OCR error (C:\Conversion\JobRoot\00000CDD\tiff\20013J3F.tif): Unspecified error
-------
COVER: Ground-water flow routes and pollutant dispersal in the
Horse Cave and Cave City area, Kentucky, just east of Mammoth
Cave National Park. Dye-tracing and mapping of the potentio-
metric surface (the water table) has made it possible to
determine: 1) the boundaries of ground-water basins, 2) the
flow-routing of sewage effluent discharged into the aquifer,
3) the flow-routing of other of pollutants that might be dis-
charged into the aquifer anywhere else in the map area, and 4)
the recharge area of springs. Reliable monitoring for pollu-
tants here and in most karst terranes can only be done at
springs, wells drilled to intercept known cave streams, wells
known to become turbid after heavy rains, and wells drilled on
photo-lineaments — but only if each site proposed for moni-
toring has been shown by dye-tracing to drain from the site to
be monitored. (reproduced from: Quinlan, James F. , Special
problems of groundwater monitoring in karst terranes, in
Neilsen, David M. , and Quinlan, James, F., eds., Symposium on
Standards Development for Groundwater and Vadose Zone Moni-
toring Investigations. American Society for Testing and
Materials, Special Technical Paper, 1988. (in press)
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
REGION IV
343 COUMTLAND STREET
ATLANTA. GEORGIA SOifS
Application Of Dye-Tracing Techniques For Determining Solute-Transport
Characteristics Of Ground Water In Karst Terranes
Prepared By
U.S. Environmental Protection Agency
Ground-Water Protection Branch
Region IV - Atlanta, Georgia
and
U.S. Geological Survey
Water Resources Division
Kentucky District
Louisville, Kentucky
Approximately 20Z of the United States is underlain by karst aquifers.
This approximation Includes roughly 50% of both Kentucky and Tennessee,
substantial portions of northern Georgia and Alabama, and parts of other
Region IV states. The prevalence of karst aquifers in the southeast, the
common use of karst aquifers as drinking water sources and the vulner-
ability of these aquifers to contamination highlighted the need to provide
a mechanism to assist in ground-water management and protection in karst
terranes. In an attempt to meet this need, the U.S. Environmental
Protection Agency (EPA) - Region IV and the Kentucky District of the U.S.
Geological Survey (USGS), have been cooperating to document the applica-
tion of dye tracing techniques and concepts to ground-water protection in
karst aquifers. I am pleased to announce that these efforts have resulted
in the preparation of this manual, entitled, "Application Of Dye-Tracing
Techniques For Determining Solute-Transport Characteristics Of Ground
Water In Karst Terranes." The information presented herein should be
viewed as another analytical "tool" to assist in the management and
protection of karst water supplies.
Ofeer C. Tidwell•
Regional Administrator
SEP 27198*
-------
APPLICATION OF DYE-TRACING TECHNIQUES FOR DETERMINING
SOLUTE-TRANSPORT CHARACTERISTICS OF GROUND WATER
IN KARST TERRANES
by
D.S. Mull, T.D. Liebermann, J.L. Smoot,
and L.H. Voosley, Jr., of the
U.S. Geological Survey
Water Resources Division
Louisville, Kentucky
EPA Project Officer
Ronald J. Mikulak
Ground-Water Protection Branch
Region IV, U.S. Environmental Protection Agency
Atlanta, Georgia 30365
-------
FOREWORD
The dawning of the "Age of Environmental Awareness" has been
accompanied by great advances in the study of karst hydrology —
its methodology, results, and understanding. This manual is
another of those advances. It describes a clever application to
subsurface streams of empirical techniques for study of surface
streams. It makes possible a good approximation of the time of
travel, peak concentration, and flow duration of harmful contami-
nants accidentally spilled into many karst aquifers and flowing
to a spring or well, and it does so for various discharge condi-
tions. As such, these techniques for analysis and interpretation
of dye-recovery curves are powerful, useful tools for the protec-
tion of water supplies. These techniques are an important
complement to those essential for delineation of wellheads and
springheads. For karst terranes characterized by conduit-flow,
such delineation can only be done by dye-tracing, preferably
guided by inferences from good potentiometric maps.
Interpretation of dye-recovery curves can yield much informa-
tion about the nature of groundwater flow in a karst aquifer .and
the structure of its conduit system, as shown in this manual, by
Maloszewski and Zuber (1985), Zuber (1986), Caspar (1987), Smart
(1988), and other investigators. Similarly, much can be learned
from interpretation of flood-discharge hydrographs of springs, as
shown by Wilcock (1968), Brown (1972), Sara (1977), Podobnik
(1987), White (1988, p. 183-186), Meiman et al. (1988), and oth-
ers.
An unstated, implicit assumption made by the authors of this
manual in their analysis of dye-recovery curves is that flow is
from a single input to a single output. This is a reasonable
assumption in many karst terranes, but there are five other sig-
nificant types of karst networks, those with:
1. One or more additional unknown inputs;
2. One or more additional unknown outputs;
3. Both one or more additional unknown inputs and outputs;
4. Distributary flow to multiple outlets;
5. Cutarounds and braided passages. (A cutaround is a pas-
sage bifurcation in which flow diverges from and then con-
verges to the main conduit. The flow routes rejoin one
another, but one of them is longer and/or less hydraulic-
ally efficient than the other. Where convergence occurs,
a second pulse, lagging behind the initial pulse, is
formed in the main conduit.)
The first three types of karst network are illustrated and
discussed by Brown and Ford (1971), Brown (1972), and Caspar
(1987) . If flow in an aquifer is through the first type, pre-
dictions based on the procedures recommended in this manual will
be accurate. If flow is through the second or third types, the
accuracy of the predictions will tend to be inversely proportion-
al to the amount of dye diverted to unknown discharge points.
Distributary flow is a subtype of the second and third types of
iii
-------
network; it is common in karsts of the midwestern United States.
Flow through cutarounds and braided passages induces bimodal-
ity and polymodality in dye-recovery curves, as illustrated and
discussed by Caspar (1987) and Smart (1988). If such curves de-
part significantly from those for a relatively simple system like
that described in this manual, the accuracy of predictions in the
fifth type of network will be inversely proportional to the time
between the maxima and will also be affected by the number of
maxima and the extent to which they are similar to the greatest
maximum.
Another factor that influences the shape of a dye-recovery
curve is the extent to which tracer penetrates the bedrock matrix
deeply enough to be influenced by adjacent fractures (Maloszewski
and Zuber, 1985; Zuber, 1986). Such penetration is strongly in-
fluenced by the porosity of the matrix and is inversely propor-
tional to the flow velocity; matrix penetration by tracer can
affect both the duration of a test and the shape of its dye-
recovery curve.
A karst aquifer may or may not lend itself to the dye-
recovery analysis proposed in this manual. But one will not find
out — or determine the type of karst network present — until
and unless dye-tests are run and interpreted. Results of the
dye-recovery analysis are vital for wellhead and springhead pro-
tection, especially if flow is through a relatively simple con-
duit system like that studied by the authors.
It was the intention of the authors to only briefly summarize
a few principles and descriptions of karst hydrology and geomor-
phology; more would be beyond the intended scope of this manual.
For a fuller discussion of these topics the reader is referred to
the recent excellent textbook by White (1988). Similarly, and
although there are much data and unique information on dye-trac-
ing in this manual, the reader desiring a comprehensive how-to
(and how-not-to) handbook on various dye-tracing techniques and
instrumentation is referred to the enchiridion by Aley et al.
(1989).
Flow velocities in karst aquifers may be tens of thousands to
many millions of times faster than flow in granular aquifers.
Therefore, it would be prudent for managers of water supplies in
karst terranes to have the recharge areas for their springs and
wells delineated and to have dye-recovery analysis done for the
sites most susceptible to accidental spills of harmful contami-
nants. They should do so before it is too late. Although dye-
injection immediately after a spill can sometimes be used to
monitor the probable arrival time of pollutants, the tracing
necessary for applying the dye-recovery analysis described in
this manual can only be done before the spill occurs.
Dye-tracing, like neurosurgery, can be done by anyone. But
when either is needed, it is judicious and most cost-efficient to
iv
-------
have it done by experienced professionals, those who have already
made the numerous mistakes associated with learning or those who
have trained under the tutelage of an expert and learned to avoid
numerous procedural errors that could have economically and
physically fatal consequences.
The analytical technique for dye-recovery analysis described
in this manual is best applied to systems similar to the rela-
tively simple but common one studied by the authors. According-
ly, their approach should be widely applicable. When combined
with basin delineation by dye-tracing, the two techniques repre-
sent the best pre-spill hydrologic preparations available for
spill-response in karst terranes. Their technique is a signifi-
cant advance in the evaluation of karst aquifers. Use of it and
testing of it is encouraged and recommended.
References Cited
Aley, T., Quinlan, J.F., Vandike, J.E., and Behrens, H., 1989.
The Joy of Dyeing: A Compendium of Practical Techniques for
Tracing Groundwater, Especially in Karst Terranes. National
Water Well Association, Dublin, Ohio. [in prep.]
Brown, M.C., 1972. Karst hydrology in the lower Maligne Basin,
Jasper, Alberta. Cave Studies, no. 13. 84 p.
Brown, M.C., and Ford, D.C., 1971. Quantitative tracer methods
for investigation of karst hydrologic systems. Cave Research
Group of Great Britain, Transactions, v. 13, p. 37-51.
Caspar, E., 1987. Flow through hydrokarstic structures, in
Caspar, E., ed., Modern Trends in Tracer Hydrology, v. 2. CRC
Press, Boca Raton, Florida, p. 31-93.
Maloszewski, P., and Zuber, A., 1985. On the theory of tracer
experiments in fissured rocks with a porous matrix. Journal
of Hydrology, v. 79, p. 333-358.
Meinman, J., Ewers, R.O., and Quinlan, J.F., 1988. Investigation
of flood pulse movement through a maturely karstified aquifer
at Mammoth Cave, Kentucky. Environmental Problems in Karst
Terranes and Their Solutions (2nd Conference, Nashville,
Tenn.) Proceedings. National Water Well Association, Dublin,
Ohio. [in press]
Podobnik, R., 1987. Rezultati poskusov z modeli zaganjalk (Ex-
perimental results with ebb and flow spring models) . Acta
Carsologica, v. 17, p. 141-165. [with English summary]
Sara, M.N., 1977. Hydrogeology of Redwood Canyon, Tulare County,
California. M.S. thesis (Geology), University of Southern
California. 129 p.
Smart, C.C., 1988. Artificial tracer techniques for the deter-
mination of the structure of conduit aquifers. Ground Water,
V. 26, p. 445-453.
White, W.B., 1988. Geomorphology and Hydrology of Karst Ter-
rains. Oxford University Press, N.Y. 464 p.
Wilcock, J.D., 1968. Some developments in pulse-train analysis.
Cave Research Group of Great Britain, Transactions, v. 10, no.
2, p. 73-98.
-------
Zuber, A., 1986. Mathematical models for the interpretation of
environmental isotopes in groundwater systems, in Fritz, P.,
and Fontes, J.C., eds., Handbook of Environmental Isotope Geo-
chemistry, v. 2. Elsevier, Amsterdam, p. 1-59.
James F. Quinlan
National Park Service
Mammoth Cave, KY 42259
VI
-------
APPLICATION OF DYE-TRACING TECHNIQUES FOR DETERMINING
SOLUTE-TRANSPORT CHARACTERISTICS OF GROUND WATER
IN KARST TERRANES
CONTENTS
Page
Foreword ill
Abstract 1
1. Introduction 3
1.1 Background 3
1.2 Purpose and scope 5
1.3 Acknowledgments 6
2. Hydrogeology of karst terrane 6
2.1 Karst features 9
2.1.1 Sinkholes 9
2.1.2 Karst windows 13
2.1.3 Losing, sinking, gaining, and underground
streams, and blind valleys 14
2.1.4 Karst springs 15
2.2 Occurrence and movement of ground water 17
2.3 Vulnerability of karst aquifers to contamination 20
3. Dye tracing concepts, materials, and techniques 22
3.1 Introduction 22
3.1.1 Dye characteristics and nomenclature 23
3.1.2 Fluorescent tracers 24
3. 2 Qualitative dye tracing 26
3.2.1 Introduction 26
3.2.2 Selecting dye for injection 27
3.2.3 Selecting quantity of dye for injection 28
3.2.4 Dye-handling procedures 29
3.2.5. Dye - recovery equipment and procedures 30
3.3 Quantitative dye tracing 35
3.3.1 Introduction 35
3.3.2 Selecting dye for injection 37
3.3.3 Selecting quantity of dye for injection 37
3.3.4 Dye-handling and recovery procedures 38
3.3.5 Fluorometer use and calibration 39
3.3.6 Calculating mass of injected and recovered dye... 42
3.3.7 Sampling procedures 43
3.3.8 Sample handling and analysis 44
3.3.9 Adjustment of dye-recovery data 44
3.4 Quality-control procedures 46
3.5 Summary decision charts 47
4. Analysis and application of dye-trace results 50
4.1 Introduction 50
4.2 Definition of quantitative characteristics 51
4.2.1 Dye - recovery curve 52
4.2.2 Normalized concentration and load 54
4.2.3 Time-of-travel characteristics 56
4.2.4 Dispersion characteristics 59
4.2.5 Summary of quantitative terms 61
4.2.6 Example of computation of quantitative
characteristics 61
vii
-------
CONTENTS- -Continued
Page
5. Application of quantitative dye-trace results for predicting
contaminant transport 66
5.1 Introduction 66
5.2 Relations among quantitative characteristics 66
5.3 Development and use of dimensionless dye-recovery curve... 68
5.4 Prediction of contaminant transport 75
6. Use of computer programs 79
6.1 Program DYE 79
6.2 Program SCALE 80
6. 3 Program SIMULATE 81
Selected references 83
Appendix
A. Computer programs DYE 90
A. 1 Programming code 90
A. 2 Sample of input 94
A. 3 Sample of output 95
B. Computer program SCALE 96
B. 1 Programming code 96
B. 2 Sample of input 97
B. 3 Sample of output 98
C. Computer program SIMULATE 99
C. 1 Programming code 99
C. 2 Sample of input 100
C. 3 Sample of output 102
viii
-------
ILLUSTRATIONS
Page
Figure 1. Hap showing distribution of karst areas in relation to
carbonate and other soluble rock in the conterminous
United States 7
2. Diagram showing diffuse, mixed, and conduit flow in a
hypothetical ground-water basin, showing the
sequence of its evolution 9
3. Diagram of common types of karst springs 16
4. Block diagram showing components of ground-water flow in a
mature karsted aquifer 19
5. Sketch of anchor used to suspend dye detectors (bugs) in
springs or streams.. : 31
6. Plot showing a typical dye-recovery (time-
concentration) curve 36
7. Sketch of basic structure of most filter fluorometers 41
8. Plots of typical response curves observed laterally and at
different distances downstream from a dye-injection point... 45
9. Decision chart for qualitative dye tracing 48
10. Decision chart for quantitative dye tracing 49
11.-15. Graphs showing:
11. Dye-recovery curve illustrating measures of elapsed
time since injection 53
12. Normalized dye-recovery curves for seven dye traces
made for various discharges at Dyers Spring 55
13. Normalized dye-load curves for seven dye traces made
for various discharges at Dyers Spring 57
14. Development of selected quantitative characteristics
for dye trace at Dyers Spring, May 30, 1985 64
15. Relation of selected quantitative characteristics
to discharge based on seven dye traces to
Dyers Spring 69
16. Development of a standardized, dimensionless recovery
curve for Dyers Spring 73
17. Comparison of normalized dye-recovery curve to simulated
curve for Dyers Spring, May 30, 1985 76
18. Simulation of a dye-recovery (time-concentration) response
curve resulting from a hypothetical contaminant spill
near Dyers Spring 78
TABLES
Table 1. Characteristics of commonly used tracer dyes 25
2. Sources of materials and equipment for dye tracing 32
3. Three-step serial dilution for preparation of standards
for fluorometer calibration 42
4. Summary of terms used in definition of quantitative
characteristics 62
5. Measured and computed data for dye trace at Dyers Spring,
May 30, 1985 63
6. Quantitative characteristics of seven dye traces at
Dyers Spring 68
ix
-------
Factors for Converting Inch-Pound Units to
International System of Units (SI)
The inch-pound units used in this report may be converted to metric
(International System) units by the following factors:
Multiply inch-pound units By
inch (in.) 25.4
feet (ft) 0.3048
mile (mi) 1.609
square mile 2.590
(mi2)
gallon per minute 0.06309
(gal/min)
million gallons per day 0.04381
(Mgal/d)
3,785
foot per second
(ft/s)
cubic foot per second
(fts/s)
foot squared per second
(ft2/s)
micromhos per centimeter
at 25° Celsius
(pmhos/cm at 25* C)
0.305
0.02832
0.09290
1.000
To obtain metric unit
millimeter (mm)
meters (m)
kilometer (km)
square kilometer
(km')
liters per second
cubic meter per second
cubic meter per day
(ms/d)
meter per second
(m/s)
cubic meter per second
meter squared per second
(ma/s)
microsiemens per centimeter
at 25* Celsius
(uS/cm at 25* C)
Temperature in degrees Fahrenheit (*F) can be converted to degrees Celsius
(°C) as follows:
°C -
32)/1.8
Sea level: In this report "sea level" refers to the National Geodetic
Vertical Datum of 1929 (NGVD of 1929)--a geodetic datum derived from a general
adjustment of the first-order level nets of both the United States and Canada,
formerly called "Sea Level Datum of 1929."
Use of brand/firm/trade names in this report is for identification purposes
only and does not constitute endorsement by the U.S. Geological Survey or the
U.S. Environmental Protection Agency.
-------
APPLICATION OF DYE-TRACING TECHNIQUES FOR DETERMINING
SOLUTE-TRANSPORT CHARACTERISTICS OF GROUND WATER
IN KARST TERRANES
By D.S. Mull, T.D. Liebermann, J.L. Smoot,
and L.H. Woosley, Jr.
ABSTRACT
Some of the most serious incidents of ground-water contamination,
nationwide, have been reported in karst terranes. Karst terranes are
characterized by sinkholes; karst windows; springs; caves; and losing,
sinking, gaining, and underground streams. These karst features are
environmentally significant because they are commonly directly connected to
the ground-water system. If these ground-water systems are used as a drinking
water source, their environmental significance is increased. Sinkholes are
especially significant because they can funnel surface runoff to the ground-
water system. Thus, any pollutant carried by surface runoff across a karst
terrane has the potential for rapid transport to the ground-water system.
Because of the extreme vulnerability of karst ground-water systems to
contamination, water-management and protection agencies need an understanding
of the occurrence of ground water, including the extent of the recharge areas
for specific karst aquifers, a knowledge of the inherent vulnerabilities of
the systems, and an understanding of the characteristics of pollutant
transport within the systems. To provide water managers (those responsible
for providing and managing water supplies) and protection agencies (those
responsible for regulating water supplies and water quality) with a tool for
the management and protection of their karst water resources, the U.S.
Geological Survey in cooperation with Region IV of the U.S. Environmental
Protection Agency has prepared this manual to illustrate the application of
dye-tracing results and the related predictive techniques that could be used
for the protection of ground-water supplies in karst terrane. This manual
will also be useful for State and local agencies responsible for implementing
Wellhead Protection Programs pursuant to the Safe Drinking Water Act as
amended in 1986.
This manual includes a brief review of karst hydrogeology, summarizes
dye-tracing concepts and selected techniques, lists sources for equipment and
materials, and includes an extensive list of references for more detailed
information on karst hydrogeology and on various aspects of dye tracing in
karst terrane. Both qualitative and quantitative dye-tracing techniques are
described and quantitative analyses and interpretation of dye-trace data are
demons trated.
Qualitative dye tracing with various fluorescent dyes and passive dye
detectors, consisting of activated coconut charcoal or surgical cotton, can be
used to identify point-to-point connections between points of ground-water
recharge, such as sinkholes, sinking streams, and karst windows and discharge
points, such as water-supply springs and wells. Results of qualitative
tracing can be used to confirm the direction of ground-water flow inferred
-------
from water-level contour maps, and to help delineate the recharge area
draining to a spring or well. Qualitative dye tracing is, generally, the
first step in the collection and interpretation of quantitative dye trace
data.
Quantitative dye tracing usually requires automatic samplers, discharge
measurements at the ground-water resurgence, and fluorometric or
spectrofluorometric analysis to quantify passage of the dye cloud. These
results can be used to determine solute-transport characteristics such as
traveltime for arrival of the leading edge of the dye cloud, peak dye
concentration, trailing edge, and persistence of the dye cloud at the
discharge point, which may be a spring or well used for public water supply.
