-------
20-1
U
o>
o
Q.
O
15-
10-
Spring G
Spring I
i
Spring B
Spring K
JANlFEB I MAR I APR I MAY I JUN I JUL I AUG I SEP I OCT I NOV I DEC
1991
Figure 3. Thermograph of Variable-Temperature Springs
(Figures 2 and 3 are plotted with the same temperature scale)
However, there are similarities and differences in the data for
all the springs. The similarities are mostly related to the
lag-times, and the differences are mostly related to the
amplitude of temperature variation through the year. It is
important to point out that, from observations in the study and
the CVs of temperature, there is no evidence of any relationship
(similarity or difference) that is related to relative locations
of springs, or the amount of their flow, or the physical
appearance of the spring outlets.
The temperature lag-time, when it occurs, provides
information about the thermal diffusivity of the system or the
time required to displace the resident thermal signature of the
bedrock. The thermal diffusivity of a substance can be defined
as the thermal conductivity (cal/sec. cm. °C) divided by the
product of the heat capacity (cal/g. °C) and the density (g/cm3)
(Oke, 1978, p. 39) which can be reduced to and expressed in
units of cm2/sec. The degree of thermal diffusivity is
described as the depth of temperature penetration into the rock
(or the system) of each pulse of recharge water temperature as
it flows through the aquifer. Another way of relating to
thermal diffusivity would be to recognize that it is a measure
of how far each molecule of recharge water can "carry" its
temperature signature into and through the system. A purely
conduit-flow system would have a high thermal diffusivity and
would result in a short lag-time. A purely diffuse-flow system
205
-------
would have a low thermal diffusivity and would result in a long
lag-time.
Wigley and Brown (1976) describe a parameter called
the relaxation length, which is associated with the decay of, or
damping, of outside air temperature to an asymptotic (constant)
value within a cave system. The relaxation length is related to
the size of the passage and the velocity of the fluid moving
through it, and its roughness factor. More simply put, the
relaxation length is a distance inside a cave that the outside
air temperature effect can be detected (or measured) . Figure 4
shows a representative thermograph for a variable-temperature
spring (Spring G) , and a curve representing the daily mean air
temperature for Oak Ridge in 1991 (NOAA, 1991) . The daily mean
air temperature curve has been smoothed by using a seven-point
moving-average procedure for each data point. The thermograph
shows that maximum temperatures at the springs occurred about 50
days after the maximum daily mean air temperature for Oak Ridge
(NOAA, 1991) . Fifty days is thought to be more reasonable than
415 days (50 + 365); there are several possible temperature
oscillation cycles as explained by Lange (1954) .
Temperature (°C)
-•• ro c
..?.,,.!. .,.?,.,,!, ...1
V
- •'"'/•" '*" '*
> ' ' " ^ - ~
*r" f • ^ Spring G
•
• *
,» ^^x-Mean Air Temperature ""
• ^ •
JAN 1 FEB 1 MAR ' APR I MAY ' JUN ' JUL 1 AUG I SEP ' OCT ' NOV ' DEC
1991
N
Figure 4. Thermograph for Spring G Compared with Mean Air
Temperature for Oak Ridge, 1991
From the relaxation length theory of Wigley and Brown,
assuming a system with flow pathways of roughly equal
dimensions, a spring draining a conduit-flow system (a quick-
flow system) would transmit a temperature pulse deeper through
206
-------
the system, possibly to the spring; there would be a long
relaxation length. In the opposite way, a diffuse-flow system
(a slow-flow system) would not transmit the temperature pulse as
far, or would have a short relaxation length; the temperature
pulse would decay or be damped within the system before reaching
the spring.
As previously stated, the lag-times for all the
springs are essentially the same (with the exception of the
"noisy" data for Spring J) ; as Figure 5 shows, the constant
temperature springs have lag-times similar to the variable-
temperature springs, but, the amplitude of temperature variation
is smaller. Interpretation of various conceptual models of the
aquifer system suggests that the differences in the data could
be related to the systems being either conduit flow or diffuse
flow systems.
16'
o
o
1
2
&
.4)
13
Spring J
Spring A
Spring E
Spring D
Spring C
Spring H
12~1 JAN I FEB I MAR' APR ' MAY ' JUN I JUL ' AUG ' SEP ' OCT I NOV I DEC
1991
Figure 5. Thermagraph of Constant-Temperature Springs
(Showing expansion, for clarity of temperature
scale used in Figures 2 and 3)
Relating thermal diffusivity to the lag-times, if the
systems are different (i.e., conduit-flow or diffuse-flow) then
the thermal diffusivities of each system should be different, so
lag-times would be different also. The data do not fit such a
model, the lag-times are similar, so this hypothesis must be
rejected. Also, the lack of significant turbidity in the
springs after rain makes it obvious that they are predominantly
diffuse-flow springs.
207
-------
A second hypothesis applies the theory of relaxation
length. If it were valid here, then the differences in each
system could be explained by the different ratios of aquifer
length versus relaxation length. With such a hypothesis, the
variable-temperature springs would be closer to the recharge
area as compared to the constant-temperature springs. But as
previously stated, thermal diffusivity is a time-and distance
related unit, so a shorter aquifer length would cause a change
in thermal diffusivity, and would probably result in different
lag-times and different ratios between aquifer length and
relaxation length. Again, the variation in the data cannot be
explained by this hypothesis.
A third hypothesis involves mixing a slow-flow
(diffuse-flow, and at a constant temperature) component with
different amounts of a quick-flow (conduit-flow - variable-
temperature) component. This mixture would always include a
diffuse-flow component with a variable proportion of the
conduit-flow component present within each system. A mixture of
different proportions of the two components could produce a
certain temperature at the output, and still maintain acceptable
thermal diffusivity and relaxation length conditions for the
system.
Multiple temperature measurements at several springs
can aid in characterizing flow at a spring, or within an
aquifer, but is it likely that whatever impression is gained
about flow at them, i.e., whether a spring or an aquifer is
dominated by mostly diffuse flow or mostly conduit flow, the
flow is probably always a mixture of quick-flow and slow-flow
components. The response of a spring or springs to rainfall or
temperature change is only the end-product of responses of the
whole system and is only a gross generalization of the sum of
those responses; they include individual and different
responses to recharge, storage, and flow. It is highly likely
that no system is predominantly conduit-flow or diffuse-flow.
Rather, all systems are different forms of mixed-flow systems.
Temperature variation should be studied at springs
because it is a physical parameter and would be expected to vary
somewhat differently to specific conductance, which is a
chemical parameter, but would provide equally useful information
and would give insight into aquifer mechanics. Measurements of
specific conductance were not made concurrently with the 1991
temperature measurements and in retrospect they should have
been. The temperature measurements and study of their variation
should be supplemented with concurrent measurements of specific
conductivity and study of its variation. Such data are now
being collected for all the springs to allow comparison between
the CV of temperature with that of specific conductance. These
208
-------
additional data will be available from the author in mid 1992.
However, it can be seen from the (temperature) data described in
this paper that if many measurements are made at several springs
that drain the same or adjacent geologic unit(s) and/or
aquifer(s), such information can yield valuable information
about the characteristics of that aquifer. Such data that can
not be obtained from studying temperature variation at only one
spring, and probably can not be obtained at randomly located
wells, in part because of scale differences between monitoring-
well effectiveness and typical karst aquifer organization, as
suggested by Quinlan and Ewers (1985) and Ford and Williams
(1989, p. 210-211).
ACKNOWLEDMENTS
The author would like to thank Jim Quinlan, Ralph
Ewers, and Tom Aley for many useful discussions about the study.
Particular thanks is given to Chris Smart for his indispensable
help with the theoretical aspects of interpretation of the
results. The drafting skills of Brenda Altom are also greatly
appreciated.
REFERENCES
Aley, T. J., 1970. Temperature fluctuations of a small Ozark
spring, Caves and Karst, 12(4):25-30
Brown, M. C., 1972. Karst hydrology of the lower Maligne basin,
Jasper, Alberta, Cave Studies 13, Cave Research Association of
California.
Ford, D. C., and P. W. Williams, 1989. Karst Geomorphology and
Hydrology, Unwin Hyman, London, 601 p.
Lange, A., 1954. Rock temperature distribution underground,
Part II, Caves and Karst (7): 26-32.
McMaster. W., 1963. Geologic map of the Oak Ridge Reservation,
Tennessee. Prepared for Oak Ridge National Laboratory, (ORNL-
TM-713) .
Meiman, J. J., Ewers, R.O, and Quinlan, J.F. 1988.
Investigation of flood pulse movement through a maturely
karstified aquifer at Mammoth Cave National Park: A new
approach, Environmental Problems in Karst Terranes and their
Solutions Conference (2nd, Nashville Tenn.) Proceedings,
National Water Well Association, Dublin, Ohio, p. 227-263
209
-------
National Oceanic and Atmospheric Administration (NOAA), 1991.
Local Climatological Data for Oak Ridge, Tennessee
Oke, T. R., 1978. Boundary Layer Climates, Metheun, London, 372
P-
Pitty, A. F., 1979. Underground contributions to surface flow,
as estimated by water temperature variability, (in) Pitty, A.
F., (ed.) Geographical Approaches to Fluvial Processes, Geo
Abstracts, Norwich, p. 163-172
Pitty, A. F., 1976. Water temperatures in the limestone areas
of the central and Southern Pennines. Proceedings of the
Yorkshire Geological Society, 79:153-177
Quinlan, J. F., and R. 0. Ewers, 1985. Ground water flow in
limestone terranes : Strategy, rationale, and procedure for
reliable, efficient monitoring of ground-water quality in karst
areas. National Symposium and Exposition on Aquifer Restoration
and Ground Water Monitoring, Proceedings, 197-234, Worthington,
Ohio, National Water Well Association
Quinlan, J. F., Smart, P. L., Schindel, G. M., Alexander, E. C.,
Jr., Edwards, A. J., and Smith, A. R., 1991. Recommended
administrative/regulatory definition of karst aquifer,
principles for classification of carbonate aquifers, and
practical evaluation of vulnerability of karst aquifers.
Hydrology, Ecology, Monitoring, and Management of Ground water
in Karst Terranes Conference (3rd, Nashville, Tenn.)
Proceedings. National Ground Water Association, Dublin, Ohio.
[in this volume]
Roy, B., and Benderitter, Y. 1986. Transferts thermiques
naturels dans un systeme aquifere carbonate fissure peu profond,
Bulletin Societe geologique France, (8)t. II, no. 4, p. 661-666
Shuster, E. T., and White, W. B. 1971. Seasonal fluctuations in
the chemistry of limestone springs: a possible means of
characterizing carbonate aquifers, Journal of Hydrology, 14:93-
128
Smart, C. C., and Ford, D. C., 1986. Structure and function of
a conduit aquifer, Canadian Journal of Earth Sciences, 23(7):
919-929
Smart, P. L., and Hobbs, S.L. 1986. Characterization of
carbonate aquifers; A conceptual base. Environmental Problems
in Karst Terranes and Their Solutions Conference (Bowling Green,
Ky.) Proceedings. National Water Well Association, Dublin, Ohio.
p. 1-14
210
-------
Wigley, T. M. L., and Brown, S.L., 1976. The physics of caves,
p. 329-358 (in) T. D. Ford and C. H. D. Cullingford, eds., The
Science of Speleology, London, Academic Press.
BIOGRAPHICAL SKETCH
Gareth J. Davies is a hydrogeologist and quality assurance
manager with Geraghty & Miller Inc., Oak Ridge, Tennessee. He
received a B.S. from Millsaps College, and an M.S. from the
University of Southern Mississippi. He is a task group and
sub-committee member of the American Society for Testing and
Materials and has been interested in caves and karst research
for over 25 years, with particular interest in the karst of the
Upper Swansea Valley, South Wales, Great Britain.
Gareth J. Davies
Geraghty & Miller Inc.,
97 Midway Lane,
Oak Ridge, TN 37830
(615) 481-3000
211
-------
WATER TEMPERATURE VARIATION AT SPRINGS IN THE KNOX GROUP NEAR
OAK RIDGE, TENNESSEE
By Gareth J. Davies
Do you think the time lags you observed are actual travel times?
If so, why? If not, why not?
The lag times are not travel times. Rather, they provide
information about the thermal diffusivity of the system. The
lag-time is an average for the centroid of the whole basin that
drains to a particular spring. A dye-trace provides information
about travel time from one point to another, and is not
necessarily representative of the whole basin. If a lag-time
happens to coincide with a travel time inferred from a dye-
trace, e.g., Pitty (1976, 1979) the relationship is just that, a
coincidence. Two different processes are being monitored.
212
-------
THE EFFECT OF PETROLEUM HYDROCARBONS ON THE SORPTION OF
FLUORESCENT DYES BY ACTIVATED COCONUT CHARCOAL
Scott A. Recker
Delta Environmental Consultants, Inc.
Charlotte, North Carolina
Michael J. Carey
Delta Environmental Consultants, Inc.
Charlotte, North Carolina
Joe Meiman
National Park Service
Mammoth Cave National Park, Kentucky
ABSTRACT
The use of fluorescent dyes to delineate ground water flow-routes
and basin boundaries in karst aquifers has dramatically increased
in the past decade. Dye tracing techniques have been used in
conjunction with contaminant assessment investigations to determine
the extent of soil and ground water contamination caused by leaking
underground storage tanks.
Fluorescein and rhodamine WT, the two most common tracers, are
passively collected on activated coconut charcoal at discharge
points or in wells . The presence of petroleum hydrocarbons in
ground water may retard the sorptive efficiency of the charcoal by
competing for carbon attachment sites or by reducing the actual dye
concentration due to chemical alteration of the dye.
Solutions were prepared from ground water containing known
concentrations of benzene, toluene, ethylbenzene and xylenes (BTEX)
and known concentrations of both fluorescein and rhodamine WT. Half
of the solutions were analyzed for florescence intensity particular
to these dyes, using a scanning spectrofluorophotometer. The
second half of the solutions were used to immerse a sample of
activated coconut charcoal (6-14 mesh), mimicking a passive
detector. After a contact period of 48 hours, the charcoal was
eluted with a mixture of 42% 1-propanol, 38% ammonia hydroxide, and
20% distilled water for a period of 24 hours. The elutriant was
also analyzed for dye concentration intensity characteristics of
the dyes using a scanning spectrofluorophotometer. Comparison of
the fluorescent intensity in the dye/hydrocarbon solutions and the
elutriants enabled a determination of sorption retardation and
chemical alteration caused by the presence of the petroleum
hydrocarbons.
213
-------
INTRODUCTION
Fluorescent dyes are commonly used tracing agents for hydrogeologic
investigations. Heightened environmental awareness and expanding
government regulations have increased the need to conduct dye
tracing investigations in contaminated hydrologic systems.
Rhodamine WT (RWT) and fluorescein,two of the most commonly used
dyes, are often employed in conjunction with passive activated
charcoal detectors. Dye is introduced into a hydrologic system at
an input point and is collected at discreet discharge points
(springs) or wells. Historically, tracers have primarily been used
to amass information concerning the flow velocity and basin
boundary characteristics of free flow karst aquifers.
Investigations were rarely conducted in response to the loss of
organic contaminants. The need to conduct tracer investigations in
the presence of petroleum hydrocarbon contamination has increased
with the promulgation of CFR 40, Part 280, requiring hydrologic
investigations in response to releases from underground storage
tanks (UST's).
Activated charcoal is useful in tracer studies as it readily
adsorbs organic molecules. However, the charcoal is indifferent to
which organic it will sorb in the presence of multiple organic
compounds. Rhodamine WT and fluorescein are organic compounds that
adsorb readily to charcoal in a contaminant free environment. An
environment riddled with organic contaminants may compromise the
sorption efficiency of the charcoal wi€h respect to the dyes. The
result of a tracer study conducted in the presence of organic
contaminants may be inaccurate because the charcoal may
preferentially sorb petroleum hydrocarbons, rather than fluorescent
dyes.
The presence of organics, particularly petroleum hydrocarbons, may
also chemically alter RWT and fluorescein producing erroneous
tracer results versus if the organics were not present. The
purpose of this study was to assess the sorption capabilities of
activated charcoal in the presence of light petroleum hydrocarbons
(gasoline) and how hydrocarbons may possibly chemically alter the
respective dyes.
METHODOLOGY
Representative ground water and liquid phase hydrocarbon (LPH)
samples were collected from a gasoline contamination site in the
coastal plain of North Carolina. Samples were immediately cooled
to 4°C for shipment to a certified laboratory. Samples were
analyzed according to EPA SW-846 Method 602 modified for benzene,
toluene, ethylbenzene and total xylenes (BTEX). The samples
contained varying concentrations of total BTEX as listed in Table
1. Although a sample was also collected from a monitoring well
which contained floating LPH, its total BTEX concentration was not
determined.
214
-------
Table 1. Total BTEX concentrations (mg/L)
TOTAL BTEX CONCENTRATIONS
BTEX Free 0.0
Low Dissolved 3.17
High Dissolved 9.95
LPH 132*
* Based on similar samples from monitoring wells containing
LPH.
Representative ground water samples were split during collection.
One was analyzed by a laboratory and the other was used to prepare
the dye mixtures. Dye solutions of 50,000, 1,000, 1.0 and 0.5 ug/1
were prepared from ground water samples and distilled water. The
various dye solutions were split into two bottles. Half of the
solution was transferred into a bottle and kept in total darkness,
at a constant temperature of approximately 4° C. The other half of
the solution was transferred into a bottle with a packet containing
approximately 3.0 grams of activated charcoal and kept in total
darkness, at a constant temperature of 4° C. After a period of 48
hours the charcoal packets were removed and shipped with the
dye/ground water solutions for dye analysis.
Groundwater/dye solutions were analyzed on a Shimadzu scanning
spectrofluorophotometer with respect to RWT and fluorescein. The
50,000 ug/L solution was not analyzed because its fluorescent
intensity exceeded the capacity of the instrument. Results of the
analyses are shown in Table 2.
Table 2. Concentration of rhodamine WT and fluorescein elutriant.
RHODAMINE SOLUTION
(Concentration total BTEX (mg/L)
Dye Distilled 0.0 3.17 9.95 LPPH
Concentration Water
500 0.988 0.898 0.946 0.892 0.916
0.5 0.006 0.005 0.005 0.004 0.007
0.25 0.003 0.002 0.003 0.003 0.004
FLUORESCEIN SOLUTION
Dye Distilled 0.0 3.17 9.95 LPPH
Concentration Water
500 0.772 0.313 0.063 0.073 0.081
0.5 0 0 0.001 0.001 0.001
0.25 0 0 0.001 0.001 0.001
215
-------
Activated charcoal packets were placed in a mixture of 42% 1-
propanol, 38% ammonia hydroxide and 20% distilled water for a
period of 24 hours. The resultant solution, known as elutriant,
was analyzed on the Shimadzu scanning spectrofluorophotometer with
respect to RWT and fluorescein. Results of the analyses are
included in Table 3.
Table 3. Concentration of rhodamine WT and fluorescein solutions.
RHODAMINE ELUTRIANT
(Concentration total BTEX (mg/L)
0.0 3.17 9.95 LPPH
Dye Distilled
Concentration Water
500
0.5
0.25
1.455
0.004
0.004
0.892
0.008
0.003
1.599
0.005
0.004
Dye Distilled
Concentration Water
FLUORESCEIN ELUTRIANT
0.0 3.17
500
0.5
0.25
5.91
0.009
0.006
4.141
0.008
0.007
6.353
0.007
0.005
1.036
0.006
0.003
9.95
5.389
0.009
0.006
1.85
0.01
0.004
LPPH
3.473
0.1
0.007
RESULTS AND DISCUSSION
The 500 ug/L dye concentration used in this study is a higher
concentration than commonly occurs at aquifer discharge points.
Dye concentration of initial solutions versus dye concentration
measurements of resultant solutions were plotted as bar graphs for
each BTEX value. Graphs are presented in Figures 1 through 4.
Review of these figures indicate that at low dye solution
concentrations (0.5 and 1.0 ug/L), resultant dye concentration is
insignificant for both the dye solutions and for the elutriant.
Measured dye concentration appears to be within the range of
instrument fluctuation. Dye solutions at these chosen
concentrations were not useful in determining the chemical
alteration of the dye or the sorption capability of the charcoal.
Concentration from elutriant though cannot be considered purely
quantitative. Although care was taken to standardize charcoal
volume, no two dye packets will have the same sorptive
characteristics.
216
-------
50.8
5 0.7
K
£ 0.6-j
| 0.5-
c
u 0.4
$ 0.3-
Q
h 0.2-
t°:
K
0.5 1 1000
DYE CONCENTRATION OF SOLUTIONS (ug/L)
Distilled Water
High Dissolved
Clean Ground Water
LPH
Low Dissolved
Figure 1. Resultant dye concentration versus dye concentration in
prepared RWT solutions.
217
-------
0.8-
E °-5^
z
w
UQ.4
o
U0.3-
O 0.2-
5 0.1-
I °'
K
0.5 1 1000
DYE CONCENTRATION OF SOLUTIONS (u&/L)
Distilled Water
High Dissolved
Clean Ground Water
LPH
Low Dissolved
Figure 2. Resultant dye concentration in prepared fluorescein
solutions.
218
-------
0.5 1 1000
DYE CONCENTRATION OF SOLUTIONS (ug/L)
Distilled Water
High Dissolved
Clean Ground Water
LPH
Low Dissolved
Figure 3. Dye concentration of elutriant versus RWT concentration
of prepared solutions.
219
-------
1-1
7n
6-
5-
w
4-
r
K2-
W
0.5 1 1000
DYE CONCENTRATION OF SOLUTIONS (ug/L)
Distilled Water
High Dissolved
Clean Ground Water
LPH
Low Dissolved
Figure 4. Dye concentration of elutriant versus fluorescein
concentration in prepared solutions.
220
-------
RWT Solution
Dye concentration versus BTEX concentration was also plotted for
the 500 ug/L RWT ground water solutions. The plot is shown by the
bar graph which is Figure 5. Figure 5 illustrates the relationship
of chemical alteration of RWT with increasing concentrations of
BTEX. Rhodamine WT concentrations for these solutions range from
0.988 mg/L in distilled water to 0.892 mg/L in the 9.95 mg/L BTEX
ground water solution. The deviation in dye concentration between
these samples is not significant indicating little to no chemical
alteration of the dye with increasing concentration of dissolved
BTEX. The highest concentration of BTEX shows an intermediate RWT
concentration, also indicating that chemical alteration of RWT is
not a function of BTEX concentration.
0.0 3.17 9.95
BTEX CONCENTRATION (mg/1)
132
Figure 5. Dye concentration versus BTEX concentration in 1000 ug/L
RWT solution.
221
-------
RWT Elutriant
Results relating to the concentration of 1000 ug/L of each dye with
differing concentrations of BTEX were quite visible. Figure 6
illustrates a bar graph of dye concentration versus BTEX
concentrations for the 1000 ug/L elutriant. This figure
illustrates that dye concentration appears to be independent of
BTEX concentration. The sorption capability of the activated
charcoal, therefore, does not appear to have been affected by an
increase in BTEX concentration. This is intuitively contradictory
with the theoretical situation where charcoal sites should be used
up by the presence of additional organics other than RWT. The
highest dye concentration is evidenced by the elutriant from
charcoal in the highest BTEX concentration and the lowest in the
"clean" ground water sample.
0
0.0 3.17 9.95
BTEX CONCENTRATION (mg/L)
132
7
Figure 6. Dye concentration versus BTEX concentration in 1000 ug/L
RWT elutriant.
222
-------
Fluorescein Solutions
Figure 7 is a plot of dye concentration versus BTEX concentrations
for the 1000 ug/L fluorescein dye/ground water solutions. A
decrease in dye concentration is noted between the solution of
distilled water and dye versus the solutions of ground water and
dye. The difference in dye concentration may be an indication that
naturally occurring organics in ground water may act upon the dye
and reduce its concentration through chemical alteration.
Dye concentration ranges widely with a maximum of 0.772 mg/L in the
distilled water/dye solution to a minimum of 0.063 in the 3.17 mg/L
BTEX ground water/dye solution. Notice the marked decrease in dye
concentration in the presence of detectable concentrations of BTEX.
This drastic decrease signifies that the presence of petroleum
hydrocarbons may have a significant effect on the chemical
alteration of fluorescein in ground water. In other words the
concentration of fluorescein may be decreased as a result of the
chemical breakdown due to the presence of elevated levels of
petroleum hydrocarbons in the ground water.
0.0 3.17 9.95
BTEX CONCENTRATION (mg/L)
132
Figure 7- Dye concentration versus BTEX concentration in 1000 ug/L
fluorescein solution.
223
-------
Fluorescein Elutriant
Figure 8 shows the plot of dye concentration versus BTEX
concentration for the elutriant from charcoal soaked in the 1000
ug/L dye/ground water solutions. Values for dye concentration in
elutriant range from a maximum of 6.535 mg/L in the 3.17 mg/L total
BTEX solution to a minimum of 3.473 mg/L in elutriant from the
solution containing LPH. Review of Figure 7 also indicates a
decrease in dye concentration of 2.880 mg/L from the highest to the
lowest BTEX solution concentration. This decrease in dye
concentration with an increase in BTEX concentration suggests that
the efficiency of charcoal decreased with an increase in BTEX
concentration. This is contradictory to results from RWT. Each
dye, however, has a different molecular composition and structure
and may, therefore, be sorped to the charcoal at different rates
and/or different efficiencies in the presence of multiple organics.
In other words the charcoal may "prefer" RWT over petroleum
hydrocarbons. Fluorescein, conversely, may take a "back seat" to
the hydrocarbons and be sorped less efficiently had the
hydrocarbons not been present.
0.0 3.17 9.95
BTEX CONCENTRATION (mg/L)
132
Figure 8. Dye concentration versus BTEX concentration in 1000 ug/L
fluorescein elutriant.
224
-------
CONCLUSIONS
Rhodamine WT
The 1000 ug/L RWT dye/groundwater solution shows little fluctuation
in dye concentration. The presence of increasing concentrations of
petroleum hydrocarbons does not appear to affect or chemically
alter RWT. Elutriant from charcoal immersed in the 1000 ug/L RWT
solutions shows no distinctive dye concentration pattern. It is
unlikely, therefore, that the sorptive effectiveness of charcoal
has been compromised by the presence of the petroleum hydrocarbons.
Even in the presence of hydrocarbons it appears that carbon
effectively sorbs RWT. Charcoal sorption sites may "prefer" RWT
over constituent compounds of the petroleum hydrocarbons.
Use of RWT as a ground water tracer in an environment affected by
petroleum hydrocarbons is recommended. There appears to be little
if any chemical alteration of the dye and charcoal readily sorbs
the dye in the presence of hydrocarbons.
Fluorescein
Elutriant from charcoal immersed in the 1000 ug/L fluorescein
solutions exhibited a distinctive decrease in dye concentration
with increasing BTEX concentrations. This observation may be
significant in evaluating the sorption capabilities of the charcoal
with respect to this dye; however, the\apparent chemical alteration
of the dye must first be considered. " The 1000 ug/L fluorescein
dye/groundwater solutions displayed a drastic reduction in dye
concentration with increasing concentrations of BTEX. Based on
this information the dye seems to be drastically compromised in the
presence of even low concentrations of petroleum hydrocarbons.
Use of fluorescein as a ground water tracer in environments
affected by petroleum hydrocarbons is not recommended. Chemical
alteration of the dye may result in a negative trace that may have
been a positive one had the contaminant not been present. The
contaminant, therefore, may affect an area of the aquifer that was
previously traced as "safe" using fluorescein. A dye concentration
of 1000 ug/L is perhaps a higher concentration than is commonly
recovered at collection points. Perhaps at lower concentrations
more common to real situations, petroleum hydrocarbons may have a
more detrimental alteration effect on the dyes.
FUTURE STUDY
As with any decent real world scientific investigation, this study
seems to raise more questions than it has answered. Future
research using dyes in contaminated environments is necessary to
promote accurate, dependable, and reproducible tracer tests to
ensure protection of the environment and human health. Here are
several suggestions for further studies:
225
-------
A similar study should be conducted using a wider range and
number of dye concentrations. A range from 100 ug/1 to 1000
ug/1 in increments of 100 ug/1 is recommended;
Studies should be conducted using a wide range of organic
contaminants such as heavier hydrocarbons (diesel fuel, fuel
oil, jet fuel, kerosene, waste oil) organic solvents, and
organic metals complexes. The fluorescent wavelengths of the
contaminants should be assessed prior to the study; and
An actual ground water trace should be conducted in an aguifer
affected by a contaminant that was previously traced while
unaffected to assess the differences in dye interaction with
contaminants at various stage. Flow routes should be retraced
and confirmed using several different dye types.
226
-------
ACKNOWLEDGEMENTS
The authors would like to extend their thanks to Paula Recker for
providing the clerical support to make this publication a reality.
We would also like to thank Hiram Rutledge who took time out of his
already overloaded schedule to collect duplicate ground water
samples necessary to perform this research.
BIOGRAPHICAL SKETCHES
Scott A. Recker, a native of Cincinnati, completed a Bachelors
Degree in geology at the University of Cincinnati in 1985 and
received a Master of Science in hydrogeology from Eastern Kentucky
University in 1990. Scott is a member of the National
Speleological Society and his research centers around ground water
flow in the subcutaneous zone of karst aquifers. He is currently
residing in Charlotte, North Carolina and is employed as a
consulting hydrogeologist for Delta Environmental Consultants, Inc.
Scott Recker
3300 Circles End Road
Charlotte, North Carolina 28226
(704) 541-8191
Michael J. Carey is a Project Hydrogeologist at Delta
Environmental Consultants, Inc., in Charlotte, North Carolina. He
earned his B.S. in Geology (1985) at the University of Rhode
Island, and his M.S. in Geology (1990) at Eastern Kentucky
University. Most of his experience is related to subsurface
assessment and remediation of petroleum hydrocarbons. His
specialties include the investigation of karst aquifers utilizing
dye tracing techniques.
Michael J. Carey
8107 Tifton Road
Charlotte, NC 28226
(704) 542-6776
Joe Meiman, a Kentuckian, is currently the hydrologist at Mammoth
Cave National Park. His B.S. in Geology (1985) and M.S. in
Geology/Hydrogeology (1989) were earned at Eastern Kentucky
University. At the present, he is kept busy directing hydrologic
research at the park which includes: water quality monitoring, dye
tracing, three-dimensional schematic karst aquifer modeling,
research and development of monitoring equipment, and the general
quest of scientific knowledge.
Joe Meiman
P.I 0. Box 95
Mammoth Cave National Park, Kentucky 42259
(512) 758-2339
227
-------
THE EFFECT OF PETROLEUM HYDROCARBONS ON THE SORPTION OF
FLUORESCENT DYES BY ACTIVATED COCONUT CHARCOAL
Scott A. Recker
Delta Environmental Consultants, Inc.
Charlotte, North Carolina
Michael J. Carey
Delta Environmental Consultants, Inc.
Charlotte, North Carolina
Joe Meiman
National Park Service
Mammoth Cave, Kentucky
1. You exposed the charcoal to dye and BTEX simultaneously.
Would int not be more relevant to expose the charcoal to the
BTEX for several days and then expose it to the dye?
Yes, we agree that BTEX exposure may be the preferred method as
it would better simulate a real world situation. During every
dye trace it is necessary to place the detectors into
contaminated ground water prior to the dye drop. The exposure of
the detectors to BTEX during the dye's travel time is certainly a
factor which should be considered.
2. The fluorescence of fluorescein is widely known to be
greatly suppressed at pH's below about 4.5 and maximized at
pH's above 9.0. These reactions are reversible, with no
destruction of the dye. What data, if any, do you have on
the pH of your mixtures of BTEX in water? Without such data
is it possible that at least some of your apparent loss of
fluorescein is due to lowered pH? Also, why not dilute
samples with high concentrations of dye — as is standard
analytical procedure, with high concentrations of any
substance with any analytical technique?
Monthly pH measurements have been recorded in ground water
samples from this location at monthly intervals for a period of
three years. The range of pH is small and generally remains
between 7.6 and 8.0. It is unlikely that the prepared solutions
exhibited pH below this expected range.
Yes, the higher concentration samples should have been diluted to
allow for quantification. This information may have provided
more concrete results on the reaction of the dyes in these
environments.
3. I am distressed by the sloppiness of the design of the
experiments described. so many variables are unaccounted
for that it is highly questionable whether there is any
significance or reliability in any of your conclusions. Why
228
-------
did you present this paper NOW, rather than after a properly
designed series of tests? By presenting possible erroneous
results, you may inspire someone to conduct the testes with
adequate rigor, but you may also mislead both investigators
and administrators to perpetuate your possibly false
conclusions and innocently make wrong decisions concerning
spill investigations — how to conduct them and how to
interpret them. Rigorous experimental design requires more
thought than money or time.
Any and all constructive criticism is welcome, however, this
question does not appear to fall into the constructive category.
This paper was submitted at this time because of our strong
belief that these questions need to be addressed in the dye
tracing community. We have simply touched on the fundamental
issues using the resources we have available. Scientific study
is evolutionary and many workers fail to present experimentation
which yield anything but presupposed results. We felt it
necessary to break this mold and present our information at this
time. We would hope that our work would not mislead any workers
in that our conclusions are tentative and we have simply raised
questions for other workers to consider.
229
-------
230
-------
CONTINUOUS-FLOW FLUOROMETRY IN
GROUNDWATER TRACING
C.C. Smart
Department of Geography
University of Western Ontario
London, Ontario, N6A 5C2
Canada
S-E. Lauritzen
Department of Geology
University of Bergen
Allegt 41, 5000 Bergen
Norway
ABSTRACT
In dye tracing, continuous flow fluorometry limits contamina-
tion, significantly improves precision and accuracy and allows
immediate response to tracer appearance. However, it is logis-
tically more complex than simpler methods, and provides data rather
than actual water samples which may be critically important should
corrective processing be required. Disadvantages outweigh
advantages for most tracer applications. However, an attempt to
investigate the detailed evolution of tracer dye through a defined
water-filled conduit lent itself to continuous flow fluorometry and
led to some developments of the method.
The Jordtulla system in Svartisen, northern Norway drains a
27 km2 subarctic catchment through a 580 m largely phreatic conduit
averaging 5 m in diameter. The system drains directly from a large
lake, has clear well-buffered water and has been mapped by cave
divers. It thus provides an ideal site in which to determine the
relationship between conduit hydraulics, morphology and tracer
behaviour. Sampling occurred at the resurgence (Fiskepole), and at
a window (Mellomgrotta) part way through the system. Portable
alternators and automobile batteries powered pumps and standard
Turner Designs Model 10 Series Fluorometers. Water temperature,
fluorescence and stage were recorded on Campbell CR21X Micro-
loggers. Thirty three traces were made over 9 days, providing a
total of 50 breakthrough curves.
Fluorometer response was inadequate on the unusually steep
231
-------
rising edge of the breakthrough curve, demanding a complex logging
algorithm. The resulting data required substantial processing
before satisfactory breakthrough curves could be defined.
Fiskepole breakthrough curves are consistent and reproducible, but
the Mellomgrotta data showed a distinctive structure indicating
incomplete mixing. Subtle detail in individual curves appears to
reflect the style of tracer injection.
The heavy power demand of pump and fluorometer limit the
portability of the present system. Sophisticated data loggers
permit excellent control and measurement capability, but add yet a
further level of logistic complexity. Continuous flow fluorometry
remains justified only by stringent experimental or legal demands.
INTRODUCTION: SAMPLING STRATEGY AND PROJECT DESIGN
Strategies adopted in groundwater tracing are usually a
compromise between objectives and resources available. The tracer
preferred by most workers in karst terrain is slug injected dye
because it meets the fundamental requirements of safety, availabil-
ity (including legality in most cases), cost and detectability. A
range of sampling strategies are available for determining dye
return characteristics. These range in order of increasing
technology and information content from visual observation, through
charcoal detectors, discrete water sampling and continuous flow
fluorometry. The analytical system similarly ranges from visual
detection, through filter fluorometers to spectrofluorometers.
As a general rule, the more advanced technology can deliver
more precise and accurate data, but only at greater logistical and
financial cost. The fundamental question to be posed at the
inception of any tracer experiment is the information required.
Resource availability is a secondary question, although it may
prevent investigation of some questions.
The resources available should not dictate the methodology
adopted. On the one hand this may result in data inadequate to
answer the question being posed. On the other hand, the data may
be obtained at excessive expense. In addition, an over-technical
trace may be less reliable, because more complex tests present
more opportunity for error.
The majority of important tracer tests will be simple point-
to-point detector traces with the hydrogeographical objective of
determining tracer trajectories. More sophisticated traces demand
prior knowledge of the hydrogeography derived from point-to-point
tracing or well-judged hydrogeological reconnaissance.
However, there are legal and scientific situations where very
high quality data may be required. First, as hydrogeological
litigation evolves, demonstrably higher quality data will become
more important. This will provide a means of assuaging the
increasingly expensive confrontation, scrutiny and derogation of
232
-------
"expert" witness. The technology required, along with simple
quality-assurance-quality-control requirements, may make ground-
water tracing considerably more expensive. However, it seems more
rational to expend resources on a high calibre trace than in
vacuous litigation.
Scientific work may also demand high quality data either to
verify the technique or the system under scrutiny. In field
experiments, the difficulty of exerting effective scientific
control makes it critically important to minimise any error masking
the experimental variance. The remainder of this paper is a report
on the design and implementation of a continuous-flow fluorometric
system for field studies in remote areas.
CONTINUOUS-FLOW FLUOROMETRY
Modern field fluorometers are designed to allow water to pass
directly through the analytical cuvette, permitting a continuous
reading of water fluorescence. Normally, this water is drawn
directly from a spring or stream and a continuous trace of
fluorescence is printed on an analogue recorder (e.g. Turner
Designs 1978). There are problems and advantages inherent in this
system. These will be addressed with special reference to the
Turner Designs Model 10 Series Fluorometer which is the most common
and, despite its age, arguably the most suitable instrument
currently available for field fluorometry-
Calibration of the continuous flow cell requires rather large
quantities of dye standard, perhaps a litre or more. Usually, a
short hose length is used to fill the flow cell from the bottom.
Several rinsings may be required until a stable reading is obtained
on successive flushes. With high temperature coefficient dyes
(e.g. the "Rhodamines") it is essential to record the temperature
of the calibration standard.
In remote areas without A.C. line voltage, power consumption
is a serious consideration, which can significantly increase
logistical load. The Model 10 draws some 2 amps of power when
operating from a 12 volt source, and thus requires a rechargeable
battery source (usually one or more automobile batteries) and a
portable alternator with fuel (usually gasoline). Such instrumen-
tal energy consumption is appalling by contemporary standards, but
is dictated by the "chopper" and high voltage photomultiplier
technology which gives the instrument its impressive analytical
capability.
Instrument power demands are often overshadowed where it
becomes necessary to pump water up from a stream at a suitably
rapid rate. A heavy duty 12 v marine "bilge" pump will often draw
3-5 amps. A siphon can supply water at a suitable rate for zero
power, but such systems are inherently unreliable and their flow
rate is strongly dependent on stream stage. A combination pump-
assisted siphon is very efficient, allowing (once primed) a much
233
-------
lower power pump, providing a priming system is incorporated, it
is generally suggested that pumps should be placed downstream of
the fluorometer to avoid over-pressurisation. However, we have
found that in small diameter pipe (e.g. V garden hose), this can
result in degassing of water. Instrument noise increases and dye
detectability suffers greatly from the stream of bubbles produced.
The lightest fluorometer-alternator package with a lightweight
motorcycle battery weighs at least 30 kg. The system may operate
for only an hour or so without recharging. It is imperative in
these circumstances to have prior knowledge of likely travel times.
Where A.C. line power is available, the fluorometer will run
indefinitely. Line power instability may cause some interference,
e.g. low pressure mercury lamps (Alexander pers. comm. 1991). it
may be worthwhile to run the fluorometer from a 12 volt battery
continuously charged from an A.C. battery charger. This isolates
the fluorometer from line noise and improves precision. Fouling of
the analytical cuvette and calibration drift become the primary
long-term problems. These can be corrected on an itinerant basis.
Although it is possible to collect an excessive quantity of data,
over long duration tests, or those requiring high frequency
sampling, continuous flow fluorometry may be much less expensive
and less prone to error than equivalent discrete sampling pro-
cedures .
RECORDING
The simplest recording system is an analogue chart recorder,
matched to the instrument telemetry. A chart record provides real-
time "hard-copy" of a trace. This is valuable as a statutory
record. It also allows an immediate impression of trace progress
and is invaluable in managing sampling strategy, e.g. chart speed,
regional sampling patterns, sampling frequency and sampling
termination. However, the chart record includes all instrument
noise, can be mis-matched to the fluorometer and is often unreli-
able.
Modern digital data loggers are less expensive than analogue
chart recorders and considerably more reliable and powerful. They
are no more technically demanding than chart recorders, although
they do not produce any real-time hard copy; their memory and
recording media can be volatile. Work with both Campbell Scien-
tific and Aanderaa data loggers suggests that with suitable memory
back-up facilities, this is a mature technology with any modern
field logger system.
The single greatest problem with continuous fluorometry occurs
where the direct fluorescence reading is dependent on other
variables than tracer concentration. For example, cross-fluor-
escence from other dyes, background variations, temperature, pH and
turbidity controls on fluorescence. These problems normally
require a discrete sample so that some sample preparation or
further analysis can be undertaken before fluorescence can be
234
-------
determined. Many data loggers also permit data processing,
logical process control and data management. It may thus be
possible to correct readings to obtain a direct record of tracer
concentration. For example, it is a simple matter to record sample
temperature and correct to standard temperature. In general, it is
advisable to obtain required data on site, but to avoid on-site
processing because errors become more probable as processing
becomes more complex. In the worst case, this may invalidate the
entire record. Providing sufficient memory is available, it is
better to record essential parameters such as temperature, mass
injected etc. and to process the data after downloading to a
suitable computer system or spreadsheet.
A great advantage to logger processing is the capacity to
smooth the data by averaging. Measurements can be made with very
high frequency (perhaps every 0.1 s) and the average and standard
deviation recorded (perhaps every 5 minutes). Each recorded value
in this example is thus the average of 3000 readings. The signal
noise which obfuscates analogue chart traces is removed and,
providing there is no significant trend over the recording
interval, a very smooth (precise) record is generated. This
facility is of such great value at low tracer concentrations that
it is now being applied to the analysis of discrete samples.
Loggers can also be programmed to respond to measured data.
Typically, this will involve higher frequency measurement during
rapid evolution of tracer concentration and lower sampling rates in
the tail of a breakthrough curve.
Continuous flow fluorometry can generate a very large data
base compared to that possible with discrete sample analysis. This
is advantageous, unless there is a need for collection of physical
samples. Such a need may arise for reasons of legal evidence, or
more commonly because some form of sample processing is necessary
to determine correct dye concentrations. Background, turbidity,
pH, temperature and cross-fluorescence are examples of factors
which may require some pre-analytical processing.
The Model 10 combined with a data logger can provide excellent
field data in a typical trace. However, the Turner Designs
fluorometer is unable to automatically switch over its full dynamic
range, a major constraint. It is also possible to use only a
single filter combination with each fluorometer which limits its
value in multi-tracing. Fountain (1989 pers comm.) has developed
a logger-controlled hydraulic switching system which can route
water from a number of intakes into a single fluorometer. In
applications with very short rise times, the instrument is unable
to respond sufficiently quickly to rapid increases in fluorescence.
The need to switch through all ranges on rising fluorescence while
only a single range change is necessary on falling fluorescence is
a fundamental design flaw, albeit one of limited impact.
Logger data collected from continuous-flow fluorometers
provides the best data currently available on tracer breakthrough
235
-------
curves. In sites with high background or complex traces requiring
careful interpretation, spectrofluorometric analysis of grab
samples is advisable. Data acquired from very rapid sampling and
very rapid changes in fluorescence can be extremely complex to
process. It has proven worthwhile to record at two levels during
critical rapid changes; the first is the typical averaging
approach, the other is a direct recording of readings which can be
invaluable in interpreting the reliability of the averaged data.
A visual interactive editing programme has been found the most
useful way to process fluorometric data.
FIELD EXPERIMENTS
Earlier work on continuous flow fluorometry in the Canadian
Karst systems at Castleguard and Maligne are reported in Smart
(a,b). These and other (multi-tracer) experiments showed clear
advantages in continuous flow fluorometry, especially for revealing
precise detail in the breakthrough curve and for eradicating
contamination.
More recent work on has taken place on the Jordtulla conduit,
Svartisen, Norway (Lauritzen et al. 1985), an exceptional natural
laboratory for experiments on conduit flow. It is a 5 m in
diameter, 580 m long, singular conduit, which has been completely
explored and mapped by divers and (draining an extensive lake)
varies gradually in discharge between 1 and 50 m3s~1. A series of
33 tracer experiments were made in May 1989, a period of transition
from winter base flow to spring floods (Figure 1). Injections of
-20 g Rhodamine WT (Acid Red 338) were made at the sink point or at
one of two karst windows, Mellomgrotta and Waltergrotta. Fluor-
escence was measured continuously on two Turner Designs Model 10
Series Fluorometers powered from 12 v automobile batteries.
Samples were collected at Fiskepole, the resurgence and at
Mellomgrotta a karst window 150 m from the sink point. Fluor-
escence and sample temperature data were recorded on Campbell
Scientific CR21X Microloggers as one minute averages or 60 readings
and as 10-second direct readings during periods of high fluor-
escence. The data were segregated into discrete traces, corrected
to a standard injected mass and standard temperature and edited
using an interactive sorting routine.
Figure 2 shows an example of tracer breakthrough curves from
Mellomgrotta and Fiskepole. Note the extremely rapid rise of the
Mellomgrotta curve and clear instability compared to the Fiskepole
trace. When inspected in detail (Figure 3), the Mellomgrotta trace
shows a rapid decline in fluorescence background prior to arrival
of the dye. The latter problem results from desorption of dye from
the vinyl hose used to feed the fluorometer. It is significant
because of the high dye concentrations involved and the cessation
of pumping between traces. Adequate flushing time is essential
before dye arrival. The impact of such adsorption can be minimised
by continuous flushing of the shortest possible length of intake
hose. Provided each subsequent dye impulse is comparable in
236
-------
hO
U)
-J
n (D rt
o cn y
3 o ro
a ro
f- 3 o
ro
p-3 MI
N H"
D) ro
n o
3-
PI
rt ^-^ PJ
« 3 cn 3
n 3
rt •
C
O
5'« 2 ro
8g?"
•§8.eS
* * £ o
5.0) ®
rr ^1
CD *" El D)
a [J ro H-
o H ^ w
n a ro x1
g|8-8
ro IT ro o
3 ^ cn M
r+rt <">
Q, ^ H"*
• « cj:
!-•• Hi M
3 M ro
C rt
rt O
3- f-l O
ro i M)
JQ
Q.
Q.
C
O
o
0)
Q
30~
20-
10-
0
-4
138
140
142 144
Julian Date
12
-0
cn
L_
o
JZ
o
cn
Q
146
-------
o n n>
cr c:
to
LO
00
fl>
3
O
cn o
(D n
o
aw g
H- J
'"'• 3 O
W ^Q h-h
t->
g
1 — i
i~b i-j
8-S*
t?g s
H" 0) (B
3 ^ M
(D i u
\V ^-1
a o o
rt ^ iM
3- < n
t» ro o
3 ft
cn rt
rt 3* CD
3-0
(D £ CT
tn >-<
o> o DJ
rt O X
rt 3 rt
(D W 3*
• a o
(D C
I"X »/O
1 3"
Rn
DU
'S"
CL
v x^O -
C
o
•43 30 H
^•T
D
L_
-4-»
C
(D 20-
o
C
O
0 10-
CD
^^N
Q
0
1 1 1
FP17 & MG17
*• • \
. •
•
;
.
.
•
/^~\
•
\
N»
^ •' '.
* '.
• '^
' I 'I ' I I |
0
1 2 3
Time from Injection (h)
4
-------
K3
OJ
rt o a to
3* 0) fD
(o n o ,.
I-1- M
Ifl M H- *
ft 01 3
••< rt n> a
(D H-
OJ O
0) H-
.rt
0)
H-
M
(D
a
Q.
n
H)
o>
a> 5*
o- 1-
.^< o
(o g
C rt
3 O Qj
rt C o
rt
H- 01 tJ1
3 rfp fD
0.0 0.5 1.0 1.5
Time from Injection (h)
-------
magnitude to its predecessor, there should be no significant
impact. However, it may be possible to mimic subtle variations in
the detailed form of the breakthrough curve by adsorptive exchange.
The gross instability of the Mellomgrotta curve was reproduc-
ible in frequency and amplitude, but not phase on all traces, it
is considered to be an indication of incomplete mixing in the
system. It is thus a measure of the scale of turbulence in the
stream. Such irregularity in a continuous record can be used to
indicate inadequate mixing length in a tracer study.
The Fiskepole breakthrough curves were in general well-mixed
and showed little deviation from the anticipated form. Apart from
discharge, the single largest source of variation in form results
from the injection style. Although injection technique was
reasonably standardised, the exact pattern of secondary flows at
the injection site varied somewhat with discharge. A dye-trapping
eddy at the sink thus resulted in slight shoulders on the tracer
recession under higher flows.
Our technical conclusions from this experiment are that
despite the excellent technical procedures, there remain practical
limitations to the precision of field fluorometry, although the
error resulting from imprecision is far from significant relative
to overall variance. However, changes in the breakthrough curve in
response to discharge are very subtle over the range of discharges
investigated. For this system, therefore, extremely precise
procedures are required. Smaller diameter conduits or greater
changes in discharge are necessary for more satisfactory experi-
ments, but of course this may prevent complete exploration.
Further work will investigate the relationship of tracer
dispersion and advection relative to injection site and discharge.
Preliminary studies indicate that the limited free-surface
associated with the karst windows have a detectable effect on the
breakthrough curve. Models developed for a pure conduit, will thus
require modification to accurately represent the system.
CONCLUSIONS
A continuous flow fluorometer coupled to a good data logger
can provide excellent tracer breakthrough curves. The high
resolution of the record and lack of sample handling provide
precise curve definition and can minimise labour. However, in
remote sites, or where tracer concentration is not simply measured
the technique can become logistically complex, expensive and
inaccurate. However, recent development of less precise, but quite
adequate submersible fluorometers ( 1991) suggests that
multi-tracing to multiple springs may soon become affordable,
although still requiring considerable logistical effort.
240
-------
REFERENCES
Lauritzen, S.E., J. Abbott, R. Arnesen, G. Crossley, D. Grepperud,
A, Ive, and S. Johnson, 1985. Morphology and hydraulics
of an active phreatic conduit. Cave Science 12(4) 153-
61.
C.C. Smart, 1988a. Quantitative tracing of the Maligne Karst
Aquifer, Alberta, Canada. Journal of Hydrology. V.98
pp.185-204.
C.C. Smart, 1988b. Artificial tracer techniques for the determina-
tion of the structure of conduit aquifers. Groundwater.
V.26 pp.445-453.
PERSONAL RESUMES
Chris Smart obtained a BSc from the University of Bristol
(UK) in 1974 where his interest in karst and hydrology was sparked.
The statistical properties of meandering cave passages provided the
focus of an MSc gained from the University of Alberta in 1977. A
PhD. was awarded by McMaster University in 1983 following a study
on the hydrology of the partially subglacial Castleguard Karst
aquifer in the Canadian Rockies. Present research interests focus
on instrumented field studies in alpine terrain, including karst,
till, glacier and fracture aquifers. This work is complimented by
work on instrument development, data processing, mapping and
numerical modelling.
An organic chemist by training, Stein-Eric Lauritzen converted
an amateur interest in caves and karst into a profession. He is
currently involved in projects in karst hydrology and hydro-
chemistry, quaternary geology and cave archaeology, glacier
hydrology and karst resource conservation. He runs the Norwegian
Uranium-Thorium dating facility in the University of Bergen.
241
-------
Chris Smart - Continuous Flow Fluorometry
1. Given that most applications of continuous flow fluorometry will
probably be at sites where there is ready access to electric
power from the mains, I think you have overemphasized the
difficulties of the technique. Please comment.
Most springs and cave streams are remote from mains power. Therefore,
the arbitrary application of continuous flow fluorometry will be
constrained by the logistics of power supply. At those few sites
with safe mains power, the technique is straightforward; however,
other problems remain. First are technical problems of robust
data recording. Second are irrecoverable analytical problems which
may not be evident in the data record. Finally, problems associated with
lack of physical samples will also remain unless a sampler is operated
concurrently. Therefore, provision of mains power only overcomes
the most superficial, albeit immediate difficulties of continuous
flow fluorometry.
242
-------
DEVELOPMENT OF A FLOW-THROUGH FILTER FLUOROMETER FOR USE IN
QUANTITATIVE DYE TRACING AT MAMMOTH CAVE NATIONAL PARK
Martin Ryan
Division of Science and Resources Management
Mammoth Cave National Park
Mammoth Cave, Kentucky 42259
ABSTRACT
A series of quantitative traces were completed in the Buffalo
Spring ground water basin in Mammoth Cave National Park as part
of the field testing of a newly developed filter fluorometer.
The RME flow-through filter fluorometer is an inexpensive, labor-
saving, battery-operated, submersible device that, when
interfaced with a digital datalogger, is capable of precisely
measuring the travel time of two dye slugs (rhodamine WT and
fluorescein) simultaneously. It is also able to measure the
approximate dye concentrations passing a recovery point.
Interpretation of RME data yielded unprecedented information
concerning the hydrology of the Buffalo Spring basin—including
the unanticipated discovery of a major flow-route.
INTRODUCTION
Fluorescent tracer dyes are commonly used in the study of ground
water movement in karst terraries. Qualitative dye tracing, using
passive dye-detectors like cotton and activated charcoal to
recover the dye, is frequently employed to approximate ground
water flow-routes and define ground water basin boundaries.
Quantitative dye tracing, which requires the measuring of
changing dye concentrations at a recovery point, is useful in the
determination of ground water velocities, conduit condition
(phreatic or vadose), unexpected flow routes, and water "budgets"
(for basins with multiple discharge points). If the flow of dye
through an aquifer is closely documented using quantitative
tracing, models of soluble point source contamination events may
be generated. Such models may be used by ground water managers
to aid in drafting contingency plans for dealing with acute
ground water pollution.
243
-------
Quantitative dye tracing—much more expensive and labor intensive
than qualitative tracing—is generally performed only as a
supplement to qualitative tracing. Typical methods used to
recover dye include grab sampling, automatic sampling, and flow-
though fluorometry. Each of these methods has inherent
drawbacks. Grab sampling is enormously labor intensive.
Automatic sampling is moderately labor intensive and costly—
automatic samplers cost over $2000 each. Samples obtained using
grab or automatic sampling must be quantitatively analyzed on a
fluorometer. Fluorometer prices start at around $7000. Flow-
through fluorometers are able to directly measure concentrations
of dye at a recovery point; however, in addition to being
expensive, they require a pump to generate flow through the
instrument. The energy requirements of the pump and fluorometer
make this method impractical in remote areas where electrical
service is not available. A need exists for the development of
cheaper and easier techniques capable of obtaining results of
similar precision.
THE RME FILTER FLUOROMETER
An alternative to conventional dye recovery methods has been
developed and is being used extensively at Mammoth Cave National
Park. The RME flow-through filter fluorometer is an inexpensive,
battery-operated, submersible probe, supported by a digital
datalogger. It may be deployed in the field for extensive
periods of time, and requires only occasional servicing. The RME
uses 6 volts DC and draws less than 100 ma/hr. Since it is
submerged into the spring and is designed to slowly draw water
through itself, no pump is required. The RME is capable of
continuously measuring small concentrations of two dyes
simultaneously—rhodamine WT (C.I. Acid Red 388) down to 0.5 ppb
and fluorescein (C.I. Acid Yellow 73) down to 5 ppb. The
material cost of building an RME is approximately $175 per unit.
A datalogger with versatile programming is required to execute
and record the data measurements. The datalogger and RME battery
power supply, attached to the RME through waterproof wire, must
be placed above the highest possible water level in a
weatherproof enclosure.
The RME fluorometer is two filter fluorometers in one package.
It has one light source, a 4-watt clear quartz mercury
ultraviolet lamp, sandwiched between two flow-through sample
tubes (Figure 1). The flow-through tubes are made of 6 inch
sections of 1-inch ID aluminum box tubing. Two elongate windows
are milled into each tube at right angles to each other. Clear
microscope slide glass is mounted across each window from the
inside using a silicone sealant.
LIGHT FILTER SETS
Situated between the lamp and the rhodamine sample tube is an
244
-------
PHOTORESISTOR ARRAY
PHOTORESISTOR
4 WATT Hg LAMP
RHODAMINE TUBE
EXCITATION FILTER SET
FLUORESCEIN TUBE
EXCITATION FILTER SET
I INCH
Figure 1. Diagrammatic cross-section of the RME filter
fluorometer.
excitation filter set, composed of a Kodak Wratten 61 gel filter
sealed between two Corning 1-60 colored glass filters
(recommended in Smart and Laidlaw, 1977). This filter set is
designed to allow only the 546nm mercury line light to illuminate
the inside of the sample tube. The other major spectral lines
emitted by the mercury lamp (578nm, 436nm, 405nm, 365nm, and
254nm) are absorbed by the filter set. The 546nm light
illuminating the interior of the rhodamine tube is within the
excitation spectrum for rhodamine WT (its excitation maximum is
about 555nm), so if that dye were present in the sample tube, it
would be induced to fluoresce. An emission filter set, composed
of a Corning 3-66 and a Corning 4-97, is located between the
other window and the photodetective array. The secondary filter
set is designed to transmit a spectrum that has peak nearly
coinciding with the emission maximum of rhodamine WT (about
580nm); the filters are nearly opaque to wavelengths outside this
relatively narrow spectrum.
The excitation filter set for the fluorescein tube is a
combination Wratten 2A and a Wratten 47B. It transmits the 436nm
mercury line, which is within the excitation spectrum of
fluorescein (the excitation maximum is 490) and is nearly opaque
to the other lines. The emission filter, located between the
emission window and a photodetective array, is composed of a
Wratten 2A, a Wratten 12, and a Corning 4-97 (recommended in
Turner Designs, 1983). This filter set transmits a portion of
the excitation spectrum of fluorescein (the maximum is about
520nm) to the photodetective array. All Wratten filters are
sealed inside clear glass to help preserve them.
THE PHOTODETECTORS
The RME uses cadmium sulfide photoresistors as photodetectors.
The electrical resistance of a CdS photoresistor varies—in an
245
-------
inverse log relationship—with the intensity of light striking
it. The photoresistors are extremely sensitive even to tiny
changes in light intensity—especially in the 500nm to 600nm
range. Both of the RME's sample tubes have an array of three
photoresistors, connected in parallel and located outside the
emission filter (Figure 1). Another photoresistor, along with a
protective neutral density filter, is used to monitor intensity
fluctuations in the mercury lamp.
PROGRAMMING THE RME
To conserve battery power and to extend the life of the heat
sensitive Wratten filters, the lamp, and the lamp circuitry, the
lamp is only operated periodically. The datalogger (Campbell
Scientific 2IX microloggers were used by this investigator), via
a relay, switches the lamp on for one minute out of every ten.
At the end of that one minute the resistances of the rhodamine
array, fluorescein array, and the lamp reference are measured,
using a DC half bridge, and stored. If dye above a pre-chosen
concentration is sensed, the sampling interval will change to
once per five minutes and then revert back to ten minutes when
that concentration is no longer exceeded. This insures a better
probability of documenting short duration features.
Cadmium sulfide photoresistors have an undesirable inherent
behavior called a memory or light history. If they are placed in
total darkness for even a brief period of time (as they are when
the lamp is off), they will become "stuck" in this very high
resistance dark state; small increases in illumination will not
cause any change in electrical resistance. Minute increases in
dye concentration above background would therefore go unnoticed.
To counter this, an LED that keeps the rhodamine array slightly
illuminated is switched on when the lamp is switched off. The
fluorescein array receives enough exciting light through its
secondary filter (an otherwise negative trait) that, when the
lamp is switched on, the array is quickly "snapped out" of the
memory state.
RME ENCLOSURE
The electronics and optics of the RME are enclosed in a
watertight 4-inch schedule 40 PVC pipe compartment. The sample
tubes are connected to 3/4-inch PVC pipes which pass through the
endcaps of the compartment. To prevent ambient light from
reaching the sample tubes, light baffles made of 45 and 90 degree
ells are inserted between the sample tubes and the outside
(Figure 2). The front light baffles are removable to facilitate
cleaning the sample tubes. All piping and the enclosure itself
are painted black as further protection against ambient light.
On the downstream end of the RME exterior is an inverted funnel-
shaped feature called a drag inducer. When the RME is properly
oriented in a flowing stream, the drag inducer produces a vacuum
246
-------
— --- L'lVK, INDUCEfl
SAMPLE TUBES
(SHOWING WINDOWS)
WIRE CONDUIT
FRONT LIGHT
'-—' ~ BAFFLE
DIAGRAMMATIC CUT-AWAY VIEWS
OF RME
Figure 2. Diagrammatic cut-away views of the RME.
effect which draws water through the sample tubes. This insures
that the RME is taking a sample representative of the water
around it at any given time.'
SUMMARY OF HOW THE RME MEASURES DYE CONCENTRATION
The following is a very basic summary of the physical
relationships employed by the datalogger and the RME to measure
dye concentration:
1). A change in dye concentration results in a directly
proportional change in fluorescence.
2). A change in fluorescence results in a directly
proportional change in the amount of illumination striking
the CdS photodetector.
3). A change in the amount of illumination striking the
photodetector results in an inverse logarithmic change in
the electrical resistance of the photodetector.
4). A change in the electrical resistance of the array is
measured as a proportional change in the output voltage of a
DC half bridge by the datalogger.
The datalogger records the output voltages of DC half bridge
measurements, which are downloaded from the logger onto a
cassette tape or into a data can and then loaded into a PC
spreadsheet where the following transformations may be made to
it:
1). Using a conversion formula, output voltages are
converted to resistances.
2). Resistances are converted into dye concentrations by
interpolating from a calibration curve—calibration curves
are created prior to field deployment by plugging one end of
an RME's flow-through tubes, pouring in a series of
247
-------
standards, and recording the resultant resistances.
3). Concentrations are temperature compensated using the
formula provided by Smart and Laidlaw (1977) .
4). Instrumentational background is subtracted out by
"zeroing" data immediately proceeding the leading edge of a
dye slug.
5). Temperature compensated dye concentrations are
multiplied by the discharge of the spring or stream to
determine dye load.
6). The area under a dye load curve is calculated to
determine the total amount of dye recovered.
Because of deficiencies in the present RME design (primarily in
the light filter sets), fluorescein concentrations may only be
roughly determined; therefore, fluorescein may only be reliably
used with the RME in ground water time of travel study.
THE DETERMINATION OF THE HYDROLOGY OF THE BUFFALO SPRING
GROUND WATER BASIN USING RME FLUOROMETRY
INTRODUCTION
The Buffalo Spring ground water basin occupies about a 20Km2
portion of Mammoth Cave National Park, Kentucky. It is located
north of the Green River, just west of its confluence with the
Nolin River within the Hilly Country of the Chester Upland
(George, 1989). Buffalo Spring is stratigraphically located near
the middle of the Girkin Formation, the uppermost unit of a thick
section of highly karstifiable Mississippian limestone. An
alternating sequence of relatively thin Mississippian sandstones
and limestones and the basal Pennsylvanian Caseyville Formation
are located above the Girkin Formation. The regional dip is a
relatively gentle 5 to 15 m/Km to the west-northwest. The Buffalo
Creek surface drainage splits into its two main tributaries, the
Wet Prong and the Dry Prong, about IKm from the Green River.
Surface flow is absent in both of these branches where the Girkin
Formation crops out, except under high flow conditions. The
surface streams are lost through a series of sequential ponors
downstream from the upper Girkin contact. Many of the
tributaries to both the Wet and Dry Prongs also sink into the
upper Girkin.
Buffalo Spring is a rise pit type spring. It has a highly
variable discharge that ranges between about 60 and 1800 1/s,
with an average discharge of about 500 1/s. Qualitative dye
tracing by Meiman and Ryan (1990) confirmed that sinking water
from the Wet Prong and Dry Prong sequential ponors resurges at
Buffalo Spring (Figure 3). A large tributary to the Dry Prong,
Mill Branch, and numerous smaller tributaries to both Prongs were
also traced to Buffalo Spring. Qualitative dye tracing showed
that Confluence Spring was an overflow spring for Buffalo Spring.
Fort's Funnel is a cave located on the flank of Collie Ridge just
northwest of the Dry Prong (Figure 3) and containing a large
248
-------
DP332, DPIOO. AND
DPI03 TRACES
INJECTION POINT
-X- (^ L- WP33O TRACE
\ <^ ^ INJECTION POINT <-£,
\&
A i / CP^
KELLVS CUT-OFF
• MBI02 AND MBI03 TRACES
,.~ - INJECTION POINT
', FF330 TRACE
INJECTION POINT
«» BUFFALO SPRING V,. -
% - -' ^
' ', ', BUFFALO CREEK
RHSEIOI AND
RHSEI05 TRACES
INJECTION POINT
DRY STREAM BED
TERMINAL SINKPOINT
QUANTITATIVE DYE TRACE
I KM
» SINGLE CHANNEL RME
»«OUAL CHANNEL RME
Figure 3. Map of the Buffalo Spring ground water basin.
stream. The discharge of the cave stream is roughly half that of
Buffalo Spring. Every qualitative dye trace performed in the
basin was recovered with positive results at Fort's Funnel as
well as at Buffalo Spring and Confluence Spring (if Confluence
Spring was flowing). Since the discharge at Fort's Funnel was
considerably less than Buffalo Spring, the exact relationship
between the Wet and Dry Prongs and Fort's Funnel remained
problematic even after recovering numerous qualitative traces.
QUANTITATIVE TRACES
Table 1 is a summary of the quantitative tracer tests performed
in the Buffalo Spring basin using the RME filter fluorometer. A
dual channel RME, capable of recovering rhodamine and fluorescein
simultaneously, was placed at Buffalo Spring during all the
traces. _ Single channel RMEs, rhodamine sensitive only, were
placed in Fort's Funnel and Confluence Spring only during the
April, 1991 traces. Figure 3 shows the injection points
recovery points, and straight line travel routes for each trace
and Table 2 summarizes the results of each trace.
249
-------
i-n
O
Table 1. Summary of quantitative dye traces
performed in the Buffalo Spring basin
using the RME.
Table 2. RME-determined results of the
Buffalo Spring basin quantitative
traces.
Trace Injection Injection
number date site
DP100 4-10-91
DP103 4-13-91
DP332 11-28-90
MB102 4-12-91
MB103 4-13-91
RHSE101 4-11-91
RHSE105 4-15-91
WP330 11-26-90
FF330 11-26-90
Dry Prong
base flow
sink point
Dry Prong
base flow
sink point
Dry Prong
base flow
sink point
Mill Branch
quarry
ponor
Mill Branch
quarry
ponor
Raymer Hoi.
SE terminal
sink point
Raymer Hoi.
SE terminal
sink point
Wet Prong
base flow
sink point
Fort's Fun-
nel cave
stream
Recovery
point (s)
Buffalo Spring
Confluence Spring
Fort ' s Funnel
Buffalo Spring
Confluence Spring
Fort's Funnel
Buffalo Spring
Buffalo Spring
Confluence Spring
Fort's Funnel
Buffalo Spring
Buffalo Spring
Buffalo Spring
Confluence Spring
Fort's Funnel
Buffalo Spring
Buffalo Spring
Dye used/
Quantity
Rhod. WT/
150g
Rhod. WT/
150g
Rhod. WT/
119g and
Fluor./
150g
Rhod. WT/
150g
Fluor./
15 Og
Fluor./
150g
Rhod. WT/
150g
Rhod. WT/
119g
Fluor./
lOOg
Trace *Recovery Apparent
number point (s) travel
distance
Discharge Time to
(1/s) leading
edge
(meters)
DP100
DP103
DP332
MB102
MB103
RHSE101
RHSE105
WP330
FF330
BS
CS
FF
BS
CS
FF
BS
BS
CS
FF
BS
BS
BS
CS
FF
BS
BS
4320
4000
3120
4320
4000
3120
4320
3980
3630
2780
3980
4850
4850
4600
3750
2000
1200
875
150
395
950
460
770
280
940
360
700
960
930
1090
330
595
280
280
(hours)
12
11
8
10
8
5
28
11
9
7
9
14
12
11
7
11
18
.83
.13
.23
.00
.90
.75
.75
.27
.77
.67
.50
.72
.25
.85
.25
.92
.20
Time to
peak
cone.
(hours)
14.38
14.18
9.38
11.90
10.65
7.50
34.75
12.37
11.02
8.37
10.50
17.22
16.75
14.75
11.85
13.82
20.75
Approx.
peak
cone.
(mg/1)
.002
.005
.004
. 002
. 006
.010
.007
.004
.011
.024
-
_
.0005
.001
.0025
.100
-
*Buffalo Spring = BS
Confluence Spring = CS
Fort's Funnel = FF
-------
November Traces
Three quantitative traces were recovered at Buffalo Spring in
November, 1990. Flow conditions were low and relatively stable
during this period, and Confluence Spring was not flowing.
Figures 4 and 5 show the recovery curves of traces initiated
simultaneously from Fort's Funnel and the Wet Prong terminal
sinkpoint. The dye slugs were recovered using both a dual
channel RME and an automatic sampler. Surprisingly, the
rhodamine injected in the Wet Prong (WPS30 trace) arrived at
Buffalo Spring more than six hours before the fluorescein from
the much closer Fort's Funnel (FF330 trace) (Figures 4 and 5, and
Table 2). This shows that a primary flow-route exists between
the Wet Prong sink and Buffalo Spring with a gradient that is
significantly steeper than the flow-route between Fort's Funnel
and Buffalo Spring. Conseguently, a difference in head must
exist between this newly discovered trunk conduit carrying Wet
Prong water and the Dry Prong trunk visible at Fort's Funnel.
Enough of the Wet Prong trunk is apparently pirated by the Dry
Prong trunk above Fort's Funnel to be detected using qualitative
dye tracing methods, but not enough to cause a noticeable
secondary dye slug to appear at Buffalo Spring while using
quantitative methods.
The RME results (Figure 4) compared favorably with the ISCO
sampler/Shimadzu spectrofluorophotometer results (Figure 5). The
ISCO sampler, which was programmed to draw a sample hourly,
failed to sample the peak rhodamine concentration. The higher
resolution RME data shows that the peak rhodamine concentration
was considerably greater than what was determined by the
ISCO/Shimadzu methods. Figures 4 and 5 prove that the RME is a
capable alternative to conventional dye recovery methods. RME
and ISCO/Shimadzu results were also similar for a simultaneous
two-dye trace from the Dry Prong sinkpoint to Buffalo Spring
(DP332) .
April Traces
Six quantitative traces—two using fluorescein and four using
rhodamine—were performed in the Buffalo Spring basin in April,
1991. Flow conditions were much higher than in November and
fluctuated due to several moderate rainfall events received
during the study period. During the six day study period the
discharge was measured, using a tape measure, a survey stick, and
a Marsh-McBirney flow meter, five times at Buffalo Spring, four
times at Fort's Funnel, and five times at Confluence Spring.
Discharges listed in Table 2 for each dye slug at each recovery
site were interpolated from these measurements and are presumed
to be only moderately accurate.
If automatic sampling had been used as the dye recovery method at
all three sites for six days with a sampling interval of one
hour, it would have required changing 432 sample bottles and
251
-------
0.00
SI
M
200000 ^
240000>
70
280000 ^
3-
320000 »
360000
Z
O
M
O
•*)
Figure 4.
5 10 15 20 25 30 35
TIME FROM INJECTION (hours)
Recovery curves of FF330 and WP330 simultaneous traces
created from RME data.
x~x U.U /
X
gjo.06
£"J 2
| 2 0.05
C "^J
E *
Q £ 0.04
P> g
Q 5 0.03
< o
g | 0.02
o 3
to 0 0.01
i-^ O
T*
J-i
K 0.00
1 1
i i i i
RHODAMINE
-
_
-
_
-
. ~^^ WET PRONG
I "
i • -. FLUORESCEIN
; J FORT'S FUNNEL ~
*
.; .
\ M
\ • ^v
I ^ ^
\ ! V
J \ "•-
lNy •.
VJ. 1- ^i
O ~
ra S
o ^~
S a
z 5
£.
o fe
o a
0.2 O G
2 S
H 3
si
0.1 0 S
•y Z
p]
?g
(D
X
n n>-^
5 10 15 20 25 30
TIME FROM INJECTION (hours)
35
Figure 5
Recovery curves of FF330 and WP330 simultaneous traces
created from automatic sampler/spectrofluorometer
results.
analyzing 576 samples on the spectrofluorometer. All this toil
would have produced only mediocre results because each dye slug
would have been sampled only a few times (due to temporal
compactness of the slugs) and only a vague picture of a slug's
true shape would have resulted. The ten (or five) minute
sampling freguency of the RME insures that a more realistic view
of the dye recovery curve will be recorded.
MB102 Trace
Rhodamine trace MB102 was initiated at a discrete sinkpoint in
Mill Branch and was recovered at Fort's Funnel, Confluence
252
-------
Spring, and Buffalo Spring. The resultant recovery curves are
shown in Figure 6. It is apparent in that the peak
concentrations of Confluence Spring and Buffalo Spring are much
lower than Fort's Funnel. The Wet Prong trunk converges with the
Dry Prong trunk somewhere between Fort's Funnel and Buffalo
Spring and dilutes the dye-laden Dry Prong waters. Further
longitudinal dispersion was probably also a factor in this
reduction. Hubbard et al. (1982) contains an excellent
description of dye dispersion in streams during quantitative
traces. Evidently, an input from the Wet Prong trunk exists
between Fort's Funnel and Confluence Spring as well, and it is
responsible for the sizable drop in concentration at Confluence
Spring. Helpful "black box" type models of conduit systems like
these were presented and discussed by Brown (1973).
The total mass of dye recovered at Fort's Funnel during the MB102
trace was 122g, or 81% of the 150g injected. Only 45% of that
122g was recovered at the two terminal springs; the remaining 55%
was unaccounted for. Since the total discharge of the Buffalo
Spring system was above average and increasing, a plausible
explanation is that dye-laden conduit water moved into diffuse
storage adjacent to the conduit—like river bank storage.
Atkinson et al. (1973) describes this analogy in some detail.
Significantly, very strong positive results for rhodamine and
fluorescein were still being found on the passive dye detectors
at Buffalo Spring under low flow conditions more than three
months later. It is possible that dye in conduit adjacent
diffuse storage was slowly released as the summertime base flow
condition was approached. Other factors that may have
contributed to the apparent dye loss may have been adsorption or
the use of inaccurate (too small) discharge values.
g
h-
<
h-
Z
UI
O
Z
o
o
S
o:
I
0.024
0.020
0.016
0.012
0.008
0.004
0.000
FORT'S TUNNEL
CONFLUENCE SPRING
- BUFFALO SPRING
': 700
\ 360 l/i
940 l/i
6 9 12
HOURS FROM INJECTION
15
Figure 6. Recovery curves from MB102 trace,
253
-------
MBI02 TRACE
WEI PRONG TRUNK
I50g
FORT'S FUNNEL 0=700 l/s
CAVE I22g
CONFLUENCE 0=360 l/s
SPRING 2Sg
BUFFALO 0=940 l/s
SPRING 29g
GREEN RIVER
Figure 7. Summary of MB102 trace.
Figure 7 summarizes the dye recovery results for the MB102 trace.
The percentage of dye (and flow) going to each spring was
computed by considering the total mass of dye recovered at both
terminal springs 100%, then the mass recovered at either one of
the springs was divided by the total recovered at both springs
and multiplied by 100. The discharge of the Wet Prong trunk was
computed by subtracting the discharge at Fort's Funnel (which was
assumed to be the entire Dry Prong trunk flow) from the combined
Confluence Spring and Buffalo Spring discharges. Based on this,
approximately 47% of the discharge from Fort's Funnel resurged at
the overflow route—so about 329 l/s of Confluence Spring's 360
l/s discharge came from the Dry Prong trunk. The remainder of
the Dry Prong trunk flow and nearly all the Wet Prong trunk flow
resurged at Buffalo Spring.
DP103 Trace
Figure 8 shows the recovery curves at the three recovery sites
for the DP103 dye trace. When compared with the recovery curves
from the MB102 trace, in which the same amount of rhodamine was
injected, several differences are discernable: the peak
concentrations are all lower, the travel times are less, and the
slugs are more dispersed longitudinally. Because the discharge
was greater during this trace, the first two are believable—even
though the reduction in concentration was larger than such an
increase in discharge would warrant. An increased longitudinal
dispersion, however, is the exact opposite of what would
typically be expected for a trace initiated under higher flow
conditions.
The reason the DP103 slugs were more dispersed than the MB102
slugs may be related to the fact that DP103 dye entered the
subsurface through three widely spaced sequential ponors instead
of through one discrete ponor. All three Dry Prong traces were
254
-------
g
5
o
o
o
g
i
CC
a
Ld
0.010
0.008
0.006
0.004
2 Q.Q02
2
en
0.000
FONT'S FUNNEL
CONFLUENCE SPRING
BUFFALO SPRING
450 l/s
950 l/s
0
3 6 9 12 15
HOURS FROM INJECTION
Figure 8. Recovery curves from DP103 trace.
initiated from the same point—just above the Dry Prong base flow
terminal sinkpoint. However, the flow conditions were very
different for each trace: the terminal sinkpoint for the DP332
trace was the base flow sinkpoint, the terminal sinkpoint for the
DP100 trace was about 600 meters downstream at a ponor called
Kelly's Cut-off, and for the DP103 trace the terminus of surface
flow was a huge ponor 250 meters further downstream called Norain
Cave (Figure 3). Dye from the DP103 injection entered the
subsurface at all three of these major ponors. As a result of
this the injected slug was split into three separate slugs, each
with a slightly different route to follow at first. Flow from
the three separate inputs eventually reunited and the three
slightly out of phase dye slugs were fused back together—
slightly more dispersed and with a lower amplitude than a single
input slug would have been.
Only 75g of the 150g of dye injected (50%) was recovered at
Fort's Funnel. The amount of that dye which was recovered at the
terminal springs was 73%. The results, summarized in Figure 9,
suggests that 55% of the total discharge passing Fort's Funnel
went to the Confluence Spring (about 423 l/s) and 45% went to
Buffalo Spring (about 347 l/s). The remainder of each was
supplied by the Wet Prong trunk.
When the DP103 trace results (Figure 9) are compared to the MB102
results (Figure 7) several important insights into the behavior
of this aquifer may be gleaned: the Confluence Spring waters are
mostly derived from the Dry Prong trunk, and the Wet Prong trunk
is not well connected to Confluence Spring. Thus the Confluence
Spring is predominately an overflow spring for the Dry Prong
trunk. If the discharge were increased in both the trunks
simultaneously, hydraulic damming by Wet Prong waters, which
basically have no place else to go but Buffalo Spring, would
cause a decrease in the percentage of Dry Prong water resurging
255
-------
DPI03 TRACE I5°8
WET PRONG TRUNK
DRY PRONG TRUNK
Xr
J FORTS FUNNEL 0=770 l/s
CAVE 75g
CONFLUENCE 0=45O l/s
SPRING 30g
BUFFALO 0 = 950 1,'s
SPRING 25q
GREEN RIVER
Figure 9. Summary of DP103 trace.
at Buffalo Spring and an increase in the percentage overflowing
at Confluence Spring.
Repeated Quantitative Traces as a Predictive Tool
The input-to-resurgence travel time of a dye slug decreases with
increasing discharge. Peak concentrations often decrease with
increasing discharge because dilution increases, and longitudinal
dispersion decreases due to decreased dye slug travel time.
Figure 10 illustrates the results of a trace from Dry Prong to
the terminal spring(s) repeated three times under different flow
conditions. The aforementioned effects of increased discharge
are very clear. Using results from these three traces, reliable
predictions for almost any set of flow conditions could be made
concerning travel time, peak concentration, and dispersion for a
soluble contaminant accidentally injected into the Dry Prong.
Mull et al. (1988) gives a detailed discussion of this important
topic.
Determination of Conduit Condition
Conduit condition may be resolved eveji if a conduit is
inaccessible by using quantitative tracing; this was done for
segments of the Dry Prong trunk using the RME. The discharge of
a vadose conduit is increased by increasing the flow velocity
and/or the cross-sectional area of the channel (by increasing
stage). The only way to increase the discharge of a phreatic
conduit, since it is completely full and stage cannot be
increased, is by increasing the flow velocity. So, when log
discharge (X) is plotted versus the log travel time (Y) for a
series of traces through a phreatic conduit the result would be a
line with a slope of nearly -1 (Smart, 1981). A plot of traces
through a vadose conduit would be a line with a slope of less
256
-------
than -1.0 (but probably greater than -0.3). Figure 11 shows
first order linear regressions of log discharge versus log travel
time for the three Dry Prong traces recovered by the RME for, the
entire Dry Prong, the segment of the Dry Prong trunk upstream
from Fort's Funnel, and the segment of Dry Prong trunk downstream
from Fort's Funnel. Judging from their slopes, which are
admittedly based on a paucity of data, the segment downstream
from Fort's Funnel is apparently mostly phreatic and the segment
upstream is mostly vadose.
0.008
BS=BUFFALO SPRING
CS=CONFLUENCE SPRING
Figure 10. Recovery curves of Dry Prong to the terminal
spring(s) traces.
LJ
o
Q
U
o 10
UJ
2
10U
_0 ENTIRE DRY PRONG CONDUIT
-T— UPSTREAM FROM FORT'S FUNNEL
-V-- DOWNSTREAM FROM FORT'S FUNNEL
1CT 10J 10*
BUFFALO SPRING DISCHARGE - (l/s)
Figure 11. Log-Log plots of discharge vs. time of travel for
various segments of the Dry Prong trunk.
257
-------
Hydrologic Structure of Buffalo Creek
A pictorial summary of the hydrologic structure of the Buffalo
Spring karst ground water basin was generated by synthesizing all
the qualitative and quantitative trace data and geomorphological
data collected (Figure 12). Smart (1988) and Smart and Ford
(1986) presented a structural model of the Castleguard conduit
aquifer and laid the groundwork for this type of aquifer
representation. Models like these could be quite useful to
ground water managers charged with determining a course of action
during an accidental contamination event.
SUMMARY AND CONCLUSIONS
The RME is an inexpensive alternative to conventional
quantitative dye recovery methods. Extensive fieldwork in the
Buffalo Spring ground water basin, including some in conjunction
with traditional dye recovery methods for comparison, proved that
the RME is a useful dye quantification tool for field study.
Through use of the RME, subtle details concerning the hydrology
of Buffalo Spring basin were recognized and described including
several previously unknown ground water flow routes. Also
generated was new information about the relationships of the
primary spring and the over-flow spring to the two primary feeder
trunks and the response of aquifer transmissivity to changes in
discharge. Interpretation of RME data helped to identify the
phreatic and the vadose portions of the Dry Prong trunk conduit.
A structural model of the Buffalo Spring basin was produced using
all the available dye tracing data.
BASE FLOW
SWtPOCHT
WET PRONG TRUNK
1—1"
MILL BRANCH TRUNK
DRY PRONG TRUNK
/""RAYMER HOLLOW
SE TRUNK
:EU.Y'5 CUT-OFF
-BUFFALO SPRING
Figure 12. Hydrologic structure of the Buffalo Spring ground
water basin.
258
-------
ACKNOWLEDGEMENTS
Dr. Ralph Ewers of Eastern Kentucky University provided to me the
initial idea of developing a submersible fluorometer supported by
a digital datalogger. Joe Meiman, my colleague at Mammoth Cave,
wrote the datalogger program used by the RME and contributed many
hours of assistance during development and testing. Mammoth Cave
National Park provided financial support, laboratory facilities,
and other resources. I am extremely grateful to them all.
REFERENCES
Atkinson, T.C., Smith, D.I., Whitaker, R.J., and Lavis J.J.,
1973, Experiments in tracing underground waters in
limestone: Journal of Hydrology, v. 19, p. 323-349.
Brown, M.C., 1973, Mass balance and spectral analysis applied to
karst hydrologic networks: Water Resources Research, v. 9,
p. 749-752.
George, A.I., 1989, Caves and drainage north of the Green River:
in W.B. White and E.L. White (eds.), Karst Hydrology:
Concepts from the Mammoth Cave Area, Van Nostrand Reinhold,
New York, p. 189-221.
Hubbard, E.F., Kilpatrick, F.A., Martens, L.A., and Wilson J.F.
Jr., 1982, Measurement of time of travel and dispersion in
streams by dye tracing: U.S. Geological Survey Techniques
of Water-Resources Investigations, Book 3, chap. A9, 44p.
Meiman, Joe and Ryan, M.T., 1990, Preliminary results of
groundwater dye-tracer studies north of the Green River,
Mammoth Cave National Park: Proceedings of Mammoth Cave
National Park's First Annual Science Conference: Karst
Hydrology, p. 137-142.
Mull, D.S., Liebermann, T.D., Smoot, J.L., and Woosley, L.H.,
Jr., 1988, Application of dye-tracing techniques for
determining solute-transport characteristics of ground water
in karst terranes: EPA 904/6-88-001, Atlanta, Ga., U.S.
Environmental Protection Agency, 103p.
Smart, C.C., 1988, Artificial tracer techniques for the
determination of the structure of conduit aquifers: Ground
Water, v- 26, No. 4, p. 445-453.
Smart, C.C. and Ford, D.C., 1986, Structure and function of a
conduit aquifer: Canadian Journal of Earth Sciences, v. 23,
p. 919-929.
Smart, P.L., 1981, Variations of conduit flow velocities with
discharge in the Longwood to Cheddar Rising system, Mendip
Hills: in Beck, B.F. (ed.), Proceedings of the VIII
259
-------
International Congress for Speleology, Bowling Green,
Kentucky, p. 333-335.
Smart, P.L. and Laidlaw, I.M.S., 1977, An evaluation of some
fluorescent dyes for water tracing: Water Resources
Research, v. 13, no. 1, p. 15-33.
Turner Designs, 1983, Fluorometric Facts—Fluorescein, Bulletin
No. 103, 3 p.
BIOGRAPHIC SKETCH
Martin Ryan is a Hydrologic Technician at Mammoth Cave National
Park. Current research includes developing and implementing new
quantitative fluorometric techniques to model flow in karst
aquifers, delineating ground water basins north of the Green
River in the Mammoth Cave area, and performing water quality
monitoring. He received his BS in geology from Eastern Illinois
University in 1987 and he is currently pursuing a MS in geology
at Eastern Kentucky University-
260
-------
Development of a Flow-through Filter Fluorometer for Use in
Quantitative Dye-tracing at Mammoth Cave National Park
by Martin T. Ryan
I understand that inclusion of a turbidity measurement was
contemplated for an earlier model of your flow-through
fluorometer. Why did you decide not to monitor turbidity?
The nephelometer/turbidity meter prototype we were experimenting
with had severe problems with water leakage. Additionally,
laboratory experiments had showed that the design of the
instrument, which used a jumbo LED with a Fresnel lens and two
photoresistors, was adequate only for approximation of turbidity
(which is itself only an approximation of suspended sediment
concentration—the maligner of dye fluorescence.) Even if a
successful turbidity meter had been deployed alongside the
fluorometer, I do not believe that a turbidity-compensating
mathematical formula could have been generated. Rather,
turbidity data may have been useful in helping to explain
anomalies in the fluorometric record by intuitive comparison—for
example, if the mass of dye recovered during a quantitative trace
is unusually low, a partial explanation may be available if the
turbidity data is examined and found to be exceptionally high
(eg. because a brief storm event) during the time period
including the recovery of the dye slug.
We are still trying to develop a nephelometer/turbidity meter
that is interfaceable with a digital datalogger, but it is not
urgently required to validate data acquired through RME
fluorometry — rather it may, at times, be a helpful supplement.
261
-------
262
-------
Session IV:
Hydrogeology and Processes Occurring in
Karst Aquifers
-------
264
-------
Stable Isotope Separation of Spring Discharge in a Major
Karst Spring, Mitchell Plain, Indiana, U.S.A.
Barbara L. Gruver and Boel C. Krothe
Dept. of Geological Sciences
Indiana University
ABSTRACT
Major-ion chemistry and isotopes of oxygen and deuterium
where used to identify components of water resurging at
Orangeville Rise, a large perennial spring. The Orangeville R^se
basin IB recharged by precipitation that infiltrates a 125 km'
sinkhole plain and upland area developed on Mississippian
limestone and clastic sequences.
In October, 1990, 53 mm of rain fell on the basin in a 40
minute period. Isotopic composition of the rain was -8.2%0and
-56%-for 6 I8O and^D respectively. Prior to this, at baseflow,/
TD and 6D of Orangeville Rise was -6.27oc.and -407oorespectively,
and specific conductance was 770 uS. Discharge increased rapidly
from 0.3 nrs to 3.4 m s , however major ion concentrations did
not show a rapid drop which would have been the case had an
influx of rainwater been responsible for the increase in
discharge. Rather, the most dilute waters were found 24 hours
after peak discharge. Isotopic composition of waters also
reflected this trend. Isotopic separation of storm hydrograph
Into pre-storm and storm water components revealed that 75% of
discharging water consisted of pre-storm water.
Similar monitoring was conducted during April, 1991, when 4
rains, totaling 38 mm produced an increase in discharge from 1.6
to 4.9 ills'1. £T) of these rains ranged from -7 %o to -45%,,
and 5 TD ranged from -2.67oo to —6.8%0. Again, majoi—ion and
isotope chemistry did not reflect a rapid influx of rainwater,
and isotopic hydrograph separation showed that rain water
comprised only 20% of the total volume of discharge. These
results seem to indicate that water stored in the epikarst may
discharge at Orangeville Rise prior to the arrival of the mass of
rain water.
UTTRODUCTIOH
Discharge from springs in karst terrains responds to
influxes of precipitation in many ways. The response is governed
by the type of recharge (concentrated or dispersed), as well as
the routes of subsurface flow within the basin (conduit or
diffuse). It is also highly dependent upon the volumes of water
stored in the unsaturated and saturated zones.
Historically, individuals have studied karst systems by
265
-------
mapping cave passageways, and by conducting dye-tracing studies.
Analysis of storm hydrographs from karst springs has also been
performed, concentrating on recession limb behavior CMangin,
1974, 1975; Torbarov, 1976; Milanovic, 1976; and Atkinson, 1977).
Major—ion analyses have proven useful in discerning
properties of a karst system, and have been used by numerous
investigators, (e.g., Shuster and White, 1971; Ford and Ewers,
1978; Friedrich and Smart, 1982; Gunn, 1983; and Scanlon and
Thrailkill, 1989). Changes in water qiaality associated with
storm hydrographs of karst springs have been helpful in
interpreting sources of water which contribute to discharge
(Williams, 1983).
Stable isotopes have been used to show that new rainwater
can displace old water in storage in karst systems (Bakalowicz et
al., 1974). The two were found to mix in different proportions
to create fluctuations in A O across the storm hydrograph.
Dreiss (1989) separated storm discharge from karst springs in
Missouri, into new event water and pre-storm water on the basis
of calcium and magnesium ion concentrations.
This study looks at water quality changes associated with
storm hydrographs at Orangeville Rise, a perennial spring in a
karst region of Indiana. Goals of this study include quantifying
the proportion of rainwater contributing to storm discharge
through the use of oxygen and deuterium isotopes, accounting for
observed water chemistry changes across the hydrograph, and
attempting to identify subsurface storage zones which contribute
to discharge at Orangeville Rise.
Geologic and Hydrologic Setting of the Study Area
Orangeville Rise is a perennial spring located in south-
central Indiana, approximately 150 km south of Indianapolis
(Figure 1). The ground water basin supplying water to
Orangeville Rise, as defined through dye-tracing studies (Murdock
and Powell, 1968; Bassett, 1974), has an areal extent of 115 W
and lies within two physiographic provinces, the Mitchell Plain
and the Crawford Upland.
The Mitchell Plain is a flat-to-gently-rolling lowland
surface developed on middle Mississippian limestones of the Blue
River and Sanders Groups (Figure 2). These formations consist of
dense, thinly bedded, highly fractured limestones which have
experienced profound solutional modification to produce features
typical of karstic terrains. Indeed, the Mitchell Plain is known
for its extensive karst topography expressed In a myriad of
sinkholes, caverns and blind valleys.
The Crawford Upland occupies a region where alternating
formations of sandstone, shale and limestone of the Mississippian
Chester Series crop out at the surface. These formations are
more resistant to erosion than underlying Blue River and Sanders
limestones and hence form an upland surface that stands 45 to
60 m above the Mitchell Plain. The Crawford Upland is a rugged,
deeply dissected surface. Karst features (i.e., sinkholes and
sinking streams) have developed where stream erosion has out
through Chesterian Strata to expose Blue River limestones. As a
result, the hydrologic behavior of Orangeville Rise is dictated
266
-------
I
048 kilometers
Mitchell (Sinkhole) Plain
Crawford Upland
Figure 1. Location of the Orangeville Rise, and physiographic
provinces within the basin.
West Baden/ Slephensport Groups
Blue River Group
Sanders Group
0 5 miles
i i i i i
5 milt
048 kilometers
Figure 2. Geologic Map of the Orangeville Rise basin.
267
-------
by the karstic nature of the Blue River bedrock.
% - 1
Discharge at Orangeville Rise ranges from 0.06 nt s (2 cfs)
to 5.1 m3s"' (180 cfs). The Rise is situated within a bedrock
alcove forming a pool that is approximately 15 m in diameter, and
at low-flow is 6 ra deep. The Rise is fed by two vertical
conduits which are 6 m long. Scuba divers descending these
shafts, found them to connect to narrow horizontal conduits at
their base. Discharge from Orangeville Rise forms a perennial
surface stream that occupies an alluviated channel 5 m deep.
At low flow, when suspended sediment loads are low, waters
in the discharge pool take on a greenish cast. After
precipitation events, discharge increases rapidly. Depending on
the magnitude of the storm, the pool surface may rise 3 m or
more. The waters become heavily laden with suspended sediment,
and a bailing of the pool surface can be seen as water is forced
up through the vertical conduits.
Ho perennial surface streams discharge at Orangeville Rise.
.Rather, discharge is derived predominantly from precipitation
that infiltrates soils of the sinkhole plain, with a secondary
component coming from sinking intermittent streams draining the
Crawford Upland.
Soils of the Orangeville Rise basin consist of clay and clay
loam derived as residuum from underlying limestones, as well as
residual material produced during erosional retreat of the
Crawford Upland, and thin loess deposits (Ruhe, 1975). These
soils are susceptible to fracturing due to their high clay
content, creating macropores for rapid infiltration of
precipitation to the subsurface (Wells and Krothe, 1989). Soils
are up to 5 m thick, but are locally much thinner In many
places, particularly around sinkholes, soil material has been
completely removed by erosion, creating another avenue for rapid
subsurface recharge.
Southern Indiana has a humid-temperate climate, with a mean
annual rainfall of 1130 mm. Precipitation Is distributed through
the year, such that there is a pronounced dry time in the fall
when monthly average precipitation is 70 mm, and a wetter time in
late winter through spring, when average monthly precipitation is
115 mm. July is the warmest month on average, with mean daily
maximum and minimum temperatures of 90°and 64° F respectively.
In contrast, January is the coldest month, when mean daily
maximum and minimum temperatures are 43°and 25 ° F respectively.
Previous Investigations at Orangeville Rise
Bassett (1974) has identified the waters from Orangeville
Rise to be calcium-bicarbonate in character, and he saw an
inverse relationship between discharge and total dissolved
solids. Subsequent studies have confirmed these observations
(Libra, 1981; Tweddale, 1987). The carbonate aquifer(s)
supplying water to Orangeville Rise have a strong conduit
component, however, it is more complicated than a simple system
of conduits. Based on major-ion chemistry and sulfur isotopic
composition, Krothe and Libra (1983) differentiated two flow
systems within the ground water basin. One is a shallow conduit
system, dominated by surface flow entering the subsurface through
268
-------
large fractures. Discharge at Orangeville Rise responds rapidly
to precipitation by water entering the aquifer via this route. A
second system consists of a deeper diffuse network of
interconnected fine fractures and joints which respond more
slowly to inputs from precipitation.
BACKGROUND
Conceptual Model for Ground Water Storage
A simplified model illustrating water storage compartments
for Orangeville Rise is shown in Figure 3. The model is simple
in that it includes no impermeable bedrock or caprock, and
recharge occurs only from precipitation that falls within the
basin. These identified compartments could exist in many karst
systems, and have important ramifications for supplying water as
discharge to any fcarst spring.
Recharging precipitation may be temporarily stored in the
vadose zone (unsaturated zone) as soil moisture, epikarst
storage, diffuse vadose storage, or vadose conduit storage.
The volume of water stored as soil moisture at any one
moment is highly variable and dependent on soil properties of
depth, texture and composition, as well as on the timing and
volume of precipitation events. After extensive drought
conditions, soil moisture will drop to a minimum. When
precipitation events are large, or follow in rapid succession,
soils may become completely saturated. Recharging water from
precipitation moves through soil horizons as an advancing front,
acting like a piston to displace and replace previous soil
moisture.
01
Figure 3. Model of water storage compartments of the vadose and
phreatic zones in the vicinity of Orangeville Rise.
269
-------
Water which passes through the soil zone may reside for some
time in epikarst storage. Mangin (1974, 1975) defined epikarst
as a region in the upper weathered layers of rock at the base of
soil horizons, lying above the permanently saturated (phreatic)
zone. Williams (1983) described the formation of epikarst as
follows: Water flowing through soil dissolves COj from the soil
atmosphere, and becomes aggressive toward underlying limestones.
As infiltrating water encounters bedrock, its ability to dissolve
the rock is at a maximum. These aggressive waters preferentially
flow along prominent fractures, dissolving bedrock along the
fracture, making thus a zone of maximum dissolution, and
producing substantial secondary porosity. As the water flows
deeper into the fracture, its capacity to dissolve is
diminished, and the fracture narrows with depth, limiting the
rate of water transmission to deeper zones. Following
precipitation events of high intensity and/or large volume, a
substantial body of water can accumulate within the epifcarst, and
form a region where water saturation exists, resulting in a
perched water table.
Water will drain from the epikarst along solutionally
modified fractures, but will also move into the rock mass through
the system of interconnected fine joints and fractures.
Depending on the amount of primary porosity and the extent of
fracturing and jointing within the bedrock, the volume of water
within diffuse vadose storage may be minimal or qtiite large. The
degree of interconnectedness of the Joints and fractures will
also govern the rate at which water can move into and out of this
compartment.
There exists a transition zone between diffuse vadose and
diffuse phreatic storage, as well as between vadose conduit and
phreatic conduit storage. It is well documented that changes in
water table elevation in karst terrains can be quite dramatic,
and can occur on a seasonal cycle, or following precipitation
events. The magnitude of change in conduits can be on the order
of tens of meters or more, and can occur at an alarming rate
after heavy rainfall. The response is slower and more subdued
within the diffuse bedrock compartments. Because conduits flood
quickly and drain more rapidly than the diffuse zones, the two
areas do not appear to be in equilibrium, and it is therefore
difficult to speak of a water table in karst terrains. •
nonetheless, it is helpful to visualize water storage in terms of
the vadose and phreatic zones since it lends an element of
elevation or position within the subsurface.
Stable Isotopes in Ground Water Stxidies
Stable isotopes of oxygen and hydrogen bound in a water
molecule, under low temperature conditions, are altered only by
physical processes such as diffusion, dispersion, mixing and
evaporation, and may therefore behave in a conservative manner.
As a result, these isotopes can act as tracers and their analysis
may aid in determining the geochemical history of a water mass.
Stable isotopic analysis of waters may also provide information
on movement and mixing of water bodies, provided that the
isotopic composition of each water body differs by at least the
270
-------
analytical error of Isotoplc ratio identification.
In this study, stable isotopes of oxygen and hydrogen are
used in an attempt to identify two time-source components of
discharge at Orangeville Rise: 1) water that existed within the
basin prior to a storm event and 2) new rainwater that fell
within the basin.
If it is assumed that discharge at Orangeville Rise is
derived by simple mixing between two components: pre-storm water
(Qpr) and rainwater (Q >, then two mass balance equations can be
written that describe the water flux and the isotope flux at the
Rise:
where Q. le the total measured discharge at any instant in time
and is defined by the sum of the two components. Delta notation
represents the °O or deuterium (D) composition of the
instantaneously measured discharge G>^>, the rainwater ($p ), or
pre-storm water (Spe) • By combining these two equations, a third
equation can be written which identifies the rainwater
contribution to discharge at any instant in time, in terms of the
isotopic composition of the measured discharge, pre-storm water
and rainwater.
To use this equation as a means of separating discharge into
rainwater and pre-storm water components, the isotopic
composition of rainwater must be significantly different than
pre-storm water, and the pre-storm water should have an
identifiable and uniform isotopic composition throughout the
basin. This technique has been applied to surface streams by
Fritz et al. (1976) and Sklash and Farvolden (1979), where the
pre-storm water was interpreted to be ground water.
In this report, stable isotopic composition of water is
reported relative to standard mean ocean water (SMOW) , in parts
per thousand (or premil) notation, such that:
"
O.D]
p
where R is the ratio of D/H or ' O/ 0, depending on which ratio
is being identified.
271
-------
FIELD IITVESTIGATION
The field investigation was conducted in two stages- During
the first stage, water was sampled on a bi-weekly schedule to
establish baseflow chemical characteristics. It was critical
during this stage to determine the baseflow or pre-storm isotopic
composition to be used in the mase-balance equations Water fro.
domestic wells was also sampled during this stage of the
investigation to establish isotopic composition of the ground
water within the basin.
The second stage of sampling concentrated on short-term
changes in water chemistry associated with changes in discharge
brought on by two storm events in the Orangeville Rise basin: in
October, 1990, corresponding to the driest time of the year; and
in April, 1991, during the wettest time of the year ^en
recharge to the carbonate aquifer is a maximum (Bassett 1974).
Sampling frequency during this stage of monitoring was keyed to
thePrat! of discharge increase and decrease: every two to four
hours during the time of rapid discharge increase, peak flow and
initial discharge recession, and less frequently during later
stages when discharge decreased less rapidly.
RESULTS
Results from the bi-weekly sampling confirm the results o±
previous investigations. Discharge at Orangeville Rise is
calcium-bicarbonate in character, and there is an inverse
relationship between discharge and total dissolved solids. It
was also found that as discharge approached baseflow, the water s
isotopic composition repeatedly approached -40 7<*>and -6.2&o for
4'D and b 18O respectively. These values were used as the isotopic
signature of pre-storm water in hydrograph separation of spring
discharge.
Ground water isotopic composition was found to be quite
uniform, even though there was variable major-ion chemistry in
the basin (Table 1). The average 6'D and S I8O of ground water was
found to be -44 %0 and -6.6%0 respectively, and is similar to that
of the average isotopic composition of rainwater in southern
Indiana CSheppard et al., 1969).
October, 1990 Storm
On October 4, 1990, 53 mm of rain fell on the Orangeville
Rise basin in a 40-minuter period. Rain was sampled from two
locations in the basin to check for areal variability in the
amount and isotopic composition of the rain. The amounts of
rainfall at each station were identical, and isotopic
compositions were within analytical error, and were averaged.
Table 1. Selected chemical data from domestic wells located in
the Orangeville Rise basin. < n = 26 )
PARAMETER
Specific Conductance (uS)
Calcium (mg/l)
Sulfate (tng/l)
£°smow <*•>
* *0»mow «J
RANGE
408 (o 1985
70 to 388
13.9 to 976
-43to-46
-6.4to -63
AVERAGE
668
113
IZ5
-44
-6.6
272
-------
Discharge
(m3/s)
0 12 24 36 48 60 72 84 96 108
Time - Hours
Figure 4. Stnrm hydrograph for Orangeville Rise,
October 4, 1990.
leotoplc composition of the rainwater was -56%oand -8.2?c,ofor
SD and &I6O respectively. Isotopic composition of the rain may
have shifted during the course of the 40 minute storm, however
since the duration of the storm was short in comparison to
variable travel time for water in the entire basin, the potential
shift was considered insignificant for the purpose of this study,
and a bulk isotopic composition was determined.
Prior to this storm, baseflow discharge at Orangeville Rise
was 0.3 nTs1 and specific conductance of discharging waters was
770 uS. Calcium, magnesium, sulfate, and bicarbonate
concentrations were at the highest levels measured at any time
during the study. . .
Discharge increased rapidly to a maximum of 3.4 itfs within
6 hours after the storm (Figure 4). There was no precipitation
during the next four days, and discharge underwent a steady
exponential decrease to 0.3 nrs ' , with on exception between 26
and 36 hours, where there was a small perturbation in the
otherwise smooth recession.
Results of chemical analyses for waters collected across
rising and receding limbs of the hydrograph are shown in Figure
5. It can be seen that major-ion concentrations slowly decreased
during the first 8 hours, even though discharge increased
dramatically. Concentrations were not at their minimum until 24
hours after peak discharge, when the HCOj concentration had
decreased to 75% of its baseflow value and SO^ dropped to only
21% of its baseflow value. These minimums were preceded by a
small yet significant increase in ionic concentration for several
species, which corresponds to the small irregularity in
recession. Ion concentrations remained low during much of the
recession, and only started to recover in the last 24 hours of
monitoring.
Isotopic ratios remained fairly constant during the rapid
rise in discharge, and showed a small decrease during peak
273
-------
Discharge (m3/s)
3.5
72
96
BOO
750
- 700
650
-600
^550
500
450
Specific
Conductance
Time - Hours
B
Discharge
(rrvVs)
Discharge
-20
-16
-12
-8
- 4
72
Time - Hours
96
HCO3
(mg/T)
20
16
•12
-8
g21, Na*
(mg/l)
,cr
(mg/l)
Figure 5. Chemical data from Orangeville Rise, October 4-8, 1990:
A) specific conductance, B) cation concentrations, and
C) anion concentrations.
274
-------
^ OS
Discharge 2 0.
(tn'/s)
24 48 72 96
Time - Hours
Figure 6. Isotopie composition of discharge at Orangeville Rise,
October 4-8, 1990.
discharge and Initial discharge recession (Figure 6). Oxygen and
deuterium shifts mirrored one another during a period of rapid
isotopic fluctuations between 18 and 40 hours, when two pulses of
Isotopically light water discharged at the Rise. During much of
the discharge recession, isotopic composition remained steady,
and approached that of baseflow values.
April 1991 Storm
Chemical variability of discharge was monitored during a
second storm event in the spring, 1991 (Figure 7). Because of
frequent rainfall in the winter and spring, discharge at
Orangeville Rise does not achieve baseflow conditions.
Therefore, to establish pre-storm conditions for this sampling,
Orangeville Rise was monitored for 6 days prior to the storm
event. Major-Ion concentrations and isotopic composition of
discharge remained fairly steady at the values illustrated in
Figures 8 and 9 for time equal to zero-hours.
As in the October storm, precipitation was collected at two
locations within the basin. Again, the amount and isotopio
composition of the two rainwater samples were similar, and were
averaged. Rainfall history during this event was much more
complicated than the October storm, and is outlined in Table 2.
Because of the complex rainfall history, discharge response was
significantly different from that observed in October (Figure 7).
275
-------
Table 2.
Isotoplc composition of rainfall, April 12-15, 1991,
Orangevllie Rise basin.
Date Time
4-12 / 1930
4-1 2/2330
4-13/0430
4-14/0645
4-15 /0500
Amount
(mm)
3
7
19
3
6
•ID
(X.)
-12
-18
-7
-9
-45
i'«0
<%.)
-2.6
-3.4
-2.3
-2.7
-6.8
Volume Averaged
iD
«.)
-12
-16
-12
-II
-16
SO
(X)
-2.6
-3.4
-2.7
-2.7
-3.3
_3 "1 3 -I
Discharge increased rapidly from 1.6 nrs ^o 42 ms In 12
hours, in response to 29 mm of precipitation. During the next 24
hours discharge leveled off, then rose again following a short
rainfall, to a maximum of 4.9 ms". After the peak, discharge
slowly decreased, following an irregular linear recession.
Behavior of major-ion concentrations during this storm were
markedly different from that observed during the October
monitoring (Figure 8). Concentration of several ions increased
during an increase in discharge. Calcium and magnesium increased
by 5 and 10% respectively during the first 6 hours, while sodium
fluctuated between 8 and 12 mg/1 during the first 25 hours.
Bicarbonate remained constant during the first 12 hours of
monitoring when discharge increased from 1.6 ms to 3.2 ms ,
and after an initial decrease, sulfate concentration increased to
a maximum at 18 hours. There was also a Jump in ionic
concentrations at 25 hours when Ca , Ha* and SO,~ increased by 13
to 31% from the previous sample. Minimum specific conductance
and ionic concentrations occurred 4 hours after peak discharge.
Concentrations increased from this minimum, however the trend was
marked by erratic fluctuations.
5.0-
4.0-
Dlscharge 3.0 -
(nrVs)
2.0-
1.0-
11
1
i
B
._ ~0<5o_
• 0
• 5
• 10
• 15
I- 20
Precipitation
^ i I 1 1 1 I 1 r—i 1 1—
0 12 24 36 48 60 72 84 96 108 120 132
Tims - Hours
144
Figure 7- Storm hydrograph and rainfall history for Orangeville
Rise, April 12-18, 1991.
276
-------
Discharge
2.0-
48 72 96
Time - Hours
120
r 4-10
•••too
•360
Specific
^320 Conductance
(uS)
-200
T-L240
144
5.0-
Discharge
(m3/s)
Gallons
24
48 72 96
Time - Hours
40
20
-10
Mg2', Na1
(tnoyl)
144
HCOi
(m3/s)
Anions
48 72 96
Time - Hours
120 144
-39
-33
-27
^21
-15
r9
-3
SO4=
Oa", ci
(tng/l)
Figure 8. Chemlonl data fTom Orangoville RlRe, April 12 18, 1991;
A> Bpoclfic conductance, B) cation eoncentrationR, and
C) anion concentrations.
277
-------
Discharge
(m3/s)
5.0-
4.0-
3.0-
2.0-
10
Average 518ORa|n- -3.3%o
Discharge
—I 1 1 r
24 48
72
Tims • Hours
96
—I—
120
144
Figure 9. Isotopic composition of discharge at Orangeville Rise,
April 12-18, 1991.
Isotopic composition of discharge Just prior to the onset of
rainfall was -44 foe, and -6.57oofor ^D and i °O respectively
(Figure 9). During the first 12 hours, when discharge increased
dramatically, & D increased by only 2 7oo and & O increased by
only Q.2%o. During the next 24 hours when discharge increased
only slightly, the isotopic shift was more rapid. The heaviest
discharge occurred from 62 to 65 hours, 24 hours after peak
discharge, when & D reached ~307o0 and & O climbed to -4.5%o.
Isotopic composition of discharge became lighter during the
remainder of recession but did not obtain values seen in pre-
storm discharge.
Hydrograph Separation
As a means of investigating components of discharge at
Orangeville Rise, storm hydrographs were separated into pre-storm
and rainwater components. Separations were performed using both
deuterium and oxygen isotopic data as a means of checking for
consistency in the data.
Hydrograph separation of the October storm is shown In
Figure 10 along with the equation used to solve for instantaneous
rainwater discharge. At peak discharge, rainwater comprised 12
to 15% of the total discharge. The greatest percentage of
rainwater at any time on the hydrograph occurred 18 hours after
peak discharge, where it made up 39 to 46% of the total.
Separation revealed that a secondary pulse of pre-storm water
278
-------
Hydrograph Separation based on Deuterium Isotope Data
Discharge
(rrrVs)
3.0-
2.0'
1.0'
Q,-^ ^-40)
16
24 48 72
Time - Hours
96
Discharge
(m3/s)
3.0 H
2.0 A
I.CH
Hydrograph Separation based on Oxygen Isotope Data
6.2)
24 48 72
Time - Hours
96
Figure 10. Hydrograph separation of discharge
from Orangeville Rise into pre—
storm and rainwater components,
October 4-8, 1990.
Hydrograph Separation based on Deuterium Isotopic Data
Discharge
(m3/s)
Discharge
(m3/s)
(5DM - 44)
24
144
Time - Hours
Hydrograph Separation based on Oxygen Isotope Data
24
144
Time - Hours
Figure 11. Hydrograph separation of discharge
from Orangeville Rise into pre-
storm and rainwater components,
April 12-18, 1991.
-------
arrived just after the time of maximum rainwater contribution.
For the entire five days of monitoring, rainwater made up 20 to
25% of the total discharge at Orangeville Rise.
Separation of the April, 1991 storm hydrograph into pie-
storm and rainwater components is shown in Figure 11. The
volume-averaged isotopic signatures of rainwater shown in Table 2
were used to calculate the rainwater component of discharge at
any one time (McDonnell, et al., 1990).
Hydrograph separation shows that at peak discharge, the
rainwater component made up 27 to 33% of the total discharge.
Twenty-four hours after peak discharge, the rainwater
contribution to total discharge was at a maximum and equaled 47
to 52%. The proportion of rainwater decreased from this point to
a very small to non-identifiable component, 110 hours into
monitoring. As in the October storm, rainwater made up 20% of
the total discharge during the six days of monitoring. This
separation also showed pulses of pre—storm water, but the pulses
.were more subdued than that observed in the October hydrograph.
DISCUSS I OF and COUCLUSIOJTS
Vhile discharge increased rapidly in both storms, ionic
concentrations decreased only moderately or even increased as in
the April storm, indicating that rainwater could not have
contributed to a majority of discharge at peak flow. Isotopic
data also indicates that there was not a large rainwater pulse
arriving at peak discharge. Hydrograph separations reveal that
rainwater made up only a small proportion of discharge at peak
flow, and the largest proportion of rain water contribution to
instantaneous discharge occurred 18 to 24 hours after peak flow.
If the observed isotopic composition of discharge at
Orangeville Rise were a product of simple mixing between two end
members, namely rainwater, and pre-storm water identified by the
isotopic composition of discharge at baseflow, then all data
should fall on a line connecting the two extremes. Figure 12
shows the global meteoric water line (Craig, 1961) and the
isotopic composition of rain samples collected in the Orangeville
Rise basin. The average ground water and baseflow also fall on
the meteoric water line, and hence if two component mixing were
responsible for shifts in the isotopic composition of discharge,
then all data would lie along the line. Data from the October
and April storms do not fall on the meteoric water line, but
rather define another unique line. These results identify an
element of complexity not accounted for in the simple two
component mixing model. Since the isotopic composition of all
phreatic waters in the basin lies on the meteoric water line, one
may be compelled to look to the vadose zone as a source of water
which could cause discharge to deviate from the meteoric water
line.
The rapidly increasing discharge indicates that there must
be substantial concentrated recharge through sinkholes, and rapid
transmission of water through soils via macropore flow. Large
volumes of water must flow quickly through conduits to discharge
at Orangeville Rise, however, this water cannot be rainwater.
Rather "old" water must be displaced by new water. The storage
280
-------
-50
-80
Rain : April, 1991
5D- 85"O + 10
5D-55"0-9
' Rain : October. 1990
• Rain
i Orangeville Rise: October. 1990
o Orangevilla Rise: Apnl. 1991
+ Basetlow
* Average Ground Water
•2
Figure 12. Isotopie data from Orangeville Rise and rainwater
collected within the basin, relative to the meteoric
water line of Craig, 1961.
compartments which could respond most quickly would be
which contain water stored in conduits, or contain water that is
in direct contact with conduits, including epikarst storage,
vadose conduit storage and phreatic conduit storage. An influx
of rainwater through concentrated recharge could be quickly fed
to the epikarst, increasing hydraulic heads there, and causing a
rapid discharge of water stared within the epikarst to
Orangeville Else.
The Increased hydratilic heads associated with recharge will
also result in an increased flow of water through more diffuse
compartments of storage. However, transmissivity of the diffuse
stores are orders of magnitude lower than that associated with
conduit flow. It is difficult, therefore, to see how diffuse
water could arrive at the Rise ahead of the water transmitted via
conduit flow. As a result, we believe that epikarst and conduit
water contributed to the bulk of discharge at peak flow, and
diffuse stores of water may be responsible for the pulses of pre-
storm water seen during discharge recession.
The rain of October 4 broke drought conditions in the basin.
Because it had been so dry, vadose—s?one water storage was at a
minimum, which can be seen In the rapid rate of discharge
recession. The April monitoring occurred at a time of high
recharge to the ground water systems. Water storage compartments
contained large volumes of water, as can be seen in the slow
discharge recession. Ion concentrations remained constant, or
increased with initiation of storm flow, and may represent the
flushing of waters from epikarst and/or conduit storage.
It is interesting to note that in both storms, the tjme of
281
-------
minimum epecific conductance did not correspond to the time of
maximum rain water discharge as identified by the isotopic data.
Discharge at any moment of time is a product of all the
contributing sources. It is possible that at the time of maximum
rainwater arrival, water was also arriving from sources with
higher than average total dissolved solids, the resultant
discharge having a greater conductivity than might be anticipated
for the time of peak rainwater discharge.
Aknowledements
This study was funded by the American Geophysical Union,
Horton Research Fellowship.
References Cited
Atkinson, T.C., 1977, Diffuse flow and conduit flow in limestone
terrain in the Mendip Hills, Somerset (Great Britain); Jour.
of Hydrology, v.35, p.93-110.
Bakalowicz, M. ,B. Blavoux, and A. Mangin, 1974, Apports du
tracage isotopique nature1 a la connaissance du
fonctionnement d'un de trois systemes des Pyrenees France*
Jour, of Hydrology, v.23, p.141-158.
Bassett, J.L., 1974, Hydrology and geochemistry of karst terrain,
upper Lost River drainage basin, Indiana; M.A. Thesis,
Dept. of Geology, Indiana University, 102p..
Craig, H., 1961, Isotopic variations in meteoric waters; Science,
v.133, p.1702-1703.
Dreiss, S.J., 1989, Regional scale transport in a karst aquifer
1. Component separation of spring flow hydrographs; Water
Resources Research, v.25, n.l, p. 117-125.
Fritz, P., J.A. Cherry, K.U. Veyer , and H. Sklash, 1976, Storm
run-off analysis using environmental isotopes and major
ions; Interpretation of environmental Isotope and
Hydrochemical Data in Groundwater Hydrology, International
Atomic Energy Agency, Vienna, p.111-130.
Gunn, J., 1983, Point recharge of limestone aquifers - a model
from Hew Zealand karst; Jour, of Hydrology, v.61, p. 19-29.
Krothe, H.C., and R.D. Libra, 1983, Sulfur isotopes and
hydrochemical variations of spring waters of southern
Indiana, USA; Jour, of Hydrology, v.81, p.267-283.
Libra, R. D. , 1981, Hydrogeology and sulfur isotope variation of
spring systems; south-central Indiana; M.A. Thesis, Dept. of
Geology, Indiana University, 109p.
Mangin, A., 1974, 1975, Contribution a 1*etude hydrodynamique des
aquiferes karstiques. DES thesis, Univ. Dijon, France; (Ann.
Speleology, 1974, v.29, n.3, p.283-332; v.29, n.4, p.495-
601; 1975, v.30, n.l, p. 21-124.
McDonnell, J.J., M. Bonell, M.K. Stewart, and A.J. Pearce, 1990,
Deuterium variations in storm rainfall: Implications for
stream hydrograph separation; Water Resources Research,
v.26, n.3, p.455-458.
Milanovic, P.T., 1976, Water regime in deep karst. Case study of
the Ombla spring drainage area; in Karst Hydrology and Water
Resources, v.l, Karst Hydrology, V.YevJevich (ed.), Fort
Collins, Colorado, Water Resources Publication, p. 165-191.
282
-------
Murdock, S. H. , and R.L. Powell, 1968, Subterranean drainage
routes of Lost River, Orange County, Indiana; Indiana
Academy of Science Proceedings n.77, p.250-255.
Rune, R.V., 1975, Geohydrology of karst terrain, Lost River
watershed, southern Indiana; Indiana University Water
Resources Research Center, Report of Investigation,
n.7, 91p.
Sheppard, S.F.M., R.L. Nielsen, and H.P. Taylor, 1969, Oxygen and
hydrogen isotope ratios of clay minerals from porphyry
copper deposits; Economic Geology, v.64, p.755-777.
Shuster, E.B., and V.B. White, 1971, Seasonal fluctuations in the
chemistry of limestone springs: A possible means of
characterizing carbonate aquifers; Jour- of Hydrology, v.14,
p.93-128.
Sklash.M. G. , and R.IT. Farvolden, 1979, The role of groundwater in
storm runoff; Jour, of Hydrology, v.43, p.45-65.
Torbarov, K. , 1976, Estimation of permeability and effective
porosity in karst on the basis of recession curve analysis;
in Karst Hydrology and Water Resources, v.l, Karst
Hydrology, V.YevJevich (ed.), Fort Collins, Colorado,
Water Resources Publication, p.121-136.
Tweddale, J.B. , 1987, The relationship of discharge to
hydrochemical and sulfur isotope variations in spring waters
of south—central Indiana; M.S. Thesis, Dept. of Geology,
Indiana University, 165p.
Wells, E.R. , and N.C. Krothe, 1989, Seasonal fluctuations in ^15IT
of ground water nitrate in a mantled karst aquifer due to
macropore transport of fertilizer-derived nitrate; Jour, of
Hydrology, v.112, p.191-201.
Williams, P. W. , 1983, the role of the subcutaneous zone in karst
hydrology; Jour, of Hydrology, v.61, p.45-67.
283
-------
Gruver - Stable Isotope Separation
1. I want to congratulate the speaker for her particularly thoughtful
discussion, and her stimulating suggestions. The stable isotope records
presented are suggestive of piston-flow (at least in part). Piston-flow
- response is difficult to reconcile with the suggestion that the source of the
expelled (pre-storm) water deviating from the meteoric water line in the
epikarst. Please comment.
2. Can you suggest a method for spiking clean injection water with heavy
oxygen or deuterium in order to perform a tracer test with stable isotopes?
This could be useful in areas where, for political reasons, it was not
practical to use dyes in a karst aquifer. Do you know of such studies?
Would light water be more easily obtained? What is the cost per liter (or
kg) for heavy water and light water? Having bought same, how much of it
is ordinary water? (This is asked in order to get an idea of how much
tracer might be heeded for ah "average" trace and what the cost would
be.)
3. The delta 1 &O - delta D relations of the spring response suggests a third
hydrological component, presumably in the vadose zone. However, the
end member occupied by the two storms are different, although both lie on
the "evaporite line." Why are they different, and why are they displaced to
the left and right of the "meteoric water line" for the two storms?
284
-------
Answer to Questiontfl:
Water collected during baseflow at Orangeville Rise, and ground waters
sampled from domestic water wells lie on the meteoric water line (MWL). As all
phreatic and rain water samples lie on or near the MWL, one may need to look to the
vadose zone for a source of water that could cause discharge to shift from the MWL.
The first few samples collected during each storm event lie very close to the
MWL. I believe these samples represent displaced phreatic conduit water. Six to
twelve hours into storm flow, the isotopic signature of discharge begins to deviate
from the MWL, as well as in the direction of the rainwater isotopic signature. It
seems reasonable to interpret these waters as rain and epikarst water that has been
rapidly transmitted via shaft, and conduit flow. Very little work has been done on
chemical processes within waters of the epikarst, or what fractionation processes
might occur there. This study points to the need for research which addresses these
questions.
Answer to Question #2:
I am unfamiliar with any investigation which has used "light" or "heavy"
water in place of traditional dyes in karst aquifer tracing. If one feels compelled to
use this technique, however, they would be advised to use a water tracer that has the
greatest isotopic difference from that of the natural system. The amount of "spiked"
water used would be dependent upon the volume of water in the natural system. I
could not begin to guess the price of "spiked" water. I can say that a large expense
in a stable isotope study is the cost of analysis. Count on at least $35 per sample for
deuterium analyses, and $45 for oxygen analyses.
Answer to Question #3:
The delta ISO-delta D relationships suggest at least a third component
contributing to total discharge. I would suggest that, while the data seem to lie along
a common line with a slope of 5, upon closer examination, the October data lie along a
line with slope 4, while the April storm lies along a line with slope 6. I believe that
the cause of the shifts for each storm is different. Fractionation processes acting in
the dry fall-time, compared with those of the spring may be different and may be
responsible for shifts in the isotopic signature of discharge in opposite directions
from the meteoric water line.
285
-------
286
-------
The Transmission of Light Hydrocarbon Contaminants
In Limestone (Karst) Aquifers
Ralph 0. Ewers Ph.D./ Eastern Kentucky University
and Ewers Water Consultants Inc.
Anthony J. Duda, Dames & Moore
Elizabeth K. Estes M.S., University of Minnesota
Peter J. Idstein M.S., Eastern Kentucky University
Katherine M. Johnson, U. S. Army (USATHAMA)
ABSTRACT
Hydrocarbon contaminants with densities less than water may
enter karst aquifers through soil percolation, directly by way
of surface runoff into open sinks, or by a combination of these
two routes. Once in the aquifer, the movement of these
hydrocarbons is governed by four factors: (1) the level of the
principal dissolution conduits relative to the zone of
saturation, (2) the flow regime (turbulent or non-turbulent) in
the conduits which contain the contaminant, (3) the vertical
complexity of the framework of dissolution porosity in the
aquifer, and (4) the nature of the recharge to the aquifer.
