la zone  non-saturee d'un
     aquifer karstique:  etude  experimentale.    Doctoral  thesis,
     Universite des Sciences et Techniques du Languedoc.  212 p.

Christiansen, T.H., Cossu, R. ,  and Stegmann,  R.   1989.  Sanitary
     Landfilling:   Process,  Technology, and Environmental Impact.
     Academic,  London.   592 p.

Committee on Ground Water Quality Protection.  1986.  Ground Water
     Quality Protection:  State and  Local Strategies.    National
     Academy Press, Washington, D.C.   309 p.

Crawford, N.C.   1984.  Sinkhole flooding associated with urban de-
     velopment upon karst terrain: Bowling Green,  Kentucky.  Multi-
     disciplinary  Conference  on Sinkholes  (1st, Orlando,  Fla.),
     Proceedings.   Balkema, Rotterdam,  The Netherlands.   p.  283-
     292.

Crawford, N.C.   1991.  Emergency response procedures  for spills of
     hazardous waste liquids upon karst terranes.   Hydrogeology,
     Ecology, Monitoring,  and Management of Ground Water in Karst
     Terranes Conference  (3rd,  Nashville,  Tenn.),  Proceedings.
     National Ground Water  Association, Dublin,  Ohio.   [in this
     volume]

Dalgleish, J.,  and  Alexander,  E.G., Jr.  1984.  Sinkhole distribu-
     tion in Winona County,  Minnesota.  Multidisciplinary Confer-
     ence on Sinkholes (1st, Orlando,  Fla.), Proceedings.  Balkema,


                                617

-------
     Rotterdam, The Netherlands,  p. 79-85.

David, M.  1977.  Geostatistical Ore Reserve Estimation.  Elsevier,
     Amsterdam.  364 p.

Davies, G.J.  1991.   Water temperature  variation in springs  in the
     Knox Group near Oak Ridge,  Tennessee.  Hydrogeology, Ecology..
     Monitoring, and Management of Ground Water  in  Karst Terranes
     Conference  (3rd,  Nashville,  Tenn.),  Proceedings.   National
     Ground Water Association, Dublin,  Ohio.   [in this volume]

Davies, W.E.,  Simpson,  J.H.,  Ohlmacher,  G.C.,  Kirk,  W.S.,  and
     Newton, E.G.  1984.  Engineering Aspects  of Karst.  National
     Atlas of the U.S., U.S. Geological Survey Map US 05572.  Scale
     1:7.5 million.  1 sheet  in color,  with text by W.E. Davies.
     [Supercedes Open File Map 76-623,  1976, which  is illegible]

Davis, D.S.   1955.   Nomography and Empirical Equations.  Reinhold,
     New York.  236 p.

Davis, G.A., and Quinlan, J.F.   1991.   Legal Tools  for the Protec-
     tion of Groundwater  in Karst Terranes.  Hydrogeology, Ecology,.
     Monitoring, and Management of Ground Water  in  Karst Terranes
     Conference  (3rd,  Nashville,  Tenn.),  Proceedings.   National
     Ground Water Association, Dublin,  Ohio.   [in this volume]

Davis, J.C.   1986.   Statistics and Data Analysis in Geology.  2nd
     ed.  Wiley, New York.  646 p.

Dawson, K.J., and Istok,  J.D.  1991.  Aquifer Testing:  Design and
     Analysis  of Pumping and Slug Tests.   Lewis,  Chelsea,  Mich.
     344 p.

DeHan, R.S.   1988.   New approach to protection of sensitive aqui-
     fers in Florida.  Hydrological  Science and Technology.  2(2):
     51-59.

Devinny, J.S., Everett, L.G.,  Lu, J.C.S., and Stollar, R.L.  1990.
     Subsurface Migration of Hazardous  Wastes.  Van  Nostrand Rein-
     hold, New York.  387 p.

Domenico,  P.A.,  and Schwartz,  F.    1990.  Physical and Chemical
     Hydrogeology.   Wiley, New York.  824 p.

Douglas, J., Jr., and Arbogast,  T.   1990. Dual  porosity models for
     flow in naturally  fractured reservoirs, in Cushman, J.H., ed.
     Dynamics  of Fluids  in  Hierarchical Porous Media.  Academic,
     London.  p. 177-221.

Downey, J.S.   1984.   Geohydrology of  the  Madison  and associated
     aquifers in parts  of Montana, North Dakota,  South Dakota, and
     Wyoming.  U.S. Geological Survey,  Professional Paper 1273-G.
     47 p.
                                618

-------
Dreybrodt,  W.   1988.   Processes  in Karst  Systems.   Springer,
     Berlin.   288  p.

Driscoll, F.   1986.  Groundwater and Wells.  2nd ed.  Johnson Divi-
     sion,  St. Paul,  Minn.   1089 p.

Edmunds,  W.M.  1981.  Hydrogeochemical  investigations,  in Lloyd,
     J.W.,  ed.  Case-studies in Groundwater Resources Evaluation.
     Clarendon,  Oxford,   p.  87-112.

Edwards, A.J., and Smart, P.L.  1989. Waste disposal on karstified
     Carboniferous  Limestone  aquifers  of  England  and  Wales.
     Engineering and Environmental Aspects of Sinkholes and Karst
     Conference (3rd, St. Petersburg, Fla.), Proceedings.  Balkema,
     Rotterdam,  The Netherlands.  p. 165-182.

Erdelyi,  M.,  and Galfi,  J.   1988.  Surface and Subsurface Mapping
     in Hydrogeology-   Wiley, New York.   384  p.

Everts, C.J., Kanwar, R.S.,  Alexander,  E.G.,  Jr.,  and Alexander,
     S.C.   1989.  Comparison of tracer mobilities under laboratory
     and field  conditions.    Journal  of Environmental  Quality.
     18(4):491-498.

Ewers,  R.D.,  1982.  Cavern development in the dimensions of length
     and breadth.   Ph.D. thesis, Geography.   McMaster University.
     398 p.

Fetter, C.W.   1988.   Applied Hydrogeology.    2nd ed.   Merrill,
     Columbus, Ohio.   592 p.

Field,  M.S.  1988.  The vulnerability of  karst aquifers to chemical
     contamination.   Hydrological  Science and  Technology.   4(1-
     4):130-142.

Field,  M.S.   1990.   Transport of  chemical contaminants  in karst
     terranes: Outline  and summary, in Selected Papers on Hydroge-
     ology from the 28th International  Geological Congress, Wash-
     ington,  D.C.,  1989.  International Association of Hydrogeol-
     ogists,  Selected Papers.  Heise, Hannover.  1:17-27.

Fookes, P.G., and Vaughan,  P.R.   1986.   A Handbook of Engineering
     Geomorphology.   Surry University Press,  London.  343 p.

Ford,  D.C.,  and  Williams,  P.W.   1989.   Karst Geomorphology and
     Hydrology.   Unwin Hyman, Boston.  601 p.

Franklin,  B.   1781.   A letter to a Royal Academy, in Japikse, C.,
     ed.  1990.  Fart Proudly.  Enthea,  Columbus, Ohio.  p. 13-17.

Freeze, R.A.,  and Cherry, J.A.  1979. Groundwater.  Prentice-Hall,
     Englewood Cliffs,  N.J.   604 p.

Friedrich,  H., and Smart, P.L.   1981.   Dye tracer studies in the


                               619

-------
     unsaturated-zone  recharge  of  the  Carboniferous  Limestone
     aquifer of the Mendip Hills,  England.  International Congress
     of Speleology (8th,  Bowling Green,  Ky.),  Proceedings.   1:283-
     286.

Gams, I.  1991.  The origin of the term  karst  in the transition of
     karst (kras)  from deforestation to forestation.  International
     Conference on Environmental  Changes in Karst Areas  (Padova,
     Italy), Proceedings,  p. 1-8.

Gardner, W.A.  1988.  Statistical  Spectral Analysis: A nonprobabi-
     listic Theory-  Prentice-Hall, Englewood Cliffs, N.J.   566 p.

Gilbert, R.O.  1987.  Statistical  Methods for  Environmental  Pollu-
     tion Monitoring.  Van Nostrand Reinhold, New York.  320 p.

Goldman, B.A.,  Hulme, J.A.,  and Johnson, C.  1986.  Hazardous Waste
     Site Management: Reducing the Risk.  Island Press, Washington,
     D.C.  314 p.

Gottman,  J.M.    1981.     Time Series  Analysis:  A  Comprehensive
     Introduction  for  Social  Scientists.    Cambridge  University
     Press, Cambridge.   400 p.

Gupta, R.S.   1989.   Hydrology and Hydraulic  Systems.   Prentice-
     Hall, Englewood Cliffs, N.J.  739 p.

Hall, F.R.  1968.   Baseflow recessions — A review.  Water Resour-
     ces Research.  4:978-983.

Hamburg, M.  1977.  Statistical Analysis for Decision Making.  2nd
     ed.  Harcourt Brace Jovanovich, New York.  801 p.

Hannah,  E.D.,  Pride,  T.E.,  Ogden,  A.E., and Taylor,  R.   1989.
     Assessing groundwater flow paths from pollution sources in the
     karst of Putnam County, Tennessee.  Engineering and Environ-
     mental  Aspects  of  Sinkholes and Karst  Conference  (3rd,  St.
     Petersburg, Fla.),  Proceedings.  p. 183-188.

Hansen,  W.R.   1991.   Suggestions to  Authors of Reports  of the
     United States Geological Survey.  7th ed.  Government Printing
     Office, Washington, D.C.  289 p.

Hem, J.D.  1985.   Study  and interpretation of  the chemical charac-
     teristics of natural water.   3rd ed.  U.S. Geological Survey,
     Water Supply Paper 2254.  264 p.

Henley,  S.   1981.  Nonparametric  Geostatistics.  Applied Science,
     London.   145 p.

Henry, J.G.,  and  Heinke,  G.W.  1989.   Environmental Science and
     Engineering.  Prentice-Hall, Englewood Cliffs, N.J.  728 p.

Henry, M.F.,  ed.   1989.   Hazardous  Materials Response Handbook.
                               620

-------
     National Fire Protection Association, Quincy, Mass.  561 p.

Hess,  A.E.,  and Paillet, F.L.  1990.  Application of the thermal-
     probe flowmeter in the hydraulic characterization of fractured
     rocks,  in Paillet, F.L., and Saunders, W.R.,  eds.  Geophysical
     Investigations for Geotechnical Investigations.  ASTM Special
     Technical Publication 1101.  American Society for Testing and
     Materials (ASTM), Philadelphia.  p. 99-112.

Hess,  J.W.,  and White, W.B.  1988.   Storm response of the karstic
     carbonate  aquifer of  southcentral  Kentucky.    Journal  of
     Hydrology.  99:235-252.

Hickey,  J.J. 1984.  Field  testing  the hypothesis of Darcian flow
     through a carbonate aquifer.  Ground Water.   22:544-547.

Hoenstine, R.W.,  Lane, E.,  Spencer,  S.M.,  and O'Carroll, T.  1987.
     A  landfill  site  in  a karst  environment,   Madison  County,
     Florida —  A case study.   Multidisciplinary  Conference  on
     Sinkholes  and the  Environmental Impacts   of  Karst  (2nd,
     Orlando, Fla.), Proceedings.   Balkema, Rotterdam, The Nether-
     lands,   p. 253-258.

Idstein, P.J., and Ewers,  R.O.   1991.  Deducing karst aquifer re-
     charge, storage,  and  transfer mechanisms  through continuous
     electronic monitoring.  Hydrogeology,  Ecology, Monitoring, and
     Management of Ground  Water in  Karst Terranes Conference (3rd,
     Nashville, Tenn.), Proceedings.  National Ground Water Associ-
     ation,  Dublin, Ohio.    [in this volume]

Johns, R.A., and  Roberts, P.V.  1991.  A solute transport model for
     channelized flow  in  a fracture.   Water Resources Research.
     27(8):1797-1808.

Karl,  J.H.   1989.   An Introduction to Digital  Signal Processing.
     Academic Press, San Diego,  Calif.  341 p.

Kerfoot, W.B., Beaulieu, G., and  Kieley, L.  1991.  Direct-reading
     borehole  flowmeter  results  in  field applications.   Ground
     Water Management.  (Proceedings of the Fifth National Outdoor
     Action  Conference on Aquifer Restoration,  Ground Water Moni-
     toring, and Geophysical Methods)  5:1073-1084.

Kiraly,  L.  1975.  Rapport sur 1'etat actuel des connaissances dans
     le  domaine des caracteres physiques des roches karstiques, in.
     Burger, A.,  and Dubertret, L.,  eds. Hydrology of Karstic Ter-
     rains.   International Association of Hydrogeologists, Paris.
     International Union of Geological  Sciences, Ser. B, no.  3.  p.
     53-67.

Kiraly,  L.,  and Muller, I.  1979.  Heterogeneite de la permeabilite
     et  de I1 alimentation dans le karst: effect  sur la variation du
     chimisme des  sources  karstiques.  Universite  de Neuchatel,
     Centre  d'Hydrologie,  Bulletin.  3:237-285.
                                621

-------
Krumbein, W.C.  1937.  Sediments and exponential curves.  Journal
     of Geology.  45:577-601.

Krumbein, W.C., and Pettijohn, F.J.  1938.  Manual of Sedimentary
     Petrography.  Appleton-Century-Crofts, New York.  549 p.

Kruseman, G.P., and de Ridder, N.A.  1990.   Analysis and Evaluation
     of Pumping Test Data.   2nd  ed.   International Institute for
     Land  Reclamation  and  Improvement  (ILRI),  Publication 47.
     ILRI, Wageningen,  The Netherlands.  377 p.

Lallemand-Barres, A., and Roux, J.C.  1989.  Guide methodologique
     d1establishment des perimeters des captages d'eau souterraine
     destinee  a  la consommation  humaine.   Bureau  de Recherches
     Geologiques et Minieres  (BRGM), Manuals et Methodes, no. 19.
     Editions du BRGM,  Orleans.  221 p.

Lepiller, M. ,  and Mondain, P.H.  1986.  Les  tracages articifiels en
     hydrogeologie  karstique:  Mise  en  oevre  et  interpretation.
     Hydrogeologie.  2(1):32-52.

Lundgren, L.  1986.   Environmental  Geology-  Prentice-Hall, Engle-
     wood Cliffs, N.J.   576 p.

Mackay, D.M.  1990.  Characterization of the distribution and be-
     havior of contaminants  in the subsurface,  in Water Sciences
     and  Technology  Board.   Ground Water  and  Soil Contamination
     Remediation:  Toward Compatible Science,  Policy  and Public
     Perception.  National Academy Press, Washington,  B.C.  p. 70-
     90.

Mackey, C.O.  1936.  Graphical Solutions.  Wiley, New York.  153 p.

McCann, M.R.,  and Krothe, N.C. 1991.  Development of a monitoring
     program at a Superfund site  in a karst terrane near Blooming-
     ton, Indiana.   Hydrogeology,  Ecology,  Monitoring, and Manage-
     ment  of   Ground  Water  in  Karst  Terranes Conference   (3rd,
     Nashville, Tenn.), Proceedings.  National Ground Water Associ-
     ation, Dublin, Ohio.  [in this volume]

Maloszewski,  P., and Zuber,  A.   1990.  Mathematical modeling of
     tracer behavior in short term experiments in fissured rocks.
     Water Resources Research.  26(7):1517-1528.

Mangin, A.  1982.   Determination du comportment hydrodynamique des
     aquiferes  karstiques a  partir  de 1'etude des informations
     fournies par leurs exutoires,  in Les discontinus en hydrogeo-
     logie. Bureau de Recherches Geologiques et Minieres, Orleans.
     Documents du BRGM, no. 45.  p. 397-403.

Mangin, A.   1984.   Ecoulement en  milieu  karstique.   Annales des
     Mines.   191(5-6):135-142.

Matheron, G.   1965.  Les variables regionalisees  et leur estima-


                                622

-------
     tion.   Masson, Paris.  305 p.

Mazor,  E.  1991.  Applied Chemical and Isotopic Groundwater Hydrol-
     ogy.   Wiley, New York.  274 p.

Mazor,  E.,  Jaffe, F.C., and Vuataz, F.D.  1990.  Time-data series
     as a  major  diagnostic tool in  groundwater hydrology,  in
     Parriaux,  A., ed.   Water Resources  in  Mountainous Regions.
     International Association of Hydrogeologists,  22nd Congress
     (Lausanne, Switzerland), Proceedings.  Memoir, 22 (1):273-280.

Meade,  M.L., and Dillon, C.R.   1991.   Signals and Systems: Models
     and Behavior.  2nd ed.  Chapman & Hall,  London.   167 p.

Meiman, J.   1991.  The effects of recharge basin land-use practices
     on water  quality at  Mammoth Cave National  Park,  Kentucky.
     Hydrogeology, Ecology, Monitoring, and  Management  of Ground
     Water in Karst Terranes Conference (3rd,  Nashville,  Tenn.),
     Proceedings.  National Ground Water Association, Dublin, Ohio.
     [in this volume]

Meiman, J. ,  Ewers,  R.O., and Quinlan,  J.F.   1988.  Investigation of
     flood-pulse movement through a maturely karstified aquifer at
     Mammoth Cave National Park:   A  new approach.   Conference on
     Environmental Problems  in  Karst  Terranes  (2nd,  Nashville,
     Tenn.), Proceedings.  National Water Well Association, Dublin,
     Ohio.   p.  227-263.

Milanovic,  P.T.   1981.  Karst Hydrogeology-  Water Resources Publi-
     cations, Littleton, Colo.   434 p.

Miller, R.L., Bradford,  W.L.,  and Peters, N.E.   1988.   Specific
     conductance:  theoretical  considerations and  application to
     analytical quality  control.   U.S. Geological Survey,  Water
     Supply Paper 2311.  16 p.

Moltyaner,  G.L.  1990a.  Field studies of dispersion: Radioactive
     tracer experiments  on scales of 20,  40, and  260  meters, in
     Cushman, J.H., ed.  Dynamics of  Fluids in Hierarchical Porous
     Media.   Academic, London,   p. 7-36.

Moltyaner,  G.L.,  ed.  1990b. Transport  and Mass Exchange Processes
     in Sand  and Gravel  Aquifers:  Field  and  Modelling Studies.
     (Proceedings of  the  workshop held in Ottawa, Canada)  AECL-
     10308.   Atomic Energy of Canada Limited [AECL],  Chalk River,
     Ottawa.  2 v-  402 + 464  p.

Molz, F.J.,  Morin,  R.H. ,  Hess, A.E.,  Melville, J.G. , and Giiven, O.
     1989.   The impeller meter for measuring aquifer permeability
     variations:  Evaluation and comparison with other  tests.  Water
     Resources Research.  25(7):1677-1683.

Montgomery,  C.  1992.  Environmental Geology.   3rd ed.  W.C. Brown,
     Dubuque, Iowa.  465 p.
                                623

-------
Moore, G.K.   1988a.   Occurrence,  frequency,  and significance  of
     cavities in  fractured-rock aquifers near Oak Ridge National
     Laboratory, Tennessee.  Conference  on Environmental Problems
     in  Karst  Terranes  (2nd,   Nashville,  Tenn.),  Proceedings.
     National Water Well Association, Dublin, Ohio.  p. 171-182.

Moore, G.K.   1988b.   Uses of probability graphs for the  charac-
     terization of fractured rock aquifers near  Oak Ridge,  Tennes-
     see.  Hydrological Science  and Technology.  4(1-4):57-65.

Murray, J.P., Rouse,  J.V-,  and Carpenter, A.B.   1981.  Groundwater
     contamination by sanitary landfill  and domestic wastewater  in
     carbonate terrain:  principal source diagnosis, chemical trans-
     port  characteristics,  and  design  implications.    Water Re-
     search.  15:745-757.

Newson, M.D.  1971.  A model of  subterranean limestone  erosion  in
     the British  Isles  based  on hydrology.   Institute  of  British
     Geographers, Transactions.   54:51-70.

Nielsen, D.M., ed. 1991.  Practical Handbook of  Ground-Water Moni-
     toring.  Lewis,  Chelsea, Mich.  717 p.

Nonveiller,  E.   1989.   Grouting Theory  and  Practice.   Elsevier,
     Amsterdam.  250 p.

Nyquist, J.E.,  Moore,  G.K.,  Young, S.C., and Clapp,  R.B.   1991.
     Use  of  electromagnetic  borehole  flowmeter  to  delineate
     groundwater  producing fractures.   Tennessee Water Resources
     Symposium  (4th,  Knoxville), Extended Abstracts,  p. 26-28.

O'Brien  and Gere Engineers,  Inc.    1988.   Hazardous  Waste Site
     Remediation.  Van Nostrand Reinhold, New York.  422 p.

O'Leary, P.R., Canter,  L.,  and Robinson,  W.D.  1986.  Land  dispos-
     al, in Robinson, W.D., ed.  The Solid Waste  Handbook:   A Prac-
     tical Guide.  Wiley-Interscience, New York.  p. 259-376.

Onder, H.  1986.   Interaction between conduit type flow and  diffuse
     flow in karst formations, in Giiltekin, G. ,  and Johnson, A.I.,
     eds.   Karst Water Resources.   International  Association  of
     Hydrological Sciences, Publication no. 161 (Proceedings of the
     Ankara-Antalya Symposium,  1985).  p. 371-385.

Palmer, A.N.   1990.  Groundwater processes in karst terranes,  in
     Higgins, C.G., and Gates, D.R., eds.  Groundwater Geomorpholo-
     gy-   Geological  Society of America, Special  Paper 252.    p.
     177-209.

Palmer,  A.N.   1991.    Origin and  morphology  of limestone caves.
     Geological Society of American, Bulletin.   103(1):1-21.

Papadopulos, I.S.  1965.  Nonsteady flow to a well in an infinite
     anisotropic  aquifer.    Dubrovnik  Symposium on  Hydrology  of
                                624

-------
     Fractured Rocks, Proceedings.   International  Association of
     Scientific Hydrology-   1:21-31.

Patrick,  R.,  Ford,  E., and Quarles, J.  1987.  Groundwater Contam-
     ination in the United States.  2nd ed.  University of Pennsyl-
     vania,  Philadelphia.   513 p.

Pedler,  W.H., Barvenik,  M.J., Tsang,  C.F.,  and  Hale,  F.V.   1990.
     Determination of bedrock hydraulic conductivity and hydrochem-
     istry using a wellbore  fluid logging method.   Ground Water
     Management.   (Proceedings  of the  Fourth  National  Outdoor
     Action Conference on Aquifer Restoration,  Ground Water Moni-
     toring,  and Geophysical  Methods)  2:39-53.

Pfeffer,  J.T.   1992.  Solid Waste  Management Engineering.   Pren-
     tice-Hall, Englewood Cliffs, N.J.  307 p.

Phillips, O.J.   1991.   Flow and  Reactions in Permeable  Rocks.
     Cambridge University Press,  Cambridge.  285 p.

Pierce,   J.R.    1980.   An  Introduction to Information  Theory:
     Symbols, Signals and Noise.  2nd ed.  Dover, New York.  305 p.

Plummer,  L.N.,  Busby, J.F.,  Lee, R.W., and Hanshaw,  B.B.   1990.
     Geochemical modeling of  the Madison aquifer in parts of Mon-
     tana, Wyoming, and South Dakota.  Water Resources Research.
     26(9):1981-2014.

Puig, J.M.   1990.    Le  systeme  karstique  de  la Fountaine  de la
     Vaucluse.   Bureau  de  Recherches Geologiques  et  Minieres
     (BRGM),  Document no.  180.  208 p.  [doctoral  thesis]

Quinlan,  J.F.  1978.   Types of karst, with  emphasis on cover beds
     in their classification  and development.   Ph.D.  dissertation
     (Geology), The University of Texas at  Austin.   325 p.

Quinlan,  J.F.   1986.   Recommended  procedure  for responding to
     spills of hazardous materials  in  karst terranes.  Environmen-
     tal Problems in  Karst Terranes and Their Solutions Conference
     (1st, Bowling Green,  Ky.),  Proceedings. National Water Well
     Association,  Dublin,  Ohio.   p. 183-196.

Quinlan,  J.F.   1989.  Ground-Water Monitoring  in  Karst Terranes:
     Recommended Protocol and  Implicit Assumptions.  U.S. Environ-
     mental Protection Agency,  Environmental Monitoring Systems
     Laboratory, Las Vegas, Nev.   EPA/600/X-89/050  88 p. [DRAFT;
     final version to be published 1992]

Quinlan,  J.F.   1990a.  Special problems of ground-water monitoring
     in karst terranes,  in Nielsen, D.M.,  and Johnson, A.I., eds.
     Ground Water and Vadose  Zone Monitoring.   ASTM Special Tech-
     nical Publication  1053.   American  Society for  Testing and
     Materials  (ASTM), Philadelphia,   p.  275-304.
                               625

-------
Quinlan, J.F.  1990b.  Types of karst, in Quinlan,  J.F.,  ed., Prac-
     tical  Karst  Hydrogeology,  with  Emphasis  on  Ground Water
     Monitoring.   [Course manual]  5th  ed.   Association of Ground
     Water Scientists and Engineers, Dublin, Ohio.   9:H-1 to H-9.

Quinlan, J.F.  1992.   Interpretation of dye-concentration and dye-
     recovery curves from four simultaneous interwell traces in a
     karst aquifer.   Privately  published Christmas card.   [Nash-
     ville, Tenn.]  1 p.

Quinlan, J.F.,  and Alexander, E.G.,  Jr.   1987.   How often should
     samples be taken  at relevant locations for reliable monitoring
     of pollutants from an agricultural, waste disposal, or spill
     site. Multidisciplinary Conference on Sinkholes and the Envi-
     ronmental Impacts of Karst  (2nd, Orlando, Fla.), Proceedings.
     Balkema, Rotterdam, The Netherlands.  p. 277-286.

Quinlan, J.F., and Alexander,  E.G., Jr.  1990.  Practical Tracing
     of Groundwater, with Emphasis on  Karst Terranes.   Quinlan &
     Associates, Nashville,  Tenn.  2 v.  146 + 91 p.

Quinlan, J.F., and Aley, T.   1987.  Discussion of "A new approach
     to the  disposal  of solid waste on  land", by R.C.  Heath and
     J.H. Lehr.  Ground Water.  25(5):615-616.

Quinlan, J.F., and Ewers,  R.O.   1981.  Hydrogeology  of the Mammoth
     Cave Region, Kentucky,  in Roberts, T.G., ed.   GSA Cincinnati
     '81  Field  Trip Guidebooks.   American Geological Institute,
     Washington, D.C.   3:457-506.

Quinlan, J.F., and Ewers, R.O.  1985.  Ground water flow in lime-
     stone terranes:  Strategy  rationale and procedure for reli-
     able, efficient monitoring  of ground-water  quality in karst
     areas.  National  Symposium  and Exposition on Aquifer Restora-
     tion  and  Ground Water  Monitoring  (5th,  Columbus,  Ohio),
     Proceedings.  National Water Well Association, Worthington,
     Ohio.  p.  197-234.

Quinlan,  J.F.,  Ewers,  R.O. ,  and Field,  M.S.   1988.   How to use
     ground-water  tracing  to  "prove"  that leakage  from  a karst
     terrane will not  occur.  Conference  on Environmental Problems
     in  Karst  Terranes  (2nd,   Nashville,  Tenn.),  Proceedings.
     National Water Well Association, Dublin, Ohio.  p. 289-301

Quinlan,  J.F.,  and Ray, J.A.    1989.  Groundwater  Basins in the
     Mammoth  Cave Region, Kentucky.  Friends of the Karst, Occa-
     sional Publication No. 2.  Friends of the Karst, Mammoth Cave,
     Ky.   [map]

Quinlan, J.F., and Ray,  J.A.  1991.  Groundwater remediation may be
     achievable in some karst aquifers that are contaminated, but
     it ranges  from unlikely to  impossible in most:  I. Implica-
     tions of long-term tracer tests for universal failure  in goal
     attainment  by   scientists,   consultants,   and  regulators.
                                626

-------
     Hydrogeology,  Ecology, Monitoring, and  Management of Ground
     Water in Karst Terranes Conference  (3rd,  Nashville,  Tenn.)/
     Proceedings.  National Ground Water Association, Dublin, Ohio.
     [in this volume]

Quinlan,  J.F., and Rowe,  D.R.   1977.   Hydrology and Water Quality
     in the  Central Kentucky  Karst:  Phase  I.   Water Resources
     Research Institute,  University of Kentucky-  Research Report
     101.   93 p.

Robinson,  E.A.  1980.   Physical Application of Stationary Time-
     Series,  with  Special Reference to Digital  Data Processing of
     Seismic Signals.   Charles  Griffin, London.  302 p.

Robinson,  J.L.,  and Hutchinson, C.B.   1991.  Ground-water tracer
     tests in west-central Florida [abs.].   American Institute of
     Hydrology, Annual  Meeting  (Orlando, Fla.), Abstracts,  p. 39.

Roscoe Moss Company.  1990.  Handbook of Ground Water Development.
     Wiley, New York.   493 p.

Rubin, P.A.  1991.   Land use planning and watershed protection in
     karst  terranes.    Hydrogeology,  Ecology,  Monitoring,  and
     Management of  Ground  Water in Karst Terranes Conference (3rd,
     Nashville,  Tenn.),   Proceedings.     National   Ground  Water
     Association,  Dublin,  Ohio.  [in this volume]

Samama, J.C.   1986.  Ore  Fields and  Continental Weathering.   Van
     Nostrand Reinhold, New York.   326 p.

Scanlon,  B.R., and  Thrailkill,  J.   1987.   Chemical similarities
     among physically  distinct  spring  types in  a  karst terrain.
     Journal of Hydrology.  89:259-279.

Shuster,  E.T., and White, W.B.  1971.  Seasonal fluctuations in the
     chemistry of  limestone springs:  A possible means for charac-
     terizing carbonate aguifers.  Journal of Hydrology. 14:93-128.

Smart, P.L.,  Edwards,  A.J., and Hobbs,  S.L.   1991.   Heterogeneity
     in carbonate aguifers:  Effect of scale, fissuration, litholo-
     gy,  and  karstification.   Hydrogeology,  Ecology,  Monitoring,
     and Management of Ground Water  in Karst Terranes Conference
     (3rd, Nashville,  Tenn.),  Proceedings.   National Ground Water
     Association,  Dublin,  Ohio.  [in this volume]

Smart, P.L.,  and  Friedrich, H.    1986.   Water movement  in the
     unsaturated zone of a maturely karstified  carbonate aquifer,
     Mendip  Hills,  England.    Environmental  Problems in  Karst
     Terranes and  Their Solutions  Conference (1st,  Bowling Green,
     Ky.), Proceedings.  National  Water Well Association,  Dublin,
     Ohio.  p. 59-87.

Smart, P.L.,  and Hobbs, S.L.  1986.  Characteristics of carbonate
     aguifers: A conceptual base.  Environmental Problems in Karst
                               627

-------
     Terranes and Their Solutions Conference  (1st, Bowling Green,
     Ky.), Proceedings.  National Water Well Association, Dublin,
     Ohio.  p. 1-14.

Sun,  R.J.,  and  Weeks, J.B.    1991.    Bibliography  of Regional
     Aquifer-System Analysis Program of the U.S.  Geological Survey,
     1978-91.  U.S. Geological Survey, Water-Resources  Investiga-
     tions Report 91-4122.  92 p.

Thiery, D.,  Vandenbeusch,  M., and Vaubourg,  P-   1983.  Interpreta-
     tion des pompages d'essai en milieu fissure aquifere.  Docu-
     ments du BRGM, No. 57.  53 p.

Tsang, C.F-, Tsang, Y.W.,  and Hale, F.V-  1991.  Tracer transport
     in fractures:   Analysis of field data based  on a variable-
     aperture channel  model.   Water Resources Research.  27(12):
     3095-3106.

Tukey, J.W.   1962.   The future of data analysis.  Annals of Mathe-
     matical Statistics.  33:1-67.   [Reprinted in Jones, L.V., ed.
     1986.  The Collected  Works of John W. Tukey, Vol  3:  Philoso-
     phy and Principles of Data Analysis,  1949-1964. Wadsworth and
     Brooks/Cole, Monterey, Calif.  p. 391-484.]

U.S. EPA  [U.S.  Environmental Protection Agency].   1991.   40 CFR
     Parts 257  and 258, Solid Waste  Disposal Facility Criteria;
     Final Rule.   Federal Register, October  9.    56(196):50,977-
     51,119.

van der Leeden, F., Troise,  F.L.,  and  Todd,  O.K.  1990.  The Water
     Encyclopedia.   Lewis, Chelsea, Mich.  808 p.

Viessman, W., Jr.,  Lewis,  G.L., and Knapp, J.W.  1989.  Introduc-
     tion to Hydrology.  3rd ed.  Harper & Row, New York.  780 p.

Ward, R.C.,  Loftis, J.C., and McBride,  G.B.  1990.   Design of Water
     Quality Monitoring Systems.   Van Nostrand Reinhold, New York.
     231 p.

Water Science and  Technology Board.   1988.   Hazardous Waste Site
     Management: Water Quality Issues.    National  Academy Press,
     Washington, D.C.  212 p.

Water Science and Technology Board.  1990a.  Ground Water and Soil
     Contamination Remediation:  Toward Compatible  Science, Policy
     and Public  Perception.   National Academy Press, Washington,
     D.C.  261 p.

Water Science and Technology Board.  199Ob.  Ground Water Models:
     Scientific  and Regulatory  Applications.    National Academy
     Press,  Washington, D.C.  303 p.

Watson, K.W., and Luxmore, R.J.  1986.  Estimating macroporosity in
     a forest watershed by  use of a tension infiltrometer.  Soil
                               628

-------
     Science Society Journal.  50:578-582.

Weeks,  E.P.  1969.  Determining the ratio of horizontal to vertical
     permeability  by  aquifer-test  analysis.    Water  Resources
     Research.   5(1):196-214.

Wells,  E.R., and Krothe,  N.C.  1989.   Seasonal fluctuation in d15N
     of groundwater  nitrate in  a mantled karst aquifer  due to
     macropore transport of fertilizer-derived nitrate.    Journal
     of Hydrology.   112:191-201.

White,  W.B.  1977.  Conceptual models  of  karst aquifers: Revisited,
     in Dilamarter,  R.R.,  and Csallany,  S.C.,  eds.   Hydrologic
     Problems  in  Karst Terrains.   Western Kentucky  University,
     Bowling Green.  p. 176-187.

White,  W.B.   1988.  Geomorphology  and Hydrology of Karst Terrains.
     Oxford University Press, New York.   464  p.

Williams,  P.W.   1983.  The role of the subcutaneous zone in karst
     hydrology.  Journal of Hydrology-  61:45-67.

Williams,  P.W.   1985.  Subcutaneous hydrology and the development
     of doline and  cockpit  karst.  Zeitschrift fur Geomorphologie.
     29(4):463-482.

Wilson, G.V-,  Jardine, P.M., Luxmoore, R.J.,  Todd,  D.E., Zelazny,
     L.W. ,  and  Lietzke,  D.A.   1991.  Hydrogeochemical  processes
     controlling subsurface transport from an upper subcatchment of
     Walker Branch watershed during  storm events.   1.  Hydrologic
     transport  processes,   and 2.  Solute  transport  processes.
     Journal of Hydrology-   123(3-4):297-336.

Wisconsin Geological and Natural  History Survey.   1991.   Delinea-
     tion of Wellhead  Protection  Areas  in Fractured Rocks.  U.S.
     Environmental Protection Agency, Office  of  Ground  Water and
     Drinking Water,  Washington,  D.C.  EPA 570/9-91-009.  144 p.

Wright, V.P.,  Esteban, M.,  and  Smart,  P.L.  1991.  Paleokarsts and
     Paleokarstic Reservoirs.   University of Reading, Postgraduate
     Research Institute for  Sedimentology, Contribution No. 152.
     158 p.

Wynne,  B.  1987.  Risk Management  and  Hazardous Waste:  Implementa-
     tion and  the  Dialectics of   Credibility-   Springer,  Berlin.
     447 p.
                            APPENDIX

  PROCEDURE FOR PREVENTION OF ALIASING AND FOR DETERMINATION OF
APPROPRIATE SAMPLING FREQUENCY FOR A SPRING OR CONDUIT-FED WELL

     The chemical character of a discharging spring can be consi-
                                629

-------
dered to be a signal, a hydrologic time series that can be recorded
(monitored) continuously or sampled regularly.  The main problem in
sampling, however, is how to choose the sampling interval,  At.  It
is obvious that sampling leads to some loss of information and that
this loss  gets worse as  At  increases.  Aliasing  occurs.   It is
expensive  to make At very  small,  so a compromise value  must be
sought.  Determination of the maximum frequency of the signal in a
spring chemograph, f^,  is quite  simple if one is dealing with a
sinusoidal function with a constant frequency, but hydrologic func-
tions are less regular  (less periodic or non-periodic)  and have a
highly variable frequency.   Accordingly,  we propose a method for
obtaining a practical approximation of  f^, a method based on the
common relationship in which specific conductance (conductivity; C)
varies inversely with discharge (Q).   Our  rationale for the method
will be similar to and complementary with that used for calculating
the response time, tR in the section of this paper concerned with
characterizing hydrological behavior as a measure of aquifer sensi-
tivity-  Although we  can compensate for the fact that  some springs
are characterized by  a short-lived increase in conductivity at the
onset  of  a storm event  (because  relatively mineralized water is
pushed out by piston-flow),  we  will ignore these short-lived peaks
in this description and utilize the well-known inverse relationship
between conductivity and discharge of springs and streams, a rela-
tionship caused by the relatively low conductivity of precipitation
and its dilution of groundwater.

     The appropriate sampling interval, free of  aliasing,  will be
determined by a technique we propose here.   This technique consists
of evaluation of a semi-log plot of conductivity vs. time, empiri-
cally determining the coefficients of the exponential equation des-
cribing the relationship between these data, and use of a coeffici-
ent in this equation to  calculate the aquifer  (spring) recovery
time, TR,  using a  procedure devised by  Hess and White  (1988).  The
sampling interval is  then calculated  from  this recovery time.  [We
recognize the irony and seeming redundancy of designing a sampling
plan to design a sampling plan.]

     It is more reliable, more  convenient, and  far  less expensive
and less labor-intensive to monitor the conductivity continuously
for the purpose of designing the sampling plan.  It would be wise
to do  so  through  several storms,  and for several months,  so that
the variation,  if any,  in spring behavior  can be detected,  so that
fine structure  (if present) in the chemograph can also  be detected,
and so that subsequent possible aliasing by discrete  sampling can
be prevented.  Accordingly, the following description  is based on
the use of data from continuous records of conductivity rather than
discrete samples for it.

     We classify  conductivity  chemographs into three  gradational
groups. A, B, and C.  For group A,  chemographs with a smooth reces-
sion limb, perform the following 7 steps:

     1. On a plot of  conductivity  vs.  time (extending  over  several
        months), sketch a smooth curve for background conductivity,
                                630

-------
   CB.   [The background  is likely to  show  a broad seasonal
   trend.]

2.  On the  chemograph for each storm event with a good record,
   pick the lowest point after the trough and on the recession
   limb where it begins to be linear.   This  is  Day 0  (t0) and
   C0 for each  storm event.

3.  For each of  the storm events selected for  step  #2, pick the
   post-C0  values  of C at 6 am and 6 pm,  record them, the cor-
   responding value for CB, the difference between them  (AC) ,
   and the values  for time, t.  Do so only for the near-linear
   part of the curve.

4.  For each data set, plot the following on semi-log paper: AC
   (on the y-axis,  the log scale)  vs.  t (on the x-axis, the
   arithmetic scale).

5.  Determine b,  the  conductivity coefficient, in the following
   exponential  equation in  which b   is  the  slope  of  its
   straight line on semi-log paper:

                         AC = C0e"bt

   A procedure  for easily solving this equation graphically is
   given by Krumbein (1937) and Krumbein and  Pettijohn (1938,
   p. 209-210).  Alternative graphical procedures are given by
   Mackey (1936, p.  113-115) and Davis (1955, p. 16-20).  [An-
   other alternative is to perform a computer-aided regression
   analysis.]

6.  The reciprocal  of b is the  recovery time,  TR,  for each
   storm event, as described  by Hess and White (1988, p. 250-
   251) . This  is  a  measure of how long  it takes the conducti-
   vity of a spring (aquifer) to rise to a value that is ap-
   proximately 63%  (= 1  - 1/e) of its pre-storm value.   The  TR
   for several  storms  can be averaged.  Alternatively, one
   might select the appropriate TR for  the  recession limb of
   a particular storm being monitored by comparing its nascent
   hydrograph or chemograph shape with one already studied.

7.  As  a  first  approximation,  we  recommend sampling  every
   0.05TR from the onset of a storm until 0.2TR after its con-
   ductivity minimum, Cmjn,  or stage maximum, O.^, then every
   0.1 TR for the  duration  of TR,  and followed by  0.2TR for as
   long as desired.   One might change to sampling every  0.4TR
   at 4TR after Cmjn  or Qmax.  These  sampling frequencies are a
   guide;   sampling  should generally be done  at reasonable
   hours of the day rather than at 3:21  a.m.  because that was
   the calculated value  —  unless,  of course, one  is trying to
   characterize a high-frequency  signal. Experience may show
   that use of  different coefficients of TR yield results more
   desirable.
                            631

-------
     For group B,  conductivity  chemographs with spikey or  finely
undulatory structure, select a sample interval, At, that  is twice
the highest frequency to be resolved.  For example, a signal with
10-minute  frequency peaks  should be  sampled  at  least  every 5
minutes.  TR  is  ignored.

     For group C,  conductivity chemographs with a coarsely undula-
tory structure in the recession  limb, one can  follow the procedure
for group B or sketch a smooth line that  curvilinearly straightens
them and then follow the  procedure for group A.  Use this  sketched
line to determine the values for t and C described in steps 2 and
3,  above.

     Although there  is usually  a  slight  phase-lag and an inverse
correlation between chemograph and hydrograph  structure, sometimes
there is no relationship  — as when,  for  example, there is a semi-
sinusoidal spring hydrograph from  a siphon  having  a  20-minute
period.  Sometimes also,  there may be a fine structure in  the con-
ductivity  chemograph that  is  not present  in  the  hydrograph or
thermograph (Meiman et al., 1988).

     The procedures  described above minimize  the  possibility of
aliasing of samples.  If followed, they yield an objective, reli-
able, non-aliased,  lowest  cost, legally  defensible sampling plan
that maximizes the probability of accurate characterization of an
aquifer with conductivity data and/or other water-quality parame-
ters.  We are excited at the prospect of testing these plans at a
wide range of spring types and conditions.
                      BIOGRAPHICAL SKETCHES

Dr. James F. Quinlan, P.G., is president of Quinlan & Associates,
Inc., a consulting firm specializing  in problems of carbonate ter-
ranes.  He was Research Geologist for the National Park Service at
Mammoth Cave, Kentucky,  for  16  years and  has been an independent
consultant on karst for more than 10 years.  He earned a Ph.D. in
geology at The University of Texas at Austin (1978).  His experi-
ence includes 36 years of research and observations in karst ter-
ranes of 26  states  and  25  countries,  and  work as a consultant in
many of them.  He has written or co-written more than 180 publica-
tions on karst-related topics.  He is co-chairman of ASTM Subcom-
mittee D18.21.09  on Special  Problems of Monitoring  in Karst and
Fractured Rock Terranes.

Dr. Peter L. Smart is a lecturer in geography at the University of
Bristol, England, from which  he  also  received his first degree and
doctorate.  He has extensive research experience in karst hydrology
and geomorphology in many different areas  of the world, including
the  Bahamas, China,  Indonesia,  Malaysia,  and  several  European
countries.

Geary M.  Schindel,  P.G.,  has been Project Manager and Senior Hydro-
geologist  for Eckenfelder Inc.  since 1990.   He  was previously
                                632

-------
Manager of the Environmental Division of ATEC Associates, Inc., in
Nashville.  Prior to joining ATEC,  he was Manager of the Ground-
water Branch of the Kentucky Division of Water.   He has also worked
as a  research assistant  at Mammoth  Cave National Park  and the
Center for Cave and Karst Studies at Western Kentucky University in
Bowling Green. Geary has  a  M.S. degree from Western Kentucky Uni-
versity and a B.S. degree from West Virginia University.