Repeated quantitative dye traces between the same recharge and discharge
points, under different flow conditions, can be used to develop predictive
relations between ground-water discharge, apparent ground-water flow velocity,
and solute-transport characteristics. Normalized peak-solute concentration,
mean traveltime, and standard deviation of time of travel can be used to
produce a composite, dimensionless recovery curve that is used to simulate
solute-transport characteristics for selected discharges. Using this curve
and previously developed predictive relations, a water manager can estimate
the arrival time, peak concentration, and persistence of a soluble
conservative contaminant at a supply spring or well, on the basis of discharge
and the quantity of spilled contaminant.
-------
1. INTRODUCTION
1.1 Background
The primary objective of the manager of a public water-supply system is
to provide the consumer with a safe, dependable supply of drinking water. In
large areas of many states, ground water is the exclusive or primary source of
drinking water. Often, disinfection is the only treatment used to meet
applicable public drinking-water standards. However, reports of ground-water
contamination nationwide, combined with increasing dependence on ground water,
have led to a growing awareness of the potential for degradation of this
valuable source of drinking water.
Almost all ground water is vulnerable to contamination, whether the
contamination is caused by natural geologic or hydrologic conditions or by
man's activities. Karst ground-water supplies are particularly vulnerable to
contamination because of the relatively direct connection to surface
activities and the rapid transport of surface runoff and contaminants to karst
ground-water systems. The potential for contamination of karst ground-water
systems from man-made sources is particularly great where urban areas and
major transportation corridors are built in the recharge areas of karst
aquifers.
Karst terrane is characterized by surface and subsurface features, such
as sinkholes; karst windows; springs; caves; and losing, sinking, gaining and
underground streams. Sinkholes are environmentally significant land forms
because they can provide a direct path for surface runoff to recharge karst
aquifers. They commonly lead directly to the aquifer system through pipe-like
openings in residuum and bedrock. Some sinkholes may also act as collection
and retention basins for surface runoff. Thus, depending upon the size of the
area draining to the sinkhole and the nature of the subsurface openings,
relatively large quantities of water may enter the aquifer system in a short
period of time. Where sinkholes occur, any pollutant carried by surface
runoff has the potential for rapid transport to ground water.
Public water-supplies in karst terranes may be more vulnerable to
detrimental effect than nonkarst, ground-water supplies. Because of the
variability of soil cover and the likelihood of overlying soils being shallow
or absent in karst areas, the potential exists for little or no enhancement of
water quality before surface water is recharged to the aquifer system. Also,
pollutant traveltime in a karst aquifer can be rapid, on the order of miles
per day in contrast to feet per year in most non-karst aquifers. Therefore,
the managers of a water supply derived from ground water in a karst terrane
need to have a detailed understanding of the extent of the aquifer recharge
area, a knowledge of the inherent vulnerabilities of the aquifer, and an
understanding of how pollutants move through the system. Accordingly,
specialized qualitative techniques are required to delineate the recharge
areas of karst aquifer and to identify continuity between potential recharge
and discharge points of aquifers. In addition, predictive techniques are
needed in order that the water-supply manager can effectively respond to the
presence of contaminants in the karst aquifer.
-------
In response to the widespread need for the protection of vulnerable
ground-water supplies, Congress enacted the 1986 Amendments to the Safe
Drinking Water Act. Prior to 1986, the Federal statutes available to the U.S.
Environmental Protection Agency (EPA), although designed for more general
purposes, provided substantial protection for ground water (U.S. Environmental
Protection Agency, 1984, p. 23). These statutes are designed to protect
ground water by focusing on controlling specific contaminants or sources of
contamination. However, with the enactment of the 1986 Amendments to the Safe
Drinking Water Act, there is for the first time a Federal statutory goal for
the protection of ground water as reflected by the establishment of the
Wellhead Protection Program. This goal represents a significant change in the
roles and relations of Federal, state, and local governments with regard to
ground-water protection.
The Wellhead Protection Program (Section 1428 of the Safe Drinking Water
Act) is a state program designed to prevent contamination of public water-
supply wells and well fields that may adversely affect human health. Although
springs that supply public drinking water were not specifically mentioned in
the statute, it has been interpreted by the EPA that the protection of public
water-supply springs should be included in the program.
The Act requires states to develop programs to protect the wellhead
protection area of all public water-supply systems from contaminants that may
have any adverse effects on the health of humans. A wellhead protection area
is defined by statute as the surface or subsurface areas surrounding
wellfields through which contaminants are reasonably likely to move toward and
reach such wells or wellfields.
The Act specifies that the following elements be incorporated into state
programs:
A description of the duties and responsibilities of state and local
agencies charged with the protection of public water-supplies:
Determination of wellhead protection areas for each public water-
supply well.
Identification of all potential man-made sources within each
wellhead protection area.
As appropriate, technical assistance; financial assistance;
implementation of control measures; and education, training,
and demonstration projects to protect the wellhead areas.
Contingency plans for alternative water supplies in case of
contamination.
Siting considerations for all new wells.
Procedures for public participation.
States electing to develop a Wellhead Protection Program need to submit
their program proposal to the EPA by June 1989 and the program needs to be
implemented within two years of approval. The EPA has established a policy
that states shall have considerable flexibility in carrying out the program.
-------
Guidance available from the EPA to states in developing their programs
include "Guidance for Applicants for State Wellhead Protection Program
Assistance Funds Under the Safe Drinking Water Act" (U.S. Environmental
Protection Agency, 1987a) and "Guidelines for Delineation of Wellhead
Protection-Areas" (U.S. Environmental Protection Agency, 1987b).
Recognizing the occurrence of water supplies in karst aquifers of the
southeastern United States, the U.S. Geological Survey in cooperation with
Region IV of the Environmental Protection Agency, prepared this manual to
assist Federal, State, and local agencies in ground-water management and
protection in karst terranes to support the Wellhead Protection Program. The
manual demonstrates the application of dye-tracing concepts and techniques for
determining solute-transport characteristics of ground water in karst terranes
and illustrates the development of predictive techniques for ground-water
protection.
With the serious potential for ground-water contamination and the need to
identify the areas most likely to drain directly to karst ground-water
systems, numerous investigations by state and Federal agencies, university
researchers, and environmental consulting firms have been conducted to better
define the hydraulic nature of these systems. Some objectives of these
investigations include, but are not limited to, the location and
classification of sinkholes most susceptible to surface runoff; the
identification of point-to-point hydrologic connections by dye traces between
selected sinkholes, losing and sinking streams, and public water-supply
springs and wells; and the definition of the relation between precipitation,
storm-water drainageways, streams, sinkhole drainage, ground-water movement,
and downgradient springs and wells.
Information gained from these studies has been helpful to local, state,
and Federal water supply management and protection agencies and researchers in
their efforts to develop aquifer and well-head protection plans. The results
have been useful for developing land-use controls around sinkholes whose
drainage has been traced to public water-supply springs or wells. Recent
studies have demonstrated the use of predictive techniques for estimating
solute transport in karst terranes.
1.2 Purpose and Scope
The purposes of this manual are to provide a review of the hydrogeology
of karst terranes, summarize concepts and techniques for dye tracing, and
describe and demonstrate the application of dye-trace data to determine
solute-transport characteristics of ground-water in karst terranes. The
manual was prepared in support of the Wellhead Protection Program pursuant to
the 1986 Amendments to the Safe Drinking Water Act.
The dye-tracing procedures and the analysis and application of dye-trace
data provided in this manual were used to determine ground-water flow
characteristics in the Elizabethtown area, Kentucky (Mull, Smoot, and
Liebermann, 1988). In general, these techniques may be used in other karst
areas with similar hydrologic characteristics. The quantitative analyses of
dye-recovery data and the development of prediction capabilities are most
useful in areas where ground-water flow occurs primarily in conduits that
drain to a spring or springs where discharge can be measured.
5
-------
1.3 Acknowledgments
The authors are grateful to James F. Quinlan of the U.S. National Park
Service; and Ronald J. Mikulak of the EPA, and Robert E. Faye and John K.
Carmichael of the U.S. Geological Survey; each of whom performed a technical
review of the manual. Their constructive criticism was beneficial to the
technical content and accuracy of the manual.
The authors are also grateful to the City of Elizabethtown, which
provided much of the dye-trace data used in this manual, for its cooperative
support of a dye-trace investigation by the U.S. Geological Survey.
2. HYDROGEOLOGY OF KARST TERRANE
Karst is an internationally used word for terranes with characteristic
hydrology and landforms. Most karst terranes are underlain by limestone or
dolomite, but some are underlain by gypsum, halite, or other relatively
soluble 'rocks in which the topography is chiefly formed by the removal of rock
by dissolution. As a result of the rock solubility and other geological
processes operating through time, karst terranes are characterized by unique
topographic and subsurface features. These include sinkholes; karst windows;
springs; caves; and losing, sinking, gaining, and underground streams, but in
some terranes one or more of these features may be dominant. The hydrology of
aquifers underlying karst terranes is markedly different from that of most
granular or fractured-rock aquifers because of the abundance, size, and
integration of solutionally enlarged openings in karst aquifers.
There are many different geologic settings and hydrologic conditions that
influence the development and hydrology of karst terranes. For the purposes
of this manual, many important aspects of the hydrogeology of karst terrane
can be mentioned only briefly. A comprehensive guide to the hydrogeology of
karst terranes and its literature has been published by White (1988). It and
other references cited herein will direct the reader to more complete
discussions of various aspects of the hydrogeology of karst terranes.
According to Davies and LeGrand (1972), about 15 percent of the
conterminous United States, consists of limestone, gypsum, and other soluble
rock at or near the land surface (fig. 1). Karst terranes are particularly
well developed in the following areas: (1) Tertiary Coastal Plain of Georgia
and Florida, (2) Paleozoic belt of the Appalachian Mountains stretching from
Pennsylvania to Alabama, (3) nearly flat-lying Paleozoic rocks of Alabama,
Tennessee, Kentucky, Ohio, Indiana, Illinois, Wisconsin, Minnesota, Arkansas
and Missouri, (4) nearly flat-lying Cretaceous carbonate rocks in Texas, (5)
nearly flat-lying Permian rocks of New Mexico, and (6) the Paleozoic belt of
folded rocks in South Dakota, Wyoming, and Montana (LeGrand, Stringfield, and
LaMoreaux, 1976). Much of the subsurface of the Coastal Plain in South
Carolina and Alabama is a karst aquifer, but there is minimal surface
expression of typical karst features. Most of the karst areas are underlain
by carbonate rocks that have varying amounts of fractures. The fractures
usually are enlarged by solution where they are in the zone of ground-water
circulation. The enlargement of the fractures is controlled, in part, by
geologic structure and lithology.
-------
110°
fv T^^ »^ i»
I *N» i ^ • ^ C
: h:!''"»l \ ^C ~-~~?
^ r"""'^^-\^^ ;
** /• * * ." • 4 > «».^Sw> '
I J* f'e I ^ . ^%~^^!^-—G$
100°
9(T
80"
40°
'.'/'
:V^
- » »v ,,
30°
(from W.E. Davis, in Komatina, 1975, figure 7, p. 23)
EXPLANATION
A "Atlantic and Gulf Coastal Plain regions
B SP East-central region
C * Great Plains region
D -Western mountain region
Hi Karst areas
200
400 «oo MILES
0 200 400 «00 KILOMETER!
Carbonate and other
soluble rock at or near
the surface
Figure 1.—Distribution of karst areas in relation to carbonate and other
soluble rock in the conterminous United States.
-------
There are five key elements necessary for a ground-water basin to develop
in carbonate rocks. It must have: (1) an area of intake or recharge, (2) a
system of interconnected conduits that transmit water, (3) a discharge point,
(4) rainfall, and (5) relief. If any one of these elements is missing, the
rock mass is hydrologically inert and likely cannot function as a ground-water
basin.
Ground-water recharge occurs as infiltration through unconsolidated
material overlying bedrock or as direct inflow from sinking streams and open
swallets. Infiltrated water moves vertically until it intercepts relatively
horizontal conduits that have been enlarged by the solutional and erosive
action of flowing water. Springs are the discharge points of the ground-water
basin and usually are located at or near the regional base level or where
insoluble rocks or structural barriers such as faults, impede the solutional
development of conduits.
Adequate rainfall is necessary for the solution of limestone to take
place. Karst development tends to be absent if precipitation is less than 10-
12 inches per year. Maximum karstification occurs in regions of heavy
precipitation and in regions with marked seasons of heavy precipitation and
drought (Sweeting, 1973, p. 6).
The development of ground-water basins requires vertical and horizontal
circulation of ground-water. Such development is enhanced if available relief
places the soluble rock above the regional base level.
The presence of solutionally enlarged fractures presents unique problems
for water managers in karst terrane because of the velocity of ground-water
flow and the possibility that relatively little water-quality enhancement
occurs while the water is in transit within the karst aquifers. Ground-water
velocities in conduits may be as high as 7,500 feet per hour (ft/hr) where the
potentiometric gradient is as steep as 1:4 (Ford, 1967). Under fairly typical
ground-water gradients of 0.5 to 100 feet per mile (ft/mi), velocities range
from 30 ft/hr during base flow to 1,300 ft/hr during flood flow within the
same conduit (Quinlan and others, 1983, p.11). Under such conditions,
pollutants can impact water quality more than 10 miles away in Just a week
during base flow (Vandike, 1982) and much sooner during flood flow.
Where fractures within a bedrock aquifer are well developed and ground-
water flow is convergent to major springs via well developed conduits, the
aquifer is considered to be mature. Mature carbonate aquifers are generally
developed beneath mature karst terrane, having well-developed sinkholes that
collect and drain surface runoff directly into the subsurface conduit system.
Streams can also drain to the subsurface through swallets developed in the
stream bed or disappear into a swallet at the end of a valley.
In maturely karstified terranes, springs in a given area generally have
similar flow and water-quality characteristics. Spring discharge is normally
flashy, responding rapidly to rainfall. Flow is turbulent and turbidity,
discharge, and temperature are highly variable. Also, hardness is usually low
but highly variable. Springs with these characteristics are the outlets for
conduit-flow systems (Schuster and White, 1971, 1972) that usually drain a
discrete ground-water basin. Flow in a conduit system is similar to flow in a
-------
surface stream in that both are convergent through a system of tributaries and
both receive diffuse (non-concentrated) flow through the adjacent bedrock or
sediment.
If the aquifer is less mature, water moves through small bedrock openings
that have undergone only limited solutional enlargement. Flow velocities are
low and ground water may require months to travel a few tens of feet through
the carbonate bedrock (Freidrich, 1981; Freidrich and Smart, 1981). Discharge
from springs fed by slow-moving water in less mature karst is non-flashy,
relatively uniform, and responds slowly to storms. Flow is usually laminar,
turbidity is very low, and water temperature is very near the mean annual
surface-water temperature. Springs with these characteristics are typical of
ground-water outlets from diffuse-flow systems (Schuster and White, 1971,
1972).
Quinlan and Ewers (1985) propose that the major portion of ground-water
movement in a diffuse-flow (less mature karst) system is also through a
tributary network of conduits. Only in the headwaters of a ground-water basin
and adjacent to a conduit is flow actually diffuse. Their inspections in
quarries and caves show that the smallest microscopic solutional enlargements
of bedding planes and joints function as tributary conduits. They also
propose that conduit and diffuse flow in carbonate aquifers are end members of
a flow continuum (fig. 2). Although most carbonate aquifers are characterized
by both types of flow, as discussed in detail by Atkinson (1977), generally,
one type of flow predominates. Smart and Hobbs (1986) state that flow in
massive carbonate aquifers tends to be either predominately diffuse or
predominately conduit, depending on the degree of solutional development.
2.1 Karst Features
The most noticeable and direct evidence of karstification is the
landforms that are unique to karst regions. Most karst landforms are the
direct consequence of dissolution of soluble carbonate bedrock and are
characteristic of areas having vertical and horizontal underground drainage.
Although some karst landforms, such as sinkholes, may develop within a
relatively thick zone of unconsolidated regolith overlying bedrock, it is the
presence of solutionally enlarged openings in bedrock that ultimately controls
the development of such features. Karst features include sinkholes; karst
windows; springs; caves; and losing, gaining, sinking, and underground
streams. These features are discussed in detail by White (1988), Jennings
(1985), Milanovic (1981), and Sweeting (1973). The following discussion is
limited to karst features that are most related to the ground-water system,
primarily those that collect and discharge into, store and transmit, or
discharge water from the ground-water system.
2.1.1 Sinkholes
The term sinkhole has been used to identify a variety of topographic
depressions resulting from a number of geologic and man-induced processes. In
geologic research, the term doline is synonymous with sinkhole and has become
standard in the literature. However, Beck (1984, p. IX), cautions against the
use of the term sinkhole to describe depressions caused by mine collapse or
other man-induced activities which are not caused by karst processes.
-------
O
oc
O
I 3 CU
Is* co co
O 3
Q
O
O
o
* p
>i
§f*
• tii * M
j c o
Sss
s*
Uj
f)
?*
^J
t
Q
3
u
•« "I--
— 2
!
s
X
MM
S
e
£
!B
X
u
>
0
o
K
0
UBRIAN-
•<
U
ni
H
o
o
o
a
^
u
U
*»
»
m
I
-a-
CO
CU
u
3
00
•H
M-l
*
in
00
ON
*H
CO
u
CU
£
•o
e
cd
e
cd
i-i
c
•H
5-
e
o
h
vw
•O
CU
u
O.
cd
•a
cd
N*X
CO
01
e
•H
r-l
cd
9
•
CO
00
c
1-1
u
a
CO
CU
Wi
cd
CO
0)
rH
u
M
•H
u
•o
0)
e
CU
^
CJ
cd
rH
ca
•
CO
u
3
a.
C
•H
Wi
CU
4J
2
9
CU
M
cd
V-
o
14-1
S
o
j=
ca
CU
M
cd
ca
«
c
•rt
i-l
i-l
cd
•H
4J
e
CU
u
o
a
•H
3
cr
CU
T3
C
cd
CO
CU
e
•H
i-H
g
~4
fa,
•
.n
OJ
oc
Cd
ca
ca
cd
cu
•
B
Si
4J
»
>»
CO
u
.^4
3
T3
C
O
U
>>
f*
u
u
3
B.
•o
§
o
(.1
bO
i-l
cd
u
1-1
4J
CO
J3
t3
o
(X
1
cd
c
•H
»
0
I-l
(4-1
4-1
i-l
3
•5
c
0
u
•0
B
S
co
CU
c
1-1
ca
CU
4J
cd
u
•H
T3
U
U
CU
CO J=
U U
CU CO
b
cd **
o
ca
cu u
c e.
•H Cl
iH O
C
.* a
u u
ca
r-l CU
A .c
u
>,
> W Q
10
-------
Therefore, as used in this manual, the term sinkhole refers to an area of
localized land surface subsidence, or collapse, due to karst processes which
result in closed depressions (Monroe, 1970 and Sweeting, 1973, p. 45).
Sinkholes are depressions that can provide a direct path for surface
runoff to drain to the subsurface. They can occur singly or in groups in
close proximity to each other. In size, diameter is usually greater than
depth, and average sinkholes range from a few feet to hundreds of feet in
depth and to several thousand feet in diameter (Sweeting, 1973, p. 44).
Sinkholes are generally circular or oval in plan view but have a wide variety
of forms such as dish or bowl shaped, conical, and cylindrical (Jennings,
1985, p. 106). Increasing size is usually accompanied by increasing
complexity of form. Sinkholes frequently develop preferentially along joints
in the underlying bedrock. This often results in sinkholes which have a
distinct long dimension which corresponds with local Joint patterns.
The number of sinkholes generally depends on the nature of the land
surface and the depth of karstification. In general, plains tend to have a
large number of sinkholes of small size, and hilly terranes tend to have fewer
sinkholes but of larger size. On steep slopes, sinkholes are uncommon, but if
found they are likely the result of collapse of cave roofs rather than gradual
subsidence of surficial materials (Milanovic, 1981, p. 60).
Several natural processes contribute to the formation of sinkholes
including solution, cave-roof collapse, and subsidence. Sinkholes are
generally formed by the collapse or migration of surface material into the
subsurface that results in the typical funnel-shaped depressions. In general,
there are two types of collapse that form sinkholes: (1) slumping of surface
material (regolith collapse) into solutionally enlarged openings in limestone
bedrock and (2) collapse of carbonate (limestone) cave roofs. Sinkholes
caused by the collapse of cave roofs may be either shallow or deep-seated and
may develop suddenly when the cave roof can no longer support itself above the
underlying cave passage. Although dramatic and at times damaging, this
process is considered by some investigators to be the least common cause of
collapse sinkholes.