Field investigations indicate that light hydrocarbons can move
in these aquifers as gross concentrations of free product at
speeds of the order of kilometers per hour. Under other
circumstances, the contaminants may be trapped in the aquifer so
completely that spring water originating from the point of gross
hydrocarbon release shows none of the contaminant.
Trapping of the hydrocarbons occurs so long as two conditions
exist. First, the active conduits must remain, to a large
extent, in a completely water-filled (phreatic) condition. If
a free surface (an air-water interface) develops throughout a
large percentage of the active conduits the floating fluid may
move as "free product". Second, the groundwater velocity in the
vicinity of the floating product must remain slow. Gross
turbulence entrains the "free product" in the swift flow of the
active conduits.
INTRODUCTION
Contaminant movement in karst aquifers often appears strange
and erratic to hydrogeologists more familiar with granular
287
-------
aquifers. Light hydrocarbon contaminants, those with densities
lower than water, appear especially capricious. The behavior of
these materials in karst aquifers is related to the dissolution
porosity, often referred to as conduit porosity, which they
possess.
Dissolution porosity in paleozoic carbonate rock normally
occurs along partings such as bedding planes, joints, and
faults. In areas of flat lying or gently dipping carbonates,
conduits formed along bedding planes are commonly most important
in forming the master conduits which permit long distance
transmission of groundwater. Joint and fault related conduits
provide routes for admitting meteoric water to the aquifer. The
total length of explorable conduits primarily guided by bedding
planes in such areas commonly exceeds the total length of those
primarily guided by joints by 10 to 1 (Ewers, 1972). On a
worldwide basis, which includes karst aquifers in dipping rocks,
the ratio of bedding plane to joint and fault controlled
conduits is of the order of 1.3 to 1 (Palmer, 1991) This is the
case, not only because bedding planes commonly extend over
greater horizontal distances than joints, but also because they
can transmit water in any direction, unlike joints and faults.
Unlike the diffuse, dispersive, laminar flow of granular
aquifers, groundwater movement in karst aquifers is
concentrated, convergent toward the conduits, and at times
turbulent (Quinlan and Ewers, 1985). Surface water and
contaminants may have direct access to the conduits by way of
sinkholes with swallet openings and with sinking streams. The
presence of conduits may permit rapid movement of liquid phase
hydrocarbons. Because of the turbulent flow regime in these
conduits, contaminant movement may be in the form of globules of
entrained "free product" as well as a dissolved phase.
Groundwater and contaminant velocities in these conduits are
commonly as high as several kilometers per day (Quinlan and
Ewers, 1983).
FACTORS AFFECTING LIGHT HYDROCARBON MOVEMENT IN KARST
The movement of immiscible floating hydrocarbons in karst
aquifers is controlled by factors relating to the
characteristics of the framework of dissolution porosity and to
the type of flow regime found in these conduits. These factors
can be conveniently grouped into the following four categories.
1.- The level of the principal conduits relative to the zone of
saturation
Conduits in karst aquifers occur deep within the phreatic zone
(the zone of saturation beneath the potentiometric surface) and
well above normal levels of groundwater circulation. Their
range in a specific area is related to the thickness of soluble
carbonate rock, the local geomorphic history, and climatic
factors. Light hydrocarbon contaminant transmission is less
288
-------
likely in master conduits located deep in the phreatic zone than
in those which are along the upper part of this zone.
Phreatic Conduits
Where the master conduits responsible for horizontal
transmission of groundwater lie permanently within the phreatic
zone, floating hydrocarbons should not move easily. The entry
point for the contaminants should form at least a temporary trap
for this free product at the top of the phreatic zone (Fig. 1,
point "A"). Likewise, any other portion of the master conduit
Hydrocarbon Traps
Water Table
i ^ Jo
Master
Conduit
T
Free Hydrocarbons
FIGURE 1 Light hydrocarbons can be trapped above phreatic
master conduits in ("A") aquifer entry points, ("B")
solution widened joints, ("C") horizontal conduits above
bedding planes, and ("D") the soil filling of grikes
(solution widened joints) and in the soil around nearby
clints (bedrock pinnacles).
which extends above the potentiometric surface should form an
additional trap for any free product which becomes entrained in
the flow of the conduit, only that portion of the hydrocarbon
which dissolves in the groundwater should move easily in
phreatic conduits.
289
-------
Epiphreatic Conduits
Where the master conduits are in the epiphreatic zone (the
zone in the immediate vicinity of the potentiometric surface),
floating hydrocarbons tend to move more freely. Under base-flow
conditions, some master conduits will have an air-water
interface along their entire length. This provides a
continuously sloped surface for the movement of floating "free
product", identical to a surface stream. The hydrocarbons will
move at a speed nearly the same as the water in this
circumstance. More commonly, epiphreatic conduits have air-
water interface conditions spaced intermittently along their
length. To the degree that this interface diminishes, floating
hydrocarbons will move with increased difficulty, their mobility
becoming more and more dependant upon the nature of the flow
regime in the conduit.
2.- The flow regime in the conduits which contain the
contaminant
Turbulent Flow
During periods of turbulent flow, floating hydrocarbons would
be readily entrained in conduit water. Groundwater velocities
in typical karst aquifers range from 30 ft/hr at base flow to
1300 ft/hr in periods of high flow (Quinlan and Ewers, 1985).
The hydrocarbons, normally excluded from phreatic master
conduits, could be mobilized by the turbulence and would travel
at nearly the velocity of the water.
Laminar Flow
When laminar flow prevails, floating free hydrocarbons are
unlikely to be entrained in the groundwater. The "free product"
will separate from the water and collect at the highest points
along the conduit ceiling. Diffusion of the hydrocarbons into
the water will occur, but the concentrations of contaminants
carried in this way may be very low. The large quantities of
groundwater moving in the conduits may limit these
concentrations as well as the relatively high octanol-water
partition coefficients for many of these substances. Both
trapping of the floating free product and diffusion of the
hydrocarbons into the water are dependant upon the vertical
complexity of the conduits.
3.- The vertical complexity of the framework of dissolution
porosity in the aquifer
Vertical conduits in carbonate aquifers have two roles in the
transmission of light hydrocarbons. They provide entry routes
for these contaminants and they also provide traps for the free
product. These conduits together with abandoned horizontal
conduits are collectively referred to as the epikarst (Mangin,
1974-75). This term is used here in a broad sense to include
any solution porosity lying above the level of the active
290
-------
drainage trunks, including that which may be above or below the
normal potentiometric surface. Karst aquifers with an extensive
system of interconnected vertical and horizontal phreatic
conduits above deep phreatic master conduits have an increased
ability to trap light hydrocarbons and a reduced likelihood of
transmitting these contaminants. Those with simple shallow
conduit systems are more prone to contaminant movement.
Hydrocarbon Entry
Sinkhole
With Swallet Opening
Leaking Underground
Storage Tank
With Cover of Residuum
Water Table
t
FIGURE 2 Light hydrocarbons can enter karst aquifers in
several ways. They may enter directly with concentrated
surface recharge entering swallets (point "A"), by seepage
through sinkhole bottoms covered by residuum (point "B"),
or through the soil cover and minor solution widened joints
in the carbonate rock (point "C").
Hydrocarbon Entry
Hydrocarbons may gain entry to a karst aquifer through
sinkholes with open swallets which give direct access to the
conduit system or by seepage through the residuum in a sinkhole
bottom (Fig. 2, points "A" and "B"). In either case, a vertical
conduit, typically developed along a joint, provides both the
means for the sinkhole to form and the entry route for the
contaminants.
291
-------
A much larger number of vertical epikarst conduits are
associated with features called grikes. These are linear
depressions in the bedrock surface with narrow bottoms and wide
tops which develop along joints. They can conduct hydrocarbon
contaminants from the bottom of the soil horizon to master
conduits in the phreatic or epiphreatic zone. These conduits
are often common in areas where no depressions are visible at
the surface of the overlying soil.
Hydrocarbon Trapping
Master conduits are frequently intersected by joints. Where
these joints are enlarged by dissolution, they form traps in
their phreatic portions for floating hydrocarbons (Fig. 1, point
"B"). Vertical conduits of this type intersecting master
conduits which are accessible to direct observation frequently
have volumes in excess of 1000 gallons. Where the carbonate
rock is heavily jointed, these potential trapping situations are
seen to occur at intervals of a few meters or less. Thus, they
may provide a considerable storage capacity for floating
hydrocarbons.
Vertical conduits may connect the master conduit with higher
horizontal conduits in the phreatic portion of the epikarst
(Fig. 1, point "C"). These conduits are often laterally
discontinuous due to truncation by soil filled grikes. Conduits
of this type may provide additional storage for light
hydrocarbons.
Vertical conduits may extend from the master conduit to the
upper bedrock surface and the base of the soil. In such cases
the hydrocarbons may collect at the soil bedrock interface if
the potentiometric surface reaches above this level either
permanently or intermittently (Fig. 1, point "D").
Where the interval between the level of the master conduit and
the level of the uppermost phreatic part of the vertical conduit
elements is great, the groundwater in the master conduit may be
in poor communication with the floating contaminants. Both
entrainment of the hydrocarbons by master conduit turbulence and
their diffusion into this flow will be very limited.
4.- The nature of the recharge to the aquifer
Two basic types of recharge occur in carbonate aquifers,
diffuse recharge and concentrated recharge. The nature of this
recharge has a significant effect upon the movement of light
hydrocarbon contaminants. Diffuse recharge which passes through
a soil cover and enters the aquifer through a large number of
joints is less likely to produce gross turbulence in the aquifer
conduits than concentrated recharge. This latter includes the
concentrated runoff which enters swallets in sinkholes and both
292
-------
allogenic and autogenic sinking streams. Concentrated storm
runoff from these sources regularly produce stage changes in
conduit systems of 30 feet, and changes of 75 feet are not
unusual. Flows of these magnitudes produce gross turbulence in
the conduit systems which propel coarse gravel from spring
orifices. A modest amount of turbulent recharge entering the
top of a free product trap may entrain the collected
contaminant. Similarly, hydrocarbons trapped close to the
master conduit may be mobilized by master conduit turbulence
induced by the storm flow of a sinking stream.
Karst aquifers with thick soil covers, few open sinks, and
without sinking streams have a reduced likelihood of
experiencing movement of trapped light hydrocarbon contaminants.
Those with the opposite characteristics are at higher risk for
contaminant movement.
CASE STUDY lr AN AQUIFER WITHOUT CONTAMINANT MOVEMENT
Quarles Spring Groundwater Basin
Quarles Spring, located in Christian County, Kentucky, is used
as a potable and irrigation water supply. Dye tracing by Ewers
and his students (Ewers Water Consultants, 1989; Carey, 1990;
Greene, 1990) has shown that this spring is the principal
discharge point for a groundwater basin which includes Campbell
Army Airfield (Fig. 3). Jet fuel spillage at the airfield has
appeared in several wells in the karst aquifer at the airfield.
As much as 16 feet of "free product" has been measured in
monitoring well MCI-2 (Point "C", Fig. 3; Fig. 4) by Ewers
(Ewers Water Consultants, 1991). Dames & Moore report free
product in four other wells (Dames & Moore, 1991).
Investigations by Duda (Dames & Moore, 1991) suggest that the
fuel entered the aquifer through a sinkhole which takes drainage
from the airfield (Point "B", Fig. 3). Repeated sampling of
Quarles Spring by Duda (Dames & Moore, 1991) has failed to show
jet fuel, dissolved or free. Thus, it appears that floating
hydrocarbons are not mobile in this aquifer. The known and
inferred aquifer characteristics fit the criteria for low
mobility outlined above.
The principal conduit system-
Quarles Spring possesses a deep rise-pool, suggesting a
phreatic condition in the master conduit which is tributary to
it. Information from Well MCI-2 (Fig. 4) at the airfield gives
further evidence that the conduit system is well within the
phreatic zone. This well intersects a solution conduit at about
459 feet msl. This conduit has a minimum vertical extent of 16
feet (MCI, 1987), and is a possible source of the jet fuel
contaminant which is seen in the annulus of this well. The well
bore entry point into the conduit is approximately 29 feet lower
than the normal level of Quarles Spring, well within the
phreatic zone (Table 1; Fig. 4).
293
-------
Quarles Spring Groundwater Basin
Quarles Spring
~r - '.n
*
55
FIGURE 3 Quarles Spring is the principal discharge point for a groundwater
basin which includes Campbell Army Airfield. Jet fuel spillage at the airfield
has appeared in several wells in the karst aquifer. Point "C" is well MCI-2.
Points "B", "C", and "H" are the location of dye inputs. The tracer dyes were
recovered at Quarles Spring, number 55. No jet fuel has been recovered from
this spring.
-------
Vertical Section - Quarles Spring Basin
Elevation MSL
57 0\-
Well I
Airfield Elevation 559 msl.
520
470
N5
VO
420
Quarles Spring
488 \rnsl.
/Product reported by
/ . Dames & Moore
Product Level 6/90
494' msl. (approx)
Water Interface 6/90
478' msl.
Cavity
j Top 459'- Bot. 443'
(width inferred)
Surface Elev
Open Hole
Bedrock Interface
Conduit
Casing
Free JP-4
0000 Epikarst (position inferred)
FIGURE 4 This schematic vertical section between well MCI-2 (point "C", Fig.
3) and Quarles Spring is based upon information from 37 wells and numerous soil
investigations. The conduits shown are inferred, except for that in the
vicinity of well MCI-2. The typical and extreme range of the soil bedrock
interface is shown.
-------
Table 1 ELEVATION OF FEATURES RELATED TO MONITORING WELL
MCI-2, THE BEDROCK, AND QUARLES SPRING
Well Features Elevation (ft.) Bedrock & Spring Features
Top of Casing 560.16
Ground Level 559.65
Bedrock/Soil Interface-
534 Shallowest (MCI, 1987)
Product Level 6/90 494
491 Floodplain at Quarles Spr.
Water Level Without-
Floating JP-4 490.25
Bottom of Casing 489.74
488 Water level in Quarles Spr.
Water Interface 6/90— 478
Top of Cavity 459
Bottom of Cavity 443
Bedrock/Soil Interface-
433 Deepest (MCI, 1987)
The conduit intersected by this well was shown by dye tracing
to be connected to Quarles Spring (EWC, 1989). It is reported
to have significant amounts of fine sediment moving through it.
The driller recorded that a five foot layer of mud was swept
into this conduit during a severe rainfall event which occurred
during the installation of the well (MCI, 1987). We may assume
that this conduit is a part of the active conduit system,
tributary to Quarles spring. If this conduit were significantly
lower than the active conduit system and if it were not a part
of that system, then it would probably be sediment filled. The
groundwater gradient to Quarles Spring is very low. The head
loss from the well to the spring is approximately 2 feet at base
flow (Table 1) over a distance of 11,800 feet, a typical value
for a phreatic conduit. These observations suggest that the
conduit system beneath Campbell Army Airfield and tributary to
Quarles Spring is phreatic in nature, even under extreme low-
flow circumstances.
The flow regime and the aquifer recharge-
Flow in the master conduits of the Quarles Spring groundwater
basin are almost certainly turbulent much of the time. The
spring discharges water even when heavy withdrawal for
irrigation occurs during very dry periods. Base flow discharge
is estimated at 0.5 ft3/sec. However, it discharges primarily
fine sand-sized sediment and is typically quite clear. It does
not have the flashy response and wide range of flow velocities
exhibited by many karst aquifers. Flow velocities in the region
of most of its concentrated recharge points is probably not
turbulent. The reason for the lack of turbulence can be found
in the nature of the aquifer recharge.
296
-------
Recharge to the Quarles Spring groundwater basin does not
include sinking streams. All of its recharge is autogenic, it
occurs directly over the groundwater basin where only soluble
bedrock with chert bands and cherty residuum overlie the bedrock
aquifer. Sinkholes are common, but they rarely contain direct
openings to the bedrock. A layer of cherty residuum, 25 to 127
feet in thickness, mantles the bedrock and sinkholes restricting
most of the recharge to soil percolation (Dames & Moore, 1991).
Low-flow and high-flow time of travel studies utilizing dye
tracers were undertaken between the drywell at point "H" and
Quarles Spring (Fig. 3)(EWC, 1991). The velocities obtained
were 95 feet per hour and 226 feet per hour respectively. These
values are quite different from those obtained from the Mammoth
cave region where the aquifer is recharged directly through
swallets and by sinking streams. Values of 9 meters per hour
and 390 meters per hour are quoted by Quinlan & Ewers (1985) for
that area.
The vertical complexity of the conduits-
If the conduit encountered in the well at point "C" (Fig. 3)
is indicative of the level of the master conduit, 31 feet of
space in the vertical dimension are available for the trapping
of contaminants (Table 1; Fig, 4). Fracture trace analysis
(Dames and Moore, 1991) and the extremely variable elevation of
the bedrock soil interface (Dames & Moore, 1991; MCI, 1987)
strongly suggest that fractures enlarged by dissolution are
common in the bedrock aquifer. Therefore, vertical and
horizontal conduits should be common in the epikarstic zone.
These should be quite effective at trapping floating
contaminants in the Campbell Army Airfield setting. The wide
range of the soil bedrock interface, reported by drillers,
should insure that there are several horizons of horizontal
epikarstic solution porosity which are poorly integrated
laterally (Fig. 4, Table 1).
The 16 foot layer of floating jet fuel in the well reported
by Ewers is associated with the conduit porosity encountered
when this well was drilled. The free product reported by Duda
was discovered at the soil bedrock interface, and is probably
fuel which has migrated upward through solution widened joints
from the master conduit below. These horizons are likely to be
poorly connected to sources of concentrated recharge because of
the relative lack of open sinkholes. Without this turbulent
recharge, accumulated product is unlikely to be flushed from
these reservoirs. Because these reservoirs extend well above
the active flow system there would be little opportunity for
measurable quantities of the floating immiscible contaminants to
dissolve in the groundwater circulating in the main conduits.
297
-------
CASE STUDY 2, AN AQUIFER WITH GROSS CONTAMINANT MOVEMENT
Little Sinking Creek - Big Sinking Creek
This site is located near the Big Sinking Creek oil field in
Lee County, Kentucky. Both Little Sinking Creek and Big Sinking
Creek are, in part, subsurface streams. Portions of their
surface valleys are perennially streamless (Fig. 5). Conduits
of explorable dimensions with multiple entrances pass beneath
the dry portions of these valleys.
Little Sinking Creek - Big Sinking Creek, KY
Legend
-<•• Spring
-—^ Sinking Stream
A Oil Tanks
1000 0 1000 ?000 3000 4000
FIGURE 5 Little Sinking Creek and Big Sinking Creek are,
in part, subsurface streams. Several smaller sinking
streams are tributary to the master conduits beneath these
valleys. oil spills at several points have been
transmitted quickly to the springs at points "A" and "B".
This karst aquifer is developed in the Mississippian age
Newman Limestone, oil production is from units 700 feet beneath
the karst aquifer. These horizons are isolated from the aquifer
298
-------
by several hundred feet of shale. Pipeline breaks, storage tank
cleaning, and brine discharges have introduced chlorides and
crude oil into these sinking streams and into the conduit
entrances. The interior surfaces of the conduits are coated
with petroleum and tar balls are common along the surface and
subsurface portions of the streams. A spring, perched upon a
shale layer beneath the Newman Limestone discharges from the
lower end of each of these valleys near their confluence (Fig.
5). Floating free product is discharged with the groundwater
from these spring points after spills have occurred. Thus, it
appears that floating hydrocarbons are extremely mobile in this
aquifer. The known and inferred aquifer characteristics fit the
criteria for high mobility outlined above.
The principal conduit systems-
The master conduits in both valleys are located in the
epiphreatic zone. Deeper phreatic circulation is prevented in
this case by the presence of insoluble shale beds in the Renfro
Member of the Borden Formation. Virtually the entire length of
conduit beneath the valley of Little Sinking Creek shown in
figure 6 is explorable and occupied by a stream with an air-
water interface above. The conduit which forms the spring
orifice has been observed to maintain an air-water interface
under very high flow conditions. The master conduit beneath the
valley of Big Sinking Creek does not possess an air-water
interface over its entire length, but the crude oil is
transmitted along the total conduit system (Fig. 7).
The flow regime and the aquifer recharge-
The aquifer recharge at this site is almost entirely from
allogenic sources, by way of sinking streams. This concentrated
recharge is clearly turbulent, and during heavy rainfall events
it is grossly turbulent. The conduit flow in the aquifer is
visible at several points along each valley and turbulent flow,
similar to that in the sinking streams and the spring
resurgences, can be directly observed.
The portion of the Big sinking Creek conduit above mist cave,
shown in Figure 7, may not possess an air-water interface even
under base flow conditions. Water ponds over the swallet of Big
Sinking Creek during high-flow conditions and the crude oil is
temporarily trapped at the surface. When the flow subsides, the
ponded water lowers and turbulence entrains the floating
contaminant, and it is conducted through several spring and sink
points to its final resurgence.
The vertical complexity of the conduits-
Only a few feet of limestone overlie the master conduits at
this site. Therefore, the opportunities for trapping of
hydrocarbons along the few sections of phreatic flow are
severely limited.
299
-------
Vertical Section - Little Sinking Creek
Swallet of Little Sinking Creek
c
CO
E
0)
z
Oil Tanks
c
0)
13
O
CQ
Arrow Cave
Flood Cave
Resurgence
of
Little Sinking Creek
FIGURE 6 Vertical section through Little Sinking Creek
The Swallet of Little Sinking Creek and Flood Cave are
points "A" and "C" in Figure 5, respectively.
Vertical Section - Big Sinking Creek
Dune Buggy Cave
Swallet of Big Sinking Creek
Mist Cave
Resurgence
of
Big Sinking Creek
V)
2
_i
c
(0
E
FIGURE 7 Vertical section through Big Sinking Creek
The Swallet of Big Sinking Creek and Dune Buggy Cave are
points "B" and "D" in Figure 5, respectively.
300
-------
CASE STUDY 3, AN AQUIFER WITH MIXED CONTAMINANT MOVEMENT
Lancaster Road Gasoline Station - Richmond Kentucky
An underground storage tank at this site released gasoline
into a karst aquifer consisting of Ordovician Limestones. Free
product was discharged into Tennis Court Spring, an intermittent
seep 150 ft from the leak site (Fig. 8). Fluorescent tracer dye
injected into the storage tank pit showed that groundwater
flowed to Tennis Court Spring and also to Little Caesar Spring
a perennial spring 900 ft distant. Gas chromatograph analysis
of samples from these springs showed that the free product
discharging from Tennis Court Spring was nearly identical to the
tank contents. No gasoline constituents could be discerned in
the samples from Little Caesar Spring either in high-flow or
low-flow conditions.
There are no data from wells or other sources which can give
direct insight into the nature of the conduits which connect
these two springs to the leak site. However, the two springs
clearly indicate the presence in the aquifer of a natural light
non-aqueous phase liquid separation system. This system
operates in an environment where water flow clearly occurs in
discrete conduits. The tracer introduced into the tank pit
leaves no doubt about this assertion. It appeared at both
springs overnight. As expected, the level of the spring
discharging product was significantly higher than the one which
discharged water only (Table 2). This karst aquifer has no
sinking streams and few sinkholes, none with swallet openings
which could admit water directly to the aquifer.
Table 2 ELEVATION OP FEATURES RELATED TO THE LANCASTER ROAD
GASOLINE STATION
Elevation (ft.) Spring Features
995 Top of gasoline storage tank pit (dye injection
point)
985 Tennis Court Spring - Groundwater, gasoline, and
dye were discharged at this point.
935 Little Caesar Spring - Only groundwater was
discharged at this point.
CONCLUSIONS
The empirical studies suggest that the following conditions
decrease the probability of movement of light hydrocarbons in
karst aquifers:
1-The limestone formations are thick and conduits are well
within the phreatic zone.
301
-------
Lancaster Road Site, Richmond, KY
Gasoline Leak Site
Tennis Court Spring
/
0 100 200 300
I I I I
Feet
Little Caesar Spring
935' msl
Eastern Bypass
FIGURE 8 An underground storage tank at this site released
gasoline into a karst aquifer. Free product was discharged
at Tennis Court Spring, an intermittent seep. Fluorescent
tracer dye injected into the storage tank pit showed that
groundwater flowed to Tennis Court Spring and also to
Little Caesar Spring, although no gasoline could be
detected at this latter spring. The numbers indicate the
elevation of the leak site and the springs.
302
-------
2-The limestones are covered by a thick mantle of soil causing
recharge to the aquifer to be slow and non-turbulent.
3-The aquifer is not recharged by sinking streams so that
turbulence in the master conduits is minimized.
4-An extensive system of epikarstic porosity overlies the master
conduit providing cavities for trapping of the hydrocarbons.
Given all of these characteristics, hydrocarbon mobilization
could occur if drought conditions can lower the potentiometric
surface sufficiently to alter the phreatic condition in the
master conduit.
Human activities or natural events which would permit storm
water to flush the epikarstic zone could possibly mobilize
contaminants stored there. These would include intense highly
localized rainfall events, developing sinkholes, excavations
near bedrock, and dry-wells. Attempts to retrieve contaminants
through recovery wells by creating depression cones extending
near or below the level of the spring outlets could create
conditions for their transport, particularly if this occurred at
a time when gross turbulence existed in the trunk conduits.
REFERENCES CITED
Carey, M,J., 1990, The delineation of Karst drainage to identify
possible contaminant migration routs from the Campbell Army
Airfield, Fort Campbell, Kentucky. Master of Science Thesis,
Eastern Kentucky University, Department of Geology, Richmond
Kentucky, 6Op.
Dames & Moore, 1991, Remedial Investigation Report, Contract
DAAA 15-88-D-0008 to USATHAMA
Ewers, R.O., 1972, A model for the development of subsurface
drainage routes along bedding planes, M.S. Thesis, University of
Cincinnati, Cincinnati, Ohio, 84p.
Ewers Water Consultants (EWC), 1991, Groundwater Time of Travel
Study, Campbell Army Airfield, Implications for Contaminant
Movement, R.O. Ewers. Contract to Ft. Campbell and USATHAMA.
Ewers Water Consultants (EWC), 1989, Fort Campbell Groundwater
Study, Final Report, R.O. Ewers, M.J. Carey, and D.L. Greene.
Contract to Ft. Campbell and USATHAMA.
Green, D.L., 1990, Hydrogeology of the potable water supply
spring (Boiling Spring) on the Fort Campbell Military
Reservation, southwestern Kentucky - northwestern Tennessee.
Master of Science Thesis, Eastern Kentucky University,
Department of Geology, Richmond Kentucky, 43p.
303
-------
Mangin, A., 1974-75, Contribution a 1'etude hydrodynamic des
aquiferes karstiques, Annals de Speleologie, 29(3).
MCI, 1987, Subsurface investigation for JP-4 contamination, POL
facility 7226, Fort Campbell Army Airfield, Fort Campbell KY, to
US Army Corps of Engineers, Nashville Dist. by C.S. Higgins,
M.J. Levy, and C.B. Huggins, MCI Consulting Engineers.
Quinlan, J.F. and R.o. Ewers, 1985, Groundwater flow in
limestone terranes: strategy rational and procedure for
reliable, efficient monitoring of groundwater quality in karst
areas. Proceedings of the Fifth National Symposium anrj
Exposition on Aquifer Restoration and Groundwater Monitoringr
National Water Well Association.
Quinlan, J.F. and R.o. Ewers, 1985, Groundwater Hydrology
geomorphology of the Mammoth Cave region, Kentucky, and of the
Mitchell plain, Indiana (Field Trip 7); with J.A. Ray, N.C.
Krothe, and R.L. Powell; in Shaver, R.H. and Sunderman J.A.,
eds., Field Trips in Midwestern Geology: Bloomington. Ind.f
Geological Society of America and Indiana Geological Survey, v.
2, p. 1-85.
BIOGRAPHICAL SKETCHES
Ralph o. Ewers is professor of geology and director of the
Groundwater Research Laboratory at Eastern Kentucky University,
and a principal in Ewers Water Consultants, a consulting firm
specializing in carbonate aquifers. His B.S. and M.S. degrees
in geology were earned at the University of Cincinnati and his
Ph.D. was earned at McMaster University (1982). Professor
Ewers' special interests include the applications of tracer and
electronic monitoring techniques to the solution of practical
environmental problems in karst groundwaters. He was co-
recipient of the 1986 E.B. Burwell Award from the Geological
Society of America for a "work of distinction in engineering
geology."
Ralph O. Ewers Ph.D.
Ewers Water Consultants Inc.
160 Redwood Drive
Richmond, KY 40475
(606) 623-8464
Anthony J. Duda is senior hydrogeologist with Dames & Moore,
where he has worked since 1980. He received his B.A. degree in
geology from Rutgers University. Mr Duda has been involved in
planning and implementing groundwater remediation strategies,
soil gas sampling programs, and geophysical surveys. His
experience includes investigations at more than 30 governmental
and private sites in more than 10 states. He has been lead
hydrogeologist at the Ft. Campbell, Campbell Army Airfield site
where he was involved in a major UST investigation.
304
-------
Elizabeth K. Estes is a doctoral candidate at the Department
of Geology and Geophysics at the university of Minnesota. She
received her M.S. degree from Eastern Kentucky University.
Among her many accomplishments is a five year study of the area
in the vicinity of Big Sinking Creek in Lee County, Kentucky,
and several studies with the Cave Research Foundation at Mammoth
Cave Kentucky- Her research interests include, karst
geomorphology, speleogenesis, and groundwater geochemistry.
Peter Idstein received his Bachelor of Science degree in
Geology at Eastern Illinois University. He is completing his
Master of Science degree in Geology at Eastern Kentucky
University. Mr. Idstein has spent one year working at the
Florida Sinkhole Research Institute conducting studies on
conduit dominated groundwater flow. He has also spent a year
working for Ewers Water Consultants conducting dye tracing
studies and continuous electronic monitoring studies in many
karst dominated and non-karst terranes.
Catherine A. Johnson is a project officer for the United
States Army Toxic and Hazardous Materials Agency (USATHAMA).
She graduated with General Honors from the University of
Maryland with a B.S. degree in Chemical Engineering. Her duties
at USATHAMA involve managing Preliminary Assessments/Site
Inspections, Remedial Investigations/Feasibility Studies, and
RCRA Facility Assessments/Investigation studies at Army sites.
Ms. Johnson has been the project officer at the Fort Campbell,
Campbell Army Airfield site for the last two years. During that
time she has managed extensive field work including; geophysical
investigations, sediment, Water, and soil sampling, and
monitoring well installation.
305
-------
The Transmission of Light Hydrocarbon Contaminants In Limestone
(Karst) Aquifers
Ralph O. Ewers, Anthony J Duda, Elizabeth K. Estes, Peter J Idstein
and Ratherine M. Johnson.
1. What are the major difference between the physical and
chemical properties of the dyes and the light hydrocarbons used in
you study? Why not try to use a tracer that has properties
similar to the hydrocarbons? Has there been a chemical
transformation of the hydrocarbons in the limestone which affects
their properties?
ANSWER- The dyes are water soluble substances that are non-toxic.
The light hydrocarbons are toxic immiscible organic substances.
I know of no non-toxic immiscible hydrocarbons that regulatory
agencies would allow me to place in an aquifer in the quantities
that would be required. In any case, tracers with the properties
of the light hydrocarbons would not give us the answers that we
need in the time required. They would be poor tracers.
I suspect that there are many chemical and biochemical
transformations which have occurred. However, I do not believe
that such changes are the principal reason for their apparent
retention in many karst situations. I believe that immiscibility
and density are the primary reasons. The demonstrated travel times
are too fast for these processes to be effective.
2. What were the dissolved BTEX levels associated with the Little
Caesar Spring? Was dilution sufficient to limit benzene level to
below 5 micrograms per liter. What were dissolved levels at low
flow conditions?
ANSWER- I do not know the BTEX levels because they were below the
level of detection at the time.
Demonstrably, the separation system in the aquifer and the
dilution were sufficient. Hundreds of gallons of gasoline were
probably involved, and the spring flow was relatively small, a 2-
3 cubic feet per second at maximum. Separation of the floaters
seems to be very important.
306
-------
VELOCITIES OF PIEZOMETRIC WAVES CAUSED BY
PUMPING IN KARSTIC AQUIFERS
C. Drogue
Laboratoire d'Hydrogeologie, Universite de Montpellier II, Place E. Bataillon, 34095 Montpellier
Cedex 05, France
Abstract
The velocities induced by pumping in a karstic area were studied at a test site in 20 boreholes at
5m intervals. The site was 4.6 km from a spring. Pumping of 1 to 2 m^ s ~1. was performed at the
spring. The time for the effect of pumping at the spring to reach the boreholes varied from 6 to 8
minutes, corresponding to an apparent velocity of 9.6 to 12.8 m s~V Wave propagation probably
occurs in drainage "conduits" (connected to the spring) at a velocity related to physical
parameters such as the water (modulus of elasticity), rock (Young's modulus) and the
characteristics of the conduits (equivalent diameter) and the type of flow (under head or open
surface). The recorded velocities show that flow was not entirely under head since they would
have been much higher in this case because of the rigidity of the limestone. At the other
boreholes (which boreholes) impact times varied from 26 minutes to 5 hours, corresponding to an
apparent velocity of 2.9 to 0.2 m s . These observations are interpreted using a conceptual model
with double diffusivity taking into account low diffusivity fissured blocks and high diffusivity
cracks. The boreholes directly affected by pumping at the spring penetrated a fracture in the
cracks with blocks in the model. The boreholes with low velocity are in the blocks themselves and
far from the drain. One of the results of this difference in piezometric behaviour is that certain
apparent piezometric gradients are reversed between pumping and periods of stoppage.
Introduction
Measurements of the velocity of piezometric waves caused by pumping in a karstic aquifer can
contribute to better surveying of the heterogeneity of the medium. This is particularly fruitful
when numerous piezometers are installed in an aquiferous zone and the fissure structure can be
surveyed.
General considerations
In porous media, pressure transfers can be interpreted by solving the diffusivity equation (Bear,
1972). There are two possible procedures for karstic aquifers:
307
-------
a) The karst is identified as a continuous medium equivalent, as has been proposed for certain
fissured rocks (Maini & Ockins, 1977). The problems are handled like those of porous media. This
is possible if the area considered is very large in relation to a basic grid representative of the
high conductivity network. However, even more than in fractured rock, heterogeneity is found at
all scales in karst (Drogue, 1988) and it is difficult to consider it as a porous medium.
b) The second procedure takes into account the geometrical and hydraulic characteristics of
aquiferous karstic networks. However, schematization is necessary because of the extreme
complexity. This is the case of the conceptual models proposed by fissured aquifers. In these
models, matrix porosity is combined with fracture porosity (Barenblatt et al, 1960; Wilson et al,
1974; Shapiro & Anderson, 1983; Huyakorn et al, 1983; Moench, 1984).
A double porosity fissure model is used for karstic aquifers (Drogue, 1980, 1988). Little-karstified
fissure blocks with low hydraulic conductivity (K' = 10"7 to 10~9 m s'1) are separated by cracks or
karstic channels with high hydraulic conductivity (K' = 10~3 to 10'1 m s'1) (Fig. 1). The model can
be used in many karstic aquifers to interpret the phenomena observed: piezometry,
hydrochemistry and temperature patterns (Drogue, 1985, Drogue, forthcoming).
The channels between the blocks form the aquifer drainage network which converges on the
channel feeding the spring. If the channel is under head it can be considered as a conduit. The
velocity of a pressure wave can then be expressed as follows:
c =
where: p = voluminal mass of water, K = modulus of elasticity of the water, D = diameter of the
conduit, E = Young's modulus of the conduit material, e = thickness of the wall
In the extreme case of an indeformable conduit, the ratio D/Ee is zero and C = 1425 m s~* (at 15°C),
where p = 10'2 and K = 2 10~8 kg f mr2.
This velocity that of sound under water - is certainly not attained in karstic aquifers. Indeed,
these rocks undergo deformation by the piezometric variations caused by rain water alone. This
was demonstrated by extensometric measurements performed in caverns (Crochet et al., 1983). In
addition, the channels are often combined with cracks with open surfaces corresponding to the
blocks in the model. These cracks contain damping volumes which slow the pressure waves.
Aquiferous non-karstic soil cover may also play the same role through the drainage effect.
Experimental conditions
The experimental site consists of 20 boreholes 60 m deep laid out in a 500 m2 area (Fig. 2). The
boreholes are 5 to 7 m apart. They run through much-fractured, karstified Upper Jurassic and
Lower Cretaceous limestone forming a free aquifer. All the boreholes reached the saturated zone.
Piezometric levels vary from 20 to 50 m according to the season. Various work at the site - test
pumping, tracing, logging, piezometric monitoring over a period of several years and
hydrochemical investigations have verified that there is hydraulic continuity between all the
boreholes. It has also been possible to site each borehole in relation to the drains and blocks of the
conceptual model (Drogue & Grillot, 1976).
The spring at which pumping is carried out is 4.6 km from the experimental site. The natural flow
ranges from 0.4 m3 s'1 at low water to 8 m3 s'1 during flood periods; average flow is 2 m3 s"1. It
drains a 400 km2 aquifer within which the experimental site is located. Numerous faults between
the spring and the experimental site place karst and non-karstic formations in contact with each
other.
Exploitation of the spring with discontinuous pumping started in 1968 at 0.8 m3 s"1. Pumping
discharge has been up to 1.5 m3 s'1 in recent years.
308
-------
Fig. 1. Conceptual model of a double
fissural porosity karst. Blocks with low
permeability (B) are bounded by open
cracks with high hydraulic conductivity
(C). When the karst outcrops, the upper
part of the aquifer is extremely fissured (A)
because of decompression and surface
weathering.
Meteorological station
Laborato
Fig. 2. The experimental site
(Terrieu, karst formation in
the South of France) equipped
with 20 piezometric boreholes
60 m deep. Only the boreholes
mentioned in the text are
numbered. They are repres-
entative of the various
piezometers monitored at the
site.
Phreolic high
Phreatlc low
Ground water flow to the spring
Fig. 3. Piezometric evolution in
several boreholes during the
first pumping test: A, pumping
at the spring; B, pumps
stopped.
I9;5 .
I
N
e
5.
20.0
hours
309
-------
48-
2h 3h 4h
Time (hours)
Fig. 4. Piezometric evolution during the second pumping test. 1, piezometry at the spring; 2,
piezometry at borehole P20; 3, piezometry at borehole P9; a, 2-minute response time; b, 24-minute
response time.
Drain
PIO
X
Block
P20,
1
I I I I I1
LJ _L'._i
rt+4
Spring
Z*- (pumping
Fig. 5. Diagram (based on piezometer velocity) of the spring and boreholes P20, P9 and PIO in a
double porosity fissure model.
block
drain
Time
Fig. 6. Piezometric data for a drain and an adjacent block during pumping. Dephasing of the
piezometric movement accounts for the inversion of the piezometric gradient (a).
310
-------
Experimental results
Preliminary experiment
This preliminary experiment was performed with natural flow of 0.5 m^ s~* at the spring.
Pumping discharge was 0.8 m^ s~l with 0.5 m maximum drawdown at the spring.
Piezometric variations at the experimental site displayed the following features (Fig. 3):
- drawdown was considerable at some boreholes: 0.3 to 0.4 m. In contrast, drawdown was slight or
non-existent at other boreholes only 5 or 10 m from the former;
- the piezometric levels in the boreholes little-affected by pumping were higher than those of the
boreholes in which there was considerable drawdown. When the pumps were stopped, the
apparent piezometric gradients between these boreholes were reversed.
In order to confirm and complete these observations, pumping was performed at high discharges to
cause greater drawdown at the sight and enable better observation of the phenomena.
High discharge pumping
After pump stoppages limited to 30 min because of water requirements, piezometric variations
were triggered by instantaneous pump starting at discharges of 1.5 and then 2 m^ s~l. Three
piezometers representative of the different domains of the double porosity model were observed:
P20: on a drain,
P9: in a block and near a drain,
P10: in a block.
The results are shown in the table below.
Response time (min)
Apparent velocity
(ms'1)
P20
2
38
P9
24
3.2
P10
approx. 120
approx. 0.6
Drawdown was too slight for accurate measurement at borehole P10.
Discussion
It should first be noted that the hydraulic distance between the spring and experimental site is
doubtless greater than the geographical distance of 4.6 km because of twists and turns of the
network. Real wave velocities were thus greater than the apparent velocities in the table above.
Borehole P20 responded very rapidly. It cut through a drain which must be connected to the
karstic network which feeds the spring. P9, which is on a block but near a drain reacted more
slowly. P10, located in a block, reacted much later. These results thus confirm the suitability of
the model for the interpretation of the velocities of piezometric waves. The respective positions
of the spring and the piezometer boreholes can be schematized in the model (Fig. 5).
Dephasing times are related to reductions in the amplitude of piezometric fluctuations. Thus,
stopping pumping between discharges of 1.5 m3 s-1 and 2 m3 s-1 caused piezometric rises of 0.65 m
in borehole P20, 0.25 m in P9 and approximately 0.03 m in P10. There were thus apparent
piezometric gradients between boreholes. This is explained by the dephasing of the piezometric
variations between a drain and an adjacent block during short duration pumping (Fig. 6). This
phenomenon might be a characteristic of the karst. It is interesting to observe that these
piezometric inversions were similar to those observed in the same boreholes during a flood caused
by infiltration (Drogue, 1980). In the latter case, rapid piezometric recharging caused a
311
-------
temporary rise in level in the drains in relation to the piezometric levels in the blocks. This
piezometric inversion phenomenon also occurred between the spring and borehole P10 and was
caused by the low apparent piezometric gradient between the site and the spring (8.7 10~2) when
pumping was stopped.
Conclusion
Piezometric wave velocities thus provide extremely accurate information about the aquiferous
structure of the karst and its degree of heterogeneity. The representativeness of a piezometer in
aquifers of this kind should thus be examined attentively. The hydrodynamic calculations made
using data from such a piezometer only apply to a certain domain of the aquifer.
References
BARENBLATT, G.I., IV.P. ZHELTOV and I.N. KOCINA, 1960 Basic concepts in the theory of
seepage of homogeneous liquids in fissured rocks.
J. Appl. Math. Mech. Engl. Transl., v. 24, p.1286-1303.
BEAR, J., 1972 - Dynamics of fluids in porous media.
Elsevier, New-York, 764 p.
CROCHET Ph., LESAGE Ph., BLUM P.A., VADELL M., 1983 - Extensometric deformations linked
with rainfalls.
Annales Geophysical, 1, 4-5, p.329-334.
DROGUE C. and GRILLOT J.C., 1976 Structure geologique et premieres observations
piezometriques & la limite du sous-systeme karstique de Terrieu (Perimetre experimental).
2e coll. Hydrologie en pays calcaire. Besangon, France. Ann. Sc. Universite de Besangon,
fasc. 25, 3e serie, p.195-209.
DROGUE C., 1980 - Essai d'identification d'un type de structure de magasins carbonates, fissures.
Application £ 1'interpretation de certains aspects du fonctionnement hydrogeologique.
Mem. Soc. Geol. France, No.ll, p.101-108.
DROGUE C., 1985 Geothermal gradients and ground-water circulation in fissured and karstic
rocks : the role played by the structure of the permeable network.
J. of Geodynamics, 4, p.219-231.
DROGUE C., 1988 - Scale effect on rock fissuration porosity.
Environ. Geol. Water Sci. vol.11, No.2, p.135-140.
DROGUE C., forthcoming, 1992 Hydrodynamics of karstic aquifer, results from experimental
sites (Mediterranean karsts in the South of France)
"Hydrogeology of Selected Karst Regions, I.A.H., vol.13.
HUYA KORN P.S., LESTER B.H. and FAUST C.R., 1983 Finite element techniques for modeling
groundwater flow in fractured aquifers.
Water Resour. Res., v.19, No.4, p.1019-1035.
MAIM and OCKINS, 1977 An examination of the feasibility of Hydrologie isolation of a high
level waste repository in crystalline rocks. Invited paper, geological disposal of high
radioactive waste session, annual meeting of Geological Society of America, Seattle.
MOENCH, A.F., 1984 - Double porosity models for a fissured groundwater reservoir with fracture
skin.
Water Resour. Res., v. 20, No.7, p.831-846.
SHAPIRO A.M. and ANDERSON J., 1983 Steady state fluid response in fractured rock: a
boundary element solution for a compled discrete fracture continuum model.
Water Resour. Res., v.10, No.2, p.328-335.
312
-------
Solution Mining And Resultant Evaporite
Karst Development In Tully Valley, New York
Paul A. Rubin1, John C. Ayers2, and Kristin A. Grady3
'Oak Ridge National Laboratory, Oak Ridge, Tennessee
2Vanderbilt University, Nashville, Tennessee
3Law Environmental, Inc., Albany, New York
The views and conclusions put forth in this paper
are expressly those of the authors and do not
necessarily reflect those of the State of New York,
its employees, experts and/or agencies.
Abstract
A solution mining operation was conducted in Tully Valley, New York
from 1889 to 1988. In excess of 37 million m3 of halite was removed from
335 to 518 meters below the ground surface. An interbedded sequence of
gypsum, shales, limestones, and sandstone overlie the halite beds. This
sequence is capped by thick, unconsolidated deposits of till, sand and
gravel, and lacustrine clay.
As a result of this mining, large void cavities were created, followed
by numerous fractures extending upward to the ground surface. The resulting
settlement area is in excess of 550 hectares. Within this area sinkholes
formed, gaping fractures developed and streams were pirated into the
subsurface. Interformational mixing of groundwater now occurs between
formerly separate flow systems, providing substantial recharge to deep
formations.
Some 2 kms downvalley of the brine fields, in a smaller settlement
area, mud "volcanos" effuse weakly saline groundwater that flows into
Onondaga Creek. The clay fraction of the effluent gives Onondaga Creek the
appearance of chocolate milk for the =26 kms it takes to reach Onondaga
Lake. The location of the mud volcanos appears to coincide with an up-
valley moving salt front.
The number of mud boils and their areal extent has substantially
increased since the onset of brining operations. By characterizing the
chemistry of groundwaters in local formations and performing mixing
calculations based on mass balance, the volcano effluents were shown to
represent a mixture of groundwaters from 3-4 formations. Several working
hypotheses are advanced and critically evaluated in an effort to define the
dynamics necessary for rapid mud volcano growth in a karst setting.
Location and Geology
Tully Valley is a glacially scoured Finger Lake valley located 24 km
south of Syracuse, New York (Fig. 1). The valley floor is blanketed by
Pleistocene ice marginal deposits 18 to 134 m thick consisting of
gravel/clayey gravel, a middle thick rhythmic clay sequence and a basal
sandy gravel (Getchell, 1983; Mullins et al., 1991). Bedrock consists of an
interbedded sequence of shales, limestones, and sandstone overlying four
evaporite beds ranging up to 21 ± 8 m in thickness (Fig. 2). All beds dip
approximately 2° to the south.
313
-------
SURFACE HYDROLOGY AND LOCATION
MAP OF TULLY VALLEY, NY
. BRINE WELL OR
CORE HOLE
VESPER
FIELD
WEST FIELD
Sand Volcano U Volcano Area Ram Pum
0 500 1000 meters
Figure 1:
General location map of the features in Tully Valley, including
brining fields and associated collapse.
-------
SCHEMATIC CROSS-SECTION
SOUTH
Showing Collapse Structures and Flow
Regime of Tully Valley, New York
NORTH
CARDIFF
WELL
50
100
1000
2000 meters
D
POST GLACIAL LACUSTRINE AND GLACIAL OVTWASH
Clayn and Sllte, Sand and Grovel
DEVONIAN SHALES - Hamilton Croup
Skeneatellea and Marcalliu »h*lfi
ONQNDAGA LIMESTONE
ORISKANY SANDSTONE
DEVONIAN/SILURIAN LfUESTONE AND DOWIOTSS
Uanliua Lm./Readout and
Coblesldll Dolomites and Bertie Fm.
SILURJAN-CAHILWS SHALE
with lowvr gypsum bed
GYPSUM BED
SALT- *
* B*"1", 3 ««1 4"1
Beda plotUd tc6elher
- Syracuse Pm.
SUurimn Veroon Shale
Disrupted beds =
shale, gypsum
Infiltrating meteoric
and ground water
flow direction
of brine waters
Regional Groundwater flow
Brine Potentiometric
Surface
1 •*. *~ FRACTURE
Original
land surface
Brine Cavities
and brine /tow path
Figure 2: Generalized cross-section of Tully Valley showing stratigraphy, collapse
structures and flow regime. See text for further discussion.
-------
Solution Mining Operation and Resulting Subsidence
A solution mining operation was conducted in Tully Valley from 1889
until 1988. Until the late 1950's salt was usually mined by injecting fresh
water down one well to the formerly dry salt horizons and then pumping or
air lifting brine out another well. Brine was piped downvalley to Syracuse
for the production of soda ash. Some 150 wells were drilled, many in close
proximity (= 40 m) to each other. Continuing brine extraction caused the
solution cavities to expand and interconnect. It was not uncommon to drill
new wells and find that uncontrolled solutioning had already removed some
or all of the upper halite beds (Fig. 3).
Seven large sinkholes formed catastrophically in the Tully brine fields
between 1949 and 1980. A monument network was installed and periodically
surveyed from 1959 to monitor subsidence. Although the extent of early
subsidence will never be known, a steady growth in the subsidence area has
been documented. Today, this settlement area is in excess of 550 hectares,
has subsided up to 14 meters vertically, and is progressively expanding
outward from the cavities. A comparison between the 37 million m of halite
extracted and the total settlement volume of 5 million m3 calculated as of
1982 (Tully, 1983) indicates that only 14 percent of the total possible
settlement volume has occurred.
Such extensive settlement resulted from fracturing and subsequent
collapse of beds overlying the brine cavities. Evidence supporting this
includes the continued growth of the large surficial settlement area,
individual sinkhole formation, the presence of large gaping fractures open
to the surface, sheared well casings, and the piracy of streams (e.g.,
Emerson Gulf) into the subsurface. In essence, extensive brine cavity roof
collapse and bedrock disruption has resulted in a fracture network of high
permeability.
E WELL GROUPING OF THE EAST FIELD
YEAR'DRILLING COMPLETED
1889 1889 1B90 1890 1B82 1930 1934 1937 1940 1941 1958
3
1
w
g-iao
2
-ISO
Aver, top of
-- 1st Salt Bed
Aver, top of
2nd Salt Bed
Aver, bottom of
2nd Salt Bed
Aver, top of
3rd Salt Bed
Aver, bottom of
Salt Beds
WELL LOCATIONS
RE-1
S-2
o
S-1
NE-2
o 4 E-2
t
NE-3 '
~^J4E
0™^""ToOmel
E-l E-2 E-3 £-4 E-5 _j NE-2 NE-3 NE~» NE-1 RE-1
•
B
KEY
Depth to first SALT
encountered
SALT
Figure 3:
Time versus depth plot of the E well grouping. By 1937 when NE-3 was
drilled, the first salt bed was already uncontrollably dissolutioned by
brining activities.
316
-------
Hydrologic evidence supports the presence of at least one major
confining bed near the top of the Manlius Limestone (Fig. 2). Large
artesian pressures have been encountered when penetrating this horizon
during drilling. Recent subsidence and fracturing has led to
interformational mixing of formerly separate hydrologic flow systems and
greatly increased the natural recharge to all subsurface formations.
Increased recharge meant that by the late 1950's it was no longer necessary
to pump fresh water down into the salt horizons to obtain brine during
mining operations.
Further evidence of the extent of subsidence-induced fracturing and
interformational mixing is found by comparing potentiometric surfaces
between pumping brine wells and glacial aquifer wells. For example, brine
wells H-7 and MV-5 are 432 and 428 meters deep respectively, while CATO-4
penetrates deep glacial deposits some 1.8 kilometers from well H-7.
Measurement of the potentiometric surfaces of these wells over a period of
four years by Allied-Signal and the U.S.G.S. reveals similar mimicking
responses to brine pumping, although they are some 1.2 kilometers apart and
differ in depth by approximately 300 meters. During a period of high brine
field pumping (August 1985) the water table in CATO-4 was drawn down 15.7
meters below its "normal" flowing elevation. At the same time the
potentiometric surface in H-7 was drawn down approximately 38 meters.
Continued observations revealed that the magnitude of fluid level
fluctuations in these wells also significantly decreased when pumping rates
decreased.
The similar hydrologic response of bedrock and soil ag_uifers to pumping
suggests that the ==300 m between them is now hydraulically connected by the
increased permeability resulting from subsidence-induced fracturing. The
damage to the structural integrity of beds beneath and adjacent to the
brine fields and the resultant interformational mixing may now be so
extensive that they are irreparable. Similarly, the piracy of surface water
from Emerson Gulf and the increased recharge through formerly competent
lacustrine clays and bedrock must increase the flow through subsurface
formations.
Appearance of Mud Boils and Sand Volcanos
Some 2 kms downvalley of the brine fields, in a smaller settlement
area, sand "volcanos" effuse weakly saline groundwater that flows into the
Onondaga Creek. Projection of geologic units described in distant wells
into this area and observation of the growth pattern (N to S, and E and w)
of these mud boils suggest that they may mark the present location of the
subsurface salt front. Figure 4, time sections 1-3, depict the natural
southward migration of the salt front.
An early newspaper documents the presence of a sand volcano at Otisco
Road (Fig. 1) in 1899. These early sand volcanos were areally limited.
Perhaps they were a reflection of the earliest down valley effects of the
solution mining operation. The number of mud boils and their areal extent
have substantially increased since the onset of brining operations.
Individual effusion features vary in size from small mud boils to broad,
flat cones (sand volcanos) measuring 12 m in diameter and 1 m in height.
Historic accounts and the interpretation of historic aerial photography
reveal that the mud boil areas expanded rapidly during solution mining
operations. Sometime between July 27, 1936 and October 15, 1951, the clay
fraction of the mud boil effluent caused Onondaga Creek to change from
clear to turbid. Onondaga Creek now has the appearance of chocolate milk
for the =26 kms it takes to reach Onondaga Lake.
317
-------
Time Section 1 -- (>125,000 years ago)
d Tully Valley pre-Wisconsinan glaciation d
a: a:
SOUTH | |g NORTH SOUTH <
8 BRINE FIELD u3 *
sS R
Approx. 15 km
BENJAMIN £
ARM WELL C o
Down valley
Time Section 2 — (<10,000 years ago)
Tully Valley Postglacially
(soil deposits in place)
NORTH
SOUTH <
Time Section 3--(circa 1900)
The beginning of salt mining
operations in Tully Valley
Shdles
10
Limestones
Lacustrine Clays — Otisco Clay
Sand and Gravel or Till
- inferred contact between clay deposits
and sand and gravel deposits
Oriskany Sandstone
V*
\
Subsidence Fracture with
infiltrating groundwaters
Confined groundwater
Infiltrating surface "waters
Mixed groundwaters
Salt Front - headward
retreat due to dissolution
Brine
Time Section 4—(circa 1950)
Collapse of the brine field area.
New circulation flow path(s) established
NORTH SOUTH <
Cm
60
100
SCALE
1000
2000 meters
Figure 4: Time sequence diagrams showing the flow regimes and sequence of events
that led to the present day. See text for further discussion.
NORTH
CARDIFF
WELL
-------
Massive subsidence in the recharge area coincides in time with the
greatest sand volcano activity. The largest and most prolifically effusing
sand volcano field, approximately 400 meters south of Otisco Road, was not
present as of April 28, 1967. By April 29, 1972 it measured an irregular =
60 m by 25 m; by March 28, 1981, = 160 m by 73 m; and on November 29, 1989,
191 by 175 m. Continued settlement, to a depth of 6 m, has further enlarged
this area in the last two years. Much of the growth of the mud boil areas
appears to develop along the strike of the inferred location of the salt
front.
The physical setting of the mud boils is inconsistent with natural
springs. Springs typically occur in one or a few closely spaced discharge
points and are stable in size and discharge location over time. The
artesian Tully Valley mud boils have migrated rapidly southward along
Onondaga Creek, and southwestward along a tributary of Onondaga Creek in
the last 80 years. This is not in keeping with known short term geologic
phenomenon.
Chemistry of Groundwaters and Sand Volcano Effluents
The chemical relationships among uncontaminated groundwaters in Tully
Valley and sand volcano effluents (SVE) can be used to identify the source
of SVE. Stiff diagrams were found to have limitations when trying to define
the chemical relationships between groundwaters and SVE. Because there are
many groundwaters that are potential sources, and most have distinctive
chemistries, multivariate models were necessary to differentiate among
them. The multivariate models give quantitative results in the form of
significance levels of chemical discrimination and calculated mixing
proportions of groundwaters.
The chemical analyses were taken from three main sources; the New
York State Department of Law; the New York State Department of
Environmental Conservation; and a thesis by Noble (1990). Many samples from
Noble (1990) were taken just outside Tully Valley, but from the same
geological formations. Chemical analyses with > 5% error in ionic balance
and > 10% error in mass balance were deleted from the sample database.
Water samples were classified into hydrostratigraphic units based on
geologic interpretation and well logs. The five units present in Tully
Valley in order of increasing depth (Fig. 2) are Glacial Till, Devonian
Shale, Silurian-Devonian (SD) Carbonate (Onondaga-Devonian/Silurian
carbonates), Evaporite (Silurian shale-Syracuse Fm.), and Vernon Shale
(Noble, 1990). The compositions of dilute groundwaters in the Glacial Till
and Devonian Shale units are similar, therefore samples from these units
are grouped together with the name Near-Surface waters. Average
compositions of water from these four units and the SVE are given in Table
1. The SVE contain significant halite and gypsum components.
Statistical Tests
If SVE derive from a single hydrostratigraphic unit, their compositions
must be the same within error as samples from that unit. Because samples
were analyzed for > 8 chemical species, a multivariate technique called
multiple discriminant analysis was used to test for statistical differences
in composition between each unit and SVE. Multivariate statistics give at
least as good a discrimination as any univariate technique. Although some
variables (species concentrations) have non-normal distributions for some
hydrostratigraphic units (log transformations do little to correct this),
this should not affect the results of the discriminant analysis, although
the significance levels obtained may be questionable (Le Maitre, 1982).
319
-------
TABLE 1
Average Chemical Compositions of Waters in Tully Valley*
Element
Near-
Surface
n=48
SD Car-
bonate
n=5
Evap-
orite
n=5
Vernon
Shale
n=6
SVE
n=6
Na
Ca
Cl
S04
Sr*
Mg'
HCO3*
1.05
(0.66)
37.4
(64.2)
73.5
(34.3)
54.4
(80.9)
35.1
(35.7)
0.69
(0.96)
17.3
(9.15)
271
(85.3)
0.94
(0.46)
18.2
(12.2)
207
(180)
51.6
(30.2)
320
(405)
3.0
(1.3)
32.6
(17.2)
314
(35.1)
232
(20.2)
110,000
(10,500)
1880
(198)
171,000
(19,900)
2300
(1150)
76
(11)
187
(106)
75.7
(50.9)
11.1
(12.4)
51.0
(51.6)
392
(144)
117
(130)
829
(390)
5.4
(2.5)
32.3
(16.0)
274
(73.4)
6.47
(2.00)
1800
(963)
241
(64.1)
3500
(1610)
329
(136)
84
(16)
126
(29.5)
90.0
(51.4)
Model
#15
6.47
1900
241
2980
329
4.0
37.1
401
Model
#2*
6.48
2650
239
4160
350
4.6
35.4
306
Concentrations in mg/1; standard deviations in parentheses.
Element not included in model calculations (not conserved).
Model #1 proportions of end-members (calculated using non-linear least
squares): SD Carbonate = 0.91, Near-Surface = 0.062, Evaporite =
0.024, SSR (sum of squares of residuals) = 2.3 E6.
Model #2 proportions of end-members (calculated using linear
programming and multiple linear regression): SD Carbonate = 0.38, Near-
Surface = 0.53, Evaporite = 0.017, Vernon Shale = 0.11, SSR = 3.9 E5.
The objective of discriminant analysis is to find functions that
maximize the differences between chemical groups. Specifically, the
functions are linear combinations of variables that maximize the ratio of
between-group to within-group variances (Le Maitre, 1982). Here the groups
are the four hydrostratigraphic units and SVE and the variables are the
concentrations in mg/1 of the ions Na, K, Ca, Mg, Cl, and SO4. Use of
element ratios did not improve the discrimination between groups.
Calculations were made using the program DISCRIM in the SAS software
package (SAS Institute, © 1988) using equal weighting for each group.
Over 99% of the variability in the sample data set is accounted for by
the two discriminant functions CAN1 and CAN2:
CAN 1 = 0.0011*Na + 0.045*K - 0.00033*Ca - 0.0063*Mg -
0.00043*C1 - 0.0016*SO4
CAN 2 = 0.000025*Na - 0.0075*K + 0.011*Ca + 0.0083*Mg
-0.00014*C1 + 0.00070*SO4
320
-------
Species concentrations are in mg/1. The angle between these two vectors is
90°, which allows plotting of the values of CAN1 and CAN2 for each sample
on an x-y diagram (Fig. 5). Waters from Near-Surface and SD Carbonate are
not significantly different. However, compositions of SVE and waters from
the four hydrostratigraphic units are distinctive. The F-statistic shows
that SVE are significantly different from all units at the 99% significance
level. The SVE therefore cannot form by direct sampling of a single unit.
If waters in SD Carbonate are more saline than our analyses indicate,
SVE may be derived directly from this unit. Samples taken from the SD
Carbonate unit during drilling of the M-series wells (Phalen, 1928) were in
fact more saline than modern samples taken from wells penetrating SD
Carbonate. However, multiple discriminant analysis shows that the SVE and
M-series samples are compositionally different at the 99% significance
level.
Since SVE cannot be derived from a single hydrostratigraphic unit, they
must represent a mixture of waters from the various units (chemical
end-members). In Fig. 5, lines connecting the end-members define a mixing
polygon. Any mixtures formed from the end-members must lie within the
polygon. The location of average SVE suggests mixing of Vernon Shale + SD
Carbonate + Near-Surface ± Evaporite. The next section discusses
quantitative tests of these general conclusions.
8
7
6
CN 5
O 3
2
1
Sand Volcanos
Surface Brines
SD Carb
Near—Surface
-3 -2 -1 0
CAN1
Figure 5: Chemical variation diagram relating compositions of the sand
volcano effluents and surface brines to the end-member groundwaters in
Tully Valley (Evaporite is off-scale). CAN1 and CAN2 are the functions
obtained from the multiple discriminant analysis (see text). Mean values
for SVE and surface brines (filled circles) and end-members (open circles)
are plotted with ± 1-sigma error bars. Lines connecting the end-members
define the mixing polygon.
321
-------
Mass-Balance Calculations
The process of groundwater mixing can be modelled by mass-balance if
species behave conservatively. In the mass balance model, the average
composition of SVE represents the mixture and the average compositions of
the four units represent the end-members (Table 1). Species are conserved
if no dissolution or precipitation of minerals occurs between mixing and
sampling. On short time scales precipitation is more likely to occur, but
only if the mixture is saturated in a mineral, i.e., the saturation index
(SI) is > 1. Calculation of SI for various minerals (Truesdell and Jones,
1974) shows that SVE are saturated only in calcite, dolomite, and
strontianite. Species other than Ca, Mg, Sr and HCO3 should behave
conservatively. Because there must be as many species as end-members in the
mixing calculations, we were forced to use Ca as a variable, but the fact
that the Ca data fit the model suggests that Ca behaves conservatively.
Magnesium and Sr have anomalously high concentrations in SVE, and
comparison of analyses of filtered and unfiltered samples suggest this
results from clay particles that remain suspended even after filtering to
0.45 urn.
Two different models were used to calculate the mixing proportions, Xjf
of the end-members. The required constraints are that all Xj > 0 and ZXj =
1. Model #1 is a non-linear model based on the Marquardt-Levinson algorithm
(Press et al., 1985). The objective function minimized was S(M; - SX^)2,
where Mi is the concentration of element i in the mixture and Cy is the
concentration of element i in end-member j. Model #2 is a linear model that
uses multiple linear regression to minimize the same objective function.
However, since there is no way to constrain all Xj > 0 using multiple
linear regression, the first step is to choose those end-members that can
mix in positive proportions to form the mixture (Wright and Doherty, 1970).
This is accomplished using linear programming based on the simplex
algorithm (Press et al., 1985). The optimal proportions of the chosen
end-members are then calculated using multiple linear regression. The
disadvantage of this approach is that the objective function for the linear
programming step (minimize E(M; -SXjCy)) is different from that for the
multiple linear regression step. However, this discrepancy is unlikely to
have any effect when choosing the best mixing model.
Table 1 shows the results from Models #1 and #2. Although the
calculated proportions of end-members are different, both models fit the
data within error. Model #1 gives the proportions 0.91 SD Carbonate, 0.062
Near-Surface, and 0.024 Evaporite with SSR (sum of squares of residuals) =
2.3 E6. Model #2 gives the proportions 0.38 SD Carbonate, 0.53
Near-Surface, 0.017 Evaporite, and 0.11 Vernon Shale with SSR = 3.9 E5.
Note that the relative proportions of SD Carbonate and Near-Surface have no
meaning, since the compositions of these two end-members are not
significantly different (Fig.5). Also, the stated proportions of
end-members were calculated for average SVE, but the same calculations for
each sand volcano effluent show highly variable proportions of end-members.
This is to be expected in such a complex mixing scenario.
The fact that the solution to the mass balance problem is non-unique
(i.e., averaged SVE chemistry can be explained by several different models)
is not surprising, since there are too few high quality constraints on the
mixing problem. This results from having too many end-members and not
enough elements. Also, the variance of each element is too large,
distributions of some variable are non-normal, some elements are
correlated, and some end-members have similar chemistries. More data are
required before a choice can be made between mixing models. However, we can
conclude that SVE waters form from a mixture of groundwaters that are
present beneath the sand volcanos, apparently to a depth of at least 335 m.
This figure represents the depth to the Vernon Shale, as projected into
this area from distant wells.
322
-------
Many wells down valley of the sand volcanos at Otisco Road have
encountered dilute brine waters after penetrating the confining Otisco
Clay. Similar waters effuse from a large spring called the Sulfur Hole (=
3500 m north of Otisco Rd. and = 500 m west of Tully Farms Rd.), which has
been present near the valley flank for at least 130 years. We assume that
its waters probably rise along the clay/valley flank contact or through
fractures in the clay where it thins out. Discriminant analysis of these
surface brines reveals that they are chemically similar to the Vernon Shale
(Fig. 5), yet distinct from SVE at the 99% significance level. The average
chemistry of these waters is modelled as a mixture of 97.4% Vernon Shale
and 2.6% Evaporite. Thus, the natural brines appear to rise undiluted from
the Vernon Shale and up-valley Evaporite units because these units are
closer to the surface than at Otisco Road and because the uppermost
confining beds have been erosionally removed.
Valley Axial Groundwater Flow
Discharge from the sand volcanos responds dynamically with significant
infiltration in Tully Valley. This may result from rapid flow through
highly transmissive bedrock conduits or by exertion of a large pressure
head on confined porous media.
A more complete geologic column to the south limits natural recharge to
brine-rich bedrock formations. Similarly, the thick lacustrine clays near
the surface of the valley act as a confining bed, severely limiting natural
surface infiltration. Several possibilities for the saline discharge flow
routes exist; all of which require substantial system recharge to highly
transmissive formations (Fig. 2).
Groundwater in deep regional flow systems moves slowly (Fetter, 1988).
Rapid flow through deep-seated formations is rare, unless there is
substantial fracturing at both the recharge and discharge ends of the flow
system. Naturally-occurring fractures are present at the up-dip end of the
Tully system, but the mining has provided access routes at the down-dip
(brine field) end (Fig. 2). To generate rapid flow through intervening
tight bedrock formations requires pathways such as fractures and/or the
dissolution of soluble geologic units (e.g., carbonates or evaporites).
A surcharged pressure head acting on highly fractured limestones is one
flow route option (e.g., brine field to volcanos). However, the rapid
increase in volcano discharge following system recharge would, if
transmitted from the brine field to the volcanos, require solution
conduits. The natural transmissivity of fractured limestone would not be
expected to communicate a hydraulic pressure wave over any appreciable
distance due to pulse damping under laminar/non-open conduit flow
conditions. Neither should the increased discharge from the sand volcanos
originate from valley flank sources, because recharge waters would have to
breach density stratified waters and one or more confining beds (e.g., clay
and/or bedrock). Fracturing of the thick Otisco clay along the valley flank
contact (e.g., Sulfur Hole area) would, if present, probably serve to
discharge artesian valley flank and axial pressures rather than serve as a
recharge site.
In evaluating whether the limestones would be preferred flow routes one
would have to assess the likelihood that solutional conduits were present
pre-glacially between the brine field and the volcanos. Solutional conduits
develop in limestone only where a pre-existing network of integrated
openings connect the recharge and discharge areas (Palmer, 1991).
Furthermore, carbonate dissolution requires that undersaturated water
remain in contact with soluble fracture or conduit walls. A minimum of
10,000 years is required to widen joints and partings enough to produce
significant permeability (Palmer, 1984; Dreybrodt, 1990; Palmer, 1991). A
typical water sample obtained from the Devonian shales contained 51.9 mg/1
calcium, or 129.6 mg/1 CaCO3 equivalents. For this concentration, the rate
323
-------
of wall retreat (e.g., joint enlargement) in limestone is calculated to be
0.05 cm'/liter in 10,000 years (Palmer, pers. comm.)• Based on an
evaluation of these considerations, it is our opinion that geologic
conditions have not been favorable pre- or post-glacially for the
development of significant solution conduits in the limestones.
The authors evaluated several other lines of reasoning in assessing the
likelihood that the main flowpath between the brine fields and the sand
volcanos occurs in the limestones. This included an evaluation of natural
infiltration pathways for calcite undersaturated waters. Geologic logs
reveal that little or no exposed limestone was present up valley of the
sand volcanos pre-glacially. Furthermore, much of the mud boil subsidence
has occurred along bedrock strike, which is inconsistent with typical
downdip vadose cavern development in this type of geologic setting (Palmer,
1991). Not only are the effusion areas migratory in nature, they exceed the
largest natural sinkhole size in New York State by well over an order of
magnitude. Conditions often attendant to sinkhole formation, such as
aquifer dewatering, were also not present. The variable mud boil discharge
in recent time would not be expected from a naturally deep-fractured karst
aquifer system if it was physically separated from large scale recharge.
The hypothesis might be advanced that the artesian pressures and
volcano effluents may derive from infiltration entering alluvial fans at
the outlets of either Rattlesnake Gulf or Rainbow Creek (Fig. 1). The
chemistry of the SVE would require this water to breach density stratified
brines and discharge up-valley against the regional hydraulic gradient. In
addition, significant recharge would have to completely penetrate the thick
confining Otisco clays. It is the combination of these unlikely events and
the areally wide and migrating distribution of the mud boils that virtually
removes all potential sources of system recharge other than the brine field
settlement area. It is more likely that any infiltration into the
Rattlesnake Gulf alluvial fan flows a short distance below the ground
surface, daylighting in the Ram Pump Spring area (Fig. 1).
The variable discharge rate and the chemistry of SVE are critical to
our understanding of likely source areas. The presence of gypsum and halite
components in the effluent is a clear signature telling us that at least
some volcano effluent must derive from the deep evaporite unit. The salt
content of SVE is diluted by shallow groundwater in sediments and bedrock
in the valley. Geochemically, the SVE represent mixtures of waters from all
of the hydrostratigraphic units beneath the sand volcanos to a depth of 335
m. SVE have moderately variable salinities, but groundwaters from the
hydrostratigraphic units show much less variation. The chemistry of SVE
suggests fluctuating amounts of dilution of deep saline and gypsiferous
waters by near-surface fresh waters. The brine must originate at a location
having a higher hydrostatic head than at the sand volcanos. Mixing occurs
during ascent along vertical pipes (joint or fracture flow paths) with
discordant hydraulic efficiencies.
Two separate flow regimes are probably responsible for the response of
volcano discharge following recharge events; deep evaporite and shallow
confined sediments. The high salinity of SVE observed during storm events
may indicate the development of a subsurface flow route, first through
fractures in the Vernon shale and then through salt dissolution and conduit
formation along the shale/salt contact (Fig. 2). Density-driven circulation
initiating at the brine fields may now rise along the fractured U-shaped
(areally) salt front beneath the volcanos. Jenyon (1986), discusses the
natural circulation systems attendant to salt fronts. Formation of
fractures at the salt front (Fig. 4, sections 1-3 and Fig. 2) normally
results in infiltration and downvalley flow. Instead, water rises at the
mud boils, almost as if the flow system was artificially short-circuited
(Fig. 4, section 4). The large sub-vertical arrow in Time Section 4 of
Figure 4 indicates upward flow (not against the regional flow gradients)
along salt front fractures resulting from this short-circuited circulation.
324
-------
Alternately, the rise of saline waters south of Otisco Road may directly
relate to the erosional removal of critical confining beds near the sand
volcanos. Whatever the flow path of the saline volcano component, the storm
discharge of the volcanos requires conduits or a confined porous media
capable of rapidly transmitting a recharge pressure pulse.
Several flow routes have been hypothesized to explain the rapid
increase in volcano effluent following recharge events, but the actual
pathway is unimportant. Visual observation of increased volcano discharge
during periods of high infiltration indicate that the source area responds
in tandem with surface recharge, operating through and beneath the thick
Otisco clay confining bed. It is unlikely that significant storm
infiltration through this clay can naturally occur to cause the increased
discharge observed at the mud boils. It is more likely that the sudden
surcharge on the hydrologic system from extensive infiltration in the brine
field settlement basin acts as a pressure pulse. Rubin has proposed several
means of verifying these relationships.
The sudden increase in pressure head, exerted through one or more
saturated formations, is partially released at the mud boils. While some of
the volcano discharge originates in the evaporite units, much of the
observed discharge may reflect a pressure wave being transmitted through
the sandy gravel aquifer confined beneath the Otisco clay. As sand is
entrained and removed from below the clay, a void develops until the
mechanical strength of the roof is exceeded. Ring fractures develop, the
subsidence area grows, and the efficiency of the exit pathways continues to
increase. The SVE are piped upward to an elevation well in excess of the
shallow water table common to valley centers. The only viable explanation
for the storm discharge observed at the sand volcanos (Fig. 6) is from
recharge through severely fractured brine field clay and/or bedrock. Here,
interformational mixing of groundwaters from massively fractured clay and
bedrock formations has resulted in a vast pressure head on a saturated flow
network.
Fluids from the deep saline beds are successfully diluted at this time.
It is the quantity of discharge that is of concern. This turbid water
degrades Onondaga Creek, Onondaga Lake, the aquatic ecosystem, and the
aesthetic quality of the valley.
Conclusion
Solution mining operations resulted in large-scale subsidence.
Subsidence-induced fractures serve to significantly recharge formerly
confined flow systems. This increased recharge has resulted in the
formation of sand volcanos downgradient. The bulk of the sand volcanos
appear to have formed by a sequential combination of: 1) increased recharge
to formations above solutioned out evaporite beds that has unnaturally
surcharged the flux through geologic units below the brine fields. This
flow through has created an unnatural circulation pattern that vents upward
at the sand volcanos. Pumping of brine increased recharge through formerly
competent and confining beds leading to the 2) solution of salt and the
transmission of a recharge pressure pulse through one or more geologic
horizons. This water 3) rises under artesian pressure through pre-existing
fractures underneath the sand volcanos, 4) mixing with other groundwaters
during ascent. Rising waters 5) suspend unconsolidated sediments (e.g.,
clay, silt, and sand) and discharge to Onondaga Creek via the sand
volcanos.
325
-------
lit
Figure 6: Typical Tully Valley sand volcano.
Diameter is approximately 2 meters.
Acknowledgement
Preparation of this manuscript occurred while the senior author was
employed in the Environmental Sciences Division at the Oak Ridge National
Laboratory. ORNL is managed by Martin Marietta Systems, Inc., for the U.S.
Department of Energy under contract DE-AC05-84OR21400.
"The submitted manuscript has been
authored by a contractor of the U.S.
Government under contract No. DE-
AC05-84OR21400. Accordingly, the U.S.
Government retains a nonexclusive.
royalty-free license to publish or reproduce
the published form of this contribution, or
allow others to do so. for U.S. Government
purposes "
References Cited
Dreybrodt, W., 1990, The role of dissolution kinetics in the development of
karst aquifers in limestone: A model simulation of karst evolution:
Journal of Geology, v. 98, no. 5, p. 639-655.
Fetter, C.W., (1988), Applied Hydrogeology, Columbus, Merrill Publishing
Company.
Getchell, F.A., 1983, Subsidence in the Tully Valley, New York, M.S.
Thesis, Syracuse University, 140p.
Jenyon, 1986, Salt Tectonics, Elsevier Applied Science Publishers Ltd.,
London, 191 pp.
326
-------
Le Maitre, R.W., 1982, Numerical Petrology: Statistical Interpretation of
Geochemical Data. Elsevier Publ. Co., Amsterdam.
Mullins, H.T., Wellner, R.W., Petruccione, J.L., Hinchey, E.J., and Wanzer,
S., 1991, Subsurface geology of the Finger Lakes region, Ebert, J.R.
(ed.), New York State Geological Association 63rd Annual Meeting
Field Trip Guidebook, p. 1-54.
Noble, J.M., 1990, A reconnaissance of the natural groundwater
geochemistry of Onondaga County, New York, Unpublished MS Thesis,
Syracuse University, 129 pp.
Palmer, A.N., 1984, Geomorphic interpretation of karst features in
Groundwater as a geomorphic agent, LaFleur, R.G. (ed.), Allen and
Unwin. Inc., London, p. 173-209.
Palmer, A.N., 1991, Origin and morphology of limestone caves: Geological
Society of America Bulletin; v. 103, p. 1-21.
Phalen, W.C., 1928, Memo MK-87, Underground water conditions in new wells
at Tully, The Solvay Process Company.
Press, W.H., Flannery, B.P., Teukolsky, S.A., and Vetterling, W.T., 1985,
Numerical Recipes: The Art of Scientific Computing, Cambridge
University Press, New York.
SAS, 1988, SAS Institute Inc., Gary, N.C., 27512, USA.
Truesdell, A.H., Jones, B.F., 1974, WATEQ, a computer program calculating
chemical equilibria of natural waters, U.S. Geological Survey Journal
of Research, v. 2, p. 233-248.
Tully, W.P., 1983, Relationship of brining operations in the Tully Valley
to the behavior of groundwater and geologic resources, Summary
Consultants Report, prepared for The Solvay Process Allied Chemical
Corporation, Camillus, NY.
Wright, T.L., Doherty, P.C., 1970, A linear programming and least squares
computer method for solving petrologic mixing problems, Geol. Soc.
Amer. Bull., v. 81, p. 1995-2008.
327
-------
Solution Mining And Resultant Evaporite
Karst Development In Tully Valley, New York
Paul A. Rubin, John C. Ayers, and Kristin A. Grady
1. What is the likelihood that the sand boils are a result of the
expulsion of water due to progressive collapse of cavities created
by ongoing solution along the upper portion of the salt by water
entering at cracks?
The process of cavity collapse, exemplified by sinkhole formation in
limestone karst settings, occurs at either an extremely slow rate as
soil is sapped downward and removed or as a catastrophic event. The
downward sapping of sediments and rock debris from above the salt
horizons would occur so slowly that it could not be expected to
continuously displace quantities of water. If catastrophic cavity
collapse were the driving force behind sand volcano effusion, then
sand boils would form instantaneously, resulting in a high effusion
rate immediately following cavity collapse. This effusion rate
would rapidly drop to zero. Without a large and continuous fresh
supply of rock or sediment to displace groundwater, the sand boils
would soon be removed by erosion. Therefore, the likelihood that
the sand boils are a result of the expulsion of water due to
progressive collapse of cavities is unlikely. Instead, we see a
slow, steady growth of the sand boils, and although the effusion
rate may be variable, there is always significant flow.
2. There is an incredibly vast, international literature on the effects
of both natural and anthropogenic solution of salt in the
subsurface. I know of at least 100 references and could probably
find 1000. Some of the most useful that come to mind are:
Calvert, A.F., 1915. Salt in Cheshire, Spon, London, 1206 p.
(includes many photos).
Evans, W.B. et. al., 1988. Geology of the country around
Macclesfield — Geological Survey of Great Britain, Memoir (plus a
more recent memoir on Cheshire published 1987±).
Reuter, F. and Tolmanacer, V.V. 1990. Bauen and Bergbau in
Senkungs und Erdfallgebieten. Schriftenreihe fur Geogische
Wissenschaften, Akademie Verlag, Berlin, No. 28., 176 p.
Weber, H. 1952. Pliozan und Auslaugung im Gebiet der Oberen
Werra, Geologica Akademie Verlag, Berlin, v.8, 136 p.
328
-------
Session V:
Case Histories
-------
330
-------
Hydrogeology and Ground-Water Monitoring of an Ash-Disposal
Site at a Coal-Fired Power Plant in a Karst System
Shelley A. Minns1, Arsin M. Sahba1, Lyle V.A. Sendlein1,
James C. Currens2 and James S. Dinger2
Department of Geological Sciences and
2Kentucky Geological Survey
University of Kentucky, Lexington, Kentucky
Abstract
A coal-fired power plant located along a lake in south-
central Kentucky is situated on karstified Mississippian-age
carbonates. Ground-water flow and chemistry are evaluated to
assess impacts of current and past coal-ash-disposal practices.
Ash disposal in an unlined, active slurry pond is allowing ash
and pond water to enter the ground-water system through
underlying sinkholes and fractures. Water infiltrating through
20-year-old ash in a closed-out slurry pond enters the ground-
water system similarly- Ground water discharges at springs
near lake pool elevation. Sinkholes, swallets, and springs
were field mapped, and a dye-tracing plan was implemented. Dye
traces indicate separate flow systems for each disposal site.
Dry fly ash was leached to determine constituents present in
fresh ash. Water-quality analyses were obtained from
upgradient and downgradient springs and surface-water sources.
Wells installed in the closed-out slurry pond were sampled.
Monitoring data indicate that major ion composition in the
active slurry pond is more similar to the lake water used for
the ash slurry than natural ground water. Active-pond water is
elevated in sulfate, barium, boron, copper, iron, manganese,
and zinc compared to background waters. Spring discharge
associated with the active slurry pond shows major ion
composition intermediate between pond water and natural ground
water. Boron, iron, manganese, and zinc are elevated in these
springs. Spring discharge associated with the closed-out
slurry pond is similar to background springs except for
increased concentrations of boron and iron. Water samples
collected from wells drilled into the ash are similar in major
ion composition to background springs but show elevated
concentrations of barium, boron, iron, manganese, and vanadium.
331
-------
Introduction
The purpose of this research is to study the ground-water
flow system and chemistry in a karst aquifer underlying a coal-
fired power plant. The results of this research will be used
by the power company to develop a ground-water-monitoring
program which meets the requirements of the state regulatory
authority.
Coal ash is an aluminosilicate glass which forms from the
burning of coal at temperatures of approximately 1500 degrees
celsius (Elseewi, 1980). Ash is composed primarily of silicon,
aluminum, iron, and calcium oxides. Lesser amounts of sodium,
magnesium, potassium, and sulfur are typically present
(Cherkauer, 1980; Elseewi, 1980). The major constituents are
derived from thermal breakdown of minerals, primarily illite
and kaolinite. As a result, the composition of coal ash is
highly dependent on the composition of the pre-combusted coal.
Coal ash consists of two products. Fly ash is fine
grained and often forms concentrically layered spheres. This
fraction accumulates in electrostatic precipitators and
comprises approximately 20 percent of the total ash fraction at
the study site (Rai and Zachara, 1990, personal commumication).
Bottom ash is coarser grained material which falls to the
bottom of the combustion chamber.
Elements associated with ash may be incorporated into the
aluminosilicate matrix of individual particles or may
accumulate on the outer surface of the spheres. Many elements
volatilize during combustion and subsequently condense on the
outer surface of the cooling particles. Antimony, boron,
fluoride, selenium, arsenic, cadmium, lead, zinc, barium,
chromium, copper, chlorine, sulfur, and vanadium are often
preferentially enriched on ash particles in this manner
(Ainsworth and Rai, 1987; Rai, 1987). Aluminum, cobalt, iron,
manganese, nickel, strontium, and vanadium are also associated
with coal ash, though not generally enriched on the surface of
the particles (Wu and Chen, 1987; Rai, 1987).
Setting
The power plant was constructed in 1964 and houses two
generating units which produce a total of 320 megawatts of
electricity. Maximum coal usage is 5,000 tons per 24-hour
period.
The site has two ash-disposal facilities (Figure 1). From
1964 to 1978, coal ash was sluiced to a 40-acre slurry pond.
This disposal facility was retired in 1978 when its storage
capacity was reached. The maximum thickness of ash in this
facility is 35 feet.
Since 1978, ash has been disposed of in an unlined pond
that has an area of approximately 50 acres. Fly ash and bottom
ash are transported to this facility as an ash-water slurry via
a 3,500-foot-long pipeline.
332
-------
ACTIVE
ASH PDND
4
CLDSED-DUT
SLURRY PDND
Figure 1. Map showing location of power station and ash-disposal
facilities.
The plant is located in the Mississippian Plateau
carbonate region of Kentucky, adjacent to a major impounded
river. The topography is a rolling, upland, karst plain which
exhibits subsurface drainage. In the immediate vicinity of the
plant, the land surface is dissected by small v-shaped valleys.
Surface flow in these channels is ephemeral and often
disappears into swallets in the stream bed. The land surface
drops steeply, often vertically, from the upland karst area on
which the plant is situated to the surface of the lake.
Rocks underlying the site are primarily limestone and
dolomite (Figure 2). Structures associated with the plant are
constructed on the Ste. Genevieve Limestone and St. Louis
Limestone. The Salem and Warsaw Formations (undifferentiated)
crop out a few feet above permanent lake pool elevation.
The Salem and Warsaw Formations, which are considered
together because of their lithologic similarity, are the
stratigraphically lowest formations that crop out near the
plant. Total thickness ranges from 30 feet to 75 feet (Lewis,
1974; Taylor, 1975). The two formations represent an
intertonguing sequence of dolomite, limestone, shale,
siltstone, and sandstone, which shows significant lateral
variation. The exposed section of Salem and Warsaw near the
333
-------
Mbha
N
(PbO Breathltt Fn.
(Mp) Pennlngton Fn,
(Mbha) Bangor Ls. &
Hartselle Fn,
(Mnk) Kldder Ls, Mbr,
-------
ACTIVE
ASH PDND
LEGEND
DYE INPUT
o •
ACTIVE-POND •
SPRINGS •
WELLS
SPRINGS
CLUSED-DUT
SLURRY PDND
SAMPLED
SPRING
O SINKHOLE
POWER
STATION
Figure 3. Hydrogeologic features map.
A survey was conducted to locate springs potentially
related to subsurface drainage from the site. The survey began
in late winter when lake levels were highest. As summer
approached, lake levels declined and additional springs became
visible. Mapped springs are shown on Figure 3.
A submerged spring situated about 22 feet below permanent
pool elevation was noted by plant employees during the time the
closed-out slurry pond was an active facility- It was thought
that this spring may represent a major discharge point from the
conduit system associated with the closed-out slurry pond and
that stratigraphically higher springs may be wet-weather
overflows. During early spring of 1990, the spring was
documented when a conductivity plume was delineated relative to
ambient lake conductivity at the same depth. In July 1990, an
attempt was made to locate this spring using SCUBA but the
attempt was not successful.
By October 1990, the lake level had fallen and exposed a
small seep at the top of the agillaceous dolomite zone in the
Salem and Warsaw Formations. It was concluded that this spring
maintains a very low flow in the summer and fall and is not a
major discharge point.
A dye-tracing program was initiated in the spring of 1990
to determine hydraulic connections between swallets and springs
335
-------
located around the periphery of the site. The program was
developed with the assistance of Dr. John Thrailkill, Professor
of Geology at the University of Kentucky, and James Currens, a
hydrogeologist at the Kentucky Geological Survey.
The fluorescent tracers Fluorescein LT (Acid yellow 73,
CI 45350) and optical brightener (Tinopal CBS-X) were chosen
for this study. Fluorescein imparts a fluorescent green color
to water, and optical brightener fluoresces blue-white under
ultraviolet light. These dyes have been used in the karst
regions of Kentucky by both Thrailkill and Currens.
Fluorescein and optical brightener do not interfere with one
another and can be used concurrently.
Passive detectors were used to adsorb the tracers from the
water. Laboratory-grade activated charcoal in nylon screen
bags was used for fluorescein adsorption. Cotton broadcloth
test fabric stretched over a wire frame was used for adsorbing
optical brightener.
Prior to the addition of dye into the conduit system, a
background test of fluorescence was conducted using both types
of detectors. No evidence of potential interference with
either dye was detected.
Several measures were taken to prevent accidental
contamination of detectors with dye. Dye and detector
materials were not stored or handled in the same room, nor were
dye and detectors handled on the same day. To minimize
handling of dye during the course of the project, all bulk dye
products were prepackaged into 50, 100, and 200 gram bottles
before any dye tracing was initiated. As an extra precaution,
bottles of dye were sealed inside plastic bags. Once the
amount of dye needed for a trace was calculated, the
appropriate amount was chosen from the prepackaged selection.
Detector placement and dye injection were always performed on
different days.
The amount of fluorescein introduced into each point was
calculated from the following formula (Aley and Fletcher,
1976):
Wd = 1.478 D x (Q/V) where:
Wd = Weight of dye in kg
D = Distance travelled in km
Q = Approximate discharge in mVsec
V = Approximate velocity in m/hr
The calculated amount was rounded to the next even 50 grams.
For optical brightener, Quinlan (1986) suggests that
doubling the amount calculated from the above formula works
well for the average Kentucky spring. This rule of thumb was
used for determining the amount of optical brightener to
inject.
Eight sites were selected for dye injection, four near
each disposal facility. One detector site was located at the
submerged spring. Detectors were attached to a rope anchored
to a 30-pound weight on the lake bottom and could be raised and
336
-------
lowered using small pulleys. A float was tied to the anchor in
order to mark the site. Use of this method allowed easy access
from a small boat.
Seven of the eight dye traces were successful. One
fluorescein dye trace was not detected at any of the detectors.
Since laboratory tests showed that fluorescein is strongly
adsorbed onto coal ash, it is possible that this trace was lost
as a result of adsorption onto ash which may be present in the
conduit system. It is also possible that not enough water was
used to flush the dye through the system.
Three ground-water-monitoring wells were installed in the
closed-out slurry pond. Well locations are shown in Figure 3.
These wells were used to monitor water levels and to collect
water-quality data in the dewatered, closed-out facility- Each
well was completed in ash using 10-foot-long screens, the
bottoms of which were placed at the top of bedrock.
Water samples were obtained from each monitoring well
during March and May to analyze for non-metals. One set of
field-filtered samples was collected for dissolved-metals
analyses. Each well was purged prior to sampling. When
possible, three well volumes were removed. Low-yield wells
were purged until dry, then allowed to recover enough to permit
sampling. A bladder pump was used for sampling when there was
sufficient water volume. Wells with little water were sampled
using a stainless steel bailer with a bottom-emptying device.
Springs were sampled by filling a sample bottle with water
obtained from near the spring mouth.
Water samples for dissolved-metals analyses were preserved
immediately upon collection using nitric acid. All samples
were placed on ice immediately after collection and delivered
to the laboratory the next working day. Water-quality analyses
were performed by the Kentucky Geological Survey laboratory.
Analyses were done according to EPA procedures described in
"Methods for Chemical Analyses of Water and Wastes." Metals
analyses were performed using the Inductively Coupled Argon
Plasma method (ICAP).
Fresh, dry ash samples were collected directly from the
hoppers within the plant. Numerous leaching tests of varying
duration were conducted by Battelle-Pacific Northwest
Laboratory on fly ash and bottom ash through a range of
solid/solution ratios. Data were obtained on elements
associated with the ash fractions and their leachability.
Results
Dye-trace results show that the active disposal pond and
the closed-out pond are located in different ground-water-flow
systems. Figure 4 shows the dye input points and the resulting
positive traces.
The lower boundary of karst development in the closed-out
slurry pond flow system is located approximately 30 feet below
the top of the Salem and Warsaw Formations. The lower boundary
is marked by an argillaceous dolomite layer (Figure 5). Spring
337
-------
CTIVE
'• ASH PDND
CLDSED-DUT
SLURRY PDND
Figure 4. Locations of positive dye traces.
ft
ELEV.
(Ft.)
1000 .
950 .
900 .
850 -
800
750 _
700
650
\ BANGDR LS. &.
-V^RARTSELLE FM,
-------
flow in the closed-out pond system varies seasonally. The
greatest flows generally occur in the winter and spring and the
lowest flows are in the summer and fall.
The lower boundary of karst development in the active-ash
pond flow system appears to be argillaceous zones at the top of
the Salem and Warsaw Formations. All springs associated with
the active pond occur in the lower section of the St. Louis
Limestone (Figure 6). Spring discharge from this system
remains constant throughout the year.
I
ELEV
-------
Table 1. Maximum and minimum concentrations for dissolved
metals (mg/L).
Constituent
aluminum
antimony
arsenic
barium
beryllium
boron
cadmium
calcium
chromium
cobalt
copper
iron
gold
lead
lithium
magnesium
manganese
nickel
phosphorus
potassium
selenium
silicon
silver
sodium
strontium
sulfur
thallium
tin
vanadium
zinc
Background
springs
Max
.139
<.017
<.028
.032
•C.OOl
.020
<.003
123
.010
<.004
.008
.116
•e.004
<.03B
•e.006
8.62
.010
<.012
•c.068
<1.32
•e.052
3.01
<.003
6.67
.028
14.7
<-026
<.026
.006
.009
Min
<.027
•e.017
<.028
.028
<.001
<.006
<.003
85.1
<.006
•c.004
.005
<-004
-C.004
•c.038
<.006
5.55
•c.OOl
•c.012
<.068
.400
<.052
.074
<-003
1.77
.252
11.0
<.026
<-026
<.003
.006
Wells in
closed-out
. ,2
ash pond
Max
.057
<.017
.071
.125
.001
1.06
<.003
70.2
.006
.004
.005
.050
.007
•C.038
.118
8.35
.654
<.012
.463
5.79
<.052
10.5
<.003
13.2
2.29
27.6
<.026
<.026
.031
.010
Min
.030
<.017
<.028
.082
<.001
.657
<.003
53.3
<.006
<.004
<.004
.028
.004
<.038
.110
6.95
.160
•C.012
<.060
5.67
<.052
4.62
•c.003
8.55
1.71
23.5
<.026
<.026
<-003
.008
Springs down-
gradient of
closed-out
ash pond
Max
.118
<.017
•c.028
.044
•C.OOl
.175
.003
73.9
.009
•e.004
.011
.157
•c.004
<.038
.028
7.88
.040
•e.012
<.068
2.80
•e.052
1.44
<.003
11.84
.399
20.9
<.026
.066
<.003
.088
Min
.101
<.017
<.028
•c.OOl
•c.OOl
.175
<-003
71.7
.008
•c.004
.008
.142
<.004
•C.038
.024
7.74
.015
<.012
•c.006
2.71
<.052
1.27
<.003
10.1
.398
20.1
•C.026
<.026
•c.003
.011
Springs down-
gradient of
active pond
Max
.117
.020
.029
•e.OOl
<.001
.081
<.003
82.9
.010
<.004
.010
.198
•c.004
<.038
.013
6.12
.005
.013
•e.068
2.14
<.052
1.38
•C.003
5.63
.404
11.5
•c.026
<.026
<.003
.014
Min
.099
•e.017
<.028
<.001
<.001
.061
•c.003
82.2
.010
<.004
•C.004
.154
<.004
<.038
.011
6.02
.001
•c.012
<.068
1.70
•c.052
.992
<.003
2.98
.391
9.30
•c.026
<.026
<.003
.010
Surface water
samples of
active pond
Max
.047
•e.017
•c.028
<.001
<.001
.231
.006
28.7
<.006
.005
.095
.730
<.004
•c.038
.138
8.42
.100
.035
•c.068
4.64
<.052
2.55
<.003
14.08
.246
25.6
.029
.116
.005
.038
Min
•e.027
•c.017
•C.028
<.001
<-001
.125
<.003
26.1
•e.006
<.004
.009
.006
•C.004
<.038
.119
7.20
.043
.012
•c.068
4.20
<.052
.581
•e.003
6.94
.233
25.3
<.026
•c.026
<.003
.007
Lake
wa-
ter
1
value
.086
•c.017
•C.028
<.001
<.001
.011
•e.003
16.9
•c.006
.004
<.004
.008
•c.004
<.038
<.006
7.40
.006
•e.012
<.060
<1.32
<.052
.073
•e.003
12.2
.082
15.1
<.026
.062
•e.003
.008
1 Three analyses except for Sb, Be, Cd, Co, Au, Pb, Se, Ag, S, and Sn which have two analyses.
2. Two analyses.
3. Three analyses except for Sb, Be, Cd, Co, Au, Pb, Se, Ag, S, and Sn which have two analyses.
340
-------
Table 2. Maximum and minimum concentrations for non-metals
(mg/L).
Constituent
pH
alkalinity
as CaCO?
bicarbonate
total
dissolved
solids
chloride
fluoride
aulfate
Background
springe
Max
8.08
249
304
314
2.85
.15
32.6
Min
7.47
225
274
258
2.31
.10
21.6
Wells in
closed-out
ash pond
Max
7.18
208
254
294
2.10
.37
22.8
Min
6.55
122
149
146
1.43
.16
15.4
Springs
down-
gradient
of closed-
out ash
pond
Max
7.61
216
263
310
4.03
.20
95.6
Min
7.34
122
149
224
3.03
.17
19.8
Springs down-
gradient of
j*
active pond
Max
7.42
221
270
290
2.84
.17
33.7
Min
6.93
210
256
274
2.18
.14
20.9
Surface water
samples of
active pond
Max
7.79
10
12
202
3.99
.20
116
Min
5.92
8
10
128
2.76
.10
80.7
Lake
water
1 value
6.94
39
48
128
2.21
.06
45.4
1 Five analyses except for alkalinity, bicarbonate, and TDS which have four analyses.
2 Five analyses.
3 Nine analyses.
4 Five analyses. >
5 Six analyses except for alkalinity, bicarbonate, and TDS which have three analyses.
Table 3 contains a partial summary of the ash-leaching
results using deionized water. After 15 minutes of contact
time, high concentrations of some elements were present in the
leachate. Leaching continued for most elements after one week
of contact time. In general, the most enriched leachate
resulted from the fine-grained fly-ash fraction.
Table 3. Concentrations (mg/L) of dry ash leachate leached
with deionized water at different solid solution ratios for
different time increments.
Constituent
(mg/L)
aluminum
arsenic
barium
boron
cadmium
chromium
chloride
cobalt
1:1
Fly ash;
15 min.
700
40
.84
62
.18
14.2
1.9
1.6
1:1
Bottom
ash; 15
min.
239
7.9
.240
14.1
<.010
2.86
8.07
.44
1:10
Fly ash;
15 min.
65
8.0
.262
5.5
•e.010
1.4
.89
.16
1:10
Bottom
ash;15
min.
20.3
.86
.454
1.38
<.010
.22
1.53
.04
1:1
Fly
ash; 1
week
619
19
.08
66
.24
1.2
2.8
1.7
1:1
Bottom
ash; 1
week
287
<.050
.15
21
.12
.3
22.37
.6
1:10
Fly
ash; 1
week
81
2.7
.072
7.82
.023
.139
.10
.206
1:10
Bottom
ash; 1
week
69.2
<.050
.033
2.34
.014
.020
3.50
.088
341
-------
Table 3 continued.
Constituent
(mg/L)
copper
fluoride
iron
lithium
manganese
nickel
selenium
strontium
vanadium
zinc
1:1
Fly ash;
15 min.
29.1
12.61
19.6
30.5
4.06
4.65
.78
20.4
33.6
7.86
1:1
Bottom
ash; 15
min.
6.44
4.61
25.5
7.1
4.11
1.27
<.010
7.69
1.10
2.15
1:10
Fly ash;
15 min.
2.81
2.33
2.53
2.6
.379
.434
.39
2.77
3.34
.761
1:10
Bottom
ash; 15
min.
.718
.54
2.32
1.04
.308
.115
.018
.913
.15
.186
1:1
Fly
ash; 1
week
23.5
8.80
5.69
24.2
4.77
4.9
16.4
.72
7.95
1:1
Bottom
ash; 1
week
4.54
6.39
6.48
10.2
2.10
7.19
2.90
1:10
Fly
ash; 1
week
3.02
3.89
1.38
3.0
.540
.597
4.40
.219
1.00
1:10
Bottom
ash; 1
week
.602
1.18
3.01
1.11
2.18
.253
1.78
.324
Summary
Dry-ash samples collected from the site indicate that coal
ash generated at the plant is enriched in most elements tested.
Dry-ash leach tests show that significant leaching of the ash
particles occurs after a 15-minute contact time in deionized
water and that continued leaching occurs. The system of ash
disposal at the plant which incorporates large volumes of water
with a 9 to 15 minute transit time, effectively leaches many
trace elements from the surface of the ash particles during
sluicing.
Dye traces in the vicinity of the closed-out slurry pond
define a flow system which is approximately 2,000 feet wide and
4,000 feet long. The closed-out pond lies within the boundary
of the flow system. Anecdotal evidence suggests that spring
flow in this system was much greater when it was an active
slurry pond. Numerous sinkholes reportedly opened while water
was impounded, allowing water to enter the ground-water system.
Ash was reportedly observed at some spring outlets when the
pond was active.
Major-ion chemistry of the springs connected to the
closed-out pond is very similar to background spring chemistry
(Table 4). Comparison of mean values for background and down-
gradient springs for trace elements associated with coal ash
shows that leaching of the ash has already occurred (Table 5).
On the average, constituent values for springs below the
closed-out pond are similar to background values. Only boron
and iron are more than two times greater than background for
these springs.
Wells completed directly in the ash show that boron and
manganese exceed background by more than one order of
magnitude. Barium and fluoride are two to three times
342
-------
Table 4. Mean concentrations of major ions (mg/L)
Constituent
bicarbonate
Bulfate
sodium
potassium
magnesium
calcium
Back-
ground
springs
286
27.8
4.00
<1.32
6.97
99.4
Lake
water
48
45.4
12.2
<1.32
7.40
16.9
Hells in
closed-
out ash
pond
196
18.4
10.8
5.73
7.65
61.8
Springs
down-
gradient of
closed-out
ash pond
264
28.5
4.31
1.92
6.07
82.6
Springs
down-
gradient of
active pond
173
80.14
10.93
2.76
7.81
72.8
Surface
water
samples
of active
pond
11
95.9
9.49
4.38
8.13
27.4
Table 5. Mean concentrations of trace constituents (mg/L)
Constituent
aluminum
antimony
arsenic
barium
boron
cadmium
chromium
chloride
cobalt
copper
fluoride
iron
lead
lithium
manganese
nickel
selenium
strontium
sulfur
vanadium
zinc
Back-
ground
springs
.079
<.017
<.028
.030
.015
<.003
.008
2.60
<.004
.006
.13
.057
<.038
<.006
.005
<.012
<.052
.276
12.9
<.003
.008
Lake
water
1 value
.086
<.017
<.028
.021
.011
•e.003
<.006
2.21
.004
<.004
.06
.008
<.038
<.006
.006
•e.012
<.052
.082
15.1
•e.003
.008
Wells in
closed-
out ash
pond
.044
<.017
.049
.104
.859
<.003
.006
1.66
.004
.005
.28
.039
•e.038
.114
.407
<.012
<-052
2.00
25.6
.017
.009
Springs
down-
gradient of
closed-out
ash pond
.108
.019
.029
.036
.071
<.003
.010
2.44
<.004
.007
.16
.176
<.038
.012
.003
.013
<-052
.398
10.4
<.003
.012
Springs
down-
gradient of
active pond
.110
<.017
<.028
.044
.175
.003
.009
3.53
•c.004
.010
.19
.150
•c.038
.026
.028
<.012
•e.052
.399
20.5
•C.003
.050
Surface
water
samples
of active
pond
.035
<.017
<.028
.203
.168
.005
<.006
3.43
.005
.031
.15
.204
<.038
.128
.071
.024
<.052
.237
25.5
.003
.024
343
-------
background levels. Vanadium and strontium values are
approximately six times greater than background concentrations.
Other constituents are similar to background measurements.
Dye traces define a wedge-shaped flow system which
encompasses the active pond and is at least 2,900 feet wide and
6,200 feet long. Total ground-water spring flow in this system
is approximately two orders of magnitude greater than spring
flow observed in the closed-out pond system. The source of
much of this flow is apparently the active disposal pond.
Water from the pond enters the ground-water system through
fractures and sinkholes.
Comparison of water quality from springs and the disposal
pond with background quality shows that leaching has occurred
in this system. Most trace constituents are similar to
background spring and lake water quality for the elements
listed (Table 5).
Barium, boron, iron, and manganese are more than 10 times
greater than raw lake water in the surface-water samples taken
from the active pond. Copper and zinc are eight and three
times greater, respectively. The remaining constituents are
similar to raw lake water concentrations. Water quality in the
active pond reflects the major-ion composition of the raw lake
water used for sluicing (Table 4). The pH, alkalinity,
bicarbonate, total dissolved solids, and calcium are depressed
compared to background springs.
Boron values in the springs below the active pond are
approximately 10 times greater than background spring
measurements. Manganese and zinc concentrations are
approximately six times the background values in springs
associated with the active pond. Iron is three to four times
greater than background spring values. Major-ion composition
for these springs is intermediate between background and active
pond water.
Conclusions
1. The ground-water flow system in which the disposal
sites occur have been defined relatively well. Even though
deep wells have not been installed at the site, the base of the
flow system can be placed near the contact between the St.
Louis Limestone and Salem and Warsaw Formations.
2. Total dissolved solids are relatively low with the
maximum value occurring in the background springs.
3. Mean values for all constituents analyzed are within
U.S. EPA standards, even though some elements are 10 times
background spring concentrations.
References Cited
Ainsworth, C. C., and Rai, p., 1987, Chemical characterization
of fossil fuel combustion wastes: Palo Alto, California,
Electric Power Research Institute, EA-5321.
344
-------
Aley, T., and Fletcher, M. W., 1976, The water tracer's
cookbook: Missouri Speleology, v-16, no. 3.
Cherkauer, D. A., 1980, The effect of fly ash disposal on a
shallow ground water system: Ground Water, v. 18, no. 6,
p. 544-550.
Elseewi, A. A., Page, A. L., and Grimm, S. R., 1980, Chemical
characterization of fly ash aqueous systems: Journal of
Environmental Quality, v. 9, no. 3, p. 424-428.
Lewis, R. Q., Sr., 1974, Geologic map of the Somerset
Quadrangle, Pulaski County, Kentucky: U.S. Geological
Survey Geologic Quadrangle Map GQ-1196, scale 1:24,000.
Quinlan, J. F., 1986, Practical karst hydrology, with emphasis
on groundwater monitoring (course manual): National Water
Well Association, Dublin, Ohio, 898 p.
Rai, D., 1987, Inorganic and organic constituents in fossil
fuel combustion residues, volume 1: a critical review:
Palo Alto, California, Electric Power Research Institute,
EA-5176.
Taylor, A. A., Lewis, L. Q., Sr., and Smith, J. H., 1975,
Geologic map of the Burnside Quadrangle, south-central
Kentucky: U.S. Geological Survey Geologic Quadrangle Map
GQ-1253, scale 1:24,000.
Wu, E. J., and Chen, K. Y., 1987, Chemical form and
leachability of inorganic trace elements in coal ash: Palo
Alto, California, Electric Power Research Institute,
EA-5115.
Biographical Sketch
Shelley A. Minns obtained the B.S. degree in geology from
West Virginia University in 1977. She worked for 10 years
throughout the southern Appalachian coal field region,
primarily for coal-related consulting firms. She obtained the
M.S. degree in environmental studies from Ohio University in
1989 and is currently a Ph.D. candidate in geology,
specializing in hydrogeology, at the University of Kentucky,
Lexington, Kentucky- Her research interests are coal-ash
disposal and coal hydrology. She is currently employed with
the Kentucky Geological Survey as a research assistant.
345
-------
Hydrogeology and Ground-Water Monitoring of an Ash-Disposal
Site at a Coal-Fired Power Plant in a Karst System
Shelley A. Minns1, Arsin M. Sahba1, Lyle V. A. Sendlein1,
James C. Currens2 and James S. Dinger2
Department of Geological Sciences and
2Kentucky Geological Survey
University of Kentucky, Lexington, Kentucky
Question 1: Did you define two flow systems or two drainage
basins?
The two disposal ponds are located in different surface
water drainage basins. Dye traces were conducted on all
available sinkholes considered upgradient of each pond. The dye
traces defined the flow systems and as it turned out, they also
reflected the surface water basins.
Question 2: What was the rationale for selecting 5 times
background concentration as the criterion for determining whether
the levels were elevated above background?
Concentrations of metals for water associated with the two
disposal sites range from nearly identical to background spring
concentrations to about two orders of magnitude greater than
background concentrations. The ranges of five times greater than
background and ten times greater than background were selected to
use during the presentation of the paper at the conference as a
way to get a feel for how metals were distributed when compared
to background spring concentrations. These numbers to do not
imply any significance relative to background concentrations and
are not used in the paper. The paper contains detailed data of
constituents in the water associated with the two disposal sites.
Question 3: What leaching tests were found to be most
representative of actual processes presumably operative within
the ponds? Was there a significant difference in the trace
element nature of the coals burned at the site between the old
and the active ponds? Were there other significant waste streams
contributing to the ponds?
Leaching tests were not conducted as part of our research
effort. Battelle-Pacific Northwest Laboratory conducted the
leaching tests as part of a separate investigation. We have not
received the complete results of these tests. We do know that
the solid/solution ratio, which is generally representative for
the sluicing system used at the plant, is ten parts ash to one
part lake water.
346
-------
Trace element analyses of coal burned at the plant when the
closed-out pond was an active facility are not reported.
No other waste streams impact the disposal ponds. The
contributing watershed is mainly undisturbed forest. The coal
stockpile area is located downgradient of both ponds and does not
affect the ash disposal site.
347
-------
348
-------
DEVELOPMENT OF A MONITORING PROGRAM AT A SUPERFUND SITE
IN A KARST TERRANE NEAR BLOOMINGTON, INDIANA
Michael R. McCann, Westinghouse Electric Corp.
Noel C. Krothe, Indiana University
Bloomington, Indiana
Abstract
Lemon Lane, a former municipal dump from 1933 to 1964, is
located entirely within two sinkholes on a ridgetop over
Mississippian limestones. Materials containing PCBs were discarded
there from 1957 to 1964.
Twenty-two monitoring wells have been installed in stages into
two separate ground-water zones. Results from three tracer tests
conducted from 1987 through 1990, and interpretations of limited
potentiometric data, indicate that the site is drained by a
convergent conduit system that, during low to moderate flow, drains
through a main conduit which exits at a series of springs. A
quantitative tracer test showed that 97% of the tracer recovered
flowed to one spring. During high flows, when the conduit is
flooded, some water is temporarily stored in the epikarst and the
fractured bedrock adjacent to the conduit. Small amounts of water
in storage may spillover into adjacent spring drainage basins,
suggesting the site is near the ground-water drainage divides of
several subterranean basins.
Fluorescein injected as a tracer in an on-site well, in 1989,
was still detectable at the main spring outlets 19 months later.
Monitoring at selected on-site wells has shown tracers at levels
near the detection limit in water (0.005 ppb) of a scanning
spectrofluorophotometer.
The springs, indicated by tracing to be directly connected to
the site, will provide the most time-sensitive and relevant
monitoring locations for potential off-site migration of
contaminants. Existing monitoring wells should effectively monitor
for contaminants in temporary storage. As a result, the
installation of additional monitoring wells is not recommended
because the probability of any of them intersecting conduits
draining the site is minimal.
'349
-------
Description of Lemon Lane Landfill
Lemon Lane Landfill is located on the west-central edge of
Bloomington in Monroe County, Indiana. Bloomington is located in
the Mitchell Plain physiographic unit. The Mitchell Plain is a low
plateau underlain with carbonates of Mississippian age on which a
characteristic karst terrane has developed. The Lemon Lane
landfill, comprising about 9 acres, was located entirely within two
sinkholes on a ridgetop at or near several surface water or
topographic divides. However, all of the natural surface drainage
at the site is into several sinkholes. The 1908 topographic map
for the area (Figure 1) , shows the two sinkholes that were used for
the disposal of solid waste, which are actually part of a large
compound sink extending to the northwest and southwest of Lemon
Lane.
Lemon Lane was operated as a municipal dump from 1933 to 1964.
A nearby plant manufactured capacitors containing PCBs as an
insulating fluid beginning in 1957. Materials containing PCBs were
discarded at Lemon Lane from 1957 to 1964 by local contract waste
haulers.
Geology of Lemon Lane Landfill
The site is underlain by Mississippian age limestones,
dolostones, and shales of the St. Louis Formation. The Salem
Limestone, which has strata similar to the St. Louis in its upper
part underlies the St. Louis Formation. The lower part of the
Salem is a thick, cross-bedded calcarenite which is famous as a
building stone known as the "Indiana Limestone" (Shaver et al.,
1986) . The contact between the St. Louis and the Salem is not well
defined lithologically in south-central Indiana.
The strata generally dip westward from their outcrop on the
flank of the Cincinnati Arch into the Illinois Basin. The dip
averages about 30 feet to the mile in the Bloomington area (Gates,
1962) .
Monitoring Well Installation
Twenty-two monitoring wells were installed in two phases at
Lemon Lane. Seventeen wells were installed, from 1982 to 1983, in
the first phase and five additional wells were installed in 1987
(Figure 2) . Seventeen of the wells were installed to monitor
ground water in shallow bedrock and five of the wells monitor a
deeper zone. Figure 3 shows the stratigraphy encountered. The
shallow zone occurs between elevations 798 ft. and 820 ft. amsl and
the deeper zone occurs between elevations 760 ft. and 770 ft. amsl.
Monitoring wells MW-4D, MW-5, MW-8D, MW-12, and MW-14 are installed
in the deeper zone and the rest are installed in the shallow zone.
350
-------
uo
Ul
R2W R1W
0 2000 FEET
i i i
Figure 1.
Location of Lemon Lane Landfill relative to
sinkholes prior to use as a landfill.
(Bloomington, Indiana, Quadrangle 15 minute
series, 1908)
-------
N
MW-12
O
MW-1D
°0
MW-1S
MW-8S
MW-8D
MW-B4
O
(not to acale)
(well la off mop)
MW-B3
O
/\
\fc
\°»
\\
X^
Vfe
r
I
i
i
1
MW-9
Oi
^rf 1
^ MW-2
X,g
~^JMW-13
MW-3
or
MW-14O |
MW-4S
O I
MW-41 OMW-10
O O
MW-40 1
SCALE 1 inch - 200 FEET
MW-B2
MW-11 O
O O
MW-5
E
_t
Figure 2 - Monitoring Well and Cross-Section Location
352
-------
ELEVATION
ABOVE M8L
oo
Ul
CO
Figure 3 - Geologic Cross-Section
BLACK SLACEOUS CORAL
LS
DOLOSTONE
LS
IpLOSTONE
TONE
SILT STONE WITH PYRITE
POROUS ZONE
BLACK SHALE IN UW-40
L.J
-------
The two zones appear to be separated by a shale unit that is at
elevation 782 ft. to the east and 778 ft. to the west (see Cross-
section, Figure 3).
Packer tests and slug tests conducted during installation
indicated the wells in the deeper zone encountered fractures much
less transmissive than those encountered by the wells in the
shallow zone.
Water Level and Water Quality Measurements
Potentiometric surface maps for water levels taken on 2/2/88
and 6/14/88 are shown for the upper and lower zones in Figures
4,5,6,and 7. The interpretation of ground water flow directions in
the lower zone appears to be relatively simple. During the higher
flow period (2/2/88) recharge is from the west and flow directions
are to the east. During the lower flow period (6/14/88) recharge
is from the west and east and flows are generally to the north and
northwest. Interpretations of flow directions in the upper zone,
however, are problematic. During the higher-flow period, ground
water appears to be mounded in the vicinity of wells MW-7, MW-B3,
MW-4I, MW-8S, and MW-11. Water levels are depressed in the
vicinity of MW-9, giving the appearance of a sink. During the low-
flow period, mounding is apparent near MW-B3, MW-3, and MW-11. In
neither flow regime is there an unambiguous flow direction
indicating the probable direction of ground-water movement.
Table 1 shows the analytical results of PCB sampling at the 17
monitoring wells installed in 1982 and some springs in the vicinity
of Lemon Lane. Wells MW-10, MW-11, MW-12, MW-13, and MW-14 were
installed in 1987 and have not yet been sampled. The results are
difficult to interpret. For example, although PCBs were found in
wells MW-8S and MW-8D, the potentiometric maps for both flow
regimes seem to indicate these wells are upgradient. Illinois
Central and Quarry springs, located about 2000 feet southeast of
Lemon Lane, have concentrations of PCBs one to two orders of
magnitude higher than the on-site wells.
The original monitoring plan for Lemon Lane required on-site
monitoring wells to be selected on the basis of ground-water flow
directions and the results of initial PCB sampling. Locations for
off-site wells were to be selected on the basis of preferred flow
directions as interpreted from on-site wells. The original
monitoring plan was based on the assumption that an aquifer was
present that behaved as a porous-media-equivalent and that ground-
water flow direction could be predicted by idealized flow nets
constructed from water level measurements. Contaminant movement
could thus be predicted from aquifer properties deduced from
pumping tests which were analyzed by standard equations of ground-
water flow. In an attempt to resolve the ambiguities, tracer tests
were conducted.
354
-------
I Site Boundary
MW-5
Figure 4 - Lower Zone Potentiometric Map 6/14/88
en ° Site Boundary
MW-5
Figure 5 - Lower Zone Potentiometric Map 2/Z/88
355
-------
MW-1S
'Big
MW-4S
MW-11
Figure 6 - Upper Zone Potentiometric Map 6/14/88
MW-1S
MW-8S
N
MW-11
Figure 7 - Upper Zone Potentiometric Map 2/2/88
356
-------
Location
Springs/Surface Water
Sargent Pond #1
Sargent Pond #2
Stout Creek
Robertson Spring
Snoddy Spring
Stoney Spring-East
Stoney Spring-West
P.H. Road Spring
Detmer Spring
Packing Plant Spring
III. Central Spring
Quarry Spring
Slaughter House Spring
Hinkle Wet Weather Rise
Table 1
PCB ANALYTICAL DATA FOR GROUND WATER AND SURFACE WATER
LEMON LANE LANDFILL
BLOOMINGTON, INDIANA
6/25-29/81 7/1-2/81 8/17/81 10/7-8/81 7/28-29/82 10/19/82
ND
ND
0.9*
ND
ND
ND
5.7***
ND
ND
ND
ND
ND
ND
ND
6.8*
ND
ND
12/9,15/82
ND
ND
1.85**
ND
ND
ND
ND(ND)
12.2**
2.7***
ND
ND
ND
ND
ND
10.**
5.5***
Location
Monitoring Wells
MW-B1
MW-B2
MW-B3
MW-B4
MW-1S
MW-1D
MW-2
MW-3
MW-4S
MW-4I
MW-4D
MW-5
MW-6
MW-7
MW-8S
MW-8D
MW-9
ND
ND
ND
ND
12/17/82-1/5/83 2/2/83-2/14/83 6/8/83
ND
ND
ND
ND
0.4
0.3
<0.01
0.5
0.8
2.4
0.2
<0.01
0.6
0.6
1.4
1.1
0.11
0.13(.05 dup)
0.02*
0.04*
0.07
1.5*
0.09
0.02*
0.02(.05 dup)
0.10
0.11*
0.07
0.40*
NOTES: All analyses for mixed Aroclors unless noted by * or **
All analyses in ug/l or parts per billion (ppb)
* Aroclor 1016/1242 _
** Aroclor 1248 Arcolor 1
-------
1987 Low Flow Tracer Test
The first tracer test at Lemon Lane was conducted under dry,
fall conditions. Injection occurred on November 10, 1987 and
sampling continued until November 27, 1987. Monitoring wells MW-7,
MW-1D, and MW-10 were selected as injection locations. Springs to
be sampled were: Crestmont A and B, Detmer A, Illinois Central,
Quarry, Packinghouse Road (PH Road), Pumping Station, Snoddy A and
B, Stony East, Stony West A and B, and Urban (Figure 8) .
Monitoring wells to be sampled were: MW-5, MW-11, MW-8S, and MW-8D.
Lithium bromide (Br-) was chosen as the tracing agent and two sets
of background samples were taken prior to injection. Background
concentrations of Br- were in the range of 0.01 to 0.39 ppm with a
detection limit of 0.01 ppm.
Sampling frequency was every 6 hours for the first 72 hours
and then every 24 hours for the remainder of the sampling period.
Only Illinois Central Spring and Quarry Springs had definitive
breakthrough curves for Br- that had a classic rising limb, a peak,
and a receding tail. The peak at Illinois Central was 5.11 ppm and
the peak at Quarry was 4.83 ppm. The time to peak from first
detection at both springs was about 250 hours. Illinois Central
rises and flows along the surface for 600 feet to where it sinks
about 200 feet above where Quarry Springs resurges. Based on this
and the tracer detection times, Quarry Springs was presumed to be
second resurgence of the same water. This was later confirmed by
a dye trace run from where the Illinois Central waters sink to
their outlet at Quarry Springs. Crestmont A had one Br- sample
within 15 hours of injection that was 0.50 ppm and Urban spring had
a similar one time occurrence of Br- at 0.66 ppm within 1.5 hours
after injection. Well MW-8D had 1.49 ppm Br- detected in it on
November 25, 1987. It is possible that these Br- detections may be
the tracer injected at Lemon Lane, but it is not considered likely
because they are isolated occurrences.
1989 High Flow Tracer Test
A tracer test during high flow periods was scheduled to be
performed in the spring of 1988. However, drought conditions
prevailed and the test was rescheduled for the spring of 1989. A
different fluorescent dye was injected in each well along with an
aliquot of Br- in each well. Color Index (C.I.) Fluorescent
Brightner 28 (FB28) was injected in MW-7. MW-10 received
Fluorescein (C.I. Acid Yellow 73; AY73). Instead of MW-1D, well
MW-1S was injected with Direct Yellow 96 (DY96) (Figure 2).
Tracers were distributed throughout the bedrock column in these
wells because injection occurred during rising water levels and
because there was interconnection between nested wells. The trace
was to be qualitative with detection to be accomplished visually
for the fluorescent dyes. FB28 and DY96 are adsorbed onto sterile
cotton fabric which is suspended in the resurgence. The cotton is
washed and then observed under a hand-held ultraviolet lamp. FB28
will fluoresce blue, and DY96 will fluoresce yellow. If both
358
-------
V
RESIDENCE WITH
WELL IN USE
SPRING
BACKUP STATION
STREAM
DYE RECOVERED
LESS THAN 0.01%
2000 Feet
4WEIMER ROAD
Figure 8 - 1987 Low Flow Tracer Test
-------
tracers are present, the cotton will fluoresce a characteristic
blue-white. AY73 is adsorbed onto activated charcoal which is
suspended in screen packets in the resurgences. The dye is eluted
from the charcoal with a hydroxide solution and will appear as a
yellow-green layer in the elutant above the charcoal. Background
samples taken before injection indicated minor amounts of FB28 at
some stations. This is expected since FB28 is a component of
laundry detergents. However, no background samples were taken
during storm flows.
Injection of tracers took place on May 26, 1989 after a 1.35
inch rainfall. Several storms of 1" or greater rainfall occurred
during sampling, including a 2.36" rain on July 12, 1989. Figure
9 shows the location of springs and streams that were sampled
during the test until July 22, 1989. Monitoring wells that were
sampled were MW-8S, MW-6, MW-5, and MW-11. Sampling began within
6 hours after injection, and Illinois Central and Quarry springs
had distinct positives for FB28 on the first samples. Breakthrough
curves for Br- were also detected at Illinois Central and Quarry
springs with peak concentrations at Illinois Central of 8.7 ppm and
10.5 ppm for Quarry. Distinct visual positives for AY73 were
detected at Illinois Central and Quarry springs beginning 55 hours
after injection and continued throughout the 57 day sampling
period. The tracer DY96 was not recovered visually at any station
during the sampling period. Apparent detections of AY73 were
recorded for all stations on samples collected following rain
storms. However, the coloration was not the distinctive AY73
yellow-green and it was suspected other organic constituents were
being adsorbed by the charcoal and making the definite
identification of AY73 impossible by visual means. There was no
visual detection of dye at any of the monitoring wells, although
Br- was detected at MW-8D. AY73 was detected visually at
residential well #83 (Figure 9) . There were Br- detections at
Clear Creek of 2.96 ppm at 839 hrs (35 days) and 1.45 ppm at 1147
hrs (47 days) after injection, but since they are isolated
occurrences they are not considered conclusive.
