Journal of Hydrology 515 (2014) 223–236

Contents lists available at ScienceDirect

Journal of Hydrology
journal homepage: www.elsevier.com/locate/jhydrol

Characterization of the porosity distribution in the upper part

of the karst Biscayne aquifer using common offset ground
penetrating radar, Everglades National Park, Florida
Gregory J. Mount a,⇑, Xavier Comas a, Kevin J. Cunningham b
a
Department of Geosciences, Florida Atlantic University, Davie, FL 33314, United States
b
U.S. Geological Survey, Davie, FL 33314, United States

a r t i c l e i n f o s u m m a r y

Article history: The karst Biscayne aquifer is characterized by a heterogeneous spatial arrangement of porosity and
Received 19 February 2014 hydraulic conductivity, making conceptualization difficult. The Biscayne aquifer is the primary source of
Received in revised form 14 April 2014 drinking water for millions of people in south Florida; thus, information concerning the distribution of
Accepted 19 April 2014
karst features that concentrate the groundwater flow and affect contaminant transport is critical. The prin-
Available online 28 April 2014
This manuscript was handled by Corrado
cipal purpose of the study was to investigate the ability of two-dimensional ground penetrating radar
Corradini, Editor-in-Chief, with the (GPR) to rapidly characterize porosity variability in the karst Biscayne aquifer in south Florida. An 800-
assistance of Juan V. Giraldez, Associate m-long GPR transect of a previously investigated area at the Long Pine Key Nature Trail in Everglades
Editor National Park, collected in fast acquisition common offset mode, shows hundreds of diffraction hyperbolae.
The distribution of diffraction hyperbolae was used to estimate electromagnetic (EM) wave velocity at each
Keywords: diffraction location and to assess both horizontal and vertical changes in velocity within the transect. A
Ground penetrating radar petrophysical model (complex refractive index model or CRIM) was used to estimate total bulk porosity.
Karst aquifers A set of common midpoint surveys at selected locations distributed along the common-offset transect also
Florida were collected for comparison with the common offsets and were used to constrain one-dimensional (1-D)
Porosity distributions of porosity with depth. Porosity values for the saturated Miami Limestone ranged between
CRIM petrophysical model
25% and 41% for common offset GPR surveys, and between 23% and 39% for common midpoint GPR surveys.
Laboratory measurements of porosity in five whole-core samples from the saturated part of the aquifer in
the study area ranged between 7.1% and 41.8%. GPR estimates of porosity were found to be valid only under
saturated conditions; other limitations are related to the vertical resolution of the GPR signal and the vol-
ume of the material considered by the measurement methodology. Overall, good correspondence between
GPR estimates and the direct porosity values from the whole-core samples confirms the ability of GPR com-
mon offset surveys to provide rapid characterization of porosity variability in the Biscayne aquifer.
The common offset survey method has several advantages: (1) improved time efficiency in comparison
to other GPR acquisition modes such as common midpoints; and (2) enhanced lateral continuity of porosity
estimates, particularly when compared to porosity measurements on 1-D samples such as rock cores. The
results also support the presence of areas of low EM wave velocity or high porosity under saturated con-
ditions, causing velocity pull-down areas and apparent sag features in the reflection record. This study
shows that GPR can be a useful tool for improving understanding of the petrophysical properties of highly
heterogeneous systems such as karst aquifers, and thus may assist with the development of more accurate
groundwater flow models, such as those used for restoration efforts in the Everglades.
Ó 2014 Elsevier B.V. All rights reserved.

1. Introduction
The Biscayne aquifer provides water for approximately 6
million people in Miami-Dade, Broward, and southern Palm Beach
Counties (McPherson et al., 2000). Increases in the population of
⇑ Corresponding author. Tel.: +1 954 236 1569.
E-mail address: Gmount1@fau.edu (G.J. Mount).
Florida since 1970 average between 280,000 and 320,000 persons
per year; the population is projected to increase by 2.5 million dur-
ing the next 25 years (Murley et al., 2006). Population growth will
continue to deplete the supply of groundwater in south Florida,
where the Biscayne aquifer is a sole source aquifer dependent on
surficial recharge from the Everglades and the Lake Okeechobee
drainage basin (Fish and Stewart, 1990). In addition, restoration

http://dx.doi.org/10.1016/j.jhydrol.2014.04.048
0022-1694/Ó 2014 Elsevier B.V. All rights reserved.
224 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236