Sinkholes can also form as a result of man's activities which tend to
accelerate natural processes. Such induced sinkholes are caused primarily by
ground-water withdrawals and diversion or impoundment of surface water which
accelerates the downward migration of unconsolidated deposits into
solutionally enlarged openings in bedrock (Newton, 1987). Induced sinkholes
may provide convenient sites for dye injection and monitoring similar to
natural sinkholes. However, care must be used because the area around these
sites can be much less stable than in the vicinity of natural sinkholes.
There are numerous systems for the classification of sinkholes based on
characteristics as varied as their processes of formation, size and
orientation, and relation to surface runoff and the water table. The
classification system proposed for use in this manual is based on the relative
ability of the sinkhole to transmit water to the subsurface. This system
emphasizes the interrelation between sinkholes and the ground-water system and
especially the potential for sinkholes to funnel surface runoff directly into
the subsurface. The system was used by Mull, Smoot, and Liebermann (1988) to
identify those sinkholes with the greatest potential for contaminating the
11
-------
ground water in the Elizabethtown area, Kentucky. The criteria for sinkhole
classification are based on the material in which the sinkhole is developed
(sedimentary rock) and the presence or absence of drain holes (swallets).
These criteria yield four types of sinkholes:
(1) sinkholes developed in unconsolidated material overlying bedrock with
no bedrock exposed in the depression, but with well developed, open
swallets that empties into bedrock,
(2) sinkholes that have bedrock exposed in the depression and a well
developed swallet that empties into bedrock,
(3) sinkholes or depressions in which the bottom is covered or plugged
with sediment and in which bedrock is not exposed, and
(4) sinkholes in which bedrock is exposed but the bottom is covered or
plugged with sediment.
Sinkholes of types 1 and 2, having a well-developed, open drain or
swallet, are thought to have the greatest potential for polluting ground water
because the open drain is usually connected to subsurface openings that lead
directly to the ground-water system. Thus, there is no potential for
enhancement of water quality by processes such as filtration, as may be the
case if water percolates through soil or other unconsolidated material at the
bottom of types 3 and 4 sinkholes that do not have open swallets. Because of
the open drains, types 1 and 2 sinkholes offer the most direct method for
injecting tracers (dye) into the ground-water flow system because the tracers
can be added to water draining directly to the subsurface through the open
swallet.
Although types 3 and 4 sinkholes may be hydraulically connected to the
aquifer system, the potential for pollution is generally less than from
sinkholes with open drains because the percolation of water through sediments
may provide some enhancement of quality before the water reaches the aquifer.
Types 3 and 4 sinkholes can also be used as dye injection points during tracer
tests, but the quantity and type of dye used must reflect the fact that the
dye must first percolate through the soil plug before reaching the subsurface
flow system. Also, dye traveltimes from types 3 and 4 sinkholes are difficult
to predict because of the time required for water and dye to percolate through
the soil plug.
The nature of the swallet and the hydraulic characteristics of the
underlying aquifer will, in part, control the rate and quantity of water
draining from the sinkhole. Ponding or sinkhole flooding can occur in types 1
and 2 sinkholes if runoff exceeds the drainage capacity of the swallet or if
the subsurface system of conduits which receives sinkhole drainage is blocked.
Drainage from sinkholes is also controlled by the nature of sediment and
debris washed into the swallet. Heavy sediment loads, such as are common from
freshly disturbed construction or cultivated sites, coupled with large debris,
can partially obstruct or plug the swallet resulting in the ponding of water
in the sinkhole. Soil and debris plugs can be temporary and be flushed open
with subsequent heavy runoff. Mull and Lyverse (1984, p. 17) reported several
instances following periods of rapid runoff in which water collected in and
overflowed sinkholes because the drain was partially plugged. In some cases,
12
-------
water ponded in a sinkhole having a plugged swallet may overflow the sinkhole
and flow to other sinkhole swallets or surface streams. However, depending on
the shape and size of the sinkhole and the nature of the area draining to it,
the ponded water may not drain to an adjacent sinkhole or surface stream but
may remain in the drainage basin of the sinkhole and eventually drain to the
subsurface through the swallet.
Because virtually all surface runoff that is collected by sinkholes is
eventually funneled directly into the ground-water system, drainage through
sinkholes can seriously impact water supplies developed from underlying
carbonate aquifers. Thus, it is imperative that water managers identify those
sinkholes and sinkhole areas that recharge a particular karst water-supply
spring or well in order to develop adequate ground-water protection
procedures. Such procedures can be preventive, such as the application of
land-use restrictions around selected sinkholes, or reactive, such as the
determination of traveltimes and other aquifer flow characteristics developed
from quantitative dye tracing.
Sinkholes are often the most ubiquitous evidence of karstification.
Although some authors consider the absence of sinkholes sufficient evidence to
classify an area as non-karat, Dalgleish and Alexander (1984) point out that
in some areas of the Midwest more than 60 percent of the existing sinkholes
are not shown on 7 1/2-minute (1:24,000) topographic maps. Also, Quinlan and
Eweri (1985) state that karst cannot be defined solely in terms of the
presence or absence of sinkholes. As a generalization, almost any terrane
underlain by near-surface carbonate rocks has some degree of karst and can,
therefore, exhibit water supply and protection problems that are
characteristic of classic karst terranes.
2.1.2 Karst Windows
A karst window is a landform that has features of both springs and
sinkholes. It is a depression with a stream flowing across its floor: it may
be an unroofed part of a cave. Thrailkill (1985, p. 39) adds the criteria
that karst windows are deep sinkholes in which major subsurface flow surfaces
and that the streamflow is near the level of major subsurface flow.
Most karst-window streams issue as a spring at one end of the sinkhole,
flow across its floor, and sink into the subsurface through a swallet. The
length of the surface flow varies from what may appear to be a pool in the
bottom of a sinkhole to a stream several hundred feet long (Thrailkill, 1985,
p. 39). As with a sinkhole that drains to the subsurface, the karst window
may flood and overflow its depression if the openings draining to the
subsurface become blocked or if the subsurface conduits receiving that
drainage are filled. Karst windows are hydrologically significant because the
exposed streams provide a direct path to the subsurface for any contaminant
deposited in or near the sinkhole. Also, they serve as convenient ground-
water sampling and monitoring points.
13
-------
2.1.3 Losing, Sinking, Gaining, and Underground Streams, and Blind Valleys
Underground streams are common in karst terranes and, in contrast to
ground water in granular aquifers, exhibit many characteristics of surface
streams. Losing streams have streambeds above the water table and recharge
the underlying karst aquifers. Losing streams may lose all or part of their
flow to the ground-water system at points or intervals along their course
through openings in bedrock that intersect the streambed or as seepage through
alluvium overlying bedrock. Losses of flow can also occur through well
developed swallets in the streambed that occur singly or as a group of
sinkpoints developed over a particular part of the aquifer. Streams may gain
flow in one reach but lose in another reach depending on local geologic and
hydrologic conditions. A losing stream may be perennial or intermittent and
may either gain or lose at different times of the year depending on the
seasonal fluctuations of the water table. Although difficult to quantify,
recharge to the aquifer by losing streams may constitute a major portion of
the total ground-water recharge in some areas.
Some sink points in streams in karst terrane behave in a way that are
unique to streams with major connections to the karst ground-water system.
During time of flood, they may discharge ground water rather than receive
water from the stream. This happens because the conduit or cave system fed by
the stream also receives water from other underground feeders, and during
flood events the total flow underground cannot pass through conduit
constrictions and backs up to the point where it issues from a former
sinkpoint. The French word "estavelle" is the commonly used term for such
points of reversing flow (Jennings, 1985 p. 46). The term is also applied to
sinkholes that alternately receive or discharge water and are not necessarily
located in a streambed (Milanovic, 1981 p. 102). Estavelles may serve a dual
purpose for the water manager; as a sinkpoint for potential contaminants or
dye injection point during periods of low flow and a water-quality monitoring
or dye-recovery point during periods of high flow.
The authors distinguish losing streams from sinking streams on the basis
of the quantity of streamflow that drains underground and the presence of a
swallet through which water drains underground. Losing streams are those that
lose part or all of their flow by seepage. In contrast, sinking streams are
those that terminate and usually drain underground through one or more well
developed swallets, usually at the end of a valley.
A valley that ends suddenly at the point where its stream disappears
underground is termed a blind valley (Monroe, 1970). Some blind valleys have
no present-day streams and are suggestive of an earlier stage of
karstification. Thrailkill (1985, p. 40) uses the term "paleovalleys" to
describe valleys that contain no active surface stream channel. Paleovalleys
usually contain a series of sinkholes in their bottoms, and apparently formed
when their surface streams were diverted underground at several points.
Eventually all stream flow was underground except possibly during high
discharge events.
Relatively large sinking streams are commonly called "lost rivers"
because they drain underground through caves of relatively large size and
issue from caves as relatively large springs. A well known lost river occurs
14
-------
in Bowling Green, Kentucky where flow in a major conduit surfaces at the Lost
River Blue Hole, flows on the surface for about 400 feet, and sinks into the
large entrance to Lost River Cave. The underground river flows beneath the
City of Bowling Green, eventually resurfacing at the Lost River Rise, a
straight-line distance of about 2.8 miles. From this point, the river flows
on the surface to the Barren River, the major base level stream of the area
(Crawford, 1981).
The subsurface course of a lost river is described as an underground
stream. A cave stream is another example of an underground stream. Flow
patterns of underground streams do not always obey topographic drainage
patterns as do surface streams. Using qualitative dye traces in the Bowling
Green, Kentucky, area, Able (1986, p. 37) demonstrated that two adjacent
valleys with separate surface streams, transferred ground water beneath their
topographic drainage divide through underground streams.
Because water in losing, sinking, and underground streams is subject to
contamination from the surface, the contaminants may be transported directly
into the ground-water system. Thus, it is important to identify these
features in areas where karst aquifers are the source of public water
supplies.
Gaining streams occur where the streambed descends to an altitude low
enough to intersect the zone of saturation. Such streams may be gaining from
headwaters to their mouth or for only a short distance. Because stream flow
in a gaining stream is mainly ground-water discharge, the water quality of the
stream can be affected. Thus identification of the area draining to gaining
streams is needed, especially where gaining streams are used for public water
supplies.
2.1.4 Karst Springs
A karst spring is a natural point of discharge from a karst ground-water
aquifer. Discharge from a karst spring may issue from openings ranging in
size from less than an inch to many feet in diameter. Water may either flow
out under gravity or rise under pressure to form seeps or surface streams.
The discharge openings may be visible, or invisible where covered by
unconsolidated material or submerged below the surface of a lake or stream.
Different discharge points within the same system can be active during
different flow conditions. Some of the more common types of karst springs are
shown in figure 3.
A karst spring may occur at local or regional base levels or at a point
where the land surface intersects the water table or water-bearing cavities.
Underlying impervious bedrock can cause springs at the interface of the karst
aquifer and the bedrock. Springs can occur either in valley bottoms or at
sharp breaks in slope. Karst springs may occur at any point where impermeable
rock or structural features, such as faults, impedes ground-water flow and
thus restricts the formation of conduits in the soluble bedrock. Karst
springs can occur high, along valley sides, without obvious topographic or
geologic cause. This may be caused by rapid down cutting of the main valley
which has exceeded the downward development of solutionally enlarged openings
sufficient to lower the karst spring outlets to the local base level
(Jennings, 1985 p. 50).
15
-------
/.I.I.I
./Ill
CONDUIT SPRING
Flood overflow
spring
Normal flow
spring
™ I I ' - '
Z£
FLOOD OVERFLOW SPRING
ill i \\^=-
| [Bedrock
III I—I—T\r^^
ii i
RISE PIT SPRING
(modified from Jennings, 1985, figure 15, p. 47)
Figure 3.—Common types of karst springs.
16
-------
A karst spring is generally the principal discharge point for a karst
ground-water basin. A particular karst spring may discharge from a master
conduit and represent drainage from a system of integrated conduits that
converge to that conduit (convergent flow) or be one of a group of springs
(terminal distributaries) that function much like the delta of a surface
stream in which ground water is dispersed to many springs from a tributary
system of conduits (distributary flow) (Quinlan, Saunders, and Ewers, 1978).
Quinlan and Ewers (1985) state that terminal distributaries and their related
springs are a common feature of carbonate aquifers. Quinlan and Rowe (1977)
discuss a distributary system of Hidden River Cave, Kentucky in which 46
springs at 16 locations in a 5-mile reach of river, discharge water to the
Green River.
Because karst springs are usually the principal discharge points for
drainage from an entire ground-water basin, they represent the most likely and
convenient points to monitor the recovery of dye injected upgradient from the
springs. Thus, it is mandatory that all springs that may drain a particular
injection point, be located and monitored during dye tracing.
2.2 Occurrence and Movement of Ground Water
Ground water in karst terrane occurs in both unconsolidated sediment and
bedrock that may comprise a complex interrelated aquifer system. The nature
of ground-water movement in karst terrane varies considerably from place to
place depending on the nature of the aquifer system.
Ground water in the unconsolidated surficial material overlying bedrock
is generally thought to occur in intergranular (primary) openings and thus, to
behave according to the theories of ground-water movement in porous media.
Although this is generally true, there is evidence that concentrated flow also
occurs in enlarged openings (macropores) in the unconsolidated material (Sevan
and Germann, 1982, p. 1311). Quinlan and Aley (1987) state that concentrated
flow in macropores (root channels, cracks or fissures, animal burrows, and
textural transitions) is commonly several orders of magnitude more rapid than
in the adjacent unconsolidated sediment. Mull, Smoot, and Liebermann (1988)
report the presence of pipe-like openings and numerous conduits in the
unconsolidated material overlying bedrock in the Elizabethtown area, Kentucky.
Many of these conduits contained water, in some cases, as much as 40 feet
above bedrock. This indicates that, in places, there is a conduit system in
the unconsolidated material that can collect and funnel water to water bearing
conduits in the underlying bedrock. Because of this conduit system, potential
ground-water contaminants placed on the surface or in depressions such as
sinkholes, can enter the ground-water flow system fairly quickly, despite the
fact that the thickness of unconsolidated, surficial material may be 50 feet
or more.
The occurrence and movement of ground water in bedrock underlying karst
terrane is quite different from that underlying non-karst terrane, primarily
because of the presence of conduits that permit relatively rapid transmission
of ground water. The predominant rock types in most karst terrane are
limestone and dolomite. These rocks may be relatively impervious except where
fractures and bedding planes have been enlarged by circulating ground water.
The circulating water dissolves carbonate bedrock and enlarges the openings.
17
-------
The enlarged openings may be vertical or horizontal and range in size from a
fraction of an inch to tens of feet, such as at Mammoth Cave, Kentucky. As
mentioned earlier, ground-water flow in a mature karst aquifer is primarily
through conduits and can be described by pipe and channel flow equations
(Gale, 1984). Components of ground-water flow in a mature karst aquifer are
shown by the generalized block diagram (Gunn, 1985) in figure 4.
Ground water in karst terrane, as in other terranes, moves in response to
hydraulic gradients from points of recharge to points of discharge. The
horizontal gradient of the ground-water surface, the general shape of the
water table, and the general direction of movement can be determined from a
water-level contour or potentiometric map. The contours are based on the
altitude of the water level in wells, springs, and streams. The general
direction of ground-water movement can be estimated by drawing flow lines
perpendicular to the water-level contours. Results from dye traces can also
be used to confirm the direction of ground-water movement shown by the water-
level contour map.
The rate of ground-water movement is also important to the understanding
and solution of many ground-water problems in karst terrane, especially to
those related to contamination of the ground-water system in karst terranes.
Results of dye tracing can define the rate of ground-water movement and the
fact that the rate is not constant, but varies with hydrologic conditions.
Karst aquifers have been classified into three types based on the nature
of ground-water flow (White, 1969, 1988, p. 171). Each type contains
subvarieties and variations. The classification system applies mainly to
karst regions with low and gentle relief and is intended to establish some
useful criteria for determining the nature of the underground-flow system from
observable geologic characteristics. The following discussion of the three
types of karst aquifers is adapted from Fetter (1980), Milanovic (1981), and
White (1969).
(1) Diffuse-flow karst aquifers. Diffuse-flow aquifers commonly develop
in dolomitic rocks or shaly limestones where the solutional activity
of moving ground water has been retarded by lithologic factors.
Water movement is laminar, along joints and bedding planes that have
been only slightly enlarged by solution. Ground-water flow is
usually not concentrated in certain zones in the aquifer and, if
present, caves are limited in size and not interconnected. Discharge
from the aquifer occurs through many small springs and seeps. The
water table is well defined and can rise well above the regional base
level. Typical karst landforms are absent or poorly developed. An
example of a diffuse-flow aquifer is the dolomite aquifer in Silurian
rocks of the Dupage County-Chicago Region of Illinois (Zeizel and
others, 1962).
(2) Free-flow karst aquifers. Free-flow or conduit-flow aquifers are
developed in thick and massive soluble rocks where ground-water flow
is concentrated in a well-defined and integrated system of enlarged
conduits which behave hydraulically as a system of pipes. Flow
velocities are similar to surface streams and are often turbulent.
The regional discharge may occur through a single large spring.
Because of the rapid drainage, the water table can be virtually flat
18
-------
U1
ao
z
O
0.
X
IU
(II
U-l
•H
3
cr
re
•o
ai
OJ
M
3
?
TJ
C
3
O
00
c
01
c
o
a.
o
u
I
I
oo
•H
= £ = 3 p
o u. 0 to I
19
-------
for miles, with only a slight elevation above the regional base
level. Water levels in the conduit network and spring discharge
respond rapidly to recharge events, and during periods of heavy
precipitation the spring hydrograph may resemble the flood peak of a
surface stream. Examples of this type of aquifer may be found in the
Missouri and Arkansas Ozarks, southern Indiana, and central Kentucky.
(3) Confined-flow karst aquifers. Confined-flow karst aquifers contain
beds of low hydraulic conductivity, caused by stratigraphic or
structural conditions, that control the rate and direction of ground-
water flow. Flow in karst aquifers bounded by such confining beds
can occur at great depths in solution openings that are much deeper
than in free-flow aquifers. The flow is not concentrated in master
conduits but occurs in a system of relatively dense solution joints
that form characteristic network caves. The Floridan aquifer
(Stringfeld, 1966) is an example of a confined-flow karst aquifer.
In summary, two types of ground-water flow generally occur in karst
terrane, diffuse (slow, laminar flow) and conduit (rapid, turbulent flow).
Diffuse flow occurs predominantly in primary openings, whereas conduit flow
occurs mostly in secondary openings that range from a fraction of an inch to
tens of feet. Water that moves through conduits can enter the subsurface
through discrete points, such as sinkholes or sinking streams. Flow in a
diffuse*flow aquifer can be concentrated into discrete conduits in the
subsurface. The hydrologic significance of conduit flow is the rapid
transmission of water through the aquifer. For further discussions on the
classification and subdivision of karst aquifers see Thrailkill (1986), Smart
and Hobbs (1986), and White (1977, 1988).
2.3 Vulnerability of Karst Aquifers to Contamination
As previously stated, ground water in karst terrane can be extremely
vulnerable to contamination. This vulnerability varies according to the
nature of the contaminant, karst features, occurrence of ground water in karst
terrane, the degree of contact of infiltrating water with the soil zone, and
the opportunity for transported pollutants to enter the aquifer system.
In general, water quality can be considered in terms of physical,
chemical, and biological characteristics. Contaminants of concern are
generally of a chemical or biological nature. The chemical contaminants that
may be transported in a karst ground-water flow system can be classified as
inorganic or organic, dissolved or suspended (particulate), or volatile.
Contaminants may originate from a variety of land-use activities, such as
agriculture, mining, construction, urban stormwater and spills, and waste and
wastewater management practices, such as municipal or industrial wastewater
discharges, septic tank leachate, and landfill leachate. Crawford (1982) has
identified many of these contaminants and contaminant sources in the Lost
River drainage basin near Bowling Green, Kentucky. Leaking underground
storage tanks and industrial discharges have entered the cave system beneath
Bowling Green, causing explosive fumes to rise into homes and other buildings.
Dissolved contaminants in conduit-flow aquifers can be readily
transported under all types of flow conditions. Examples include many
industrial organic compounds, herbicides, nutrients, and trace metals.
20
-------
Constituents associated with suspended matter generally require more energy
(generated by high velocities and turbulence) for transport. The energy
required for transport is related to the density, size, and shape of the
suspended particles. Contaminants could include sediment with attached
insecticides, nutrients, and heavy metals. Contaminants associated with
suspended material can also be mechanically filtered out in diffuse-flow
conditions by the small pore openings in the aquifer matrix. In contrast, in
the large, well-developed solutional openings in some karst aquifers, it is
common for large-sized sediment and other particulate material with associated
contaminants to be readily transported. These contaminants may enter the
aquifer from a sinking stream or sinkhole, move rapidly through the conduit
system, and exit at a spring or well.