Since the visual detection of the dyes at many stations was
inconclusive, random samples were submitted to Quinlan and
Associates for analysis on a scanning spectrofluorophotometer
(SSFP). These analyses indicated AY73 and DY96 were present in
some samples, even though they were not recognizable visually. A
SSFP was purchased in February 1990, and most of the grab water
samples taken for Br- analysis were analyzed for fluorescence.
Table 2 shows the results of that analysis. These results
confirmed the suspicion that Lemon Lane was at or near the divides
of several subterranean drainage basins, and, that upper level
conduits exist that would divert ground water, during high flow
storm events, into those adjacent spring drainage basins. Actual
dye concentrations were not calculated for these samples, but the
relative intensity of fluorescence as measured by the SSFP
indicated that Illinois Central and Quarry Springs received the
majority of the tracers and all the other stations received only
minor amounts. It was obvious that the proportional distribution
360
-------
•SSSVVSSS*
*ILLINOl'sYTHaADAMS
^CENTRAL :
RESIDENCE WITH
WELL IN USE
SPRING
BACKUP STATION
STREAM
DYE RECOVERED
LESS THAN 0.01%
2000 Feet
^WEIMER ROAD '
Figure 9 - 1989 High Flow Tracer Test
-------
Table 2 - Results of Spectrofluorophotometer Analysis
1989 High Flow Tracer Test
Station
Detmer A
Detmer B
Defeat E.
Defeat W.
K1rby Rd.
Stony E.
Stony M.
Sinking Cr.
WN-1
WS-2
ICG-1
ICG-Z
ICG-3
ICG- 6
Pump St.
Urban
Bypass 37
S. House
PHRoad
Hinkle
Snoddy ASB
17th St.
IL Centra]
8 Allen
Weiner Rd.
Stouts U.
Stouts E.
HW-8S
MU-6
MM- 11
MW-5
5/26
ND
FB28
DY96
FB28
ND
ND
—
.FB28/
AY73
iV7T
HI / J
FB28/
DY96
ND
ND
FB28
FB28
DY96
ND
FB28/
AY73
FB28
FB28
FB28
FB28
FB28
FB28
DY96
FB28
ND
NO
FB28
SS
S§
5/27
ND
FB28
DY96
DY96
DY96
DY96
DY96
iY7T
ft I / 0
ND
ND
FB28
FB28
DY96
ND
DY96
ND
ND
FB28
ND
DY96
ND
ND
FB28/
AY73
ND
ND
ND
ND
o o
So
o
5/27
DY96
DY96
—
ND
ND
-
ND
FB28/
AY73
ND
ND
ND
FB28/
AY73
AY73
ND
ND
DY96
ND
DY96
ND
ND
ND
ND
DY96
ND
FB28
ND
ND
FB28
ND
ND
0 o
o o
0 tNJ
o o
5/28
DY96
ND
DY96
DY96
ND
FB28
ND
AY73
AY73
ND
DY96
ND
ND
ND
ND
ND
ND
DY96
ND
ND
DY96/
AY73
ND
ND
ND
ND
ND
ND
o o
o o
<£> CO
0 0
5/28
DY96
ND
DY96
ND
FB28
ND
DY96
AY73
un
nu
ND
ND
NO
ND
NO
AY73
ND
ND
FB28
--
AY73
ND
ND
ND
NO
FB28
ND
0 O
0 O
5/28
ND
ND
DY96
DY96
ND
DY96
AY73
ND
NO
--
ND
ND
ND
ND
AY73
AY73
ND
DY96
ND
DY96/
AY73
ND
ND
ND
ND
ND
ND
o o
§§
i— i CO
5/28
DY96
DY96
ND
ND
DY96
ND
DY96
AY73
wn
nu
ND
AY73
ND
ND
ND
ND
AY73
AY73
ND
FB28
DY96
ND
ND
NO
NO
ND
ND
5/29
ND
NO
DY96
DY96
DY96
ND
ND -
DY96/
AY73
un
nu
ND
AY73
ND
--
DY96
NO
AY73
AY 73
NO
ND
DY96/
AY73
ND
ND
ND
ND
ND
FB28
ND
5/30
ND
NO
DY96
DY96/
AY73
DY96
ND
ND
FB28/
AY73
un
nu
ND
AY73
ND
ND
ND
ND
AY73
AY73
DY96
ND
DY96/
AY73
DY96/
AY73
ND
ND
NO
ND
NO
ND
5/31
ND
OY96
DY96
ND
DY96
AY73
AY73
DY96/
AY73
ND
ND
DY96/
AY73
ND
ND
ND
AY73
AY73
DY96
AY73
ND
NO
ND
ND
AY73
ND
ND
6/1
ND
DY96
OY96
ND
DY96
AY73
ND
AY73
MD
nu
AY73
—
ND
ND
ND
AY73
AY73
DY96
AY73
AY73
ND
ND
ND
ND
NO
ND
6/2 6/9 6/16
OY96 DY96 ND
ND DY96 DY96
DY96 - DY96
DY96 ND
ND DY96 ND
DY96 ND ND
AY73 DY96 ND
AY73 ND DY96/
AY73
ND
AY73 DY96/ AY73
AY73
AY73 ND ND
ND ND ND
DY96/ DY96/ AY73
AY73 AY73
ND ND ND
AY73
AY73
DY96 —
AY73 ND ND
DY96/ —
AY73
ND ND OY96
ND DY96 ND
ND NO OY96
ND ND ND
ND FB28/ ND
AY73
NO FB28 ND
ND ND ND
6/23
ND
--
ND
OY96
ND
DY96
DY96
—
AY73
ND
ND
AY73
ND
—
ND
AY73
ND
ND
ND
ND
ND
ND
ND
6/30
DY96
-
OY96
ND
ND
DY96
--
—
AY73
AY73
ND
DY96/
AY73
ND
-.
ND
AY73
DY96
AY73
ND
ND
ND
ND
ND
7/7
DY96
DY96
DY96
NO
DY96
--
--
DY96
AY73
--
AY73
DY96
DY96/
AY73
DY96
—
DY96
AY73
NO
DY96
ND
ND
ND
ND
7/13
DY96
DY96
DY96
ND
ND
--
--
AY73
ND
—
ND
ND
DY96
ND
ND
DY96
AY73
ND
ND
ND
ND
ND
ND
7/22
NO
ND
ND
ND
DY96
--
-
AY73
—
--
ND
ND
ND
ND
—
ND
AY73
ND
ND
ND
ND
ND
ND
ND - Not Detected
Not sampled or not analysed
FB28 Optical brightners detected
AY73 Fluorescein detected
OY96 Direct yellow 96 detected
362
-------
of groundwater would have to be ascertained. A quantitative high-
flow tracer test was scheduled for the spring of 1990.
1990 High Flow Tracer Test
Background sampling was conducted at all monitoring stations
to ascertain which fluorescent tracer would have the least amount
of background interference and be most suitable for the
quantitative trace. The sampling indicated that Rhodamine WT,
(C.I. Acid Red 388; RWT) would be appropriate. Some stations
however, did have background concentrations of RWT in grab samples
of water analysed on the SSFP. The detection limit of RWT on the
SSFP was 0.01 ppb (Table 3).
RWT was detected at 7th & Adams on 4/19/90 in the elutant from
a charcoal packet. In addition AY73 was detected in Quarry Springs
at 0.51 ppb. Eighty-two residential wells in the vicinity of
Lemon Lane were monitored during the tracer test by teams from
federal, state, and local agencies (Figure 10). The one residence,
#83, that had AY73 detected in its well during the 1989 test, was
subsequently connected to city water and this well was not
monitored during 1990. Residential wells #82 and #212 had AY73
detected in them during background monitoring for the 1990 test.
Spring and stream stations were gaged, where possible, and
Table 4 lists the stations and the flow monitoring method.
Illinois Central spring, per se, was not monitored for this test,
rather a gaging station was installed below Quarry Springs.
Table 3 - Background Determination of
Rhodamine WT in Water Samples
in parts per billion (ppb)
STATION
Quarry Spring
Urban Spring
Bypass 37
Stony East
Clear Cr @ 1st St.
ICG-2 Spring
Cascade Br.
Weimer Rd.
7th & Adams
3/20/90
0.075
0.095
0.075
0.058
0.31
0.15
0.095
0.058
ND
4/2/90
ND
ND
ND
ND
0.125
ND
ND
ND
ND
4/19/90 5/7/90
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
0.11
ND
ND
ND
ND
363
-------
STOUTS^STOUTS
WEST rl FAST
*
-------
TABLE 4 - MONITORING STATIONS FOR 1990 TRACER TEST
Station is downstream
of a Spring
13. Bypass 37 (re)
26. Defeat Cr. E. (re)
27. Defeat Cr. W. (re)
28. Kirby Road (re)
29. Sinking Creek (re)
Station is near
a Spring
l. Quarry Spring (w)
3. ICG-1 (re)
4. ICG-2 (w)
5. ICG-3 (w)
6. ICG-6 (w)
7- Fell Iron (nm)
9. Crestmont (w)
10.Pumping Stn. (nm)
11.17th Street (nm)
12.Urban (w)
14.S.House (w)
15.PH Culvert (w)
16.PH Road (w)
17.Snoddy A (w)
IS.Snoddy B (w)
19.Hinkle (w)
20.Abrams (w)
21.Walcott A (w)
22.Walcott B (w)
23.Robertson (re)
24.Detmer A (nm)
25.Detmer B (w)
30.Stony West (re)
31.Stony East (re)
32.WN-1 (w)
33.WS-2 (w)
(w) - 90° v - notch weir
(re) - rating curve with staff gage
(nm) - not measured
Station is a backup
point downstream
of several primary
of several Springs
2. 7th & Adams (re) Cascade Br. (re)
8. Clear Cr. @ 1st (re) Stouts Cr. W. (re)
Stouts Cr. E. (re)
111 Cen.@ Allen (re)
Weimer Road (nm)
Clear Cr. @ 7th (nm)
365
-------
Injection of tracers began on May 12, 1990 after a 1.45" rain
fell. RWT was injected as a 20% solution and the injected amounts
are corrected for this dilution. Well MW-1S received 9,545 grams
of RWT, MW-10 received 9,545g, and MW-7 received 8,183g for a total
of 27,273 grams of RWT injected. Large amounts of RWT were
injected because RWT is known to suffer large adsorption losses and
high suspended clay and organic matter loads were observed during
storm flows at Quarry Springs.
RWT was first detected at Quarry Springs 3.75 hours after
injection and reached a peak concentration of at least 4300 ppb
(distinctly visible) at 5.25 hours. By 10 hours after injection
the concentration had receded to 40 ppb and remained at or below
that level through the last sample taken on August 24, 1990.
Five other stations had repeated detections of RWT. They were ICG-
1, Slaughterhouse, PH Road, PH Culvert, and Clear Creek at First
St. However, Clear Creek did not have a breakthrough curve during
the first 72 hours of sampling. Only two samples from Clear Creek
were above the background level of 0.31 ppb. They were a 2.6 ppb
detection at 584 hours after injection and a 0.7 ppb detection at
593 hours. The Clear Creek watershed is urbanized and the specific
spring heads could not be located or monitored. A determination of
Clear Creek's connection with Lemon Lane above its confluence with
the Illinois Central-Quarry Spring Branch is not conclusive.
Ten stations had sporadic and low (near background level)
detections of RWT in the grab water samples. They are listed in
Table 5, along with the date, time, concentration detected, and
amount detected in background samples. In addition the following
stations on the following dates had RWT detected in the back-up
activated charcoal samples: Fell Iron Spring on 6/8/90 and 6/15/90,
Snoddy A on 7/6/90, Snoddy B on 7/6/90, Detmer A on 7/13/90, and
Stony East on 7/14/90. Three residential wells, #82, #86, and
#212, had RWT detected in them. Residential well #86 had been
abandoned when the owner connected to city water in 1986, but was
elected to be sampled anyway by state agency representatives. Well
#212 also had AY73 detected in it. Wells #82 and #212 were
analyzed for PCBs and none were found to a detection limit of 0.1
ppb.
Calculations for mass balance were performed to determine the
dye recovery at the six stations that had repeatable RWT
detections. Those results are listed in Table 6. Clear Creek at
First St. was included even though the connection is problematic.
Only 18% of the total amount of RWT injected was recovered. The
amount not recovered is attributed to losses by adsorption,
underestimation of peak flows, underestimation of peak
concentrations due to quenching effects of high concentrations on
the SSFP, and tracer in ground-water storage. Sampling frequency
was sufficient to catch all peaks of dye detections and is not
considered to be a significant contributor to the lack of dye
recovery. Slightly less than 97% of the amount recovered occurred
at Quarry Springs. ICG-1 appears to be related to Quarry, perhaps
366
-------
TABLE 5 - RHODAMINE WT APPEARANCES AS OF 6/29/90
Station
Kirby Rd.
Stony West
WN-1
Stouts W.
ICG-6
Clear Cr.
@ 7th
Cascade Br.
WS-2
Bypass 37
Defeat E.
Date
5/19/90
5/12/90
5/12/90
5/12/90
5/18/90
5/29/90
5/13/90
5/14/90
5/15/90
5/25/90
5/15/90
5/28/90
6/15/90
Time
0148
2230
2250
2033
2233
1130
1130
0520
0720
0930
0525
0520
1215
0641
0920
0930
Concentration
360 ppt
260 ppt
290 ppt
115 ppt
280 ppt
58 ppt
58 ppt
85 ppt
75 ppt
120 ppt
42 ppt
44 ppt
275 ppt
54 ppt
42 ppt
65 ppt
Background
ND
ND
85 ppt
ND
ND
ND
95 ppt
ND
75 ppt
ND
367
-------
as a subsidiary overflow route during storm events or as a
distributary resurgence. Based on their morphology, elevations,
and comparison of dye breakthrough curves, it is concluded that
Slaughterhouse, PH Road, and PH Culvert are distributary outlets of
a main conduit draining in that direction. The conclusion then, is
that about 98% of the ground water from Lemon Lane resurges in the
vicinity of Quarry Springs, about 1% or less flows to the three
distributaries to the northwest, and another 1% or less may flow to
other headwaters of Clear Creek above the Illinois Central-Quarry
Spring Branch confluence. Some of ground water from the vicinity
of Lemon Lane may be diverted to numerous other spring basins
during storm flows, but the amount of ground water is probably no
more than 0.01% of the total from Lemon Lane based on the tracer
recovery ratios (Figure 10).
Hydrogeology of Lemon Lane Landfill
An examination of the Cross-section in Figure 3 shows that
shallow strata on the east dip to the south-west and strata on the
west slope trend to the south-east. This is interpreted to be an
effect of sinkhole subsidence and collapse, as well as development
of the conduit along a structural sag and trough. Strata adjacent
to the conduit that collapsed also subsided in the direction of the
collapse. A zone of interbedded limestone, dolomites, and shales
at approximately the 798'-805' elevation is where the major water
producing fractures occur. The different strain properties of the
different interbedded lithologies caused bedding plane openings,
vertical jointing, and general fracturing of the rock units in
response to the subsidence stresses. This produced the zone of
enhanced permeability noted on the cross-section. A higher density
of fracturing also occurs within the first 10' of the bedrock.
This is in accordance with Williams (1983), Ford and William (1989,
p. 206) , and others, who report that the uppermost layers of
bedrock constitute a suspended aquifer termed the subcutaneous zone
(Williams, 1983) or the epikarst (Mangin 1975).
TABLE 6 - RHODAMINE WT RECOVERY
Station
Quarry
ICG-1
Clear Cr. @ 1st
Slaughterhouse
PH Road
Ph Culvert
Mass (grams)
Recovered
4744.243
64.135
60.192
22.032
5.589
2.562
4898.753
% of
Injected
17.40%
0.24%
0.22%
0.08%
0.02%
0.01%
17.97%
% of
Recovered
96.85%
1.31%
1.23%
0.45%
0.11%
0.05%
100%
368
-------
MW-1S
MW-11
Figure 11 - Hypothetical Base Flow Map
MW-1S
MW-10
*&'
vMW/7^--!ite.!!.unlary
-838
\
£
MW-11
Figure 12 - Hypothetical High Flow Map
369
-------
ground-water flow and contaminant, movement based on that assumption
erroneous. The original monitoring plan, which was predicated on
the basis of intercepting a contaminant plume, with wells, is
incapable of yielding reliable or accurate monitoring - except by
improbably good luck.
(2) The aquifer has been shown to be a karst aquifer, one
dominated by conduits in which flow is convergent; therefore,
efficient, reliable, and time-sensitive monitoring of off-site
movement of contaminants is best accomplished through the sampling
of springs shown to be connected to the landfill by tracing as
described by Quinlan and Ewers (1985) and Quinlan (1989). The
tracer tests have clearly shown that most of the ground water flows
rapidly via conduits to Illinois Central-Quarry springs during high
flows. Off-site releases of contaminants in directions other than
the Illinois Central-Quarry system would more efficiently and
reliably be detected at some, or all, of those 32 springs shown by
tracing to have intermittent connection with Lemon Lane rather than
off-site wells.
(3) Since movement of ground water occurs in different
directions during different flow regimes, monitoring schedules must
be adjusted on the basis of hydrologic criteria. For example,
during pre-excavation baseline sampling, it is proposed to collect
samples during storm events as well as during base flow periods.
When material is being excavated from Lemon Lane, it is proposed to
sample Illinois Central Spring and Quarry Springs weekly and also
during every storm event of sufficient magnitude and duration to
elevate the discharge of those springs above their previous week's
base flow discharge.
(4) On-site monitoring wells will provide water level data and
samples to accomplish monitoring of ground water in storage near
the site. This is a necessary component of a complete monitoring
plan but, at Lemon Lane, the on-site wells were shown by tracing to
be unreliable monitors for ground water leaving the site. This is
because wells tend to intersect ground water in the diffuse bedrock
aquifer and miss the conduits that convey most of the water. Part
of that ground water is recharge from the backflooding of the
conduits. However, the probability of wells intersecting conduits
conveying the majority of the ground water, and, hence, the
contaminants leaving the site is low. For the same reason, the use
of proposed off-site wells to intercept contaminants that have left
the site is considered to be inefficient and unreliable, especially
since tracing has shown the springs to be time-sensitive and
accurate monitoring stations for the potential off-site release of
contaminants.
370
-------
DEVELOPMENT OP A MONITORING PROGRAM AT A SUPERFUND SITE
IN A KARST TERRANE NEAR BLOOMINGTON, INDIANA
Michael R. McCann, Westinghouse Electric Corp.
Noel C. Krothe, Indiana University
Bloomington, Indiana
Q: Do you have any reason to believe that all the dye is out of the
system?
A: On the contrary, we have evidence the dyes are still in the
system. As late as September 23, 1991 Illinois Central Spring had
8.7 ppb Fluorescein and 12.3 ppb Rhodamine WT and Quarry Spring had
1.8 ppb Fluorescein and 8.0 ppb Rhodamine WT detected in grab water
samples. We believe this is due to two reasons. One is that
injection wells MW-1S and MW-10 were relatively "tight" and a
portion of the dye simply remains in the fractures near the wells,
draining slowly into the conduit system. This is particulary true
of well MW-1S; during purging of this well for PCB sampling a great
deal of concentrated Rhodamine WT was evacuated. Apparently, very
little of the dye from this well entered the system at all.
Secondly, a portion of the dye is thought to be in storage in the
epikarst and adjacent fractured bedrock. This storage is either an
artificial result of injection into rock borings under rising water
levels, or a result of the backflooding of the conduit system
during the storm event.
Q: When is it best to inject dye? Why didn't you catch the dye at
its peak concentration?
A: If time and money will allow, it is best to inject dye first on
a moderate flow event, followed by repeat injections on a low flow
and then high flow event. At a minimum, one low flow event
followed by one high flow event should be conducted. The low flow
event is necessary to formulate your base-flow baseline sampling
schedule. The high flow event is necessary, not only to establish
your high-flow baseline sampling schedule, but also to make sure
all off-site flow routes have been established, because upper level
overflow routes may exist that operate only at the higher flows.
The dye pulse came through quickly, and even though the
sampling frequency was every 15 minutes, it may not have been
enough to catch the peak. In retrospect, when dye begins to appear
visually, it is probably necessary to increase one's sampling
frequency to every 5 minutes or less.
371
-------
372
-------
HETEROGENEITY IN CARBONATE AQUIFERS;
EFFECTS OF SCALE, FISSURATION, LITHOLOGY AND KARSTIFICATION
SMART, P.L., A.J. EDWARDS AND S.L. HOBBS+
Department of Geography, University of Bristol,
BRISTOL, BS8 1SS, England.
+Aspinwall and Co. Ltd, Walford Manor,
SHREWSBURY, SY4 2HH, England.
Abstract
The flow transmission properties of carbonate aquifers may be determined and inter-
compared using cumulative probability plots of log hydraulic conductivity obtained from
slug and bailer tests in observation boreholes. Such tests are simple and inexpensive, and
allow determination of both average hydraulic conductivity, and an indication of the extent
of aquifer heterogeneity. Karstified aquifers, such as the Carboniferous Limestone of the
Mendip Hills, exhibit a large range of hydraulic conductivities, while fissure flow aquifers,
such as the Great Oolite of the adjacent Cotswold Hills, have both higher hydraulic
conductivities and are less heterogeneous. The importance of fissure density and
karstification can be illustrated using caliper logs and aquifer tests of different scales.
Spatial variations of aquifer properties also occur, and can be related to variations in
lithology and structure, and to the effects of quarrying.
Introduction
It is generally recognised that the behaviour of carbonate aquifers is dependent on the type
of groundwater flow occurring, although the recharge and storage properties are also of
importance (Smart and Hobbs, 1986). Various authors suggest a continuum of flow types
ranged between two end members (conduit and diffuse flow; White (1977), Smart and
Hobbs (1986)), three end members (conduit, diffuse and fissure network flow; Atkinson
(1985)) and even four end-members (granular, fracture, diffuse and conduit flow; Quinlan
(1988)). Hydrodynamic (Aley, 1975; Atkinson, 1977) hydrochemical (Schuster and White,
1971) methods have been used to characterise flow type for particular aquifers, and a list
of possible characteristic criteria was presented by Smart and Hobbs (1986). Another
approach is to consider the spatial variation of aquifer transmission properties which can
be obtained from borehole aquifer tests (specific yield, transmissivity or hydraulic
conductivity). Diffuse flow aquifers would be expected to exhibit only limited spatial
373
-------
variations while karstified aquifers are characteristically heterogeneous and would show
much greater variation (Daoxian, 1986; Hobbs and Smart, 1988), in this paper we present
data derived from simple slug and bailer tests, which can conveniently be displayed using
cumulative probability plots, a tool often used by hydrogeologists to display well specific
capacity data (Csallany and Walton, 1963). The data may also assist in understanding the
factors controlling the spatial distribution of aquifer hydraulic conductivity, and provide a
data-base suitable for parametarisation of numerical aquifer models.
Methods
Slug and bailer tests (Ferris and Knowles, 1963) provide a simple, quick and cheap method
for estimating aquifer hydraulic conductivity. Unlike the pumping tests, more
conventionally employed to determine aquifer parameters, they do not require installation
of submersible pumps, and can thus be undertaken in small diameter observation boreholes.
The logistic support required to install, power and discharge water from the pump is thus
eliminated. Typically slug and bailer tests can be completed in a few hours and require
only one or at the most two operators they are therefore ideal for geotechnical and other
low costs investigations in which development of a pumping supply borehole is not the
prime objective. In these tests, a slug of water is either introduced into or removed from
the borehole, and the return of the water level to pre-existing conditions is monitored,
preferably using a sensitive pressure transducer and chart recorder or data logger. The
initial change in water level should theoretically be instantaneous, and it is often easier to
approximate this condition by either inserting or removing a long weighted float to
generate the disequilibrium in head conditions. A number of theoretical solutions to this
problem have been presented which permit direct calculation of hydraulic conductivity in a
variety of different situations (see review by Karasaki et al, 1988). In our study of water-
table aquifers the solutions of Bouwer and Rice (1976) and Van der Kamp (1976) for the
underdamped case were employed. It should be noted that these solutions are based on the
assumption of laminar flow within the aquifer, wheras in aquifers with dissolutionally
enlarged voids turbulent flow may develop under the test conditions.
Hydraulic conductivity data follow a log-normal distribution and thus can be linearised by
plotting log hydraulic conductivity on cumulative frequency probability paper. The mean
and standard deviation are determined for the normally distributed log data, and can be
used to calculate the coefficient of variability. The standard deviation can also be
converted back to non-logarithmic form to provide upper and lower limits which are
asymmetric about the mean, but are expressed directly in the units of hydraulic
conductivity (here m/d).
Comparison of Hydraulic Conductivities from Four Contrasting Carbonate Aquifers
Hydraulic conductivities have been obtained from slug and bailer tests (henceforth slug
tests) in four contrasting aquifers. The Carboniferous Limestone aquifer of the Mendip
Hills is a maturely karstified aquifer developed in massively bedded, well jointed
limestones of low primary porosity (< 0.1 %) (Green and Welch, 1965). Flow in the aquifer
is predominantly via conduits, but fractures and fissures provide the majority of the
saturated zone storage in the diffuse flow zone adjacent to the conduits (Atkinson, 1977).
Abstraction is wholly from spring sources. The slug tests were conducted in the eastern
Mendips (Hobbs and Smart, 1988; Atkinson et al, 1973), where surficial karst features are
less well developed than in the better known and more mature central Mendips area
374
-------
Fractured Quarry Sub-floor
• (Carboniferous Limestone)
: Pleistocene
Lucayan
•\ Limestone
Fissured
Great
Oolite
Karstified
Carboniferous
Limestone
EB73 Caliper extension (cm)
10 15 20 25 30
35
2-
4-
6-
Q.
0
Q
8-
10-2
50 70 90
Cumulative probability (%)
Figure 1 Cumulative probability plot of log hydraulic
conductivity determined from slug tests for quarry sub-floor,
Lucayan Limestone, Great Oolite and Carboniferous Limestone
aquifers.
10-
12-
14J
'Intact' sections - Black
'Fissured' sections - White
-
Total saturated thickness 13.70m
Total'fissured'sections 11.55m
Total Intact' sections 2.15m
Flssuration = '5S x 100 = 84.3%
Figure 2 Caliper log for Grand Bahama borehole EB 73, and
calculation of fissuration index.
-------
described by Atkinson (1977). Within the East Mendip area, the Carboniferous Limestone
has been extensively extracted by hard-rock quarries, which extend below the natural
water table (Edwards et al; this volume). Modern blasting techniques, which are designed
to comminute the rock, also cause extensive fracturing of the Carboniferous Limestone
both below the quarry floor and laterally away from the walls (Holmberg and Maki, 1981;
Gunn and Gagen, 1987). Where it is below the water table the sub-floor blast zone can be
considered a separate aquifer characterised by high fracture density, low primary porosity
and negligible conduit development. Our data are derived from shallow (< 10m) boreholes,
drilled into the sub-watertable quarry floor.
The Great Oolite aquifer of the southern Cotswolds comprises massive oolitic and bioclastic
limestones (Kellaway and Welch, 1980), and has been extensively developed by large
abstraction boreholes (> 10 ML/d). Transmissivities are high and are dominated by
development of dissolutional fissures upon the bedding planes and extensive joint systems.
These fissures provide the majority of the storage (specific yield c 3%), but there is also
significant intergrannular porosity (up to 15%). Test data has been derived from a network
of observation boreholes. Unlike the Great Oolite and Carboniferous Limestone aquifers,
the Pleistocene Lucayan Limestone aquifer of the Bahamas has never suffered deep burial
(Beach and Ginsberg, 1980), and has a high intergrannular porosity (typically 40%) and
specific yield (5-6 %). It has however been karstified, with vadose meteoric leaching
during times of low Quaternary sea-level, and dissolution in the freshwater phreatic and
mixing zones at times of high sea-stands (Smart and Whitaker, 1988), resulting in vuggy
porosity and development of coastal (dune-flank) caves (Mylroie and Carrew, 1990). The
data presented here have been derived from observation boreholes and drainage wells on
the island of Grand Bahama.
Hydraulic conductivity data from these four contrasting carbonate aquifers is presented in
Figure 1, while summary statistics are given in Table 1. The densely fractured blast zone
aquifer has the highest mean hydraulic conductivity, and is also spatially the most
homogeneous of the aquifers. The Lucayan Limestone and Great Oolite aquifers attain
similar maximum hydraulic conductivities but exhibit a much greater range of test values,
indicating greater heterogeneity. In the case of the fissured Great Oolite this greater
heterogeneity results from a relatively small proportion of sites having either unusually
high, or very low hydraulic conductivities, the lower quartile spanning two orders of
magnitude. The former are associated with fissure network zones which are responsible for
the rapid transmission observed in the aquifer from tracer tests (Smart 1976), while the
latter represent tight zones where bedding plane and fracture density are low, and
extensive dissolution to generate fissures has not occurred. A similar pattern is seen for
the karstified Carboniferous Limestone, the data from which also span four orders of
magnitude. None of the boreholes tested intersects on open conduit. Had this been the
case, much higher maximum apparent hydraulic conductivities would have been obtained,
significantly increasing the range. The mean hydraulic conductivity of the Carboniferous
Limestone is much less than that of the Great Oolite, both because of the more massive
bedding, and because concentration of flow into conduits at an early state in the aquifer
development limits the extent of dissolution in the diffuse zone (Ewers 1978). In the more
densely fractured Great Oolite, competition between flow routes of similar size ensures a
more uniform distribution of dissolution (Palmer, 1984).
Thus conduit flow aquifers are characterised by high heterogeneity and a rather low
transmissivity dominated by the diffuse flow zone. Fissured aquifers have a higher
transmissivity and significantly lower heterogeneity, as indicated by the coefficient of
variation (Table 1), and the gradient of the log probibility plot. Both the Pleistocene
Lucayan Limestone, in which intergrannular flow is significant, and the fracture flow
quarry sub-floor have high transmissivity and low heterogeneity. The mean transmissivity
376
-------
Hydraulic Conductivity (m/d)
Quarry Sub-Floor
Lucayan Limestone
65 70 75 80 85 90 95 100
% Fissuration
Figure 3 Relation between fissuration and log hydraulic conductivity for Lucayan Limestone
and quarry sub-floor aquifers.
Hydraulic Conductivity (m/d)
10l==
10'= =
10;
10==
n Quarry Sub-Floor
0 Carboniferous Lstn.
A Lucayan Limestone
njf]
<\A
D
A
A
A
EE 0 0
0 0
IOH 1 1 1 1 1 1 1 h—
10 20 30 40 50 60 70 80 90 100
% Fissuration
Figure 4 Comparison of relation between fissuration and log hydraulic conductivity for
Carboniferous Limestone and Lucayan Limestone and quarry sub-floor aquifers.
377
-------
is thus an indication of the extent of dissolution enlargement of voids within the aquifer,
while the coefficient of variation (or slope of the log probability plot) indicates the degree
of flow concentration occuring on the continuum, from diffuse to conduit flow (Smart and
Hobbs, 1986).
Role of Dissolution and Void Integration
If the aquifer hydraulic conductivity at a particular site is controlled by the development
of dissolutional fissures, it should be possible to predict hydraulic conductivity directly
from the aperture and number of fissures penetrated by the borehole. In order to
investigate this relationship, selected boreholes were caliper logged using a Gerhart T-450
borehole logging system (Keys and Maccary, 1971). An index of fissuration was then
derived from the caliper logs by summing the length of borehole segments which were
greater than the nominal drilled diameter, and expressing this sum as a percentage of the
total saturated borehole depth (Figure 2). The index should thus depend both on the
number and size of fissure openings, although in practice the latter will be overestimated
by 'break-out' of the drill bit into the fissure. It is also important to remember that
widening of the hole may not simply be due to the presence of open fissures, but can
occur in softer rock units, such as poorly cemented limestones in the diagenetically
inmature Lucayan Limestone, or shales in the Carboniferous Limestone. Furthermore, the
amount of 'break-out' which occurs into a fissure is strongly dependent on method of
drilling, drilling rate and rock matrix strength, which are not controlled in our data set.
Figure 3 presents data from the Lucayan Limestone aquifer of Grand Bahama, and the
quarry sub-floor aquifer. For these there is a remarkably good relation between
percentage fissuration and the measured hydraulic conductivity (which is not dependent on
the coincidence of the quarry and Lucayan Limestone relationships). Thus even in young
limestones with high intergrannular porosity, the transmission properties of the aquifer are
governed by development of dissolutional fissures. In the Bahamas these form along the
frequent exposure surfaces which terminate early carbonate sub-unit and are developed
during low sea-stands; such surfaces are laterally continuous and thus generate very high
horizontal hydraulic conductivities.
In Figure 4, the Bahamas and quarry sub-floor data are supplemented by a data from the
karstified Carboniferous Limestone aquifer. Because the intergrannular porosity of this
limestone is very low, secondary voids would be expected to dominate the transmissivity.
In fact there is not a significant relationship between the fissuration index and borehole
hydraulic conductivity (it should however be noted that unfortunately the logged boreholes
did not include any sites with above average hydraulic conductivities). The probable
explanation for this unexpected finding is related to the connectivity of individual fissure
openings. Packer testing has shown previously that some fissures yield hydraulic
conductivities similar to the bulk rock, indicating that they do not interconnect laterally
with other voids in the aquifer (Hobbs and Smart, 1988). They thus contribute to storage
within the aquifer by delayed leakage, but are not important for flow. Thus as emphasized
by Sendlein and Palmquist (1977) void intergeration is critical in controlling transmission in
karstified rocks. Whilst in the blast zone and Bahamas fissure voids integrate laterally, in
the Carboniferous Limestone this is not the case and the resulting aquifer is both less
transmissive and more heterogeneous (Figure 1).
Careful examination of the data for the individual aquifers (Figure 1) suggests that there
may be sub-populations present in the data, as indicated by linear segments of differing
gradient on the cummulative probability plot. This is illustrated for the Carboniferous
378
-------
10'
o
•§ 10°-
o
o
o
1
t
10-'
10-
'Well Connected Fissures'
'Fissures'
Fractures
Unconnected
^ Fractures'
10 30 50 70 90
Cumulative probability (%)
99
Figure 5 Slug test data for the karstified Carboniferous Limestone sub-divided into linear
components.
Table 1 Comparison of hydraulic conductivity data for the four aquifers
Carbonate
Aquifer Type
Fractured Quarry
Sub-Floor
Pleistocene Lucayan
Limestone
Fissured Great Oolite
Karstified Carboniferous
Mean
(m/d)
305
96.2
31.8
0.214
Coefficient
of Variation
(%)
10.4
29.5
70.1
179
Total Number
of Boreholes
11
44
24
46
Limestone
379
-------
Limestone aquifer (for which we have most data) in Figure 5, which shows four sub-
populations represented by straight line segments. These sub-populations may simply be a
statistical artefact, but could also represent the combined effects of void integration and
dissolution. Fractures (width mm) are unmodified joints and beddings developed by
unloading an re-exposure of the limestone, while fissures (width cm) have suffered
dissolutional modification, and in some cases may have turbulent flow. They have,
however, not expanded to the size of conduits (width tens of cm to m). Thus while
fractures give lower hydraulic conductivities than the larger aperture fissures, the degree of
void integration is also critical. Unconnected fractures give very low hydraulic
conductivities, which increase significantly once integration of the net occurs, for instance
by continued unloading (or in the quarry sub-floor case by blasting). The enhanced
circulation permits dissolution generating fissures, but initially these are interconnected
only by fractures which effectively limit the increase in hydraulic conductivity. Continued
circulation expands these fracture links giving a net of highly connected fissures with high
hydraulic conductivity. In fact it is precisely such a net which characterise the Great
Oolite aquifer. These suggestions could be more rigourously tested by application of
simulation models, such as those developed by Foster and Milton (1974) for the Chalk
aquifer of southern England.
Controls on the Spatial Variation of Aquifer Hydraulic Conductivity
Because slug tests are relatively rapid and simple they can be used to generate sufficiently
large data sets that controls on the spatial variations of aquifer hydraulic conductivity can
be investigated (provided sufficient suitable boreholes are available). For the East Mendip
Carboniferous Limestone data set, an initial sub-division was made into boreholes on the
steeply dipping (c 70°) northern limb of the Beacon Hill pericline, and those from the more
gentle southern limb (dip c 30°) (Figure 6). These differ significantly in hydraulic
conductivity (at the 95% confidence interval), the northern limb having an average
hydraulic conductivity on order of magnitude less than the southern limb. The northern
limb also has markedly less sites in the 'fissure' range (10° to 10 ^ m/d). These differences
are paralleled by a much greater degree of fissuration on the southern (average fissuration
index 78.6%) than the northern limb (34.6%). As both areas have a similar geomorphic
history, this difference may be attributed to the difference in dip which controls the
number of bedding planes intersected by a vertical borehole. This is much lower in the
steeply dipping beds, of the northern limb, than for the more gently dipping southern
limb. Because it is the bedding planes which provide laterally continuous and thus
interconnected openings, the steep dips yield lower hydraulic conductivities on the northern
limb.
The Carboniferous Limestone data was also sub-divided with respect to the lithological
group within the Carboniferous Limestone Series (Green and Welch, 1965). Again
significant differences are apparent, both with regard to the average hydraulic conductivity
and the heterogeneity indicated by the coefficient of variation (Figure 7, Table 2).
Although there are lithological differences between the groups, (the Vallis and Hotwells
Limestones are coarse, pure bioclastic limestone, the Clifton Down contains a major
mudstone unit, and the coarse bioclastic Blackrock Limestone has both significant shales
and chert beds), these lithological factors appear to be less important than disposition in
the aquifer. Thus the impure Blackrock Limestone would be expected on lithological
grounds to be less transmissive than the pure Vallis Limestone. The converse is in fact the
case, because the Blackrock Limestone has been exposed to much greater dissolution as it is
at the base of the limestones, and receives aggressive allogenic runoff and diffuse leakage
from the topographically higher siliclastic rocks forming the core of the anticline. The
380
-------
CO
00
101
o
o
o
o
10°-
10-'
Northern Limb
Dip 70° (34.6%)\
(Percentage fissuration
in brackets)
VSouthern L/nrt>
30° (78.6%)
10 30 50 70 90
Cumulative probability (%)
99
Figure 6 Comparison o) hydraulic conductivity data for
boreholes from the northern and southern limbs of the Beacon
Hill pericline. Carboniferous Limestone aquifer.
10
• Black Rock
o Clifton Down
• Hotwells
• Vallis
10-2
10 30 50 70 90
Cumulative probability (%)
Figure 7 Slug test data for the Carboniferous Limestone
aquifer sub-divided into lithologic groups.
-------
Vallis Limestone in contrast has received only autogenic recharge, and dissolutional
integration of the initial fractures has been limited.
Interestingly, the Blackrock Limestone has also been more extensively quarried than the
other limestone groups, and some of the tested boreholes are in fact drilled within the
quarries. If we only consider sites which have water-levels well below (< 10 m) the sub-
floor blast zone, these have hydraulic conductivities much higher than boreholes which are
remote from quarries (>150 m; Figure 8). Whilst this may be explained by the lithological
controls discussed above, the implications for geotechnical aquifer parameterisation studies
which draw data only from the quarry sites are potentially serious.
Table 2 Hydraulic conductivity data for the Carboniferous Limestone sub-divided
into different lithological groups
Group
Blackrock
Clifton Down
Hotwells
Vallis
Mean
(m/d)
0.97
0.123
0.102
0.025
Coefficient
of variation
(%)
9330
120
88.4
43.8
Total Number
of Boreholes
22
7
7
9
Effect of Scale
Kiraly (1975) has suggested that in karstified aquifers there is a systematic relationship
between the effective transmissivity of the aquifer and the scale considered. Thus at the
catchment scale conduits dominate transmission, which at the other extreme in core samples
only intergrannular voids are present, and measured hydraulic conductivities are very low.
Slug tests involve displacement of only a limited volume of water, and thus 'sample' only a
relatively small zone of the aquifer within 1 to 2 m of the test borehole (depending on slug
size and aquifer hydraulic conductivity). In contrast pump tests may cause more extensive
drawdown over distances from tens to hundreds of metres from the test borehole. Figure 9
contrasts our slug test results with pump test data for Grand Bahama obtained by Little et
al (1975). Over 60% of the pump test results have higher hydraulic conductivities than the
maximum recorded in the slug tests, but the two data sets give similar minimum values.
This suggests that in pump tests the cone of depression expands to intersect dissolutional
conduits, whose spacing is sufficiently wide that they are not penetrated directly by
random boreholes. It should also be born in mind that at high hydraulic conductivities
such as these turbulent flow may develop and the transmissivities determined may be
erroneous.
382
-------
100%
75%
50%
25%
0%
10-3 10-2 10-1 10 0 10 1 10 2 10 3
Hydraulic Conductivity (m/d)
Borehole Location
•• Outside Quarry I I Within Quarry
I I Quarry Sub-Floor
Figure 8 Comparison of hydraulic conductivities for boreholes within quarries, in the quarry
sub-floor, and greater than 150 m from quarries,Carboniferous Limestone aquifer.
Slug and
Bailer Tests
10'
99
1 10 30 50 70 90
Cumulative probability (%)
Figure 9 Comparison of hydraulic conductivity data for the Lucayan Limestone aquifer,
Grand Bahama derived from pump and slug tests.
383
-------
Conclusions
In our study we have made use of small diameter observation boreholes which do not have
pumps installed, and are therefore readily tested by slug and bailer methods. In other areas
where large numbers of domestic abstraction wells fitted with pumps are available, use
could be made of borehole specific capacity data (or transmissivity derived from this data)
see for instance (Sallany and Walton, 1963). In both cases log probability plots offer an
easy and effective method for quantification of the position of the aquifer on the flow
spectrum from difuse to conduit within the conceptual model of Smart and hobbs (1986).
Different carbonate aquifers can be readily compared and contrasted, and insight gained on
the flow properties. Hydraulic conductivity in carbonate aquifers is controlled by void size
and void integration. In the more densely fractured aquifers, void integration is effectively
complete, giving a relatively homogeneous aquifer, and the density of fissuration and
degree of dissolution enlargement are the main factors controlling mean hydraulic
conductivity. In less densely fractured aquifers, void integration becomes much more
important, and where dissolutional development is not extensive, the aquifer is much more
heterogeneous, especially where flow concentration into conduits has occured. Lithology
and structure may be important in controlling aquifer transmissivity, but disposition of the
unit within the aquifer may be equally important where dissolutional void enlargement is
important. Thus, the approach to aquifer characterisation exemplified in this paper has
two important applications. On a practical basis it provides a means of describing the
transmission properties of carbonate aquifers, providing data in a format suitable for
stochastic parameterisation of aquifer models. It also provides a simple design tool useful
for prediction of borehole yields, and the preferred mode of aquifer development (Smart
and Hobbs, 1986). On a more academic level it provides data to elucidate the controls on
transmission and structure of carbonate aquifers. This also has implications for aquifer
management, for instance transmission of pollutants from waste disposal sites via the
diffuse flow zone in karstified limestones (Edwards and Smart, 1989) and the validity of
geotechnical measurements of aquifer parameters.
Acknowledgements
The authors would like to thank the following for access to, and in some cases drilling,
boreholes: ARC Southern Ltd, Foster Yoeman Ltd, Wimpey Hobbs Ltd, ECC Quarries Ltd,
National River Authority; Wessex Region, Wessex Water Pic., the Government of the
Bahamas, the Bahamian Ministry of Works and Utilities and the Freeport Water Company.
The authors would also like to thank the British Geological Survey for the loan of the
geophysical calliper log equipment. A.J. Edwards and S.L. Hobbs were supported by
NERC training awards in collaboration with ARC Southern Ltd, and Bristol Waterworks
Company and Wessex Water Authority respectively. The Bahamas fieldwork was supported
by Amoco UK Ltd.
References Cited
Aley, T. 1975. A predictive hydrological model for evaluating the effects of land use and
management on the quantity and quality of water from Ozark Springs. Ozark
underground Laboratory, Springfield Missourri.
384
-------
Atkinson, T.C. 1977. Diffuse flow and conduit flow in limestone terrain in the Mendip
Hills, Somerset. Journal of Hydrology, 35,93-110.
Atkinson, T.C. 1985. Present and future directions in karst hydrogeology. Annales de la
Societe Geologique de Belgique, 108, 293-296.
Atkinson, T.C., Smith, D.I., Lavis, J.J. and Whitaker, R.J. 1973. Experiments in tracing
underground water in limestones. Journal of Hydrology, 19, 323-349.
Beach, D.K. and Ginsberg, R.N. 1980. Facies and succession of Pliocene-Pleistocene
carbonates, Northwestern Great Bahama Bank. American Association of Petroleum
Geologists Bulletin, 64, 1634-1641.
Bouwer, H.F. and Rice, R.C. 1976. A slug test for determining hydraulic conductivity of
unconfined aquifers with completely or partially penetrating wells. Water Resources
Research, 12, 423-428.
Csallany, S. and Walton, W.C. 1963. Yields of Shallow Dolomite Wells in Northern Illinois.
Illinois State Water Survey, 1-43.
Dao Xian, Y. 1986. On the heterogeneity of karst water. International Association of
Hydrological Sciences Publication, 161, 281-292.
Edwards, A.J., Hobbs, S.L. and Smart, P.L. 1991. Effect of quarry dewatering on a
limestone aquifer: case study from the Mendip Hills, England. Proceedings of the
Third Conference on Hydrology, Ecology, Monitoring and Management of Ground
Water in Karst Terraines, (in press).
Edwards, A.J. and Smart, P.L. 1989. Waste Disposal on karstified Carboniferous Limestone
aquifers of England and Wales. Proceedings of the 3rd Multidisciplinary
Conference on Sinkholes and Engineering and Environmental Impacts of Karsts.
Balkema, Rotterdam, 165-182.
Ewers, R.O. 1978. A model for the development of broad scale networks of groundwater
flow in steeply dipping carbonate aquifers. Transactions of the British Cave
Research Association, 5, 121-125.
Ferris, J.G. and Knowles, D.B. 1963. The slug-injection test for estimating the coefficient
of Transmissivity of an aquifer. United States Geological Survey Water Supply
Paper, 1536-1, 299-304.
Foster, S.S. and Milton, V.A. 1974. The permeability and storage of an unconfined chalk
aquifer. Hydrological Sciences Bulletin, XIX, 485-500.
Green, G.W. and Welch, F.B.A. 1965. Geology of the Country around Wells and Cheddar.
H.M.S.O., London, pp. 225.
Gunn, J. and Gagen, P., 1987. Limestone quarrying and sinkhole development in the
English Peak district. Proceedings of the 2nd Multidisciplinary Conference on
Sinkholes and Engineering and Environmental Impacts of Karst. Balkema,
Rotterdam, 121-125.
385
-------
Hobbs, S.L. and Smart, P.L., 1988. Heterogeneity in carbonate aquifers: a case study from
the Mendip Hills, England. Proceedings of the Internation Association of
Hydrogeologist 21st Conference. Geological Publishing House, Beijing, 724-300.
Holmberg, R. and Maki, K., 1981. Case examples of blasting damage and its influence on
slope stability. Proceedings of the 3rd International Conference on Stability in
Surface Mining. Vancouver, British Columbia, 773-793.
Karasaki, K., Long, J.C.S. and Witherspoon, P.A. 1988. Analytical models of slug tests.
Water Resources Research, 24, 115-126.
Kellaway, G.A. and Welch, F.B.A. 1980. British Regional Geology, Bristol and Gloucester
District (Second Edition). H.M.S.O., London.
Keys, W.S. and Maccary, L.M. 1971. Applications of borehole geophysics to water
resources investigations. Techniques of Water-Resources Investigations of the United
States Geological Survey, chapter El, 1-26.
Kiraly, L. 1975. Rapport sur e'etat actuel des connaissanies dans le domaine des
characteres physiques des roches karstique. In Bwrger, A. and Duberet, L. (eds)
Hydrology of Karstic Terrains, International Union of Geological Sciences, Series B,
3, 53-67.
Little, B.C., Buckley, O.K., Cant, R., Jefferies, A., Stark, J. and Young, R.N. 1975. Land
Resources of the Commonwealth of the Bahamas, Grand Bahama. Overseas
Development Authority, Surbiton, Surrey, pp. 87.
Mylroie, J.E. and Carew, J.L. 1990. The flank margin model for dissolution cave
development in carbonate platforms. Earth Surface Processes and Land forms, 15,
413-424.
Palmer, A.N. 1984. Geomorphic interpretation of karst features. In La Fleur, R.G. (ed).
Groundwater as a Geomorphic Agent. George Allen and Unwin, New York, 173-
209.
Quinlan, J.F. 1988. Special Problems of groundwater monitoring in karst terraines.
Symposium on Standards Development for Groundwater and Vadose Zone
Monitoring Investigations. American Society for Testing and Materials,
Philadelphia, 1-37.
Sendlein, L.V.A. and Palmquist, R.C. 1977. Strategic placement of waste disposal sites in
karst regions. In Tolson, J.S. and Doyle, F.L., (ed). Karst Hydrogeology.
University of Alabama, Huntsville, 323-336.
Schuster, E.T. and White, W.B. 1971. Seasonal fluctuations in the chemistry of limestone
springs: a possible means for characterising carbonate aquifers. Journal of
Hydrology, 14, 93-128.
Smart, P.L. and Hobbs, S.L. 1986. Characterisation of carbonate aquifers: a conceptual
base. Proceedings of the Environmental Problems in Karst Terrains and their
Solution. National Water Well Association, Dublin, Ohio, 1-14.
386
-------
Smart, P.L. and Whitaker, F.F. 1988. Controls on the rate and distribution of carbonate
bed-rock dissolution in the Bahamas. Proceedings of the 4th Symposium on the
Geology of the Bahamas. Bahamian Field Station, Fort Lauderdale, 313-321.
Smart, P.L. 1976. Catchment delimination in karst areas by the use of quantitative tracer
method. Proceedings of the 3rd International Symposium of Underground Water
Tracing. Institute of Karst Research, Postojna, 291-298.
Van der Kamp, G. 1976. Determining aquifer transmissivity by means of well response
tests. Water Resources Research, 12, 71-77.
White, W.B. 1977. Conceptual models for carbonate aquifers: Revisited. In Dilamarter,
R.R. and Csallany, S.C., Hydrologic Problems in Karst Regions. Western Kentucky
University Press, Kentucky, 176-187.
387
-------
HETEROGENEITY IN CARBONATE AQUIFERS: EFFECTS OF SCALE, FISSURATION,
LITHOLOGY, AND KARSTIFICATION
By: Peter L. Smart, Alan J. Edwards, and Steve L. Hobbs
Q. Relative to the use of caliper logs to characterize the degree
of fissuration, how was fracture aperture determined? What
about single solutionally enlarged fractures? The caliper log
would seem to be an indicator of borehole instability and be a
measure of factors other than single fracture density.
A. Borehole caliper logs are controlled by the degree of aquifer
fissuration as well as the drilling method and borehole
instability. The percentage fissuration index described in
this paper is a crude method of measurement. However, it
offers a pragmatic and simple technique which we believe
generates interesting and useful results.
Q. Did you use different methods of slug-testing? Did you observe
any differences between the hydraulic conductivities determined
with various slug-test methods and a bailer? If so, how do you
account for any observed differences?
A. In all slug and bailer tests, weighted floats were inserted and
removed from the water within the tested borehole. This
technique was utilized in all the tests. Hydraulic conductivi-
ties determined from the slug rather than the bailer test were
generally higher. This was because the rise in water level,
due to float insertion, resulted in an immediate water loss to
the surrounding unsaturated rock. In order to overcome this
disparity, the average hydraulic conductivity from 6 slug and
bailer tests was calculated for each tested borehole.
388
-------
CAUSTIC WASTE CONTAMINATION OF KARSTIC LIMESTONE AQUIFERS
IN TWO AREAS OF JAMAICA
BASIL P. FERNANDEZ
UNDERGROUND WATER AUTHORITY
HOPE GARDENS, KINGSTON, JAMAICA
Abstract
Jamaica's bauxite deposits occur in association with the
karstified White Limestone Formation of Tertiary Age. The
refining of bauxite to alumina results in a waste known
locally as "red mud", which is stored in unsealed mined out
karstic depressions. The red mud, which is more than 70% water
is highly caustic and infiltrates to the groundwater table.
Groundwater contaminated by "red mud", shows increased sodium
concentration and increased pH. Monitoring of groundwater
quality in two areas of the island, where bauxite/alumina plants
are located, indicate significant contamination over the last
20 years. In these two areas, caustic contamination has resul-
ted in over 20 km of aquifer being unsuited for groundwater
development. The mud disposal ponds are located in the direct
path of groundwater flow and still poses a serious threat to
groundwater reservoirs and consequently the groundwater reserves
of the island. Remedial measures to reduce contamination have
included sealing of disposal ponds, thickening of mud and solar
drying and recycling of liquid fraction.
Introduction
A highly caustic waste known locally as "red mud" is a by-product
of the extraction of alumina from Jamaican bauxite, which occurs
as an interfingered blanket deposit in association with the
limestone deposits forming the exploitable aquifer of the island.
Four plants located on the south coast of the island, produce
approximately 3Mm of red mud waste annually and an increase to
4.5Mm is expected with expansion of the present plants and
construction of one new plant. The red mud waste is stored in
mined out or topographic depressions in the limestone.
389
-------
TABLE 1 CROUNDWATER RESOURCES AVAILABILITY
M»3/Yr Hied Percent of
Mm /Yr MIgd Total
GROUNDWATER SAFE YIELD 3418 2062
FROM LIMESTONE AQUIFERS 3294 1987 96
FROM ALLUVIUM AQUIFERS 124 75 4
PRODUCTION (WELLS & SPRINGS) 839 506
FROM LIMESTONE AQUIFERS '06 426 84
FROM ALLUVIUM AQUIFERS 133 80 16
UNUTILIZED RESOURCES
2579 1555 75
Limestone rocks in Jamaica have a high risk factor for pollution
because of the development of relatively large channels for
groundwater infiltration and percolation (karstification). The
limestone aquifers in Jamaica are a major source of water supply
because of the high transmissivities and storage especially
along fault and fracture lines and in areas where karstification
has produced a network of solution openings. Limestone rocks
underlie 67% of the island (figure 1) and important groundwater
reservoirs are to be found in the limestone close to major urban
and agricultural areas. There are more than 400 wells tapping
the limestone aquifer and abstracting approximately 706 Mm/ yr
of water or 8470 of the total water produced (Table 1).
Th& disposal sites (figure 1) are directly in the flow path of
groundwater to major reservoir areas. No structural or litho-
logic barriers exist in the limestone and changes in groundwater
abstraction patterns could induce a more wider.spread of the
contaminant. At sites 1 and 4 (figure 1) caustic contamination
of groundwater has been detected. Several springs and wells
have been abandoned and a significant portion of the limestone
aquifer adjoining these sites is unsuitable for further ground-
water development.
Geology/Hydrogeology of the Limestone
The White Limestone Formation is a thick series of pure calcium
carbonate ranging in age from middle Eocene to Miocene (47M.'yrs -
10M yrs). The entire sequence is estimated to be more than
3,000 meters thick and has been subdivided into ten (10) bio-
stratigraphic and four (4) hydro-stratigraphic units.
Hydrogeologic properties of the entire series are broadly similar
and there is hydraulic continuity between all the units.
Extensive tectonism has resulted in folding, faulting and frac-
turing of the rocks.
390
-------
Figure
SIMPLIFIED GEOLOGY OF JAMAICA
QroundwQtar Flow Direction
Faults
FTOURE 2
HYDRO-STRATIGRAPHIC MAP OF JAMAICA
E II 0
HTtmO-STHATOOIWHIC WITS
l::-:l Alluviim AqiHftr/AquteMt
E3 Cental AquioMl/Ai|
-------
In the lowlands, solution of the limestone has resulted in an
integrated network of solution openings, imparting to the
aquifer almost homogenous and isotopic properties. In the
lowlands conduits which originate in the uplands may continue
as cave systems or subterranean streams. Most of the recharge
of the limestone aquifer takes place in the uplands, where soil
cover is absent and rainfall is the highest, while storage is
in that portion of the aquifer underlying the coastal plains.
Lateral flow of groundwater takes place both in the vadose and
phreatic zones. In the vadose zone, lateral flow is rapid and
confined to narrow channels which may continue into the
phreatic zone. Most of the water in the phreatic zone moves
through the interconnected network of solution openings and
fracturejoints.
.Limestone Hydrostratigraphic Units
The limestone is subdivided in four (4) hydrostratigraphic
units (figure 2) viz:
Limestone Aquifer
Limestone Aquiclude
Coastal Aquifer
Coastal Aquiclude
The limestone aquifer is comprised of members of the White
Limestone Group. They form a sequence of moderately compacted,
well bedded, partially crystallised bioclastic and micritic
limestones with thickness in excess of 2,000 metres. The
aquifer exhibits mature karstic features, typified by a very
high infiltration capacity, predominant sub-surface drainage
and highly compartmentalised sub-surface conduit flow.
Its karstic nature makes the limestone aquifer very suscep-
tible to contamination.
The limestone aquiclude is a soft, fine grained chalk of low
permeability which fringes the coast ponding groundwater
within the juxta posiitoned limestone aquifer. Thickness
approaches 1000 metres.
The coastal aquifers are raised reefs patchily deposited along
the north coast of the island. These reefs form highly karsti-
fied limestone aquifers of high permeability and low groundwater
storage potential. Thickness is usually less than 50 meters.
The coastal aquicludes are soft marls fringing the coast of the
island. It dams groundwater flow which occurs along faults
through the limestone aquiclude such that springs issue along
the limestone/coastal aquicludes boundary. Thickness approaches
300 me t e r s.
392
-------
TABLE 2
RED KUD INSOLUBLE SOLIDS
TABLE3 RED MUD SOLUBLE SOLIDS
CONSTITUENT
L10
SlOj
AI2°3
F'2°3
P2°5
CiO
N,20
Tt°2
MnOj
Miscellaneous
PERCENT
11.0
5. 5
12.0
49. 5
2.0
8.0
3 . 5
5.0
1 .0
1 .5
CONSTITUENT
-AMOUNT
A12°3
NaOH
Nacl
So ec . Gravity
BOD
COD
2.5g/kg liquid
3.'g/kg
1 -6g/kg
0. 7e/ko
0.1c/kc
1 .008
10.5 units
6 pptn
1<*8 ppm
ANALYSES Of HID HUD LIQUID riACTIOH AND
•tCOKHENDCD COKCEHTR AT10H LIMITS
Cenit ltu«nt
C.Utu.
Hl|i«llu.
Sodium
tot • I • 1 ua
Cirbonctf
Sulch.t.
Chlor ld<
Ktrdncil •• C«CO.
AU.Ilnll, .1 C.COj
°H
Turbidity
Colour
a s.,,u.
•* V«t«T Quality Crlt«rlA
Fro* W.d50
10600 " 500
- 500 1300 " 700 2100 •• 100
12.6- 69.] 6.59 3.39.1
unit* 30 unit!
3 30" 1200 until
601
by Hek.«
-------
Its potential for contamination of groundwater lies in the
concentration of sodium and hydroxide ions; the presence of
iron oxide; and the organic substances which on decomposition
impart an unpleasant smell to the waste. Tables 2 and 3 show
respectively the chemical analyses of insoluble and soluble
solids of Jamaica red mud.
•For domestic, industrial and agricultural water, colour, taste,-
smell, high soda content and high pH makes the "red mud" a
potential agent for degrading groundwater quality. Table 4
shows an analysis of the red mud liquid fraction plus the
recommended concentration limits.
Contamination Criteria
Analyses to detect above average concentrations of the chemical
constituents present plus esthetic indices such as colour,
taste and smell determine the degree of contamination.
Five indices were specifically used to detect contamination.
(1) Sodium to chloride concentration ratio exceeding the
maximum ratio encountered in uncontaminated groundwater
in Jamaica of 1.5
(2) High sodium content. This alone is not a precise
indicator as sodium chloride waters are found in the
limestone aquifers as a result of saline intrusion.
However, in this form of contamination high sodium
concentrations are associated with high chloride concen-
trations, not the case with caustic contamination.
(3) Sodium to Calcium concentration ratio in excess of the
ratios generally encountered in uncontaminated ground-
wat e r of 1.0.
(4) High pH values in excess of 8.0 units, the maximum
encountered in uncontaminated groundwater.
(5) The presence of suspended solids, red discoloration,
poor smell and unpleasant taste.
Disposal Methods
The two plants, around which contamination has been detected,
have similar disposal methods. The red mud waste is piped
to and ponded in a natural karstic depression from where the
bauxite had been previously mined. The two sites are located
atop highly karstified limestone with numerous solution
openings. No special precautions were taken to prevent or
reduce infiltration before storage of waste began. Infiltra-
tion was seen as one way to extend the life of the disposal
pond. No proper geological, hydrogeologica1 and engineering
studies were done before the selection of disposal site was
made .
394
-------
Figure 3: Diagrammatic Section through a Red Mud Pond1
OUTFAU. UUO — DELTA EVAPORATION PRECIPITATION RUNOFF
i^CAUSTC LIQUID v^V;:;^^/;
FRACTURED LIMESTONE
Movement of Contaminant into Aquifer
A diagramatic section through a red mud pond is shown as
Figure 3. The pond is formed by a sloping, elongated, steep
sided karst depression with a narrow outlet at the downstream
end. Storage is improved by the construction of a rock fill
dam across the outlet with increases in dam height as necessary.
Limestone is exposed along the surface and at the base of the
pond. The depression is naturally drained by sinkholes or
other solution openings. Lateral and vertical percolation takes
place through the network of fractures and solution openings
typical of the karstified limestone. Leakage also occurs through
the rockfill dams. With time, the accumulation of solid material
will form a seal over the limestone, reducing the percolation
rate. Calculations indicate that only 30 - 40% of the waste is
retained in the pond.
The formation of vertices, rapid fall of mud level and overnight
disappearance of mud and water, indicate rapid and high infiltra-
tion losses. The formation of mud deltas at the outfall pushes
the liquid faction towards the rockfill dam where the capacity
for leakage is high. Evaporation from the pond surface concen-
trates the pollutants in the waste. The ponds act as collectors
of rainfall runoff, mixing of this water with the waste and then
infiltrating into the limestone.
395
-------
Contaminated Areas
Of the four sites shown on figure 1, three have been constructed
atop karstic limestone and one atop alluvium. Of the three
sites atop karstic limestone, contamination has been detected
at two sites - site 1 and 4. Site 1 is associated with the
Ailcan Jamaica Company (ALJAM) Ewarton Plant while site 4 is
associated with Alumina Partners (ALPART) Nain Plant.
Site 1 - ALJAM Mt. Rosser Mud Pond
This site has the largest mud disposal pond in the island with
13.17 Mm of waste in storage up to September 1991, and a
surface area of 40 hectares. The pond occupies a large elonga-
ted karst depression approximately 6.5 km north of the plant
(figure 4). This pond known as Mt. Rosser or Schwa 1 lenburgh,
had no special preparation before mud storage began. At the
lower eastern end a rockfill dam of limestone blocks was
constructed. Massive material losses were recorded during the
early disposal period. However, it was felt that the impact on
the surrounding environment would not be significant. There was
also the expectation that with increasing submergence of the
pond floor, the mud would form a sufficiently good seal against
further effluent loss. This has not been the case and effluent
losses have continued unabated. At first, attempts were made
to locate and seal all the sinkholes and fractures with clay to
stop the leakages. Attempts have have also been made to create
beaches of tailings around the face of the dam to keep the
liquid faction from seeping through but this was also unsuccess-
ful in reducing effluent loss.
The Mt . Rosser Pond is located on a faulted anticline, the
northern limb of which continues into the Moneague Sub-basin
(a syncline) and the northern limb continues into the Linstead
Sub-basin (also a syncline). The latter merges with a faulted
anticline at Bog Walk, where impervious strata are brought to
the surface and forms a barrier to groundwater flow.
The Moneague syncline forms a large depression into which ground-
water discharges through springs. This depression has no surface
water outlet and the spring discharge enters the aquifer via
sinkholes and flows northwards to the headwaters of the rivers
on the north coast of the island. Caustic contamination of
groundwater has been detected at a number of springs and wells
to the north and south of Mt. Rosser Pond. To the north the
Rio Hoe Spring and the Walkers Wood well show evidence of caustic
contamination.
The Rio ^Hoe ".Spring rises at 'the' sou-thern edge of the- Moneague
Sub-basin, flows for 2km and enters the Moneague Lake. Water
from the Moneague Lake enters the aquifer via the Walton
Sink and flows to the Walkers Wood well. Figure 5 shows the
396
-------
plot of sodium concentration
for the period 1973-1991 .for
Rio Hoe. The contamination
was first detected in 1971
and rose progressively to
peak at over 350 mg/1 sodium
in 1983, and the spring which
was a source of domestic
water has been declared unfit
for human consumption. The
Moneague Lake fed by the Rio
Hoe Spring is also contamina-
ted with maximum sodium
concentration of 65 mg/1
(Mar. 1990).
The contamination in the
Walkers Wood well was detec-
ted in 1977. The well was
drilled in 1976 and the
sodium concentration during
the pump test was 5mg/l.
The contamination has
increased and monthly monito-
ring, which began in 1990,
shows a worsening situation,
(figure 6).
The elevation of the surface
of the red mud waste is at
an ;altitude of 466.3 metres
while the spring rises at
305 meters and there seems
t o .i>e continuity between the
pond and the spring by a
system of conduits above
the regional water table.
The spring is above the
regional water table and
during high rainfall the
discharge becomes brown and
turbid, while the water table
does not exhibit this change.
The area is now being assessed and so far 5 monitoring wells vary-
ing in depth from 200-350 metres have been completed. To date,
only 3 of the monitoring wells have tapped contaminated ground-
water. The probable area of contamination is shown on figure 4.
To the south the Weatherly Spring and Alcan's Deepwell 2 have
shown evidence of caustic contamination. The plots of sodium
concentration for Weatherly Spring and Deepwell 2 for 1973-1991
are shown as figures 7 and 8 respectively.
397
-------
Figure 5 : SODIUM CONCENTRATION RIO HOE SPRING 1973 - 1991
I
I9B2 1963 196* 1985 1985
1973 197+ 1975 1976 1977 1978 1979
1937 1963 1969 1990 1931
Figure 6' SODIUM CONCENTRATION WALKER'S WOOD WELL 1976-1991
1976 1977 197B 1979 1980 1931 1982 1933 19B<
19E5 1966 1987 1983 1939 1990 1991
In 1989/90 the Underground Water Authority and ALJAM collaborated
on a project to assess the level of contamination. Six monitoring
wells were drilled and average sodium concentration varied from
9-30 mg/1. The results of the project indicate:
(i) the area of contamination has been better defined
and has been found to be smaller than originally
env i s ioned.
(2) two zones of contaminated groundwater separated by
clean water were intercepted and confirmed the more
significant role that compartmentalised flow plays in
the movement of the pollutant than diffuse flow.
(3) The upper contaminated zone has been interpreted as being
the direct result of the plant, its attendant dumps and
other adjacent effluent generating facilities.
398
-------
Figure 7 : SODIUM CONCENTRATION WEATHERLY SPRING 1973 - 1991
3000
2700
2«OC
21OQ
1800
1500.
1200.
900.
600
300.
1980 1961 1982 1983 1964 1985 1986
1973 1974 1975 1976 1977 1978 1979
1967 1988 1969 1990 1991
Figure 8 : SODIUM CONCENTRATION DEEPWELL
1973 - 1991
10QQO
9000
BOOO
7000
6000
5000
«00
33a)
2000
1003
1973 1974 1975 1976 1977 1978 1979
19BD 19B1 1932 1933 1934 1935 1935
Monfe
1937 I SEE 1933 1993 1991
To the north and south the total area of the aquifers contaminated
is approximately 18 km^ (figure 4).
Site 4 Alpart -Nain Mud Pond
There.are two (2) disposal ponds at this site, a south mud pond and
a north clear effluent pond (figure 9) . The ponds are located on
a thick sequence (>1500m) of faulted and fractured brecciated
limestone. The presence of numerous sinkholes and the absence of
surface drainage indicate the potential for rapid infiltration.
Thesites are located close to the groundwater divide which separates
northerly flow to the Upper Morass from southerly flow to the sea.
The surface area of the ponds are 90 hectares and the volume of
mud in storage is estimated at 12.8 Mm3. Red mud disposal was
being restricted to the south lake but caustic enriched clear
effluent is being disposed of in the north lake. The formation of
tailings "deltas" were being restricted to the northern and eastern
perimeters of the south lake with the resultant ponding of superna-
tant liquor directly against the limestone along the southern and
399
-------
flgur. «i AREA CONTAMINATED
BY CAUSTIC WASTE FROM
ALPART ALUHIHA
PLANT
western lake perimeter.
The plant began operations in 1969 and
five wells were drilled in the vicinity
of the plant to supply industrial and
domestic water. As early as March 1970
high sodium concentrations were noted
in the well water. This was followed
by a decrease in well yields and the
drilling of a replacement well (#6) to
satisfy demands. The sodium concentra-
tions increased until all the wells
around the plant are now contaminated.
Domestic water now comes from a new well
field established 6.4 km north of
the plant at Pepper. The Pepper
well field began operations in
1974 and has since been showing a
trend to increasing sodium concen-
tration (figures 10, 11, 12, 13).
To the south a 1990/91 investiga-
tion has indicated a low level
of contamination of domestic wells
at New Forest. This contamination
is masked by saline intrusion into
the aquifer but once the molar
ratio between sodium and chloride
have been reconciled, excess
sodium remains in the groundwater
and its source can only be the mud
pond. The area of the aquifer
contaminated is approximately 20 km2
and the domestic wells that supply
the large urban centre of Mandeville
are now threatened.
Reduction of Contamination
The Underground Water Authority has
been working with both Alpart and
ALJAM to reduce groundwater conta-
mination.
system called "mud stacking
consists of thickening the
thinly on a sloping, sealed
sun. All run-off is collec
to a sealed holding pond, f
Once the mud has dried on t
can be cleaned and returned
Company to dispose of less
in the Mt. Rosser red mud p
At ALJAM a new mud
and drying" has been imp
red mud to 22% solids, sp
drying bed, where it is
ted at the toe of the bed
rom where it is recycled
he bed (it does not reslu
to use. This has allowed
than 5% of the annual was
ond, and only in the very
disposal
1emented. This
raying it
dried by the
and transferred
into the plant.
rry) the bed
the Bauxite
te generated
high rainfall
400
-------
Figure 10 : SODIUM CONCENTRATION ALPART I WELL 1975 - 1991
1000
900
eco
TOO
sen
ra
200
ICO
1975
1976
1977
1978
1979
1330
199
1EEB
19B3
196*
19B5
1S6B
1987
19BB
1969
1990
1991
Figure II: SODIUM CONCENTRATION PEPPER I WELL 1975-1991
9Q
«
40
s
a
z
XI
IS
Illlllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll
1975 1976 1977 1978 1979 I960 199
1S62
1S83
1934-
1985
1985
1987
1983
1980
1931
Figure 12 : ANNUAL MAXIMUM Na/CI
RATIO ALPART I 1975-1991
2000
1500
1000
500
003
i f I I I I I I I I I I I I I I I
1975 1977 1979' 193 19S5 19E6 1967 1989 1991
1976 1978 I960 1932 19B4 1SBB 19B3 1SGO
Years
Figure 13: ANNUAL MAXIMUM Na/CI
RATIO PEPPER I 1975-1991
I
250
200
150
oso
oco
I l l l 1 l l l 1 r i I r r i I I
1975 1977 1979 1981 1EG3 1SB5 19B7 1S6B 1991
1976 1978 'Sffl 1962 19B* 19B6 1SGB 1980
ran
401
-------
periods when the drying cycle becomes longer. Alpart is now
investigating the suitability of this method of disposal for
implementation at its plant site.
Summary and Conclusions
The red mud wastes produced from .the bauxite alumina operations
contain sufficiently high concentrations of pollutants to
contaminate the groundwater, making it unusable. The method
of disposal of the waste utilizing natural or mined out
depressions in the limestone creates a potential hazard in view
of the high degree of karstification of the aquifer.
Contamination of the groundwater in the vicinity of.:'the two
sites have been proven. The solutions to reducing the contami-
nation lie in the sealing of the disposal pond and possible
total impoundment of the effluent. Pond management needs to
be improved to minimize the possibilities and effects of further
contaminat ion.
Acknowledgements
The Author wishes to thank the following persons/agencies who
assisted in the preparation of this paper.
(UWA)
(ALJAM)
(ALPART)
( JBI)
Mr- Dexter Lewis who assisted with the preparation of figures
Miss Michelle Wint who compiled data and
Mrs. Nerma Wynter who typed the paper-
Underground Water Authority
Alcan Jamaica Company
Alumina Partners Limited
Jamaica Bauxite Institute
References
1 .
2.
3 .
: ion
ALJAM/UWA - Preliminary report on Groundwater Pollute..
by caustic effluent from the Ewarton Plant and associated
facilities - Dec. 1989. A joint project report.
FERNANDEZ, BASIL: Caustic waste pollution of groundwater
in Jamaica. Two solutions - May 1983. An unpublished
Underground Water Authority Report.
Underground Water Authority: Water Resources Development
Master Plan Report 1 - Water Resources Inventory -
Dec. 1985. An unpublished Underground Water Authority
Report .
402
-------
Question
Are there any springs near the plants that are used
as water supplies? If so, are they hydraulically
connected to the plant?
Answer
Yes, there are 4 springs located near to the plant that
can be used for water supplies. One spring is located
to the north, and 3 are located to the south of the
pi ant (see fi gure 4).
Of the 4 springs, 3 are hydraulically linked to the mud
disposal pond, and show signs of contamination.
These are:
(a) North - Rio Hoe Spring
(b) South - Weatherly Spring
Cashew Tree Spring
The fourth spring is not hydraulically linked, and is
not contaminated. It is used as a source of domestic
water for a small village in the hills above the plant.
403
-------
404
-------
The Interaction of Flow Mechanics and Aqueous Chemistry
in a Texas Hill Country Grotto
Barbara Mahler and Phillip Bennett
Department of Geological Sciences, University of Texas at Austin
Austin, Texas
Abstract
The physical characteristics of flow, or flow mechanics, through a carbonate terrane
affect the chemical signatures of groundwater inputs to Hamilton Pool, a grotto in the
Central Texas Hill Country portion of the Trinity Aquifer. Because flow mechanics
influence geochemical processes, their identification is necessary to protect water quality at
this unique site, which is used for both recreation and habitat preservation.
During baseflow, two groundwater sources supply all the water in the pool: 1) small
perennial springs half a kilometer upstream from the pool are the headwaters for Hamilton
Creek, which enters the pool via a 25-meter waterfall (0.1-0.3 ftVs), and 2) stalactites
hanging from the ceiling of a large overhanging limestone ledge drip water directly into
Hamilton Pool (0.05 ft-Vs). The two sources exhibit differences in pH, calcium
concentration, Ca/Mg ratio, and temporal variation in chemistry. The chemistry of the drip
water is affected by degasing of carbon dioxide and equilibration with ambient temperature
upon contact with the atmosphere. Backmodeling of these environmental processes with
PHREEQE resulted in a subsurface chemistry of the drip water very similar to that of the
spring water. Analysis of conductivity variation in response to precipitation revealed that
the chemistry of the upstream springs is controlled by conduit flow, while the ceiling water
is controlled by diffuse flow.
These results suggest that a) both groundwater inputs to Hamilton Pool are from the
same recharge source, and b) the conduit-controlled upper springs render Hamilton Pool
vulnerable to contamination should the catchment area be urbanized.
405
-------
Introduction
Carbonate aquifers are distinguished from other groundwater media by their physical
flow characteristics. Flow rates at different points within a single carbonate aquifer can
vary many orders of magnitude, from centimeters per day to kilometers per day. The
mechanics controlling flow rate influence both the chemistry of the water and the
susceptibility of the water to contamination. This paper highlights the findings of a more
extensive study of geochemical response to flow mechanism in a Texas Hill Country grotto
(Mahler, 1991).
Diffuse flow and conduit flow represent opposite extremes of carbonate flow. Diffuse
flow is similar to flow through a granular, porous media; water flows through extremely
tiny joints, partings, bedding planes, and fractures, the diameters of which are on the scale
of millimeters or less. Conduit flow, in contrast, involves flow through integrated,
solutionally-developed conduit systems with diameters on the scale of centimeters to meters
(e.g., Shuster and White, 1971).
The risk of contamination of conduit-controlled water sources is high. Conduits
behave as pipes rather than as filters; water moves through them with high velocities and
short residence times, and without undergoing dilution. The processes that normally
cleanse groundwater (e.g. filtration by soils, adsorption of contaminants by mineral grains,
dilution and dispersion through large volumes of water, and decay and die-off during long
residence times) often do not affect conduit flow.
Springs controlled by the two different flow mechanisms can be distinguished by their
chemistry. Shuster and White (1971) found that diffuse flow springs showed much less
variation in hardness than conduit flow springs. Similarly, Jacobson and Langmuir (1974)
found that diffuse flow springs showed less variation in solute amounts than conduit flow
springs. They also found that after a storm, conductance remained the same or increased
slightly for diffuse flow springs due to salts and solutes swept in from soil, while
conductance decreased in conduit flow springs due to dilution with low conductivity
meteoric water. They therefore suggested using specific conductance as an indicator
variable of subsurface flow mechanism.
The objective of the investigation presented here is to characterize the aqueous
geochemistry of inflows to Hamilton Pool, Travis County, Texas, to determine the flow
mechanism controlling each source, and to investigate each inflow's genetic relationship to
the other baseflow sources.
The Study Site. Hamilton Pool is a pristine waterfall-fed grotto in rural western
Travis County, Texas, approximately 20 miles southwest of the city of Austin. The pool is
the focal point of Hamilton Pool Preserve, a county-operated facility. Water quality at the
pool is utmost important both for recreational use and species habitat. Although current
406
-------
land-use in the Hamilton Creek watershed is primarily agricultural, commercial and
residential development are rapidly expanding westward from the Austin area.
The Hamilton Creek watershed is located in the rocks of the Lower Cretaceous Trinity
Division. Five outcropping formations of the Trinity Division can be viewed at Hamilton
Pool Preserve. Sycamore Sand outcrops in the bed of Hamilton Creek from the Pedernales
River to approximately one kilometer upstream, where it slopes into Hammett Shale. The
shale forms the eroded walls of the cliff surrounding the east side of Hamilton Pool, and
Cow Creek Limestone forms the overhanging ledge. The rolling grasslands surrounding
Hamilton Pool are on Hensel Sand, which outcrops along the creek bed and smaller
drainages. Glen Rose Limestone outcrops in the hills east of the pool, and is the most
areally extensive formation cropping out within the watershed (Figures 1 and 2).
There are two principal inflows to Hamilton Pool. The main source is Hamilton Creek,
which empties into the pool via a 25-meter waterfall. During baseflow, the sole source of
this water is a group of small springs a few hundred meters upstream from the waterfall. A
second inflow consists of groundwater dripping from a large travertine formation ("Drippy
Rock") and stalactites which line the ceiling of a cave-like structure which overhangs the
pool. Both these inflows issue from the Cow Creek Limestone. During baseflow, Drippy
Rock and stalactite flow total approximately 0.035 ft3/s, while flow from the upper springs
varies from 0.1 to 0.5 ft^/s.
Approach
The aqueous chemistry within the Hamilton Pool Preserve was examined to determine
the water source, flow mechanism, and subsurface processes. Water samples were
collected weekly for six months from the four springs comprising the Upper Springs site,
from Drippy Rock, and from Hamilton Pool itself. Field measurements of pH,
temperature, alkalinity, and conductivity were recorded, and analyses of major anions and
cations performed in the laboratory. The chemistries of the four upper springs, ah1 located
within three meters of one another, were compared to see if one spring could be chosen as
representative of all four. Because all four sites displayed similar solute concentrations and
temporal variation, it was decided to use analyses of East Spring as indicative of the
chemistries of all the upper springs.
Mineralogy of solid phase samples was analyzed to estimate the contribution from
each formation to aqueous chemistry. Fresh samples of the Glen Rose Limestone, the
Hensell Sand, the Cow Creek Limestone, and the Hammett Shale were collected in the field
and analyzed for major cations.
407
-------
•XV:':VJ Glen Rose Limestone
Hensell Sand
Cow Creek Limestone
q^HHil Hammett Shale
Sycamore Sand
50 200 600ft
Figure 1. Plan view of Hamilton Pool Preserve.
408
-------
^i Glen Rose
):£j Limestone
Hensell
Sand
Cow Creek
Limestone
Hammett
m Shale
Sycamore
Sand
Hamilton
Pool
900'
850'
S5^saaasHE; ,800'
750'
L700'
600'
VERTICAL EXAGGERATION X 10
500ft.
Figure 2. Cross Section of through Hamilton Pool Preserve.
-------
Results
Environmental processes and flow mechanics cause differences in aqueous chemistry at
Drippy Rock and East Spring. The formation of travertine where Drippy Rock water
issued from the subsurface indicated that the drip water became supersaturated with respect
to aragonite upon contact with the atmosphere. Additional carbon dioxide may have been
lost to the atmosphere from the collection vessel in the amount of time that it took to
accumulate enough drip water for a sample, as suggested by White (1988). Saturation
indices for aragonite, calculated with WATEQF (Plummer et al., 1976), an equilibrium
speciation program, showed that Drippy Rock water was indeed supersaturated with
respect to aragonite, while East Spring was slightly undersaturated. Analyses of Drippy
Rock revealed that Drippy rock was depleted in calcium ion and had increased pH relative
to East Spring, as expected. In order to compare the subsurface chemistries of the two
sources, PHREEQE (Parkhurst et al., 1980) was used to back-model the environmental
processes affecting the drip water. Loss of carbon dioxide was reversed by reacting
Drippy Rock water with a fixed pCO2 equal to that of baseflow at East Spring (-1.70), the
effects of ambient temperature were reversed by holding temperature fixed at the baseflow
temperature of East Spring (23°C), and the water was held in equilibrium with calcite.
Figure 3 compares the chemical analyses for Drippy Rock and East Spring with the results
of the simulated subsurface Drippy Rock water. Modeling Drippy Rock water in the
subsurface eliminated most of the difference in pH and calcium concentration for the two
waters, but had no effect on magnesium concentrations.
Drippy Rock and East Spring had contrasting responses to rainfall. Conductivity and
rainfall for a three-month period are shown in Figure 4. Conductivity at East Spring drops
after a major rain, while the conductivity at Drippy Rock remains the same or increases
slightly.
In general, the back-modeled Drippy Rock water and East Spring water had similar
overall chemistries. Both were both calcium/magnesium bicarbonate waters. They
contained similar levels of magnesium, chlorine, sulfate, and silica, and had very low
levels of iron, phosphate, boron, and potassium. These similarities suggest that the source
of these waters was similar. To investigate possible rock-water interactions, solid samples
of Glen Rose Limestone, Hensell Sand, Cow Creek Limestone, and Hammett Shale were
analyzed for major cations. The results are displayed in Table 1, which shows that the
Cow Creek Limestone did not have any detectable magnesium, in contrast to the other
formations, which contained magnesium levels similar within an order of magnitude. All
four formations had similar proportions of calcium and strontium. These results indicate
that 1) the Cow Creek Limestone has been entirely recrystallized into a pure calcite
limestone, and 2) that the magnesium in water samples issuing from the Cow Creek
410
-------
PH
|
Drippy Rock
East Spring
Simulated Dttppy Rock
Calcium Concentration
Magnesium Concentration
Figure 3. Temporal variation in pH, calcium, and magnesium versus rainfall.
Simulated water is fixed at pCO2 = -1.70, T = 23.0° C, and is in equilibrium
with calcite.
-------
Drippy Rock
L
\
§
500
U400
E
300 f
t>
1
o
o
-200
100
S a
500
I 2
1
a:
100
3) TO
Figure 4. Conductivity versus rainfall at Drippy Rock and East Spring.
412
-------
Table 1. Cation analyses for solid phase samples.
Hammett Shale 1
Hammett Shale 2
Hammett Shale 3
Hammett Shale 4
Cow Creek Limestone 1
Cow Creek Limestone 2
Glen Rose Limestone
Hensell Sand
Ca
Mg
Sr
0.324
0.442
0.284
0.476
0.443
0.536
0.393
0.245
0.043
0.008
0.012
0.003
0.000
0.000
0.028
0.075
0.00038
0.00033
0.00051
0.00056
0.00063
0.00028
0.00047
0.00019
Units given in gram constituent per gram sample.
Limestone must be derived from other formations, most likely the Glen Rose Limestone,
which covers the majority of the catchment area. These results suggest that water at both
East Spring and Drippy Rock originate from recharge water falling on the Glen Rose
Limestone.
Discussion
The results of the chemical analyses suggest that water at Drippy Rock is chemically
similar to water from the upper springs, but that each is controlled by contrasting flow
mechanics.
The chemical behavior of East Spring suggests that it is controlled by conduit flow.
The decrease in conductivity and calcium concentrations after rainfall at this site indicates
dilution of baseflow by low-conductance meteoric water. Once baseflow water has been
diluted by rainfall in a conduit system, it moves swiftly through the aquifer and emerges
from a spring before it has reached equilibrium with the surrounding rock matrix
(Jacobson and Langmuir, 1974). An interesting physical manifestation of conduit flow at
this site was the build-up of small mounds of sand and gravel at the mouths of the upper
springs after a particularly heavy rain; the sediment had evidently moved through the
conduits as bedload. The conduit nature of flow at this site greatly increases its
susceptibility to contamination, as water which enters the conduits undergoes minimal
filtration and adsorption of contaminants.
413
-------
In contrast, diffuse flow mechanics control the water chemistry at Drippy Rock, as
evidenced by the lack of variation in conductivity. The slow rate of diffuse flow allows
water to approach equilibrium with the rock matrix, and water at this site remains
unaffected by dilution from rainfall. In diffuse flow systems, rainfall may also flush soil
water with high carbon dioxide concentrations into the rock matrix. Such an increase in
carbon dioxide could cause the increase in dissolved calcium concentrations observed at
Drippy Rock after rainfall.
Conclusion
The two groundwater inflows to Hamilton Pool are genetically very similar. The
contrasting chemistries of these inflows to Hamilton Pool are caused by physical
differences in flow as the water comes into contact with the atmosphere. Drip water from
Drippy Rock undergoes much more degasing upon atmospheric contact than spring water
at East Spring; it also responds more quickly to ambient temperature. As a result, Drippy
Rock water undergoes an increase in pH accompanied by a reduction in calcium and
bicarbonate concentrations as it emerges from the subsurface. After removing the effect of
these processes through geochemical modeling, the simulated Drippy Rock subsurface
water and the East Spring water both appear to originate as recharge through the overlying
Glen Rose Limestone.
Contrasting response to rainfall indicates that the two sources are controlled by
contrasting flow mechanics. East Spring is controlled by conduit flow. As a result, after
rainfall East Spring water shows an increase in conductivity and a decrease in
concentrations of several ionic constituents including calcium. Because conduit-controlled
springs are susceptible to contamination, Hamilton Pool is vulnerable to water quality
degradation from this source. In contrast, the second source (Drippy Rock) is controlled
by diffuse flow. After rainfall, water at Drippy Rock undergoes a slight increase in
conductivity and an increase in calcium concentration due to flushing of soil carbon dioxide
into the water.
Acknowledgements
Very special thanks go to Travis County Parks and Recreation staff at Hamilton Pool
Preserve, particularly Terri Siegenthaler and Dan Chapman. Additional thanks to Roger
Lee, Raymond Slade, and Colleen Stapleton.
414
-------
References
Jacobson, Roger L. and Donald Langmuir, 1974, Controls on the quality variations of
some carbonate spring waters, /. of Hydrology, 23, 247-265.
Mahler, Barbara J., 1991, A Geormorphic, Hydrologic, and Geochemical Study of
the Hamilton Creek Watershed, Travis County, Texas, Masters Thesis,
University of Texas at Austin.
Parkhurst, D.L., D.C. Thorstenson and L.N. Plummer, 1980, PHREEQE A
Computer Program for Geochemical Calculations, U.S. Geol. Surv. Water-
Resour. Invest. 80-96, NTIS Tech. Report PB 81-167801, Springfield, VA.
Plummer, L. N., B.F. Jones, and A.H. Truesdall, 1976, WATEQF - a FORTRAN IV
version of WATEQ, a computer program for calculating chemical equilibrium in
natural waters, U.S. Geol. Survey Water Resour. Invest. 76-13,, 61 pp.
Shuster, Evan T. and William B. White, 1971, Seasonal Fluctuations in the chemistry
of limestone springs: a possible means for characterizing carbonate aquifers,
Journal of Hydrology, 14, 93-128.
White, William B., 1988, Geomorphology and Hydrology ofKarst terraines, Oxford
University Press, N.Y.
Biographical Sketch
Barbara Mahler is a Master's Candidate in the Department of Geological Sciences at the
University of Texas at Austin. She received her Bachelor's Degree from Boston
University, and will begin working toward a Ph.D. in hydrogeology at U.T. in the
spring of 1991. Dr. Philip Bennett is an Assistant Professor in the Department of
Geological Sciences at the University of Texas at Austin.
415
-------
The Interaction of Row Mechanics and Aqueous chemistry in a Texas Hill Country Grotto.
By Barbara Mahler and Philip Bennett
Q: Do either or both of the chemistries of the diffuse and conduit-dominated flow regimes vary
seasonally?
A: Water samples were collected from June through October, 1990, a sampling period insufficient
to verify existence or lack of seasonal variation in chemistry. During this period no seasonal
variation was detected in the conduit-controlled regime. In contrast, the samples collected at the
diffuse-controlled site had calcium concentrations which were inversely related to ambient
temperature, due to rapid equilibration of the drip water with surface conditions. However, we
assume that within the subsurface the diffuse-controlled regime would display the same sensitivity
to seasonal variation as the conduit-controlled regime.
The sampling period was necessarily limited due to the scope of the project (field work for
a M. A. thesis). However, regular sampling over a two year period may reveal differences in
seasonal variation between the two regimes, which would provide more information about the
relation between the two water sources.
416
-------
THE ORONOCO LANDFILL DYE TRACE III: RESULTS FROM A SUPERFUND
REMEDIAL INVESTIGATION IN A GLACIATED, DIFFUSE-FLOW KARST.
E. Calvin Alexander, Jr. and Scott C. Alexander
Department of Geology and Geophysics
University of Minnesota, Minneapolis, MN 55455
Barbara J. Huberty
Olmsted County Public Works Department.
Rochester, MN 55904
James F. Quinlan
Quinlan & Associates, Inc., Nashville, TN 37222
ABSTRACT
The Oronoco Landfill in Olmsted County, Minn., a Superfund
site 10 miles north of Rochester, is built on the Ordovician
Prairie du Chien Formation, and important regional karst
aquifer. As part of the site's R.I., a dye trace was con-
ducted to determine: 1) the local directions(s) and speed of
ground water flow, 2) if an existing monitoring well network
effectively intercepted ground water flow beneath the site,
and 3) which local private wells were most likely to be af-
fected by potential ground water contamination attributable
to the landfill.
Samples were collected from 11 on-site monitoring wells, 12
springs, and 201 private wells. Continuous pulses of dye
were detected in 6 on-site monitoring wells, and 2 private
wells located 2.2 and 2.9 km northeast of the landfill. The
continuous pulses of dye detected in the wells have persisted
for up to two years. The concentration of dye in the wells
periodically increases after major recharge events. This
karst is a diffuse-flow system with ground water flow to the
northeast at velocities of the km/yr. The monitoring wells
417
-------
appear to effectively intercept ground water flowing under
the site. Conventional quarterly sampling will detect the
largest, chronic hypothetical landfill releases but may miss
short-term, acute pulses of pollutants. The continued pres-
ence of dye in the system has proven to be a useful guide
during the installation of additional monitoring wells at the
site.
Introduction
The Olmsted County Sanitary Landfill is located approximately
10 miles north of Rochester , Minnesota (Figure 1) and was
purchased by the City of Rochester in 1969. A permit to con-
struct a landfill on the site was granted by the Minnesota
Pollution Control Agency (MPCA) in 1970. Site development
began in 1972. The City of Rochester operated the site until
1983, when the City and Olmsted County agreed to transfer the
permit to Olmsted County.
The landfill received waste from local refuse collection
haulers serving residences throughout Olmsted and adjacent
counties. Local commercial establishments, institutions, and
industries also used the site. In addition to the usual ref-
use from these generators, the site also received incinerator
ash, utility ash, wastewater treatment plant sludge and
sludge ash, miscellaneous industrial process sludges, and
debris resulting from a 1978 flood. In March 1987, the land-
fill was closed to municipal solid waste, but continues re-
ceiving demolition, coal ash, and asbestos waste.
In 1983, analysis of water samples from monitoring wells at
the site indicated the presence of VOCs. As a result, MPCA
and EPA began Superfund actions at the site in 1986. Subse-
quently MPCA and EPA agreed to establish MPCA as the lead
agency for remedial action. The MPCA negotiated a Response
Order by Consent with the City of Rochester and Olmsted
County which became effective in December 1989.
As part of the site's Remedial Investigation (R.I.), the Olm-
sted County Landfill Dye Trace Study was conducted to deter-
mine: 1) the local directions(s) and velocity of ground
water flow, 2) if an existing monitoring well network effect-
ively intercepted ground water flow beneath the site, and 3)
which local private wells were most likely to be affected by
potential ground water contamination attributable to the
landfill. The Study was initiated in April, 1989 and con-
tinues at the present time (October, 1991). This report is
based primarily on data gathered between April, 1989 and
September, 1990 and is excerpted from Alexander et al.
(1991). The Study was divided into two phases. Phase 1
extended from the initiation of the project through March 31,
1990. Phase 2 extended from April 1, 1990 to the present.
418
-------
N
Figure 1. Map of Olmsted County Minnesota with the town-
ships, cities, major highways and County Sanitary
Landfill near Oronoco shown.
419
-------
Site Geology and Hydrology
In the vicinity of the site, glacial overburden generally is
less than 100 feet thick. Within the site, the overburden is
0 to 60 feet thick and includes sands and gravels, clay and
silt. Clay and silt units predominate in the southern part
of the site. Sand and gravel underlies the central and
northern part of the site, including the waste cells.
The first bedrock encountered under the site is the Prairie
du Chien Group. It is up to 250 feet thick and is predomi-
nantly dolomitic limestone. It is fractured, jointed, and
contains numerous karst features. Underlying the Prairie du
Chien Group is the Jordan Sandstone (Olson, 1988a, 1988b).
The water table is encountered in the Prairie du Chien Group
(Kanivetsky, 1988). In the vicinity of the waste cells,
groundwater monitoring wells completed at the water table
have static water levels at depths of 80 to 115 feet below
ground surface. Ground water flow in the vicinity is gener-
ally to the east or northeast. Residential wells in the vi-
cinity draw water from both the Prairie du Chien and the
Jordan aquifers.
Dye Trace Methods
Groundwater flow velocity and direction in limestone is aqui-
fers difficult to predict because the exact nature of the
fractures, joints and solution cavities in the limestone is
unknown. Groundwater flow in these conditions can be studied
by introducing dye at one or more locations and then monitor-
ing the groundwater at many locations to determine the speed
and direction of groundwater and dye movement.
Rhodamine WT dye was selected for use in this study princi-
pally because of its ease and economy of analysis. Dye was
introduced into carbonate bedrock in a quarry west of the
landfill in May 1989 (Phase 1). A second introduction of dye
was made in March 1990 in a sinkhole located east of the site
(Phase 2).
A study area approximately 1.5 miles square was established.
Groundwater samples were taken daily from approximately 100
residential wells and from the landfill monitoring wells.
Springs and sites along the Zumbro River were sampled less
frequently.
The residential wells were distributed throughout the study
area. They included up-gradient wells, generally west and
south of the landfill site, and wells north and east of the
Zumbro River, a hydrogeologic boundary. Residential well
samples were taken by the residents.
420
-------
Springs were sampled because they were considered to be
potential discharge points for dye. The Zumbro River was
sampled in an effort to detect dye which might seep directly
into the river. These samples were taken by County staff.
Results and Conclusions
Rhodamine WT dye reached the first monitoring well less than
24 hours after introduction (see Figure 2). Dye reaching a
total of six monitoring wells in the following weeks and
months documented a complex pattern of groundwater flow
roughly west to east under the waste cells. The continued
presence of dye in the monitoring wells documents that they
are adequately sampling the local groundwater flow. All of
the positive monitoring wells are in the Prairie du Chien
aquifer.
Rhodamine WT dye reached private well 151 north-northeast of
the landfill 77 days after introduction (see Figure 3). That
well is 1.2 miles northeast of the dye introduction point.
The minimum speed the leading edge of the dye traveled to
reach well 151 is 5.7 miles per year. Dye continues to
emerge from that well in October 1991. Dye reached private
well 108, 1.5 miles northeast of the introduction point, in
207 days. Dye also continues to emerge from that well. Both
wells are older Prairie du Chien wells. These two positive
wells document a groundwater flow path from the landfill to
the north-northeast in the Prairie du Chien aquifer.
Rhodamine WT from the first dye introduction has not been
confirmed at any of the other original monitoring wells, res-
idential wells, springs, or river stations. However, trace
levels of Rhodamine have appeared at spring 412 (see Figure
4)- Rhodamine WT from the first dye introduction has been
detected in several of the new monitoring wells drilled into
the Phase 1 dye plume. Rhodamine WT from the second intro-
duction of dye was detected during the installation of a new
monitoring well nest in June 1991. No other detections from
Phase 2 dye have occurred.
The dye pulses in five of the six positive monitoring wells
respond rapidly to recharge events producing large fluctua-
tions in the dye concentrations. The monitoring well break-
through curves illustrate the complex nature of the ground-
water system beneath the landfill. The Prairie du Chien
aquifer under and surrounding the landfill has flow charac-
teristics that are intermediate between those of mature,
conduit-flow karst aquifers and porous-media aquifers.
In addition to the continuous dye pulses discussed above, a
low level contamination problem complicated and increased the
cost of the Study- During the summer of 1989, dye detections
421
-------
PHASE I DYE MOVEMENT PHASE I DYE MOVEMENT
1 DAY
PHASE I DYE MOVEMENT
4 DAYS
PHASE I DYE MOVEMENT
7 DAYS
45 DAYS
LEGEND
DYE INTRODUCTION LOCATIONS
I • - PHASE I
II* - PHASE II
DYE PLUME
SAMPLING LOCATIONS
» - PRAIRIE DU CHIEN WELLS WITH DYE
' - PRAIRIE DU CHIEN WELLS WITH NO DYE
- JORDAN WELLS WITH NO DYE
SOLE!
Figure 2. Phase 1 dye movement during the first 45 days,
422
-------
PHASE I DYE MOVEMENT PHASE I DYE MOVEMENT
4 MONTHS
PHASE I DYE MOVEMENT
10.5 MONTHS
LEGEND
8 MONTHS
PHASE I DYE MOVEMENT
11 MONTHS
DYE INTRODUCTION LOCATIONS
I • - PHASE I
II* - PHASE II
DYE PLUME
SAMPLING LOCATIONS
A - PRAIRIE DU CHIEN WELLS WITH DYE
- PRAIRIE DU CHIEN WELLS WITH NO DYE
- JORDAN WELLS WITH NO DYE
SCALE I
Figure 3. Phase 1 dye movement, four to eleven months
423
-------
•t-
K5
c
n
o
i-(
o
en
en
en
o>
o
rt
H-
o
O
M)
CD
O
<
(D
3
(D
3
rt
LOCATION
MAP
3000
2500
\
2000 1500
DISTANCE (FEET)
THE HORIZONTAL AND VERTICAL
SCALES ARE DIFFERENT.
1000
-------
began to occur erratically at isolated, random locations in
the residential wells. The number of residential wells in
the study was expanded. Ultimately the number exceeded 200.
These scattered, short-term detections were eventually shown
to be a random phenomenon unrelated to the Rhodamine WT used
in the study.
The results of an anion survey conducted as part of the Study
indicates that the Jordan aquifer in the Study Area is hydro-
geologically isolated from the overlying Prairie du Chien
aquifer. In many parts of southeastern Minnesota these two
aquifers behave as single aquifer. In the Study Area, the
Prairie du Chien aquifer contains elevated chloride and ni-
trate contents presumably from a variety of surface sources.
The Jordan aquifer, in contrast, shows very little evidence
of surface impact. The pattern of dye movement revealed by
the positive wells is consistent with the chemical data and
with Study Area scale potentiometric mapping.
Recommendations
Based on dye trace study results through September 1990, ad-
ditional geotechnical investigations were recommended for the
site. Specific recommendations included: 1) slug and pump
tests on the existing monitoring wells, 2) investigation of
the source of the vertical head gradients at the site, 3)
gamma logging of the monitoring wells and selected nearby
private wells, and 4) investigation of the existence and size
of rapid pulses of contaminants in the monitoring wells.
It was also recommended that all of the existing monitoring
wells should be maintained as part of the expanded Environ-
mental Monitoring System and private wells 151 and 108 should
be added to the array. There is a danger that access to
samples from wells 151 and 108 could be lost as their owners
seek alternative water supplies. The ability to sample wells
151 and 108 should be maintained.
Several new monitoring wells were recommended. Several
shallow water-table, mid-level Prairie du Chien and at least
one Jordan monitoring well were needed in the northeastern
part of the landfill site. Mid-level Prairie du Chien moni-
toring wells and another Jordan monitoring well were needed
in the area between the landfill site and private well 151.
Finally, if the anion evidence for leachate pulses in the
monitoring wells is confirmed, the ability of routine quar-
terly sampling to adequately define the contamination poten-
tial will need to be examined.
425
-------
Remedial Investigation
During the summer of 1991, work was conducted on the site as
part of the Remedial Investigation (RI) phase of the Super-
fund process. The second stage of field work for the R.I. is
currently underway- Seven new monitoring wells have been
installed and geophysical logging was conducted on a subset
of the new and old monitoring wells and residential wells 151
and 108. A "maintenance mode" of sampling for dye has been
maintained throughout the R.I. The presence of dye in the
aquifer was useful during the installation of the new moni-
toring wells to confirm correct siting of the wells and to
assist in selecting the depth for screen placement.
References
Alexander, E. Calvin, Jr., Barbara J. Huberty and Keith J.
Anderson (1991) Final Report for Olmsted County Dye
Trace Investigation of the Oronoco Sanitary Landfill.
Prepared by Donohue and Associates, Inc. April 1991 in 4
volumes.
Kanivetsky, Roman (1988) Bedrock Hydrogeology- Sheet 5 in
(N.H. Balaban, ed.) Geologic Atlas Olmsted County,
Minnesota. County Atlas Series Atlas C-3, Minnesota
Geological Survey, St. Paul, MN.
Olson, B. (1988a) Bedrock Geology. Sheet 2 in (N.H.
Balaban, ed.) Geologic Atlas Olmsted County, Minnesota.
County Atlas Series Atlas C-3, Minnesota Geological
Survey, St. Paul, MN.
Olson, B. (1988b) Depth to Bedrock and Bedrock Topography.
Sheet 4 in (N.H. Balaban, ed.) Geologic Atlas Olmsted
County, Minnesota. County Atlas Series Atlas C-3,
Minnesota Geological Survey, St. Paul, MN.
BIOGRAPHICAL SKETCHES OF THE AUTHORS
E. Calvin Alexander. Jr.r is a Professor in the Department of
Geology and Geophysics at the University of Minnesota. He
has a B.S. in chemistry (1966) from Oklahoma State University
and a Ph.D. in chemistry (1970) from the University of Mis-
souri at Rolla. The central theme of his current research is
the rate of movement of fluids in hydrogeology. He and his
research group are utilizing a variety of methods to measure
flow and residence times, which can range from hours to ten
of thousands of years. His full address is as follows.
426
-------
Department of Geology and Geophysics
University of Minnesota
Minneapolis, MN 55455-0219
(612) 624-3517
Scott C. Alexander is a Senior Laboratory Technician in the
Department of Geology and Geophysics at the University of
Minnesota. He has a B.S. in geophysics (1986) from the
University of Minnesota. His current research involves the
application of chemical, isotopic, and computer modeling
techniques to the solution of ground water pollution pro-
blems. He is acting as a technical consultant for federal,
state and local research and water planning groups while
supporting the day-to-day operation of a hydrogeologic re-
search facility. His full address is as follows.
Department of Geology and Geophysics
University of Minnesota
Minneapolis, MN 55455-0219
(612) 624-3517
Barbara J. Huberty is an Environmental Analyst in the Olmsted
County Public Works Department, Solid Waste Division. She
has a B.S. in biology (1977) and a B.S. in broad-area science
education (1979), both degrees from the University of Minne-
sota. Her current responsibilities include environmental
monitoring of the Olmsted County's landfill operations and
oversight of a Superfund Investigation. Her full address is
as follows.
Olmsted County Public Works Department
2122 Campus Drive S.E.
Rochester, MN 55904-4744
(507) 285-8231
James F. Ouinlan is an independent consultant and former
research geologist for the National Park Service at Mammoth
Cave Kentucky. He has a B.S. in geology (1959) from Virginia
Polytechnic Institute and a Ph.D. in geology (1978) from the
University of Texas at Austin. His field experience in-
cludes: 32 years of research and observations in karst ter-
ranes of more than 20 countries and 25 states; more than 500
dye-traces in the Mammoth Cave region of Kentucky and in
various states; the environmental applications of dye-
tracing; evaluation of waste-disposal sites in limestone
terranes; design of ground water monitoring networks; and
analysis and remediation of sinkhole development. He and
427
-------
Ralph Ewers received the 1986 E.B. Burwell Award from the
Geological Society of America for their paper on monitoring
ground water in karst terranes. His full address is as
follows.
Quinlan & Associates, Inc.
P.O. Box 110539
Nashville, TN 37222-0539
(615) 833-4324
428
-------
Questions
1. There is an explicit vertical path followed by the first
trace. What is the possibility of there being an
additional vertical "step" and a deeper, undetected
(untested) flow system?
Response: The possibility exists. The monitoring wells
at the landfill do not sample the lower part of the Prairie
du Chien aquifer and there was no way to monitor that part of
the aquifer in this study. The major horizontal flow seems,
however, to be in the middle of the aquifer, and we did not
see any dye exiting at springs that appears to have missed
the wells. Finally, the flow system does not extend locally
into the underlying Jordan (sandstone) aquifer as evidenced
by the lack of dye detection in any of the numerous Jordan
wells monitored.
2. In retrospect, how could (should?) your investigation
have been organized to maximize efficiency and lower
costs without sacrificing reliability?
Response: Given the initial state of hydrogeologic and
dye-tracing knowledge for this site, coupled with the
administrative requirements and the residents' concerns,
there is little that we could have done to maximize
efficiency and lower costs. Armed with three years and half
a million dollars worth of experience, requirements for
future traces can be refined. We would remove the
requirements for immediate public response and pre-set,
automatic sampling plan expansion. Rather, necessary changes
in the project plan should be technically driven.
The people who pay for dye traces and those who are
affected by the results find it dificult to accept unknowns
(costs, time-spans, out-comes, etc.) However, unknowns can
not be completely removed and will continue to make dye
traces complicated and expensive.
429
-------
430
-------
Deducing Karst Aquifer Recharge, Storage, and Transfer
Mechanisms Through Continuous Electronic Monitoring -
A Confirmation With Tracers
Peter J. Idstein, Ralph O. Ewers Ph.D.
Eastern Kentucky University and Ewers Water Consultants Inc.
Richmond, Kentucky
ABSTRACT
Rainfall events produce significant changes in the stage,
conductivity, and temperature of the groundwater mass in a karst
aquifer. In theory, the arrival time of these thermal and
conductivity pulses, relative to the stage pulses, should reveal
the nature of the aquifer recharge, storage, and transfer
mechanisms above the point of measurement.
An analysis of these fundamental aquifer properties was
made utilizing continuous monitoring records from a karst
aquifer in Virginia. These records suggested that there is no
allogenic water entering this portion of the aquifer. This was
evidenced by the subdued response in temperature during
precipitation events. The monitoring record also suggested that
the basin contributing flow to the part of the aquifer under
consideration does not contain a significant conduit storage
component. This was shown by the simultaneous response of
stage, conductivity, and temperature to even the smallest of
storm events. The rapid response to storm events and the
limited rise in stage indicates that the basin is of very small
dimensions.
Dye tracing of the two sources of allogenic water, in the
area, confirmed that they do not contribute to the portion of
the aquifer that was continuously monitored. The basin that
contributes to groundwater flow at the monitoring point has been
shown by dye tracing to be an area elongated along the strike of
the local rocks, collecting only autogenic waters.
431
-------
THEORY
In theory, the nature of the flow path or paths that
groundwater in a karst aquifer traverses, to reach a particular
site will be reflected in the physical and chemical adjustments
each path imprints upon the water passing through it. Ashton
(1966) proposed that the flow at any one point in the system
will be the sum of the response of the portions tributary to
that location. Given the most simple configuration the response
of a measured parameter would be a deviation from base flow
conditions to a peak and then a recession to the base flow
condition. Response for a system that consists of two simple
inputs that combine flow up stream from the monitored location
will result in a record that reflects this as the sum of the
characteristics of the two individual components (Figure 1).
The resultant record will thus have peaks that reveal the
existence and nature of each of the inputs, including their
phase relationships. Ashton suggests that in addition to the
flow rate other parameters of utility may be turbidity, pH,
hardness,and temperature. He also suggests that specific
conductance as a measure of the ionic character of the water is
a practical alternative for hardness. The response for each of
these parameters during the passage of a flood pulse will be
modified by the various components of the aquifer that are
encountered. The changes that the original pulses undergo and
the relative timing of these pulses should reveal the nature of
the storage, transfer mechanisms and the magnitude of each
component of the aquifer system.
STUDY SITE
The Cathedral Hall Passage in Unthanks Cave has a free
surface cave stream that flows through it. The Cathedral Hall
Cave Stream contains an extraordinary assemblage of cave adapted
aquatic invertebrates. One of these troglobitic organisms is an
endangered species of Isopod Crustacean Caecidotea recurvata
that the Virginia Chapter of The Nature Conservancy is committed
to preserving. This study was conducted to determine the extent
of the drainage basin and the transfer mechanisms for the water
which contributes to the Cathedral Hall Cave Stream.
Karst by it's very nature presents certain problems that
must be overcome by the groundwater scientist. The methods that
have been used in this study have been implemented to understand
the special nature of karst groundwater flow.
LOCATION OF STUDY AREA
The study area is located in Lee County Virginia (Figure
2), within the Powell Valley. The Powell Valley is one of the
most western valleys of the Appalachian Valley and Ridge
physiographic provence. The valley is bounded on the northwest
by Cumberland Mountain and to the southeast by Wallen Ridge.
432
-------
me
Figure 1. Flood wave complexing from multiple inputs
(modified from Ashton 1966).
r.. v
'
V
T VIRGINIA
TENNESSEE
Figure 2. Location of the study area.
433
-------
The limit of the investigation was controlled by the location of
the Powell River and Wallen Ridge.
REGIONAL GEOLOGY
The rocks in the region are Ordovician carbonates with some
shales in the valley and Silurian siltstones on the ridge.
Unthanks Cave is formed in the upper portion of the Martin Creek
Limestone near the base of the Hurricane Bridge Limestone. The
Sandy Ridge Anticline follows the regional structure and is
located between the cave and the Powell River. Dip of the rocks
in the area ranges from horizontal to over 45 degrees, but
within the cave, dip never exceeds 20 degrees. There are no
major mapped faults with any surface expression in the study
area.
CONTINUOUS MONITORING
Continuous monitoring of the Cathedral Hall Cave Stream was
undertaken using a Campbell Scientific Incorporated digital
micrologger model 21X. This is the central core of a Karst
Water Instrumentation System (KWIS). The micrologger is a very
versatile device that can be used to monitor many different
kinds of sensors and physical parameters. The KWIS used in
Unthanks Cave had three sensors that were monitored by the
micrologger for the time that the system was installed in the
cave. These sensors were used to monitor the stage (water
level), temperature and conductivity (specific conductance) of
the water in the Cathedral Hall Cave Stream.
The stage of the cave stream was monitored using a Druck
pressure transducer. This device possesses a silicon wafer that
varies in resistance as the pressure around it changes. This
change is proportional to the height of water that is above the
sensor. This transducer can detect changes in stage as small as
.001 feet.
Water temperature was monitored using an Platinum
Resistance Thermometer (PRT) supplied by Omega Engineering
Incorporated. The resistance of a platinum wire changes with
the water temperature. This resistance is measured relative to
a 100 ohm precision resistor that is attached to the
micrologger. Temperature is measured in degrees Celsius with a
resolution of -01 degree.
The specific conductance (conductivity) sensor that was
used at this site was designed and constructed specifically for
this site. The new design was developed in order to avoid the
large current demand of earlier instruments. The new design has
greatly reduced current demands and is thus able to be powered
from the micrologger directly. Another advantage to the new
design is the elimination of the "ground looping" problems
associated with most other conductivity measurement systems.
The datalogger and sensors were located in the Cathedral
Hall stream passage of Unthanks Cave. This passage is oriented
434
-------
along the strike of the rocks in the area and is apparently
controlled by the fractures in the Martin Creek Limestone. The
cave stream is about one inch deep by one foot across. There
are numerous pools in the bottom of the accessible potions of
the cave stream that are up to 18 inches deep. The sensor
cluster used in this study was located in one of these pools to
ensure that they always remain covered with water. The
datalogger was located five feet above the cave stream on a
ledge.
TYPICAL RESPONSE OF THE SYSTEM
A typical response for the aquifer system at Unthanks Cave
can be seen in Figure 3. This record is related to the day 193
rain a single short uncomplicated event. The system responds
with a rapid change in stage as the pulse of water progresses
through the aquifer. Temperature of the water experiences a
rapid rise at the same time as the stage changes are seen.
Water temperature for this event responds in a increasing
manner, events during colder seasons are represented by a
decrease in temperature. There is also an immediate decrease in
the specific conductance of the water from the base flow values.
A short term reversal of the trend of temperature and
conductivity occurs within a few minutes of the initial system
response. The detail of this occurrence can be seen most
clearly in Figure 4.
The peak (the maximum deviation from base flow values
independent of direction) is reached for each of these
parameters at nearly the same time. Recession of stage is
typically a simple return to base level without any abrupt
changes in slope. The return of temperature to base flow values
usually experiences a noticeable change in slope during its
recession. Conductivity may also experiences a change of slope
during the recession to base flow values.
All of the responses to storm events take this general
form. Deviations from this form are believed to be due to
variations in antecedent conditions. These changes may be due
to the system being in a state of recovery from a previous storm
event and thus its response is that of the combined pulses.
Another factor connected with these deviations is the present
state of saturation of the soil and rock. This may vary
depending upon the length of time between precipitation events
and the seasonal variations of evapotranspiration. Further
variation may also arise from the stage increase associated with
large storm events which bring other flow paths into operation
that were not active during smaller events.
The simultaneous response of temperature, conductivity and
stage is believed to be an indication that fresh storm water is
arriving at the monitoring site from inputs very nearby and
without significant restriction of flow. Hess and White
(1973,1988) and Meiman et. al. (1988) observed a lag between the
arrival of the stage response and a response in temperature and
conductivity. This has been attributed to the propagation of
435
-------
Figure 3.
Figure 4.
300
22:00 24:00
193
194
195 196
JUUANDAYS
18:00 20:00 22:00 24,00
197
300
12:00
MOO
16:00 16:00 20:00 22.00 24:00
BOMB (MIAN MY 113)
Figure 3. Aquifer response to rain event of Julian day 193.
Representative of typical aquifer response.
Figure 4. Detailed inspection of the fluctuation in specific
conductance and temperature shortly after first
response to the Julian day 193 rain event.
436
-------
the stage pulse as a pressure wave through the phreatic portion
of the aquifer in advance of the fresh input waters. This lag
effect has not been observed at this site, thus it is asserted
that there is no significant phreatic contribution to the flow
that passes the monitoring site.
The reversal of trend in temperature and conductivity
(Figure 4) a short time after the initial response to the storm
event is presumed to be the arrival of higher conductive water
with a temperature signature resembling that of base flow
values. This water is assumed to be the flushing of stored
water as the fresh input waters affect other parts of the
system. This may be the arrival of a flow component from
further up stream or displacing water from nearby storage that
exhibits more restricted flow of extended residence time than
the initial response component. The cessation of this reversal
of trend is interpreted as the flushing of the stores of this
second component or the overwhelming of the influence by an
increasing volume of fresh input water.
Meiman et. al. (1988) saw multiple slope changes in the
conductivity response that was attributed to the input of
separate concentrated allogenic sinking streams in the basin.
There is normally only one slope change seen in the conductivity
response with the present study. The arrival of this pulse is
seen as a more conductive water than the fresh input waters.
This high conductive signature is not likely to be associated
with an allogenic sinking stream.
Continued departure from base flow values for all
parameters is seen as the fresh input waters dominate the flow.
Maximum stage occurrence is associated with the cessation of the
source of fresh input waters to the system. The return to base
flow conditions occurs as the excess water moves through the
aquifer. Conductivity and temperature peaks are reached
slightly after the stage crest is reached. Recession to base
flow conditions for conductivity and temperature is marked by an
initially rapid change that corresponds to the swift recession
of stage. The slope of the recession of conductivity and
temperature changes as the stage response approaches base flow.
This change in slope is due to reduced significance of the quick
flow component of the fresh input water and an expansion of the
influence of the restricted flow extended residence time
component.
SYSTEM RECOVERY AFTER STORM EVENT
The response to the day 259 precipitation event is similar
to the record for all single storm events. This event clearly
displays a temperature phenomenon that exists in the record of
most events that have been observed in this study (Figure 5).
After the peak response of the event has been attained for all
parameters and the recession to base flow conditions begins, the
return of temperature to base flow values lags behind that of
stage, and to a lesser extent it lags behind conductivity.
437
-------
Figure 5.
Figure 6.
2.00
252 254 258 258 260 262 264 266
2.401
2.20
2.00-
1.80-
1.60 "
0 80 •
IS
^
5 20C
L , ^ ^
205 .
U\,"''i ... •••[•••
TA....I J..1
10 215 220 225 230 235
\J \
24
1 /.U
13.5-
1.1 0 •
s^-
\
195 200 205 210 215 220 225 230 235 240
500
I 450
I 400
•150-
i
| 300
300
252 254 256 258 260 262 264 266
JUUANDU3
230
195 200 205 210 215 220 225 230 235 240
JUUiNUYS
Figure 5. System response to storm events of Julian days 259
and 255. The response to the day 259 event exhibits
the delay in recovery of temperature, conductivity
and stage relative to each other. The initial
response to the day 255 event show an initial
increase in conductivity.
Figure 6. System response to drought conditions.
438
-------
In contrast to this study Meiman et. al. (1988) observed
that the return to base flow values for temperature was more
rapid than for conductivity. Storage of thermal energy in the
rock mass in the vicinity of the active flow routes is the
likely reason for any lag effect in temperature. The release of
this thermal energy from the rock back into the water slows the
return of temperature to base level until the thermal stores
have been depleted.
This difference in response in the two studies may be
related to the presence or absence of phreatic passages or the
distance between the monitoring site and the source of input.
The diffuse stores in the subcutaneous zone and saturated rock
mass with restricted flow conditions will have a relatively high
conductance due to an extended residence time. The aquifer at
Mammoth Cave has been shown to receive flow from both
concentrated and diffuse sources under both flood flow and base
flow conditions. The combining of multiple sources will slow
the return to base flow values of the conductance at a down
stream monitoring point. Additionally, the aquifer at Mammoth
Cave has a significant phreatic zone. As the storm pulse is
transmitted through the aquifer a significant portion of the
fresh input waters may be pushed from the conduit into the
conduit adjacent stores. These short residence time waters will
be released from the conduit adjacent stores as the flood event
wanes. The low conductivity waters from the conduit adjacent
storage will slow the recession to base flow values. If an
aquifer receives the majority of its base flow from extended
residence time water the response will be such that the
conductivity may return to a base flow condition before the
thermal stores in the rock from the storm event can be depleted.
Meiman et. al. (1988) saw that as a storm pulse was
transmitted through the system, there is a lag in the initial
response of temperature relative to conductivity and stage.
They also note that the amount of lag is affected by the
antecedent conditions. The absence of a thermal lag at the
onset of the storm event, in the present study, suggests that
this initial response is dominated by a quick flow component.
This may reflect concentrated autogenic contribution into
enlarged fractures and vadose shafts. Through this route of
transit the water would not have enough time to react chemically
with the rock to significantly change its conductance. The
leading edge of the thermal response would not be delayed
relative to the other values if the input is near the monitoring
site. Additionally there is sufficient contact with the rock
that a reduction of the temperature of the water without a
significant delay in the arrival of the initial temperature
change is likely. If flow at the monitoring site received
contribution from the allogenic sinking streams in the area this
water would not have sufficient contact with the rock to subdue
the thermal response.
439
-------
CONDUCTIVITY RISE AT THE ONSET OF AN EVENT
Meiman et. al. (1988) saw an initial rise in conductivity
at the onset of storm events. This is explained as the flushing
of high conductive stores from portions of the aquifer that are
not typically active during base flow. The storm event of day
255 (Figure 5) does show a small initial rise at the onset of
the storm event. This is the most noticeable occurrence of this
effect seen during the time this site was monitored. The
typical response has been, as stated before, an immediate drop
in the conductance at the same time that the other parameters
respond. This suggests that the flow route for the fresh input
water must be very direct and possessing very limited stores in
order to dominate the initial response, and that only
periodically are there sources of high conductive water to be
flushed ahead of this storm pulse.
LOW FLOW CONDITIONS
The flow of the cave stream during the period from day 195
to day 235 was seriously reduced due to the dry conditions
(Figure 6). Starting about day 201 the stream flow was reduced
enough that the level of water in the pool, where the probes had
been located, was no longer sufficient to maintain a normal flow
out of the pool. The reduced pool level appears to be
controlled by fractures at a location lower than the normal pool
level.
In response to several small storm events, that occur
during this period of time, the pool level is temporarily raised
to the overflow point that signifies the return to stream flow.
This pool height is only maintained for a short period of time
until the flow is cut off and the surface is returned to a level
controlled by the fractures. Water temperature during this
period is very stable and there is only a limited response that
can be associated with the small storm events that have been
recorded. This thermal stability indicates the limited extent
of these storm events and/or reflects the efficiency of the
thermal stores. Conductance of the water fluctuates with each
storm event but is stable between storm events. When the stream
ceases to flow, the pool is once again isolated and so
measurement of the conductance represents the character of the
last water to be flowing in the stream before flow stopped.
During a portion of the time that the stream in the
Cathedral Hall Cave Passage had ceased to flow, the sinking
streams that flow from the ridge still maintain a low level of
flow. If these streams had contributed flow to the portion of
the cave that was continuously monitored during this time period
the flow of the cave stream would have been maintained.
440
-------
DYE TRACE RESULTS
ALLOGENIC SINKING STREAMS
Two allogenic sinking streams flow off of Wallen Ridge in
the area of the Cathedral Hall Passage. The "Scott Farm Sinking
Stream" (SFSS) was found to sink at the contact of the Hurricane
Bridge Ls. and the Martin Creek Ls. Water flowing in this
stream includes runoff from a few farms and a winding road that
cross the ridge, State Route 758. The sinking stream "SFSS"
(Figure 7) was traced at three different times due to it's
proximity to the cave stream of interest. At no time during the
study was this sinking stream found to connect to the Cathedral
Hall Cave Stream. Dye from "SFSS" has been detected in the cave
at monitoring points down stream from the Cathedral Hall
Passage.
The water flowing in the Horn Farm Sinking Stream (HFSS)
was shown to flow away from Unthanks Cave parallel to strike but
in the opposite direction of all other traces. Water Sinking at
this location on the Horn Farm was never shown, under the
conditions investigated, to contribute to any portion of
Unthanks Cave.
AUTOGENIC SOURCES
During one of the research trips into the Cathedral Hall
Passage a strong odor of "diesel" fuel was detected. After
exiting the cave an investigation of local sources of this
product was initiated. It was determined that the most likely
source was due to dumping of excess product from large metal
drums along side a sinkhole located behind "Bacons Store" (BSD).
This is the site of a small country store that also sells
gasoline. This sinkhole was later used for dye injection in
order to further understand its connection to the cave system.
Water entering at this point flowed to the Cathedral Hall
Passage and was detected at the monitoring site. Dye introduced
at this site was also detected at dye monitoring sites in the
cave down stream from the Cathedral Hall Cave Stream.
"JWD1" is a sinkhole located a short distance south of
State Route 758 in a wooded rocky portion of a farm. Dye input
at this location was detected in the Cathedral Hall Passage.
This dye was also detected at dye monitoring sites down stream
from the Cathedral Hall Cave Stream. This site was only
investigated during high flow.
Sinkhole "JWD2" is also located a short distance south of
State Route 758 on the same farm. This sinkhole is located a
few hundred feet East of JWD1. Dye introduced at this site was
detected at dye monitoring sites down stream from the Cathedral
Hall Cave Stream. This site was tested under high flow
conditions but no dye from this site was detected in the
Cathedral Hall Passage.
"RSD" is a sinkhole that is located very close to and just
south of State Route 758. It is about half way between "BSD"
441
-------
4000 FEET
DRAINAGE BASIN CONTRIBUTING TO
CATHEDRAL HALL PASSAGE CAVE STREAM
LOCATION OF CAVE
Figure 7. Map showing the location of dye insertion sites and
the basin that contributes to the Cathedral Hall
Cave Stream.