Dr. E. Calvin Alexander, Jr., is a Professor in the Department of
Geology and Geophysics  at the University  of Minnesota.   He has a
B.S.  in Chemistry (1966) from Oklahoma State University and a Ph.D.
in Chemistry (1970)  from the University of Missouri at Rolla.  The
central theme of his current research is  the rate of movement of
fluids in hydrogeology.  He  and his  research group are utilizing a
variety of methods to measure flow and residence times of water in
aquifers, which can range from hours to tens of thousands of years.

Alan J. Edwards is a research student in the Geography Department
of the University of Bristol, from which he also  obtained his first
degree.   He  is  currently  completing his  doctoral  dissertation
concerning  the hydrogeology of  quarries  and  landfills in  the
karstified limestone aquifer of the Mendip Hills.

A. Richard Smith  earned  a B.S. in Geology (1964)  from The Universi-
ty of Texas at Austin.   He has been involved in various hydrogeo-
logical activities for  two  major corporations  and  is now Senior
Geologist of the  Ground-Water Protection  Branch,  Bureau of Solid
Waste Management, Texas Department  of Health.   His experience in
karst geomorphology and hydrology includes  investigation of Texas
caves as  an associate of  the Texas  Speleological  Survey, ongoing
long-term study of gypsum  karst in West Texas, and exploration for
karst-related sulfur deposits in evaporite  rocks.
                            ADDRESSES
Dr. James F. Quinlan
Quinlan & Associates, Inc.
Box 110539
Nashville, TN  37222-0539
(615) 833-4324

Dr. Peter L. Smart
Department of Geography
University of Bristol
Bristol BS8 1SS, ENGLAND
44-272-303030

Dr. E. Calvin Alexander, Jr.
Department of Geology and
     Geophysics
University of Minnesota
Minneapolis, MN  55455-0219
(612) 624-3517
Geary M. Schindel
Eckenfelder Inc.
227 French Landing Drive
Nashville, TN 37228
(615) 255-2288

Alan J. Edwards
Department of Geography
University of Bristol
Bristol BS8 1SS, ENGLAND
44-272-303030

A. Richard Smith
Municipal Solid Waste Division
Texas Water Commission
P.O. Box 13087,  Capitol Station
Austin, TX  78711
(512) 834-6683
                                633

-------
Recommended Administrative/Regulatory Definition of Karst Aquifer,
Principles for Classification of Carbonate Aquifers,  and Practical
Evaluation of Vulnerability of Karst Aquifers

By:  James F-  Quinlan,  Peter L. Smart, Geary M.  Schindel, E. Calvin
     Alexander, Jr., Alan J. Edwards, and A. Richard Smith


Q.   Should the definition of karst aquifer include the modifier of
     rock type?  Isn't the term too broad without such a modifier?

A.   Yes and no.  More specifically,  because the great majority of
     karsts are developed in carbonate rock,  the word karst implies
     a limestone  or dolomite  karst.   If  one were referring to a
     karst developed in another rock type, one would refer to  it as
     a gypsum karst, a salt karst, a carbonatite karst, etc.


Q.   There is an inherent problem in lumping  all carbonate aquifers
     together as "karst11,  especially when the word karst is used in
     regulations.  By your broad definition, 98% of Missouri, for
     example, could be classified as a karst terrane —  in spite of
     the fact that 60% of the state is not significantly affected
     by karst.   Why not  restrict the word  karst  to areas where
     there are environmental problems and use another word for the
     other carbonate aquifer areas?

A.   Approximately  60%,  not  98%,  of Missouri has  limestone or
     dolomite cropping out at the surface, and approximately 80% of
     that 60% (50% of the  state) is karst, as we have defined  it in
     this paper.  Without trying to nitpick over semantics, a key
     phrase in  the  above  question is significantly affected.  We
     can  agree to  disagree  with the  accuracy  of the  unknown
     questioner's estimate of 60% of  the  erroneous  98% but, never-
     theless,  address  the principle  raised.    The definition of
     karst is vague enough without making it more  so, as it would
     be if the  questioner's suggestion were followed.   We do not
     believe in changing  traditional  definitions.   Those we have
     given herein are a codification, fine-tuning, and synthesis of
     traditional uses; they replace  definitions poorly worded by
     non-specialists in karst.  The new terms we have introduced,
     describing aquifer sensitivity, may achieve what the question-
     er suggested about using another word in lieu of karst. but it
     does so without changing  the meaning of karst.  Also, a key
     word in the question is are. It excluded might be.  We believe
     it is  better to err on  the side of conservatism.   Aquifer
     contamination, like death, can be so permanent.


Q.   Doesn't  the  work  of Scanlon and  Thrailkill  (1987)  in the
     Bluegrass  Karst region, near Lexington, Kentucky, contradict
     what you claim about the reliability of inferring the nature
     of an aquifer  from analysis of its spring  waters?
                                634

-------
At first glance,  the work of Scanlon  and Thrailkill (1987)
might seem  to invalidate the relations we allege between water
quality and  physical characteristics  of springs  and their
aquifers.   They claim that the two spring types studied, major
(large) and high-level (small),  could not be discriminated on
the basis  of water-quality parameters.  We agree.   But our
review  of  their  published  data and  their description  of
behavior of both spring types during storms shows unambiguous-
ly that each of  them is a conduit-flow  spring.   Their work
does not question or invalidate our conclusion  on the prac-
tical  utility  of  conductivity  measurements  for  assessing
aquifer types or aquifer vulnerability.
                           635

-------
636

-------
         LEGAL TOOLS FOR THE PROTECTION OF GROUND WATER
                        IN KARST TERRANES
                 GARY A. DAVIS, Attorney at Law
                Ray, Farmer, Eldridge, & Hickman
                      Knoxville, Tennessee

                        JAMES F. QUINLAN
                     Quinlan and Associates
                      Nashville, Tennessee
                            ABSTRACT

     Ground water in karst terranes is exceptionally vulnerable to
contamination due  to  human  activities such  as waste  disposal,
septic  tanks,   chemical  spills,   and  storage  of  gasoline  in
underground tanks.  Because of this vulnerability, legal tools for
the protection of  ground  water  are extremely  important  in karst
terranes.    These legal tools include source  controls,  land-use
controls,  which  may be utilized by state and local governments and
by several types of quasi-governmental entities.  Users of ground
water can  assert their water rights to protect ground water.

     Source controls include statutes and regulations governing the
licensing  of potential  sources  of  contamination and  design and
operating  standards to prevent release of  contaminants.  Land use
controls  include zoning  regulations   and  the  power of  eminent
domain.     These  controls   should   recognize   the   exceptional
vulnerability of ground  water  in karst terranes, but most statutes
and regulations  do not  provide for  any differential  treatment of
karst aquifers.   Unfortunately,  many  of those that address karst
only deal  with the potential for  surface collapse and not the rapid
migration  of  contaminants in solution channels.   Additionally,
wellhead and sensitive area protection programs in karst terranes
should not be based upon arbitrary circles around sources of water
supplies,  but upon well-designed studies delineating ground water
flow in solution channels.

     Finally, ground water users  in karst terrane can rely upon
principles  of water  law developed  for surface water  to protect
their water rights.   The  principle  of reasonable use  has been
applied to  ground water,  particularly ground water  flowing in
                                637

-------
discrete channels in karst terrane.
                           INTRODUCTION

     Ground water in karst terranes  is  exceptionally vulnerable to
contamination  due  to  human activities  such as  waste disposal,
septic  tanks,  chemical  spills,  and  storage  of  gasoline  in
underground tanks.   Water supplies dependent upon ground water in
karst  terranes  are  also  subject  to  disruption by  subsurface
disturbances  such  as  mining,   and  by  diversion of  flow  by  an
"upstream" user.

     Because of this exceptional vulnerability, legal tools  for the
protection  of  ground  water  are  extremely  important in karst
terranes.   These legal  tools  include  source  controls,  land-use
controls, and water rights.  They may be utilized  by federal, state
and local governments  and  by several types of quasi-governmental
entities.

     Source controls include statutes and regulations governing the
licensing of  potential sources  of  contamination and  design and
operating standards to  prevent release of contaminants.  These need
to be tailored  to recognize this special vulnerability of ground
water in karst terranes, but, unfortunately- many  of these statutes
and regulations only deal with the potential for surface collapse
and not the rapid migration of  contaminants in solution channels.

     Land use controls  include  zoning regulations and the power of
eminent domain.  In addition to  counties and municipalities, there
are other quasi-governmental entities, such as watershed districts,
that possess some of these powers. The  federal wellhead protection
provisions can  act in  the  manner of  land use controls  to prevent
the location of potential sources of ground water contamination.

     Finally, ground-water users in karst terranes may be able to
rely upon principles of water  law developed for  surface water to
protect their water rights.  The principle of reasonable use has
been applied to ground  water, particularly to ground water  flowing
in discrete channels in karst terrane.
                         SOURCE CONTROLS

Legal Framework

     The protection  of  ground water was not  a  specific focus of
concern until the 1970s, and the legal framework for ground-water
protection on the federal level and in most states is fragmented.
It  is  made up of various  federal  and state  laws,  some of which
touch only indirectly on ground water protection.  These controls
and exercises  of rights can  be  set in perspective  by  reading a
                               638

-------
recent critical review of public policy considerations that enter
into  the  assignment  of  a  regulatory  scheme  to  one level  of
government rather than another.

     Federal  laws  governing  potential  sources  of  ground-water
pollution as an  area  of major concern include the  Safe Drinking
Water Act  (underground injection control and wellhead protection);
the Resource Conservation  and Recovery Act  (solid  and hazardous
waste management and underground storage tanks); the Comprehensive
Environmental   Response,   Compensation,   and   Liability   Act
(investigation and  remediation of uncontrolled hazardous substance
sites); the  Toxic  Substance  Control  Act (disposal of  PCB's  and
asbestos); the Federal Insecticide, Fungicide, and Rodenticide Act
(pesticide use); the Uranium Mill Tailings Radiation Control Act;
the Surface  Mining Control and  Reclamation  Act,  and  the Atomic
Energy Act of 1954  (radioactive waste disposal).

     Of  these,  the  Safe  Drinking  Water   Act,   the  Resource
Conservation and Recovery Act, and the Surface Mining Control and
Reclamation Act allow  states to assume  primary management authority
to carry out the federal legislation, if  the  state  programs meet
federal guidelines and are  approved by the  federal  agency.  State
regulations under these laws can generally be more stringent than
federal regulations.   In addition to  federal  laws  and authorized
state  programs under  those  laws,  states  have  regulated  other
potential sources  of  ground-water degradation, including septic
tank systems, oil exploration, and water  withdrawal.


Specific Provisions for Karst Terranes

     Federal statutes and regulations governing potential sources
of ground  water degradation  have  generally not addressed  karst
terranes.    The only  instance  in  which  karst terranes  have been
addressed in federal source-control regulations is in the new EPA
municipal solid-waste  landfill regulations, promulgated under RCRA.
Under  these  regulations,  new and existing municipal  solid  waste
landfills in  "unstable  areas" must include a  demonstration that
"engineering measures have  been  incorporated  into  the [landfill]
design to ensure that  the integrity of  the structural components of
the [landfill] unit will not  be  disrupted."   An  "unstable area,"
under  the regulation  includes karst  terranes.3  Guidance  for the
investigation  of  site-specific  potential  for  subsidence  and
collapse is to be included in the technical guidance document the
agency plans to issue within six months.

     This regulation  deals only with the potential  for  surface
collapse  and  not  with  the  potential   for  rapid  migration  of
contaminants  through  solution channels  and  the  difficulty  of
detecting the release  of contaminants  in karst terranes.  Although
none  of  the  monitoring regulations  specifically  address  karst
terranes,  they are  sufficiently flexible to allow state agencies to
                                639

-------
require proper monitoring.

     Several state source-control statutes and regulations address
karst  terranes  specifically.    One  of the  most  comprehensive
examples is the Minnesota Solid Waste Disposal Regulations, which
deal with the difficulties of preventing and detecting movement of
contaminants in karst terranes as well as the potential for surface
collapse.

     The  Minnesota   Technical   Requirements  for   Solid  Waste
Facilities require the siting of a solid waste landfill only in an
area where  the ground water flow paths are known  in sufficient
detail to enable reliable tracking of pollutant movement and where
it  is  feasible to construct a  monitoring system with sufficient
monitoring points  to assure that pollutants can be detected and
tracked.4  This  general  siting  requirement  would make  siting in
karst  terranes  difficult without a dye-trace study to  delineate
ground-water flow paths and establish monitoring points.

     The regulations go on to state:

     A land disposal  facility must  not  be located on a site
     where: (1) there are karst features, such as sinkholes,
     solution channels, disappearing streams, and  caves, which
     may cause  failure of  the leachate  management system or
     prevent effective monitoring or containment  of a release
     of leachate;

     Kentucky   solid  waste  regulations  also   focus   on  the
difficulties  of characterizing  ground  water  flow  and  detecting
pollutant migration  in karst terranes.   In applying  for a solid
waste  landfill  permit in a  karst  environment, the applicant must
characterize both  diffuse and  discrete flow conditions,  and the
state  may  require dye trace studies  before  finalizing  a ground
water  monitoring  plan.6   In addition to the  general requirement
that a landfill may  only be sited in  an area  where the uppermost
aquifer  is  capable of being monitored  in  a manner that detects
migration of  pollutants  and where corrective action  is possible
once  pollutants  are  detected,  Kentucky  regulations  prohibit
placement of waste within 250 feet of  a  feature of karst  terrane.7
This  buffer-zone requirement,   however,  gives a false  sense of
security, since waste placed within 250 feel of a  karst feature
could  easily be still within the karst terrane.

     Tennessee  solid  waste  regulations  establishing  landfill
location requirements for  karst areas  address both potential for
surface  collapse  and migration  of contaminants  but fail to deal
with the difficulties of reliably detecting contaminant migration
in  karst terranes.   The  Tennessee regulation states:

     If a facility is proposed in an area of highly developed
     karst  terrane  (i.e.,  sink  holes, caves,  underground
                               640

-------
     conduit   flow  drainage,  and   solutionally  enlarged
     fractures)  the   applicant   must  demonstrate  to  the
     satisfaction  of  the  Commissioner [of  Environment  and
     Conservation]  that relative  to the  proposed facility
     siting:
     (i)  There  is  no  significant   potential  for  surface
          collapse;
     (ii) The ground water flow system is not a conduit flow
          which would  contribute  significant  potential  for
          surface  collapse or which  would cause significant
          degradation to the ground water; and
    (iii) Location  in  the  karst  terrane will not cause
          any  significant  degradation  to  the  local
          ground water resources.8

Tennessee's  ground-water monitoring  provisions  for solid  waste
landfills,  however,  do  not  address  the difficulty  of detecting
contaminant flow in karst terrane,  nor have dye-trace studies been
required by the state for the karst demonstration envisioned by the
regulation.   The regulation  also gives  far too much discretion to
the Commissioner to decide when  to require a karst demonstration
and whether a  conduit  flow system could cause degradation to the
ground water.

     Alabama  has   a  similar  presumption  against  the  siting  of
landfills in karst  terranes unless a site specific demonstration is
performed, but  the site-specific  demonstration  required  does not
address  the  particular difficulties  of determining  ground  water
flow paths and tracking contaminant migration in karst  terranes.10
Other states that  include specific standards  for location of solid
waste landfills in karst terranes  include Florida and Georgia, but
their requirements seem to be limited to preventing waste disposal
in or near sinkholes.
Local Government Source Control Programs

     General police power ordinances which employ such regulatory
mechanisms  as  permits, licenses,  inspections,  and  standards  to
prevent potential sources  of pollution have also  been  used by -local
governments to protect ground water.  Health codes adopted by local
governments, or  regulations imposed  by  local health departments,
have an advantage  over land-use  controls,  discussed below, since
they can be imposed retroactively to deal with existing pollution
threats as well as future  threats.  They may  also gain more public
support than zoning laws.

     In most  states,   unless  state  and federal  programs  preempt
local authority,  local source control  efforts  to  protect ground
water can proceed  by  any  of several  routes,  including county and
municipal   ordinances,  local  health   department  regulations,
watershed district regulations, and utility district regulations.
                               641

-------
For  example,  the Cape  Cod  Planning  and  Economic  Development
Commission designed  model groundwater  protection ordinances for
enactment at  the municipal level  and by  local  boards  of health
after  designation  of their  region  as  a  Sole  Source  Aquifer
protectable  under the Safe Drinking  Water Act.   One ordinance,
designed  for  adoption  by  municipalities,  regulates all  firms
handling  toxic or  hazardous  materials.   Another,  designed for
adoption  by   local   health  departments,   regulates  both  the
installation of new underground storage tanks and  the maintenance
of older ones.12

     Quasi-governmental entities, such as watershed districts and
utility districts, may also be able to regulate potential sources
of  ground water  contamination.   Many states  provide   for  the
creation  of  watershed districts that are given  broad  powers to
conserve soil and water resources within the  natural boundaries of
a watershed.13  Among the powers  delegated  to districts  by the
Tennessee legislature,  for example,   are  "to take such  steps as
deemed necessary by (the)  board  of  directors  for the promotion and
protection of public health within  the boundaries  of the district"
and "to do all things necessary and proper for the protection of
the watershed and the  lands and waters  within  the boundaries of the
district."14    Although most  watershed districts  are created to
deal with surface streams,  the regulatory powers  of  a watershed
district should be able to protect ground water in karst terranes
where there is a  clear connection between ground water and surface
streams.

     Many states  also provide for the  creation  of water utility
districts to provide  for water  supply.   Where a utility district
depends upon  ground  water  from wells  or  springs, the  power to
maintain  the  source  of  supply that  is  granted  by the  state
legislature may authorize regulations  for the control  of pollution
sources which potentially endanger a ground water.  For instance,
under Tennessee law utility districts are  empowered  to acquire,
construct, operate,  and  maintain  systems  for the furnishing of
water, and to exercise all  powers necessary for the accomplishment
of these purposes.15

     Local source-control  regulations  also need  to recognize the
particular difficulties  in ensuring  ground  water protection in
karst  terrane.    Where   the  federal  or  state   source  control
regulations  have  not  gone  far  enough  in  addressing  these
difficulties, local regulations may fill this gap.


                        LAND-USE CONTROLS

     Land-use controls are well-suited for  preventing  ground-water
contamination  in  karst   terranes.     Sensitive   area protection
programs  and  wellhead  protection  programs  are appropriately
implemented through zoning and other  land-use controls.
                                642

-------
     Most  local  governments  have  broad  zoning  powers  for  the
protection of the public health and welfare.  Some states, such as
Tennessee, have given local governments  explicit power to zone for
conservation of water  supply-    Zoning may  take  numerous forms,
including  prohibitions  of  land uses with  serious  potential  to
contaminate ground water, design or performance standards on land
uses, limitations on density or lot sizes,  and overlay districts
designating recharge or well-head protection areas.17

     Another  form  of   land-use   control   is   the  purchase  or
condemnation by local governments of restrictive easements across
land  in  critical  recharge areas.   Most  local governments  can
condemn  land  or rights  in land  for public purposes,  including
protection of water supplies.   A ground-water protection easement
allows the owner of property to retain the land and use it for any
purpose  that  does not  present a  threat  to ground  water,  while
requiring  less  compensation  by the local  government than  the
purchase or condemnation of the entire property.

     For  land-use  controls to be  effective  in karst  terranes,
careful mapping of ground water flow is necessary.   Reliance upon
arbitrary circular restrictive zones around water supply wells or
springs  is  unlikely to  protect water  supplies  in  karst,  since
contaminants may  flow long distances  in  solution  channels  with
little attenuation.

     Several communities have utilized land-use  controls to protect
ground water.18  An  example in  karst terrane is the comprehensive
watershed development  ordinances  adopted  by the City  of  Austin,
Texas, to  protect the  Edwards Aquifer.   Austin  has  delineated
special watershed zones for controlled development density and has
prohibited almost all  development  within  "critical  water quality
zones" in aquifer recharge areas.19

     Although Lexington, Kentucky,  and the surrounding county were
primarily concerned with prevention of flooding of  sinkholes  in
response to storms and blockage of sinkhole drains by sediment that
runs  off  in  response  to  erosion  triggered  by  construction
activities, their  ordinance also protects  groundwater  quality.
Similar ordinances have been enacted by Bowling Green,  Kentucky21,
and Springfield and Perryville, Missouri.

     Clinton  Township,   New   Jersey,   and  Macungie   Township,
Pennsylvania, both in the  Appalachians, in  a effort primarily to
control  sinkhole  development  but  also to  protect  ground  water
quality in the karst aquifers that underlay them, have each enacted
ordinances in 1988 that regulate but do not prohibit construction
in order  to  lessen the  probability of  sinkhole collapse.22   The
legality and validity of the New Jersey ordinance has been upheld
in the courts,  and its operation has been successful.23

     Local land use controls can be aided by  federal programs found
                                643

-------
in the Safe Drinking Water Act.  The Wellhead Protection Program,
included  in the  1986  amendments  to the  Act,  allows  states to
develop and submit to EPA programs that protect areas around public
water  supply  wells  (Wellhead  Protection Areas).  These Wellhead
Protection Programs often provide a mix of state and local source
controls with local land use controls.24
                      COMMON LAW PROTECTIONS

     Ground-water  users  can  rely upon  principles  of  water law
developed for  surface water to protect  their water  rights.   The
principle  of  reasonable  use,  where  applicable,  requires  each
property owner to make only reasonable use of the ground water so
as not  to  interfere with use of ground water  by  other property
owners.  An unreasonable use can be enjoined by the courts.

     Under English common law, water under the ground was generally
considered the exclusive property of  the  surface owner.  Thus, the
surface owner could make whatever use of the ground water he or she
wished, including  uses destructive  of  ground  water  quality and
quantity.     In the  United   States  state  courts  have  gradually
recognized  the  fact  that  ground  water  flows  and  that  one
landowner's  use of  ground  water can  affect   other  landowners.
American courts have adopted the reasonable use rule, which holds
that one user may only put ground water to reasonable use where it
may affect the use of ground water by another user.

     The first  step  in that recognition, and a  distinction which
still holds in some states, was to apply the reasonable use rule to
water that flowed underground  in a well-defined channel,  as opposed
to percolating ground water.  Since water flowing in a well-defined
channel is similar to surface water, the reasonable use rule, which
was developed  for surface waters, applied.  Such conduit flow, of
course, is an attribute of karst terranes.

     Most  states now  apply  the  reasonable use  rule  to  either
conduit flow situations or percolating  ground water where there is
proof that one property owner's use has unreasonably affected the
ground water use of another.  For instance, the  Tennessee Court of
Appeals found in 1935 that the rights of a property owner to enjoin
the unreasonable use of ground water that supplied a spring did not
depend  on proof  of  a  well-defined channel,   but  simply of  an
"intimate" ^connection  between  the  ground  water  on  the  two
properties.

     But in  states  where the well-defined channel  distinction is
still important, knowledge of the subsurface conduit flow feature
is critical  to the  protection  of the  ground water.   Texas, for
instance,  still  follows  the English rule  for  percolating ground
water,  and  there   is  a presumption that  all  ground water  is
percolating  ground  water.   In a  recent  case, the  Texas Court of
                               644

-------
Appeals  found  the  hydrogeological  testimony  insufficient  to
establish well-defined channel flow of a spring flowing through an
underground cavity before it surfaced.   This ruling permitted the
property  owner  to  utilize  all  of the  flow of  the spring  for
irrigation purposes.


                    DISCUSSION AND  CONCLUSIONS

     There are ample  legal  tools available  for  the protection of
ground water  by state  and  local  governments  and  by  individual
ground-water users.  Most statutes and regulations, however, ignore
karst aquifers because the framers  never thought about them or had
no knowledge of how they  differ  from other  aquifers.   When these
legal tools are used for karst terranes, they need to be sharpened
to address the particular features  of karst.

     There are three major problems  with many existing statutes and
regulations that specifically  address the regulation of activities
in karst terranes:

     1.   The definitions of  karst are inadequate  or  ambiguous,
          particularly when they focus solely on  the presence of
          sinkholes.

     2.   The definition  of karst, even when adequate,  is often
          misinterpreted to  apply only to areas  with sinkholes.

     3.   There is too much emphasis on the surface of  the karst
          and too little on  the aquifer itself.  Regulatory buffer
          zones of  a few hundred  feet radius around sinkholes, for
          instance, ignore  the fact that,   in most  terranes,  the
          greatest  danger  to   ground  water   is  migration  of
          contaminants  below  the  surface,  not  collapse  of  the
          surface.

     Source-control   regulations   addressing   karst   terranes,
including siting restrictions for potentially polluting activities,
should focus on more than just the  potential for surface collapse
in  karst  terranes.    Because  of   the  extreme  difficulties  in
preventing and detecting ground water contamination in karst, there
should be a presumption against the  location  of certain activities,
such as  hazardous  and solid waste  landfills, in  karst  terranes.
And  before that presumption  can  be  overcome,  there  must  be  a
demonstration that  contaminants would not enter water supplies and
that reliable monitoring  can  be  achieved as a  result of detailed
knowledge of ground-water flow paths.

     Land-use controls must  also recognize the unpredictable nature
of ground-water flow in karst terranes.  Arbitrarily drawn recharge
zones  for  wells  and springs  may not  only  be unfair  to property
owners,  but  may  miss  conduit-flow pathways  entirely.   Again,
                                645

-------
detailed knowledge of ground-water flow is necessary.

     Finally,  water  rights  in  ground water  may be  asserted by
individual ground-water users.  The extent of such rights, however,
may actually depend  on  the  ability  of the user to prove that the
ground water is flowing as a discreet underground stream.


                         REFERENCES  CITED

1.   See  Denise  D.   Fort,  Federalism  and  the  Prevention  of
     Groundwater Contamination,  Water Resources  Research, Vol. 27,
     No. 12, pp. 2811-2817  (December 1991).   See also Timothy R.
     Henderson,  The  Institutional  Framework  for  Protecting
     Groundwater in  the United  States, in G.  William Page, ed.,
     Planning for Groundwater Protection. Academic Press, Orlando,
     Florida, pp. 29-67.

2.   Safe Drinking Water Act, 42 United States Code Sections 300f-
     j; Resource Conservation and Recovery  Act, 42  United States
     Code  Sections   6901,  et seq.;   Comprehensive  Environmental
     Response  Compensation,  and  Liability Act,  42  United States
     Code Sections 9601, et  seq.; Toxic Substances Control Act, 15
     United  States   Code   Sections   2601,   et  seq.;   Federal
     Insecticide, Fungicide  and Rodenticide Act,  7  United States
     Code Sections 136, et  seq.; Uranium  Mill  Tailings Radiation
     Control Act, 42 United States  Code Sections 7901,  et seq.;
     Surface Mining Control and Reclamation Act, 30 United States
     Code Sections 1201,  et seq.;  Atomic Energy Act  of  1954,  42
     United States Code Sections  2014, 2021,  2021a,  2022, 2111,
     2113, 2114.

3.   56 Federal Register 51019-20 (Oct. 9, 1991). "Karst terranes"
     are  defined as "areas  where   karst topography,  with  its
     characteristic surface and subterranean features,  is developed
     as the result of dissolution of  limestone,  dolomite, or other
     soluble rock. Characteristic physiographic features present in
     karst terranes  include, but  are not limited  to, sinkholes,
     sinking streams, caves,  large  springs, and  blind valleys."
     This definition is repeated on  page  51047, but there "karst
     terranes"  are  twice   mistakenly  referred  to  as  "karst
     terraces".

4.   Minnesota Rules, Section 7035.2815,  Subpart 2A.  (1988).

5.   Minnesota Rules, Section 7035.2815, Subpart 2C.  (1988). Under
     Minnesota Rules, Section 7035.0300, Subpart 51,  "karst" "means
     a type of  topography that is formed  from  the dissolution of
     limestone, dolomite, or gypsum  and that  is characterized by
     closed depressions or  sinkholes,  and underground  drainage
     through conduits enlarged by dissolution."
                               646

-------
6.    Kentucky Administrative Regulations,  401 KAR .. Under 401 KAR
     30:010(104)  "karst terrain"  "means a type of topography where
     limestone, dolomite or gypsum is present and is characterized
     by naturally  occurring  closed topographic  depressions  or
     sinkholes,   caves,  disrupted  surface  drainage,   and  well
     developed underground solution channels formed by dissolution
     of these rocks by water moving underground."

7.    Kentucky Administrative Regulations, 401 KAR 48:050, Sections
     6 and 1, respectively-

8.    Tennessee Administrative Code, Rule 1200-1-7-.04(2)(q). Under
     Rule   1200-l-7-.01(2)   "karst"  "means  a  specific  type  of
     topography   that   is  formed  by dissolving  or  solution  of
     carbonate formations,  such  as  limestone or dolomite;  it  is
     characterized by closed  depressions or  sinkholes,  caves,
     sinking and reappearing streams, and/or underground conduit
     drainage flow."

9.    Rule  1200-l-7-.04(2) (q) goes  on to  state  that "the  above-
     referenced   demonstration  may  require the  installation  of
     piezometers, the  developing  of  potentiometric-surface map  of
     ground water,  conducting geophysical  surveys,  dye  tracing  or
     other   specific   requirements   deemed  necessary  by   the
     Commissioner  to  evaluate   the   proposed   site  to   his
     satisfaction." In our  experience  with solid waste  landfill
     siting in Tennessee,  the focus of the Division of Solid Waste
     has  been more on the  detection  of  potential  for  surface
     collapse through  geophysical studies,  and no dye trace studies
     have   been   required,  despite  clear   indications  of  highly
     developed karst.   Those traces that have been performed in an
     attempt to demonstrate the threat to karst aquifers have been
     ignored or  misinterpreted  by the Division.

10.  Alabama Administrative Code,  Sections 335-13-4-. 01 (6 ), 335-13-
     4-.11 through 335-13-4-.14.  (as  amended through  October  2,
     1990).

11.  Florida Administrative Code,  Section 17-701.040(2)  (as amended
     through July 19,  1990);  Rules and Regulations of the State of
     Georgia, Section  391-3-4-.05 (g)5 . (as  amended through June 29-
     1989) .

12.  Scott W. Horsley,  Beyond  Zoning: Municipal  Ordinances  to
     Protect  Groundwater,   Cape   Code  Planning  and  Economic
     Development  Commission, Barnstable, Mass.  (1982).

13.  For example, Tennessee  Code Annotated Sections  69-7-101,  et
     seq.  provide for  the creation of watershed  districts.

14.  Tennessee Code Annotated Section 69-7-118.
                               647

-------
15.   Tennessee Code Annotated Sections 7-82-302, 7-82-306.

16.   Tennessee  counties  may  be  pass  zoning ordinances  for the
     conservation of water supply. Tennessee Code Annotated Section
     13-7-101.

17.   See Douglas Yanggen and Stephen Born, Protecting Groundwater
     Quality by Managing Local Land Use,  Journal of Soil and Water
     Conservation, Vol.  45,  No. 2,  pp.  207-210 (March-April 1990);
     F. DiNovo and M. Jaffe, Local Groundwater Protection—Midwest
     Region, American Planning Association, Chicago, II. (1984).

18.   See  DiNovo   and  Jaffe,  supra;   See  also,   Committee  on
     Groundwater   Quality    Protection,    Groundwater   Quality
     Protection—State  and Local   Strategies,  National  Academy
     Press, Washington,  D.C. (1986).

19.   See Douglas Yanggen and Stephen Born, Protecting Groundwater
     Quality by Managing Local Land Use,  Journal of Soil and Water
     Conservation, Vol.  45, No. 2,  p. 207-210 (March-April 1990);
     Kent  S.   Butler,  Urban  Growth Management and  Groundwater
     Protection: Austin, Texas, in G. William Page, ed., Planning
     for Groundwater Protection,  Academic Press, Orlando, Fla., p.
     261-287;   H.D.  Smith,  Erosion  and  Sedimentation  Control
     Methodologies  for  Construction Activities over  The  Edwards
     Aquifer  in  Central  Texas.    Hydrogeology,   Ecology,  and
     Management of Ground Water in Karst Terranes Conference, (3rd,
     Nashville,  Tenn.),  Proceedings.     National  Ground  Water
     Association, Dublin, Ohio (1991).  [In this volume].

20.   See James S.  Dinger and  James R.  Rebmann,  Ordinance  for the
     Control  of  Urban   Development in   Sinkhole  Areas  in  the
     Bluegrass Region, Lexington, Kentucky, Environmental Problems
     in Karst  Terranes and their Solutions Conference (1st, Bowling
     Green, Ky.),  Proceedings.  National  Water  Well  Association,
     Dublin, Ohio, pp.  163-180.

21.   See Nicholas  C. Crawford, Sinkhole  Flooding  Associated with
     Urban  Development  Upon  Karst  Terrain:    Bowling  Green,
     Kentucky-  Multidisciplinary Conference  on Sinkholde  (1st,
     Orlando,  Fla.), Proceedings, pp. 283-292 (1984).

22.   See Joseph  A.  Fisher  and Hermia  Lechna, A Karst Ordinance,
     Clinton Township, New  Jersey,  Engineering  and Environmental
     Impacts  of  Sinkholes  and  Karst   (St.  Petersburg,  Fla.)
     Proceedings, pp.  357-361  (1989); Percy H. Dougherty, Land-Use
     Regulations  in the Lehigh  Valley:    Zoning  and  Subdivision
     Ordinances  in an  Environmentally  Sensitive  Karst  Region,
     Ibid., pp.  341-348.
                               648

-------
23.   See Joseph A.  Fisher and Robert J.  Canace,  Karst Geology and
     Ground-Water   Protection   Law,    Hydroqeoloqy,    Ecology/
     Monitoring,  and Management of Karst Terranes Conference (3rd,
     Nashville/  Tenn.),   Proceedings,   National  Ground   Water
     Association, Dublin, Ohio.  (1991)  [in this  volume].

24.   42 United States  Code Section 300h-7.

25.   Nashville,  C.  &  St.  L.  Ry.   v.   Rickert,  89  S.W.2d  889
     (Tenn.Ct.App.  1935), cert,  denied  (Tenn.  1936).

26.   Denis v. Kickapoo Land Co.,  771  S.W.2d 235 (Tex.Ct.App.  1989).

27.   See  James  F.  Quinlan,  P.L.  Smart,  G.M.  Schindel,  E.G.
     Alexander, Jr., A.J.  Edwards,  and  A.R. Smith.   Recommended
     Administrative/Regulatory  Definitions  of  Karst   Aquifer,
     Principles  for Classification  of   Carbonate  Aguifers,  and
     Practical Evaluation  of Vulnerability of  Karst Aguifers.
     Hydrogeology,  Ecology, Monitoring,  and Management of  Ground
     Water in Karst Terranes Conference  (3rd, Nashville, Tenn.),
     Proceedings. National Ground Water Association,  Dublin, Ohio
     (1991).  [in this volume]. This paper proposed the following
     definitions:

          karst  -  a landscape  and  subsurface  formed as  a
          result  of  dissolution  of bedrock (usually,  but  not
          necessarily,    limestone   and/or   dolomite)   and
          characterized by distinctive  subsurface  hydrology
          that includes  flow of water  through  caves.   The
          landscape   may  include sinkholes,  springs,   and
          sinking  streams,  but  these   features  are   not
          essential  to   the  definition   and should  not  be
          confused with the karst itself.

          karst aquifer - an aquifer in which flow of water
          is  or can  be appreciable through one or more of  the
          following:   joints,  faults,   bedding  planes  and
          cavities,  any or all of which have been enlarged by
          dissolution of bedrock.

     Each of these  definitions  is suitable for  incorporation in
     statutes,  regulations, and revisions thereof.   We recommend
     their use.
                                649

-------
                      BIOGRAPHICAL SKETCHES
     Gary  A.  Davis  is  an  environmental  attorney who  has been
practicing throughout the United States for more than  eight  years.
He  is  also  the  Director  of  the  Clean  Products  and  Clean
Technologies  Program  at  the  University  of   Tennessee  Energy,
Environment, and resources Center.  He was previously a hazardous
waste policy advisor in the California Governor's Office and  worked
as an  environmental  engineer with a major consulting firm.  Mr.
Davis received a B.S.  in Chemical Engineering from the University
of Cincinnati, and a J.D. from the University of Tennessee College
of Law.

Gary A. Davis
Attorney at Law
Two Centre Square
625 S. Gay Street
Suite 230
Knoxville, TN  37902
(615)637-7725


     Dr.  James  F.   Quinlan,  P.G.,  is  president of Quinlan  &
Associates, a consulting  firm specializing in problems  of carbonate
terranes.  He was Research Geologist for  the  National Park Service
at Mammoth Cave,  Kentucky, for 16 years and has been an independent
consultant on karst  for more than 10 years.  He  earned a Ph.D. in
geology  at  The   University  of  Texas  at  Austin  (1978).    His
experience includes 36 years  of research  and  observations in karst
terranes of 26 states and 25 countries, and work  as a consultant in
many  of  them.    He  has written  or  co-written  more  than  170
publications on  karst-related  topics.   He is co-chairman of ASTM
Subcommittee D18.21.09 on Special Problems of Monitoring in Karst
and Fractured Rock Terranes.

Dr. James F. Quinlan
Quinlan & Associates,  Inc.
Box 110539
Nashville, TN  37222-0539
(615)833-4324
                                650

-------
       Session IX:
Ground Water Management

-------
652

-------
         KARST GEOLOGY AND GROUND WATER PROTECTION LAW

 By: Joseph A. Fischer, Geoscience Services, and
     Robert J. Canace and
     Donald H. Monteverde, NJ Geological Survey, Dept. of
               Environmental Protection & Energy
ABSTRACT

     Clinton Township, Hunterdon County, New Jersey is situat-
ed partially  in a  Paleozoic outlier  within the  New  Jersey
Highlands Physiographic  Province.  Some  15%  to 20%  of  the
Township is underlain  by solution-prone,  folded and  faulted
carbonate rocks of Cambrian  and Ordovician age.  Faulting  of
possible Taconic, Alleghanian, and  Triassic age has  deformed
these carbonates.  The Flemington Fault, part of a major  bor-
der fault system, traverses the Township.

     Clinton is located at the intersection of one of New Jer-
sey's principal  east-west  interstate highways  and  a  major
north-south State highway.   Its rural  environment and  ready
access to major highways places Clinton in a growth  corridor.
The State  Planning  Commission has  recently  identified  the
Township as a regional rural growth center.

     Two state reservoirs extend onto Township lands, one  un-
derlain by carbonate bedrock.  The construction of the  Spruce
Run Reservoir provided a great amount of geologic  information
on the local carbonate deposits, including the nature and  ex-
tent of folding,  faulting, and solution-channel  development.
Extensive mapping and exploratory drilling permitted the  com-
pilation of  the  occurrence  of voids  in  specific  geologic
units, and  a better  definition of  the relationship  between
solutioning and faults.

     Enlightened Township  officials introduced  an  ordinance
governing construction in  areas underlain  by carbonate  rock
and those areas draining into carbonate lands.  The  Ordinance
has been in effect since the spring of 1988.  Although  origi-
nally met with great protestation by development interests, it
has proven to be a rational, working process that is presently
accepted by both the public and private sector.
                             653

-------
     The ordinance requires a multi-phased data collection ef-
fort at various stages of a project.  The legality and validi-
ty of this Ordinance have been tested and upheld in the  court
room.
INTRODUCTION

     New Jersey is  well known  for its number  of Super  Fund
sites and its rather  stringent State regulations in  relation
to discharges into ground water  and surface water bodies,  as
well as through injection wells.  However, the only State reg-
ulations that give consideration to  the problems of siting  a
facility in karst areas are provisos not to place septic  sys-
tems in sinkholes, and to establish trend directions of  solu-
tion channels in  carbonate rocks and  sinkholes for  landfill
sites.

     There are five counties in the northwestern corner of the
state that are underlain by major deposits of solutioned  car-
bonates (Figure 1).  However, there are no county  regulations
that address karst  related concerns and  the municipal  offi-
cials of these counties have little guidance in either  recog-
nizing or dealing with karst-related concerns.  In addition,
the potential  for ground  water pollution  from  agricultural
activities has also been neglected.

     At one time, these valley lands were primarily either ag-
ricultural in nature or  merely bucolic scenes.  However,  re-
cent population growth in these areas is rapidly changing  the
use of this  prime real estate.   Ground water extracted  from
the high yield  carbonate aquifers  is the  primary source  of
domestic and  industrial water  in this  region.   Reservoirs,
water treatment  facilities, shopping  centers, office  build-
ings, and corporate centers have been built above a variety of
carbonate terrane,  sometimes supporting  the structures  atop
large (now, usually cement-filled)  cavities.  Some have  been
well-investigated and engineered while others have not.

     The one municipality that  has recognized these  problems
is Clinton Township, a State targeted growth area in Hunterdon
County, New Jersey.  Clinton Township is located at the inter-
section of a principal east-west interstate highway and a  ma-
jor north-south State  highway, less  than 50  miles from  New
York City.  Some 10 years ago it was a rural, primarily  agri-
cultural community.  At the completion of the interstate high-
way, the area became the target of both State and private  de-
velopment planning.   The Township's  complex geology  (Figure
2), includes  solutioned carbonates  under  some 20%  of  its'
lands. Currently, wells are the primary source of domestic and
industrial water supply for  the Township.  Development  pres-
sures, led  some perceptive  municipal officials  to pass  New
Jersey's first "limestone" Ordinance to protect the ground wa-
ter supply in this environmentally sensitive setting.
                             654

-------
T4eIO'
                                                       T4°00'
                           (SCALE IN MILES)

       KNOWN CARBONATE ROCK DEPOSIT
       AREA OF SUSPECTED
       CARBONATE ROCK DEPOSIT
Figure 1.  Carbonate rock deposits of  northern New Jersey.

     The mayor, an articulate,   aware,  and strong  individual,
led the struggle to  secure a rational  approach to the  Town-
ship's development processes.  The New Jersey Geological  Sur-
vey encouraged the ordinance  and provided technical  guidance
and a simplified geologic map identifying the general location
of carbonate bedrock.  Previous  State  funded geologic informa-
tion provided a rational data  base  for the Ordinance and  its
implementation.  Problem recognition,  conceptual planning, and
Ordinance formulation  and  implementation  was  a  relatively
swift process.

     In brief, the Township, faced with increasing development
pressures in 1987, placed a 150  day  moratorium on  development
and formed a committee of lay and technical people to draft an
ordinance that would:
                              655

-------
              PCu
      Wllloughby Brook
        Fault
      LEGEND
JTru - Mesozoic rocks, undivided
Ojtu - Jutland sequence, undivided
Oj - Jacksonburg Limestone
Ow - Wantage sequence
Kirtatinny Supergroup
Obu - Beekmantown Group, upper part
Obi - Beekmantown Group, lower part
OCa - Allentown Dolomite
Cl - Leithsville Formation
Ch - Hardyston Quartzite

PCu - Proterozoic, undivided
I. !,:?;'.• j- carbonate rock
  --thrust fault
  -- normal fault
                                                              QCa
           MILES
Figure  2.   Geology  of Spruce Run Reservoir and adjacent areas
            of Clinton Township,  Hunterdon  County, New Jersey.


      1.  Protect ground water in  the carbonate rock  areas.

      2.  Address public safety  in relation  to structural integ-
         rity -

      3.  Allow conscientious development.

      The Ordinance  was  passed by the   Town Council  (although
not  unanimously)  in May, 1988.   Since that time, some 30 proj-
ects have been  reviewed by the   Township's geotechnical  con-
sultant (GTC) under the provisions of the  Ordinance.   The  re-
sults of the review process  have included the abandonment   of
projects in the exploration phase  (even prior to the  comple-
tion of the field explorations), the completion  of  the project
as envisioned by   the designers (with   little to no  changes),
and  the  completion  of projects  that   incorporated  numerous
changes from the  initial design, some during construction.

      The Ordinance has been tested in court by one  of the  de-
velopers affected by  the  moratorium.   In  a broad-based  suit
against the Township, the  Ordinance was subject  to  legal scru-
                                656

-------
tiny, was upheld, and found to be a most valuable document for
achieving the Township's desire to insure that development  is
conducted intelligently.