Biological contaminants such as viruses, bacteria, and other
microorganisms, as well as larger organisms, may be readily transported in a
karst aquifer in much the same manner as chemical contaminants. They may or
may not be associated with other suspended matter. The larger organisms and
organism/suspended matter aggregates generally require large openings and high
velocities for ground-water transport similar to those that typify ground-
water movement in karst terranes.
Almost all water that reaches a ground-water flow system percolates
through a soil zone. The soil zone can significantly enhance the quality of
percolating water by filtration, various physical and chemical reactions
(solution, precipitation, oxidation-reduction, ion exchange, adsorption-
desorption, and acid-base reactions), microbiological transformation, and
other physical, chemical, or biological processes. However, in karst terrane,
the infiltrating water may have little or no contact with the soil zone and,
thus, limited opportunity for quality enhancement before entering the ground-
water system.
Although sinkholes with plugged drains may be hydraulically connected to
the aquifer system, the potential for pollution is generally less than from
those with open drains. Because water infiltrates through sediments in the
bottom of the sinkhole, some water-quality enhancement can occur.
As described previously in this manual, the development of solutionally
enlarged openings in soluble bedrock below the soil zone and other
unconsolidated material is generally the first phase of sinkhole development.
The growth and interconnection of these openings provide the subsurface
drainage system for the infiltration and transport of unconsolidated material
from the surface and results in sinkhole formation. It is this same network
of subsurface openings that provides avenues for rapid movement of
contaminants to ground water. Because ground water in karst aquifers moves
mainly through open conduits, it typically moves much faster than in other
aquifers and may be on the order of miles per day. Therefore, any contaminant
entering the ground-water system, such as one carried by surface runoff, has
the potential for rapid transport and distribution within the system.
21
-------
3. DYE TRACING CONCEPTS, MATERIALS, AND TECHNIQUES
3.1 Introduction
Given the unique characteristics of karst ground-water systems, the
management of a water supply using such an aquifer is difficult. The water-
supply manager needs to fully understand the aquifer's geographic extent and
hydraulic characteristics in order that a safe, reliable supply of water can
be provided to the consumer. Specialized, site-specific predictive techniques
based on qualitative and quantitative dye tracing can be used to develop
appropriate plans to react to events and actions, which may jeopardize the
safety of the water source.
The practice of tracing ground-water flow by adding distinctive
substances to water draining underground and monitoring the downgradient
resurgence of that water has long been a useful tool for hydrologic
investigations. Given the hydrology of karst terranes, dye tracing is
generally the most practical and satisfactory method to provide information
for the management and protection of karst aquifers. Information from
properly conducted dye traces can identify point-to-point connections between
discrete recharge areas and discharge points such as springs or wells. In
addition, analysis of dye-recovery data can provide critical management
information, such as time of travel, peak concentration, and persistence of
potential contaminants.
One of the earliest reported water-tracing experiments took place almost
2,000 years ago when chaff was thrown into Ram Crater Lake in order to
identify springs at the headwaters of the Jordan River (Mazor, 1976).
Throughout the development of tracer techniques, the materials used for
tracers have been limited only by the creativity of the experimenters and in
recent years have included such diverse items as computer-card chips, dog
i biscuits, plastic pellets, oranges, and food coloring. Accidental or
intentional dumping of distinctive substances has often served to identify
point-to-point connections between various sinkholes and springs or wells.
For example, Quinlan and Rowe (1977, p. 9) reported the dumping of an
estimated 340 tons of whey into a sinkhole that contaminated water from
public-supply wells at Smiths Grove, Kentucky, about 5 miles away. Also, Aley
and Fletcher (1976) reported that water from a sinkhole was traced to a nearby
high school in Tennessee when revenue agents dumped 2,000 gallons of illegal
whiskey into the sinkhole. Mull, Smoot, and Liebermann (1988), reported that
drainage from a salt-storage yard that flows into a cave near Elizabethtown,
Kentucky was traced to an unused spring about 1.4 miles away.
In all cases mentioned above, the primary objective of water tracing was
successfully, but perhaps unintentionally, accomplished. The subsurface
connection between a specific recharge point and a discharge point located
some distance away was identified. Dye tracing to identify point-to-point
connections is generally known as qualitative dye tracing and may use visual
observation or a passive detector and visual observation to detect dye at a
discharge point. Time of travel and flow velocity may generally be determined
through qualitative dye tracing. However, if more precise hydrologic
information, such as time of travel and ground-water flow velocity, or
potential contaminant transport characteristics, such as persistence,
22
-------
dispersion rates, and concentration are needed, then quantitative dye tracing
using discharge measurements and precise measurements of the dye
concentrations in the water are required.
Many reports give detailed information on the theory and techniques of
qualitative and quantitative dye tracing. In particular, the manuals by Aley
and Fletcher (1976) and Quinlan (1987a), provide detailed discussions of
criteria for dye selection and qualitative dye tracing techniques. Smart and
Laidlaw (1977) provide a primary reference on properties of dyes used for
tracing. Basic information on quantitative aspects of dye tracing of surface
waters are presented by Wilson, Cobb, and Kilpatrick (1986), Hubbard and
others (1982) , Kilpatrick and Cobb (1985). Techniques for tracing ground
waters are discussed by Jones (1984). Dye-tracing techniques that may be
described as semi-quantitative are discussed by Duley (1986) and Spangler,
Byrd, and Thrailkill (1984).
The distinguishing characteristic of semi-quantitative tracing is
instrumental analysis of various elutants or detectors to confirm the recovery
of a particular dye used for tracing. The use of filter fluorometers or
scanning spectrofluorometers can identify dye concentrations well below the
limits of visual detection. Semi-quantitative tracing is not discussed in
this manual because the equipment and procedures used in this method are
similar to those of qualitative and quantitative tracing which are described
in detail in the following sections.
Investigators need to contact appropriate state water-supply and ground-
water protection program and health department offices concerning state
policies on dye tracing and to coordinate and inform these offices of tracing
activities. This will serve to determine if other dye traces have been
conducted in the same area. Discussion with other investigators may prevent
wasted effort and expense due to confusing or erroneous results caused by
simultaneous use of the same tracer or the recovery of dye from an otherwise
unknown tracer test.
3.1.1 Dye Characteristics and Nomenclature
Perhaps the most widely used tracers in karst terrane are fluorescent
dyes. These dyes are commonly used because they are readily available, are
generally the most practical and convenient tracers, and they all, to some
degree, are absorbed on activated coconut charcoal or unbleached cotton.
Fluorescent dyes are generally superior to non-fluorescent dyes because they
can be detected at concentrations ranging from one to three orders of
magnitude less than those required for visual detection of non-fluorescent
dyes. Thus, traces with fluorescent dyes usually can be completed without the
aesthetically unpleasant probability of discoloring a private or public water
supply.
Because tracing karst ground-water flow frequently involves either
private or public water supplies, the problem of toxicity of the tracers must
be considered. There is a relatively large amount of information available on
the toxicity of the most commonly used tracers. Smart (1984) presents a
review of the toxicity of 12 fluorescent dyes used for water tracing that
includes the tracers discussed in this manual, namely, rhodamine WT, optical
brighteners, Direct Yellow 96, and fluorescein. As reported by Smart, three
23
-------
dyes present minimal carcinogenic and mutagenic hazard: Tlnopal CBS-X
(brightener), fluorescein, and rhodamlne WT. Douglas and others (1983)
reported that rhodamlne WT Is non-carcinogenic but found a small but
statistically significant dose-related, mutagenic effect. However, they
concluded that the use of rhodamlne WT does not appear to represent a major
genotoxlc hazard. Stelnhelmer and Johnson (1986) have shown that, under
customary dye-study practices In surface waters, the possible formation of
carcinogenic nitrosamines from the use of rhodamine WT should not constitute
an environmental hazard. In ground water which may be enriched with nitrite,
nitrosamines could form, but high-nitrite concentrations in ground water are
uncommon (Hem, 1985, p. 124-126). Therefore, the possible formation of
nitrosamines from rhodamine WT is not likely to be a problem. Quinlan (1987b)
points out that numerous investigators (Anliker and Muller, 1975; Lyman and
others, 1975; Ganz and others, 1975; Burg and others, 1977; and Smart, 1984)
have found optical brighteners to be non-toxic, non-carcinogenic, and non-
mutagenic and therefore safe for use as a tracer. It should be pointed out;
that one dye, rhodamine B, that was earlier approved by the EPA for use as a
tracer in potable water (Cotruvo, 1980) is now not recommended because
impurities within it are known to be carcinogenic and possibly mutagenic
(Smart, 1984).
Because most dyes are available under many commercial names, the Colour
Index (CI) Generic Name or Constitution Number of a dye needs to be used to
avoid confusion and the possible use of inappropriate dyes. Positive dye
identification is especially needed if dye properties and results of dye
tracing are being discussed. The standard industrial reference to dyes is the
Colour Index (SDC & AATC, 1971-1982) which describes about 38,000 dyes and
pigments. Concise guides to dye nomenclature can be found in Giles (1974),
Abrahart (1968), and especially the paper by Smart and Laidlaw (1977). In
addition, useful discussions on dye nomenclature and the problems arising from
multiplicity of names for the same dye are presented by Quinlan (1987b and
1986b) and Quinlan and Smart (1977). The problems of dye nomenclature are
reduced if only one or two dyes are used. However, knowledge of dye
nomenclature and classification is essential in order to select multiple dyes
with similar characteristics or obtain a replacement if the original dye is no
longer available.
3.1.2 Fluorescent Tracers
Although many different fluorescent dyes are used as ground-water
tracers, present usage is centered on four: rhodamine WT (CI Acid Red 388),
fluorescein (CI Acid Yellow 73), optical brighteners, and Direct Yellow 96.
In general, rhodamine WT is not used for qualitative tracing because of the
difficulty of visually distinguishing the pink color of the dye from that of
other organic compounds that can easily be sorbed by activated coconut
charcoal. Characteristics of the fluorescent dyes discussed in this manual
are presented in table 1.
Although there are two dyes called fluorescein, and both are labeled CI
Acid Yellow 73, only one is water-soluble. The water soluble sodium salt of
fluorescein (C20H12°5^ is sodium fluorescein (C20H.005Na2) and is commonly
used for water tracing. The European name for sodium fluorescein is uranine.
However, the conventional usage in the United States and England is simply
fluorescein, which is followed in this manual.
24
-------
Table 1.—Characteristics of commonly used tracer dyes
Tracer
and
color
Index
Passive Detection
detector (elutHant)
Detectable
concentrations
Advantages
Disadvantages
Remarks
Sodium
fluoresceln
CI Add
Yellow 73
Activated
coconut
charcoal
6-14 mesh
Ethyl alcohol
and 5 percent
KOH or fluo-
rometHc
analysis of
water samples.
0.1 micro-
grams per
liter,
dependent on
background
levels and
fluctuation.
1. Does not
require con-
stant monitor-
Ing or
fluorotnetrlc
analysis.
2. Inexpensive.
1. Dye 1s photo-
chemical ly unstable.
2. Moderate
sorptlon on clay.
3. pH-sens1t1ve.
Most common dye
used In karst
terrane.
Rhodanrlne WT
CI Acid
Red 388
Activated
coconut
charcoal
6-14 mesh
1-Propanol and
NH4OH, or
ethyl alcohol,
5 percent KOH
and water, or
fluorometrlc
analysis of
water samples.
0.01 micro-
grams per
liter,
with fluorom-
eter.
1. Dye Is
photochemical ly
stable.
2. Dye may be
used 1n low
pH waters.
1. May require
fluorometrlc
analysis.
2. Moderate
clay sorptlon.
3. Difficult to
distinguish during
qualitative tracing.
Not recommended
for qualitative
tracing. Ideal
for quantitative
tracing.
Optical
bHghteners
CI Fluores-
cent
brlghtener
28
Unbleached
cotton
Visual exami-
nation of
detectors
under UV
light or
fluorometrlc
analysis.
Dependent on
background
levels, but
generally at
least 0.1
mlcrograms
per liter.
1. Inexpensive.
2. No coloring
of water.
1. Background
readings may
limit use.
2. Adsorbed onto
some organlcs.
May be used
simultaneously
with a green and
orange dye that
sorb onto
activated coconut
charcoal.
Direct Yellow
DY 96
Unbleached
cotton
Visual exami-
nation of
detectors
under UV
light or
fluorometrlc
analysis.
1.0 micro-
grams per
liter,
on cotton.
1. No natural
background.
2. Good
stability and
low sorptlon.
3. No coloring
of water 1n
usual concen-
trations.
1. Moderate cost.
2. Sensitive to pH.
Has been used
extensively 1n
Kentucky.
Table modified from Jones, 1984.
*SDC and AATCC Color Index.
Very dilute dye solutions are concentrated upon the detector over a period of time.
Use of brand/firm/trade names 1n this report Is for Identification purposes only and does not constitute endorsement by
the U.S. Geological Survey or the U.S. Environmental Protection Agency.
25
-------
Fluorescein has been widely used for ground-water tracing in karst
terrane since the late 1800's (Aley and Fletcher, 1976). It is presently one
of the most widely used water-tracers in karst areas in the United States
(Quinlan, 1986a) because of its safety, availability, and its ready adsorption
onto activated coconut charcoal. Fluorescein is a reddish-brown powder that
turns vivid yellow-green in water, is photochemically unstable, and loses
fluorescence in water with pH less than 5.5.
Optical brighteners and CI Direct Yellow 96 are suitable for dye tracing
because they are detectable in low concentrations, are non-toxic, have low
affinity for adsorption onto clays, and are readily adsorbed on undyed
surgical cotton used to recover dye in passive detectors. Optical brighteners
are, however, widely used in laundry detergents and soaps for enhancing fabric
colors and are thus a common constituent of domestic wastewater. For this
reason, the effectiveness of optical brighteners for water tracing may be
limited in areas where effluent from septic systems is present. Because, the
presence of optical brighteners indicates the presence of laundry wastewaters,
they can be used as indicators of sewage in various ground-water resurgences.
If relatively high levels of optical brighteners are present as background,
the choice and amount of brightener used for tracing can be affected. The
background levels of optical brighteners can easily be determined by placing
undyed cotton detectors in ground-water resurgences and testing for
fluorescence before tracer dyes are injected. This background check needs to
be done before the use of any dye.
CI Direct Yellow 96, because of its distinctive yellow fluorescence, is
ideal for water tracing in areas where background levels of optical
brighteners are present. Unlike optical brighteners, CI Direct Yellow 96 does
not undergo significant photochemical decay. CI Direct Yellow 96 is a powder
that needs to be mixed thoroughly with water before injection into the ground-
water system and, like optical brighteners, is readily adsorbed onto undyed
cotton detectors. Optical brighteners and CI Direct Yellow 96 have been used
with a great deal of success for water tracing in several karst areas. For
example, Quinlan and Rowe (1977), Thrailkill and others (1982), and Spangler,
Bird, and Thrailkill (1984) have reported successful tracing with optical
brighteners and CI Direct Yellow 96 in the Mammoth Cave and Inner Blue Grass
regions in Kentucky.
3.2 Qualitative Dye Tracing
3.2.1 Introduction
Although one primary purpose of this manual is to present procedures and
techniques for quantitative dye tracing, the results of qualitative dye
tracing are usually needed to design and implement the more labor intensive
quantitative phase of dye tracing. In particular, qualitative dye tracing can
delineate the boundaries of a ground-water basin and identify point-to-point
connections between input and recovery points allowing special attention to be
given to those points with the highest potential for contaminating the ground-
water system. In some cases, a qualitative dye trace may provide adequate
information to satisfy the needs of the water-supply manager. For example, if
a particular spring, used as a water-supply source, is shown to be the
resurgence of drainage from sinkholes, compatible land-use practices around
26
-------
these sinkholes may adequately protect the water supply. If, however,
qualitative dye tracing shows inputs from several widespread source areas,
then land-use controls alone may not be effective and quantitative dye trace
data and analysis may be required to more effectively manage and protect the
ground-water resource.
Qualitative dye tracing involves the tagging of a discrete sample of
water with an appropriate tracer and monitoring the arrival of that tracer-
laden water at various ground-water resurgences. The arrival of the dye may
be observed visually or the dye may be recovered by passive detectors and
identified by various chemical or instrumental analyses that reveal its
presence.
The following discussion describes the procedures and equipment used for
qualitative dye tracing by many investigators, but is based, primarily, on
those used in the Mammoth Cave area by Quinlan (1981). As Quinlan (1987a)
states, the procedures and techniques summarized here are not the only
successful techniques for dye tracing in karst terrane, but he and many
investigators, including the authors, have found that these techniques
consistently give reliable results.
3.2.2 Selecting Dye for Injection
Selecting a particular dye for water tracing in karst terrane is based on
factors such as quality of the water draining underground, nature of the
background concentrations of potential tracers such as optical brighteners,
character of the injection point, availability and cost of a particular dye,
and the availability or complexity of equipment needed to detect the dye.
Dye-selection criteria and the advantages and disadvantages of the four most
commonly used dyes are discussed below and summarized in table 1. Although
rhodamine WT is not generally considered for qualitative dye tracing because
of the difficulty of detection without the use of a spectrofluorometer or a
filter fluorometer, rhodamine WT is included because it can be used for
qualitative dye tracing under conditions of availability, convenience, or
necessity.
Although not generally a problem in karst terrane, low pH water can
seriously attenuate the fluorescence of fluorescein and Direct Yellow 96.
The fluorescence of optical brighteners is much less affected by pH, thus
these tracers can be used if the pH of the water being traced is 5.0 or less.
Quinlan (1987a) states that all tracer dyes tend to react with the
environment through which they flow and that all four tracers discussed here
are sorbed on to clays, although optical brighteners and CI Direct Yellow 96
are sorbed to a much lesser extent than fluorescein or rhodamine WT. He
further states, that sorption usually precludes the use of fluorescein or
rhodamine WT in granular aquifers. Thus, loss due to adsorption can influence
the selection or quantity of a dye if the dye must drain through
unconsolidated soil and residuum before reaching the bedrock-conduit system.
In this case, an appropriate increase in the quantity of dye injected may be
necessary.
27
-------
The level of background fluorescence can effect the selection of a
particular tracer. For example, if injection water contains a relatively high
component of domestic sewage, fluorescence from optical brighteners, a common
constituent of home-laundry products, may preclude the use of optical
brighteners as tracers. CI Direct Yellow 96 can be used where there is a high
background level of optical brighteners. Fluorescein may also be present
because of its use as a coloring agent in a number of home products such as
shampoos, bathroom cleaners, and antifreeze. Either optical brighteners or
fluorescein can be used if the background levels of that particular dye are
determined to be sufficiently low and stable before the trace is begun. In
fact, determination of background levels needs to be adopted as the first step
in most traces because this provides a standard for comparison of the dye
recovered.
Because some fluorescent dyes decay upon exposure to ultraviolet light,
these dyes are less than ideal if the water being traced repeatedly sinks and
rises and flows along the surface. Smart and Laidlaw (1977) report that the
photochemical decay rates are very high for fluorescein, which rapidly loses
its fluorescence under bright sunlight conditions. Quinlan (1987a) also
reports that optical brighteners and fluorescein are very susceptible to
pho-fcochemical decay, especially when in low concentrations, but this is only a
problem if the detectors for recovering the dye are placed in direct sunlight.
If the loss of fluorescein or optical brighteners due to photochemical decay
is possible, CI Direct Yellow 96 may be used because it does not undergo any
significant photochemical decay (Quinlan, 1987a).
An evaluation of cost and availability of a particular dye involves
factors that can be unique to a particular user. Therefore, these factors are
not discussed here. Also, the various methods and equipment required for the
recovery of the tracers commonly used for qualitative tracing are discussed in
detail in a later section and are listed in Table 1.
3.2.3 Selecting Quantity of Dye for Injection
Having selected a particular fluorescent dye, the next step is to select
the optimum quantity of dye for injection. Except in unusual circumstances,
where a highly visual appearance of the dye is desired, the quantity of dye to
be injected is selected in order to provide a detectable amount of dye at the
recovery point, but remain below visible levels. The quantity of fluorescent
dye used for each injection is based generally, on estimated flow conditions
and the straight-line distance of the trace. To some extent, the quantity of
dye will vary depending on the nature of the injection point. For example, if
the dye can be added to water draining directly into an open swallet, less dye
will be required than if the dye must infiltrate through soil and
unconsolidated material in the bottom of a plugged sinkhole. Dye studies
conducted by the U.S. Geological Survey usually limit the maximum
concentration of fluorescent dye at a water-user withdrawal point to 0.01 mg/L
(Hubbard and others, 1982).