Fig. 1. (a) Map of Florida showing the regional area in relation to the state, (b) map of the south Florida regional area showing the areal extent of the Biscayne aquifer
(crosshatched area and label) (U.S. Geological Survey, 2003), Miami Limestone (solid black outline and label) modified from Dicken et al. (2005) and the boundary of
Everglades National Park (solid brown outline) modified from Malget (2002) in relation to the south Florida regional area, (c) aerial photograph of the Long Pine Key Nature
Trail study area (http://www.labins.org) showing the location of the field site, including USGS test corehole G-3740 and test borehole G-3762 (symbolized by red crosses),
Line 1 and Line 2 common offset survey transects, and common midpoints CMP1 through 9 (symbolized by black filled circles). (For interpretation of the references to colour
in this figure legend, the reader is referred to the web version of this article.)

of the Everglades may markedly affect the hydrology of this critical
natural resource.
The Biscayne aquifer has a wedge-shaped volume underlying an
area of approximately 7770 km2, and reaching maximum depths of
61 m near the eastern coast of south Florida (Klein and Sherwood,
1961; Light and Dineen, 1994) (Fig. 1a and b). The Biscayne aquifer
is mainly composed of two lithostratigraphic formations dominated
by eogenetic karst limestone: the Miami Limestone and the Fort
Thompson Formation (Perkins, 1977; Halley and Evans, 1983;
Vacher and Mylroie, 2002). The Miami Limestone consists of two
high-frequency cycles (HFC 5e and HFC 4) that are assigned to Mar-
ine Isotope Stage (MIS) 5e and MIS 7, respectively (Cunningham
et al., 2006a,b, 2009). The base of the underlying Fort Thompson
Formation is at or near the bottom of the Biscayne aquifer through-
out much of its extent. In Miami-Dade County (Fig. 1b), the Fort
Thompson Formation consists of multiple high-frequency cycles
 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236
that are assigned to MIS 9, MIS 11, and possibly MIS 13
(Cunningham et al., 2006a). The vertical succession of these HFCs
has created a complex, heterogeneous and anisotropic distribution
of porosity with many karst features (Cunningham and Florea,
2009), including vuggy pore space, resulting in one of the most per-
meable carbonate aquifers in the world (Cunningham, 2004).
The Biscayne aquifer is conceptualized as a dual-porosity sys-
tem consisting of mainly touching-vug megaporosity, as defined
by Lucia (2007), and matrix porosity. Touching-vug flow pathways
have been observed as burrow-related megaporosity, bedding
planes, cavernous vugs, vertical solution pipes, and solution-
enhanced fractures (Cunningham et al., 2004a,b, 2009). Stratiform
layers characterized by touching-vug megaporosity related to bur-
rowing form groundwater flow pathways that commonly range in
thickness from the centimeter to meter scale and can have mea-
sured megaporosity values as high as 78% (Cunningham and
Sukop, 2011). Much of the stratiform touching-vug megaporosity
found in limestone of the Biscayne aquifer is related to the trace
fossil Ophiomorpha, where it was observed by Cunningham et al.
(2006a,b, 2009). In addition to stratiform zones characterized by
centimeter-scale touching-vug megaporosity, the Biscayne aquifer
also presents other karst features larger than centimeter scale
including small caves up to 120 m long with vertical extents of
4.5 m (Cressler, 1993; Cunningham and Florea, 2009), small sink-
holes, and rock pinnacles (Parker et al., 1955). The combination
of these karstic elements creates a heterogeneous mixture of high
megaporosity and high hydraulic conductivity areas in the subsur-
face that are difficult to characterize at a large scale.
A substantial amount of hydrogeologic research has been con-
ducted during the last 60 years on the Biscayne aquifer and the car-
bonate rocks that compose it. Studies include water quality and
aquifer characterizations (Parker et al., 1955; Klein and Hull,
1978; Schmoker and Hester, 1986; Fish and Stewart, 1990;
McPherson et al., 2000), aquifer performance tests (Parker et al.,
1955), digital borehole imaging (Cunningham, 2004; Cunningham
et al., 2004a; Manda and Gross, 2006), and geophysical well log-
ging and tracer studies (Cunningham et al., 2004b, 2009; Renken
et al., 2005; Harvey et al., 2008; Renken et al., 2008; Shapiro