     The Township was  fortunate in  that geologic  investiga-
tions both before  and during  the construction  of two  large
State-funded reservoirs  provided an  extensive geologic  data
base as well  as fueling public  awareness of karst  concerns.
It is difficult to visualize the Ordinance coming into  being,
and its successful implementation, without the State  supplied
knowledge gained through the reservoir studies and the ongoing
interest in carbonate geology of the State Survey.
TOWNSHIP GEOLOGY

     The carbonate  bedrock in  Clinton Township  consists  of
rocks of the Kittatinny Supergroup (Drake and Lyttle, 1980) of
Cambrian through Middle Ordovician age, the Middle  Ordovician
Jacksonsburg Limestone (Kummel, 1940), and the Lower to Middle
Ordovician Jutland sequence (Jutland Member of the Martinsburg
Shale of Markewicz,  1967).  The geology  of the Clinton  area
was mapped by Markewicz (1967), and is currently being revised
by Markewicz, Monteverde,  and Volkert  of the New Jersey  Geo-
logical Survey.  The geologic map provided to Clinton Township
by the New Jersey Geological Survey, which has been updated in
Figure 2, is a simplified version of the current geologic  in-
terpretation of the area.    Data derived as  a result of  geo-
technical investigations carried out  under the ordinance  has
resulted in some modifications to the initial mapping.

     As indicated  above,   geologic findings  associated  with
mapping and subsurface exploration for the proposed Spruce Run
Reservoir (Markewicz 1958-1961) provided insight into the  de-
gree of karstification of the carbonates in Clinton.  Geologic
mapping, field identification of sinkholes, seismic data,  and
numerous test borings revealed the  presence of voids both  at
the surface and  in the subsurface,  indicating the  potential
for future subsidence.  The reservoir was constructed after  a
program to grout any voids at dam locations and the  installa-
tion of a peripheral grout  curtain to prevent leakage at  the
reservoir's southeastern margin.  Since construction,  several
sinkholes have opened adjacent to the reservoir, requiring re-
mediation.

     The degree  of karstification  of the  various  carbonate
bedrock units found in Clinton, as elsewhere, is a function of
lithology and proximity  to faults, folds  and other  geologic
structures.  While most carbonates have a potential for karst-
ification, regional studies suggest  that geologic units  that
contain coarse-grained dolomite or fine-grained limestone  are
most susceptible to weathering (Rauch and White, 1970,  Siddi-
qui and Parizek, 1969, and  Dalton and Markewicz, 1972).   So-
lution-channel development  in the  Kittatinny carbonates  ap-
                             657

-------
pears to  be particularly  accentuated in  units that  contain
thin-bedded, coarse-grained dolomite.

     Table 1 presents a  summary of exploratory drilling  data
for the Spruce Run Reservoir project.  Although the number  of
drill holes installed and  total penetration varied among  the
geologic units present, the data tends to confirm that specif-
ic units may be more prone to cavity development than  others.
Figure 3, based on this drill data, presents a visual summary
Drill Data
          Geologic Unit
         Leithsville
    Allentown       Lower        Upper
                 Beekmantown  Beekmantown
Limeport   Upper
 Member  Allentown
           Member
Number 9 32 33 31
of
Borings
Total 1,035 4,574 2,770 3,852
Footage
Drilled
Number 13 108 19 80
of
Cavities
Average 80 54(*) 48
Footage
Between
Cavities
66
6,428
90
71
Total          67
Footage
of void
Penetrated

Voids as       6.5
Percentage
of Total
Depth
Penetrated

Voids as       9.4
a Percentage
of Bedrock
Penetrated
   319
66
225
   7.0
2.4
5.8
   8.3
2.7
6.6
263
4.1
4.6
Table 1.  Occurrence of voids in borings for Spruce Run Reser-
          voir (Dalton and Markewicz, 1972, Table 4).
          * - Indicates value for formation as a whole.
                             658

-------
    Upper Allentown Member
       Ground	
       Surface  \
Limeport  Member
              KEY


              Subsurface  Boring

              Overburden

              Void in  Dolomite Bedrock

         /   Member  Contact
     20  0 20 40
    vertical
    scale
feet
Figure 3.  Subsurface profile of  a portion of the  Spruce  Run
Reservoir dam  site,  showing  the contrast   in   intensity  of
weathering between the Limeport and Upper Allentown Members of
the Allentown Formation (Markewicz, 1958-1961).

of the contrast in the  occurrence of voids between the  lower
(Limeport) member and  the upper  (Upper  Allentown) member  of
the Allentown Dolomite.

     Another example of the presence of concentrated solution-
ing in the Spruce Run  area is illustrated by borings  drilled
into the  lower Beekmantown  Group of  carbonates in  which  6
borings were  installed  within 45  feet  of  one  another,  to
depths ranging from  72 to 142  feet.  These  borings  revealed
voids in the bedrock  occupying between 14  and  52  percent  of
the total footage drilled for any one of the  borings.  In  ad-
                              659

-------
dition, 100 percent of the voids in this interval occur within
the first 45  feet of  rock penetrated,  signifying that  many
near-surface voids are  present that could  contribute to  the
formation of sinkholes at the surface.

     Faults involving the carbonate units consist of imbricate
thrusts, tear faults transverse  to the structural grain,  and
normal faults most likely associated with both rifting in  the
adjacent Newark rift  basin and with  earlier Appalachian  de-
formation.  The reservoir borings showed that cavities are of-
ten well developed in the  vicinity of faults (Figure 4)  that
display brittle deformation.  Fault locations were interpreted
from mapping of the area and rock core displaying  fracturing,
recrystallization and other evidence of deformation.

     An unusual occurrence of limestone is found in the  Clin-
ton area, in rocks of the Jutland sequence (Monteverde, unpub-
lished).  A test well drilled for Clinton, into this unit,  in
proximity to a major east-west fault, yielded nearly 400  gal-
lons of  water per  minute,  attesting to  the  well-developed
permeability of  this limestone.   Additionally, test  borings
performed in the Jutland limestone  for a proposed and  subse-
quently constructed golf course/residential community revealed
the persistent  presence of  small voids  and weathered  zones
throughout the limestone section.
NATURE OF THE ORDINANCE

     There are a  number of ordinances  that have been  passed
for locales underlain  by the solutioned  carbonates found  in
the Appalachian valleys of  the eastern United States.   These
include ordinances which essentially require the  preservation
of present  surface water  flows into  existing sinkholes,  or
those that  establish  performance standards  in  relation  to
known or  suspected features.   From the  authors' reviews  of
various eastern  U.S. "limestone"  ordinances, many  different
tacks have been taken.  In the valleys of the southern Appala-
chians, where development pressures are less than in the  more
northerly valleys, it seems that the primary direction of  all
ordinances has been toward preserving sinkhole integrity as  a
means of handling storm water flows without deleterious  local
flooding (e.g., Clarke County, Virginia).  The secondary  con-
sideration seems to  be to  stop structures  from being  built
directly over existing sinkholes.

     In the northeast, some Townships have apparently tried to
essentially eliminate growth in carbonate areas.  For example,
the Bucks County, Pennsylvania "Model Ordinance" requires that
storm water  management basins  and "principal  or  accessory"
buildings will not  be located  within 100  feet of  features,
which  include;  sinkholes,  closed  depressions,  lineaments,
fracture  traces,  caverns,  ghost  lakes,  and   disappearing
streams.
                             660

-------
26CH


200-
 180 H
 160 H

 120 H
       Ground Surface
         Top of Rock
                                                          Ground Surface
                                                                    o-f Rock
      Scale = 20 Vertical
           20 Horizontal
      OS] - Overburden
      IE-Cowity (Both filled and open)
      I  | - LJme»tone
Figure 4.   Schematic  interpretation based on  drill  data  of the
             occurrence of interconnected voids in the Allentown
             Dolomite in relation to the  trace  of a transverse
             fault.
                                  661

-------
     It is obvious what  an imaginative aerial photograph  in-
terpreter could do to any project in a municipality which  has
passed an ordinance in accordance with the Bucks County recom-
mendations.  In addition, the  authors have little  understand
ing of what the magic is in the 100 foot dimension.

     The route chosen by  the Clinton Ordinance Committee  was
completely different and is still believed to be unique.   The
Ordinance was intended to help educate and provide guidance to
a Planning  Board applicant  concerning the  performance of  a
phased investigation and evaluation  of the subsurface  condi-
tions at the development site.  The investigation and  evalua-
tion process was derived from the concepts discussed in Fisch-
er, et al (1987), and were developed from experience gained in
numerous site  investigations and  facilities construction  in
karst.

     A phased approach to  the Ordinance requirements was  in-
stituted to allow  an applicant to  define site-related  flaws
and/or prepare preliminary cost estimates at an early stage of
the development without incurring significant economic  penal-
ties.  Judgement is used in defining the extent and nature  of
an investigation.  Different concepts can  be used for a  pro-
posed two-house subdivision  than might be  employed for a  59
acre, multi-use development.

     The Ordinance  encourages  the  applicant's  geotechnical
consultant to utilize any  combination of investigative  tools
and techniques that  are deemed reasonable,  but with  certain
minimum requirements such as; reviewing the available geologic
information, using aerial photography, and providing hard data
(i.e., borings, test pits, etc.) on the site subsurface.   The
GTC's experience within the Township was made available to the
applicant on an  as-requested basis.  The  results of the  ex-
plorations are submitted to the  GTC for review and  approval.
Thus, a reasonable degree  of consistency in the  explorations
performed by the various applicants has developed and any use-
ful information produced at  one site can  be fed into  future
planning processes.  The process also provides for the  appli-
cant to perform only the amount of work necessary for the  ac-
tual site conditions, i.e., the more complex the site is,  the
more detailed the exploration and evaluation is.

     The subsurface data obtained is interpreted by the appli-
cant's geotechnical consultant  and used  to develop  planning
and design solutions to  the geotechnical problems that  exist
at the site.  Again, the GTC reviews and evaluates the subsur-
face model proposed  for the site  and suggested solutions  to
any subsurface-created concerns.  The applicant is  encouraged
to use innovative  engineering solutions and,  in general,  no
set procedures are imposed upon the planning/design team.  Fi-
nal reviews and discussions are held with the Planning  Board,
the applicant's design team, and the GTC prior to giving  con-
ceptual approval to, or conversely, rejection of, the  planned
                             662

-------
development.

     As a result of the difficulty in performing a  subsurface
investigation and  reliably  interpreting the  results  for  a
karst site in  any reasonably economic  manner, the  Ordinance
provides for field inspection during construction in key areas
by the  applicant's geotechnical  consultant as  well as  spot
checks by Township forces.  In enforcing the Ordinance, it  is
expected that changes  in design  and construction  procedures
can and will be made during the building phase.

     We believe that  the investigation,  evaluation, and  re-
porting costs required as a result of the existence of the Or-
dinance are reasonably consistent with what a prudent develop-
er would ordinarily budget.  This is not surprising as the Or-
dinance procedures are predicated upon concepts developed from
commercial, not research projects.   The separate cost of  the
GTC review are prepaid into an escrow account.  The costs were
estimated at $1,000 plus an additional $500/acre.  The GTC re-
view costs have usually been less than these initial  require-
ments and only  in one instance  did the somewhat  insensitive
actions of an  applicant require replenishment  of the  escrow
fund.
EXAMPLES

     Several examples of the different manners used to  attack
the issues can be used to  illustrate some of the workings  of
the Clinton Township "Limestone" Ordinance.

Case I - The site was  first investigated by a large  national
consulting firm during the building moratorium.  Initially, no
geotechnical study was envisioned.  The firm (with the assist-
ance of an outside geologic consultant) provided a "Geotechni-
cal Feasibility Study" after the development complex plans for
the 46 acre site had  been fully formulated.  The problems  of
force-fitting the investigative results to an existing  layout
became apparent.  The existence of several sinkholes were min-
imized and none (including  one elongated elliptical  sinkhole
some 25 feet in length) were considered "major" by the  devel-
oper's consultant.  A  test pit  excavated for  the field  in-
vestigation became a sinkhole.   Faults crossing the site  and
their likely  effect on  solutioning were  ignored.  Only  the
more favorable portions of the  work performed for the  nearby
reservoir were included.  The most critical structures  (serv-
ice buildings and  office towers) were  located atop what  was
likely the most solutioned rocks at the site, the  Leithsville
Formation.  An unlined, ornamental lake was sited over a  line
of small sinkholes.  The underlying dolomites were impermeable
in one section of the report but provided a good ground  water
resource in another.  Rock Quality Designation (RQD) data  was
misapplied.  The  significance of  drilling water,  core,  and
sample losses were not understood.
                             663

-------
     These and other factors eventually resulted in an  agree-
ment to reconfigure the site and perform additional studies in
conformance with the Ordinance.  No work has been accomplished
at this site  since the agreement  was made in  1990.  As  the
business climate has been  poor since that  time, there is  no
way of telling whether the owner's eventual realization of the
site subsurface deficiencies  or the  business climate  forced
the current abandonment of development plans.

Case II - Some 10 years ago a large international organization
constructed a major research facility  upon a small area of  a
roughly 100 acre parcel.  A significant portion of the proper-
ty is underlain by  solutioned carbonates, but apparently  the
karst-related problems were not  well recognized prior to,  or
even during, the site investigation.  Two large national firms
provided geotechnical  consulting services.  Cost overruns  in
the millions of dollars were experienced in both investigation
and foundation construction.  Untold quantities of cement were
pumped into unexpected cavities  and a complicated  foundation
system was installed.  A spray irrigation field for waste  wa-
ter disposal was placed  above shallow, solutioned  limestone.
Sinkholes have developed in this field.

     The remainder of the site is currently planned for multi-
use development  by single  and multi-family  homes,  offices,
stores, open-space and recreational facilities.  A  two-phased
field investigation  has recently  been completed  by  another
well-known engineering firm.  The work was planned and  accom-
plished in complete cooperation  with the Township GTC.   Per-
sonnel experienced in carbonate rock studies were used both in
the field portion of the  work and in the report  preparation.
The owners design team is  currently using the results of  the
phased explorations  investigations,  which cover  site  areas
that are underlain by  solutioned carbonates, competent  meta-
morphics, and apparently fractured  and faulted sediments,  to
prepare a conceptual layout.  The information provided to  the
design team has blended the experience of both the  consulting
firm and the Township GTC.

Case III - A caveat  emptor situation developed in a  proposed
single-family development that was investigated under the  re-
quirements of  the  Ordinance.  The  applicant's  geotechnical
consultant and the GTC came to an agreement on layout and  de-
sign of structures, wells, and septic systems that incorporat-
ed the results of the  applicant's studies.  The revised  con-
ceptual plans were  approved by the  Planning Board, with  the
standard caveat of the need to provide knowledgeable construc-
tion inspection.

     The site was sold to a new developer.  The approved plans
were provided to the buyer but apparently without the  several
geotechnical reports  completed  by  the  original  applicant.
Site work was initiated, then abandoned by the new owner after
the first foundation wall failed as the result of the  opening
                             664

-------
of a post-construction sinkhole beneath it.
CONCLUSION

     The many  fears in  relation to  ground water  protection
that both the lay and technical person should have, or in many
instances do have, can only be exacerbated by the presence  of
solutioned carbonate rocks  below the  site development.   The
technical information needed  to address  concerns related  to
karst is available to the  interested parties in any  develop-
ment exercise.  However, the awareness of the need for techni-
cal expertise is certainly not wide-spread.  Some  communities
have undertaken to express their concerns with ordinances.

     Drill data, the mapping of caves, and field  observations
concerning the distribution of sinkholes, springs, and  disap-
pearing streams  all indicate  that certain  units within  the
Kittatinny Supergroup display a higher degree of karst  devel-
opment than other carbonate rocks.  Some units are particular-
ly prone to solutioning.  This concentration of karst features
in specific geologic  units was confirmed  by the borings  in-
stalled for  the Spruce  Run  Reservoir in  Clinton  Township.
Therefore, in planning and  carrying out site   investigations
the geotechnical expert should keep in mind the likelihood and
potential size of karst features as a function of the geologic
unit(s) found on the site.   An understanding of the  probable
severity of karstification at a development site can help dir-
ect the intensity of the  exploratory program necessary for  a
suitable site investigation.

     Despite the correlations  that probably can  be made  be-
tween specific geologic units and structures and the potential
for karst features, a decision was made to treat all carbonate
bedrock in the township equally when enforcing the  ordinance.
This decision was made because of the ultimately unpredictable
nature of subsurface voids and for the sake of equal treatment
under the law.  This approach is valid in view of the need  to
investigate for voids on a  site-specific basis, where an  un-
derstanding of the location of structures relative to voids is
vital.

     The authors believe a  Clinton Township type  "limestone"
Ordinance which draws upon the available geologic  information
to establish guidelines,  yet allows development  to occur  in
karst areas  is an  appropriate  solution in  most  instances.
Laws should not be aimed at eliminating development or putting
the developer into a strait jacket, but should incorporate ap-
propriate provisions for insuring the protection of ground wa-
ter resources and the safety of the public while allowing  for
ingenuity in design and construction.

     There are,  undoubtedly, economic  penalties that  result
from the need  to obtain  adequate geologic  information at  a
                             665

-------
site and to provide  appropriate engineering solutions.   How-
ever, these economic  penalties are relatively  small in  con-
trast to the penalties that can occur as a result of  failures
of both the infrastructure and any structures located atop so-
lutioned carbonates, during or after construction when likely,
but unforeseen problems surface.
REFERENCES

Bucks County, Pennsylvania, undated, Carbonate Geologic  Areas
Model Ordinance Amendment.

Clarke County, Virginia, undated,  Clarke County Ground  Water
Protection Plan.

Dalton, Richard and F.J. Markewicz, 1972, Stratigraphy of  and
Characteristics of Cavern Development  in the Carbonate  Rocks
of New Jersey, Bui. of the Nat. Speleological Soc., v. 34, no.
4, p. 115-128.

Drake, Avery  A. and  P.E.  Lyttle, 1980,  Alleghanian  Thrust
Faults in the Kittatinny Valley, New Jersey, Field Studies  of
NJ Geology and Guide to  Field Trips,  52nd Annual Meeting  of
the NY  State Geological  Assoc., Warren  Manspeizer, ed.,  p.
92-112.

Fischer, Joseph A.,  R.W. Greene, R.S.  Ottoson, T.C.  Graham,
1987, Planning  and Design  Considerations in  Karst  Terrain,
Env. Geol. Water Sci., v- 12, no. 2, 123-128.

Kummel, H.B., 1940,  The Geology  of New Jersey,  NJ Dept.  of
Conservation and  Economic Development,  NJ Geological  Survey
Bui. No. 50, 203 p.

Markewicz, Frank J. ,  1958-1961, Field Notes,  Haps, and  Con-
struction Profiles of the Spruce Run Reservoir Site, NJ Bureau
of Geology and Topography, unpub.

Rauch, Henry W. and W.B.  White, 1970, Lithologic Controls  on
the Development of  Solution Porosity  in Carbonate  Aquifers,
NJ Water Resources Research, v. 6, no. 4, p. 1175-1192.

Siddiqui, Shamsul  H. and  R.R. Parizek,  1969,  Hydrogeologic
Factors Influencing Well Yields in Folded and Faulted  Carbon-
ate Rocks, Central  Pennsylvania, (abs.) American  Geophysical
Union Trans., v. 50, no. 4, p. 154.

	,  1967, Geology of  the High Bridge Quadrangle,  NJ
Bureau of Geology and Topography, Bull. 69, 139 p., unpub.
                             666

-------
        KARST GEOLOGY AND GROUND WATER PROTECTION LAW

          By: Joseph A. Fischer, Robert J. Canace,
              and Donald H. Monteverde

Question

1.   What cities,  counties,   and states have ordinances  near
or similar to the one you describe?

None.  We know of  no State ordinances  of this nature.   We
also unaware of any  Counties or Cities  with such an  ordi-
nance, although our knowledge on an interstate, local  level
is obviously limited.  However, a Pennsylvania legislature's
staff member informed Clinton Township that the Commonwealth
was attempting to prepare a law and that the Township's  or-
dinance appeared to be the best available based on what they
had found.

2.  To what extent are other governmental units deciding  to
emulate the  ordinance you  describe and  adopt a  "locally-
suited" similar ordinance?

There are currently three to four New Jersey  municipalities
considering a similar ordinance.  A model ordinance is being
prepared by a  "limestone committee" of  the New Jersey  Re-
source Conservation and  Development organization  (NJRC&D),
the northern New  Jersey  branch of  a nation-wide,  county/
USDA sponsored  organization.   Information  concerning  the
NJRC&D model ordinance  will be provided  to numerous  local
municipal and private planning organizations in a series  of
spring (1992) meetings.  The RC&D, nation-wide, will also be
made aware of this and other NJRC&D work in relation to car-
bonate rock concerns.   In addition, a  number of  inquiries
from State and County level individuals, concerning the  or-
dinance, were received  by the authors  both at the  confer-
ence and subsequently.
                        667

-------
668

-------
      ANALYSIS OF DRASTIC AND WELLHEAD  PROTECTION METHODS

                   APPLIED TO A KARST SETTING


                       Lyle V.  A.  Sendlein

                Department of Geological Sciences
                     University of Kentucky
                       Lexington,  Kentucky
ABSTRACT

     DRASTIC and Wellhead Protection methodologies were used in
the analysis of an Inner Bluegrass Karst Region karst spring used
as a municipal water supply.  The area has been studied by
Thrailkill and his students (Thrailkill et.  al., 1982).  Through
dye tracing and field analysis, the area has been divided into
groundwater basins and interbasin areas.  Groundwater basins are
characterized by numerous sinkholes, conduit flow and deep water
table (33 meters).  Interbasin areas are characterized by shallow
water table, high level (upland) springs, fracture flow and
sinkholes.  Existing information was utilized in the DRASTIC
analysis with inconsistent results due to the dichotomy of a deep
water table found in the groundwater basins.  Even though high
values were used for most parameters, the deep water table
reduced the index value,  thus minimizing the potential impact.
     The Wellhead Protection analysis and the delineation of a
Wellhead Protection Area (WHPA) provides a more direct manner for
the identification of the area that needs to be protected.  The
definition of the area based on Thrailkill's concept of a ground-
water basin as defined by dye tracing provides a good planning
tool.  An important fact that must be dealt with in the planning
phase is the 80 percent of the WHPA for the municipal water
supply lies outside the jurisdiction of the community trying to
protect its water supply.   The adjacent community is currently
addressing a planning issue that involves the WHPA because a
major development is proposed for a part of the WHPA.


BACKGROUND

     This analysis is based on data gathered over the last three
                                 669

-------
years on contracts supported by the Divisions of Water and Waste
Management of the Natural Resources and Environmental Protection
Cabinet of Kentucky, the United States Environmental Protection
Agency and the University of Kentucky's Institute for Mining and
Minerals Research.
     DRASTIC maps were developed for six areas in Kentucky to
test the feasibility of producing such maps from data available
in existing data bases, files and published maps and reports.
Areas were chosen for each of the major physiographic regions of
the state including cities whose public water supply depended on
groundwater as all or part of the source.  (Sendlein, 1989)
     Wellhead protection work was conducted for three of the
areas for which DRASTIC maps had been developed.  This was a
pilot study and because of the limited funds, complete analysis
was not possible.  In fact, the study was also based on available
data and consisted of construction of the Wellhead Protection
Area maps for each area and the compilation of sources of poten-
tial contamination from state and federal files.  Each area
included spot field inspection of the groundwater water sources
and in one case a pumping test was attempted. (Sendlein and
Fitzmaurice, 1990)
     Kentucky is very fortunate because the entire state is
covered by seven and a half minute USGS quadrangle maps including
topography and geology.  In the construction of the DRASTIC maps,
the USGS maps were used as base maps, and each area studied
included two adjacent sheets, except one of the areas was
composed of four adjacent sheets.  Because the geology is know at
this level of detail, the construction of general environmental
maps such as DRASTIC maps, does not add much information for an
area that a qualified geologist could not glean from the geologic
maps.  However, the non-geologist lay person can benefit from the
conclusions reached as a result of the DRASTIC analysis.
     Considerable information is present in the files of the
Divisions of Water and Waste Management in Frankfort, Kentucky.
These data are scattered throughout the agency's various offices
with permits kept in a central file room.  Much of the tabular
information is kept in computer files, but most maps are on
paper.  An exception to this is the map information kept in the
Kentucky Natural Resources Information System (Croswell et al.,
1982) utilizing the ARCINFO software.  The scale of the
information included in this database is on the order of
1:1,000,000, considerably larger than the 1:24,000 of the
quadrangle maps used in the present studies.  The DRASTIC maps
were developed on the computers in Frankfort and are stored as
files in their system.  (Couch, 1988)
     A recent water well law in Kentucky requires that all water
wells drilled in the state be recorded and a driller's log be
filed with the Division of Water.  These logs, along with other
bore hole information, are being compiled into a database by the
Kentucky Geological Survey and will be an important database when
it comes on line.  Currently, the paper files can be reviewed in
the Division of Water and a computer printout of some of the
parameters are available from that office.
                                670

-------
STATEMENT OF THE PROBLEM

     A recent planning action of the Lexington-Fayette  Urban
County Government (LFUCG) focused attention on Royal  Spring,  the
municipal water supply for the community of Georgetown,  Kentucky.
Royal Spring's flow ranges between 30 1/s  (liters/second)  on the
low side to over 3,000 1/s on the high side with median flow
between 300 and 1,000 1/s (Thrailkill et al, 1982).   The LFUCG as
part of the comprehensive planning process identified an area for
special consideration for which the Mayor appointed the Coldstr-
eam Small Area Planning Committee in April of 1990.   This area
lies in the northern part of the city and has a large tract  of
land owned by the University of Kentucky and is used  for
agricultural research (Figure 1).  The property at one  time  was
considered to be in the country but now with the addition of 1-75
and increased development in this part of the city, there are
several reasons why land use revisions should be discussed.
     As part of the deliberations of the Committee and  the
planning staff, the fact that this tract of land was  part of the
recharge zone for Royal Spring became known and was discussed at
some length.  The final report of the Committee includes a
section that addresses this natural feature.
     This location was an area included in one of four  quadran-
gles mapped using the DRASTIC methodology.  Likewise, the Royal
                           FIGURE  1
                    COLDSTREAJ1 FARM STUDY AREA LOCATION
                           COLDSTREAn FARM

                               STUDY AREA
                               671

-------
Spring was identified as a public water supply to be included in
the Pilot Wellhead Protection Study of Kentucky.  The availabili-
ty of these maps and accompanying reports provided an opportunity
to analyze the utilization of these data in the planning process.
DRASTIC MAPS AND RESULTS

     Four quadrangles were mapped, Lexington East  (MacQuown and
Dobrovolny, 1968), Lexington West  (Miller, 1967), Georgetown
(Cressman, 1967) and Centerville,  (Kanizay and Cressman, 1967)
using the methodology defined by Aller  (1987) .  A combined
version of these four quadrangles are presented in Figure 2.  It
can be seen that DRASTIC Index values range from a low range of
100-119 to a high of 200+.  Most of the background of the maps
fall into the Index category of 180-199.  These are considered
high values and indicate a high potential for groundwater
contamination.  In constructing these maps, the work of John
Thrailkill and his students (1982) was used because they
identified areas drained by individual major springs as
groundwater basins through dye tracing studies.  These basins can
be seen on the DRASTIC map as fat, teardrop-shaped areas with
index values less than the areas surrounding them  (160-179).  The
other areas (referred to as "background" above) of the map fall
into what Thrailkill has identified as interbasin areas.
     This difference in values was expected between groundwater
basins and interbasin areas, but the fact that the groundwater
basin has a lower DRASTIC Index is just the opposite relative to
its vulnerability for contamination.  Within the groundwater
basin, there are more sinkholes and greater opportunity for
surface water to move from the surface to the conduit below.
This is an important defect in the DRASTIC method as applied to
karst areas.  The difference is produced because Thrailkill has
shown that in the groundwater basins the conduit is the main
drain and the water table is deep, at conduit level.  In the
DRASTIC method, deep water table is given less value because of
the rationale that the deeper the water table,  the less likely it
will be contaminated by surface pollutants.
     Royal Spring's groundwater basin is elongate and extends
southward from the city of Georgetown.  It is not represented by
only one DRASTIC Index value but is divided into two regions with
the higher index value (200+)  in the southern part of the basin
and the other region (160-179) occurring near the spring in the
northern part of the basin.  The difference between index values
is controlled by the difference in surface elevation.  Most of
the groundwater basin underlies the surface valley of Cane Run,
but in the lower region of the basin, closer to Georgetown and
the spring, the groundwater basin does not underlie the Cane Run
valley, because the valley abruptly turns toward the west.  The
lower end of the groundwater basin lies beneath a higher
topographic surface, thus the deeper water table and lower
DRASTIC Index value.
                                672

-------
             FIGURE 2
ROYAL SPRINSS  - DRASTIC  INDEX MAP
                                  STUDY AREA
                                  RDADS  -
                                  STREAKS
                                  CITIES  [r/  I
                 FIGURE 3
ROYAL  SPRINGS -  WELLHEAD PROTECTION AREA
                                                            B4'37'30"
                                                                                                                                     84'22'30"

-------
WELLHEAD PROTECTION FOR ROYAL SPRING

     As part of the pilot study of Kentucky, Royal Spring was
identified as a municipal water supply to be studied.  The study
included defining the Wellhead Protection Area  (WHPA) and
identifying potential sources for contamination of the aquifer.
Figure 3 shows the Wellhead Protection Area for Royal Spring.
Using U.S. Environmental Protection Agency terminology, the Zone
Of Contribution (ZOC) for this area is equal to the recharge area
to the spring.  A second zone has been defined where the Time Of
Travel (TOT) is measured in hours and days rather than months and
years common in non-karst regions.
     The boundaries of the area were determined by using
Thrailkill's approach to not only include the groundwater basin
but also the "catchment area" for the groundwater basin.  By
reference to Figure 2, one can see the other groundwater basins
that bounded the Royal Spring groundwater basin.  The area
bounded by the drainage divide for Cane Run represents the catch-
ment area.  Two deviations from this occur, one at the southern
end of the WHPA where Russell Cave Spring robs part of the
drainage of Cane Run, and the other at the northern end of the
WHPA where Cane Run abruptly turns to the west and the groundwa-
ter basin no longer underlies the drainage basin of Cane Run
(Figure 2).
     A zone (Zone 1) within the WHPA (Figure 3) that defines the
groundwater basin represents a more vulnerable region of the
recharge basin, because within this zone more than 150 openings
to the main conduit have been identified by Thrailkill and
include the swallets that have been used for tracing the basin.
Maximum travel times from the farthest swallet have been as low
as 141 hours,  but calculations of flow velocities are in the
order of 4 km/hr,  which would transfer to as little as four hours
of travel time for the Newtown Swallet that is approximately 15
kilometers from the spring.  The remaining zone represents the
rest of the area within the ZOC and falls into the interbasin
area.
     The other major part of the Wellhead Protection method is
the identification of potential contamination sources.  Sources
identified and located on maps and tabulated for the Royal Spring
WHPA  (Fitzmaurice, 1990) are: subsurface percolation from
sanitary systems,  injection wells, landfills, open dumps
(including illegal dumping), residential (or local) disposal,
graveyards, aboveground storage tanks, underground storage tanks,
containers, pipelines, material transport and transfer,
operations, animal feeding operations, production and other
wells, monitoring and exploration wells, and salt-water
intrusion/brackish water upcoming.  The information obtained on
these potential contamination sources varied for Fayette and
Scott Counties.  Fayette County is the more populated of the two
counties and more potential sources are present and thus more
information available.  The LFUCG has computerized much of their
information and have a program that converts street addresses to
latitude/longitude which made it possible to plot information on
maps.  These data are included on maps in Fitzmaurice  (1990).
                               674

-------
COLDSTREAM SMALL AREA HYDROGEOLOGY

     The topography ranges from 980 feet at the southern end to
885 feet at the point where Cane Run leaves the property.  The
floodplain of Cane Run is 910 feet at the southern end of the
site and 890 feet at the northern end, with a distance of
approximately 9,900 feet or a floodplain gradient of 20
feet/9,900 feet (0.002).  The total area lies within the upper
region of the surface drainage basin of Cane Run.  Presently,
developed areas (the two subdivisions and the industrial park)
are connected to the sanitary sewer system of LFUCG, but surface
run off is directed to the natural or constructed drainage ways
within the area.
     Karst features within the area are two swallets (one
considered an estavella by Thrailkill),  one active spring,  one
inactive spring (tufa deposits observed at this spring),  sinking
stream (most of Cane Run is dry except for major precipitation
events and spring flow terminates in one of the swallets),  and
several sinkholes.
     The geologic units present in this area are the various
members of the Lexington Limestone of Ordovician age.  The
Lexington Limestone includes argillaceous units generally less
than 6 meters thick and highly variable.  Within the area of
interest the lower unit of the Lexington Limestone is the Grier
Limestone and above that is the Tanglewood Limestone.  Interbed-
ded within this unit are the argillaceous units, some include
interbedded shale.
     Royal Spring emerges from the Grier Limestone Member about
five meters below the contact with the Tanglewood Limestone.
Within the area, one of the argillaceous units is the two meter-
thick Cane Run bed, which is mapped about 10 meters above the
Tanglewood-Grier contact.  Thrailkill believes that the main
conduit for Royal Spring may be stratigraphically controlled
within this interval and that the Cane Run bed does not perch the
subsurface conduit because most or all of the dye traced swallets
within the groundwater basin penetrate the unit.
     Surface flow in Cane Run is intermittent on the Coldstream
Small Area, but a part of the stream has been observed to flow
between precipitation events.  The source of the water is a
spring shown on Figure 5.  Because of the construction in the
area, it is not possible to determine if the source of the spring
water is related to the storm water system of LFUCG.  This water
flows on the surface for a short distance entering Cane Run and
sinking in the channel at the swallet identified by Thrailkill as
the Coldstream Swallet  (Figure 5).  Near the point where Cane Run
leaves the Coldstream Small Area, discharge from the University
of Kentucky farm operation flows into the channel.


COLDSTREAM SMALL AREA PLAN

     Land use of the 1770.9 acre area as listed in the 1988
Comprehensive Plan is shown in Table 1.   The Coldstream Small
Area plan changes are also presented in Table 1.


                                675

-------
        FIGURE  1
COLDSTREAtt SMALL  AREA PLAN
                                LEGEND

                           STUDY BOUNDARY

                           PROPERTY LINES
          FIGURE  5
COLDSTREAN SMALL AREA - TOPOGRAPHY

-------
                             TABLE 1

    EXISTING AND PROPOSED LAND USE FOR COLDSTREAM SMALL AREA

LAND USE
LOW DENSITY RESIDENTIAL
MEDIUM DENSITY RESIDENTIAL
RETAIL TRADE t PERSONAL SERVICES
PROFESSIONAL SERVICES
HIGHWAY ORIENTED COMMERCIAL
SEMI-PUBLIC USES
PARKS
OTHER PUBLIC USES
OFFICE, INDUSTRIAL RESEARCH PARK
WAREHOUSE i WHOLESALE
CIRCULATION
UTILITIES & COMMUNICATIONS
TOTAL
1988 PLAN
EXISTING
117.6
0.0
0.6
27.8
3.8
5.6
16.8
984.9

146.8
54.4
2.2
1363.5
PROPOSED
69.0
307.5
9.8
22.9
0.0
0.0
0.0
0.0

0.0
0.0
0.0
407.4
TOTAL
186.6
307.5
10.4
50.7
3.8
5.6
16.8
984.9

146.8
57.4
2.2
1770.9
1991 PLAN
PROPOSED
178.5
139.9
25.5
61.0
0.0
0.0
0.0
235.3
781.3
0.0
0.0
0.0
1421.5
TOTAL
296.1
139.9
26.1
65.3
3.8
5.6
16.8
235.3
781.3
146.8
57.4
2.2
1770.9
Figure 4 identifies the major land use of the Coldstream Small
Area,  and Figure 5 illustrates the topography and geological
features.  The major change in the area will be the development
of the 984.9 acres, listed as Other Public Uses,  owned by the
University of Kentucky-  It is the University's intent to develop
a research park in this area.  One site has already been devel-
oped by Hughes Aircraft on the property,  and the University has
plans for similar high-technology industries to locate on the
property and be related to the University of Kentucky through
faculty and student interaction with the new industries.
     The Committee has recommended the following changes to the
1988 Comprehensive Plan that are shown in Table 1.  Major changes
are the addition of 178.5 low density residential, 139.9 medium
density residential, and the addition of a new land use
identified as Office,  Industrial Research Park that would have
781.3 acres.
     While the Committee Report acknowledges that much of the
Coldstream Small Area falls in the recharge area for Royal
Spring, no specific recommendations were made to protect the
spring from water quality changes that will be caused by the
increased development.  The Georgetown Water Board presented to
the Committee the policies they plan to follow relative to
development in the recharge area under their jurisdiction (Table
2).
                                677

-------
                                TABLE 2

      AQUIFER RECHARGE PROTECTION POLICIES  FOR ROYAL  SPRING
     The Urban Service Boundary should not be extended into the Royal Spring
     Aquifer recharge area for urban development.

     No additional properties within the Aquifer Recharge Area should be
     planned industrial.

     Existing lands planned industrial but not yet zoned should be
     redesignated •environmentally sensitive light industry".  This category
     does not allow hazardous materials users/generators.

     For lands already zoned industrial, new hazardous materials users are
     prohibited, and existing hazardous materials  users may expand only if
     the design and operation of the site would protect the aquifer from
     contamination.

     Design standards are given for safe storage and spill containment.

     No new underground storage tanks should be located in the Aquifer
     Recharge Area.

     Rural residential development on septic systems should be clustered
     outside the recharge area where feasible.

     The recharge area has the highest priority in the proposed TDR/PDR
     program.

     The LFUCG is encouraged to adopt similar protection policies.
STATUS  OF KNOWLEDGE  OF THE SITE
     The long term  studies of John Thrailkill and his  students
have identified the swallets that are directly linked  to the
Royal  Spring.  Even though this  information  has been known for
the last ten years,  very little  practical application  of the
knowledge toward planning resulted.
     The DRASTIC analysis illustrates that the Coldstream Small
Area falls in the highest and second highest categories of the
DRASTIC Index indicating high vulnerability  to groundwater
contamination.  Although the study results are inconsistent on an
area basis because  it did not highlight the  most vulnerable areas
for most of the area studied, it did identify the most vulnerable
area that must be protected in  the Coldstream Small Area (200+
index  area).
     The Wellhead Protection study identified the recharge area
(ZOC)  for Royal Spring and based on previous work did  identify
Zone 1 (200+ index  which is a groundwater basin identified by
Thrailkill) as the  most vulnerable area within the Wellhead
Protection Area that must be protected.  Once Georgetown
officials became aware of this  map and related study,  they began
to include the information in their planning process.   It was the

                                   678

-------
WHPA map and Georgetown's desire to be included in the Coldstream
Small Area Planning Committee that brought the issue to the
table.
     Systematic stream hydrographs have not been developed for
Cane Run or the segment flowing across the Coldstream Small Area.
Cane Run receives flow from a spring on the site and loses water
to two swallets and fractures in the bottom of the Cane Run
channel.  Cane Run is an intermittent stream that will experience
increased discharge as development proceeds.
     The Sinkhole Ordinance (described by Dinger and Rebhman,
1986) enacted in Fayette County is an important application of
geological knowledge but is not directed toward protection of
groundwater quality.  In fact, the ordinance allows the direction
of surface runoff toward sinkholes but does exclude the construc-
tion of dwellings within them.
MANAGEMENT OPTIONS FOR THE COLDSTREAM SMALL AREA

     The proposal to rezone this area has provided an opportunity
for the analysis of current laws and policies relative to the
protection of groundwater resources utilized as a municipal water
supply.  While the city officials of Georgetown, Kentucky are
cognizant of the many potential sources of contamination to their
water supply and the fact that it is highly stressed during
periods of drought, other sources of water are being sought but
will not be in place in the near future, thus this supply is very
important to them.  The problem is more complex because George-
town does not control the whole WHPA and therefore cannot enforce
land use restrictions to protect the aquifer.
     State officials recognize the dilemma faced by Georgetown,
but feel they have no regulations that would allow them to effect
the development of a protection policy.  They were able to inject
the State into a problem during the drought of 1988 and cause a
single user to stop pumping a large volume of water from the
conduit supplying Royal Spring, thus allowing the water to be
used for the citizens of Georgetown.
     Few examples exist to use as a model for developing a
protection strategy for this area.  Austin, Texas is currently
experimenting with two policies to protect a karst aquifer that
provides an important recreational spring  (Butler, 1987) .  East
Marlborough and.West Whiteland Townships, Chester County Pennsyl-
vania have adopted zoning ordinances to protect a carbonate
aquifer (Jaffe and Dinovo, 1987).  Both of these utilize resource
protection and source control methods to protect the aquifer from
potential contamination.
     As part of the Coldstream Small Area Plan, a green space
(resource protection)  to roughly parallel the 100 year floodplain
for Cane Run on the property was proposed.  This would be an area
where no construction other than utility lines and limited roads
cross the channel.  This approach has been used in Austin, Texas
as a protection strategy for the recharge zones of the Edwards
Aquifer (a karst aquifer).  They have established three zones as
illustrated in Figure 6 based on stream hydrograph and water
                                679

-------
quality data collected for  the area.  The Pennsylvania example
limits the number of dwelling structures per acre and controls
the source of potentially contaminating substances such as home
heating oil.
                          FIGURE  6
                      MAJDR  WATERWAY
                  WATER QUALITY  ZONES
                           200' MIN,
                                     400' MAX,
                                         300'
                  CRITICAL  WATER  QUALITY  ZONE
                  WATER  QUALITY BUEEER  ZDNE
                  UPLANDS  ZONE
     Application of the water  quality zones illustrated in Figure
6 to the Coldstream Small  Area would be similar to the
designation of the green space that is under consideration at  the
present time.  The Critical  Water Quality Zone (Figure 6)  is a
zone that does not allow any improvement other than minor roads
but is generally used as a park area.  The Water Quality Buffer
Zone is a zone with limited  development activity.  For example,
very low density residential or light industry with limited toxic
substances used by them.  The  Upland Zone does not have any
restrictions.  If this were  overlaid on the sinkhole ordinance in
the Coldstream Small Area, this would be a beginning of a
protection strategy for Royal  Spring.
     To make any zoning strategy cost effective,  baseline
information on the physical  properties within the area should  be
collected before development takes place.  This information
should include the following:  1) stream flow from all channels
providing water to Cane Run, 2) water entering and leaving the
Coldstream Small Area, 3)  channel gain or loss within the Cane
Run channel on the property, and 4) water quality of water
entering Cane Run from all sources  (springs, surface water
discharges from developed  areas, and overland flow).  Information
of flow conditions will allow  for the exact location of the
boundaries of the green space.
                                680

-------
     Likewise,  consideration of the kinds of land uses that are
best suited for the area should be defined.  It is understood
that sanitary sewers are to be provided for all occupants of the
new developments.   Special consideration of the quantity and
quality of surface water leaving the developed properties must be
given so that precautions can be taken so that the quality of the
water entering Cane Run does not deteriorate.