Selecting the optimum amount of dye for qualitative tracing is, to some
extent, a matter of experience. The amounts may be adjusted depending on the
initial results. Quinlan (1987a) suggests several rules of thumb as starting
points for tracing under average conditions in the Central Kentucky karst and
using various passive detectors for dye recovery. For qualitative traces
28
-------
under moderate-flow conditions to the average spring, he suggests one pound of
fluorescein per straight-line mile of trace, up to a maximum of five pounds.
For Direct Yellow 96 and optical brighteners in powder form, two pounds per
straight-line mile of trace, and for some optical brighteners in liquid form,
one gallon per straight-line mile, up to a maximum of four gallons is
recommended. For initial traces in the Elizabethtown area, Kentucky (Mull,
Smoot, and Liebermann, 1988), the above procedure was used for
estimating the quantity of fluorescein dye for injection with satisfactory
results. However, as knowledge of the flow system developed, the amount of
fluorescein per each injection was reduced to about 1/2 pound per straight-
line mile of trace. This amount proved adequate and likely could have been
reduced further, once the conduit flow-path between the injection and recovery
points was established.
Aley and Fletcher (1976) present a nomograph for selecting the quantity
of fluorescein to be injected if the ground-water flow system under
investigation consists primarily of conduit flow in solution channels or
fracture zones. Their equation, shown here in the original units, is:
(eq. 1)
where W. - weight of fluorescein dye in kg to be injected;
D - straight-line distance in km from injection point
to recovery point;
Q - discharge at the resurgence in cubic meters per
second; and
V - estimated velocity of ground-water flow in meters
per hour.
For this equation, the velocity of ground-water flow is the most
difficult to estimate because it depends on the nature of the conduits and the
hydrologic conditions at the time of the trace. Therefore, initial dye
tracing needs to be attempted under medium base flow conditions for a study
area. Subsequent traces under extreme flow conditions are needed to better
define the flow variations and possible changes in flow paths under different
flow conditions.
3.2.4 Dye-Handling Procedures
In order to simplify dye injection when using powder such as fluorescein,
especially during windy conditions, the powder can be mixed with water before
going to the field. An approximate ratio of 0.5 pound per gallon is
suggested. The dye solution is poured into water draining directly into a
swallet in order to lessen dye loss due to photochemical decay or absorption
by organic debris at the surface. While mixing and injecting the dye, extreme
care needs to be used to avoid contaminating clothing or the area around the
injection point. The need for care in handling the dye before and during the
injection can not be overstressed because of the possibility of contaminating
the dye detectors or the area around the injection site. Such contamination
can lead to a false positive trace and erroneous interpretations. To lessen
the possibility of contamination, long-sleeve rubber or disposable, plastic
gloves needs to be worn during all dye-handling operations. In addition,
refer to other quality-control procedures described in section 3.4.
29
-------
If a trace is required during dry-weather conditions, hauled water can be
used to flush the dye into a particular injection point. Most investigators
suggest the use of two tanks of water for dry weather dye injectir.i.
Successful dry-weather traces have been completed using 1,000 to 1,600 gallons
of water per tank, (Mull, Smoot, and Liebermann, 1988, and Quinlan, 1987a) .
Jones (1984) suggests the use of at least 500 gallons of water for dye
injection and about 100 gallons should be dumped before the dye is added. In
general, the quantity of injected water seems to be less critical than the use
of the two-tank method for injection. The swallet is dosed with water from
the first tank but the dye is not injected until water from the second tank
begins to drain underground. This provides initial wetting of conduit
surfaces followed by a slug of water to flush the dye through the system. The
success of this procedure depends on the length of the trace, the period of
time the recovery site is monitored, and the nature of the conduits in
relation to the quantity of water used for the injection.
3.2.5 Dye -Recovery Equipment and Procedures
In tests using fluorescein, the dye is recovered by passive dye -detectors
consisting of packets of activated coconut charcoal suspended in all suspected
ground-water resurgence points for a particular trace. In addition, detectors
are placed in unlikely resurgences in order to define the background levels of
fluorescence and to show the condition of the dye -recovery material in the
absence of dye.
The dye -detectors, or "bugs," consist of a bag or packet of activated
coconut charcoal attached to a length of wire imbedded in a gum- drop shaped
concrete anchor about 6 inches in diameter and 3 inches high (fig. 5). The
anchor may be made by pouring concrete into plastic bowls lined with plastic
film. A piece of 9 -gage, galvanized wire, about 18 inches in length, is
imbedded in the wet concrete. Some investigators prefer to use two wires, one
for holding the bag of charcoal and the other for holding the swatch of cotton
if optical brighteners are in use (Spangler, Byrd, Thrailkill, 1984). The bag
for holding the activated coconut charcoal is fabricated from a folded 3 by 7
inch piece of aluminum, nylon, or fiberglass screening. Only activated
coconut charcoal (6-14 mesh) is used for absorbing fluorescein. It is
available from several scientific supply houses (table 2). Charcoal intended
for water -treatment processes, aquariums, or home barbecue grills does not
sorb the dye and is not acceptable for use as a detector. Because the coconut
charcoal loses its sorptive ability upon exposure to the atmosphere, it must
be stored in air-tight containers at all times. For this reason, the
preferred procedure is to place the charcoal in the fiberglass packets while
in the field at each site. Some investigators prefer to fabricate fiberglass
or aluminum packets, complete with charcoal, before going to the field. This
is acceptable provided the packets are properly stored and protected from dye
contamination.
In order to maximize exposure to the dye, the gum- drop anchor is placed
near the center of flow of the suspected dye -resurgence point. If flow
velocities are sufficiently high to cause the loss of the anchor or detector,
the anchor should be placed in the main stream but in an area of lower
velocity such as a pool. In cases where the channel is too shallow for the
anchor, a U-shaped wire pin may be driven into the channel bottom to hold the
30
-------
Vinyl-clad copper
wire, twisted to hold <
packet of charcoal
or cotton.
ylon cord
Galvanized
wire, no.9
gage
Concrete base, approximately
6 inches in diameter
(adapted from
, 1987a, figure 1, p. E-B)
Figure 5. — Anchor used to suspend dye detectors (bugs) in springs or streams,
31
-------
J
Table 2.—Sources of materials and equipment for dye tracing
It«0
Cost •• of
8-02-87
Source
Fluoreicein
(CI Acid Yellow 73)
Direct Yellow 98
(Diphonyl Brilliant
FLavine 7GFF)
Rhodanine HI
(CI Acid Red 388)
Phorwite
BBH 768
(CI F.B.A. 28)
Unbleached cotton pads
Swiu Beauty F«di
Activated coconut charcoal.
8-14 mesh
Ultra-violet Imp,
long-wave modal 11-984-8
Filter fluoromater
modal 10, 10-000
laboratory eaia 10-002
cuvette holder 10-303
rbodanine HI
filter kit 10-04
914.30 par pound
73 percent
(powder)
030 per pound
30 percent
(powder)
814.71 per pound
(powder)
20Z solution
$10.83 per pound
(9.92 poundi per
•allon)
93.90 per pound
(powder)
$72.00 per caie
caie of 3840
$18.00 per pound
$176.00
$3,302.00
$ 431.00
8 210.00
$ 314.00
Cbeme entral/Detroit
13395 Huron River Drive
Romului, MI 48174
(313) 941-4800
Pylam Product! Co. Inc.
1001 Stewart Avenue
Garden City, HY 11330
(800) 643-6096
Ciba-Geigy Corp.
Djreituffi ft Chemical Div.
P.O. Box 18300
Oreoniburg, HC 27419
(MO) 334-9481
Crompton ft Khowles
Industrial Product! Division
P.O. Box 33188
Charlotte, RC 28233
(800) 438-4122
Mobay Chemical Corp.
Dyea and Pigment! Division
Mobay Road
Plttiburs, FA 13203
(BOO) 662-2927
U.S. Cotton, Inc.
P.O. Box 387
Saratoga, CA 93071
(408) 378-7732
Fiiher Scientific Co.
341 Creek Road
Cincinatti, OB 43242
(313) 793-3100
Fiiher Scientific Co.
341 Creek Road
Cincinatti, OH 43242
(313) 793-3100
Turner Designs
2247 Old Middlefield Hay
Mountain View, CA 94043-2849
(413) 965-9800
Automatic sampler
model 2700 with
24 1000 tnL (lass bottles
HICad battery
battery charger
A/C converter and
battery charger
vinyl auction line with
itrainer 25ft, x 3/8
in diameter
teflon auction line,
25ft, x 3/8 in diameter
stainless steal strainer
$2.
$
$
$
$
$
$
170.00
195.00
38.00
175.00
64.00
92.50
63.00
Instrumentation Specialtiea
Company
P.O. Box 82531
Lincoln, HE 68501
(800) 228-4373
(402) 474-2233
lisa of brand/firm/trade names in this report is for identification purpose*
only and does not constitute endorsement by the U.S. Geological Survey or the
U.S. Environmental Protection Agency.
32
-------
packet in place. However, this technique tends to hold the packet on the
bottom of the channel where it may be covered with sediment or damaged by
bottom-dwelling organisms.
The anchor is secured to the bank with a small nylon cord. Use of a tan-
colored cord lessens the chance of discovery and tampering. If the potential
for vandalism is high, hide the anchor or place it in a location that is not
easily accessible. In addition, a business card or note containing
information about the investigations can be encased in plastic and attached to
the anchor to help satisfy the curiosity of potential vandals.
In addition to discharge points such as streams and springs, detectors
may also be suspended in toilet tanks in selected private homes served by a
well or in the stream of water discharging from private or public supply
wells. Toilet-tank placement of the detector is not acceptable if self-
dispensing toilet bowl cleaners are used. In that case, the detector can be
suspended in a container such as a 5-gallon bucket, and water from a garden
hose directed to flow into the container. The container needs to be placed
out of direct sunlight and located so that the wastewater does not create a
nuisance. A flow rate of approximately 1 gallon per minute is adequate.
The packets are left in place from 1 to 5 days but are generally changed
more often if turbidity levels are high. When retrieved, each packet is
rinsed to remove accumulated sediment and trash. Each packet is then sealed
in a lock-top plastic bag, labeled, and returned to the laboratory in a light-
tight case. If the packets are not processed immediately, they should be
dried and refrigerated in order to decrease bacterial action that could reduce
fluorescence. Packets processed this way can be stored several weeks without
adversely affecting the fluorescence of sorbed dye.
Before being removed from the fiberglass bag in the laboratory, the
charcoal is rinsed with a jet of water to remove sediment which can interfere
with the analysis. Some investigators prefer to test only about half of the
charcoal and store the remainder in case there is need to confirm the results
of the trace.
The presence of dye and, thus, a positive trace is determined by
elutriating the exposed charcoal in an alcohol solution and visually checking
for the characteristic yellow-green color above the charcoal. For rhodamine
WT, charcoal is placed in a small jar or beaker and covered with about 30 mL
of elutriant solution consisting of 38 percent ammonium hydroxide, 43 percent
1-propanol, and 19 percent distilled water (Smart, 1972). For fluorescein, a
saturated solution of about 5 percent potassium hydroxide is the most
efficient elutriant (Quinlan, 1987a). This solution consists of 6 to 7 grams
of potassium hydroxide dissolved in 100 mL of 70 percent isopropyl alcohol
(rubbing alcohol). After the potassium hydroxide dissolves, the solution
separates into a supersaturated solution and a saturated solution. The
lighter saturated solution is decanted into the containers to cover the
charcoal. Additional test solution can be made by adding potassium hydroxide
or alcohol to the original solution so long as only the lighter, saturated
part of the solution is used to elutriate the charcoal. However, Quinlan
(1987a) reports that the potassium hydroxide and alcohol solution has a
limited shelf-life and should not be used if more than a few days old. The
33
-------
ammonium hydroxide, 1-propanol, and distilled water elutriant has a shelf-life
of several months. Either elutriant can be used with the other dye, but each
is most efficient for the dye indicated.
In strongly positive tests, the maximum color intensity develops almost
immediately upon addition of the elutriant and then slowly decreases (Jones,
1984). Aley and Fletcher (1976) propose a system for qualitative assessment
of fluorescein dye concentrations that lists dye elutriating periods ranging
from 15 minutes up to 10 days. During the investigation in the Elizabethtown
area, Kentucky, Mull, Smoot, and Liebermann (1988) adopted 4 hours as the
maximum time for elutriation. In cases where the typical yellow-green color
of fluorescein is not obvious, detection can be enhanced by focusing a beam of
sunlight or light from a microscope lamp through the solution. Quinlan
(1987a) states that the light-beaming technique increases the detectability of
fluorescein so that concentrations as low as one part per billion may be seen.
Identification of a strongly positive test is generally obvious.
However, weakly positive tests require special care and experience lest the
presence of algae and organic matter, which can occur as background, be
incorrectly interpreted as fluorescein. Persons unfamiliar with the color of
dye elutriated from charcoal detectors need to prepare and elute laboratory
test solutions of various dye concentrations before interpreting first-time
dye traces. Dye identification can be confirmed by instrumental analysis of
the elutriant, as described in section 3.3.5.
Optical brighteners and Direct Yellow 96 are recovered on swatches of
undyed, surgical cotton suspended in the ground-water discharge in much the
same manner as the charcoal packets. The cotton swatches are about 4 inches
long, 2 inches wide, and 1 inch thick and are suspended from the anchored wire
by a paper clip or appropriate cord, or enclosed in aluminum, nylon, or
fiberglass packets similar to those used to hold the charcoal. Care needs to
be exercised, however, because some surgical cotton is brightened. Also, the
manufacturer may change production techniques at some time, and sell
brightened instead of unbrightened cotton. Therefore, the cotton being used
for dye detectors needs to be checked for fluorescence or contamination before
placement in the field.
The cotton swatches are rinsed to remove sediment and trash when they are
recovered from the test site. Each cotton detector is placed in a carefully
labeled, lock-top plastic bag for return to the laboratory. As with the
packets of charcoal, the cotton may be stored in a freezer for several weeks
without affecting the fluorescence. Because of the possible masking of
fluorescence by sediment, the cotton needs to be thoroughly cleaned with a
high-pressure jet of water before testing.
The presence of both optical brighteners and Direct Yellow 96 is
confirmed by viewing the cotton detectors under long-wave length ultraviolet
light. The fluorescence of these tracers is more visible if the exposed
cotton is viewed with ultraviolet light under subdued lighting such as a
darkened room or viewing box. Cotton that has adsorbed optical brighteners
will characteristically fluoresce blue-white, while the fluorescence of Direct
Yellow 96 is canary yellow. A positive trace is indicated only if the entire
cotton mass fluoresces relatively evenly. Scattered specks of fluorescence on
the cotton should not be interpreted as a positive dye recovery.
34
-------
Because different types of dye collectors and detection techniques are
used to recover fluorescein and optical brighteners, multiple qualitative dye
traces to the same recovery point may be performed simultaneously. For
example, fluorescein can be placed in one injection point and recovered on
activated coconut charcoal, while Direct Yellow 96 or an optical brightener
could be placed into a second injection point and recovered on unbleached
cotton suspended on the same anchor. If a fluorometer or spectrofluorometer
is available, rhodamine WT might also be placed into a third injection point.
As many as three separate injection points may be traced to one resurgence or
discharge point during the same test.
3.3 Quantitative Dye Tracing
3.3.1 Introduction
As used in this manual, quantitative dye tracing consists of the
injection of a known quantity of dye and the measurement of the concentration
of dye over time, at a particular ground-water discharge point such as a
spring or well. Determination of dye recovery requires measurement of both
dye content and ground-water discharge. These measurements are made at each
dye- resurgence site that was identified as being hydraulically connected to
injection locations by previous qualitative tracer tests. Water samples are
collected, usually with automatic samplers, during passage of the dye cloud
and the dye content of each sample is measured with a properly calibrated
fluorometer or spectrofluorometer. These data are plotted against time to
produce a dye-recovery (time-concentration) or breakthrough curve. A typical
dye-recovery curve is shown in figure 6.
Quantitative dye-tracing can be performed to determine potential
contaminant transport characteristics, such as persistence, dispersion rates,
and concentration. Quantitative dye tracer tests are generally more labor
intensive and require more sophisticated equipment and techniques than
qualitative dye tracing because the objective is to define dye concentration
variations during passage of the dye cloud rather than simply to determine if
the dye appeared at a particular spring or well.
The shape and magnitude of the dye-recovery curve is determined by: (1)
the quantity of dye injected, (2) the characteristics of the dye, (3)
discharge rate at the resurgence, (4) rate of dispersion of the dye, and (5)
the cross-sectional mixing of the dye before arrival at the sampling point.
The apparent shape of the dye-recovery curve can also be affected by the
sampling interval. Analysis of the dye-recovery curve provides insight into
the flow characteristics of the aquifer such as the effective time of travel
between a swallet and the resurgence and the velocity of ground-water flow.
Additional analysis of the recovery curve and discharge measurements may be
used to provide estimates of peak concentration, duration or persistence, and
dispersion. Because these data may be related to the velocity and dispersion
of & potential ground-water pollutant, quantitative dye tracing is an
especially useful tool to managers of water supplies located in karst terrane,
where springs or wells are susceptible to the introduction of contaminants
into the ground-water system. The following discussion summarizes techniques
for quantitative dye tracer tests and analysis of the results. Also included
35
-------
c o
O -3"
•t-i -a-
u —
u
o;
M JO
3 «
O N
C M
O
i-l U 0)
CO 01 -r-(
t-i C W
c • c
c
E >s -H
•H U
4-1 -H 4)
0)
o
O 0) U-H
U ^H
01 O. 4J
Vj E 01
I co 01
0) en u_i
TJ TJ O
c o
«~H en i/^
co
u E -i
•H CO —I
Q. 01
H 4J O
I W
• oo u
vO C C
•H ca
0( ^ 4J
Vj C CO
3 -H -i-l
00 U3 TJ
•H
tu
in
ro
CM
o
3AQ
36
-------
are references that describe, in detail, the procedures used. The discussion
describes each phase of quantitative dye tracing, generally in operational
sequence.
3.3.2 Selecting Dye for Injection
The primary criteria for the selection of a fluorescent dye for
quantitative tracing are its (1) water solubility, (2) detectability in low
concentrations or whether it is strongly fluorescent, (3) separability from
background fluorescence, (4) stability or conservancy in the karst
environment, (5) non-toxicity in low concentrations, and (6) cost. Generally,
detection limits are controlled by background conditions such as turbidity;
the presence of substances that may fluoresce in the same range as the dye,
such as algae; and the calibration of the fluorometer. Conservancy refers to
the stability of the dye in the environment. No dye is 100 percent
conservative because some dye is lost to sorption or chemical decay.
Therefore, dye loss must be considered during quantitative analysis of the
dye-recovery data.
Rhodamine WT ia specifically made for water tracing and has been widely
used for time of travel and dispersion tests in streams (Hubbard and others,
1982). Because of the relative similarity between ground-water flow in karst
terrane and streamflow in most areas, and because it meets the above
requirements, rhodamine WT is commonly selected for quantitative dye traces In
karst terrane. Rhodamine WT was used by Mull, Smoot, and Liebermann (1988)
for quantitative dye tracer studies in the Elizabethtown area, Kentucky, and
is the dye used for all quantitative dye trace discussions in this manual.
3.3.3 Selecting Quantity of Dye for Injection
After identifying and selecting input and resurgence points on the basis
of earlier qualitative traces, the quantity of dye to be injected for
subsequent quantitative tests needs to be determined. The quantity of dye
required for a quantitative trace depends on flow conditions, the distance of
the trace, and the peak dye concentration expected at the dye-recovery site.
Considerable experimentation may be required before the optimum quantity of
dye can be selected, consistently. In general, the quantity of dye injected
must be adequate to produce detectable dye concentrations at the monitored
resurgence but ideally remain below the levels of visual detection. As stated
earlier, dye studies conducted by the U.S. Geological Survey are usually
designed so that the amount of fluorescent dye injected into a water course
does not result in dye concentrations exceeding 0.01 mg/L at water user
withdrawal points (Hubbard and others, 1982). This value may be exceeded if
the dyed water in the water-supply spring or well can be diverted from use
during passage of the dye cloud.
For initial quantitative traces, the quantity of dye per injection can
generally be based on an equation for estimating the amount of a 20 percent
solution of rhodamine WT needed for a slug injection (Kilpatrick and Cobb,
1985). The equation adapted to dye recovery from a spring or pumped well is
as follows: s
V - 3.79 x 10'3Q (L 1.5) C (eq. 2)
*T *
37
-------
where V - the volume of rhodamine WT, 20 percent solution, in
milliliters;
Q - discharge, in ft8/s;
L - the apparent length of the trace, (map distance) in feet;
v - the apparent ground-water flow velocity, in ft/ a; and
C - the peak concentration at the sampling site, in micrograms
p per liter
For initial traces, the ground-water velocity can be estimated from the
results of qualitative traces or from spring discharge measurements. To
adjust for the meanderings of the subsurface flow passages, the apparent
length of the trace (straight- line map distance) is multiplied by 1.5
(Sweeting, 1973 p. 231).