-------
and the Cathedral Hall Passage. Dye introduced at this point
was detected in the Cathedral Hall Passage and at dye monitoring
sites further down stream.
CONCLUSIONS
The simultaneous response of all parameters at the onset of
a storm event establishes the existence of a quick flow
component to the flow of the Cathedral Hall Cave Stream. The
lack of delay in response of conductivity and temperature
relative to stage response confirms the absence of a significant
phreatic flow path for this quick flow contribution.
Variations of conductivity and temperature shortly after
the initial response from base level reflect the arrival of
fresh input quick flow waters with flushed stores from more
distant or restricted flow sources in the aquifer. Flushing of
these stores or overwhelming of their signature by the increased
volume of fresh input is considered to be the reason for the
cessation of these occurrences. The ordinary absence of high
conductance waters to be flushed from storage at the onset of
storm events suggests that the routes of quick flow are normally
drained of extended residence time waters.
Return of the waters conductance to base flow values more
rapidly than temperature, reflects the absence of concentrated
allogenic inputs and/or conduit adjacent storage of fresh input
waters. Concentrated allogenic sources and conduit adjacent
storage of fresh input water would have low conductance and
compete with the high conductance restricted flow waters that
maintain the base flow. The absence of these sources in the
flow at Unthanks Cave allows conductance to return to base flow
conditions before temperature.
The cave stream stops flowing during dry periods but the
sinking streams flowing from Wallen Ridge, for a portion of this
dry period, continued to flow and sink. There flow must not be
related to the cave stream under these conditions.
Tracing of both allogenic sinking streams in the area shows
that "HFSS" does not contribute to any portion of Unthanks Cave
and that "SPSS" contributes to portions of the cave down stream
from the Cathedral Hall Passage. Tracing of sinkholes in the
area has shown the location of the drainage basin and confirmed
the source of the water is exclusively autogenic.
The dye tracing has supported the conclusions of the nature
of the drainage basin and the continuous monitoring has revealed
more detail about the transfer mechanisms than the qualitative
dye tracing could do.
REFERENCES CITED
Ashton, Ken, 1966, The analysis of flow data from karst drainage
systems: Transactions Cave Research Group of Great
Britain, v. 7, p. 161-203.
443
-------
Hess, J.W., and White, W.B., 1974, Hydrograph analysis of karst
aquifers: Pennsylvania State University, Institute
Research Land Water Resource Publication no. 83, 63 p.
Hess, J.W-, and White, W.B., 1988, Storm response of the karstic
aquifer of Southcentral Kentucky: Journal of Hydrology,
no. 99, p. 235-252.
Meiman, J., Ewers, R.O., and Quinlan, J.F., 1988,
Investigation of flood pulse movement through a maturely
karstified aquifer at Mammoth Cave National Park: a new
approach: in Proceedings of Environmental problems in Karst
Terranes and there Solutions Conference p. 227-263
BIOGRAPHICAL SKETCHES
Peter Idstein received his Bachelor of Science degree in
Geology at Eastern Illinois University. He is completing his
Master of Science degree in Geology at Eastern Kentucky
University. Mr. Idstein has spent one year working at the
Florida Sinkhole Research Institute conducting studies on
conduit dominated groundwater flow. He has also spent a year
working for Ewers Water Consultants conducting dye tracing
studies and continuous electronic monitoring studies in many
karst dominated and non-karst terranes.
Peter Idstein
Dept. of Geology, Rm. 9 Roark
Eastern Kentucky University
Richmond, Ky. 40475
606-622-1273
Ralph O. Ewers is professor of geology and director of the
Groundwater Research Laboratory at Eastern Kentucky University,
and a principal in Ewers Water Consultants, a consulting firm
specializing in carbonate aquifers. His B.S. and M.S. degrees
in geology were earned at the University of Cincinnati and his
Ph.D. was earned at McMaster University (1982). Professor
Ewers' special interests include the applications of tracer and
electronic monitoring techniques to the solution of practical
environmental problems in karst groundwaters. He was co-
recipient of the 1986 E.B. Burwell Award from the Geological
Society of America for a "work of distinction in engineering
geology."
Ralph 0. Ewers
Ewers Water Consultants Inc.
160 Redwood Drive
Richmond, KY 40475
606-623-8464
444
-------
Deducing Karst Aquifer Recharge, Storage, and Transfer
Mechanisms Through Continuous Electronic Monitoring -
A Confirmation With Tracers
Peter J. Idstein, Ralph O. Ewers Ph.D.
Eastern Kentucky University and Ewers Water Consultants Inc.
Richmond, Kentucky
Question:
Are there temperature lag-times, and are they the same for
low flow and high flow conditions?
Response:
The study at Unthanks Cave showed no temperature lag-times,
at the onset of the storm events during the period that the
system was monitored. Work at Mammoth Cave National Park by
Meiman et.al. 1988 did show significant temperature lag-times at
the onset of storm events.
There were temperature lag-times in the return to base flow
values. Both stage and conductivity returned to base flow
conditions before temperature. The difference between low flow
and high flow conditions does not seem to be as significant as
the variability in temperature of the fresh input waters.
445
-------
446
-------
PETROLEUM HYDROCARBON REMEDIATION OF THE
SUBCUTANEOUS ZONE OF A KARST AQUIFER,
LEXINGTON, KENTUCKY
Scott A. Recker
Delta Environmental Consultants, Inc.
Charlotte, North Carolina
Investigation of karst aquifers using dye tracing techniques has
reached widespread acceptance in the hydrogeologic community.
These techniques have only recently been applied to tracing the
release of contamination from a point source. These studies fail,
however to provide suggestions for the remediation of the karst
aquifer affected by the release.
Following the release of 1300 gallons of gasoline from an
Underground Storage Tank an interceptor trench was installed and
recovered approximately 800 gallons of the lost product. A
subsequent tracer study from the loss location indicated flow
through the subcutaneous zone at 400 ft/day and discharge to the
ultimate resurgence, 1.75 miles away, in 67 days. Ground water
samples collected at the spring resurgence indicated non-detectable
levels of petroleum hydrocarbons. Residual soil and ground water
contamination still remained in the subcutaneous zone at the
release location. This residual contamination acted as a
continuing source of ground water contamination into the karst
conduit drainage network and resulted in gasoline vapors in local
businesses.
A remediation system was designed to remove the residual gasoline
contamination through the use of soil vapor extraction, in-situ
soil washing and standard pump & treat technology. The system was
put on line during April 1991. The recovery and treatment system
continues to remove gasoline contamination from the subcutaneous
zone and reintroduce treated ground water—mimicking constant
precipitation event.
INTRODUCTION
Tracer investigation in karst regimes has become an exact and well
defined technique to determine conduit drainage network boundaries
and conduit flow paths. Tracer techniques have recently been used
to trace the transport fate of contaminants from a point source
release location within the conduit drainage network. Studies of
this type have been used to estimate the residence times of the
contaminants in the aquifer but fail to address remedial options
for the affected portions of the aquifer. Releases of contaminants
are often times to the overburden and subcutaneous zone of the
aquifer rather than directly to the well developed conduit
drainage.
Remedial options for karst aquifers are often not considered due to
the fact that they are applied to the conduit system and fail to
address contamination of the subcutaneous zone. This study
447
-------
presents information collected using classical contaminant
investigative techniques for unconsolidated overburden materials
and using dye tracing techniques and how is is applied applied to
design and implement a remediation plan of the subcutaneous zone
for the case study presented.
The subject area is located at the corner of South Limestone Street
and Gazette Avenue in Lexington, Kentucky (Figure 1 & 2) . The site
lies in the Inner Blue Grass Physiographic Province of central
Kentucky. North of the site are office buildings and houses, the
University of Kentucky is to the east, office buildings and a
residence to the west, apartment buildings to the southwest, and a
bank, convenience store, and restaurant to the south.
SITE HISTORY
The site has been a gasoline station and convenience store for at
least 20 years. A map showing the former, Shop-N-Go (Super America
station), surrounding businesses and residences is included as
Figure 2. On April 12, 1986, the Shop-N-Go reported a sudden loss
of 1,300 gallons of gasoline from an underground storage tank. On
April 13, 1986 gasoline was detected in the basement of Mr. Gatti's
Pizza Restaurant, located approximately 400 feet to the south.
Because the detection of gasoline occurred approximately 24 hours
after the loss, the gasoline apparently originated at the Shop-N-Go
site.
Several remedial activities were undertaken in response to this
loss. Activities included excavation and removal of leaking
underground storage tanks, removal of contaminated soils, and
removal of liquid phase hydrocarbons (LPH) from the tank basin at
the Shop-N-Go site. A recovery trench system was installed
adjacent to Mr. Gatti's Pizza Restaurant and has been in operation
since April 1986. Initial remedial activities resulted in the
recovery of nearly 800 gallons of gasoline and the excavation of
350 cubic yards of contaminated soil from the site. Additional
gasoline was reported to have discharged to the storm water sewers
at the site. Approximately 800 gallons of the gasoline was
recovered during the first few weeks after the reported spill. No
LPH has been detected in the recovery sump since December 1986;
however, contaminated ground water has been recovered since
remediation was initiated.
GEOLOGY AND HYDROGEOLOGY
GEOLOGY
The site is situated within the Inner Bluegrass Physiographic
Province of central Kentucky. This area is characterized by flat
lying to very gently dipping limestones, dolostones and shales of
Ordovician Age. Strata beneath the site consists of limestones and
thin shales of the Lexington Limestone. Members of this formation
show an intertonguing relationship throughout the Lexington area.
The members underlying the site are, in descending order, the
448
-------
SITE
HE
KENTUCKY
0
fc
SCALE (MILES)
Figure 1. Site Location
-------
°YE INJECTI°N P°INT
SOUTH LIMESTONE AVENUE
Ul
o
.LEGEND;
—X—X- FENCE
ABANDONED INJECTION
I BORINGS
/
CHECKERS
FOOD
7
MR. GATTIS
PIZZA
OFFICES
•^
TREATMENT .^
YSTEM LOCATION V
1
Kl
o
If
0
60
SCALE FEET
Figure 2. Site Map
-------
Tanglewood, Brannon and Grier Members of the Lexington Limestone.
The Tanglewood and Grier Members are characterized by thinly bedded
argillaceous limestones with very minor and isolated shale
partings. These limestones are subject to solutional modification
by ground water flow as evidenced by solutional features along road
cuts as well as abundant closed surface depressions in the area.
The Brannon Member is characterized as argillaceous limestone
containing shale and chert beds and solutional modification is
uncommon (Miller, 1967).
A total of 36 test borings were drilled during November 1987 and
twenty three additional borings were advanced by Delta in February
1988 and May 1990. The boring locations are shown in Figures 3 and
4. Material beneath the site consists of lean to fat
unconsolidated overburden with minor amounts of chert and limestone
rock fragments increasing with depth. The depth to the Tanglewood
Member ranges from 8.5 to 16 feet below grade.
Soil was screened with a PID and soil samples were collected from
the 14 borings advanced in May 1990.
Soil screening and analytical results indicate that elevated levels
of petroleum hydrocarbons are concentrated at the soil/bedrock
interface over a large area. Levels appear to be higher in a small
area west of Mr. Gatti's Pizza Restaurant. The approximate extent
of elevated hydrocarbon levels in soil is shown in Figure 5.
HYDROGEOLOGY
Ground water was not encountered during the boring programs,
therefore, there have been no static measurements of ground water
at this location. Flow mainly occurs along the soil bedrock
interface through tiny channels and solutionally enhanced joints
and bedding planes known as the subcutaneous or epikarstic zone.
Ground Water Tracer Study
In order to determine the ultimate pat for ground water and, hence,
contaminant movement, it was necessary to conduct a dye trace from
the areA of the gasoline loss.
After initial reconnaissance on April 15, 1989, the following
observations were made:
1) Directly underlying the site is the Tanglewood Member of
the Lexington Limestone. The Tanglewood is karstif iable,
as evident by many area sinkholes.
2) Most area springs are discharged from the Grier Member of
the Lexington Limestone.
3) Between the Tanglewood and Grier is the Brannon Member
which, although contains some shale beds and chert, is
considered a semi-permeable unit.
451
-------
, DYE INJECTION POINT
SOUTH LIMESTONE AVENUE
PUMP
ISLANDS
ABANDON INJECTION
BORINGS
^\^
^
I
•^-^ <,8-6
g_2 B_7(, B-25,
B-Bn
'/
«?
«^
^
UGEND;
—X—X- FENCE
(1 TCT SOIL BORING LOCATIONS (2/BB)
«G.
8-4
26
B-27
«
1
CHECKERS
FOOD
MART
MR. GATT1S
PIZZA
R
*
H
-32
OFFICES
B-30<»
.B-M B-1
B-29
B-13- B-17-
B-10
B-23a
B-19
LIMESTONE
APARTMENTS
APARTMENTS
B-21,
B-
E
o
60
SCALE FEET
Figure 3. Borings advanced during 1987
-------
SOUTH LIMESTONE AVENUE
1ECQJQ;
—X—X- FENCE
A-1 • SOIL BORING LOCATIONS (SEPTEMBER 198B)
C-l» SOIL BORING LOCATIONS (MAY 1990)
LJ
1
A-3
C-H
SECOND
NATIONAL
BANK
C-4
'
CHECKERS
FOOD
MART
7
/
C-9»
MR. GATTIS
PIZZA
RECOVERY^
TRENCH
>C-2
A-51
A-4
A-6
C-7 I
omcES
A-B
>v C-1 C-6 /^
X
UJ
'k
A-
^
LIMESTONE
APARTMENTS
•
\S \S •*'
K-lf~
s\ s\
«.«.
C-10
[3
o
E
o
60
SCALE FEET
Figure 4. Borings advanced during 1988 and 1990.
-------
SOUTH LJMESTONE AVENUE
1EGEND;
—X—X- FENCE
—— > 20 mg/kg TOTAL BTEX
- —— - 1-20 mg/kg TOTAL BTEX
—.—.._ < 1 mg/kg TOTAL BTEX
SECOND
I NATIONAL
BANK
/
CHECKERS
FOOD
MART
7
/
/
MR. GATTIS
PIZZA
OFFICES
TRENCH
^SEPARATOR
TANKS
LIMESTONE
APARTMENTS
APARTMENTS
60
SCALE FEET
Figure 5. Estimated extent of soil contamination based on soil boring information.
-------
Table 1
Detector Locations Location Codes
Recovery Trench RT
Mr. Gatti's Sump MG
Storm Sewer SS
Storm Sewer at Railroad Culvert RR
Big Elm Country Club BE
Wolf Run WR
Vaughn's Branch VB
Town Branch TB
Hickman Creek HC
McConnell Spring MS
Preston's Cave Spring PS
455
-------
-fc-
Ul
LEXINGTON WEST. KY.
SE/4 GEORGETOWN 15' QUADRANGLE
38084- A5-TF-024
1965
PHOTOREMSED 1887
DMA 4060 n SB-SERIES VB53
IDEALIZED FLOW ROUTE
SCALE 1:24000
1 MILE
1000 20OO 3000 4000 5000
CONTOUR INTERVAL 10 FEET
Figure 6. Dye detector locations and tracer results.
-------
4) Local dip is to the northwest at approximately 20 feet
per mile. There are no mapped faults in the immediate
area of the site.
5) The storm sewer (later used as a dye detector site) which
passes the site and receives water from the separator
tank, ultimately discharges in Vaughn Branch (also later
used as a detector site).
6) Dye detector locations chosen are included in Table 1 and
shown in Figure 6.
Dye Background
Prior to dye tracer studies, an analysis for the presence of
background levels of fluorescent dyes is necessary. Dye detector
sites were chosen (Table 1 and Figure 6). Background detectors
were deployed and analyzed on two occasions. These detectors,
collected on July 14, 1989 and July 21, 1989 indicated the presence
of extremely weak concentrations of fluorescein dye in the storm
sewer and McConnell Spring sites. The dye was determined to be of
insignificant concentrations to negate the use of fluorescein for
the tracer study.
Dye Injection
On July 21, 1989 at approximately 11:00 a.m., fluorescein dye (acid
yellow 73) was injected into a fresh 10 inch diameter soil boring
at the Shop-N-Go site. The location of the boring is shown in
Figure 2. The boring encountered bedrock at approximately 12 feet.
Prior to the boring of the injection well, two additional borings
were drilled approximately 3 and 6 feet from the injection well
(Figure 2) . These wells have a total depth of 9 and 10 feet,
respectively, and would not effectively take a slug of water. The
injection well readily accepted over 100 gallons in a period of
less than 10 minutes. After the initial slug of water, 3 pounds of
fluorescein mixed with 2 gallons of water was injected into the
well. The dye was then followed with approximately 200 gallons of
water.
Dye Recovery
Dye detectors were retrieved at weekly intervals throughout the
course of the study. Dye detectors were eluted in a solution of
potassium hydroxide and isopropyl alcohol and analyzed on a
spectrofluorophotometer. Detector analysis results can be found in
Table 2.
Dye Trace Results
The first sampling point to yield a confirmed positive was the sump
in the basement of Mr. Gatti's. The dye was present on the
detector collected July 27, 1989. The second round of detector
457
-------
SITE
RT
MG
SS
RR
BE
WR
VB
TB
HC
MS
PC
Table 2
7/14 7/21 7/27 8/3 8/9 8/16 8/24 2/1
VW
+V +VB +VB
+VB +VB +VB
W
VW
DETECTOR RESULTS
= dye not detected
VW = very weak
W = possible dye detected, weak
+ = dye detected
+VB = dye detected, very strong
X = detector not collected as past + confirmed
LS = detector lost or stolen
DATE = detector collection dale
X
X
LS
LS
LS
X
X
2/9 9/15 9/26 10/10 10/19 10/27 11/S
XXX X X X X
XXX X X X X
W
- - -
LS
LS LS
-
LS
LS
LS LS
h LS LS
h
W
W
NOTE: Test was conducted 7/14/89 to 11/15/89
-------
collection on August 3, 1989 showed positive at Mr. Gatti's and in
the recovery trench. The rapid movement of the dye to these two
nearby sites demonstrates the relatively high ground water flow
velocity associated with the solutionally enhanced upper bedrock
(the epikarstic zone), which was the horizon of injection.
In the weeks following the initial confirmed positive at Mr.
Gatti's and the recovery trench, there were a few "false
positives". These false positives of fluorescein dye can be
expected when tracing in urban areas since fluorescein in the
coloring agent in many popular anti-freezes. These false positives
are not a problem in a tracer study of this design. Since a large
amount of fluorescein was injected upon the epikarst, one may
expect a rather extended dispersal period over a large lateral
area. Additionally, the detectors were retrieved at weekly
intervals and a positive was not assumed until dye was detected for
at least four continuous weeks at any particular site. Thus, false
positives can be detected and noted accordingly.
There were several false positives during the course of this study.
For example, detectors at Big Elm, Town Branch, and McConnell
Spring showed false positives on several occasions. All false
positives in the above sites were followed with negative readings
in subsequent weeks. The storm sewer also displayed false
positives throughout much of the study period. This site was
adjacent to a road with heavy traffic, and undoubtedly encountered
many urban-induced fluorescein spills. These spills were not great
enough to cause positive readings at the railroad storm sewer a
quarter mile directly downstream.
One focus of this study was to determine the ultimate resurgence of
the injected dye. The detectors collected on September 26, 1989
marked the arrival of the first confirmed positive of the dye at
McConnell Spring and Preston's Cave Spring sites. This first
confirmed positive occurred some 67 days following the dye
injection. The connection of McConnell Spring to Preston's Cave
Spring was first documented by James Rebmann of the Lexington-
Fay ette Urban County Government in January 1988.
Dye introduced at the soil-bedrock interface moved very rapidly
through the epikarst toward Mr. Gatti's and the recovery trench.
The epikarst is noted for its rapid ground water flow velocity and
wide lateral dispersion of recharge. After passing through the
epikarst, it appears that the dye moved very slowly through a
poorly developed conduit system toward McConnell Spring. After
passing through McConnell Spring, the dye resurged at Preston's
Cave Spring.
Ground Water Contamination
Ground water contamination, as detected at Mr. Gatti's and in the
recovery sump has decreased from the presence of free product in
April 1986 to dissolved contamination in the low parts per million
range in December 1989. Fluctuations in contaminant levels have
459
-------
been documented during this time period. Increases can be
attributed to heavy periods of recharge where ground water moving
along poorly developed conduits in the epikarstic zone leach
residual soil contamination. Decreases are marked by dry periods
where precipitation has little effect in leaching residual soil
contamination.
REMEDIATION DESIGN AND IMPLEMENTATION
In order to properly remediate the subcutaneous zone of this
location, an approach different from classical pump and treat was
needed. The design had to be specifically matched with the
geological and hydrogeological conditions of the subcutaneous zone,
satisfy state requirements of remeditaion to background, and
satisfy the third party landowner that every possible avenue to
remediation was being attempted using the fastest technology
available.
When designing this recovery system we had to account for:
contaminated soil above the soil/bedrock interface that
was contributing to contamination;
inaccessible contaminated soil in solutionally enlarged
pore space of the subcutaneous zone;
quickflow of contaminated ground water in poorly
developed conduits of the subcutaneous zone;
observed increases in BTEX concentration in ground water
with an increase in recharge;
available space and property boundaries; and,
the possibility of volatile organic vapors in buildings.
SELECTED TECHNOLOGIES AND APPLICATION
Soil Excavation
Elevated levels of gasoline constituents were identified in an area
of soils west of Mr. Gatti's Pizza Restaurant as outlined in Figure
5. Approximately 1960 tons of this soil was excavated from the
area and transported to the Lexington-Fayette County Municipal
Landfill.
Soil Flushing
Soil flushing was selected for this site for several reasons. The
nature of flow along the soil bedrock interface is likely guick
flow through poorly developed open and soil filled solution
channels in the subcutaneous zone. Levels of dissolved gasoline
components in ground water have shown a sharp rise following
periods of heavy recharge from precipitation. Reinfiltration of
460
-------
treated ground water was designed to mimic a constant recharge to
this zone, thereby continually flushing contaminants from the soil
in the epikarstic zone. Soil flushing is also very cost effective,
provides an option for the fate of the treated ground water and can
increase and enhance ground water flow to the interceptor trench.
An infiltration gallery was designed to accept and reinfiltrate the
volume of discharge generated by the ground water recovery and
treatment system. At this site discharge will be the result of
ground water recovered from a maximum of two continuously
operational recovery sumps.
The purpose of the infiltration gallery is two fold. First
reinfiltrated water serves to recharge the perched surficial
epikarst and move any ground water contamination to the interceptor
trench more quickly- Second, reinfiltrated ground water flushes
residual soil contamination at the soil bedrock interface causing
the leached material to become mobile in ground water for removal
at the interceptor trench.
During the tracer study conducted July 1989, a small conduit was
discovered at the soil bedrock interface at a location shown on
Figure 2. This conduit was easily able to accept at least 100
gallons of water instantaneously. This conduit was also traced by
fluorescent dye to the location of the present recovery trench in
less than 24 hours. The infiltration gallery intersects the
conduit in the epikarst therefore providing constant recharge to
the surficial flow system and constant flushing of residual
contamination from soil.
The gallery is 40 feet long, 3 feet wide and 7 feet deep. The
discharge water is introduced by means of 4" diameter perforated
pipe to allow drainage across the entire trench. The trench is
backfilled with washed gravel to allow for even percolation of the
treated ground water. Two clean-out ports have been placed on the
perforated pipe to remedy any sediment, clogs or biofouling. Two
piezometers were installed at opposite ends of the infiltration
gallery to monitor its efficiency in reintroducing ground water.
Emergency high level floats keep the infiltration gallery from over
flowing if its capacity is exceeded. A schematic of the
infiltration gallery is shown in Figure 7. The gallery was
installed during the second week of January 1991. To ensure the
gallery was hydraulically "connected" to water flow in the
epikarstic zone approximately 1000 gallons of water was introduced
to the open hole after it was excavated. The water quickly flowed
from the gallery and into the subcutaneous zone.
461
-------
,— TREATMENT SYSTEM EFFLUENT IN
/— PIEZOMETER
PIEZOMETER
• CLEANOUT
t
3'
t
X / i
C) i CO • > •• • < ' • ' ' • • ' i '(
/ X
II 1 1 I 1 1 1 I t 1 1 1 1 1 l( }
\
• o
PLAN VIEW
TREATMENT SYSTEM EFRUENT IN
CROSS SECTION
Figure 7. Infiltration gallery design.
Soil Venting
Soil venting was also a selected technology for implementation at
this site for this site for several reasons. Soil venting allows
for remediation of the residual soil contaminants with minimal
disturbance of surface structures and roads. Soil venting is a
cost-effective and field proven technology for remediating soil
contaminated with petroleum hydrocarbons. Application of a vacuum
to the interceptor trench should also increase the hydraulic yield
of the trench and enhance the rate of ground water remediation.
The primary reason, through, was to remove hydrocarbons from soil
is the epikarst that could not be removed by excavation.
462
-------
The soil vapor extraction system consists of a regenerative blower
connected via 2-inch schedule 40 PVC piping to the ground water
recovery/vapor extraction trench. The trench is screened through
the vadose zone to accommodate the vapor removal. Depression of
the potentiometric surface due to pumping of the ground water also
increases the volume of contaminated soil exposed for remediation
by venting.
Soil Remediation System Equipment
Figure 8 presents the layout for the equipment of the soil venting
system which consists of a regenerative blower connected to the
recovery well via the previously described PVC piping, an inlet
filter to remove particulates from the air stream entering the
blower, and a coalescing unit to remove moisture from the air
stream entering the blower. The effluent from the blower will be
discharged to the atmosphere via a 2-inch schedule 40 PVC pipe.
As the liquid level in the coalescing unit reservoir increases, it
activates a sensor which turns off the blower and opens a solenoid
valve in the coalescer, thereby allowing the condensate to gravity
drain into the diffused aeration tanks. A timer in the control
panel reactivates the blower and closes the solenoid valve after
all condensate has drained.
GROUND WATER RECOVERY AND TREATMENT
Ground Water Recovery Equipment
\
The ground water recovery system installed in 1986 at the Lexington
site consisted of a ground water interceptor trench and one
recovery sump. The former interceptor trench was extended and an
additional recovery sump be added as shown in Figure 9. The
ground water recovery system consists of two total fluids recovery
pumps installed in the recovery sumps of the enlarged interceptor
trench. The pumps intercept and pull contaminated ground water
towards the interceptor trench. A cross section of the trench
construction is shown in Figure 10.
The static depth to ground water in the recovery sump installed in
1986 was approximately 8 feet below grade. Both recovery pumps are
positioned at a depth of 11 feet. The depression pump on/off
floats are positioned at 11 feet and 10 feet below grade
respectively in order to establish a maintained drawdown of
approximately 2 feet in the interceptor trench. The operation of
the total fluids pump is controlled by high and low level on/off
sensors in order to facilitate the efficient removal of
contaminated ground water from the soil bedrock interface.
Ground Water Treatment Equipment
The ground water treatment technology implemented at this site is
a three step process:
463
-------
VACUUM GAUGE-
MANHOLE COVER
VACUUM ENHANCED RECOVERY
WELL
COALESCER
-VENT STACK
REGENERATIVE BLOWER
AIR F10W
BLOWER
SUPPORT
DRAIN LINE TO
DIFFUSED AERATION TANK
NOTE:
THIS DRAWING IS NOT TO SCALE
Figure 8. Soil venting system schematic.
-------
SOUTH LIMESTONE AVENUE
SECOND
NATIONAL
BANK
1
CHECKERS
FOOD
MART
/ L
/
/
v/
i /
MR. GATTIS
PIZZA
Ld
<
G.
PARKING
PARKING
TREATMENT SYSTEM
BUILDING
RECOVERY
SUMP
APARTMENTS
PARKING
LIMESTONE
APARTMENTS
LU
3
Q.
APARTMENTS
RESIDENCE
FENCE
Figure 9. Recovery trench and treatment system location.
465
-------
APPROX.
4'-5'
APPROX. 1'-2'
•|1 RISER
NATIVE BACKFILL OR
GRAVEL UNDER PARKING AREA
5'
\- 2" SLOTTED PVC
30'-0" •
30'-0"
d
I
APPROX. 1'-2'
COARSE GRAVEL
-#3 RISER
RECOVERY WELL
8" PVC
r POLY-PLASTIC UNER
\ (IMPENETRABLE)
8' —
SCREENED
5'-0"
S'-O"
BOTTOM OF TRENCH
4" PERFORATED DRAIN PIPE ''TiT'''TIT1
NOTE: DRAWING IS NOT TO SCALE.
ALL PVC IS SCHEDULE 40
Figure 10. Recovery trench construction details.
-------
1) particulate filtration as a pretreatment step to remove
suspended solids;
2) LPH/water gravity separation as a pretreatment step to provide
for the flotation separation of LPH and flow equalization; and
3) diffused aeration to remove the dissolved volatile organic
compounds.
Pretreatment Process
The ground water pretreatment system consists of an in-line
particulate filter followed by an LPH/water separator tank. Figure
11 presents the treatment system lay out.
The ground water recovery pumps transfers ground water through an
in-line filter to a 1000 gallon LPH/water separator tank to allow
for flotation separation of the LPH from the ground water and
particulate settling. A totalizing flow meter installed at the
intake of the separator tank continuously records the amount of
ground water and LPH that has been pumped.
Effluent from the separator tank flows by gravity through a contact
chlorine chamber containing solid calcium hypochlorite tablets.
This process reduces biological build up in the treatment system.
The chlorination step inhibits xthe growth of biological
microorganisms that could potentially cause the lines in the
aeration tanks to become clogged.
Diffused Aeration Equipment
Effluent from the contact chlorine chamber will flow by gravity
into a channeled diffused aeration tank. This tank will be four
feet wide by six feet long by two feet deep. A 2 hp, 230V, single
phase pressure blower will introduce air into the aeration tanks
through perforated PVC pipes to facilitate stripping of the
dissolved volatile organics.
Effluent from the diffused aeration tanks flows by gravity into a
120 gallon transfer tank. This transfer tank is equipped with a
submersible electric sump pump with a nominal operating flow rate
of 10 gpm. The sump pump moves the treated ground water from the
transfer tank to an infiltration gallery located upgradient from
the interceptor trench.
Soil Venting and Ground Water Remediation System Security
The soil venting and ground water remediation equipment are housed
inside a pre-fabricated building located at a location shown in
Figure 9.
467
-------
AERATION TANK-
I I TREATED EFFLUENT TO
I I INFILTRATION GALLERY
I I
I I
I I
JL
EXHAUST VAPORS
CHIOR1NATOR
EMERGENCY
HIGH LEVEL
FLOAT
TRANSFER
TANK WTO
SUUP PUMP
LEVEL SHUT-OFF PROBE
D
CCTBEBVATON VENT
1000 GALLON
SEPARATOR TANK
o
OBSERVATION PORT
SA14PUNC -
PORT
BALL VALVE-
^-FLOWUCTER
I INLINE FILTER
AIR INLET
FROU RECOVERY
WELL
Locxm
SATE
r— ——•———~—-\ FROU PRODUCT
I «==_.» RECOVERY WEU.
Figure 11. Treatment system layout.
468
-------
RESULTS TO DATE
Analytical results from water samples collected from the former
emergency sump and the current ground water recovery system are
plotted against time in Figure 12, illustrating a wide variation in
contaminant concentrations since recovery operations have begun.
Maximum dissolved concentrations of approximately 15,000 ug/L total
benzene, toluene, ethylbenzene and total xylenes (BTEX) have been
recorded. Although flow readings do not exist for the early
recovery system, a general correlation between recharge events and
increasing contaminant concentratons has been made. The emergency
recovery sump was taken off line from July 1989 through February
1990 to conduct the dye tracer study-
TOTAL BTEX CONCENTRATIONS (ug/L)
SHOP & GO FOOD MART
PQ
05/07/90 11/23/90 06/11/91 12/28/91
08/15/90 03/03/91 09/19/91
DATE
Figure 12.
Total BTEX concentrations in treatment system
influent since May, 1990.
The current ground water remediation system was monitored monthly
to verify that the system is operating properly. Sample ports are
provided at several locations along the treatment system. Water
from the treatment system influent and effluent sampling ports are
collected monthly, and analyzed for volatile organic constituents.
469
-------
from the treatment system influent and effluent sampling ports are
collected monthly, and analyzed for volatile organic constituents.
A totalizing flow meter is installed to document the total volume
of ground water being pumped from each recovery sump.
Plots of flow volume over time from both recovery sumps are
presented as Figures 13 and 14. These illustrations verify the
flashy nature of flow in the subcutaneous zone. Flow rates vary
from recovery well 1 from almost negligible to approximately 5500
gallons per day. Flow variations from recovery well 2 are even
more dramatic ranging from 3000 to almost 25,000 gallons per day.
This information also correlates with periods of observed heavy
rainfall events through that time period.
RECOVERY WELL 1
FLOW RATE (GALLONS PER DAY)
6000
03/03/91
05/17/91
07/31/91
10/14/91
DATE
Figure 13.
Flow in gallons per day from Recovery Well 1 since
system start-up.
A plot of BTEX concentration in the treatment system influent since
the currently system was put on line included as Figure 15. Figure
15 shows that the concentration of total BTEX varies widely but the
maximum measured levels have deceased since remediation began. The
peak concentrations are also roughly similar to maximum times of
flow or maximum recharge. This may suggests that more
contamination is being leached from soils, mobilized and removed
during times of heavy recharge and that the constant recharge from
the infiltration gallery may aid in expediting this removal
process.
470
-------
RECOVERY WELL 2
FLOW RATE (GALLONS PER DAY)
03/03/91
05/17/91
07/31/91
10/14/91
DATE
Figure 14.
Flow in gallons per day from recovery well 2 since
system start-up.
TOTAL BTEX CONCENTRATIONS (ug/L)
SHOP & GO FOOD MART
10000
U
CJ
1000
03/03/91
06/11/91
09/19/91
04/22/91
07/31/91
11/08/91
DATE
Figure 15.
Total BTEX concentration in treatment system
influeratLnce system start-up.
471
-------
The performance of the soil venting system is illustrated in Figure
16 as a plot of soil vapor concentration since the system was put
on line. Soil vapor concentrations were at a maximum when the
vacuum blower was first started and have decreased to non-
detectable since that time. The reason for the quick decrease is
believed to be removal of BTEX laden soil vapor from the immediate
area of the vapor extraction equipment. The system has since been
turned off to allow equilibration of soil vapor with BTEX on soil
particles. The system will then be put back on line and a decreased
but similar pattern of soil vapor removal is expected.
SOIL VAPOR EHRACTION SYSTEM
PHOTOIOMZATION DETECTOR READINGS
250
03/23/91
05/22/91
07/21/91
09/19/91
DATE
Figure 16.
Photoionization detector readings of the soil vapor
extraction emissions since initiation.
CONCLUSIONS
The likelihood of a release of contaminants directly into the
conduit network of a karst aquifers is low. Most releases occur
within the overburden and move into the subcutaneous zone of the
aquifer thereby decreasing the quickflow components and increasing
residence times in the intergranular overburden porosity and the
poorly developed conduit system of the subcutaneous zone. The
results of this effort have been presented as a site specific
approach to remediate the subcutaneous zone of a karst aquifer.
The remediation design was not arrived at using classical pump and
treat, intergranular flow manner approach to the problem and set
472
-------
aside the notion that this karst aquifer was not "remediateable"
as if it were conduit network contamination. A remediation system
was designed around the specific characteristics of the surficial
subcutaneous karst aquifer to be effective as a remediation tool in
this setting. The use of excavation, soil flushing, soil venting
and pump and treat has been shown effective at removing soil and
ground water contamination from the s ubcutaneous zone as a
decrease in maximum contaminant concentration over time. The
system also appears to be working as designed by mimicking a
constant recharge event thereby leaching residual soil
contamination from the subcutaneous zone for removal by the
interceptor trench. The system is currently operating as designed
and data being collected for a future submittal concerning its
efficiency and performance.
REFERENCES
Miller, R.D., 1967, Geologic Map of the Lexington West Quadrangle,
Fayette and Scott Counties, Kentucky; United States Geological
Survey Geological Quadrangle Map - GQ - 600.
BIOGRAPHICAL SKETCH
Scott A. Recker, a native of Cincinnati, completed a Bachelors
Degree in geology at the University of Cincinnati in 1985 and
received a Master of Science in hydrogeology from Eastern Kentudy
University in 1990. Scott is 'a member of the National
Speleological Society and his research^ centers around ground water
flow in the subcutaneous zone of karst aquifers. He is currently
residing in Charlotte, North Carolina and is employed as a
consulting hydrogeollogist for Delta Environmental Consultants,
Inc.
Scott Recker
3300 Circles End Road
Charlotte, North Carolina 28226
(704) 541-8191
473
-------
PETROLEUM HYDROCARBON REMEDIATION OF THE
SUBCUTANEOUS ZONE OF A KARST AQUIFER,
LEXINGTON, KENTUCKY
Scott A. Recker
Delta Environmental Consultants, Inc.
Charlotte, North Carolina
1. Has the remediation effort been evaluated at the spring?
No, the spring showed non-detctable amounts of petroleum
hydrocarbons during two sampling events. The spring has also
been documented as the ulimate resurgence for a ground water
basin which encompasses half of the City of Lexington. Remedial
evaluations at the spring would therefore be unsuccessful due to
the large amount of dilution and the presence of other potential
petroleum hydrocarbon constributors.
474
-------
CORRECTION OF BACKGROUND INTERFERENCE AND CROSS-FLUORESCENCE
IN FILTER FLUOROMETRIC ANALYSIS OF WATER-TRACER DYES
C.C. Smart,
Department of Geography
University of Western Ontario
London
Ontario, L8S 4K1
Canada
P.L. Smart
Department of Geography
University of Bristol
Bristol, BS8 1SS
England.
ABSTRACT
Filter fluorometry is a widespread analytical technique for
the quantitative analysis of fluorescent tracer dyes in
hydrogeology. However, mixtures of tracers and fluorescent
background materials may result in ambiguous results unless steps
are taken to isolate the tracer materials> Analytically it may be
possible to chemically suppress or mask interfering substances, or
to select filter combinations capable of excluding those wave-
lengths frustrating analysis.
Post-analytically, an algebraic solution is possible,
providing care has been taken to run calibration standards
({y}=l,2,3) through all filter arrays ({x}=A,B,C) to obtain a
matrix of fluorescence readings F* The matrix of cross- fluor-
escence calibration coefficients {k*} = F*/k* can be determined,
where z= A, B, C for y=l,2,3 respectively. By definition, the
positive diagonal kx values will always equal one. Off-diagonal
k* may be zero (indicating absence of cross-fluorescence), or non-
zero (indicating cross-fluorescence and the need for correction).
Note that exact values of k* require empirical determination,
because they depend on characteristics of dye stock, individual
filters, and light sources.
475
-------
Table of Cross-Fluor-
escence Calibration Coef-
ficients {k*}
Filter
Set
(x)
A
B
C
Dye (y)
1
kA
K1
kB
K1
kc
K1
2
VA
K.7
kB
K?
kc
K?
3
kA
K3
k8
*h
kc
K^
If no cross-
fluorescence
If tracer dyes 1, 2 and 3 are matched to filter sets A, B and
C respectively and calibration allows k*=k|=kij=l, then an approxi-
mate but general solution to the cross-fluorescence problem can be
derived. The actual fluorescence of dye 1 in a sample (FA) can be
expressed in terms of measured fluorescence of that sample in all
three filter sets (FA,FB,FC), and the coefficients {k*} :
- FB(k* -
_L
C-j jCii
(10a)
The relationship can be further simplified for two-component
mixtures, to produce the common correction equation given by Kass
(1967) :
(11)
Where cross-fluorescence coefficients
further simplification is possible:
are small numbers,
(12)
The model will be robust for all fluorescent materials with a
consistent spectral signature. This may allow correction for
background interference from organic materials, provided that
background variations are due to fluctuations in spectral amplitude
rather than shifts in spectral character. The background "stan-
dard" used to determine k£ must be entirely free of tracer dye, but
can be of arbitrary concentration.
The technique will not allow cross-fluorescence correction in
waters containing fluorescent contaminants closely mimicking dyes,
nor accommodate dramatic shifts in background signature. These
problems may be particularly acute in elutant from activated
charcoal, which must remain a semi-quantitative technique.
Where precise interpretation is required, a spectrofluorometer
remains the instrument of choice. The same method may be applied
to correct for cross-fluorescence.
476
-------
INTRODUCTION
Fluorescent dyes are an invaluable tracing agent, widely used
in surface and groundwater studies. They are inexpensive, benign
(Smart 1982) and readily detectable at concentrations below the
parts per billion (ppb) level. It is also possible to use several
dyes simultaneously, allowing several components of a hydrological
cycle to be monitored under a given set of flow conditions.
However, in practice, multi-tracing can be complex, expensive and
possibly ambiguous, especially if tracers can not be completely
separated from one another (cross-fluorescence) or from natural
background materials. In practical applications, this will
sometimes be seen as exact, but highly attenuated mimicking of a
primary breakthrough curve in a neighbouring spectral trace. Where
the two tracers have been injected simultaneously, it may be of
considerable importance to determine if the attenuated trace is
real,or simply the result of cross-fluorescence.
The problem is illustrated in Figure 2 where cross fluor-
escence from Dye I has caused an increase in the apparent fluor-
escence of Dye 2 (broken line) . The exact magnitude of the problem
will depend on the overlap of the spectra and the relative
concentrations of the two tracers. Also note that cross-fluor-
escence is "non-commutative"; despite the interference of dye 1
with measurements of dye 2, measurements of dye 1 are little
influenced by concentrations of dye 2.
The problem of two-component cross-fluorescence in spectro-
fluorometry has received some attention in the European literature.
The present paper first reviews a correction methodology and then
provides an initial attempt to formalise the problem for multi-
component cross-fluorescence, with particular emphasis on filter
fluorometers.
There are two strategies to solving cross-fluorescence
problems. Pre-analytical treatments are designed to mask or
suppress fluorescence of interfering tracers in mixed samples
without altering the fluorescence of a selected tracer. Behrens
(e.g. 1982) has derived an array of such treatments such as
filtration and pH control allowing spectacular segregation of dyes
prior to analysis. Further segregation may be gained using
chromatographic separation techniques (e.g. Rochat et al. 1975).
In general, these techniques require relatively sophisticated
processing and although automated techniques may be adopted, remain
prohibitively expensive for most environmental investigations.
The alternative strategy where moderate cross-fluorescence
occurs in mixtures of two or three tracers is post analytical. It
involves application of an algebraic correction (Kass 1967) based
on the theory of mixtures. This requires very simple cross-
calibration of the fluorometer using single dye standards. Cross-
calibration is best performed at the time of analysis, but in
emergencies can be done some time subsequent to the analysis if
necessary. The method is straightforward and inexpensive, and
should be adopted in any multi-tracer test.
477
-------
n M) P) M
O M
MOW
M g ti w
fD fD O
Q cn o y
rt H- rt fD
o 0)
I rt
H> H-
H1 O
O
M
O 0)
g IQ
O (D
W
cn en
I ft (D
HI o rt p)
M 3 hj g
c a H-
O 0) O O
(0 O- 0>
cn w
Hi
-fc-
^J
OO
0) s;
(D ro o
P- O
rt en g
»•• -a
(0 O
3
n 1-3 n>
B) £T 3
I- (0 rt
H-
o
o
cn
•8
o
ft
O
80
70-
60H
CD
^
0)
O
C/)
0)
30-
20-
10-
CALIBRATION
Dye 2
h
B
A
B Wavelength
-------
rt < i-3 rt rt cn N)
M Q> D- 3" *; T) .
C I-1 (D (D O (D
C O CO
?&3
°8S
83s
n
M 0)
C rt
O H-
c H n o
>-( 3" O
< to 3 a
CD ft> h-
cn tr rt 0)
If t-t iQ
rt o H-
(D 0)
3
3 O
a c
(D l-(
^ <
M fD
o p>
3
0)
3 O
0) H,
en o
1 3 3 O (D
*Q a MI en
H- o
en o o CD
Ti D) 3
ID rt *: o
T3 O
o rt
(D en '
3-a'
rt (D
en o '
• rt JT n fo OJ m
•1 2 rt m rt
» * ^ fD en
M ^ 0) rt i-t T3
— y a>
p" ^ o ro en o
ro ™ HI a> rt
H- £' 3 3 i-i
iQ ti rt fD >a C
3- 2 31 0) H- 3
rt .v fD en fD
en m e o
" rt i-( O tT
rt « fD O rt
O 2 O a 3 0>
rt H-
O "> a rt 0) 3
cr —^ n M-fo
rtj-31 fD 0) 3 a
cu en n M-
H- • fD 3 H)
3 - iQ t-!
o
3
SAMPLE
A
B Wavelength
-------
CROSS-FLUORESCENCE: SPECTROFLUOROMETERS
A fluorescent material absorbs light over a characteristic
waveband and re-emits light at a characteristically lower waveband.
In fluorometric analysis a sample is irradiated (excited) with
light of wavelength within the absorption spectrum of interest, and
emittance measured at an exclusive longer wavelength. The key to
effective measurement is provision of suitable, but mutually
exclusive excitation and emission wavelengths. In spectro-
fluorometers, this is achieved by very precise control of the
wavelengths using diffraction gratings.
A spectrofluorometric analysis is usually undertaken by
scanning with excitation and emission wavelengths a fixed distance
apart (typically " ^m) . This produces a typical spectral trace of
overlapping peaks where peak position is characteristic of
particular fluorescent materials and peak height its concentration.
Unfortunately, fluorescence is additive so that if the tail of one
spectral peak underlie the peak of another, the peak height is no
longer a simple measure of concentration of a single material.
This is the problem of cross-fluorescence. In worse cases, the
overlap may result in complete masking of smaller peaks in the
flanks of more dominant ones. The problem becomes more acute at
shorter wavelengths when the very broad fluorescence of naturally
occurring organic materials may underlie tracer peaks.
In spectrofluorometry, the problem of cross-fluorescence is
usually readily apparent in a compound form to the spectral trace.
For two component mixtures, it may be corrected using the method of
Kass (1967). As there is no formal statement readily available on
the correction procedure, it is summarised as follows. First,
single tracer standards are run, noting the height of the peak and
shoulder at the characteristic peak wavelengths for both dyes. A
matrix of trace height readings is thus obtained for each dye (1,2)
at each diagnostic wavelength (A,B). For example, Figure 1 gives
an array as follows:
Calibration Measurements
Wavelength
A
B
Dye
1
h*=39.9
1^=5.40
2
h$=0.00
h^=53.2
The convention introduced here is that the superscript (A,B...)
indicates the wavelength(s) used, and the subscript (1,2,...) the
dye of interest.
Cross-fluorescence correction factors are the ratios of
spectral readings to peak height. They are thus independent of
concentration, and are obtained by dividing each column by the
measurement obtained for the matched wavelength-dye pair (Dye 1-
480
-------
wavelength A; dye 2-wavelength B). Thus,
Cross Fluorescence
Calibration Coefficients
r vx j
4Kvl
Wavelength
(x)
A
B
Dye(y)
1
k*=i.oo
ki|=0.135
2
k*=0.00
kjHl.OO
The cross-fluorescence calibration factors are generalised as
{k*} where x indicates the wavelength (normally the spectral peak
for a dye), and y indicates the dye of interest. Note that 0 <
k* < 1, the positive diagonal members will always be unity.
Off-diagonal values will be zero where there is no cross-fluor-
escence (eg. k* above), and non-zero (usually small numbers) when
cross-fluorescence occurs. Ka'ss (1967) and others (e.g Behrens
18982) employ a and 6 to indicate these correction factors, but
their derivation may not be clear and their meaning can be
ambiguous.
Figure 2 provides an example of a sample analysis, where hA
and hB are measured peak heights from a sample and other terms
are defined above. The following estimates of actual peak
heights h1 ^ may be obtained by
i-kfkf
20.0 - 0.0 x 13.4
1 - 0.00 x 0.102
=20.0
(1)
i.e there is no cross-fluorescent interference with dye 1,
dye 2
For
hB -
13 .4 - 0. 135 x 20.0
1 - 0.00 x 0.135
= 10.7
(2)
In this case a 20% correction has been applied to the raw data.
A formal derivation of this simple case is developed below.
CROSS FLUORESCENCE: FILTER FLUOROMETERS
In contrast to spectrofluorometers, filter fluorometers employ
glass and gelatine colour filters and light sources (collectively
referred to as a "filter set") which are carefully selected to
optimise fluorescent sensitivity, although the spectral width of
both absorption and emission wavelengths are much wider and less
discriminating. The problem of cross fluorescence is thus more
likely to occur, and will be far less evident to the casual
analyst. For this reason, it will always be worthwhile obtain-
ing cross-fluorescence calibration coefficients if more than one
481
-------
fluorescent tracer is likely to be encountered in a tracer test.
Measurement of fluorescence using a filter fluorometer is
shown in Figure 3 which may be contrasted to Figure 1. The
filter cut-off is shown to be unrealistically sharp, this simpli-
fication in no way invalidates the analysis which follows.
A filter fluorometer integrates across a broad spectral band to
obtain fluorescence. It can be seen that Dye 2 now has a small
overlap with spectral band A, in contrast to the complete exclu-
sion possible with a spectrofluorometer.
Despite the increase in cross-fluorescence, the calibration
and analytical procedure is identical to that developed above.
First, calibration standards (of arbitrary concentration) are run
through both filter sets to get a matrix of fluorescence readings
Calibration Measurements
{F*}
Filter Set
A
B
Dye
1
FA=241.
FB=45.9
2
FA=1.1
FB=260
The calibration measurements are divided by the column matched
pair (FA and FB) to generate the correction factor matrix {k*}.
Cross Fluorescence
Calibration Coefficients
te\
Filter Set
(x)
A
B
Dye(y)
1
kA=1.00
kB=0.191
2
kA=0.004
kB=1.00
These factors are then employed on sample analyses to correct for
any cross-fluorescence. Figure 4 provides an example where FA
and FB are 313 and 196 respectively. Substituting FA and FB for
hA and hB in eguations 4 and 5 above gives estimated sample FA and
FB of 312 and 136 respectively, with -0.2% and -30% change from
the measured fluorescence. Fluorescence measurements may be
converted to concentrations in the usual manner.
Colour filters, light sources and electronics are specific
to each filter fluorometer. The same idiosyncracy applies to dye
stocks, even from a single batch if poorly homogenised. It is
not possible or desirable to transfer calibration data between
machines, components, dyes or over time. A final caution con-
cerns the high temperature coefficient of rhodamine dyes (Smart
and Laidlaw 1977). Where sample temperatures differ from the
temperature of the standard used in calibration, temperature
measurements will be necessary on both A and B analyses, and the
482
-------
OO
LO
o> o g w cj
3 Q) 0)
Q- rt X" i-t)
(D
c o 1 g
w o i-< i a
ro M
(D CO
W rt c
ft O i-l o
fl> fl> O
3
D
X 3 3 6
ft rtiQ g
fD (D iQ
HI ^1 i-J rt n
O (D MO)
i 01 o p- a
~ o
•O 0
0> H)
0) rt
M§
en
P- O
CO O
'I
0) W
H-
M O
W 3
o a (-(
HIS; n>
n> cn
rt en o
a>
a>
O 0)
Q) 3
M a
H"
CT NJ
ft 3 (D
3 ff (D
O (D 3
0 O
O ^
X O
M
O W
M) I
f-h 0) 01 M
H-H -a c
I-1 (D (t> O
rt o> o n
fl> W (-•• fl>
!-( HI cn
TJ (-• O
en*" o a>
rt
cn
• 3
• n
(D
H-
3
80
70-
60-
0
CALIBRATION
Dye 2
Dye 1
Waveband
-------
00
ft 0) w *>.
H- 3 0) •
O Q. 3
3 n in
*i M o
rt " n> sr
o ' ID
n 3
rt ^ rt H-
P) ID £U O
H- ,- H-
3 § 3 a
2 P- H-
rt ^3 0)
a o>
*< 3
(D
tn o
Ml
(D
cn
s»
ai
rt
0) rr
3 (D
a
Mi
£•
O
3
>Q 2
H-
(D O
n> o>
3 3
ft 0)
e O^
r 3 0)
H' |-J H-
S^"
m n o
n ^ ""
2 n> PI
2 a
m Q)
rt
CD 20^
U
C
CD
U
00
CD
L_
O
SAMPLE
Dye 1
Dye 2
Waveband
-------
correction factor applied to the estimated fluorescence of the
rhodamine dye.
THEORY OP CROSS FLUORESCENT CORRECTION
The methods above have a theoretical basis which can be
expanded to deal with a more general case of three dyes (1,2,3)
and filter sets (A,B,C).
Calibration of three dyes through three filter sets will
give a 3X3 matrix of fluorescence readings, which divided by the
matched dye-filter pair for the column will yield the following
general matrix of correction factors {k*}:
Table of Cross-Fluor-
escence Calibration Coef-
ficients {k*}
Filter
Set
(x)
A
B
C
Dye (y)
1
kA
K1
kB
K1
kc
K1
2
kA
K?
k8
K?
kc
x?
3
kA
K7
kB
KT
kc
KT
If no cross-
fluorescence
The total (i.e. measurable) fluorescence of a mixture of three
cross-fluorescent dyes in a particular filter set is the sum of
the matched dye fluorescence, plus the other dye fluorescences in
their matched filter set, corrected with the appropriate cross-
fluorescence calibration factor. Thus,
FA =
(3)
(4)
(5)
The quantity of interest is the positive diagonal F*, the
actual fluorescence due to a particular dye in its matched filter
set. Re-expressing 3-5 gives
(6)
485
-------
=F* - JtA - *BFC (7)
= Fc - JfcT - *CFB (8)
It is necessary to remove unknown terms from these express-
ions by substitution. Taking (6) as an example, substituting (7)
for F| and (8) for F3, collecting terms and then resubstituting
gives the following general expression
A - - 32 - *
- Fc(k
FT =
ABAC
1 ~~ JCp/C- ~ /C-7/C*
The product of three off -diagonal k* will be negligible.
Ignoring such terms (collected in the lower line of (9)) not only
simplifies the expression, but leaves the required fluorescence
expressed in measurable and known terms.
.*}FLit ~~ jV-jjC*
The complementary equations to solve for the true fluorescence of
the other two dyes are as follows:
F* -
J- VC « -«^p "^ ?
(lOb)
_ ,C B _ B C _ ^C, A
2 ^' V 2 1 2^ (IOC)
-L /C i -K-z "^? "^
In two component mixtures (e.g. dyes 1 and 2) all terms with
superscript C or subscript 3 will equal zero, giving
In many cases the product of two k£ will be negligible, thus
allowing a very simple cross-fluorescence correction.
486
-------
It is unlikely that a full three-component cross-fluor-
escence correction will be necessary where widely different dyes
and filter sets are employed (e.g. the Red, Green, Blue suite
outlined by Smart and Laidlaw, 1977). It may be possible to
separate dyes previously rejected because of excessive cross-
fluorescence, although no experiments have been tried as yet.
The most common application will be for Rhodamine WT interference
with fluorescein sodium (i.e. Uranine, CI 45350, Acid Yellow 73).
Very high rhodamine concentrations may mask a weak fluorescein
peak. Alexander (1991, pers comm.) reports just such an occur-
rence which was only detected following correction for cross-
fluorescence.
BACKGROUND CORRECTION
The single greatest problem in contemporary fluorometric
tracing is that of natural background fluorescence. Although
seldom a serious problem with "red" dyes, it is a common diffi-
culty with green dyes, and especially acute with blue dyes. This
consistent wavelength dependent sensitivity to background inter-
ference indicates that background has a characteristic spectral
signature (e.g. Smart et al. 1976). Smart and Laidlaw (1977) and
Smart and Friederich (1982) have employed this property to
establish the ratio of background fluorescence in dye-free water.
Significant departure from this line is an indication of a
positive sample. A critical displacement of two standard devi-
ations from the line was suggested, although a statistical
confidence band might provide a more objective criterion. The
technique will fail if the tracer dyes are present in a similar
ratio to the background fluorescences.
The viability of the ratio technique suggests that it is
possible to correct for background interference with red and
green dyes by treating the background as a third dye component.
A third filter set, sensitive to background (an ultra-violet:blue
combination will be most suitable) constitutes the C waveband
(Figure 5) . A dye-free background sample is essential to cali-
brate the method, but the concentration for all cross-fluor-
escence calibrations is arbitrary, so composition is less import-
ant than stability and representativity- Fluorescence readings
are taken using the three standards in all three filter sets to
give a 3x3 matrix {F*}.
487
-------
00
00
H> Q, (Jl
3
0)
Hi ft
C (D
O 01 W
l-i O
(D 0) 3"
cn 3
o
(D
O
(D C O
O
tn o> H-
01 0>
M O *Q
0) (D l-l
cr 3 01
fo rt g
MCT o
(D (M Hi
d O
• ^i H>
W O ft
m c a>
n> 3 M
a
ft • Hi
i>
O
a i-! o
(D O H)
rt cn
Q) 0) rt
*:
O
CALIBRATION
Background
o
B Waveband
-------
An example based on idealised data can be drawn from figure
5 to give
Table of Calibration
Measurements {F*}
Filter
Set
(x)
A
B
C
Dye (y)
243
44.1
4.75
2
1.50
269
0.00
3
85.2
6.53
238
Division of each row by its positive diagonal element will
give the {k*} matrix.
Table of Cross-Fluor-
escence Calibration Coef-
ficients {k*}
Filter
Set
(x)
A
B
C
Dye (y)
1
1
0.182
0.0195
2
0.0056
1
0.0
3
0.358
0.0274
1
Taking an imaginary example (Figure 6) of FA=292, FB=117 and
Fc=340 and applying these values with {k*} into equations lOa,
lOb, and lOc gives F*=169, F*=76.9, and F^=337, corrections of
-42, -34 and -0.9 percent respectively. It is clear that the
magnitude of the background problem can be severe and probably
accounts for many cases of excess dye recovery estimations.
The cross-fluorescence method of background correction is
the best developed to date, although it reguires considerable
analytical commitment to provide all the necessary data. The
method is currently only theoretical and will require consider-
able testing and verification before its reliability can be
assured. The simplest empirical indication of an error in
application is the estimation of negative fluorescence. The key
assumption is that the spectral signature of the background is
steady and changes only in amplitude over time. The influence of
spectral shift is unknown, but may not be too significant if a
sufficiently broad wave band is employed in characterising the
background signature. There is very little information available
on the pattern of background variation between springs and with
time at a single site. Quinlan (1991, pers. comm.) suggests that
variations between springs are greater than variations at a
single spring. This means it would be necessary to establish a
{k*} matrix for each spring if local stability were assured.
Tracer samples eluted from activated charcoal detectors
often exhibit the most serious background problem. Unfortunate-
ly, the concentration and exchange of organic molecules on tracer
489
-------
cr e rt oi
D) CO 3" D)
n (o (D 3
o
O 3
O 0)
3 rt
rt H-
d ^
H- 01 2 o> o
3 01 5 H-
rt I 2 3 &
(D H> £ I-1- H-
t~h C ^ ^ ^Q
(D O rt ^
(D (D
3 01
O O
(D (D
• 3
rt
5 « S
3 o
o
S D- ">
(D
o>
rt
tr
V^ A "-"
O, (I) O. H)
CD 2 *-•
M g B) C
ctgtr?
m
a
n
o
o
rt
o
n
01
o
ff
vQ
01
C O
' O> D)
O O
' O M
3 ><
en
3 01
(D
3 O
-rtH,
01
(D
(!) ... O
hj ft '•
(D
•< (D
i-(
50
40-
(D
U
C 30
0)
O
(f]
(U
020-
10-
SAMPLE
Background
C
FA \
\ Dye 2
B Waveband
-------
samples eluted from charcoal detectors creates dramatic vari-
ations in background composition and resulting signature (Quinlan
pers comm.), so that the cross-fluorescence method will not be
applicable to this particular problem.
Experiments are currently under way to test and evaluate the
cross-fluorescence methodology. The results and conclusions
will be reported following this completion of this work.
CONCLUSIONS
Cross-fluorescence is a potentially serious problem in the
analysis of fluorescent tracers, especially with filter
fluorometers. Cross-calibration should always be undertaken
during analysis to determine the extent of the problem and to
allow post-analytical correction if necessary.
The three-component general theory of cross-fluorescence as
presented can be extended to more complex mixes with considerable
spectral overlap, providing care is taken to validate the
approach. The method promises to provide a useful operational
correction for background interference, but substantial field
experiments will be necessary to assess the reliability of the
method.
REFERENCES
Behrens, H., 1982. Verfahren zum qualitativen Nachweis von neben-
einander vorleigenden Fluoreszenztracern. Beitrage zur
Geologie der Schweiz- Hydrologie 28 I 39-50.
Kass, W. , 1967. Erfahrungen mit Uranin bei Farberversuchen.
Steirische Beitrage zur Hydrogeologie, Jg 1966-67. 123-135.
Rochat, J. J. Alary, J. Molinari, R. Chariere, 1975. Separations
physicochemique de colorants xantheniques utilises comme
traceurs en hydrologie. Journal of Hydrology 26 277-293.
Smart, P.L., 1982. A review of the toxicity of 12 fluorescent
dyes. National Speleological Society Bulletin.
Smart, P.L., B.L. Finlayson and C.M. Ball, 1976. The relation of
fluorescence to dissolved organic carbon in surface waters.
Water Research 10 805-811.
Smart, P.L., and H. Friederich, 1982. An assessment of the
methods and results of water tracing experiments in the
Gunung Mulu National Park, Sarawak. Cave Science 9 100-112.
Smart, P.L., and I.M.S. Laidlaw, 1977. An evaluation of some
fluorescent dye for water tracing. Water Resources Research
13 p.15-33.
491
-------
492
-------
Session VI:
Ecology of Caves and Karst Terranes
-------
494
-------
ASSESSING GROUNDWATER QUALITY IN CAVES
USING INDICES OF BIOLOGICAL INTEGRITY
Thomas L. Poulson
University of Illinois
Chicago, Illinois
ABSTRACT
The states of populations and communities give an integrated record of past and
ongoing pollution whereas toxicity Dioassays and chemical analyses may detect
pollution occurring only at the moment of sampling. Of the population and
community biocriteria that are derived from field data, indicator species,
diversity indices, and community similarity metrics are either totally flawed or
have more disadvantages than advantages. The preferred biocriterion is an Index
of Biological Integrity (IBI) which combines habitat data with population and
community data. In caves there are few species and so most must be considered.
For fish, salamanders, and crayfish all individuals are located on a map, measured,
and examined for reproductive condition and lesions. The smaller species and
habitat characteristics are sampled at random stations. Since surface waters
enter caves by different routes, I sample communities from diffuse input
formation areas, from mixed input areas like terminal breakdowns, and from
direct inputs associated with sinking streams, vertical shafts, and backflooding
from surface rivers. Through 30 years of study, mainly in the Mammoth Cave
Region, I have documented or inferred the non-lethal effects on populations and
communities due to homogenizing of habitat structure by si Ration, due to
favoring of facultative over obligate cave species by excess food Input, and due to
compromising reproduction and growth of very long-lived species by
biomagnification of toxins along food chains.
495
-------
BIOLOGICAL MONITORING IS CHEAPER THAN CHEMICAL MONITORING AND GIVES A
MORE INTEGRATED RECORD OF PAST AND ONGOING POLLUTION
Chemical toxicity tests have now been flawed as a single approach because
they have been done on a minority of suspected toxicants, can not assess effects
of bioaccumulation or resulting indirect ecological effects, use only a few test
species, and toxicity of one chemical changes as the overall chemistry of water
changes. Increasing use of whole effluent ("out of the pipe") toxicity tests has
only partly mitigated these problems. Neither chemical-specific criteria nor
whole effluent tests assess impacts of habitat modification, such as water level,
or non-toxic pollutants, such as of sediments and nutrients. Furthermore both
depend on sampling at the moment of pollution input, unlike the case for field
biological criteria.
The composition of ecological communities integrates ongoing and past
pollution over space and time whereas toxicity tests usually only detect pollution
occurring at the instant of water sampling. Yearly biological surveys often
suffice to demonstrate ongoing sublethal pollution or past pulses of lethal
pollution (as with "midnight dumping") whereas these impacts are usually missed
by monthly or even weekly sampling of water for chemical and toxicity tests.
Thus biological surveys should be used as an inexpensive early warning system to
detect the presence of a problem. Expensive chemical analysis and toxicity
testing should be reserved to pinpoint the exact nature of a problem and to develop
discharge limits for particular pollutants. This combination of biological surveys
and chemical tests provides a complementary balance in the protective and
regulatory process.
In karst regions monitoring water quality is more difficult than in surface
waters because of problems of inaccessibility and complications of diffuse flow.
Quinlan and Alexander ( 1987) and Quinlan (1990) show that nearly continuous
sampling is needed to detect organic pollutant pulses (e.g. the herbicide atrazine)
coincident with hydrograph peaks during conduit flow in streams. It may be that
headwater point inputs of organic pollutants via sinking streams and open
sinkholes on the Sinkhole Plain will be diluted before they reach Green River
Springs but diffuse input of pollutants can linger and cause insidious
bioaccumulation in cave organisms (Barr 1976). Diffuse input of pollutants
through the soil and epikarst into wells and eventually into streams is slow and
complex as Quinlan and Alexander (1987) demonstrate for an in organic fertilizer
(nitrate).
496
-------
THE NEED FOR A COMMUNITY AND HABITAT BASED INDEX OF BIOLOGICAL INTEGRITY
Of the biocriteria available toxicity bioassays, Indicator species, diversity
indices, and community similarity metrics are either totally flawed or have many
more disadvantages than advantages. The preferred biocriterion is an index of
biological integrity (IBI) which combines habitat data with population and
community data (Karr 1989).
Single Species Toxicity Bioassays
EPA's acute and chronic bioassays on Ceriodaphnia and fathead minnows
have the advantage that they have been standardized and are rigorously analyzed
statistically but they are fatally flawed as single best biocriteria for assessing
impacts of pollution (Cairns 1986, Karr et al. 1986). Cairns points out that the
central flaw in toxicity testing is the "myth of the most sensitive species". One
corollary of this myth is the unwarranted assumption that the responses of the
test species correspond to those of the many species not tested. Another flawed
assumption is that there are no responses that are more sensitive than the 50%
lethal dose usually measured. For example even the standard test organism may
have sublethal responses more sensitive than LC50, such as depression of
respiratory rate. Such sublethal responses in nature could change behavior of
predators or competitors and so have reverberating efffects throughout the
community. Still another flawed assumption is that the cost savings using single
species tests more than compensates for the cost of failing to act when the single
species test does not predict negative effects in nature. In addition, Karr points
out (1988) that even in situ field toxicity tests are not sensitive to environmental
degradation due to non-toxic pollutants like sediments and nutrients. In situ
toxicity tests are more reliable at detecting intermittent pollution than
laboratory tests using water collected at one instant in time but the costs of in
situ tests and their practical difficulties are great. The most important practical
difficulty is inacessibility of the caged test organisms during flood peaks when
we expect the greatest concentrations of many organic pollutants (e.g. Quinlan
1990).
I believe that there is little hope of developing a "white mouse" toxicity
test using a cave species because of the complexity of culturing cave organisms
and the conflicting toxicological results using field collected cavernicoles.
Reliable culturing of a troglophilic amphipod has been difficult (Dan Fong personal
communication) and nobody has successfully gotten an aquatic trogobitic species
through even one generation in the laboratory. The one toxicity test using EPA
protocols on field-collected animals (Bosnak and Morgan 1981) had the
counterintuitive result that the cadmium LC50 for a troglobitic isopod was higher
497
-------
than for a troglophilic isopod. My explanation for this result is that the trogloblte
would be less sensitive to acute exposure to a toxin if it had a lower metabolic
rate than the troglophile, as is consistent with the troglobite having a 16 fold
lower body concentration of cadmium with a 15 fold higher LC50 than for the
troglophile. Barn's acute toxicity studies at Mammoth Cave (1976) are not very
sensitive because he used only 1000 ppM of several toxins and his results are
equivocal. Given the complexities in his own data and the conservation problems
of using so many individuals for tests, Barr recommended to NPS that no more
toxicity tests be done.
Problems of Using a Single Indicator Species in the Field
Use of a single indicator species or even a single genus as a field indicator
of adverse impacts has fewer advantages and some of the same disadvantages as
just discussed for single species toxicity testing. Presence or absence of an
Indicator is a simple criterion but there are many problems of interpretation. In
addition to the same problems as with the "myth of a most sensitive species"
there is the problem that a species may be absent due either to pollution or for
natural reasons. One natural constraint is inability to live in some habitats. For
example shrimp do not live at upper levels or even at the headwaters of shaft
master drains ( Leitheuser et al. 1982) for reasons that are not clear. Another
natural constraint is of predation or predation mediated competition. For example
the isopod Caecidotea stygia does not occur in lower level habitats possibly
because of predation by crayfish and/or fish but this does not appear to be a
problem for Caecidoteabicrenata (Lewis and Lewis 1980). In other situations
food and predation may interact to affect the distribution and abundance of a
species. An example is of isopods in three unimpacted master shaft drains under
Flint Ridge (Poulson 1968). Isopods are most abundant where there is a
combination of gravel and rock cover, that mitigates the impact of fish and
crayfish predation, and a lot of fine particulate organic matter, that serves as
food.
Interpretation of Species Diversity is Difficult Because Not All Species are Equal
The number of species present, species richness, is easy to understand and
the the diversity of species, H, is easy to calculate but both have critical
problems with measurement and/or interpretation.
As a single biocrlterion species richness is especially sensitive to the
recurring problem of standardizing sampling effort and efficiency. Just how long
498
-------
should one sample over how large aan area to be sure that one has found all the
species present?
The information theory index of diversity H (see Poulson and Culver 1969
for a detailed description) is not so sensitive to sampling effort but its
interpretation is made difficult because it confounds the separate effects of
numbers of species and their relative abundances. Because of this problem one
should report both numbers of species and their relative abundances and not use
the single index H.
Quantitatively both species richness and H are flawed by the implicit
assumption that all species are equal in their effects in a community. In fact the
potential impact of a species is measured by its absolute importance value
(Poulson and Kane 1981). Importance Value of a species is the composite of its
frequency, density, and impact per individual. A very frequent species, found in
all habitats and at all times, is more likely to have an impact than a rare species.
A species which is locally very dense could have a high local impact even if it is
not frequent in space and time. And, a species which is large relative to others in
its community could have a great impact even if it is infrequent and has a low
density.
Qualitatively both species richness and H are flawed by the implicit
assumption that communities with natural and exotic species are equally
indicative of biological integrity. In the case of caves this is equivalent to saying
that two communities are the same if one has many accidentals and troglophiles
and the other has only troglobites, as long as both communities have the same
number of species and the same H. An example shows that this is patent nonsense.
Clearly the pollution stressed Hidden River Cave community dominated by tubificid
sewage worms and several large protozoan species with high numbers of
troglophilic crayfish and rare unspecialized troglobitic amphipods (Crangonyx} is
very different than the unimpacted Eyeless Fish Trail under Flint Ridge with a
regular troglobitic cave fish and crayfish, one or two common copepods, uncommon
Isopods, and rare but specialized troglobitic amphipods (Stygobromus}.
Measuring Statistical Similarity of Community Species Composition Has
Advantages but it is Not A Good Single Biocriterion
The simplest measure of community similarity uses only presence-absence
data and and is calculated by dividing the species in common by the sum of species
in the two communities to get a percent similarity. This measure easily detects
the differences between a badly impacted community such as in Hidden River Cave
and an unimpacted one such as in Eyeless Fish Trail. However it does not give an
499
-------
early warning sign if there are slight changes In community composition and or
changes in abundance of species without changes in species composition.
Multivariate statistical techniques, such as Principal Components Analysis,
use both species identity and abundance data to group species along two or more
axes. The axes often do not correspond to any simple environmental variable and
so the species clusters can be difficult to interpret ecologically. None-the-less
statistical changes in species position in a cluster or in location of a cluster,
compared to the natural variability seen from time to time in baseline
communities, could signal subtle changes and provide a good early warning sign
before negative impacts proceed too far.
An Index of Biological Integrity Keeps the Advantages and Avoids the
Disadvantages of Other Biocriteria and An IBI Has Unique Advantages
A maximum value for an IBI Indicates that a sampled area has the
"capability of supporting and maintaining a balanced, integrated, adaptive
community having a species composition and functional organization comparable
to that of the natural (unimpaired, original) habitat in the area" (Karr et al. 1986).
The EPA now mandates use of both a fish and a benthic macroinvertebrate IBI in all
surface stream surveys. An IBI includes many different kinds of criteria that are
ecologically meaningful and so it is sensitive to a wide array of differing insults
to the natural environment. An IBI does a particularly good job of integrating, over
space and time, the effects of past and ongoing negative impacts. An IBIs detailed
interpretation depends on the professional judgement of an ecologist because the
use of a variety of qualitative and quantitative measures does not allow for
statistical treatment. However the methodology for measuring each metric used
is clearly enough explained that anyone can be trained to use the IBI.
The multiplicity of criteria used to calculate an IBI include major
categories of habitat heterogeneity, energy source, numbers of species and guilds,
and well-being of each species. Available IBIs and my proposed cave IBI give
positive points to sensitive natural species like cave fish and give negative
points to troglophiles, exotic species, and pollution tolerant species not normally
found in the community. The two scales of habitat heterogeneity measured are
also applicable to caves. At a large scale one measures characteristics of the
stream itself such as size metrics, meander length, presence of undercut banks,
and the proportions of pool, raceway, and riffle habitats. At a finer scale one
measures the sediment size distribution on the bottom from bedrock and cobble,
through gravels and sand, to silt and clay. In a cave there is no plant
photosynthesis and so the relevant measure of energy input is the amount of
coarse and fine particulate organic matter and the amount of non-toxic dissolved
500
-------
organic matter. Most of the energy will be on or in the substrate and so I have
used weight loss on ignition at 550C as an index of total organic matter in the
sediments (Poulson 1968). Simple indices of well-being can be applied to most
cave species. A species is doing well if it is abundant and has a size distribution
that includes the small individuals indicative of recent reproductive success. In
fish a rounded belly and subcutaneous fat deposits are good signs and in
translucent crayfish the presence of yolked eggs in the ovary is a good sign.
A CAVE IBI WILL INCLUDE MOST KINDS OF AQUATIC SPECIES
AND SHORELINE TERRESTRIAL SPECIES
Cave ecosystems have few species and so currently EPA approved I Bis for
fish (Karr et al. 1986) and for benthic macro invertebrates (Ohio EPA 1987) cannot
be used or even simply modified for use in caves. The major reason for low
species diversity in caves is that lack of photosynthesis means there is a small
energy base for food webs. The problems of low food supply mean that most cave
species must be generalists in both feeding and habitat use and this in turn makes
it difficult for new species to invade unimpacted cave communities. For example
cave crayfish feed as detritivores, scavengers, and predators and cave isopods
occur in fast and quiet water and on all kinds of substrates.
All Kinds of Species Will Be Sampled
Because there are few species, a cave IBI must include all kinds of easily
sampled organisms. Field census without collection is easily done for fish,
crayfish, shrimp, and flatworms and can be done for isopods and amphipods if
species identification is not critical (Poulson 1968). Censuses of the larger
shoreline terrestrial species is also easy in the field. Baiting and trapping
techniques will be used where organisms are rare, where much of the habitat is
inaccessible, and where turbidity and high water preclude visual observation
(Cooper and Poulson 1978, Cooper 1975, Poulson and Culver 1969). A high
intensity diving light will greatly increase efficiency of census any time that
waters are slightly turbid and/or deep.
Detailed Data Taken On All Individuals Of Large Species
For extremely long-lived species with slow growth and infrequent
reproduction (fish Poulson 1963, crayfish Cooper 1975, and shrimp Leitheuser et
al. 1982) the entire accessible stream will be searched and all individuals will be
measured. Size distributions are a good index of current and past reproductive
501
-------
success, as explained earlier. To check for the presence of fin erosion, sores or
tumors, and ectoparasites the fish must be examined in a translucent lucite box
for a few minutes. Even more detailed data will be taken on crayfish because
their hard exoskeleton protects them during handling. It is easy to sex them and
assess male reproductive condition (form I vs form II). In addition the translucent
exoskeleton of the trogobitic species allows detailed assessment of female
reproductive status even by observing them in the stream (Jegla and Poulson
1970). Unlike crayfish, shrimp are too delicate to handle but they are transparent
and so status of ovarian or hatched eggs is observable in the stream with
appropriate snorkel or SCUBA gear ( Leitheuser et al. 1982).
Stratified Random Sampling To Census Smaller Organisms and Habitat
Characteristics
Smaller organisms have shorter generation times than fish, salamanders, or
crayfish and so are better indicators of short term changes in water quality. Their
more cryptic habits and higher densities require that random station be sampled
[Permanent stations are useful for short term studies of life cycles (Lewis 1982)
but can be misleading when there are shifts of organism microdistribution as
locations of substrates or food shifts after periods of high stream flow.] At each
random station aquatic macro invertebrates are censused stream habitat
characteristics measured, abundance of surface particulate organic matter
scored, large shoreline terrestrial species censused, and tiny terrestrial species
trapped. Both large aquatic and terrestrial species will be censused using a
combination of timed and areal searches to allow comparison between areas with
abundant and sparse populations. All microhabitats will be included in the
censuses. If there is a stream current then mini-surber samples will be taken
using a small silk aquarium net (in my experience this picks up copepods as well
as some large protozoa and some bacteria ). Habitat characteristics will be
assessed just as discussed earlier for Karr's fish IBI but with special attention to
refuges from floods such as undercut banks, breakdown, and rocks and gravel in the
streambed. Cores of the stream and/or shoreline sediments may be taken for
weight loss on ignition and to assess the presence of lenses of particulate organic
matter that could be exposed by flooding. Often there is no input of new leached
organic matter for years and so redistribution of the low quality old particulate
matter may be important.
BASELINE IBIS MUST BE DEVELOPED AT ECOREGION, SUBREGION, DRAINAGE BASIN,
AND LOCAL WATERSHED SCALES
For karst, contrasting ecoregions are the Valley-Ridge of the central
Appalachians vs the Interior Low Plateau in the Mammoth Cave area. Cave streams
502
-------
have different characters In these two ecoreglons and so there are different
baseline faunas (e.g. Culver 1982). Within the Mammoth Cave ecoregion southeast
of Green River there are two subregions, the Sinkhole Plain and the Mammoth Cave
Plateau. These subregions differ greatly in geology, hydrology, aand cave
development (as reviewed by many authors in White and White 1989). Each of
these areas has its characteristic communities (Barr and Kuehne 1971, Lewis and
Lewis 1980, Lewis 1981, Poulson 1968, Poulson and Culver 1969). In addition
there is a community that includes shrimp. It appears to be restricted to the
interface of downstream flows and backflow, during times of flood from the Green
River, through alluviated springs (Lisowski and Poulson 1979, Duchon and Lisowski
1981, Leitheuser et al. 1982). The still different Hilly Country to the northwest
of Green River is now being studied hydrologically but has scarcely begun to be
studied biologically.
DIFFERENT COMMUNITY RESPONSES TO NATURAL FOOD INPUTS AND
DIFFERENT CATEGORIES OF POLLUTANTS:
A SERIES OF NATURAL EXPERIMENTS
In this final section I review what baseline data we have so far to use in
developing IBIs and discuss what natural experiments can tell us about community
change with categories of pollutants that include sediments, human and animal
wastes, pesticides, heavy metals, and a potpourri associated with oil and gas
exploration. More detail is provided In the unpublished Proceedings of the 1990
Mammoth Cave National Park Karst Conference.
General kinds of responses of cave communities to different
categories of disturbance have been considered elsewhere. Thus Poulson and Kane
(1977) discuss the general impacts of removing single species, simplifying
habitat structure by siltation, modification of the timing and extent of floods,
favoring facultative cave species (troglophiles) by excess nutrient input, and
stress on long-lived obligate cave species (trogiobites) associated with toxins
that are biomagnif fed along food chains.
Natural Variation Among Vertical Shaft Master Drain
Streams Under Flint Ridge
Vertical shaft drains and master drains under the Mammoth Cave Plateau are
almost certainly in a near natural state and base level streams under Flint and
Mammoth Ridges are probably also natural since some receive little input from the
Sinkhole Plain. Thus data on these streams (especially Poulson 1968) are a solid
503
-------
start for developing I Bis to use In assessing impacts on more impacted baselevel
streams under Joppa Ridge.
The diversity of fauna among the streams is correlated with diversity of
habitat and kinds of food supply. The stream with the highest diversity of aquatic
species and the least census to census variation is the longest stream with both
the greatest diversity of the food base and the greatest diversity of habitats. Thus
the right habitat combination for each species is to be found somewhere at all
times. In contrast the stream that is smallest in extent has a low diversity of
habitats. Even though it has by far the greatest particulate organic matter it
also has the greatest variation in aquatic organism density as the stream varies in
extent among years. On the other hand the variation in the terrestrial fauna is the
least of the four streams. The veneer of coarse particulate organic matter on the
ceiling provides a virtually unvarying food supply that supports by far the richest
terrestrial fauna of the four streams and the highest densities of troglobitic
millipede (Scoterpes copei) and harvestman {Phalangodes armata} that I have
censused anywhere In the Mammoth Cave Area.
Tradeoff of High Food Supply and Low Predation Pressure with Low Habitat
Stability in Upper Level Seeps, Drip Pools, and Shaft Streamlets
In the caves under the Mammoth Cave Plateau the upper level aquatic
habitats do not have a diverse fauna but an IBI should be developed for them
because they are likely to be the first areas to be impacted by any local pollution
on the plateau and their waters often reach the shaft stream complexes at lower
levels. In general, whether impacted or not, these upper level habitats are
dominated by amphipods and f latworms that have adaptations for surviving in the
mud when pools or seeps dry up annually and can take advantage of high seasonal
inputs of bacteria and fine particulaate organic matter. When there is permanent
water isopods are added (Caec/cfotea stygia, Lewis 1982) and a troglophilic
amphipod may occur if there is unnaturally high organic input (Crangonyxpackardi,
Lewis 1987).
Cave Base-Level Community Compromised By Historic Alterations Of Green River
Discharge Patterns And Depth
The combined effects of three dams have compromised the base-level
communities, of the Styx-Echo River areas of Mammoth Cave, through a
combination of increased siltation and decreased food availability. This
conclusion is based on historical records pre and post installation or Lock and Dam
*6 downstream on Green River and on biosurveys of fish, crayfish, and shrimp
504
-------
during development of negative Impacts of the upstream dam In the 60s and 70s
(Barrand Kuehne 1971, Llsowskl and Poulson 1979, Duchon and Llsowskl 1981) and
continuing to present. A main focus in the 70s was the apparent decline of the
Kentucky cave shrimp, PaJaemoniasganterl', in the Styx-Echo River area, a pattern
that has now been borne out by detailed study of the shrimp's distribution and
abundance (Leitheuser et a). 1982).
Short-lived Species May Be More Sensitive To Acute Exposure To Toxins
And Long-Lived Species More Sensitive To Chronic Exposure
Cave communities will respond differently to acute and chronic exposure to
heavy metal or organic toxins depending on whether the exposure is acute and/or
chronic. Single or repeated acute exposures are likely to kill small, short-lived
species with relatively high metabolic rate. This will decrease or eliminate the
food supply for the larger, longer-lived and lower metabolic rate species of upper
trophic levels. The early warning signs of this scenario are easy to detect from
biosurveys and IBIs. More subtle and Insidious are the effects of sublethal chronic
exposure to toxins because the early warning signs are first seen in a further
decrease in the normally infrequent reproductive success of fish and with no
elimination of species. With chronic toxin pollution the cave fish will have
increasing body loads of toxins due both to bioaccumulation associated with long
life and to biomagnification along the food chain. If we are lucky there will also
be obvious signs of problems such as with the hemorrhaging and impaired
swimming of "broken-back syndrome" in cave fish (Keith and Poulson 1979). Cave
crayfish are less likely than cave fish to be affected by chronic organic toxins
because they store less fat and less likely to be killed by severely acute exposures
because they can leave the stream and so decrease exposure as was observed to
occur in summer 1979 with an acute hydrocarbon exposure in Hawkins River.
If there are changes in IBI or individual species well-being due to suspected
toxin exposure then identification of the culprit(s) will depend on chemical
testing since all toxins can kill both by acute exposure and by chronic
bioccumulatlon and biomagnification during prolonged sublethal exposures.
Inorganic Fertilizers Affect Cave Communities Only If They Stimulate Primary
Productivity Of Surface Waters Which Are Subsequently Diverted Into Caves
There is no photosynthesis in caves so the stimulating effect of inorganic
fertilizers, especially phosphates and nitrates, on aquatic plant growth can only
affect cave communities indirectly. Impacts will occur when algal choked
sinkhole ponds overflow into caves or when the sinkhole pond disappears
505
-------
underground as Its drain suddenly opens. A natural experiment that simulated this
scenario was of a catastrophic leakage of the Job Corp sewage treatment lagoons
into vertical shafts that feed Eyeless Fish Trail under Flint Ridge in 1967. This
was a one time event and did not cause a detectable change in the distribution,
abundance, or well-being of cave fish or crayfish (Poulson 1968 and unpublished)
but it had a sudden depressing effect on the species diversity and abundance of the
shoreline terrestrial community (Poulson and Culver 1969). This depressing
effect was presumably due to toxic effects of the blue-green bacteria Spfru/ma
that was left as a band along the shore as the pollution laden crest of water
passed and receded. This influx of sewage lagoon waters caused the substrate
organic content to increase from 1.05-1.23% to 2.44-3.06%.
Human And Animal Wastes Impact Cave Stream Communities Mainly By Favoring
Species That Tolerate Low Oxygen And Are Reproductively Stimulated By Increased
Food Supply
The destruction of the cave community in Hidden River Cave was due
initially to a combination of past toxic effects of heavy metals pollution (Barr
1976) and the high biological oxygen demand of decomposing cheese whey and
human wastes (Austin personal communication) but the current slow recovery
seems due to residual effects of human wastes. My conclusion is based on natural
experiments in other caves where the only pollution is by human or animal wastes.
Holsinger (1966) and Lewis et al. (1982) provide qualitative observations,
Weingartner (1977) provides detailed quantitative data, and I have preliminary
quantitative data on different degrees of impact. At high impact levels, as
directly under a feed lot or a large sewage treatment plant the high BOD kills all
macroscopic organisms and leaves only strands of colonial sewage bacteria and
associated protozoa. If the BOD does not completely remove oxygen then tubificid
sewage worms become part of the community as has recently been the case in the
south branch of Hidden River Cave (Lewis et al. 1982, Poulson unpublished). If the
amount of wastes is not too great, as with septic field diffuse input or with
dilution by less polluted water as in east branch of Hidden River, the sewage fauna
drops out and the higher than normal food supply favors survival and reproduction
of shorter-lived macroscopic cave fauna, especially troglophilic isopods,
amphipods, and crayfish which may replace troglobitic species. As the input of
waste becomes less accidental chironomid midges and troglophiles survive, but do
not reproduce, and reproduction of short-lived troglobitic isopods and f latworms
is stimulated. At still lower impact levels the reproduction of larger troglobites
may also be stimulated. Far downstream from a presumed feed lot input into
Black River of Roppel Cave there are incredibly dense populations of cave crayfish
(Leitheuser personal communication) and by far the highest densities of
meiofaunal nematodes (280/liter of sediment) and harpactacoid copepods
506
-------
(858/llter of sediment) measured anywhere by Whitman (in Leitheuser et a).
1986). High meiofauna) densities may be an early warning sign of increasing
human and animal waste pollution because normally the meiofaunal species
diversity and density decreases downstream from vertical shaft inputs.
ACKNOWLEDGEMENTS
I am indebted to a host of colleagues with whom I have discussed aquatic
cave ecology over the past 30 years. My interpretations do not always match
theirs but I could not have generated my ideas without their verbal input and their
data. I cannot mention everyone who has helped to form the ideas expressed herein
and/or has helped me in the streams in the Mammoth Cave Area but a few bear
special mention. In alphabetical order they are Tom Aley, Bill Austin, Tom Barr,
Nick Crawford, Dave Culver, John Cooper, Dave Griffith, Jack Hess, Tom Kane, Jim
Keith, Terry Leitheuser, Jerry Lewis, Ed Lisowski, Eric Morgan, Rick Olson, Jim
Quinlan, Stan Sides, and both Bette and Will White. Finally I could not have learned
what I know without the continued cooperation of landowners and the help and
encouragement of the staff at Mammoth Cave National Park and of the JVs and
members of The Cave Research Foundation. I thank you one and all!
CITED LITERATURE
Barr, T. C. Jr. 1976. Ecological effects of water pollutants in Mammoth Cave.
Final Technical Report to NP5. Contract *CX-5000-50204
Barr, T. C. Jr. and R. A. Kuehne 1971. Ecological studies in the Mammoth Cave
System of Kentucky. II. The ecosystem. Annales de Speleologie, 26(1): 47-96.
Bosnak, A. D. and E. L Morgan 1981. Acute toxicity of cadmium, zinc, and total
residual chlorine to epigean and hypogean isopods (Asellidae).
National Speleological Society Bulletin, 43: 12-18.
Cairns, J. Jr. 1986. The myth of the most sensitive species: Multispecies
testing can provide valuable evidence for protecting the environment.
Bioscience, 36(10): 670-672.
Cooper, J. E. 1975. Ecological and behavioral studies in Shelta Cave, Alabama,
with special emphasis on decapod crustaceans.
Ph.D. in Zoology, University of Kentucky. 364 pp.
Cooper, J. E. and T. L. Poulson 1978. A guide for biological collecting in caves.
Caving Information Series of the National Speleological Society. 14 pp.
Culver, D. C. 1982. Cave Life: Evolution and Ecology.
Harvard University Press, Cambridge MA, USA. 189 pp.
507
-------
Duchon, K. and E. A. Llsowski 1980. Draft environmental assessment of Lock and
Dam Six, Green River Navigation Project on Mammoth Cave National Park.
Cave Research Foundation, Dallas TX
EPA, Ohio 1987. Biological criteria for the protection of aquatic life. Vol. 2.
Users manual for biological assessment of Ohio surface waters.
Division of Water Quality Management and Assessment, Columbus OH
Holsinger, J. R. 1966. A preliminary study on the effects of organic pollution
of Banners Corner Cave, Virginia.
International Journal of Speleology, 2: 75-89.
Karr, J. R., K. D. Fausch, P L. Angermeier, P R. Yant, and I. J. Schlosser 1986.
Assessing biological integrity in running waters: A method and its rationale.
Special Publication 5. Illinois Natural History Survey, Champaign, IL
Keith, J. H. and T. L. Poulson 1979. Broken-back syndrome in Amblyopsis
spelaea, Donaldson-Twin Cave, Indiana.
pp 45-48 in Cave Research Foundation Annual Report
Leitheuser, A. T. and various authors. 1982-1986. Phases I-VI. of Ecological
analysis of the Kentucky cave shrimp Palaemoniasganteri Hay,
Mammoth Cave National Park. Old Dominion University Research
Foundation, Norfolk VA. NP5 Contract * CX-5000-1 -1037.
Lewis, J. L. 1981. Observations on aquatic communities in the Historic Section of
Mammoth Cave, pp 17-19 in Cave Research Foundation Annual Report
1982. The life cycle and distribution of two troglobitic Caecidotea in
Mammoth Cave National Park.
pp 10-11 in Cave Research Foundation Annual Report
1987. Aquatic communities in the Cathedral Domes section of the
Mammoth Cave, pp 35-38 in Cave Research Foundation Annual Report
and T. M. Lewis 1980. The distribution and ecology of two species
of subterranean Caecidotea in Mammoth Cave National Park.
pp 23-27 in Cave Research Foundation Annual Report
T. M. Lewis, and J. Eckstein 1982. A biological reconnaissance of
a polluted cave stream: The Hidden River groundwater basin.
pp 9-10 in Cave Research Foundation Annual Report
Lisowski, E. A. and T. L. Poulson 1979. Impacts of Lock and Dam Six on baselevel
ecosystems in Mammoth Cave.
pp 48-54 in Cave Research Foundation Annual Report
Poulson, T. L. 1963. Cave adaptation in Amblyopsid fishes.
The American Midland Naturalist, 41: 263-290.
1957 Comparisons of cave stream communities.
pp 33-34 in Cave Research Foundation Annual Report
1985. Evolutionary reduction by neutral mutations: Plausibility
arguments and data from Amblyopsid fish and Linyphiid spiders.
National Speleological Society Bulletin, 47(2): 109-117.
508
-------
and D. C. Culver 1969. Diversity in terrestrial cave communities.
Ecology, 50(1): 153-157.
and T. C. Kane 1977. Ecological diversity and stability: Principles
and management.
pp 18-21 in T. Aley and D. Rhodes eds.,Proceedings of the 2nd Cave
Management Symposium, Speleobooks, Albuquerque, NM, USA
and W. B. White 1969. The cave environment: Limestone caves
provide unique natural laboratories for studying biological and
geological processes. Science, 165:971-981.
Quinlan, J. F. 1990. Special problems of ground-water monitoring in karst
terranes. pp 275-304 in D. M. Nielsen and A. I. Johnson eds.,
Groundwater and vadose zone monitoring. ASTM Special Technical
Paper 1053, American Society for Testing and Materials,
Philadelphia, PA, USA
and E. C. Alexander Jr. 1987. How often should samples be taken
at relevant locations for reliable monitoring of pollutants from an
agricultural, waste disposal, or spill site in a karst terrane?
A first approximation, pp 277-286 in B. Beck and W. L. Wilson eds.,
Proceedings from the 2nd Multidisciplinary Conference on Sinkholes and the
Environmental Impacts of Karst, Florida Sinkhole Research
Institute, University of Central Florida, Orlando, FL, USA
White, W. B. and E. L. White eds. 1989. Karst hydrology: Concepts from the
Mammoth Cave Area. Van Nostrand-Reinhold, NY, USA
BIOGRAPHICAL SKETCH
Tom Poulson got his B.S. at Cornell University and his Ph.D. at the University of
Michigan. After a postdoctorate at U.C.L.A. he taught 10 years at Yale and 3 years
at Notre Dame before coming to the University of Illinois at Chicago where he is
now Professor of Biological Sciences. He is Chief Scientist of The Cave Research
Foundation. He Studies morphology, physiology, life history, and behavior in
ecological and evolutionary contexts. In caves he studies fish, crayfish, spiders,
millipedes, springtails, flies, and beetles. Outside he has studied many
communities including tropical forest dung beetles, salt marsh and desert birds,
mountain and desert chipmunks, floating plants, dunes grasses, and forest trees.
509
-------
U5E OF BIOLOGICAL INDICATORS OF GROUNDWATER QUALITY
Thomas L. Poulson
Q. When using an Index of Biological Integrity (IBI) to assess groundwater
contamination in a cave, how necessary is it to have a pre-contamination
assessment? Will nearby cave streams be sufficient to provide a baseline?
A. It is sufficient to define maximum values for an IBI using the variation seen
for the most pristine cave stream(s) in the local region. Only slight adjustments
to expectations for non-impacted streams are made due to abiotic factors such as
stream order and above or below average refuges from flooding and predation.
Q. How useful and reliable might the IBI approach be for karst springs where caves
are not accessible?
A. The principles behind an IBI will hold for any aquatic community including
karst springs. I would develop an IBI using spring basin faunas and floras for the
region in question and, if possible, use a net to collect the organisms washed out
of the spring over a week long period both at times of low and high flow. The
problem with spring outwash data alone is that one cannot assess the underground
habitat to determine what it can support and so it might be important to have pre
contamination data on the outwash fauna for a particular spring system.
Q. Are the top species of the food chain sufficiently rapid in their responses to be
useful bioindicators of pollution?
A. Fish and crayfish at the end of the food chain react quickly to some and slowly
to other contaminants. Crayfish crawl out of the water when there is an acute
input of a toxin at high concentrations. Increased proportions of females becoming
fecund may indicate chronic input of human and animal waste at low
concentrations. But neither long-lived fish nor crayfish may respond quickly to
chronic input of toxins at low concentrations that take many years to
bioaccumulate enough to cause visible tumors or obviously compromise
reproduction. If toxin pollution is suspected then I would find out what is used in
the watershed (which pesticides and herbicides) and monitor the most dangerous
toxin in the water at peak flow when runoff will be greatest.
510
-------
Q. Have you seen any significant changes in the aquatic fauna of Mammoth Cave
since the 1950s and if so why have the changes occurred?
A. No species has been lost or gained but the diversity and abundance of the Echo-
Styx River communities apparently declined since the early 1900s and has clearly
declined since the 1950s when I first censused the area (Lisowski and Poulson
1979). At present cavefish, cave crayfish, and cave shrimp are either absent or
very difficult to find and the upper level species of isopod may have replaced the
usual river species. These changes were associated with changes in flow regime
and water level starting with the construction of Lock and Dam *6 just
downstream on Green River around 190 and then continued with the construction
of the Nolin River Dam just downstream and the Green River Dam far upstream in
the 1950s. I do not know the mechanism but have hypothesized that it is increased
silt deposition in the cave. Siltation does decrease substrate heterogeneity and so
may have decreased refuges of copepods, isopods, amphipods, and shrimp from
floods and/or predation by fish and crayfish. Siltation may also cover or dilute
particulate organic matter and so cause declines of the bacteria and protozoa
which are eaten by copepods, isopods, amphipods, shrimp, and small crayfish.
511
-------
512
-------
THE USE OF BENTHIC MACROINVERTEBRATES FOR ASSESSING THE
IMPACT OF CLASS V INJECTION WELLS ON CARBONATE GROUND WATERS
Albert E. Ogden1, Ronald K. Redman2 and Teresa L. Brown3
Center for the Management, Utilization and Protection
of Water Resources1 and The Department of Biology2,
Tennessee Technological University, Cookeville, TN 38505
and
First Tennessee Development District3, 207 N Boone Street,
Suite 800, Johnson City, TN 37604
ABSTRACT
The U.S. Environmental Protection Agency contracted the
authors to investigate the impact of Class V injection wells on
ground water in karst terranes to demonstrate the need for
regulations over these shallow injection methods. Any sinkhole
that has been modified to better accept drainage, including
storm water runoff, was considered a Class V injection well. In
addition, service station bay drains that lead to septic tanks,
pits, or dry wells were a focus of the study.
The study was conducted in Cookeville and Johnson City,
Tennessee, to compare contaminant transport in both flat-lying
and folded rock. A major cave system underlies both cities.
Ground water flow paths and velocities were determined using dye
tracing methods. Water and benthic macroinvertebrate samples
were taken where water entered Class V injection wells and at
the springs influenced by these waters. As a control, samples
were also gathered from a sinking stream and spring in a
forested area. Concentrations of MTBE, BTEX, TPHC, ethylene
glycol, and metals were below detection limit at most sites and,
therefore, were not useful in demonstrating an impact. This
emphasizes that grab samples in karst seldom indicate overall
water quality conditions. In contrast, the Shannon Diversity,
Shannon Evenness, and Family Biotic indices, coupled with a
cluster analysis of the benthic data, were very useful in
classifying sampling sites as to their degree of pollution where
water chemical data did not indicate degradation.
INTRODUCTION
Part C of the Safe Drinking Water Act (Public Law 93-523)
authorized the U.S. Environmental Protection Agency (EPA) to
establish regulations to assure that potable ground water is not
513
-------
endangered by the underground injection of waste. Guidelines
for underground injection and the classification of injection
wells come under Part 146.04 of the Federal Underground
Injection Control program (U.S. EPA, 1981). This program
created five classes of underground injection wells. Classes I
through IV include such categories as radioactive and hazardous
waste injection and disposal of brines from the oil and gas
industry.
Class V wells are generally defined as those which inject
only nonhazardous fluids into or above strata that contain
underground sources of drinking water (USDW). USDWs not only
include aquifers which are currently serving as drinking water
supply sources but also aquifers which are of acceptable quality
for possible future use. Class V wells include any type of
injection well not covered in the UIC definition of Classes I,
II, III, or IV. EPA has classified Class V injection wells into
six groups based in part on the expected quality of the injected
fluid. The following is a listing of these groups as found in
regulations for the State of Florida which is particularly
sensitive to karst aquifers:
Group 1 - wells associated with thermal energy exchange
processes, which include air conditioning return flow
wells and cooling water return flow wells. Cooling
water return flow wells may be part of a closed-loop
system, with no hazardous additives;
Group 2 - recharge wells, saltwater intrusion barrier wells,
connector wells, and subsidence control wells
(associated with aquifer overpumping);
Group 3 - wells which are part of domestic waste treatment
systems, swimming pool drainage wells, injection wells
used in experimental technologies, wells used to
inject spent brine into the same formation from which
it was withdrawn after extraction of halogens or their
salts;
Group 4 - nonhazardous industrial and commercial disposal wells,
which include laundry waste, dry wells, sand backfill
wells, and nuclear disposal wells used to inject
radioactive wastes, provided the concentrations of the
waste do not exceed drinking water standards contained
in Chapters 17-22, FAC, and injection wells used for
in-situ recovery of phosphate, uraniferous sandstone,
clay, sand, and other minerals extracted by the
borehole slurry mining method (lignite, tar sands, oil
shale, coal);
Group 5 - lake level drainage and stormwater drainage wells; and
Group 6 - geothermal wells and "other" wells.
The U.S. EPA funded twenty-four projects nationwide under
its Shallow Injection Well Initiatives Program to help evaluate
the impact of Class V injection wells on ground water and to
establish best management practices. The authors of this paper
514
-------
were chosen to help evaluate Class V injection well practices in
the karst of Tennessee. Any sinkhole that had been modified to
accept waste, including stormwater runoff, was considered a
Class V injection well in the study.
OBJECTIVES
The primary goal of the proposed study was to determine if
Class V injection wells have created a ground water pollution
problem. Evaluating the effects of service station bay drains
(Group 6) that lead to septic tanks, pits, or dry wells was a
primary target of the investigation. Other high priority Class
V injection wells that were investigated include: 1.
Agricultural and municipal drainage into improved and unimproved
sinkholes (Group 1), 2. Industrial drainage into sinkholes
(Group 4), and 3. Domestic wastewater drainage into sinkholes
(Group 5) .
To fulfill these objectives, samples were collected from
waters entering sinkholes and the springs hydrologically
connected to them as determined by ground water tracing methods.
A benthic macroinvertebrate study was then conducted at most of
the sites to evaluate the impact of contaminants on stream
biota. Since ground water chemistry in karst terranes changes
rapidly during storm events (Ogden, 1988 and Quinlan, 1989), it
was felt that the benthic survey might prove a more reliable
indicator of long-term water quality.
HYDROGEOLOGY OF THE STUDY SITES
Two geologically different karst terranes in Tennessee were
chosen so that the results would have applicability throughout
much of the karst in the United States (Figure 1). Around
Western Villi
Cumberland Plitfi
Mnmslppi River Vtllcy
Figure 1. Location of study areas (adapted from Miller, 1974).
515
-------
Cookeville (Putnam County), the Mississippian-aged carbonates of
the Eastern Highland Rim Province are flat-lying allowing ground
water to move along a wide range of orientations corresponding
to joint and photo-lineament trends (Ogden et al., 1989).
Streams originating on the Cumberland Plateau east of Cookeville
flow over shales and sandstones until underlying limestone beds
are intersected. Some streams sink when they reach the Bangor
Limestone while others sink into the Monteagle or St. Louis
limestones. Subterranean water moves through caves, pits, and
solution-enlarged fractures until emerging as spring flow,
commonly at the St. Louis-Upper Warsaw boundary. Within much of
the city limits of Cookeville, surface runoff is perched on the
upper Warsaw Formation which is a cross-bedded, reddish-brown,
sandy limestone. This water then sinks into the middle
limestone member. A four-mile-long cave that drains much of
Cookeville's storm water runoff occurs within this unit. The
lower Warsaw member is a relatively impermeable calcareous
siltstone and shale with some argillaceous limestone beds. The
largest springs around Cookeville are found perched above this
lower member.
Ground water conditions in Putnam County have not been
extensively studied. An early regional analysis of the aquifers
and a listing of water wells was made by Smith (1962) . A ground
water resource analysis around Center Hill Lake, which included
the Cookevile area, was made by Moore and Wilson (1972). A
water quality survey of some water wells around town was
performed by Collar and Ogden (1990). Faulkerson and Mills
(1981), Pride et al. (1988), and Hannah et al. (1989) have
performed some dye traces around Cookeville. Physiochemical
degradation of the water within the four-mile-long cave beneath
Cookeville has been documented by Smithson (1975), Faulkerson
and Mills (1981), Wilson (1985), and Pride et al. (1988). A
recent assessment of sinkhole flooding in Cookeville from storm
water drainage has been conducted by Mills et al. (1991).
The second study area used for the investigation occurs
within the city limits of Johnson City (Washington County)
located within the Appalachian Valley and Ridge Province of
eastern Tennessee (Figure 1). In the Valley and Ridge Province
around Johnson City, the Ordovician-aged Knox Group carbonates
are complexly folded and faulted, and ground water moves
predominantly along stratigraphic strike within solution-
enlarged bedding planes. The Knox Group is over 3,000 feet
thick and has not been subdivided in the study area. Much of
Johnson City is developed on a sinkhole plain. Many of the
sinkholes were designated as Class V injection wells by this
study due to storm water routing into them and/or the presence
of discharging pipes (often of unknown origin) within the
sinkhole basin.
Early reconnaissance ground water studies of the Johnson
City area were performed by Maclay (1956) and DeBuchananne and
-------
Richardson (1956). Many of the public water supplies near
Johnson City utilize springs that emerge from cavernous
limestones and dolomites. The vulnerability of these municipal
water supplies prompted the First Tennessee Development District
(Stanley Consultants, Inc., 1983; Matthews, 1986; and Brown,
1987) and TVA (Foxx, 1981) to make a survey of potential
pollution sources within close proximity of the springs. Other
springs in the area have shown probable contamination from
septic tanks (Wilson et al., 1988). Many of the eastern
Tennessee springs become turbid after storm events and show
elevated levels of fecal coliform (Brown, 1987). This reflects
the impact that natural and anthropogenic activities in the
recharge area have on ground water quality. Ogden et al. (1991)
recently completed a "wellhead" protection study of nine
municipal-used springs in eastern Tennessee that involved ground
water tracing and a time-series analysis of water quality. The
springs within the city limits of Johnson City were not included
in any of these studies.
No ground water tracing had been conducted in Johnson City
until initiation of this project. Field investigations
performed for this study showed that most of the storm drainage
in north Johnson City flows through cave systems and then
emerges at springs along Knob and Cobb creeks which flow into
the Watauga River. These creeks are being studied by the U.S.
Geological Survey as part of the nationwide effort to monitor
urban storm water runoff quality (Johnson, personal
communication).
METHODS
Samples were collected of waters entering Class V injection
wells (sinkholes) and at the springs influenced by these waters
during the 1991 wet and dry seasons. As a control, samples were
also gathered from a sinking stream and a spring with a
predominantly forested recharge area. Dissolved oxygen (DO) ,
pH, conductivity, and temperature were measured in the field
with a Hydrolab field monitor. Laboratory analyses used as
indicators of contamination from agricultural activities and
septic tanks included nitrate, chloride, fecal coliform
bacteria, and fecal streptococcus bacteria. Zinc, chromium,
lead, methyl tertiary butyl ether (MTBE, a gasoline additive),
benzene-toluene-ethylene-xylene (BTEX), total petroleum
hydrocarbon (TPHC), and ethylene glycol levels were measured as
indicators of waste products and leaks from service stations and
parking lot runoff. Seven springs and seven sinking streams
were sampled around Cookeville; whereas, six springs and three
sinking streams were sampled around Johnson City.
To enhance the interpretation of the impact of Class V
injection wells on spring waters, benthic macroinvertebrate
samples were collected at riffle areas and pools at most of the
injection points and the springs. Samples were collected with
517
-------
a modified kick net and a Surber sampler (0.09 m2) . All
organisms were preserved in a 10% formalin solution, enumerated,
and identified to the genus level when possible. Statistical
analysis of the benthic macroinvertebrate data were then
conducted.
Qualitative ground water tracing was conducted using
fluorescein dye and activated charcoal detectors. Optical
brighteners and cotton detectors were also used for tracing. A
control packet was placed at a spring known not to be
hydrologically connected to the tracer input site during each
test.
RESULTS
Water Quality Sampling
Both wet and dry season samples from nearly all of the
twenty-three sites in the two study areas showed below detection
limits for BTEX, MTBE, ethylene glycol, chromium, lead, and
zinc. One site, the storm drain for Tennessee Technological
University, showed below detection for BTEX in one sample and
high levels in another (benzene - 1200 ppb, toluene - 220 ppb,
xylene - 780 ppb) . Fumes were reported in this drain which
initiated its sampling. The drain receives runoff from the
university's garage and lies next to several gasoline
underground storage tanks. In addition, a nearby underground
storage tank used to provide fuel for a background power supply
in the basketball arena was recently pulled, and soil samples
showed TPHC levels as high as 2088 ppm. No remediation or site
characterization was performed at the site, and a new building
has since been constructed over it. The storm drainage system
beneath the university has been diverted into a large sinkhole
which also receives storm water runoff from other areas of
Cookeville. Raw sewage from the city lines is occasionally
bypassed into the same sinkhole. Ground water tracing showed a
hydrologic connection between the sinkhole and Big Spring.
MTBE was found in only two samples, and it was at low
levels (< 5 ppb) . TPHC was above the detection limit at only
one site during the wet season with a level of 309 ppb.
Sampling during the dry season showed detectable TPHC at only
the Cookeville city garage drain site with a value of 6 ppm.
Zinc was measured at levels above detection at five sites with
concentrations being up to 73 ppm.
Chloride levels were significantly higher at springs
receiving recharge from Class V injection wells compared to the
control springs. The control springs had chloride levels
between 1 and 3 ppm; whereas, the other sites ranged from 3 to
64 ppm with an average of 14 ppm. Nitrate concentrations were
below 2 ppm at all but two sites. The highest measured value
was 5 ppm. Fecal coliform and fecal streptococcus bacteria
518
-------
counts were dramatically different between the wet and dry
sampling periods. The wet season samples were relatively free
of these bacteria. Only five of the samples had fecal
streptococcus bacteria with the highest count being only 21
colonies/100 ml. All but two of the sites had fecal coliform
bacteria counts over fifty with the highest being 350
colonies/100 ml. In contrast, most of the dry season samples
had fecal streptococcus levels that were too numerous to count
(TNTC). Fecal coliform bacteria counts were greater than 500
colonies/100 ml at thirteen of the sites for the dry season
samples and only one site had less than 25 colonies/100 ml. The
storm drain at Tennessee Technological University and Big Spring
had the highest fecal coliform bacteria counts in the Cookeville
samples related to leaky sewers and bypassing of raw sewage
(2900 colonies/100 ml and TNTC, respectively). The fecal
coliform/fecal streptococcus ratios suggest that most of the
other sites are being polluted from animal versus human wastes.
Benthic Macroinvertebrate Sampling
Benthic macroinvertebrate samples were collected from
twelve sites in the two study areas. To date, identifications
have been completed for six samples collected during the wet
season. As a control, a spring with a forested recharge area
was chosen for comparison to the springs with urban recharge
areas. Macedonia Spring in Cookeville and Taylor Spring in
Johnson City served as the control springs.
Shannon Diversity indices (Washington, 1984) were
calculated on each spring and sinking stream (Table 1). Wilhm
(1970) found values between 3 and 4 for the Shannon Diversity
Index (d) in unpolluted waters, but in badly polluted waters it
was generally less than 1. All of the sites shown on Table 1
exhibited low Shannon Diversity (Base 2) values with Macedonia
Spring having the highest (2.77) indicating only slight organic
pollution. However, EPA biologists for the southeastern region
have found d not to be sensitive enough and have used
equitability e (Shannon Evenness) to better demonstrate
degradation of stream waters (Weber, 1973). E is sensitive to
slight levels of degradation. Polluted streams usually have
values between 0.0 to 0.5; whereas, unpolluted streams fall
between 0.5 to 1.0. In the Cookeville area, the Shannon
Evenness Index (Table 1) clearly distinguishes the polluted
urban springs (Big and Pigeon Roost) from the rural spring
(Macedonia). In the Johnson City area, this distinction is not
as apparent. This may be due to Taylor Spring's immediate
confluence with a surface stream which may not allow time for
the benthic environment to change from heterotrophic to
autotrophic. According to Vannote et al. (1980), this change is
necessary for increasing species diversity. Also, the substrate
at Taylor Spring is almost completely composed of sand, silt,
and clay with nearly no gravel or cobble-size material.
519
-------
Table l. Shannon Diversity, Shannon Evenness, and Family Biotic
indices calculated for the wet season samples.
Macedonia Spring
Big Spring
Pigeon Roost Spring
Taylor Spring
Shoe Spring
Car Wash Insurgence
Shannon
Diversity
(Base 2)
2.77
0.88
0.15
2.14
1.64
2.15
SP
P
P
SP
SP
SP
Shannon
Evenness
0.69
0.31
0.08
0.58
0.55
0.68
UP
P
P
UP
UP
UP
Family
Biotic
Index
3.97
5.84
6.01
6.08
7.05
5.50
VG
FP
FP
FP
P
F
P = polluted; SP = slightly polluted; UP= unpolluted; VG = very
good; F = fair; FP = fairly poor; P = poor
Macedonia Spring, Big Spring, and Pigeon Roost Spring also
emerge from caves but their substrates are composed primarily of
gravel, cobble, and boulder sizes allowing for more diversity
and greater numbers of species. Car Wash Insurgence and Shoe
Spring have substrate compositions composed of gravel and cobble
mixed with sand, silt, and clay.
Hilsenhoff's (1988) Family-Level Biotic Index (FBI) was
also calculated for each site to see how it compared to the
Shannon Base 2 values and equitability e values (Table 1). In
general, the higher the FBI, the greater is the amount of
organic pollution with 10 being the highest number possible. In
the Cookeville area, the rural spring was rated as very good
while the urban springs were ranked as fairly polluted. In the
Johnson City area, a distinction between the rural and urban
sites was not seen, probably for the reasons discussed
previously. Visual and olfactory evidence of pollution was
found at the two urban springs in Johnson City, and water
quality differences between the two groups were seen based on
chloride and conductivity results (Table 2). This water quality
data helps support the substrate and heterotrophic vs.
autotrophic hypotheses just presented.
A cluster analysis of the benthic macroinvertebrate data
was also performed for the study sites. The cluster analysis,
together with the Shannon Diversity, Evenness, and the Family
Biotic indices, gave further insight to which sites have
threatened benthic communities and possible ground water
degradation. Big, Shoe, and Pigeon Roost springs are polluted
by city runoff into Class V injection wells and thus cluster
together (Figure 2). Low number of genera at Taylor Spring,
related to substrate composition, caused it to be grouped with
520
-------
Table 2. Wet and dry season chloride and conductivity values.
Macedonia Spring
Big Spring
Pigeon Roost Spring
Taylor Spring
Shoe Spring
Car Wash Insurgence
Chloride (Mg/L)
Wet season
1.70
4.00
18.20
2.82
7.92
10.20
Dry season
2.40
15.60
56.30
2.30
7.39
11.80
Conductivity (/unhos/cm)
Wet season
364
158
286
281
478
475
Dry season
142
236
262
572
442
863
the polluted urban springs. Macedonia Spring, which has
relatively pristine water, was grouped alone. Car Wash
Insurgence, a small sinking stream, was clustered separately due
to large annual water temperature fluctuations which have a
significant effect on the biologic community structure not seen
at the springs.
Cluster Analyses
• Macedonia
1 Big Spring
Shoe Spring
Pigeon Roost
Taylor Spring
Car Wash Insr.
Linkage Clusters
1
2
3
4
5
Big Spring
Big Spring
Big Spring
Macedonia
Macedonia
2 .4
Linked
Shoe Spring
Pigeon Roost
Taylor Spring
Big Spring
Car Wash Ins.
.6
Similarity
.50446
.38494
.28472
.04615
.00098
.8
Prob
.29000
.15000
.10000
.01000
.01000
Figure 2. Cluster analysis results.
521
-------
CONCLUSIONS
The water quality tests performed for this study were not
definitive in assessing the impact on karst ground waters of
volatile chemicals entering Class V injection wells. The open
conduit flow in the karst aquifers allows contaminants to be
rapidly flushed through the ground water system. Nearly
continuous water quality monitoring would be necessary to detect
contaminants dumped or spilled into Class V injection wells such
as from service station bay drains. In addition, turbid flow in
the caves and sinking streams causes rapid degassing of the
volatile components, thus explaining why some sample sites
showing the presence of oil and grease had no measurable amounts
of volatiles. Also, if volatiles were detected, it would be
nearly impossible to determine if they were derived from bay
drains or from leaky underground storage tanks. Fecal coliform
and fecal streptococcus bacteria levels, on the other hand, were
very high during the dry season demonstrating the impact on
ground water from leaky sewers, sewage bypassing, and animal
wastes entering Class V injection wells. Chloride and
conductivity values were also useful in demonstrating ground
water degradation.
Benthic macroinvertebrate analyses may be better for
accessing the impact of Class V injection wells on ground water
since these organisms can be very sensitive to daily water
changes not detected by collecting occasional grab samples for
chemical analysis. In this study, the Shannon Diversity,
Shannon Evenness, and Family Biotic indices, in conjunction with
the cluster analysis, proved useful in distinguishing spring
waters that have been degraded by pollutants entering Class V
injection wells.
ACKNOWLEDGEMENTS
This project was funded by the U.S. Environmental
Protection Agency with matching funds provided by the Center for
the Management, Utilization and Protection of Water Resources—
Tennessee Technological University and the First Tennessee
Development District. This paper does not necessarily reflect
the views of the agencies.
REFERENCES
Brown, T., 1987, Survey of public ground water supplies: First
Tennessee Development District, Johnson City, TN, 217 p.
Collar, P.O. and A.E. Ogden, 1990, A hydrochemical survey of
carbonate groundwaters, Putnam and Jackson counties,
Tennessee: Proc. 3rd TN Hydro. Sym., U.S.G.S. Nashville,
TN, pp. 30-37.
522
-------
DeBuchananne, G.D. and R.M. Richardson, 1956, Ground water
resources of east Tennessee: Tenn. Div. of Geol., Bull.
58, Nashville, TN, 393 p.
Faulkerson, J. and H. Mills, 1981, Karst hydrology, morphology,
and water quality in the vicinity of Cookeville, Tennessee:
Report to the City of Cookeville, Tennessee Technological
University, Cookeville, TN, 67 p.
Foxx, Suzanne A., 1981, A summary of the geohydrology and of the
spring recharge zones within the Tennessee portion of the
208 study area of the First Tennessee-Virginia Development
District: TVA Office of Natural Resources, Division of
Water Resources, Report No. WR28-1-520-116.
Hannah, E.D., T.E. Pride, A.E. Ogden, and R. Paylor, 1989,
Assessing ground water flow paths from pollution sources in
the karst of Putnam County, Tennessee: Proc. of Third
Multidisciplinary Conference on Sinkholes: Their Geology,
Engineering, and Environmental Impact, Florida Sinkhole
Res. Inst., Orlando, FL, pp. 183-188.
Hilsenhoff, W.L., 1988, Rapid field assessment of organic
pollution with a family-level biotic index. J.N. Am.
Benthol. Soc., 7(1): 65-68.
Johnson, G., 1991, Personal communication regarding U.S.G.S.
activities in Johnson City, Knoxville Field Office.
Maclay, R.W., 1956, The ground water hydrology of the
Elizabethton and Johnson City, Tennessee area: Unpubl. M.S
Thesis, University of Tennessee, 54 p.
Matthews, M.R., 1986, Groundwater situation assessment of the
Tennessee Valley Region: TVA Office of Natural Resources
and Economic Development, Division of Air and Water
Resources, Chattanooga, TN.
Miller, R.A., 1974, The geologic history of Tennessee: TN
Division of Geology, Bull. 74, Nashville, TN, 63 p.
Mills, H.H., H.N. Taylor, A.E. Ogden, Y. Robinet Clark, and R.
Forde, 1991, Predicting sinkhole flooding in Cookeville,
TN, using SWM model and CIS: Proc. Appalachian Karst Sym.,
Natl. Spel. Soc., Huntsville, AL, pp. 159-167.
Moore, G.K. and J.M. Wilson, 1972, Water resources of the Center
Hill Lake region, TN, State of TN Dept. of Cons., Div- of
Water Res., Series No. 9, 77 p.
Ogden, A.E., 1988, Distinguishing flow regimes of springs in
carbonate rock terranes: Proc. Natl. Water Well Assoc.,
Ground Water Chemistry Conf., Dublin, Ohio, pp. 53-73.
523
-------
Ogden, A.E., W.A. Curry, and J.L. Cummings, 1989, Morphometric
analysis of sinkholes and caves in Tennessee comparing the
eastern Highland Rim and Valley and Ridge physiographic
provinces: Proc. of Third Conference on Sinkholes: Their
Geology, Engineering, and Environmental Impact, Florida
Sinkhole Res. Inst., Orlando, FL, pp. 135-142.
Ogden, A.E., T.L. Brown, and K.G. Hamilton, 1991, Delineation of
wellhead protection areas for municipal-used springs in
eastern Tennessee: UT - Tennessee Water Resources Research
Center, Report #124, Knoxville, TN, 134 p.
Pride, T.E., A.E. Ogden, M.J. Harvey, D.B. George, 1988, The
effect of urban development on spring water quality in
Cookeville, TN: Proc. Natl. Water Well Assoc. 2nd Karst
Conference, Dublin, Ohio, pp. 97-120.
Quinlan, J.F., 1989, Ground water monitoring in karst terranes:
Recommended protocols and implicit assumptions: U.S. EPA
600/X-89/050, Las Vegas, Nevada, 79 p.
Smith, O., 1962, Ground water resources and municipal water
supplies of the Highland Rim in Tennessee: Tennessee
Division of Water Resources, Water Resources Series Number
3, 257 p.
Smithson, K.D., 1975, Some effects of sewage effluent on the
ecology of a stream ecosystem: Unpubl. M.S. Thesis,
Tennessee Technological University, Cookeville, Tennessee,
74 p.
Stanley Consultants, Inc., 1983, Ground water protection plan -
final report: First Tennessee-Virginia Development
District 208 Water Quality Management Program, Johnson
City, TN.
Vannote, R.L., G.W. Minshall, K.W. Cummins, J.R. Sedall, and
C.E. Gushing, 1980, The river contiuum concept: Can. J.
Fish. Aquat. Sci. 37: 130-137.
Washington, H.G., 1984, Diversity, biotic and similarity
indices: A review with special relevance to aquatic
ecosystems: Water Resources Res., 19(6):653-694.
Weber, C., 1973, Biological field and laboratory methods for
measuring quality of surface waters and effluents: USEPA
670/4-73-001. Cincinnati, OH.
Wilhm, J.L., 1970, Range of diversity index in benthic
macroinvertebrate populations: JWCF., 42(5): R221-R224.
Wilson, D., 1985, The physical, chemical, and biological
recovery of Pigeon Roost Creek following the closing of a
524
-------
sewage treatment plant: Unpubl. M.S. Thesis, Tennessee
Technological University, Cookeville, Tennessee, 73 p.
Wilson, T.M., J.A. Gordon, A.E. Ogden, and J.D. Reinhard, 1990,
Detection of failing septic tanks in east Tennessee
utilizing infrared color aerial photographs: Proc. NWWA
Conf. on Eastern Regional Ground Water Issues, Dublin,
Ohio, pp. 177-188.
U.S. Environmental Protection Agency, 1981, 40 CFR Parts 122 and
146 Underground Injection Control Program Criteria and
Standards: Federal Register, V. 46, no. 190, pp. 48243-
48255.
BIOGRAPHICAL SKETCHES
Albert E. Ogden received his B.S. degree in geology from the
Pennsylvania State University and his Ph.D. in hydrogeology from
West Virginia University. He was the hydrogeology professor for
the University of Arkansas from 1976-1981. He then was
assistant director and hydrogeologist for the Edwards Aquifer
Research and Data Center in Texas from 1981-1985. Following
this, Dr. Ogden was the senior hydrogeologist for the RCRA and
Superfund programs within the Idaho Division of Environment from
1985-1987. He presently is an associate professor for the
Center for the Management, Utilization and Protection of Water
Resources, Box 5033, Tennessee Technological University,
Cookeville, Tennessee 38505, (615) 372-3353. Dr. Ogden's
research interests include site characterization of hazardous
waste sites and solving water quality problems in karst
terranes.
Ronald K. Redman received his B.S. degree in biology from
Arkansas Tech University. He then worked as a manager for a
fish hatchery where his duties included water quality
monitoring. He now is working on his M.S. in biology at
Tennessee Technological University, and this paper is a portion
of his thesis research.
Teresa L. Brown received her B.S. degree in geology from James
Madison University followed by graduate work in the Hydrology
and Environmental Science Program at Indiana University. After
college, she worked as a geologist and pollution control
specialist for the Virginia Division of Mined Land Reclamation.
Since 1986, she has been the environmental management
coordinator for the First Tennessee Development District, 207
North Boone Street, Suite 800, Johnson City, TN 37604, (615)
928-0224. Ms. Brown has experience in siting landfills,
assaying the impact of land application of sludge, conducting
water quality surveys, and delineating wellhead protection
areas.