CONCLUSIONS

     The use of scientific information for the purpose of plan-
ning is not easy to document because it becomes part of the
policy-making process and this process is not  generally document-
ed by retrospective analysis.  In this case,  Royal Spring is the
largest spring in the Commonwealth and is used as a municipal
water supply and the Kentucky Department of Environmental Protec-
tion and Natural Resources became interested in protection
strategies for the spring.  Their support of the DRASTIC and
Wellhead Protection studies focused attention on the spring and
it is fortuitus that the LFUCG began to study the Coldstream
Small Area to develop a plan for rezoning the area within the
same time frame.  From this study the following conclusions can
be made:
     1.   The Wellhead Protection analysis had more of an impact
          on the planning process because it defined the Zone Of
          Contribution for the Royal Spring.
     2.   The zone identified within the WHPA (Zone 1) is a very
          vulnerable part of the ZOC and occupies a large portion
          of the Coldstream Small Area Plan.
     3.   The previous work by Thrailkill, while known to exist
          for some time, was not directly applied to the planning
          process until the present time.
     4.   Because approximately 80% of the ZOC is outside the
          jurisdiction of the Georgetown City Council, protection
          strategies for the entire WHPA cannot be enforced.
     5.   There are no state or federal laws that give the Georg-
          etown City Council authority to protect the ZOC outside
          of Scott County.
     6.   Protection of the ZOC within the Coldstream Small Area
          will depend on the building permits  issued by the
          LFUCG.  The statement that LFUCG recognizes that the
          portion of the Cane Run drainage basin on the property
          is in recharge area for Royal Spring is a positive
          sign, and their willingness to define a green space
          along the stream channel that includes the two swallets
          that are directly linked to Royal Spring is a form of
          aquifer protection.
     7.   Baseline data must be collected to document the poten-
          tial changes in the quantity and quality of water
          flowing into Cane Run before, during and after develop-
          ment if a protection strategy is to be cost effective.
                                681

-------
REFERENCES CITED

Aller, L., Bennett, T., Lehr, J. H., Petty, R. J.,  and  Hackett,
     G., 1987, DRASTIC: A Standardized system  for  evaluating
     ground water pollution potential using hydrogeologic
     settings: U. S. EPA-600/2-87-035, 1-455
Butler, Kent S., 1987, Urban Growth Management and Groundwater
     Protection: Austin, Texas, in Planning For Groundwater
     Protection, edited by G. William Page, Academic  Press, New
     York, pages 261-287
Couch, Amber, 1988, A DRASTIC Evaluation For Part  Of  The Inner
     Bluegrass Karst Region, Kentucky: The Centerville,
     Georgetown, Lexington East, Lexington West Quadrangles,
     Unpublished MS Thesis, University of Kentucky, 95  pages
Cressman, Earle R.,1967, Geologic Map of the Georgetown
     Quadrangle, Map GQ-605, USGS
Croswell, P. L., Sanders, S. L., Dryden, W., and Schneider, W.
     L., 1982, User Guide to the data and capabilities  of the
     Kentucky Natural Resources information System: Kentucky
     Natural Resources and Environmental Protection Cabinet,
     Lands Unsuitable for Mining Program, 149 pages
Dinger, James S. and James R. Rebman, 1986, Ordinance for the
     Control of Urban Development in Sinkhole Areas in  the Blue
     Grass Karst Region, Lexington, KY, in Proceedings  of the
     Conference on Environmental Problems in Karst  Terranes and
     Their Solution, NWWA, pages 163-180
Fitzmaurice, Karen, P, 1990, An Analysis of the Kentucky Pilot
     Wellhead Protection Programs For Georgetown,  Elizabethtown,
     and Calvert City, Kentucky, Unpublished MS Thesis,
     University of Kentucky, 200 pages
Jaffe, Martin, and Frank Dinovo, 1987, Local Groundwater Protec-
     tion, American Planning Association, Washington, D.C., 262
     pages
Kanizay, S. P. and E. R. Cressman, 1967, Geologic  Map of the
     Centerville Quadrangle, Map GQ-653, USGS
Miller, Robert D., 1967, Geologic Map of the Lexington  West
     Quadrangle, Map GQ-600, USGS
MacQuown, William C. Jr. and Ernest Dobrovolny, 1968, Geologic
     Map of the Lexington East Quadrangle, Map GQ-683
Sendlein, Lyle V. A., 1989, DRASTIC Analysis For Application By
     State Government, Final Report, Groundwater Branch, Division
     of Water, Division of Waste, Natural Resources and
     Environmental Protection Cabinet, Kentucky, Frankfort, KY,
     59 pages
Sendlein, Lyle V. A. and Fitzmaurice, Karen, A, 1990, Kentucky
     Pilot Wellhead Protection Study, Draft Report  to USEPA,
     Contract # 8R-1574-NAEX, Region IV Atlanta, 95 pages
Thrailkill, J., Spangler, L. E., Hopper, W. M. Jr., McCann, M.
     R., Troester, J. W., and Gouzie, D. R., 1982,  Groundwater in
     the Inner Bluegrass Karst Region, Kentucky, University of
     Kentucky, Water Resources Research Institute,  Research
     Report No. 140, 89 pages
                                682

-------
BIOGRAPHICAL SKETCH

Lyle V.  A. Sendlein received his BS and AM in geological
engineering and geology, respectively, from Washington University
in Saint Louis, Missouri in 1958 and 1960.  He received his Ph.D.
from Iowa State University in geology and soil engineering in
1964.  He taught and conducted research for seventeen years at
Iowa State University.  In 1977 he assumed the directorship of
the Coal Research Center at Southern Illinois University at
Carbondale and in 1982 assumed his current position as the
director of the Institute for Mining and Minerals Research at the
University of Kentucky.  He is also a Professor in the Department
of Geological Sciences and is currently directing graduate
student research and conducting research in hydrogeology.   His
areas of research include aquifer definition in both the coal
fields and the karst regions of Kentucky and contaminant
transport in these hydrogeological domains related to mining,
waste disposal and agricultural practices.
                               683

-------
       ANALYSIS OF DRASTIC AND WELLHEAD PROTECTION METHODS

                         APPLIED TO A KARST SETTING


                                Lyle V. A. Sendlein

                        Department of Geological Sciences
                              University of Kentucky
                               Lexington, Kentucky
Question 1.

Because DRASTIC drastically under-estimated the vulnerability of those portions of the
karst basins dominated by conduit flow and recharged directly by swallets in sinking
streams and open sinkholes, how can it be utilized as a predictive hydrologic mapping tool in
karst?  Aren't there serious limitations and dangers in relying on DRASTIC in karsts
without performing tracer studies and doing field reconnaissance for direct recharge sites
such as swallets and open sinkholes? Won't dye-tracing give you more information (and
more accurate information) about aquifer vulnerability?

Question la.

"Because DRASTIC drastically under-estimated the vulnerability of those portions of the
karst basins dominated by conduit flow and recharged directly by swallets in sinking
streams and open sinkholes, how can it be utilized as a predictive hydrologic mapping tool in
karst?"

The key words in this question are "predictive hydrologic mapping tool". DRASTIC is not a
hydrologic mapping tool. The method is designed to assist hi the planning process, which
would address large areas and is not designed for site specific use. For the areas mapped in
this study, high DRASTIC Index values resulted which would alert public officials (including
planners) that this area is vulnerable to groundwater contamination from surface activities.
As indicated in the paper, problems arose when karst groundwater basins determined by dye
traces were introduced into the analysis because the deep groundwater table caused by
drainage through the main conduit resulted in a lower DRASTIC Index value for those areas
within the groundwater basins.

Question Ib.

"Aren't there serious limitations and dangers in relying on DRASTIC in karsts without
performing tracer studies and doing field reconnaissance for direct recharge sites such as
swallets and open sinkholes?"

If the analysis produces high DRASTIC Index values, the purpose of the analysis has been
accomplished, even though specific karst features have not been identified.  I do believe that
if one has knowledge of specific karst features that would directly transport contaminants to
the groundwater within the conduit system, they should be added to the map.
                                          684

-------
Question Ic.

"Won't dye-tracing give you more information (and more accurate information) about aquifer
vulnerability?"

Of course dye-tracing will give site specific as well as general information about the karst
groundwater system. This would have to be the next step in an analysis of an area should
one decide to proceed with the planning of a specific land use knowing that the DRASTIC
Index value is high, which indicates a high vulnerability for contamination of the
groundwater.

Question 2.

I have found that although DRASTIC is a powerful and practical mapping tool in terranes
consisting of many rock types, it is useless for distinguishing slight differences in
vulnerability if the entire area to be evaluated in a karst.  DRASTIC wasn't designed to
evaluate differences between contiguous karst areas. For Hart County, Kentucky, I found
direct interpretation of the published geologic maps to be most reliable.  Your comment is
welcome.

I agree with this statement.  For the public official or planner who may not know anything
about geology and karst, a DRASTIC map  of a karst area would alert the planner to the
high vulnerability for contamination of the groundwater by surface activities. A qualified
hydrogeologist can take a published geologic map and probably glean more valuable
information about the groundwater flow systems in the area than a lay person. But I would
remind all that this method was not developed for geologists, but for public officials and
planners.
                                          685

-------
686

-------
              USE OF DYES FOR TRACING GROUND WATER:

                      ASPECTS OF REGULATION


                         James F. Quinlan

                   Quinlan & Associates,  Inc.
                   Nashville, Tennessee 37222



                            ABSTRACT

     An affirmative EPA  policy  on  tracers is needed,  and I urge
that steps be  taken to  establish  one  as  part of  legislation or
administrative  regulations.    Suggested  wording  for  part of  a
statute is included here, but if legislation is unnecessary and a
policy can be established administratively, it should be so done.
Tracing is  a technique  uniquely  able to test and  predict flow
velocity and flow direction of water and pollutants.  When properly
used in  the settings  where they  work best  — karst aquifers,
fractured rocks, and  coarsely  granular sediments  —  tracers  can
give reliable answers that can  not be  obtained  in  any other way.
The  substances  traditionally  used for  tracing are  benign  and
harmless  in  the concentrations employed by most  knowledgeable
investigators.   The use of tracers  is justifiable on the basis of
their safety, ease of use, utility,  reliability, and unique ability
to secure results obtainable in no  other way.  Currently, EPA has
no policy on the use or non-use  of  tracers, but their use has been
endorsed by the Agency in several publications and  reports.  Some
states routinely employ tracers; a  few  regulate them; a few forbid
their use.  Most states  ignore them and  have no policy on their
use.


                          INTRODUCTION

     Tracing agents  (tracers) are added to  a medium, usually water,
to enable the medium to be recognized at a remote location.  Their
arrival concentration  and arrival time  can be used in calculations
and decisions.   The  most  common  tracers are substances  added (such
as fluorescent dyes) or physical changes (such as temperature) made
in the medium.   This paper  is concerned primarily  with tracers
added to water and used to study its movement and velocity as they
relate to environmental problems.
                               687

-------
     Tracers are fundamental tools for discovery and prediction of
the velocity and dispersal-path of pollutants in ground water and
surface water.   Interpretation of data from tracer studies  makes it
possible to protect water quality, public health, and aquatic life.
Such data are crucial to the development of wellhead and springhead
protection  strategies  and  for  high-reliability  prediction  of
contaminant flow-paths and flow-directions.  They can be essential
for the calibration of computer models of water  flow and pollutant
movement, and  they greatly enhance the  validity of such models.
Tracing  is  cost-efficient  and  is often  the only  way to  obtain
essential data.

     Two major  types  of tracers are  used most  commonly:    water-
soluble  fluorescent  dyes  that  have  been shown to be non-toxic
(Field et al.,  1992) and ions of non-toxic elements or compounds.
Specific tracers are named  in a subsequent paragraph.

     The  use of tracers  as  an  essential aid  in  the design of
monitoring  systems in karst  terranes is  critically reviewed by
Quinlan  (1989,  1990).  Most karst terranes are  areas underlain by
limestone and/or dolomite and characterized by  sinkholes, sinking
streams, caves, and springs. Approximately 20% of the U.S.  (40% of
the eastern U.S.  that includes the upper Mississippi Valley)  is
underlain by carbonate rock and is some type of karst.

     An  anonymous  wag  has  whimsically  stated that  "One  well-
designed tracer test, properly done, and  correctly  interpreted is
worth  1000  expert  opinions ... or  100 computer  simulations of
ground-water flow  in  karst terranes."   He believes that this is
because most computer models are incapable of giving valid results
for flow  in most karst  terranes — for various technical reasons
not germane to this paper.

     I do not deny that great advances have been made (and will be
made)  in  understanding and  predicting  ground-water  flow with
computer models, and  I  freely acknowledge their unique capabili-
ties.  Nevertheless, models have recognized limitations (Scarrow,
1989), about which two points must be made:

     1.  Hard  data (real  numbers,  based on measurements) have
         greater reliability than  soft,  imaginary data  (calculated
         values derived  from  computer  models   or  simulations) .
         Given  a choice between believing the  results of  a well
         designed,  properly executed, and  carefully  interpreted
         tracer test or the  answers given by a computer model which
         disagree with the tracer test, the tracer  results  should
         be given greater  weight —  for the same reason that we do
         not debate the existence of gravity.  The tangible results
         of each are obvious.

     2.  Computer models are  no better than the validity  of the
         assumptions made  in their design, the reliability (repre-
         sentativeness)  of the parameters used for data input, and
         the correctness of the analysis  and  manipulation  of data.
                                688

-------
        Their  validity  must  be  evaluated with  a  sensitivity
        analysis and by  comparison with  hard data.

I make these points because tracers  and  models do not substitute
for one another.  Tracer  results take priority over model results,
however, but each can complement the other,  doing what the other
can not.  Although I enthusiastically endorse the use of computer
models  for  analyzing  karst  aquifers  when  done  successfully
(Teutsch,  1989; Teutsch and Sauter, 1991; Sauter, 1991a, 1991b), I
demand  the  rigor   imposed  by  J.   C.  Griffiths,  the  doyen  of
quantitative geologists,  on the methodology of geologic investiga-
tion (Drew, 1990,  p. 3-10).
                     DESCRIPTION OF  PROBLEM

     The use  of tracers  for  solving environmental  problems has
become well established in recent years.  In some settings, it is
not possible to define important environmental parameters without
first conducting comprehensive tracing studies.

     In an  unrealistic effort  to  protect human health  and the
environment  from any  possible harm,  some  state agencies  have
unrealistically restricted the use of tracers,  regardless of the
information to be gained.  For  example,  some state agencies have
interpreted  tracers  used  in  ground-water  investigations to  be
pollutants  or  contaminants  and  therefore  violations  of  their
"non-degradation policy."  Strictly speaking, injection of dye or
other non-toxic tracers into a well  (or other input source, e.g.,
a sinkhole), no matter how noble the  reason for doing so, makes it
possible to construe the well to be  a Class  V injection well and
thus subject to State  and  Federal  regulations governing its use.
(If the tracer were toxic, the well would be a Class IV injection
well.)

     Part of  the problem  is  due to the misperceptions  of  many
individuals about fluorescent dyes and  other substances used for
environmental  tracing  studies,  especially with regard  to their
suspected  toxicity.     This   is  the   fallacious  logic  used:
Fluorescent dyes are synthetic substances (i.e.. ,  they do not occur
naturally); therefore,  they  must be a  pollutant  and  harmful  to
human  health   and  the   environment.     This  misperception  is
exacerbated by the lack of comprehensive toxicity test data on the
dyes.  The fallacious argument continues:  Since fluorescent dyes
must be harmful,  so probably must  be all other tracing agents.
Granted,  some   tracers   are   hazardous   (e.g.,   tritium),   but
assumptions such as these  cited are  absurd and unscientific, and
they decrease our ability to gather data needed to better protect
ground-water quality.

     Clearly, there is a  need for  a  Federal  policy endorsing the
use  of  tracers  in ground  water if the  various  misperceptions
regarding their use are to be dispelled.   Such a policy does not
need to be very extensive  or detailed, but it does need  to be


                               689

-------
flexible.
                            DISCUSSION

     Well-meaning as the  above  interpretation of regulations for
injection wells and the concern regarding suspected tracer toxicity
may be, it is  not justifiable in terms of potential benefits for
environmental  protection,  original intent of  the law-makers, or
risk  of  exposure to  pollutants.   Like boats  put  into a lake,
tracing agents are used  in  the water  for  a good  and definite
purpose,  not put into it  for disposal.   And like boats, dyes and
ions  generally used  for  tracing  ground  water  are  benign  and
harmless in  the concentrations  commonly employed (Field et al.,
1992; Smart,  1984).

     A further analogy describing the use of tracers can be made.
Doctors use vaccines and a wide  range of diagnostic techniques to
prevent and treat illnesses.  Some of these vaccines and techniques
have definite  risks associated with their use.   These  risks are
assumed by an informed patient  because the  consequences  of not
preventing or  not diagnosing an  illness  far outweigh  the much
smaller risk from use of the vaccine or diagnostic technique.

     If and when state  officials establish regulations  governing
the  use  of dyes or  any  other  ground-water tracer,  they should
require their use by knowledgeable,  experienced professionals.  If
it is  felt necessary  to give the  states regulatory authority in
tracing,   I  suggest   that  regulation  be  through   licensing  of
individuals rather than on a study-by-study basis.

     Many Federal and  State agencies have sanctioned  the use of
dye-tracing  studies  in the  study of ground-water  pollution and
time-of-travel of pollutants  in  rivers.   Several major  EPA docu-
ments endorse the use of tracers in ground-water  studies (US EPA,
1986, p.  64;  1987a, p. 2-12, 2-22 2-23; 1987b,  p.  127-148; Quinlan,
1989). Guidance manuals for tracing techniques exist and have been
sponsored by EPA (Davis et al..  1985; Mull et al.. 1988; Quinlan,
1991) and by  the Societe Geologique Suisse (Parriaux et  al. , 1988).
Updated manuals on ground-water  tracing have been written by the
U.S.  Geological Survey  (under  contract to  the EPA; Mull et al.,
1988) and are in preparation for the National  Ground Water Associa-
tion  (Aley et al.,  in prep.).  Several manuals on the  use of dyes
for  measurement of discharge, time of  travel,  and dispersion in
surface streams have  been written by the U.S.  Geological Survey
(Kilpatrick and Wilson, 1989; Wilson et al., 1986; Kilpatrick and
Cobb, 1985; Hubbard et al., 1982).

     International interest in  and  recognition  of  tracing as a
useful tool  in the kit  of the hydrogeologist is  suggested by the
content of proceedings volumes of the five International Symposiums
on Underground Water Tracing held during the past 25 years and by
the synthesis of Caspar  (1987), and by a recent meeting  on granular
aquifers (Moltyaner,  1990).  The sixth International  Symposium, to
                                690

-------
be sponsored by the Association for Tracer Hydrology, will be held
in Germany in 1992.

     It is desirable and necessary that proposed Federal ground-
water  legislation  include a  section that  sanctions the use  of
tracing agents,  prevents them from being regulated as pollutants or
contaminants, and strongly encourages their use in predicting the
possible consequences of pollutant release and dispersal.  Alterna-
tively, and  until  legislation is passed, a  policy  endorsing the
judicious use of tracers should be promulgated.


                      RECOMMENDED SOLUTION

     I suggest that legislation or policy concerning ground-
water protection include wording (originally developed by Quinlan
and Field, 1991) to the effect that:

     The  use of  organic  and/or inorganic  tracing  agents
     employed in investigations of ground-water and surface-
     water hydrology is  a desirable activity  and shall not be
     construed as the addition of a pollutant or contaminant
     to the water when deliberately  introduced as a tracer.
     Useful tracing agents include,  but are  not limited to,
     the following:  dyes  such as  fluorescein (Colour Index
     [CI] Acid Yellow 73),  eosin (CI Acid Red 87), Rhodamine
     WT  (CI  Acid Red 388) , Sulpho Rhodamine G  (CI  Acid Red
     50) ,  Sulpho  Rhodamine  B  (CI  Acid Red  52) ,  optical
     brighteners (fluorescent whitening agents with various CI
     numbers),  Diphenyl  Brilliant Flavine  7GFF  (CI  Direct
     Yellow  96) ,  Lissamine Flavine  FF  (CI  Acid  Yellow 7)  ,
     pyranine  (CI  Solvent  Green 7) ;  ions such  as  lithium,
     chloride, bromide,  iodide, nitrate, and sulfate; fluor-
     inated organic anions; Lycopodium spores; bacteriophages;
     noble gases;  other  gases such  as  sulfur hexafluoride;
     stable isotopes; particulates; hot water; and cold water.

     Sewage  effluent and  waste  products  from  industrial
     activity,  waste disposal,  or  spills may  function  as a
     tracing agent, but they are not suitable for deliberate
     addition to water as a tracer.

     Investigations regarding the vadose zone in fractured and
     karstic terranes can employ the  use of smoke and various
     gases (e.g., sulfur hexafluoride).  These can be used for
     determining subsurface interconnections  above the poten-
     tiometric  surface  when  light nonaqueous  phase liquids
     (LNAPLS)  and   other  highly volatile contaminants  are
     residing in the vadose zone.

     The U.S. Environmental Protection Agency, U.S. Geological
     Survey, American Society for Testing and Materials, and
     professional societies of hydrologists, hydrogeologists,
     and  engineers are  authorized  to  establish  voluntary
                                691

-------
     protocols for the selection and use of tracing agents.

     Tracing should be conducted only by those  individuals who
     can demonstrate evidence of substantial training and/or
     experience in the use of tracer agents. The training may
     have been acquired through  academic  study at a univer-
     sity, courses  offered by  professional societies,  or by
     experience gained while  working under the supervision of
     an experienced water-tracing professional.


                           CONCLUSIONS

     The importance of establishing a  set of  regulations for the
use  of  tracers   in the  environment,   particularly the  use  of
fluorescent dyes in ground water cannot be overemphasized.  Their
utility  in  defining  the  true  flow paths  and  flow  rates  are
invaluable, especially in pollution studies.

     Failure  to  develop  coherent  regulations will continue  to
result in confusion regarding the alleged  toxicity of the dyes and
other  tracer  substances,  their  use  by  professionals  versus
amateurs, and their importance in the realm of  ground-water quality
monitoring.   It  is  essential that the  Federal government support
the appropriate use of water-tracing substances so that much of the
confusion regarding their proper use can be overcome.
                            REFERENCES

Aley, T., Quinlan, J. F., Alexander, E. L., Jr., and Behrens, H.,
     [in prep.].   The  Joy of  Dying:  A Compendium  of Practical
     Techniques  for  Tracing  Ground Water,  Especially  in Karst
     Terranes.  National Ground Water Association, Dublin, Ohio.

Davis, S. N. ,  Campbell,  D. J. ,  Bently, H. W. ,  and Flynn, T. J.;
     1985.  Ground Water Tracers.  National Water Well Association,
     Worthington, Ohio.  200 p.   [Reprint of report prepared under
     contract  to the  U.S.  EPA. ;  there are  many errors  in the
     chapter on dye tracing, p.  97-118; see discussion by Quinlan
     (1986)]

Drew,  L. J. ,  1990.    Oil  and  Gas  Forecasting.    International
     Association for Mathematical Geology, Studies in Mathematical
     Geology, No. 2.  Oxford, New York.  252 p.

Field, M. S., Wilhelm,  R. G., and Quinlan,  J-  F., 1992.  Selection,
     toxicity,  and  use  of fluorescent dyes for  tracing ground
     water:  1. What  is safe,  why,  where,  and when? [submitted to
     Ground Water]

Caspar,  E.,  ed.,  1987.  Modern Trends  in  Tracer  Hydrology.   CRC
     Press, Boca Raton, Fla.  2 v.  145 and 136 p.
                                692

-------
Hubbard,  E. F.,  Kilpatrick, F. A., Martens, L. A., and Wilson, J.
     F.,  Jr., 1982.  Measurement of time of travel and dispersion
     in streams by dye-tracing.  U.S.  Geological Survey, Techniques
     of Water-Resources Investigations,  Book 3, Chapter A9.  44 p.

Kilpatrick, F. A.,  and Cobb, E. D. , 1985. Measurement of discharge
     using tracers.  U.S.   Geological Survey, Techniques of Water-
     Resources Investigation, Book 3, Chapter A16.  52 p.

Kilpatrick, F. A. and J. F. Wilson, Jr.   1989.  Measurement of time
     of travel in streams  by dye tracing.  U.S. Geological Survey,
     Techniques of  Water-Resources Investigations, Book 3, Chapter
     A9.   27 p.   [Revision of Hubbard et. al. , 1982]

Moltyaner, G., ed.  1990.  Transport and  Mass Exchange Processes in
     Sand and Gravel  Aquifers: Field and  Modelling Studies.  Atomic
     Energy of Canada Limited, Ottawa.   AECL-10308.   2 v.  866 p.
     [Includes 46 papers presented at an international conference
     and  workshop]

Mull, D.  S., Liebermann, T. D.,  Smoot,  J. L., and Woosley, L. H.,
     Jr., 1988.   Application of dye-tracing techniques for deter-
     mining  solute-transport  characteristics of  ground  water in
     karst terranes.  U.S.  Environmental Protection Agency, Region
     IV,  Atlanta, Ga.,  EPA 904/6-88-001.  103 p.

Parriaux, A., Liszkay,  M. , Miiller, I.,  and della Valle,  G. , 1988.
     Guide  practique pour  1'usage  des  traceurs artificiels  en
     hydrogeologie.     Societe   Geologique   Suisse,   Groupe  des
     Hydrogeologues.   GEOLEP  EPFL, Laussane.   51 p.   [bilingual;
     also  published  in the  same  volume as:  Leitfaden  fur  den
     Gebrauch Kvinstlicher Tracer in der Hydrogeologie.  49 p. ]

Quinlan,  J. F.,  1986.  Discussion of  GROUND  WATER TRACERS by Davis
     et al. ,  (1985), with  emphasis on dye-tracing,  especially in
     karst  terranes.   Ground Water.   v.  24,  p. 253-259  and p.
     396-397.  (Reply by Davis:  V. 24,  p. 398-399)

Quinlan,  J. F.,  1989.  Ground-water monitoring in karst terranes:
     Recommended protocols and  implicit assumptions.   Internal
     report  prepared for  U.S.  Environmental Protection Agency,
     Environmental  Monitoring  Systems  Laboratory,  Las  Vegas,
     Nevada.  EPA/600/X-89/050.   88 p.  [In review; to be published
     1992]

Quinlan,  J. F.,  1990.  Special problems  of ground-water monitoring
     in karst terranes.  Nielsen, D. M., and Johnson, A. I., eds.
     Ground  Water  and  the Vadose  Zone.   American  Society  for
     Testing and Materials,  Special  Technical  Paper 1053.  ASTM,
     Philadelphia,  Penn.  p. 275-304.

Quinlan,  J. F.,  1991.  Delineation of Wellhead Protection Areas in
     Karst  Terranes,  with  Emphasis  on   Puerto  Rico.    Quinlan  &
     Associates, Nashville, Tennessee,   ca.  10 p.  [Course Manual]


                                693

-------
Quinlan, J. F.,  and M.  S. Field,  1991.   Use of dyes for  tracing
     ground water: Aspects of regulation.  Internal report prepared
     for  U.S.  Environmental  Protection  Agency,   Environmental
     Monitoring Systems Laboratory, Las Vegas,  Nevada.   EPA/XXX/X-
     XX/XXX.  7 p.   [In review]

Sauter, M., 1991a.  Double porosity models in karstified limestone
     aquifers: Field validation and data provision.   International
     Association  of Hydrogeologists,  Karst  Symposium  (Antalya,
     Turkey, 1990), Proceedings.   27 p.   [in press]

Sauter, M., 1991b.  Assessment of  hydraulic conductivity in karst
     aquifers at  local  and regional scale.   Quinlan,  J.  F. , and
     Stanley, A.,  eds.    Hydrogeology,  Ecology,  Monitoring,  and
     Management of Ground Water in Karst Terranes  Conference  (3rd,
     Nashville), Proceedings.  National  Ground Water Association,
     Dublin, Ohio.   [in press]

Scarrow, J. W., 1989.  The use of  modeling in groundwater  contam-
     ination cases.  Virginia Environmental Law Journal, v. 9, p.
     185-205.

Smart, P- L. ,  1984. A review of the toxicity of twelve fluorescent
     dyes used in  water tracing.   National Speleological Society,
     Bulletin.  v. 46, p. 21-33.

Teutsch, G., 1989. Groundwater models in karstified terranes:  Two
     practical  examples  in  the  Swabian Alb,  southern Germany.
     Solving Ground-Water Problems with Models (4th,  Indianapolis) ,
     Proceedings.  National Water  Well Association,  Dublin, Ohio.
     p. 929-953.

Teutsch, G, and M. Sauter.,  1991.  Groundwater modeling in karst
     terranes:  Scale effects,  data acquisition, and field  valida-
     tion,  in Quinlan,  J.  F., and Stanley, A.,  eds.   Hydrogeology,
     Ecology, Monitoring, and Management of Ground  Water in Karst
     Terranes Conference  (3rd, Nashville), Proceedings.  National
     Ground Water  Association, Dublin, Ohio.   [in press]

U.S.  Environmental Protection Agency,   1986.   RCRA Ground-Water
     Monitoring Technical Enforcement Guidance Document.  Office of
     Waste  Programs Enforcement  and Office  of  Solid  Waste and
     Emergency Response, Washington, D.C.  319 p.

U.S.  Environmental  Protection Agency,  1987a.    Guidelines for
     Delineation  of Wellhead Protection  Areas.  Office  of  Ground-
     Water  Protection, Washington, D.C.   EPA 440/6-87/-010.  204 p.

U.S. Environmental Protection Agency, 1987b.  Groundwater Handbook.
     Robert S. Kerr Environmental  Research Laboratory,  Ada, Okla.
     EPA/625/6-87/016.    212  p.     [The  section  on ground water
     tracers,  p.   127-148,  is  adapted  from Davis  et  al. .  1985;
     republished as:  Barcelona, M. , Wehrman, A., Keely, J.  F., and
     Pettyjohn, W. A.,  1990. Contamination of  Ground Water:   Pre-
                                694

-------
     vention, Assessment,  Restoration.   Noyes  Data Corporation,
     Park Ridge, N.J.  213 p.]

Wilson,  J.  F.,  Jr., Cobb,  E. D.,  and  Kilpatrick, F.  A.,  1986.
     Fluorometric procedures  for dye  tracing.    U.S.  Geological
     Survey, Techniques for Water-Resources  Investigation, Book 3,
     Chapter A12.  34 p.
                        ACKNOWLE DGEMENTS

     This paper has  been  critically reviewed by  Thomas Aley,  E.
Calvin Alexander,  Jr. ,  and A.  Richard Smith,  each  of  whom made
substantial improvements in its content and clarity.  The inval-
uable contributions of Malcolm S. Field have helped make  this paper
what it is and are gratefully acknowledged.


                       BIOGRAPHICAL SKETCH

     Dr.  James  F. Quinlan  is a  consultant specializing  in the
hydrology of karst aquifers and techniques  for tracing ground water
with fluorescent dyes.   He  has more than 36  years  of  experience in
karsts of 26 states and 23 countries.  For  16 years he  directed a
hydrologic  research  program  for the  National  Park Service  at
Mammoth Cave National Park, Kentucky.  He has written  or  co-written
more than 170 publications on ground-water monitoring,  evaluation
of waste disposal sites, and spill response in karst  terranes.  In
1985  he  and  Ralph  Ewers  received  the  Burwell  Award  from the
Geological Society of  America for a pioneering  paper on ground-
water monitoring  in  karst terranes.  He is a chairman of an ASTM
Task Group on  ground-water  monitoring in karst and other fractured
rocks, has  been  a Director of the  Association  of  Ground Water
Scientists  and  Engineers,  has  organized  numerous   scientific
meetings and,  since 1986, has annually  co-taught an N.G.W.A. short
course on practical karst hydrogeology.

James F.  Quinlan
Quinlan & Associates, Inc.
Box 110539
Nashville, TN  37222
(615) 833-4324
                                695

-------
USE OF DYES FOR TRACING GROUND WATER:  ASPECTS OF REGULATION

By:  James F. Quinlan


Both Missouri and Kentucky are in the process of establishing dye-
trace data bases.   This will require sharing of tracing results.
Do you/ as a  consultant,  support this  type of additional regula-
tion?

Assuming  the data  base  is  available to  all,  yes,  absolutely.
Everyone benefits.


Has there been extensive acute and chronic testing of tracers?

Yes, but there is never as much  as one might wish for.  The best
review of this topic  is by Pete  Smart  in the NSS Bulletin  (1984.
46(2):21-33).


I have heard  that it  is dangerous to use Rhodamine WT because it
forms  nitrosamine  compounds  which  are carcinogenic.   Would you
please comment on this issue.

The  concern  about nitrosamines  from Rhodamine WT was  raised by
Abidi (1982,  Water Research,  v. 16, p. 199-204).  She reported that
diethylnitrosamine  (NDEA)   formed  in  river waters treated with ni-
trite ions (NO2~)  and  either  Rhodamine  WT or B.  Nitrite is  an un-
stable intermediate ion formed in the reduction of nitrate ions or
the oxidation of ammonium  and organic nitrogen.  Nitrite ions are
found, in small concentrations, primarily in waters heavily impact-
ed by animal  wastes  such as the discharge from  trout-rearing ponds
or  sewage  wastes.   Nitrite  levels  are  very  low in  most  ground
waters, even those with elevated  levels of nitrate ions.

Steinheimer and Johnson (1986, U.S.G.S. Water Supply Paper 2290, p.
37-49) did a  very detailed and  careful follow-up on Abidi's work
and found that in four different  river waters  "nitrosamine  forma-
tion did not  occur in the river samples  at concentrations typically
encountered  during  dye injection studies."   They  then concluded
"under customary dye-study practices, NDEA resulting from the use
of Rhodamine WT does not constitute  an environmental hazard."

No diagnostic tool is without risk,  however.  An x-ray examination
of a broken  bone  carries  a measurable  risk of inducing a cancer,
for  example.   The  advantage of  diagnosing and  setting the bone
correctly outweighs the risk and  diagnostic tool.  It should only
be used  when there  is a need for the  information  it will  yield.
The danger associated with long-term chronic pollution of a water
supply will  usually outweigh any risk associated with an improb-
able,  transient, potential exposure to a reaction product  of the
dye.   The dye is  not  the  problem here, it is  the conditions that
necessitate the dye trace.
                                696

-------
                                            This paper is included
                                            within the Proceedings of the
                                            National Cave Conservation
                                            Association
        THE EFFECTS  OF  RECHARGE BASIN LAND-USE PRACTICES

                       ON WATER QUALITY AT

              MAMMOTH CAVE  NATIONAL PARK,  KENTUCKY
                            Joe Meiman
                   Mammoth  Cave National Park
                     Mammoth  Cave,  Kentucky
                             ABSTRACT

A water quality monitoring program was designed at Mammoth Cave
National Park to determine if there exists any influence on the
water quality of the Mammoth Cave karst aquifer within the park
from various land-use practices of the recharge area.  These  land
uses primarily include: heavy agriculture  (row crops and
livestock),  logging, oil and gas production, and residential
areas.  The program, initiated in March 1990 and extending
through September 1992, samples two rivers, and eight springs
recharged by lands with varying land-use.  Monthly non-
conditional synoptic sampling monitors 36 parameters, including
site discharge.  The first 19 months of data demonstrate a strong
correlation between drainage basin land-use and water quality.
Contaminant entrainment mechanisms and relative pollutant input
rates can be discerned when the mass-flux of selected parameters
is calculated.  By use of these data, effective resource
management decisions can, and are being made to conserve and
protect the irreplaceable natural resources of Mammoth Cave
National Park.
                           INTRODUCTION

For what purpose do we monitor the quality of water at Mammoth
Cave National Park?  Aside from pure stoichiometric data  to
satisfy our curiosity of the water's chemical composition,
spatially and temporally, the fundamental mission of this
monitoring program is to better understand, and thus better
manage, the aquatic natural resources  of the park.  During the
three year course of this program, data will be collected and
interpreted to provide information on  the current state of the
surface and subsurface water of the park.  This data set  will be
                                697

-------
used as a datum from which to compare past and future studies.
As the author is not a biologist, no claims, speculations,
conjectures, or theories pertaining to the present health or
future of the aquatic ecosystems will be made.  However, before
trained personnel can accurately assess the condition of the
park's aquatic life, a broad database of the physical and
chemical properties must be available.

Although this phase of the monitoring program is far from
complete, there appear to exist a few trends and correlations
which deserve mention.  The following pages will concern the
first nineteen rounds of monthly sampling.


             BRIEF DESCRIPTION OF MONITORING PROGRAM

The monitoring program is largely based upon synoptic samplings.
Synoptic, as defined by Webster, is "relating to or displaying
conditions as they exist simultaneously over a broad area".
Although the water quality monitoring program includes two
different synoptic approaches, conditional and non-conditional,
the later comprises by far the bulk of monitoring activities for
the first years of the study.  The program also includes topical
sampling which provides a detailed evaluation of a particular
flow condition, contaminant, basin or river reach.

Choosing synoptic stations within a karst aquifer differs greatly
from the same task preformed on a surface drainage.  In a surface
drainage one can choose sites based upon stream reaches (every 20
miles for example) to improve spatial distribution, or install a
station exactly where a known pollutant source is located. The
monitoring sites of this program were chosen with respect to
land-use practices of the various recharge basins.  These
practices range from the naturally wooded park-land groundwater
basins where human influence has been absent for at least 50
years, to highly agricultural lands with a share of urban use and
oil and gas exploration.

The ten non-conditional synoptic stations are sampled, regardless
of flow or weather conditions, on the 10th of each month for the
duration of the study (Figure 1).  The sites are sampled during a
single day by one field crew.  The need for repetitive sampling
(each month for three years) at each non-conditional synoptic
station arises from the considerable temporal variability of
karst and surface water quality.  This variability is largely a
result of sudden changes in discharge, and seasonal availability
of contaminant sources.  Over the course of the study each end of
the flow continuum  (base and flood conditions) and each growing
season will have been encountered several times, as by program
design.

The primary use of conditional synoptic surveys is to provide a
finer degree of spatial resolution to the descriptions of
discrete water quality and flow conditions than would be


                                698

-------
attainable from the non-conditional synoptic station network.
One of the goals of the conditional synoptic surveys is to
identify relatively short reaches of drainage basins which have
demonstrated (by data from the non-conditional synoptic station
network)  chronic water quality problems.

Although the program is designed to allow the park to determine
the effects of la'nd-use practices on water quality after three
years of sampling,  we can, at this juncture, observe various
traits which may be attributed to types of land-use in the
recharge areas.  Each time a sample is extracted, discharge at
the site is recorded.  Parameter concentration, coupled with
discharge will yield flow-weighted values.  These values will
allow us to determine the mass flux (loading) of a particular
parameter at various flow conditions.   These data will allow us
to better determine contaminant source as it pertains to
constituent availability, release, and entrainment into the
water, and mechanisms of transfer from the surface to the
subsurface.
       A DISCUSSION CONCERNING MASS  FLUX AND  FLOOD  PULSES

It would be difficult to continue this discussion without first
examining mass flux and flood pulses.  Mass flux is simply the
amount (mass)  of a particular parameter passing a point in a
given time interval (flux).   A flood pulse is the portion of
water propagated along a channel and/or conduit as result of a
recharge event,  most commonly, rainfall.  With an understanding
of flood pulse movement, one might better understand mass flux
signatures of various contaminants.
MASS FLUX

If a contaminant is released into a stream of water at a constant
rate, and at some point downstream its concentration and the
stream's flow can be measured, the mass flux of the contaminant
can be calculated (Figure 1).   If flow (Discharge 1) decreases or
increases, the contaminant's concentration will proportionally
increase or decrease, respectively (Concentration).   That is, at
times of high flow the contaminant will experience a greater
amount of dilution.   The resultant mass flux signature (mass flux
over time) of a constant source release will consist of a
relatively low amplitude disturbance (Mass Flux 1).   One may
think of this mass flux signature as a type of destructive wave
form interference.

Suppose a contaminant is released into a stream only when
specific hydrologic conditions are met, a rainfall event of a
certain intensity and volume for example.  Therefore, if flow
(Discharge 2) increases, contaminant concentration also increases
(Concentration) as these stores are displaced into the streams
during flood pulse activity.  That is during the times of peak
                               699

-------
                      MASS  FLUX
           O
           Z
           CO
           <
           LJ
           LT
           o
                            MASS FLUX 2 \
                            MASS FLUX 1


                           CONCENTRATION
                            DISCHARGE 1
                               I
                             TIME 	-
     Figure 1. Hypothetical mass flux signatures.

flow, peak (or near peak) concentrations also occur.  The
resultant mass flux signature of a precipitation-triggered
release will be of relatively high amplitude, perhaps several
orders of magnitude greater than the pre-pulse mass flux (Mass
Flux 2).   This mass flux signature may be likened to constructive
wave form interference.
FLOOD PULSES

Flood pulses in the Mammoth Cave area may raise a basin's
discharge a couple liters per second following minor rainfall, to
several thousand liters per second after major rainfall.
Research by Meiman (1988 and 1989) has demonstrated that a flood
pulse is comprised of two chemically and physically distinct
components: displaced stores and freshly input recharge.  The
former, which usually occurs as the leading edge of a flood pulse
and is characterized by high specific conductances, can be
thought of as easily displaced vadose storage.  Freshly input
recharge, which comprises the bulk of a flood pulse, is
characterized by low conductances, as there is little time for
interaction between its waters and ionic sources.  These
relationships are also manifested in water temperature, as
displaced stores, with longer residence times, will reflect the
antecedent system temperature, and freshly input recharge
correlative to surface temperature (Meiman, 1988 and 1989).
                                700

-------
Consider the flood pulse displayed in Figure  2.   This  pulse,
documented over approximately 42 hours  in the fall  of  1987  at a
sinking creek of the Turnhole Spring groundwater  basin,  clearly
indicates the arrival of three highly conductive  sources during
the course of flood pulse activity.  The majority of flow
generated from this rainfall event was  of the low-conductance,
run-off variety.  If water provenance suddenly changes,  one may
expect to see a similar change in water quality with respect  to
available water-borne constituents.  Hallberg, et al  (1985)
identified an acute, albeit brief, water quality  degradation
associated with this run-off component  in the karst of
northeastern Iowa.
WATER QUALITY, EXPRESSED BY MASS FLUX, AS IT
RELATES TO FLOOD PULSES

Important water quality information may be gained  if  knowledge of
flood pulses is combined with mass flux signatures.   As  rainfall
occurs, flood pulses are generated and propagated  through  the
karst aquifer.  Just as stage may suddenly vault from its  base
condition, water quality may also undergo rapid and drastic
change as a flood pulse passes.  If a significant  amount of
     70 r
     60
(n 50

I

W 40
    30
     20
     10
                                                stage
                                                            450
                                                            400
                                                            350
                                                            300
                                                            250
                                                            200
                                                            150
               100
                                                             50
                                                                en
                                                                •
                                                                m
                                                                O
                  o
                  o
                  o
                  o
                  o
                  o
                  m
                  c
                  3
                               12

                            TIME IN HOURS
24
Figure 2. Stage and specific conductance responses  of  a flood
          pulse in a surface stream, Turnhole  Spring groundwater
          basin.
                                701

-------
constituents are released  by  the precipitation event (entrained
in run-off), the mass  flux of these elements may rise
tremendously.

Contemplate the two hypothetical mass  flux signatures of
Figure 3.  It  is vital  to  note the  relative,  unitless scales of
the two graphs.  Although  the X-axis scales are equal,  the Y-axis
of 3a is I/8th that of  3b.  Also note  that "Time 0"  indicates the
advent of precipitation.   The same  discharge hydrograph is
employed for both graphs.   Remember, data  used in these graphs
are hypothetical.  Numbers were derived  by a noting  the timing,
duration, and wave-form characteristics  of years of  continuous
data  (stage, specific conductance,  water temperature and
discharge) and months of water quality data.   Mass fluxes are
actual products of discharges and concentrations.

If a constant source parameter is monitored through  a flood
pulse, oil-field brine  chlorides from  a  leaking well casing for
example, a response similar to that of Graph  3A might be
expected.  Following an initial  upward spike  in concentration,
perhaps caused by a flushing  of  vadose stores,  chloride
concentration is diluted as the  pulse's  freshly input recharge
component dominates the flow.   The  resultant  mass  flux  signature,
although not without structure,  displays relatively  low amplitude
disturbance.

If the same discharge is used with  a precipitation triggered
release parameter, certain  pesticide residues for  instance,  a
totally different mass  flux signature  results (Graph 3b).   Again,
note the Y-axes scales.  The  effects of constructive wave-form
           CONSTANT SOURCE
                                           PRECIPITA TION- TRIGGERED
                                                             3B
      -20  2  4  6 8 10 12 14 16  18 20 22

               TIME IN HOURS
-20 2 4  6  8  10 12 14 16 18 20 22

         TIME IN HOURS
Figure 3.  Hypothetical mass flux signatures of constant source
          release (3A) and precipitation-triggered release  (3B).
                                702

-------
interference are noticeable as a high amplitude mass flux
signature is generated.  The passage of the leading edge of the
pulse may be reflected in a sharp drop in pesticide residue
concentration,  as long-residing stores are displaced.  This
effect will be far overshadowed by the arrival of the freshly
input recharge.  Not only does this flow component comprise the
majority of the flood pulse discharge, it also contains the bulk
of surface run-off with entrained herbicides.  The resultant mass
flux signature may be several orders of magnitude higher than
pre-pulse values.

There are many factors that may control the shape of the mass
flux signature:  availability of constituents, entrainment
method, transfer mechanism from surface to subsurface, rainfall
volume and areal distribution, time since last rainfall, and
conduit condition, to name a few.  It should be noted that the
conduit condition used in this example is highly vadose.  A
different signature, especially with respect to temporal lags of
concentrations and discharge peaks, will occur when dealing with
a phreatic conduit system (Meiman 1988, 1989).  Current research
at Mammoth Cave specifically addresses flood-pulse water quality.