3.3.4 Dye-Handling and Recovery Procedures
Handling, transporting, and injecting dye for quantitative tracer tests
requires even more care than that for qualitative tests. Considering that dye
concentrations are commonly measured in parts per billion, or less, the
•lightest contamination can cause erroneous or misleading data. Rubber or
disposable plastic gloves need to be worn during all handling of the tracer
and extreme care used to avoid contaminating clothing or the area around
injection or recovery points. Ideally, in both qualitative and quantitative
traces, care needs to be taken to ensure that dye collection devices are not
handled by the same person that previously handled the dye. Different people
need to handle each, or if only one person is used, the collectors need to be
placed before the dye is handled. The tracer needs to be stored and
transported to the injection site in an opaque, non-breakable container. The
quantity of dye to be injected needs to be measured with a graduated cylinder
or equivalent and mixed with water in a metal pail for injection. A metal
pail is used to lessen dye loss due to adhesion to plastic. The dye needs to
be diluted about 10:1 before injection. The mixture needs to be carefully
poured into water draining directly into a swallet or into the center of flow
if water is ponded over the swallet.
Rhodamine WT normally comes from the manufacturer as a 20 percent
solution in bulk containers. Over time, some dye may settle out of solution.
Therefore, before the dye is withdrawn from the container, the contents need
to be thoroughly mixed to ensure that the dye is appropriately mixed.
The collection of water samples over time and fluorometric or
spectrofluorometric analysis of their dye content is necessary to adequately
define the dye -recovery curve at a monitored spring or well. The samples can
be collected by hand or by automatic samplers. Equipment required for
sampling by hand will vary depending on conditions at the sampling site.
Samples could be obtained from a bridge, a boat, or by wading.
Although hand sampling can suffice for quantitative tracing in certain
situations, automatic samplers are more efficient because sampling may be
required for a prescribed frequency over a period of several days or during
inclement weather. In addition, the chance of missing part of the dye cloud
is reduced because automatic samplers can sample for long periods prior to the
arrival of the dye cloud. This may be necessary during initial quantitative
dye tracing when ground-water velocities are not well defined. Also,
38
-------
simultaneous sampling from multiple sites is generally more efficient and may
be less costly with automatic samplers because only one person is required to
service several samplers. Because the passage of the dye cloud must be
defined, the use of an automatic sampler may permit more precise detection of
dye concentrations than might be achieved using the random hand sampling
technique (Crawford, 1979, p. 43).
The use of automatic samplers is not without some disadvantages, however.
A problem with both automatic and manual sampling is the potential for poor
definition of the dye cloud if the selected sampling interval is too long.
Accurate definition of the dye-cloud peak generally requires relatively
frequent samples, but continuation of the same sampling frequency can lead to
excessive sampling during the dye-cloud recession. In general, unnecessarily
frequent sampling is preferred to inadequate definition of the dye-cloud peak
due to long sampling intervals. Repeat traces may be needed to determine the
most efficient sampling interval.
•
The samplers may be subject to vandalism and could malfunction,
especially during severe winter weather. Covers made from boxes or barrels
that are lockable and insulated can improve automatic sampler reliability
under these conditions.
There are several types of automatic samplers commercially available.
Also, Crawford (1979, p. 43) describes a "homemade" sampler consisting of a
pump, pump activation clock, power source, and sample distribution and storage
box. Automatic samplers are generally of two types: (1) bank-mounted
samplers having a small diameter suction tube leading to the sampling point in
the water, and (2) a floating, boat-like sampler (Kilpatrick, 1972) that is
partially immersed in the flow. Commercially produced, bank-mounted samplers,
also called sequential waste-water samplers, which collect samples in glass or
plastic bottles, are available from several manufacturers. An 1SCO sampler,
model 2700 (table 2), was used during the dye tracing in the Elizabethtown,
Kentucky area (Mull, Smoot, and Liebermann, 1988). Although superseded, this
model can collect and hold as many as 24 samples with a sampling interval
adjustable from 1 to 999 minutes and is powered by either self-contained
batteries or an external power source.
Regardless of the method of sampling, glass sample bottles are
preferable to plastic bottles because the dye may have a slight affinity for
the plastic, resulting in dye loss. One ounce (approximately 32 milliliters)
polyseal-cap glass bottles are commonly used for hand sampling. The dye
content of each sample is then measured with a properly calibrated fluorometer
or spectrofluorometer.
3.3.5 Fluorometer Use and Calibration
A filter fluorometer is an instrument that measures the intensity of
light at a selected wavelength from a water sample containing a fluorescent
substance. The intensity of fluorescence is proportional to the amount of
fluorescent substance present. Fluorescence can also be measured with
fluorescence spectrometers or spectrofluorometers (Udenfriend, 1962 p. 62-86
and Duley, 1986). These instruments are especially useful if several
different dyes are being simultaneously used for tracing to the same point or
if there is a high background level of dye from an earlier trace. In this
39
-------
manual, however, reference will only be made to the use of a fluorometer.
Although a spectrofluorometer offers many capabilities not available with a
fluorometer, the capital cost is about three times greater. A portable
instrument, such as the Turner Designs Model 10 (table 2) filter fluorometer,
provides direct reading of dye content as determined by prior calibration.
This instrument reaches operational temperature in about 15 minutes, can be
powered with either 115 volt a.c. or 12 volt d.c. current, does not adversely
increase the temperature of the test sample, and can be equipped for
continuous flow-through monitoring and recording.
A filter fluorometer consists of six basic components as shown in figure
7 (Wilson and others, 1986). The light source and filters are selected for
maximum sensitivity to the particular fluorescent tracer being used according
to the manufacturers' specifications. Detailed information about the
operation and calibration of fluorometers, is given by Wilson and others
(1986). The calibration procedure is reiterated here because dye-recovery
data acceptable for quantitative analysis is dependent on the use of a
correctly calibrated fluorometer.
The preparation of calibration standards is basically the process of step
by step reduction of the stock dye solution until concentrations that are
expected during dye recovery are reached. This reduction is generally known
as a serial dilution and is explained in detail by Kilpatrick and Cobb (1985).
Precise measurements of the initial volume of dye and added diluent in each
step of the procedure are necessary in order to prepare a set of standards for
an accurate calibration of the fluorometer. The serial dilution procedure is
based on the equation:
W. V
Cn ' Ci — Ci Sg —*— ' (eq' 3)
v + v. 8 v + v.
w d w d
where C - the new dye concentration, in micrograms per liter (ug/L);
C. - the initial dye concentration, in micrograms per liter;
V - the volume of the added diluent, in milliliters (mL);
V. - the volume of the dye solution added, in milliliters;
W, - the weight of the initial solution, in grams (g); and
S - the specific gravity of the initial dye solution,
8 1.19 grams per cubic centimeter (g/cm3) for
rhodamine WT, 20-percent solution,
For rhodamine WT, 20-percent solution, (200 g/L) 3 serial dilutions are
required to obtain concentrations on the order of 100 jig/I, (table 3). In each
step, the preceding C becomes the new initial concentration, C.. This third
dilution (100 .ug/L) needs to be retained in quantity as a "working solution"
and is used for further dilution if standards below 100 jig/L are needed. The
use of the working solution eliminates the necessity to perform complete
serial dilutions each time the fluorometer is calibrated so long as the same
dye lot is being used. The same working solution needs to be used throughout
an investigation as long as all dye used is from the same dye lot. The
working solution needs to be sealed and stored out of light when not in use.
40
-------
6. Readout device
Gives values proportional to the
light reaching the sensing device.
5. Sensing device
Responds to the spectral band
passed by the secondary litter.
1. Energy source
2. Primary filter
Passes only a selected band of the
source's output spectrum matching
a selected band of the dye's
excitation spectrum.
3. Sample holder
Sight angle in light path minimizes
amount of scattered light reaching
sensing device.
4. Secondary filter
Passes only a selected band of the dye's
emission spectrum and preferably none
of the light passed by the primary filter.
(from Wilson, Cobb, and Kilpatrick, figure 4, p. 10)
Figure 7.—Basic structure of most filter fluorometers.
41
-------
Table 3.--Three-step serial dilution for preparation of
standards for fluorometer calibration.
First dilution
Second dilution
Third dilution
Initial
concen-
tration
(micro -
grams
per
liter)
200,000,000
2,280,000
15,100
Specific
Gravity
(grams
per
cubic
centi-
meter)
1.190
1.002
1.000
Volume
of dye
(milli-
liter)
20
20
20
Volume
of diluent
(milli-
liter)
2,068
3,000
3,000
New
concen-
tration
(micro -
grams
per
liter)
2,280,000
15,100
100
Although an infinite number of combinations of water and dye can be used
to prepare standards, table 3 shows the combinations for three dilutions to
produce the working solution of 100 ug/L, a commonly used calibration
standard. Distilled water is used for calibrating the fluorometer to zero
background. Fluorometer calibration should be checked with temperature
equilibrated solutions before the final measurements for all dye samples.
3.3.6 Calculating Mass of Injected and Recovered Dye
Analysis of dye-recovery data is based on a mass balance relation between
the injected and recovered dye (Hubbard and others, 1982). Therefore, a
critical element of this analysis is the determination of the mass of dye
injected. It is calculated by the following equation:
mass - volume x density x purity.
(eq. 4)
Thus, for a 10 mL injection of rhodamine WT, 20 percent solution, the mass is
computed by multiplying the volume (10 mL) by the density (1.19 g/cm3) and by
the strength or purity (20 percent). In this case the mass injected is 2.38
grams.
Calculation of the mass of dye recovered is based on the dye-recovery
(time-concentration) curve and on the discharge of the monitored resurgence.
The time-concentration curve is a plot of time versus the dye content of water
samples collected during the passage of the dye cloud. The curve is typically
bell-shaped but skewed to the right, such that it is steeper on the rising
limb than on the falling or recession limb of the curve as shown in figure 6.
For the conditions of a variable flow rate the mass of dye recovered is:
M -
where
/QCdt,
(eq. 5)
M - mass of dye recovered,
Q - discharge, and
C - dye concentration at time t
42
-------
If the flow rate is steady, the mass of the recovered dye is:
•*
/•
M - Q ICdt. (eq. 6)
CT
Equations 5 and 6 are given for conceptual purposes only, computational
formulas are given in sections 4.2.1, and an example of the computation of the
mass of recovered dye is given in section 4.2.6. Discharge is determined by
current-meter measurements. If discharge varies, such as during a storm
event, several discharge measurements are necessary to accurately characterize
flow conditions and calculate dye recovery.
3.3.7 Sampling Procedures
Adequate definition of the time -concentration curve is dependent on an
appropriate sampling interval that is based on an assessment of the flow
conditions and hydrology of the test site. The sampling interval chosen
during initial traces to a particular resurgence is frequently adjusted for
better definition of the time-concentration curve during repeat traces. For
example, the sampling interval varied from 20 to 60 minutes during traces in
the Elizabethtovm area (Mull, Smoot, and Liebermann, 1988). The adequacy of
the sampling interval to define the dye -recovery curve can be determined by
measuring the fluorescence with a properly calibrated fluorometer and plotting
the results in the field.
The degree to which the dye is laterally and vertically mixed in the flow
at the sampling site affects the sampling strategy. Sampling at a single
point from a laterally unmixed concentration distribution cannot be used to
properly calculate the mass of recovered dye for the sampling interval.
An acceptable mixing length may be defined as the distance needed for
nearly complete lateral mixing of the dye tracer in the test stream. In the
application of dye- tracer techniques to surface-water investigations, the
mixing length can be estimated using techniques described by Kilpatrick and
Cobb (1985) and the distance to sampling sites can be selected accordingly.
In ground-water applications of quantitative dye- tracer techniques, however,
this calculation has little application. In a single conduit system, the
mixing length is typically the distance between the dye injection point, such
as a sinkhole, and the point of dye recovery from a ground-water resurgence,
such as a spring. This distance is fixed unless dye is sampled downstream of
the resurgence point. In a converging flow system, the mixing length would
not begin until the last convergence of flow took place. In this case, an
adequate mixing length could, theoretically, extend beyond the resurgence
spring or well.
Incomplete lateral mixing is usually the result of a short mixing length.
Incomplete mixing can also occur when undyed ground water enters the flow
system between the injection point and the sampling point. The addition of
undyed water to the flow system upstream of the sampling point, may cause
inaccurate estimates of recovered dye.
During quantitative dye tracing, a preliminary trace is usually needed to
assess the degree of lateral mixing of dye at the recovery point. Dye is
recovered from at least three points in the cross section at the resurgence
and time -concentration curves are plotted for each sampling point in the
43
-------
section. Complete lateral mixing is reached when the areas under the time-
concentration curves for each lateral position sampled are the same,
regardless of curve shape or magnitude of the peaks. Kilpatrick and Cobb
(1985 p. 3) considered that mixing was adequate for single point dye sampling
when the mixing was about 95 percent complete at the resurgence or other
sampling point. The relation of mixing length to curve areas is illustrated
by figure 8.
In order for samples to represent the dye cloud with incomplete lateral
mixing, the dye cloud needs to be sampled at several points laterally, such as
at points a, b, and c as shown in figure 8. Selection of these sampling
points is based on equal increments of flow as determined from a stream
discharge measurement made with a current meter. For example, if three points
in a section are to be sampled, the samples need to be collected at 1/6, 3/6,
and 5/6 points of cumulative flow across the section (Kilpatrick and Cobb,
1985). However, in most karst ground-water situations, the recovery distance
is hundreds to thousands of times greater than the diameter of the stream, or
conduit width. Therefore, single point sampling should be adequate.
3.3.8 Sample Handling and Analysis
Specific procedures are required for the handling of water •ample* after
collection because improper handling can affect the accuracy of the
determination of dye content. All samples need to be transported and stored
in light-proof containers to prevent reduction of fluorescence due to
photochemical decay. Although some sample* may be checked for fluorescence in
the field, primarily as confirmation that the dye cloud was sampled, the
samples need to be returned to the laboratory for fluorometric analysis of
their dye content.
Also, if results of the dye trace may be used as evidence for litigation,
it may be necessary to establish chain-of-custody control of the samples.
This procedure ensures that the samples are secure at all times and a written
log is maintained that identifies each person having access to the samples
(Quinlan, 1986a).
The dye content of each water sample needs to be measured by determining
its fluorescence in the laboratory using a properly calibrated fluorometer.
Because fluorescence activity is significantly affected by temperature
(fluorescence is inversely proportional to temperature), temperature effects
must be accounted for in data analysis (Wilson and others, 1986). However,
temperature effects can usually be ignored if the fluorometer used for
measuring dye content does not increase the temperature of the sample during
analysis and if the temperature of the sample and calibration standards are
allowed to equilibrate, usually overnight, before analysis.
The results of the fluorometric analyses of the dye samples should be
plotted as a dye-recovery (time-concentration) curve, which is discussed in
section 4.2.1 of this manual.
3.3.9 Adjustment of Dye-Recovery Data
Because the dye is not completely conservative (that is, some of the dye
is lost to decay and sorption during the trace), the measured dye
44
-------
tr
t-
z
Ui
O
8
Flow
a •
b •
c •
Definition sketch of
sample points
TIME
Short Distance
Curve areas not the same,
lateral mixing incomplete.
Optimum Distance
Curve areas about the same,
mixing nearly complete.
Long Distance
Curve areas identical,
perfect mixing
(from Kilpatrick and Cobb, 1985, figure 1, p. 3)
Figure 8.—Typical response curves observed laterally and at different distances
downstream from a dye-injection point.
45
-------
concentrations are adjusted to eliminate the effects of dye loss. This
adjustment is required if the results of different traces are to be compared.
In this procedure, the measured dye concentrations in each sample for a given
dye trace are multiplied by the ratio of the dye mass injected to the mass of
dye recovered.
Because the concentration of dye (tracer) recovered is affected by the
quantity of dye injected, it may be desirable to "normalize" the dye-recovery
data by dividing observed concentrations of each sample by the mass of the
injected dye. The resulting concentrations are therefore adjusted to a
particular unit of dye, for example, milligrams per liter per kilogram. This
procedure compensates for variations in the mass of dye injected during
different traces and simplifies analysis of the time-concentration data.
To adjust for dye loss and normalize to a unit mass of tracer, it is
necessary to assume that none of the dye is trapped in the subsurface or flows
to resurgences other than those sampled and included in the computation of the
mass of dye recovered. This fact can usually be established with qualitative-
tracer tests and confirmed with discharge measurements during the time of the
quantitative test. The complexities of ground-water flow in karst terrane are
such that some dye may escape detection by flowing to an unmonitored
resurgence, especially during traces performed under high-flow conditions. In
addition to incomplete mixing and nonconservative dye, the addition or
subtraction of water to the flow stream might greatly affect the recovery and
computed characteristics. The equations and methodology for these adjustments
are shown in section 4.2.
3.4 Quality-Control Procedures
Extreme care must be used whenever handling dyes in order to avoid
contamination of the dye detectors and sampling equipment. For this reason,
detectors or "bugs" need to be fabricated and installed before the dye is
handled and injected. This is also true for all sampling equipment including
automatic samplers. Long-sleeved rubber or disposable plastic gloves need
to be used during all dye handling operations, including recovery operations.
This limits exposure to the dye and also reduces the possibility of cross-
contamination between exposed detectors and samples through handling. The
lock-top plastic bags used while returning the detectors to the laboratory
should not be reused. The dye needs to be stored in a separate room from the
detectors and sampling equipment and transported to the injection site in
lock-top plastic boxes or in plastic jugs, in the case of liquid dye.
First-time users of a particular tracer should conduct laboratory tests
to familiarize themselves with the characteristics of that tracer. In
particular, learning to recognize weakly positive concentrations of dyes such
as fluorescein requires practice, especially in locations where the background
fluorescence may be increased by traces of fluorescein (or substances that may
look like it) which are found in such products as antifreeze, dishwashing
liquids, and toilet-bowl cleaners (Duley, 1983).
Repeat traces, both qualitative and quantitative, to the same point using
the same dye should not be performed until after the occurrence of a major
storm in order to allow any residual dye to be flushed from the system. This
46
-------
reduces the possibility that a later storm might flush slow-moving or residual
dye-laden water from the first trace and cause an incorrect positive trace.
Repeat traces during low-flow conditions must be delayed sufficiently to
insure that residual dye has drained from the system. In the case of
quantitative traces, this may be confirmed by near zero dye concentrations at
a spring or well measured with a fluorometer prior to the start of each trace.
In addition, fresh dye detectors may be kept in place for at least a week
after the arrival of the initial dye cloud. This will help to define the
level of background fluorescence and monitor the arrival of a second dye cloud
from a single dye injection.
Quinlan (1987a) states that all dye-tracer tests be designed so that
there is always a positive dye recovery somewhere. Dye traces without
positive results generate questions such as, was enough dye used? Were the
correct resurgences monitored? And were the resurgences monitored long
enough? Also, subsequent traces in the same area with the same dye may be
questionable until the specific point-to-point connections are defined.
During fluorometric analysis, careful handling of the cuvette (sample
holder) is required to avoid an increase in sample temperature and the
resulting incorrect determination of its dye content. The cuvettes need to be
handled only by the open end and cleaned with distilled water or water from
the next sample between each reading. All water and equipment used during
fluorometric analysis needs to be temperature equilibrated.
In addition to the collection of water samples for fluorometric analysis
during quantitative-tracer tests, passive detectors can also be used to
confirm passage of the dye cloud. This check is useful in case the selected
sampling interval misses the dye passage or in case the automatic sampler
malfunctions. Passive detectors are used for a period of one to two weeks,
after the passage of the dye cloud, in order to identify passage of a
secondary dye cloud, or confirm that all dye has drained from the system.
The time that the detectors are installed is a convenient time to review
changes in water use from resurgence points such as intermittent withdrawals
by farmers for irrigation purposes. This is also a convenient time to explain
to the landowner or water user the nature and purpose of the pending dye
trace, which is especially important if the possibility exists that the dye
may discolor the water. Ideally, this will simply be a review of earlier
conversations when permission to monitor for the arrival of dye in privately
owned springs or wells was obtained.
3.5 Summary Decision Charts
Decision charts are presented which summarize the material presented in
the discussion of qualitative (fig. 9) and quantitative (fig. 10) dye tracing.
These charts present key steps, in sequential order, for completion of
qualitative and quantitative dye tracing in karst terrane.
47
-------
si?
SIS
u:
5**
t or coordinate
tate agencies
0 M
• ft
'§ *
O'
kvst features U
>,
s
*
A
1
00
0)
PQ
Si
e
ft
l!