525
-------
The Use of Benthic Macroinvertebrates for Assessing the Impact of Class V Injection
Wells on Carbonate Groundwaters
Albert E. Ogden, Ronald K. Redman, and Teresa L Brown
Question 1. The Shannon-Wiener diversity indices are problematic because they:
a) confound the numbers of species and evenness of numbers of
individuals among species; and
b) count tolerant species, e.g., surface chironomids, the same as
obligate cave species (e.g., Caecidotea isopods). So, what were
the species and their degrees of restriction to caves and
groundwater in springs with outflows from control and contaminated
basins?
Response: No samples were taken in caves.
Question 2. Most, if not all, of the insect orders of your control surface streams are not
normally found in caves. What orders of Crustacea and what kinds of
species (Facultative vs. obligate cave species) did you find in outflows from
your control and contaminated basins?
Response: In the control springs, we found decapods, amphipods, and isopods. We
also found them in some contaminated springs, but amphipods and/or
isopods were usually absent.
526
-------
Session VII:
Ground Water Monitoring
-------
528
-------
The Response of Landfill Monitoring Wells
In Limestone (Karst) Aquifers
To Point Sources and Non Point Sources of Contamination
Ralph O. Ewers Ph.D.
Director, Groundwater Research Laboratory
Department of Geology Eastern Kentucky University,
Ewers Hater Consultants Inc.
Richmond, Kentucky
Abstract
The terms "point source" and "non-point source" in the context
of groundwater contaminants are relative terms. Landfills can be
considered non-point sources, relative to a monitoring well at the
boundary of a typical landfill, if contaminants are released at
hundreds of points, Conversely, the landfill is a point source
relative to these wells if contaminants are released at one or two
points.
This author and several of his colleagues have maintained that
the convergent flow in karst aguifers should make monitoring wells
unreliable. Implicit in these arguments was the assumption that
the landfill was a point source of contamination.
In the non-point source case, virtually every joint and
bedding plane would be exposed to contamination. Thus, any down-
gradient monitoring will which intersects the groundwater should
show contaminants, although they may be unrepresentative of
leachate from the entire landfill.
In the point source case, the monitoring well may not
intersect the conduit which carries the contaminant. This
contention was confirmed at a western Kentucky landfill where
rhodamine-WT was injected into an up-gradient well, simulating a
point source of contamination. Two down-gradient monitoring wells
did not show the dye. although the dye appeared at three off site
springs approximately three miles distant.
Therefore, neither contaminated nor uncontaminated landfill
monitoring wells in karst aguifers should give us confidence in
their reliability.
529
-------
Introduction
Landfills As Non Point Sources Of Contamination
When a monitoring well is placed down-gradient from a landfill
an assumption is made that it will intercept a plume of
contamination if the landfill leaks. If the leachate escapes
uniformly over the entire liner or at hundreds of points, the
traditional system of one monitoring well up gradient and three
down-gradient will signal that this has occurred (Fig. 1).
Relative to the monitoring wells, such a landfill is a non-point
source of contamination.
Although the leakage in this case is homogenous the leachate
may not be. Should the most hazardous waste be located in a single
cell, and if that cell is located along a flow vector midway
between two of the monitoring wells, dispersion may be insufficient
to bring this substance to these monitoring points. The
contamination detected by the monitoring system would not be
representative of the chemical nature of the leachate as a whole.
Traditional monitoring wells may not provide a representative
sample of leachate unless the landfill waste is homogenous.
Landfills As Point Sources Of Contamination
If the leachate passes through the landfill liner at a small
number of discrete points, the traditional monitoring well system
may not signal that this has occurred or it may give an incomplete
understanding of the problem (Fig. 2). Only favorably located
points of leachate escape may develop plumes which intersect the
wells. Relative to the traditional monitoring wells, such a
landfill is a point source of contamination.
This scenario also presents problems of obtaining a
representative sample. Should the landfill waste be less than
homogenous, the most troublesome leachate may go undetected. This
would be the case unless each point of release is sufficiently
distant from the well so that dispersion has time to spread the
contaminants, or unless each point of release is directly up-
gradient from a monitoring well.
Landfills In Karst
An additional complication arises when the aquifer beneath
the landfill is composed of carbonate rocks. These aquifers often
possess very little primary (intergranular) porosity but they
exhibit significant fracture porosity (secondary porosity) and
tertiary conduit porosity. While contaminant plumes may develop
in the primary and secondary porosity, the plumes may be truncated
and their contaminants entrained by the conduit porosity (Fig. 3).
530
-------
Background Monitoring Well
Landfill Boundary
Figure 1. The contaminant plume at a hypothetical landfill which
releases Iqachate uniformly over its entire liner. The stippled
area indicates the plume.
Background Monitoring Well
Landfill Boundary
Monitoring Wells
Figure 2. The contaminant plumes at a hypothetical landfill which
releases leachate at two discrete points. The stippled area
indicates the plume.
531
-------
Landfill Boundary
Figure 3. The contaminant plume at a hypothetical landfill in a
karst aquifer with a high density of solution conduits. Leachate
is released at two discrete points. The presence of conduit
porosity is indicated by the black branching line.
Figure 4. The contaminant plume at a hypothetical landfill in a
karst aquifer with a low density of solution conduits. Leachate
is released at two discrete points. The presence of conduit
porosity is indicated by the black branching line.
532
-------
If the landfill leaks at discrete points, the point source
case, there is a considerable likelihood that the traditional
monitoring wells will not function as they were intended. Quinlan
and Ewers (1985) made this point, although they did not
specifically describe the development of plumes and their spatial
relationship to the conduits. The probability of the wells working
as intended in carbonate aquifers will depend not only upon the
geometric relationship of the points of leachate release and the
monitoring wells, but also upon the density of conduits in the
aquifer. If the conduits are widely spaced, the probability of
their being located between the contaminant source and the
monitoring wells will be less and the monitoring wells are more
likely to be efficacious (Fig. 4). Conversely, if the density of
conduits in the aquifer is great, it would be unlikely that the
wells would be useful.
If the landfill is a non-point source, virtually every joint
and bedding plane beneath the landfill and immediately down-
gradient would be affected by leachate. Although some portion of
the leachate may be intercepted by conduits, any nearby down-
gradient monitoring well will intercept some leachate. However,
samples from such a well are unlikely to be representative of the
landfill leachate unless the waste is homogenous.
A Kentucky Landfill Example
A recent study at a Western Kentucky landfill confirmed the
contention that traditional monitoring wells may be ineffective in
limestone aquifers. Two down-gradient monitoring wells at this
site were tested for efficacy by injection of a tracer into the up-
gradient background monitoring well. The landfill is located in
a karst region. The aquifer beneath the site is a Mississippian
age limestone, the upper member of the St. Louis Formation.
Four pounds of 20 percent Rhodamine-WT dye was injected into
the up gradient well. The dye was diluted in 50 gallons of water
before injection. The well was first tested by an injection of
several hundred gallons of potable water. This insured that the
well would accept the dye and also served to wet the pathways that
the dye could follow. After the dye injection several hundred
gallons of potable water was injected into the well to flush the
dye into the aquifer.
The results of the dye study are depicted in Figure 5.
Rhodamine-WT from the well injection appeared in the Town Spring
at Cadiz, Kentucky and an adjacent spring. These two sites are
probably part of a local distributary system. This was the
principle location of dye recovery from the well injection.
Rhodamine-WT was also found in very low concentrations at two other
springs, one north and the other southeast of the landfill. The
533
-------
N
Muddy Fork, Little River
LEGEND
• Spring
O Swallet
Q Monitoring Well
A Dye Injection Well
Miles
Figure 5. Results of a dye study at a Western Kentucky landfill.
The long arrows indicate the schematic path of the dye. The
principal dye recovery point was at the Cadiz town spring. The
local gradient at the landfill is indicated by the short arrows.
534
-------
dye reached all of these sites nearly three miles distant in less
than four days.
No dye was recovered from charcoal dye detectors in the two
proposed down-gradient monitoring wells located 1100 and 1300 feet
from the dye injection well. Both of these wells were equipped
with airlifts which continuously circulated the water in the well
over the dye detectors. Dye detectors were exchanged regularly
over a period of 38 days. Apparently the tracer was intercepted
by conduit porosity before reaching the monitoring wells.
Appropriate Methods Of Monitoring In Karst
Quinlan and Ewers (1985) suggest that appropriate monitoring
methods in karst should utilize springs. When it can be
demonstrated by proper dye-tracing procedures that a particular
spring drains a landfill site it may be the most appropriate and
dependable monitoring site. However, springs are, in some
instances, demonstrably inappropriate monitoring sites, even though
they are shown to be connected to the area of potential
contamination release. This is the case at the cited Kentucky
example for two reasons. First, the spring is quite large and
therefore the dilution of landfill leachate would be very great.
Only a near catastrophic release would likely be detectable.
Second, the probable recharge basin for the spring contains many
potential sources of contamination. It could be very difficult to
assure that a given contaminant was the result of landfill leakage
or due to some other problem.
A alternative monitoring procedure which is more likely to be
specific to the landfill would be to employ a pumped monitoring
well at the site. During dye tracing programs in karst, wells are
frequently employed as dye recovery sites. Those that are even
modestly pumped during the tracing period often show the dye, those
which are not pumped rarely do. An excellent example can be found
in the work of Aley at a proposed landfill site near Pindall,
Arkansas (Aley, 1986). In this study, dye was recovered from five
wells, ranging up to a mile from both the dye injection point and
the spring resurgences where the dye appeared. Apparently, local
gradients serve to carry dye, and presumably contaminants, to the
conduits, establishing the convergent flow which is characteristic
of these aquifers. Pumping stress apparently reverses these local
gradients, extracting the entrained dye from the conduit (Fig. 6).
If tracer dyes can be recovered from pumped wells in karst,
by inference, contaminants could also be recovered from such wells.
A protocol for establishing a pumped monitoring well would be as
follows.
1- Locate the lowest point on the potentiometric surface, down-
gradient from the site to be monitored. This could be done with
535
-------
Pumped Monitoring Well
Figure 6. A pumped monitoring well lowers the potentiometric
surface (dotted line) near a conduit which is carrying
contaminants. This pumping stress draws contaminants to the
monitoring well that would otherwise remain entrained in the
conduit.
536
-------
several small-bore temporary piezometers. Low points of the
potentiometric surface correlate well with the location of conduits
(Quinlan and Ewers, 1981).
2- Install a well at this point and outfit it with a submersible
domestic well pump having a capacity sufficient to create drawdown
in nearby piezometers in a short time. A site so located should
be near conduits which may drain from the landfill.
3- Qualify the well with a dye test from an appropriate up gradient
site or sites. This would help to define the capture zone of the
proposed pumped monitoring well. If the capture zone is sufficient
to encompass the site it could be used as a monitoring point.
4- Sample the well for contaminants on the required schedule after
an appropriate amount of pumping has occurred. The amount of
pumping required could be determined from "3" above.
Two monitoring wells of this type are ready to be installed at two
sites, one in Kentucky and one in Puerto Rico.
Conclusions
Clearly, the traditional monitoring wells in the cited example
cannot be relied upon to accurately indicate the performance of the
landfill's clay liner. If one of the wells should show the
presence of leachate, one cannot be confident that it monitors the
entire landfill or that it fairly represents the character of all
of the leachate that is released. Neither contaminated nor
uncontaminated traditional monitoring wells in karst aquifers
should give us confidence in this method of monitoring. Springs
or pumped monitoring wells offer a more reliable alternative.
REFERENCES CITED
Aley, T., 1986/1989. Assessing presumptions of negligibly slow
downward water movement through deep residuum beneath a proposed
landfill site, Pindall Arkansas. National Water Well Association
Karst Hydrology Course Manual, 1989.
Quinlan, J.F. and R.O. Ewers, 1985. Groundwater flow in limestone
terranes: strategy rational and procedure for reliable efficient
monitoring of ground water quality in karst areas. Proceedings of
the Fifth National Symposium and Exposition on Aquifer Restoration
and Ground Water Monitoring Proceedings, pp 197-234.
Quinlan J.F. and R.O. Ewers, 1981. Hydrogeology of the Mammoth Cave
Region, Kentucky, in Roberts, T,G., ed. , 1981 GSA Cincinnati '81
Field Trip Guidebooks: Washington D.C., Am. Geol. Inst., v. 3, p.
496-501.
537
-------
Biographical Sketch
Ralph O. Ewers is professor of geology and director of the
Groundwater Research Laboratory at Eastern Kentucky University, and
a principal in Ewers Water Consultants, a consulting firm
specializing in carbonate aquifers. His B.S. and M.S. degrees in
geology were earned at the University of Cincinnati and his Ph.D.
was earned at McMaster University (1982) . Professor Ewers's
special interests include the applications of tracer and electronic
monitoring techniques to the solution of practical environmental
problems in karst groundwater. He has more than 30 years of
research and consulting experience gained throughout much of North
America and Europe and he serves on several state groundwater
advisory boards. He is author or co-author of more than 100 papers
and reports dealing with groundwater in karst. For one of the
papers concerning procedures for monitoring ground water in karst
terraces, he and co-instructor James Quinlan received the 1986 E.B.
Burwell Award from the Geological Society of America for a "work
of distinction which advances knowledge concerning principles of
practice of engineering geology."
538
-------
The Response of Landfill Monitoring Wells In Limestone (Karst)
Aquifers To Point Sources and Non-Point Sources of Contamination
Ralph O. Ewers
1. In England and Wales, 56 domestic (municipal?) waste sites are
located on the karstified carboniferous limestone. Unpumped
monitoring wells have been successfully used to monitor contaminant
plumes leaving the sites, yet only 4 springs have been affected.
Why do you think so few springs have been adversely affected?
Why do you think the monitoring wells have been so successful?
ANSWER- I suspect that the reason so few springs seem to be
contaminated is that dilution of the contaminants is too great.
This is one of the reasons I believe that springs are often
inappropriate monitoring points.
A successful monitoring well can be defined several ways. If
you define success as having no contamination, you may be deluding
yourself. The aquifer may be contaminated you are just unaware of
that fact. If you define success as showing some contamination,
you may be unaware of the real scope of the problem.
2. Once you have determined the most appropriate well location
(at the lowest point on the potentiometric surface), how do you
determine the "appropriate" pumping rate?
ANSWER- I would judge the pumping rate and time on the basis of the
tracer tests that are used to qualify the well. If pumping at a
given rate gives a strong dye concentration which appears to
plateau after pumping 2000 gallons of water, I would assume that
would be an appropriate rate and quantity-
3. In many county landfills in Tennessee, no municipal sewer
system is available to receive water pumped as you propose. The
volume of water you imply would be pumped would be a problem. Do
you have any suggestions for disposal of such pumped water?
ANSWER- I would store the water until after the analysis was
complete and then release it, if it is clean, to some appropriate
receiving stream, assuming the appropriate discharge permits were
obtained. Alternatively, it could be trucked to a disposal
facility.
4. How often do you propose to continue pumping to determine if
there is leakage? What do you propose to do with the volume of
water? Once the initial pumping test has been conducted and, for
example, the results indicated that the landfill is not leaking
are your proposing to cease pumping at the monitoring point until
the next scheduled monitoring date, at which time you would begin
the entire pumping and sampling procedure again? Would additional
539
-------
wells be installed to enhance interception of the leakage?
ANSWER- I would pump as often as required by law or good science.
Perhaps quarterly, I would prefer to pump at during low-flow
conditions. I am not sure that high-flow samplings are necessary
as they may be for spring monitoring.
See 3 above regarding disposal of the water.
I see no reason at this time to continue pumping between
sampling periods.
Additional pumped monitoring wells may be required if the
landfill is large or if the dye test results indicate that it is
infeasible to extend the capture zone over the required area.
5. I suspect that the cost of continuing to analyze for
pollutants in your well pumped to draw water from the conduit, plus
the cost of water handling and disposal, would make it cheaper to
combine a natural potential survey with your potentiometric map and
drill several "pilot" holes (let's say 5) as potential monitoring
wells. If one of the 5 holes hits the conduit (as proposed by
tracing) you have maximum reliability and no problems with water
disposal. Convince me that I am wrong!
ANSWER- I doubt that the natural potential method is developed
sufficiently at this time to find the smaller conduits which are
likely to be associated with the average landfill. If a large
conduit is suspected, the type investigated by A. Lang and others,
I would certainly advise using this technique.
540
-------
DEVELOPMENT OF AN ASTM STANDARD GUIDE
FOR THE DESIGN OF GROUND-WATER MONITORING SYSTEMS
IN KARST AND FRACTURED ROCK TERRANES
Michael R. McCann, Westinghouse Electric Corp.
Bloomington, Indiana
James F. Quinlan, Quinlan & Associates
Nashville/ Tennessee
ABSTRACT
Permeability in karst terranes and many fractured rock
terranes is developed in joints, fractures, and bedding planes. In
karst terranes some of these three discontinuities have been
solutionally enlarged, sometimes to form conduit-like caves. In
both terranes the assumptions underlying the description of ground-
water flow as taking place in a homogeneous, isotropic, porous
medium are usually invalid. As a result, ground-water flow and
pollutant flow in most karst terranes and many fractured rock
terranes is not described by the radially dispersive charac-
teristics of flow in a granular medium. Monitoring systems
designed with the assumption of dispersive flow will not produce
accurate, reliable, or efficient monitoring of either ground-water
flow or pollutant flow.
A Standard Guide, developed under the auspices of ASTM Sub-
committee D18.21.09 on Special Problems of Monitoring Karst and
Fractured Rock Terranes, has been deemed necessary to assist in the
design and implementation of accurate and reliable monitoring
systems in those hydrogeologic settings which depart significantly
from the characteristics of a porous medium. This Guide is based
on recognized methods of monitoring system design and imple-
mentation for the purpose of collecting representative ground water
data. The design guidelines are applicable both to the detection
of contaminant transport from existing facilities and to the
assessment of proposed facilities. The recommended procedures of
the Guide are designed to obtain representative aquifer and ground-
water information. Use of the Guide will assist in the development
of accurate conceptual hydrogeological models which are integral to
the design of a time-sensitive, relevant, and reliable monitoring
program.
541
-------
INTRODUCTION
The American Society for Testing and Materials, better known
as ASTM, was organized in 1898 and has grown into one of the
largest voluntary standards development systems in the world. It
is a non-profit organization that provides a forum for producers,
users, and all others having a general interest in materials,
products, systems, and services. Its 134 committees and additional
subcommittees write standard test methods, specifications,
practices, terminology, guides, and classifications for metals,
paints, plastics, textiles, petroleum, construction, energy, the
environment, consumer products, medical devices, computerized
systems, electronics, and many other areas. More than 8500 ASTM
standards are published each year in the 68 volumes of the Annual
Book of ASTM Standards.
One of the 134 ASTM Committees, D18, is concerned with soil
and rock. One of its subcommittees, .21, is concerned with
groundwater monitoring, and it has been divided into 10 Task
Groups, one of which, 21.09, is concerned with monitoring in karst
and fractured rock terranes. This paper is written to tell what
that task group is doing and why. But first, we wish to briefly
establish that the problems of karst and fractured rock aquifers
are different from those of other aquifers.
HOW KARST AND FRACTURED ROCK AQUIFERS PHYSICALLY DIFFER FROM OTHERS
The main physical difference between aquifers in karst and
fractured rock terranes and aquifers in other terranes is the type
of permeability (effective porosity). In aquifers consisting of
unconsolidated material, and in many consolidated aquifers, the
permeability is a primary feature of composition and deposition.
In fractured and karst aquifers, however, the permeability is
secondary, a result of fractures, bedding planes, and commonly
dissolution along them. Underlying the description and analysis of
ground water and contaminent flow is the implicit assumption that
the aquifer is behaving as a porous medium. Even though it is
widely recognized that ground-water flow in many rock aquifers is
through secondary openings, the assumption that it is behaving as
a porous medium is still made at some working scale of
representation. The problem in many fractured rock aquifers, and
most karst aquifers, is that this assumption is made uncritically.
HYDRAULIC AND HYDROLOGIC DIFFERENCES
A result of the porous media assumption is the further
assumption that the aquifer is homogeneous and isotropic, and that
flow is laminar. The range in aquifer characterstics within the
spatial limits of the aquifer is frequently large in fractured
rocks, and most karst aquifers are strictly non-homogeneous.
Fractured rock and karst aquifers are typically highly anisotropic
542
-------
in three directions. Hydraulic conductivity and ground-water
velocity can range over several orders of magnitude, depending upon
the direction of ground-water movement. For example, in extreme
cases in some karst aquifers, ground water appears to be moving
parallel to the equipotential lines of a mapped potentiometric
surface (Arnow, 1963) . This is, of course, due to an inadequate
amount of data points used to describe the surface. Nevertheless,
until another means is utilized to check the ground-water flow
direction, a potentiometric map can be misinterpreted to be a
correct representation under the porous media assumption.
Flow in porous media is assumed to be laminar over the
macroscopic scale of investigation. Although this assumption is
reasonable for the average of hydraulic properties in a porous
medium, water movement in fractures, especially solutionaly
enlarged fractures, is often rapid and turbulent. This is due not
only to size of the apertures in the rock, but also to the many
direct inputs from the surface to the aquifer. This rapid and
turbulent flow in both recharge and movement through the aquifer
often results in large variations in head in response to
precipitation, and large spatial and temporal variations in water
chemistry. Monitoring wells installed in these terranes vary in
the number and type of fractures encountered from well to well.
Some wells encounter larger fractures with good connection to
surface inputs or major conveyances; others encounter relatively
"tight" and non-transmissive fractures. The result is poten-
tiometric maps that can be difficult to interpret or are entirely
misleading. For example, Figure 1 shows a potentiometric map for
a site in an Indiana karst terrane during the February wet season.
In fact, ground water from this site drains to the southwest, as
demonstrated by tracer tests. Such a conclusion, however, could
not be interpreted from this map. The tracer tests were needed to
test and establish flow directions and destinations.
DIFFERENCES IN DESCRIPTION OF GROUND-WATER AND CONTAMINANT FLOW
Standard equations describing ground-water flow are based on
Darcy's Law which assumes laminar flow in a porous medium, and is
invalid in nonlaminar flow and other media. The equations are not
valid for karst and fractured rock aquifers because flow in them is
neither laminar nor in a porous medium. This means that the
parameters for transmissivity and storativity which may be
calculated from pumping tests are not representative of the aquifer
as a whole. Consequently, predictions of contaminant movement
based on assumed radial dispersion and advection, and calculated
using these aquifer parameters, will be erroneous and misleading in
both direction and magnitude. Monitoring systems designed on false
assumptions will not produce accurate, reliable, efficient, or
time-sensitive monitoring. Examples from case studies of the
inadvisability of applying the porous-media assumption to the
design of ground-water monitoring systems are given by: Aley
(1988), Bradbury et al. (1990), Crawford (1988), Field (1988),
McCann and Krothe (1991), Quinlan and Ewers (1985), and Quinlan
543
-------
N
1W-4S
0 200
Scale ir\ feet
Figure 1 - Lemon Lane Landfill Potentiometric Map, February 1988
(after McCann and Krothe, 1991)
544
-------
(1989). Table 1 gives a comparision of characteristics of porous
media, fractured rock aquifers, and karst aquifers, and highlights
their significant differences.
WHY A STANDARD GUIDE IS NECESSARY
Quinlan (1989), in the foreword of his document written for
the EPA, "Ground-Water Monitoring in Karst Terranes: Recommended
Protocols and Implicit Assumptions", stated four major reasons why
that document was necessary. They are also four good reasons why
an ASTM Standard Guide is necessary: 1) The hydrology of karst
terranes is significantly different from that of terranes
characterized by granular and fractured rocks — flow velocities in
karst may be several orders of magnitude higher than in other
ground-water settings; Darcy's Law describing flow is rarely
applicable; 2) For monitoring to be relevant and reliable in karst
terranes, monitoring procedures must be radically different from
those in non-karst terranes; 3) There is a need for a practical
guide that tells engineers, geologists, hydrologists, and
regulators what the monitoring problems are in karst terranes and
how to solve them; and 4) Create awareness of the state-of-the-art
in monitoring in karst terranes — and provoke thought and
discussion about the subject and its implications for ground-water
protection strategy. Although Quinlan (1989) focused specifically
on karst terranes, ASTM Subcommittee D18.21.09 recognized that many
fractured rock aquifers share similar problems in monitoring design
and decided to include them so the Guide would have the widest
practical application possible.
Almost 20% of the United States and 40% of the country east of
Tulsa, Oklahoma, is underlain by carbonate rocks. The extreme
vulnerabilty of carbonate aquifers in these terranes to contam-
ination make it imperative that a reliable guide to monitoring be
made available. Very few state agencies have guidelines that take
into account aquifers that do not behave as porous media. However,
some states such as Alabama, Indiana, Kentucky, Missouri, and
Tennessee have begun to require water tracing studies to confirm
ground-water flow directions in their karst terranes. Although
some EPA documents do address the problem of monitoring in non-
porous-media aquifers, most do not. In particular, the Technical
Enforcement Guidance Document (TEGD) which governs monitoring at
RCRA sites assumes a porous media exists in most applications of
its monitoring guidance. Its one brief discussion of karst, p. 66-
68, is ambiguous and probably erroneous; no data are given for the
conclusions reached. Few universities, in their hydrogeology
programs, specifically address how to characterize aquifers that do
not behave as porous media.
A Standard Guide, developed through the extensive peer review
and consensus process of ASTM, will provide a much needed and
nationally disseminated standard on which to base design of
monitoring systems in karst and fractured rock terranes. Such a
Standard Guide does not yet exist. The thorough and complete ASTM
545
-------
AQUIFER
CHARACTER-
ISTICS
Permeability
Flow
Isotropy
Homogeneity
Flow
predictions
Storage
Head
variation
Water
chemistry
variation
AQUIFER TYPE
POROUS MEDIA FRACTURED ROCK
Mostly primary
Slow, laminar
Most
isotropic
Most
homogeneous
Darcy's law
usually applies
Within
saturated zone
Minimal
variation
Minimal
variation
Mostly secondary
Possibly fast
and turbulent
Less
isotropic
Less
homogeneous
Darcy's law
may not apply
Within
saturated zone
More
variation
More
variation
KARST
Almost
entirely
secondary
Likely fast
and turbulent
Highly
anisotropic
Non-
homogeneous
Darcy's law
rarely applies
Both in
saturated &
un-saturated
(epikarstic)
zone
Can have
extreme
variation
Can have
extreme
variation
Table l - Comparison of porous, fractured, and karst aquifers
(adapted from Bradbury et al., 1990)
546
-------
process would best produce the necessary document which would have
the widest practical application and acceptance. Such a document
would be most useful to three groups of people: 1) Administrators
who must evaluate existing or proposed networks for monitoring
water quality in karst or similar terranes, but have minimal
experience in non-porous-media hydrogeology; 2) Consultants and
others who must design monitoring networks but may or may not have
extensive experience in karst or similar terranes; and 3) The
well-experienced water tracer who is already familiar with the
hydrology and geomorphology of karst or similar terranes but has
minimal familiarity with monitoring problems.
HOW A STANDARD GUIDE WILL ASSIST IN DEVELOPING A RELIABLE
MONITORING SYSTEM
As to be explained within the Standard Guide, there are three
main objectives to be accomplished in assisting the development of
a reliable monitoring system: 1) The Guide will aid in the
accurate characterization of karst- and fractured-rock aquifers
because the development of a conceptual hydrogeological model that
identifies and defines the various components of the flow system is
a necessary first step to the design of a monitoring system; 2)
The Guide will be based on recognized methods of monitoring system
design and implementation for the purpose of collecting
representative ground-water data. The design guidelines are
applicable both to the determination of ground-water flow and
contaminant transport from existing sites and assessment of
proposed sites; 3) The objectives of the Guide are to recommend
procedures for obtaining information on aquifer characteristics and
representative water quality.
In order to accomplish these objectives, the Guide will
discuss and compare qualitative differences in aquifer
characteristics between porous, fractured, and karst media. The
special characteristics of karst and fractured-rock aquifers will
be pointed out, and explanations of why typical investigative
procedures often fail to give accurate information will be
provided. The Guide will recommend procedures to obtain repre-
sentative aquifer and ground-water information. These recommended
procedures include: 1) Criteria for determining whether or not an
aquifer under investigation is behaving as a porous medium; 2)
Procedures for potentiometric mapping; 3) Procedures for
interpreting aquifer tests; 4) Criteria for determining when
tracing is necessary; 5) Procedures for conducting tracer tests;
6) Criteria for determining monitoring station location; 7)
Criteria for establishing sampling protocol; 8) Procedures for
interpreting chemical and hydrologic data; and 9) Meeting
regulatory requirements.
It is not the intent of the Guide to provide a rulebook on how
monitoring must be conducted in non-porous-media aquifers. Rather,
the members of ASTM Sub-committee D18.21.09 wish to provide the
users of the Guide with the criteria and procedures to conduct an
547
-------
investigation so that they may determine the most accurate,
reliable, efficient, and time-sensitive monitoring program
possible.
People interested in assisting in formulating and reviewing
the Guide described herein are invited to contact either of its
Task Group co-chairman, the authors of this paper. Such parti-
cipation involves attendance at meetings, review of documents, and
voting on drafts proposed for adoption. Membership in sub-
committees developing Standards or Standard Guides is open to all,
but only members of ASTM can vote on them.
REFERENCES CITED
Aley, T., 1988. Complex radial flow of ground water on flat-lying
residuum - mantled limestone in the Arkansas Ozarks. In:
Environmental Problems in Karst Terranes and Their Solutions
Conference (2nd, Nashville, Tenn.), Proceedings. National
Water Well Association, Dublin, Ohio. pp. 159-170.
Arnow, T., 1963. Ground water geology of Bexar County, Texas.
U.S. Geological Survey, Water Supply Paper 1588. 36 pp.
Bradbury, K.R., Muldoon, M.A., Zaporozec, A., and Levy, J.. 1991.
Delineation of wellhead protection areas in fractured rocks.
Guidelines for Delineation of Wellhead Protection Areas, U.S.
Environmental Protection Agency, Office of Ground-Water
Protection, Washington, D.C.. EPA 570/9-91-009. 144 pp.
Crawford, N.C., 1988. Karst ground water contamination from
leaking underground storage tanks: Prevention, monitoring
techniques, emergency response procedures, and aquifer
restoration. In: Environmental Problems in Karst Terranes and
Their Solutions Conference (2nd, Nashville, Tenn.), Proceed-
ings. National Water Well Association. Dublin, Ohio. pp.
213-226.
Field, M.R., 1988. U.S. Environmental Protection Agency's strategy
for ground-water quality monitoring at hazardous waste land
disposal facilities located in karst terranes. In: Inter-
national Association of Hydrogeologists, Congress (21st,
Guilin, China), Proceedings, vol. 2, pp. 1006-1011.
McCann, M.R., and Krothe, N.C., 1991. Development of a monitoring
program at a Superfund site in a karst terrane near
Bloomington, Indiana. In: Hydrogeology, Ecology, Monitoring,
and Management of Ground Water in Karst Terranes Conference
(3rd, Nashville, Tenn.), Proceedings. National Ground Water
Association. Dublin, Ohio. (in press).
548
-------
Quinlan, J.F., 1989. Ground-water monitoring in karst terranes:
Recommended protocols, regulatory problems, and implicit
assumptions. U.S. Environmental Protection Agency,
Environmental Monitoring Systems Laboratory, Las Vegas,
Nevada. EPA/600/X-89/050. [Draft; final version to be
released 1991] 79 pp.
Quinlan, J.F. and Ewers, R.O., 1985. Ground water flow in lime-
stone terranes: Strategy, rationale, and procedure for
reliable, efficient monitoring of ground-water quality in
karst areas. In: National Symposium and Exposition on
Aquifer Restoration and Ground Water Monitoring (5th,
Columbus, Ohio), Proceedings. National Water Well Associ-
ation. Worthington, Ohio. pp. 197-234.
BIOGRAPHICAL SKETCHES
Michael R. McCann is Senior Project Geologist for Westinghouse
Electric Corporation's Bloomington Project. He supervises geologic
and hydrologic investigations and ground water monitoring for the
six PCB-contaminated Superfund sites near Bloomington, Indiana.
Previously he was a consultant with Skelly and Loy Engineers in
Lexington, Kentucky, and prior to that he was hydrogeologist for
the Kentucky Division of Water. Mr. McCann was also assistant
geologist at Mammoth Cave National Park. He holds a B.S. degree
from Indiana University, a M.S. degree from the University of
Kentucky, and is a Certified Professional Geologist with the state
of Indiana. Mr. McCann is currently serving as co-chairman of ASTM
subcommittee D18.21.09 on Special Problems of Monitoring Karst and
Other Fractured Rock Terranes.
Michael R. McCann
Westinghouse Electric Corp.
P.O. Box 997
Bloomington, IN 47402
812-334-0030
Dr. James F. Quinlan, P.G., is president of Quinlan and
Associates, a consulting firm specializing in problems of carbonate
terranes. He was Research Geologist for the National Park Service
at Mammoth Cave, Kentucky, for 16 years and has been an independent
consultant on karst for more than 10 years. He earned a Ph.D. in
geology at the University of Texas at Austin (1978) . His
experience includes 36 years of research and observations in karst
terranes of 26 states, 2 territories, and 23 countries and work as
a consultant in many of them. This experience has included
environmental applications of dye-tracing, evaluation of waste-
disposal sites, design of ground-water monitoring networks, peer-
review of reports and both analysis and remediation of sinkhole
development. He has written or co-written more than 160 publi-
cations on karst-related topics. For one of the papers concerning
549
-------
procedures for monitoring ground water in karst terranes, he and
co-author Ralph Ewers received the 1986 E.B. Burwell Award from the
Geological Society of America. For the past 6 years, he and two
others have annually taught a NWWA course on Practical Karst
Hydrogeology, with Emphasis on Ground-Water Monitoring. He served
for 4 years as a Director of the Association of Ground Water
Scientists and Engineers and is co-chairman of ASTM Subcommittee
D18.21.09 on Special Problems of Monitoring Karst and Fractured
Rock Terranes.
Dr. James F. Quinlan
Quinlan & Associates, Inc.
P.O. Box 110539
Nashville, TN 37222
615-833-4324
550
-------
DEVELOPMENT OF AN ASTM STANDARD GUIDE FOR THE DESIGN OF GROUND
WATER MONITORING SYSTEMS IN KARST AND OTHER FRACTURED ROCK TERRANES
by Michael R. McCann and James F. Quinlan
It is appropriate to mention that funding of the entire ASTM ground
water monitoring standards effort — all 10 committees — is being
paid for by the U.S. Navy, the U.S. Environmental Protection
Agency, and the U.S. Geological Survey.
We gladly acknowledge this support.
551
-------
552
-------
GROUND-WATER REMEDIATION MAY BE ACHIEVABLE IN SOME KARST AQUIFERS
THAT ARE CONTAMINATED, BUT IT RANGES FROM UNLIKELY TO IMPOSSIBLE
IN MOST: I. IMPLICATIONS OF LONG-TERM TRACER TESTS
FOR UNIVERSAL FAILURE IN GOAL ATTAINMENT
BY SCIENTISTS, CONSULTANTS, AND REGULATORS
JAMES F. QUINLAN and JOSEPH A. RAY*
Quinlan & Associates, Inc.
Nashville, Tennessee
*Now with
Kentucky Division of Water
Frankfort, Kentucky
EXTENDED ABSTRACT
Not a single contaminated aquifer in the United States has
been confirmed to have been successfully restored (remediated) by
conventional "pump and treat" technology! At present, "No matter
how much money the federal government is willing to spend, contami-
nated aquifers can not be restored to a condition compatible with
health-based standards." (Travis & Doty, 1990). Numerous distin-
guished authorities agree (Feenstra and Cherry, 1988; Freeze and
Cherry, 1989; Mackay and Cherry, 1989; Mackay, 1990; Rowe, 1991).
Many karst hydrogeologists, including the authors, have
believed that aquifer remediation in karst terranes might be
quicker and more easily achieved as a consequence of the rapid
flow-through times and resultant short residence-times characteris-
tic of most karst aquifers. This might be so. This might be
especially so in highly integrated, maturely karsted aquifers
typified by much of the Mammoth Cave area, much of western Kentucky
and southern Indiana, and much of the midwest and southeast. We
now, however, doubt the probability of quick and easy remediation
and self-cleansing — except where contaminants might enter the
aquifer through swallets and sinkholes draining directly to
underground streams and to springs. Recent multi-tracer investiga-
tions, begun in the epikarst zone of karst aquifers characterized
553
-------
chiefly by diffuse flow, strongly suggest that the probability of
remediation in most karst aquifers, diffuse- or conduit-flow, may
range from unlikely to nil. (The tracer dyes were used as
surrogate pollutants.)
Interpretation of judicious extrapolations of tracer-recovery
data from a total of 9 randomly-located dye-injection wells at
sites in the folded Appalachians of Pennsylvania and Tennessee —
as contrasted with discrete inputs at sinking streams or cave
streams — data from more than a thousand water samples and
hundreds of activated-charcoal-detectors, all regularly collected
over periods ranging from 6 to 20 months, strongly indicates that
dye-residency times in the sub-water-table part of many karst
aguifers can range from several years to many tens of years
(Quinlan, 1992; Quinlan and Ray, 1991; Quinlan, et al. , 1990a,
1990b, 1991) .
Application of "pump and treat" technology at an Appalachian
superfund site in Beekmantown (Knox) dolomites has prevented some
migration of dye from it, but has not diminished the long-term
concentration of dye in the aquifer. Similar long-term residency
of dye (and, therefore, of pollutants) in parts of epikarst
aquifers propinquitous to those parts with rapid flow-through and
short-term residency has been recognized by others also (Smart and
Friedrich, 1986; Alexander et al., 1991).
An assumption that rapid remediation might be characteristic
of karst and epikarst aquifers in, for example, the Mammoth Cave
area, ignores the following facts: 1) More than 99% of the
approximately 600 dye-tests performed there have been from sinking
streams and cave streams rather than from wells randomly drilled
into the diffuse-flow part of the aquifer (between conduits and
their tributaries), and 2) No tracer tests have been performed in
the epikarst there. The time necessary for possible remediation of
an aquifer in the Mammoth Cave area would be highly variable and a
function of where pollutant input occurs above the aquifers.
Remediation there could be practically impossible.
These conclusions are applicable to most karst areas and are
relevant to interpretation of traces performed as part of spill
response in most karst terranes.
REFERENCES CITED
Alexander, E.G., Jr., Alexander, S.C., Huberty, B.J., and Quinlan,
J.F. The Oronoco Landfill dye trace III: Results from a
Superfund remedial investigation in a glaciated, diffuse-flow
karst. Hydrogeology, Ecology, Monitoring, and Management of
Ground Water in Karst Terranes Conference (3rd, Nashville,
Tenn.), Proceedings. National Ground Water Association,
Dublin, Ohio. [in this volume]
Feenstra, S., and Cherry, J.R. 1988. Subsurface contamination by
554
-------
dense non-aqueous phase liquid (DNAPL) chemicals. Interna-
tional Association of Hydrogeologists, International Groundwa-
ter Symposium (Halifax, N.S.), Proceedings, p. 16-69.
Freeze, R.A., and Cherry, J.A. 1989. What has gone wrong? Ground
Water. 27 (4) :458-464.
Mackay, D.M. 1990. Characterization of the distribution and be-
havior of contaminants in the subsurface, in Water Sciences
and Technology Board. Ground Water and Soil Contamination
Remediation: Toward Compatible Science, Policy and Public
Perception. National Academy Press, Washington, B.C. p. 70-
90.
Mackay, D.M., and Cherry, J.A. 1989. Groundwater contamination:
Pump-and-treat remediation. Environmental Science and Tech-
nology. 23:630-636.
Quinlan, J.F. 1992. Interpretation of dye-concentration and dye-
recovery curves from four simultaneous interwell traces in a
karst aquifer. Privately published Christmas card, [Nash-
ville, Tenn.] 1 p.
Quinlan, J.F., and Ray, J.A. 1991. Application of dye-tracing to
evaluation of a landfill site in a karst terrane: Rationale
and an Appalachian case study (abs.). Geological Society of
America, Abstracts with Programs. 23(3):54.
Quinlan, J.F., Ray, J.A., and Elliott, W.G. 1990a. Flowpath de-
lineation, interpretation, and control in Pennsylvania karst
terrane: Utilization of dye-tracing to successfully accomplish
and monitor these objectives (abs.). Geological Society of
America, Abstracts with Programs. 22 (7):371.
Quinlan, J.F., Ray, J.A., Elliott, W.G., Smith, A.R., and Behrens,
H., 1990b. Interwell tracing with simultaneous use of four
different fluorescent dyes in a Pennsylvania karst terrane
(abs.) Geo , 17(2-3):82.
Quinlan, J.F-, Ray, J.A., and Schindel, G.M. 1991. Application of
dye tracing to evaluation of a landfill site (abs.), in Kast-
ning, E.H., and Kastning, K.M., eds. Appalachian Karst. Na-
tional Speleological Society, Huntsville, Ala. p. 168.
Rowe, W.D., Jr. 1991. Superfund and groundwater remediation:
Another perspective. Environmental Science and Technology.
25(3):370-371.
Smart, P.L., and Friedrich, H. 1986. Water movement in the un-
saturated zone of a maturely karstified carbonate aquifer,
Mendip Hills, England. Environmental Problems in Karst Ter-
ranes and Their Solutions Conference (Bowling Green, Ky.),
Proceedings. National Water Well Association, Dublin, Ohio.
p. 59-87.
555
-------
Travis, C.C., and Doty, C.B. 1990. Can contaminated aquifers at
Superfund sites be remediated? Environmental Science and
Technology. 24 (10):1464-1466.
BIOGRAPHICAL SKETCHES
Dr. James F- Quinlan, P.G., is president of Quinlan & Associates,
Inc., a consulting firm specializing in problems of carbonate
terranes. He was Research Geologist for the National Park Service
at Mammoth Cave, Kentucky, for 16 years and has been an independent
consultant on karst for more than 10 years. He earned a Ph.D in
geology at the University of Texas at Austin (1978). His experi-
ence includes 36 years of research and observations in karst
terranes of 26 states and 25 countries, and work as a consultant in
many of them. He has written or co-written more than 170 publica-
tions on karst-related topics. He is co-chairman of ASTM Subcom-
mittee D18.21.09 on Special Problems of Monitoring Karst and Other
Fractured Rock Terranes.
Dr. James F. Quinlan
Quinlan & Associates, Inc.
Box 110539
Nashville, TN 37222-0539
(615) 833-4324
Joseph A. Ray is now Senior Hydrologist for the Groundwater Branch
of the Kentucky Division of Water where he is involved in evaluat-
ing vulnerability of aquifers throughout the state, but especially
karst aquifers. Previously, he was a karst hydrogeologist with
Quinlan and Associates, Inc., and the National Park Service at
Mammoth Cave, Kentucky. His graduate work on environmental change
in Columbia, South America, earned a Master's degree in geography
at Western Kentucky University. He is extensively experienced in
karst field investigations and dye tracing. His other research
interests include the origin of prairies and savannas in humid
regions, and geomorphology.
Joseph A. Ray
Groundwater Branch
Kentucky Division of Water
18 Reilly Road
Frankfort, KY 40601
(502) 564-3410
556
-------
Ground-water remediation may be achievable in some karst aquifers
that are contaminated, but it ranges from unlikely to impossible in
most: I. Implications of long-term tracer tests for universal fail-
ure in goal attainment by scientists, consultants, and regulators
By: James F. Quinlan and Joseph A. Ray
Q. Have you seriously questioned how many of your fluorescein
recoveries are your dye and how many are from other sources?
Also, there can occur what some believe to be "breakdown
products" of Rhodamine WT that have special characteristics
similar to those of fluorescein. We have seen this in long-
term tests in Missouri. Also, we sometimes find a long-term
background for fluorescein.
A. Yes. Paranoia is a healthy state of mind. For procedural
reasons too lengthy to summarize here, we are satisfied that no
tests we have run have had recovery of fluorescein from the
tests of others, from industrial or household background, or
from Rhodamine WT breakdown products. In brief, however, the
possibility of such false positives can be minimized by repeat-
edly monitoring for them in background before the tests and by
sampling often enough so that one can closely and reliably
monitor the breakthrough, peaking, and recession of all dyes
used, and by recognizing that landfills and gasoline stations
can be a source of fluorescein (from the colorant in anti-
freeze) .
The breakdown of Rhodamine WT to something that might be con-
fused with fluorescein was first reported by Jim Duley in the
proceedings of the first NWWA karst meeting, in 1986 (p. 396-
397) , but we are unaware of any further study of such com-
pounds. Dave Sabatini has studied the properties of Rhodamine
WT (Ground Water, 1991. 29(2):341-349) and shown that it actu-
ally consists of two structural isomers with different sorption
tendencies (EOS, 1991, 72(44):154). These isomers explain why
Rhodamine WT breakthrough curves commonly have a plateau at a
C/C0 value of about 0.4.
As for the long-term fluorescein background in Missouri, all
one can do is repeatedly monitor with a scanning spectro-
fluorophotometer, not trace indiscriminately, and (seemingly
contradictorily) use enough dye to unambiguously exceed back-
ground and the known fluctuations of it. The problem of
accumulation of fluorescein background within a karst aquifer
has also been recognized in Switzerland and is the subject of
a recent paper by Aurele Parriaux and two colleagues (Hydrogeo-
logie, 1990. (3):183-194).
Q. When will regulatory agencies recognize the futility of com-
plete remediation in Appalachian karsts? How do we make them
realize this? They insist on pumping and treating for up to 10
to 20 years. Costs skyrocket and frustration rules.
557
-------
We wish we knew. We are fully sympathetic to the physical
impossibility of remediation of some aquifers and agree with
the authors we cite. Nevertheless, regulatory agencies are
required to develop, implement, and enforce regulations as
enabled by state or federal legislation. Many states have a
non-degradation policy or legislation that requires remediation
to background or detection limits. Such legislation is created
by politicians, many of which do not necessarily understand the
complex and technical issues involving hazardous waste remedi-
ation problems. They do understand that the voting public has
zero tolerance for any level of contamination — irrespective
of risks, potential exposure, or cost of remediation. Some-
times, the fault is with well-meaning regulators, legislators,
and legislative staff who, for various reasons, have written
rules or statutes more stringent than is necessary or intended.
We do not mean to seem naive, but we see the first steps in
improvement of communication between regulatory agencies and
both the regulated and their consultants are to be education of
all parties as to the nature of reality, establishment of
mutual respect for the technical competence and the integrity
of both consultants and regulators, and creation of a non-
adversarial climate. If mutual good faith is established, and
if there are alternative interpretations of the regulations,
the more reasonable one can be sanctioned. These desirable
goals are more easily stated than obtained.
558
-------
Session VIII:
Emergency Response and Ground-Water
Management
-------
560
-------
The Use of Groundwater-Level Measurements and Dye Tracing to Determine the Route of Groundwater
Flow from a Hazardous Waste Site in an Area of Karst
in Hardin County, Kentucky
C. Britton Dotson1, Nicholas C. Crawford2, and Mark J. Rigatti3
'Roy F. Weston, Inc.
Florence, Kentucky
2Western Kentucky University
Bowling Green, Kentucky
3O.H. Remediation Services Corporation
Atlanta, Georgia
ABSTRACT
The nature of karst groundwater flow dictates that groundwater investigations in karst areas be approached
in a different manner than those in more traditional hydrogeologic settings. Water-level measurements over
relatively large areas and dye tracing have proven to be suitable techniques for determining characteristics
of groundwater flow in karst areas. This paper presents a case study of a two-phase investigation consisting
of, first, water-level measurement and identification of hydrogeologic features and, second, dye tracing to
determine the groundwater flow route from the Middleton Hazardous Waste Site in Hardin County,
Kentucky.
Groundwater measurements were obtained over an area of approximately 70 square miles. From these
measurements a water-table contour map was constructed. Based on the slope of the water-table gradient
and the form of the water-table contours, a hypothesis was formed: groundwater flowing from the Middleton
Hazardous Waste Site would resurge at a spring 4.5 miles to the west. This hypothesis was tested by
performing dye traces.
Passive dye receptors were placed at all groundwater resurgences, interface locations and, where possible,
cave streams in the study area. An automatic water sampler was placed at the hypothesized resurgence.
Dyes were injected into a sinkhole collapse adjacent to the site and at a nearby sinking stream. After
allowing sufficient time for the dye to flow through the system, the passive dye receptors and water samples
were analyzed for presence of the dyes. The hypothesis was proven correct by the dye traces.
Based on interpretation of the water-table contour map and results of the dye traces (both qualitative and
quantitative) the direction, destination, velocity, and general route of groundwater flow from the Middleton
Hazardous Waste Site were determined.
561
-------
INTRODUCTION
Groundwater investigations typically involve the use of monitor wells to measure the movement of
groundwater and/or contaminants (Fetter, 1988) (Driscoll, 1986). In karst aquifers the majority of
groundwater flow is turbulent (White, 1988) compared to flow through porous-media aquifers which is
generally laminar. Relatively rapid transmittance of groundwater through discrete conduits in karst aquifers
precludes the use of monitor wells as a means of accurately measuring the movement of groundwater and/or
contaminants. In karst settings water and/or contaminants are typically on the ground surface for only a
short tune before sinking into the subsurface through features such as sinkholes, sulking streams, and soil
macropores. These features provide direct access for liquids and solids to migrate from the surface into the
subsurface with little or no filtration by soils. Once in the subsurface, the turbulently flowing groundwater
can rapidly transport liquids and solids considerable distances. Most water wells and springs in karst areas
become muddy after hard rains because solid material transported by surface water flows from the ground
surface into the aquifer through conduits large enough to transmit the particles. The subsurface conduit
system will then transport solids (sediments) much like surface flow systems, the more turbulent the water,
the greater its capacity to transport solid material. Due to the unique characteristics of groundwater flow
in karst areas, investigations should be designed to encompass a larger geographical area of study than the
site itself and incorporate water-table measurement data to identify general flow directions and dye tracing
to determine general flow routes.
This paper presents a case study of a two-phase investigation designed to determine characteristics of
groundwater flow in the vicinity of a hazardous waste site in a karst area. The initial phase of this
investigation involved the identification of all applicable hydrogeologic features in the study area and the
preparation of a water-table contour map based on water-level measurements from private water wells and
elevations of surface and subsurface streams. From the information acquired through these activities, the
second phase of the investigation, dye tracing, was designed and performed. The results of both phases of
the investigation determined the direction, destination, velocity and general route of groundwater flow from
the hazardous waste site.
SITE BACKGROUND
Several hundred electric transformers were burned over a period of years at a rural "junkyard" located 10
miles south of Elizabethtown in Hardin County, Kentucky (Figure 1). The "junkyard" is referred to as the
Middleton Hazardous Waste Site (MHWS). The transformers were burned in order to facilitate the
extraction of copper wiring housed inside the transformers. As a result of the activities at the site, oil laden
with polychlorinated biphenyls (PCBs) and associated metals were released into the soil. PCBs have a strong
affinity for organic materials in soils and therefore constitute a threat to groundwater in karst areas due to
the direct surface-subsurface link provided by karst features. The uppermost aquifer in the vicinity of the
MHWS is a karst aquifer. At the time of this investigation the aquifer served as a water supply for the
majority of residences in the area. Due to the potential impact of the "junkyard" activities on the quality of
the area's groundwater, a karst groundwater investigation was performed (Crawford et al., 1989) as a
component to the overall site remediation activities of the Region IV U.S. EPA.
Physical Setting
Hardin County, Kentucky is located in the Mississippian Plateaus Physiographic Province. The area of study
encompasses approximately 70 square miles of mature karst topography which is bounded on the east, north
and west by the Noun River as shown in Figure 1. This river serves as the base-level stream for
groundwater flow in the area. Another surface stream, Dorsey Run, flows toward the west and sinks 2 miles
south of the MHWS.
The geology of the study area is comprised of two slightly westward-dipping (<1 degree) limestone
formations of Mississippian Age. The oldest unit is the St. Louis Limestone, which is bedrock in the eastern
562
-------
HARDIN COUNTY
1-69
STUDY AREA
MIDDLETON HAZARDOUS
WASTE SITE
FIGURE 1 - Location map for the Middleton Hazardous Waste Site
Groundwater Investigation
563
-------
two-tMrds of the study area. Bedrock in the western one-third of the study area is the younger Ste.
Genevieve Limestone (Moore, 1964,1965). The study area exhibits features common to areas underlain by
relatively homogeneous carbonate rock.
METHODOLOGY
The groundwater investigation consisted of two phases. The initial phase involved determining the water-
table elevation in as many water wells as possible in the area and collecting information concerning locations
of applicable hydrogeologic features from landowners, area residents, and members of the Fort Knox Grotto
of the National Speleological Society. The second phase involved dye tracing from a sinkhole adjacent to
the MHWS and the sinking stream located south of the MHWS.
Phase One - Identification of Hydrogeologic Features and Determination of the Water-Table Gradient
A door-to-door survey was conducted to identify residences with accessible water wells and to acquire
permission to obtain water-level measurements. Measurements were made with water-level indicators. Only
data representing the water-table aquifer were collected. Information from well owners indicated the
majority of the wells were cased only to bedrock. The wells were uncased from bedrock to the total depth
of the well. Due to the size of the study area and the distribution of wells, it was impractical to determine
the precise elevation of the wells. Surface elevations were estimated from the Sonora and Summit 7.5
minute topographic quadrangle maps (USGS, 1967 and 1972). The contour interval of the topographic
quadrangles in the vicinity of the site was 20 feet. Estimates of surface elevations from these large contour
interval maps decreased the accuracy of the water-table elevations in the area but did not preclude the use
of the technique in providing a generalized map of the water table.
During the acquisition of water-level measurements from wells, inquiries were made of landowners and area
residents concerning locations of various hydrogeologic features such as springs, surface streams, sulking
streams, sinkholes with water flow, and caves, in particular those caves with streams. These features, as well
as those identified by local cavers, were investigated and located on topographic maps. In anticipation of
the second phase of the investigation, passive dye receptors were placed at the aforementioned hydrogeologic
features. These receptors (strips of untreated surgical cotton and packets of activated coconut charcoal)
once analyzed, served as an indicator of the presence or absence of fluorescent dyes in the groundwater flow
system. Placement of receptors at the identified hydrogeologic features throughout the study area prior to
dye tracing provided a measure of background fluorescence to which the post dye-trace receptors would be
compared. Before initiating the dye traces, all background dye receptors were removed and replaced with
fresh receptors. Locations of measured water wells and passive dye receptors are identified in Figure 2.
The information acquired from water-level measurements and the estimated elevations of surface and cave
streams were used to prepare a water-table contour map (Figure 3). Interpretation of the water-table
contour map influenced the second phase of the investigation. Due to its proximity to the MHWS, a spring
located 1.3 miles east of the site (Location 1) was initially hypothesized as the primary resurgence of water
from the site. However, the slope of the water-table gradient and the form of the water-table contours
suggested the most likely resurgence was at Waddell Spring, 4.5 miles to the west (Location 8). Anticipating
the resurgence of dye at this spring, an ISCO automatic water sampler was installed and programmed to
collect a sample every four hours. Collection and analysis of these samples provided information necessary
to make a quantitative interpretation of groundwater flow from the MHWS.
Phase Two - Dye Tracing
Six pounds of fluorescein dye (C.I. Acid Yellow 73) was injected into a collapsed sinkhole approximately 200
feet east of the site (Figure 4). Fifteen-hundred gallons of water was injected to insure the drainage capacity
of the collapse while at the same time wetting the soil to reduce dye sorption. Nine thousand-five hundred
gallons of water was then used to flush the dye into the system.
564
-------
MIDDLETON HAZARDOUS
WASTE SITE
432
MIDDLETON HAZARDOUS WASTE SITE
GROUNDWATER INVESTIGATION
1 0.5 0
1 MILE
LEGEND
• MEASURED WATER WELL
A PASSIVE DYE-RECEPTOR LOCATION
Figure 2
-------
NOUN RIVER
i\
II
MIDDLETON HAZARDOUS
WASTE SITE
sao
MIDDLETON HAZARDOUS WASTE SITE
GROUNDWATER INVESTIGATION
WATER-TABLE CONTOUR MAP
1 0.5
1 MILE
LEGEND
. MEASURED WATER WELL
A PASSIVE DYE-RECEPTOR LOCATION
«•«- GENERAL GROUNDWATER FLOW DIRECTION
WATER-TABLE CONTOURS ARE IN
FEET ABOVE SEA LEVEL
Figure 3
-------
NOLIN RIVER
MIDDLETON HAZARDOUS
WASTE SITE
MIDDLETON HAZARDOUS WASTE SITE
GROUNDWATER INVESTIGATION
WATER-TABLE CONTOUR MAP
AND
DYE TRACES
1 0.5
LEGEND
MEASURED WATER WELL
PASSIVE DYE-RECEPTOR LOCATION
DYE INJECTION LOCATION
DYE TRACE AND DIRECTION OF FLOW
WATER-TABLE CONTOURS ARE IN
FEET ABOVE SEA LEVEL
Figure 4
-------
To further characterize the groundwater flow of the area, 15 pounds of optical brightener (Tinopal 5BM GX,
F.B A. 22) were injected into Dorsey Run Swallet (Figure 4). This sinking stream provided an ideal location
to inject dye, and receptors capable of capturing the optical brightener had been placed along with those
being used to intercept fluorescein dye.
Every four days water samples being taken by the automatic sampler were removed and fluorometrically
analyzed hi the laboratory for presence of fluorescein dye. After allowing sufficient time for the dyes to flow
through the system, all passive dye receptors were replaced with fresh receptors. The dye receptors removed
from the locations were analyzed for the presence of both fluorescein and optical brightener. Analysis for
the presence of fluorescein involved rinsing the coconut charcoal with high-pressure water to remove
sediment and elutriating the charcoal with a solution of 9.5:1 isopropyl alcohol and potassium hydroxide.
Both visual and fluorometric analysis of the elutriant and comparison to the background receptor elutriant
indicated which hydrogeologic features the dye flowed through. Analysis for optical brightener involved
observing the spray-cleaned strips of untreated cotton under a long-wave ultraviolet light. Receptors which
fluoresced bright blue-white, as compared to background receptors from the same locations, were identified
as locations which were hi the flow path of the optical brightener introduced into the system.
RESULTS
During the inventory of the hydrogeologic features a number of springs, cave streams, and karst windows
were located (Figure 2, Locations 1-9). Dye receptors were placed at each of these features. Because the
Nolin River is the base-level stream hi the area, it was unnecessary to include hydrogeologic features north
of the river in the investigation. Passive dye receptors were positioned at several features south of the area
depicted hi Figure 2. However, these locations did not prove to be a part of the groundwater flow system
in the vicinity of the MHWS and, therefore, are not indicated hi this figure.
Groundwater elevations obtained by measuring water wells were combined with estimated elevations of the
surface and cave streams to produce a water-table contour map (Figure 3). Data from forty-three wells were
used hi preparation of the water-table contour map. From this map it is apparent that groundwater in the
vicinity of the MHWS flows west, contrary to the original hypothesis of flow to the east. Based on the
relative size of Waddell Spring and the direction of groundwater flow, it was selected as the most probable
location for the resurgence of groundwater from the site. The water-table contour map indicates that water
sulking at Dorsey Run Swallet is part of the groundwater flow found at Locations 2, 3, 4, and 5. This water
also flows hi a westerly direction toward Waddell Spring. The flow of water from the spring at Location 1
originates south and southeast of the spring rather than from the west as originally hypothesized.
Qualitative Dye Trace Results
Dye receptors were collected and analyzed for the presence of fluorescein and optical brightener. The
results of the dye traces and the water-table contours are shown in Figure 4. Activated charcoal dye
receptors positioned at Locations 6, 7, and 8 were positive for the presence of fluorescein. All other
receptors were negative for the presence of fluorescein. Surgical cotton dye receptors positioned at
Locations 2, 3, 4, 5, 6, 7, and 8 tested positive for the presence of optical brightener, and all other receptors
tested negative.
Quantitative Dye Trace Results
Water samples collected at Waddell Spring by the automatic sampler were fluorometrically analyzed for
fluorescein. Results of this analysis are illustrated in Figure 5. The first arrival of dye at Waddell Spring
occurred approximately 86 hours after injection. The concentration of dye at the spring peaked
approximately 12 hours after first appearance and was no longer detectable 80 hours later.
568
-------
FLOW-THROUGH CURVE FOR FLUORESCEIN DYE AT WADDELL SPRING
Dye Injected on September 7, 1989 at the MHWS
c
g
'•*-•
O 0_
§^
O
0)
>.
Q
0.08
0.06-
0.04
Peak Dye Concentration
First Arrival of Dye
Dye Injected
Dye No Longer
Detectable
0.02-
0
T i r
80 100 120 140 160 180 200
20 40 60
Time (Hours From Injection)
Figure 5 - Dye Flow-Through Curve
Sample Data Points
Distance from Waddell
Spring to the MHWS is 4.5
miles. First arrival of dye
was approximately 86 hours
after injection. Velocity of
flow was 0.052 mph.
-------
DISCUSSION OF RESULTS
Groundwater from the MHWS flows toward the west through karst windows at Locations 6 and 7 before
resurging at Waddell Spring (Figure 4). Dorsey Run, which sinks 2 miles south of the site, flows through
Locations 2, 3, 4, and 5 before joining the flow from the MHWS. The convergence of these two cave
streams occurs between Locations 5 and 6. The rate of flow from the MHWS to Waddell Spring is
approximately 0.052 miles per hour. This flow rate is based on linear distances between positive fluorescein
detection points. The actual flow rate is higher given the tendency for cave streams to meander in a manner
similar to surface streams.
CONCLUSIONS
A two-phase approach to the Middleton Hazardous Waste Site groundwater investigation determined the
direction, destination, velocity, and general route of groundwater flow from, and in the vicinity of the site.
The use of water-table measurement data produced a generalized water-table map of the study area and was
beneficial in directing the second phase of the investigation. Dye traces identified the destination and
general flow routes of groundwater from the site and in the vicinity of the site.
Error introduced to the water-table measurements by estimating surface elevations from 20-foot contour
interval topographic maps would not typically be acceptable in hydrogeologic investigations. However, in
this investigation the generalized water-table map proved useful in identifying the most probable resurgence
of groundwater from the site. Based on this information, a water sampler was placed at that resurgence.
Information gained from the fluorometric analysis of the samples was used to prepare a dye flow-through
curve. The added certainty of the dye flow-through curve and the associated information it provides were
necessary given the nature of the investigation. A dye flow-through curve more positively identifies the
connection between injection point and sampling point than visual or fluorometric analysis of elutriant from
passive dye receptors. From the dye flow-through curve it was possible to estimate the flow rate from the
site to the spring. The water-table contour map and dye-trace information obtained from this study will also
serve to focus any groundwater sampling activities associated with the site investigation.
If the water-table map had not been prepared prior to dye tracing, the water sampler would have been
incorrectly located at the spring east of the site. Analysis of samples collected at that spring would not have
identified the resurgence of groundwater from the site. Since passive dye receptors had been placed at all
hydrogeologic features in the area, Waddell Spring would have been identified as the resurgence of
groundwater from the site. However, it would have been necessary to perform another dye trace with the
water sampler housed at Waddell Spring to provide information for a quantitative interpretation of
groundwater flow from the site. An additional dye trace would have required more time and resources, and
an accurate interpretation of the second dye trace may have been hindered by dye from the first trace
remaining in the system.
This case study reinforces the need for hydrogeological investigations at hazardous waste sites in karst
terrains to include regional groundwater data to determine general flow directions and dye tracing to identify
the destination and general route of groundwater flow. The more traditional approach at hazardous waste
sites using site monitor well data may result in a completely false characterization of groundwater flow.
REFERENCES
Crawford, Nicholas C. and Dotson, C. Britton. 1989. Groundwater Flow in the Vicinity of the Middleton
Toxic Waste Site, Hardin County, Kentucky. Prepared for O.H. Materials Corporation and U.S.
Environmental Protection Agency, Region IV. 15 pp.
570
-------
Driscoll, Fletcher G. 1986. Groundwater and Wells. Second Edition. Johnson Filtration Systems Inc., St.
Paul, Minnesota. 1089 pp.
Fetter, C.W. 1988. Applied Hydrogeology. Second Edition. Merrill Publishing Company, Columbus, Ohio.
592pp.
Moore, F.B. 1964. Geology of the Summit Quadrangle, Kentucky. Kentucky Geological Survey and USGS
Geologic Quadrangle Map.
Moore, F.B. 1965. Geologic Map of the Sonora Quadrangle, Hardin and Larue Counties, Kentucky.
Kentucky Geological Survey and USGS Geologic Quadrangle Map.
USGS. 1967. Sonora Quadrangle, Kentucky, 7.5 minute series (topographic).
USGS. 1972. Summit Quadrangle, Kentucky, 7.5 minute series (topographic).
White, William B. 1988. Geomorphology and Hydrology of Karst Terrains. Oxford University Press, New
York. 464 pp.
C. Britton Dotson
Britton Dotson has a B.S. in Geology and a M.S. in Geography from Western Kentucky University. At the
time of this research he was employed by the Center for Cave and Karst Studies at Western as a research
hydrogeologist. He is now employed as a geologist by Roy F. Weston, Inc., under the Response,
Engineering, and Analytical Contract (REAC) for the Environmental Protection Agency's Environmental
Response Team, in Florence, Kentucky.
Dr. Nicholas Crawford
Nicholas C. Crawford, Ph.D. is a Professor in the Department of Geography and Geology and Director of
the Center for Cave and Karst Studies at Western Kentucky University. He has written over 140 articles
and technical reports dealing primarily with groundwater contamination of carbonate aquifers. The recipient
of 25 grants for hydrologic research on environmental problems of karst regions, he was awarded Western's
highest award for Outstanding Achievement in Research in 1985. As a consultant specializing in carbonate
aquifers for the past fifteen years, Dr. Crawford has performed over 600 dye traces and worked on numerous
groundwater contamination problems for private firms and for federal, state and local government agencies.
Mark J. Rigatti
Mark J. Rigatti is the program manager for the U.S. EPA Region IV Emergency Response Cleanup Services
(ERGS) contract for O.H. Remediation Services Corporation. Mr. Rigatti holds a B.S. in aquatic biology
and has been associated with several studies dealing with groundwater contamination in carbonate aquifers
in Kentucky and Tennessee as a component to hazardous waste site remediation activities.
571
-------
The Use of Groundwater-level Measurements and Dye Tracing to Determine
the Route of Groundwater Flow from a Hazardous Waste Site
in an Area of Karst in Hardin County, Kentucky
C. Britton Dotson, Nicholas C. Crawford, Mark J. Rigatti
Question 1 Was there any indication of site contaminants in the local wells or springs?
Question 2 - What assurances do you have that dye-tracing would intersect the same conduit system with the
sinks at the waste site?
Answer 1 The primary objective of Crawford and Associates was the identification of flow direction and
destination of groundwater in the vicinity of the site. Crawford and Associates was not involved in any sampling
activities. To date there has been no extensive sampling of water wells and springs in the study area. A
domestic water well located on the site property was sampled during the initial EPA response. There were no
site contaminants detected in this well.
Answer 2 - The uppermost aquifer in the vicinity of the hazardous waste site is a karst aquifer. Sinkholes on
the site and in the vicinity drain into this aquifer. Since individual groundwater drainage basins exist within the
aquifer, as evidenced by multiple groundwater resurgences, all sinkholes in the study area do not contribute to
the same drainage basin. It is possible that water flowing into one sinkhole may contribute to a different
drainage basin than water flowing into another sinkhole only 200 feet away. Based on the hydrogeologic
inventory only two springs (Locations 1 and 8) were likely candidates for resurgence of dye injected at the site
or in the vicinity. Dye that was injected into the sinkhole 200 feet east of the site resurged 4.5 miles west of the
site at Location 8 (Waddell Spring). This fact, accompanied by information provided by the water-table contour
map, indicated that the sinkhole where dye was injected contributed to the same drainage basin as the sinkholes
on site. If dye had been injected into a sinkhole on the site it is unlikely that it would have flowed east to
Location 1. To do this, flow would have been against the gradient, as indicated by the water-table contour map,
and would be opposite the general flow direction identified by the dye trace.
572
-------
RECOMMENDED ADMINISTRATIVE/REGULATORY DEFINITION OF KARST AQUIFER,
PRINCIPLES FOR CLASSIFICATION OF CARBONATE AQUIFERS,
PRACTICAL EVALUATION OF VULNERABILITY OF KARST AQUIFERS, AND
DETERMINATION OF OPTIMUM SAMPLING FREQUENCY AT SPRINGS
JAMES F. QUINLAN
Quinlan & Associates, Inc.
Nashville, Tennessee
PETER L. SMART
University of Bristol
Bristol, England
GEARY M. SCHINDEL
Eckenfelder Inc.
Nashville, Tennessee
E. CALVIN ALEXANDER, JR.
University of Minnesota
Minneapolis, Minnesota
ALAN J. EDWARDS
University of Bristol
Bristol, England
A. RICHARD SMITH
Municipal Solid Waste Division
Texas Water Commission
Austin, Texas
SUMMARY
A major problem of karst hydrology is an inadequate under-
standing by many regulators, planners, attorneys, geologists,
hydrologists, engineers, and other non-karst specialists of what a
karst aquifer is and how it differs significantly from other aqui-
fers. As a first step in solving this problem, we offer the fol-
lowing definition. A karst aquifer is an aquifer in which flow of
water is or can be appreciable through one or more of the follow-
573
-------
ing: joints, faults, bedding planes, and cavities — any or all of
which have been enlarged by dissolution of bedrock.
The above definition is in terms of hydraulics, not landforms
(such as sinkholes), grikes (soil-filled joints), or hydrologic
features (such as swallets, sinking streams, springs, and caves).
The presence of sinkholes and any or all of these other features is
diagnostic of a subjacent karst aquifer, but the absence (or un-
recognized presence) of any or all of them does not mean that a
karst aquifer is not present. As a generalization, however, if
carbonate rocks such as limestone, marble, or dolomite, or even
more soluble rocks such as salt or gypsum, are present, assume that
the terrane is a karst — until or unless the opposite can be con-
vincingly proven. Almost all carbonate terranes are underlain by
one or more karst aquifers, but some terranes and aquifers are more
karstic than others.
Having defined karst aquifers, we classify them according to
three fundamental, independent attributes governing their behavior
— recharge, storage, and flow — each of which can be visualized
as forming orthogonal edges of a cube. Each of the three attri-
butes is part of a continuum between two end-members, yielding a
classification based on a total of 6 karst end-members, many inter-
mediate members, and 2 non-karst end-members. The cube, modified
after one proposed by Smart and Hobbs (1986), can be visualized as
a 3-dimensional conceptual model within the boundaries of which
various aquifers can be plotted (mapped) and their relations to one
another can be seen, as shown in Figure 1.
All of the many types of karst aquifers can be grouped into a
fourth continuum based on the sensitivity of their response to
variations in recharge, storage, and flow, and their attendant
relative vulnerability to the adverse effects of spills and impro-
per disposal of waste: hypersensitive aquifers, very sensitive
aquifers, moderately sensitive aquifers, and slightly sensitive
non-karst aquifers. These new terms are plotted on the same cube.
Hypersensitive aquifers are characterized by concentrated re-
charge at discrete points, low to moderate storage, and conduit
flow. Consequently, they are extremely vulnerable to groundwater
contamination. Very sensitive aquifers are characterized by the
occurrence of either conduit flow, point recharge, or low storage,
or a combination of two of these, as shown in Figure 1. Again,
they are sensitive to pollution, but less so than hypersensitive
aquifers. Moderately sensitive aquifers have dispersed, slow
recharge, high storage, and diffuse flow, as also shown in
Figure 1; such aquifers are therefore less vulnerable to rapid
transmission of pollutants than are very sensitive karst aquifers.
The continuum from hypersensitive to moderately sensitive
karst aquifers can be discussed in terms of two end-members which
have traditionally, but misleadingly, been called "conduit-flow
aquifers" and "diffuse-flow aquifers". Recognition of the position
of an individual aquifer in this continuum is essential to provid-
574
-------
ing adequate aquifer protection and appropriate management of waste
disposal and also to the design of reliable groundwater monitoring
schemes. In general, movement of pollutants in hypersensitive
karst aquifers is very rapid, commonly tens to thousands of feet
per hour, depending upon stage. Reaction time, the time between
the start of a recharge event (or when a spill occurs) and when it
is first sensed by the aquifer, is short. Duration of the adverse
effects of a pollution incident may be relatively brief (depending
upon the properties of the contaminant and how and where in the
system the contamination occurs). In contrast, movement of pollu-
tants in moderately sensitive aquifers is usually several orders of
magnitude less rapid than in hypersensitive aquifers, but it is
usually orders of magnitude faster than in most granular aquifers.
Reaction time is much longer than in hypersensitive aquifers. The
results of a pollution incident can persist for tens of years.
The terms conduit flow and diffuse flow, as descriptors of
aquifer type used for classification, should be abandoned because
they prevent necessary consideration of the important roles of
recharge and storage. Conduit flow and diffuse flow should be
retained, however, as descriptors of flow type.
The distinctions between the two vulnerability end-members in
karst aquifers can be made in several ways; they are most easily
determined by interpreting variations in specific conductance
(hereafter called conductivity), as characterized by its coeffi-
cient of variation (CV). The variations in conductivity are more
important than the absolute values. The distinctions between aqui-
fer vulnerability can also be determined from fluctuations in one
or more of the following alternative water quantity or quality
parameters: discharge, turbidity, temperature, and carbonate hard-
ness. These distinctions can only be determined by site-specific
fieldwork. The necessary data for these conductivity and alterna-
tive vulnerability-type discrimination measurements are easily and
inexpensively obtained, but they are not included in geologic or
hydrologic publications or reports.
The spring is the pulse of a karst aquifer. Monitor it in
order to understand the nature of an aquifer. While springs pro-
vide the most accessible and perhaps the most representative moni-
toring points, alternative monitoring points may include cave
streams, seeps, and wells that may intercept fracture zones, dis-
solution zones, or conduits. If monitoring is done merely to
attempt to characterize an aquifer — as contrasted with attempting
to detect pollutants from a specific site — the wells and seeps
used as alternative monitoring points are likely to undergo much
smaller fluctuations in chemical and physical characteristics than
are springs.
Interpretation of variations in conductivity of a spring as an
indicator of the vulnerability of a karst aquifer is probably reli-
able for the vast majority of North American karst aquifers and, we
believe, essentially all those with a drainage-basin area smaller
than about 500 square miles. Although the controls on conductivity
575
-------
are many and complex, it is a reliable measure and integrator of
hydrologic and water-quality response to the dynamics of an aqui-
fer. Conductivity senses and integrates the covariance of the
complex response and allows a simple measure of it.
Six important caveats must be remembered:
1. An aquifer can not be classified as being "not a karst"
simply because one can successfully model it numerically
over so large an area that it seems to behave with the
properties of a porous medium. The local perturbations
that so dramatically affect site characterization and
remediation are beyond the ability of analysis by present-
day computer models. Such model codes are most often
applicable only to granular aquifers where aquifer parame-
ters do not vary as dramatically and over as short a
distance as they can in karst aquifers.
2. Data from wells alone might not show karstification, even
if it exists, because the scale of organization of flow in
a karst aquifer is not likely to be adequately sensed by
just a few wells. If wells alone are used for monitoring,
they must be shown by rigorously designed and executed
tracer tests to have hydrologic connection to the site to
be monitored.
3. A karst aquifer within a single groundwater basin may be
locally characterized by either of the two end-member types
of vulnerability, or by an intermediate member. In part,
this is because contaminants introduced into a sinkhole
connected to a conduit will move more rapidly than if
spilled on the ground or introduced in the area between
swallets or open sinkholes, or introduced in an area where
dispersed recharge and diffuse flow may predominate and
conduits may be rare or relatively inaccessible.
4. Just because a spring flows from a conduit, it does not
follow that it is a hypersensitive spring or that it is a
discharge point of an aquifer dominated by conduit flow.
It could be a moderately sensitive spring draining a simi-
lar aquifer, as determined by interpretation of variations
in its water quality and discharge. For nomenclature deci-
sions, aquifer response to recharge events takes priority
over physical appearance.
5. Site-specific fieldwork may require examination and evalu-
ation of geology, hydrology, and karst features within a
radius of a mile from a site to perhaps as much as 40 miles
away, depending upon the structural setting and probable
size of groundwater basins.
6. The conductivity CV is not an infallible indicator of the
nature and vulnerability of a spring and the aquifer it
drains. Nevertheless, the CV is extremely useful and
576
-------
probably reliable so often that it can be employed with a
high degree of confidence. Its reliability can be tested
by tracing.
An alternative to using conductivity at discharge points such
as springs to characterize the nature of a karst aquifer is to
obtain hydraulic conductivity or transmissivity data from aquifer
tests (slug, bailer, or pump tests). Such data are characterized
by a low mean and a high CV in aquifers dominated by conduit flow
and by a moderate to high mean but a much lower CV in aquifers
dominated by diffuse flow. Non-karstic carbonate aquifers gener-
ally have a low hydraulic conductivity mean and low CV (Smart et
al., 1991). Interpretation of data from aquifer tests in karst
terranes, however, should be done cautiously, for the reasons
discussed.
Aquifer tests may yield valuable information (albeit inaccu-
rate) about the flow and quantity of water in the portion of karst
aquifer tested but say nothing about its recharge or storage beyond
the immediate vicinity of a well. Such tests are complemented by
interpretation of conductivity CV's which say nothing about aquifer
parameters but do imply much about recharge, storage, and flow.
An appendix outlines two procedures for using conductivity
data to prevent aliasing (loss of recognition of the signal from
rapidly changing parameters, a consequence of insufficient frequ-
ency of sampling) and to determine an appropriate, practical,
aliasing-free sampling procedure for using water quality data to
characterize an aquifer, its spring, or a conduit-fed well, and for
the establishment of sampling frequency during tracer tests.
CONTENTS
Summary 573
Introduction 578
Importance of aquifer vulnerability to design of
groundwater monitoring plans for karst aquifers . . . 583
Recognition of karst terranes 584
Definition of karst aquifer 586
Recognition of karst aquifers 587
Principles for classification of carbonate aquifers .... 588
Relation between types of karst aquifer and
sensitivity to groundwater contamination 593
Characterizing carbonate aquifer sensitivity 594
Introduction 595
Background 596
Geochemical variation 597
Hydrological behavior 598
Limitations 599
Practical evaluation of vulnerability of karst aquifers .601
Procedure 601
Difficulties and limitations of vulnerability
evaluation 602
577
-------
Use of aquifer test data to characterize flow in and
vulnerability of karst aquifers 604
A cautionary tale 607
Caveats 608
Discussion 608
Solicitation 615
Acknowledgements 615
References cited 615
Appendix: Procedure for prevention of aliasing and for
determination of appropriate sampling frequency for
a spring or a conduit-fed well 629
Biographical sketches 632
Questions and answers 634
INTRODUCTION
This paper is written primarily for geologists, hydrologists,
and engineers who are not specialists in karst but who work in car-
bonate terranes and deal with regulators, planners, and attorneys
who must make important decisions concerning such terranes. It is
also written for regulators and planners who are not likely to be
trained in karst hydrology but who must educate some of the consul-
tants they deal with. The discourse is written as a practical
guide for "greater utility" (Franklin, 1781) . Accordingly, we have
tried to avoid the Feynman Effect (directing the content toward our
colleagues, karst specialists, when we are supposed to be focusing
on communication with non-karst specialists; Bartlett, 1992), but
we believe that karst specialists will find it to be a valuable
synthesis which revolutionizes concepts of karst aquifer classifi-
cation and of what constitutes necessary sampling frequency. A
less technical summary of this paper will be published elsewhere.
The Summary, the sections titled Definition of karst aquifer,
Principles for classification of carbonate aquifers, and Practical
evaluation of vulnerability of karst aquifers — as well as the Ap-
pendix and the penultimate indented paragraph of the Discussion —
are the most important parts of this paper; the rest comprise the
foundation for it and the necessary rationale. The Discussion is
a potpourri of important points that, although extremely relevant,
might have detracted from the clarity and continuity of the paper
itself.
About 20% of the conterminous U.S. and 40% of the U.S. east of
Tulsa, Oklahoma, is underlain by carbonate rock, mostly limestone.
We maintain that almost all of these areas are, therefore, some
type of karst and probably include some type of karst aquifer. It
is far more judicious to operate on the assumption that they are
karst aquifers — until and unless conclusively proven otherwise.
An understanding of the extraordinary hydrology of such areas is
necessary if their waters are to be protected and used most wisely.
Groundwater movement in karst aquifers differs from that in
granular media. It is commonly orders of magnitude faster, seem-
578
-------
ingly (but not actually) less predictable, and commonly through
conduits or dissolutionally enlarged joints and bedding planes.
Water movement in karst aquifers can be described by known physical
laws but generally not the same ones that successfully describe and
predict flow in granular media.
Karst terranes can be usefully classified according to eight
major attributes: cover, rock-type, climate, geologic structure,
physiography, hydrology, modification during or after karstifica-
tion, and dominant landforms, each of which forms the basis for a
separate classification that can be combined with the other seven
(Quinlan, 1978, as partially summarized by White, 1988, p. 117-118;
1990b). Simple as the above classifications are, and useful as
each of them and a combination of them may be for geomorphological
and regional analysis, they are obviously much more than is needed
for day-to-day use by professionals concerned with environmental,
regulatory, and administrative problems of terranes potentially af-
fected by hazardous materials.
Karst is not rare. Its hydrology is not unpredictable. When
the appropriate methods of investigation are used, the properties
and flow within a karst are more predictable than most profession-
als realize (Quinlan and Ray, 1989).
Although various sections of the long-awaited, newly issued
Subtitle D regulations for RCRA (Resources Conservation and Recov-
ery Act) and the CWA (Clean Water Act) can be easily interpreted to
sanction special monitoring protocols that may be necessary for
karst aquifers — as they should — there is only one section
(Section 258.15) that specifically mentions karst, in referring to
unstable areas. Karst is mentioned in the appendices that discuss
this section and others. Unfortunately, in Appendix C (Supplemen-
tal Information), the discussion repeatedly and mistakenly refers
to karst terraces where, from context, it clearly means karst
terranes (U.S. EPA, 1991, p. 51,020, 51,047-48). Sinkhole develop-
ment can be extremely important at waste disposal sites, but it is
our considered opinion that the potential for groundwater contami-
nation at such sites is far more common and nearly ubiquitous.
One of the more practical recent applications of academically-
inspired karst research has been the recognition that springs and
cave streams, shown by tracing to drain from a site, may be optimal
points for monitoring groundwater quality and that only wells shown
by tracing to drain from a site can be considered to be monitoring
wells (Quinlan and Ewers, 1985; Quinlan, 1989, 1990a). These con-
cepts have been well received by the hydrogeologic community (Beck
et al., 1987) and the regulatory community (Quinlan, 1989) and are
implicit in this paper.
There is a need for recognition of degrees of vulnerability of
the water quality that can occur in different types of karst aqui-
fers. Such recognition enables the user to think in terms of why
one is concerned with the karstic nature of a terrane (Field, 1988,
1990). The "why" is answered in terms of sensitivity to effects of
579
-------
groundwater contamination and in terms of how to reliably monitor
the various types of karst so that the monitoring is relevant,
reliable, and time-sensitive.
Although much of this paper is concerned with the potentially
severe consequences of spills of hazardous materials, no attempt is
made here to discuss procedures for response to them. See Quinlan
(1986) and Crawford (1991).
How, then, can karst aquifers be adequately protected? We
will discuss why degree of vulnerability of karst aquifers is rele-
vant to groundwater monitoring plans and suggest how to recognize
a karst terrane and/or a karst aquifer. Then we will state princi-
ples for classification of karst aquifers. This will be followed
by a discussion of flow, recharge, and storage within karst aqui-
fers, relations between karst aquifer types and sensitivity to
groundwater contamination, and characterization of carbonate aqui-
fer sensitivity- We then give a simple technique for reliably pre-
dicting aquifer vulnerability insofar as it is relevant to problems
of groundwater protection and monitoring, and close with a review
of the use of aquifer tests to characterize flow in and vulnerabil-
ity of karst aquifers, some caveats, a concluding discussion, and
an appendix that describes a new procedure for reliably establish-
ing the optimal frequency for sampling a spring.
A cautionary quotation is appropriate: "Of all the words in
the hydrologic vocabulary, there are probably none with more shades
of meaning than the term aquifer. It means different things to
different people, and perhaps different things to the same person
at different times. It is used to refer to individual geologic
layers, to complete geologic formations, and even to groups of geo-
logic formations. The term must always be viewed in terms of the
scale and context of its usage." (Freeze and Cherry, 1979, p. 47).
An aquifer can be defined as a saturated permeable geologic
unit that can transmit significant quantities of water under ordi-
nary hydraulic gradients (Freeze and Cherry/ 1979, p. 47). Stated
another way, an aquifer is a body of rock or sediment capable of
yielding usable quantities of water to wells and springs. If one
uses or could use the water, its host is an aquifer. With this
less precise definition, the rock adjacent to a cave stream that is
perched above the water table could be classified as and regulated
as an aquifer, even though most of the rock is not saturated.
[Some state regulations would consider this cave stream as surface
water because it is in a defined channel; there is case law to sup-
port such interpretation (Davis and Quinlan, 1991).]
The term aquifer is purposely imprecise with respect to hy-
draulic conductivity; it refers only to the ability to yield a
usable quantity of water. A rock or sediment mass yielding less
than one gallon per minute to a well and supplying one household is
just as much an aquifer as one yielding thousands of gallons per
minute to a well for a municipality or industry- Obviously, the
latter is a better aquifer (if one needs a high yield), but the
580
-------
former is no less an aquifer.
By water we mean both phreatic water (water below the water
table — groundwater in the traditional sense) and vadose water
(water in the unsaturated zone between the water table and the
ground surface), as discussed by Domenico and Schwartz (1990, p. 9-
11) . Indeed, for the past 15 years, the nature and movement of
water and contaminants in the vadose zone has been a major frontier
in research of both groundwater hydrology in general (Devinny et
al.. 1990, p. 42-43) and karst hydrology (Smart and Friedrich,
1986; Bonacci, 1987, p. 28-35; Chevalier, 1988). In karst ter-
ranes, it is especially important to monitor and characterize the
flow of vadose water because the vadose zone in karsts provide sig-
nificant storage; flow from it to the saturated zone can be quite
rapid. Monitoring of groundwater in the unsaturated zone, particu-
larly in karst terranes, is an as yet unexplored frontier, both
practically and administratively.
Several relevant definitions are given in recently issued
Federal regulations concerning landfills (EPA, 1991, p. 51,017-18).
"Aquifer" means a geological formation, group of formations, or
port[i]on of a formation capable of yielding significant quantities
of ground water to wells or springs." [The problem words here are
use of ground water rather than water and the definition of signi-
ficant. What is the difference between significant and apprecia-
ble? We would argue that significant, as defined by Webster's
Ninth New Collegiate Dictionary, means "of a noticeable or measur-
ably large amount", and that large is a relative term. In the
southwestern U.S., a spring may flow a few quarts per hour, an
insignificant quantity only by the standards of the humid east, but
enough to sustain an entire local ecosystem and to be used as a
water supply.] "Ground water means water below the land surface in
a zone of saturation. ... Saturated zone means that part of the
earth's crust in which all voids are filled with water."
The emphasis of monitoring sections of these new regulations
is on water below a water table or in a confined aquifer. EPA
proposed in 1988 to require monitoring elsewhere, including unsatu-
rated zones. The agency is currently evaluating comments on that
proposal and is preparing a final rule. The current regulations,
however, do not preclude states from requiring monitoring in the
unsaturated zone (U.S. EPA, 1991, p. 51,067).
An encyclopedic compilation of aquifer and oroundwater defini-
tions could be obtained from hydrology texts and case law, but such
a compilation is beyond the scope of this paper. We are not obses-
sed with definitions, but we believe it is necessary to agree on
them in order to achieve clear, unambiguous communication.
Four excellent texts in English provide thorough expositions
on the hydrology of karst aquifers (Milanovic, 1981; Bonacci, 1987;
White, 1988; Ford and Williams, 1989), each with a different
emphasis.
581
-------
Even though there are numerous excellent texts on groundwater
hydrology (Freeze and Cherry, 1979; Driscoll, 1986; Fetter, 1988;
Domenico and Schwartz, 1990), applied chemical and isotopic ground-
water hydrology (Mazor, 1991), and on hydrology in general (e.g.,
Gupta, 1989; Viessman et al.. 1989), none of these texts or others
describe the hydrology of karst aquifers in more than a superficial
fashion, if at all. Most hydrology texts completely ignore karst.
Otherwise excellent, state-of-the-art texts on environmental sci-
ence and engineering and on watershed management don't even mention
karst (Henry and Heinke, 1989; Brooks et al., 1991). Recent defi-
nitive books on groundwater development, on groundwater monitoring,
on landfill design, and on hazardous-waste site management or re-
mediation mention karst only in passing (Roscoe Moss Company, 1990,
p. 14, 167; Nielsen, 1991, p. 64, 407; O'Brien and Gere Engineers,
Inc., 1988, p. 51, 53, 56, 78), ignore it (Ward et al., 1990;
Bagchi, 1990; Pfeffer, 1992; Goldman et al., 1986; Christiansen et
al., 1989; O'Leary et al., 1986), or erroneously refer to karst
terranes as being "rare" (Devinny et al.. 1990, p. 145). A recent
environmental geology text says nothing about the special charac-
teristics of groundwater pollution in karst terranes (Montgomery,
1992) .
Although the second edition of an authoritative water encyclo-
pedia includes data on velocities characteristic of conduit flow
and diffuse flow (van der Leeden et al., 1990, p. 286), little else
of its content suggests the importance of karst aquifers in Ameri-
ca. An otherwise excellent text on environmental geology includes
a brief discussion of sinkhole development but says nothing about
groundwater movement in karst aquifers, especially as related to
groundwater contamination (Lundgren, 1986). A highly readable com-
pendium on groundwater contamination in the United States says
nothing about karst (Patrick et al., 1987) and neither does an
otherwise excellent introduction to groundwater contamination
written for executives, plant managers, and politicians (Bailey and
Ward, 1990). A fascinating synthesis of geochemistry and fluid
mechanics of permeable rocks ignores karst (Phillips, 1991) as does
a reference handbook by the National Fire Protection Association
written as guidance for response to hazardous materials incidents
(Henry, 1989). Even the National Research Council totally ignores
karst (Committee on Ground Water Quality Protection, 1986; Water
Science and Technology Board, 1988, 1990a) or gives it only one
paragraph in a book on groundwater models (Water Science and Tech-
nology Board, I990b, p. 99-100) . And the National Ground Water
Information Center, operated by the National Ground Water Associ-
ation, does not include karst as a Descriptor/Keyword on its index
sheet for Proceedings volumes. The situation with respect to re-
cognition of karst-related problems by non-karst specialists in
Europe is little better, but there are some notable exceptions
(Erdelyi and Galfi, 1988; Fookes and Vaughan, 1986; Lallemand-
Barres and Roux, 1989; Nonveiller, 1989; Wynne, 1987).
The books cited above are some of the more recent on various
topics. The same could be said about the inadequacy of exposition
of karst-related topics in earlier books. Accordingly, most con-
582
-------
sultants, whether hydrogeologists or engineers, and most regula-
tors, whatever their education, lack formal training in karst
hydrology and rarely have the opportunity to learn about it. Many
are unaware of the significance of karst. Too many others deny its
existence or importance.
Only six North American universities and the National Ground
Water Association regularly offer formal courses devoted entirely
to various aspects of karst hydrogeology. Training in karst is
also available at several additional universities where the topic
is a research interest of a faculty member. All of these educa-
tional efforts are taught by well-experienced staff, but they reach
an inadequate number of potential users of the concepts espoused.
IMPORTANCE OF AQUIFER VULNERABILITY TO
DESIGN OF GROUNDWATER MONITORING PLANS FOR KARST AQUIFERS
In an ideal world, the concepts advocated in this paper would
be regularly applied to planning and management. In the real
world, however, the concepts can and should be so applied, but most
applications will be to groundwater monitoring and responding to
contamination situations that already exist.
All aquifers are sensitive to groundwater contamination, hence
the need for aquifer protection policies and groundwater monitoring
to ensure compliance with them. Knowledge of whether a spring (and
the aquifer it drains) is hypersensitive rather than moderately
sensitive to contamination will affect the design of its monitoring
plan and can affect the design and operation of tracer tests used
to establish that plan. Hypersensitive springs typically respond
rapidly and possibly greatly to recharge events; they will require
frequent sampling, beginning early in an event. In contrast,
moderately sensitive springs typically respond slowly and in a
relatively subdued manner; sampling frequency can be less and can
start later in an event but sampling duration will be longer.
These sampling matters are discussed in more detail by Quinlan and
Alexander (1987) and Quinlan (1989, 1990a), where we used the
traditional (but not equivalent) terms conduit flow and diffuse
flow, respectively, and by Blavoux and Mudry (1988).
In this paper, we offer a simple technique for aquifer charac-
terization on the basis of its relative vulnerability and the vari-
ability of its specific conductance because there is a great need
for reliable, rapid, and inexpensive methods for site characteriza-
tion and monitoring during remediation (Mackay, 1990) and, more
important, before remediation is ever necessary. We suggest three
other techniques that yield complementary results and may or may
not give a higher reliability in assessment and characterization,
but they involve aquifer testing and a significantly greater cost.
We recognize that evaluation of aquifer vulnerability is just
one aspect of land-use planning and watershed protection often
necessary for karst terranes (Rubin, 1991).
583
-------
RECOGNITION OF KARST TERRANES
When environmental professionals think of karst, many of them
think of the Mammoth Cave area of Kentucky with its tens of thou-
sands of sinkholes, hundreds of sinking streams and springs, and
hundreds of miles of accessible cave passages, some so large that
two garbage trucks could be driven side-by-side through them. Most
of these same people are unwilling to recognize, for example, the
Nashville, Tennessee, area as a karst. To judge a carbonate ter-
rane not to be a karst because it does not resemble the Mammoth
Cave area is to judge the National Enquirer not to be English prose
because it lacks the style, grace, and wit of a Shakespearean
comedy. Indeed, the classic examples of karst in the Mammoth Cave
area and The Taming of the Shrew, however delightful and fascinat-
ing they may be, are trivially small non-representative examples of
the spectrum of karst types and the spectrum of English prose.
Others have defined a karst terrane in terms of the number of
sinkholes shown on a topographic map — "X" sinkholes per square
mile or per ninth of a 7.5-minute quadrangle. This is objective
but naive, and not recommended. Aside from the fact that sinkholes
are rare or only subtly present in some karst terranes, in others
most sinkholes are not shown on the topographic map for many
reasons, ranging from non-intersection of them by contours, to non-
recognition of them by photo-interpreters making topographic maps,
to farmers having partially filled them. Indeed, 85% of the 535
sinkholes in Winona Co., Minnesota, are not on the 7.5-minute topo-
graphic maps and were found only by field work (Dalgleish and
Alexander, 1984).
A karst is not a landscape. Rather, it is a variable aggre-
gate of diagnostic surface and subsurface features that are known
to be genetically related to its development. [Admittedly, this
definition is a bit circular.] We use karst terrane interchange-
ably with karst to refer to both the surface and the subsurface,
the lithologic expanse considered as a whole. Strictly speaking,
however, karst terrain refers only to the surface. [Consult Hansen
(1991, p. 177) for discussion of the distinctions between terrane
and terrain.]
The necessary (essential) elements of a definition of karst
are that it be composed of both the landscape and the subsurface
features formed as a result of dissolution of bedrock (usually, but
not necessarily limestone and/or dolomite), that it be character-
ized by a distinctive subsurface hydrology, and that it include any
of a suite of distinctive surface and subsurface features that are
sufficient (non-essential) for fulfillment of the definition.
The presence of any one of the following types of distinctive
surface and subsurface features is sufficient for diagnosing a
terrane as a karst. No specific one of them is necessary. Lack
(or apparent lack) of nearly all of them does not mean a terrane is
not a karst:
584
-------
1. Sinkholes (any closed depressions, with or without a dis-
crete opening at their bottom, formed by dissolution and/or
collapse of bedrock, with flushing and/or collapse of soil
into a subjacent cavity) and internal drainage to them;
2. Dry valleys (in humid climates);
3. Springs (draining carbonate, sulfate, or halide rocks);
4. Caves (open to the surface or accidentally encountered by
drilling);
5. Sinking streams (that sink at a hole known as a swallet);
6. Dissolutionally enlarged joints and/or bedding planes (as
seen in cores or outcrops);
7. Grikes (soil-filled, dissolutionally enlarged joints or
grooves; also known as cutters or soil karren);
8. Karren (dissolutionally, subaerially, water-carved grooves
on rock, commonly subparallel).
A few examples of types 1 through 5 may be shown on a topo-
graphic map, but most of them and types 6 through 8 can only be
found by field work. If one looks in the field, and does so dili-
gently, some of these eight features, perhaps all of them, will be
found. We are so confident of this that we say "If there is car-
bonate rock, there is almost certainly some type of karst."
The apparent lack of some or all of the above eight features
in a carbonate terrane, or their non-recognition, does not mean
that a karst terrane is not present. It probably means that more
field work or drilling is needed to find them. Similarly, we con-
sider the presence of just one of the features to be diagnostic of
a karst because that one is usually representative of a much larger
population.
The above definition of karst in terms of what is necessary
and what is sufficient is totally (and independently) consistent
with current European thought on the subject, thought which has
evolved during the past 150 years. To paraphrase and abbreviate
Gams (1991), karst is a territory formed by dissolution of bedrock
and in which underground drainage of precipitation prevails. These
he identifies as the absolute (necessary) elements of the defini-
tion. The so-called karst landforms and other surface features are
the possible (sufficient) elements. Gam's definition of karst and
ours were independently conceived but are remarkably similar.
Some people confuse a sinkhole with the karst itself. The
senior author has reviewed reports that describe proposed landfill
sites in flat-lying limestones such as the Mississippian-age St.
Louis Limestone. These reports state, for example: "There is no
karst on the property except for one, in the southeast corner,
where there is an open hole." The report writers ignored the other
closed contours (indicators of sinkholes) without an open hole at
their bottom, and they truly confused an open sinkhole with the
karst itself. This is like confusing Cyrano de Bergerac's nose
with the man himself. All will agree that there was a lot more to
Cyrano than just his nose. So also, there is far more to a karst
than just a sinkhole or a cave. In the example cited here, the
585
-------
whole property is part of a karst terrane.
Like sinkholes, springs are commonly omitted on topographic
maps. Fewer than 5% of springs relevant to regional movement of
groundwater in the Mammoth Cave area of Kentucky are shown on 7.5-
minute (1:24,000) topographic maps. The remaining 95% had to be
found by field work (Quinlan, 1989, 1990a), but the presence of a
few large springs not shown on the maps could be inferred from
diagnostic contours suggestive of their alcove. Non-indication of
springs is characteristic of topographic maps of the eastern and
midwestern U.S. MORAL: In consideration of the unreliability of
topographic map depiction of sinkholes and springs, absence of
recognizable karst landforms or karst hydrologic features on a
topographic map must not be interpreted to mean a karst terrane is
absent. Topographic maps are useful — but not reliable — indica-
tors of the karstic nature of a terrane.
Obviously, there are degrees of karstification, just as there
is a range in the quality of English prose. One can quibble over
the subjective judgement as to degree of karstification or quality
of writing, but nearly all carbonate terranes are a karst and the
National Enquirer is English prose.
The best map showing the distribution of American karst ter-
ranes, albeit incomplete, is the hitherto almost unknown useful map
(in color) by Davies et al. (1984). Discussion of our strong dis-
agreement with many of their classification criteria, however, is
beyond the scope of this paper.
DEFINITION OF KARST AQUIFER
A karst aquifer is an aquifer in which flow of water is or can
be appreciable through one or more of the following: joints,
faults, bedding planes, and cavities — any or all of which have
been enlarged by dissolution of bedrock. This definition is in
terms of hydraulics, not landforms (such as sinkholes), grikes
(soil-filled joints), or hydrologic features (such as swallets,
sinking streams, springs, and caves). The presence of sinkholes
and any or all of these other features is diagnostic of the
presence of a karst aquifer, but the absence (or unrecognized
presence) of any or all of them — on the ground or on a topograph-
ic map — does not mean that a karst aquifer is absent.
The definition of appreciable (as related to the quantity of
flow), like the definition of aquifer itself, may vary as the use
varies. Appreciable is a relative term and does not refer to the
same quantity of water in all aquifers. According to the standard
American law dictionary, appreciable means "capable of being esti-
mated, weighed, judged of, or recognized by the mind. Capable of
being perceived or recognized by the senses. Perceptible, but not
a synonym of substantial" (Black, 1989, p. 92).
Dissolutional enlargement of joints and cavities in limestone
586
-------
is initially extremely slow. Dissolution rates and their kinetics
are described by Palmer (1991) and Dreybrodt (1988, p. 229-240).
Routing of much of the initial flow may be like that described by
Ewers (1982; partially summarized by Ford and Williams, 1989, p.
249-261, and Dreybrodt, 1988, p. 223-229, 240-245), Johns and
Roberts (1991), and Tsang et al. (1991).
It is probable that most flow in most karst aquifers is turbu-
lent (Ford and Williams, 1989, p. 142-148). We believe, however,
that it would be a mistake to use such a criterion to distinguish
a karst aquifer from one that is not. Aside from the fact that
some flow in some karsts is probably laminar (Hickey, 1984), accur-
ate measurement of flow velocities, fissure widths, and hydrostatic
heads at inaccessible parts of an aquifer is difficult and ambigu-
ous. To routinely spend considerable time and resources trying to
make such problematic, unnecessary measurements distracts from the
likely reasons for assessing an aquifer.
The presence of non-carbonate beds such as sand, clay, or
sandstone at the surface of a terrane does not mean that the under-
lying carbonate beds do not comprise a karst aquifer. They usually
do. Such mantled, covered, or caprock-protected aquifers, be they
shallow or deep and possibly confined, can be regionally important.
RECOGNITION OF KARST AQUIFERS
As a generalization, if carbonate rocks such as limestone,
marble, or dolomite are present, or if more soluble rocks such as
salt or gypsum are present, assume that the water moving through
these rocks is in a karst aquifer — until or unless convincingly
proved otherwise. This statement is correct probably 95% of the
time. Assume also an Orwellian nature to the definition of a karst
aquifer: All carbonate terranes are karstic and underlain by one
or more karst aquifers, but some are more karstic than others.
If there is evidence for a karst terrane, assume that there is
also a karst aquifer. The aquifer, however, may be several orders
of magnitude less active than it once was, because it is a relict
karst and no longer has its original climate or hydrographic set-
ting, or because it is a buried karst which is completely or
largely de-coupled from the present hydrogeochemical system (Ford
and Williams, 1989, p. 507-512; Quinlan, 1978, 1990b; Samama, 1986,
p. 200-242; Bosak et al., 1989; Wright et al., 1991).
The karstic nature of an aquifer can often be inferred from
pumping tests in which the drawdown is stepped, even though pumping
is continuous. Such stepped drawdown curves, however, can also be
a consequence of flow through fractures that have not been dissolu-
tionally enlarged. All one can be confident of is that flow is
non-Darcian, so one still must be cautious. Conversely, however,
a smooth drawdown curve does not mean an aquifer is not karstic.
Other indicators of the karstic nature of an aquifer — or
587
-------
that it is fractured, or both — are:
• Irregular cone of depression or no cone of depression, as
defined by multiple observation wells around a pumped well;
• Non-linearity of the drawdown vs. discharge plot;
• Stair-stepped, irregular configuration of potentiometric
surface (where there are enough data to depict the surface
with reasonable accuracy);
• Strongly bimodal distribution of the logarithm of hydraulic
conductivity of a suite of wells completed in the same for-
mation (Smart et al.. 1991). Caution: Karst aquifers may
also have a unimodal log-normal distribution of hydraulic
conductivity (Moore, 1988a, 1988b);
• Non-coincidence of water levels in closely adjacent wells;
• Bimodal or polymodal distribution of daily or continuous
measurements of specific conductance (Ford and Williams,
1989, p. 213-214; Bakalowicz and Mangin, 1980);
• Significant variation in specific conductance and hydraulic
conductivity, as interpreted from wellbore fluid logs (Ped-
ler et al., 1990);
• Significant variations in the distribution of discharge, as
measured by an impeller meter moved vertically during pump-
ing at a constant rate (Molz et a!., 1989);
• Significant variations in the distribution of flow in a
pumped or unpumped well, as measured by a movable electro-
magnetic or thermal flowmeter in the borehole (Nyquist et
al.. 1991; Kerfoot et al.. 1991; Hess and Paillet, 1990);
• Significant differences in the breakthrough curves in a
well for different dyes injected into several wells and re-
covered in that well (Quinlan, 1992).
The first four of these indicators are illustrated and briefly
discussed by Wisconsin Geological and Natural History Survey
(1991). Proof of appreciable flow through joints, faults, bedding
planes, and cavities can also be obtained by inference from cores,
video logs, caliper logs, and temperature logs. Lacking such
proofs, if the bedrock is limestone or dolomite, one would be well-
advised to believe in the probable presence of a karst aquifer.
PRINCIPLES FOR CLASSIFICATION OF CARBONATE AQUIFERS
The behavior of an aquifer is dependent on three essentially
independent properties: the mode of recharge, extent of storage,
and type of groundwater transmission (flow) (Smart and Hobbs,
588
-------
1986). In karst terranes, recharge may range between predominately
point recharge (for instance, via sinking streams) to dispersed
recharge (for instance, where water enters the karst aquifer over
a broad area and through an overlying permeable sandstone, thick
soil, or perhaps densely fractured rock with a thin soil). Simi-
larly, flow can be considered as a continuum between conduit flow
in large-diameter, pipe-like caves and diffuse flow via fractures
and intergranular pores. Storage is largely controlled by porosity
and geological disposition of the aquifer, but ranges from high,
where there is a deep saturated zone, to low, where the aquifer is
perched and has a limited extent and thickness. In order to char-
acterize an aquifer, it is necessary to define its position on each
of the three continua, effectively mapping it into a three-
dimensional space defined by recharge, storage, and flow (Figure
1) . Also shown in this figure are the general types of recharge,
storage, and flow which characterize each continuum. These are
discussed further below and the rationale for them is given by
Smart and Hobbs (1986) , but it is important to stress here that the
boundaries are fuzzy. Each type grades into and overlaps with
those adjacent.
Conduit flow, in the strict sense, has been and should be used
to refer to flow through dissolution passages with diameters of
centimeters to meters, following the usage of Shuster and White
(1971), Atkinson and Smart (1981), and Smart and Hobbs (1986).
Velocities are commonly high; flow is commonly turbulent. Diffuse
flow, in the strict sense, is confined to tight fractures and pores
with small, interconnected openings, with diameters of centimeters
or less. Velocities are low; flow is commonly laminar and may be
Darcian. These two terms, conduit flow and diffuse flow, have been
subsequently (and erroneously) broadened by Quinlan and Ewers,
1985; Quinlan 1989, 1990a) and by numerous other writers to include
implicit characteristics of water-quality and discharge variabili-
ty. Between the two end-members, we can intuitively recognize a
continuum of intermediate flow types. Pipe flow in open cave pass-
ages represents a flow type close to the conduit end-member. In
other karst aquifers, dissolution has resulted in dense networks of
open fissures which are intermediate in character between conduit
and diffuse flow. Finally, flow through intergranular pores and
fractures essentially unmodified by dissolution approximates the
diffuse-flow end-member.
Diffuse flow, as used in describing karst aquifers, should not
be construed to be the laminar, dispersed flow characteristic of
granular aquifers and described by Darcy's Law. The term diffuse.
as used here, is intended to mean slow, both laminar and slightly
turbulent flow of water through a system of small, discrete path-
ways that are being dissolutionally enlarged, albeit extremely
slowly. These pathways range from fractures and near-microscopic
conduits to intergranular pores. Hence, a diffuse-flow karst aqui-
fer may be described as having flow generally restricted to dis-
crete pathways, with minimal turbulence rather than the extremes of
turbulence characteristic of conduit-flow aquifers. The principle
difference between a diffuse-flow part of a karst aquifer and a
589
-------
CLASSIFICATION OF CARBONATE AQUIFERS
VULNERABILITY
HYPERSENSITIVE
KARST AQUIFERS
VERY SENSITIVE
KARST AQUIFERS
MODERATELY SENSITIVE
KARST AQUIFERS
SLIGHTLY SENSITIVE
NON-KARST AQUIFERS
(In Fractured Rock)
NON-AQUIFERS
Figure 1. Conceptual model for carbonate aquifers which recognizes
independence of recharge, storage, and flow. Volumes #1 through #4
depict the relative vulnerability of karst and non-karst carbonate
aquifers, ranging from hypersensitive to slightly sensitive. Each
of the three axes represents a different non-arithmetic continuum
that ranges from 0% to 100% for each end-member. Boundaries be-
tween vulnerability fields are approximate, intuitive, and transi-
tional. The relationships shown and implied are depicted within a
cube because there are three independent variables, each of which
may have a value up to 100%. The coefficient of variation (CV) of
hydraulic conductivity, as discussed by Smart et al. (1991), and
the CV of specific conductance (conductivity), as discussed in this
paper, are each at a maximum at the upper far vertex of the cube
and a minimum at the lower near vertex; the CV of both parameters
decreases along the diagonal connecting them. Not all the boun-
daries of volume #5 (non-aquifers) are shown. This solid extends
been used in two senses: the strict sense, referring to water
transmission through variously sized apertures, and in the broad
along part of the back edge of the cube but its edges have been
omitted, for clarity. (Greatly modified after Smart and Hobbs,
1986)
590
-------
Darcian granular aquifer is that the flow pathways in the former
have been and are being significantly enlarged by dissolution.
To partially summarize, conduit flow and diffuse flow have
been used in two senses: the strict sense, referring to water
transmission through variously sized apertures, and in the broad
sense, referring to aquifer type as represented by the sum of all
responses to recharge, storage, and flow. The sense intended by a
writer must be interpreted from context. We henceforth use them in
the strict sense, to describe transmission.
Recharge ranges from point to dispersed. Point is used in the
same sense that Smart and Hobbs (1986) used concentrated. but the
former term was selected to be in conformity with its widespread
usage in America. However, we have deliberately not used its
false, apparent antonym, non-point, because it does not mean aeri-
ally distributed. Although large quantities of "high-strength"
agricultural runoff may enter an aquifer at the swallet of a sink-
ing stream, i.e., at a point, it is technically classified by most
state agencies as non-point pollution. Accordingly, we have re-
tained dispersed in the same sense as Smart and Hobbs (1986).
Point recharge is characterized by sinking and losing streams.
However, even if these are absent, concentration of recharge may
still occur in karst terranes with closed depressions where drain-
age is often via a shaft system (Smart and Friedrich, 1986). This
may also occur in limestone aquifers overlain by a thick regolith
and lacking obvious depressions. Aquifers with recharge through
shaft systems occupy the center of the recharge continuum, grading
into recharge through fissures toward the dispersed end-member.
The latter is characterized by recharge through fractures and
intergranular voids.
Storage ranges between high and low end-members (defined in
terms of the ratio of storage volume to annual recharge) . Two
factors control storage, the effective porosity of the aquifer and
its geological disposition in relation to regional base-level and
any confining beds (White, 1977). In practice, the aquifer config-
uration dominates. Hence we divide the continuum into three tran-
sitional groups. The first comprises aquifers with limited stor-
age, predominantly in the unsaturated zone, primarily in soils and
the subcutaneous (epikarst) zone (Smart and Friedrich, 1986). The
second is similar to the first but also has a seasonally drained
saturated zone that may be part of a shallow water-table aquifer.
Finally, the high-storage end-member comprises aquifers with an un-
saturated zone, a seasonally inundated saturated zone, and either
perennial storage in a confined aquifer or a water-table aquifer
which extends well below spring level. Examples of such high-
storage karst aquifers are the Madison aquifer, chiefly in Montana
and Wyoming (Downey, 1984; Plummer et al., 1990), that in the
carbonate-rock province of the Great Basin, chiefly in Nevada and
Utah (Burbey and Prudic, 1991), and in the Floridan aquifer (Sun
and Weeks, 1991, p. 5-9; see also U.S. Geological Survey Profes-
sional Papers 1403-A through I).
591
-------
Storage in conduits (pipes), as a fraction of the total stor-
age within an entire aquifer, is always very small. Thus aquifers
having storage only in conduits would represent the extreme low-
storage end-member.
Each of the three independent variables that determine the
nature of an aquifer — recharge, storage, and flow — ranges from
the extremes shown in Figure 1 by upper-case letters, POINT to
DISPERSED, LOW to HIGH, and DIFFUSE to CONDUIT. The three major
types of each of the three independent variables are shown in
lower-case letters indented between their two extremes. These
three sets of three major types are each part of a non-arithmetic
continuum that extends from 0 to 100% fracture and intergranular
recharge, 0 to 100% soil + subcutaneous + seasonally saturated +
perennial storage, and 0 to 100% dominance by conduit-flow, as
shown.
The eight corners of the cube correspond to the following end-
members, most of which probably do not exist as pure end-members
and two of which are not karst. As with any continuum, however, it
is easier to think in terms of the end-members while recalling that
most of reality lies between them. The pure end-member vertices
are:
1. Point recharge, high storage, conduit-flow karst aquifer;
2. Point recharge, low storage, conduit-flow karst aquifer;
3. Point recharge, high storage, diffuse-flow karst aquifer;
4. Point recharge, low storage, diffuse-flow karst aquifer;
5. Dispersed recharge, high storage, conduit-flow karst aqui-
fer;
6. Dispersed recharge, low storage, conduit-flow karst aqui-
fer;
7. Dispersed recharge, high storage, diffuse-flow non-karst
aquifer;
8. Dispersed recharge, low storage, diffuse-flow non-aquifer.
Obviously, there are numerous intermediate types of karst
aquifers, but we do not propose to describe them or the end-members
in detail here or to give examples of each of them. Instead, we
have stated and illustrated the principles for classification of
carbonate aquifers.
Remember, Figure 1 is a classification of carbonate aquifers,
not karst terranes. As mentioned in the Introduction, there are
seven other criteria by which karst can be usefully classified. To
be practical, however, we use here only the criterion that is most
relevant to the movement of and monitoring of pollutants — aquifer
hydrology.
Moderately sensitive karst aquifers may behave in a less
flashy manner than hypersensitive karst aquifers because of either
diffuse flow or high storage or dispersed recharge or any combina-
tion of them, not simply because of diffuse flow. In fact, chem-
ograph/hydrograph criteria do not separate which of these is the
case, only that aquifer responses are slower and more damped.
592
-------
RELATION BETWEEN TYPES OF KARST AQUIFER AND
SENSITIVITY TO GROUNDWATER CONTAMINATION
The particular susceptibility of karst aquifers to groundwater
pollution is related to the rapid transmission of pollutants with
little attenuation, dilution, or dispersion via dissolutionally
enlarged fractures and cave conduits (Aley, 1986; Hoenstine et al. ,
1987; Hannah et al., 1989; Murray et al., 1981; Quinlan and Ewers,
1985; Edwards and Smart, 1989; Quinlan and Rowe, 1977). However,
recharge and storage processes are also important. If recharge is
via dispersed routes such as a porous sandstone caprock overlying
the limestone, any pollutant entering from the surface will be con-
siderably attenuated and delayed before reaching the cave conduit
in the underlying limestone. The risks of severe contamination of
a spring or other discharge point are thus somewhat reduced, al-
though they could still be high if the contaminant had direct
access to the subsurface, for instance, via a drainage well (Craw-
ford, 1984). If there were a substantial body of water stored in
phreatic fissures and pores adjacent to a conduit, further dilution
and attenuation are possible before the pollutant is discharged at
the spring or other discharge point. Thus, moderate to high stor-
age and dispersed recharge will reduce susceptibility of discharge
points to contamination.
Intuitively, we can infer the susceptibility of carbonate
aquifers to groundwater contamination by considering the type of
aquifer defined with reference to the recharge, flow, and storage
continua described above. Accordingly, in Figure 1, we introduce
and define four new terms for zones representing relative sensitiv-
ity to groundwater pollution: hypersensitive karst aquifer (at the
upper far vertex, volume #1), very sensitive karst aquifers (volume
#2), moderately sensitive karst aquifers (volume #3), and slightly
sensitive non-karst aquifers (at the lower near vertex, volume #4) .
There is also a non-aquifer class (volume #5) in which the lack of
concentrated flow and low saturation preclude practical abstraction
of water. Effectively, we have added a fourth continuum to Figure
1 which ranges from hypersensitive to groundwater contamination to
slightly sensitive thereto. Note that all aquifers of whatever
type are sensitive to groundwater contamination and thus require
proper management. For this reason, we rejected the end-member
terms non-sensitive and robust.
Hypersensitive aquifers are the most karstic of all karst
aquifers and are characterized by conduit flow in pipes (caves),
concentrated recharge (via swallets, stream beds, or shaft drains),
and low storage (limited saturated zone). There is a high prob-
ability that pollutants introduced at the surface will pass rapidly
downward, without attenuation, directly into conduits where they
will be transmitted rapidly via pipe-flow to springs (generally the
only easily found high-yield source of water in such aquifers).
Even if there is limited saturated storage, little dilution or
attenuation of pollutants will occur as point recharge generally
bypasses the storage sites (Atkinson, 1977; Smart and Friedrich,
1986) .
593
-------
Very sensitive aquifers are characterized by the presence of
either well-developed flow in pipe-like caves (conduits), point
recharge via swallets, or storage ranging from very low (unsaturat-
ed) to very high (saturated), or a combination of two of these.
Moderately sensitive aquifers do not have pipe-flow, point
recharge through swallets and losing streams, or low storage.
However, there is still a significant risk of groundwater contami-
nation, with the possibility of rapid recharge through closed
depressions, rapid transmission through dissolutionally-enlarged
fissure networks, and limited dilution when seasonal storage is at
a minimum.
As emphasized in our earlier definition, karst aquifers are
characterized by appreciable flow through dissolutionally developed
voids. It must be recognized, however, that for several reasons,
some carbonate aquifers (extremely few, in our experience) have not
undergone significant dissolution. These reasons include limited
exposure time, presence of overlying carbonate-rich regolith, and
low solubility of bedrock. These non-karst carbonate aquifers are,
therefore, identical in character to clastic aquifers in which flow
is via intergranular pores or fractures developed during unloading
or exposure in the near-surface environment. They are shown as
volume #4 in Figure 1. For these non-karst carbonate aquifers,
special attention is not needed; conventional techniques of aquifer
development, protection, and monitoring are probably applicable,
but may have to be modified because of fracturing.
CHARACTERIZING CARBONATE AQUIFER SENSITIVITY
Introduction
The scheme described above uses observable criteria and pro-
vides a simple, intuitive basis for assessing the sensitivity of
carbonate aquifers to groundwater contamination. However, we are
not yet able to adequately fix the position of individual aquifers
within the continua. Clearly, for administrative and regulatory
purposes, we need a much simpler scheme, one which, although less
rigorous, is capable of differentiating aquifers which have a high
pollution risk from those which are less susceptible.
A potential solution to this problem is to use the geochemical
or hydrological behavior of a major spring to characterize the
nature of the aquifer it drains. We treat the aquifer as a black-
box which transforms the "signal" from natural storm recharge
(which has a distinctive chemical signature) into an output res-
ponse (hydrograph or chemograph) at the spring, or possibly at a
seep or well shown to intersect fracture zones, dissolution zones,
or conduits. If this signal is substantially smoothed, delayed,
and attenuated by passage through the aquifer, then the same will
happen to a pulse of pollutant; the aquifer will be only moderately
sensitive to contamination. If it is transmitted rapidly, with
little or no attenuation or dilution, the aquifer is hypersensi-
tive.
594
-------
Background
Fortunately, characterization and interpretation of geochemi-
cal and hydrological variation of springs has been the basis for
many studies of karst aquifers (see the summary in Ford and
Williams, 1989, p. 193-214, and the analysis by Kiraly and Muller,
1979). In a classic paper, Shuster and White (1971) discussed the
variations of water hardness, degree of saturation with respect to
calcite and dolomite, and Ca/Mg ratios of springs in the Appal-
achians of Pennsylvania. They recognized two types of springs:
(1) those with highly variable water quality, wide fluctuations in
discharge, and rapid response to recharge events (which they called
conduit-flow springs) and (2) those with much smaller changes in
water quality parameters, higher saturation state, small fluctua-
tions in discharge, and lagged, prolonged hydrograph response
(which they called diffuse-flow springs). Similar studies by
Newson (1971) in the Mendip Hills, England, supported the observed
continuum in geochemical response but suggested that the highly
variable springs there were dominated by point recharge at swal-
lets, while the less variable springs were recharged predominantly
by diffuse percolation. The proportion of diffuse to point re-
charge was similarly used by Atkinson (1977) in explaining Mendip
spring hydrograph response. In contrast, Aley (1975, 1977) dis-
cussed the same hydrograph features in terms of the proportions of
transit and storage water.
Despite the disparity in these explanations of an essentially
identical range of spring behavior, all of them are correct. As we
have discussed previously, the most karstified aquifers (hypersen-
sitive aquifers) are characterized by low storage, pipe flow, and
point recharge. There is a strong tendency for these three charac-
teristics to be genetically and functionally linked. Thus maximal
concentrated dissolution of the bedrock occurs where aggressive
surface water drains into limestone as recharge from adjacent im-
permeable rocks. Caves develop (Palmer, 1990). Conduits eventu-
ally link the swallets directly to the springs which discharge the
sometimes large flow entering at the swallets (Ford and Williams,
1989, p. 249-271). Where conduits are developed, rapid discharge
of groundwater can occur, and the amount entering into long-term
storage is relatively limited. The flashy, highly variable behav-
ior of Shuster and White's conduit-flow springs is a consequence of
the fact that much of their recharge is rapid, at discrete points,
and into conduits, and the fact that the conduit flow reduces the
effective aquifer storage. However, the more subdued response of
their diffuse-flow springs could equally well be explained either
by lack of point recharge, or by a geological disposition favoring
substantial storage, or by the dominance of diffuse flow.
In Figure 1, variability in the chemistry and flow of a spring
(or in the chemical composition and water level of a well that
intersects a fracture zone, dissolution zone, or conduit) decreases
systematically along all axes of the cube as one moves away from
the most karstified end-member, at the upper far vertex. The most
damped response is expected at the farthest distance from the
hypersensitive vertex. As one might expect, this is the diagonally
595
-------
opposite lower near vertex — aquifers that are not appreciably
karstified and are thus only slightly sensitive to groundwater
contamination.
Although the above generalizations by Shuster and White were
based primarily on water hardness and related to discharge and lag-
time, they are equally valid for numerous other water quality and
quantity parameters of a spring or an aquifer. The easiest para-
meters to monitor, however, are specific conductance (hereafter
referred to as conductivity), temperature, turbidity, discharge,
and lag-time between onset of recharge and the time of hydrograph
peak. The conductivity, an easily determined electrical property
of water, correlates with the sum of dissolved major-ion concentra-
tions in water and often with a single dissolved-ion concentration
(Hem, 1985; Miller et al.. 1988). It is a quick, extremely inex-
pensive surrogate for a chemical analysis when it is not necessary
to know what specific major ions are present.
The nature of a spring or other suitable monitoring point (and
the aquifer that feeds them) can be adequately determined by
systematically measuring its conductivity and by observing and es-
timating its turbidity, discharge, and lag-time. These latter
three parameters, as well as temperature, can be quantified, but it
is not always necessary to do so. Indeed, variations in turbidity,
discharge, and lag-time can often be estimated accurately enough
for reliable characterization by interviewing local residents.
Geochemical Variation
The direct relationship between water hardness and conductivi-
ty allows us to use the same numerical values for discriminating
spring type (or aquifer type) as did Shuster and White (1971).
Diffuse-flow aquifers (as defined by them) are characterized by a
conductivity in which the coefficient of variation (CV; standard
deviation x 100 •*• mean) is less than 5%. Conduit-flow aquifers (as
also defined by them) are characterized by a range of conductivity
in which the CV is greater than 10%. Aquifers that have a con-
ductivity CV between 5 to 10% are interpreted to be a mixture of
diffuse flow and conduit flow. [These empirical class boundaries
are arbitrary but utilitarian. The general validity of using hard-
ness at all, however, has been challenged, as discussed in the two
paragraphs following the next. Nevertheless, conductivity CV has
seemingly worked well to date.] We have assumed the same CV
boundaries for conductivity as an index of aquifer sensitivity.
For moderately sensitive aquifers the conductivity CV is 5% or
less. For very sensitive aquifers it is between 5 and 10%, and for
hypersensitive aquifers the CV is greater than 10%. Work is in
progress to verify and, if necessary, adjust these boundaries.
We estimate that most of the other chemical parameters will
probably have similar CV's.