Perhaps by close examination of mass flux signatures of fecal
coliform bacteria, dominant waste sources, human or animal, may
be discerned.  Human waste,  for the most part, should behave as a
constant release source.  Human waste is injected directly into
the aquifer via leach fields, leaking septic tanks, or dry-wells,
at a relatively constant rate.  A constant mass flux signature
should result.   Animals, not nearly intelligent enough to
defecate down wells, will deposit waste on the surface.  Without
rainfall (or a major snow-melt), this waste will not be
transferred into the aquifer.  Following a significant recharge
event, animal waste will be washed into the aquifer, producing a
mass flux signature characterized by a very high amplitude
disturbance.
          RESULTS OF MARCH  1990 THROUGH  SEPTEMBER  1991

The following data (based upon non-conditional samples)  run from
March 1990 through September 1991.  The summer months, which
comprise a disproportionally large percentage of the data, will
skew the data toward low-flow conditions.  Non-conditional
synoptic sampling covers a wide spectrum of flow conditions,
ranging from flood-pulses to flow-reversals.  If river water is
back-flooded into the spring, it is considered to be
representative of the spring's water at that moment in time.  The
sample will be taken and analyzed regardless of water provenance
(river or cave derived).  The aquatic communities of the spring
and related conduit must live in the waters, regardless the
source, therefore the sample is representative of their
environment.
                                703

-------
The presentation  of  water quality data in the following
discussion will be in  two forms:  statistical graphs  (bar and
whisker) and XY graphs depicting trends at selected  sites of
selected parameters.   The four selected sampling sites for this
document are:  Light agriculture (Pike Spring, PSPS), Park/heavy
agriculture  (Echo River Spring,  ERES), Heavy agriculture
(Turnhole Spring  area,  THNS),  and Park lands  (Buffalo Creek
Spring, BCGR).
DISCHARGE

Discharge depends,  of  course,  upon precipitation events.  The
summer and fall months are  traditionally characterized by low
discharge, with higher discharge through the winter and spring
(Figure 4).  Overall the  largest discharges during sampling
occurred on April  10,  1991.   On this date flood-pulse activity
was high as the aguifer quickly responded to the rains of the
previous day.  This sampling round is of specific importance as
samples were extracted near pe^k discharge times of the flood-
pulses.  Although  other rounds saw relatively high discharges,
samples were, as dictated by monitoring program, taken either
well before or well after pulse peaks.

During the first nineteen months of the study a major back-
flooding event was sampled  on June 10,  1990.  At this time all
       10000
       8000
    Q
    •z.
    8  6000
    UJ
    GO
    01
    D.

    a:  4000
    LU
       2000
-•4-- HEAVY AGRICULTURE
-•— LIGHT AGRICULTURE
-f-	 PARK / HEAVY AGRICULTURE
-••--	 PARK LANDS
           MAMJJASONDJFMAMJJAS
                            MONTHS  1990 - 1991
Figure 4. Discharge  values,  March 1990 through September 1991, of
          four monitoring sites.
                                704

-------
springs,  with the exception of THNS, were  in a state of flow
reversal  - water from the Green River  flowing back into the
aquifer.

Notice that Echo River Spring is  referred  to as "Park/heavy
agriculture".  During times of high discharge,  flow from the
heavy agriculture basin is shunted through a high-level overflow
route into the Echo River basin,  which is  normally recharged by
park lands.  When this route is activated,  water quality in Echo
River may, nearly instantaneously, degrade.   Research in the next
year will document the conditions needed to conduct flow through
the overflow route.
TURBIDITY

Turbidity, correlative to the amount  of  suspended sediment in the
water, is highly variable through all non-conditional synoptic
sites (Figure 5).  As expected, basins dominated by agricultural
land-use, discharge more turbid waters than  those dominated by
undisturbed forests.  Generally one would  expect the higher
turbidities associated with areas of  high  soil  loss.  Although
the Turnhole basin  (Heavy agriculture) contains by far the
greatest area of tilled crop-lands, its  turbidities, albeit high,
were not the highest recorded; that honor  goes  to the light
agriculture Pike Spring basin.  Although containing far fewer
        90 p


        80 i-


        70


        60


        50
--- HEAVY AGRICULTURE
— LIGHT AGRICULTURE
	 PARK / HEAVY AGRICULTURE
	 PARK LANDS
          MAMJJASONDJFMAMJJAS
                            MONTHS 1990 - 1991
Figure 5. Turbidity values, March  1990  through September 1991, of
          four monitoring sites.
                                 705

-------
acres of tilled land, the rugged topography of the Pike Spring
basin amplifies soil loss when disturbed.

Displayed in turbidity are the back-flooding and overflow
described in the preceding section.  The back-flooding event of
June 1990 is evident in turbidity, as back-flooded springs
display turbidities close to that of the Green River.  The Echo
River basin (Park/heavy agriculture) exhibits low turbidities,
associated with low to moderate discharges when the spring is
recharge by park lands, and high turbidities when the overflow
route from the heavy agriculture basin is activated.
CHLORIDE

Chloride may be indicative of animal/human waste and oil field
brines.  Figure 6a shows the chloride concentration trends of the
four selected sites, while Figure 6b demonstrates the mass flux
of the chloride ions.  Both graphs exhibit interesting data.
Figure 6a shows elevated concentrations of chloride in the heavy
agriculture basin.

Oil field brines seem the prime suspect for two reasons: presence
of associated brine ions, and mass flux signatures.  Within the
headwaters of the basin, and adjacent to the Park City oil-field,
is Parker Cave.  On a low-flow conditional synoptic survey
(September, 1990), Parker River (a stream passage within Parker
Cave) had chloride levels of 1476.1 ppm.  Further down-basin,
Mill Hole chloride was 59.9 ppm, as the Parker River water was
diluted.  At the basin's terminal spring, chloride was further
diluted to 31.6 ppm.  Bromide and sulphate, also suggestive of
brines, were found decreasing at similar rates at the same sites.
Similar results were reported by Meiman (1989), and Quinlan and
Rowe (1978).

The mass flux signature of chloride may also indicate brine
contamination instead of animal waste.  Figure 6a displays a
variable, yet predictable pattern of chloride concentrations.
One may assume that the chloride source is of relatively constant
delivery as chloride concentrations are higher during low flow
periods of summer and early autumn, and lower, more dilute,
during the high flow periods of winter and spring.  Figure 6b
indicates an apparently dramatic increase on the mass flux of
chloride during months of high discharge.  This increase, some
eight times the mass flux of low discharge periods, may not be as
severe as it may seem.  This variation may be normal, even for
this relatively constant source parameter.  One might expect a
much greater (several orders of magnitude) rise in the mass flux
signature if a source is released by run-off from a precipitation
event.

A certain portion of chloride can be considered as a natural,
background concentration.  Observe the chloride trends of the
Park land basin.  Not only are chloride concentrations low
                                706

-------
         32 r
                                    -->-- HEAVY AGRICULTURE

                                    —•— LIGHT AGRICULTURE

                                    - V	 PARK / HEAVT AGRICULTURE

                                    	•	 PARK LANDS
       E
       Q-
       Q_

       Ld
       Q

       Ct
       O

       I
       O
           M  A  M   J   J
                     ASONDJ   FM

                          MONTHS  1990 - 1991
                                                   A  M  J  J  A  S
Figure 6a. Chloride  concentrations,  March  1990 through September
            1991, of  four monitoring  sites.
        40000
        30000
O
in

cr>


UJ
O
C£

3

O
        20000
        10000
           0
                      HEAVY AGRICULTURE

                      LIGHT AGRICULTURE

                      PARK / HEAVY AGRICULTURE

                      PARK LANDS
                     ^*

                     -T '
            MAMJJASONDJFMAMJjAS

                                MONTHS  1990 - 1991


Figure  6b. Chloride mass flux,  March  1990 through  September 1991,
            of  four sampling  sites.
                                    707

-------
(Figure 6a),  they remain at approximately the same mass flux
throughout the year (Figure 6b).   Upon closer examination, notice
the slight increase in mass flux through the winter months.
Although seemingly small and insignificant, the relative changes
between seasonal mass fluxes in the Park land and the Heavy
agriculture basins are very similar.  This trend may be an
inherent wave-form signature due to the vast increase in
discharge.

Road salts as a potential chloride source must be recognized.
Although the use of road salts have been prohibited within the
park since 1987, they are used throughout several of the park's
groundwater basins.  The amount of road salt contributing to
chloride levels found in parks waters is yet unknown.  Since the
first month of sampling there has not been a significant snow
fall to warrant the use of much salt.  Perhaps this unnatural
source will be manifested in a "unique" mass flux signature,
representative of seasonal application and recharge.
FECAL COLIFORM

Fecal coliform bacteria is found at all sampling sites
(Figure 7).  These bacteria are common in wastes of all healthy
warm-blooded animals.  By far, basins with high occurrences of
dairies, feed-lots and urban areas are characterized by high
levels of fecal coliform.

A certain amount of fecal coliform can be attributed to wildlife.
Note that the Park land basin (BCGR), representative of pristine
conditions, contains a fair amount of fecal coliform bacteria
(mean of 67 colonies per 100 ml),  and discharges a relatively
stable 1.5 million colonies per second (not shown).  Although the
latter number may appear high, a single gram of feces may contain
tens of millions to tens of billions of cells (Feachem, et al.,
1983) .

The heavy agriculture basin (THNS),  with hundreds of homes
without proper waste treatment facilities, and scores of dairies
and feed-lots where live-stock waste flows as sinking-creeks into
the aquifer yielded the highest overall fecal coliform levels.
Feachem et al. (1983) reports that although fecal coliform
density per gram of feces of man and livestock are comparable, a
human may excrete 150 grams of feces per day compared to 15 to 20
kilograms for a cow.  The highest flow weighted value, greater
than 535 million colonies/second,  was observed at this spring on
April 10, 1991.  As samples are taken on a set monthly date,
regardless of weather or flow conditions, over the three-year
period of monitoring some flood pulses are likely to be sampled.
April 1991 was such an occurrence.  It is important to note the
relative temporal position within the flood pulse from which the
sample was taken.  A great amount of variance in parameter
concentration may exist throughout high discharge periods of a
flood pulse.  It is not possible to tell from one sample its
                                708

-------
         5000
         4000
      E
      o
      o
      o  3000
      cc
      o
      U-
      o  2000
      o
      o
      UJ
         1000
            0
                                              maximum
                                           T  95 percentile
                         75 percentile


                         mean

                         med ian
                      LTJ 2.5 percenlile
                       -L  5  percentile
                       *  minimum
         ±
A
J.
i.
PSPS    ERES    THNS

       MONITORING SITES
                                             BCGR
Figure 7. Statistical examination of  fecal coliform bacteria
          levels, March 1990 through  September  1991,  of four
          monitoring sites.

temporal relationship to concentration or mass  flux peaks  of a
particular parameter.

The chance occurrence of flood pulse  activity coinciding with a
predetermined sampling date tends to  create a large variance in
reported concentrations and mass fluxes of fecal  coliform
bacteria.  A high variance may indicate, as in  the  heavy
agriculture basin, the presence of large amounts  of animal waste
stored at the surface, awaiting release by a rainfall event.

Notice that low values dominate the data set in the park
land/heavy agriculture basin (ERES).  Occasionally  these low
levels of fecal coliform are interrupted by brief periods  of very
high concentrations, as expressed in  the elevated mean and
maximum values.  Recall the overflow  route mentioned earlier.
During high-stage times, a portion of the bacteria-laden waters
of the heavy agriculture basin are shunted into the relatively
clean park basin.
                                709

-------
TRIAZINE-CLASS HERBICIDES

The monitoring program indicates the presence of triazine-class
herbicides (greater than 1 part per billion)  within the  surface
and ground waters of the park.   The occurrence of these  compounds
generally coincided with the peak application period.  To  avoid
costly organic laboratory testing for these compounds, the
program employs assay screening tests.   Although gas
chromatography analysis would indeed be desired,  laboratory  costs
of a couple of sampling rounds would destroy the monitoring
budget.  Assay-screening can not be thought of as a quantitative
analysis.  It is used primarily as a "hit-or-miss" technique,
with semi-quantitative values (ie, greater than 1 ppb).

The spatial and temporal occurrence of triazine-class  herbicides
reflect land-use, herbicide application periods,  and perhaps the
mechanism of transfer between the surface and subsurface
 (Figure 8).  With the exception of a back-flooding of  triazine-
tainted river water, the only groundwater sampling site  in which
triazines were found was the heavy agriculture basin (THNS)
spring.  Additionally, triazines were only found in months (June
1990, June and July 1991) following peak application periods
within the basin.  Triazines are also found at both river  sites
following peak application.

For the remainder of the year no triazine-class herbicide
residues were found in the sampled springs.  Although  rapid


              TRIAZINE  CLASS HERBICIDES

               TRIAZINES ppb
A M
                 I  I  • I • I ' I  • I
                 J  J  A 8  O N
                    1990
i   i  i  t   i  i
D  J  F  M A  M
MONTHS
T—^— II
J  J  A  S
1991
            PARK LAND BASIN

            GREEN RIVER
       HEAVY AGRICULTURE

       NOLIN RIVER
 Figure 8.  Temporal occurrences  of triazine-class herbicides using
           immuno-assay methods, March  1990 through September
           1991,  at four monitoring  sites.
                                710

-------
transport of these residues through the karst system is expected,
one may not assume that all, or even the majority of these
compounds that will move through the aquifer have done so.
Research in Iowa by Hallberg et al. (1985) found that although
large amounts of herbicides are quickly transported through the
karst system via run-off following rainfall, the bulk of these
materials are slowly released through infiltration in low
concentrations.  It would not be surprising to see a similar
pattern of pesticide transfer through the Mammoth Cave aquifer.

Aside from occurrence following peak application periods,
triazine-class herbicides were found in both the Green and Nolin
rivers in the fall of 1990, and possibly in the fall of 1991.
Two scenarios may be possible:  1)  There was a late application
of these compounds in the fall, or, 2)  The residues were slowly
transferred through a less permeable media (clastic strata).

As triazines are applied as pre-plant or pre-emergence
herbicides, there is no reason to believe that there were a late
applications, as crops that receive triazines (corn, and to a
lesser degree, soybeans) were near harvest.

River flood plains, with associated unconsolidated fluvial
deposits, are favored lands for row-crop production.  It may be
possible that these persistent compounds may: 1)  become entrained
in run-off shortly after application,  and 2) slowly infiltrate
through the fluvial materials and leached into the river
following fall rains.  The "half-life"  of these compounds  (3-12
months) is certainly sufficient to cause such persistency.
                        REFERENCES  CITED

Feachem, R.G., Bradley, D.J., Garelick, H.,  and Mara, D.D., 1983,
     Sanitation and disease, health aspects of excreta and
     wastewater management, Chapter 13: John Wiley and Sons
     Publ., p.199-242.

Hallberg, G. R.,  Libra, R. D.,  and Hoyer,  B. E., 1985, Nonpoint
     source contamination of ground water in karst — carbonate
     aquifers in Iowa: Perspectives on Nonpoint Source Pollution,
     U.S. Environmental Protection Agency, 440/5-85-001, p. 109-
     114.

Meiman,  J., Ewers,  R.O., and Quinlan,  J.E.,  1988; Investigation
     of Flood Pulse Movement Through a Maturely Karstified
     Aquifer at Mcimmoth Cave National Park,  a New Approach, NWWA
     Conference on Environmental Problems in Karst Terranes and
     Their Solutions, Nashville, TN, 1988, p 227-262.

Meiman,  J., 1989, Investigation of flood pulse movement through a
     maturely karstified aquifer at Mammoth Cave Kentucky:  M.S.
     thesis, Eastern Kentucky University,  343 p.
                                711

-------
Quinlan, J.F., and Rowe, D.R.,  1978, Hydrology and water quality
     in the Central Kentucky Karst:  Phase II, Part A:
     Preliminary summary of the hydrogeology of the Mill
     Hole Sub-basin of the Turnhole Spring Groundwater
     Basin:  Kentucky Water Resource Research Institution,
     Research  Report no. 109,  41 p.
                        BIOGRAPHIC SKETCH

Joe Meiman, a Kentuckian, is currently the Hydrologist at Mammoth
Cave National Park.  His B.S. in Geology (1985) and M.S. in
Geology/Hydrogeology (1989)  were earned at Eastern Kentucky
University.  At the present he is kept busy by directing
hydrologic research at the park, which includes:  water quality
monitoring, dye-tracing, three-dimensional schematic karst
aquifer modeling, research and development of monitoring
equipment, and the general quest of scientific knowledge.
                                712

-------
    The Effects of Recharge Basin Land-use Practices on Water
         Quality at Mammoth Cave National Park, Kentucky

                            Joe  Meiman

Your study is interesting and has some novel, admirable
interpretations of the data.  Having said that, however, I am
surprised that samples were not also collected frequently
(perhaps at 1- to 4-hour intervals)  and analyzed at several sites
throughout several storm events.  This is more important than
monthly non-conditional synoptic sampling—if you want to
understand what is happening in the groundwater basins.  The 3-
year monthly sampling program is incapable of having any
significant statistical validity because it is not tied into
analysis of storm events and has too short a duration.  There
already exist hundreds of analysis of samples collected weekly to
monthly at many of your sites by Jack Hess (Ph.D. dissertation,
1974), the U.S. Geological Survey, and others.  Before your
sampling began, the National Park Service was repeatedly advised
in writing, by specialists in monitoring water quality in karst
terranes of the existence of such data, and that sampling of
storm events could yield more useful data and insights than
monthly sampling.  Why has the Park Service ignored sampling and
characterization of storm events?
As I stated in the presentation, monthly non-conditional synoptic
sampling is only a portion of the Park's water quality monitoring
program.  Indeed, storm event sampling will comprise a large
measure of the monitoring during the coming year, as was
discussed.  I am currently using storm-event (flood-pulse)  data,
recorded as often as every two minutes, in our present study.  As
work in this area continues, we will report our results when
sufficient flood-pulse water quality data have been collected.  I
am aware of the excellent study by Dr. Hess, who primarily
focused upon carbonate chemistry variations during flood-pulse
activity.  We have expanded our scope of parameters to include
bacteria, nutrients and brine-associated ions.   Synoptic data
generated by the studies of the U.S. Geological Survey during the
1950's and 1960's are rather spatially limited.  We have
increased sampling sites to capture data related to the myriad of
lands uses within the recharge basins.  Past works serve as
skeletal datums of which data of the current study will be
compared.  Temporal comparison is the essence of monitoring.  Be
assured that the National Park Service has not entered this
program without careful consideration of past work in the region,
or the dynamic nature of the karst aquifer.
                               713

-------
714

-------
         THE  SINKHOLE COLLAPSE  OF  THE  LEWISTON,  MINNESOTA

               WASTE  WATER  TREATMENT  FACILITY  LAGOON


    Nancy O. Jannik, E. Calvin Alexander,  Jr,  and Lawrence J.  Landherr

              Department of Geology,  Winona State University
                          Winona, MN 55987-5838

      Department  of Geology and Geophysics, University of Minnesota
                        Minneapolis,  MN 55455-0219

          Minnesota Pollution Control Agency, Regional Director
                           Rochester, MN  55904
ABSTRACT

On February 20,  1991,  city workers discovered a  sinkhole  collapse  in  the
Lewiston,  MN waste water treatment facility (WWTF)  lagoons.  The  collapse
apparently occurred during the preceding few days  and drained an estimated
7.7 million gallons of partially treated effluent  into the  local ground
water system. A temporary dike was constructed to  isolate the sinkhole
from the rest of the lagoon.  Subsequent,  ad hoc  testing for coliform
bacteria and nitrates  did not detect  evidence of effluent from the lagoon
in nearby residential  wells.  Following a shallow (20  foot penetration)
geophysical investigation using ground penetrating radar  and an
electromagnetic survey,  the city decided to fill the  sinkhole and  to  erect
a dike around the collapse. The collapse was repaired in  May,  1991 and the
lagoon returned to full operation.

The 1991 Lewiston collapse follows the  nearby,  1974  and  1976 collapses  of
the Altura,  Minnesota  WWTF  lagoon   (Liesch,  1977;   Alexander  and  Book,
1984) .    Two of the  7  to 10 WWTF lagoons constructed  on  the  Ordovician
Prairie du Chien Group carbonates  in  the  southeastern Minnesota  karst
terrain  have  catastrophically  failed  in  less  than  20  years.   That
corresponds to a failure rate of over 20% for these million dollar WWTFs --
so far. The federal programs that cost-shared the  bulk of the construction
expenses for these  WWTFs  no  longer exist. The  cost  of potential  damages,
remediation,  and/or replacement of these WWTFs falls  directly on the  state
and local units of government.
                                  715

-------
INTRODUCTION

  Southeastern  Minnesota is an  active  karst area.   Geomorphic features
associated with the karst  include sinkholes, enlarged joints, numerous
springs, disappearing  streams,  cave systems, and  dry valleys.  There are
problems with ground-water quality ranging from occasional high levels of
selected parameters to  chronic  sub-standard drinking water conditions in
the hydrogeologically sensitive  area.

  The  region  is characterized by farms,  small towns, and a few moderate-
sized  cities.  Many  of  the community  centers  have  waste  treatment
facilities  that  consist of  a  series   of  settling ponds,  or lagoons.
Lewiston, Minnesota is  one  of  the small towns located within this karst
region.  This paper documents the  failure  of one of Lewiston's ponds due
to the instantaneous collapse of a  sinkhole.
PHYSICAL  SETTING
Topography

  Lewiston  is located in southeast Minnesota,  in Winona  County  (Figure 1).
The region  surrounding  Lewiston  is  characterized by gently rolling hills
and swales with local relief of about 20 m  (Figure 1).  Sinkholes and dry
valleys  are  evident at  the  surface.    A very thin  soil,   ranging  in
thickness from  0  to about 15 m,  covers  the bedrock.  The source material
for the soil is both residuum and/or glacial tills and loess.
                                                           I   QUADRANGLE LOCATION
 Figure  1.   Portion  of  the topographic map  in  the vicinity of  Lewiston,
 Lewiston, Minnesota Quadrangle.
                                  716

-------
Climate

   At  present,  the  area  has  a  temperate  climate,  with  a mean annual
temperature of 7.6°,  and an annual  precipitation of  75 cm  (NOAA,  1978).
This  region has experienced  climatic changes, most  recently, the changes
that  resulted in Pleistocene  glaciation.


Geology

   The region is underlain by  a  series  of  lower Ordovician  and Cambrian
sandstone  and  carbonate units  (Figure  2) .  The units were  deposited in
nearshore  to shallow-sea  environments,  and exhibit typical  vertical  and
lateral facies  changes associated with sedimentation during Transgression
and regression  of  the  shallow sea.   The strata dip gently to the southwest
towards the center of the  Hollandale Embayment  (Mossier and Book, 1984) .
The  units  of  most concern  are  the  Jordan  Sandstone and the  Oneota
Dolomite.
SYSTEM
SERIES
|
_i
UPPER CAMBRIAN
GROUP OR
FORMATION
NAME
PRAIRIE DU CHIEN GROUP
SHAKOPEE
FORMATION
ONEOTA
DOLOMITE
JORDAN
SANDSTONE
ST. LAWRENCE1
FORMATION
FRANCON1A1
FORMATION
KONTON &
GALESVILLE
SANDSTONES
EAU CLAIRE2
FORMATION
MT. SIMON2
SANDSTONE
PRECAMBRIAN3
SYM
BOL
Opm
4
*--

€1
.ft
f-i-.
|
do
€t
/**, "
/ /CO/ /CV
. A s'cr> / &/~,
-,*••' :,•••',• /A
'-.' ', * , ',-}
<7 -^->rx^f
• ;&[.••*( ,/&)
S«a?i»£Sik
-ii .^ — ".'^r... ~j~ -=
-±,-:..^---J-i
- L i. 1 'i -/
.,6T°77
:'• o > -G I./
•'Vfrif . ' 'mrrf
.•XT : '•••
:-T:G ••'•'> -':\
&s/:
a^si^^sE^
- 6 - . G- )
•.".•'" - G '
4-0 .; G -6 y
•"t: »>&ss
-------
  The Jordan Sandstone is Cambrian  in age  and  averages  30  m in thickness.
The Jordan is a  massive,  upward-grading,  fine- to coarse-grained  friable
sandstone.  Upward in the unit,  it becomes progressively more  indurated
with carbonate  and siliceous  cements,  first forming lenses  and concretions
and then well-bedded,  highly  lithified strata.

  The Ordovician Prairie du Chien  Group  conformably overlies the  Jordan
(Figure 2).  The   Prairie  du  Chien is composed of the Oneota  Dolomite and
upper Shakopee  Formation.   The  Oneota Dolomite  is  about  60  m thick,  and is
fine- to medium-grained,  thick-  to  thin-bedded to massive, with calcite-
filled vugs in the  upper  portion,  and minor chert nodules throughout the
unit (Mossier and Book, 1984).   Both drill  cores and outcrops  reveal that
the dolomite is highly jointed and  has undergone extensive solution.  The
dolomite is  vuggy to cavernous  particularly in  the  upper portion.

  The  Shakopee  Formation is  subdivided  into  the  lower New  Richmond
Sandstone member  and  the  upper Willow River Dolomite member.  The  latter
is not present  in the area and  is not  discussed further.  The  New Richmond
Sandstone of the Shakopee Formation  is a fine-  to medium-grained quartzose
sandstone with infrequent interbedded medium-grained arenaceous carbonate
beds.  This sandstone unit  averages about  6 m in thickness, is friable,
extensively  jointed, easily eroded,  and does not form many  outcrops.
Hydrology

  Surface flow is in small headwater channels of the Whitewater  and Root
Rivers.  The channels are characterized by meander development and easily
credible banks.   Some  surface  run off  flows  into sinkholes.   Regional
ground-water flow is east-northeast toward  the  Mississippi River.   Local
ground-water flow is toward discharge points such as small tributaries or
springs.

  Joints are  common throughout the  Jordan  Sandstone and  springs  in the
well-lithified portions  tend  to discharge directly from joints.   In the
more friable lower part, springs are often a combination of discrete flow
from joints and diffuse  flow  from  numerous  seeps.   The  Jordan is a major
source of water for  wells in the area.

  Only a  few  springs,  confined to discharge  from well-developed joints,
have been mapped in the Oneota.   Few wells in the area rely solely on the
Oneota as  a water  supply.   However, many older  wells are  open  holes
through the Oneota.   The New Richmond Sandstone member of  the Shakopee
Formation has  a few springs  which emerge  form the New Richmond/Oneota
contact.   The  New Richmond is  not a significant  aquifer in the area.
                                  718

-------
Karst   Features

  Numerous karst features such as sinkholes, enlarged joints, springs, dry
valleys,  and small  caves  have  formed  in the  Oneota  Formation of the
Prairie du Chien Group  (Figure 3).   Sediment-filled solution cavities are
common  features in  outcrops  and quarry walls  that expose the Oneota
Dolomite.   The karstification  of the  Oneota probably  began during the
Ordovician and has continued intermittently until the present.  The region
would  be classified as  fluviokarst  according  to  the scheme  used  by
Sweeting  (1973)  because both karst and  fluvial processes have contributed
to the  development of the  features  that are evident,  or that are  being
exhumed.

  Sinkholes are by far  the  most  dominant  karst feature.   Historically,  if
the holes are left in  the  natural state,  they  are either  fenced-in and
left to be  naturally vegetated, or several  have  been  used as  backyard
landfills. In the past,  several  have been filled  with debris and  soil and
then used as  farm  land.   Many of the sinkholes in the area  have developed
catastrophically,  often  in the spring of the year in response to unusually
wet conditions.   It appears  that  the sinkholes  develop through the New
Richmond Sandstone into  the underlying  Oneota  Dolomite.
                  sinkholes
dry valley
      .JORDAN • •.'.' .-,-  •. ••
                                                            springs
         Typical relationships in Winona County between karst and topography
Figure 3.  Block  diagram of  karst  landforms  in  southeast  Minnesota
            (Dalgleish and Alexander, 1984).
                                  719

-------
  Lewiston,   and  the  immediate  vicinity,   were  classified  as  high
probability  of  sinkhole development  by Dalgleish  and Alexander  (1984).
The  classification  was based on  the  observed density of  sinkholes,
together with information on  the  bedrock geology,  surficial geology, and
hydrogeology.  Dalgleish and Alexander  (1984)  conclude that the carbonate
bedrock is the primary control on sinkhole formation.  Secondary controls
include the  type and thickness of  the overburden,  and the  depth  of the
water table.   Areas where the  Oneota Dolomite  is  overlain  by the sandstone
member of  the Shakopee Formation,  such as in the  vicinity of Lewiston,
are  the  most  susceptible to  sinkhole  development.    Fractures  in the
noncalcareous sandstone  act as conduits to preferentially direct  surface
water into the Oneota.
THE   SINKHOLE  COLLAPSE  OF   THE  LEWISTON  WASTE   WATER  TREATMENT
FACILITY  LAGOON
Background

  The Waste Water  Treatment  Facility  at  Lewiston,  (population -1300)   is
constructed in an area that overlies the New Richmond Sandstone member of
the Shakopee Formation and the Oneota  Dolomite,  and has less than 30 m of
regolith.   The Waste Water  Treatment Facility consists  of a  series  of
settling ponds or lagoons  which  is commonly known as a "natural" treatment
system  (Figure  4) .   These  types  of systems  are common in  southeast
Minnesota.  The  one  at  Lewiston is  about 20 years old.   Exposure of the
waste water  allows oxygen and  sunlight,  together with microorganism to
"treat"  the   effluent  after  initial  screening  for  large-sized  solids.
Machines or chemicals are not used.   The treated water is  then discharged
to  a  surface-water  channel. This  natural  purification  takes  about  6
months.
       FENCE OH PROPERTY UNE
                                                                 t«' DOUflU CAIE
Figure 4.  Schematic of the Lewiston WWTF ("From Braun,  Intertec,  1991)
                                  720

-------
  According to the superintendent of the Lewiston's  sewer  and water system
Lewiston had wanted a mechanical  system 20  years  ago,  but was  denied the
request  by the  Minnesota  Pollution Control  Agency.    Supposedly,  the
decision was based on the consensus that small towns could not afford the
upkeep and  the  operating expenses of a  mechanical  plant  (Rochester Post
and Bulletin;  February 27, 1991).  Today, that opinion is  not held by any
state government agency.   The change of opinion is based  not on a town's
new-found  ability to afford a  plant,  but because of the  ground-water
quality  problems associated with  a  karst terrain,  and  the  documented
failure  of  a  similar lagoon  system in Altura, Minnesota (Alexander and
Book, 1984), which is  about  10 km northwest  of  Lewiston.
Sinkhole  collapse

  On February  20,  1991,  it was discovered that  a  sinkhole had opened on
the edge of sewage lagoon Number 2  (Figure  4) at  the Waste Water Treatment
Facility at Lewiston, Minnesota.  The sinkhole collapse caused a break in
the  dike  enclosing  the  lagoon.   The  collapse  left  a  hole that  was
approximately 12 m in diameter and  2-4 m  in depth (Figure 5).
Figure 5.   Photo  of  dike-side of sinkhole,  on February 20, 1991. View  is
            northwest. Photo courtesy of R. Dunsmoor,  Winona County.
                                  721

-------
  It  is  estimated that  the collapse  occurred on or  about  February 14,
1991.   According to the records of city workers,  approximately  7.7 million
gallons of  semi-treated  sewage effluent were  lost  from Lagoon Number  2.
The loss  occurred over  several  hours  to  perhaps  a  day.   The effluent
entered the ground through  a conduit  at the  bottom of the sinkhole.  The
waste water had been in the lagoon only about  two months,  and probably
still contained  bacteria and/or viruses because  it was covered with ice
which would prevent sunlight and heat  from destroying them.

  The  sinkhole  collapse  at  Lewiston has  striking  similarity to  the
collapse at the Altura  Waste  Water Treatment Facility  as  documented by
Alexander and Book  (1984) .   Both collapses  formed in  the Oneota Dolomite
where it  was  overlain with the  basal  sandstone member  of  the Shakopee
Formation.   At both locations,  sinkhole collapse  was catastrophic.  One of
major differences  is that  the failure of the Altura lagoon  was  due to
several sinkholes in the  bottom of the lagoon,  whereas, the failure of the
Lewiston lagoon was due to a single sinkhole  collapse near the  edge of the
lagoon which led to breaching of the dike.
Response  to  the  problem

  The water level  in  Lagoon  Number  was  2  was lowered and continued to be
monitored so that  further  semi-treated  water did not spill over into the
sinkhole.   A dam was built  around the  sinkhole in order to prevent surface
run off from entering the hole.
  Water-quality tests were performed  on the  city wells  and on 11 private
wells in  the  area.   The results did not detect contamination  from the
effluent.   Residents  were  advised to drink  water from  hot water heaters
that had been cooled, until they could have their well tested.  They were
further advised to chlorinate their  wells,  and drink  bottled water if they
had any concerns.
Remediation  of  the problem

  The city hired a consulting firm in early March  to  determine  the  size of
the  sinkhole  and propose  short  term  remediation.    A  shallow  (6  m
penetration) geophysical investigation used ground-penetrating radar and
an electromagnetic survey to determine the  limits  of  the  observed sinkhole
and possible fractures in the vicinity.  An independent   proposal to run a
dye  trace form the   sinkhole  in  order  to  determine ground-water  flow
patterns was not adopted by the  city.  This was due in part  to  the  concern
of liability.

  By mid May 1991, it was decided by city  employees that  the best response
to the  sinkhole collapse was to  repair  the dike and seal the sinkhole.
Beginning May 21, 1991, the  site was cleaned-up,  and new dike  was  created
about 15 m from the surface expression of  the sinkhole.   According  to city
employees, the  sinkhole was excavated to within 1 m of bedrock.  The hole
was then filled and sealed  by May  24,  1991.
                                  722

-------
POTENTIAL  FAILURE  OF  OTHER  WASTE  WATER  TREATMENT  FACILITIES

  The Minnesota Pollution  Control  Agency  has  compiled a  list of towns in
southeast Minnesota with   waste-water  pond  facilities similar to that at
Lewiston.   The screening  for the  list included  those  sites  which  are
situated over  karstic bedrock and   had less than  30 m of  soil  or till
above the bedrock.   The initial  list, which  is being refined at this time,
includes  14  sites.    Of  those  14,    10    are  considered  to  have high
potential for  failure,  and 4,  low   potential.  Of the 10  that have high
potential for  failure, 2  have  had failures  within  the  past   20 years.
This corresponds  to a failure  rate  of about  20%.

  Costs  for potential  damages,  repair  and/or  remediation for these Waste
Water Treatment Facilities  is  now the responsibility of the local units of
government.    The  Federal  programs that  cost-shared the  bulk  of  the
original construction costs have  been severely reduced or phased out.   It
is recognized that the  use  of a  lagoon system in  a karst region can lead
to catastrophic  failures  and  potential health concerns.    However,  the
sealing  of the sinkholes not only  in,  or next to, sewage  ponds,  but in
other sensitive  areas,  is  often the  remedial method of  choice  due  to
economics.  A new sewage  treatment plant  would  cost about  1.5  million
dollars  which is an  exorbitant financial  burden for a small town.   These
small towns   are  forced  either to spend  millions  of  dollars   on  new
construction  or risk potential liability suits.  Neither choice is good.
REFERENCES

Braun Intertec  Environmental,  Inc.  1991. Hydrogeologic  investigation of
the  sinkhole  at the  City of  Lewiston Waste  Water Treatment  Facility.
Project No.  BCBX-91-033A.

Dalgleish, J.B.  and E.G.  Alexander, Jr.  1984.  Geologic Atlas  of Winona
County,  Minnesota,  Plate 5.  In: N.H. Balaban and B.M. Olsen  (eds.), County
Atlas Series  C-2, Minnesota  Geological  Survey. University of Minnesota.

Liesch, B.A.  1977. Hydrogeologic  Investigation of  the  Altura  area  for
Altura,  Minnesota.  Report  dated November 1974. 18 p.

Mossier,  J.H.,  and P.R.  Book.  1984.  Geologic Atlas  of Winona County,
Minnesota, Plate 2. In: N.H. Balaban and  B.M.  Olsen (eds.),  County Atlas
Series C-2, Minnesota  Geological Survey. University of Minnesota.

Sweeting,  M.M.  1973. Karst Landforms. Columbia Press, New York.  362 p.
                                  723

-------
BIOGRAPHICAL   SKETCHES

Nancy 0. Jannik is an Associate  Professor  in  the  Department of Geology at
Winona State University. She has a B.S. in geology (1976)  from the College
of William and Mary,  a M.S. in geology (1980)  from Rutgers University,  and
a Ph.D. in  geosciences/hydrology (1988)  from the New Mexico  Institute of
Mining and  Technology.  Her current research  involves  determining aquifer
characteristics, aquifer interactions, and the inter-relationships between
surface and ground water  in the karst terrain of southeastern Minnesota.
Her full address is as follows.

                           Department  of Geology
                          Winona  State University
                           Winona, MN  55987-5838
                              (507)  457-5267
E. Calvin Alexander, Jr., is a Professor  in  the  Department  of Geology and
Geophysics  at  the  University  of Minnesota.  He  has a  B.S.  in  chemistry
(1966) from Oklahoma State University and a Ph.D. in chemistry (1970)  from
the University of  Missouri  at Rolla.  The central  theme  of his  current
research  is  the rate of  movement  of fluids  in  hydrogeology. He  and his
research  group are utilizing  a  variety  of  methods to  measure flow and
residence times, which can range from hours  to ten  of thousands  of years.
His full address is as follows.

                   Department of Geology and Geophysics
                          University  of Minnesota
                        Minneapolis,  MN 55455-0219
                               (612)  624-3517
Lawrence J.  Landherr is the Regional Director of the  Minnesota Pollution
Control Agency  (MPCA)  Southeast Region. He has  a  B.S. in biology  (1967)
from the University of Minnesota. He has  24  years  working experience  in
environmental programs in the karst of southeastern  Minnesota.  His current
responsibilities involve managing the activities  of  the eight  person staff
involved in all program areas of the MPCA.  His full  address  is  as follows.

                          MPCA Regional Director
                           2116 Campus Dr.  S.E.
                            Rochester, MN 55904
                               (507) 285-7343
                                  724

-------
     Erosion  and  Sedimentation Control  Methodologies  for
   Construction Activities over the Edwards Aquifer in Texas
                      Hank B. Smith, P.E.

                     Texas Water  Commission
                         Austin,  Texas
I. Abstract
Through its implementation  of  the  Edwards  Aquifer Program,  the
Texas Water Commission strives to preserve the quality of water
produced from  this  karst aquifer.   The Texas  Water Commission
(TWC) has rules regarding development  over the Edwards Aquifer
Recharge Zone which require erosion and sedimentation controls.

During    construction-related    activities,    erosion    and
sedimentation   are   the   primary  mechanisms   for   aquifer
contamination.   Temporary  erosion and  sedimentation  control
structures are necessary to ensure that suspended sediments in
stormwater remain within the construction  area.   These control
structures  filter  sediments  and   include hay  bales,  filter
fences, rock berms, and  brush berms.

Sediments that are deposited in streams and lakes impact benthic
organisms, reduce  storage  capacity of lakes,  and reduce water
quality.    Sedimentation   also  impairs   natural  groundwater
recharge which results in the reduction of base flow in creeks,
destruction  of seeps and  springs,  and   loss  of  groundwater
supplies.

Erosion occurs when natural  vegetation that binds soils together
is disturbed.   During rainfall events,  natural vegetative cover
acts  as a  filter  to remove  pollutants,   and reduces  runoff
velocity-     During  construction  related  activities,  natural
vegetative  cover  is  disturbed.    As  a  consequence,  Total
Suspended  Solids   (TSS)  in  stormwater  runoff  is   greatly
increased.

Stormwater runoff  from  urban areas  has  increased levels  of
nutrients  (nitrogen,  phosphorus,  etc.)  from  fertilizers,  and
pesticides.    Other  pollutants   come  from  asphalt   roofing
materials, sewage disposal  facilities, and other  forms of non-
point  source  pollution.    Stormwater  runoff  from  roadways
                             725

-------
typically  has  increased   levels   of   lead,   total  petroleum
hydrocarbons, oil & grease, and metals.

The  Edwards  Aquifer Program  requires permanent  sedimentation
control  structures which  improve  the quality  of  stormwater
runoff and  provide protection of  water resources.   Permanent
control  structures  include sand  filtration and  sedimentation
basins and hazardous materials traps.


II. The Edwards Aquifer Protection Program in Texas

Effective March 21,  1990, the  Texas Water  Commission  (TWC)
adopted revised rules requiring the  preparation  and submission
of a Water  Pollution Abatement Plan (WPAP)  for any  regulated
developments over the  Edwards Aquifer recharge zone  in  Texas.
The  purpose of  the rules is  to regulate  activities  having
potential for  polluting the  Edwards Aquifer.   The  activities
addressed are those that pose direct threats to  water quality.
Nothing in the rules is intended to  restrict the powers  of  the
TWC  or  any  other governmental entity  to  prevent,  correct,  or
curtail  activities  that  result  or  potentially  result   in
pollution of the Edwards Aquifer.   Each WPAP must contain, at a
minimum, the following information:

1)   name,  address,  and   telephone  number  of  owner,  agent,
     developer and contact persons;
2)   general location  map which identifies  the  limits  of  the
     recharge   zone,   and  the   location   of  the   proposed
     development;
3)   site   plan  with   proposed  development,   flood  plain
     information, location of  known wells, existing and proposed
     drainage  patterns,   and  identification  of  any  known
     significant recharge features;
4)   detailed  assessment  of area  geology   which  includes:
     geologic map  at site  plan  scale; stratigraphic  columns;
     narrative description of surface geology,  soil profile  and
     units;  and,  a  narrative description  of  all  significant
     recharge features; and
5)   detailed technical report which includes:   description of
     the  proposed   development;   volume   and  character   of
     wastewater and stormwater expected to be generated from the
     site;  description  of methods  to  be  taken  to prevent
     pollution of stormwater originating on-site and upstream of
     the  site;  description of methods to be  used to prevent
     pollutants   from   entering  recharge   features;   and   a
     description of method of disposal of wastewater.

Upon receiving  final  approval of  the WPAP, the  applicant must
deed  record  the  approval in  the  County  Courthouse.     Any
modifications to the plans must be submitted to and approved by
the TWC prior to construction.  The  TWC must be  notified prior
to initiation  of construction.   If any  significant  recharge
features  are identified,  all construction  activities  in  the
                              726

-------
immediate area  must  be halted upon discovery until  protection
plans can be developed, reviewed and approved by the TWC.

All organized sewage collection systems must also be approved by
the TWC prior to initiation of construction.   At a minimum,  the
plans must comply with the following design criteria.

1)   all manholes must be watertight with watertight rings  and
     covers.  Alternate means of venting must be  provided  for
     every third manhole.   All manholes must  also be tested  for
     water-tightness prior to use.
2)   lift stations  must be designed to ensure that bypassing
     does not occur.   All  lift stations must have  audible  and
     visual alarms, as well as, auto-dial telemetry to  notify a
     series of locations with 24-hour response coverage.
3)   all gravity  and pressure sewage lines,  including  private
     service laterals and manholes, must be inspected and tested
     to ensure that no leakage occurs after construction.
4)   all wastewater systems must be tested every  five  years to
     determine the types and locations  of structural damage  and
     defects  such as  offset  or  open  joints,  or  cracked  or
     crushed lines that could allow exfiltration to occur.

Additional requirements provide that new or increased discharges
of  treated  effluent  that  would  create increased loading  of
treated wastewater are prohibited over  the recharge zone.  Land
application systems  that  rely  on  percolation are  prohibited.
However, irrigation systems may be allowed and are reviewed on
a case-by-case basis.

All  WPAP's  and  sewage collection  system plans  must  include
erosion and sedimentation control  plans.    These plans  shall
provide designs  for  control  structures that ensure  stormwater
contaminated during  construction  activities  is not  allowed to
discharge from the site without treatment.  Some  of the effects
of  uncontrolled erosion  and  sedimentation  from  construction
areas include:

1)   changes in habitat and localized natural land uses;
2)   increased flood plain depths;
3)   reduced  volumes  of  water  storage in  rivers  and  lakes
     resulting in increased down-stream flooding;
4)   increased  eutrophication  rates  from excessive  nutrient
     loadings; and
5)   loss of top soil.