•2
= ;
V
I
1
I
u
!i'
?P
i"
00
•H
U
cd
*
CO
cr
I
co
T-l
U
•
ON
0)
48
-------
Begin
Saltct «nd procure
wppllu end iqulpmml
to
timpllns
Irawvcl
dtquit* to dil
4y» cloud?
End
Figure 10.—Decision chart for quantitative dye tracing.
49
-------
4. ANALYSIS AND APPLICATION OF DYE-TRACE RESULTS
4.1 Introduction
Ground-water tracing has long been a valuable tool for investigating
ground-water conditions in karst terrane. Atkinson and Smart (1981) classify
the application of tracers into three use -categories: (1) determination of
flow paths and residence times, (2) measurement of aquifer characteristics,
and (3) mapping and characterization of karst conduit systems. Aley (1984)
lists several applications of ground-water tracing that are especially useful
in water pollution investigations. Specifically, tracing provides (1) direct
proof of movement of water from one point to another, (2) results that are
easily understood by the public, regulatory agencies, and the courts, (3) a
qualitative indication of whether or not effective natural cleansing of water
contaminants could occur along the flow path, and (4) an indication of
underground travel rates, which tend to be underestimated by persons
unfamiliar with ground-water flow conditions in karst terrane.
Analysis and application of the results of dye tracing is controlled, in
part, by the design and implementation of the tracer tests and can generally
be classed as qualitative or quantitative according to the nature of the trace
and subsequent analysis. Perhaps the simplest and most frequent use of the
data from dye tracing is to identify point-to-point connection between a
discrete ground-water input point such as a sinkhole, and a downgradient
discharge point such as a spring or well. Numerous investigators have used
these results to define the ground-water basin or catchment area for a
particular spring or group of springs (Aley, 1972; Atkinson and Smith, 1974;
Quinlan and Ray, 1981; Quinlan and Rowe, 1977; Skelton and Miller, 1979;
Smart, 1977; Crawford, 1981; Thrailkill and others, 1982). These data are
useful for a variety of purposes but, from the standpoint of water-supply
protection, are especially useful for the identification of potential sources
of contaminants detected in springs or wells and are almost always a necessary
first step to quantitative dye tracing.
Results of qualitative dye tracing can confirm the validity of water-
level contours and thus the direction of ground-water movement. Qualitative
dye tracing can also identify the nature of the system draining to a
particular spring, that is, whether the flow system is convergent or
distributary to the spring. If distributary, qualitative dye tracing is the
most practical technique to identify the springs that drain the system.
These techniques can also efficiently show changes in flow routes when
drainage from a particular sinkhole or karst window is traced to different or
additional springs during high- flow conditions.
With the aid of the filter fluorometer and conservative fluorescent dyes,
the analysis of data from ground-water tracer tests can be quantitative in
that mass-balance analysis of dye-recovery data is possible. These data, in
combination with a carefully measured discharge, permits a hydraulic
characterization of the underground conduit network and the definition and
estimation of flow characteristics that is not generally available from other
techniques. For example, repeated traces between the same input and recovery
points showed the relation between traveltimes and discharge (Smart, 1981;
Stanton and Smart, 1981; Mull and Smoot, 1986). Smart and others (1986)
50
-------
demonstrated the advantages gained from quantitative tracing over the
qualitative methods for definition of a karst conduit system on the basis of
mass balance of the fluorescent tracers.
In recent years, significant advances have been made in the application
of quantitative dye tracing to the description of flow characteristics such as
dispersion. The analysis of results of quantitative dye tracing was used by
Mull, Smoot, and Liebermann (1988) to describe predictive relations between
discharge, mean traveltime, apparent ground-water flow velocity, and solute
transport characteristics in the Elizabethtown area, Kentucky. Because of the
similarity between a water-soluble contaminant and the fluorescent dyes used
for dye tracing, the results of these studies can be applied to predict
probable impacts on a water supply arising from the spill of water-soluble
contaminants. Contaminants that are not soluble in water do not migrate at
the same rate as do soluble contaminants. Quantitative dye tracing is
therefore, not applicable in determining flow rates of immiscible
contaminants. Crawford (1986) presented the hypothesis that floating
contaminants carried by cave streams accumulate upgradient from the point
where cave passages become completely filled. Section 6 describes, in detail,
the application of results of quantitative dye tracing to predicting flow and
transport characteristics of potential contaminants in karst water supplies.
4.2 Definition of Quantitative Characteristics
The initial result of quantitative dye tracing is a set of measured dye
concentrations, each sampled at a selected time and place. For dye tracing in
streams, it is possible to sample at many locations within a cross section and
at many stations along the flow path, measuring concentration as a function
both of cross-sectional location, longitudinal distance, and elapsed time
since injection. For karst ground-water flow, however, one is generally
restricted to sampling at one or a few fixed locations, measuring
concentration only as a function of time.
Quantitative characteristics are the numeric descriptors of a dye-
response curve. They may be derived from quantitative dye trace data. Some
characteristics, such as elapsed time to peak dye concentration, may be taken
directly from the measured data. Others, such as mean traveltime and
normalized dye concentration, are calculated from the measured data. Still
others, notably the dispersion coefficient, may be estimated from the data
only if simplifying assumptions are made.
This section describes the primary quantitative characteristics for dye-
tracing analysis and how they are derived. A summary table of these
characteristics is given in section 4.2.5. An example of the computation of
quantitative characteristics for an actual dye trace is given in section
4.2.6. A computer program code that calculates quantitative characteristics
is discussed in section 6, and is listed in Appendix A.
Application and interpretation of these quantitative characteristics is
discussed in. section 5 of this manual. It includes the development of
relations among characteristics, and their use in the prediction of
contaminant transport.
51
-------
4.2.1 Dye-Recovery Curve
The development of quantitative information from dye traces is based
primarily on the analysis of dye-recovery curves, which are plots of
concentration versus elapsed time since injection. Such curves are also known
as time-concentration, breakthrough, or dye-response curves. Plotted dye-
recovery data typically gives a positively skewed, bell-shaped curve that is
steeper on the rising limb than on the falling limb.
A dye-recovery curve yields several quantitative characteristics by
simple inspection. The time of injection (t_) is used as the beginning of the
dye-recovery curve (fig. 11) and is taken as zero. The measured dye
concentration at tQ becomes the background concentration. Time to leading
^ edge (t.) is the elapsed time, or time of travel, before concentration
increases above background. Time to trailing edge (t_) is the time of travel
until concentration decreases to the background level, and may be called the
elapsed time for dye passage. Time to peak concentration (t , ) is the time
of travel from dye injection to the peak of the dye-recovery^curve.
Persistence is the length of time that any given concentration is exceeded.
Other characteristics may be defined, such as time of travel from the leading
edge until the concentration has decreased to 10 percent of the peak value
(Hubbard and others, 1982).
-j The shape and magnitude of dye-recovery curves are influenced, primarily,
by the amount of dye injected, the velocity and magnitude of flow, the mixing
characteristics within the flow system, the sampling interval, and whether the
discharge is diluted by non-dyed waters. The data from a single dye trace
generally reflect conditions for that particular test, and especially for that
particular discharge. Repeated quantitative dye traces between the same
injection and recovery points are needed to describe the dye-trace
characteristics under different flow conditions.
v The mass of dye recovered, is summed from the time-concentration data for
a dye trace, by the following relation:
n , .
-0.1019 T 0.(C. - C_)
-------
1
fl
0
R
0
^
(
_ C
I
1
•5
V
(
u
<
1
i
4
"" •
r
I
*
—
k
)
a
i
1}
J
5.
1)
»»
4
3
-i
1
H
J
g
U
J)
a.
-t
1
1 ' I ' 1 ' 1 '
0)
•0 >
CO 01 -0
at co IM
63 0
•rl CT
4-1 01 T3
i-l
14-1 CO O
O -H Vd
^
*O C C
00 in
•H i-l C,
l-l 4J
0) (0 u-
Ou U C
4-1
• C 0
01 (U 13 g
O O 01 i-
C C T) 4.
01 O 0)
4-J O 0) *-
CO U g
•H C X S
CO 0) 0) r
1- > \.
0) i-l V4 g.
PL, 60 0
»"*•*"""*"
.....•••• w
""*"•••"—.....
""*"**"*"—....
"••••••—...
g •— ^— ^—— ^ »^—
o
•H 1
4-1 1
CO 1
^ "j
Ub^
1
C
0) ^^
tj
c
0 S
u
0) g
fl •'"I
1 1
II . 1 . 1 . 1 1
1
4J" /]
* ^"N ^^ /
'^ 01 '
eo ^
g 1! j
•^ 1
4J eo ;
C j
i-H i-l j
> -H I
CO CO t
U M •
e o ?
CO U *
1 0) •
i S 01 •'
1 l-i -H ;
O H i _
' • tf
• CO .•
3 W
J I ....-••
*"
-
'•"••«..
"''*-.
O
4J
N-'
0)
Tl
U^
0
C
o
•H
u
0)
*1™1
c
1 1 I \
CM
CO
ID
O
00 J-
O
co o
Ld
—3
O
TI- ct:
LU
P
CO
c
o
o
0)
0)
u
at
T>
0)
CO
a
Cfl
01
t-l
CO
CO
00
c
CO
S
o
M
0)
o
u
0)
M
I
0)
01
M
3
00
NI
3AQ
53
-------
If the unit of concentration used in equation 7 is ug/L, then the mass of dye
recovered will be computed in grams. Note that the sampling interval is the
period of time between two successive samples or observations, and that the
summation was computed from the mean values within each sampling interval. In
this manual, concentrations that pertain to a sampling interval are identified
with the subscript "i"; concentrations that pertain to a given sample are not
subscripted.
For verification, total dye recovery needs to be compared with the amount
of dye injected. If the dye is known or suspected to move from the injection
point along more than one conduit, it is necessary to sample each resurgence
and sum the masses of dye recovered. If the difference between total dye
recovery and the amount injected is appreciably more than what is expected to
be lost to decay and sorption, the results need to be re-examined. Such
results may occur if the conduit system has not been well-defined and all
resurgences have not been sampled. Also, calculated dye recoveries may exceed
the injection quantity if the dye is not well mixed in the cross-section, if
water samples are not representative of the discharge cross section, or if
discharge measurements are erroneous.
Measured dye-recovery concentration needs to be adjusted for background
concentration and reasonable dye loss by:
*in
(eq. 8)
where C, - adjusted concentration, in mg/L;
C - measured concentration, in mg/L; and
M. - mass of dye injected, in kg.
4.2.2 Normalized Concentration and Load
Because the quantity, or mass, of injected dye may be different for
different dye traces, dye concentrations for each trace are normalized to give
the concentration that would have occurred if a standard mass of one kilogram
of dye had been injected. Normalized dye concentrations are calculated as:
C - C (C - Cn) , (eq. 9)
where C - normalized dye concentration, in milligrams per liter
per kilogram of dye injected, in (mg/L)/kg.
Normalized concentrations are useful for deriving relations among
quantitative characteristics and discharge, and are used for the prediction of
concentrations resulting from the injection of a known mass of contaminant.
A plot of normalized concentration gives a true picture of concentration
versus time, for a particular hydrologic condition. Dye-recovery curves of
normalized concentration for seven different dye traces between the same
sites, but under different hydrologic conditions, are shown in figure 12.
Because the concentrations have been scaled for a standard injection mass, one
54
-------
I I I 1 I I I I I I I I I I I I I I ' ' III
O
(N
00
o:
Z)
o
X
o
o
LJ
o
C£
u_
LU
10
o
co
01
T>
CO
e
CO
01
o
cd
0>
>^
T3 •
60
C C
0) i-l
> M
CU O.
CO CO
M CO
O M
<4-l 01
co a
01
> *J
M CO
O 10
0)
>> 60
V-i M
O CJ
u en
0) -H
»-i 13
01
•a
cu
o
z
i
-------
T!
*;'
1
might expect the seven curves to have equal areas under their curves. This is
not the case because there is increased dilution with increased discharge; the
faster traveltimes and higher discharges, hence, greater dilutions are
measured by the response curves to the left in figure 12. The effects of
discharge variation between traces may be removed, however, by converting
normalized dye concentrations to normalized dye loads.
The normalized dye load, or mass flux, is the amount of dye per kilogram
injected passing the sampling point at a given time. Load is dye
concentration times discharge, and normalized load is normalized concentration
times discharge, calculated as:
L - 28.32 C Q, (eq. 10)
where L - normalized dye load, in milligrams per second per
kilogram of dye injected, in (mg/s)/kg; and
28.32 - a unit conversion factor; and
C - normalized dye concentration in (mg/L)/kg,
Q - discharge, in fts/s.
Areas under different normalized dye- load curves are equal and by definition
sum to 1 kg per kilogram injected over the duration of each test. Normalized
load curves from different dye traces may be plotted on the same graph, to
illustrate the passage of the dye cloud under different hydrologic conditions.
The normalized dye load for seven dye traces made between the same sites is
shown in figure 13. Curves toward the left of the graph were obtained under
high- flow conditions, and show that the dye cloud traveled more rapidly and
with less spreading than the curves toward the right, which represent low- flow
conditions .
4.2.3 Time -of -Travel Characteristics
Time of travel is the time required for movement of the dye cloud between
the injection site and the sampling site. The dye cloud spreads as it moves,
however, so the leading edge of the dye cloud can pass the sampling point long
before the trailing edge, with the time to peak dye concentration somewhere in
between. The time to peak concentration gives an indication of the time of
travel, but because of the typical asymmetry of the dye -recovery curve, it is
not representative of the time of travel for the bulk of the dye cloud. For
quantitative analysis, the time of travel is best represented by the centroid,
or mass-weighted mean, of the dye-recovery curve. The time of travel of the
centroid of the dye mass, or simply the mean traveltime, is computed as:
j^i Ci 4ti
-------
— 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 1 1
"
" «•» o
CO OO
O\ *f\ CN
-'"]
I
i
•
«.
o
m
(0
3
O
•H
M
fO
_ OO —
m ir> o
• CO */^ U^ 33 "^ •
oo sr» ao aa a% o
CM _ S% C> -> • CN|
U — - * O U
ea n O — « 3
3 J: SN n « 3
u u x 3 u
J3 U 5
I 1
O
m
CM
o
o
CM
O
O
CG103PNI
QN003S
57
-------
Computation of the apparent mean velocity of ground-water flow follows
directly from the mean traveltime:
d
u - , (eq.
3600 t
12)
where u - mean-flow velocity, in feet per second (ft/s);
d - map distance of the trace, in feet;
3600 - a unit conversion factor, and
t - mean traveltime, in hours.
Because the exact flow path and distance between the dye input and recovery
points is not known, d and TT are apparent values, computed as if the dye
followed a straight line between the points. Thus, the actual flow velocity
is probably somewhat greater than the apparent velocity because of the longer_,
meandering nature of the actual flow path in most karst aquifers. The term u
does give, however, a general idea of ground-water velocity that may be
compared between dye traces for different flow conditions. It may also be
used in estimating the dispersion coefficient discussed in section 4.2.4.
The standard deviation of the time of travel of the dye mass is a
temporal measure of the amount of dispersion of the dye mass that occurred
during the dye trace. In other words, it indicates how much the dye cloud has
spread out in time, between the injection and the sampling site. It is
related to the time of travel and the rate of dispersion, and is computed by
the equation:
n
Z
. 2
n
Z
i-l
0.5
(eq. 13)
where
O -
standard deviation of the traveltime of the dye mass,
in hours.
The standard deviation of traveltime is used in creating the dimensionless
dye-recovery curve, in calculating the dispersion coefficient, and in
developing relations among the quantitative characteristics and discharge.
Because it is a measure of the total amount of dispersion that has taken
place, Q generally increases when Hie time of travel is long or when the rate
of dispersion is great.
The skewness coefficient is a measure of the lateral asymmetry of the
time-concentration curve. This non-dimensional statistic is computed as:
n 3
Z (t.-t) c 4t Q
i-l 111
y -
(eq. 14)
where
- skewness coefficient.
58
-------
A symmetrical curve would have a skewness coefficient of zero. The dye curves
shown in figures 12 and 13 are positively skewed. Positive skewness indicates
that the time-concentration curve is weighted to the right, and that
concentration recedes more slowly than it rises.
As presented, the skewness coefficient is not applied directly for predictive
purposes for ground-water protection. Rather, it is used to compare dye-
recovery curves, to interpret the way in which the dye cloud disperses, and to
synthesize standardized dye-recovery curves, as discussed in section 5.2.
4.2.4 Dispersion Characteristics
The longitudinal dispersion coefficient is a measure of the rate at which
the concentrated dye mass spreads out along the flow path, and is defined as
the temporal rate of change of the variance of the dye-trace cloud (Fischer,
1968). Longitudinal means lengthwise, or along the flow path; dispersion
means spread; and a dispersion coefficient is a rate of spread. In this
manual, dispersion is considered only in one dimension, and longitudinal
dispersion is referred to simply as dispersion. A consideration of dispersion
is appropriate because it is the general term to describe the spreading of the
dye mass that results both in increasing persistence and decreasing
concentration as time passes. In surface-water studies, determination of the
dispersion coefficient is one of the primary goals of tracer tests. Although
the dispersion coefficient is not used directly for predictive purposes in
ground-water studies, it is useful for making comparisons among dye traces of
the relative rate of spreading and, more importantly, as a conceptual tool for
understanding observed dye-trace results, as seen in section 5.2.
Two methods of estimating the dispersion coefficient are presented. Both
are based on equations presented by Fischer (1968), with the assumptions of
constant velocity and uniform flow characteristics between the injection and
sampling points for the entire duration of the dye trace. The first and more
general equation is based on the definition of dispersion coefficient from a
slug injection of dye:
2
a
3600 2 t
D u , (eq. 15)
2 t
where D.. - first dispersion coefficient, in (ft2/s).
The second equation is based on the further assumption that the dye-response
curve is normally distributed, with zero skew along the flow path. When
sampled at the peak of the dye-recovery curve,
588.5
C
peak
A 4 7T D0 t T
2 peak
59
-------
where C . - peak of the normalized concentration curve, in mg/L/kg;
588.5 - a unit conversion factor;
7T - 3.1416;
A - the effective cross-sectional area of the flow medium, in
square feet (ft2), estimated as discharge divided by
mean flow velocity; and
Dg - second dispersion coefficient, in ft2/s.
On the basis of equation 16, the second estimate of the dispersion coefficient
may be computed as: 346,400
D2 ' , (eq. 17)
4 TTt . (C .A)*
peak x peak '
where 346,400 - a unit conversion factor.
The value of the dispersion coefficient is affected by the flow velocity
and the characteristics of the flow medium. In a water body, the observed
amount of dispersion is affected by turbulent forces as well as by simple,
hydrodynamic dispersion, and the computed dispersion coefficient reflects the
combination of both processes. In the case of simple dispersion alone, the
dye spreads outward by molecular diffusion, and concentration within the dye
cloud is normally distributed in space. Even so, the dye-recovery or time-
concentration curve for the case of simple dispersion is expected to have a
slightly positive skew. This slight skew results from the transformation of a
dye cloud that is normally distributed longitudinally at any given time into a
set of dye samples from a fixed location at different times. In other words,
by the time the trailing point of the dye cloud passes the sampling point, it
has had more time to disperse, thus lagging and extending the tail of the
observed curve. This is, however, a relatively small effect. In practice,
the skewness of the observed dye-response curves is typically much greater
than that expected from this axis transformation. Most of the skewness, or
lengthened recession of the dye-recovery curve is not accounted for by simple
dispersion alone, but results mainly from unequal flow lengths and velocities
along and across the flowpath, or in other words, turbulence.
If the dye-recovery curve were normally distributed and perfectly
sampled, values of D.. and D. computed from the curve would be equal. In
practice, however, several factors typically lead to different values of D.
and D? computed from the same curve. The assumption of zero skewness, used in
computing D., usually does not hold. The quantities u and A are not simple
constants, But rather apparent average values. The computed dispersion
coefficient is a composite term for the combined action of simple dispersion
and turbulent action. Turbulence not only introduces skewness, but also
increases the computed value of standard deviation of traveltime, thus
increasing the value of D.. . If the sampling interval is not adequate to
precisely measure the peak concentration and the time it occurred, the value
of D- will be affected. An inadequate sampling interval also affects the
computed values of the other quantitative characteristics from which the
dispersion coefficients are estimated. Based on this discussion, it is
60
-------
difficult to determine which of the two values is more correct. Both are
relative values, and the reader is cautioned not to attach undue numeric
accuracy to ttve computed values of DI and D_ When comparing dye-recovery
curves, however, the use of either coefficient gives a relative idea of the
rate of spread of the dye cloud along the flowpath. At present, consideration
of the dispersion coefficient is appropriate mainly as a conceptual tool for
understanding relations among the quantitative characteristics under different
hydrologic conditions.