Shuster and White's use of carbonate hardness for defining the
nature of an aquifer was strongly criticized by Bakalowicz (1976)
596
-------
who emphatically asserted that conductivity is a regionalized vari-
able — in the sense of Matheron (1965) and as the concept is dis-
cussed by David (1977, p. 91-114) and by Henley (1981, p. 9-34) —
and that it provides a more accurate and sensitive indicator of the
characteristics of an aquifer. Regular measurements by Bakalowicz
(1979), cited by Bakalowicz and Mangin (1980) and Ford and Williams
(1989, p. 213-214, Fig. 6.24), have shown the following about the
frequency distribution of conductivity at springs in various types
of aquifers:
Granular aquifers: unimodal, relatively high conductivity
Fractured aquifers: unimodal, relatively low conductivity
Karst aquifers: polymodal, wide range of conductivity
Bakalowicz and Mangin (1980) and Mangin (1984) stated that the
CV of hardness is not a valid measure of the complexity of the
polymodal distribution of hardness. This is partly because the
accuracy of the standard deviation used to calculate CV is based on
the validity of the assumption of a normal distribution (Hamburg,
1977, p. 42-44). They are technically correct, but we disagree
with the necessity for their rejection of its use. The standard
deviation can still be calculated for a polymodal distribution, and
it is still a measure of dispersion about the mean. However, its
meaning is different. For example, the mean + two standard devia-
tions might include only 80% of the conductivity values rather than
95.5% of them. We justify use of the CV of conductivity, imprecise
as it is for a polymodal distribution, with a quote from the emi-
nent statistician John W. Tukey: "Far better an approximate answer
to the right question which is often vague, than an exact answer to
the wrong question, which can always be made precise" (Tukey, 1962,
p. 11) . [This same rationale could be employed to justify cautious
use of standard techniques for interpreting slug, bail, and pump
tests in karst aquifers.]
Bakalowicz (1979) also showed that the frequency distribution
for daily measurements of conductivity at a particular karst spring
in the Pyrenees varied significantly from year to year over a 4-
year period and was trimodal; the amplitude and frequency of the
maxima changed from year to year and, as one might expect, from
storm to storm. We have used the daily data in his dissertation to
calculate that the CV of conductivity at this spring for the 4-year
period was 6.4%, 7.9%, 6.4%, and 5.7% (mean =6.6%) — in spite of
its high variability — suggesting that it is a very sensitive
spring but not greatly different from a hypersensitive spring. We
interpret the slightly more than low value of its CV to be a func-
tion of fluctuations of recharge, storage, and flow in the aquifer
studied. We do not have comparable long-term data for American or
British karst springs, but a review of existing data and a study of
other springs are underway.
The use of conductivity as a measure of the nature of an aqui-
fer is not an issue here. The issues are whether its CV is a reli-
able measure thereof and what arbitrary CV boundaries can (or
should) be drawn between spring and aquifer types. These matters
are being evaluated by us and will be described in future publica-
597
-------
tions. In the meantime, we consider CV < 5%, CV = 5 to 10%, and
CV > 10% to be practical boundaries for discrimination between
aquifer vulnerability types.
Conductivity CV's have been used to identify and quantify the
difference between two groundwater basins (Quinlan and Ewers, 1981,
p. 488, 490). A hypersensitive spring (mean = 326 micromhos per
cm; CV = 14%) was clearly differentiated from a moderately sensi-
tive spring (mean = 488 micromhos per cm; CV = 2%) with about one-
thirtieth the drainage area of the hypersensitive spring and dis-
charging in the middle of a distributary. The latter is used for
water supply; the former is not.
Two generalizations, based on the field experience of the
senior author, can be made about the relations between conductiv-
ity, spring and aquifer type, and groundwater basin size. First,
for equivalent basin size, a moderately sensitive spring has a
significantly higher conductivity than a hypersensitive spring; 20
to 30% higher is not unusual. Second, for both spring types, con-
ductivity is directly proportional to basin size, which is a
control on residency time (time for reaction) of water.
Hydrological Behavior
For hypersensitive aquifers, the ratio between flood-flow and
base-flow discharge may well be 10:1 to perhaps 100:1. Lag-times
between storm events and hydrograph maxima will range from a few
hours to a few days and be directly proportional to basin size.
Duration of response may be a few days to a few weeks. In moder-
ately sensitive aquifers, the ratio between flood-flow and base-
flow discharge may be 3:1 or less; lag-time will range from days to
weeks or even months, again in direct proportion to basin size, as
discussed by White (1988, p. 186), but as conduit-flow and diffuse-
flow aquifers. Duration of response may be several weeks to sever-
al months.
One could use just the ratio of flood-flow discharge to base-
flow discharge to determine flow-type (and, by inference, probable
spring type or aquifer type), as discussed by White (1988, p. 184-
187) concerning aquifer type, but such data are essentially non-
existent in either published or open-file form.
One might also analyze the exponentially decaying recession
limb of several spring hydrographs and calculate the response time,
tR, that is characteristic of the spring and its aquifer type.
Physically, tR is the time it takes for the hydrograph to fall to
about 37% (= 1/e) of its maximum. This time ranges from a few days
to a few thousand days and depends upon many factors not yet fully
understood. More accurately, tR is the reciprocal of epsilon (what
has been called the exhaustion coefficient), the constant in the
variable exponent of e in the empirically derived equation that
describes the relationship between spring discharge and time, as
discussed by White (1988, p. 186). Epsilon is the slope of the
straight line describing that relationship when the data are plot-
598
-------
ted on semi-log paper, as explained by Krumbein and PettiJohn
(1938, p. 209-210). [This graphical procedure is mentioned here
because a similar technique will be proposed, in the Appendix, for
interpreting the recession limb of storm-related conductivity data
and design of an optimal plan for reliably sampling springs and
conduit-fed wells.] Such an analysis of discharge may be useful if
one has the luxury of time and money for instrumentation of a
spring and if it is physically possible to gauge it reliably.
There are, however, other problems, as discussed by White (1988, p.
186) :
"Precipitation events are variable in intensity, spacing
and duration. If tR is much less than the mean spacing
of precipitation events, hydrographs will be flashy and
the yearly record will consist of a series of sharp
peaks. If tR is on the same order of magnitude as the
mean spacing of precipitation events, individual storms
will tend to overlap but seasonal changes in precipita-
tion will appear. If tR is much longer than the mean
spacing of precipitation events, the hydrographs will be
broad and relatively featureless. Large systems tend to
have longer response times than small systems simply be-
cause of the longer times needed to transmit the storm
impulse from input to output. Thus, short response times
and high discharge ratios are indicators of conduit
systems only for small basins."
White's analysis, above, is generally valid and we agree with
it, but he assumes basin homogeneity and ignores the possible
influence of local, near-spring input of storm runoff in a large
basin. By inference, his conclusion about short response times and
high discharge ratios would be also applicable to interpretation of
conductivity variations. The validity of this inference depends
upon the distinction between small basins and large basins.
Within our experience in the eastern, midwestern, and south-
western U.S., we estimate that damping of spring hydrograph and
chemograph response caused by basin size is not significant at
springs draining less than about 500 square miles. We conclude
that, for whatever the several reasons that control a spring's be-
havior, monitoring of its conductivity and determination of its
conductivity CV are reliable measures of spring (aquifer) variabil-
ity and response to recharge which may include pollutants. Clear-
ly, research is needed on basin hydraulics and response as a func-
tion of numerous variables.
Limitations
Using only the variation in conductivity CV and discharge for
spring (aquifer) characterization is not infallible. For example,
the fourth author of this paper, E.G. Alexander, Jr., and his stu-
dents have been studying the hydrology of Lanesboro Spring, near
Lanesboro, Minnesota. Its hydrographs, chemographs, and conductiv-
ity plots are those of classic diffuse-flow springs — more than
599
-------
99% of the time. Yet dye injected at either of two wet-weather
swallets, 2.5 miles away, during base-flow conditions was recovered
in only 8.5 hours and with a breakthrough curve having a very steep
rising limb. The recession limb is complex, has multiple peaks
that are reproducible in repeated tests, and takes 2 weeks to re-
turn to background levels. The occasional muddying of the Lanes-
boro spring waters (3 times in 2 years) and the results of these
repeated tracer tests* during base flow can be explained in terms
of infrequent storm-related point-recharge into an otherwise dis-
persed recharge, moderate to high storage, conduit-flow aquifer
that is probably very sensitive to pollutants for more than 99% of
the year and hypersensitive for less than 1% of the year. The be-
havior of this spring is totally consistent with the ideas expres-
sed here and by Smart and Hobbs (1986), but at variance with those
of Shuster and White (1971) if one thinks of the aquifer just in
terms of diffuse flow (as suggested by the very low CV for its con-
ductivity) rather than recharge, storage, and flow. The spring
behavior also brings into question the reliability of conductivity
CV in assessing aquifer vulnerability.
We do not have enough data to know how common such inconsis-
tency is between low CV of conductivity and occasional hypersensi-
tivity of a spring and its aquifer, or whether it occurs as a gen-
eral phenomena. We do believe, however, that such storm-related
behavior is probably uncommon and contributory to a distinctive-
mode on a conductivity-frequency plot. Nevertheless, we also
believe that use of the conductivity CV, as we advocate, will usu-
ally be reliable for an overwhelming majority of springs. This
reliability justifies its use.
The valuable concepts of conduit-flow aquifer and diffuse-flow
aquifer (as defined by Shuster and White, 1971) have been usefully
applied for more than 30 years, even though Bakalowicz (1976),
Bakalowicz and Mangin (1980), Mangin (1984), and Smart and Hobbs
(1986) questioned them and called attention to their deficiencies.
We recommend here that these two terms be retired as the basis for
classification of aquifer type — even though there is usually a
* Briefly stated, most tracer tests involve injection of a recog-
nizable non-toxic substance at one point and recovery of it at
another, yielding information about flow-direction, -dispersion,
-destination, and -velocity (Quinlan and Alexander, 1990;
Atkinson and Smart, 1981; Lepiller and Mondain, 1986). A well-
designed, properly performed, and correctly interpreted tracer
test, most commonly employing one or more fluorescent dyes, is
the most powerful technique for studying karst aquifers (Aley,
1986; Ford and Williams, 1989, p. 219-241; Quinlan, 1989; Quinlan
and Ray, 1989) but, until recently, they have rarely been used,
except by karst specialists. The value of tracing in all types
of aquifers, however, is now widely recognized (Moltyaner, 199Oa,
1990b; Maloszewski and Zuber, 1990; Tsang et al. , 1991; Adams and
Davis, 1991).
600
-------
partial overlap of so-called conduit-flow aquifers with hypersensi-
tive aquifers and usually a partial overlap of so-called diffuse-
flow aquifers with moderately sensitive aquifers. The terms
conduit-flow and diffuse-flow should be retained, however, as des-
criptors of flow type and to refer to aquifers dominated by one
type of flow or the other.
PRACTICAL EVALUATION OF VULNERABILITY OF KARST AQUIFERS
Procedure
A spring is the pulse of the aquifer it drains. Understanding
the variations in its discharge and water quality (or those of
other suitable monitoring points) is the key to understanding the
hydraulics of its (their) aquifer and assessing aquifer vulner-
ability -
We believe that interpretation of water quality and quantity
parameters, as described herein, even though less meticulous than
proposed by Smart and Hobbs (1986), is practical and justifiable
for the purposes we propose — to identify aquifer vulnerability —
and adequate for use by non-specialists. The classification pro-
cedure we advocate is less subjective than what they proposed.
Smart and Hobbs are more rigorous, but we are more practical.
Quantitative characterization of recharge, storage, and flow pro-
portion are difficult, time-consuming tasks best left to special-
ists.
The conductivity CV of a spring (and, therefore, of a ground-
water basin presumed to be representative of the aquifer it drains)
should be determined from at least 20 to 40 daily, weekly, or other
regular or random measurements of its conductivity. The measure-
ments should be taken at least a day apart and must include both
base flow and flood flow during the estimated peak discharge. We
can not say how many samples are required; we can only advise that
caution must be used to make the samples statistically representa-
tive of the various conditions within the aquifer. This is easier
said than done.
It is critically important to include conductivity measure-
ments collected under both base-flow and storm-flow conditions.
Twenty conductivity measurements taken only under base-flow condi-
tions, perhaps under the pressure of an arbitrary 30-day deadline,
will probably have a CV of less that 5% and will "prove" that the
spring is characterized by diffuse flow. Such a conclusion is
wrong because it is based on data that do not represent the true
variability of the system. Reliable interpretation of conductivity
is aided by also having daily measurements of rainfall, the maximum
daily rainfall, and the maximum rainfall per storm.
We reiterate: If the conductivity CV is 5% or less, the
spring (aquifer) is moderately sensitive to pollution and is prob-
ably characterized by mostly diffuse flow. If the conductivity CV
is greater than 10%, the spring (aquifer) is hypersensitive and is
601
-------
characterized by a much greater proportion of conduit flow. If the
conductivity CV is between 5 and 10%, the spring (aquifer) is very
sensitive and is probably a mixture of both diffuse flow and con-
duit flow. Also, reliable inferences can be made about whether an
aquifer is dominated by moderate sensitivity or is hypersensitive,
both by observing the turbidity and discharge during, immediately
after, and shortly after storms of different intensities, and by
interviewing landowners who are familiar with the springs and their
behavior.
Not a single published American report known to us has enough
data that would enable an investigator to run a few simple conduc-
tivity CV calculations and determine the vulnerability (sensitivi-
ty) of the groundwater basins in a given area. Therefore, if one
needs such data for a specific area, it is highly likely that they
can be acquired only by field work.
Difficulties and Limitations of Vulnerability Evaluation
The conductivity CV is a useful method for characterizing the
vulnerability of aquifers; it must be remembered, however, that the
hydrograph or chemograph (a plot of chemical concentration or con-
ductivity vs. time) of a spring reflects the recharge, flow, and
storage characteristics of the groundwater basin that it drains.
For site-specific investigations in karst aquifers, the springs
used to characterize an aquifer will probably not be on a site to
be evaluated and may even be several miles away. Their possible
relevance to the hydrology of the site must be evaluated with the
aid of geologic maps of the area adjacent to the site; dye tracing
may also be necessary to prove or disprove their relevance of
particular springs to a given site (Quinlan et al., 1988).
The conductivity CV method has some limitations in areas where
there are few springs and where springs are only located several
miles from a site under investigation and tracing has demonstrated
that flow velocities are, for example, 1 mile per year or slower.
An alternative approach in these areas is to interview well owners
and determine if their well water turns turbid or muddy after heavy
rains. Be cautious. Approximately 90 to 95% of the wells in
karsts dominated by conduit flow are in the diffuse-flow part of
the aquifer that feeds the conduits. [These are usually hypersen-
sitive aquifers.] Accordingly, many well owners, perhaps 150 to
200, need to be interviewed before one can be confident about using
wells to ascertain whether an aquifer is hypersensitive rather than
moderately sensitive. Some of these 5 to 10% of wells that turn
turbid or muddy are the ones most likely to be positive for dye
during a tracer test.
We hear the argument that "My client can't afford to interview
150 to 200 well-owners." We reply that "If you need the truth, you
can't afford not to do so."
Evaluation of conductivity CV has severe limitations for some
karst terranes, such as those in much of Florida, that are charac-
602
-------
terized primarily by recharge and flow rather than also by dis-
charge. Discharge at springs may be several tens of miles away,
perhaps beneath the ocean; flow velocities may be no more than a
few hundred feet per year. The springs, if accessible, can be
monitored for conductivity, but they will probably be inaccessible.
Well-to-well tracer tests may be practical for determination of
flow velocities for either the design of monitoring systems or for
determination of wellhead and springhead protection areas based on
time of travel, but tracer tests would be unsuccessful in delinea-
tion of wellhead or springhead protection areas based on basin
boundaries because it would take too long to do so. Conductivity
CVs of wells in such terranes are not likely to be meaningful.
Currently, some of the serious limitations in applying the
conductivity CV at springs as a method of assessing aquifer vulner-
ability are: 1) the false assumption (by some agencies, clients,
and consultants), that monitoring wells alone are sufficient to
characterize a karst aquifer, and 2) an unwillingness by some of
the same to extend monitoring efforts beyond the boundaries of a
site under investigation, to all possibly relevant springs, wells,
and base-level stream segments. [If such short-sighted views and
restrictions are allowed to prevail, and they should not be, the
quality and credibility of a tracing investigation is severely
compromised. So also is the integrity of the investigator compro-
mised. ]
We argue that a limited number of boreholes, for example five
at a given site, do not provide a statistically valid confirmation
of the existence or probable non-existence of conduits in the sub-
surface. Benson and La Fountain (1984) have shown that drilling
1,000 3-cm holes per acre is necessary in order to have a 90%
chance of intercepting a cavity about 2.3 m in diameter. A lesser
number would be needed for finding a conduit of the same diameter,
but many of the conduits of interest at a given site are likely to
be only a tenth as wide (or smaller) . A thorough survey of springs
and existing wells (see above discussion), followed by a well-
designed tracer test, properly performed, and correctly interpret-
ed, will provide much more cost-effective and accurate information
about the site.
We also hear the argument that "My client (and his attorneys)
does not wish to perform any investigations off-site." In the real
world, water-flows (and contaminants) do not respect property
boundaries; they travel beyond them. A consultant may have to
convince clients and their attorneys that judicious investigation
that minimizes potential liability requires following the flow
wherever it may go. Off-site field work is probably always cheaper
than the off-site courtroom work it can prevent.
Well-meaning supervisors, clients, regulators, and consultants
with whom the reader interacts may exert pressure to assess an
aquifer from the office or without the amount of field work we are
recommending as minimal. In response, ask whether he or she would
respect or trust a C.P.A. who was willing to audit accounts over
603
-------
the telephone. A field problem can only be resolved by field work.
We wish to emphasize that:
1. If a terrane is underlain by limestone or dolomite, as
indicated by an accurate geologic map and/or site-specific
observations, the site is extremely likely to be part of a
karst and have a karst aquifer;
2. With one exception, there is no way without fieldwork, as
described, to determine whether the aquifers in a site area
are hypersensitive rather than moderately sensitive to con-
tamination and probably dominated by conduit flow rather
than diffuse flow. The sole exception is when so much
field work has already been done in the area that one can
confidently interpolate or extrapolate on the basis of
experience or pre-existing state-of-the-art literature.
USE OF AQUIFER TEST DATA TO CHARACTERIZE
FLOW IN AND VULNERABILITY OF KARST AQUIFERS
In some karst areas, where domestic wells are sufficiently
abundant and accessible, it may be possible to characterize flow
within an aquifer by using aquifer tests and to make inferences
about its vulnerability from such data. One advantage of this
approach is that it directly measures the transmission (flow)
properties of an aquifer, whereas the conductivity CV and hydro-
graph analysis methods (discussed above) integrate flow, recharge,
and storage properties of the aquifer. In addition, slug or bailer
tests and single-well pump tests are simple, rapid, and cost-effec-
tive methods for determining average of the vertical distribution
of aquifer horizontal properties.
Briefly, the aquifer-test method of assessing aquifer vulner-
ability involves determination of hydraulic conductivity* from slug
or bailer tests, single-well pumping tests, or multi-well pumping
tests, calculation of CV of log hydraulic conductivity, and inter-
pretation of cumulative probability plots of log hydraulic conduc-
ivity. The distribution of hydraulic conductivity tends to be log-
normal. Therefore, the logarithm of hydraulic conductivity data
should be used for calculation of statistics, and the geometric
mean of hydraulic conductivity should be used to best represent the
average of log-normally distributed values (Domenico and Schwartz,
* Hydraulic conductivity, the volume of fluid (usually water) that
will move through a medium in a unit of time, T, under a unit
hydraulic gradient through a unit area measured perpendicular to
the flow^ has the dimensions L/T because the L of volume divided
by the L of area cancels to L, a length. Consequently, hydrau-
lic conductivity, a volume, is often erroneously thought to be a
velocity.
604
-------
1990, p. 66-67; Bradbury and Rothschild, 1985). The geometric mean
and CV of the hydraulic conductivity distribution are then used to
make inferences about aquifer vulnerability. As a rule of thumb,
we suggest that at least 20 data points should be used to calculate
these statistics. Additional data points, however, will provide a
more representative estimate of the distribution of hydraulic con-
ductivity within an aquifer.
For hypersensitive aquifers (those probably dominated by con-
duit flow), the CV will be very high (>500%), with wells penetrat-
ing both highly transmissive fissures and conduits plus non-conduit
blocks with low permeability- [The probability of a well inter-
cepting a conduit is extremely small, but doing so would simply
increase the CV of the sample] . However, the geometric mean of the
hydraulic conductivity will be low. For moderately sensitive aqui-
fers (those probably dominated by diffuse flow), the mean hydraulic
conductivity will be moderate to high, reflecting the dissolutional
widening of openings in the limestone, but the CV will be much
lower, reflecting the more homogenous distribution of flow within
the aquifer (Smart et al. , 1991) . In both cases, the log-normal
probability plot of hydraulic conductivity may well be polymodal.
This hydraulic conductivity approach has some advantages over
the indirect analysis of specific conductance CV and hydrographs,
discussed above, because it allows direct determination of flow
characteristics of carbonate aquifers. The specific conductance
and the hydrograph are a response to the processing by the aquifer
of a signal that is not uniquely controlled by its flow character-
istics and is also dependent on the nature of recharge to and
storage in the aquifer. Fortunately, many of the most karstified
(and therefore most troublesome) carbonate aquifers have conduit
flow, concentrated recharge, and low storage, thus providing an
unambiguous signal. Other aquifers show a more damped signal, im-
plying diffuse flow, but this may result from an absence of concen-
trated recharge (for instance, where recharge is through a
permeable caprock or thick soil mantle not breached by sinkholes)
or high storage (for instance, in a deep, saturated zone of an
aquifer).
There may be a methodological problem with the use of hydra-
ulic conductivity for characterizing a carbonate aquifer. The
thickness of the open interval of a borehole tested may bias an
evaluation of its hydraulic conductivity- More specifically, long
intervals may tend to average and dampen its true extremely high
and low values. The measured high hydraulic conductivity of a well
is always less than the actual value for a fracture or other cavity
intercepted by it (Maureen Muldoon, Wisconsin Geological and
Natural History Survey, written communication, January 1992). The
possible relationship between the thickness of developed interval
in a borehole and the hydraulic conductivity in different types of
carbonate aquifers needs to be evaluated.
As appealing, objective, useful, and diagnostic as interpreta-
tion of, say, 30 hydraulic conductivity measurements may be, in
605
-------
actual practice obtaining them may be difficult, time consuming,
and infeasible in some areas. There is little doubt, however, that
evaluation of the CV of hydraulic conductivity from slug or bail
tests is superior in many ways (but not all ways) to evaluation of
the CV of specific conductance. The latter technique can be
cheaper but it takes longer and may have to be delayed because of
climatic seasonality.
Slug and bailer tests are reviewed by Fetter (1988, p. 196-
200) , Domenico and Schwartz (1990, p. 164-168), and Dawson and
Istok (1991). There are many limitations to applying such tests,
both theoretical and logistical. When an investigator uses the
standard techniques for interpretation of bailer or slug tests
(Hvorslev, Papadopulos, and many others) or for single-well pumping
tests, he must be familiar with and willing to accept the assump-
tions inherent in those analytical solutions. All of these methods
assume Darcian flow and assume also that the hydraulic conductivity
is homogeneous over the interval tested. Since these assumptions
are rarely, if ever, valid for karst aquifers, the resulting hydra-
ulic conductivities should not be used to calculate flow velocities
or times of travel or to calibrate computer models. The variation
in the hydraulic conductivity distribution, however, may be used to
draw conclusions about relative vulnerability of an aquifer.
Logistically, slug tests are probably the easiest aquifer
tests to perform. The ideal method of characterizing the variation
of hydraulic conductivity of a karst aquifer would be to drill
wells specifically for that purpose, then perform packer tests at
a set interval in them (Milanovic, 1981, p. 237-251; Domenico and
Schwartz, 1990, p. 163-164). This is not practical for most
studies and, as a result, slug tests are commonly performed in
monitoring wells and in existing domestic wells where the pump has
either not yet been installed, or it has been pulled in order to
perform the tests.
If an investigator is using domestic wells as points to
measure hydraulic conductivity, single-well pumping tests (those
without an observation well) can not be used to reliably determine
hydraulic conductivity, but they do provide an advantage over slug
tests in that the pump does not have to be pulled. The drawdown
and recovery data (easily recorded with a downhole pressure trans-
ducer) from a single-well pump test can, however, be used to calcu-
late residual drawdown and transmissivity (Domenico and Schwartz,
1990, p. 162-163; Kruseman and de Ridder, 1990, p. 193-197; Dawson
and Istok, 1991, p. 59-61). A major problem, however, with any
attempt to reliably determine transmissivity or even hydraulic
conductivity from any test but a packer test, is that in a karst
aquifer it is often difficult, if not impossible, to do so from a
well that partially penetrates an aquifer in which the thickness is
unknown or (in some cases) indeterminable.
The hydraulic conductivity measured by a slug test, a single-
well pump test, or a multi-well pump test is the average hydraulic
conductivity measured over the entire test interval. Using this
606
-------
"average" hydraulic conductivity to predict flow velocities or
travel times will result in significant errors, however, because
karst aquifers are characterized by highly heterogeneous hydraulic
conductivity distribution (i.e., low-permeability blocks, and high-
conductivity conduits or dissolution zones) and most of the flow is
through the high-permeability zones.
Hydraulic conductivity can also be measured using multi-well
pump tests. The use of these tests is limited to those projects
with budgets that are sufficient to install a pumping well and
several observation wells. The standard techniques for the inter-
pretation of data from any type of pump test (Theis, Hantush-Jacob,
etc.) require the erroneous assumption of isotropic conditions;
other analysis techniques, however, can be used to calculate both
the horizontal anisotropy ratio (Papadopulos, 1965), as well as the
horizontal to vertical anisotropy ratio (Weeks, 1969). The assump-
tions of the different pump test interpretation techniques are
clearly discussed by Kruseman and de Ridder (1990) and Dawson and
Istok (1991) . They also discuss techniques developed by Streltsova
and Boulton and by Streltsova that may be more suitable for karst
aquifers. Thiery et al. (1983) have also tried to develop a reli-
able technique for interpreting pump-test data from karst aquifers,
but they too have been unsuccessful. The discharge of a well, par-
ticularly one in a karst aquifer, is the integral of the hydraulic
character (hydraulic conductivity, dispersivity, storage, and
tortuosity) of all the different dissolutionally-enlarged fissure-
systems that feed it (Quinlan, 1992). The hydraulic conductivity
calculated from pump test data still represents a value averaged
over the interval tested and should be not used in calculations of
Darcian or non-Darcian flow.
In spite of the potential errors in hydraulic conductivities
calculated from slug and pump tests in karst aquifers, and their
possible misuse, we do believe in their value for evaluation of
aquifer hydraulic properties and probable vulnerability. The CV of
specific conductance represents a response to the combined effects
of recharge, storage, and flow, and gives a direct, integrated
measure of the vulnerability of a karst aquifer to pollutants. In
addition, the data are relatively inexpensive to collect. However,
it may take longer to collect or be delayed because of climatic
seasonality. Both types of information are complementary and
needed. Neither type is intrinsically superior to the other.
A CAUTIONARY TALE
A detailed study, using geophysical logs, borehole video logs,
lithologic core analyses, and unspecified aquifer tests, was made
of the dissolution porosity and permeability at a Florida site
(Robinson and Hutchinson, 1991). [The thoroughness of their well-
designed and carefully executed study was much more than we have
recommended here.] Tracer tests were performed by injecting a
fluorescent dye into the open-hole interval of a well and measuring
the movement induced by pumping 1000 gallons per minute at a well
607
-------
200 feet away. Dye concentration at the pumped well was measured
continuously with a filter fluorometer. Based on standard assump-
tions about flow in porous media and a porosity of 20%, the theo-
retical arrival time for the dye should have been approximately 40
days, with a 4-day persistence time for the dye. Actual break-
through time was 5 hours. The peak was at 22 hours and the persis-
tence time was 28 days. A second breakthrough of dye occurred at
36 days after injection and its persistence time was 8 days. This
second breakthrough is not attributable to response to a storm
(James Robinson, U.S. Geological Survey, Tampa, Fla., oral communi-
cation, December 1991). The bimodal distribution of tracer arrival
time was interpreted to indicate dual porosity in the aquifer, but
we suggest that bifurcation of a flow route, followed by rejoining
of the trunk route, is a more plausible explanation (Tsang et al.,
1991; Ford and Williams, 1989, p. 226-228).
This Florida study illustrates the heterogeneity of the aqui-
fer and the impossibility of using carefully (and expensively)
acquired aquifer parameters to make correct predictions about
travel times in karst aquifers.* Most important, it shows that the
assumptions of uniform distribution of porosity and attendant dif-
fuse flow, as has been assumed and proposed in current strategies
of wellhead protection for the Floridan aquifer (DeHan, 1988), are
probably not valid. Again, the conceptual model inadequately
corresponds to reality.
CAVEATS
The summary of this paper lists six important caveats. They
require no additional explanation. Please go back and reread them.
DISCUSSION
Experience has shown that, in the midwest and south, there is
sometimes a rough correlation between spring morphology and flow
type. Springs dominated by conduit flow tend to have high-energy
discharge from a large alcove that is eroding headward, commonly
have dry high-level orifices that function only in response to
storms, and may have a deeply eroded channel. In contrast, springs
dominated by diffuse flow commonly have low-energy discharge from
a very small alcove, shallow channels, and fontaphilic (spring-
loving) vegetation. These correlations are guides, for first ap-
* The discrepancy between actual tracer velocities and those based
on predictions from analysis of cores and logs is partly a
consequence of extrapolation from a borehole scale of study to a
field scale, as discussed for karst aquifers by Kiraly (1975;
summarized by Ford and Williams, 1989, p. 134, Fig. 5.4) and
Smart et al. (1991) , and as recognized in other aquifers by
Domenico and Schwartz (1990, p. 84-87, 371-84) and many others.
608
-------
proximations during the first few days of field work.
A review of post-1960 classifications of karst aquifers was
made by Ford and Williams (1989, p. 166-170). No attempt has been
made here to repeat it. The aquifer classification we have pro-
posed is consistent with and complementary to those of Mangin
(1984) and Ford and Williams (1989, p. 169). It is an update and
revision of one first proposed by Smart and Hobbs (1986).
Figure 1 depicts a generalized conceptual hydrogeologic model
of karst aquifers. As such, it incorporates all the essential
features of their physical system. It is not, however, a site-
specific, accurate representation of the conditions beneath and
adjacent to a particular site. Although the degree of detail
necessary for various sites differs, presentation of a site-
specific conceptual model should be accompanied by maps and cross
sections. Conceptualization is a means of achieving a graphic
idealization of the actual geologic/hydrologic conditions, and it
deliberately ignores minor features that are not important to the
overall picture.
The arbitrary but convenient dividing points in the continuum
between conduit flow and diffuse flow are 20% and 80%. The inter-
mediate members of the continuum, having 20 to 80% conduit flow and
80 to 20% diffuse flow, have been known as mixed flow aquifers
(Atkinson and Smart, 1981; Quinlan and Ewers, 1985) . This term,
too, should be retired. [Strictly speaking, all aquifers that are
not 100% one flow-type or the other are mixed flow, but it has been
convenient to draw intermediate boundaries at 20% and 80%.]
Diffuse flow can be better understood by making an analogy
with the flow of water through a stack of bricks; flow is chiefly
between the bricks rather than through them. We recognize, of
course, that water also moves within the rock matrix of a karst
aquifer, just as it would within the bricks, but we consider the
intergranular Darcian component of the flow to be unimportant to
this discussion. The importance of such dual porosity, especially
during tracer tests, has been analyzed in a series of papers by
Maloszewski and Zuber, the most recent of which is Maloszewski and
Zuber (1990). A recent mathematical analysis of dual porosity has
been made by Douglas and Arbogast (1990) and its role in water
quality distribution within fractures and intergranular pore space
has been briefly described by Edmunds (1981).
The complex relationships between conduit flow and diffuse
flow, particularly as related to the role of dual porosity during
recharge events, have been mathematically analyzed by Onder (1986).
He showed that base-flow recession through a conduit may be related
to aquifer properties, as might be expected.
A substantial amount of the water in an aquifer dominated by
conduit flow moves as diffuse flow, as explained by Atkinson
(1977). In the British aquifer he analyzed, 60 to 80% of the dis-
charge is via conduits; active storage in the diffuse-flow part of
609
-------
the aquifer that feeds them is 30 times greater than in the con-
duits. Approximately 50% of the recharge is from sinking streams
and sinkholes; the remaining 50% is via slow percolation through
the soil.
There is a perception among some hydrologists and engineers
that, because conduits comprise such a small percentage of an
aquifer and because most of the rock surrounding them may be
characterized by diffuse flow, overall, there is no significant
problem of aquifer vulnerability. Wrong! This is like saying that
the hole blown into the bottom of a battleship by a torpedo is only
10 feet in diameter but, because the rest of the steel is imperme-
able, there is no problem of sinking. Wrong again!
There are two further complications. First, almost all of the
discharge from all aquifers dominated by conduit flow occurs at a
point or group of points along a line. A small component of the
discharge is as diffuse flow, as described by Atkinson (1977).
Second, most karst aquifers dominated by diffuse flow also dis-
charge at a point or group of points, but some may have almost all
of their discharge over a strip or a broad area, as discussed by
Quinlan (1989) as an exception to his Implicit Assumption #1. Our
experience suggests that point discharge is predominant in most
carbonate terranes. Accordingly, no consideration to such non-
point discharge has been given in this paper.
In lieu of daily or twice- or thrice-weekly measurements of
specific conductivity, it is likely to be far more cost-efficient,
and could be more reliable, to continuously record it with a trans-
ducer and a data-logger. Also, having a continuous record of con-
ductivity facilitates making an analysis that enables optimization
of sampling frequency, as discussed below, and if such would be
useful for a given project. The utility of the conductivity data
could be inexpensively enhanced by using the same data-logger to
also continuously record temperature with a platinum-resistance
thermometer and turbidity. Interpretation of temperature data,
although it can be quite subtle, can give significant useful
insights into aquifer hydraulics (Meiman, et al. , 1988; Davies,
1991).
Continuous recording of water-quality parameters also elimi-
nates the possibility of aliasing, a sampling error that can intro-
duce significant bias into estimates of the magnitude and frequency
variation of a parameter and reliable characterization of it — all
as a result of sampling with insufficient frequency. Aliasing is
a phenomenon in which a high-frequency component of a signal takes
on the identity of one with a lower frequency. [The familiar illu-
sion of stagecoach wheels in a movie first going forward, then
apparently backward, then forward again is an example of aliasing.]
Signal processing theory, information theory, spectral analysis
theory, and time-series analysis theory all include a theorem which
states that a continuous, periodic signal, with frequency compo-
nents in the range / = 0 to f = fmax per unit of time, can be recon-
structed from a series of equally spaced samples if the sampling
610
-------
frequency exceeds 2/max samples per unit of time. /^ = 1/2At,
where At = the sampling interval and fmax = the Nyquist number, the
fastest frequency detectable. The slowest frequency detectable is
zero; it never repeats. Trend lines have zero frequency and infi-
nite period. These concepts and many others relevant to interpre-
tation of water-quality data are discussed by Meade and Dillon
(1991, p. 16-19), Karl (1989, p. 4-8), Bandat and Piersol (1986, p.
335-339), Pierce (1980, p. 35-39), Bras and Iturbe (1984, p. 172-
177), Davis (1986, p. 257-258, 349), Gardner (1988, p. 49-52),
Chatfield (1975, p. 155-158), Gottman (1981, p. 15-16). Aliasing
can be prevented by continuously recording a parameter or by taking
samples extremely frequently, perhaps every 30 seconds. A pro-
cedure for using conductivity data to prevent aliasing of water
quality data and for determination of an appropriate aliasing-free
sampling protocol for using water-quality data to characterize an
aquifer or its spring is given in the accompanying appendix. The
procedure is similar to that already discussed as used to analyze
exponential decay of the recession limb of a hydrograph. We
believe the procedure is also applicable to the design and inter-
pretation of tracer tests.
Aliasing of water-quality data can be a serious problem be-
cause it can inadvertently and seriously bias their interpretation,
but it is not mentioned by Gilbert (1987) or in any other water-
quality publications we have reviewed. Only Davis (1986), of the
several geostatistical books known to us, discuss aliasing of
spectral data, but there are several signal-processing texts such
as that by Robinson (1980) which recognize aliasing of seismic
data.
It is highly likely that most water-quality studies and moni-
toring of springs and wells characterized by conduit flow are
flawed by aliasing unless they have been sampled frequently (per-
haps daily) and/or as related to storm events. Unless they have,
such effects are incapable of accurately characterizing the dynamic
changes in water quality except by improbably good luck. Most
assuredly, this is a fatal defect of the monitoring studies at
Mammoth Cave National Park described by Meiman (1991); as designed,
such studies are intrinsically incapable of achieving their pur-
pose. Probable aliasing of water quality parameters is also a
deficiency of the recent synthesis of data on the largest and most
famous spring in France, Vaucluse (Puig, 1990).
Time-series analysis and spectral analysis of data from well-
designed studies of regularly sampled and continuously sampled
spring waters was first applied in various papers by Mangin and by
Bakalowicz, as partially summarized by Ford and Williams (1989, p.
210-214). Application of these data-analytical techniques to other
studies, like those described by Mazor et al. (1990) and those
cited in the following paragraph, will lead to significant improve-
ments in understanding of the processes operating in karst aqui-
fers.
The complexity of flow within an aquifer dominated by conduit
611
-------
flow, plus interpretation of the lack of coincidence among the
pulses of stage, conductivity, and temperature, are discussed by
Meiman et al. (1988) and, to a lesser extent, by Ford and Williams
(1989, p. 204-214). Interpretation of coincidence among the pulses
associated with a different type of conduit flow is discussed by
Idstein and Ewers (1991).
Further discussion of the interpretation of hydrographs of
springs and streams is given by Bonacci (1987, p. 75-81), Ford and
Williams (1989, p. 193-203), Domenico and Schwartz (1990, p. 15-
17), and Hall (1968). Spectral analysis and time-series analysis
of spring discharge have been used by Bakalowicz and Mangin (1980),
and Mangin (1982, 1984) as a basis for aquifer classification.
Chemograph separation is discussed by Ford and Williams (1989, p.
204-214). Detailed summary of these topics is beyond the scope of
this paper, but some are discussed in the accompanying appendix.
Spills in karst terranes frequently occur on surfaces with a
soil profile 5 to perhaps 50 feet thick and are sometimes confined
to the soil. With luck, the entire spill is confined to the soil.
Many workers would assume that Darcian flow equations adequately
describe the movement of liquids through the soil. Unfortunately,.
non-Darcian macropore flow (Quinlan and Aley, 1987; Everts et al.,
1989; Wells and Krothe, 1989) commonly occurs in the soil; contami-
nants can move rapidly through the soil to, and perhaps into, the
bedrock. Fluid movement through soil with macropores can not be
analyzed as though it were through a granular medium. Indeed,
Watson and Luxmore (1986) found that macropores and mesopores in an
Appalachian karst area, which together constitute only 0.3% of the
soil volume, account for 96% of all infiltration. This topic is
discussed in great detail by Wilson et al. (1991). Further, the
pollutant may be perched at the soil-bedrock contact and may move
hundreds of feet along it before intercepting a dissolutionally
enlarged joint which allows access into bedrock of the karst
aquifer below.
One of the more challenging frontiers in contemporary karst
hydrogeology is the epikarst (also known as the subcutaneous zone).
This is the dissolutionally well-developed zone immediately beneath
the soil and above the relatively dry zone above the phreatic zone.
In a sense, it is much like a perched aquifer, but usually there is
no discrete perching bed. As discussed by Ford and Williams (1989,
p. 120-121 and in many other places throughout their book),
Williams (1983, 1985), Bonacci (1987, p. 28-35), Friedrich and
Smart (1981), Smart and Friedrich (1986), Field (1990), Quinlan and
Ray (1991), and McCann and Krothe (1991), there is significant
water-storage capacity and sufficient interconnection to diffuse
the movement of pollutants and tracer dyes widely. Drainage from
the epikarst is not uniform but is down preferential pathways to
the subsurface. Residence time for pollutants or tracers in the
same epikarst can range from hours to tens of years (Quinlan and
Ray, 1991). Much is being learned about flow and storage in epi-
karst, both in the U.S. and in Europe (Smart and Friedrich, 1986;
Chevalier, 1988), but most of it has not yet been published. We
612
-------
believe that epikarst hydrology is the most enigmatic in all of
karst. Assume that an epikarst exists beneath any soil mantle —
until you prove otherwise. Hydrologists, engineers, and regulators
should be prepared to encounter epikarst. They should be prepared
also to be confused and frustrated by it.
Some experienced hydrologists believe that slug and bailer
tests measure little more than skin-effects and a small, local
volume adjacent to the well bore. The results of such tests can be
non-representative and misleading. They may be in vogue chiefly
because they are quick, inexpensive, and readily create a favorable
response in some regulators. All this is more true than not. But,
as such tests are orders of magnitude superior to laboratory deter-
minations of permeability of a core sample and more representative
of aquifer hydraulic properties in a karst or any other aquifer, so
also, pump tests are significantly superior to slug and bail tests.
Nevertheless, tens to perhaps hundreds of pump tests are necessary
to accurately characterize an aquifer. All this is to be expected.
As pointed out by Kiraly (1975) and quoted by Ford and Will-
iams (1989, p. 134, Fig. 5.4), karst aquifers become both more het-
erogeneous and more anisotropic with passing time. Their average
total porosity is directly proportional to the reference volume of
rock considered and may range over three orders of magnitude —
depending upon the scale at which investigation is performed. Ac-
cordingly, it could be convincingly argued that a well-designed,
carefully interpreted tracer test, because it may sample a far
larger volume of an aquifer, can be superior to a pump test for
characterizing some properties of an aquifer. The two types of
test are complementary, however. Neither is intrinsically superior
to the other.
Many topics have been discussed in this paper. Its most
important points are that:
• Clear, accurate definitions of karst and karst aquifer —
as well as practical techniques for recognizing them — are
needed. We have fulfilled this need.
• Regulations concerning karst should focus on protecting
waters of karst aquifers. Sinkhole collapse is a relative-
ly minor problem.
• Karst aquifers should be classified according to the three
highly variable major controls on their nature: recharge,
storage, and flow, as shown on the cube depicted in Figure
1, and grouped into three classes of vulnerability: hyper-
sensitive, very sensitive, and moderately sensitive.
• Karst aquifers should no longer be classified as conduit-
flow aquifers or diffuse-flow aquifers. These terms should
be retained, however, as useful descriptors of flow type.
• Systematic measurement of the specific conductance of
613
-------
springs during base flow and during and after flood flow,
with evaluation of the coefficient of variation (CV) of
such data, is a valuable and probably reliable predictor of
not only the net effect of an aquifer's recharge, storage,
and flow characteristics but also its relative vulnerabili-
ty to the adverse effects of contaminants and mismanage-
ment.
• A major deficiency of most attempts to monitor water qual-
ity of springs and wells in karst terranes is aliasing, the
failure to detect and recognize high-frequency fluctuations
(which can also be high-magnitude fluctuations) in water
quality because the water is not sampled often enough. The
highest frequency signal that can be resolved has a wave
length that is twice the distance (time) between successive
regular observations.
• Two objective, sensitive, flexible procedures for determin-
ing a cost-efficient, aliasing-free sampling frequency for
reliable characterization of springs, aquifers, and
conduit-fed wells now exist. They are described in the
accompanying appendix. We believe these procedures are
applicable to: 1) Reliable assessment of water quality, 2)
Sampling for and interpretation of tracer tests in which
dye-concentration is not monitored continuously, and 3)
Design, operation, and interpretation of pulse tests (Ford
and Williams, 1989, p. 226-228).
• The empirically established values for the CV of conduc-
tivity, recommended for use in separating the three types
of vulnerability of karst aquifers, are < 5%, 5 to 10%, and
> 10%. The reliability of these values is being reviewed
by the authors, but we do not expect any substantial
changes in them.
• Imperfect as their results in karst aquifers usually are,
slug, bail, and pump tests should be performed in an effort
to characterize the aquifer. An individual test has little
significance, but evaluation of the geometric mean and the
CV of a representative sample (at least tens) is a useful
technique for approximating the hydraulics of karst aqui-
fers and their differences.
Finally, a few words should be said about how this paper was
written. Although a manuscript was completed in time for the
meeting, it was constructive dialogue with participants there that
forced all of us to rethink our assumptions, interpretations, and
conclusions, and to reformulate our ideas. The three authors of
the original manuscript believe that this final version has been
significantly enhanced by the input and active participation of the
others and the reviewers we have acknowledged. We know other meet-
ing participants have had similar experiences in completing their
manuscript. This is a good, satisfying way to do better science
and is one reason why we believe that conference proceedings should
614
-------
be distributed after a meeting, not before or at it.
SOLICITATION
Readers of this paper are invited to correspond with the
authors and suggest improvements in the definition of karst aquifer
and ways to classify such aquifers meaningfully, usefully, and
practically.
ACKNOWLEDGEMENTS
Tom Aley, Richard C. Benson, Gareth J. Davies, Gary A. Davis,
Malcolm S. Field, Michael R. McCann, Maureen Muldoon, Joseph A.
Ray, Paul A. Rubin, and Joe P. Sparks made numerous useful sugges-
tions. We think that the content and clarity of expression of the
ideas discussed herein has been significantly enhanced by their
critical review. William B. White graciously clarified mathemati-
cal interpretation of discharge and conductivity data during hydro-
graph recession. We are indebted to them. The typesetting talents
of Marcia Williams in lettering Figure 1 are greatly appreciated.
We thank them all.
REFERENCES CITED
Adams, M.C., and Davis, J. 1991. Kinetics of fluorescein decay
and its application as a geothermal tracer. Geothermics.
20(1/2):53-66.
Aley, T. 1975. A predictive hydrologic model for evaluating the
effects of land use and management on the quantity and quality
of water from Ozark springs. Ozark Underground Laboratory,
Protem, Mo. [Republished in Missouri Speleology, 1978. 18:1-
185. ]
Aley, T. 1977. A model for relating land use and groundwater
quality in southern Missouri, in Dilamarter, R.R., and
Csallany, S.C., eds. Hydrologic Problems in Karst Areas.
Western Kentucky University, Bowling Green, p. 323-332.
Aley, T. 1986. Complex radial flow of ground water in flat-lying,
residuum-mantled limestone in the Arkansas Ozarks. Environ-
mental Problems in Karst Terranes and Their Solutions Confer-
ence (1st, Bowling Green, Ky.), Proceedings. National Water
Well Association, Dublin, Ohio. p. 159-170.
Atkinson, T.C. 1977. Diffuse flow and conduit flow in limestone
terrain in the Mendip Hills, Somerset (Great Britain).
Journal of Hydrology. 35:93-110
Atkinson, T.C., and Smart, P.L. 1981. Artificial tracers in karst
hydrogeology, in A Survey of British Hydrogeology. The Royal
615
-------
Society, London, p. 173-190.
Bagchi, A. 1990. Design, Construction and Monitoring of Sanitary
Landfill. Wiley-Interscience, New York. 284 p.
Bailey, P-, and Ward, W.D., eds. 1990. Understanding Ground-Water
Contamination: An Orientation Manual. Executive Enterprises
Publications, New York. 223 p.
Bakalowicz, M. 1976. Geochemie des eaux karstiques. Une methode
d1etude de 1"organization des ecoulements souterrains. Col-
loque d'Hydrologie en Pays Calcaire (2nd, Besancon-Neuchatel) ,
Proceedings. Annales Scientifique de 1'Universite de Besan-
con. Series 3, 25:49-59.
Bakalowicz, M. 1979. Contribution de la geochemie des eaux a la
connaissance de 1'aquifere karstique et de la karstification.
Doctoral thesis, L'Universite Pierre et Marie Curie - Paris 6.
269 p.
Bakalowicz, M., and Mangin, A. 1980. L'aquifere karstique. Sa
definition, ses characteristiques et son identification, in
Hydrogeologie: Interactions entre 1'eau souterrain et son
milieu. Societe Geologique de France, Memoire hors-serie
no. 11. p. 71-79.
Bartlett, A.A. 1992. The Feynman Effect and the boon docs.
Physics Today. 45(1):67,69.
Beck, B.F., Quinlan, J.F., and Ewers, R.O. 1987. Presentation of
the E.B. Burwell, Jr., Memorial Award to James F. Quinlan and
Ralph O. Ewers. Geological Society of America, Bulletin.
97:141-143.
Bendat, J.S., and Piersol, A.G. 1986. Random Data: Analysis and
Measurement Procedures. 2nd ed. Wiley-Interscience, New
York. 566 p.
Benson, R.C., and La Fountain, L.J. 1984. Evaluation of subsi-
dence or collapse potential due to subsurface cavities.
Multidisciplinary Conference on Sinkholes Conference (1st,
Orlando, Fla.), Proceedings. Balkema, Rotterdam, The Nether-
lands, p. 201-215.
Black, H.C. 1979. Black's Law Dictionary. 5th ed. West, St.
Paul, Minn. 1511 p.
Blavoux, B., and Mudry, J. 1988. Importance of the sampling
rhythm in the hydrochemical study and hydrocinematical knowl-
edge of karstic aquifers. Karst Hydrology and Karst Environ-
mental Protection (Guilin, China), Proceedings. (Internation-
al Association of Hydrogeologists, 21st Congress). Geological
Publishing House, Beijing. Memoir 21. 2:864-869.
616
-------
Bonacci, O. 1987. Karst Hydrology, With Special Reference to the
Dinaric Karst. Springer, Berlin. 184 p.
Bosak, P-, Ford, D., Glazek, J., and Horafiek, I., eds. 1989.
Paleokarst: A Systematic and Regional Review. Elsevier,
Amsterdam. 725 p.
Bradbury, K.R., and Rothschild, E.R. 1985. A computerized tech-
nique for estimating the hydraulic conductivity of aquifers
from specific capacity data. Ground Water. 23(2):240-246.
Bras, R.L., and Rodriguez-Iturbe, I. 1985. Random Functions in
Hydrology. Addison-Wesley, Reading, Mass. 559 p.
Brooks, K.N., Ffolliott, P.F., Gregersen, H.M., and Thames, J.L.
1991. Hydrology and the Management of Watersheds. Iowa State
University, Ames. 392 p.
Burbey, T.J., and Prudic, D.E. 1991. Conceptual evaluation of
regional ground-water flow in the carbonate-province of the
Great Basin, Nevada, Utah, and adjacent states. U.S. Geologi-
cal Survey, Professional Paper 1409-D. 84 p.
Chatfield, C. 1984. The Analysis of Time Series: An Introduc-
tion. 3rd ed. Chapman and Hall, London. 286 p.
Chevalier, J. 1988. Hydrodynamique de la zone non-saturee d'un
aquifer karstique: etude experimentale. Doctoral thesis,
Universite des Sciences et Techniques du Languedoc. 212 p.
Christiansen, T.H., Cossu, R. , and Stegmann, R. 1989. Sanitary
Landfilling: Process, Technology, and Environmental Impact.
Academic, London. 592 p.
Committee on Ground Water Quality Protection. 1986. Ground Water
Quality Protection: State and Local Strategies. National
Academy Press, Washington, D.C. 309 p.
Crawford, N.C. 1984. Sinkhole flooding associated with urban de-
velopment upon karst terrain: Bowling Green, Kentucky. Multi-
disciplinary Conference on Sinkholes (1st, Orlando, Fla.),
Proceedings. Balkema, Rotterdam, The Netherlands. p. 283-
292.
Crawford, N.C. 1991. Emergency response procedures for spills of
hazardous waste liquids upon karst terranes. Hydrogeology,
Ecology, Monitoring, and Management of Ground Water in Karst
Terranes Conference (3rd, Nashville, Tenn.), Proceedings.
National Ground Water Association, Dublin, Ohio. [in this
volume]
Dalgleish, J., and Alexander, E.G., Jr. 1984. Sinkhole distribu-
tion in Winona County, Minnesota. Multidisciplinary Confer-
ence on Sinkholes (1st, Orlando, Fla.), Proceedings. Balkema,
617
-------
Rotterdam, The Netherlands, p. 79-85.
David, M. 1977. Geostatistical Ore Reserve Estimation. Elsevier,
Amsterdam. 364 p.
Davies, G.J. 1991. Water temperature variation in springs in the
Knox Group near Oak Ridge, Tennessee. Hydrogeology, Ecology..
Monitoring, and Management of Ground Water in Karst Terranes
Conference (3rd, Nashville, Tenn.), Proceedings. National
Ground Water Association, Dublin, Ohio. [in this volume]
Davies, W.E., Simpson, J.H., Ohlmacher, G.C., Kirk, W.S., and
Newton, E.G. 1984. Engineering Aspects of Karst. National
Atlas of the U.S., U.S. Geological Survey Map US 05572. Scale
1:7.5 million. 1 sheet in color, with text by W.E. Davies.
[Supercedes Open File Map 76-623, 1976, which is illegible]
Davis, D.S. 1955. Nomography and Empirical Equations. Reinhold,
New York. 236 p.
Davis, G.A., and Quinlan, J.F. 1991. Legal Tools for the Protec-
tion of Groundwater in Karst Terranes. Hydrogeology, Ecology,.
Monitoring, and Management of Ground Water in Karst Terranes
Conference (3rd, Nashville, Tenn.), Proceedings. National
Ground Water Association, Dublin, Ohio. [in this volume]
Davis, J.C. 1986. Statistics and Data Analysis in Geology. 2nd
ed. Wiley, New York. 646 p.
Dawson, K.J., and Istok, J.D. 1991. Aquifer Testing: Design and
Analysis of Pumping and Slug Tests. Lewis, Chelsea, Mich.
344 p.
DeHan, R.S. 1988. New approach to protection of sensitive aqui-
fers in Florida. Hydrological Science and Technology. 2(2):
51-59.
Devinny, J.S., Everett, L.G., Lu, J.C.S., and Stollar, R.L. 1990.
Subsurface Migration of Hazardous Wastes. Van Nostrand Rein-
hold, New York. 387 p.
Domenico, P.A., and Schwartz, F. 1990. Physical and Chemical
Hydrogeology. Wiley, New York. 824 p.
Douglas, J., Jr., and Arbogast, T. 1990. Dual porosity models for
flow in naturally fractured reservoirs, in Cushman, J.H., ed.
Dynamics of Fluids in Hierarchical Porous Media. Academic,
London. p. 177-221.
Downey, J.S. 1984. Geohydrology of the Madison and associated
aquifers in parts of Montana, North Dakota, South Dakota, and
Wyoming. U.S. Geological Survey, Professional Paper 1273-G.
47 p.
618
-------
Dreybrodt, W. 1988. Processes in Karst Systems. Springer,
Berlin. 288 p.
Driscoll, F. 1986. Groundwater and Wells. 2nd ed. Johnson Divi-
sion, St. Paul, Minn. 1089 p.
Edmunds, W.M. 1981. Hydrogeochemical investigations, in Lloyd,
J.W., ed. Case-studies in Groundwater Resources Evaluation.
Clarendon, Oxford, p. 87-112.
Edwards, A.J., and Smart, P.L. 1989. Waste disposal on karstified
Carboniferous Limestone aquifers of England and Wales.
Engineering and Environmental Aspects of Sinkholes and Karst
Conference (3rd, St. Petersburg, Fla.), Proceedings. Balkema,
Rotterdam, The Netherlands. p. 165-182.
Erdelyi, M., and Galfi, J. 1988. Surface and Subsurface Mapping
in Hydrogeology- Wiley, New York. 384 p.
Everts, C.J., Kanwar, R.S., Alexander, E.G., Jr., and Alexander,
S.C. 1989. Comparison of tracer mobilities under laboratory
and field conditions. Journal of Environmental Quality.
18(4):491-498.
Ewers, R.D., 1982. Cavern development in the dimensions of length
and breadth. Ph.D. thesis, Geography. McMaster University.
398 p.
Fetter, C.W. 1988. Applied Hydrogeology. 2nd ed. Merrill,
Columbus, Ohio. 592 p.
Field, M.S. 1988. The vulnerability of karst aquifers to chemical
contamination. Hydrological Science and Technology. 4(1-
4):130-142.
Field, M.S. 1990. Transport of chemical contaminants in karst
terranes: Outline and summary, in Selected Papers on Hydroge-
ology from the 28th International Geological Congress, Wash-
ington, D.C., 1989. International Association of Hydrogeol-
ogists, Selected Papers. Heise, Hannover. 1:17-27.
Fookes, P.G., and Vaughan, P.R. 1986. A Handbook of Engineering
Geomorphology. Surry University Press, London. 343 p.
Ford, D.C., and Williams, P.W. 1989. Karst Geomorphology and
Hydrology. Unwin Hyman, Boston. 601 p.
Franklin, B. 1781. A letter to a Royal Academy, in Japikse, C.,
ed. 1990. Fart Proudly. Enthea, Columbus, Ohio. p. 13-17.
Freeze, R.A., and Cherry, J.A. 1979. Groundwater. Prentice-Hall,
Englewood Cliffs, N.J. 604 p.
Friedrich, H., and Smart, P.L. 1981. Dye tracer studies in the
619
-------
unsaturated-zone recharge of the Carboniferous Limestone
aquifer of the Mendip Hills, England. International Congress
of Speleology (8th, Bowling Green, Ky.), Proceedings. 1:283-
286.
Gams, I. 1991. The origin of the term karst in the transition of
karst (kras) from deforestation to forestation. International
Conference on Environmental Changes in Karst Areas (Padova,
Italy), Proceedings, p. 1-8.
Gardner, W.A. 1988. Statistical Spectral Analysis: A nonprobabi-
listic Theory- Prentice-Hall, Englewood Cliffs, N.J. 566 p.
Gilbert, R.O. 1987. Statistical Methods for Environmental Pollu-
tion Monitoring. Van Nostrand Reinhold, New York. 320 p.
Goldman, B.A., Hulme, J.A., and Johnson, C. 1986. Hazardous Waste
Site Management: Reducing the Risk. Island Press, Washington,
D.C. 314 p.
Gottman, J.M. 1981. Time Series Analysis: A Comprehensive
Introduction for Social Scientists. Cambridge University
Press, Cambridge. 400 p.
Gupta, R.S. 1989. Hydrology and Hydraulic Systems. Prentice-
Hall, Englewood Cliffs, N.J. 739 p.
Hall, F.R. 1968. Baseflow recessions — A review. Water Resour-
ces Research. 4:978-983.
Hamburg, M. 1977. Statistical Analysis for Decision Making. 2nd
ed. Harcourt Brace Jovanovich, New York. 801 p.
Hannah, E.D., Pride, T.E., Ogden, A.E., and Taylor, R. 1989.
Assessing groundwater flow paths from pollution sources in the
karst of Putnam County, Tennessee. Engineering and Environ-
mental Aspects of Sinkholes and Karst Conference (3rd, St.
Petersburg, Fla.), Proceedings. p. 183-188.
Hansen, W.R. 1991. Suggestions to Authors of Reports of the
United States Geological Survey. 7th ed. Government Printing
Office, Washington, D.C. 289 p.
Hem, J.D. 1985. Study and interpretation of the chemical charac-
teristics of natural water. 3rd ed. U.S. Geological Survey,
Water Supply Paper 2254. 264 p.
Henley, S. 1981. Nonparametric Geostatistics. Applied Science,
London. 145 p.
Henry, J.G., and Heinke, G.W. 1989. Environmental Science and
Engineering. Prentice-Hall, Englewood Cliffs, N.J. 728 p.
Henry, M.F., ed. 1989. Hazardous Materials Response Handbook.
620
-------
National Fire Protection Association, Quincy, Mass. 561 p.
Hess, A.E., and Paillet, F.L. 1990. Application of the thermal-
probe flowmeter in the hydraulic characterization of fractured
rocks, in Paillet, F.L., and Saunders, W.R., eds. Geophysical
Investigations for Geotechnical Investigations. ASTM Special
Technical Publication 1101. American Society for Testing and
Materials (ASTM), Philadelphia. p. 99-112.
Hess, J.W., and White, W.B. 1988. Storm response of the karstic
carbonate aquifer of southcentral Kentucky. Journal of
Hydrology. 99:235-252.
Hickey, J.J. 1984. Field testing the hypothesis of Darcian flow
through a carbonate aquifer. Ground Water. 22:544-547.
Hoenstine, R.W., Lane, E., Spencer, S.M., and O'Carroll, T. 1987.
A landfill site in a karst environment, Madison County,
Florida — A case study. Multidisciplinary Conference on
Sinkholes and the Environmental Impacts of Karst (2nd,
Orlando, Fla.), Proceedings. Balkema, Rotterdam, The Nether-
lands, p. 253-258.
Idstein, P.J., and Ewers, R.O. 1991. Deducing karst aquifer re-
charge, storage, and transfer mechanisms through continuous
electronic monitoring. Hydrogeology, Ecology, Monitoring, and
Management of Ground Water in Karst Terranes Conference (3rd,
Nashville, Tenn.), Proceedings. National Ground Water Associ-
ation, Dublin, Ohio. [in this volume]
Johns, R.A., and Roberts, P.V. 1991. A solute transport model for
channelized flow in a fracture. Water Resources Research.
27(8):1797-1808.
Karl, J.H. 1989. An Introduction to Digital Signal Processing.
Academic Press, San Diego, Calif. 341 p.
Kerfoot, W.B., Beaulieu, G., and Kieley, L. 1991. Direct-reading
borehole flowmeter results in field applications. Ground
Water Management. (Proceedings of the Fifth National Outdoor
Action Conference on Aquifer Restoration, Ground Water Moni-
toring, and Geophysical Methods) 5:1073-1084.
Kiraly, L. 1975. Rapport sur 1'etat actuel des connaissances dans
le domaine des caracteres physiques des roches karstiques, in.
Burger, A., and Dubertret, L., eds. Hydrology of Karstic Ter-
rains. International Association of Hydrogeologists, Paris.
International Union of Geological Sciences, Ser. B, no. 3. p.
53-67.
Kiraly, L., and Muller, I. 1979. Heterogeneite de la permeabilite
et de I1 alimentation dans le karst: effect sur la variation du
chimisme des sources karstiques. Universite de Neuchatel,
Centre d'Hydrologie, Bulletin. 3:237-285.
621
-------
Krumbein, W.C. 1937. Sediments and exponential curves. Journal
of Geology. 45:577-601.
Krumbein, W.C., and Pettijohn, F.J. 1938. Manual of Sedimentary
Petrography. Appleton-Century-Crofts, New York. 549 p.
Kruseman, G.P., and de Ridder, N.A. 1990. Analysis and Evaluation
of Pumping Test Data. 2nd ed. International Institute for
Land Reclamation and Improvement (ILRI), Publication 47.
ILRI, Wageningen, The Netherlands. 377 p.
Lallemand-Barres, A., and Roux, J.C. 1989. Guide methodologique
d1establishment des perimeters des captages d'eau souterraine
destinee a la consommation humaine. Bureau de Recherches
Geologiques et Minieres (BRGM), Manuals et Methodes, no. 19.
Editions du BRGM, Orleans. 221 p.
Lepiller, M. , and Mondain, P.H. 1986. Les tracages articifiels en
hydrogeologie karstique: Mise en oevre et interpretation.
Hydrogeologie. 2(1):32-52.
Lundgren, L. 1986. Environmental Geology- Prentice-Hall, Engle-
wood Cliffs, N.J. 576 p.
Mackay, D.M. 1990. Characterization of the distribution and be-
havior of contaminants in the subsurface, in Water Sciences
and Technology Board. Ground Water and Soil Contamination
Remediation: Toward Compatible Science, Policy and Public
Perception. National Academy Press, Washington, B.C. p. 70-
90.
Mackey, C.O. 1936. Graphical Solutions. Wiley, New York. 153 p.
McCann, M.R., and Krothe, N.C. 1991. Development of a monitoring
program at a Superfund site in a karst terrane near Blooming-
ton, Indiana. Hydrogeology, Ecology, Monitoring, and Manage-
ment of Ground Water in Karst Terranes Conference (3rd,
Nashville, Tenn.), Proceedings. National Ground Water Associ-
ation, Dublin, Ohio. [in this volume]
Maloszewski, P., and Zuber, A. 1990. Mathematical modeling of
tracer behavior in short term experiments in fissured rocks.
Water Resources Research. 26(7):1517-1528.
Mangin, A. 1982. Determination du comportment hydrodynamique des
aquiferes karstiques a partir de 1'etude des informations
fournies par leurs exutoires, in Les discontinus en hydrogeo-
logie. Bureau de Recherches Geologiques et Minieres, Orleans.
Documents du BRGM, no. 45. p. 397-403.
Mangin, A. 1984. Ecoulement en milieu karstique. Annales des
Mines. 191(5-6):135-142.
Matheron, G. 1965. Les variables regionalisees et leur estima-
622
-------
tion. Masson, Paris. 305 p.
Mazor, E. 1991. Applied Chemical and Isotopic Groundwater Hydrol-
ogy. Wiley, New York. 274 p.
Mazor, E., Jaffe, F.C., and Vuataz, F.D. 1990. Time-data series
as a major diagnostic tool in groundwater hydrology, in
Parriaux, A., ed. Water Resources in Mountainous Regions.
International Association of Hydrogeologists, 22nd Congress
(Lausanne, Switzerland), Proceedings. Memoir, 22 (1):273-280.
Meade, M.L., and Dillon, C.R. 1991. Signals and Systems: Models
and Behavior. 2nd ed. Chapman & Hall, London. 167 p.
Meiman, J. 1991. The effects of recharge basin land-use practices
on water quality at Mammoth Cave National Park, Kentucky.
Hydrogeology, Ecology, Monitoring, and Management of Ground
Water in Karst Terranes Conference (3rd, Nashville, Tenn.),
Proceedings. National Ground Water Association, Dublin, Ohio.
[in this volume]
Meiman, J. , Ewers, R.O., and Quinlan, J.F. 1988. Investigation of
flood-pulse movement through a maturely karstified aquifer at
Mammoth Cave National Park: A new approach. Conference on
Environmental Problems in Karst Terranes (2nd, Nashville,
Tenn.), Proceedings. National Water Well Association, Dublin,
Ohio. p. 227-263.
Milanovic, P.T. 1981. Karst Hydrogeology- Water Resources Publi-
cations, Littleton, Colo. 434 p.
Miller, R.L., Bradford, W.L., and Peters, N.E. 1988. Specific
conductance: theoretical considerations and application to
analytical quality control. U.S. Geological Survey, Water
Supply Paper 2311. 16 p.
Moltyaner, G.L. 1990a. Field studies of dispersion: Radioactive
tracer experiments on scales of 20, 40, and 260 meters, in
Cushman, J.H., ed. Dynamics of Fluids in Hierarchical Porous
Media. Academic, London, p. 7-36.
Moltyaner, G.L., ed. 1990b. Transport and Mass Exchange Processes
in Sand and Gravel Aquifers: Field and Modelling Studies.
(Proceedings of the workshop held in Ottawa, Canada) AECL-
10308. Atomic Energy of Canada Limited [AECL], Chalk River,
Ottawa. 2 v- 402 + 464 p.
Molz, F.J., Morin, R.H. , Hess, A.E., Melville, J.G. , and Giiven, O.
1989. The impeller meter for measuring aquifer permeability
variations: Evaluation and comparison with other tests. Water
Resources Research. 25(7):1677-1683.
Montgomery, C. 1992. Environmental Geology. 3rd ed. W.C. Brown,
Dubuque, Iowa. 465 p.
623
-------
Moore, G.K. 1988a. Occurrence, frequency, and significance of
cavities in fractured-rock aquifers near Oak Ridge National
Laboratory, Tennessee. Conference on Environmental Problems
in Karst Terranes (2nd, Nashville, Tenn.), Proceedings.
National Water Well Association, Dublin, Ohio. p. 171-182.
Moore, G.K. 1988b. Uses of probability graphs for the charac-
terization of fractured rock aquifers near Oak Ridge, Tennes-
see. Hydrological Science and Technology. 4(1-4):57-65.
Murray, J.P., Rouse, J.V-, and Carpenter, A.B. 1981. Groundwater
contamination by sanitary landfill and domestic wastewater in
carbonate terrain: principal source diagnosis, chemical trans-
port characteristics, and design implications. Water Re-
search. 15:745-757.
Newson, M.D. 1971. A model of subterranean limestone erosion in
the British Isles based on hydrology. Institute of British
Geographers, Transactions. 54:51-70.
Nielsen, D.M., ed. 1991. Practical Handbook of Ground-Water Moni-
toring. Lewis, Chelsea, Mich. 717 p.
Nonveiller, E. 1989. Grouting Theory and Practice. Elsevier,
Amsterdam. 250 p.
Nyquist, J.E., Moore, G.K., Young, S.C., and Clapp, R.B. 1991.
Use of electromagnetic borehole flowmeter to delineate
groundwater producing fractures. Tennessee Water Resources
Symposium (4th, Knoxville), Extended Abstracts, p. 26-28.
O'Brien and Gere Engineers, Inc. 1988. Hazardous Waste Site
Remediation. Van Nostrand Reinhold, New York. 422 p.
O'Leary, P.R., Canter, L., and Robinson, W.D. 1986. Land dispos-
al, in Robinson, W.D., ed. The Solid Waste Handbook: A Prac-
tical Guide. Wiley-Interscience, New York. p. 259-376.
Onder, H. 1986. Interaction between conduit type flow and diffuse
flow in karst formations, in Giiltekin, G. , and Johnson, A.I.,
eds. Karst Water Resources. International Association of
Hydrological Sciences, Publication no. 161 (Proceedings of the
Ankara-Antalya Symposium, 1985). p. 371-385.
Palmer, A.N. 1990. Groundwater processes in karst terranes, in
Higgins, C.G., and Gates, D.R., eds. Groundwater Geomorpholo-
gy- Geological Society of America, Special Paper 252. p.
177-209.
Palmer, A.N. 1991. Origin and morphology of limestone caves.
Geological Society of American, Bulletin. 103(1):1-21.
Papadopulos, I.S. 1965. Nonsteady flow to a well in an infinite
anisotropic aquifer. Dubrovnik Symposium on Hydrology of
624
-------
Fractured Rocks, Proceedings. International Association of
Scientific Hydrology- 1:21-31.
Patrick, R., Ford, E., and Quarles, J. 1987. Groundwater Contam-
ination in the United States. 2nd ed. University of Pennsyl-
vania, Philadelphia. 513 p.
Pedler, W.H., Barvenik, M.J., Tsang, C.F., and Hale, F.V. 1990.
Determination of bedrock hydraulic conductivity and hydrochem-
istry using a wellbore fluid logging method. Ground Water
Management. (Proceedings of the Fourth National Outdoor
Action Conference on Aquifer Restoration, Ground Water Moni-
toring, and Geophysical Methods) 2:39-53.
Pfeffer, J.T. 1992. Solid Waste Management Engineering. Pren-
tice-Hall, Englewood Cliffs, N.J. 307 p.
Phillips, O.J. 1991. Flow and Reactions in Permeable Rocks.
Cambridge University Press, Cambridge. 285 p.
Pierce, J.R. 1980. An Introduction to Information Theory:
Symbols, Signals and Noise. 2nd ed. Dover, New York. 305 p.
Plummer, L.N., Busby, J.F., Lee, R.W., and Hanshaw, B.B. 1990.
Geochemical modeling of the Madison aquifer in parts of Mon-
tana, Wyoming, and South Dakota. Water Resources Research.
26(9):1981-2014.
Puig, J.M. 1990. Le systeme karstique de la Fountaine de la
Vaucluse. Bureau de Recherches Geologiques et Minieres
(BRGM), Document no. 180. 208 p. [doctoral thesis]
Quinlan, J.F. 1978. Types of karst, with emphasis on cover beds
in their classification and development. Ph.D. dissertation
(Geology), The University of Texas at Austin. 325 p.
Quinlan, J.F. 1986. Recommended procedure for responding to
spills of hazardous materials in karst terranes. Environmen-
tal Problems in Karst Terranes and Their Solutions Conference
(1st, Bowling Green, Ky.), Proceedings. National Water Well
Association, Dublin, Ohio. p. 183-196.
Quinlan, J.F. 1989. Ground-Water Monitoring in Karst Terranes:
Recommended Protocol and Implicit Assumptions. U.S. Environ-
mental Protection Agency, Environmental Monitoring Systems
Laboratory, Las Vegas, Nev. EPA/600/X-89/050 88 p. [DRAFT;
final version to be published 1992]
Quinlan, J.F. 1990a. Special problems of ground-water monitoring
in karst terranes, in Nielsen, D.M., and Johnson, A.I., eds.
Ground Water and Vadose Zone Monitoring. ASTM Special Tech-
nical Publication 1053. American Society for Testing and
Materials (ASTM), Philadelphia, p. 275-304.
625
-------
Quinlan, J.F. 1990b. Types of karst, in Quinlan, J.F., ed., Prac-
tical Karst Hydrogeology, with Emphasis on Ground Water
Monitoring. [Course manual] 5th ed. Association of Ground
Water Scientists and Engineers, Dublin, Ohio. 9:H-1 to H-9.
Quinlan, J.F. 1992. Interpretation of dye-concentration and dye-
recovery curves from four simultaneous interwell traces in a
karst aquifer. Privately published Christmas card. [Nash-
ville, Tenn.] 1 p.
Quinlan, J.F., and Alexander, E.G., Jr. 1987. How often should
samples be taken at relevant locations for reliable monitoring
of pollutants from an agricultural, waste disposal, or spill
site. Multidisciplinary Conference on Sinkholes and the Envi-
ronmental Impacts of Karst (2nd, Orlando, Fla.), Proceedings.
Balkema, Rotterdam, The Netherlands. p. 277-286.
Quinlan, J.F., and Alexander, E.G., Jr. 1990. Practical Tracing
of Groundwater, with Emphasis on Karst Terranes. Quinlan &
Associates, Nashville, Tenn. 2 v. 146 + 91 p.
Quinlan, J.F., and Aley, T. 1987. Discussion of "A new approach
to the disposal of solid waste on land", by R.C. Heath and
J.H. Lehr. Ground Water. 25(5):615-616.
Quinlan, J.F., and Ewers, R.O. 1981. Hydrogeology of the Mammoth
Cave Region, Kentucky, in Roberts, T.G., ed. GSA Cincinnati
'81 Field Trip Guidebooks. American Geological Institute,
Washington, D.C. 3:457-506.
Quinlan, J.F., and Ewers, R.O. 1985. Ground water flow in lime-
stone terranes: Strategy rationale and procedure for reli-
able, efficient monitoring of ground-water quality in karst
areas. National Symposium and Exposition on Aquifer Restora-
tion and Ground Water Monitoring (5th, Columbus, Ohio),
Proceedings. National Water Well Association, Worthington,
Ohio. p. 197-234.
Quinlan, J.F., Ewers, R.O. , and Field, M.S. 1988. How to use
ground-water tracing to "prove" that leakage from a karst
terrane will not occur. Conference on Environmental Problems
in Karst Terranes (2nd, Nashville, Tenn.), Proceedings.
National Water Well Association, Dublin, Ohio. p. 289-301
Quinlan, J.F., and Ray, J.A. 1989. Groundwater Basins in the
Mammoth Cave Region, Kentucky. Friends of the Karst, Occa-
sional Publication No. 2. Friends of the Karst, Mammoth Cave,
Ky. [map]
Quinlan, J.F., and Ray, J.A. 1991. Groundwater remediation may be
achievable in some karst aquifers that are contaminated, but
it ranges from unlikely to impossible in most: I. Implica-
tions of long-term tracer tests for universal failure in goal
attainment by scientists, consultants, and regulators.
626
-------
Hydrogeology, Ecology, Monitoring, and Management of Ground
Water in Karst Terranes Conference (3rd, Nashville, Tenn.)/
Proceedings. National Ground Water Association, Dublin, Ohio.
[in this volume]
Quinlan, J.F., and Rowe, D.R. 1977. Hydrology and Water Quality
in the Central Kentucky Karst: Phase I. Water Resources
Research Institute, University of Kentucky- Research Report
101. 93 p.
Robinson, E.A. 1980. Physical Application of Stationary Time-
Series, with Special Reference to Digital Data Processing of
Seismic Signals. Charles Griffin, London. 302 p.
Robinson, J.L., and Hutchinson, C.B. 1991. Ground-water tracer
tests in west-central Florida [abs.]. American Institute of
Hydrology, Annual Meeting (Orlando, Fla.), Abstracts, p. 39.
Roscoe Moss Company. 1990. Handbook of Ground Water Development.
Wiley, New York. 493 p.
Rubin, P.A. 1991. Land use planning and watershed protection in
karst terranes. Hydrogeology, Ecology, Monitoring, and
Management of Ground Water in Karst Terranes Conference (3rd,
Nashville, Tenn.), Proceedings. National Ground Water
Association, Dublin, Ohio. [in this volume]
Samama, J.C. 1986. Ore Fields and Continental Weathering. Van
Nostrand Reinhold, New York. 326 p.
Scanlon, B.R., and Thrailkill, J. 1987. Chemical similarities
among physically distinct spring types in a karst terrain.
Journal of Hydrology. 89:259-279.
Shuster, E.T., and White, W.B. 1971. Seasonal fluctuations in the
chemistry of limestone springs: A possible means for charac-
terizing carbonate aguifers. Journal of Hydrology. 14:93-128.
Smart, P.L., Edwards, A.J., and Hobbs, S.L. 1991. Heterogeneity
in carbonate aguifers: Effect of scale, fissuration, litholo-
gy, and karstification. Hydrogeology, Ecology, Monitoring,
and Management of Ground Water in Karst Terranes Conference
(3rd, Nashville, Tenn.), Proceedings. National Ground Water
Association, Dublin, Ohio. [in this volume]
Smart, P.L., and Friedrich, H. 1986. Water movement in the
unsaturated zone of a maturely karstified carbonate aquifer,
Mendip Hills, England. Environmental Problems in Karst
Terranes and Their Solutions Conference (1st, Bowling Green,
Ky.), Proceedings. National Water Well Association, Dublin,
Ohio. p. 59-87.
Smart, P.L., and Hobbs, S.L. 1986. Characteristics of carbonate
aguifers: A conceptual base. Environmental Problems in Karst
627
-------
Terranes and Their Solutions Conference (1st, Bowling Green,
Ky.), Proceedings. National Water Well Association, Dublin,
Ohio. p. 1-14.
Sun, R.J., and Weeks, J.B. 1991. Bibliography of Regional
Aquifer-System Analysis Program of the U.S. Geological Survey,
1978-91. U.S. Geological Survey, Water-Resources Investiga-
tions Report 91-4122. 92 p.
Thiery, D., Vandenbeusch, M., and Vaubourg, P- 1983. Interpreta-
tion des pompages d'essai en milieu fissure aquifere. Docu-
ments du BRGM, No. 57. 53 p.
Tsang, C.F-, Tsang, Y.W., and Hale, F.V- 1991. Tracer transport
in fractures: Analysis of field data based on a variable-
aperture channel model. Water Resources Research. 27(12):
3095-3106.
Tukey, J.W. 1962. The future of data analysis. Annals of Mathe-
matical Statistics. 33:1-67. [Reprinted in Jones, L.V., ed.
1986. The Collected Works of John W. Tukey, Vol 3: Philoso-
phy and Principles of Data Analysis, 1949-1964. Wadsworth and
Brooks/Cole, Monterey, Calif. p. 391-484.]
U.S. EPA [U.S. Environmental Protection Agency]. 1991. 40 CFR
Parts 257 and 258, Solid Waste Disposal Facility Criteria;
Final Rule. Federal Register, October 9. 56(196):50,977-
51,119.
van der Leeden, F., Troise, F.L., and Todd, O.K. 1990. The Water
Encyclopedia. Lewis, Chelsea, Mich. 808 p.
Viessman, W., Jr., Lewis, G.L., and Knapp, J.W. 1989. Introduc-
tion to Hydrology. 3rd ed. Harper & Row, New York. 780 p.
Ward, R.C., Loftis, J.C., and McBride, G.B. 1990. Design of Water
Quality Monitoring Systems. Van Nostrand Reinhold, New York.
231 p.
Water Science and Technology Board. 1988. Hazardous Waste Site
Management: Water Quality Issues. National Academy Press,
Washington, D.C. 212 p.
Water Science and Technology Board. 1990a. Ground Water and Soil
Contamination Remediation: Toward Compatible Science, Policy
and Public Perception. National Academy Press, Washington,
D.C. 261 p.
Water Science and Technology Board. 199Ob. Ground Water Models:
Scientific and Regulatory Applications. National Academy
Press, Washington, D.C. 303 p.
Watson, K.W., and Luxmore, R.J. 1986. Estimating macroporosity in
a forest watershed by use of a tension infiltrometer. Soil
628
-------
Science Society Journal. 50:578-582.
Weeks, E.P. 1969. Determining the ratio of horizontal to vertical
permeability by aquifer-test analysis. Water Resources
Research. 5(1):196-214.
Wells, E.R., and Krothe, N.C. 1989. Seasonal fluctuation in d15N
of groundwater nitrate in a mantled karst aquifer due to
macropore transport of fertilizer-derived nitrate. Journal
of Hydrology. 112:191-201.
White, W.B. 1977. Conceptual models of karst aquifers: Revisited,
in Dilamarter, R.R., and Csallany, S.C., eds. Hydrologic
Problems in Karst Terrains. Western Kentucky University,
Bowling Green. p. 176-187.
White, W.B. 1988. Geomorphology and Hydrology of Karst Terrains.
Oxford University Press, New York. 464 p.
Williams, P.W. 1983. The role of the subcutaneous zone in karst
hydrology. Journal of Hydrology- 61:45-67.
Williams, P.W. 1985. Subcutaneous hydrology and the development
of doline and cockpit karst. Zeitschrift fur Geomorphologie.
29(4):463-482.
Wilson, G.V-, Jardine, P.M., Luxmoore, R.J., Todd, D.E., Zelazny,
L.W. , and Lietzke, D.A. 1991. Hydrogeochemical processes
controlling subsurface transport from an upper subcatchment of
Walker Branch watershed during storm events. 1. Hydrologic
transport processes, and 2. Solute transport processes.
Journal of Hydrology- 123(3-4):297-336.
Wisconsin Geological and Natural History Survey. 1991. Delinea-
tion of Wellhead Protection Areas in Fractured Rocks. U.S.
Environmental Protection Agency, Office of Ground Water and
Drinking Water, Washington, D.C. EPA 570/9-91-009. 144 p.
Wright, V.P., Esteban, M., and Smart, P.L. 1991. Paleokarsts and
Paleokarstic Reservoirs. University of Reading, Postgraduate
Research Institute for Sedimentology, Contribution No. 152.
158 p.
Wynne, B. 1987. Risk Management and Hazardous Waste: Implementa-
tion and the Dialectics of Credibility- Springer, Berlin.
447 p.
APPENDIX
PROCEDURE FOR PREVENTION OF ALIASING AND FOR DETERMINATION OF
APPROPRIATE SAMPLING FREQUENCY FOR A SPRING OR CONDUIT-FED WELL
The chemical character of a discharging spring can be consi-
629
-------
dered to be a signal, a hydrologic time series that can be recorded
(monitored) continuously or sampled regularly. The main problem in
sampling, however, is how to choose the sampling interval, At. It
is obvious that sampling leads to some loss of information and that
this loss gets worse as At increases. Aliasing occurs. It is
expensive to make At very small, so a compromise value must be
sought. Determination of the maximum frequency of the signal in a
spring chemograph, f^, is quite simple if one is dealing with a
sinusoidal function with a constant frequency, but hydrologic func-
tions are less regular (less periodic or non-periodic) and have a
highly variable frequency. Accordingly, we propose a method for
obtaining a practical approximation of f^, a method based on the
common relationship in which specific conductance (conductivity; C)
varies inversely with discharge (Q). Our rationale for the method
will be similar to and complementary with that used for calculating
the response time, tR in the section of this paper concerned with
characterizing hydrological behavior as a measure of aquifer sensi-
tivity- Although we can compensate for the fact that some springs
are characterized by a short-lived increase in conductivity at the
onset of a storm event (because relatively mineralized water is
pushed out by piston-flow), we will ignore these short-lived peaks
in this description and utilize the well-known inverse relationship
between conductivity and discharge of springs and streams, a rela-
tionship caused by the relatively low conductivity of precipitation
and its dilution of groundwater.
The appropriate sampling interval, free of aliasing, will be
determined by a technique we propose here. This technique consists
of evaluation of a semi-log plot of conductivity vs. time, empiri-
cally determining the coefficients of the exponential equation des-
cribing the relationship between these data, and use of a coeffici-
ent in this equation to calculate the aquifer (spring) recovery
time, TR, using a procedure devised by Hess and White (1988). The
sampling interval is then calculated from this recovery time. [We
recognize the irony and seeming redundancy of designing a sampling
plan to design a sampling plan.]
It is more reliable, more convenient, and far less expensive
and less labor-intensive to monitor the conductivity continuously
for the purpose of designing the sampling plan. It would be wise
to do so through several storms, and for several months, so that
the variation, if any, in spring behavior can be detected, so that
fine structure (if present) in the chemograph can also be detected,
and so that subsequent possible aliasing by discrete sampling can
be prevented. Accordingly, the following description is based on
the use of data from continuous records of conductivity rather than
discrete samples for it.
We classify conductivity chemographs into three gradational
groups. A, B, and C. For group A, chemographs with a smooth reces-
sion limb, perform the following 7 steps:
1. On a plot of conductivity vs. time (extending over several
months), sketch a smooth curve for background conductivity,
630
-------
CB. [The background is likely to show a broad seasonal
trend.]
2. On the chemograph for each storm event with a good record,
pick the lowest point after the trough and on the recession
limb where it begins to be linear. This is Day 0 (t0) and
C0 for each storm event.
3. For each of the storm events selected for step #2, pick the
post-C0 values of C at 6 am and 6 pm, record them, the cor-
responding value for CB, the difference between them (AC) ,
and the values for time, t. Do so only for the near-linear
part of the curve.
4. For each data set, plot the following on semi-log paper: AC
(on the y-axis, the log scale) vs. t (on the x-axis, the
arithmetic scale).
5. Determine b, the conductivity coefficient, in the following
exponential equation in which b is the slope of its
straight line on semi-log paper:
AC = C0e"bt
A procedure for easily solving this equation graphically is
given by Krumbein (1937) and Krumbein and Pettijohn (1938,
p. 209-210). Alternative graphical procedures are given by
Mackey (1936, p. 113-115) and Davis (1955, p. 16-20). [An-
other alternative is to perform a computer-aided regression
analysis.]
6. The reciprocal of b is the recovery time, TR, for each
storm event, as described by Hess and White (1988, p. 250-
251) . This is a measure of how long it takes the conducti-
vity of a spring (aquifer) to rise to a value that is ap-
proximately 63% (= 1 - 1/e) of its pre-storm value. The TR
for several storms can be averaged. Alternatively, one
might select the appropriate TR for the recession limb of
a particular storm being monitored by comparing its nascent
hydrograph or chemograph shape with one already studied.
7. As a first approximation, we recommend sampling every
0.05TR from the onset of a storm until 0.2TR after its con-
ductivity minimum, Cmjn, or stage maximum, O.^, then every
0.1 TR for the duration of TR, and followed by 0.2TR for as
long as desired. One might change to sampling every 0.4TR
at 4TR after Cmjn or Qmax. These sampling frequencies are a
guide; sampling should generally be done at reasonable
hours of the day rather than at 3:21 a.m. because that was
the calculated value — unless, of course, one is trying to
characterize a high-frequency signal. Experience may show
that use of different coefficients of TR yield results more
desirable.
631
-------
For group B, conductivity chemographs with spikey or finely
undulatory structure, select a sample interval, At, that is twice
the highest frequency to be resolved. For example, a signal with
10-minute frequency peaks should be sampled at least every 5
minutes. TR is ignored.
For group C, conductivity chemographs with a coarsely undula-
tory structure in the recession limb, one can follow the procedure
for group B or sketch a smooth line that curvilinearly straightens
them and then follow the procedure for group A. Use this sketched
line to determine the values for t and C described in steps 2 and
3, above.
Although there is usually a slight phase-lag and an inverse
correlation between chemograph and hydrograph structure, sometimes
there is no relationship — as when, for example, there is a semi-
sinusoidal spring hydrograph from a siphon having a 20-minute
period. Sometimes also, there may be a fine structure in the con-
ductivity chemograph that is not present in the hydrograph or
thermograph (Meiman et al., 1988).
The procedures described above minimize the possibility of
aliasing of samples. If followed, they yield an objective, reli-
able, non-aliased, lowest cost, legally defensible sampling plan
that maximizes the probability of accurate characterization of an
aquifer with conductivity data and/or other water-quality parame-
ters. We are excited at the prospect of testing these plans at a
wide range of spring types and conditions.
BIOGRAPHICAL SKETCHES
Dr. James F. Quinlan, P.G., is president of Quinlan & Associates,
Inc., a consulting firm specializing in problems of carbonate ter-
ranes. He was Research Geologist for the National Park Service at
Mammoth Cave, Kentucky, for 16 years and has been an independent
consultant on karst for more than 10 years. He earned a Ph.D. in
geology at The University of Texas at Austin (1978). His experi-
ence includes 36 years of research and observations in karst ter-
ranes of 26 states and 25 countries, and work as a consultant in
many of them. He has written or co-written more than 180 publica-
tions on karst-related topics. He is co-chairman of ASTM Subcom-
mittee D18.21.09 on Special Problems of Monitoring in Karst and
Fractured Rock Terranes.
Dr. Peter L. Smart is a lecturer in geography at the University of
Bristol, England, from which he also received his first degree and
doctorate. He has extensive research experience in karst hydrology
and geomorphology in many different areas of the world, including
the Bahamas, China, Indonesia, Malaysia, and several European
countries.
Geary M. Schindel, P.G., has been Project Manager and Senior Hydro-
geologist for Eckenfelder Inc. since 1990. He was previously
632
-------
Manager of the Environmental Division of ATEC Associates, Inc., in
Nashville. Prior to joining ATEC, he was Manager of the Ground-
water Branch of the Kentucky Division of Water. He has also worked
as a research assistant at Mammoth Cave National Park and the
Center for Cave and Karst Studies at Western Kentucky University in
Bowling Green. Geary has a M.S. degree from Western Kentucky Uni-
versity and a B.S. degree from West Virginia University.
Dr. E. Calvin Alexander, Jr., is a Professor in the Department of
Geology and Geophysics at the University of Minnesota. He has a
B.S. in Chemistry (1966) from Oklahoma State University and a Ph.D.
in Chemistry (1970) from the University of Missouri at Rolla. The
central theme of his current research is the rate of movement of
fluids in hydrogeology. He and his research group are utilizing a
variety of methods to measure flow and residence times of water in
aquifers, which can range from hours to tens of thousands of years.
Alan J. Edwards is a research student in the Geography Department
of the University of Bristol, from which he also obtained his first
degree. He is currently completing his doctoral dissertation
concerning the hydrogeology of quarries and landfills in the
karstified limestone aquifer of the Mendip Hills.
A. Richard Smith earned a B.S. in Geology (1964) from The Universi-
ty of Texas at Austin. He has been involved in various hydrogeo-
logical activities for two major corporations and is now Senior
Geologist of the Ground-Water Protection Branch, Bureau of Solid
Waste Management, Texas Department of Health. His experience in
karst geomorphology and hydrology includes investigation of Texas
caves as an associate of the Texas Speleological Survey, ongoing
long-term study of gypsum karst in West Texas, and exploration for
karst-related sulfur deposits in evaporite rocks.
ADDRESSES
Dr. James F. Quinlan
Quinlan & Associates, Inc.
Box 110539
Nashville, TN 37222-0539
(615) 833-4324
Dr. Peter L. Smart
Department of Geography
University of Bristol
Bristol BS8 1SS, ENGLAND
44-272-303030
Dr. E. Calvin Alexander, Jr.
Department of Geology and
Geophysics
University of Minnesota
Minneapolis, MN 55455-0219
(612) 624-3517
Geary M. Schindel
Eckenfelder Inc.
227 French Landing Drive
Nashville, TN 37228
(615) 255-2288
Alan J. Edwards
Department of Geography
University of Bristol
Bristol BS8 1SS, ENGLAND
44-272-303030
A. Richard Smith
Municipal Solid Waste Division
Texas Water Commission
P.O. Box 13087, Capitol Station
Austin, TX 78711
(512) 834-6683
633
-------
Recommended Administrative/Regulatory Definition of Karst Aquifer,
Principles for Classification of Carbonate Aquifers, and Practical
Evaluation of Vulnerability of Karst Aquifers
By: James F- Quinlan, Peter L. Smart, Geary M. Schindel, E. Calvin
Alexander, Jr., Alan J. Edwards, and A. Richard Smith
Q. Should the definition of karst aquifer include the modifier of
rock type? Isn't the term too broad without such a modifier?
A. Yes and no. More specifically, because the great majority of
karsts are developed in carbonate rock, the word karst implies
a limestone or dolomite karst. If one were referring to a
karst developed in another rock type, one would refer to it as
a gypsum karst, a salt karst, a carbonatite karst, etc.
Q. There is an inherent problem in lumping all carbonate aquifers
together as "karst11, especially when the word karst is used in
regulations. By your broad definition, 98% of Missouri, for
example, could be classified as a karst terrane — in spite of
the fact that 60% of the state is not significantly affected
by karst. Why not restrict the word karst to areas where
there are environmental problems and use another word for the
other carbonate aquifer areas?
A. Approximately 60%, not 98%, of Missouri has limestone or
dolomite cropping out at the surface, and approximately 80% of
that 60% (50% of the state) is karst, as we have defined it in
this paper. Without trying to nitpick over semantics, a key
phrase in the above question is significantly affected. We
can agree to disagree with the accuracy of the unknown
questioner's estimate of 60% of the erroneous 98% but, never-
theless, address the principle raised. The definition of
karst is vague enough without making it more so, as it would
be if the questioner's suggestion were followed. We do not
believe in changing traditional definitions. Those we have
given herein are a codification, fine-tuning, and synthesis of
traditional uses; they replace definitions poorly worded by
non-specialists in karst. The new terms we have introduced,
describing aquifer sensitivity, may achieve what the question-
er suggested about using another word in lieu of karst. but it
does so without changing the meaning of karst. Also, a key
word in the question is are. It excluded might be. We believe
it is better to err on the side of conservatism. Aquifer
contamination, like death, can be so permanent.
Q. Doesn't the work of Scanlon and Thrailkill (1987) in the
Bluegrass Karst region, near Lexington, Kentucky, contradict
what you claim about the reliability of inferring the nature
of an aquifer from analysis of its spring waters?
634
-------
At first glance, the work of Scanlon and Thrailkill (1987)
might seem to invalidate the relations we allege between water
quality and physical characteristics of springs and their
aquifers. They claim that the two spring types studied, major
(large) and high-level (small), could not be discriminated on
the basis of water-quality parameters. We agree. But our
review of their published data and their description of
behavior of both spring types during storms shows unambiguous-
ly that each of them is a conduit-flow spring. Their work
does not question or invalidate our conclusion on the prac-
tical utility of conductivity measurements for assessing
aquifer types or aquifer vulnerability.
635
-------
636
-------
LEGAL TOOLS FOR THE PROTECTION OF GROUND WATER
IN KARST TERRANES
GARY A. DAVIS, Attorney at Law
Ray, Farmer, Eldridge, & Hickman
Knoxville, Tennessee
JAMES F. QUINLAN
Quinlan and Associates
Nashville, Tennessee
ABSTRACT
Ground water in karst terranes is exceptionally vulnerable to
contamination due to human activities such as waste disposal,
septic tanks, chemical spills, and storage of gasoline in
underground tanks. Because of this vulnerability, legal tools for
the protection of ground water are extremely important in karst
terranes. These legal tools include source controls, land-use
controls, which may be utilized by state and local governments and
by several types of quasi-governmental entities. Users of ground
water can assert their water rights to protect ground water.
Source controls include statutes and regulations governing the
licensing of potential sources of contamination and design and
operating standards to prevent release of contaminants. Land use
controls include zoning regulations and the power of eminent
domain. These controls should recognize the exceptional
vulnerability of ground water in karst terranes, but most statutes
and regulations do not provide for any differential treatment of
karst aquifers. Unfortunately, many of those that address karst
only deal with the potential for surface collapse and not the rapid
migration of contaminants in solution channels. Additionally,
wellhead and sensitive area protection programs in karst terranes
should not be based upon arbitrary circles around sources of water
supplies, but upon well-designed studies delineating ground water
flow in solution channels.
Finally, ground water users in karst terrane can rely upon
principles of water law developed for surface water to protect
their water rights. The principle of reasonable use has been
applied to ground water, particularly ground water flowing in
637
-------
discrete channels in karst terrane.
INTRODUCTION
Ground water in karst terranes is exceptionally vulnerable to
contamination due to human activities such as waste disposal,
septic tanks, chemical spills, and storage of gasoline in
underground tanks. Water supplies dependent upon ground water in
karst terranes are also subject to disruption by subsurface
disturbances such as mining, and by diversion of flow by an
"upstream" user.
Because of this exceptional vulnerability, legal tools for the
protection of ground water are extremely important in karst
terranes. These legal tools include source controls, land-use
controls, and water rights. They may be utilized by federal, state
and local governments and by several types of quasi-governmental
entities.
Source controls include statutes and regulations governing the
licensing of potential sources of contamination and design and
operating standards to prevent release of contaminants. These need
to be tailored to recognize this special vulnerability of ground
water in karst terranes, but, unfortunately- many of these statutes
and regulations only deal with the potential for surface collapse
and not the rapid migration of contaminants in solution channels.
Land use controls include zoning regulations and the power of
eminent domain. In addition to counties and municipalities, there
are other quasi-governmental entities, such as watershed districts,
that possess some of these powers. The federal wellhead protection
provisions can act in the manner of land use controls to prevent
the location of potential sources of ground water contamination.
Finally, ground-water users in karst terranes may be able to
rely upon principles of water law developed for surface water to
protect their water rights. The principle of reasonable use has
been applied to ground water, particularly to ground water flowing
in discrete channels in karst terrane.
SOURCE CONTROLS
Legal Framework
The protection of ground water was not a specific focus of
concern until the 1970s, and the legal framework for ground-water
protection on the federal level and in most states is fragmented.
It is made up of various federal and state laws, some of which
touch only indirectly on ground water protection. These controls
and exercises of rights can be set in perspective by reading a
638
-------
recent critical review of public policy considerations that enter
into the assignment of a regulatory scheme to one level of
government rather than another.
Federal laws governing potential sources of ground-water
pollution as an area of major concern include the Safe Drinking
Water Act (underground injection control and wellhead protection);
the Resource Conservation and Recovery Act (solid and hazardous
waste management and underground storage tanks); the Comprehensive
Environmental Response, Compensation, and Liability Act
(investigation and remediation of uncontrolled hazardous substance
sites); the Toxic Substance Control Act (disposal of PCB's and
asbestos); the Federal Insecticide, Fungicide, and Rodenticide Act
(pesticide use); the Uranium Mill Tailings Radiation Control Act;
the Surface Mining Control and Reclamation Act, and the Atomic
Energy Act of 1954 (radioactive waste disposal).
Of these, the Safe Drinking Water Act, the Resource
Conservation and Recovery Act, and the Surface Mining Control and
Reclamation Act allow states to assume primary management authority
to carry out the federal legislation, if the state programs meet
federal guidelines and are approved by the federal agency. State
regulations under these laws can generally be more stringent than
federal regulations. In addition to federal laws and authorized
state programs under those laws, states have regulated other
potential sources of ground-water degradation, including septic
tank systems, oil exploration, and water withdrawal.
Specific Provisions for Karst Terranes
Federal statutes and regulations governing potential sources
of ground water degradation have generally not addressed karst
terranes. The only instance in which karst terranes have been
addressed in federal source-control regulations is in the new EPA
municipal solid-waste landfill regulations, promulgated under RCRA.
Under these regulations, new and existing municipal solid waste
landfills in "unstable areas" must include a demonstration that
"engineering measures have been incorporated into the [landfill]
design to ensure that the integrity of the structural components of
the [landfill] unit will not be disrupted." An "unstable area,"
under the regulation includes karst terranes.3 Guidance for the
investigation of site-specific potential for subsidence and
collapse is to be included in the technical guidance document the
agency plans to issue within six months.
This regulation deals only with the potential for surface
collapse and not with the potential for rapid migration of
contaminants through solution channels and the difficulty of
detecting the release of contaminants in karst terranes. Although
none of the monitoring regulations specifically address karst
terranes, they are sufficiently flexible to allow state agencies to
639
-------
require proper monitoring.
Several state source-control statutes and regulations address
karst terranes specifically. One of the most comprehensive
examples is the Minnesota Solid Waste Disposal Regulations, which
deal with the difficulties of preventing and detecting movement of
contaminants in karst terranes as well as the potential for surface
collapse.
The Minnesota Technical Requirements for Solid Waste
Facilities require the siting of a solid waste landfill only in an
area where the ground water flow paths are known in sufficient
detail to enable reliable tracking of pollutant movement and where
it is feasible to construct a monitoring system with sufficient
monitoring points to assure that pollutants can be detected and
tracked.4 This general siting requirement would make siting in
karst terranes difficult without a dye-trace study to delineate
ground-water flow paths and establish monitoring points.
The regulations go on to state:
A land disposal facility must not be located on a site
where: (1) there are karst features, such as sinkholes,
solution channels, disappearing streams, and caves, which
may cause failure of the leachate management system or
prevent effective monitoring or containment of a release
of leachate;
Kentucky solid waste regulations also focus on the
difficulties of characterizing ground water flow and detecting
pollutant migration in karst terranes. In applying for a solid
waste landfill permit in a karst environment, the applicant must
characterize both diffuse and discrete flow conditions, and the
state may require dye trace studies before finalizing a ground
water monitoring plan.6 In addition to the general requirement
that a landfill may only be sited in an area where the uppermost
aquifer is capable of being monitored in a manner that detects
migration of pollutants and where corrective action is possible
once pollutants are detected, Kentucky regulations prohibit
placement of waste within 250 feet of a feature of karst terrane.7
This buffer-zone requirement, however, gives a false sense of
security, since waste placed within 250 feel of a karst feature
could easily be still within the karst terrane.
Tennessee solid waste regulations establishing landfill
location requirements for karst areas address both potential for
surface collapse and migration of contaminants but fail to deal
with the difficulties of reliably detecting contaminant migration
in karst terranes. The Tennessee regulation states:
If a facility is proposed in an area of highly developed
karst terrane (i.e., sink holes, caves, underground
640
-------
conduit flow drainage, and solutionally enlarged
fractures) the applicant must demonstrate to the
satisfaction of the Commissioner [of Environment and
Conservation] that relative to the proposed facility
siting:
(i) There is no significant potential for surface
collapse;
(ii) The ground water flow system is not a conduit flow
which would contribute significant potential for
surface collapse or which would cause significant
degradation to the ground water; and
(iii) Location in the karst terrane will not cause
any significant degradation to the local
ground water resources.8
Tennessee's ground-water monitoring provisions for solid waste
landfills, however, do not address the difficulty of detecting
contaminant flow in karst terrane, nor have dye-trace studies been
required by the state for the karst demonstration envisioned by the
regulation. The regulation also gives far too much discretion to
the Commissioner to decide when to require a karst demonstration
and whether a conduit flow system could cause degradation to the
ground water.
Alabama has a similar presumption against the siting of
landfills in karst terranes unless a site specific demonstration is
performed, but the site-specific demonstration required does not
address the particular difficulties of determining ground water
flow paths and tracking contaminant migration in karst terranes.10
Other states that include specific standards for location of solid
waste landfills in karst terranes include Florida and Georgia, but
their requirements seem to be limited to preventing waste disposal
in or near sinkholes.
Local Government Source Control Programs
General police power ordinances which employ such regulatory
mechanisms as permits, licenses, inspections, and standards to
prevent potential sources of pollution have also been used by -local
governments to protect ground water. Health codes adopted by local
governments, or regulations imposed by local health departments,
have an advantage over land-use controls, discussed below, since
they can be imposed retroactively to deal with existing pollution
threats as well as future threats. They may also gain more public
support than zoning laws.
In most states, unless state and federal programs preempt
local authority, local source control efforts to protect ground
water can proceed by any of several routes, including county and
municipal ordinances, local health department regulations,
watershed district regulations, and utility district regulations.
641
-------
For example, the Cape Cod Planning and Economic Development
Commission designed model groundwater protection ordinances for
enactment at the municipal level and by local boards of health
after designation of their region as a Sole Source Aquifer
protectable under the Safe Drinking Water Act. One ordinance,
designed for adoption by municipalities, regulates all firms
handling toxic or hazardous materials. Another, designed for
adoption by local health departments, regulates both the
installation of new underground storage tanks and the maintenance
of older ones.12
Quasi-governmental entities, such as watershed districts and
utility districts, may also be able to regulate potential sources
of ground water contamination. Many states provide for the
creation of watershed districts that are given broad powers to
conserve soil and water resources within the natural boundaries of
a watershed.13 Among the powers delegated to districts by the
Tennessee legislature, for example, are "to take such steps as
deemed necessary by (the) board of directors for the promotion and
protection of public health within the boundaries of the district"
and "to do all things necessary and proper for the protection of
the watershed and the lands and waters within the boundaries of the
district."14 Although most watershed districts are created to
deal with surface streams, the regulatory powers of a watershed
district should be able to protect ground water in karst terranes
where there is a clear connection between ground water and surface
streams.
Many states also provide for the creation of water utility
districts to provide for water supply. Where a utility district
depends upon ground water from wells or springs, the power to
maintain the source of supply that is granted by the state
legislature may authorize regulations for the control of pollution
sources which potentially endanger a ground water. For instance,
under Tennessee law utility districts are empowered to acquire,
construct, operate, and maintain systems for the furnishing of
water, and to exercise all powers necessary for the accomplishment
of these purposes.15
Local source-control regulations also need to recognize the
particular difficulties in ensuring ground water protection in
karst terrane. Where the federal or state source control
regulations have not gone far enough in addressing these
difficulties, local regulations may fill this gap.
LAND-USE CONTROLS
Land-use controls are well-suited for preventing ground-water
contamination in karst terranes. Sensitive area protection
programs and wellhead protection programs are appropriately
implemented through zoning and other land-use controls.
642
-------
Most local governments have broad zoning powers for the
protection of the public health and welfare. Some states, such as
Tennessee, have given local governments explicit power to zone for
conservation of water supply- Zoning may take numerous forms,
including prohibitions of land uses with serious potential to
contaminate ground water, design or performance standards on land
uses, limitations on density or lot sizes, and overlay districts
designating recharge or well-head protection areas.17
Another form of land-use control is the purchase or
condemnation by local governments of restrictive easements across
land in critical recharge areas. Most local governments can
condemn land or rights in land for public purposes, including
protection of water supplies. A ground-water protection easement
allows the owner of property to retain the land and use it for any
purpose that does not present a threat to ground water, while
requiring less compensation by the local government than the
purchase or condemnation of the entire property.
For land-use controls to be effective in karst terranes,
careful mapping of ground water flow is necessary. Reliance upon
arbitrary circular restrictive zones around water supply wells or
springs is unlikely to protect water supplies in karst, since
contaminants may flow long distances in solution channels with
little attenuation.
Several communities have utilized land-use controls to protect
ground water.18 An example in karst terrane is the comprehensive
watershed development ordinances adopted by the City of Austin,
Texas, to protect the Edwards Aquifer. Austin has delineated
special watershed zones for controlled development density and has
prohibited almost all development within "critical water quality
zones" in aquifer recharge areas.19
Although Lexington, Kentucky, and the surrounding county were
primarily concerned with prevention of flooding of sinkholes in
response to storms and blockage of sinkhole drains by sediment that
runs off in response to erosion triggered by construction
activities, their ordinance also protects groundwater quality.
Similar ordinances have been enacted by Bowling Green, Kentucky21,
and Springfield and Perryville, Missouri.
Clinton Township, New Jersey, and Macungie Township,
Pennsylvania, both in the Appalachians, in a effort primarily to
control sinkhole development but also to protect ground water
quality in the karst aquifers that underlay them, have each enacted
ordinances in 1988 that regulate but do not prohibit construction
in order to lessen the probability of sinkhole collapse.22 The
legality and validity of the New Jersey ordinance has been upheld
in the courts, and its operation has been successful.23
Local land use controls can be aided by federal programs found
643
-------
in the Safe Drinking Water Act. The Wellhead Protection Program,
included in the 1986 amendments to the Act, allows states to
develop and submit to EPA programs that protect areas around public
water supply wells (Wellhead Protection Areas). These Wellhead
Protection Programs often provide a mix of state and local source
controls with local land use controls.24
COMMON LAW PROTECTIONS
Ground-water users can rely upon principles of water law
developed for surface water to protect their water rights. The
principle of reasonable use, where applicable, requires each
property owner to make only reasonable use of the ground water so
as not to interfere with use of ground water by other property
owners. An unreasonable use can be enjoined by the courts.
Under English common law, water under the ground was generally
considered the exclusive property of the surface owner. Thus, the
surface owner could make whatever use of the ground water he or she
wished, including uses destructive of ground water quality and
quantity. In the United States state courts have gradually
recognized the fact that ground water flows and that one
landowner's use of ground water can affect other landowners.
American courts have adopted the reasonable use rule, which holds
that one user may only put ground water to reasonable use where it
may affect the use of ground water by another user.
The first step in that recognition, and a distinction which
still holds in some states, was to apply the reasonable use rule to
water that flowed underground in a well-defined channel, as opposed
to percolating ground water. Since water flowing in a well-defined
channel is similar to surface water, the reasonable use rule, which
was developed for surface waters, applied. Such conduit flow, of
course, is an attribute of karst terranes.
Most states now apply the reasonable use rule to either
conduit flow situations or percolating ground water where there is
proof that one property owner's use has unreasonably affected the
ground water use of another. For instance, the Tennessee Court of
Appeals found in 1935 that the rights of a property owner to enjoin
the unreasonable use of ground water that supplied a spring did not
depend on proof of a well-defined channel, but simply of an
"intimate" ^connection between the ground water on the two
properties.
But in states where the well-defined channel distinction is
still important, knowledge of the subsurface conduit flow feature
is critical to the protection of the ground water. Texas, for
instance, still follows the English rule for percolating ground
water, and there is a presumption that all ground water is
percolating ground water. In a recent case, the Texas Court of
644
-------
Appeals found the hydrogeological testimony insufficient to
establish well-defined channel flow of a spring flowing through an
underground cavity before it surfaced. This ruling permitted the
property owner to utilize all of the flow of the spring for
irrigation purposes.
DISCUSSION AND CONCLUSIONS
There are ample legal tools available for the protection of
ground water by state and local governments and by individual
ground-water users. Most statutes and regulations, however, ignore
karst aquifers because the framers never thought about them or had
no knowledge of how they differ from other aquifers. When these
legal tools are used for karst terranes, they need to be sharpened
to address the particular features of karst.
There are three major problems with many existing statutes and
regulations that specifically address the regulation of activities
in karst terranes:
1. The definitions of karst are inadequate or ambiguous,
particularly when they focus solely on the presence of
sinkholes.
2. The definition of karst, even when adequate, is often
misinterpreted to apply only to areas with sinkholes.
3. There is too much emphasis on the surface of the karst
and too little on the aquifer itself. Regulatory buffer
zones of a few hundred feet radius around sinkholes, for
instance, ignore the fact that, in most terranes, the
greatest danger to ground water is migration of
contaminants below the surface, not collapse of the
surface.
Source-control regulations addressing karst terranes,
including siting restrictions for potentially polluting activities,
should focus on more than just the potential for surface collapse
in karst terranes. Because of the extreme difficulties in
preventing and detecting ground water contamination in karst, there
should be a presumption against the location of certain activities,
such as hazardous and solid waste landfills, in karst terranes.
And before that presumption can be overcome, there must be a
demonstration that contaminants would not enter water supplies and
that reliable monitoring can be achieved as a result of detailed
knowledge of ground-water flow paths.
Land-use controls must also recognize the unpredictable nature
of ground-water flow in karst terranes. Arbitrarily drawn recharge
zones for wells and springs may not only be unfair to property
owners, but may miss conduit-flow pathways entirely. Again,
645
-------
detailed knowledge of ground-water flow is necessary.
Finally, water rights in ground water may be asserted by
individual ground-water users. The extent of such rights, however,
may actually depend on the ability of the user to prove that the
ground water is flowing as a discreet underground stream.
REFERENCES CITED
1. See Denise D. Fort, Federalism and the Prevention of
Groundwater Contamination, Water Resources Research, Vol. 27,
No. 12, pp. 2811-2817 (December 1991). See also Timothy R.
Henderson, The Institutional Framework for Protecting
Groundwater in the United States, in G. William Page, ed.,
Planning for Groundwater Protection. Academic Press, Orlando,
Florida, pp. 29-67.
2. Safe Drinking Water Act, 42 United States Code Sections 300f-
j; Resource Conservation and Recovery Act, 42 United States
Code Sections 6901, et seq.; Comprehensive Environmental
Response Compensation, and Liability Act, 42 United States
Code Sections 9601, et seq.; Toxic Substances Control Act, 15
United States Code Sections 2601, et seq.; Federal
Insecticide, Fungicide and Rodenticide Act, 7 United States
Code Sections 136, et seq.; Uranium Mill Tailings Radiation
Control Act, 42 United States Code Sections 7901, et seq.;
Surface Mining Control and Reclamation Act, 30 United States
Code Sections 1201, et seq.; Atomic Energy Act of 1954, 42
United States Code Sections 2014, 2021, 2021a, 2022, 2111,
2113, 2114.
3. 56 Federal Register 51019-20 (Oct. 9, 1991). "Karst terranes"
are defined as "areas where karst topography, with its
characteristic surface and subterranean features, is developed
as the result of dissolution of limestone, dolomite, or other
soluble rock. Characteristic physiographic features present in
karst terranes include, but are not limited to, sinkholes,
sinking streams, caves, large springs, and blind valleys."
This definition is repeated on page 51047, but there "karst
terranes" are twice mistakenly referred to as "karst
terraces".
4. Minnesota Rules, Section 7035.2815, Subpart 2A. (1988).
5. Minnesota Rules, Section 7035.2815, Subpart 2C. (1988). Under
Minnesota Rules, Section 7035.0300, Subpart 51, "karst" "means
a type of topography that is formed from the dissolution of
limestone, dolomite, or gypsum and that is characterized by
closed depressions or sinkholes, and underground drainage
through conduits enlarged by dissolution."
646
-------
6. Kentucky Administrative Regulations, 401 KAR .. Under 401 KAR
30:010(104) "karst terrain" "means a type of topography where
limestone, dolomite or gypsum is present and is characterized
by naturally occurring closed topographic depressions or
sinkholes, caves, disrupted surface drainage, and well
developed underground solution channels formed by dissolution
of these rocks by water moving underground."
7. Kentucky Administrative Regulations, 401 KAR 48:050, Sections
6 and 1, respectively-
8. Tennessee Administrative Code, Rule 1200-1-7-.04(2)(q). Under
Rule 1200-l-7-.01(2) "karst" "means a specific type of
topography that is formed by dissolving or solution of
carbonate formations, such as limestone or dolomite; it is
characterized by closed depressions or sinkholes, caves,
sinking and reappearing streams, and/or underground conduit
drainage flow."
9. Rule 1200-l-7-.04(2) (q) goes on to state that "the above-
referenced demonstration may require the installation of
piezometers, the developing of potentiometric-surface map of
ground water, conducting geophysical surveys, dye tracing or
other specific requirements deemed necessary by the
Commissioner to evaluate the proposed site to his
satisfaction." In our experience with solid waste landfill
siting in Tennessee, the focus of the Division of Solid Waste
has been more on the detection of potential for surface
collapse through geophysical studies, and no dye trace studies
have been required, despite clear indications of highly
developed karst. Those traces that have been performed in an
attempt to demonstrate the threat to karst aquifers have been
ignored or misinterpreted by the Division.
10. Alabama Administrative Code, Sections 335-13-4-. 01 (6 ), 335-13-
4-.11 through 335-13-4-.14. (as amended through October 2,
1990).
11. Florida Administrative Code, Section 17-701.040(2) (as amended
through July 19, 1990); Rules and Regulations of the State of
Georgia, Section 391-3-4-.05 (g)5 . (as amended through June 29-
1989) .
12. Scott W. Horsley, Beyond Zoning: Municipal Ordinances to
Protect Groundwater, Cape Code Planning and Economic
Development Commission, Barnstable, Mass. (1982).
13. For example, Tennessee Code Annotated Sections 69-7-101, et
seq. provide for the creation of watershed districts.
14. Tennessee Code Annotated Section 69-7-118.
647
-------
15. Tennessee Code Annotated Sections 7-82-302, 7-82-306.
16. Tennessee counties may be pass zoning ordinances for the
conservation of water supply. Tennessee Code Annotated Section
13-7-101.
17. See Douglas Yanggen and Stephen Born, Protecting Groundwater
Quality by Managing Local Land Use, Journal of Soil and Water
Conservation, Vol. 45, No. 2, pp. 207-210 (March-April 1990);
F. DiNovo and M. Jaffe, Local Groundwater Protection—Midwest
Region, American Planning Association, Chicago, II. (1984).
18. See DiNovo and Jaffe, supra; See also, Committee on
Groundwater Quality Protection, Groundwater Quality
Protection—State and Local Strategies, National Academy
Press, Washington, D.C. (1986).
19. See Douglas Yanggen and Stephen Born, Protecting Groundwater
Quality by Managing Local Land Use, Journal of Soil and Water
Conservation, Vol. 45, No. 2, p. 207-210 (March-April 1990);
Kent S. Butler, Urban Growth Management and Groundwater
Protection: Austin, Texas, in G. William Page, ed., Planning
for Groundwater Protection, Academic Press, Orlando, Fla., p.
261-287; H.D. Smith, Erosion and Sedimentation Control
Methodologies for Construction Activities over The Edwards
Aquifer in Central Texas. Hydrogeology, Ecology, and
Management of Ground Water in Karst Terranes Conference, (3rd,
Nashville, Tenn.), Proceedings. National Ground Water
Association, Dublin, Ohio (1991). [In this volume].
20. See James S. Dinger and James R. Rebmann, Ordinance for the
Control of Urban Development in Sinkhole Areas in the
Bluegrass Region, Lexington, Kentucky, Environmental Problems
in Karst Terranes and their Solutions Conference (1st, Bowling
Green, Ky.), Proceedings. National Water Well Association,
Dublin, Ohio, pp. 163-180.
21. See Nicholas C. Crawford, Sinkhole Flooding Associated with
Urban Development Upon Karst Terrain: Bowling Green,
Kentucky- Multidisciplinary Conference on Sinkholde (1st,
Orlando, Fla.), Proceedings, pp. 283-292 (1984).
22. See Joseph A. Fisher and Hermia Lechna, A Karst Ordinance,
Clinton Township, New Jersey, Engineering and Environmental
Impacts of Sinkholes and Karst (St. Petersburg, Fla.)
Proceedings, pp. 357-361 (1989); Percy H. Dougherty, Land-Use
Regulations in the Lehigh Valley: Zoning and Subdivision
Ordinances in an Environmentally Sensitive Karst Region,
Ibid., pp. 341-348.
648
-------
23. See Joseph A. Fisher and Robert J. Canace, Karst Geology and
Ground-Water Protection Law, Hydroqeoloqy, Ecology/
Monitoring, and Management of Karst Terranes Conference (3rd,
Nashville/ Tenn.), Proceedings, National Ground Water
Association, Dublin, Ohio. (1991) [in this volume].
24. 42 United States Code Section 300h-7.
25. Nashville, C. & St. L. Ry. v. Rickert, 89 S.W.2d 889
(Tenn.Ct.App. 1935), cert, denied (Tenn. 1936).
26. Denis v. Kickapoo Land Co., 771 S.W.2d 235 (Tex.Ct.App. 1989).
27. See James F. Quinlan, P.L. Smart, G.M. Schindel, E.G.
Alexander, Jr., A.J. Edwards, and A.R. Smith. Recommended
Administrative/Regulatory Definitions of Karst Aquifer,
Principles for Classification of Carbonate Aguifers, and
Practical Evaluation of Vulnerability of Karst Aguifers.
Hydrogeology, Ecology, Monitoring, and Management of Ground
Water in Karst Terranes Conference (3rd, Nashville, Tenn.),
Proceedings. National Ground Water Association, Dublin, Ohio
(1991). [in this volume]. This paper proposed the following
definitions:
karst - a landscape and subsurface formed as a
result of dissolution of bedrock (usually, but not
necessarily, limestone and/or dolomite) and
characterized by distinctive subsurface hydrology
that includes flow of water through caves. The
landscape may include sinkholes, springs, and
sinking streams, but these features are not
essential to the definition and should not be
confused with the karst itself.
karst aquifer - an aquifer in which flow of water
is or can be appreciable through one or more of the
following: joints, faults, bedding planes and
cavities, any or all of which have been enlarged by
dissolution of bedrock.
Each of these definitions is suitable for incorporation in
statutes, regulations, and revisions thereof. We recommend
their use.
649
-------
BIOGRAPHICAL SKETCHES
Gary A. Davis is an environmental attorney who has been
practicing throughout the United States for more than eight years.
He is also the Director of the Clean Products and Clean
Technologies Program at the University of Tennessee Energy,
Environment, and resources Center. He was previously a hazardous
waste policy advisor in the California Governor's Office and worked
as an environmental engineer with a major consulting firm. Mr.
Davis received a B.S. in Chemical Engineering from the University
of Cincinnati, and a J.D. from the University of Tennessee College
of Law.
Gary A. Davis
Attorney at Law
Two Centre Square
625 S. Gay Street
Suite 230
Knoxville, TN 37902
(615)637-7725
Dr. James F. Quinlan, P.G., is president of Quinlan &
Associates, a consulting firm specializing in problems of carbonate
terranes. He was Research Geologist for the National Park Service
at Mammoth Cave, Kentucky, for 16 years and has been an independent
consultant on karst for more than 10 years. He earned a Ph.D. in
geology at The University of Texas at Austin (1978). His
experience includes 36 years of research and observations in karst
terranes of 26 states and 25 countries, and work as a consultant in
many of them. He has written or co-written more than 170
publications on karst-related topics. He is co-chairman of ASTM
Subcommittee D18.21.09 on Special Problems of Monitoring in Karst
and Fractured Rock Terranes.
Dr. James F. Quinlan
Quinlan & Associates, Inc.
Box 110539
Nashville, TN 37222-0539
(615)833-4324
650
-------
Session IX:
Ground Water Management
-------
652
-------
KARST GEOLOGY AND GROUND WATER PROTECTION LAW
By: Joseph A. Fischer, Geoscience Services, and
Robert J. Canace and
Donald H. Monteverde, NJ Geological Survey, Dept. of
Environmental Protection & Energy
ABSTRACT
Clinton Township, Hunterdon County, New Jersey is situat-
ed partially in a Paleozoic outlier within the New Jersey
Highlands Physiographic Province. Some 15% to 20% of the
Township is underlain by solution-prone, folded and faulted
carbonate rocks of Cambrian and Ordovician age. Faulting of
possible Taconic, Alleghanian, and Triassic age has deformed
these carbonates. The Flemington Fault, part of a major bor-
der fault system, traverses the Township.
Clinton is located at the intersection of one of New Jer-
sey's principal east-west interstate highways and a major
north-south State highway. Its rural environment and ready
access to major highways places Clinton in a growth corridor.
The State Planning Commission has recently identified the
Township as a regional rural growth center.
Two state reservoirs extend onto Township lands, one un-
derlain by carbonate bedrock. The construction of the Spruce
Run Reservoir provided a great amount of geologic information
on the local carbonate deposits, including the nature and ex-
tent of folding, faulting, and solution-channel development.
Extensive mapping and exploratory drilling permitted the com-
pilation of the occurrence of voids in specific geologic
units, and a better definition of the relationship between
solutioning and faults.
Enlightened Township officials introduced an ordinance
governing construction in areas underlain by carbonate rock
and those areas draining into carbonate lands. The Ordinance
has been in effect since the spring of 1988. Although origi-
nally met with great protestation by development interests, it
has proven to be a rational, working process that is presently
accepted by both the public and private sector.
653
-------
The ordinance requires a multi-phased data collection ef-
fort at various stages of a project. The legality and validi-
ty of this Ordinance have been tested and upheld in the court
room.
INTRODUCTION
New Jersey is well known for its number of Super Fund
sites and its rather stringent State regulations in relation
to discharges into ground water and surface water bodies, as
well as through injection wells. However, the only State reg-
ulations that give consideration to the problems of siting a
facility in karst areas are provisos not to place septic sys-
tems in sinkholes, and to establish trend directions of solu-
tion channels in carbonate rocks and sinkholes for landfill
sites.
There are five counties in the northwestern corner of the
state that are underlain by major deposits of solutioned car-
bonates (Figure 1). However, there are no county regulations
that address karst related concerns and the municipal offi-
cials of these counties have little guidance in either recog-
nizing or dealing with karst-related concerns. In addition,
the potential for ground water pollution from agricultural
activities has also been neglected.
At one time, these valley lands were primarily either ag-
ricultural in nature or merely bucolic scenes. However, re-
cent population growth in these areas is rapidly changing the
use of this prime real estate. Ground water extracted from
the high yield carbonate aquifers is the primary source of
domestic and industrial water in this region. Reservoirs,
water treatment facilities, shopping centers, office build-
ings, and corporate centers have been built above a variety of
carbonate terrane, sometimes supporting the structures atop
large (now, usually cement-filled) cavities. Some have been
well-investigated and engineered while others have not.