Ill. Controlling Erosion and Sedimentation

During construction  related  activities  erosion is the  primary
concern.  The major part of the erosion  potential exists between
the time when the native vegetation is removed and when  the site
construction is complete and the vegetation is restored.   These
impacts can be minimized with the proper design,  installation,
                             727

-------
and maintenance of temporary  erosion  and sedimentation control
structures.    In  addition to  temporary  control  structures,
impacts can be reduced by limiting the amount of disturbed areas
by identifying a construction area  and  limiting access to this
area.  Temporary erosion controls can be designed and installed
within the construction areas to provide four basic functions:

Diversion - by diverting  off-site flows around  disturbed areas
the amount of runoff  collecting  sediment and the volume of water
that must be treated  can  be minimized.   Diversions can also be
used to direct runoff from drainage areas to a location where
sedimentation can occur more  cost effectively.

Ponding  -  by  installing  controls  that  reduce  the  runoff
velocities the sediments  are  allowed  to settle out and  can be
removed from the pond on a routine basis.  Ponding is one of the
primary  methods  of   sediment  removal  used  during  and  after
construction.

Filtration - when stormwater flow is routed through a filtration
structure, soil particles  are trapped on the  filter media,  and
sediment  settles  out  up-gradient  of  the  filter  structure.
Filter structures can be used to capture the fine portions of
the sediment and are used  extensively during construction.

Flow spreading - by  installing  controls that  change the  runoff
characteristics from concentrated to sheet flow, the velocity of
the runoff  is  reduced which  increases  local  sedimentation  and
minimizes erosion.

The  effectiveness of any  temporary erosion and  sedimentation
control  structure  is  dependent upon proper  installation  and
adequate maintenance.   The following  guideline can be used to
approximate  the  type of  structure  that  is  appropriate  for
various conditions.
Type of Structure
Hay Bales
Silt Fence
Brush Berm
Rock Berm
Reach Length
< 100 feet
n/a
< 200 feet
< 100 feet
<  50 feet
n/a
< 150 feet
<  75 feet

500 feet
Drainage area
  1/2 Acre
  2   Acres
  2   Acres
  1   Acre
  1/2 Acre
  1 1/2 Acres
  1     Acre
  1/2   Acre

  5 Acres
 Slope
 0 - 10%
 0 - 10%
10 - 20%
20 - 30%
> 30%
  0 - 10%
 10 - 20%
 20 - 30%
> 30%
 0 - 10%
For  drainage  areas  greater  than  5  acres,   the  design  of
structural  features  are reviewed on  a  case-by-case  basis.
Approval  is  based on the anticipated quantity and  quality  of
stormwater,  and  the  location  of any  critical  environmental
features downstream.
                              728

-------
Hay  bales are  an  inexpensive  and  readily  available  control
structure; however,  they are  difficult to  maintain,  must  be
replaced following every rainfall, and  are ineffective  for any
drainage area larger than 0.5 acres.  The TWC does not recommend
the use of hay bales except in limited areas  where construction
times will be short and rainfall is not anticipated.

Filter  fabric  may  be  required  within  the  rock  berms  if
determined necessary  by the TWC.   These structures have  been
proven  effective  in  steep  canyons,  above  permanent springs,
pools or  other  environmentally  sensitive areas.  This  type  of
structure combines  the  removal  efficiency of silt  fences  with
the structural stability of rock berms.


IV-  Design Calculations

Temporary erosion and sedimentation  control  plans must  include
calculations  to  verify  that  the   proposed   structures   are
adequately designed.   The calculations involve  establishing  a
storm flow volume for the drainage area for each structure, and
an appropriate  barrier flow-through  rate.   The  computational
procedure for establishing a flow-through rate includes  several
calculations  such   as   coefficient  of  roughness,  time  of
concentration, and rainfall intensity.

Using the "Rational Method"  (Q  =  C * i * A)  the  volume  "Q"  of
flow  can  be  calculated for various drainage  areas with  the
following parameters:

1)   determine  the  drainage  area "A"  for  each structure  or
     drainage basin based on the site plan;
2)   determine rainfall intensity "i";
3)   estimate the runoff coefficient  "C" for a two year rainfall
     storm for each specific soil type and channel slope;
4)   compute peak flow rate of runoff "Q" for a two year storm
     using the rational method;
5)   determine frontal area for flow "Af" at  the barrier; and
6)   determine the  flow  through  rate at the barrier using the
     following   "q = Q / A,"


Based on  the flow  through rate  "q" the proper structure  is
chosen from the following Table 1:
                              729

-------
Table 1

Sheet; Flow
Hay Bales
Filter
Dikes
Brush Berm
Silt Fence
Rock Berm
Diversion
Dike, etc.
Swales
Concentrated
Flow
Stone
Outlets
Diversion
Dike, etc.
Swales
Rock Berm
Slope Drain
Sediment
Traps
Sediment
Basins
Stream Flow
Rock Berm
Sandbag
Berm
Maximum Flow
Through Rate
(Gal/Min/SgFt}

5
20
40
40
60



40


60






Trenched
or
Secured

Yes
Yes
Yes
Yes
No



Yes


Wire
Mesh
Sheath




Wire
Mesh
Sheath
No
Recommended
Maximum
Drainage
Area

. 5 acre
1 acre
2 acre
2 acre
5 acres
5 acres
5 acres

5 acres
5 acres
5 acres
5 acres
5 acres
5 acres
100 acre

+5 acres
+5 acres
Recommended
Maximum slope
Length

50 feet for
slopes exceeding
10%
50 feet for
slopes exceeding
10%
















 730

-------
V.  Additional Structures

Temporary erosion  and sedimentation control plans  should  also
provide  for  stabilized  construction  entrances,  equipment  and
petroleum  storage  areas,  and  spoils  disposal  sites.     An
inspection program  and a maintenance plan and schedule  should
also be included.

Stabilized construction entrances consist of a stabilized pad of
open  graded  rock  located at  any point where traffic may  be
entering or  leaving a construction site to a public  right-of-
way,  street,  alley,  etc.    The  purpose  of  a   stabilized
construction entrance is to reduce or eliminate the  tracking or
flowing of sediment onto  public  roadways where they contribute
to non-point source pollution.

Equipment storage  areas,  and temporary fuel and   oil  storage
facilities  should  be  designated  on  the  site  plans.    All
construction vehicles  should  be  parked  in protected areas  when
not in use,  and all maintenance of vehicles  should occur within
these limits.   These  areas  should  be placed at topographical
highs, or berms  should be provided to divert stormwater  flow
around  the   site.     Structural  controls   must  be   placed
downgradient of the site  to  filter  stormwater runoff.   Special
precautions  should  be taken  to prevent the spill of  petroleum
products from storage  facilities or from emergency  maintenance
activities.   General  scheduled maintenance  is not  allowed  on-
site.

Spoil  disposal  and  material storage  areas  should  also  be
designated  on  the  site  plans.    Spoils  material  from  the
construction   activity,    sediment   removed   from   temporary
protective  structures,  and   construction material   (backfill,
sand, base material,  etc.) must  all be stored in these  areas.
These  areas  should   be   protected from  stormwater  run-on.
Stormwater discharges from the site  should be filtered to remove
potential pollutants.

Temporary erosion and sedimentation  control structures should be
inspected routinely to ensure effective  control  of  stormwater.
Weekly inspections should be  conducted of the entire  area,  and
inspections should be conducted within 24 hours of any rainfall
event.   Inspection  forms should be  developed  and should  be
maintained at the construction site.

Typically, the contractor should be required  to routinely clean
paved surfaces with appropriate power  brooms, or  vacuum devices
to minimize  the amount of  sediment within the construction area.

A temporary  erosion and  sedimentation control maintenance  plan
should be developed prior to  initiation of construction.   These
maintenance plans should provide a schedule  for the  removal and
replacement  of  temporary  structures   that  are  no   longer
effective.   Provisions should be  made for the removal of  silt,
                             731

-------
and sediments following rainfalls.  Troublesome areas should be
identified and evaluated for improved structural controls.
VI.  Cooperative Agreements

The TWC also enters into Cooperative Agreements with other State
agencies to  ensure that  the  agencies involved  in  development
activities can be  leaders in environmental protection.

In September  1991, the Texas Department of  Transportation  and
the  TWC entered  into a  cooperative  agreement regarding  the
construction  of a  major  highway intersection over  the  Edwards
Aquifer.  The agreement  required that the Texas Department of
Transportation comply  with  all rules and requirements  for  the
submission  of  a  Water  Pollution  Abatement  Plan  prior   to
construction.  The agreement also  provided additional specific
requirements  applicable  only   to the  development   at this
location, and included the following.

1)   Provide  a detailed  temporary erosion  and  sedimentation
     control maintenance plan and schedule  that would  establish
     guidelines for inspection  of structures,   replacement  of
     inadequate or damaged structures, and  increased protection
     of any impacted areas.
2)   Restore  and maintain vegetative  cover prior to completion
     of construction.
3)   Restrict  construction  vehicle  access   from  unprotected
     areas, provide routine  street  cleaning and employee parking
     in stabilized areas.
4)   Designate contact persons for the TWC, Texas Department of
     Transportation, and the contractor.
5)   Develop  an interagency training program.
6)   Develop  a  remediation  plan and  schedule  in the event  an
     impacted area is identified.

The  TWC also encourages input  from  other  entities such  as
municipalities, river authorities, conservation districts,  and
environmental  groups.    Citizen  monitoring  can also  provide
valuable information regarding impacted areas and problem areas
within existing developments.


VII.  Estimated Annual Pollutant Load

The amount of total suspended solids,  total phosphorus,  and  oil
and grease that may be discharged  from a site  before  and after
development can be approximated.   The annual pollutant load is
the pollutant concentration times the  annual  runoff volume.  The
following  calculation  can  be  used  to  estimate  the  annual
pollutant load for a tract of land:
                              732

-------
          L=AxRFxRvx 0.2266 x C,  where:

          L  = Annual pollutant load in pounds
          A  = Area of tract in acres
          RF = Annual rainfall amount in inches
          RV =  runoff  to rainfall  ratio
          RV =  0.05 +  (0.009 *  1C)  where:
               1C = impervious cover for site in percent
          C  = Pollutant concentration in milligrams per liter
                                Background "C"   Developed "C"
            Total Suspended Solids 48             130
            Total Phosphorus        0.08            0.26
            Oil and Grease          0              15

The area used in the annual pollutant load calculation for total
suspended solids and total phosphorus should represent only  that
portion of  the  tract  which is  to be developed, not the  entire
site area.   Typically,  this is the limits of construction for
the area.  For oil and grease,  the load  calculation should  only
use the paved area.


VIII.  Permanent Erosion and Sedimentation Control

Sedimentation  basins  can  be  effective at  removing  suspended
solids   from  developed  areas   but  are   very   limited   in
effectiveness of removing dissolved solids.  The basic design of
this type of basin is  to  capture and isolate a percentage of the
rainfall or  a  certain volume of stormwater, and discharge the
flow over a vegetative strip over a prolonged period of about 24
hours.  A design to capture the first l" of stormwater flow  will
generally be able to  remove about  50%  of the total  suspended
solids, 15%-20% of the total phosphorus  and about  8% of the oil
and grease content of the stormwater.

Sand filtration basins have been used in the Austin and Central
Texas area to effectively remove suspended solids,  bacteria, and
other  pollutants  from developed  areas.   The  basic design  of
these structures is to capture  and isolate up to the first  inch
of stormwater runoff and release the flow  through  a porous  sand
media over  a prolonged  period  of  about  24 hours.   A  design  to
capture  the   first  1"   of  stormwater   flow   will   remove
approximately 55%-70%  of the total suspended solids,  25%-33% of
the total phosphorus  and about 23%-30% of  the oil and  grease
content of the stormwater.

Sand  filtration and  sedimentation basins  are  often  designed
together with the sedimentation basin discharging  into the  sand
filtration basin for optimum efficiency.

A  formal maintenance plan  and  schedule  is  required  to  be
submitted to  the  TWC for  review,  possible modification,   and
approval prior  to completion of  construction.   The plan  must
                             733

-------
include a  responsible party  and  the anticipated  cleaning  and
monitoring schedule.  The plan must also include a disposal site
for  any  material removed from the  basins  and an anticipated
removal schedule.

Plans should be developed to ensure that a copy of the WPAP and
approval letter  is  maintained at the construction site.   The
requirements contained within the design plans and the approval
letter from the TWC should be reviewed with the contractor.

Other factors  to  consider during the development  design phase
include  land   grading,   stormwater  outfall  designs,   grade
stabilizing   structures,   ponds  and  retarding   structures,
streambank protection, and storm sewer design.
IX.  Other Options

Street sweeping and parking lot sweeping activities can be used
to reduce levels of pollutant run-off from developed areas.  The
effectiveness of street sweeping  and vacuuming  is  difficult to
measure and is a function  of how frequently it is performed, and
the percent of site that can be cleaned.  Consideration must be
given to  the  pollutant load removed by non-sweeping  processes
such  as  wind  and biochemical  processes.     The  following
efficiencies for sweeping activities are appropriate:

                              Broom Sweeper  Vacuum Sweeper
     Parameter                Efficiency f%) Efficiency  f%)
     Total Solids                  55             93
     Total Phosphorus              40             74
     Total Nitrogen               42             77
     Chemical Oxygen Demand        31             63
     Biochemical Oxygen Demand     43             77
     Lead                          35             76
     Zinc                          47             85

An important consideration in reviewing street sweeping options
is that 78% of the  solids load is contained  within an area six
inches from the curb, and that 97% of the solids load is within
40  inches from the curb.   Some  street sweeping devices only
redistribute the solids and actually remove  only about half of
the solids.  A significant percentage of the solid particles are
also of sizes which are not effectively removed by  conventional
street sweepers.


X.  Hazardous Spill Consideration

Spills  of hazardous materials  from roadways  and  other areas
where  hazardous materials  are  stored, transported,  or used
create  potentially major  impacts to  surface  and  groundwater
quality.  Hazardous materials traps  (HMT's) can  be at stormwater
outfalls and other locations that  will protect our states' water
                              734

-------
resources from spill contamination.

HMT design typically incorporates concrete basins with a minimum
storage volume  of 10,000 gallons.   The design allows  for  the
first  10,000  gallons of  runoff  to be  directed to a  concrete
basin where it can be isolated.  The average truck carries less
than 10,000 gallons of fuel, petroleum product or waste, sewage,
or other hazardous material.  Any flow greater than the capacity
of the basin would initiate a siphon flow which would drain  the
basin.   In theory, any hazardous  material  spill will be of  a
volume less than 10,000 gallons and would therefore be isolated
in the concrete basin.   A series of valves could then be opened
or closed  by  emergency crews  which would divert  future  flows
around the basin until the hazardous material can be removed.

Most designs  for  these  basins are  ineffective  during  rainfall
events since it is not practical to contain all rainfall or  try
to  differentiate  between  contaminated  and   uncontaminated
stormwater flows.

A  formal  maintenance  plan  and  schedule  is  required  to be
submitted  to  the  TWC  for review,  possible modification,  and
approval prior  to completion of construction.   The plan must
include a responsible party for monitoring the  siphons and  the
hazardous  materials traps.   The  plan  must  also provide  for
notification and training of local emergency response personnel.
XI.  Summary

The  TWC reviews  water pollution  abatement plans  and  sewage
collection systems prior to initiation of construction to ensure
adequate protection  of environmental features.   As a part  of
this   process,   the   TWC   requires  temporary   erosion   and
sedimentation  control structures  be designed,  approved,  and
installed prior to initiation of construction  over  the  Edwards
aquifer.  The TWC also requires adequate maintenance and  routine
inspection of these structures.

Erosion and  sedimentation  from construction areas  is a  major
factor leading toward contamination of karst aquifers.   By  the
proper  design,  installation,   and  maintenance   of  temporary
erosion and  sedimentation control structures the  impacts  of
construction can be minimized.

Permanent sedimentation  and filtration  basins,  and  hazardous
materials  traps can  be  used  to  mitigate  the  impacts from
developed areas.  The amount of  pollutants from developed areas
can  be estimated  and using  appropriate  removal  rates  for
sedimentation   and  filtration  structures   the   anticipated
additional loading to receiving waters can be  estimated  during
the design phase of developments.
                              735

-------
                         SILT FENCE
                                         48" Min. Heavy Weight T-Post

                                         24* Tall Mln., *"x 4' 12 G«.
1

i
T
48' 2




4 -^V
¥' <



\
4



r
/^&»J&

^ '
\
\
\
\
\
\
\
\
s
s
\
\
.^ \>Biv«nizca «*iic n »-
^^
,,* o* Mln. Non-Wove
^X^ Filter Fabric 36* Wl<
^ —

.SKIRT ANCHORS
/
1 Soil Level

*-ll-->-
                           Cross-Section
            GEOTEXTILE
Installation
                                                WOVEN WIRESUPPORT
                                                  2'X 4'  WIRE MESH
     Layout the silt fence following as closely as possible to the
     contour.
     Clear the ground of debris, rocks,  plants (including grasses
     taller than 2") so that a smooth surface can be utilized for
     anchoring the skirt.
     Drive the  heavy duty T posts  at least 12 inches in  to the
     ground and at a slight angle towards the flow.
     Attach the 2" x 4"  12  Gauge  welded  wire mesh to the T posts
     with  11  1/2  gauge galvanized  T post clips.  The top  of the
     wire  should be  24"  above  ground level.  The  welded  wire mesh
     shall be  overlapped 6" and  tied at least 6  times  with hog
     rings.
     The silt  fence  will be installed with  a  skirt  a minimum of
     11" wide placed on the uphill side of the fence in a direction
     towards the anticipated runoff. The  fabric should overlap the
     top of the wire by  1".
     Anchor  the silt fence on alternating 2' centers  front and
     back  so that there  is only 1'  between anchor  points.
     Geotextile splices  should be a minimum of 18" wide attached
     in at least 6 places. Splices in concentrated  flow areas will
     not be accepted.       o

-------
                      ROCK  BERM
                       16' Min.
                                            Woven Wire Sheathing
                                             3' TO 5* OPEN
                                             GRADED ROCK
                             Cross-Section
    Woven Wire Sheathing
Installation
     Layout the rock berm following as closely as possible  to the
     contour.
     Clear the area of debris, rocks or plants that will interfere
     with installation.
     Place woven  wire fabric on the ground  along the proposed
     installation with enough overlap to completely encircle the
     finished size of the berm.
     Place the rock along the center of the wire to the designated
     height.
     Wrap the structure with the  previously placed wire mesh  secure
     enough so that when walked  across the  structure retains it's
     shape.
     Secure with tie wire.
                          737

-------
          STABILIZED CONSTRUCTION
                      ENTRANCE
Mutt Be Properly Graded
To Prevent Runoff From Leaving
The Construction Site \
                                          Public
                                          Riflht-of-Way
\

.., so* Min.



                         Cross-Section
                                  GeotcxtlU
                                  Undcrllncr
                                 (If Required)
                             50* Mln.
                                                 Public
                                                Rlflht-of-Way
                         Plan View
Installation
     Clear the area of debris, rocks or plants that will interfere
     with installation.
     Grade  the area  for the entrance  to flow back on to  the
     construction site. Runoff from the S.C.E. onto a public street
     will not be accepted.
     Place geotextile  fabric if  required.
     Place rock as  required.
                               738

-------
XII.  References

Regional Resource Division; Northern Division Planning District
Commission,  1979,  Guidebook  for  Screening  Urban  Non-point
Pollution Management Strategies

Hydroscience  Inc.;  1977,  Water  Quality  Management  Planning
Methodology for Urban and Industrial Stormwater Needs

Lower Colorado  River Authority  (LCRA);  Environmental  Quality
Division,  January  1991,   LCRA  Lake  Travis  Non-point  Source
Pollution Control Ordinance

City of Austin; 1988, Erosion and Sedimentation Control Manual

City of Austin; Drainage Criteria Manual

City of  Austin;  Design  Guidelines for  Water Quality  Control
Basins
                             739

-------
       Erosion and Sedimentation Control Methodologies for
    Construction Activities over the Edwards Aquifer  in Texas


                       Hank B. Smith, P.E.

                      Texas Water Commission
                          Austin, Texas

Question - On environmental geology field trips in the  Austin area,
I have noticed severe erosion problems  in the stream channels as a
result of ephemeral storm runoff caused by urbanization.  What is
the Texas Water Board doing about this problem?

Answer -

The Texas Water Commission (TWC)  has rules regarding development
over the  Edwards  Aquifer  Recharge  Zone  which  require temporary
erosion  and  sedimentation  controls.   These control structures
filter sediments and include hay bales,  filter fences, rock berms,
and brush berms.

The major part of the erosion potential  exists  between the time
when  the   native   vegetation  is   removed  and  when  the  site
construction  is complete and  the vegetation is  stabilized.   Such
impact can be minimized with the proper design,  installation, and
maintenance   of   temporary  erosion  and  sedimentation  control
structures.    Additionally,  negative  impact  can  be  reduced  by
limiting  the  amount  of  disturbed   areas  by  identifying  a
construction  area and limiting access to this area.

Contractors  are required  to routinely  clean paved surfaces with
appropriate power brooms,  or vacuum  devices to minimize the amount
of sediment within the construction area.

The  Edwards  Aquifer  Program  requires  permanent  sedimentation
control structures which improve the quality of stormwater runoff
and provide  protection of  water resources.  Sedimentation basins
can be effective at removing suspended solids from developed areas,
and sand filtration basins  have been used in the Austin and Central
Texas area  to effectively  remove suspended solids, bacteria, and
other pollutants from developed areas.
                                740

-------
       Case Histories  of  Several Approaches  to  Stormwater

    Management in Urbanized Karst Terrain, Southwest Missouri


                        Wendell L.  Earner
                    ICF Kaiser Engineers Inc.
              Pittsburgh,  Pennsylvania 15205-1017

                         Patricia Miller
             California State University-Sacramento
               Sacramento, California 95819-6049


                            Abstract

     Because sinkholes in karst terrain are dynamic components of
surface-ground water interactions,  they present particular
challenges in the design of urban stormwater systems.  Sinkholes
may function as either drains or ponds for surface runoff,  and
their hydrologic behavior may change markedly with long term
urban growth,  as impacts of earlier development and off-site
influences are added to natural drainage patterns at proposed
construction sites.  Comparison of three sites in Springfield,
Missouri, in the Springfield Plateau Karst, illustrates the
effects of contrasting approaches to stormwater design.

     A wooded tract with unmodified sinkholes was cleared and
developed for residential use.  Discharge of stormwater was
directed into sinkholes, and erosion control consisted of hydro-
mulching and sedimentation fences in sinkhole areas.  East of
this location are two parcels which differ in removal of
vegetation and off-site drainage relationships.   Stormwater
design at these sites was adapted for modifications made to
sinkholes during railroad and highway construction several
decades earlier.  Sediment fencing, hydro-mulching, and detention
berms augment infiltration, restrict erosion, retard discharge to
sinkholes, and incorporate off-site considerations.

     Ongoing observations of stormwater behavior indicate
problems of flooding and sediment control at the western site but
minimal disruption of existing drainage patterns at the eastern
sites.  Difficulties in ascertaining all potential sources of
groundwater contamination underscore the need for stormwater
regulations which regard sinkholes as integral components of
                                741

-------
entire karst drainage systems.

                           Introduction

     Springfield, the county seat of Greene County, is located in
the Springfield Karst Plain in southwest Missouri  (Figure 1).
Dominant bedrock underlying this plain consists of horizontal
strata of the Mississippian Burlington-Keokuk Limestone, a light
gray, coarsely crystalline, fossiliferous unit.

     Because the Burlington-Keokuk Limestone is susceptible to
solution, karst features are extensive throughout the county.
Figure 2 illustrates areas of extensive sinkhole development.
Two major concentrations of sinkholes are found within the city
limits of Springfield.  These areas are located in northwest
Springfield near the airport and in a part of east Springfield
referred to as the East Cherry Street Sinkhole Area.

     These two sinkhole regions are situated on topographic
highs, near surface water drainage divides.  Apparent controls on
sinkhole occurrence in these areas include facies differences
within the Burlington-Keokuk Limestone and intersections of large
scale lineaments.  Stormwater drainage of unaltered sinkholes
ranges from rapid drainage to long-term ponding.

               Past  Stormwater Management  Practices

     Springfield and its Metropolitan Statistical Area comprise
the fastest growing urban region in the state of Missouri.  Rapid
population growth has created additional stresses to existing
land use, including reduction of open areas and increased
Stormwater runoff.  Control of Stormwater in sinkhole areas has
utilized several standard engineering approaches: concrete lined
channels draining into sinkholes; installation of drain pipes
into the sinkhole "eye"; filling of sinkholes; elaborate drains
or pumps removing Stormwater from one sinkhole and discharging
into another drainage basin or sinkhole; enlargement of sinkhole
eyes by excavation to increase drainage capacity.

     Several problems may arise in using these standard
practices:

     1.  Runoff may exceed drainage capacity of sinkholes.
     2.  Sediment and debris may plug sinkholes and associated
         karst conduits.
     3.  Sinkholes are not isolated entities, but are integral
         parts of dynamic drainage systems.  Stormwater directed
         into one sinkhole may resurface in another part of the
         drainage.
     4.  Direct disposal of stormwater into a karst aquifer may
         degrade water quality by introducing contaminants.
         Existing practices may violate 1988 federal regulations
         governing urban stormwater quality.
                                742

-------

-------
  KARST AREAS
        OF
  SPRINGFIELD
       AND
EENE COUNTY. MO
                                                    CIT?
                                                     OF
                                              SPRINGFI
   Figure 2.  Major karst areas of Springfield and Greene County, Missouri.  Two areas of
         extensive sinkhole development within the city are located at the Springfield Airport
         and near East Cherry Street. Scale is approximately 1"=3.5 mi.

-------
These problems have been recognized in previous studies and
master plans for stormwater control in the Springfield area.
In the 1950's a comprehensive stormwater master plan addressed
existing practices and recommended discontinuing the use of
sinkholes as stormwater drains.  Because of the total costs
necessary to accomplish its goals, the master plan was never
adopted (Turner, 1989).

     In the 1970's William Hayes, Environmental Geologist for the
City of Springfield, prepared a report sponsored jointly by the
city and the Federal Department of Housing and Urban Development
(Hayes, 1977).  This report discussed the origin of local
sinkholes, their drainage areas and hydrologic characteristics,
and geologic factors controlling their alignment.  Although the
study addressed the adverse effects of urban development on
sinkholes and karst groundwater systems, it emphasized many
standard structural controls for containment and disposal of
stormwater.

     Having recognized the inadequacy of existing designs to
control flooding and the need to accommodate increased runoff
from future development, the City of Springfield adopted an
ordinance  (effective June 19, 1989) to protect sinkholes and
prevent flooding.  This ordinance, based on similar regulations
enacted by Lexington, Kentucky, requires a permit for any
proposed construction within a sinkhole drainage area.
Application for the permit must include a hydrogeologic report
evaluating the sensitivity of soil to erosion; relationship of
the sinkhole to the overall drainage area; existing sinkhole
drainage characteristics; capability of the sinkhole to accept
stormwater; type of work and equipment to be used, impacts from
proposed activities, and requirements to protect the sinkhole
during construction.

     Although the ordinance attempts a balance between urban
growth and protection of natural drainage of sinkholes, several
serious ambiguities exist.  Description of enforcement procedures
is indistinct. The ordinance does not address the piecemeal
nature of development, in which earlier construction on adjacent
sites may affect drainage at the site being considered.  Lack of
recognition of sinkholes as integral parts of dynamic hydrologic
systems may result in problems with on-site/off-site drainage.

           Case Histories of Recent Urban Developments

     Three sites were analyzed to examine the effectiveness of
contrasting design approaches to stormwater management.  These
sites differ in vegetation, on-site/off-site considerations, and
types of development proposed.  All three sites are located
within the East Cherry Street Sinkhole Area previously described.
Figure 3 illustrates the locations of these sites relative to
important hydrologic and cultural features.  Dye tracing (Hayes
and Vineyard, 1969; Hayes, 1977; Aley and Thomson, 1981; and
Thomson, 1986) has indicated that the East Cherry Sinkhole
                                745

-------
                           CHESTN JT
                                                      KH&L&FIL
                                                     C.HWA Y
ONEBREAK
  ING
       CATALPA STREET
.SINKHOLE DRAINAGE AREA
 fiQR JONES AND BONEBREAK S$jj?IN
 AF\ER HA YES AND	-*^~  	

     SX_^X
  Figure 3. Map of East Cherry Street Sinkhole Area (dashed line), showing locations of study
        sites relative to hydrologic and cultural features.  S = Sundance Estates; CN —
        Chestnut Plaza Phase I; and CS = Chestnut Plaza Phase II. Scale is approximately
        1"= 3 mi.                                                          y

-------
drainage provides recharge to Jones and Bonebreak Springs.  These
springs in turn discharge to Pierson Creek, a major tributary
which joins the James River approximately 3 miles upstream from
the intake of the Blackman Treatment Plant municipal water
supply.  Sinkholes on two of the sites, Chestnut Plaza North and
South, drain by way of the Steury Cave system to Jones and
Bonebreak Springs.

     Urban development in the East Cherry Street Area began in
the early 1900's.  Residential development, usually on large
parcels, continued into the 1950's and 1960's, when industrial
and commercial uses were targeted for this area. By the late
1980's few open areas remained.  These undeveloped parcels are
attractive because of their proximity to downtown Springfield and
major regional transportation routes.  During the past several
decades, this area has experienced sinkhole flooding, exacerbated
by increased runoff accompanying development.

                   Sundance  Estates  Subdivision

     Sundance Estates Subdivision is located on East Grand
Street,  southwest of the other two study areas (Figure 3).  This
ten-acre was a densely wooded tract with one farm house and
several outbuildings. Existing structures were all located near
Grand Street.

     Sinkholes on this tract are not apparent on either the
U.S.G.S. (10-foot contour interval)  or the City of Springfield
(5-foot contour interval) topographic maps.  Preliminary plans
for the subdivision were approved in May 1989 based on the
possible existence of one sinkhole near Grand Street.  Evidence
of other sinkholes on the site was not encountered until design
topographic surveys were conducted.

     As a result of this new evidence, the city specified several
additional design considerations for the site.  Buildings were
prohibited within the floodplains of individual sinkholes.
Construction of basements was prohibited, additional
reinforcement was required for the foundations, and wider
easements between houses were required for possible future
enhancement of flood control.  The city, the developer, and the
design consultant agreed to several other precautionary measures:
installation of silt fences around sinkholes; planting of
vegetation to control erosion; prohibited use of heavy equipment
in the sinkholes; clearing of vegetation by backhoes situated
outside the sinkhole rim and extending only the bucket into the
sinkholes.  Figure 4 illustrates the site layout for the
subdivision and the location of the sinkholes.

     Prior to development, observations made during moderate to
heavy storm events indicated that the sinkholes were free
draining. During a 4 to 5 inch rainfall in approximately 10 to 12
hours  (May 1989), ponding was not observed in the large sinkhole
(S-l)  and only very moist soil conditions were evident in several
                                747

-------
00
                                                    ILLSFORDSJME*
                                	^	r	i L	.  ?

                                GRAND STREET  C	^
                                                                                             DRAINAGE
                                                                                               DIVIDE
                 Figure 4.  Sundance Estates subdivision, showing locations of sinkholes S-l through S-6.
                       Scale is approximately 1"=100'.

-------
other sinkholes (S-4 and S-3).  Stormwater design used the
Rational Method to determine peak flows and detention volumes
necessary for each sinkhole.  Total drainage to each sinkhole is
small, and for the proposed land-use, calculated volumes
accommodate this additional runoff.  Table 1 lists a summary of
sinkhole drainage characteristics. A retention basin was designed
at the north end of the property in order to prevent potential
off-site flooding. Recommended stormwater design measures were
preventive and not remedial in nature.

     Problems at the site began after clearing the sinkholes of
debris.  Cleared sinkholes were hydro-mulched but were not
watered to facilitate germination of the seed.  Work was
performed during the late fall, and temperatures dropped below
freezing before germination.  Sediment and seed were washed into
the centers of the sinkholes, and sinkholes remained void of
vegetation for several months.  Contrary to recommendations, the
contractor used heavy equipment in sinkhole areas for grading,
thus compacting the soil structure.  Lack of silt fence
maintenance allowed sediment to wash into the sinkholes. Heavy
equipment used in the large retention area compacted soils and
reduced infiltration capacity.  One house, constructed between
sinkholes S-2 and S-3, encroaches on sinkhole floodplains, and a
second house is located within the floodplain of sinkhole S-l.
Instead of enforcing design agreements, the city placed
responsibility on the design consultant to oversee all phases of
construction.

     Sundance Estates Subdivision is almost completely developed,
and all sinkholes have adequate vegetative cover.  During 1989
and 1990, flooding occurred repeatedly at the site.  Moderate
rainfall (1 to 2-inch rain in 6 hours) caused ponding of water in
the sinkholes, with water remaining up to 3 or 4 days in all
sinkholes.   Because of the flooding and possible structural
hazards, HUD has refused loans for this subdivision on lots with
sinkholes or adjacent to sinkholes.  Further evaluation of the
site during 1991 has been limited because of sparse rainfall in
the Springfield area.

                  Chestnut  Plaza  Phase I  (North)

     Chestnut Plaza Phase I, the first site affected by the new
sinkhole ordinance, is an approximately 30 acre tract located at
the corner of East Chestnut Expressway and U.S. Highway 65
(Figure 3).  Chestnut Plaza Phase I is planned for commercial
development, which is anticipated to include fast food
restaurants, a hotel, retail stores, and light industry or
manufacturing.

     During the 1930's this site was used as pasture land.
Review of 1936 aerial photos reveals that sinkholes at the site
were clear of any trees and that the railroad and the highway had
not been constructed.  Shortly after this time, construction of
the Burlington Northern railroad bisected the large sinkhole on
                                749

-------
Sinkhole Drainage Characteristics
SINKHOLE
S-1
S-2
S-3
S-4
S-5
S-6
SINKHOLE
AREA
0.12 ACRES
0.09 ACRES
0.11 ACRES
0.03 ACRES
0.01 ACRES
0.05 ACRES
SINKHOLE
DRAINAGE
VOLUME
5,800 CU. FT.
7,240 CU. FT.
2,600 CU. FT.
6,964 CU. FT.
SINKHOLE
DRAINAGE
AREA
1.60 ACRES
1.87 ACRES
1.53 ACRES
1.57 ACRES
CALCULATED
RUNOFF
VOLUME
2,400 CU. FT.
2,800 CU. FT.
2,300 CU. FT.
2,400 CU. FT.
       Table 1

-------
Sinkhole Drainage Characteristics
SINKHOLf
CN-1
CN-2
CN-3
CN-4
CS-1A
CS-1B
CS-2
SINKHOLE
AREA
3.6 ACRES
0.81 ACRES
21 6 FT2
0.15 ACRES
5 ACRES
10 ACRES
0.68 ACRES
SINKHOLE
DRAINAGE
VOLUME
535,000 CU. FT.
60,000 CU. FT.
N/A
N/A
800,000 CU. FT.
1, 600,000 CU. FT.
50,000 CU. FT.
SINKHOLE
DRAINAGE
AREA
27.5 ACRES
3.9 ACRES
N/A
N/A
50 ACRES
50 ACRES
10 ACRES
CALCULATED
RUNOFF
VOLUME
1 00,000 CU. FT.
1 5,000 CU. FT.
N/A
N/A
741, 000 CU. FT.
1, 263,000 CU. FT.
130,000 CU. FT.
        Table 2

-------
Ln
1-0
                                                           CHESTNUT PLAZA
                                                           PHASE I (NORTH)
                     STORMWATER
                     DETENTION
                        IN
                                                     SlpRMWATE
                                                     DfTJENTION BA
                 Figure 5. Chestnut Plaza Phase I (north), showing location of sinkholes CN-1 through CN-4
                      Scale is approximately 1"=160'

-------
the site.  Later construction of East Chestnut Street covered
another part of the sinkhole.

     During construction of the railroad and Chestnut Street,
excavated soil and other material were dumped into the sinkhole.
When Chestnut Expressway was widened, excavated soil, pavement,
and other materials were dumped into the sinkhole.

     Alteration of the large sinkhole severely restricted its
drainage capacity.  The sinkhole floods and ponds water for
several days after moderate rainfalls.  Observations made during
a 4 to 5-inch rainfall in January, 1990, revealed that stormwater
runoff rose to the 1354 elevation within the sinkhole.  This
level is below the established floodplain elevation of 1359, and
flood waters drained after 5 to 6 days.  This large sinkhole also
receives off-site drainage from a 30 to 40 acre site to the
north.

     Because sinkholes at this site had already been altered, it
was imperative that development would not further degrade
drainage of the sinkholes.  Although it contained a great amount
of fill material, the larger sinkhole still had the capacity and
drainage to accommodate increased runoff if some protective
measures were taken into consideration.

     Design of the stormwater detention areas utilized the
Rational Method for calculating stormwater runoff and peak flows
(Table 2).  FEMA had already established floodplain limits in the
larger sinkhole.  To protect the sinkhole and allow for future
development, stormwater detention berms were designed around the
two larger sinkholes (CN-1 and CN-2 in Figure 5).  No building is
allowed in the sinkhole areas.  To prevent soil compaction by
heavy equipment, clearing is permitted by hand only.

     Additional stormwater detention will be required for
individual sites along the northwest side of sinkhole CN-1 as
development occurs along Chestnut Expressway.  Minor filling has
been approved for the area adjacent to Chestnut Expressway.
Although two smaller sinkholes, CN-3 and CN-4, are not important
in the overall drainage of the site, extreme thickness of their
soils and lack of solid bedrock beneath them could lead to
potential collapse.  For these reasons CN-3 and CN-4 have been
designated "non-buildable" and will remain in their natural
state.

     Chestnut Plaza Phase I was approved for construction during
the summer of 1990.  Prior to any construction or clearing at the
site, sediment control methods were established.  These methods
included silt fences around sinkhole areas and also outside the
perimeters of their established floodplains. Completion of
detention berms around sinkholes CN-1 and CN-2 and establishment
of vegetation in these berms was required before construction
could begin.
                                753

-------
     Recommended protective measures have been established, and
the street and one warehouse have been constructed.  Additional
construction is expected within the next several months.
Observations made during a 2 to 3 inch, 12 hour storm indicate
confinement of drainage to the sinkhole and berms and
infiltration of stormwater within 3 to 4 days.  Additional
monitoring of runoff and sampling of stormwater are planned for
the site.

                     Chestnut Plaza Phase II

     Chestnut Plaza Phase II is located at the intersection of
East Cherry Street and U.S. Highway 65 and south of Chestnut
Plaza Phase I (Figure 6). This site encompasses approximately 30
acres and surface runoff drains southwest to an off-site sinkhole
(CS-1B).  Sinkholes CS-1A and CS-1B are two halves of an
originally continuous sinkhole which was bisected by construction
of the Burlington Northern Railroad.  Other modifications include
partial filling of sinkhole CS-1B for industrial development
during the 1970's.

     Figure 6 illustrates the complexities of the drainage
system.  Drainage area shown in this figure is approximately 100
acres.  All runoff is directed toward an outlet constructed in
sinkhole CS-1B.  When this modification was made in 1972, the
city's environmental geologist stated that the Chestnut Plaza
Phase II site was "undevelopable" (Giles, 1990, personal
communication).

     Initially sinkhole SC-1B was able to accept runoff from
1970's and early 1980's construction.  However, because design
modification included only a metal grate to stop debris, the
underlying cavern system has filled with sediment and is greatly
reduced in drainage capacity (Thomson, 1990, personal
communication).  Plans for Chestnut Plaza Phase II had to address
this problem.

     Figure 7 illustrates the design for Chestnut Plaza Phase II,
including the sinkhole area and proposed stormwater controls.
Because the developer desired the largest possible area for
building, stormwater detention basins such as those at Chestnut
Plaza Phase I were not used.  Instead, a two tier stormwater
network was designed to provide maximum buildable area while
controlling stormwater runoff.

     Because FEMA had not established a floodplain for the site,
floodplain information was calculated using the Soil Conservation
Service TR-55 Stormwater Method.  Pre-existing major alterations
of sinkhole drainage and sensitivity of the adjacent site to
flooding suggested the use of the TR-55 method, which allows more
detailed analysis of stormwater behavior.  This method utilizes
soil characteristics, land use, hydrologic characteristics, and
rainfall distribution and duration to develop runoff hydrographs.
Calculated floodplain elevation is 1353.5 feet and volume is
                                754

-------
LEGEND
   ,	i
SINKHOLES

SATELLITE SINKHOLES
PENBROKESOIL SERIES
APPROX. LOCATION OF WO
YEAR FLOOD PLAIN.
ELEV=1353.5
APPROX. LIMITS OF
SINKHOLE (EAST SIDE) BASED
ON TOPOGRAPHIC SURVEY
SINKHOLE DRAINAGE BASIN
                                      CRELD
                                      SOIL AERIES
                                                                               PROJECT SITE
                                                                               KEENO-ELDON
                                                                               SOIL SERIES
                        CRELDON
                        SOIL S5RIES
                                                                        SINKHOLE
                                                                        OUTLET TYP.

                                                                        KEENO-ELDON
                                                                        SOIL SERIES
           Figure 6. Map of Chestnut Plaza area, showing drainage basin for large bisected sinkhole
                (CS-1A and CS-1B). Scale is approximately 1"=500'.

-------
                                 HIGH WAY 65
                               HESTNUT PL

                               HASEIV (SOU
                                                                                      Uj

                                                                                      Uj
                                                                                      ce
                                                                                      Uj

                                                                                      a:
                                                                     fcw^:^^53atj«gs,a«y>i5ffl38«--
Figure 7.  Chestnut Plaza Phase II (south), showing sinkholes CS-1A and CS-1B  Scale is

      approximately 1"—170'.

-------
estimated at 1,000,000 cubic feet for the 100 year  flood storage
basin (Table 2).

     In addition to the 100 year flood volume, drainage volume
for projected complet development was computed using the Rational
Method,  and a detention basin was designed for this volume.  As
in Chestnut Plaza Phase I, both basins will be grass lined and
sediment control measures will be used during and after
construction.

     Outflow to the off-site drain was determined from field
measurement, and values obtained were further reduced to
accommodate runoff from future development.  Drainage design for
the site west of Chestnut Plaza Phase II allowed a discharge of
93 cfs from sinkhole CS-1B during a 10 year storm (Hayes and
Thomson, 1973).  Mapping of the Steury Cave system suggests that
sedimentation has reduced this discharge capacity by 90 to 95
percent (Thomson, 1990, personal communication).   Field
observations of sinkhole CS-1B made during a 2 inch, 8 hour
rainfall support this estimate.  Off-site discharge to CS-1B from
Chestnut Plaza II was reduced accordingly by 95 percent to a
design outflow of 5 cfs.

     Preliminary design for Chestnut Plaza Phase II was approved
in the summer of 1990.  Vegetation has been cleared at the site,
and silt fences have been placed at sinkhole CN-1B.  Construction
is expected to begin in 1992.  Runoff and water quality will be
monitored before and during construction.

                           Conclusions

     Case histories of three sites in Springfield, Missouri
illustrate results of contrasting approaches to stormwater design
and regulation in urban karst terrain.  Residential development
planned for a 10-acre wooded tract (Sundance Estates) required
minimal preventive measures to protect free-draining sinkholes
from construction activities and to maintain the drainage
capacities of those sinkholes.  Lack of adherence to design
recommendations has resulted in repeated flooding of sinkholes at
this site.

     Two adjoining commercial parcels are located in an area
which experienced extensive alteration of drainage during
development of neighboring sites. Because of existing flood
hazards and inflow of stormwater from an adjacent site, design
for the northern parcel (Chestnut Plaza Phase I)  included land
use restrictions, sediment control, and construction of detention
basins which approximate natural landforms.  Close communication
among the city, the developer, and the design consultant ensured
completion of stormwater measures as planned.  Observations made
during recent storms suggest that the design maintains adequate
drainage.  Future studies will examine the impact of the
detention basins on stormwater quality-
                                757

-------
     Because of off-site releases of stormwater and a desire to
maximize commercial land use, design for the southern parcel
(Chestnut Plaza Phase II) includes a two-tier stormwater network
as well as vegetative and sediment control measures. Construction
at this site is incomplete, and continued monitoring of runoff
and stormwater quality are planned.