4.2.5 Summary of Quantitative Terms
Table 4 gives a summary of the main terms used in the development of
quantitative characteristics. Measured data are plotted to make a time-
concentration or dye-recovery curve of & particular dye trace. The area under
the time-load curve is summed to give the total mass of dye recovered and
percent recovery is computed. Concentrations are adjusted by subtracting the
background concentration and dividing by the percent recovery. Concentrations
are normalized to a standard injection by dividing by the mass injected.
Loads may be normalized by multiplying normalized concentration by discharge.
Statistical descriptors of the dye-recovery data are based on mass-
weighted moments of the data. Mean traveltime is the best estimate of the
time of travel. Standard deviation of traveltime is a measure of the amount
of dispersion that occurred during the trace. The skewness coefficient
measures the asymmetry of the curve, and is indicative of the time delay for
passage of the trailing edge of the dye cloud.
Dispersion simply means the gradual outward spread of the dye from an
initial slug to an expanding dye cloud. It is caused by simple molecular
diffusion, known as hydrodynamic dispersion, and the turbulence of the water
body. The dispersion coefficient is an estimate of the rate at which the dye
cloud expands. Its application to ground-water flow in karst terranes is
presently limited to conceptual use in interpretation of repeated dye traces
under differing hydrologic conditions.
4.2.6 Example of Computation of Quantitative Characteristics
A dye trace performed on May 30, 1985 at Dyers Spring near Elizabethtown,
Kentucky (Mull, Smoot, and Liebermann, 1988) produced the time-concentration
data shown on the left side of table 5. Including the end points, 26
observations of dye concentration were made. Map distance from injection to
sampling point was 3,000 ft. Discharge of 1.14 fts/s was measured by current
meter at the spring, and was constant throughout the dye-trace period. A
quantity of 15 mL of 20 percent rhodamine WT was injected. From equation 4,
the mass of dye injected was:
Mass - volume x density x purity
Min - (15 cm3) (1.19 g/cm3) (0.20) - 3.57 g.
Background concentration (C ) was 0.01 ug/L, time of injection (t ) was 1015
hours. The measured time-concentration data are displayed in figure 14a.
The data were processed using the computer program DYE, described in
section 6.1. Output from the program is shown on the right side of table 5.
61
-------
Table 4—Summary of terms used in definition of quantitative characteristics
Characteristic
Method of
Computation
Definition or primary use
Dye-recovery curve
Time to leading edge
Ma«« of dye recovery
Adjusted dye concentration
Normalised dye concentration
Normalised dye load
Time of travel
Mean traveltime
Mean flow velocity
Standard deviation of traveltime
Skewness coefficient
Dispersion
Dispersion coefficient
Measured
Measured
Figure 11
Equation 7
Equation 8
Equation 9
Equation 10
Equation 11
Equation 12
Equation 13
Equation 14
Equations IS, 17
Measured concentration as a function of
time. Basis for all further computations.
Comparison of curves, synthesis of
'Standardized curves.
Computation of percent recovery, correction
for dye loss.
Correction for back-ground and dyt loss.
Correction for unit mass of injection.
Analysis and prediction.
Compensation for discharge dilution,
for comparison of repeat traces.
Time required for passage of the dye cloud.
Also called traveltime.
Best estimate of time of travel. Analysis
and prediction.
Computation of dispersion coefficients.
Analysis and prediction.
Amount of spread of the dye cloud.
Analysis and prediction.
Asymmetry of the dye-recovery curve.
Synthesis of standardised curves.
General term for the gradual spreading of
the dye cloud. Strictly speaking,
longitudinal dispersion.
Rate of dispersion. Conceptual analysis.
62
-------
Table 5.—Measured and computed data for dye trace at Dyers Spring, May 30, 1985
[(mg/D/kg, milligrams per liter per kilogram injected; (mg/§)/kg, milligrama
per tecond per kilogram injected]
Time
of day
10 15
21 45
22 IS
22 45
23 IS
23 45
15
45
1 15
1 45
2 IS
2 45
3 IS
3 45
4 IS
4 45
5 15
5 45
6 15
6 45
7 15
7 45
8 IS
8 45
13 45
22 45
Measured data
Measured
concentration,
in micrograms
per liter
0.010
.010
.060
.500
1.320
2.050
3.900
4.200
4.200
3.400
3.050
2.4SO
2.000
1.500
1.200
.950
.800
.600
.550
.500
.420
.370
.350
.300
.200
.010
Discharge,
in cubic
feet per
second
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
1.140
Observation,
in hours
1
2
3
4
S
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Elapsed
time,
in hours
0.0
11.5
12
12.5
13
13.5
14
14.5
15
15.5
16
16.5
17
17.5
18
18.5
19
19. S
20
20.5
21
21.5
22
22.5
27.5
36.5
Computed data1
Concentration
minus back-
ground, in
micrograms
per liter
0.000
.000
.050
.490
1.310
2.040
3.890
4.190
4.190
3.390
3.040
2.440
1.990
1.490
1.190
.940
.790
.590
.540
.490
.410
.360
.340
.290
.190
.000
Normalized
concentration
Umg/D/kg]
0.000
.000
.022
.220
.587
.914
1.743
1.877
1.877
1.519
1.362
1.093
.892
.668
.533
.421
.354
.264
.242
.220
.184
.161
.152
.130
.085
.000
Normalized
load
[mg/s)/kg]
0.000
.000
.723
7.086
18.945
29.502
56.257
60.596
60.596
49.026
43.964
35.287
26.779
21.548
17.210
13.594
11.425
8.533
7.809
7.086
5.929
S.206
4.917
4.194
2.748
.000
1Th« following summary statistics were computed from the data:
Mass of dye recovered, 2.232 grams;
Mean discharge, 1.140 (ft3/s);
Mean traveltime, 17.146 hours;
Elapsed time to peak concentration, 14.500 hours;
Standard deviation of traveltime, 4.370 hours;
Peak normalized concentration, 1.877 (mg/L)/kg;
Peak normalized load, 60.596 (mg/s)/kg;
Apparent mean velocity, 0.049 ft/a; _
First dispersion coefficient (01), 4.737 ftVs;
Second dispersion coefficient (D2), 0.981 ftz/s;
Skewness coefficient, 1.998.
63
-------
10 20 30
TIME FROM INJECTION. IN HOURS
0 10 20 30
TIME FROM INJECTION. IN HOURS
Figure 14.—Development of selected quantitative characteristics for
dye trace at Dyers Spring, May 30, 1985: A. Measured data prior to any
computations; B. Normalized dye concentration versus elapsed time from
injection; C. Normalized dye load versus elapsed time from injection.
The shaded region under the curve in 14C has an area of exactly one.
64
-------
Summary statistics computed from the data are shown at the bottom of table 5.
A summary of the method used in computing the quantitative characteristics
follows.
For computational purposes, 25 sampling intervals were obtained by
averaging values between the 26 consecutive observations, or samples. For
example, for the fifth sampling interval--the period of time between the fifth
and sixth observations--concentration was the mean of 1.32 and 2.05, less the
background, giving 1.67 ug/L or 1.67 x 10" mg/L. For a duration of 0.5 hours
and a mean discharge of 1.14 fts/s, the mass of dye recovered during the
interval, from equation 7, was 9.73 x 10" kg, or 0.0973 g. The same general
procedure was used for all subsequent calculations involving sampling
intervals.
The total mass of dye recovered during the trace, from equation 7, was
2.23 x 10" kg, or 2.23 g. Percent recovery, based on 3.57 g injected, was
62.5 percent.
Normalized dye concentration and load were computed for each observation,
using equations 9 and 10. Note that a quantity expressed in units of ug/L/g
has the same magnitude as in units of mg/L/kg. Thus, for convenience, in
equation 9 ug and g may be used for the computation, with the result given in
mg/L/kg. For example; for the sixth observation in table 5,
C - (2.05 - 0.01) 1/2.23 - 0.914 mg/LAg
and L - 28.32 (0.914) (1.14) - 29.5 mg/sAg
Graphs of normalized concentration and load versus time are shown in figures
14b and 14c, respectively.
Mean traveltime, standard deviation of traveltime, and skewness are
calculated from equations 11, 13, and 14 respectively. These equations are
based on the statistical definitions of mean, standard deviation, and skewness
coefficient. In equation 11, the time term t, is used to give the first
moment about the origin or, in other words, trie mean. In equations 13 and 14,
the time term (t. - t) refers to the n central moment of the data. The
term C. AtjQj is used to weight the time term by the corresponding mass of dye
for the interval.
As shown in table 5, the mean traveltime for the example was 17.15 hours.
Note that the mean traveltime occurred later than the normalized peak
concentration, which reached 1.88 mg/L/kg at 14.5 hours. A skewness
coefficient of 2.0 reflects the asymmetry of the curves in figure 14, which
shows the lagging effect of the passage of the trailing edge of the dye cloud.
Based on the apparent flow length of 3,000 ft, the mean-flow velocity was
computed from equation 12 as 0.049 ft/s, or 176 ft/hr.
Computation of the two estimates of the dispersion coefficient follows
from equations 15, 16, and 17, using quantities that have already been
determined.
65
-------
In this example,
D, -
2 17.15
A - 1.14/0.049 - 23.5 ft2, and
D2 - 346,400/(47T(14.5) [(1.88)(23.5)]2) - 0.98 ft2/s.
The different values for D. and D., assuming that sampling was adequate,
is primarily due to the skewness, or asymmetry, of the curve introduced by
turbulent action of the water body. The value of D. is primarily dependent on
the peak of the dye trace; D^ on the standard deviation, which is influenced
by the asymmetry of the curve. D- may be thought of as being more
representative of the molecular diffusion and D- of the turbulent forces. It
should be remembered that these are not precise estimates and that neither
value is necessarily more correct, but that these values may be compared to
values obtained from other dye traces.
5. APPLICATION OF QUANTITATIVE DYE-TRACE RESULTS
FOR PREDICTING CONTAMINANT TRANSPORT
5.1 Introduction
Because the movement of dye used for tracer studies is similar to most
conservative soluble contaminants, dye-trace results can be used to predict
the transport characteristics of such contaminants. For example, normalized
dye concentrations from a dye trace, in (mg/L)/kg, can be multiplied by a
given mass of contaminant to estimate the time-concentration response at the
sampling site that would result from a contaminant spill at the injection site
under the same hydrologic conditions.
The use of results from repeated dye traces extends the range of
hydrologic conditions for which the time-contaminant response may be
estimated. Furthermore, results from repeated dye traces may be compared and
analyzed to derive relations among the quantitative characteristics and the
hydrologic conditions as shown by Mull, Smoot, and Liebermann (1988), and
Smart (1981). If such relations can be established, then the interpretation
of the results, the understanding of the karst ground-water flow system, and
the ability to predict contaminant response are all strengthened.
5.2 Relations ffflPTIg Quantitative Characteristics
In most cases, discharge is the variable that characterizes the
hydrologic conditions associated with a dye trace. By use of data from
repeated dye traces, each quantitative characteristic can be plotted as a
function of discharge. Least-squares regression or another reliable method of
fitting curves to data may be used to establish a relation between each
characteristic and discharge. If a statistically significant relation can be
established for a given characteristic (such as mean traveltime), then its
value can be predicted for a given discharge. As with any empirically derived
relation, predictions should be used with caution, and may not be valid if
discharges are outside the range of the original data.
66
-------
As an example, relations were developed for quantitative characteristics
from seven dye traces at Dyers Spring (Mull, Smoot, Liebermann, 1988). A
summary of the characteristics is given in table 6. In the table, the second
estimate of the dispersion coefficient (D.) is shown. By use of standard
least-squares regression techniques, best-fit models were developed, that
used several transformations of discharge as the independent variable. The
best transformation for a given characteristic was chosen, based on the
greatest value of R2, where R is the correlation coefficient of the
regression. If several transformations resulted in close fits, the simplest
or most readily understood was chosen. For example, the inverse
transformation was chosen for mean traveltime, because traveltime is inversely
proportional to velocity. Figure 15 shows examples of the relations of the
quantitative characteristics to discharge developed from seven dye traces to
Dyers Spring. When appropriate, it may also be useful to derive relations
among various quantitative characteristics such as peak dye load as a function
of mean traveltime.
Once established, the relations may be used for prediction and to gain an
understanding of the movement and spread of solutes in the karst ground-water
system. Interpretations of the quantitative characteristics should be based
on the established relations and on an understanding of solute transport, both
in general and in a specific karst ground-water system. The quantitative
characteristics computed for a given dye trace result from complex
interactions within the flow system, mainly between the length of time that
the dye is in the system and the rate and pattern with which the dye
disperses. Although the best empirical fit among variables may differ for
different locations, it is likely that certain general relations will hold for
most karst ground-water flow systems. For a specific flow path, if discharge
increases from one dye trace to another, then:
(1) Apparent flow velocity will increase and the mean traveltime will
decrease;
(2) The dispersion coefficient will increase, because of increased
turbulence and mixing brought about from the increased velocity;
(3) Standard deviation of traveltime will probably decrease, because
even though the dye cloud disperses more rapidly, it has less time
to disperse;
(4) Peak normalized load may increase, because the time of passage is
shorter and the response curve is steeper and narrower; and
(5) Peak normalized concentration may decrease, because the dye is
diluted by the increased discharge.
To illustrate interpretation of a specific set of dye-trace results, the
relations derived from the dye traces at Dyers Spring, as shown in figure 15,
may be interpreted as follows. Discharge is the controlling factor of the
quantitative characteristics for the seven dye traces. The relations between
discharge, mean traveltime, and the dispersion coefficient may be used to
explain other dye-response characteristics. As discharge increases, velocity
increases (fig. 15a) and time of travel decreases (fig. 15b). Thus the dye
cloud has less time to disperse, which tends to decrease the standard
deviation. In the same circumstance, however, the dispersion coefficient
increases (fig. 15c); thus the dye cloud disperses more rapidly, which tends
to increase the standard deviation. The net result of these contrary
tendencies is that as discharge increases, standard deviation decreases
67
-------
Table 6.—Quantitative characteristics of seven dye traces at Dyers Spring
[(mg/L)/kg, milligrams p«r liter par kilogram injected;
(mg/i)/kg, milligrams par second par kilogram injactad]
Apparent
Mean mean
Mass Discharge, travel- velocity, Normalized
Data of dye in cubic time of in feet peak dye
of dye injected, feet per dye mass, per concentration
trace in grams second in hours second [(mg/L)/kg]
2-28-85 2.38 4.72 6. OS 0.138 1.12
3- 1-85 2.38 4.59 8.19 .135 1.55
5-23-85 3.57 1.35 14.6 .057 2.53
5-30-85 3.57 1.14 17.1 .049 1.88
7-16-85 2.38 .72 25.4 .033 2.23
8-12-85 2.38 .53 31.4 .027 2.48
2-26-86 7.14 2.88 7.18 .116 2.06
Standard Pispersion
Normalized deviation of coefficient
peak dye traveltime in square
load of dye mass, feet per
[(mg/s)/kg] in hours second
150 1.48 3.07
201 .82 1.61
96.5 2.17 .52
60.6 4.37 .98
45.5 3.69 .46
35.1 4.68 .37
168 1.46 1.46
(fig. 15d). This indicates that traveltime exerts a greater influence on the
relative amount of dispersion than does the dispersion coefficient. The same
interaction of traveltime and dispersion coefficient also affects the peak
value of the normalized dye load, or peak load. As discharge increases, the
peak load increases (fig. 15e); because the traveltime is shorter, the rate of
mass transport is greater. When considered as mass per unit volume, or
concentration, this increased load is diluted by the increased volume of
water, resulting in an overall decrease in peak concentration (fig. 15f). At
this location, the amount of downstream dispersion is dependent mainly on the
length of time since injection, which in turn is dependent on the travel
distance and discharge.
Interpretations such as these allow one to conceptualize the transport of
dye or other soluble materials within the karst system. This understanding
may then be applied to predict the behavior of solutes in the karst system
under different flow conditions.
5.3 Development and Use of Dimensionless Dve-Recovery Curve
Knowledge of quantitative response characteristics is useful in
predicting the peak concentration or mean traveltime of the dye cloud for a
given injection, but it is knowledge of the dye-recovery (time-concentration)
curve itself that allows prediction of variables such as the time to leading
edge and persistence--variables of importance to water-supply managers. This
section describes one procedure for developing a generalized curve that can be
used for predictive purposes.
Simulations of dye-recovery curves that are based on simple dispersion
theory yield dye-cloud concentrations that are normally distributed along the
68
-------
0.16
•t
CO
2 Q 0.12
£ O
0.08
u r-
O UJ
_i ui
UJ U.
— 0.04
0.00
LU O
2 2
uj
1s
35
30
25
20
15
10
u - 0.019 + 0.027 Q
RZ - 0.960
where u • »ean velocity of dye ••••
q • discharge
R • correlacion coefficient
I Computed
' Least-squares regression
1234
DISCHARGE, IN CUBIC FEET PER SECOND
I • 2.71 + 15.67 / Q
R2- 0.535
where 7 - mean travultlmc of dye mass
Q • discharge
R - correlation coefficient
B
I Computed
Least-squares regression
I
1234
DISCHARGE, IN CUBIC FEET PER SECOND
Figure 15.—Relation of selected quantitative characteristics to
discharge based on seven dye traces to Dyers Spring: A. Apparent
mean velocity; B. Mean traveltime;
69
-------
1
la
o to
UJ
CO O
a"
z
UJ
> en
IS
fel
M
°%
&
02 - 0.054 + 0.513 Q
R2 - 0.834
where D. • iceond dliper«lon coefficient
Q • diicharge
R - correlation coefficient
Computed
Leait-equerei regreftion
1234
DISCHARGE, IN CUBIC FEET PER SECOND
ff( - 3.50 - 1.62 In Q
R2 . 0.836
where at * »t*nderd deviation of
tr«veltlM
Q • dlecherge
R » correlation coefficient
• Computed
— Lea«t»equ*re» regrenlon
012345
DISCHARGE, IN CUBIC FEET PER SECOND
Figure 15 (continued).—Relation of selected quantitative characteristics
to discharge based on seven dye traces to Dyers Spring: C. Dispersion
coefficient; D. Standard deviation of traveltime;
70
-------
< a:
UI Q
>- 2
Q O
^j t.1,1
r*, on
ui
Q.
250
200
150
100
50
Lp • 71.7 + 71.0 In 0
R2 - 0.906
where L - peak dye load
Q - discharge
R • correlation coefficient:
I Computed
Least-squares regression
I • •
1234
DISCHARGE, IN CUBIC FEET PER SECOND
CO
O 8
— ui
og
ui ui
O Q.
il
UI —'
>- or
Q LJ
ui
Q_
2.5
1.5
0.5
C . 2.53 - 0.2d4 Q
R2 - 0.739
where C • peak dye concentration
Q - discharge
R - correlation coefficient
• Computed
— Least-squares regression
I , ... I
1234
DISCHARGE, IN CUBIC FEET PER SECOND
Figure 15 (continued).—Relation of selected quantitative characteristics
to discharge based on seven dye traces to Dyers Spring: E. Normalized
peak load; F. Normalized peak concentration.
71
-------
flow path. Transformation of a dye cloud that is normally-distributed in
space at a given time into a set of dye samples from a fixed location will
introduce a slight positive skewness but will not adequately reproduce the
skewed shape of observed time-concentration curves. The skewed shape results
mainly from unequal flow lengths and velocities along and across the flow
path. If in general, the shapes of a set of observed curves are similar, a
standardized recovery curve can be developed to simulate time-concentration
curves for the site under study.
In the absence of an adequate theoretical model or frequency distribution
that generates curves of the proper shape, a graphical solution can be
applied. Individual dye-response curves for which discharge is constant
should be rescaled so that the peak value of each curve equals one, the mean
] traveltime is zero, and the standard deviation of traveltime is one. A sample
of computer program code that will accomplish this is the program SCALE,
discussed in section 6.2. The resulting standardized curves all should be
plotted on the same axis. If the shapes and the skewness coefficients are
similar, then the curves can be overlaid and redrawn into a single,
representative dimensionless curve. This standardized, dimensionless recovery
curve can be used to derive response curves of dye or contaminant
concentrations or loads under constant discharge conditions.
An example of the development of a dimensionless recovery curve is shown
- in figure 16, using data from five of the dye traces at Dyers Spring.
Figure 16a shows normalized dye concentration as a function of time for the
five traces. For each trace, concentration and time were standardized using
the computer program SCALE (section 6.2), based on :
Cs ' C/Cpeak <•«• 18)
where C - standardized concentration
and ts - (t - r)/ |