The one municipality that has recognized these problems
is Clinton Township, a State targeted growth area in Hunterdon
County, New Jersey. Clinton Township is located at the inter-
section of a principal east-west interstate highway and a ma-
jor north-south State highway, less than 50 miles from New
York City. Some 10 years ago it was a rural, primarily agri-
cultural community. At the completion of the interstate high-
way, the area became the target of both State and private de-
velopment planning. The Township's complex geology (Figure
2), includes solutioned carbonates under some 20% of its'
lands. Currently, wells are the primary source of domestic and
industrial water supply for the Township. Development pres-
sures, led some perceptive municipal officials to pass New
Jersey's first "limestone" Ordinance to protect the ground wa-
ter supply in this environmentally sensitive setting.
654
-------
T4eIO'
T4°00'
(SCALE IN MILES)
KNOWN CARBONATE ROCK DEPOSIT
AREA OF SUSPECTED
CARBONATE ROCK DEPOSIT
Figure 1. Carbonate rock deposits of northern New Jersey.
The mayor, an articulate, aware, and strong individual,
led the struggle to secure a rational approach to the Town-
ship's development processes. The New Jersey Geological Sur-
vey encouraged the ordinance and provided technical guidance
and a simplified geologic map identifying the general location
of carbonate bedrock. Previous State funded geologic informa-
tion provided a rational data base for the Ordinance and its
implementation. Problem recognition, conceptual planning, and
Ordinance formulation and implementation was a relatively
swift process.
In brief, the Township, faced with increasing development
pressures in 1987, placed a 150 day moratorium on development
and formed a committee of lay and technical people to draft an
ordinance that would:
655
-------
PCu
Wllloughby Brook
Fault
LEGEND
JTru - Mesozoic rocks, undivided
Ojtu - Jutland sequence, undivided
Oj - Jacksonburg Limestone
Ow - Wantage sequence
Kirtatinny Supergroup
Obu - Beekmantown Group, upper part
Obi - Beekmantown Group, lower part
OCa - Allentown Dolomite
Cl - Leithsville Formation
Ch - Hardyston Quartzite
PCu - Proterozoic, undivided
I. !,:?;'.• j- carbonate rock
--thrust fault
-- normal fault
QCa
MILES
Figure 2. Geology of Spruce Run Reservoir and adjacent areas
of Clinton Township, Hunterdon County, New Jersey.
1. Protect ground water in the carbonate rock areas.
2. Address public safety in relation to structural integ-
rity -
3. Allow conscientious development.
The Ordinance was passed by the Town Council (although
not unanimously) in May, 1988. Since that time, some 30 proj-
ects have been reviewed by the Township's geotechnical con-
sultant (GTC) under the provisions of the Ordinance. The re-
sults of the review process have included the abandonment of
projects in the exploration phase (even prior to the comple-
tion of the field explorations), the completion of the project
as envisioned by the designers (with little to no changes),
and the completion of projects that incorporated numerous
changes from the initial design, some during construction.
The Ordinance has been tested in court by one of the de-
velopers affected by the moratorium. In a broad-based suit
against the Township, the Ordinance was subject to legal scru-
656
-------
tiny, was upheld, and found to be a most valuable document for
achieving the Township's desire to insure that development is
conducted intelligently.
The Township was fortunate in that geologic investiga-
tions both before and during the construction of two large
State-funded reservoirs provided an extensive geologic data
base as well as fueling public awareness of karst concerns.
It is difficult to visualize the Ordinance coming into being,
and its successful implementation, without the State supplied
knowledge gained through the reservoir studies and the ongoing
interest in carbonate geology of the State Survey.
TOWNSHIP GEOLOGY
The carbonate bedrock in Clinton Township consists of
rocks of the Kittatinny Supergroup (Drake and Lyttle, 1980) of
Cambrian through Middle Ordovician age, the Middle Ordovician
Jacksonsburg Limestone (Kummel, 1940), and the Lower to Middle
Ordovician Jutland sequence (Jutland Member of the Martinsburg
Shale of Markewicz, 1967). The geology of the Clinton area
was mapped by Markewicz (1967), and is currently being revised
by Markewicz, Monteverde, and Volkert of the New Jersey Geo-
logical Survey. The geologic map provided to Clinton Township
by the New Jersey Geological Survey, which has been updated in
Figure 2, is a simplified version of the current geologic in-
terpretation of the area. Data derived as a result of geo-
technical investigations carried out under the ordinance has
resulted in some modifications to the initial mapping.
As indicated above, geologic findings associated with
mapping and subsurface exploration for the proposed Spruce Run
Reservoir (Markewicz 1958-1961) provided insight into the de-
gree of karstification of the carbonates in Clinton. Geologic
mapping, field identification of sinkholes, seismic data, and
numerous test borings revealed the presence of voids both at
the surface and in the subsurface, indicating the potential
for future subsidence. The reservoir was constructed after a
program to grout any voids at dam locations and the installa-
tion of a peripheral grout curtain to prevent leakage at the
reservoir's southeastern margin. Since construction, several
sinkholes have opened adjacent to the reservoir, requiring re-
mediation.
The degree of karstification of the various carbonate
bedrock units found in Clinton, as elsewhere, is a function of
lithology and proximity to faults, folds and other geologic
structures. While most carbonates have a potential for karst-
ification, regional studies suggest that geologic units that
contain coarse-grained dolomite or fine-grained limestone are
most susceptible to weathering (Rauch and White, 1970, Siddi-
qui and Parizek, 1969, and Dalton and Markewicz, 1972). So-
lution-channel development in the Kittatinny carbonates ap-
657
-------
pears to be particularly accentuated in units that contain
thin-bedded, coarse-grained dolomite.
Table 1 presents a summary of exploratory drilling data
for the Spruce Run Reservoir project. Although the number of
drill holes installed and total penetration varied among the
geologic units present, the data tends to confirm that specif-
ic units may be more prone to cavity development than others.
Figure 3, based on this drill data, presents a visual summary
Drill Data
Geologic Unit
Leithsville
Allentown Lower Upper
Beekmantown Beekmantown
Limeport Upper
Member Allentown
Member
Number 9 32 33 31
of
Borings
Total 1,035 4,574 2,770 3,852
Footage
Drilled
Number 13 108 19 80
of
Cavities
Average 80 54(*) 48
Footage
Between
Cavities
66
6,428
90
71
Total 67
Footage
of void
Penetrated
Voids as 6.5
Percentage
of Total
Depth
Penetrated
Voids as 9.4
a Percentage
of Bedrock
Penetrated
319
66
225
7.0
2.4
5.8
8.3
2.7
6.6
263
4.1
4.6
Table 1. Occurrence of voids in borings for Spruce Run Reser-
voir (Dalton and Markewicz, 1972, Table 4).
* - Indicates value for formation as a whole.
658
-------
Upper Allentown Member
Ground
Surface \
Limeport Member
KEY
Subsurface Boring
Overburden
Void in Dolomite Bedrock
/ Member Contact
20 0 20 40
vertical
scale
feet
Figure 3. Subsurface profile of a portion of the Spruce Run
Reservoir dam site, showing the contrast in intensity of
weathering between the Limeport and Upper Allentown Members of
the Allentown Formation (Markewicz, 1958-1961).
of the contrast in the occurrence of voids between the lower
(Limeport) member and the upper (Upper Allentown) member of
the Allentown Dolomite.
Another example of the presence of concentrated solution-
ing in the Spruce Run area is illustrated by borings drilled
into the lower Beekmantown Group of carbonates in which 6
borings were installed within 45 feet of one another, to
depths ranging from 72 to 142 feet. These borings revealed
voids in the bedrock occupying between 14 and 52 percent of
the total footage drilled for any one of the borings. In ad-
659
-------
dition, 100 percent of the voids in this interval occur within
the first 45 feet of rock penetrated, signifying that many
near-surface voids are present that could contribute to the
formation of sinkholes at the surface.
Faults involving the carbonate units consist of imbricate
thrusts, tear faults transverse to the structural grain, and
normal faults most likely associated with both rifting in the
adjacent Newark rift basin and with earlier Appalachian de-
formation. The reservoir borings showed that cavities are of-
ten well developed in the vicinity of faults (Figure 4) that
display brittle deformation. Fault locations were interpreted
from mapping of the area and rock core displaying fracturing,
recrystallization and other evidence of deformation.
An unusual occurrence of limestone is found in the Clin-
ton area, in rocks of the Jutland sequence (Monteverde, unpub-
lished). A test well drilled for Clinton, into this unit, in
proximity to a major east-west fault, yielded nearly 400 gal-
lons of water per minute, attesting to the well-developed
permeability of this limestone. Additionally, test borings
performed in the Jutland limestone for a proposed and subse-
quently constructed golf course/residential community revealed
the persistent presence of small voids and weathered zones
throughout the limestone section.
NATURE OF THE ORDINANCE
There are a number of ordinances that have been passed
for locales underlain by the solutioned carbonates found in
the Appalachian valleys of the eastern United States. These
include ordinances which essentially require the preservation
of present surface water flows into existing sinkholes, or
those that establish performance standards in relation to
known or suspected features. From the authors' reviews of
various eastern U.S. "limestone" ordinances, many different
tacks have been taken. In the valleys of the southern Appala-
chians, where development pressures are less than in the more
northerly valleys, it seems that the primary direction of all
ordinances has been toward preserving sinkhole integrity as a
means of handling storm water flows without deleterious local
flooding (e.g., Clarke County, Virginia). The secondary con-
sideration seems to be to stop structures from being built
directly over existing sinkholes.
In the northeast, some Townships have apparently tried to
essentially eliminate growth in carbonate areas. For example,
the Bucks County, Pennsylvania "Model Ordinance" requires that
storm water management basins and "principal or accessory"
buildings will not be located within 100 feet of features,
which include; sinkholes, closed depressions, lineaments,
fracture traces, caverns, ghost lakes, and disappearing
streams.
660
-------
26CH
200-
180 H
160 H
120 H
Ground Surface
Top of Rock
Ground Surface
o-f Rock
Scale = 20 Vertical
20 Horizontal
OS] - Overburden
IE-Cowity (Both filled and open)
I | - LJme»tone
Figure 4. Schematic interpretation based on drill data of the
occurrence of interconnected voids in the Allentown
Dolomite in relation to the trace of a transverse
fault.
661
-------
It is obvious what an imaginative aerial photograph in-
terpreter could do to any project in a municipality which has
passed an ordinance in accordance with the Bucks County recom-
mendations. In addition, the authors have little understand
ing of what the magic is in the 100 foot dimension.
The route chosen by the Clinton Ordinance Committee was
completely different and is still believed to be unique. The
Ordinance was intended to help educate and provide guidance to
a Planning Board applicant concerning the performance of a
phased investigation and evaluation of the subsurface condi-
tions at the development site. The investigation and evalua-
tion process was derived from the concepts discussed in Fisch-
er, et al (1987), and were developed from experience gained in
numerous site investigations and facilities construction in
karst.
A phased approach to the Ordinance requirements was in-
stituted to allow an applicant to define site-related flaws
and/or prepare preliminary cost estimates at an early stage of
the development without incurring significant economic penal-
ties. Judgement is used in defining the extent and nature of
an investigation. Different concepts can be used for a pro-
posed two-house subdivision than might be employed for a 59
acre, multi-use development.
The Ordinance encourages the applicant's geotechnical
consultant to utilize any combination of investigative tools
and techniques that are deemed reasonable, but with certain
minimum requirements such as; reviewing the available geologic
information, using aerial photography, and providing hard data
(i.e., borings, test pits, etc.) on the site subsurface. The
GTC's experience within the Township was made available to the
applicant on an as-requested basis. The results of the ex-
plorations are submitted to the GTC for review and approval.
Thus, a reasonable degree of consistency in the explorations
performed by the various applicants has developed and any use-
ful information produced at one site can be fed into future
planning processes. The process also provides for the appli-
cant to perform only the amount of work necessary for the ac-
tual site conditions, i.e., the more complex the site is, the
more detailed the exploration and evaluation is.
The subsurface data obtained is interpreted by the appli-
cant's geotechnical consultant and used to develop planning
and design solutions to the geotechnical problems that exist
at the site. Again, the GTC reviews and evaluates the subsur-
face model proposed for the site and suggested solutions to
any subsurface-created concerns. The applicant is encouraged
to use innovative engineering solutions and, in general, no
set procedures are imposed upon the planning/design team. Fi-
nal reviews and discussions are held with the Planning Board,
the applicant's design team, and the GTC prior to giving con-
ceptual approval to, or conversely, rejection of, the planned
662
-------
development.
As a result of the difficulty in performing a subsurface
investigation and reliably interpreting the results for a
karst site in any reasonably economic manner, the Ordinance
provides for field inspection during construction in key areas
by the applicant's geotechnical consultant as well as spot
checks by Township forces. In enforcing the Ordinance, it is
expected that changes in design and construction procedures
can and will be made during the building phase.
We believe that the investigation, evaluation, and re-
porting costs required as a result of the existence of the Or-
dinance are reasonably consistent with what a prudent develop-
er would ordinarily budget. This is not surprising as the Or-
dinance procedures are predicated upon concepts developed from
commercial, not research projects. The separate cost of the
GTC review are prepaid into an escrow account. The costs were
estimated at $1,000 plus an additional $500/acre. The GTC re-
view costs have usually been less than these initial require-
ments and only in one instance did the somewhat insensitive
actions of an applicant require replenishment of the escrow
fund.
EXAMPLES
Several examples of the different manners used to attack
the issues can be used to illustrate some of the workings of
the Clinton Township "Limestone" Ordinance.
Case I - The site was first investigated by a large national
consulting firm during the building moratorium. Initially, no
geotechnical study was envisioned. The firm (with the assist-
ance of an outside geologic consultant) provided a "Geotechni-
cal Feasibility Study" after the development complex plans for
the 46 acre site had been fully formulated. The problems of
force-fitting the investigative results to an existing layout
became apparent. The existence of several sinkholes were min-
imized and none (including one elongated elliptical sinkhole
some 25 feet in length) were considered "major" by the devel-
oper's consultant. A test pit excavated for the field in-
vestigation became a sinkhole. Faults crossing the site and
their likely effect on solutioning were ignored. Only the
more favorable portions of the work performed for the nearby
reservoir were included. The most critical structures (serv-
ice buildings and office towers) were located atop what was
likely the most solutioned rocks at the site, the Leithsville
Formation. An unlined, ornamental lake was sited over a line
of small sinkholes. The underlying dolomites were impermeable
in one section of the report but provided a good ground water
resource in another. Rock Quality Designation (RQD) data was
misapplied. The significance of drilling water, core, and
sample losses were not understood.
663
-------
These and other factors eventually resulted in an agree-
ment to reconfigure the site and perform additional studies in
conformance with the Ordinance. No work has been accomplished
at this site since the agreement was made in 1990. As the
business climate has been poor since that time, there is no
way of telling whether the owner's eventual realization of the
site subsurface deficiencies or the business climate forced
the current abandonment of development plans.
Case II - Some 10 years ago a large international organization
constructed a major research facility upon a small area of a
roughly 100 acre parcel. A significant portion of the proper-
ty is underlain by solutioned carbonates, but apparently the
karst-related problems were not well recognized prior to, or
even during, the site investigation. Two large national firms
provided geotechnical consulting services. Cost overruns in
the millions of dollars were experienced in both investigation
and foundation construction. Untold quantities of cement were
pumped into unexpected cavities and a complicated foundation
system was installed. A spray irrigation field for waste wa-
ter disposal was placed above shallow, solutioned limestone.
Sinkholes have developed in this field.
The remainder of the site is currently planned for multi-
use development by single and multi-family homes, offices,
stores, open-space and recreational facilities. A two-phased
field investigation has recently been completed by another
well-known engineering firm. The work was planned and accom-
plished in complete cooperation with the Township GTC. Per-
sonnel experienced in carbonate rock studies were used both in
the field portion of the work and in the report preparation.
The owners design team is currently using the results of the
phased explorations investigations, which cover site areas
that are underlain by solutioned carbonates, competent meta-
morphics, and apparently fractured and faulted sediments, to
prepare a conceptual layout. The information provided to the
design team has blended the experience of both the consulting
firm and the Township GTC.
Case III - A caveat emptor situation developed in a proposed
single-family development that was investigated under the re-
quirements of the Ordinance. The applicant's geotechnical
consultant and the GTC came to an agreement on layout and de-
sign of structures, wells, and septic systems that incorporat-
ed the results of the applicant's studies. The revised con-
ceptual plans were approved by the Planning Board, with the
standard caveat of the need to provide knowledgeable construc-
tion inspection.
The site was sold to a new developer. The approved plans
were provided to the buyer but apparently without the several
geotechnical reports completed by the original applicant.
Site work was initiated, then abandoned by the new owner after
the first foundation wall failed as the result of the opening
664
-------
of a post-construction sinkhole beneath it.
CONCLUSION
The many fears in relation to ground water protection
that both the lay and technical person should have, or in many
instances do have, can only be exacerbated by the presence of
solutioned carbonate rocks below the site development. The
technical information needed to address concerns related to
karst is available to the interested parties in any develop-
ment exercise. However, the awareness of the need for techni-
cal expertise is certainly not wide-spread. Some communities
have undertaken to express their concerns with ordinances.
Drill data, the mapping of caves, and field observations
concerning the distribution of sinkholes, springs, and disap-
pearing streams all indicate that certain units within the
Kittatinny Supergroup display a higher degree of karst devel-
opment than other carbonate rocks. Some units are particular-
ly prone to solutioning. This concentration of karst features
in specific geologic units was confirmed by the borings in-
stalled for the Spruce Run Reservoir in Clinton Township.
Therefore, in planning and carrying out site investigations
the geotechnical expert should keep in mind the likelihood and
potential size of karst features as a function of the geologic
unit(s) found on the site. An understanding of the probable
severity of karstification at a development site can help dir-
ect the intensity of the exploratory program necessary for a
suitable site investigation.
Despite the correlations that probably can be made be-
tween specific geologic units and structures and the potential
for karst features, a decision was made to treat all carbonate
bedrock in the township equally when enforcing the ordinance.
This decision was made because of the ultimately unpredictable
nature of subsurface voids and for the sake of equal treatment
under the law. This approach is valid in view of the need to
investigate for voids on a site-specific basis, where an un-
derstanding of the location of structures relative to voids is
vital.
The authors believe a Clinton Township type "limestone"
Ordinance which draws upon the available geologic information
to establish guidelines, yet allows development to occur in
karst areas is an appropriate solution in most instances.
Laws should not be aimed at eliminating development or putting
the developer into a strait jacket, but should incorporate ap-
propriate provisions for insuring the protection of ground wa-
ter resources and the safety of the public while allowing for
ingenuity in design and construction.
There are, undoubtedly, economic penalties that result
from the need to obtain adequate geologic information at a
665
-------
site and to provide appropriate engineering solutions. How-
ever, these economic penalties are relatively small in con-
trast to the penalties that can occur as a result of failures
of both the infrastructure and any structures located atop so-
lutioned carbonates, during or after construction when likely,
but unforeseen problems surface.
REFERENCES
Bucks County, Pennsylvania, undated, Carbonate Geologic Areas
Model Ordinance Amendment.
Clarke County, Virginia, undated, Clarke County Ground Water
Protection Plan.
Dalton, Richard and F.J. Markewicz, 1972, Stratigraphy of and
Characteristics of Cavern Development in the Carbonate Rocks
of New Jersey, Bui. of the Nat. Speleological Soc., v. 34, no.
4, p. 115-128.
Drake, Avery A. and P.E. Lyttle, 1980, Alleghanian Thrust
Faults in the Kittatinny Valley, New Jersey, Field Studies of
NJ Geology and Guide to Field Trips, 52nd Annual Meeting of
the NY State Geological Assoc., Warren Manspeizer, ed., p.
92-112.
Fischer, Joseph A., R.W. Greene, R.S. Ottoson, T.C. Graham,
1987, Planning and Design Considerations in Karst Terrain,
Env. Geol. Water Sci., v- 12, no. 2, 123-128.
Kummel, H.B., 1940, The Geology of New Jersey, NJ Dept. of
Conservation and Economic Development, NJ Geological Survey
Bui. No. 50, 203 p.
Markewicz, Frank J. , 1958-1961, Field Notes, Haps, and Con-
struction Profiles of the Spruce Run Reservoir Site, NJ Bureau
of Geology and Topography, unpub.
Rauch, Henry W. and W.B. White, 1970, Lithologic Controls on
the Development of Solution Porosity in Carbonate Aquifers,
NJ Water Resources Research, v. 6, no. 4, p. 1175-1192.
Siddiqui, Shamsul H. and R.R. Parizek, 1969, Hydrogeologic
Factors Influencing Well Yields in Folded and Faulted Carbon-
ate Rocks, Central Pennsylvania, (abs.) American Geophysical
Union Trans., v. 50, no. 4, p. 154.
, 1967, Geology of the High Bridge Quadrangle, NJ
Bureau of Geology and Topography, Bull. 69, 139 p., unpub.
666
-------
KARST GEOLOGY AND GROUND WATER PROTECTION LAW
By: Joseph A. Fischer, Robert J. Canace,
and Donald H. Monteverde
Question
1. What cities, counties, and states have ordinances near
or similar to the one you describe?
None. We know of no State ordinances of this nature. We
also unaware of any Counties or Cities with such an ordi-
nance, although our knowledge on an interstate, local level
is obviously limited. However, a Pennsylvania legislature's
staff member informed Clinton Township that the Commonwealth
was attempting to prepare a law and that the Township's or-
dinance appeared to be the best available based on what they
had found.
2. To what extent are other governmental units deciding to
emulate the ordinance you describe and adopt a "locally-
suited" similar ordinance?
There are currently three to four New Jersey municipalities
considering a similar ordinance. A model ordinance is being
prepared by a "limestone committee" of the New Jersey Re-
source Conservation and Development organization (NJRC&D),
the northern New Jersey branch of a nation-wide, county/
USDA sponsored organization. Information concerning the
NJRC&D model ordinance will be provided to numerous local
municipal and private planning organizations in a series of
spring (1992) meetings. The RC&D, nation-wide, will also be
made aware of this and other NJRC&D work in relation to car-
bonate rock concerns. In addition, a number of inquiries
from State and County level individuals, concerning the or-
dinance, were received by the authors both at the confer-
ence and subsequently.
667
-------
668
-------
ANALYSIS OF DRASTIC AND WELLHEAD PROTECTION METHODS
APPLIED TO A KARST SETTING
Lyle V. A. Sendlein
Department of Geological Sciences
University of Kentucky
Lexington, Kentucky
ABSTRACT
DRASTIC and Wellhead Protection methodologies were used in
the analysis of an Inner Bluegrass Karst Region karst spring used
as a municipal water supply. The area has been studied by
Thrailkill and his students (Thrailkill et. al., 1982). Through
dye tracing and field analysis, the area has been divided into
groundwater basins and interbasin areas. Groundwater basins are
characterized by numerous sinkholes, conduit flow and deep water
table (33 meters). Interbasin areas are characterized by shallow
water table, high level (upland) springs, fracture flow and
sinkholes. Existing information was utilized in the DRASTIC
analysis with inconsistent results due to the dichotomy of a deep
water table found in the groundwater basins. Even though high
values were used for most parameters, the deep water table
reduced the index value, thus minimizing the potential impact.
The Wellhead Protection analysis and the delineation of a
Wellhead Protection Area (WHPA) provides a more direct manner for
the identification of the area that needs to be protected. The
definition of the area based on Thrailkill's concept of a ground-
water basin as defined by dye tracing provides a good planning
tool. An important fact that must be dealt with in the planning
phase is the 80 percent of the WHPA for the municipal water
supply lies outside the jurisdiction of the community trying to
protect its water supply. The adjacent community is currently
addressing a planning issue that involves the WHPA because a
major development is proposed for a part of the WHPA.
BACKGROUND
This analysis is based on data gathered over the last three
669
-------
years on contracts supported by the Divisions of Water and Waste
Management of the Natural Resources and Environmental Protection
Cabinet of Kentucky, the United States Environmental Protection
Agency and the University of Kentucky's Institute for Mining and
Minerals Research.
DRASTIC maps were developed for six areas in Kentucky to
test the feasibility of producing such maps from data available
in existing data bases, files and published maps and reports.
Areas were chosen for each of the major physiographic regions of
the state including cities whose public water supply depended on
groundwater as all or part of the source. (Sendlein, 1989)
Wellhead protection work was conducted for three of the
areas for which DRASTIC maps had been developed. This was a
pilot study and because of the limited funds, complete analysis
was not possible. In fact, the study was also based on available
data and consisted of construction of the Wellhead Protection
Area maps for each area and the compilation of sources of poten-
tial contamination from state and federal files. Each area
included spot field inspection of the groundwater water sources
and in one case a pumping test was attempted. (Sendlein and
Fitzmaurice, 1990)
Kentucky is very fortunate because the entire state is
covered by seven and a half minute USGS quadrangle maps including
topography and geology. In the construction of the DRASTIC maps,
the USGS maps were used as base maps, and each area studied
included two adjacent sheets, except one of the areas was
composed of four adjacent sheets. Because the geology is know at
this level of detail, the construction of general environmental
maps such as DRASTIC maps, does not add much information for an
area that a qualified geologist could not glean from the geologic
maps. However, the non-geologist lay person can benefit from the
conclusions reached as a result of the DRASTIC analysis.
Considerable information is present in the files of the
Divisions of Water and Waste Management in Frankfort, Kentucky.
These data are scattered throughout the agency's various offices
with permits kept in a central file room. Much of the tabular
information is kept in computer files, but most maps are on
paper. An exception to this is the map information kept in the
Kentucky Natural Resources Information System (Croswell et al.,
1982) utilizing the ARCINFO software. The scale of the
information included in this database is on the order of
1:1,000,000, considerably larger than the 1:24,000 of the
quadrangle maps used in the present studies. The DRASTIC maps
were developed on the computers in Frankfort and are stored as
files in their system. (Couch, 1988)
A recent water well law in Kentucky requires that all water
wells drilled in the state be recorded and a driller's log be
filed with the Division of Water. These logs, along with other
bore hole information, are being compiled into a database by the
Kentucky Geological Survey and will be an important database when
it comes on line. Currently, the paper files can be reviewed in
the Division of Water and a computer printout of some of the
parameters are available from that office.
670
-------
STATEMENT OF THE PROBLEM
A recent planning action of the Lexington-Fayette Urban
County Government (LFUCG) focused attention on Royal Spring, the
municipal water supply for the community of Georgetown, Kentucky.
Royal Spring's flow ranges between 30 1/s (liters/second) on the
low side to over 3,000 1/s on the high side with median flow
between 300 and 1,000 1/s (Thrailkill et al, 1982). The LFUCG as
part of the comprehensive planning process identified an area for
special consideration for which the Mayor appointed the Coldstr-
eam Small Area Planning Committee in April of 1990. This area
lies in the northern part of the city and has a large tract of
land owned by the University of Kentucky and is used for
agricultural research (Figure 1). The property at one time was
considered to be in the country but now with the addition of 1-75
and increased development in this part of the city, there are
several reasons why land use revisions should be discussed.
As part of the deliberations of the Committee and the
planning staff, the fact that this tract of land was part of the
recharge zone for Royal Spring became known and was discussed at
some length. The final report of the Committee includes a
section that addresses this natural feature.
This location was an area included in one of four quadran-
gles mapped using the DRASTIC methodology. Likewise, the Royal
FIGURE 1
COLDSTREAJ1 FARM STUDY AREA LOCATION
COLDSTREAn FARM
STUDY AREA
671
-------
Spring was identified as a public water supply to be included in
the Pilot Wellhead Protection Study of Kentucky. The availabili-
ty of these maps and accompanying reports provided an opportunity
to analyze the utilization of these data in the planning process.
DRASTIC MAPS AND RESULTS
Four quadrangles were mapped, Lexington East (MacQuown and
Dobrovolny, 1968), Lexington West (Miller, 1967), Georgetown
(Cressman, 1967) and Centerville, (Kanizay and Cressman, 1967)
using the methodology defined by Aller (1987) . A combined
version of these four quadrangles are presented in Figure 2. It
can be seen that DRASTIC Index values range from a low range of
100-119 to a high of 200+. Most of the background of the maps
fall into the Index category of 180-199. These are considered
high values and indicate a high potential for groundwater
contamination. In constructing these maps, the work of John
Thrailkill and his students (1982) was used because they
identified areas drained by individual major springs as
groundwater basins through dye tracing studies. These basins can
be seen on the DRASTIC map as fat, teardrop-shaped areas with
index values less than the areas surrounding them (160-179). The
other areas (referred to as "background" above) of the map fall
into what Thrailkill has identified as interbasin areas.
This difference in values was expected between groundwater
basins and interbasin areas, but the fact that the groundwater
basin has a lower DRASTIC Index is just the opposite relative to
its vulnerability for contamination. Within the groundwater
basin, there are more sinkholes and greater opportunity for
surface water to move from the surface to the conduit below.
This is an important defect in the DRASTIC method as applied to
karst areas. The difference is produced because Thrailkill has
shown that in the groundwater basins the conduit is the main
drain and the water table is deep, at conduit level. In the
DRASTIC method, deep water table is given less value because of
the rationale that the deeper the water table, the less likely it
will be contaminated by surface pollutants.
Royal Spring's groundwater basin is elongate and extends
southward from the city of Georgetown. It is not represented by
only one DRASTIC Index value but is divided into two regions with
the higher index value (200+) in the southern part of the basin
and the other region (160-179) occurring near the spring in the
northern part of the basin. The difference between index values
is controlled by the difference in surface elevation. Most of
the groundwater basin underlies the surface valley of Cane Run,
but in the lower region of the basin, closer to Georgetown and
the spring, the groundwater basin does not underlie the Cane Run
valley, because the valley abruptly turns toward the west. The
lower end of the groundwater basin lies beneath a higher
topographic surface, thus the deeper water table and lower
DRASTIC Index value.
672
-------
FIGURE 2
ROYAL SPRINSS - DRASTIC INDEX MAP
STUDY AREA
RDADS -
STREAKS
CITIES [r/ I
FIGURE 3
ROYAL SPRINGS - WELLHEAD PROTECTION AREA
B4'37'30"
84'22'30"
-------
WELLHEAD PROTECTION FOR ROYAL SPRING
As part of the pilot study of Kentucky, Royal Spring was
identified as a municipal water supply to be studied. The study
included defining the Wellhead Protection Area (WHPA) and
identifying potential sources for contamination of the aquifer.
Figure 3 shows the Wellhead Protection Area for Royal Spring.
Using U.S. Environmental Protection Agency terminology, the Zone
Of Contribution (ZOC) for this area is equal to the recharge area
to the spring. A second zone has been defined where the Time Of
Travel (TOT) is measured in hours and days rather than months and
years common in non-karst regions.
The boundaries of the area were determined by using
Thrailkill's approach to not only include the groundwater basin
but also the "catchment area" for the groundwater basin. By
reference to Figure 2, one can see the other groundwater basins
that bounded the Royal Spring groundwater basin. The area
bounded by the drainage divide for Cane Run represents the catch-
ment area. Two deviations from this occur, one at the southern
end of the WHPA where Russell Cave Spring robs part of the
drainage of Cane Run, and the other at the northern end of the
WHPA where Cane Run abruptly turns to the west and the groundwa-
ter basin no longer underlies the drainage basin of Cane Run
(Figure 2).
A zone (Zone 1) within the WHPA (Figure 3) that defines the
groundwater basin represents a more vulnerable region of the
recharge basin, because within this zone more than 150 openings
to the main conduit have been identified by Thrailkill and
include the swallets that have been used for tracing the basin.
Maximum travel times from the farthest swallet have been as low
as 141 hours, but calculations of flow velocities are in the
order of 4 km/hr, which would transfer to as little as four hours
of travel time for the Newtown Swallet that is approximately 15
kilometers from the spring. The remaining zone represents the
rest of the area within the ZOC and falls into the interbasin
area.
The other major part of the Wellhead Protection method is
the identification of potential contamination sources. Sources
identified and located on maps and tabulated for the Royal Spring
WHPA (Fitzmaurice, 1990) are: subsurface percolation from
sanitary systems, injection wells, landfills, open dumps
(including illegal dumping), residential (or local) disposal,
graveyards, aboveground storage tanks, underground storage tanks,
containers, pipelines, material transport and transfer,
operations, animal feeding operations, production and other
wells, monitoring and exploration wells, and salt-water
intrusion/brackish water upcoming. The information obtained on
these potential contamination sources varied for Fayette and
Scott Counties. Fayette County is the more populated of the two
counties and more potential sources are present and thus more
information available. The LFUCG has computerized much of their
information and have a program that converts street addresses to
latitude/longitude which made it possible to plot information on
maps. These data are included on maps in Fitzmaurice (1990).
674
-------
COLDSTREAM SMALL AREA HYDROGEOLOGY
The topography ranges from 980 feet at the southern end to
885 feet at the point where Cane Run leaves the property. The
floodplain of Cane Run is 910 feet at the southern end of the
site and 890 feet at the northern end, with a distance of
approximately 9,900 feet or a floodplain gradient of 20
feet/9,900 feet (0.002). The total area lies within the upper
region of the surface drainage basin of Cane Run. Presently,
developed areas (the two subdivisions and the industrial park)
are connected to the sanitary sewer system of LFUCG, but surface
run off is directed to the natural or constructed drainage ways
within the area.
Karst features within the area are two swallets (one
considered an estavella by Thrailkill), one active spring, one
inactive spring (tufa deposits observed at this spring), sinking
stream (most of Cane Run is dry except for major precipitation
events and spring flow terminates in one of the swallets), and
several sinkholes.
The geologic units present in this area are the various
members of the Lexington Limestone of Ordovician age. The
Lexington Limestone includes argillaceous units generally less
than 6 meters thick and highly variable. Within the area of
interest the lower unit of the Lexington Limestone is the Grier
Limestone and above that is the Tanglewood Limestone. Interbed-
ded within this unit are the argillaceous units, some include
interbedded shale.
Royal Spring emerges from the Grier Limestone Member about
five meters below the contact with the Tanglewood Limestone.
Within the area, one of the argillaceous units is the two meter-
thick Cane Run bed, which is mapped about 10 meters above the
Tanglewood-Grier contact. Thrailkill believes that the main
conduit for Royal Spring may be stratigraphically controlled
within this interval and that the Cane Run bed does not perch the
subsurface conduit because most or all of the dye traced swallets
within the groundwater basin penetrate the unit.
Surface flow in Cane Run is intermittent on the Coldstream
Small Area, but a part of the stream has been observed to flow
between precipitation events. The source of the water is a
spring shown on Figure 5. Because of the construction in the
area, it is not possible to determine if the source of the spring
water is related to the storm water system of LFUCG. This water
flows on the surface for a short distance entering Cane Run and
sinking in the channel at the swallet identified by Thrailkill as
the Coldstream Swallet (Figure 5). Near the point where Cane Run
leaves the Coldstream Small Area, discharge from the University
of Kentucky farm operation flows into the channel.
COLDSTREAM SMALL AREA PLAN
Land use of the 1770.9 acre area as listed in the 1988
Comprehensive Plan is shown in Table 1. The Coldstream Small
Area plan changes are also presented in Table 1.
675
-------
FIGURE 1
COLDSTREAtt SMALL AREA PLAN
LEGEND
STUDY BOUNDARY
PROPERTY LINES
FIGURE 5
COLDSTREAN SMALL AREA - TOPOGRAPHY
-------
TABLE 1
EXISTING AND PROPOSED LAND USE FOR COLDSTREAM SMALL AREA
LAND USE
LOW DENSITY RESIDENTIAL
MEDIUM DENSITY RESIDENTIAL
RETAIL TRADE t PERSONAL SERVICES
PROFESSIONAL SERVICES
HIGHWAY ORIENTED COMMERCIAL
SEMI-PUBLIC USES
PARKS
OTHER PUBLIC USES
OFFICE, INDUSTRIAL RESEARCH PARK
WAREHOUSE i WHOLESALE
CIRCULATION
UTILITIES & COMMUNICATIONS
TOTAL
1988 PLAN
EXISTING
117.6
0.0
0.6
27.8
3.8
5.6
16.8
984.9
146.8
54.4
2.2
1363.5
PROPOSED
69.0
307.5
9.8
22.9
0.0
0.0
0.0
0.0
0.0
0.0
0.0
407.4
TOTAL
186.6
307.5
10.4
50.7
3.8
5.6
16.8
984.9
146.8
57.4
2.2
1770.9
1991 PLAN
PROPOSED
178.5
139.9
25.5
61.0
0.0
0.0
0.0
235.3
781.3
0.0
0.0
0.0
1421.5
TOTAL
296.1
139.9
26.1
65.3
3.8
5.6
16.8
235.3
781.3
146.8
57.4
2.2
1770.9
Figure 4 identifies the major land use of the Coldstream Small
Area, and Figure 5 illustrates the topography and geological
features. The major change in the area will be the development
of the 984.9 acres, listed as Other Public Uses, owned by the
University of Kentucky- It is the University's intent to develop
a research park in this area. One site has already been devel-
oped by Hughes Aircraft on the property, and the University has
plans for similar high-technology industries to locate on the
property and be related to the University of Kentucky through
faculty and student interaction with the new industries.
The Committee has recommended the following changes to the
1988 Comprehensive Plan that are shown in Table 1. Major changes
are the addition of 178.5 low density residential, 139.9 medium
density residential, and the addition of a new land use
identified as Office, Industrial Research Park that would have
781.3 acres.
While the Committee Report acknowledges that much of the
Coldstream Small Area falls in the recharge area for Royal
Spring, no specific recommendations were made to protect the
spring from water quality changes that will be caused by the
increased development. The Georgetown Water Board presented to
the Committee the policies they plan to follow relative to
development in the recharge area under their jurisdiction (Table
2).
677
-------
TABLE 2
AQUIFER RECHARGE PROTECTION POLICIES FOR ROYAL SPRING
The Urban Service Boundary should not be extended into the Royal Spring
Aquifer recharge area for urban development.
No additional properties within the Aquifer Recharge Area should be
planned industrial.
Existing lands planned industrial but not yet zoned should be
redesignated •environmentally sensitive light industry". This category
does not allow hazardous materials users/generators.
For lands already zoned industrial, new hazardous materials users are
prohibited, and existing hazardous materials users may expand only if
the design and operation of the site would protect the aquifer from
contamination.
Design standards are given for safe storage and spill containment.
No new underground storage tanks should be located in the Aquifer
Recharge Area.
Rural residential development on septic systems should be clustered
outside the recharge area where feasible.
The recharge area has the highest priority in the proposed TDR/PDR
program.
The LFUCG is encouraged to adopt similar protection policies.
STATUS OF KNOWLEDGE OF THE SITE
The long term studies of John Thrailkill and his students
have identified the swallets that are directly linked to the
Royal Spring. Even though this information has been known for
the last ten years, very little practical application of the
knowledge toward planning resulted.
The DRASTIC analysis illustrates that the Coldstream Small
Area falls in the highest and second highest categories of the
DRASTIC Index indicating high vulnerability to groundwater
contamination. Although the study results are inconsistent on an
area basis because it did not highlight the most vulnerable areas
for most of the area studied, it did identify the most vulnerable
area that must be protected in the Coldstream Small Area (200+
index area).
The Wellhead Protection study identified the recharge area
(ZOC) for Royal Spring and based on previous work did identify
Zone 1 (200+ index which is a groundwater basin identified by
Thrailkill) as the most vulnerable area within the Wellhead
Protection Area that must be protected. Once Georgetown
officials became aware of this map and related study, they began
to include the information in their planning process. It was the
678
-------
WHPA map and Georgetown's desire to be included in the Coldstream
Small Area Planning Committee that brought the issue to the
table.
Systematic stream hydrographs have not been developed for
Cane Run or the segment flowing across the Coldstream Small Area.
Cane Run receives flow from a spring on the site and loses water
to two swallets and fractures in the bottom of the Cane Run
channel. Cane Run is an intermittent stream that will experience
increased discharge as development proceeds.
The Sinkhole Ordinance (described by Dinger and Rebhman,
1986) enacted in Fayette County is an important application of
geological knowledge but is not directed toward protection of
groundwater quality. In fact, the ordinance allows the direction
of surface runoff toward sinkholes but does exclude the construc-
tion of dwellings within them.
MANAGEMENT OPTIONS FOR THE COLDSTREAM SMALL AREA
The proposal to rezone this area has provided an opportunity
for the analysis of current laws and policies relative to the
protection of groundwater resources utilized as a municipal water
supply. While the city officials of Georgetown, Kentucky are
cognizant of the many potential sources of contamination to their
water supply and the fact that it is highly stressed during
periods of drought, other sources of water are being sought but
will not be in place in the near future, thus this supply is very
important to them. The problem is more complex because George-
town does not control the whole WHPA and therefore cannot enforce
land use restrictions to protect the aquifer.
State officials recognize the dilemma faced by Georgetown,
but feel they have no regulations that would allow them to effect
the development of a protection policy. They were able to inject
the State into a problem during the drought of 1988 and cause a
single user to stop pumping a large volume of water from the
conduit supplying Royal Spring, thus allowing the water to be
used for the citizens of Georgetown.
Few examples exist to use as a model for developing a
protection strategy for this area. Austin, Texas is currently
experimenting with two policies to protect a karst aquifer that
provides an important recreational spring (Butler, 1987) . East
Marlborough and.West Whiteland Townships, Chester County Pennsyl-
vania have adopted zoning ordinances to protect a carbonate
aquifer (Jaffe and Dinovo, 1987). Both of these utilize resource
protection and source control methods to protect the aquifer from
potential contamination.
As part of the Coldstream Small Area Plan, a green space
(resource protection) to roughly parallel the 100 year floodplain
for Cane Run on the property was proposed. This would be an area
where no construction other than utility lines and limited roads
cross the channel. This approach has been used in Austin, Texas
as a protection strategy for the recharge zones of the Edwards
Aquifer (a karst aquifer). They have established three zones as
illustrated in Figure 6 based on stream hydrograph and water
679
-------
quality data collected for the area. The Pennsylvania example
limits the number of dwelling structures per acre and controls
the source of potentially contaminating substances such as home
heating oil.
FIGURE 6
MAJDR WATERWAY
WATER QUALITY ZONES
200' MIN,
400' MAX,
300'
CRITICAL WATER QUALITY ZONE
WATER QUALITY BUEEER ZDNE
UPLANDS ZONE
Application of the water quality zones illustrated in Figure
6 to the Coldstream Small Area would be similar to the
designation of the green space that is under consideration at the
present time. The Critical Water Quality Zone (Figure 6) is a
zone that does not allow any improvement other than minor roads
but is generally used as a park area. The Water Quality Buffer
Zone is a zone with limited development activity. For example,
very low density residential or light industry with limited toxic
substances used by them. The Upland Zone does not have any
restrictions. If this were overlaid on the sinkhole ordinance in
the Coldstream Small Area, this would be a beginning of a
protection strategy for Royal Spring.
To make any zoning strategy cost effective, baseline
information on the physical properties within the area should be
collected before development takes place. This information
should include the following: 1) stream flow from all channels
providing water to Cane Run, 2) water entering and leaving the
Coldstream Small Area, 3) channel gain or loss within the Cane
Run channel on the property, and 4) water quality of water
entering Cane Run from all sources (springs, surface water
discharges from developed areas, and overland flow). Information
of flow conditions will allow for the exact location of the
boundaries of the green space.
680
-------
Likewise, consideration of the kinds of land uses that are
best suited for the area should be defined. It is understood
that sanitary sewers are to be provided for all occupants of the
new developments. Special consideration of the quantity and
quality of surface water leaving the developed properties must be
given so that precautions can be taken so that the quality of the
water entering Cane Run does not deteriorate.
CONCLUSIONS
The use of scientific information for the purpose of plan-
ning is not easy to document because it becomes part of the
policy-making process and this process is not generally document-
ed by retrospective analysis. In this case, Royal Spring is the
largest spring in the Commonwealth and is used as a municipal
water supply and the Kentucky Department of Environmental Protec-
tion and Natural Resources became interested in protection
strategies for the spring. Their support of the DRASTIC and
Wellhead Protection studies focused attention on the spring and
it is fortuitus that the LFUCG began to study the Coldstream
Small Area to develop a plan for rezoning the area within the
same time frame. From this study the following conclusions can
be made:
1. The Wellhead Protection analysis had more of an impact
on the planning process because it defined the Zone Of
Contribution for the Royal Spring.
2. The zone identified within the WHPA (Zone 1) is a very
vulnerable part of the ZOC and occupies a large portion
of the Coldstream Small Area Plan.
3. The previous work by Thrailkill, while known to exist
for some time, was not directly applied to the planning
process until the present time.
4. Because approximately 80% of the ZOC is outside the
jurisdiction of the Georgetown City Council, protection
strategies for the entire WHPA cannot be enforced.
5. There are no state or federal laws that give the Georg-
etown City Council authority to protect the ZOC outside
of Scott County.
6. Protection of the ZOC within the Coldstream Small Area
will depend on the building permits issued by the
LFUCG. The statement that LFUCG recognizes that the
portion of the Cane Run drainage basin on the property
is in recharge area for Royal Spring is a positive
sign, and their willingness to define a green space
along the stream channel that includes the two swallets
that are directly linked to Royal Spring is a form of
aquifer protection.
7. Baseline data must be collected to document the poten-
tial changes in the quantity and quality of water
flowing into Cane Run before, during and after develop-
ment if a protection strategy is to be cost effective.
681
-------
REFERENCES CITED
Aller, L., Bennett, T., Lehr, J. H., Petty, R. J., and Hackett,
G., 1987, DRASTIC: A Standardized system for evaluating
ground water pollution potential using hydrogeologic
settings: U. S. EPA-600/2-87-035, 1-455
Butler, Kent S., 1987, Urban Growth Management and Groundwater
Protection: Austin, Texas, in Planning For Groundwater
Protection, edited by G. William Page, Academic Press, New
York, pages 261-287
Couch, Amber, 1988, A DRASTIC Evaluation For Part Of The Inner
Bluegrass Karst Region, Kentucky: The Centerville,
Georgetown, Lexington East, Lexington West Quadrangles,
Unpublished MS Thesis, University of Kentucky, 95 pages
Cressman, Earle R.,1967, Geologic Map of the Georgetown
Quadrangle, Map GQ-605, USGS
Croswell, P. L., Sanders, S. L., Dryden, W., and Schneider, W.
L., 1982, User Guide to the data and capabilities of the
Kentucky Natural Resources information System: Kentucky
Natural Resources and Environmental Protection Cabinet,
Lands Unsuitable for Mining Program, 149 pages
Dinger, James S. and James R. Rebman, 1986, Ordinance for the
Control of Urban Development in Sinkhole Areas in the Blue
Grass Karst Region, Lexington, KY, in Proceedings of the
Conference on Environmental Problems in Karst Terranes and
Their Solution, NWWA, pages 163-180
Fitzmaurice, Karen, P, 1990, An Analysis of the Kentucky Pilot
Wellhead Protection Programs For Georgetown, Elizabethtown,
and Calvert City, Kentucky, Unpublished MS Thesis,
University of Kentucky, 200 pages
Jaffe, Martin, and Frank Dinovo, 1987, Local Groundwater Protec-
tion, American Planning Association, Washington, D.C., 262
pages
Kanizay, S. P. and E. R. Cressman, 1967, Geologic Map of the
Centerville Quadrangle, Map GQ-653, USGS
Miller, Robert D., 1967, Geologic Map of the Lexington West
Quadrangle, Map GQ-600, USGS
MacQuown, William C. Jr. and Ernest Dobrovolny, 1968, Geologic
Map of the Lexington East Quadrangle, Map GQ-683
Sendlein, Lyle V. A., 1989, DRASTIC Analysis For Application By
State Government, Final Report, Groundwater Branch, Division
of Water, Division of Waste, Natural Resources and
Environmental Protection Cabinet, Kentucky, Frankfort, KY,
59 pages
Sendlein, Lyle V. A. and Fitzmaurice, Karen, A, 1990, Kentucky
Pilot Wellhead Protection Study, Draft Report to USEPA,
Contract # 8R-1574-NAEX, Region IV Atlanta, 95 pages
Thrailkill, J., Spangler, L. E., Hopper, W. M. Jr., McCann, M.
R., Troester, J. W., and Gouzie, D. R., 1982, Groundwater in
the Inner Bluegrass Karst Region, Kentucky, University of
Kentucky, Water Resources Research Institute, Research
Report No. 140, 89 pages
682
-------
BIOGRAPHICAL SKETCH
Lyle V. A. Sendlein received his BS and AM in geological
engineering and geology, respectively, from Washington University
in Saint Louis, Missouri in 1958 and 1960. He received his Ph.D.
from Iowa State University in geology and soil engineering in
1964. He taught and conducted research for seventeen years at
Iowa State University. In 1977 he assumed the directorship of
the Coal Research Center at Southern Illinois University at
Carbondale and in 1982 assumed his current position as the
director of the Institute for Mining and Minerals Research at the
University of Kentucky. He is also a Professor in the Department
of Geological Sciences and is currently directing graduate
student research and conducting research in hydrogeology. His
areas of research include aquifer definition in both the coal
fields and the karst regions of Kentucky and contaminant
transport in these hydrogeological domains related to mining,
waste disposal and agricultural practices.
683
-------
ANALYSIS OF DRASTIC AND WELLHEAD PROTECTION METHODS
APPLIED TO A KARST SETTING
Lyle V. A. Sendlein
Department of Geological Sciences
University of Kentucky
Lexington, Kentucky
Question 1.
Because DRASTIC drastically under-estimated the vulnerability of those portions of the
karst basins dominated by conduit flow and recharged directly by swallets in sinking
streams and open sinkholes, how can it be utilized as a predictive hydrologic mapping tool in
karst? Aren't there serious limitations and dangers in relying on DRASTIC in karsts
without performing tracer studies and doing field reconnaissance for direct recharge sites
such as swallets and open sinkholes? Won't dye-tracing give you more information (and
more accurate information) about aquifer vulnerability?
Question la.
"Because DRASTIC drastically under-estimated the vulnerability of those portions of the
karst basins dominated by conduit flow and recharged directly by swallets in sinking
streams and open sinkholes, how can it be utilized as a predictive hydrologic mapping tool in
karst?"
The key words in this question are "predictive hydrologic mapping tool". DRASTIC is not a
hydrologic mapping tool. The method is designed to assist hi the planning process, which
would address large areas and is not designed for site specific use. For the areas mapped in
this study, high DRASTIC Index values resulted which would alert public officials (including
planners) that this area is vulnerable to groundwater contamination from surface activities.
As indicated in the paper, problems arose when karst groundwater basins determined by dye
traces were introduced into the analysis because the deep groundwater table caused by
drainage through the main conduit resulted in a lower DRASTIC Index value for those areas
within the groundwater basins.
Question Ib.
"Aren't there serious limitations and dangers in relying on DRASTIC in karsts without
performing tracer studies and doing field reconnaissance for direct recharge sites such as
swallets and open sinkholes?"
If the analysis produces high DRASTIC Index values, the purpose of the analysis has been
accomplished, even though specific karst features have not been identified. I do believe that
if one has knowledge of specific karst features that would directly transport contaminants to
the groundwater within the conduit system, they should be added to the map.
684
-------
Question Ic.
"Won't dye-tracing give you more information (and more accurate information) about aquifer
vulnerability?"
Of course dye-tracing will give site specific as well as general information about the karst
groundwater system. This would have to be the next step in an analysis of an area should
one decide to proceed with the planning of a specific land use knowing that the DRASTIC
Index value is high, which indicates a high vulnerability for contamination of the
groundwater.
Question 2.
I have found that although DRASTIC is a powerful and practical mapping tool in terranes
consisting of many rock types, it is useless for distinguishing slight differences in
vulnerability if the entire area to be evaluated in a karst. DRASTIC wasn't designed to
evaluate differences between contiguous karst areas. For Hart County, Kentucky, I found
direct interpretation of the published geologic maps to be most reliable. Your comment is
welcome.
I agree with this statement. For the public official or planner who may not know anything
about geology and karst, a DRASTIC map of a karst area would alert the planner to the
high vulnerability for contamination of the groundwater by surface activities. A qualified
hydrogeologist can take a published geologic map and probably glean more valuable
information about the groundwater flow systems in the area than a lay person. But I would
remind all that this method was not developed for geologists, but for public officials and
planners.
685
-------
686
-------
USE OF DYES FOR TRACING GROUND WATER:
ASPECTS OF REGULATION
James F. Quinlan
Quinlan & Associates, Inc.
Nashville, Tennessee 37222
ABSTRACT
An affirmative EPA policy on tracers is needed, and I urge
that steps be taken to establish one as part of legislation or
administrative regulations. Suggested wording for part of a
statute is included here, but if legislation is unnecessary and a
policy can be established administratively, it should be so done.
Tracing is a technique uniquely able to test and predict flow
velocity and flow direction of water and pollutants. When properly
used in the settings where they work best — karst aquifers,
fractured rocks, and coarsely granular sediments — tracers can
give reliable answers that can not be obtained in any other way.
The substances traditionally used for tracing are benign and
harmless in the concentrations employed by most knowledgeable
investigators. The use of tracers is justifiable on the basis of
their safety, ease of use, utility, reliability, and unique ability
to secure results obtainable in no other way. Currently, EPA has
no policy on the use or non-use of tracers, but their use has been
endorsed by the Agency in several publications and reports. Some
states routinely employ tracers; a few regulate them; a few forbid
their use. Most states ignore them and have no policy on their
use.
INTRODUCTION
Tracing agents (tracers) are added to a medium, usually water,
to enable the medium to be recognized at a remote location. Their
arrival concentration and arrival time can be used in calculations
and decisions. The most common tracers are substances added (such
as fluorescent dyes) or physical changes (such as temperature) made
in the medium. This paper is concerned primarily with tracers
added to water and used to study its movement and velocity as they
relate to environmental problems.
687
-------
Tracers are fundamental tools for discovery and prediction of
the velocity and dispersal-path of pollutants in ground water and
surface water. Interpretation of data from tracer studies makes it
possible to protect water quality, public health, and aquatic life.
Such data are crucial to the development of wellhead and springhead
protection strategies and for high-reliability prediction of
contaminant flow-paths and flow-directions. They can be essential
for the calibration of computer models of water flow and pollutant
movement, and they greatly enhance the validity of such models.
Tracing is cost-efficient and is often the only way to obtain
essential data.
Two major types of tracers are used most commonly: water-
soluble fluorescent dyes that have been shown to be non-toxic
(Field et al., 1992) and ions of non-toxic elements or compounds.
Specific tracers are named in a subsequent paragraph.
The use of tracers as an essential aid in the design of
monitoring systems in karst terranes is critically reviewed by
Quinlan (1989, 1990). Most karst terranes are areas underlain by
limestone and/or dolomite and characterized by sinkholes, sinking
streams, caves, and springs. Approximately 20% of the U.S. (40% of
the eastern U.S. that includes the upper Mississippi Valley) is
underlain by carbonate rock and is some type of karst.
An anonymous wag has whimsically stated that "One well-
designed tracer test, properly done, and correctly interpreted is
worth 1000 expert opinions ... or 100 computer simulations of
ground-water flow in karst terranes." He believes that this is
because most computer models are incapable of giving valid results
for flow in most karst terranes — for various technical reasons
not germane to this paper.
I do not deny that great advances have been made (and will be
made) in understanding and predicting ground-water flow with
computer models, and I freely acknowledge their unique capabili-
ties. Nevertheless, models have recognized limitations (Scarrow,
1989), about which two points must be made:
1. Hard data (real numbers, based on measurements) have
greater reliability than soft, imaginary data (calculated
values derived from computer models or simulations) .
Given a choice between believing the results of a well
designed, properly executed, and carefully interpreted
tracer test or the answers given by a computer model which
disagree with the tracer test, the tracer results should
be given greater weight — for the same reason that we do
not debate the existence of gravity. The tangible results
of each are obvious.
2. Computer models are no better than the validity of the
assumptions made in their design, the reliability (repre-
sentativeness) of the parameters used for data input, and
the correctness of the analysis and manipulation of data.
688
-------
Their validity must be evaluated with a sensitivity
analysis and by comparison with hard data.
I make these points because tracers and models do not substitute
for one another. Tracer results take priority over model results,
however, but each can complement the other, doing what the other
can not. Although I enthusiastically endorse the use of computer
models for analyzing karst aquifers when done successfully
(Teutsch, 1989; Teutsch and Sauter, 1991; Sauter, 1991a, 1991b), I
demand the rigor imposed by J. C. Griffiths, the doyen of
quantitative geologists, on the methodology of geologic investiga-
tion (Drew, 1990, p. 3-10).
DESCRIPTION OF PROBLEM
The use of tracers for solving environmental problems has
become well established in recent years. In some settings, it is
not possible to define important environmental parameters without
first conducting comprehensive tracing studies.
In an unrealistic effort to protect human health and the
environment from any possible harm, some state agencies have
unrealistically restricted the use of tracers, regardless of the
information to be gained. For example, some state agencies have
interpreted tracers used in ground-water investigations to be
pollutants or contaminants and therefore violations of their
"non-degradation policy." Strictly speaking, injection of dye or
other non-toxic tracers into a well (or other input source, e.g.,
a sinkhole), no matter how noble the reason for doing so, makes it
possible to construe the well to be a Class V injection well and
thus subject to State and Federal regulations governing its use.
(If the tracer were toxic, the well would be a Class IV injection
well.)
Part of the problem is due to the misperceptions of many
individuals about fluorescent dyes and other substances used for
environmental tracing studies, especially with regard to their
suspected toxicity. This is the fallacious logic used:
Fluorescent dyes are synthetic substances (i.e.. , they do not occur
naturally); therefore, they must be a pollutant and harmful to
human health and the environment. This misperception is
exacerbated by the lack of comprehensive toxicity test data on the
dyes. The fallacious argument continues: Since fluorescent dyes
must be harmful, so probably must be all other tracing agents.
Granted, some tracers are hazardous (e.g., tritium), but
assumptions such as these cited are absurd and unscientific, and
they decrease our ability to gather data needed to better protect
ground-water quality.
Clearly, there is a need for a Federal policy endorsing the
use of tracers in ground water if the various misperceptions
regarding their use are to be dispelled. Such a policy does not
need to be very extensive or detailed, but it does need to be
689
-------
flexible.
DISCUSSION
Well-meaning as the above interpretation of regulations for
injection wells and the concern regarding suspected tracer toxicity
may be, it is not justifiable in terms of potential benefits for
environmental protection, original intent of the law-makers, or
risk of exposure to pollutants. Like boats put into a lake,
tracing agents are used in the water for a good and definite
purpose, not put into it for disposal. And like boats, dyes and
ions generally used for tracing ground water are benign and
harmless in the concentrations commonly employed (Field et al.,
1992; Smart, 1984).
A further analogy describing the use of tracers can be made.
Doctors use vaccines and a wide range of diagnostic techniques to
prevent and treat illnesses. Some of these vaccines and techniques
have definite risks associated with their use. These risks are
assumed by an informed patient because the consequences of not
preventing or not diagnosing an illness far outweigh the much
smaller risk from use of the vaccine or diagnostic technique.
If and when state officials establish regulations governing
the use of dyes or any other ground-water tracer, they should
require their use by knowledgeable, experienced professionals. If
it is felt necessary to give the states regulatory authority in
tracing, I suggest that regulation be through licensing of
individuals rather than on a study-by-study basis.
Many Federal and State agencies have sanctioned the use of
dye-tracing studies in the study of ground-water pollution and
time-of-travel of pollutants in rivers. Several major EPA docu-
ments endorse the use of tracers in ground-water studies (US EPA,
1986, p. 64; 1987a, p. 2-12, 2-22 2-23; 1987b, p. 127-148; Quinlan,
1989). Guidance manuals for tracing techniques exist and have been
sponsored by EPA (Davis et al.. 1985; Mull et al.. 1988; Quinlan,
1991) and by the Societe Geologique Suisse (Parriaux et al. , 1988).
Updated manuals on ground-water tracing have been written by the
U.S. Geological Survey (under contract to the EPA; Mull et al.,
1988) and are in preparation for the National Ground Water Associa-
tion (Aley et al., in prep.). Several manuals on the use of dyes
for measurement of discharge, time of travel, and dispersion in
surface streams have been written by the U.S. Geological Survey
(Kilpatrick and Wilson, 1989; Wilson et al., 1986; Kilpatrick and
Cobb, 1985; Hubbard et al., 1982).
International interest in and recognition of tracing as a
useful tool in the kit of the hydrogeologist is suggested by the
content of proceedings volumes of the five International Symposiums
on Underground Water Tracing held during the past 25 years and by
the synthesis of Caspar (1987), and by a recent meeting on granular
aquifers (Moltyaner, 1990). The sixth International Symposium, to
690
-------
be sponsored by the Association for Tracer Hydrology, will be held
in Germany in 1992.
It is desirable and necessary that proposed Federal ground-
water legislation include a section that sanctions the use of
tracing agents, prevents them from being regulated as pollutants or
contaminants, and strongly encourages their use in predicting the
possible consequences of pollutant release and dispersal. Alterna-
tively, and until legislation is passed, a policy endorsing the
judicious use of tracers should be promulgated.
RECOMMENDED SOLUTION
I suggest that legislation or policy concerning ground-
water protection include wording (originally developed by Quinlan
and Field, 1991) to the effect that:
The use of organic and/or inorganic tracing agents
employed in investigations of ground-water and surface-
water hydrology is a desirable activity and shall not be
construed as the addition of a pollutant or contaminant
to the water when deliberately introduced as a tracer.
Useful tracing agents include, but are not limited to,
the following: dyes such as fluorescein (Colour Index
[CI] Acid Yellow 73), eosin (CI Acid Red 87), Rhodamine
WT (CI Acid Red 388) , Sulpho Rhodamine G (CI Acid Red
50) , Sulpho Rhodamine B (CI Acid Red 52) , optical
brighteners (fluorescent whitening agents with various CI
numbers), Diphenyl Brilliant Flavine 7GFF (CI Direct
Yellow 96) , Lissamine Flavine FF (CI Acid Yellow 7) ,
pyranine (CI Solvent Green 7) ; ions such as lithium,
chloride, bromide, iodide, nitrate, and sulfate; fluor-
inated organic anions; Lycopodium spores; bacteriophages;
noble gases; other gases such as sulfur hexafluoride;
stable isotopes; particulates; hot water; and cold water.
Sewage effluent and waste products from industrial
activity, waste disposal, or spills may function as a
tracing agent, but they are not suitable for deliberate
addition to water as a tracer.
Investigations regarding the vadose zone in fractured and
karstic terranes can employ the use of smoke and various
gases (e.g., sulfur hexafluoride). These can be used for
determining subsurface interconnections above the poten-
tiometric surface when light nonaqueous phase liquids
(LNAPLS) and other highly volatile contaminants are
residing in the vadose zone.
The U.S. Environmental Protection Agency, U.S. Geological
Survey, American Society for Testing and Materials, and
professional societies of hydrologists, hydrogeologists,
and engineers are authorized to establish voluntary
691
-------
protocols for the selection and use of tracing agents.
Tracing should be conducted only by those individuals who
can demonstrate evidence of substantial training and/or
experience in the use of tracer agents. The training may
have been acquired through academic study at a univer-
sity, courses offered by professional societies, or by
experience gained while working under the supervision of
an experienced water-tracing professional.
CONCLUSIONS
The importance of establishing a set of regulations for the
use of tracers in the environment, particularly the use of
fluorescent dyes in ground water cannot be overemphasized. Their
utility in defining the true flow paths and flow rates are
invaluable, especially in pollution studies.
Failure to develop coherent regulations will continue to
result in confusion regarding the alleged toxicity of the dyes and
other tracer substances, their use by professionals versus
amateurs, and their importance in the realm of ground-water quality
monitoring. It is essential that the Federal government support
the appropriate use of water-tracing substances so that much of the
confusion regarding their proper use can be overcome.
REFERENCES
Aley, T., Quinlan, J. F., Alexander, E. L., Jr., and Behrens, H.,
[in prep.]. The Joy of Dying: A Compendium of Practical
Techniques for Tracing Ground Water, Especially in Karst
Terranes. National Ground Water Association, Dublin, Ohio.
Davis, S. N. , Campbell, D. J. , Bently, H. W. , and Flynn, T. J.;
1985. Ground Water Tracers. National Water Well Association,
Worthington, Ohio. 200 p. [Reprint of report prepared under
contract to the U.S. EPA. ; there are many errors in the
chapter on dye tracing, p. 97-118; see discussion by Quinlan
(1986)]
Drew, L. J. , 1990. Oil and Gas Forecasting. International
Association for Mathematical Geology, Studies in Mathematical
Geology, No. 2. Oxford, New York. 252 p.
Field, M. S., Wilhelm, R. G., and Quinlan, J- F., 1992. Selection,
toxicity, and use of fluorescent dyes for tracing ground
water: 1. What is safe, why, where, and when? [submitted to
Ground Water]
Caspar, E., ed., 1987. Modern Trends in Tracer Hydrology. CRC
Press, Boca Raton, Fla. 2 v. 145 and 136 p.
692
-------
Hubbard, E. F., Kilpatrick, F. A., Martens, L. A., and Wilson, J.
F., Jr., 1982. Measurement of time of travel and dispersion
in streams by dye-tracing. U.S. Geological Survey, Techniques
of Water-Resources Investigations, Book 3, Chapter A9. 44 p.
Kilpatrick, F. A., and Cobb, E. D. , 1985. Measurement of discharge
using tracers. U.S. Geological Survey, Techniques of Water-
Resources Investigation, Book 3, Chapter A16. 52 p.
Kilpatrick, F. A. and J. F. Wilson, Jr. 1989. Measurement of time
of travel in streams by dye tracing. U.S. Geological Survey,
Techniques of Water-Resources Investigations, Book 3, Chapter
A9. 27 p. [Revision of Hubbard et. al. , 1982]
Moltyaner, G., ed. 1990. Transport and Mass Exchange Processes in
Sand and Gravel Aquifers: Field and Modelling Studies. Atomic
Energy of Canada Limited, Ottawa. AECL-10308. 2 v. 866 p.
[Includes 46 papers presented at an international conference
and workshop]
Mull, D. S., Liebermann, T. D., Smoot, J. L., and Woosley, L. H.,
Jr., 1988. Application of dye-tracing techniques for deter-
mining solute-transport characteristics of ground water in
karst terranes. U.S. Environmental Protection Agency, Region
IV, Atlanta, Ga., EPA 904/6-88-001. 103 p.
Parriaux, A., Liszkay, M. , Miiller, I., and della Valle, G. , 1988.
Guide practique pour 1'usage des traceurs artificiels en
hydrogeologie. Societe Geologique Suisse, Groupe des
Hydrogeologues. GEOLEP EPFL, Laussane. 51 p. [bilingual;
also published in the same volume as: Leitfaden fur den
Gebrauch Kvinstlicher Tracer in der Hydrogeologie. 49 p. ]
Quinlan, J. F., 1986. Discussion of GROUND WATER TRACERS by Davis
et al. , (1985), with emphasis on dye-tracing, especially in
karst terranes. Ground Water. v. 24, p. 253-259 and p.
396-397. (Reply by Davis: V. 24, p. 398-399)
Quinlan, J. F., 1989. Ground-water monitoring in karst terranes:
Recommended protocols and implicit assumptions. Internal
report prepared for U.S. Environmental Protection Agency,
Environmental Monitoring Systems Laboratory, Las Vegas,
Nevada. EPA/600/X-89/050. 88 p. [In review; to be published
1992]
Quinlan, J. F., 1990. Special problems of ground-water monitoring
in karst terranes. Nielsen, D. M., and Johnson, A. I., eds.
Ground Water and the Vadose Zone. American Society for
Testing and Materials, Special Technical Paper 1053. ASTM,
Philadelphia, Penn. p. 275-304.
Quinlan, J. F., 1991. Delineation of Wellhead Protection Areas in
Karst Terranes, with Emphasis on Puerto Rico. Quinlan &
Associates, Nashville, Tennessee, ca. 10 p. [Course Manual]
693
-------
Quinlan, J. F., and M. S. Field, 1991. Use of dyes for tracing
ground water: Aspects of regulation. Internal report prepared
for U.S. Environmental Protection Agency, Environmental
Monitoring Systems Laboratory, Las Vegas, Nevada. EPA/XXX/X-
XX/XXX. 7 p. [In review]
Sauter, M., 1991a. Double porosity models in karstified limestone
aquifers: Field validation and data provision. International
Association of Hydrogeologists, Karst Symposium (Antalya,
Turkey, 1990), Proceedings. 27 p. [in press]
Sauter, M., 1991b. Assessment of hydraulic conductivity in karst
aquifers at local and regional scale. Quinlan, J. F. , and
Stanley, A., eds. Hydrogeology, Ecology, Monitoring, and
Management of Ground Water in Karst Terranes Conference (3rd,
Nashville), Proceedings. National Ground Water Association,
Dublin, Ohio. [in press]
Scarrow, J. W., 1989. The use of modeling in groundwater contam-
ination cases. Virginia Environmental Law Journal, v. 9, p.
185-205.
Smart, P- L. , 1984. A review of the toxicity of twelve fluorescent
dyes used in water tracing. National Speleological Society,
Bulletin. v. 46, p. 21-33.
Teutsch, G., 1989. Groundwater models in karstified terranes: Two
practical examples in the Swabian Alb, southern Germany.
Solving Ground-Water Problems with Models (4th, Indianapolis) ,
Proceedings. National Water Well Association, Dublin, Ohio.
p. 929-953.
Teutsch, G, and M. Sauter., 1991. Groundwater modeling in karst
terranes: Scale effects, data acquisition, and field valida-
tion, in Quinlan, J. F., and Stanley, A., eds. Hydrogeology,
Ecology, Monitoring, and Management of Ground Water in Karst
Terranes Conference (3rd, Nashville), Proceedings. National
Ground Water Association, Dublin, Ohio. [in press]
U.S. Environmental Protection Agency, 1986. RCRA Ground-Water
Monitoring Technical Enforcement Guidance Document. Office of
Waste Programs Enforcement and Office of Solid Waste and
Emergency Response, Washington, D.C. 319 p.
U.S. Environmental Protection Agency, 1987a. Guidelines for
Delineation of Wellhead Protection Areas. Office of Ground-
Water Protection, Washington, D.C. EPA 440/6-87/-010. 204 p.
U.S. Environmental Protection Agency, 1987b. Groundwater Handbook.
Robert S. Kerr Environmental Research Laboratory, Ada, Okla.
EPA/625/6-87/016. 212 p. [The section on ground water
tracers, p. 127-148, is adapted from Davis et al. . 1985;
republished as: Barcelona, M. , Wehrman, A., Keely, J. F., and
Pettyjohn, W. A., 1990. Contamination of Ground Water: Pre-
694
-------
vention, Assessment, Restoration. Noyes Data Corporation,
Park Ridge, N.J. 213 p.]
Wilson, J. F., Jr., Cobb, E. D., and Kilpatrick, F. A., 1986.
Fluorometric procedures for dye tracing. U.S. Geological
Survey, Techniques for Water-Resources Investigation, Book 3,
Chapter A12. 34 p.
ACKNOWLE DGEMENTS
This paper has been critically reviewed by Thomas Aley, E.
Calvin Alexander, Jr. , and A. Richard Smith, each of whom made
substantial improvements in its content and clarity. The inval-
uable contributions of Malcolm S. Field have helped make this paper
what it is and are gratefully acknowledged.
BIOGRAPHICAL SKETCH
Dr. James F. Quinlan is a consultant specializing in the
hydrology of karst aquifers and techniques for tracing ground water
with fluorescent dyes. He has more than 36 years of experience in
karsts of 26 states and 23 countries. For 16 years he directed a
hydrologic research program for the National Park Service at
Mammoth Cave National Park, Kentucky. He has written or co-written
more than 170 publications on ground-water monitoring, evaluation
of waste disposal sites, and spill response in karst terranes. In
1985 he and Ralph Ewers received the Burwell Award from the
Geological Society of America for a pioneering paper on ground-
water monitoring in karst terranes. He is a chairman of an ASTM
Task Group on ground-water monitoring in karst and other fractured
rocks, has been a Director of the Association of Ground Water
Scientists and Engineers, has organized numerous scientific
meetings and, since 1986, has annually co-taught an N.G.W.A. short
course on practical karst hydrogeology.
James F. Quinlan
Quinlan & Associates, Inc.
Box 110539
Nashville, TN 37222
(615) 833-4324
695
-------
USE OF DYES FOR TRACING GROUND WATER: ASPECTS OF REGULATION
By: James F. Quinlan
Both Missouri and Kentucky are in the process of establishing dye-
trace data bases. This will require sharing of tracing results.
Do you/ as a consultant, support this type of additional regula-
tion?
Assuming the data base is available to all, yes, absolutely.
Everyone benefits.
Has there been extensive acute and chronic testing of tracers?
Yes, but there is never as much as one might wish for. The best
review of this topic is by Pete Smart in the NSS Bulletin (1984.
46(2):21-33).
I have heard that it is dangerous to use Rhodamine WT because it
forms nitrosamine compounds which are carcinogenic. Would you
please comment on this issue.
The concern about nitrosamines from Rhodamine WT was raised by
Abidi (1982, Water Research, v. 16, p. 199-204). She reported that
diethylnitrosamine (NDEA) formed in river waters treated with ni-
trite ions (NO2~) and either Rhodamine WT or B. Nitrite is an un-
stable intermediate ion formed in the reduction of nitrate ions or
the oxidation of ammonium and organic nitrogen. Nitrite ions are
found, in small concentrations, primarily in waters heavily impact-
ed by animal wastes such as the discharge from trout-rearing ponds
or sewage wastes. Nitrite levels are very low in most ground
waters, even those with elevated levels of nitrate ions.
Steinheimer and Johnson (1986, U.S.G.S. Water Supply Paper 2290, p.
37-49) did a very detailed and careful follow-up on Abidi's work
and found that in four different river waters "nitrosamine forma-
tion did not occur in the river samples at concentrations typically
encountered during dye injection studies." They then concluded
"under customary dye-study practices, NDEA resulting from the use
of Rhodamine WT does not constitute an environmental hazard."
No diagnostic tool is without risk, however. An x-ray examination
of a broken bone carries a measurable risk of inducing a cancer,
for example. The advantage of diagnosing and setting the bone
correctly outweighs the risk and diagnostic tool. It should only
be used when there is a need for the information it will yield.
The danger associated with long-term chronic pollution of a water
supply will usually outweigh any risk associated with an improb-
able, transient, potential exposure to a reaction product of the
dye. The dye is not the problem here, it is the conditions that
necessitate the dye trace.
696
-------
This paper is included
within the Proceedings of the
National Cave Conservation
Association
THE EFFECTS OF RECHARGE BASIN LAND-USE PRACTICES
ON WATER QUALITY AT
MAMMOTH CAVE NATIONAL PARK, KENTUCKY
Joe Meiman
Mammoth Cave National Park
Mammoth Cave, Kentucky
ABSTRACT
A water quality monitoring program was designed at Mammoth Cave
National Park to determine if there exists any influence on the
water quality of the Mammoth Cave karst aquifer within the park
from various land-use practices of the recharge area. These land
uses primarily include: heavy agriculture (row crops and
livestock), logging, oil and gas production, and residential
areas. The program, initiated in March 1990 and extending
through September 1992, samples two rivers, and eight springs
recharged by lands with varying land-use. Monthly non-
conditional synoptic sampling monitors 36 parameters, including
site discharge. The first 19 months of data demonstrate a strong
correlation between drainage basin land-use and water quality.
Contaminant entrainment mechanisms and relative pollutant input
rates can be discerned when the mass-flux of selected parameters
is calculated. By use of these data, effective resource
management decisions can, and are being made to conserve and
protect the irreplaceable natural resources of Mammoth Cave
National Park.
INTRODUCTION
For what purpose do we monitor the quality of water at Mammoth
Cave National Park? Aside from pure stoichiometric data to
satisfy our curiosity of the water's chemical composition,
spatially and temporally, the fundamental mission of this
monitoring program is to better understand, and thus better
manage, the aquatic natural resources of the park. During the
three year course of this program, data will be collected and
interpreted to provide information on the current state of the
surface and subsurface water of the park. This data set will be
697
-------
used as a datum from which to compare past and future studies.
As the author is not a biologist, no claims, speculations,
conjectures, or theories pertaining to the present health or
future of the aquatic ecosystems will be made. However, before
trained personnel can accurately assess the condition of the
park's aquatic life, a broad database of the physical and
chemical properties must be available.
Although this phase of the monitoring program is far from
complete, there appear to exist a few trends and correlations
which deserve mention. The following pages will concern the
first nineteen rounds of monthly sampling.
BRIEF DESCRIPTION OF MONITORING PROGRAM
The monitoring program is largely based upon synoptic samplings.
Synoptic, as defined by Webster, is "relating to or displaying
conditions as they exist simultaneously over a broad area".
Although the water quality monitoring program includes two
different synoptic approaches, conditional and non-conditional,
the later comprises by far the bulk of monitoring activities for
the first years of the study. The program also includes topical
sampling which provides a detailed evaluation of a particular
flow condition, contaminant, basin or river reach.
Choosing synoptic stations within a karst aquifer differs greatly
from the same task preformed on a surface drainage. In a surface
drainage one can choose sites based upon stream reaches (every 20
miles for example) to improve spatial distribution, or install a
station exactly where a known pollutant source is located. The
monitoring sites of this program were chosen with respect to
land-use practices of the various recharge basins. These
practices range from the naturally wooded park-land groundwater
basins where human influence has been absent for at least 50
years, to highly agricultural lands with a share of urban use and
oil and gas exploration.
The ten non-conditional synoptic stations are sampled, regardless
of flow or weather conditions, on the 10th of each month for the
duration of the study (Figure 1). The sites are sampled during a
single day by one field crew. The need for repetitive sampling
(each month for three years) at each non-conditional synoptic
station arises from the considerable temporal variability of
karst and surface water quality. This variability is largely a
result of sudden changes in discharge, and seasonal availability
of contaminant sources. Over the course of the study each end of
the flow continuum (base and flood conditions) and each growing
season will have been encountered several times, as by program
design.
The primary use of conditional synoptic surveys is to provide a
finer degree of spatial resolution to the descriptions of
discrete water quality and flow conditions than would be
698
-------
attainable from the non-conditional synoptic station network.
One of the goals of the conditional synoptic surveys is to
identify relatively short reaches of drainage basins which have
demonstrated (by data from the non-conditional synoptic station
network) chronic water quality problems.
Although the program is designed to allow the park to determine
the effects of la'nd-use practices on water quality after three
years of sampling, we can, at this juncture, observe various
traits which may be attributed to types of land-use in the
recharge areas. Each time a sample is extracted, discharge at
the site is recorded. Parameter concentration, coupled with
discharge will yield flow-weighted values. These values will
allow us to determine the mass flux (loading) of a particular
parameter at various flow conditions. These data will allow us
to better determine contaminant source as it pertains to
constituent availability, release, and entrainment into the
water, and mechanisms of transfer from the surface to the
subsurface.
A DISCUSSION CONCERNING MASS FLUX AND FLOOD PULSES
It would be difficult to continue this discussion without first
examining mass flux and flood pulses. Mass flux is simply the
amount (mass) of a particular parameter passing a point in a
given time interval (flux). A flood pulse is the portion of
water propagated along a channel and/or conduit as result of a
recharge event, most commonly, rainfall. With an understanding
of flood pulse movement, one might better understand mass flux
signatures of various contaminants.
MASS FLUX
If a contaminant is released into a stream of water at a constant
rate, and at some point downstream its concentration and the
stream's flow can be measured, the mass flux of the contaminant
can be calculated (Figure 1). If flow (Discharge 1) decreases or
increases, the contaminant's concentration will proportionally
increase or decrease, respectively (Concentration). That is, at
times of high flow the contaminant will experience a greater
amount of dilution. The resultant mass flux signature (mass flux
over time) of a constant source release will consist of a
relatively low amplitude disturbance (Mass Flux 1). One may
think of this mass flux signature as a type of destructive wave
form interference.
Suppose a contaminant is released into a stream only when
specific hydrologic conditions are met, a rainfall event of a
certain intensity and volume for example. Therefore, if flow
(Discharge 2) increases, contaminant concentration also increases
(Concentration) as these stores are displaced into the streams
during flood pulse activity. That is during the times of peak
699
-------
MASS FLUX
O
Z
CO
<
LJ
LT
o
MASS FLUX 2 \
MASS FLUX 1
CONCENTRATION
DISCHARGE 1
I
TIME -
Figure 1. Hypothetical mass flux signatures.
flow, peak (or near peak) concentrations also occur. The
resultant mass flux signature of a precipitation-triggered
release will be of relatively high amplitude, perhaps several
orders of magnitude greater than the pre-pulse mass flux (Mass
Flux 2). This mass flux signature may be likened to constructive
wave form interference.
FLOOD PULSES
Flood pulses in the Mammoth Cave area may raise a basin's
discharge a couple liters per second following minor rainfall, to
several thousand liters per second after major rainfall.
Research by Meiman (1988 and 1989) has demonstrated that a flood
pulse is comprised of two chemically and physically distinct
components: displaced stores and freshly input recharge. The
former, which usually occurs as the leading edge of a flood pulse
and is characterized by high specific conductances, can be
thought of as easily displaced vadose storage. Freshly input
recharge, which comprises the bulk of a flood pulse, is
characterized by low conductances, as there is little time for
interaction between its waters and ionic sources. These
relationships are also manifested in water temperature, as
displaced stores, with longer residence times, will reflect the
antecedent system temperature, and freshly input recharge
correlative to surface temperature (Meiman, 1988 and 1989).
700
-------
Consider the flood pulse displayed in Figure 2. This pulse,
documented over approximately 42 hours in the fall of 1987 at a
sinking creek of the Turnhole Spring groundwater basin, clearly
indicates the arrival of three highly conductive sources during
the course of flood pulse activity. The majority of flow
generated from this rainfall event was of the low-conductance,
run-off variety. If water provenance suddenly changes, one may
expect to see a similar change in water quality with respect to
available water-borne constituents. Hallberg, et al (1985)
identified an acute, albeit brief, water quality degradation
associated with this run-off component in the karst of
northeastern Iowa.
WATER QUALITY, EXPRESSED BY MASS FLUX, AS IT
RELATES TO FLOOD PULSES
Important water quality information may be gained if knowledge of
flood pulses is combined with mass flux signatures. As rainfall
occurs, flood pulses are generated and propagated through the
karst aquifer. Just as stage may suddenly vault from its base
condition, water quality may also undergo rapid and drastic
change as a flood pulse passes. If a significant amount of
70 r
60
(n 50
I
W 40
30
20
10
stage
450
400
350
300
250
200
150
100
50
en
•
m
O
o
o
o
o
o
o
m
c
3
12
TIME IN HOURS
24
Figure 2. Stage and specific conductance responses of a flood
pulse in a surface stream, Turnhole Spring groundwater
basin.
701
-------
constituents are released by the precipitation event (entrained
in run-off), the mass flux of these elements may rise
tremendously.
Contemplate the two hypothetical mass flux signatures of
Figure 3. It is vital to note the relative, unitless scales of
the two graphs. Although the X-axis scales are equal, the Y-axis
of 3a is I/8th that of 3b. Also note that "Time 0" indicates the
advent of precipitation. The same discharge hydrograph is
employed for both graphs. Remember, data used in these graphs
are hypothetical. Numbers were derived by a noting the timing,
duration, and wave-form characteristics of years of continuous
data (stage, specific conductance, water temperature and
discharge) and months of water quality data. Mass fluxes are
actual products of discharges and concentrations.
If a constant source parameter is monitored through a flood
pulse, oil-field brine chlorides from a leaking well casing for
example, a response similar to that of Graph 3A might be
expected. Following an initial upward spike in concentration,
perhaps caused by a flushing of vadose stores, chloride
concentration is diluted as the pulse's freshly input recharge
component dominates the flow. The resultant mass flux signature,
although not without structure, displays relatively low amplitude
disturbance.
If the same discharge is used with a precipitation triggered
release parameter, certain pesticide residues for instance, a
totally different mass flux signature results (Graph 3b). Again,
note the Y-axes scales. The effects of constructive wave-form
CONSTANT SOURCE
PRECIPITA TION- TRIGGERED
3B
-20 2 4 6 8 10 12 14 16 18 20 22
TIME IN HOURS
-20 2 4 6 8 10 12 14 16 18 20 22
TIME IN HOURS
Figure 3. Hypothetical mass flux signatures of constant source
release (3A) and precipitation-triggered release (3B).
702
-------
interference are noticeable as a high amplitude mass flux
signature is generated. The passage of the leading edge of the
pulse may be reflected in a sharp drop in pesticide residue
concentration, as long-residing stores are displaced. This
effect will be far overshadowed by the arrival of the freshly
input recharge. Not only does this flow component comprise the
majority of the flood pulse discharge, it also contains the bulk
of surface run-off with entrained herbicides. The resultant mass
flux signature may be several orders of magnitude higher than
pre-pulse values.
There are many factors that may control the shape of the mass
flux signature: availability of constituents, entrainment
method, transfer mechanism from surface to subsurface, rainfall
volume and areal distribution, time since last rainfall, and
conduit condition, to name a few. It should be noted that the
conduit condition used in this example is highly vadose. A
different signature, especially with respect to temporal lags of
concentrations and discharge peaks, will occur when dealing with
a phreatic conduit system (Meiman 1988, 1989). Current research
at Mammoth Cave specifically addresses flood-pulse water quality.
Perhaps by close examination of mass flux signatures of fecal
coliform bacteria, dominant waste sources, human or animal, may
be discerned. Human waste, for the most part, should behave as a
constant release source. Human waste is injected directly into
the aquifer via leach fields, leaking septic tanks, or dry-wells,
at a relatively constant rate. A constant mass flux signature
should result. Animals, not nearly intelligent enough to
defecate down wells, will deposit waste on the surface. Without
rainfall (or a major snow-melt), this waste will not be
transferred into the aquifer. Following a significant recharge
event, animal waste will be washed into the aquifer, producing a
mass flux signature characterized by a very high amplitude
disturbance.
RESULTS OF MARCH 1990 THROUGH SEPTEMBER 1991
The following data (based upon non-conditional samples) run from
March 1990 through September 1991. The summer months, which
comprise a disproportionally large percentage of the data, will
skew the data toward low-flow conditions. Non-conditional
synoptic sampling covers a wide spectrum of flow conditions,
ranging from flood-pulses to flow-reversals. If river water is
back-flooded into the spring, it is considered to be
representative of the spring's water at that moment in time. The
sample will be taken and analyzed regardless of water provenance
(river or cave derived). The aquatic communities of the spring
and related conduit must live in the waters, regardless the
source, therefore the sample is representative of their
environment.
703
-------
The presentation of water quality data in the following
discussion will be in two forms: statistical graphs (bar and
whisker) and XY graphs depicting trends at selected sites of
selected parameters. The four selected sampling sites for this
document are: Light agriculture (Pike Spring, PSPS), Park/heavy
agriculture (Echo River Spring, ERES), Heavy agriculture
(Turnhole Spring area, THNS), and Park lands (Buffalo Creek
Spring, BCGR).
DISCHARGE
Discharge depends, of course, upon precipitation events. The
summer and fall months are traditionally characterized by low
discharge, with higher discharge through the winter and spring
(Figure 4). Overall the largest discharges during sampling
occurred on April 10, 1991. On this date flood-pulse activity
was high as the aguifer quickly responded to the rains of the
previous day. This sampling round is of specific importance as
samples were extracted near pe^k discharge times of the flood-
pulses. Although other rounds saw relatively high discharges,
samples were, as dictated by monitoring program, taken either
well before or well after pulse peaks.
During the first nineteen months of the study a major back-
flooding event was sampled on June 10, 1990. At this time all
10000
8000
Q
•z.
8 6000
UJ
GO
01
D.
a: 4000
LU
2000
-•4-- HEAVY AGRICULTURE
-•— LIGHT AGRICULTURE
-f- PARK / HEAVY AGRICULTURE
-••-- PARK LANDS
MAMJJASONDJFMAMJJAS
MONTHS 1990 - 1991
Figure 4. Discharge values, March 1990 through September 1991, of
four monitoring sites.
704
-------
springs, with the exception of THNS, were in a state of flow
reversal - water from the Green River flowing back into the
aquifer.
Notice that Echo River Spring is referred to as "Park/heavy
agriculture". During times of high discharge, flow from the
heavy agriculture basin is shunted through a high-level overflow
route into the Echo River basin, which is normally recharged by
park lands. When this route is activated, water quality in Echo
River may, nearly instantaneously, degrade. Research in the next
year will document the conditions needed to conduct flow through
the overflow route.
TURBIDITY
Turbidity, correlative to the amount of suspended sediment in the
water, is highly variable through all non-conditional synoptic
sites (Figure 5). As expected, basins dominated by agricultural
land-use, discharge more turbid waters than those dominated by
undisturbed forests. Generally one would expect the higher
turbidities associated with areas of high soil loss. Although
the Turnhole basin (Heavy agriculture) contains by far the
greatest area of tilled crop-lands, its turbidities, albeit high,
were not the highest recorded; that honor goes to the light
agriculture Pike Spring basin. Although containing far fewer
90 p
80 i-
70
60
50
--- HEAVY AGRICULTURE
— LIGHT AGRICULTURE
PARK / HEAVY AGRICULTURE
PARK LANDS
MAMJJASONDJFMAMJJAS
MONTHS 1990 - 1991
Figure 5. Turbidity values, March 1990 through September 1991, of
four monitoring sites.
705
-------
acres of tilled land, the rugged topography of the Pike Spring
basin amplifies soil loss when disturbed.
Displayed in turbidity are the back-flooding and overflow
described in the preceding section. The back-flooding event of
June 1990 is evident in turbidity, as back-flooded springs
display turbidities close to that of the Green River. The Echo
River basin (Park/heavy agriculture) exhibits low turbidities,
associated with low to moderate discharges when the spring is
recharge by park lands, and high turbidities when the overflow
route from the heavy agriculture basin is activated.
CHLORIDE
Chloride may be indicative of animal/human waste and oil field
brines. Figure 6a shows the chloride concentration trends of the
four selected sites, while Figure 6b demonstrates the mass flux
of the chloride ions. Both graphs exhibit interesting data.
Figure 6a shows elevated concentrations of chloride in the heavy
agriculture basin.
Oil field brines seem the prime suspect for two reasons: presence
of associated brine ions, and mass flux signatures. Within the
headwaters of the basin, and adjacent to the Park City oil-field,
is Parker Cave. On a low-flow conditional synoptic survey
(September, 1990), Parker River (a stream passage within Parker
Cave) had chloride levels of 1476.1 ppm. Further down-basin,
Mill Hole chloride was 59.9 ppm, as the Parker River water was
diluted. At the basin's terminal spring, chloride was further
diluted to 31.6 ppm. Bromide and sulphate, also suggestive of
brines, were found decreasing at similar rates at the same sites.
Similar results were reported by Meiman (1989), and Quinlan and
Rowe (1978).
The mass flux signature of chloride may also indicate brine
contamination instead of animal waste. Figure 6a displays a
variable, yet predictable pattern of chloride concentrations.
One may assume that the chloride source is of relatively constant
delivery as chloride concentrations are higher during low flow
periods of summer and early autumn, and lower, more dilute,
during the high flow periods of winter and spring. Figure 6b
indicates an apparently dramatic increase on the mass flux of
chloride during months of high discharge. This increase, some
eight times the mass flux of low discharge periods, may not be as
severe as it may seem. This variation may be normal, even for
this relatively constant source parameter. One might expect a
much greater (several orders of magnitude) rise in the mass flux
signature if a source is released by run-off from a precipitation
event.
A certain portion of chloride can be considered as a natural,
background concentration. Observe the chloride trends of the
Park land basin. Not only are chloride concentrations low
706
-------
32 r
-->-- HEAVY AGRICULTURE
—•— LIGHT AGRICULTURE
- V PARK / HEAVT AGRICULTURE
• PARK LANDS
E
Q-
Q_
Ld
Q
Ct
O
I
O
M A M J J
ASONDJ FM
MONTHS 1990 - 1991
A M J J A S
Figure 6a. Chloride concentrations, March 1990 through September
1991, of four monitoring sites.
40000
30000
O
in
cr>
UJ
O
C£
3
O
20000
10000
0
HEAVY AGRICULTURE
LIGHT AGRICULTURE
PARK / HEAVY AGRICULTURE
PARK LANDS
^*
-T '
MAMJJASONDJFMAMJjAS
MONTHS 1990 - 1991
Figure 6b. Chloride mass flux, March 1990 through September 1991,
of four sampling sites.
707
-------
(Figure 6a), they remain at approximately the same mass flux
throughout the year (Figure 6b). Upon closer examination, notice
the slight increase in mass flux through the winter months.
Although seemingly small and insignificant, the relative changes
between seasonal mass fluxes in the Park land and the Heavy
agriculture basins are very similar. This trend may be an
inherent wave-form signature due to the vast increase in
discharge.
Road salts as a potential chloride source must be recognized.
Although the use of road salts have been prohibited within the
park since 1987, they are used throughout several of the park's
groundwater basins. The amount of road salt contributing to
chloride levels found in parks waters is yet unknown. Since the
first month of sampling there has not been a significant snow
fall to warrant the use of much salt. Perhaps this unnatural
source will be manifested in a "unique" mass flux signature,
representative of seasonal application and recharge.
FECAL COLIFORM
Fecal coliform bacteria is found at all sampling sites
(Figure 7). These bacteria are common in wastes of all healthy
warm-blooded animals. By far, basins with high occurrences of
dairies, feed-lots and urban areas are characterized by high
levels of fecal coliform.
A certain amount of fecal coliform can be attributed to wildlife.
Note that the Park land basin (BCGR), representative of pristine
conditions, contains a fair amount of fecal coliform bacteria
(mean of 67 colonies per 100 ml), and discharges a relatively
stable 1.5 million colonies per second (not shown). Although the
latter number may appear high, a single gram of feces may contain
tens of millions to tens of billions of cells (Feachem, et al.,
1983) .
The heavy agriculture basin (THNS), with hundreds of homes
without proper waste treatment facilities, and scores of dairies
and feed-lots where live-stock waste flows as sinking-creeks into
the aquifer yielded the highest overall fecal coliform levels.
Feachem et al. (1983) reports that although fecal coliform
density per gram of feces of man and livestock are comparable, a
human may excrete 150 grams of feces per day compared to 15 to 20
kilograms for a cow. The highest flow weighted value, greater
than 535 million colonies/second, was observed at this spring on
April 10, 1991. As samples are taken on a set monthly date,
regardless of weather or flow conditions, over the three-year
period of monitoring some flood pulses are likely to be sampled.
April 1991 was such an occurrence. It is important to note the
relative temporal position within the flood pulse from which the
sample was taken. A great amount of variance in parameter
concentration may exist throughout high discharge periods of a
flood pulse. It is not possible to tell from one sample its
708
-------
5000
4000
E
o
o
o 3000
cc
o
U-
o 2000
o
o
UJ
1000
0
maximum
T 95 percentile
75 percentile
mean
med ian
LTJ 2.5 percenlile
-L 5 percentile
* minimum
±
A
J.
i.
PSPS ERES THNS
MONITORING SITES
BCGR
Figure 7. Statistical examination of fecal coliform bacteria
levels, March 1990 through September 1991, of four
monitoring sites.
temporal relationship to concentration or mass flux peaks of a
particular parameter.
The chance occurrence of flood pulse activity coinciding with a
predetermined sampling date tends to create a large variance in
reported concentrations and mass fluxes of fecal coliform
bacteria. A high variance may indicate, as in the heavy
agriculture basin, the presence of large amounts of animal waste
stored at the surface, awaiting release by a rainfall event.
Notice that low values dominate the data set in the park
land/heavy agriculture basin (ERES). Occasionally these low
levels of fecal coliform are interrupted by brief periods of very
high concentrations, as expressed in the elevated mean and
maximum values. Recall the overflow route mentioned earlier.
During high-stage times, a portion of the bacteria-laden waters
of the heavy agriculture basin are shunted into the relatively
clean park basin.
709
-------
TRIAZINE-CLASS HERBICIDES
The monitoring program indicates the presence of triazine-class
herbicides (greater than 1 part per billion) within the surface
and ground waters of the park. The occurrence of these compounds
generally coincided with the peak application period. To avoid
costly organic laboratory testing for these compounds, the
program employs assay screening tests. Although gas
chromatography analysis would indeed be desired, laboratory costs
of a couple of sampling rounds would destroy the monitoring
budget. Assay-screening can not be thought of as a quantitative
analysis. It is used primarily as a "hit-or-miss" technique,
with semi-quantitative values (ie, greater than 1 ppb).
The spatial and temporal occurrence of triazine-class herbicides
reflect land-use, herbicide application periods, and perhaps the
mechanism of transfer between the surface and subsurface
(Figure 8). With the exception of a back-flooding of triazine-
tainted river water, the only groundwater sampling site in which
triazines were found was the heavy agriculture basin (THNS)
spring. Additionally, triazines were only found in months (June
1990, June and July 1991) following peak application periods
within the basin. Triazines are also found at both river sites
following peak application.
For the remainder of the year no triazine-class herbicide
residues were found in the sampled springs. Although rapid
TRIAZINE CLASS HERBICIDES
TRIAZINES ppb
A M
I I • I • I ' I • I
J J A 8 O N
1990
i i i t i i
D J F M A M
MONTHS
T—^— II
J J A S
1991
PARK LAND BASIN
GREEN RIVER
HEAVY AGRICULTURE
NOLIN RIVER
Figure 8. Temporal occurrences of triazine-class herbicides using
immuno-assay methods, March 1990 through September
1991, at four monitoring sites.
710
-------
transport of these residues through the karst system is expected,
one may not assume that all, or even the majority of these
compounds that will move through the aquifer have done so.
Research in Iowa by Hallberg et al. (1985) found that although
large amounts of herbicides are quickly transported through the
karst system via run-off following rainfall, the bulk of these
materials are slowly released through infiltration in low
concentrations. It would not be surprising to see a similar
pattern of pesticide transfer through the Mammoth Cave aquifer.
Aside from occurrence following peak application periods,
triazine-class herbicides were found in both the Green and Nolin
rivers in the fall of 1990, and possibly in the fall of 1991.
Two scenarios may be possible: 1) There was a late application
of these compounds in the fall, or, 2) The residues were slowly
transferred through a less permeable media (clastic strata).
As triazines are applied as pre-plant or pre-emergence
herbicides, there is no reason to believe that there were a late
applications, as crops that receive triazines (corn, and to a
lesser degree, soybeans) were near harvest.
River flood plains, with associated unconsolidated fluvial
deposits, are favored lands for row-crop production. It may be
possible that these persistent compounds may: 1) become entrained
in run-off shortly after application, and 2) slowly infiltrate
through the fluvial materials and leached into the river
following fall rains. The "half-life" of these compounds (3-12
months) is certainly sufficient to cause such persistency.
REFERENCES CITED
Feachem, R.G., Bradley, D.J., Garelick, H., and Mara, D.D., 1983,
Sanitation and disease, health aspects of excreta and
wastewater management, Chapter 13: John Wiley and Sons
Publ., p.199-242.
Hallberg, G. R., Libra, R. D., and Hoyer, B. E., 1985, Nonpoint
source contamination of ground water in karst — carbonate
aquifers in Iowa: Perspectives on Nonpoint Source Pollution,
U.S. Environmental Protection Agency, 440/5-85-001, p. 109-
114.
Meiman, J., Ewers, R.O., and Quinlan, J.E., 1988; Investigation
of Flood Pulse Movement Through a Maturely Karstified
Aquifer at Mcimmoth Cave National Park, a New Approach, NWWA
Conference on Environmental Problems in Karst Terranes and
Their Solutions, Nashville, TN, 1988, p 227-262.
Meiman, J., 1989, Investigation of flood pulse movement through a
maturely karstified aquifer at Mammoth Cave Kentucky: M.S.
thesis, Eastern Kentucky University, 343 p.
711
-------
Quinlan, J.F., and Rowe, D.R., 1978, Hydrology and water quality
in the Central Kentucky Karst: Phase II, Part A:
Preliminary summary of the hydrogeology of the Mill
Hole Sub-basin of the Turnhole Spring Groundwater
Basin: Kentucky Water Resource Research Institution,
Research Report no. 109, 41 p.
BIOGRAPHIC SKETCH
Joe Meiman, a Kentuckian, is currently the Hydrologist at Mammoth
Cave National Park. His B.S. in Geology (1985) and M.S. in
Geology/Hydrogeology (1989) were earned at Eastern Kentucky
University. At the present he is kept busy by directing
hydrologic research at the park, which includes: water quality
monitoring, dye-tracing, three-dimensional schematic karst
aquifer modeling, research and development of monitoring
equipment, and the general quest of scientific knowledge.
712
-------
The Effects of Recharge Basin Land-use Practices on Water
Quality at Mammoth Cave National Park, Kentucky
Joe Meiman
Your study is interesting and has some novel, admirable
interpretations of the data. Having said that, however, I am
surprised that samples were not also collected frequently
(perhaps at 1- to 4-hour intervals) and analyzed at several sites
throughout several storm events. This is more important than
monthly non-conditional synoptic sampling—if you want to
understand what is happening in the groundwater basins. The 3-
year monthly sampling program is incapable of having any
significant statistical validity because it is not tied into
analysis of storm events and has too short a duration. There
already exist hundreds of analysis of samples collected weekly to
monthly at many of your sites by Jack Hess (Ph.D. dissertation,
1974), the U.S. Geological Survey, and others. Before your
sampling began, the National Park Service was repeatedly advised
in writing, by specialists in monitoring water quality in karst
terranes of the existence of such data, and that sampling of
storm events could yield more useful data and insights than
monthly sampling. Why has the Park Service ignored sampling and
characterization of storm events?
As I stated in the presentation, monthly non-conditional synoptic
sampling is only a portion of the Park's water quality monitoring
program. Indeed, storm event sampling will comprise a large
measure of the monitoring during the coming year, as was
discussed. I am currently using storm-event (flood-pulse) data,
recorded as often as every two minutes, in our present study. As
work in this area continues, we will report our results when
sufficient flood-pulse water quality data have been collected. I
am aware of the excellent study by Dr. Hess, who primarily
focused upon carbonate chemistry variations during flood-pulse
activity. We have expanded our scope of parameters to include
bacteria, nutrients and brine-associated ions. Synoptic data
generated by the studies of the U.S. Geological Survey during the
1950's and 1960's are rather spatially limited. We have
increased sampling sites to capture data related to the myriad of
lands uses within the recharge basins. Past works serve as
skeletal datums of which data of the current study will be
compared. Temporal comparison is the essence of monitoring. Be
assured that the National Park Service has not entered this
program without careful consideration of past work in the region,
or the dynamic nature of the karst aquifer.
713
-------
714
-------
THE SINKHOLE COLLAPSE OF THE LEWISTON, MINNESOTA
WASTE WATER TREATMENT FACILITY LAGOON
Nancy O. Jannik, E. Calvin Alexander, Jr, and Lawrence J. Landherr
Department of Geology, Winona State University
Winona, MN 55987-5838
Department of Geology and Geophysics, University of Minnesota
Minneapolis, MN 55455-0219
Minnesota Pollution Control Agency, Regional Director
Rochester, MN 55904
ABSTRACT
On February 20, 1991, city workers discovered a sinkhole collapse in the
Lewiston, MN waste water treatment facility (WWTF) lagoons. The collapse
apparently occurred during the preceding few days and drained an estimated
7.7 million gallons of partially treated effluent into the local ground
water system. A temporary dike was constructed to isolate the sinkhole
from the rest of the lagoon. Subsequent, ad hoc testing for coliform
bacteria and nitrates did not detect evidence of effluent from the lagoon
in nearby residential wells. Following a shallow (20 foot penetration)
geophysical investigation using ground penetrating radar and an
electromagnetic survey, the city decided to fill the sinkhole and to erect
a dike around the collapse. The collapse was repaired in May, 1991 and the
lagoon returned to full operation.
The 1991 Lewiston collapse follows the nearby, 1974 and 1976 collapses of
the Altura, Minnesota WWTF lagoon (Liesch, 1977; Alexander and Book,
1984) . Two of the 7 to 10 WWTF lagoons constructed on the Ordovician
Prairie du Chien Group carbonates in the southeastern Minnesota karst
terrain have catastrophically failed in less than 20 years. That
corresponds to a failure rate of over 20% for these million dollar WWTFs --
so far. The federal programs that cost-shared the bulk of the construction
expenses for these WWTFs no longer exist. The cost of potential damages,
remediation, and/or replacement of these WWTFs falls directly on the state
and local units of government.
715
-------
INTRODUCTION
Southeastern Minnesota is an active karst area. Geomorphic features
associated with the karst include sinkholes, enlarged joints, numerous
springs, disappearing streams, cave systems, and dry valleys. There are
problems with ground-water quality ranging from occasional high levels of
selected parameters to chronic sub-standard drinking water conditions in
the hydrogeologically sensitive area.
The region is characterized by farms, small towns, and a few moderate-
sized cities. Many of the community centers have waste treatment
facilities that consist of a series of settling ponds, or lagoons.
Lewiston, Minnesota is one of the small towns located within this karst
region. This paper documents the failure of one of Lewiston's ponds due
to the instantaneous collapse of a sinkhole.
PHYSICAL SETTING
Topography
Lewiston is located in southeast Minnesota, in Winona County (Figure 1).
The region surrounding Lewiston is characterized by gently rolling hills
and swales with local relief of about 20 m (Figure 1). Sinkholes and dry
valleys are evident at the surface. A very thin soil, ranging in
thickness from 0 to about 15 m, covers the bedrock. The source material
for the soil is both residuum and/or glacial tills and loess.
I QUADRANGLE LOCATION
Figure 1. Portion of the topographic map in the vicinity of Lewiston,
Lewiston, Minnesota Quadrangle.
716
-------
Climate
At present, the area has a temperate climate, with a mean annual
temperature of 7.6°, and an annual precipitation of 75 cm (NOAA, 1978).
This region has experienced climatic changes, most recently, the changes
that resulted in Pleistocene glaciation.
Geology
The region is underlain by a series of lower Ordovician and Cambrian
sandstone and carbonate units (Figure 2) . The units were deposited in
nearshore to shallow-sea environments, and exhibit typical vertical and
lateral facies changes associated with sedimentation during Transgression
and regression of the shallow sea. The strata dip gently to the southwest
towards the center of the Hollandale Embayment (Mossier and Book, 1984) .
The units of most concern are the Jordan Sandstone and the Oneota
Dolomite.
SYSTEM
SERIES
|
_i
UPPER CAMBRIAN
GROUP OR
FORMATION
NAME
PRAIRIE DU CHIEN GROUP
SHAKOPEE
FORMATION
ONEOTA
DOLOMITE
JORDAN
SANDSTONE
ST. LAWRENCE1
FORMATION
FRANCON1A1
FORMATION
KONTON &
GALESVILLE
SANDSTONES
EAU CLAIRE2
FORMATION
MT. SIMON2
SANDSTONE
PRECAMBRIAN3
SYM
BOL
Opm
4
*--
€1
.ft
f-i-.
|
do
€t
/**, "
/ /CO/ /CV
. A s'cr> / &/~,
-,*••' :,•••',• /A
'-.' ', * , ',-}
<7 -^->rx^f
• ;&[.••*( ,/&)
S«a?i»£Sik
-ii .^ — ".'^r... ~j~ -=
-±,-:..^---J-i
- L i. 1 'i -/
.,6T°77
:'• o > -G I./
•'Vfrif . ' 'mrrf
.•XT : '•••
:-T:G ••'•'> -':\
&s/:
a^si^^sE^
- 6 - . G- )
•.".•'" - G '
4-0 .; G -6 y
•"t: »>&ss
-------
The Jordan Sandstone is Cambrian in age and averages 30 m in thickness.
The Jordan is a massive, upward-grading, fine- to coarse-grained friable
sandstone. Upward in the unit, it becomes progressively more indurated
with carbonate and siliceous cements, first forming lenses and concretions
and then well-bedded, highly lithified strata.
The Ordovician Prairie du Chien Group conformably overlies the Jordan
(Figure 2). The Prairie du Chien is composed of the Oneota Dolomite and
upper Shakopee Formation. The Oneota Dolomite is about 60 m thick, and is
fine- to medium-grained, thick- to thin-bedded to massive, with calcite-
filled vugs in the upper portion, and minor chert nodules throughout the
unit (Mossier and Book, 1984). Both drill cores and outcrops reveal that
the dolomite is highly jointed and has undergone extensive solution. The
dolomite is vuggy to cavernous particularly in the upper portion.
The Shakopee Formation is subdivided into the lower New Richmond
Sandstone member and the upper Willow River Dolomite member. The latter
is not present in the area and is not discussed further. The New Richmond
Sandstone of the Shakopee Formation is a fine- to medium-grained quartzose
sandstone with infrequent interbedded medium-grained arenaceous carbonate
beds. This sandstone unit averages about 6 m in thickness, is friable,
extensively jointed, easily eroded, and does not form many outcrops.
Hydrology
Surface flow is in small headwater channels of the Whitewater and Root
Rivers. The channels are characterized by meander development and easily
credible banks. Some surface run off flows into sinkholes. Regional
ground-water flow is east-northeast toward the Mississippi River. Local
ground-water flow is toward discharge points such as small tributaries or
springs.
Joints are common throughout the Jordan Sandstone and springs in the
well-lithified portions tend to discharge directly from joints. In the
more friable lower part, springs are often a combination of discrete flow
from joints and diffuse flow from numerous seeps. The Jordan is a major
source of water for wells in the area.
Only a few springs, confined to discharge from well-developed joints,
have been mapped in the Oneota. Few wells in the area rely solely on the
Oneota as a water supply. However, many older wells are open holes
through the Oneota. The New Richmond Sandstone member of the Shakopee
Formation has a few springs which emerge form the New Richmond/Oneota
contact. The New Richmond is not a significant aquifer in the area.
718
-------
Karst Features
Numerous karst features such as sinkholes, enlarged joints, springs, dry
valleys, and small caves have formed in the Oneota Formation of the
Prairie du Chien Group (Figure 3). Sediment-filled solution cavities are
common features in outcrops and quarry walls that expose the Oneota
Dolomite. The karstification of the Oneota probably began during the
Ordovician and has continued intermittently until the present. The region
would be classified as fluviokarst according to the scheme used by
Sweeting (1973) because both karst and fluvial processes have contributed
to the development of the features that are evident, or that are being
exhumed.
Sinkholes are by far the most dominant karst feature. Historically, if
the holes are left in the natural state, they are either fenced-in and
left to be naturally vegetated, or several have been used as backyard
landfills. In the past, several have been filled with debris and soil and
then used as farm land. Many of the sinkholes in the area have developed
catastrophically, often in the spring of the year in response to unusually
wet conditions. It appears that the sinkholes develop through the New
Richmond Sandstone into the underlying Oneota Dolomite.
sinkholes
dry valley
.JORDAN • •.'.' .-,- •. ••
springs
Typical relationships in Winona County between karst and topography
Figure 3. Block diagram of karst landforms in southeast Minnesota
(Dalgleish and Alexander, 1984).
719
-------
Lewiston, and the immediate vicinity, were classified as high
probability of sinkhole development by Dalgleish and Alexander (1984).
The classification was based on the observed density of sinkholes,
together with information on the bedrock geology, surficial geology, and
hydrogeology. Dalgleish and Alexander (1984) conclude that the carbonate
bedrock is the primary control on sinkhole formation. Secondary controls
include the type and thickness of the overburden, and the depth of the
water table. Areas where the Oneota Dolomite is overlain by the sandstone
member of the Shakopee Formation, such as in the vicinity of Lewiston,
are the most susceptible to sinkhole development. Fractures in the
noncalcareous sandstone act as conduits to preferentially direct surface
water into the Oneota.
THE SINKHOLE COLLAPSE OF THE LEWISTON WASTE WATER TREATMENT
FACILITY LAGOON
Background
The Waste Water Treatment Facility at Lewiston, (population -1300) is
constructed in an area that overlies the New Richmond Sandstone member of
the Shakopee Formation and the Oneota Dolomite, and has less than 30 m of
regolith. The Waste Water Treatment Facility consists of a series of
settling ponds or lagoons which is commonly known as a "natural" treatment
system (Figure 4) . These types of systems are common in southeast
Minnesota. The one at Lewiston is about 20 years old. Exposure of the
waste water allows oxygen and sunlight, together with microorganism to
"treat" the effluent after initial screening for large-sized solids.
Machines or chemicals are not used. The treated water is then discharged
to a surface-water channel. This natural purification takes about 6
months.
FENCE OH PROPERTY UNE
t«' DOUflU CAIE
Figure 4. Schematic of the Lewiston WWTF ("From Braun, Intertec, 1991)
720
-------
According to the superintendent of the Lewiston's sewer and water system
Lewiston had wanted a mechanical system 20 years ago, but was denied the
request by the Minnesota Pollution Control Agency. Supposedly, the
decision was based on the consensus that small towns could not afford the
upkeep and the operating expenses of a mechanical plant (Rochester Post
and Bulletin; February 27, 1991). Today, that opinion is not held by any
state government agency. The change of opinion is based not on a town's
new-found ability to afford a plant, but because of the ground-water
quality problems associated with a karst terrain, and the documented
failure of a similar lagoon system in Altura, Minnesota (Alexander and
Book, 1984), which is about 10 km northwest of Lewiston.
Sinkhole collapse
On February 20, 1991, it was discovered that a sinkhole had opened on
the edge of sewage lagoon Number 2 (Figure 4) at the Waste Water Treatment
Facility at Lewiston, Minnesota. The sinkhole collapse caused a break in
the dike enclosing the lagoon. The collapse left a hole that was
approximately 12 m in diameter and 2-4 m in depth (Figure 5).
Figure 5. Photo of dike-side of sinkhole, on February 20, 1991. View is
northwest. Photo courtesy of R. Dunsmoor, Winona County.
721
-------
It is estimated that the collapse occurred on or about February 14,
1991. According to the records of city workers, approximately 7.7 million
gallons of semi-treated sewage effluent were lost from Lagoon Number 2.
The loss occurred over several hours to perhaps a day. The effluent
entered the ground through a conduit at the bottom of the sinkhole. The
waste water had been in the lagoon only about two months, and probably
still contained bacteria and/or viruses because it was covered with ice
which would prevent sunlight and heat from destroying them.
The sinkhole collapse at Lewiston has striking similarity to the
collapse at the Altura Waste Water Treatment Facility as documented by
Alexander and Book (1984) . Both collapses formed in the Oneota Dolomite
where it was overlain with the basal sandstone member of the Shakopee
Formation. At both locations, sinkhole collapse was catastrophic. One of
major differences is that the failure of the Altura lagoon was due to
several sinkholes in the bottom of the lagoon, whereas, the failure of the
Lewiston lagoon was due to a single sinkhole collapse near the edge of the
lagoon which led to breaching of the dike.
Response to the problem
The water level in Lagoon Number was 2 was lowered and continued to be
monitored so that further semi-treated water did not spill over into the
sinkhole. A dam was built around the sinkhole in order to prevent surface
run off from entering the hole.
Water-quality tests were performed on the city wells and on 11 private
wells in the area. The results did not detect contamination from the
effluent. Residents were advised to drink water from hot water heaters
that had been cooled, until they could have their well tested. They were
further advised to chlorinate their wells, and drink bottled water if they
had any concerns.
Remediation of the problem
The city hired a consulting firm in early March to determine the size of
the sinkhole and propose short term remediation. A shallow (6 m
penetration) geophysical investigation used ground-penetrating radar and
an electromagnetic survey to determine the limits of the observed sinkhole
and possible fractures in the vicinity. An independent proposal to run a
dye trace form the sinkhole in order to determine ground-water flow
patterns was not adopted by the city. This was due in part to the concern
of liability.
By mid May 1991, it was decided by city employees that the best response
to the sinkhole collapse was to repair the dike and seal the sinkhole.
Beginning May 21, 1991, the site was cleaned-up, and new dike was created
about 15 m from the surface expression of the sinkhole. According to city
employees, the sinkhole was excavated to within 1 m of bedrock. The hole
was then filled and sealed by May 24, 1991.
722
-------
POTENTIAL FAILURE OF OTHER WASTE WATER TREATMENT FACILITIES
The Minnesota Pollution Control Agency has compiled a list of towns in
southeast Minnesota with waste-water pond facilities similar to that at
Lewiston. The screening for the list included those sites which are
situated over karstic bedrock and had less than 30 m of soil or till
above the bedrock. The initial list, which is being refined at this time,
includes 14 sites. Of those 14, 10 are considered to have high
potential for failure, and 4, low potential. Of the 10 that have high
potential for failure, 2 have had failures within the past 20 years.
This corresponds to a failure rate of about 20%.
Costs for potential damages, repair and/or remediation for these Waste
Water Treatment Facilities is now the responsibility of the local units of
government. The Federal programs that cost-shared the bulk of the
original construction costs have been severely reduced or phased out. It
is recognized that the use of a lagoon system in a karst region can lead
to catastrophic failures and potential health concerns. However, the
sealing of the sinkholes not only in, or next to, sewage ponds, but in
other sensitive areas, is often the remedial method of choice due to
economics. A new sewage treatment plant would cost about 1.5 million
dollars which is an exorbitant financial burden for a small town. These
small towns are forced either to spend millions of dollars on new
construction or risk potential liability suits. Neither choice is good.
REFERENCES
Braun Intertec Environmental, Inc. 1991. Hydrogeologic investigation of
the sinkhole at the City of Lewiston Waste Water Treatment Facility.
Project No. BCBX-91-033A.
Dalgleish, J.B. and E.G. Alexander, Jr. 1984. Geologic Atlas of Winona
County, Minnesota, Plate 5. In: N.H. Balaban and B.M. Olsen (eds.), County
Atlas Series C-2, Minnesota Geological Survey. University of Minnesota.
Liesch, B.A. 1977. Hydrogeologic Investigation of the Altura area for
Altura, Minnesota. Report dated November 1974. 18 p.
Mossier, J.H., and P.R. Book. 1984. Geologic Atlas of Winona County,
Minnesota, Plate 2. In: N.H. Balaban and B.M. Olsen (eds.), County Atlas
Series C-2, Minnesota Geological Survey. University of Minnesota.
Sweeting, M.M. 1973. Karst Landforms. Columbia Press, New York. 362 p.
723
-------
BIOGRAPHICAL SKETCHES
Nancy 0. Jannik is an Associate Professor in the Department of Geology at
Winona State University. She has a B.S. in geology (1976) from the College
of William and Mary, a M.S. in geology (1980) from Rutgers University, and
a Ph.D. in geosciences/hydrology (1988) from the New Mexico Institute of
Mining and Technology. Her current research involves determining aquifer
characteristics, aquifer interactions, and the inter-relationships between
surface and ground water in the karst terrain of southeastern Minnesota.
Her full address is as follows.
Department of Geology
Winona State University
Winona, MN 55987-5838
(507) 457-5267
E. Calvin Alexander, Jr., is a Professor in the Department of Geology and
Geophysics at the University of Minnesota. He has a B.S. in chemistry
(1966) from Oklahoma State University and a Ph.D. in chemistry (1970) from
the University of Missouri at Rolla. The central theme of his current
research is the rate of movement of fluids in hydrogeology. He and his
research group are utilizing a variety of methods to measure flow and
residence times, which can range from hours to ten of thousands of years.
His full address is as follows.
Department of Geology and Geophysics
University of Minnesota
Minneapolis, MN 55455-0219
(612) 624-3517
Lawrence J. Landherr is the Regional Director of the Minnesota Pollution
Control Agency (MPCA) Southeast Region. He has a B.S. in biology (1967)
from the University of Minnesota. He has 24 years working experience in
environmental programs in the karst of southeastern Minnesota. His current
responsibilities involve managing the activities of the eight person staff
involved in all program areas of the MPCA. His full address is as follows.
MPCA Regional Director
2116 Campus Dr. S.E.
Rochester, MN 55904
(507) 285-7343
724
-------
Erosion and Sedimentation Control Methodologies for
Construction Activities over the Edwards Aquifer in Texas
Hank B. Smith, P.E.
Texas Water Commission
Austin, Texas
I. Abstract
Through its implementation of the Edwards Aquifer Program, the
Texas Water Commission strives to preserve the quality of water
produced from this karst aquifer. The Texas Water Commission
(TWC) has rules regarding development over the Edwards Aquifer
Recharge Zone which require erosion and sedimentation controls.
During construction-related activities, erosion and
sedimentation are the primary mechanisms for aquifer
contamination. Temporary erosion and sedimentation control
structures are necessary to ensure that suspended sediments in
stormwater remain within the construction area. These control
structures filter sediments and include hay bales, filter
fences, rock berms, and brush berms.
Sediments that are deposited in streams and lakes impact benthic
organisms, reduce storage capacity of lakes, and reduce water
quality. Sedimentation also impairs natural groundwater
recharge which results in the reduction of base flow in creeks,
destruction of seeps and springs, and loss of groundwater
supplies.
Erosion occurs when natural vegetation that binds soils together
is disturbed. During rainfall events, natural vegetative cover
acts as a filter to remove pollutants, and reduces runoff
velocity- During construction related activities, natural
vegetative cover is disturbed. As a consequence, Total
Suspended Solids (TSS) in stormwater runoff is greatly
increased.
Stormwater runoff from urban areas has increased levels of
nutrients (nitrogen, phosphorus, etc.) from fertilizers, and
pesticides. Other pollutants come from asphalt roofing
materials, sewage disposal facilities, and other forms of non-
point source pollution. Stormwater runoff from roadways
725
-------
typically has increased levels of lead, total petroleum
hydrocarbons, oil & grease, and metals.
The Edwards Aquifer Program requires permanent sedimentation
control structures which improve the quality of stormwater
runoff and provide protection of water resources. Permanent
control structures include sand filtration and sedimentation
basins and hazardous materials traps.
II. The Edwards Aquifer Protection Program in Texas
Effective March 21, 1990, the Texas Water Commission (TWC)
adopted revised rules requiring the preparation and submission
of a Water Pollution Abatement Plan (WPAP) for any regulated
developments over the Edwards Aquifer recharge zone in Texas.
The purpose of the rules is to regulate activities having
potential for polluting the Edwards Aquifer. The activities
addressed are those that pose direct threats to water quality.
Nothing in the rules is intended to restrict the powers of the
TWC or any other governmental entity to prevent, correct, or
curtail activities that result or potentially result in
pollution of the Edwards Aquifer. Each WPAP must contain, at a
minimum, the following information:
1) name, address, and telephone number of owner, agent,
developer and contact persons;
2) general location map which identifies the limits of the
recharge zone, and the location of the proposed
development;
3) site plan with proposed development, flood plain
information, location of known wells, existing and proposed
drainage patterns, and identification of any known
significant recharge features;
4) detailed assessment of area geology which includes:
geologic map at site plan scale; stratigraphic columns;
narrative description of surface geology, soil profile and
units; and, a narrative description of all significant
recharge features; and
5) detailed technical report which includes: description of
the proposed development; volume and character of
wastewater and stormwater expected to be generated from the
site; description of methods to be taken to prevent
pollution of stormwater originating on-site and upstream of
the site; description of methods to be used to prevent
pollutants from entering recharge features; and a
description of method of disposal of wastewater.
Upon receiving final approval of the WPAP, the applicant must
deed record the approval in the County Courthouse. Any
modifications to the plans must be submitted to and approved by
the TWC prior to construction. The TWC must be notified prior
to initiation of construction. If any significant recharge
features are identified, all construction activities in the
726
-------
immediate area must be halted upon discovery until protection
plans can be developed, reviewed and approved by the TWC.
All organized sewage collection systems must also be approved by
the TWC prior to initiation of construction. At a minimum, the
plans must comply with the following design criteria.
1) all manholes must be watertight with watertight rings and
covers. Alternate means of venting must be provided for
every third manhole. All manholes must also be tested for
water-tightness prior to use.
2) lift stations must be designed to ensure that bypassing
does not occur. All lift stations must have audible and
visual alarms, as well as, auto-dial telemetry to notify a
series of locations with 24-hour response coverage.
3) all gravity and pressure sewage lines, including private
service laterals and manholes, must be inspected and tested
to ensure that no leakage occurs after construction.
4) all wastewater systems must be tested every five years to
determine the types and locations of structural damage and
defects such as offset or open joints, or cracked or
crushed lines that could allow exfiltration to occur.
Additional requirements provide that new or increased discharges
of treated effluent that would create increased loading of
treated wastewater are prohibited over the recharge zone. Land
application systems that rely on percolation are prohibited.
However, irrigation systems may be allowed and are reviewed on
a case-by-case basis.
All WPAP's and sewage collection system plans must include
erosion and sedimentation control plans. These plans shall
provide designs for control structures that ensure stormwater
contaminated during construction activities is not allowed to
discharge from the site without treatment. Some of the effects
of uncontrolled erosion and sedimentation from construction
areas include:
1) changes in habitat and localized natural land uses;
2) increased flood plain depths;
3) reduced volumes of water storage in rivers and lakes
resulting in increased down-stream flooding;
4) increased eutrophication rates from excessive nutrient
loadings; and
5) loss of top soil.