                            References

Aley, T. and K. C. Thomson, 1981.  Hydrogeologic Mapping of
     Unincorporated Greene County, Missouri, To Identify Areas
     Where Sinkhole Flooding and Serious Groundwater
     Contamination Could Result From Land Development.  Report
     repared for Greene County Sewer District/U.S.Environmental
     Protection Agency.  Ozark Underground Laboratory, Protem,
     Missouri, lip.

Hayes, W. C. and J. D. Vineyard, 1969.  Environmental Geology
     of Towne and Country:  Missouri Geologic Survey and Water
     Resources, Rolla, Missouri, Educational Series Report No. 2.
     42 p.

Hayes, W. C. and K. C. Thomson, 1973.  Engineering and
     Environmental Geology of The Springfield Urban Area.
     Association of Missouri Geologists Twentieth Annual Meeting,
     Branson, Missouri, Sept. 21-22, 1973.  Geography and Geology
     Department, Southwest Missouri State University,
     Springfield, Missouri, 27 p.

Hayes, W. C., 1977. Urban Development in a Karst Terrain -
     Springfield, Missouri.  Report prepared for the Department
     of Housing and Urban Development, 64 p.

Thomson, K. C., 1986.  Geology of Greene County, Missouri.
     Report prepared for Watershed Management Coordinating
     Committee, Springfield, Missouri.  Southwest Missouri State
     University, Springfield, Missouri, 86 p.

Turner, C. R., 1989.  Stormwater Management in Springfield, Now
     and in the Future.  Proceedings of the Third Annual
     Watershed Conference, Springfield, Missouri, June 14-15, pp.
     51-54.
                                758

-------
TITLE:  CASE HISTORIES OF SEVERAL APPROACHES TO STORMWATER
MANAGEMENT IN AN URBANIZED KARST TERRAIN, SOUTHWEST MISSOURI

AUTHORS:  Wendell L. Earner - ICF Kaiser Engineers
          Patricia Miller - California State University,
          Sacramento

Question:  Is dye-tracing a reguirement  (or is it done) prior to
zoning or development?

Response:   The sinkhole ordinance adopted in 1989 does not
specifically require or mention dye tracing. However, the
ordinance does specify substantial "state of the art" field
studies and evaluation of the sinkhole system.  Therefore, dye
tracing may be performed by the developer or the consultant to
determine sinkhole drainage characteristics and impacts on the
groundwater system from the proposed development.  Other
investigators have performed dye tracing in the Springfield area,
and their results are on file with the Missouri Geological
Survey, as well as with local governing agencies.  Many traces in
the Springfield area were executed prior to, during, or after
modifications to sinkholes for stormwater drainage.
                                759

-------
760

-------
                     CONTAMINANT INVESTIGATION IN A KARST REGION
                                     J. E. BENTKOWSKI, P.O.

                                     METCALF & EDDY, INC.
                                       ATLANTA, GEORGIA
A series of springs in the karst region of north central Kentucky appeared to have been contaminated.
These springs are within a half mile of two sinkholes, which were filled in as permitted landfills for inert
waste and then developed into a medium industrial park. A preremedial site inspection was performed under
the authority of the Superfund laws in late 1989. A preliminary site visit included site reconnaissance and
geologic field work to locate the springs. A review of historical aerial photos aided in the planning of the
investigation program, which consisted of magnetic and soil gas surveys and environmental soil and water
sampling.  The magnetic survey indicated the presence of buried ferrous objects. The soil gas survey points
were laid out incorporating this information. Soil sample locations were selected based on the results of the
soil gas survey. Seventeen surface and subsurface soil samples were taken. Eleven water  samples were
taken from various springs, rivers, and the local public water supply. The analytical results of the soil
samples taken over the sinkholes matched 20 compounds also found in the water and sediment samples
taken from the spring. The location of the springs roughly coincided with the strike of the major fracture
systems reported in the literature. The success of this investigation emphasizes the  importance of proper
geologic consideration for contaminant monitoring  in karst regions.

Introduction

This investigation was part of the phased investigative approach that is employed by the USEPA to
determine whether sites warrant being placed on the National Priorities List. The purpose of this particular
investigation was to determine the nature of the contaminants present at the site and to determine if a
release of these substances has occurred or is likely to occur (USEPA, 1991). Additionally, the potential
pathways for contaminant migration were investigated.  This paper describes the methodology that was
employed  to meet the stated purpose.

It is important to note several limitations, both organizational and geologic,  that were key in formulating the
approach to this investigation. This Screening Site Inspection (SSI) was  done under the authority of the
USEPA, Region IV,  as part of a preremedial Superfund investigation. These SSIs are limited in scope and
concentrate on source characterization and identification of potential pathways for contaminant migration.
Permanent groundwater monitoring wells are not part of normal SSIs. Temporary groundwater monitoring
wells were not considered for two reasons. First, the site is located adjacent to a bluff that overlooks Valley
Creek and Valley Creek reservoir (the closest surface water bodies). The top of the bluff is approximately
                                             761

-------
65 feet above the water level in Valley Creek reservoir. This physically precluded the installation of the
temporary wells by the investigation personnel.  Second, in mature karst areas,  such as around
Elizabethtown,  Kentucky, the majority of the groundwater,  from 60% to 80%, would be expected to flow
in the dissolution features in the limestone (conduit flow), not as diffuse flow through the limestone itself
(Quinlin and Ewers, 1985). These dissolution features cannot be readily encountered when drilling. A much
more reliable approach towards monitoring potentially contaminated groundwater is to inventory and
monitor springs in the areas  as they are the discharge points for the groundwater (Quinlin and Ewers,
1985). This idea is the basis for the technical approach taken in this investigation.

Background

From 1969 to 1977, a landfill was permitted to receive inert waste to be disposed of in a series of sinkholes
located approximately 1/2 mile southeast of Elizabethtown. The site is approximately ten acres in size (1800
feet by 1000 feet) and contains at least three sinkholes.  The site is bounded approximately on the west by
Interstate  65 and on the north by the valley of Valley Creek (Figure 1). This creek flows to the west
through Elizabethtown.
                                                                                              N
                                        Figure 1. Study Area
                                       Elizabethtown, Kentucky

The permitted wastes for this landfill included Neoprene, cardboard, wood, paper, foundry sand, bricks,
wood palettes, and empty barrels. An inspection in 1972 by the Kentucky Department of Environmental
Protection indicated that the landfill was receiving liquid wastes and domestic garbage in violation of its
permit. Regarding the liquid wastes, a galvanizing company reportedly had a pipeline for the disposal of
spent sulfuric acid into one of the sinkholes.  USEPA files raise the question of the disposal method of 1200
drums of dross zinc ash by the galvanizing company. These files also raise the possibility that a tanker
truck was buried in the largest of the sinkholes (USEPA, 1991).

Contamination of a spring on a farm immediately across 1-65 from the landfill was noted in the early  1970s.
Investigations by both State and Federal agencies indicated the presence of elevated values for zinc, nickel,
chlorinated compounds, and other organic compounds as well as low pH values for the spring water. The
landfill was officially closed in 1977. This investigation took place in 1989 (USEPA, 1991).
                                               762

-------
Geology

The Elizabethtown area is in the Interior Low Plateaus physiographic province (Fenneman, 1938). The
Pennyroyal Plain is a gently rolling mature plain that slopes towards the southwest and is characterized by
such karst features as sinkholes, springs, and sinking streams (Sauer, 1928). Valley Creek, the only
perennial river in this area, flows along the northern boundary of the study area.

The discussion of the geology of this area will be limited to describing the surficial soils and the carbonates
of Mississippian age which control the development of the karst features. The alluvium in this area is
confined to Valley Creek and its feeder streams. The bedrock of the area is covered by a blanket of
residuum of clay and chert fragments. This layer is of variable thickness; it fills in the dissolution and
collapse features of the underlying limestone (Mull et  al, 1988).

The primary limestones of the area are the Ste. Genevieve and the underlying St. Louis. These formations
are separated by a relatively thin bed of limestone  and silicified fossils known as the Lost River Chert. The
Ste. Genevieve is typified by occasionally shaly oolitic limestone and dolomite beds varying from 0.5 to 4.0
feet in thickness (Mull et al, 1988). The St.  Louis  is the most significant formation with regards to the
development of the dissolution features. The upper portion of the formation consists of primarily massive to
thin-bedded limestone and dolomite.  In the lower portion of the formation, there is an increase in the
amount of gypsum and anhydrite. Noger and Kepferle (1985) believe that the presence of these soluble
minerals is what lead to the extensive development of the mature karst features  seen in the study area.

Locally, the confirmation of the rock units was not possible due to the limited outcrop in the area  and the
lack of subsurface or bedrock drilling. More important is the hydrogeology of the site as typified by the
filled-in sinkholes, the adjacent springs and the elevation difference  between the land around the sinkhole
and the water  table, as represented by the springs and Valley Creek. As seen in Figure 1, the landfill was
located in the  largest of the three sinkholes east of Interstate-65 and south of Valley Creek. This plateau is
bordered on the north by the alluvial valley of Valley  Creek and rises approximately 65 feet above the
valley floor. Across the Interstate to the west are five  springs that originate from the base of a hill and flow
over the alluvium to Valley Creek.

The groundwater flow patterns of this area have been  the subject of considerable study. A recent work by
the USGS (Mull et  al, 1988) describes dye tracing studies that were performed in the Elizabethtown area.
Their studies include the documentation of groundwater flow direction towards Valley Creek and the
suggestion that shallow groundwater flow follows a major fracture joint set that trends N40W. The springs
in this study are approximately along this trend from the landfill/sinkhole.
Technical Approach

The basis of the technical approach for this investigation is a combination of modified procedures for
monitoring groundwater quality in karst areas presented by Quinlin and Ewers (1985) and the standard
investigation methods used for preremedial investigations conducted under authority of the USEPA. The
essence of Quinlin's and Ewer's recommendations is that the majority of the groundwater flow in karst
areas  is in the dissolution features. The probability of intersecting these dissolution features with a
monitoring well is very low. Therefore, the only effective method for groundwater monitoring is to locate
all the springs and seeps  along riverbanks and hillsides, plus other accessible points of intersection with the
water table, and monitor those locations. Dye tracing studies should be an integral part of every sinkhole
investigation, but they were not possible in this case, becuse the sinkhole had been filled in. There was no
surface expression of the dissolution or collapse features.
                                               763

-------
This investigation consisted of the following parts:

        1. Site reconnaissance

        2. Project planning

        3. Magnetic survey

        4. Soil gas survey

        5. Sampling of environmental media to include:
                a. surface soil
                b. subsurface soil
                c. spring water
                d. spring sediment
                e. municipal supply
                f.  appropriate background for each media

        6. Data reduction and report preparation

Site Reconnaissance

After reviewing the file, a trip was arranged to Elizabethtown to view the site, perform field geology
studies, and talk to local officials. The field work was done in October, 1988, following a dry summer. The
topographic map indicated the location of the sinkholes prior to being filled. The area is now a small
industrial park with the main sinkhole area being a storage area for large cement drainage pipes. Across the
Interstate, four springs were located at the base of the hill. A spring house was noted at one of the springs.
Also noted was a dry spring. Two large sinkhole depressions were noted at the top of the hill, above the
springs. Contact was made with the Elizabethtown Water Department to arrange permission to take water
samples from the municipal  springs and  wells later in the year. The local office of the Department of
Transportation was contacted regarding the availability of historical aerial photos that might aid in the
planning of the investigation. While this office had some coverage, the investigators were referred to the
headquarters office in Lexington for the  most complete historical coverage of the area.

Project Planning

With the information provided by the file material and the field reconnaissance, the project plans were
formulated. Because the sinkholes were used for illegal dumping and the contents and depth of burial were
not documented, a magnetic survey would provide an nonintrusive method of investigation.
Groundpenetrating radar would have been the first choice, but it was ruled out due to the high clay content
of the soils. Additionally, the file materials indicated that a tanker truck may have been buried in the largest
sinkhole. A magnetic survey would give some indication of its possible location, as well as provide some
measure of safety for the subsurface soil samples to be taken later. The file material  indicated that
chlorinated compounds had been detected in the adjacent springs. It was assumed that these compounds
were leaching from the sinkhole contents and that a soil gas survey would detect the  presence of these
compounds. This would ensure that soil  samples were taken in the most appropriate location and allow the
efficient investigation of such a large area. The presence of chlorinated compounds, specifically vinyl
chloride, demanded that the sampling of the spring water and sediment be planned as a Level B exercise.
Appropriate access  was requested and the field work was performed in early December, 1988.

Magnetic Survey

The magnetic survey was conducted with a proton precision magnetometer. Survey grids were laid out
above two sinkholes with 28 stations for the eastern sinkhole and 20 stations for the  northern sinkhole.
                                              764

-------
(Access permission was not granted in time to perform a survey over the southern sinkhole). All
appropriate calibrations were performed and standard methodologies were followed. The results of the
eastern sinkhole indicated two positive anomalies in the center of the sinkhole and four negative anomalies
around the edges of the grid. These results may be interpreted as indicating the presence of buried man-
made, ferrous objects in the center of the sinkhole. The two positive anomalies  align in such a way that
they may be interpreted as indicating the presence of the suspected tanker truck. The results of the northern
grid only indicated one positive anomaly along the southwest corner of the grid. This was interpreted as
indicating buried man-made, ferrous objects only in the southwest area,  with the rest being nonferrous fill
material (USEPA,  1991).

Soil Gas Survey

The soil gas survey was conducted using a photoionization detector  (HNu), which is designed for the direct
reading of the presence of volatile organic compounds. Soil gas probes were installed to an approximate
depth of three feet in eleven locations in the eastern sinkhole and five locations  in the northern sinkhole. On
the basis of this survey, locations of the soil samples were selected. A third survey was attempted in the
area of the southern sinkhole. Access problems delayed the performance of work in this area. The weather
became extremely unfavorable with temperatures in the 20s, blowing snow, and winds in excess of 15 miles
per hour. An attempt was made to perform a soil gas survey, but the author believes that more favorable
conditions would have allowed a more successful attempt.

Taking of Environmental Samples

A total of 17 soil and 11  water  samples were taken during this investigation. The breakdown of the sample
types follows:
* A total of nine surface  and subsurface soil samples were taken from the three sinkholes.
* A total of four water and two sediment samples were taken from the  five springs located at the base of
the hill across the interstate. Lack of water or lack of sediment, combined with temperatures in the mid-
teens, precluded the completion of the entire planned sampling program in this area.  The cold temperatures
did allow the downgrading to modified Level D due to the lack of volatilization as indicated by the air
monitoring equipment.
* Six water samples were taken from the Elizabethtown municipal water system.
* A water sample and a sediment sample were taken as background from Valley Creek Reservoir located
northeast of the site.  A total of  two surface and two subsurface soil samples were taken as background
samples: one set onsite near the sinkholes and one set on the  farm where the springs were located.

All samples were taken observing USEPA Region IV sampling protocol.
Analytical Results

Over 150 compounds, including volatiles, semivolatiles, pesticides, PCBs, acids/base/neutrals, and
inorganics were analyzed as part of the USEPA-approved Contract Lab Program. There were numerous
compounds present above detection limits. Some were present only in the soil samples taken above the
sinkholes. Some were present only in the samples taken at the springs. The set of compounds that were
detected in both locations numbered 20:  11 organic compounds  (Table 1) and nine inorganic compounds
(Table 2). The analytical results for this set of compounds are listed below with their respective sample type
and maximum concentration detected.
                                              765

-------
                                         TABLE 1
                      COMPARISON OF ORGANIC ANALYTICAL RESULTS
                                 SOIL AND WATER SAMPLES
                               ELIZABETHTOWN, KENTUCKY
PARAMETERS
(ug/kg)
ACETONE
CARBON BISULFIDE
1 , 1-DICHLOROETH ANE
1 ,2-DICHLOROETHENE
1,1,1-
TRICHLOROETHANE
TRICHLOROETHENE
BENZENE
METHYL ISOBUTYL
KETONE
TETRACHLOROETHENE
TOTAL XYLENES
TOLUENE
BACKGROUND
SOIL
12UJ
6U
6U
6U
6U
6U
6U
12UJ
6U
6U
6U
BACKGROUND
WATER
10U
5U
5U
5U
5U
5U
5U
10U
5U
5U
5U
SINKHOLE
SOIL
(MAXIMUM)
320J
2J
8
45J
6
2J
42
220J
20
66N
190
SPRING
SEDIMENT
(MAXIMUM)
630J
13J
11J








SPRING
WATER
(MAXIMUM)


110
520
67
110
4J
3J
26
8J
4J
                                         TABLE 2
                      COMPARISON OF INORGANIC ANALYTICAL RESULTS
                                 SOIL AND WATER SAMPLES
                               ELIZABETHTOWN, KENTUCKY
PARAMETERS
(ug/kg)
CADMIUM
CHROMIUM
COPPER
IRON
LEAD
MANGANESE
NICKEL
VANADIUM
ZINC
BACKGROUND
SOIL
1.1U
21J
6.7
17,000
15J
410
13J
SOU
30U
BACKGROUND
WATER

10U
22U
160U
60UJ
65
16U
16U

SINKHOLE
SOIL
(MAXIMUM)
2.6
34J
190
63,000
1100J
410
130J
92
860
SPRING
SEDIMENT
(MAXIMUM)
19,000
33J
11,000
24,000
9700
340
24J

12,000
SPRING
WATER
(MAXIMUM)

18
26
76,000
170J
4500
26
39
39
NOTES
       Material analyzed for but not detected above minimum quantitation limit (MQL)
i      Estimated value
U      Material analyzed for but not detected. The number given is the MQL
N      Presumptive evidence for the presence of material
                                           766

-------
Conclusions

The objectives of this investigation included the source characterization of suspected waste areas, the
identification of releases or suspected releases to the environment of hazardous materials, and the
establishment of a hydrogeologic interconnection between the sinkholes/landfill and the adjacent springs.
The varied analyses did characterize  the sources as to the type of material that was placed in the
sinkhole/landfill. The spring water and sediment analyses did indicate that hazardous materials were being
released into the environment. The matching of 20 compounds between the sinkhole/landfill and the springs
does indicate a connection between the two areas. Quinlin's and Ewer's basic premise that the key to
monitoring groundwater quality in karst areas is to monitor the springs has been demonstrated.

References

Fenneman, N.M.,  1938, Physiography of the Eastern United States: New York, McGraw-Hill Book
Publishing Company, Inc., p. 714.

Mull, D.S.,  J.L. Smoot, and T.D.Leiberman, 1988, Dye Tracing Techniques Used to Determine
Groundwater Flow in a Carbonate Aquifer System Near  Elizabethtown, Kentucky, Water Resources
Investigations Report 87-4174 (Louisville, Kentucky: U.S. Geologic Survey), p. 95.

Noger, Martin C. and Roy C. Kepferle, 1985, Stratigraphy along and adjacent to  the Bluegrass Parkway:
Geological Society of Kentucky Guidebook, October 18-19, 1985, Kentucky Geological Survey, p. 24.

Quintan, James F.  and Ralph O. Ewers, 1985, Ground Water Flow in Limestone  Terranes: Strategy,
Rational, and Procedure for Reliable, Efficient Monitoring of Ground Water Quality in Karst Areas. In:
National Symposium and Exposition on Aquifer Restoration and Ground Water Monitoring (5th, Columbus,
Ohio, 1985), Proceedings, National Water Well Association, Worthington, Ohio,  p. 197-234.

Sauer, C.O., 1927, Geography of the Pennyroyal: Kentucky Geological Survey, Series IV, v. 25, p. 303.
USEPA Region IV, 1991. Screening Site Inspection, Phase II, Kentucky  Industrial Haulers, Elizabethtown,
Hardin County, Kentucky, prepared by NUS Corporation, pp. 44.
                                             767

-------
Contaminant Investigation in a Karst Region

J. E. Bentkowski
Q.      How do you know each or any of the springs or surface water sites is hydraulically connected to the
        landfills?
A.      There have been no die traces performed in this exact area near Elizabeth town.  Given the hazardous
        nature of the spring water, the flushing of the karst system would be a questionable procedure.  The
        connection between springs and the landfills is demonstrated by the empirical evidence of the 20
        compounds which were detected in both the springs and the landfill soils.
                                                  768

-------
         LAND-USE PLANNING AND WATERSHED PROTECTION IN KARST TERRANES

                                 Paul A.  Rubin
              Oak Ridge  National  Laboratory,  Oak Ridge,  Tennessee

                                   Abstract

    It is critical that the special characteristics of a karst watershed
(groundwater basin) be examined when any development within it is to be
considered.  Local planning boards are most likely not aware of the radically
different flow dynamics of karsts, and therefore they may unwittingly allow
development that leads to mas£ive contamination.  A practical approach for
ascertaining the sensitivity of a basin to groundwater contamination,
utilizing relatively easily implemented and practical procedures, is
described; it will aid land-use planners.  Although the approach is kept
fairly basic, it is comprehensive.  Not only is the hydrology of the karst
aquifer characterized, but any present and future sources of contamination are
evaluated and compared with the estimated assimilative capacity of the
aquifer.

    An evaluation of the potential impacts of development should also examine
the types and amounts of contaminants which could be introduced into a system,
the likely chemical partitioning of these contaminants, and the combined
loading to the aquifer or stream receptor.  Consideration of the present
health of receptor streams should be part of the process.  Water chemistry
parameters, combined with indices of biotic integrity and ecosystem health,
can be used to evaluate the present condition of the ecosystem.  This
important base-line information should then be incorporated into a
comprehensive and practical management strategy.

    The Mill Pond karst basin, situated in east-central New York State, is an
example of an environmentally sensitive karst aquifer.  Recently zoned as
Rural Commercial, this groundwater basin will be used to illustrate the
practical application of accepted hydrologic techniques in land-use planning.
Reference to several other localities will provide further insight into the
evaluation process.


                                 Organization

    This paper is organized as follows: 1) Introduction, 2) The land-use
planners point of view:  why does a karst region deserve special consideration,
3) Assimilative capacity of the karst aquifer: how much is too much, 4) A
typical development proposal, 5)  Characterization necessary for zoning
decisions, 6) Practical land-use planning: two examples, 7) Advanced testing
and analysis to define a karst system requiring a karst hydrologist, 8)
Political realities of the land-use permitting process, and 9) Who to look to
for help.


                                 Introduction

    Groundwater in soil and most fractured bedrock aquifers moves slowly,
enabling contaminants to be partially treated and diluted.  Karst aquifers
(comprised of dissolutionally enlarged fractures and cave passages), on the
other hand, are often characterized by appreciable and sometimes rapid
                                     769

-------
groundwater flow (Quinlan, et al., 1991).  They have virtually no ability to
treat water-borne contaminants, instead they merely transmit contaminants,
much as a sewer pipe would (Ford and Williams, 1989).  In some karst aquifers,
contaminated groundwater can move from one end of the aquifer to the other in
less than one day.   Failure to recognize the nature of karst systems, and to
plan accordingly, is likely to lead to the degradation of the aquifer, its
springs, and its dependent aquatic ecosystem.  Chemical loading, and possible
resulting deoxygenation of karst groundwaters during periods of base flow, may
result in destruction of fauna, including game fish, downstream of aquifer
discharge points (i.e., springs).  Contaminant loading of a karst system and
its receiving stream may also result in cave habitat destruction, impairment
of the character or quality of important aesthetic and recreational water
resources enjoyed by residents and the community (e.g., streams), and the
possible infringement of riparian rights of domestic and agricultural users.
It is a difficult,  expensive, and sometimes an impossible proposition to
remediate a karst aquifer after it and its receiving stream have been
degraded.  The development of an area must be within the natural constraints
of its geology and hydrology.

    Urban expansion is causing greater pressure for land development in many
karst regions.  It is critical that the special concerns of karst watersheds
be addressed by local planning boards, environmental conservation departments,
and other state agencies.  For example, towns in the Helderberg Plateau area
of New York continue to receive applications for single residences and planned
unit developments to be atop maturely karstified limestone aquifers.  Optimal
land-use planning would dictate the evaluation of an entire karst system
before permitting additional development in any isolated segment of an
aquifer.  This would include characterization of the extent of the aquifer
watershed (its basin boundaries), geology, subsurface flow paths, depth to
groundwater, discharge area, and base flow.  Only after such characterization
is complete, can the net effects of multiple contaminant inputs on the
assimilative capacity of an aquifer be judiciously evaluated.  It is important
that evaluations of this nature not only define the dynamics of a karst
system, but that they then utilize the knowledge gained to protect subsurface
and surface water resources.  Unfortunately, many of the conclusions about
what is necessary for protection of the aquifer will, like zoning, be regarded
by some to be an illegal taking of their rights.

    Successful land-use planning in karst terranes involves two important
criteria.  First, there needs to be detailed analysis of the area's geology,
hydrology, and biology.  Of equal or greater importance, however, is the
willingness of land-use planners and their counterparts to become
knowledgeable and involve themselves with the actual scientific issues and to
resist strong pressures to permit land-use beyond the assimilative capacity of
the karst aquifer.   Without these elements, the analytical process may amount
to little more than good intentions.  A positive outcome of this process can
be obtained when karst hydrologists work with land-use planners.


                     The Land-Use Planners Point Of View:
             Why  Does A Karst Region Deserve Special  Consideration?

    The cleanup or remediation of contaminated water supplies or streams can
cost a municipality upwards of a million dollars.  Prudent land-use planning
can avoid this; not only eliminating the unnecessary and avoidable expenditure
of taxpayers' dollars, but also saving a great deal of environmental
destruction of a precious resource.  Excessive development in environmentally
sensitive karst groundwater basins may lead to contamination, flooding, and
subsidence problems.  This paper discusses water quality concerns.  Too much
of any contaminant introduced into a karst basin is likely to result in
degradation of groundwater resources and the water body (i.e., stream, river,
or lake) receiving the basin's discharge.  Contaminant inputs to karst
groundwater basins from sources such as leach fields, landfills, sinkhole
dumps, animal feed lots, and industrial disposal operations may combine to
                                      770

-------
exceed the assimilative capacity of the aquifer and receiving springs and
streams.  Since anything that goes into a karst system comes out virtually
untreated, dissolution conduits (e.g., cave passages) can function as large
flowing sewer pipes.  Worst-case examples of this problem are numerous and
include visual and olfactory evidence of septic waste and industrial
contaminants (i.e., gasoline, toilet paper), destroying ecosystems so that
they support little to no life.  This now-polluted water is often critical for
water supply, land value, and recreational purposes (e.g., fishing, swimming,
aesthetic quality).  A further consideration may be adverse human and animal
health impacts that may result from ingestion or contact.

    Once particular land-uses are permitted and in place (e.g., large housing
clusters, factories, dry cleaners, service stations, waste treatment centers,
feed lots, liquid waste spreading, etc.) it is difficult, if not impossible to
terminate or relocate them and retro-fitting facilities for reduced effluent
generation can be costly.  Yet, as history has repeatedly shown, failure to
properly consider the ramifications of excessive contaminant loading can
result in multi-million dollar evaluation and remediation costs.  In such
settings, where karst springs and related surface streams are used for
residential, agricultural, and municipal water supplies, the first action
required is the replacement of the primary water supply.  This involves a
lengthy, and potentially costly, process of alternative water supply source
and site selection, full geologic, hydrologic and chemical characterization,
design and construction of necessary treatment and delivery systems.  In some
cases, the contaminated supply can be treated or piping extended to
alternative existing sources.  This situation is happening with increasing
frequency and often costs millions of dollars.  The replacement of a water
supply can sometimes require years of planning, exploring funding options,
evaluation, and set-up.  One potential solution, perhaps of equal cost, is the
funding, planning, and installation of a sewer district.  Another reality
which may bear on the land-use planning issue is the legal responsibility of
the planners.  This consideration may weigh heavily in the future as
litigation and pollution-related costs rise.  Even now, corporate executives
are being held personally responsible for their actions; with some spending
time in jail.  The best time for planning for responsible land-use is prior to
significant developmental pressures.


       Assimilative Capacity  Of The  Karst Aquifer:  How  Much  Is  Too  Much?

    This is one of the more important questions in land-use planning and
watershed protection decisions.  The answer to this question should logically
dictate what are reasonable and prudent land-uses in a karst basin.
Ultimately, the relative assimilative (carrying) capacity of individual karst
basins must be determined on a case by case basis.  In many aquifers,
especially those capable of limited contaminant dilution, the assimilative
capacity is nearly zero.  Many factors must be considered when addressing what
types and amounts of contaminants should be permissible in a groundwater basin
within a karst terrane.  Some of these include: 1) presence or absence of
sewers, 2) the type and amount of contaminants (present and planned) incident
to the karst system, 3) the amount of waste pre-treatment, if any, 4) the
amount of natural cleansing,  if any, imparted by overlying soil or bedrock
units prior to conduit entry, 5) an assessment of artificial diversion of
water into or out of the basin, 6) the maturity and sensitivity of the karst
aquifer, including the extent and importance of the upper part of the
percolation zone (epikarst),  7) the physical characteristics of conduit flow
paths during base flow conditions (e.g., sluggish movement through
intermittent pools), 8) residence and travel time of contaminants in
subsurface and surface flow paths, 9) water temperature, 10) the base flow and
draught flow conditions in the aquifer, 11) the relative sensitivity and
discharge of the receiving waterway (does the spring comprise the total flow
of a stream in the headwaters of a watershed, or does it enter a stream or
water body capable of providing significant dilution?), 12) the rate of
deoxygenation and reaeration of the receiving water body, 13) the contaminant
                                     771

-------
loading and biologic health present in the receiving water body, and 14) the
current and planned uses of aquifer and receiving stream water.

    The dynamics of the contaminant degradation and dilution which occurs,
both within and downstream of springs, is critical in the assessment of the
assimilative capacity of a karst basin.  The quality of streams and rivers
used for waste-water dilution depends on natural self-purification to
assimilate wastes and restore its own quality.  A waterway's capacity to
recover from influent waste is largely determined by channel morphology and
climatic conditions (Hammer, 1975).  Standard environmental calculations
inclusive of the rates of biochemical oxygen demand (BOD) and rates of
deoxygenation and reaeration can be used as an aid in establishing acceptable
aquifer loading criteria for some contaminants; although it must be recognized
that both flow and atmospheric conditions in the karst system may vary
appreciably from that of surface water.  This type of procedure can be
particularly useful for back-calculating likely effects of sewage disposal,
such as fish kills and eutrophication.  Base line chemical and hydrologic data
is essential for this type of calculation.


                        A Typical Development Proposal

    In order to address the types of impacts existing or planned development
may have on a karst area, a number of facts and interpretations must be
considered.  Once an isolated area is proposed for a specific land-use, the
planner must determine whether it is hydrologically reasonable and prudent to
base planning decisions on a site-specific assessment.  More often than not
the parcel in question is really only a small segment of a much larger
geologic and hydrologic framework.

    In the rural Town of New Scotland, New York, for example, a 28 acre parcel
was proposed for Rural Commercial zoning.  The parcel is situated proximal to
the Stewart's label on Figure 1, and incorporates a portion of a junk yard
property and a segment of the Onesquethaw Creek channel.  The proposed zoning
allowed for multiple uses of the parcel including: shopping plazas,
laundromats, public works garages, solid waste disposal facilities, sewerage
systems, car washes, restaurants, etc.  The maximum allowable building size
proposed was 9 acres.  This rural area does not have a sewer district.

    The area is maturely karstified as shown by a lack of surface drainage for
most of the year.  Much of the watershed that includes this parcel consists of
exposed limestone, with a thin and sometimes absent soil-mantle.  An extensive
cave system has been explored and mapped throughout part of the watershed.
There are also wells which intercept water flowing in cave passages which are
accessible through natural fissures.  These wells and others that encounter
caves during drilling, are referred to as conduit wells.  Cave streams and the
related spring (aquifer discharge point) have flashy hydrograph responses to
storm and runoff events, thus indicating turbulent flow in conduits typical of
karst aquifers.

    In an effort to promote a convenience store (Stewart's), the Town decided
to advance the Rural Commercial segment of the Comprehensive Land Use Plan
still under preparation.  The Town did not contract for any detailed studies
of the hydrology and geology of the site or of its surroundings.  A public
hearing was held, with the author presenting the results of his geologic and
hydrologic study of the karst basin.  The basin was characterized by him as
environmentally extremely sensitive.  Soon thereafter the Rural Commercial
district was approved, Stewart's applied for a building permit, which was
granted and the convenience store has now been constructed.  Although no
comprehensive basin-wide zoning approach was adapted, a written and oral
record of known environmental concerns was created.
                                      772

-------
                      — 550
  1670^
                                              MIL
                                              POND
                                                   GAGING
                                                   STATION
                                                   -605
                                               BENNETT
       FEET
          O
          A
ELEVATION MSL (FT)

CAVE OR SPRING
WATERSHED IDENTIFIER

WATERSHED BOUNDARY

HELDERBERG
ESCARPMENT

RISE DIRECTING
MELTWATER FLOW
COPELAND
  HILL
FIGURE 1: TOPOGRAPHY, DRAINAGE BASINS, AND SELECTED FEATURES ALONG THE HELDERBERG
        ESCARPMENT, ALBANY COUNTY, NEW YORK. WATERSHED BOUNDARIES BASED ON TOPO-
        GRAPHY, GEOLOGY, AND DYE TRACES. CONTOUR INTERVALS VARIABLE FOR CLARITY.
                                     773

-------
                Characterization Necessary For Zoning Decisions

Evidence of the Presence of Subterranean Karst Networks

    The first step toward protecting karst aquifers is to determine whether
carbonate or other soluble (i.e., halite, gypsum) bedrock is present.  When
the answer is in the affirmative, an assessment of the nature of the aquifer
is in order.  Physical evidence of the presence of subterranean karst networks
includes: 1) large and small enterable caves; 2) sinkholes; 3) areas of
exposed carbonate bedrock with numerous dissolutionally enlarged joints
(fractures); 4) limited to non-existent surface drainage in select areas; 5)
piracy (sinking) of surface water into sinkholes and streambeds; 6) presence
of open conduit and alluviated springs; 7) presence of overflow springs; 8)
variable chemical composition of springs and conduit wells; 9) rapid
subsurface flow as documented by tracer tests; 10) boreholes having
encountered large and/or small conduits; 11) loss of circulation fluids into
conduits found during drilling operations; 12) highly variable and abnormally
large water table fluctuations in some bedrock wells (especially those with
documented cavities).


Karst Inventory

    A full inventory of karst features (as indicated above) is needed to
monitor the karst systems recharge, discharge, water quality, and
environmental sensitivity.  Such an inventory is important in focussing future
site work, in land-use planning, and in determining appropriate sites for
aquifer characterization (e.g., sinkholes, springs, conduit wells).  A number
of additional sources of information should be explored in assessing the karst
features of an area.

    Published geologic and hydrologic reports are the most valuable sources of
information, but recent and historic topographic maps are also useful.
Historic topographic maps often show sinkholes or sinking streams which recent
land-use practices may have obliterated.  Examples include the filling in of
sinkholes with soil or waste materials and physical alteration of sinking
streams to create ponds.  Although topographic maps are good sources of
information, they often have many errors.  Errors can include the connecting
of stream segments where streams are pirated into the subsurface or the
failure to portray numerous shallow sinkholes because they fall within the
map's contour interval, have been filled in, or were not recognized by the
photo-interpreter making the topographic map.

    When available, historic and recent aerial photography can supplement or
correct topographic map errors.  Low-cost stereo pairs can be obtained from a
variety of sources including the United States Department of Agriculture, the
United States Geological Survey (USGS), the National Archives, and assorted
state and local offices.  Often, local aerial photography firms can provide
previously flown stereo coverage.  Ideally, large scale photography taken at a
time of year without leaf or cloud cover is desirable. Interpretation of karst
features is best achieved through the use of a mirror stereoscope with
attached binoculars, although a simple lens stereoscope will suffice.  Aerial
photographs can also be used to define the topographic boundaries of a
watershed and assist in determining sinkhole growth rates, but topographic
boundaries do not necessarily coincide with groundwater basin boundaries.
Usually they do not.

    One of the more important resources for learning of karst features in an
area is people: local homeowners, farmers, hunters, fishermen, well drillers,
geologists, and children.  Those individuals who roam or work an area for many
years can often provide information which might otherwise be overlooked.
Examples include small sinkholes filled in by farmers, turbid well water
coincident with storm events, caves played in by youths, and wells
intersecting conduits.  Drillers' and geologists' logs from well and
                                      774

-------
foundation studies may provide valuable insight into a karst system.
Occasionally a well may encounter a cavity, a zone of increased permeability,
or a zone where drilling fluid circulation was lost.  These locations, when
properly monitored, may provide important information on the hydraulics and
water chemistry in a system.  This is particularly true when they are
determined to be within a cave conduit.  Such conduits may also be used as
tracer injection or monitoring points.

    Homeowners can often provide important information about a karst aquifer
system.  In the Clarksville, New York area they identified two "wells" which
were actually pipes placed within cave passages (see Cave Well on Fig. 1).
Both such wells, and their ultimate discharge point at the Mill Pond, had
historic diesel fuel contaminant problems.

    Finally, there is no substitute for hydrogeologic assessment and objective
field reconnaissance.  Some of the important features and items which should
be included in a karst inventory are the location of: 1) sinkholes; 2) caves;
3) springs; 4) sinking streams; 5) seasonally active overflow springs; 6)
limited to non-existent surface drainage; 7) conduit wells; and 8) evaluation
of existing chemical analyses of wells, springs, and streams.

    As karst features are added to the inventory, it is necessary to
comprehensively evaluate the importance of each and how they combine to define
the hydrology of an area.  Base maps are valuable for plotting these features.
It is also important to place known or suspected contaminant sources on this
base map.


Karst Basin Evaluation

    Evaluating the hydrology and geology in any karst basin requires a multi-
phased approach.  Prioritization of these phases is dependent on: (1) the
urgency of the particular land-use plan (2) the potential threat to aquifer
and surface flow systems (3) on the amount of information already known about
the karst terrane (4) time, and last but not least, (5) money.  Evaluation of
a karst groundwater basin should include all or many of the components
discussed in this paper, depending upon the scope of the anticipated project,
and its budget.  Often, different aspects of the evaluation may be conducted
concurrently.  It is important that evaluations of this nature not only define
the dynamics of a karst system, but that the information then be utilized to
protect the subsurface and surface water resources.

    The entire groundwater basin needs to be characterized, not only the small
segment in question.  Land-use planners who do not take this broad view tend
to have serious unforeseen problems.  Small land-use projects with limited or
no waste streams may not require complete definition of all these elements.
However, this is generally only the case where a large base flow is maintained
year round and/or a significant soil-mantle is present.


Watershed Delineation

    In order to assess the potential impacts of waste streams leaving an area
of proposed development, it is first necessary to define the surface and
subsurface watershed boundaries.  This assessment includes at least a partial
definition of surface and subsurface flow components.  Although the ultimate
definition of a groundwater basin will be portrayed as a single irregular
watershed boundary, it can be useful in assessing vulnerable aquifer segments
and prudent land-uses to determine this boundary via a three phased approach;
the topographic basin, the lithologic basin, and the hydrologic or groundwater
basin.   The term "lithologic basin" is introduced here for its utility in
locating particularly vulnerable portions of karst basins, where carbonates
are bounded by less soluble and possibly deformed geologic units, and for its
function in isolating directions to search for potential aquifer exit
                                      775

-------
pathways.  These boundaries may be defined through the use of published
geologic maps, low-altitude stereo aerial photography, USGS topographic maps,
tracer studies and, where necessary, by detailed geologic mapping.


    Topographic Basin

    A preliminary assessment of an area's surface water basin may easily be
made with topographic maps and supplemented with low-altitude stereo aerial
photography.  The standard USGS 7.5-minute maps at a scale of 1 inch = 2,000
feet is practical for most problems.  Annotations may be made on the
topographic map while using a stereoscope to interpret aerial photography.  It
is always prudent to field check all areas, especially those which are
questionable.  The surface watershed boundaries determined from this method
can be useful as a first step in delineating the actual groundwater basin
boundaries.  Further refinement can be achieved through the evaluation of
other geologic and hydrologic factors; the lithologic and hydrologic basin.


    Lithologic Basin

    The geology of an area may make groundwater basin boundaries non-
coincident with surface water basin boundaries; meaning that subsurface flow
routes may extend far beyond, or even in another direction from that of a
surface stream.  The lower permeability of non-carbonate bedrock units
generally constrains groundwater flow, regardless of whether it results from a
stratigraphic or structural break (e.g., faulting, folding).  It is often
possible to expedite the watershed evaluation process by maximizing the use of
existing geologic and hydrologic maps of the area.  These maps can then be
used to predict structural or lithologic (relating to the physical character
of a rock unit) controls on watershed boundaries.  If no work has been
conducted, it may be necessary to do geologic mapping.  For example, detailed
geologic mapping outside the K-25 area at Oak Ridge, Tennessee has revealed
that relatively impermeable non-carbonate units (Rome Formation and Nolichucky
Shale) form impermeable boundaries to aquifer waters within the thick
carbonate sequence surrounding most of the K-25 complex area (Fig. 2).
Upturned Rome and Nolichucky beds force all subsurface water discharging from
the drainage basin to move along valley trend which is coincident with strike.
Interpretations like this are essential for predicting exit pathways for
contaminants and for limiting the area to be searched for spring resurgence
points (Palmer, 1986).

    The map in figure 2 illustrates the end product of such a geologic
assessment for the Oak Ridge, Tennessee area.  The original geologic map was
subdivided into 33 stratigraphic units.  Based on lithology, carbonate bed
thickness, and relative bedrock solubility, geologic formations were divided
into 2 hydrologic units; carbonate and non-carbonate.  Thus, the potential for
major karst development within a topographic-based watershed boundary was
further subdivided.  The cross section in figure 2 shows how bedding or faults
may further constrain karst development and contaminant exit pathways.  This
map delineates several parallel and distinct karst flow systems.  Analysis of
geologic logs in a portion of one of these carbonate bands revealed that 34
percent of coreholes encountered voids, providing evidence for mature
karstification.

    In another example, detailed geologic mapping in the Clarksville, New York
area revealed that the Esopus Shale had been thrust up against the Onondaga
Limestone  (Fig. 3).  This uplifted wall of impermeable shale (sub-parallel to
the east border of the map) forces all subsurface water draining southeast in
drainage basin B of Figure 1 to flow south proximal to this fault zone (Rubin,
1991a), resurging at the Mill Pond spring.  Without this knowledge, residents
east of the fault incorrectly believed that septic contamination of their well
water was from liquid farm waste spreading which frequently enters the cave
system.
                                      776

-------
             Eblen Cave
           N34*W
           A > True North

           V
Miles
     AQUITARD
 AQUIFER
                                                                           Index Map
                   Non-Carbonate Formations


                   CarbonateAquifer                            Odk Ridge


                   Stratigraphic Contact


                   Fault


                   Known caves


                   Arrows indicate possible strike-controlled subsurface flow
                   direction. Some or all paleo and recent flow in large solution
                   conduits initially graded to paleo-Clinch River baselevels.
Figure 2: Cave development and possible contaminant exit pathways in structurally deformed regions are
         often controlled by bedrock lithology and former or recent baselevels. One aspect of
         characterizing subsurface flow in carbonate aquifers requires locating and monitoring springs.
         Spring locations may be constrained by strike-controlled lithologic boundaries and river
         baselevels.  In Oak Ridge, TN, discharge points may be located adjacent to the Clinch River
         towards the southwest and/or northeast.  Buried or relict karst, which may still function as
         dominant subsurface drainage routes, can be defined through geophysical, boring, and tracer
         studies.  (PAR / PJL 10/28/91 ORNL Drawing)
                                                   777

-------
                                        STOVE PIPE PALEOGORGE
                                     "<  "-    > J/
                                     .^-  ^ '-;: V
                                                     CLARKSVILLE CAVE */" *
                                                    LOW FLOW CONDUIT
               HYPOTHETICAL LOCATION
               OF CAVE PASSAGES AS
               INFERRED FROM DYE TRACES

      mmmitfm:  PALEOGORGE

      /
-------
    Hydrologic (Groundwater) Basin

    The ultimate definition of a karst groundwater basin requires specific
knowledge of the physical boundaries of all water entering and leaving the
basin.  Definition of topographic and lithologic basins aids in defining the
boundaries of groundwater basins, however, topographic divides rarely coincide
with subsurface divides in karst terranes (Quinlan, 1989).  Sometimes, during
and after floods, some groundwater may be diverted from one groundwater basin
to one or more others adjacent to it (Quinlan, 1990, Figs. 2 and 3).  Regions
with complex geomorphic histories, such as Oak Ridge, Tennessee (Fig. 2), may
require advanced testing and analysis to define karst systems grading to
multiple baselevels.  Structural and lithologic controls, however, will still
impose certain limits on the extent and boundaries of karst basins, as may
hydrologic baselevels.