Ill. Controlling Erosion and Sedimentation
During construction related activities erosion is the primary
concern. The major part of the erosion potential exists between
the time when the native vegetation is removed and when the site
construction is complete and the vegetation is restored. These
impacts can be minimized with the proper design, installation,
727
-------
and maintenance of temporary erosion and sedimentation control
structures. In addition to temporary control structures,
impacts can be reduced by limiting the amount of disturbed areas
by identifying a construction area and limiting access to this
area. Temporary erosion controls can be designed and installed
within the construction areas to provide four basic functions:
Diversion - by diverting off-site flows around disturbed areas
the amount of runoff collecting sediment and the volume of water
that must be treated can be minimized. Diversions can also be
used to direct runoff from drainage areas to a location where
sedimentation can occur more cost effectively.
Ponding - by installing controls that reduce the runoff
velocities the sediments are allowed to settle out and can be
removed from the pond on a routine basis. Ponding is one of the
primary methods of sediment removal used during and after
construction.
Filtration - when stormwater flow is routed through a filtration
structure, soil particles are trapped on the filter media, and
sediment settles out up-gradient of the filter structure.
Filter structures can be used to capture the fine portions of
the sediment and are used extensively during construction.
Flow spreading - by installing controls that change the runoff
characteristics from concentrated to sheet flow, the velocity of
the runoff is reduced which increases local sedimentation and
minimizes erosion.
The effectiveness of any temporary erosion and sedimentation
control structure is dependent upon proper installation and
adequate maintenance. The following guideline can be used to
approximate the type of structure that is appropriate for
various conditions.
Type of Structure
Hay Bales
Silt Fence
Brush Berm
Rock Berm
Reach Length
< 100 feet
n/a
< 200 feet
< 100 feet
< 50 feet
n/a
< 150 feet
< 75 feet
500 feet
Drainage area
1/2 Acre
2 Acres
2 Acres
1 Acre
1/2 Acre
1 1/2 Acres
1 Acre
1/2 Acre
5 Acres
Slope
0 - 10%
0 - 10%
10 - 20%
20 - 30%
> 30%
0 - 10%
10 - 20%
20 - 30%
> 30%
0 - 10%
For drainage areas greater than 5 acres, the design of
structural features are reviewed on a case-by-case basis.
Approval is based on the anticipated quantity and quality of
stormwater, and the location of any critical environmental
features downstream.
728
-------
Hay bales are an inexpensive and readily available control
structure; however, they are difficult to maintain, must be
replaced following every rainfall, and are ineffective for any
drainage area larger than 0.5 acres. The TWC does not recommend
the use of hay bales except in limited areas where construction
times will be short and rainfall is not anticipated.
Filter fabric may be required within the rock berms if
determined necessary by the TWC. These structures have been
proven effective in steep canyons, above permanent springs,
pools or other environmentally sensitive areas. This type of
structure combines the removal efficiency of silt fences with
the structural stability of rock berms.
IV- Design Calculations
Temporary erosion and sedimentation control plans must include
calculations to verify that the proposed structures are
adequately designed. The calculations involve establishing a
storm flow volume for the drainage area for each structure, and
an appropriate barrier flow-through rate. The computational
procedure for establishing a flow-through rate includes several
calculations such as coefficient of roughness, time of
concentration, and rainfall intensity.
Using the "Rational Method" (Q = C * i * A) the volume "Q" of
flow can be calculated for various drainage areas with the
following parameters:
1) determine the drainage area "A" for each structure or
drainage basin based on the site plan;
2) determine rainfall intensity "i";
3) estimate the runoff coefficient "C" for a two year rainfall
storm for each specific soil type and channel slope;
4) compute peak flow rate of runoff "Q" for a two year storm
using the rational method;
5) determine frontal area for flow "Af" at the barrier; and
6) determine the flow through rate at the barrier using the
following "q = Q / A,"
Based on the flow through rate "q" the proper structure is
chosen from the following Table 1:
729
-------
Table 1
Sheet; Flow
Hay Bales
Filter
Dikes
Brush Berm
Silt Fence
Rock Berm
Diversion
Dike, etc.
Swales
Concentrated
Flow
Stone
Outlets
Diversion
Dike, etc.
Swales
Rock Berm
Slope Drain
Sediment
Traps
Sediment
Basins
Stream Flow
Rock Berm
Sandbag
Berm
Maximum Flow
Through Rate
(Gal/Min/SgFt}
5
20
40
40
60
40
60
Trenched
or
Secured
Yes
Yes
Yes
Yes
No
Yes
Wire
Mesh
Sheath
Wire
Mesh
Sheath
No
Recommended
Maximum
Drainage
Area
. 5 acre
1 acre
2 acre
2 acre
5 acres
5 acres
5 acres
5 acres
5 acres
5 acres
5 acres
5 acres
5 acres
100 acre
+5 acres
+5 acres
Recommended
Maximum slope
Length
50 feet for
slopes exceeding
10%
50 feet for
slopes exceeding
10%
730
-------
V. Additional Structures
Temporary erosion and sedimentation control plans should also
provide for stabilized construction entrances, equipment and
petroleum storage areas, and spoils disposal sites. An
inspection program and a maintenance plan and schedule should
also be included.
Stabilized construction entrances consist of a stabilized pad of
open graded rock located at any point where traffic may be
entering or leaving a construction site to a public right-of-
way, street, alley, etc. The purpose of a stabilized
construction entrance is to reduce or eliminate the tracking or
flowing of sediment onto public roadways where they contribute
to non-point source pollution.
Equipment storage areas, and temporary fuel and oil storage
facilities should be designated on the site plans. All
construction vehicles should be parked in protected areas when
not in use, and all maintenance of vehicles should occur within
these limits. These areas should be placed at topographical
highs, or berms should be provided to divert stormwater flow
around the site. Structural controls must be placed
downgradient of the site to filter stormwater runoff. Special
precautions should be taken to prevent the spill of petroleum
products from storage facilities or from emergency maintenance
activities. General scheduled maintenance is not allowed on-
site.
Spoil disposal and material storage areas should also be
designated on the site plans. Spoils material from the
construction activity, sediment removed from temporary
protective structures, and construction material (backfill,
sand, base material, etc.) must all be stored in these areas.
These areas should be protected from stormwater run-on.
Stormwater discharges from the site should be filtered to remove
potential pollutants.
Temporary erosion and sedimentation control structures should be
inspected routinely to ensure effective control of stormwater.
Weekly inspections should be conducted of the entire area, and
inspections should be conducted within 24 hours of any rainfall
event. Inspection forms should be developed and should be
maintained at the construction site.
Typically, the contractor should be required to routinely clean
paved surfaces with appropriate power brooms, or vacuum devices
to minimize the amount of sediment within the construction area.
A temporary erosion and sedimentation control maintenance plan
should be developed prior to initiation of construction. These
maintenance plans should provide a schedule for the removal and
replacement of temporary structures that are no longer
effective. Provisions should be made for the removal of silt,
731
-------
and sediments following rainfalls. Troublesome areas should be
identified and evaluated for improved structural controls.
VI. Cooperative Agreements
The TWC also enters into Cooperative Agreements with other State
agencies to ensure that the agencies involved in development
activities can be leaders in environmental protection.
In September 1991, the Texas Department of Transportation and
the TWC entered into a cooperative agreement regarding the
construction of a major highway intersection over the Edwards
Aquifer. The agreement required that the Texas Department of
Transportation comply with all rules and requirements for the
submission of a Water Pollution Abatement Plan prior to
construction. The agreement also provided additional specific
requirements applicable only to the development at this
location, and included the following.
1) Provide a detailed temporary erosion and sedimentation
control maintenance plan and schedule that would establish
guidelines for inspection of structures, replacement of
inadequate or damaged structures, and increased protection
of any impacted areas.
2) Restore and maintain vegetative cover prior to completion
of construction.
3) Restrict construction vehicle access from unprotected
areas, provide routine street cleaning and employee parking
in stabilized areas.
4) Designate contact persons for the TWC, Texas Department of
Transportation, and the contractor.
5) Develop an interagency training program.
6) Develop a remediation plan and schedule in the event an
impacted area is identified.
The TWC also encourages input from other entities such as
municipalities, river authorities, conservation districts, and
environmental groups. Citizen monitoring can also provide
valuable information regarding impacted areas and problem areas
within existing developments.
VII. Estimated Annual Pollutant Load
The amount of total suspended solids, total phosphorus, and oil
and grease that may be discharged from a site before and after
development can be approximated. The annual pollutant load is
the pollutant concentration times the annual runoff volume. The
following calculation can be used to estimate the annual
pollutant load for a tract of land:
732
-------
L=AxRFxRvx 0.2266 x C, where:
L = Annual pollutant load in pounds
A = Area of tract in acres
RF = Annual rainfall amount in inches
RV = runoff to rainfall ratio
RV = 0.05 + (0.009 * 1C) where:
1C = impervious cover for site in percent
C = Pollutant concentration in milligrams per liter
Background "C" Developed "C"
Total Suspended Solids 48 130
Total Phosphorus 0.08 0.26
Oil and Grease 0 15
The area used in the annual pollutant load calculation for total
suspended solids and total phosphorus should represent only that
portion of the tract which is to be developed, not the entire
site area. Typically, this is the limits of construction for
the area. For oil and grease, the load calculation should only
use the paved area.
VIII. Permanent Erosion and Sedimentation Control
Sedimentation basins can be effective at removing suspended
solids from developed areas but are very limited in
effectiveness of removing dissolved solids. The basic design of
this type of basin is to capture and isolate a percentage of the
rainfall or a certain volume of stormwater, and discharge the
flow over a vegetative strip over a prolonged period of about 24
hours. A design to capture the first l" of stormwater flow will
generally be able to remove about 50% of the total suspended
solids, 15%-20% of the total phosphorus and about 8% of the oil
and grease content of the stormwater.
Sand filtration basins have been used in the Austin and Central
Texas area to effectively remove suspended solids, bacteria, and
other pollutants from developed areas. The basic design of
these structures is to capture and isolate up to the first inch
of stormwater runoff and release the flow through a porous sand
media over a prolonged period of about 24 hours. A design to
capture the first 1" of stormwater flow will remove
approximately 55%-70% of the total suspended solids, 25%-33% of
the total phosphorus and about 23%-30% of the oil and grease
content of the stormwater.
Sand filtration and sedimentation basins are often designed
together with the sedimentation basin discharging into the sand
filtration basin for optimum efficiency.
A formal maintenance plan and schedule is required to be
submitted to the TWC for review, possible modification, and
approval prior to completion of construction. The plan must
733
-------
include a responsible party and the anticipated cleaning and
monitoring schedule. The plan must also include a disposal site
for any material removed from the basins and an anticipated
removal schedule.
Plans should be developed to ensure that a copy of the WPAP and
approval letter is maintained at the construction site. The
requirements contained within the design plans and the approval
letter from the TWC should be reviewed with the contractor.
Other factors to consider during the development design phase
include land grading, stormwater outfall designs, grade
stabilizing structures, ponds and retarding structures,
streambank protection, and storm sewer design.
IX. Other Options
Street sweeping and parking lot sweeping activities can be used
to reduce levels of pollutant run-off from developed areas. The
effectiveness of street sweeping and vacuuming is difficult to
measure and is a function of how frequently it is performed, and
the percent of site that can be cleaned. Consideration must be
given to the pollutant load removed by non-sweeping processes
such as wind and biochemical processes. The following
efficiencies for sweeping activities are appropriate:
Broom Sweeper Vacuum Sweeper
Parameter Efficiency f%) Efficiency f%)
Total Solids 55 93
Total Phosphorus 40 74
Total Nitrogen 42 77
Chemical Oxygen Demand 31 63
Biochemical Oxygen Demand 43 77
Lead 35 76
Zinc 47 85
An important consideration in reviewing street sweeping options
is that 78% of the solids load is contained within an area six
inches from the curb, and that 97% of the solids load is within
40 inches from the curb. Some street sweeping devices only
redistribute the solids and actually remove only about half of
the solids. A significant percentage of the solid particles are
also of sizes which are not effectively removed by conventional
street sweepers.
X. Hazardous Spill Consideration
Spills of hazardous materials from roadways and other areas
where hazardous materials are stored, transported, or used
create potentially major impacts to surface and groundwater
quality. Hazardous materials traps (HMT's) can be at stormwater
outfalls and other locations that will protect our states' water
734
-------
resources from spill contamination.
HMT design typically incorporates concrete basins with a minimum
storage volume of 10,000 gallons. The design allows for the
first 10,000 gallons of runoff to be directed to a concrete
basin where it can be isolated. The average truck carries less
than 10,000 gallons of fuel, petroleum product or waste, sewage,
or other hazardous material. Any flow greater than the capacity
of the basin would initiate a siphon flow which would drain the
basin. In theory, any hazardous material spill will be of a
volume less than 10,000 gallons and would therefore be isolated
in the concrete basin. A series of valves could then be opened
or closed by emergency crews which would divert future flows
around the basin until the hazardous material can be removed.
Most designs for these basins are ineffective during rainfall
events since it is not practical to contain all rainfall or try
to differentiate between contaminated and uncontaminated
stormwater flows.
A formal maintenance plan and schedule is required to be
submitted to the TWC for review, possible modification, and
approval prior to completion of construction. The plan must
include a responsible party for monitoring the siphons and the
hazardous materials traps. The plan must also provide for
notification and training of local emergency response personnel.
XI. Summary
The TWC reviews water pollution abatement plans and sewage
collection systems prior to initiation of construction to ensure
adequate protection of environmental features. As a part of
this process, the TWC requires temporary erosion and
sedimentation control structures be designed, approved, and
installed prior to initiation of construction over the Edwards
aquifer. The TWC also requires adequate maintenance and routine
inspection of these structures.
Erosion and sedimentation from construction areas is a major
factor leading toward contamination of karst aquifers. By the
proper design, installation, and maintenance of temporary
erosion and sedimentation control structures the impacts of
construction can be minimized.
Permanent sedimentation and filtration basins, and hazardous
materials traps can be used to mitigate the impacts from
developed areas. The amount of pollutants from developed areas
can be estimated and using appropriate removal rates for
sedimentation and filtration structures the anticipated
additional loading to receiving waters can be estimated during
the design phase of developments.
735
-------
SILT FENCE
48" Min. Heavy Weight T-Post
24* Tall Mln., *"x 4' 12 G«.
1
i
T
48' 2
4 -^V
¥' <
\
4
r
/^&»J&
^ '
\
\
\
\
\
\
\
\
s
s
\
\
.^ \>Biv«nizca «*iic n »-
^^
,,* o* Mln. Non-Wove
^X^ Filter Fabric 36* Wl<
^ —
.SKIRT ANCHORS
/
1 Soil Level
*-ll-->-
Cross-Section
GEOTEXTILE
Installation
WOVEN WIRESUPPORT
2'X 4' WIRE MESH
Layout the silt fence following as closely as possible to the
contour.
Clear the ground of debris, rocks, plants (including grasses
taller than 2") so that a smooth surface can be utilized for
anchoring the skirt.
Drive the heavy duty T posts at least 12 inches in to the
ground and at a slight angle towards the flow.
Attach the 2" x 4" 12 Gauge welded wire mesh to the T posts
with 11 1/2 gauge galvanized T post clips. The top of the
wire should be 24" above ground level. The welded wire mesh
shall be overlapped 6" and tied at least 6 times with hog
rings.
The silt fence will be installed with a skirt a minimum of
11" wide placed on the uphill side of the fence in a direction
towards the anticipated runoff. The fabric should overlap the
top of the wire by 1".
Anchor the silt fence on alternating 2' centers front and
back so that there is only 1' between anchor points.
Geotextile splices should be a minimum of 18" wide attached
in at least 6 places. Splices in concentrated flow areas will
not be accepted. o
-------
ROCK BERM
16' Min.
Woven Wire Sheathing
3' TO 5* OPEN
GRADED ROCK
Cross-Section
Woven Wire Sheathing
Installation
Layout the rock berm following as closely as possible to the
contour.
Clear the area of debris, rocks or plants that will interfere
with installation.
Place woven wire fabric on the ground along the proposed
installation with enough overlap to completely encircle the
finished size of the berm.
Place the rock along the center of the wire to the designated
height.
Wrap the structure with the previously placed wire mesh secure
enough so that when walked across the structure retains it's
shape.
Secure with tie wire.
737
-------
STABILIZED CONSTRUCTION
ENTRANCE
Mutt Be Properly Graded
To Prevent Runoff From Leaving
The Construction Site \
Public
Riflht-of-Way
\
.., so* Min.
Cross-Section
GeotcxtlU
Undcrllncr
(If Required)
50* Mln.
Public
Rlflht-of-Way
Plan View
Installation
Clear the area of debris, rocks or plants that will interfere
with installation.
Grade the area for the entrance to flow back on to the
construction site. Runoff from the S.C.E. onto a public street
will not be accepted.
Place geotextile fabric if required.
Place rock as required.
738
-------
XII. References
Regional Resource Division; Northern Division Planning District
Commission, 1979, Guidebook for Screening Urban Non-point
Pollution Management Strategies
Hydroscience Inc.; 1977, Water Quality Management Planning
Methodology for Urban and Industrial Stormwater Needs
Lower Colorado River Authority (LCRA); Environmental Quality
Division, January 1991, LCRA Lake Travis Non-point Source
Pollution Control Ordinance
City of Austin; 1988, Erosion and Sedimentation Control Manual
City of Austin; Drainage Criteria Manual
City of Austin; Design Guidelines for Water Quality Control
Basins
739
-------
Erosion and Sedimentation Control Methodologies for
Construction Activities over the Edwards Aquifer in Texas
Hank B. Smith, P.E.
Texas Water Commission
Austin, Texas
Question - On environmental geology field trips in the Austin area,
I have noticed severe erosion problems in the stream channels as a
result of ephemeral storm runoff caused by urbanization. What is
the Texas Water Board doing about this problem?
Answer -
The Texas Water Commission (TWC) has rules regarding development
over the Edwards Aquifer Recharge Zone which require temporary
erosion and sedimentation controls. These control structures
filter sediments and include hay bales, filter fences, rock berms,
and brush berms.
The major part of the erosion potential exists between the time
when the native vegetation is removed and when the site
construction is complete and the vegetation is stabilized. Such
impact can be minimized with the proper design, installation, and
maintenance of temporary erosion and sedimentation control
structures. Additionally, negative impact can be reduced by
limiting the amount of disturbed areas by identifying a
construction area and limiting access to this area.
Contractors are required to routinely clean paved surfaces with
appropriate power brooms, or vacuum devices to minimize the amount
of sediment within the construction area.
The Edwards Aquifer Program requires permanent sedimentation
control structures which improve the quality of stormwater runoff
and provide protection of water resources. Sedimentation basins
can be effective at removing suspended solids from developed areas,
and sand filtration basins have been used in the Austin and Central
Texas area to effectively remove suspended solids, bacteria, and
other pollutants from developed areas.
740
-------
Case Histories of Several Approaches to Stormwater
Management in Urbanized Karst Terrain, Southwest Missouri
Wendell L. Earner
ICF Kaiser Engineers Inc.
Pittsburgh, Pennsylvania 15205-1017
Patricia Miller
California State University-Sacramento
Sacramento, California 95819-6049
Abstract
Because sinkholes in karst terrain are dynamic components of
surface-ground water interactions, they present particular
challenges in the design of urban stormwater systems. Sinkholes
may function as either drains or ponds for surface runoff, and
their hydrologic behavior may change markedly with long term
urban growth, as impacts of earlier development and off-site
influences are added to natural drainage patterns at proposed
construction sites. Comparison of three sites in Springfield,
Missouri, in the Springfield Plateau Karst, illustrates the
effects of contrasting approaches to stormwater design.
A wooded tract with unmodified sinkholes was cleared and
developed for residential use. Discharge of stormwater was
directed into sinkholes, and erosion control consisted of hydro-
mulching and sedimentation fences in sinkhole areas. East of
this location are two parcels which differ in removal of
vegetation and off-site drainage relationships. Stormwater
design at these sites was adapted for modifications made to
sinkholes during railroad and highway construction several
decades earlier. Sediment fencing, hydro-mulching, and detention
berms augment infiltration, restrict erosion, retard discharge to
sinkholes, and incorporate off-site considerations.
Ongoing observations of stormwater behavior indicate
problems of flooding and sediment control at the western site but
minimal disruption of existing drainage patterns at the eastern
sites. Difficulties in ascertaining all potential sources of
groundwater contamination underscore the need for stormwater
regulations which regard sinkholes as integral components of
741
-------
entire karst drainage systems.
Introduction
Springfield, the county seat of Greene County, is located in
the Springfield Karst Plain in southwest Missouri (Figure 1).
Dominant bedrock underlying this plain consists of horizontal
strata of the Mississippian Burlington-Keokuk Limestone, a light
gray, coarsely crystalline, fossiliferous unit.
Because the Burlington-Keokuk Limestone is susceptible to
solution, karst features are extensive throughout the county.
Figure 2 illustrates areas of extensive sinkhole development.
Two major concentrations of sinkholes are found within the city
limits of Springfield. These areas are located in northwest
Springfield near the airport and in a part of east Springfield
referred to as the East Cherry Street Sinkhole Area.
These two sinkhole regions are situated on topographic
highs, near surface water drainage divides. Apparent controls on
sinkhole occurrence in these areas include facies differences
within the Burlington-Keokuk Limestone and intersections of large
scale lineaments. Stormwater drainage of unaltered sinkholes
ranges from rapid drainage to long-term ponding.
Past Stormwater Management Practices
Springfield and its Metropolitan Statistical Area comprise
the fastest growing urban region in the state of Missouri. Rapid
population growth has created additional stresses to existing
land use, including reduction of open areas and increased
Stormwater runoff. Control of Stormwater in sinkhole areas has
utilized several standard engineering approaches: concrete lined
channels draining into sinkholes; installation of drain pipes
into the sinkhole "eye"; filling of sinkholes; elaborate drains
or pumps removing Stormwater from one sinkhole and discharging
into another drainage basin or sinkhole; enlargement of sinkhole
eyes by excavation to increase drainage capacity.
Several problems may arise in using these standard
practices:
1. Runoff may exceed drainage capacity of sinkholes.
2. Sediment and debris may plug sinkholes and associated
karst conduits.
3. Sinkholes are not isolated entities, but are integral
parts of dynamic drainage systems. Stormwater directed
into one sinkhole may resurface in another part of the
drainage.
4. Direct disposal of stormwater into a karst aquifer may
degrade water quality by introducing contaminants.
Existing practices may violate 1988 federal regulations
governing urban stormwater quality.
742
-------
-------
KARST AREAS
OF
SPRINGFIELD
AND
EENE COUNTY. MO
CIT?
OF
SPRINGFI
Figure 2. Major karst areas of Springfield and Greene County, Missouri. Two areas of
extensive sinkhole development within the city are located at the Springfield Airport
and near East Cherry Street. Scale is approximately 1"=3.5 mi.
-------
These problems have been recognized in previous studies and
master plans for stormwater control in the Springfield area.
In the 1950's a comprehensive stormwater master plan addressed
existing practices and recommended discontinuing the use of
sinkholes as stormwater drains. Because of the total costs
necessary to accomplish its goals, the master plan was never
adopted (Turner, 1989).
In the 1970's William Hayes, Environmental Geologist for the
City of Springfield, prepared a report sponsored jointly by the
city and the Federal Department of Housing and Urban Development
(Hayes, 1977). This report discussed the origin of local
sinkholes, their drainage areas and hydrologic characteristics,
and geologic factors controlling their alignment. Although the
study addressed the adverse effects of urban development on
sinkholes and karst groundwater systems, it emphasized many
standard structural controls for containment and disposal of
stormwater.
Having recognized the inadequacy of existing designs to
control flooding and the need to accommodate increased runoff
from future development, the City of Springfield adopted an
ordinance (effective June 19, 1989) to protect sinkholes and
prevent flooding. This ordinance, based on similar regulations
enacted by Lexington, Kentucky, requires a permit for any
proposed construction within a sinkhole drainage area.
Application for the permit must include a hydrogeologic report
evaluating the sensitivity of soil to erosion; relationship of
the sinkhole to the overall drainage area; existing sinkhole
drainage characteristics; capability of the sinkhole to accept
stormwater; type of work and equipment to be used, impacts from
proposed activities, and requirements to protect the sinkhole
during construction.
Although the ordinance attempts a balance between urban
growth and protection of natural drainage of sinkholes, several
serious ambiguities exist. Description of enforcement procedures
is indistinct. The ordinance does not address the piecemeal
nature of development, in which earlier construction on adjacent
sites may affect drainage at the site being considered. Lack of
recognition of sinkholes as integral parts of dynamic hydrologic
systems may result in problems with on-site/off-site drainage.
Case Histories of Recent Urban Developments
Three sites were analyzed to examine the effectiveness of
contrasting design approaches to stormwater management. These
sites differ in vegetation, on-site/off-site considerations, and
types of development proposed. All three sites are located
within the East Cherry Street Sinkhole Area previously described.
Figure 3 illustrates the locations of these sites relative to
important hydrologic and cultural features. Dye tracing (Hayes
and Vineyard, 1969; Hayes, 1977; Aley and Thomson, 1981; and
Thomson, 1986) has indicated that the East Cherry Sinkhole
745
-------
CHESTN JT
KH&L&FIL
C.HWA Y
ONEBREAK
ING
CATALPA STREET
.SINKHOLE DRAINAGE AREA
fiQR JONES AND BONEBREAK S$jj?IN
AF\ER HA YES AND -*^~
SX_^X
Figure 3. Map of East Cherry Street Sinkhole Area (dashed line), showing locations of study
sites relative to hydrologic and cultural features. S = Sundance Estates; CN —
Chestnut Plaza Phase I; and CS = Chestnut Plaza Phase II. Scale is approximately
1"= 3 mi. y
-------
drainage provides recharge to Jones and Bonebreak Springs. These
springs in turn discharge to Pierson Creek, a major tributary
which joins the James River approximately 3 miles upstream from
the intake of the Blackman Treatment Plant municipal water
supply. Sinkholes on two of the sites, Chestnut Plaza North and
South, drain by way of the Steury Cave system to Jones and
Bonebreak Springs.
Urban development in the East Cherry Street Area began in
the early 1900's. Residential development, usually on large
parcels, continued into the 1950's and 1960's, when industrial
and commercial uses were targeted for this area. By the late
1980's few open areas remained. These undeveloped parcels are
attractive because of their proximity to downtown Springfield and
major regional transportation routes. During the past several
decades, this area has experienced sinkhole flooding, exacerbated
by increased runoff accompanying development.
Sundance Estates Subdivision
Sundance Estates Subdivision is located on East Grand
Street, southwest of the other two study areas (Figure 3). This
ten-acre was a densely wooded tract with one farm house and
several outbuildings. Existing structures were all located near
Grand Street.
Sinkholes on this tract are not apparent on either the
U.S.G.S. (10-foot contour interval) or the City of Springfield
(5-foot contour interval) topographic maps. Preliminary plans
for the subdivision were approved in May 1989 based on the
possible existence of one sinkhole near Grand Street. Evidence
of other sinkholes on the site was not encountered until design
topographic surveys were conducted.
As a result of this new evidence, the city specified several
additional design considerations for the site. Buildings were
prohibited within the floodplains of individual sinkholes.
Construction of basements was prohibited, additional
reinforcement was required for the foundations, and wider
easements between houses were required for possible future
enhancement of flood control. The city, the developer, and the
design consultant agreed to several other precautionary measures:
installation of silt fences around sinkholes; planting of
vegetation to control erosion; prohibited use of heavy equipment
in the sinkholes; clearing of vegetation by backhoes situated
outside the sinkhole rim and extending only the bucket into the
sinkholes. Figure 4 illustrates the site layout for the
subdivision and the location of the sinkholes.
Prior to development, observations made during moderate to
heavy storm events indicated that the sinkholes were free
draining. During a 4 to 5 inch rainfall in approximately 10 to 12
hours (May 1989), ponding was not observed in the large sinkhole
(S-l) and only very moist soil conditions were evident in several
747
-------
00
ILLSFORDSJME*
^ r i L . ?
GRAND STREET C ^
DRAINAGE
DIVIDE
Figure 4. Sundance Estates subdivision, showing locations of sinkholes S-l through S-6.
Scale is approximately 1"=100'.
-------
other sinkholes (S-4 and S-3). Stormwater design used the
Rational Method to determine peak flows and detention volumes
necessary for each sinkhole. Total drainage to each sinkhole is
small, and for the proposed land-use, calculated volumes
accommodate this additional runoff. Table 1 lists a summary of
sinkhole drainage characteristics. A retention basin was designed
at the north end of the property in order to prevent potential
off-site flooding. Recommended stormwater design measures were
preventive and not remedial in nature.
Problems at the site began after clearing the sinkholes of
debris. Cleared sinkholes were hydro-mulched but were not
watered to facilitate germination of the seed. Work was
performed during the late fall, and temperatures dropped below
freezing before germination. Sediment and seed were washed into
the centers of the sinkholes, and sinkholes remained void of
vegetation for several months. Contrary to recommendations, the
contractor used heavy equipment in sinkhole areas for grading,
thus compacting the soil structure. Lack of silt fence
maintenance allowed sediment to wash into the sinkholes. Heavy
equipment used in the large retention area compacted soils and
reduced infiltration capacity. One house, constructed between
sinkholes S-2 and S-3, encroaches on sinkhole floodplains, and a
second house is located within the floodplain of sinkhole S-l.
Instead of enforcing design agreements, the city placed
responsibility on the design consultant to oversee all phases of
construction.
Sundance Estates Subdivision is almost completely developed,
and all sinkholes have adequate vegetative cover. During 1989
and 1990, flooding occurred repeatedly at the site. Moderate
rainfall (1 to 2-inch rain in 6 hours) caused ponding of water in
the sinkholes, with water remaining up to 3 or 4 days in all
sinkholes. Because of the flooding and possible structural
hazards, HUD has refused loans for this subdivision on lots with
sinkholes or adjacent to sinkholes. Further evaluation of the
site during 1991 has been limited because of sparse rainfall in
the Springfield area.
Chestnut Plaza Phase I (North)
Chestnut Plaza Phase I, the first site affected by the new
sinkhole ordinance, is an approximately 30 acre tract located at
the corner of East Chestnut Expressway and U.S. Highway 65
(Figure 3). Chestnut Plaza Phase I is planned for commercial
development, which is anticipated to include fast food
restaurants, a hotel, retail stores, and light industry or
manufacturing.
During the 1930's this site was used as pasture land.
Review of 1936 aerial photos reveals that sinkholes at the site
were clear of any trees and that the railroad and the highway had
not been constructed. Shortly after this time, construction of
the Burlington Northern railroad bisected the large sinkhole on
749
-------
Sinkhole Drainage Characteristics
SINKHOLE
S-1
S-2
S-3
S-4
S-5
S-6
SINKHOLE
AREA
0.12 ACRES
0.09 ACRES
0.11 ACRES
0.03 ACRES
0.01 ACRES
0.05 ACRES
SINKHOLE
DRAINAGE
VOLUME
5,800 CU. FT.
7,240 CU. FT.
2,600 CU. FT.
6,964 CU. FT.
SINKHOLE
DRAINAGE
AREA
1.60 ACRES
1.87 ACRES
1.53 ACRES
1.57 ACRES
CALCULATED
RUNOFF
VOLUME
2,400 CU. FT.
2,800 CU. FT.
2,300 CU. FT.
2,400 CU. FT.
Table 1
-------
Sinkhole Drainage Characteristics
SINKHOLf
CN-1
CN-2
CN-3
CN-4
CS-1A
CS-1B
CS-2
SINKHOLE
AREA
3.6 ACRES
0.81 ACRES
21 6 FT2
0.15 ACRES
5 ACRES
10 ACRES
0.68 ACRES
SINKHOLE
DRAINAGE
VOLUME
535,000 CU. FT.
60,000 CU. FT.
N/A
N/A
800,000 CU. FT.
1, 600,000 CU. FT.
50,000 CU. FT.
SINKHOLE
DRAINAGE
AREA
27.5 ACRES
3.9 ACRES
N/A
N/A
50 ACRES
50 ACRES
10 ACRES
CALCULATED
RUNOFF
VOLUME
1 00,000 CU. FT.
1 5,000 CU. FT.
N/A
N/A
741, 000 CU. FT.
1, 263,000 CU. FT.
130,000 CU. FT.
Table 2
-------
Ln
1-0
CHESTNUT PLAZA
PHASE I (NORTH)
STORMWATER
DETENTION
IN
SlpRMWATE
DfTJENTION BA
Figure 5. Chestnut Plaza Phase I (north), showing location of sinkholes CN-1 through CN-4
Scale is approximately 1"=160'
-------
the site. Later construction of East Chestnut Street covered
another part of the sinkhole.
During construction of the railroad and Chestnut Street,
excavated soil and other material were dumped into the sinkhole.
When Chestnut Expressway was widened, excavated soil, pavement,
and other materials were dumped into the sinkhole.
Alteration of the large sinkhole severely restricted its
drainage capacity. The sinkhole floods and ponds water for
several days after moderate rainfalls. Observations made during
a 4 to 5-inch rainfall in January, 1990, revealed that stormwater
runoff rose to the 1354 elevation within the sinkhole. This
level is below the established floodplain elevation of 1359, and
flood waters drained after 5 to 6 days. This large sinkhole also
receives off-site drainage from a 30 to 40 acre site to the
north.
Because sinkholes at this site had already been altered, it
was imperative that development would not further degrade
drainage of the sinkholes. Although it contained a great amount
of fill material, the larger sinkhole still had the capacity and
drainage to accommodate increased runoff if some protective
measures were taken into consideration.
Design of the stormwater detention areas utilized the
Rational Method for calculating stormwater runoff and peak flows
(Table 2). FEMA had already established floodplain limits in the
larger sinkhole. To protect the sinkhole and allow for future
development, stormwater detention berms were designed around the
two larger sinkholes (CN-1 and CN-2 in Figure 5). No building is
allowed in the sinkhole areas. To prevent soil compaction by
heavy equipment, clearing is permitted by hand only.
Additional stormwater detention will be required for
individual sites along the northwest side of sinkhole CN-1 as
development occurs along Chestnut Expressway. Minor filling has
been approved for the area adjacent to Chestnut Expressway.
Although two smaller sinkholes, CN-3 and CN-4, are not important
in the overall drainage of the site, extreme thickness of their
soils and lack of solid bedrock beneath them could lead to
potential collapse. For these reasons CN-3 and CN-4 have been
designated "non-buildable" and will remain in their natural
state.
Chestnut Plaza Phase I was approved for construction during
the summer of 1990. Prior to any construction or clearing at the
site, sediment control methods were established. These methods
included silt fences around sinkhole areas and also outside the
perimeters of their established floodplains. Completion of
detention berms around sinkholes CN-1 and CN-2 and establishment
of vegetation in these berms was required before construction
could begin.
753
-------
Recommended protective measures have been established, and
the street and one warehouse have been constructed. Additional
construction is expected within the next several months.
Observations made during a 2 to 3 inch, 12 hour storm indicate
confinement of drainage to the sinkhole and berms and
infiltration of stormwater within 3 to 4 days. Additional
monitoring of runoff and sampling of stormwater are planned for
the site.
Chestnut Plaza Phase II
Chestnut Plaza Phase II is located at the intersection of
East Cherry Street and U.S. Highway 65 and south of Chestnut
Plaza Phase I (Figure 6). This site encompasses approximately 30
acres and surface runoff drains southwest to an off-site sinkhole
(CS-1B). Sinkholes CS-1A and CS-1B are two halves of an
originally continuous sinkhole which was bisected by construction
of the Burlington Northern Railroad. Other modifications include
partial filling of sinkhole CS-1B for industrial development
during the 1970's.
Figure 6 illustrates the complexities of the drainage
system. Drainage area shown in this figure is approximately 100
acres. All runoff is directed toward an outlet constructed in
sinkhole CS-1B. When this modification was made in 1972, the
city's environmental geologist stated that the Chestnut Plaza
Phase II site was "undevelopable" (Giles, 1990, personal
communication).
Initially sinkhole SC-1B was able to accept runoff from
1970's and early 1980's construction. However, because design
modification included only a metal grate to stop debris, the
underlying cavern system has filled with sediment and is greatly
reduced in drainage capacity (Thomson, 1990, personal
communication). Plans for Chestnut Plaza Phase II had to address
this problem.
Figure 7 illustrates the design for Chestnut Plaza Phase II,
including the sinkhole area and proposed stormwater controls.
Because the developer desired the largest possible area for
building, stormwater detention basins such as those at Chestnut
Plaza Phase I were not used. Instead, a two tier stormwater
network was designed to provide maximum buildable area while
controlling stormwater runoff.
Because FEMA had not established a floodplain for the site,
floodplain information was calculated using the Soil Conservation
Service TR-55 Stormwater Method. Pre-existing major alterations
of sinkhole drainage and sensitivity of the adjacent site to
flooding suggested the use of the TR-55 method, which allows more
detailed analysis of stormwater behavior. This method utilizes
soil characteristics, land use, hydrologic characteristics, and
rainfall distribution and duration to develop runoff hydrographs.
Calculated floodplain elevation is 1353.5 feet and volume is
754
-------
LEGEND
, i
SINKHOLES
SATELLITE SINKHOLES
PENBROKESOIL SERIES
APPROX. LOCATION OF WO
YEAR FLOOD PLAIN.
ELEV=1353.5
APPROX. LIMITS OF
SINKHOLE (EAST SIDE) BASED
ON TOPOGRAPHIC SURVEY
SINKHOLE DRAINAGE BASIN
CRELD
SOIL AERIES
PROJECT SITE
KEENO-ELDON
SOIL SERIES
CRELDON
SOIL S5RIES
SINKHOLE
OUTLET TYP.
KEENO-ELDON
SOIL SERIES
Figure 6. Map of Chestnut Plaza area, showing drainage basin for large bisected sinkhole
(CS-1A and CS-1B). Scale is approximately 1"=500'.
-------
HIGH WAY 65
HESTNUT PL
HASEIV (SOU
Uj
Uj
ce
Uj
a:
fcw^:^^53atj«gs,a«y>i5ffl38«--
Figure 7. Chestnut Plaza Phase II (south), showing sinkholes CS-1A and CS-1B Scale is
approximately 1"—170'.
-------
estimated at 1,000,000 cubic feet for the 100 year flood storage
basin (Table 2).
In addition to the 100 year flood volume, drainage volume
for projected complet development was computed using the Rational
Method, and a detention basin was designed for this volume. As
in Chestnut Plaza Phase I, both basins will be grass lined and
sediment control measures will be used during and after
construction.
Outflow to the off-site drain was determined from field
measurement, and values obtained were further reduced to
accommodate runoff from future development. Drainage design for
the site west of Chestnut Plaza Phase II allowed a discharge of
93 cfs from sinkhole CS-1B during a 10 year storm (Hayes and
Thomson, 1973). Mapping of the Steury Cave system suggests that
sedimentation has reduced this discharge capacity by 90 to 95
percent (Thomson, 1990, personal communication). Field
observations of sinkhole CS-1B made during a 2 inch, 8 hour
rainfall support this estimate. Off-site discharge to CS-1B from
Chestnut Plaza II was reduced accordingly by 95 percent to a
design outflow of 5 cfs.
Preliminary design for Chestnut Plaza Phase II was approved
in the summer of 1990. Vegetation has been cleared at the site,
and silt fences have been placed at sinkhole CN-1B. Construction
is expected to begin in 1992. Runoff and water quality will be
monitored before and during construction.
Conclusions
Case histories of three sites in Springfield, Missouri
illustrate results of contrasting approaches to stormwater design
and regulation in urban karst terrain. Residential development
planned for a 10-acre wooded tract (Sundance Estates) required
minimal preventive measures to protect free-draining sinkholes
from construction activities and to maintain the drainage
capacities of those sinkholes. Lack of adherence to design
recommendations has resulted in repeated flooding of sinkholes at
this site.
Two adjoining commercial parcels are located in an area
which experienced extensive alteration of drainage during
development of neighboring sites. Because of existing flood
hazards and inflow of stormwater from an adjacent site, design
for the northern parcel (Chestnut Plaza Phase I) included land
use restrictions, sediment control, and construction of detention
basins which approximate natural landforms. Close communication
among the city, the developer, and the design consultant ensured
completion of stormwater measures as planned. Observations made
during recent storms suggest that the design maintains adequate
drainage. Future studies will examine the impact of the
detention basins on stormwater quality-
757
-------
Because of off-site releases of stormwater and a desire to
maximize commercial land use, design for the southern parcel
(Chestnut Plaza Phase II) includes a two-tier stormwater network
as well as vegetative and sediment control measures. Construction
at this site is incomplete, and continued monitoring of runoff
and stormwater quality are planned.
References
Aley, T. and K. C. Thomson, 1981. Hydrogeologic Mapping of
Unincorporated Greene County, Missouri, To Identify Areas
Where Sinkhole Flooding and Serious Groundwater
Contamination Could Result From Land Development. Report
repared for Greene County Sewer District/U.S.Environmental
Protection Agency. Ozark Underground Laboratory, Protem,
Missouri, lip.
Hayes, W. C. and J. D. Vineyard, 1969. Environmental Geology
of Towne and Country: Missouri Geologic Survey and Water
Resources, Rolla, Missouri, Educational Series Report No. 2.
42 p.
Hayes, W. C. and K. C. Thomson, 1973. Engineering and
Environmental Geology of The Springfield Urban Area.
Association of Missouri Geologists Twentieth Annual Meeting,
Branson, Missouri, Sept. 21-22, 1973. Geography and Geology
Department, Southwest Missouri State University,
Springfield, Missouri, 27 p.
Hayes, W. C., 1977. Urban Development in a Karst Terrain -
Springfield, Missouri. Report prepared for the Department
of Housing and Urban Development, 64 p.
Thomson, K. C., 1986. Geology of Greene County, Missouri.
Report prepared for Watershed Management Coordinating
Committee, Springfield, Missouri. Southwest Missouri State
University, Springfield, Missouri, 86 p.
Turner, C. R., 1989. Stormwater Management in Springfield, Now
and in the Future. Proceedings of the Third Annual
Watershed Conference, Springfield, Missouri, June 14-15, pp.
51-54.
758
-------
TITLE: CASE HISTORIES OF SEVERAL APPROACHES TO STORMWATER
MANAGEMENT IN AN URBANIZED KARST TERRAIN, SOUTHWEST MISSOURI
AUTHORS: Wendell L. Earner - ICF Kaiser Engineers
Patricia Miller - California State University,
Sacramento
Question: Is dye-tracing a reguirement (or is it done) prior to
zoning or development?
Response: The sinkhole ordinance adopted in 1989 does not
specifically require or mention dye tracing. However, the
ordinance does specify substantial "state of the art" field
studies and evaluation of the sinkhole system. Therefore, dye
tracing may be performed by the developer or the consultant to
determine sinkhole drainage characteristics and impacts on the
groundwater system from the proposed development. Other
investigators have performed dye tracing in the Springfield area,
and their results are on file with the Missouri Geological
Survey, as well as with local governing agencies. Many traces in
the Springfield area were executed prior to, during, or after
modifications to sinkholes for stormwater drainage.
759
-------
760
-------
CONTAMINANT INVESTIGATION IN A KARST REGION
J. E. BENTKOWSKI, P.O.
METCALF & EDDY, INC.
ATLANTA, GEORGIA
A series of springs in the karst region of north central Kentucky appeared to have been contaminated.
These springs are within a half mile of two sinkholes, which were filled in as permitted landfills for inert
waste and then developed into a medium industrial park. A preremedial site inspection was performed under
the authority of the Superfund laws in late 1989. A preliminary site visit included site reconnaissance and
geologic field work to locate the springs. A review of historical aerial photos aided in the planning of the
investigation program, which consisted of magnetic and soil gas surveys and environmental soil and water
sampling. The magnetic survey indicated the presence of buried ferrous objects. The soil gas survey points
were laid out incorporating this information. Soil sample locations were selected based on the results of the
soil gas survey. Seventeen surface and subsurface soil samples were taken. Eleven water samples were
taken from various springs, rivers, and the local public water supply. The analytical results of the soil
samples taken over the sinkholes matched 20 compounds also found in the water and sediment samples
taken from the spring. The location of the springs roughly coincided with the strike of the major fracture
systems reported in the literature. The success of this investigation emphasizes the importance of proper
geologic consideration for contaminant monitoring in karst regions.
Introduction
This investigation was part of the phased investigative approach that is employed by the USEPA to
determine whether sites warrant being placed on the National Priorities List. The purpose of this particular
investigation was to determine the nature of the contaminants present at the site and to determine if a
release of these substances has occurred or is likely to occur (USEPA, 1991). Additionally, the potential
pathways for contaminant migration were investigated. This paper describes the methodology that was
employed to meet the stated purpose.
It is important to note several limitations, both organizational and geologic, that were key in formulating the
approach to this investigation. This Screening Site Inspection (SSI) was done under the authority of the
USEPA, Region IV, as part of a preremedial Superfund investigation. These SSIs are limited in scope and
concentrate on source characterization and identification of potential pathways for contaminant migration.
Permanent groundwater monitoring wells are not part of normal SSIs. Temporary groundwater monitoring
wells were not considered for two reasons. First, the site is located adjacent to a bluff that overlooks Valley
Creek and Valley Creek reservoir (the closest surface water bodies). The top of the bluff is approximately
761
-------
65 feet above the water level in Valley Creek reservoir. This physically precluded the installation of the
temporary wells by the investigation personnel. Second, in mature karst areas, such as around
Elizabethtown, Kentucky, the majority of the groundwater, from 60% to 80%, would be expected to flow
in the dissolution features in the limestone (conduit flow), not as diffuse flow through the limestone itself
(Quinlin and Ewers, 1985). These dissolution features cannot be readily encountered when drilling. A much
more reliable approach towards monitoring potentially contaminated groundwater is to inventory and
monitor springs in the areas as they are the discharge points for the groundwater (Quinlin and Ewers,
1985). This idea is the basis for the technical approach taken in this investigation.
Background
From 1969 to 1977, a landfill was permitted to receive inert waste to be disposed of in a series of sinkholes
located approximately 1/2 mile southeast of Elizabethtown. The site is approximately ten acres in size (1800
feet by 1000 feet) and contains at least three sinkholes. The site is bounded approximately on the west by
Interstate 65 and on the north by the valley of Valley Creek (Figure 1). This creek flows to the west
through Elizabethtown.
N
Figure 1. Study Area
Elizabethtown, Kentucky
The permitted wastes for this landfill included Neoprene, cardboard, wood, paper, foundry sand, bricks,
wood palettes, and empty barrels. An inspection in 1972 by the Kentucky Department of Environmental
Protection indicated that the landfill was receiving liquid wastes and domestic garbage in violation of its
permit. Regarding the liquid wastes, a galvanizing company reportedly had a pipeline for the disposal of
spent sulfuric acid into one of the sinkholes. USEPA files raise the question of the disposal method of 1200
drums of dross zinc ash by the galvanizing company. These files also raise the possibility that a tanker
truck was buried in the largest of the sinkholes (USEPA, 1991).
Contamination of a spring on a farm immediately across 1-65 from the landfill was noted in the early 1970s.
Investigations by both State and Federal agencies indicated the presence of elevated values for zinc, nickel,
chlorinated compounds, and other organic compounds as well as low pH values for the spring water. The
landfill was officially closed in 1977. This investigation took place in 1989 (USEPA, 1991).
762
-------
Geology
The Elizabethtown area is in the Interior Low Plateaus physiographic province (Fenneman, 1938). The
Pennyroyal Plain is a gently rolling mature plain that slopes towards the southwest and is characterized by
such karst features as sinkholes, springs, and sinking streams (Sauer, 1928). Valley Creek, the only
perennial river in this area, flows along the northern boundary of the study area.
The discussion of the geology of this area will be limited to describing the surficial soils and the carbonates
of Mississippian age which control the development of the karst features. The alluvium in this area is
confined to Valley Creek and its feeder streams. The bedrock of the area is covered by a blanket of
residuum of clay and chert fragments. This layer is of variable thickness; it fills in the dissolution and
collapse features of the underlying limestone (Mull et al, 1988).
The primary limestones of the area are the Ste. Genevieve and the underlying St. Louis. These formations
are separated by a relatively thin bed of limestone and silicified fossils known as the Lost River Chert. The
Ste. Genevieve is typified by occasionally shaly oolitic limestone and dolomite beds varying from 0.5 to 4.0
feet in thickness (Mull et al, 1988). The St. Louis is the most significant formation with regards to the
development of the dissolution features. The upper portion of the formation consists of primarily massive to
thin-bedded limestone and dolomite. In the lower portion of the formation, there is an increase in the
amount of gypsum and anhydrite. Noger and Kepferle (1985) believe that the presence of these soluble
minerals is what lead to the extensive development of the mature karst features seen in the study area.
Locally, the confirmation of the rock units was not possible due to the limited outcrop in the area and the
lack of subsurface or bedrock drilling. More important is the hydrogeology of the site as typified by the
filled-in sinkholes, the adjacent springs and the elevation difference between the land around the sinkhole
and the water table, as represented by the springs and Valley Creek. As seen in Figure 1, the landfill was
located in the largest of the three sinkholes east of Interstate-65 and south of Valley Creek. This plateau is
bordered on the north by the alluvial valley of Valley Creek and rises approximately 65 feet above the
valley floor. Across the Interstate to the west are five springs that originate from the base of a hill and flow
over the alluvium to Valley Creek.
The groundwater flow patterns of this area have been the subject of considerable study. A recent work by
the USGS (Mull et al, 1988) describes dye tracing studies that were performed in the Elizabethtown area.
Their studies include the documentation of groundwater flow direction towards Valley Creek and the
suggestion that shallow groundwater flow follows a major fracture joint set that trends N40W. The springs
in this study are approximately along this trend from the landfill/sinkhole.
Technical Approach
The basis of the technical approach for this investigation is a combination of modified procedures for
monitoring groundwater quality in karst areas presented by Quinlin and Ewers (1985) and the standard
investigation methods used for preremedial investigations conducted under authority of the USEPA. The
essence of Quinlin's and Ewer's recommendations is that the majority of the groundwater flow in karst
areas is in the dissolution features. The probability of intersecting these dissolution features with a
monitoring well is very low. Therefore, the only effective method for groundwater monitoring is to locate
all the springs and seeps along riverbanks and hillsides, plus other accessible points of intersection with the
water table, and monitor those locations. Dye tracing studies should be an integral part of every sinkhole
investigation, but they were not possible in this case, becuse the sinkhole had been filled in. There was no
surface expression of the dissolution or collapse features.
763
-------
This investigation consisted of the following parts:
1. Site reconnaissance
2. Project planning
3. Magnetic survey
4. Soil gas survey
5. Sampling of environmental media to include:
a. surface soil
b. subsurface soil
c. spring water
d. spring sediment
e. municipal supply
f. appropriate background for each media
6. Data reduction and report preparation
Site Reconnaissance
After reviewing the file, a trip was arranged to Elizabethtown to view the site, perform field geology
studies, and talk to local officials. The field work was done in October, 1988, following a dry summer. The
topographic map indicated the location of the sinkholes prior to being filled. The area is now a small
industrial park with the main sinkhole area being a storage area for large cement drainage pipes. Across the
Interstate, four springs were located at the base of the hill. A spring house was noted at one of the springs.
Also noted was a dry spring. Two large sinkhole depressions were noted at the top of the hill, above the
springs. Contact was made with the Elizabethtown Water Department to arrange permission to take water
samples from the municipal springs and wells later in the year. The local office of the Department of
Transportation was contacted regarding the availability of historical aerial photos that might aid in the
planning of the investigation. While this office had some coverage, the investigators were referred to the
headquarters office in Lexington for the most complete historical coverage of the area.
Project Planning
With the information provided by the file material and the field reconnaissance, the project plans were
formulated. Because the sinkholes were used for illegal dumping and the contents and depth of burial were
not documented, a magnetic survey would provide an nonintrusive method of investigation.
Groundpenetrating radar would have been the first choice, but it was ruled out due to the high clay content
of the soils. Additionally, the file materials indicated that a tanker truck may have been buried in the largest
sinkhole. A magnetic survey would give some indication of its possible location, as well as provide some
measure of safety for the subsurface soil samples to be taken later. The file material indicated that
chlorinated compounds had been detected in the adjacent springs. It was assumed that these compounds
were leaching from the sinkhole contents and that a soil gas survey would detect the presence of these
compounds. This would ensure that soil samples were taken in the most appropriate location and allow the
efficient investigation of such a large area. The presence of chlorinated compounds, specifically vinyl
chloride, demanded that the sampling of the spring water and sediment be planned as a Level B exercise.
Appropriate access was requested and the field work was performed in early December, 1988.
Magnetic Survey
The magnetic survey was conducted with a proton precision magnetometer. Survey grids were laid out
above two sinkholes with 28 stations for the eastern sinkhole and 20 stations for the northern sinkhole.
764
-------
(Access permission was not granted in time to perform a survey over the southern sinkhole). All
appropriate calibrations were performed and standard methodologies were followed. The results of the
eastern sinkhole indicated two positive anomalies in the center of the sinkhole and four negative anomalies
around the edges of the grid. These results may be interpreted as indicating the presence of buried man-
made, ferrous objects in the center of the sinkhole. The two positive anomalies align in such a way that
they may be interpreted as indicating the presence of the suspected tanker truck. The results of the northern
grid only indicated one positive anomaly along the southwest corner of the grid. This was interpreted as
indicating buried man-made, ferrous objects only in the southwest area, with the rest being nonferrous fill
material (USEPA, 1991).
Soil Gas Survey
The soil gas survey was conducted using a photoionization detector (HNu), which is designed for the direct
reading of the presence of volatile organic compounds. Soil gas probes were installed to an approximate
depth of three feet in eleven locations in the eastern sinkhole and five locations in the northern sinkhole. On
the basis of this survey, locations of the soil samples were selected. A third survey was attempted in the
area of the southern sinkhole. Access problems delayed the performance of work in this area. The weather
became extremely unfavorable with temperatures in the 20s, blowing snow, and winds in excess of 15 miles
per hour. An attempt was made to perform a soil gas survey, but the author believes that more favorable
conditions would have allowed a more successful attempt.
Taking of Environmental Samples
A total of 17 soil and 11 water samples were taken during this investigation. The breakdown of the sample
types follows:
* A total of nine surface and subsurface soil samples were taken from the three sinkholes.
* A total of four water and two sediment samples were taken from the five springs located at the base of
the hill across the interstate. Lack of water or lack of sediment, combined with temperatures in the mid-
teens, precluded the completion of the entire planned sampling program in this area. The cold temperatures
did allow the downgrading to modified Level D due to the lack of volatilization as indicated by the air
monitoring equipment.
* Six water samples were taken from the Elizabethtown municipal water system.
* A water sample and a sediment sample were taken as background from Valley Creek Reservoir located
northeast of the site. A total of two surface and two subsurface soil samples were taken as background
samples: one set onsite near the sinkholes and one set on the farm where the springs were located.
All samples were taken observing USEPA Region IV sampling protocol.
Analytical Results
Over 150 compounds, including volatiles, semivolatiles, pesticides, PCBs, acids/base/neutrals, and
inorganics were analyzed as part of the USEPA-approved Contract Lab Program. There were numerous
compounds present above detection limits. Some were present only in the soil samples taken above the
sinkholes. Some were present only in the samples taken at the springs. The set of compounds that were
detected in both locations numbered 20: 11 organic compounds (Table 1) and nine inorganic compounds
(Table 2). The analytical results for this set of compounds are listed below with their respective sample type
and maximum concentration detected.
765
-------
TABLE 1
COMPARISON OF ORGANIC ANALYTICAL RESULTS
SOIL AND WATER SAMPLES
ELIZABETHTOWN, KENTUCKY
PARAMETERS
(ug/kg)
ACETONE
CARBON BISULFIDE
1 , 1-DICHLOROETH ANE
1 ,2-DICHLOROETHENE
1,1,1-
TRICHLOROETHANE
TRICHLOROETHENE
BENZENE
METHYL ISOBUTYL
KETONE
TETRACHLOROETHENE
TOTAL XYLENES
TOLUENE
BACKGROUND
SOIL
12UJ
6U
6U
6U
6U
6U
6U
12UJ
6U
6U
6U
BACKGROUND
WATER
10U
5U
5U
5U
5U
5U
5U
10U
5U
5U
5U
SINKHOLE
SOIL
(MAXIMUM)
320J
2J
8
45J
6
2J
42
220J
20
66N
190
SPRING
SEDIMENT
(MAXIMUM)
630J
13J
11J
SPRING
WATER
(MAXIMUM)
110
520
67
110
4J
3J
26
8J
4J
TABLE 2
COMPARISON OF INORGANIC ANALYTICAL RESULTS
SOIL AND WATER SAMPLES
ELIZABETHTOWN, KENTUCKY
PARAMETERS
(ug/kg)
CADMIUM
CHROMIUM
COPPER
IRON
LEAD
MANGANESE
NICKEL
VANADIUM
ZINC
BACKGROUND
SOIL
1.1U
21J
6.7
17,000
15J
410
13J
SOU
30U
BACKGROUND
WATER
10U
22U
160U
60UJ
65
16U
16U
SINKHOLE
SOIL
(MAXIMUM)
2.6
34J
190
63,000
1100J
410
130J
92
860
SPRING
SEDIMENT
(MAXIMUM)
19,000
33J
11,000
24,000
9700
340
24J
12,000
SPRING
WATER
(MAXIMUM)
18
26
76,000
170J
4500
26
39
39
NOTES
Material analyzed for but not detected above minimum quantitation limit (MQL)
i Estimated value
U Material analyzed for but not detected. The number given is the MQL
N Presumptive evidence for the presence of material
766
-------
Conclusions
The objectives of this investigation included the source characterization of suspected waste areas, the
identification of releases or suspected releases to the environment of hazardous materials, and the
establishment of a hydrogeologic interconnection between the sinkholes/landfill and the adjacent springs.
The varied analyses did characterize the sources as to the type of material that was placed in the
sinkhole/landfill. The spring water and sediment analyses did indicate that hazardous materials were being
released into the environment. The matching of 20 compounds between the sinkhole/landfill and the springs
does indicate a connection between the two areas. Quinlin's and Ewer's basic premise that the key to
monitoring groundwater quality in karst areas is to monitor the springs has been demonstrated.
References
Fenneman, N.M., 1938, Physiography of the Eastern United States: New York, McGraw-Hill Book
Publishing Company, Inc., p. 714.
Mull, D.S., J.L. Smoot, and T.D.Leiberman, 1988, Dye Tracing Techniques Used to Determine
Groundwater Flow in a Carbonate Aquifer System Near Elizabethtown, Kentucky, Water Resources
Investigations Report 87-4174 (Louisville, Kentucky: U.S. Geologic Survey), p. 95.
Noger, Martin C. and Roy C. Kepferle, 1985, Stratigraphy along and adjacent to the Bluegrass Parkway:
Geological Society of Kentucky Guidebook, October 18-19, 1985, Kentucky Geological Survey, p. 24.
Quintan, James F. and Ralph O. Ewers, 1985, Ground Water Flow in Limestone Terranes: Strategy,
Rational, and Procedure for Reliable, Efficient Monitoring of Ground Water Quality in Karst Areas. In:
National Symposium and Exposition on Aquifer Restoration and Ground Water Monitoring (5th, Columbus,
Ohio, 1985), Proceedings, National Water Well Association, Worthington, Ohio, p. 197-234.
Sauer, C.O., 1927, Geography of the Pennyroyal: Kentucky Geological Survey, Series IV, v. 25, p. 303.
USEPA Region IV, 1991. Screening Site Inspection, Phase II, Kentucky Industrial Haulers, Elizabethtown,
Hardin County, Kentucky, prepared by NUS Corporation, pp. 44.
767
-------
Contaminant Investigation in a Karst Region
J. E. Bentkowski
Q. How do you know each or any of the springs or surface water sites is hydraulically connected to the
landfills?
A. There have been no die traces performed in this exact area near Elizabeth town. Given the hazardous
nature of the spring water, the flushing of the karst system would be a questionable procedure. The
connection between springs and the landfills is demonstrated by the empirical evidence of the 20
compounds which were detected in both the springs and the landfill soils.
768
-------
LAND-USE PLANNING AND WATERSHED PROTECTION IN KARST TERRANES
Paul A. Rubin
Oak Ridge National Laboratory, Oak Ridge, Tennessee
Abstract
It is critical that the special characteristics of a karst watershed
(groundwater basin) be examined when any development within it is to be
considered. Local planning boards are most likely not aware of the radically
different flow dynamics of karsts, and therefore they may unwittingly allow
development that leads to mas£ive contamination. A practical approach for
ascertaining the sensitivity of a basin to groundwater contamination,
utilizing relatively easily implemented and practical procedures, is
described; it will aid land-use planners. Although the approach is kept
fairly basic, it is comprehensive. Not only is the hydrology of the karst
aquifer characterized, but any present and future sources of contamination are
evaluated and compared with the estimated assimilative capacity of the
aquifer.
An evaluation of the potential impacts of development should also examine
the types and amounts of contaminants which could be introduced into a system,
the likely chemical partitioning of these contaminants, and the combined
loading to the aquifer or stream receptor. Consideration of the present
health of receptor streams should be part of the process. Water chemistry
parameters, combined with indices of biotic integrity and ecosystem health,
can be used to evaluate the present condition of the ecosystem. This
important base-line information should then be incorporated into a
comprehensive and practical management strategy.
The Mill Pond karst basin, situated in east-central New York State, is an
example of an environmentally sensitive karst aquifer. Recently zoned as
Rural Commercial, this groundwater basin will be used to illustrate the
practical application of accepted hydrologic techniques in land-use planning.
Reference to several other localities will provide further insight into the
evaluation process.
Organization
This paper is organized as follows: 1) Introduction, 2) The land-use
planners point of view: why does a karst region deserve special consideration,
3) Assimilative capacity of the karst aquifer: how much is too much, 4) A
typical development proposal, 5) Characterization necessary for zoning
decisions, 6) Practical land-use planning: two examples, 7) Advanced testing
and analysis to define a karst system requiring a karst hydrologist, 8)
Political realities of the land-use permitting process, and 9) Who to look to
for help.
Introduction
Groundwater in soil and most fractured bedrock aquifers moves slowly,
enabling contaminants to be partially treated and diluted. Karst aquifers
(comprised of dissolutionally enlarged fractures and cave passages), on the
other hand, are often characterized by appreciable and sometimes rapid
769
-------
groundwater flow (Quinlan, et al., 1991). They have virtually no ability to
treat water-borne contaminants, instead they merely transmit contaminants,
much as a sewer pipe would (Ford and Williams, 1989). In some karst aquifers,
contaminated groundwater can move from one end of the aquifer to the other in
less than one day. Failure to recognize the nature of karst systems, and to
plan accordingly, is likely to lead to the degradation of the aquifer, its
springs, and its dependent aquatic ecosystem. Chemical loading, and possible
resulting deoxygenation of karst groundwaters during periods of base flow, may
result in destruction of fauna, including game fish, downstream of aquifer
discharge points (i.e., springs). Contaminant loading of a karst system and
its receiving stream may also result in cave habitat destruction, impairment
of the character or quality of important aesthetic and recreational water
resources enjoyed by residents and the community (e.g., streams), and the
possible infringement of riparian rights of domestic and agricultural users.
It is a difficult, expensive, and sometimes an impossible proposition to
remediate a karst aquifer after it and its receiving stream have been
degraded. The development of an area must be within the natural constraints
of its geology and hydrology.
Urban expansion is causing greater pressure for land development in many
karst regions. It is critical that the special concerns of karst watersheds
be addressed by local planning boards, environmental conservation departments,
and other state agencies. For example, towns in the Helderberg Plateau area
of New York continue to receive applications for single residences and planned
unit developments to be atop maturely karstified limestone aquifers. Optimal
land-use planning would dictate the evaluation of an entire karst system
before permitting additional development in any isolated segment of an
aquifer. This would include characterization of the extent of the aquifer
watershed (its basin boundaries), geology, subsurface flow paths, depth to
groundwater, discharge area, and base flow. Only after such characterization
is complete, can the net effects of multiple contaminant inputs on the
assimilative capacity of an aquifer be judiciously evaluated. It is important
that evaluations of this nature not only define the dynamics of a karst
system, but that they then utilize the knowledge gained to protect subsurface
and surface water resources. Unfortunately, many of the conclusions about
what is necessary for protection of the aquifer will, like zoning, be regarded
by some to be an illegal taking of their rights.
Successful land-use planning in karst terranes involves two important
criteria. First, there needs to be detailed analysis of the area's geology,
hydrology, and biology. Of equal or greater importance, however, is the
willingness of land-use planners and their counterparts to become
knowledgeable and involve themselves with the actual scientific issues and to
resist strong pressures to permit land-use beyond the assimilative capacity of
the karst aquifer. Without these elements, the analytical process may amount
to little more than good intentions. A positive outcome of this process can
be obtained when karst hydrologists work with land-use planners.
The Land-Use Planners Point Of View:
Why Does A Karst Region Deserve Special Consideration?
The cleanup or remediation of contaminated water supplies or streams can
cost a municipality upwards of a million dollars. Prudent land-use planning
can avoid this; not only eliminating the unnecessary and avoidable expenditure
of taxpayers' dollars, but also saving a great deal of environmental
destruction of a precious resource. Excessive development in environmentally
sensitive karst groundwater basins may lead to contamination, flooding, and
subsidence problems. This paper discusses water quality concerns. Too much
of any contaminant introduced into a karst basin is likely to result in
degradation of groundwater resources and the water body (i.e., stream, river,
or lake) receiving the basin's discharge. Contaminant inputs to karst
groundwater basins from sources such as leach fields, landfills, sinkhole
dumps, animal feed lots, and industrial disposal operations may combine to
770
-------
exceed the assimilative capacity of the aquifer and receiving springs and
streams. Since anything that goes into a karst system comes out virtually
untreated, dissolution conduits (e.g., cave passages) can function as large
flowing sewer pipes. Worst-case examples of this problem are numerous and
include visual and olfactory evidence of septic waste and industrial
contaminants (i.e., gasoline, toilet paper), destroying ecosystems so that
they support little to no life. This now-polluted water is often critical for
water supply, land value, and recreational purposes (e.g., fishing, swimming,
aesthetic quality). A further consideration may be adverse human and animal
health impacts that may result from ingestion or contact.
Once particular land-uses are permitted and in place (e.g., large housing
clusters, factories, dry cleaners, service stations, waste treatment centers,
feed lots, liquid waste spreading, etc.) it is difficult, if not impossible to
terminate or relocate them and retro-fitting facilities for reduced effluent
generation can be costly. Yet, as history has repeatedly shown, failure to
properly consider the ramifications of excessive contaminant loading can
result in multi-million dollar evaluation and remediation costs. In such
settings, where karst springs and related surface streams are used for
residential, agricultural, and municipal water supplies, the first action
required is the replacement of the primary water supply. This involves a
lengthy, and potentially costly, process of alternative water supply source
and site selection, full geologic, hydrologic and chemical characterization,
design and construction of necessary treatment and delivery systems. In some
cases, the contaminated supply can be treated or piping extended to
alternative existing sources. This situation is happening with increasing
frequency and often costs millions of dollars. The replacement of a water
supply can sometimes require years of planning, exploring funding options,
evaluation, and set-up. One potential solution, perhaps of equal cost, is the
funding, planning, and installation of a sewer district. Another reality
which may bear on the land-use planning issue is the legal responsibility of
the planners. This consideration may weigh heavily in the future as
litigation and pollution-related costs rise. Even now, corporate executives
are being held personally responsible for their actions; with some spending
time in jail. The best time for planning for responsible land-use is prior to
significant developmental pressures.
Assimilative Capacity Of The Karst Aquifer: How Much Is Too Much?
This is one of the more important questions in land-use planning and
watershed protection decisions. The answer to this question should logically
dictate what are reasonable and prudent land-uses in a karst basin.
Ultimately, the relative assimilative (carrying) capacity of individual karst
basins must be determined on a case by case basis. In many aquifers,
especially those capable of limited contaminant dilution, the assimilative
capacity is nearly zero. Many factors must be considered when addressing what
types and amounts of contaminants should be permissible in a groundwater basin
within a karst terrane. Some of these include: 1) presence or absence of
sewers, 2) the type and amount of contaminants (present and planned) incident
to the karst system, 3) the amount of waste pre-treatment, if any, 4) the
amount of natural cleansing, if any, imparted by overlying soil or bedrock
units prior to conduit entry, 5) an assessment of artificial diversion of
water into or out of the basin, 6) the maturity and sensitivity of the karst
aquifer, including the extent and importance of the upper part of the
percolation zone (epikarst), 7) the physical characteristics of conduit flow
paths during base flow conditions (e.g., sluggish movement through
intermittent pools), 8) residence and travel time of contaminants in
subsurface and surface flow paths, 9) water temperature, 10) the base flow and
draught flow conditions in the aquifer, 11) the relative sensitivity and
discharge of the receiving waterway (does the spring comprise the total flow
of a stream in the headwaters of a watershed, or does it enter a stream or
water body capable of providing significant dilution?), 12) the rate of
deoxygenation and reaeration of the receiving water body, 13) the contaminant
771
-------
loading and biologic health present in the receiving water body, and 14) the
current and planned uses of aquifer and receiving stream water.
The dynamics of the contaminant degradation and dilution which occurs,
both within and downstream of springs, is critical in the assessment of the
assimilative capacity of a karst basin. The quality of streams and rivers
used for waste-water dilution depends on natural self-purification to
assimilate wastes and restore its own quality. A waterway's capacity to
recover from influent waste is largely determined by channel morphology and
climatic conditions (Hammer, 1975). Standard environmental calculations
inclusive of the rates of biochemical oxygen demand (BOD) and rates of
deoxygenation and reaeration can be used as an aid in establishing acceptable
aquifer loading criteria for some contaminants; although it must be recognized
that both flow and atmospheric conditions in the karst system may vary
appreciably from that of surface water. This type of procedure can be
particularly useful for back-calculating likely effects of sewage disposal,
such as fish kills and eutrophication. Base line chemical and hydrologic data
is essential for this type of calculation.
A Typical Development Proposal
In order to address the types of impacts existing or planned development
may have on a karst area, a number of facts and interpretations must be
considered. Once an isolated area is proposed for a specific land-use, the
planner must determine whether it is hydrologically reasonable and prudent to
base planning decisions on a site-specific assessment. More often than not
the parcel in question is really only a small segment of a much larger
geologic and hydrologic framework.
In the rural Town of New Scotland, New York, for example, a 28 acre parcel
was proposed for Rural Commercial zoning. The parcel is situated proximal to
the Stewart's label on Figure 1, and incorporates a portion of a junk yard
property and a segment of the Onesquethaw Creek channel. The proposed zoning
allowed for multiple uses of the parcel including: shopping plazas,
laundromats, public works garages, solid waste disposal facilities, sewerage
systems, car washes, restaurants, etc. The maximum allowable building size
proposed was 9 acres. This rural area does not have a sewer district.
The area is maturely karstified as shown by a lack of surface drainage for
most of the year. Much of the watershed that includes this parcel consists of
exposed limestone, with a thin and sometimes absent soil-mantle. An extensive
cave system has been explored and mapped throughout part of the watershed.
There are also wells which intercept water flowing in cave passages which are
accessible through natural fissures. These wells and others that encounter
caves during drilling, are referred to as conduit wells. Cave streams and the
related spring (aquifer discharge point) have flashy hydrograph responses to
storm and runoff events, thus indicating turbulent flow in conduits typical of
karst aquifers.
In an effort to promote a convenience store (Stewart's), the Town decided
to advance the Rural Commercial segment of the Comprehensive Land Use Plan
still under preparation. The Town did not contract for any detailed studies
of the hydrology and geology of the site or of its surroundings. A public
hearing was held, with the author presenting the results of his geologic and
hydrologic study of the karst basin. The basin was characterized by him as
environmentally extremely sensitive. Soon thereafter the Rural Commercial
district was approved, Stewart's applied for a building permit, which was
granted and the convenience store has now been constructed. Although no
comprehensive basin-wide zoning approach was adapted, a written and oral
record of known environmental concerns was created.
772
-------
— 550
1670^
MIL
POND
GAGING
STATION
-605
BENNETT
FEET
O
A
ELEVATION MSL (FT)
CAVE OR SPRING
WATERSHED IDENTIFIER
WATERSHED BOUNDARY
HELDERBERG
ESCARPMENT
RISE DIRECTING
MELTWATER FLOW
COPELAND
HILL
FIGURE 1: TOPOGRAPHY, DRAINAGE BASINS, AND SELECTED FEATURES ALONG THE HELDERBERG
ESCARPMENT, ALBANY COUNTY, NEW YORK. WATERSHED BOUNDARIES BASED ON TOPO-
GRAPHY, GEOLOGY, AND DYE TRACES. CONTOUR INTERVALS VARIABLE FOR CLARITY.
773
-------
Characterization Necessary For Zoning Decisions
Evidence of the Presence of Subterranean Karst Networks
The first step toward protecting karst aquifers is to determine whether
carbonate or other soluble (i.e., halite, gypsum) bedrock is present. When
the answer is in the affirmative, an assessment of the nature of the aquifer
is in order. Physical evidence of the presence of subterranean karst networks
includes: 1) large and small enterable caves; 2) sinkholes; 3) areas of
exposed carbonate bedrock with numerous dissolutionally enlarged joints
(fractures); 4) limited to non-existent surface drainage in select areas; 5)
piracy (sinking) of surface water into sinkholes and streambeds; 6) presence
of open conduit and alluviated springs; 7) presence of overflow springs; 8)
variable chemical composition of springs and conduit wells; 9) rapid
subsurface flow as documented by tracer tests; 10) boreholes having
encountered large and/or small conduits; 11) loss of circulation fluids into
conduits found during drilling operations; 12) highly variable and abnormally
large water table fluctuations in some bedrock wells (especially those with
documented cavities).
Karst Inventory
A full inventory of karst features (as indicated above) is needed to
monitor the karst systems recharge, discharge, water quality, and
environmental sensitivity. Such an inventory is important in focussing future
site work, in land-use planning, and in determining appropriate sites for
aquifer characterization (e.g., sinkholes, springs, conduit wells). A number
of additional sources of information should be explored in assessing the karst
features of an area.
Published geologic and hydrologic reports are the most valuable sources of
information, but recent and historic topographic maps are also useful.
Historic topographic maps often show sinkholes or sinking streams which recent
land-use practices may have obliterated. Examples include the filling in of
sinkholes with soil or waste materials and physical alteration of sinking
streams to create ponds. Although topographic maps are good sources of
information, they often have many errors. Errors can include the connecting
of stream segments where streams are pirated into the subsurface or the
failure to portray numerous shallow sinkholes because they fall within the
map's contour interval, have been filled in, or were not recognized by the
photo-interpreter making the topographic map.
When available, historic and recent aerial photography can supplement or
correct topographic map errors. Low-cost stereo pairs can be obtained from a
variety of sources including the United States Department of Agriculture, the
United States Geological Survey (USGS), the National Archives, and assorted
state and local offices. Often, local aerial photography firms can provide
previously flown stereo coverage. Ideally, large scale photography taken at a
time of year without leaf or cloud cover is desirable. Interpretation of karst
features is best achieved through the use of a mirror stereoscope with
attached binoculars, although a simple lens stereoscope will suffice. Aerial
photographs can also be used to define the topographic boundaries of a
watershed and assist in determining sinkhole growth rates, but topographic
boundaries do not necessarily coincide with groundwater basin boundaries.
Usually they do not.
One of the more important resources for learning of karst features in an
area is people: local homeowners, farmers, hunters, fishermen, well drillers,
geologists, and children. Those individuals who roam or work an area for many
years can often provide information which might otherwise be overlooked.
Examples include small sinkholes filled in by farmers, turbid well water
coincident with storm events, caves played in by youths, and wells
intersecting conduits. Drillers' and geologists' logs from well and
774
-------
foundation studies may provide valuable insight into a karst system.
Occasionally a well may encounter a cavity, a zone of increased permeability,
or a zone where drilling fluid circulation was lost. These locations, when
properly monitored, may provide important information on the hydraulics and
water chemistry in a system. This is particularly true when they are
determined to be within a cave conduit. Such conduits may also be used as
tracer injection or monitoring points.
Homeowners can often provide important information about a karst aquifer
system. In the Clarksville, New York area they identified two "wells" which
were actually pipes placed within cave passages (see Cave Well on Fig. 1).
Both such wells, and their ultimate discharge point at the Mill Pond, had
historic diesel fuel contaminant problems.
Finally, there is no substitute for hydrogeologic assessment and objective
field reconnaissance. Some of the important features and items which should
be included in a karst inventory are the location of: 1) sinkholes; 2) caves;
3) springs; 4) sinking streams; 5) seasonally active overflow springs; 6)
limited to non-existent surface drainage; 7) conduit wells; and 8) evaluation
of existing chemical analyses of wells, springs, and streams.
As karst features are added to the inventory, it is necessary to
comprehensively evaluate the importance of each and how they combine to define
the hydrology of an area. Base maps are valuable for plotting these features.
It is also important to place known or suspected contaminant sources on this
base map.
Karst Basin Evaluation
Evaluating the hydrology and geology in any karst basin requires a multi-
phased approach. Prioritization of these phases is dependent on: (1) the
urgency of the particular land-use plan (2) the potential threat to aquifer
and surface flow systems (3) on the amount of information already known about
the karst terrane (4) time, and last but not least, (5) money. Evaluation of
a karst groundwater basin should include all or many of the components
discussed in this paper, depending upon the scope of the anticipated project,
and its budget. Often, different aspects of the evaluation may be conducted
concurrently. It is important that evaluations of this nature not only define
the dynamics of a karst system, but that the information then be utilized to
protect the subsurface and surface water resources.
The entire groundwater basin needs to be characterized, not only the small
segment in question. Land-use planners who do not take this broad view tend
to have serious unforeseen problems. Small land-use projects with limited or
no waste streams may not require complete definition of all these elements.
However, this is generally only the case where a large base flow is maintained
year round and/or a significant soil-mantle is present.
Watershed Delineation
In order to assess the potential impacts of waste streams leaving an area
of proposed development, it is first necessary to define the surface and
subsurface watershed boundaries. This assessment includes at least a partial
definition of surface and subsurface flow components. Although the ultimate
definition of a groundwater basin will be portrayed as a single irregular
watershed boundary, it can be useful in assessing vulnerable aquifer segments
and prudent land-uses to determine this boundary via a three phased approach;
the topographic basin, the lithologic basin, and the hydrologic or groundwater
basin. The term "lithologic basin" is introduced here for its utility in
locating particularly vulnerable portions of karst basins, where carbonates
are bounded by less soluble and possibly deformed geologic units, and for its
function in isolating directions to search for potential aquifer exit
775
-------
pathways. These boundaries may be defined through the use of published
geologic maps, low-altitude stereo aerial photography, USGS topographic maps,
tracer studies and, where necessary, by detailed geologic mapping.
Topographic Basin
A preliminary assessment of an area's surface water basin may easily be
made with topographic maps and supplemented with low-altitude stereo aerial
photography. The standard USGS 7.5-minute maps at a scale of 1 inch = 2,000
feet is practical for most problems. Annotations may be made on the
topographic map while using a stereoscope to interpret aerial photography. It
is always prudent to field check all areas, especially those which are
questionable. The surface watershed boundaries determined from this method
can be useful as a first step in delineating the actual groundwater basin
boundaries. Further refinement can be achieved through the evaluation of
other geologic and hydrologic factors; the lithologic and hydrologic basin.
Lithologic Basin
The geology of an area may make groundwater basin boundaries non-
coincident with surface water basin boundaries; meaning that subsurface flow
routes may extend far beyond, or even in another direction from that of a
surface stream. The lower permeability of non-carbonate bedrock units
generally constrains groundwater flow, regardless of whether it results from a
stratigraphic or structural break (e.g., faulting, folding). It is often
possible to expedite the watershed evaluation process by maximizing the use of
existing geologic and hydrologic maps of the area. These maps can then be
used to predict structural or lithologic (relating to the physical character
of a rock unit) controls on watershed boundaries. If no work has been
conducted, it may be necessary to do geologic mapping. For example, detailed
geologic mapping outside the K-25 area at Oak Ridge, Tennessee has revealed
that relatively impermeable non-carbonate units (Rome Formation and Nolichucky
Shale) form impermeable boundaries to aquifer waters within the thick
carbonate sequence surrounding most of the K-25 complex area (Fig. 2).
Upturned Rome and Nolichucky beds force all subsurface water discharging from
the drainage basin to move along valley trend which is coincident with strike.
Interpretations like this are essential for predicting exit pathways for
contaminants and for limiting the area to be searched for spring resurgence
points (Palmer, 1986).
The map in figure 2 illustrates the end product of such a geologic
assessment for the Oak Ridge, Tennessee area. The original geologic map was
subdivided into 33 stratigraphic units. Based on lithology, carbonate bed
thickness, and relative bedrock solubility, geologic formations were divided
into 2 hydrologic units; carbonate and non-carbonate. Thus, the potential for
major karst development within a topographic-based watershed boundary was
further subdivided. The cross section in figure 2 shows how bedding or faults
may further constrain karst development and contaminant exit pathways. This
map delineates several parallel and distinct karst flow systems. Analysis of
geologic logs in a portion of one of these carbonate bands revealed that 34
percent of coreholes encountered voids, providing evidence for mature
karstification.
In another example, detailed geologic mapping in the Clarksville, New York
area revealed that the Esopus Shale had been thrust up against the Onondaga
Limestone (Fig. 3). This uplifted wall of impermeable shale (sub-parallel to
the east border of the map) forces all subsurface water draining southeast in
drainage basin B of Figure 1 to flow south proximal to this fault zone (Rubin,
1991a), resurging at the Mill Pond spring. Without this knowledge, residents
east of the fault incorrectly believed that septic contamination of their well
water was from liquid farm waste spreading which frequently enters the cave
system.
776
-------
Eblen Cave
N34*W
A > True North
V
Miles
AQUITARD
AQUIFER
Index Map
Non-Carbonate Formations
CarbonateAquifer Odk Ridge
Stratigraphic Contact
Fault
Known caves
Arrows indicate possible strike-controlled subsurface flow
direction. Some or all paleo and recent flow in large solution
conduits initially graded to paleo-Clinch River baselevels.
Figure 2: Cave development and possible contaminant exit pathways in structurally deformed regions are
often controlled by bedrock lithology and former or recent baselevels. One aspect of
characterizing subsurface flow in carbonate aquifers requires locating and monitoring springs.
Spring locations may be constrained by strike-controlled lithologic boundaries and river
baselevels. In Oak Ridge, TN, discharge points may be located adjacent to the Clinch River
towards the southwest and/or northeast. Buried or relict karst, which may still function as
dominant subsurface drainage routes, can be defined through geophysical, boring, and tracer
studies. (PAR / PJL 10/28/91 ORNL Drawing)
777
-------
STOVE PIPE PALEOGORGE
"< "- > J/
.^- ^ '-;: V
CLARKSVILLE CAVE */" *
LOW FLOW CONDUIT
HYPOTHETICAL LOCATION
OF CAVE PASSAGES AS
INFERRED FROM DYE TRACES
mmmitfm: PALEOGORGE
/
-------
Hydrologic (Groundwater) Basin
The ultimate definition of a karst groundwater basin requires specific
knowledge of the physical boundaries of all water entering and leaving the
basin. Definition of topographic and lithologic basins aids in defining the
boundaries of groundwater basins, however, topographic divides rarely coincide
with subsurface divides in karst terranes (Quinlan, 1989). Sometimes, during
and after floods, some groundwater may be diverted from one groundwater basin
to one or more others adjacent to it (Quinlan, 1990, Figs. 2 and 3). Regions
with complex geomorphic histories, such as Oak Ridge, Tennessee (Fig. 2), may
require advanced testing and analysis to define karst systems grading to
multiple baselevels. Structural and lithologic controls, however, will still
impose certain limits on the extent and boundaries of karst basins, as may
hydrologic baselevels.
All karst aquifers have at least one discharge point. Discharge points
(resurgences) generally occur as springs at a local or regional baselevel
feature such as a stream, river, lake, or ocean. Spring locations may be
predicted by evaluating the combined topographic and geologic map. Palmer
(1986), documents how bedding orientation may be used to predict likely
directions of contaminant transport and groundwater flow. In many settings,
it is essential to document the actual subsurface flow direction, destination,
and velocity with one or more tracer tests. Where little surface flow is
present, it is often possible to force flush a tracer through the karst
system. A pre-injection systematic search of potential baselevel outlet areas
may serve to locate springs. Local farmers, homeowners, hunters, and
fishermen are often excellent sources of information, but there is no
substitute for field work to search for them.
The type and thickness of soil cover in an area will influence the amount
and extent of infiltration (downward percolating water) reaching an underlying
karst aquifer. A detailed examination of all surface streams and their
tributaries may reveal that little to no flow is pirated (diverted) to the
subsurface. In many karst areas, however, soils perch surface streams for
short distances prior to their partial or complete loss to the subsurface.
Infiltration areas may be detected by conducting seepage velocity (discharge)
measurements at successive points along streams where a losing component is
suspected. One common situation found in karst terranes is that of surface
flow over impermeable soil or bedrock formations present in the higher reaches
of a karst watershed. This water flows downhill until it sinks near an
exposed limestone contact, where numerous sinkholes are often aligned.
When considering acceptable land-uses and their resultant contaminant
loading, it is important, when possible, to evaluate the ability of the
bedrock or soil present to treat infiltrating septic or other waste.
Practical Land Use Planning: Two Examples
An excellent example of a karst area requiring specialized land-use
planning in order to avoid both significant aquifer and surface stream
degradation is a broad karst aquifer present in the Clarksville area (Fig. 1),
specifically the Mill Pond karst basin which forms the headwaters of the
Onesquethaw Creek. Since urban expansion was extending into the surface water
basin, a study was undertaken to define potential development concerns. No
funding was available, so the steps followed had to be capable of reliably
characterizing potential contaminant concerns with limited manpower and
resources. This type of limited approach may be all that many towns will ever
have the resources to have performed. Proper characterization of a karst
groundwater basin will require a qualified individual or firm, as discussed in
the closing "Who to look to for help" section of this paper.
The key issue for planners in karst areas should be the water quality of
the aquifer and its receiving springs and streams (e.g., including water
779
-------
supply, ecosystem, and aesthetic concerns). In order to objectively appraise
land-uses and their potential impacts in karst basins, it is necessary to
define the hydrologic boundaries (including all discharge points), ag_uifer and
receiving stream vulnerability, and many of the items discussed in the
assimilative capacity section of this article. To evaluate potential impacts
in the Mill Pond karst basin, a stepwise strategy was followed, with various
steps occurring concurrently. These are numbered for ease of reference.
1) Topographic maps were obtained of the area. A first approximation
of the surface water basin was defined to determine approximate
boundaries of the area of concern.
2) Existing published and unpublished geologic and hydrologic reports,
chemical analyses, cave maps, and stream flow data were collected to
provide a current basis for evaluating existing conditions.
3) Maps of the bedrock geology and soil thickness were obtained from
existing sources. The area of carbonate bedrock was overlain on the
surface water basin obtained in step 1), above. It was determined
that the geology was such that no subsurface flow in carbonates was
likely to enter from outside the topographic basin. This possibility
must be evaluated on a case by case basis.
4) All known karst features were inventoried. These included features
readily apparent in the basin (e.g., sinking streams, sinkholes,
caves, springs) and others learned of from local residents (e.g.,
conduit wells, caves). Past and present contaminant problems were
also noted as part of the inventory. Most state regulatory programs
maintain a file of regulated environmental problem sites.
5) Existing geologic and hydrologic reports and maps were evaluated for
likely aquifer discharge points. Baselevel discharge areas were
searched for springs.
6) The location of significant karst features were field-checked and
placed on the working base map.
7) Caves were mapped. (This step may not always be necessary.)
8) A major fault was identified in a cave. This discovery suggested
that a faulted shale unit may be forming a groundwater basin
boundary to the east. The fault was not present on published
geologic maps, but was found and readily mapped in the field (Fig. 3).
9) Existing stereo aerial photographs were obtained and utilized to
further refine the geologic contacts in the basin. Several
additional karst features recognized on the aerial photographs were
accurately plotted on the base map.
10) Known or suspected contaminant sources were plotted on the base map
(e.g., junk yard).
11) Areas of dense housing (likely contaminant sources) were also
plotted.
12) Several qualitative tracer tests were conducted to confirm
suspected flow paths and roughly approximate subsurface travel
times. It was determined that the gentle southwesterly dip of the
bedrock present in the Mill Pond aquifer fails to direct
all subsurface flow in this direction. Instead, the significantly
higher surface topography to the southwest (Wolf Hill and Cass Hill)
retards dissolution in this direction, in favor of the 1.3° apparent
dip between Wolf Hill Dam and a baselevel discharge point at a
spring in the Mill Pond. Tracer studies generally verified
780
-------
predicted flow paths, at least during periods of low discharge when
the entire headwaters of the Onesquethaw Creek resurge from the Mill
Pond. However, tracer studies also documented the unexpected
easterly diversion of moderate to high discharge waters through
Pauley Avenue in Clarksville Cave (Fig. 3).
13) A gaging station was established downstream of all karst springs
and at a point beyond which no subsurface flow from carbonates was
possible. This was monitored daily for 15 months, and periodically
thereafter during periods of low and high flow. Monitoring for this
duration may be unnecessary, since it is the lowest flows that were
of the greatest concern for estimating the likely assimilative
capacity for contaminants. Base flow from the basin was gaged at
less than 0.1 cubic feet per second (50 gallons per minute).
14) The karst system was evaluated based on the results of the
investigations conducted and analysis of stream hydrographs.
Additional information could also be obtained on the nature of the
system using relatively simple physical properties (e.g., specific
conductance, temperature, and turbidity) of spring waters. Quinlan
et al. (1991) have put forth a simple, practical, and reliable
evaluation technique for classifying the vulnerability of carbonate
aquifers based on the coefficient of variation of specific conductance
of spring waters. The technique evaluates the net effect of an
aquifer's recharge, storage and subsurface flowpath characteristics. A
fourth category, which might be evaluated as part of the vulnerability
classification, is aquifer base flow. In the nomenclature of Quinlan
et al. (1991), the Mill Pond karst aquifer may be classified as
hypersensitive.
15) The environmental sensitivity of the karst basin was ascertained,
placing particular emphasis on the vulnerability and likely
assimilative capacity of the groundwater basin. As part of this
process, consideration was given to what adverse effects might
result from contaminant types which would be permissible within the
broad Rural Commercial District guidelines.
16) Findings of the investigation and concerns regarding contaminant
loading were shared with the town planning board (designated lead
agency), town engineers, local and state agencies, local
environmental groups, and the press via reports, letters, and
telephone discussions. The findings were also presented at a
public hearing.
17) A final step in the process is to work with the agencies
involved to draft a management plan adopting land-uses within the
geologic and hydrologic constraints of the karst basin.
The Mill Pond watershed was subdivided into two parts: A) the 3,075 acre
portion located upstream of Wolf Hill Dam, and B) the 2,050 acre portion
located downstream of Wolf Hill Dam. The boundaries of the aquifer were
determined to extend to the north and northwest of the basin's resurgence
point at the Mill Pond (Fig. 3). The farthest boundary of the Mill Pond
groundwater basin lies some 2.4 miles to the northwest, proximal to the Wolf
Hill Dam on the Onesguethaw Creek. The boundaries of the catchment basin are
depicted in bold dashed lines. The upstream part is comprised of insoluble
rock, with much or all discharge being seasonally diverted to the Vly Creek
Reservoir. The downstream part of the Mill Pond watershed has features
characteristic of karst terranes. These include sinking streams, limited
surface drainage, solutionally enlarged joints, sinkholes, and the Clarksville
Cave system. During most of the year, water entering that portion of the
watershed that is downstream of the Wolf Hill Dam is pirated into the Onondaga
Limestone. Tracer tests and in-cave stream gaging indicate that flow is
turbulent and has a maximum straight-line flow velocity of 5.3 km/hr (3.3
781
-------
mph). During periods of high discharge, contaminants have the ability to
travel from one end of the aquifer to the other in less than one hour.
A detailed analysis of the flow dynamics present in the Mill Pond karst
basin was conducted. A gaging station was established in Onesguethaw Creek
(Fig. 1) in order to examine the relationship between in-cave discharge and
surface-watershed discharge. This was monitored twice daily for 15 months,
more frequently during flood events, and periodically for 4 years thereafter
during major runoff events (Fig. 4). Stream discharge was measured at 13
different stages. Curvilinear regression was then utilized to establish a
series of multi-order equations that could be used to correlate stage height
with discharge. The greatest discharge recorded for Onesquethaw Creek during
the course of this study was approximately 1340 cfs (600,000 gpm). This
occurred on March 15, 1986 at 3:00 am following heavy rains (=2.7 in.) on a
15-inch snow pack. Daily monitoring of stream stage in Clarksville Cave for
the same 15-month period revealed that a direct correlation exists between
this discharge and that in Onesquethaw Creek. Approximately 8 percent of
flood-peak discharge in Onesquethaw Creek flows through Clarksville Cave.
In this example, it is clear that major land-use in the central portion of
the basin (e.g., Rural Commercial District), where almost no soil-mantle is
present above heavily jointed limestone pavement, may have a significant
environmental impact on the basin's receiving stream. Land-uses here which
directly and/or continuously discharge contaminants to the conduit system
should be carefully evaluated. In this setting, contaminants will quickly
enter the fractured limestone and shallow underlying cave passage, where they
will be directly piped to the Mill Pond spring and the Onesquethaw Creek.
Continuous waste streams are more likely to adversely impact this geologic
setting than those which enter the aquifer during times of saturated soil
moisture conditions, high runoff, and infiltration. However, it should be
noted that one-time or infrequent chemical slugs can, in some karst aquifers,
cause greater problems than lower controlled or permitted releases.
The findings of this investigation were not used by the Town to develop a
new overall master plan. Instead, the Town ultimately decided to continue
issuing building permits on a case by case basis, presumably now taking into
account the known environmental sensitivity of the aquifer. All the desired
land-uses cited in the Typical Development Proposal section of this paper
remain intact. Towns in situations such as this may ultimately have to answer
legally and politically to their constituents. The importance and need of a
comprehensive karst educational program at local and state levels is apparent
and cannot be overemphasized.
Another example of land-use planning on a large raised karst watershed is
also situated in upstate New York. A number of private homes were proposed
for development upslope of a karst spring used by a village for water supply.
The bedrock geology was well documented, with most formations within the
watershed including caves. A relatively thin soil-mantle was present, along
with several areas of dissolutionally enlarged joints and sinkholes. After
many meetings with health officials and politicians, a preliminary decision
was made to "protect" the aquifer by imposing a several hundred foot set-back
distance from all sinkholes. The set back zone concept incorrectly
presupposes that sinkholes are independent from the area-wide karst system
(Quinlan et al., 1991). Unfortunately, this decision did not take into
account the important fact that any infiltration of septage through the
surrounding permeable soil-mantle would probably quickly enter the underlying
karst system. Although special considerations may be necessary, development
in areas such as this is still possible. The question is really what is
reasonable and prudent and within the assimilative capacity of the aquifer.
782
-------
0.40
F£
co a
|a!
LU£
m*
0.80
1.20
1.60
STEADY DECLINE
IN CREEK STAGE
2.00 ' I i I i I i I ' I M i I i I i I i I i I i I i I i I i I i I ' I i I t I ' I i
11/1811/2211/2611/3012/4 12/8 12/1212/1612/2012/2412/28 1/1 1/5 1/9 1/13 1/17 1/21 1/25 1/29 2/2
I 1985 I 1986
i I i I i I i I i I i I i I
2/10 2/14 2/18 2/22 2/26 3/2 3/6
89
46
32
O
LU
24.5 g
9.6 <
5.3 a
2.5 co
a
oo
CO
14
•w 12
I10
* 8
O
£ 6
5
co
Figure 4: Onesquethaw Creek hydrograph. Stream discharge was measured at 13 different stages. Curvilinear regression was then utilized to
establish a series of multi-order equations that were used to correlate stage height with discharge.
4.43 i
140
^+^
SHADED AREA
EQUALS
TEMPERATURE
, ABOVE 32'F
10 14 18 22 26 30 3
7 11 15 19 23 27 2 6 10 14 18 22 26 30 3
JANUARY 1984-MAY 1984
7 11 15 19 23 27
4200
3600 _
3000 g.
2400 LU
a
1800 g
1200 i
O
600
0
Figure 6: Operable dimatic conditions influencing salt contaminant transport from a roadside near Saugerties, New York. Salt infiltrates soil, passes
through fractured limestone (against dip), and contaminates wells (A °. and n) and a baselevel spring.
-------
Advanced Testing And Analysis To Define A Karst System
A number of additional steps may be incorporated into a program to define
a given karst network. Some may not be necessary if existing or planned land-
uses pose little or no threat to an aquifer or its receiving stream. These
components may include cave surveying, chemical and hydraulic monitoring, well
drilling, geophysics, and tracer tests. Some of these methodologies are
discussed below.
Physical Entry Of Karst Networks
The survey and characterization of caves in karst aquifers is valuable
because the actual location and extent of segments of the karst system can be
examined. There is no substitute for entry and survey of enterable segments
of the karst system. Visual reconnaissance of the inner workings of the karst
aquifer can provide information of primary transport pathways and can serve to
locate monitoring stations. Cave explorers and cave-divers represent one of
the best resources when conducting land-use studies in karst terranes.
Trained cave surveyors are almost always willing to help towns and federal
agencies with this integral aspect of defining karst systems, generally at no
charge. Potential access points include cave entrances and springs.
Sometimes access can be gained by excavating sinkholes or overflow springs.
Large diameter boreholes are good entry points where the conduits are
seasonally above the saturated zone. More recently, trained cave divers have
penetrated spring resurgences and brought back valuable survey and geologic
assessment. Sometimes it may be worthwhile to utilize heavy equipment to dig
or drill into blocked portions of a cave system.
Cave Detection With Geophysical Techniques
Geophysical techniques can sometimes be used for identifying conduit
locations and connectivity. Micro-gravity and electrical resistivity are two
more successful techniques, but no technique is capable of resolving caves of
small size at great depth. The work of Kilty and Lange (1991) and Lange and
Kilty (1991), however, appears encouraging for locating streams at depth.
Geophysical investigation can be followed with coreholes, dye traces, and
continuous hydraulic and chemical monitoring of wells and springs.
Water Quality, Hydraulic And Biologic Monitoring In Karst Aquifers
Great emphasis should be placed on learning more about the hydraulics and
contaminant distribution, if any, in accessible portions of a karst system.
Because a representative chemical characterization of the bedrock flow system
can only be obtained from mature conduit segments, storm water should be
sampled for contaminant transport after either, 1) karst springs are located,
or 2) hydrographs have been obtained from conduit wells, indicating that they
are part of the active conduit system.
Water Quality Monitoring Of A Karst System
Select water quality indicators, as well as suspected or planned
contaminants should be monitored during periods of low flow and most
importantly, during short-term flood events. Monthly or quarterly monitoring
is typically conducted in porous and fractured bedrock aquifers; but, the
highly transmissive nature and flashy response of some mature karst aquifers
requires a more frequent sampling schedule. An infrequent sampling schedule
in karst aquifers having rapid flow may miss the short periods of time when
chemicals are actively being transported (Quinlan, 1989, 1990). However,
different types of contaminant releases often require different sampling
schedules. Sampling schedules must be tailored to the geologic and physical
784
-------
setting of contaminant sources, taking into consideration the hydrologic
conditions most likely to result in contaminant dispersal, infiltration, and
movement. If a contaminant source is an unregulated spill into the soil zone
above a karst aquifer, storm flow monitoring may detect it; however, if the
contaminant source is a continuous discharge from a point source (e.g., from a
package plant) it may well not be detected during flood flow conditions.
Traditional monitoring frequencies may thus fail to indicate the true
distribution of a contaminant problem in a karstified aquifer. Sampling of
water chemistry in monitoring wells placed in the diffuse (non-conduit) flow
zone of a karst aquifer will typically yield unreliable data, not
representative of the system as a whole (Quinlan, 1989, 1990; Ewers, 1991).
Figure 5 illustrates the water level difference in a physical setting of
wells placed only a few meters apart, where substantially different hydraulic
conductivity and transmissivity values and chemical concentrations may be
obtained. The term "tertiary porosity" has been introduced by Teutsch and
Sauter (1991) to refer to dissolutional porosity, rather than secondary
fracture porosity. The failure to find contaminants in wells not within the
active conduit portions of the karst system may falsely suggest that no
contaminant problem is present. It is reasonable to focus contaminant
characterization and exit pathway studies on the mature zones of well-
karstified aquifers because significant remediation of contaminants found in
diffuse-flow zones of karst aquifers is unlikely to impossible (Quinlan and
Ray, 1991).
Monitoring must be conducted in locations that intersect the active
conduits of a flow system. While wells demonstrated to be in conduits may
provide reliable water chemistry data for their sub-watershed areas, the best
and most representative sampling locations for a system as a whole are at
springs (Quinlan and Ewers, 1985).
An example of the relationship between soluble contaminant movement and
assorted climatic conditions in a shallow highly fractured limestone is
illustrated in Figure 6. In this example, road salt infiltrates soil, passes
through the bedrock (against dip), and contaminates wells and a baselevel
spring. Periods of snowmelt, high rainfall, and high temperatures can
significantly increase contaminant transport rates. Lack of these conditions
can result in low level contamination. Frequent measurements of conductivity
would serve as an effective screening tool to select optimal sampling times.
Again, the frequency of groundwater sampling in karst terranes must mimic the
flow dynamics and the contaminant source type.
NW
Soil
Figure 5: Schematic cross section showing hypothetical low K (hydraulic
conductivity) and T (transmissivity) diffuse-flow zone secondary
porosity (SE well) and high K and T conduit tertiary porosity
(NW well) within carbonate aquifer.
785
-------
Continuous Chemical Monitoring Of Solution Conduit Waters
Quinlan and Ewers (1985), Quinlan (1989), and Quinlan et al. (1991) have
documented practical strategies for monitoring karst aquifers and contaminant
transport within them. Basic to these strategies is the understanding that
conduit flow may be turbulent and is often quite rapid. An entire contaminant
pulse might pass by a monitoring site in a matter of hours, rather than months
(see Fig. 7). The rapid response of the conduit system versus that of the
diffuse system is a unique feature of some karst aquifers. A sudden influx of
water within a conduit system may result in water level rises of tens of feet
in a few hours. These sudden rises signal the filling of a groundwater
trough, which may bring flood levels up to or above the water table observed
in the diffuse flow portion of the karst aquifer. Special efforts are
required in karst aquifers in order to detect and characterize episodic or
pulse-driven contaminant movement.
Monitoring the relevant water chemistry, water levels, and flow components
of a karst system is virtually impossible unless wells have deliberately or
accidently intersected active conduits or, as is usually the case, springs
have been found by field reconnaissance efforts and shown by tracing to be
draining from the source pollutants, and thus to be relevant.
Once relevant springs and conduit wells have been identified,
representative sampling of the aquifer can be conducted at times of suspected
contaminant transport (i.e., flood pulses). First round sampling might be
conducted for major ions, nitrate, ammonia, phosphates, fecal coliform,
specific conductance, dissolved oxygen, biochemical oxygen demand, turbidity,
pH, temperature, and known or suspected contaminants.
=5 790 —
I 770
55
750 -
730
.Storm-flow response
Base-flow conditions
0
I
2
\
3
Days
Figure 7: Hypothetical hydrograph of flashy response of a conduit well in
a mature karst system.
Hydraulic Monitoring Of Wells Intersecting Solution Cavities
Karst groundwater basins typically contain both diffuse and conduit flow
components. Diffuse zone in-feeders, tributary to highly transmissive
conduits, would typically exhibit slower water movement until fracture
aperture was sufficient to permit groundwater flow to cross from laminar to
turbulent flow conditions. Even poorly integrated fractures in the karst
system still ultimately drain toward a groundwater trough where transmissivity
is relatively higher.
786
-------
Transducers and data loggers can provide the continuous recording
necessary to properly define the hydraulics of wells known to be in the
conduit system. Conduit flood pulses may be short-lived (i.e., hours or
days), depending on the maturity of the conduit system and the quantity of
recharge incident upon the system (Fig. 7). Thus, in order to understand the
hydraulics and characterize the water chemistry of the system, it is sometimes
necessary to monitor on an hourly rather than monthly or quarterly sampling
schedule (Quinlan and Alexander, 1987). In a karst network, long intervals
between sample collection are likely to miss storm events and critical
contaminant transport entirely. Frequent monitoring and sampling during flood
pulses is necessary.
Wells which have large fluctuations of their piezometric surface,
especially when followed by a short recession curve, may be interpreted as
being part of an active and mature karst flow system (Fig. 7). Water flowing
through these wells, and their associated dissolution conduits ultimately must
discharge to some baselevel.
Karst Groundwater Basin Flow Components
Continuous monitoring and quantification of discharge entering and
leaving a karst basin is important for contaminant characterization and some
land-use studies. For larger projects, where contamination is or may be a
problem, this may include stream, spring, and conduit well discharge
monitoring via weirs and stage recorders (or other devices). Seepage velocity
(discharge) measurements on tributaries suspected of being losing streams
should also be considered. One or more weather stations are critical for
realistic water budget appraisal. For small projects, it may be sufficient to
monitor only the base flow discharge leaving the karst system during drought
or low-return periods.
Knowledge of expected flood-return intervals and their magnitude can be
important in assessing contaminant transport potential. One example of where
flood-return information might be applied is in predicting mobilization
probabilities of contaminants poorly bound to soils; which may occur as a
result of soil-flushing stemming from large precipitation events. In other
settings, large point source quantities of hazardous waste may be released
from elevated flow system segments only during extreme flood conditions. This
technique, which might be applied towards roughly assessing a maximum rate of
contaminant movement in a karst system, may be applicable in karst areas where
all or most of the drainage exits via springs or where sufficient in-conduit
hydrograph data is available. Where time is an element for a particular land-
use option being contemplated, and reasonable but limited stream discharge
data is available, it is possible to perform a flood probability analysis.
Similarly, knowledge of base flow and drought-return intervals, when the
assimilative capacity of an aquifer is at its lowest, can be critical to
prudent land-use determinations.
An example of such an analysis was conducted for the Mill Pond karst basin
in east-central New York (Rubin, 1991a, 1991b) where much of the basin's flow
is subsurface for most of the year. The limited data available for
statistical comparison among hydrologic years necessitated examination of
another roughly comparable basin in order to assess flood-return intervals.
The farthest headwater gaging station on Schoharie Creek at Prattsville was
selected. Many inherent differences occur between the basins, notably
elevation, geology, soil thickness, size, and location. The Prattsville and
Onesquethaw Creek gaging stations are approximately 48 kilometers apart.
However, the Prattsville and Mill Pond watersheds are comparable under
conditions of a saturated soil-moisture bank, high runoff, and similar storm
systems. Eighty-two years of data at the Prattsville station were examined.
787
-------
100
I I I I I
III I I II
• 1903-1987 LOG PEARSON
O 1903-1951 LOG PEARSON
A 1903-1987 GUMBEL
A 1903-1951 GUMBEL
1925 AND 1929 WATER
YEARS NOT AVAILABLE
I I I I I
I i i i i
10 102
FLOOD RETURN INTERVAL (Tr in yr)
103
Figure 8: Schoharie Creek at Prattsville, New York. Log-Pearson type III
and Gumbel distributions utilizing historic water year peak flow
data. Flood return interval for largest Onesquethaw Creek peak
discharge of record correlated to Prattsville March 15, 1986
flood of 54,900 cfs. Range of two methods shows a Tr of 30 to 47
years. Similar comparisons may be made for low flows in
geologically comparable basins. Analyses of this type can be
useful tools in predicting flow velocities, -—••-—«•!« = «<- = *--•:<
times, and contaminant dilutions.
contaminant arrival
A Log-Pearson Type III and Gumbel-distribution statistical comparison
(Linsley et al., 1975) of historic peak flow of Schoharie Creek gaging data
with this study's hydrograph information for Onesquethaw Creek indicates that
the largest Onesquethaw Creek peak of record (March 15, 1986) has a return
interval on the order of 30 to 47 years (Fig. 8). This corresponds to a
Prattsville hydologic-year peak discharge of 54,900 cfs. Thus, if 40 years
was the expected flood return interval, 25 floods of this magnitude could be
expected every 1000 years. These infrequent storm or runoff events reasonably
represent a near-maximum quantity of water available in the watershed under
ideal, thin-soil-mantled, rapid infiltration conditions. Knowledge of peak or
base flow return periods can sometimes be correlated with water chemistry
results to help assess chemical loading both in the karst system and to stream
receptors.
Biologic Monitoring Of The Karst System
Fauna, both within caves and at springs, provide the best indicator of
health of the karst system. Various organisms, such as benthic invertebrates,
benthic macroinvertebrates, fish, amphipods, isopods, flatworms, and copepods
can provide excellent indices of biological integrity. These organisms may be
subjected to episodic contaminant pulses and any background contaminant
chemistry. Their health, or sometimes absence, as monitored at one or more
points in time, is a true measure of the health of the groundwater system
(Poulson, 1991; Preddice, T., NYS Dept. of Environmental Conservation, pers.
comm.). Acute exposure to various chemicals can influence the ability of some
788
-------
species to reproduce or survive. This can have far-reaching effects on the
food chain. The assimilative capacity of karst waters and its biota for
contaminants is dependent upon factors such as chemical type and toxicity,
chemical injection or leakage rates, discharge in the system (e.g., low flow),
and reaeration rates along the flow path. Similarly, plant species can also
aid in the health-based assessment of a receptor stream.
Tracer Studies
Tracer studies are the best technique for determining flow direction,
destination, and velocity within a karst groundwater basin (Quinlan, et al.,
1991). State-of-the-art tracing techniques are capable of simultaneously
running multiple tracers to assess different suspected flow routes, or to
verify low quantitative results from previous traces. Tracer tests may be
successfully used to: 1) determine discharge locations of sinking streams
which may flow within shallow or deep karst systems; 2) determine aquifer
discharge locations necessary for characterizing contaminant exit pathways; 3)
determine groundwater seepage velocity rates and discharge in conduit portions
of the karst flow systems. (This is important for water budget and model
assessment, as well as preventative planning for episodic contaminant or spill
releases); 4) define drainage divides and catchment basin boundaries; 5)
define subsurface flow and transport routes within conduit and diffuse zones;
6) assess stream infiltration/piracy to shallow surficial discharge points;
and 7) realistically characterize groundwater seepage velocity rates in the
diffuse zone between boreholes.
Wells
Wells which are part of the active conduit system, as determined by
transducer/hydrograph response, or their locations in known cave passages, may
provide the best monitoring locations when springs cannot be located (e.g., as
a result of flooding). Wells have value for 1) lithologic characterization,
2) structural characterization, 3) piezometric monitoring, 4) chemical
monitoring, and 5) tracer studies.
Areas where an unusually high percentage of boreholes have encountered
voids (e.g., Oak Ridge, Tennessee) are indicative of maturely karstified
terranes. Such settings are uncommon. Quinlan and Ewers (1985) and Ewers
(1991) have estimated the odds of encountering a dissolutional conduit in a
karst aquifer by drilling a well at about 1:2600.
Political Realities Of The Land Use Permitting Process
Responsible land-use planning requires a willingness of all parties
involved in the planning and permitting process to look beyond the physical
boundaries of a particular building proposal which may be in front of them.
This may ultimately mean that the scope and scale of development needs to be
reduced. This, in turn, may make landowners unhappy that they can no longer
subdivide or use their land as they might wish. Unhappy landowners do not
reelect their politicians, who are thus hesitant to spend dollars on
consultants. Some consultants, who desire continued work with towns, may wish
to insure their client's approval. This process may be further complicated by
town planning board members who are themselves, or have relatives and friends
who are, developers. A further complication in some states may result when
lead agency/permitting status is granted to towns which may have vested
interests. In instances such as this, a cooperative agreement for technical
exchange between knowledgeable state and government people and planning boards
might be beneficial to all parties involved. Thus, it should be recognized
that prudent land-use planning may, in some instances, be fraught with
political conflict and pressures. All to often, the local residents and the
environment loses.
789
-------
Sound land-use planning is best conducted in the absence of specific
building proposals. It is important to recognize impending problems before
building permits are on the table. Ideally, land-use planners in karst
terranes will be able to conceptualize the subsurface flow systems present in
their basins and incorporate geologic and hydrologic information into a zoning
master plan. Such master plans require 1) planners with some scientific
background, 2) sufficient capital and expertise to characterize a karst basin,
and 3) an educational process, such as that now being undertaken by the
National Speleological Society (NSS) and the American Cave Conservation
Association (ACCA) in the United States. Public hearings and board meetings,
which provide a forum for the educational process, can be used to good
advantage with receptive audiences. Members of these organizations and karst
hydrologists are often willing to help in this process when requested. More
emphasis also needs to be placed on educating state, federal and local
officials.
Whereas development within karst basins has historically occurred on an
individual application basis, it may now be prudent to initiate a more broad-
based master planning process. Planning should take into account the likely
contaminant loading into the karst system, and a reasonable measure of its
assimilative capacity. In many respects, the strategy employed should mimic
the discharge requirements many states impose on industrial effluents to
surface waterways. Instead of granting discharge permits with unlimited
contaminant ceilings, permits are only granted when individual contaminant
contributions do not exceed the assimilative capacity of collective discharges
from multiple facilities. In this manner, a strategy is employed to insure
the continued health of the waterway. In karst groundwater basins, a
discharge permit of this nature should be based on the assimilative capacity
of the receiving aquifer and stream at times of low flow.
Who To Look To For Help
Characterization of the hydrology of a karst basin is beyond the technical
expertise of most land-use planners, environmental consultants, or even
experts within State and Federal agencies. The flow dynamics in karst
terranes are very different from those in porous or fractured media.
Significant scientific advances in karst hydrology have occurred in the last
two decades. Unfortunately, many of the technical papers have not been
published in mainstream journals or received widespread distribution. This
poor dissemination of knowledge in karst hydrology has hindered the land-use
planning process. More recently, this trend has been broken (Ford and Ewers,
1978; Quinlan and Ewers, 1985; White, 1988; Ford and Williams, 1989; Quinlan,
1989, 1990; Palmer, 1991; and Quinlan et al. 1991).
Plainly stated, extremely few consulting firms, universities, and
government agencies with experienced hydrogeologists, geologists and engineers
possess the specialized qualifications of karst hydrologists. Finding an
experienced consultant knowledgeable about karst hydrology is difficult.
Specialists in karst hydrology may be contacted through two organizations:
Karst Waters Institute National Ground Water Association
c/o John Mylroie; President 6375 Riverside Drive
Box 2194 Dublin, Ohio 43017
Mississippi State Univ., MS 39762 614-761-1711
601-325-8774
The wise consultant will use the unique talents and knowledge of state
cave surveys and cave explorers, who are often associated with local chapters
of the National Speleological Society (NSS). Such individuals can be a
tremendous asset in the karst inventory evaluation process, as well as
valuable sources of cave maps depicting segments of karst aquifers. The NSS
may be contacted at Cave Ave., Huntsville, Alabama 35810 (205-852-1300). It
is important to verify that the firm or individual claiming expertise in karst
hydrology actually has it.
790
-------
Ac knowl edgement. s
This manuscript was written while the author was employed in the
Environmental Sciences Division at the Oak Ridge National Laboratory. ORNL is
managed by Martin Marietta Energy Systems, Inc., under contract No. DE-AC05-
84OR21400 with the U.S. Department of Energy. Publication No. 3856,
Environmental Sciences Division, ORNL.
"The submitted manuscript has been authored by a contractor of the U.S.
Government under contract No. DE-AC05-84OR21400. Accordingly, the U.S.
Government retains a nonexclusive, royalty-free license to publish or
reproduce the published form of this contribution, or allow others to do so,
for U.S. Government purposes."
Thanks are extended to Peter Lemiszki for his assistance in the
preparation of Figure 2, Linda Armstrong, Marc Casslar, Dale Huff, James F.
Quinlan, and Geary Schindel whose constructive review greatly enhanced this
paper, and to Thorn Engel who ably assisted with stream gaging and geologic
survey work.
References Cited
Ewers, R.O., 1991, The response of landfill monitoring wells in limestone
(karst) aquifers to point sources and non-point sources of contamination.
Hydrogeology, Ecology, Monitoring, and Management of Ground Water in
Karst Terranes Conference (3rd, Nashville, Tenn.), Proceedings. National
Ground Water Association, Dublin, Ohio, [in this volume]
Ford, D.C. and Ewers, R.O., 1978, The development of limestone cave systems in
the dimensions of length and depth. Canadian Journal of Earth Science, v.
15, p. 1783-1798.
Ford, D.C. and Williams, P.W., 1989, Karst Geomorpholoqy and Hydrology.
London, Unwin Hyman, 601 p.
Hammer, M.J., 1975, Water and Waste-Water Technology. New York, John Wiley &
Sons, 482 p.
Kilty, K.T. and Lange, A.L., 1991, The electrochemistry of natural-potential
processes in karst. Hydrogeology, Ecology, Monitoring, and Management of
Ground Water in Karst Terranes Conference (3rd, Nashville, Tenn.),
Proceedings. National Ground Water Association, Dublin, Ohio, [in this
volume]
Lange, A.L. and Kilty, K.T., 1991, Natural potential responses of karst
systems at the ground surface. Hydrogeology, Ecology, Monitoring, and
Management of Ground Water in Karst Terranes Conference (3rd, Nashville,
Tenn.), Proceedings. National Ground Water Association, Dublin, Ohio, [in
this volume]
Linsley, R.K., Jr., Kohler, M.A., and Paulhus, J.L.H., 1975, Hydrology for
Engineers. 2nd Edition, New York, McGraw-Hill Book Company.
Palmer, A.N., 1986, Prediction of contaminant paths in karst aquifers. In
Environmental Problems in Karst Terranes and their Solutions Conference,
Proceedings. National Water Well Association, Dublin, Ohio, p. 32-51.
Palmer, A.N., 1991, Origin and morphology of limestone caves: Geological
Society of America Bulletin, v. 103, p. 1-21.
Poulson, T.L., 1991, Assessing groundwater quality in caves using indices of
biological integrity. Hydrogeology, Ecology, Monitoring, and Management
of Ground Water in Karst Terranes Conference (3rd, Nashville, Tenn.),
Proceedings. National Ground Water Association, Dublin, Ohio, [in this
volume]
Quinlan, J.F., 1989, Ground-water monitoring in karst terranes: recommended
protocol and implicit assumptions. U.S. Environmental Protection Agency,
Environmental Monitoring Systems Laboratory, Las Vegas, Nev. EPA/600/X-
89/050 88 p. [DRAFT; final version to be published in 1992]
791
-------
Quinlan, J.F., 1990, Special problems of ground-water monitoring in karst
terranes. In Nielsen, D.M., and Johnson, A.I., eds. Ground Water and
Vadose Zone Monitoring. ASTM Special Technical Publication 1053. American
Society for Testing and Materials (ASTM), Philadelphia, p. 275-304.
Quinlan, J.F. and Ewers, R.O., 1985, Ground water flow in limestone: Rationale
for a reliable strategy for efficient monitoring of ground water quality
in karst areas. National Symposium on Aquifer Restoration and Ground
Water Monitoring, 5th, Proceedings, p. 197-234.
Quinlan, J.F. and Alexander, E.G., Jr., 1987, How often should samples be
taken at relevant locations for reliable monitoring of pollutants from an
agricultural, waste disposal, or spill site. Multidisciplinary Conference
on Sinkholes and the Environmental Impacts of Karst (2nd, Orlando, Fla.),
Proceedings. Balkema, Rotterdam, The Netherlands, p. 277-286.
Quinlan, J.F. and Ray, J.A., 1991, Ground-water remediation may be achievable
in some karst aquifers that are contaminated, but it ranges from unlikely
to impossible in most: I. Implications of long-term tracer tests for
universal failure in goal attainment by scientists, consultants, and
regulators. Hydrogeology, Ecology, Monitoring, and Management of Ground
Water in Karst Terranes Conference (3rd, Nashville, Tenn.), Proceedings.
National Ground Water Association, Dublin, Ohio, [in this volume]
Quinlan, J.F., Smart, P.L., Schindel, G.M., Alexander, E.G., Jr., Edwards,
A.J., and Smith, A.R., 1991, Recommended administrative/regulatory
definition of karst aquifer, principles for classification of carbonate
aquifers, practical evaluation of vulnerability of karst aquifers, and
determination of optimum sampling frequency at springs. Hydrogeology,
Ecology, Monitoring, and Management of Ground Water in Karst Terranes
Conference (3rd, Nashville, Tenn.), Proceedings. National Ground Water
Association, Dublin, Ohio, [in this volume]
Rubin, P.A., 1991a, Flow characteristics and scallop-forming hydraulics within
the Mill Pond karst basin, East-Central New York. In Kastning, E.H. and
Kastning, K.M., (eds), Appalachian Karst Symposium, National
Speleological Society, Huntsville, Ala., p. 101-108.
Rubin, P.A., 1991b, Modification of preglacial caves by glacial meltwater
invasion in East-Central New York. In Kastning, E.H. and Kastning, K.M.,
(eds), Appalachian Karst Symposium, National Speleological Society,
Huntsville, Ala., p. 91-100.
Teutsch, G. and Sauter, M., 1991, Groundwater modelling in karst terranes:
scale effects, data acquisition and field validation. Hydrogeology,
Ecology, Monitoring, and Management of Ground Water in Karst Terranes
Conference (3rd, Nashville, Tenn.), Proceedings. National Ground Water
Association, Dublin, Ohio, [in this volume]
White, W.B., 1988, Geomorpholoqy and Hydrology of Carbonate Terrains. Oxford:
Oxford University Press, 464 p.
Biographical Sketch
Paul Rubin is a Research Scientist in the Environmental Sciences Division
of the Oak Ridge National Laboratory in Tennessee where he is involved in
characterizing groundwater flow. Mr. Rubin served as a hydrogeologist for the
Environmental Protection Bureau of the NYS Attorney General's Office for 8.5
years prior to his present position. He received his Master's degree in
Geology from SUNY New Paltz. He has taken an active role in research and
contaminant investigations, as well as land-use planning issues, in karst
terranes.
Paul A. Rubin
Bldg. 1503, MS: 6352
ORNL, P.O. Box 2008
Oak Ridge, TN 37831-6352
(615) 576-3476
792
-------
Land Use Planning And Watershed Protection In Karst Terranes
Paul A. Rubin
1. You mentioned chlorinating septic tank outflow. While this should take
care of fecal coliforms, I don't see any impact of chlorination on the
problem of affecting nutrient level and nutrient type in cave
ecosystems. How would you deal with the elevated organic carbon and
phosphate levels that are produced by septic tanks?
Chlorinating septic waste, as is currently being conducted by Stewart's
in the Mill Pond groundwater basin, will reduce viral and bacteriological
problems, but not eliminate the problems associated with elevated
nitrogen and phosphorus levels. Thus, practical resolution of the
problem of excessive nutrient loading and eutrophication lies in 1)
sufficient dilution of waste accompanied by adequate stream residence
time, temperature, and reaeration; 2) groundwater basin characterization
leading to responsible land-use planning; and 3) groundwater basin
management (e.g., no sinkhole dumping, removal of existing pollution
point-sources); and/or 4) waste collection and treatment prior to
discharge. Cave ecosystems are particularly vulnerable to contaminant
inputs since they are often in the zone of degradation immediately
downstream of pollution sources. Elevated organic carbon and phosphate
levels, as well as fecal coliform, will initially lead to a reduction in
dissolved oxygen used in satisfying the biochemical oxygen demand (BOD) in
streams. Farther downstream of the contaminant source, bacteria and
fungi thrive on the decomposition of organics, thus increasing ammonia
nitrogen and decreasing the BOD (Hammer, 1975). The resulting increase
in inorganic nutrients downstream of the zones of degradation and
active decomposition can result in an unnaturally nutrient-rich
environment. An increase in inorganic compounds, such as nitrate,
phosphates, and dissolved salts is common prior to the clear water zone
(Hammer, 1975). In severely contaminated streams, the clear water zone
may be far downstream of areas degraded by contaminants.
2. In light of your statements regarding the importance of evaluating
carbonate hydrology by using "karst" thinking and techniques, are the
methods you discussed being used to evaluate environmental problems on
the Oak Ridge Reservation? If not, should they?
It is recognized that karst dissolution conduits may be a significant
factor in groundwater flow in portions of the Oak Ridge Reservation
and surrounding area. Figure 2 and the Clarksville example discussed in
the text, are illustrative of the type of "karst" thinking and techniques
required to evaluate environmental problems on the Oak Ridge Reservation.
However, field work and characterization remains to be conducted on the
carbonate flow systems at the three Oak Ridge Reservation plant sites and
their catchment basins. While some isolated tracer studies have been
conducted, a more detailed and systematic approach to defining the nature
and extent of the mature karst systems present is necessary. In one such
example, the author and groundwater program managers at the Reservation
are currently spearheading a groundwater characterization program which
emphasizes the karst flow dynamics proximal to the K-25 plant site. The
methods which will be employed are consistent with the methods discussed
in this paper, and will employ sophisticated characterization techniques,
including geophysics, continuous monitoring of hydrographic response in
conduit wells, and tracer tests.
793
-------