    All karst aquifers have at least one discharge point.  Discharge points
(resurgences) generally occur as springs at a local or regional baselevel
feature such as a stream, river, lake, or ocean.  Spring locations may be
predicted by evaluating the combined topographic and geologic map.  Palmer
(1986), documents how bedding orientation may be used to predict likely
directions of contaminant transport and groundwater flow.  In many settings,
it is essential to document the actual subsurface flow direction, destination,
and velocity with one or more tracer tests.   Where little surface flow is
present, it is often possible to force flush a tracer through the karst
system.  A pre-injection systematic search of potential baselevel outlet areas
may serve to locate springs.  Local farmers, homeowners, hunters, and
fishermen are often excellent sources of information, but there is no
substitute for field work to search for them.

    The type and thickness of soil cover in an area will influence the amount
and extent of infiltration (downward percolating water) reaching an underlying
karst aquifer.  A detailed examination of all surface streams and their
tributaries may reveal that little to no flow is pirated (diverted) to the
subsurface.  In many karst areas, however, soils perch surface streams for
short distances prior to their partial or complete loss to the subsurface.
Infiltration areas may be detected by conducting seepage velocity (discharge)
measurements at successive points along streams where a losing component is
suspected.  One common situation found in karst terranes is that of surface
flow over impermeable soil or bedrock formations present in the higher reaches
of a karst watershed.  This water flows downhill until it sinks near an
exposed limestone contact, where numerous sinkholes are often aligned.

    When considering acceptable land-uses and their resultant contaminant
loading, it is important, when possible, to evaluate the ability of the
bedrock or soil present to treat infiltrating septic or other waste.


                  Practical Land Use Planning:  Two Examples

    An excellent example of a karst area requiring specialized land-use
planning in order to avoid both significant aquifer and surface stream
degradation is a broad karst aquifer present in the Clarksville area  (Fig. 1),
specifically the Mill Pond karst basin which forms the headwaters of the
Onesquethaw Creek.  Since urban expansion was extending into the surface water
basin, a study was undertaken to define potential development concerns.  No
funding was available, so the steps followed had to be capable of reliably
characterizing potential contaminant concerns with limited manpower and
resources.  This type of limited approach may be all that many towns will ever
have the resources to have performed.   Proper characterization of a karst
groundwater basin will require a qualified individual or firm, as discussed  in
the closing "Who to look to for help"  section of this paper.

    The key issue for planners in karst areas should be the water quality of
the aquifer and its receiving springs and streams  (e.g., including water
                                      779

-------
supply, ecosystem, and aesthetic concerns).  In order to objectively appraise
land-uses and their potential impacts in karst basins, it is necessary to
define the hydrologic boundaries (including all discharge points), ag_uifer and
receiving stream vulnerability, and many of the items discussed in the
assimilative capacity section of this article.  To evaluate potential impacts
in the Mill Pond karst basin, a stepwise strategy was followed, with various
steps occurring concurrently.  These are numbered for ease of reference.

    1) Topographic maps were obtained of the area.  A first approximation
       of the surface water basin was defined to determine approximate
       boundaries of the area of concern.

    2) Existing published and unpublished geologic and hydrologic reports,
       chemical analyses, cave maps, and stream flow data were collected to
       provide a current basis for evaluating existing conditions.

    3) Maps of the bedrock geology and soil thickness were obtained from
       existing sources.  The area of carbonate bedrock was overlain on the
       surface water basin obtained in step 1), above.  It was determined
       that the geology was such that no subsurface flow in carbonates was
       likely to enter from outside the topographic basin.  This possibility
       must be evaluated on a case by case basis.

    4) All known karst features were inventoried.  These included features
       readily apparent in the basin (e.g., sinking streams, sinkholes,
       caves, springs) and others learned of from local residents (e.g.,
       conduit wells, caves).  Past and present contaminant problems were
       also noted as part of the inventory.  Most state regulatory programs
       maintain a file of regulated environmental problem sites.

    5) Existing geologic and hydrologic reports and maps were evaluated for
       likely aquifer discharge points.  Baselevel discharge areas were
       searched for springs.

    6) The location of significant karst features were field-checked and
       placed on the working base map.

    7) Caves were mapped.  (This step may not always be necessary.)

    8) A major fault was identified in a cave.  This discovery suggested
       that a faulted shale unit may be forming a groundwater basin
       boundary to the east.  The fault was not present on published
       geologic maps, but was found and readily mapped in the field (Fig. 3).

    9) Existing stereo aerial photographs were obtained and utilized to
       further refine the geologic contacts in the basin.  Several
       additional karst features recognized on the aerial photographs were
       accurately plotted on the base map.

   10) Known or suspected contaminant sources were plotted on the base map
       (e.g., junk yard).

   11) Areas of dense housing (likely contaminant sources) were also
       plotted.

   12) Several qualitative tracer tests were conducted to confirm
       suspected flow paths and roughly approximate subsurface travel
       times.  It was determined that the gentle southwesterly dip of the
       bedrock present in the Mill Pond aquifer fails to direct
       all subsurface flow in this direction.  Instead, the significantly
       higher surface topography to the southwest (Wolf Hill and Cass Hill)
       retards dissolution in this direction, in favor of the 1.3° apparent
       dip between Wolf Hill Dam and a baselevel discharge point at a
       spring in the Mill Pond.  Tracer studies generally verified
                                      780

-------
       predicted flow paths, at least during periods of low discharge when
       the entire headwaters of the Onesquethaw Creek resurge from the Mill
       Pond.  However, tracer studies also documented the unexpected
       easterly diversion of moderate to high discharge waters through
       Pauley Avenue in Clarksville Cave (Fig. 3).

   13) A gaging station was established downstream of all karst springs
       and at a point beyond which no subsurface flow from carbonates was
       possible.  This was monitored daily for 15 months, and periodically
       thereafter during periods of low and high flow.  Monitoring for this
       duration may be unnecessary, since it is the lowest flows that were
       of the greatest concern for estimating the likely assimilative
       capacity for contaminants.  Base flow from the basin was gaged at
       less than 0.1 cubic feet per second (50 gallons per minute).

   14) The karst system was evaluated based on the results of the
       investigations conducted and analysis of stream hydrographs.
       Additional information could also be obtained on the nature of the
       system using relatively simple physical properties (e.g., specific
       conductance, temperature, and turbidity) of spring waters.  Quinlan
       et al. (1991) have put forth a simple, practical, and reliable
       evaluation technique for classifying the vulnerability of carbonate
       aquifers based on the coefficient of variation of specific conductance
       of spring waters.  The technique evaluates the net effect of an
       aquifer's recharge, storage and subsurface flowpath characteristics.  A
       fourth category, which might be evaluated as part of the vulnerability
       classification, is aquifer base flow.  In the nomenclature of Quinlan
       et al. (1991), the Mill Pond karst aquifer may be classified as
       hypersensitive.

   15) The environmental sensitivity of the karst basin was ascertained,
       placing particular emphasis on the vulnerability and likely
       assimilative capacity of the groundwater basin.  As part of this
       process, consideration was given to what adverse effects might
       result from contaminant types which would be permissible within the
       broad Rural Commercial District guidelines.

   16) Findings of the investigation and concerns regarding contaminant
       loading were shared with the town planning board (designated lead
       agency), town engineers, local and state agencies, local
       environmental groups, and the press via reports, letters, and
       telephone discussions.  The findings were also presented at a
       public hearing.

   17) A final step in the process is to work with the agencies
       involved to draft a management plan adopting land-uses within the
       geologic and hydrologic constraints of the karst basin.

    The Mill Pond watershed was subdivided into two parts: A) the 3,075 acre
portion located upstream of Wolf Hill Dam, and B) the 2,050 acre portion
located downstream of Wolf Hill Dam.  The boundaries of the aquifer were
determined to extend to the north and northwest of the basin's resurgence
point at the Mill Pond (Fig. 3).  The farthest boundary of the Mill Pond
groundwater basin lies some 2.4 miles to the northwest, proximal to the Wolf
Hill Dam on the Onesguethaw Creek.  The boundaries of the catchment basin are
depicted in bold dashed lines.  The upstream part is comprised of insoluble
rock, with much or all discharge being seasonally diverted to the Vly Creek
Reservoir.  The downstream part of the Mill Pond watershed has features
characteristic of karst terranes.  These include sinking streams, limited
surface drainage, solutionally enlarged joints, sinkholes, and the Clarksville
Cave system.  During most of the year, water entering that portion of the
watershed that is downstream of the Wolf Hill Dam is pirated into the Onondaga
Limestone.  Tracer tests and in-cave stream gaging indicate that flow is
turbulent and has a maximum straight-line flow velocity of 5.3 km/hr  (3.3
                                     781

-------
mph).   During periods of high discharge, contaminants have the ability to
travel from one end of the aquifer to the other in less than one hour.

    A detailed analysis of the flow dynamics present in the Mill Pond karst
basin was conducted.  A gaging station was established in Onesguethaw Creek
(Fig.  1) in order to examine the relationship between in-cave discharge and
surface-watershed discharge.  This was monitored twice daily for 15 months,
more frequently during flood events, and periodically for 4 years thereafter
during major runoff events (Fig. 4).  Stream discharge was measured at 13
different stages.  Curvilinear regression was then utilized to establish a
series of multi-order equations that could be used to correlate stage height
with discharge.  The greatest discharge recorded for Onesquethaw Creek during
the course of this study was approximately 1340 cfs (600,000 gpm).  This
occurred on March 15, 1986 at 3:00 am following heavy rains (=2.7 in.) on a
15-inch snow pack.  Daily monitoring of stream stage in Clarksville Cave for
the same 15-month period revealed that a direct correlation exists between
this discharge and that in Onesquethaw Creek.  Approximately 8 percent of
flood-peak discharge in Onesquethaw Creek flows through Clarksville Cave.

    In this example, it is clear that major land-use in the central portion of
the basin (e.g., Rural Commercial District), where almost no soil-mantle is
present above heavily jointed limestone pavement, may have a significant
environmental impact on the basin's receiving stream.   Land-uses here which
directly and/or continuously discharge contaminants to the conduit system
should be carefully evaluated.  In this setting, contaminants will quickly
enter the fractured limestone and shallow underlying cave passage, where they
will be directly piped to the Mill Pond spring and the Onesquethaw Creek.
Continuous waste streams are more likely to adversely impact this geologic
setting than those which enter the aquifer during times of saturated soil
moisture conditions, high runoff, and infiltration.  However,  it should be
noted that one-time or infrequent chemical slugs can,  in some karst aquifers,
cause greater problems than lower controlled or permitted releases.

    The findings of this investigation were not used by the Town to develop a
new overall master plan.  Instead, the Town ultimately decided to continue
issuing building permits on a case by case basis, presumably now taking into
account the known environmental sensitivity of the aquifer.  All the desired
land-uses cited in the Typical Development Proposal section of this paper
remain intact.  Towns in situations such as this may ultimately have to answer
legally and politically to their constituents.  The importance and need of a
comprehensive karst educational program at local and state levels is apparent
and cannot be overemphasized.

    Another example of land-use planning on a large raised karst watershed is
also situated in upstate New York.  A number of private homes were proposed
for development upslope of a karst spring used by a village for water supply.
The bedrock geology was well documented, with most formations within the
watershed including caves.  A relatively thin soil-mantle was present, along
with several areas of dissolutionally enlarged joints and sinkholes.  After
many meetings with health officials and politicians, a preliminary decision
was made to "protect" the aquifer by imposing a several hundred foot set-back
distance from all sinkholes.  The set back zone concept incorrectly
presupposes that sinkholes are independent from the area-wide karst system
(Quinlan et al., 1991).  Unfortunately, this decision did not take into
account the important fact that any infiltration of septage through the
surrounding permeable soil-mantle would probably quickly enter the underlying
karst system.  Although special considerations may be necessary, development
in areas such as this is still possible.  The question is really what is
reasonable and prudent and within the assimilative capacity of the aquifer.
                                      782

-------
                            0.40
F£
co a


|a!
LU£
m*
                            0.80
                            1.20
                            1.60
                                                                          STEADY DECLINE
                                                                          IN CREEK STAGE
                            2.00  '  I  i  I  i  I  i  I  '  I  M  i  I  i  I  i  I  i  I  i  I  i  I  i  I i  I  i  I  i  I  '  I  i  I t  I  '  I  i

                                  11/1811/2211/2611/3012/4  12/8 12/1212/1612/2012/2412/28 1/1   1/5  1/9  1/13  1/17 1/21 1/25 1/29  2/2

                                  I                     1985                      I                                1986
                                                                                                          i I  i  I  i  I  i  I  i  I  i  I  i  I

                                                                                                           2/10 2/14 2/18  2/22 2/26  3/2  3/6
89

46
32
                                                                                                                                                               O
                                                                                                                                                LU

                                                                                                                                          24.5  g
                                                                                                                                          9.6   <
                                                                                                                                          5.3   a
                                                                                                                                          2.5   co
                                                                                                                                                a
oo
CO
   14

•w 12


I10
*  8
O

£  6

5
                co
                        Figure 4: Onesquethaw Creek hydrograph. Stream discharge was measured at 13 different stages. Curvilinear regression was then utilized to
                                 establish a series of multi-order equations that were used to correlate stage height with discharge.
                                                                           4.43 i
                                                                                                                                                  140
                                                                                                   ^+^
                                                                                                    SHADED AREA
                                                                                                    EQUALS
                                                                                                    TEMPERATURE
                                                                                                   , ABOVE 32'F
                                   10  14 18  22  26  30   3
                                                             7   11  15  19 23  27  2   6  10  14  18  22  26 30  3

                                                                           JANUARY 1984-MAY 1984
                                                                                                                    7  11  15  19  23  27
                                                                                                                                        4200

                                                                                                                                        3600 _

                                                                                                                                        3000  g.

                                                                                                                                        2400 LU
                                                                                                                                              a
                                                                                                                                        1800 g

                                                                                                                                        1200 i
                                                                                                                                              O
                                                                                                                                        600

                                                                                                                                        0
                        Figure 6: Operable dimatic conditions influencing salt contaminant transport from a roadside near Saugerties, New York. Salt infiltrates soil, passes
                                 through fractured limestone (against dip), and contaminates wells (A °. and n) and a baselevel spring.

-------
            Advanced Testing And Analysis To Define A Karst System

    A number of additional steps may be incorporated into a program to define
a given karst network.  Some may not be necessary if existing or planned land-
uses pose little or no threat to an aquifer or its receiving stream.  These
components may include cave surveying, chemical and hydraulic monitoring, well
drilling, geophysics, and tracer tests. Some of these methodologies are
discussed below.


Physical Entry Of Karst Networks

    The survey and characterization of caves in karst aquifers is valuable
because the actual location and extent of segments of the karst system can be
examined.  There is no substitute for entry and survey of enterable segments
of the karst system.  Visual reconnaissance of the inner workings of the karst
aquifer can provide information of primary transport pathways and can serve to
locate monitoring stations.  Cave explorers and cave-divers represent one of
the best resources when conducting land-use studies in karst terranes.
Trained cave surveyors are almost always willing to help towns and federal
agencies with this integral aspect of defining karst systems, generally at no
charge.  Potential access points include cave entrances and springs.
Sometimes access can be gained by excavating sinkholes or overflow springs.
Large diameter boreholes are good entry points where the conduits are
seasonally above the saturated zone.  More recently, trained cave divers have
penetrated spring resurgences and brought back valuable survey and geologic
assessment.  Sometimes it may be worthwhile to utilize heavy equipment to dig
or drill into blocked portions of a cave system.


Cave Detection With Geophysical Techniques

    Geophysical techniques can sometimes be used for identifying conduit
locations and connectivity.  Micro-gravity and electrical resistivity are two
more successful techniques, but no technique is capable of resolving caves of
small size at great depth.  The work of Kilty and Lange (1991) and Lange and
Kilty (1991), however, appears encouraging for locating streams at depth.
Geophysical investigation can be followed with coreholes,  dye traces, and
continuous hydraulic and chemical monitoring of wells and springs.


Water Quality, Hydraulic And Biologic Monitoring In Karst Aquifers

    Great emphasis should be placed on learning more about the hydraulics and
contaminant distribution, if any, in accessible portions of a karst system.
Because a representative chemical characterization of the bedrock flow system
can only be obtained from mature conduit segments, storm water should be
sampled for contaminant transport after either, 1) karst springs are located,
or 2) hydrographs have been obtained from conduit wells, indicating that they
are part of the active conduit system.


    Water Quality Monitoring Of A Karst System

    Select water quality indicators, as well as suspected or planned
contaminants should be monitored during periods of low flow and most
importantly, during short-term flood events.  Monthly or quarterly monitoring
is typically conducted in porous and fractured bedrock aquifers; but, the
highly transmissive nature and flashy response of some mature karst aquifers
requires a more frequent sampling schedule.  An infrequent sampling schedule
in karst aquifers having rapid flow may miss the short periods of time when
chemicals are actively being transported (Quinlan, 1989, 1990).  However,
different types of contaminant releases often require different sampling
schedules.  Sampling schedules must be tailored to the geologic and physical
                                      784

-------
setting of contaminant sources, taking into consideration the hydrologic
conditions most likely to result in contaminant dispersal, infiltration, and
movement.  If a contaminant source is an unregulated spill into the soil zone
above a karst aquifer, storm flow monitoring may detect it; however, if the
contaminant source is a continuous discharge from a point source  (e.g., from a
package plant) it may well not be detected during flood flow conditions.
Traditional monitoring frequencies may thus fail to indicate the true
distribution of a contaminant problem in a karstified aquifer.  Sampling of
water chemistry in monitoring wells placed in the diffuse (non-conduit) flow
zone of a karst aquifer will typically yield unreliable data, not
representative of the system as a whole (Quinlan, 1989, 1990; Ewers, 1991).

    Figure 5 illustrates the water level difference in a physical setting of
wells placed only a few meters apart, where substantially different hydraulic
conductivity and transmissivity values and chemical concentrations may be
obtained.  The term "tertiary porosity" has been introduced by Teutsch and
Sauter (1991) to refer to dissolutional porosity, rather than secondary
fracture porosity.  The failure to find contaminants in wells not within the
active conduit portions of the karst system may falsely suggest that no
contaminant problem is present.  It is reasonable to focus contaminant
characterization and exit pathway studies on the mature zones of well-
karstified aquifers because significant remediation of contaminants found in
diffuse-flow zones of karst aquifers is unlikely to impossible (Quinlan and
Ray, 1991).

    Monitoring must be conducted in locations that intersect the active
conduits of a flow system.  While wells demonstrated to be in conduits may
provide reliable water chemistry data for their sub-watershed areas, the best
and most representative sampling locations for a system as a whole are at
springs  (Quinlan and Ewers, 1985).

    An example of the relationship between soluble contaminant movement and
assorted climatic conditions in a shallow highly fractured limestone is
illustrated in Figure 6.  In this example, road salt infiltrates soil, passes
through the bedrock (against dip), and contaminates wells and a baselevel
spring.  Periods of snowmelt, high rainfall, and high temperatures can
significantly increase contaminant transport rates.  Lack of these conditions
can result in low level contamination.  Frequent measurements of conductivity
would serve as an effective screening tool to select optimal sampling times.
Again, the frequency of groundwater sampling in karst terranes must mimic the
flow dynamics and the contaminant source type.
                                                         NW
                                                      Soil
  Figure 5: Schematic cross section showing hypothetical low K  (hydraulic
            conductivity) and T  (transmissivity) diffuse-flow zone  secondary
            porosity (SE well) and high K and T conduit tertiary porosity
            (NW well) within carbonate aquifer.
                                     785

-------
    Continuous Chemical Monitoring Of Solution Conduit Waters

    Quinlan and Ewers (1985), Quinlan (1989), and Quinlan et al.  (1991)  have
documented practical strategies for monitoring karst aquifers  and  contaminant
transport within them.  Basic to these strategies is the understanding that
conduit flow may be turbulent and is often quite rapid.  An entire contaminant
pulse might pass by a monitoring site in a matter of hours, rather than  months
(see Fig. 7).  The rapid response of the conduit system versus that of the
diffuse system is a unique feature of some karst aquifers.  A  sudden influx of
water within a conduit system may result in water level rises  of tens of feet
in a few hours.  These sudden rises signal the filling of a groundwater
trough, which may bring flood levels up to or above the water  table observed
in the diffuse flow portion of the karst aquifer.  Special efforts are
required in karst aquifers in order to detect and characterize episodic  or
pulse-driven contaminant movement.

    Monitoring the relevant water chemistry, water levels, and flow components
of a karst system is virtually impossible unless wells have deliberately or
accidently intersected active conduits or, as is usually the case,  springs
have been found by field reconnaissance efforts and shown by tracing to  be
draining from the source pollutants, and thus to be relevant.

    Once relevant springs and conduit wells have been identified,
representative sampling of the aquifer can be conducted at times of suspected
contaminant transport (i.e., flood pulses).  First round sampling  might  be
conducted for major ions, nitrate, ammonia, phosphates, fecal  coliform,
specific conductance, dissolved oxygen,  biochemical oxygen demand,  turbidity,
pH, temperature, and known or suspected contaminants.
                 =5  790 —
                 I  770
                 55
                    750 -
                    730
.Storm-flow response
                                              Base-flow conditions
                         0
  I
 2
\
3
                                         Days
    Figure 7: Hypothetical hydrograph of flashy response of a conduit well in
              a mature karst system.


    Hydraulic Monitoring Of Wells Intersecting Solution Cavities

    Karst groundwater basins typically contain both diffuse and conduit  flow
components.  Diffuse zone in-feeders, tributary to highly transmissive
conduits, would typically exhibit slower water movement until fracture
aperture was sufficient to permit groundwater flow to cross from  laminar to
turbulent flow conditions.  Even poorly integrated fractures in the  karst
system still ultimately drain toward a groundwater trough where transmissivity
is relatively higher.
                                      786

-------
    Transducers and data loggers can provide the continuous recording
necessary to properly define the hydraulics of wells known to be in the
conduit system.  Conduit flood pulses may be short-lived (i.e., hours or
days), depending on the maturity of the conduit system and the quantity of
recharge incident upon the system (Fig. 7).  Thus, in order to understand the
hydraulics and characterize the water chemistry of the system, it is sometimes
necessary to monitor on an hourly rather than monthly or quarterly sampling
schedule (Quinlan and Alexander, 1987).  In a karst network, long intervals
between sample collection are likely to miss storm events and critical
contaminant transport entirely.  Frequent monitoring and sampling during flood
pulses is necessary.

    Wells which have large fluctuations of their piezometric surface,
especially when followed by a short recession curve, may be interpreted as
being part of an active and mature karst flow system (Fig.  7).  Water flowing
through these wells, and their associated dissolution conduits ultimately must
discharge to some baselevel.


    Karst Groundwater Basin Flow Components

    Continuous monitoring and quantification of discharge entering and
leaving a karst basin is important for contaminant characterization and some
land-use studies.  For larger projects, where contamination is or may be a
problem, this may include stream, spring, and conduit well discharge
monitoring via weirs and stage recorders (or other devices).  Seepage velocity
(discharge) measurements on tributaries suspected of being losing streams
should also be considered.  One or more weather stations are critical for
realistic water budget appraisal.  For small projects,  it may be sufficient to
monitor only the base flow discharge leaving the karst system during drought
or low-return periods.

    Knowledge of expected flood-return intervals and their magnitude can be
important in assessing contaminant transport potential.  One example of where
flood-return information might be applied is in predicting mobilization
probabilities of contaminants poorly bound to soils; which may occur as a
result of soil-flushing stemming from large precipitation events.  In other
settings, large point source quantities of hazardous waste may be released
from elevated flow system segments only during extreme flood conditions.  This
technique, which might be applied towards roughly assessing a maximum rate of
contaminant movement in a karst system, may be applicable in karst areas where
all or most of the drainage exits via springs or where sufficient in-conduit
hydrograph data is available.  Where time is an element for a particular land-
use option being contemplated, and reasonable but limited stream discharge
data is available, it is possible to perform a flood probability analysis.
Similarly, knowledge of base flow and drought-return intervals, when the
assimilative capacity of an aquifer is at its lowest, can be critical to
prudent land-use determinations.

   An example of such an analysis was conducted for the Mill Pond karst basin
in east-central New York (Rubin, 1991a, 1991b) where much of the basin's flow
is subsurface for most of the year.   The limited data available for
statistical comparison among hydrologic years necessitated examination of
another roughly comparable basin in order to assess flood-return intervals.
The farthest headwater gaging station on Schoharie Creek at Prattsville was
selected.  Many inherent differences occur between the basins, notably
elevation, geology, soil thickness,  size, and location.  The Prattsville and
Onesquethaw Creek gaging stations are approximately 48 kilometers apart.
However, the Prattsville and Mill Pond watersheds are comparable under
conditions of a saturated soil-moisture bank, high runoff,  and similar storm
systems.  Eighty-two years of data at the Prattsville station were examined.
                                     787

-------
                 100
                                             I  I I I  I
                                                             III I I II
                                               • 1903-1987 LOG PEARSON
                                               O 1903-1951 LOG PEARSON
                                               A 1903-1987 GUMBEL
                                               A 1903-1951 GUMBEL

                                                 1925 AND 1929 WATER
                                                 YEARS NOT AVAILABLE
                                                 I  I I I I
                                                          I   i  i  i i
                                    10              102
                                 FLOOD RETURN INTERVAL (Tr in yr)
          103
    Figure 8: Schoharie Creek at Prattsville, New York. Log-Pearson type III
              and Gumbel distributions utilizing historic water year peak  flow
              data. Flood return interval for largest Onesquethaw Creek peak
              discharge of record correlated to Prattsville March 15, 1986
              flood of 54,900 cfs. Range of two methods shows a Tr of 30 to 47
              years. Similar comparisons may be made for low flows in
              geologically comparable basins. Analyses of this type can be
              useful tools in predicting flow velocities, -—••-—«•!« = «<- = *--•:<
              times, and contaminant dilutions.
contaminant arrival
    A Log-Pearson Type III and Gumbel-distribution statistical  comparison
(Linsley et al., 1975) of historic peak flow of Schoharie Creek gaging  data
with this study's hydrograph information for Onesquethaw Creek  indicates that
the largest Onesquethaw Creek peak of record (March 15, 1986) has  a  return
interval on the order of 30 to 47 years (Fig. 8).  This corresponds  to  a
Prattsville hydologic-year peak discharge of 54,900 cfs.  Thus,  if 40 years
was the expected flood return interval, 25 floods of this magnitude  could be
expected every 1000 years.  These infrequent storm or runoff  events  reasonably
represent a near-maximum quantity of water available in the watershed under
ideal, thin-soil-mantled, rapid infiltration conditions.  Knowledge  of  peak  or
base flow return periods can sometimes be correlated with water chemistry
results to help assess chemical loading both in the karst system and to stream
receptors.

    Biologic Monitoring Of The Karst System

    Fauna, both within caves and at springs, provide the best indicator of
health of the karst system.  Various organisms, such as benthic invertebrates,
benthic macroinvertebrates, fish, amphipods, isopods, flatworms, and copepods
can provide excellent indices of biological  integrity.  These organisms may  be
subjected to episodic contaminant pulses and any background  contaminant
chemistry.  Their health, or sometimes absence, as monitored at one  or  more
points in time, is a true measure of the health of the groundwater system
(Poulson, 1991; Preddice, T., NYS Dept. of Environmental Conservation,  pers.
comm.).  Acute exposure to various chemicals can influence the ability of some
                                       788

-------
species to reproduce or survive.  This can have far-reaching effects on the
food chain.  The assimilative capacity of karst waters and its biota for
contaminants is dependent upon factors such as chemical type and toxicity,
chemical injection or leakage rates, discharge in the system (e.g., low flow),
and reaeration rates along the flow path.  Similarly, plant species can also
aid in the health-based assessment of a receptor stream.


Tracer Studies

    Tracer studies are the best technique for determining flow direction,
destination, and velocity within a karst groundwater basin (Quinlan, et al.,
1991).  State-of-the-art tracing techniques are capable of simultaneously
running multiple tracers to assess different suspected flow routes, or to
verify low quantitative results from previous traces.  Tracer tests may be
successfully used to: 1) determine discharge locations of sinking streams
which may flow within shallow or deep karst systems; 2) determine aquifer
discharge locations necessary for characterizing contaminant exit pathways; 3)
determine groundwater seepage velocity rates and discharge in conduit portions
of the karst flow systems.  (This is important for water budget and model
assessment, as well as preventative planning for episodic contaminant or spill
releases); 4) define drainage divides and catchment basin boundaries; 5)
define subsurface flow and transport routes within conduit and diffuse zones;
6) assess stream infiltration/piracy to shallow surficial discharge points;
and 7) realistically characterize groundwater seepage velocity rates in the
diffuse zone between boreholes.


Wells

    Wells which are part of the active conduit system, as determined by
transducer/hydrograph response, or their locations in known cave passages, may
provide the best monitoring locations when springs cannot be located (e.g., as
a result of flooding).  Wells have value for 1) lithologic characterization,
2) structural characterization, 3) piezometric monitoring, 4) chemical
monitoring, and 5) tracer studies.

    Areas where an unusually high percentage of boreholes have encountered
voids (e.g., Oak Ridge, Tennessee) are indicative of maturely karstified
terranes.  Such settings are uncommon.  Quinlan and Ewers (1985) and Ewers
(1991) have estimated the odds of encountering a dissolutional conduit in a
karst aquifer by drilling a well at about 1:2600.

            Political Realities Of The Land Use Permitting Process

    Responsible land-use planning requires a willingness of all parties
involved in the planning and permitting process to look beyond the physical
boundaries of a particular building proposal which may be in front of them.
This may ultimately mean that the scope and scale of development needs to be
reduced.  This, in turn, may make landowners unhappy that they can no longer
subdivide or use their land as they might wish.  Unhappy landowners do not
reelect their politicians, who are thus hesitant to spend dollars on
consultants.  Some consultants, who desire continued work with towns, may wish
to insure their client's approval.  This process may be further complicated by
town planning board members who are themselves, or have relatives and friends
who are, developers.  A further complication in some states may result when
lead agency/permitting status is granted to towns which may have vested
interests.  In instances such as this, a cooperative agreement for technical
exchange between knowledgeable state and government people and planning boards
might be beneficial to all parties involved.  Thus, it should be recognized
that prudent land-use planning may, in some instances, be fraught with
political conflict and pressures.  All to often, the local residents and the
environment loses.
                                     789

-------
    Sound land-use planning is best conducted in the absence of specific
building proposals.  It is important to recognize impending problems before
building permits are on the table.  Ideally, land-use planners in karst
terranes will be able to conceptualize the subsurface flow systems present in
their basins and incorporate geologic and hydrologic information into a zoning
master plan.  Such master plans require 1) planners with some scientific
background, 2) sufficient capital and expertise to characterize a karst basin,
and 3) an educational process, such as that now being undertaken by the
National Speleological Society (NSS) and the American Cave Conservation
Association (ACCA) in the United States.  Public hearings and board meetings,
which provide a forum for the educational process, can be used to good
advantage with receptive audiences.  Members of these organizations and karst
hydrologists are often willing to help in this process when requested.  More
emphasis also needs to be placed on educating state, federal and local
officials.

    Whereas development within karst basins has historically occurred on an
individual application basis, it may now be prudent to initiate a more broad-
based master planning process.  Planning should take into account the likely
contaminant loading into the karst system, and a reasonable measure of its
assimilative capacity.  In many respects, the strategy employed should mimic
the discharge requirements many states impose on industrial effluents to
surface waterways.  Instead of granting discharge permits with unlimited
contaminant ceilings, permits are only granted when individual contaminant
contributions do not exceed the assimilative capacity of collective discharges
from multiple facilities.  In this manner, a strategy is employed to insure
the continued health of the waterway.  In karst groundwater basins, a
discharge permit of this nature should be based on the assimilative capacity
of the receiving aquifer and stream at times of low flow.

                           Who To  Look  To  For Help

    Characterization of the hydrology of a karst basin is beyond the technical
expertise of most land-use planners, environmental consultants, or even
experts within State and Federal agencies.  The flow dynamics in karst
terranes are very different from those in porous or fractured media.
Significant scientific advances in karst hydrology have occurred in the last
two decades.  Unfortunately,  many of the technical papers have not been
published in mainstream journals or received widespread distribution.  This
poor dissemination of knowledge in karst hydrology has hindered the land-use
planning process.  More recently, this trend has been broken (Ford and Ewers,
1978; Quinlan and Ewers, 1985; White, 1988; Ford and Williams, 1989; Quinlan,
1989, 1990; Palmer, 1991; and Quinlan et al. 1991).

    Plainly stated, extremely few consulting firms,  universities, and
government agencies with experienced hydrogeologists, geologists and engineers
possess the specialized qualifications of karst hydrologists.  Finding an
experienced consultant knowledgeable about karst hydrology is difficult.
Specialists in karst hydrology may be contacted through two organizations:

   Karst Waters Institute                 National Ground Water Association
   c/o John Mylroie; President            6375 Riverside Drive
   Box 2194                               Dublin, Ohio 43017
   Mississippi State Univ., MS 39762      614-761-1711
   601-325-8774

    The wise consultant will use the unique talents and knowledge of state
cave surveys and cave explorers, who are often associated with local chapters
of the National Speleological Society (NSS).  Such individuals can be a
tremendous asset in the karst inventory evaluation process, as well as
valuable sources of cave maps depicting segments of karst aquifers.  The NSS
may be contacted at Cave Ave., Huntsville, Alabama 35810 (205-852-1300).  It
is important to verify that the firm or individual claiming expertise in karst
hydrology actually has it.
                                      790

-------
                               Ac knowl edgement. s

    This manuscript was written while the author was employed in the
Environmental Sciences Division at the Oak Ridge National Laboratory.  ORNL is
managed by Martin Marietta Energy Systems, Inc., under contract No. DE-AC05-
84OR21400 with the U.S. Department of Energy.  Publication No. 3856,
Environmental Sciences Division, ORNL.

    "The submitted manuscript has been authored by a contractor of the U.S.
Government under contract No. DE-AC05-84OR21400.  Accordingly, the U.S.
Government retains a nonexclusive, royalty-free license to publish or
reproduce the published form of this contribution, or allow others to do so,
for U.S. Government purposes."

    Thanks are extended to Peter Lemiszki for his assistance in the
preparation of Figure 2, Linda Armstrong, Marc Casslar, Dale Huff, James F.
Quinlan, and Geary Schindel whose constructive review greatly enhanced this
paper, and to Thorn Engel who ably assisted with stream gaging and geologic
survey work.


                                References Cited

Ewers, R.O., 1991, The response of landfill monitoring wells in limestone
     (karst) aquifers to point sources and non-point sources of contamination.
     Hydrogeology, Ecology, Monitoring, and Management of Ground Water in
     Karst Terranes Conference (3rd, Nashville, Tenn.), Proceedings. National
     Ground Water Association, Dublin, Ohio, [in this volume]
Ford, D.C. and Ewers, R.O., 1978, The development of limestone cave systems in
     the dimensions of length and depth. Canadian Journal of Earth Science, v.
     15, p. 1783-1798.
Ford, D.C. and Williams, P.W., 1989, Karst Geomorpholoqy and Hydrology.
     London, Unwin Hyman, 601 p.
Hammer, M.J., 1975, Water and Waste-Water Technology. New York, John Wiley &
     Sons, 482 p.
Kilty, K.T. and Lange, A.L., 1991, The electrochemistry of natural-potential
     processes in karst. Hydrogeology, Ecology, Monitoring, and Management of
     Ground Water in Karst Terranes Conference (3rd, Nashville, Tenn.),
     Proceedings. National Ground Water Association, Dublin, Ohio, [in this
     volume]
Lange, A.L. and Kilty, K.T., 1991, Natural potential responses of karst
     systems at the ground surface. Hydrogeology, Ecology, Monitoring, and
     Management of Ground Water in Karst Terranes Conference (3rd, Nashville,
     Tenn.), Proceedings. National Ground Water Association, Dublin, Ohio, [in
     this volume]
Linsley, R.K., Jr., Kohler, M.A., and Paulhus,  J.L.H., 1975, Hydrology for
     Engineers. 2nd Edition, New York, McGraw-Hill Book Company.
Palmer, A.N., 1986, Prediction of contaminant paths in karst aquifers. In
     Environmental Problems in Karst Terranes and their Solutions Conference,
     Proceedings. National Water Well Association, Dublin, Ohio, p. 32-51.
Palmer, A.N., 1991, Origin and morphology of limestone caves: Geological
     Society of America Bulletin, v. 103, p. 1-21.
Poulson, T.L., 1991, Assessing groundwater quality in caves using indices of
     biological integrity. Hydrogeology, Ecology, Monitoring, and Management
     of Ground Water in Karst Terranes Conference (3rd, Nashville, Tenn.),
     Proceedings. National Ground Water Association, Dublin, Ohio, [in this
     volume]
Quinlan, J.F., 1989, Ground-water monitoring in karst terranes: recommended
     protocol and implicit assumptions. U.S. Environmental Protection Agency,
     Environmental Monitoring Systems Laboratory, Las Vegas, Nev. EPA/600/X-
     89/050  88 p. [DRAFT; final version to be published in 1992]
                                     791

-------
Quinlan, J.F., 1990, Special problems of ground-water monitoring  in karst
     terranes. In Nielsen, D.M., and Johnson, A.I., eds. Ground Water and
     Vadose Zone Monitoring. ASTM Special Technical Publication 1053. American
     Society for Testing and Materials  (ASTM), Philadelphia, p. 275-304.
Quinlan, J.F. and Ewers, R.O., 1985, Ground water flow in limestone: Rationale
     for a reliable strategy for efficient monitoring of ground water quality
     in karst areas.  National Symposium on Aquifer Restoration and Ground
     Water Monitoring, 5th, Proceedings, p. 197-234.
Quinlan, J.F. and Alexander, E.G., Jr., 1987, How often should samples be
     taken at relevant locations for reliable monitoring of pollutants from an
     agricultural, waste disposal, or spill site. Multidisciplinary Conference
     on Sinkholes and the Environmental Impacts of Karst (2nd, Orlando, Fla.),
     Proceedings. Balkema, Rotterdam, The Netherlands, p. 277-286.
Quinlan, J.F. and Ray, J.A., 1991, Ground-water remediation may be achievable
     in some karst aquifers that are contaminated, but it ranges  from unlikely
     to impossible in most: I. Implications of long-term tracer tests for
     universal failure in goal attainment by scientists, consultants, and
     regulators. Hydrogeology, Ecology, Monitoring, and Management of Ground
     Water in Karst Terranes Conference (3rd, Nashville, Tenn.),  Proceedings.
     National Ground Water Association, Dublin, Ohio, [in this volume]
Quinlan, J.F., Smart, P.L., Schindel, G.M., Alexander, E.G., Jr., Edwards,
     A.J., and Smith, A.R., 1991, Recommended administrative/regulatory
     definition of karst aquifer, principles for classification of carbonate
     aquifers, practical evaluation of vulnerability of karst aquifers, and
     determination of optimum sampling frequency at springs.   Hydrogeology,
     Ecology, Monitoring, and Management of Ground Water in Karst Terranes
     Conference (3rd, Nashville, Tenn.), Proceedings. National Ground Water
     Association, Dublin, Ohio,  [in this volume]
Rubin, P.A., 1991a, Flow characteristics and scallop-forming hydraulics within
     the Mill Pond karst basin, East-Central New York. In Kastning, E.H. and
     Kastning, K.M., (eds), Appalachian Karst Symposium, National
     Speleological Society, Huntsville, Ala., p. 101-108.
Rubin, P.A., 1991b, Modification of preglacial caves by glacial meltwater
     invasion in East-Central New York. In Kastning, E.H. and Kastning, K.M.,
     (eds), Appalachian Karst Symposium, National Speleological Society,
     Huntsville, Ala., p. 91-100.
Teutsch, G. and Sauter, M., 1991, Groundwater modelling in karst  terranes:
     scale effects, data acquisition and field validation.  Hydrogeology,
     Ecology, Monitoring, and Management of Ground Water in Karst Terranes
     Conference (3rd, Nashville, Tenn.), Proceedings. National Ground Water
     Association, Dublin, Ohio,  [in this volume]
White, W.B., 1988, Geomorpholoqy and Hydrology of Carbonate Terrains. Oxford:
     Oxford University Press, 464 p.


                              Biographical  Sketch

    Paul Rubin is a Research Scientist in the Environmental Sciences Division
of the Oak Ridge National Laboratory in Tennessee where he is involved in
characterizing groundwater flow.  Mr. Rubin served as a hydrogeologist for the
Environmental Protection Bureau of the NYS Attorney General's Office for 8.5
years prior to his present position.  He received his Master's degree in
Geology from SUNY New Paltz.  He has taken an active role in research and
contaminant investigations, as well as land-use planning issues,  in karst
terranes.

     Paul A. Rubin
     Bldg. 1503, MS: 6352
     ORNL, P.O. Box 2008
     Oak Ridge, TN 37831-6352
     (615) 576-3476
                                      792

-------
         Land Use Planning And Watershed Protection In Karst Terranes

                                Paul A. Rubin

1.   You mentioned chlorinating septic tank outflow.  While this should take
    care of fecal coliforms,  I don't see any impact of chlorination on the
    problem of affecting nutrient level and nutrient type in cave
    ecosystems.  How would you deal with the elevated organic carbon and
    phosphate levels that are produced by septic tanks?

    Chlorinating septic waste, as is currently being conducted by Stewart's
    in the Mill Pond groundwater basin, will reduce viral and bacteriological
    problems, but not eliminate the problems associated with elevated
    nitrogen and phosphorus levels.  Thus, practical resolution of the
    problem of excessive nutrient loading and eutrophication lies in 1)
    sufficient dilution of waste accompanied by adequate stream residence
    time, temperature, and reaeration;  2)  groundwater basin characterization
    leading to responsible land-use planning; and 3) groundwater basin
    management (e.g., no sinkhole dumping, removal of existing pollution
    point-sources); and/or 4) waste collection and treatment prior to
    discharge.  Cave ecosystems are particularly vulnerable to contaminant
    inputs since they are often in the zone of degradation immediately
    downstream of pollution sources.  Elevated organic carbon and phosphate
    levels, as well as fecal  coliform,  will initially lead to a reduction in
    dissolved oxygen used in  satisfying the biochemical oxygen demand (BOD) in
    streams.  Farther downstream of the contaminant source,  bacteria and
    fungi thrive on the decomposition of organics, thus increasing ammonia
    nitrogen and decreasing the BOD (Hammer, 1975).  The resulting increase
    in inorganic nutrients downstream of the zones of degradation and
    active decomposition can  result in an unnaturally nutrient-rich
    environment.  An increase in inorganic compounds, such as nitrate,
    phosphates, and dissolved salts is common prior to the clear water zone
    (Hammer, 1975).  In severely contaminated streams, the clear water zone
    may be far downstream of  areas degraded by contaminants.

2.   In light of your statements regarding the importance of evaluating
    carbonate hydrology by using "karst" thinking and techniques, are the
    methods you discussed being used to evaluate environmental problems  on
    the Oak Ridge Reservation?  If not, should they?

    It is recognized that karst dissolution conduits may be a significant
    factor in groundwater flow in portions of the Oak Ridge Reservation
    and surrounding area.  Figure 2 and the Clarksville example discussed in
    the text, are illustrative of the type of "karst" thinking and techniques
    required to evaluate environmental problems on the Oak Ridge Reservation.
    However, field work and characterization remains to be conducted on  the
    carbonate flow systems at the three Oak Ridge Reservation plant sites and
    their catchment basins.  While some isolated tracer studies have been
    conducted, a more detailed and systematic approach to defining the nature
    and extent of the mature  karst systems present is necessary.  In one such
    example, the author and groundwater program managers at the Reservation
    are currently spearheading a groundwater characterization program which
    emphasizes the karst flow dynamics proximal to the K-25 plant site.   The
    methods which will be employed are consistent with the methods discussed
    in this paper, and will employ sophisticated characterization techniques,
    including geophysics, continuous monitoring of hydrographic response in
    conduit wells, and tracer tests.
                                      793

-------