et al., 2008). Other studies also have used high resolution X-ray
computed tomography methods in small (less than 1 m diameter)
samples of Miami Limestone and Fort Thompson Formation to
characterize pore distribution in the laboratory (Cunningham
et al., 2009). Near-surface geophysical methods, specifically GPR,
have proven to be effective tools for investigating karst features
such as cavities and collapse structures (Ballard, 1984) or deposi-
tional features (Barr, 1993; Kruse et al., 2006). GPR has been used
to estimate the porosity of materials in the subsurface when cou-
pled with a petrophysical model (Lai et al., 2006; Turesson, 2006;
Bradford et al., 2009; Harbi and McMechan, 2011; Dafflon and
Barrash, 2012). In addition, GPR has been used for characterization
of shallow stratigraphy, and fractures in limestone (Dominic et al.,
1995; Kruse et al., 2000; Grasmueck et al., 2004).
Several studies have used GPR to investigate the carbonates of
the Biscayne aquifer. Cunningham (2004) applied GPR to delineate
a high-frequency carbonate cyclostratigraphy and the distribution
of megaporosity in the Miami Limestone and Fort Thompson For-
mation, and demonstrated a relationship between spatial distribu-
tions of porosity and reflection amplitude values. Grasmueck et al.
(2004) utilized three-dimensional (3-D) survey techniques to
examine depositional structures within the Miami Limestone.
Truss et al. (2007) conducted time-lapse GPR measurements to
investigate the geometry of sand-filled dissolution structures and
surface water infiltration, whereas Neal et al. (2008) conducted a
3-D GPR survey of Miami Limestone to investigate sedimentary
architecture, and lithostratigraphic facies at the decimeter scale.
In summary, previous GPR investigations of the Miami Limestone
225
and Fort Thompson Formation have focused on stratigraphic stud-
ies, in conjunction with laboratory methods, to investigate porosity
and hydraulic conductivity, but no work, to our knowledge, has
specifically investigated porosity variability using diffraction
hyperbolae in GPR common offset surveys.
The purpose of this investigation is to assess the ability of GPR
to conceptualize the spatial distribution of the porosity in the
Miami Limestone and Fort Thompson Formation using rapid acqui-
sition common offset measurements. This approach is important
since traditional methods such as borehole techniques, well log-
ging and other GPR techniques such as common midpoint surveys,
tend to be time consuming. By increasing surveying efficiency, lar-
ger areas can be investigated to better delineate the heterogeneous
distribution of high porosity areas in the Biscayne aquifer and
identify potential zones of concentrated groundwater flow. The
approach is possible due to the unique distribution of megaporous
karst features in the Biscayne aquifer that act as point reflectors for
the electromagnetic (EM) wave and result in the presence of
hyperbolic diffractions in the GPR record that can be used for
developing two-dimensional (2-D) models of EM wave velocity.
By applying a petrophysical model (i.e. the CRIM model, explained
below) changes in EM wave velocity below the water table can be
expressed as changes in porosity. Estimates of porosity are then
compared to: (1) common midpoint measurements, and (2)
laboratory measurements of porosity using rock core samples pre-
viously acquired in the study area (Cunningham, 2004). Further-
more, Cunningham (2004) identified an anomalous structural
‘‘sag’’ tens of meters wide in a GPR profile that was attributed to
a velocity pulldown. The study also is designed to investigate
whether common offset GPR can be used to: (1) elucidate the ori-
gin of the anomalous sag structure; and (2) determine whether
similar sag features can be identified elsewhere. Results of this
study demonstrate the potential of GPR to quickly characterize
subsurface heterogeneities of the Biscayne aquifer related to con-
centrated groundwater flow and potential contaminant transport,
and the approach also can be applied to characterize areas of high
permeability in other similar shallow karstic systems.
2. Field site
The Everglades National Park protects nearly one-fifth of the
remaining Everglades land area, and includes a variety of natural fea-
tures and ecosystems that are unique to the south Florida area
(Fig. 1b). The study area is located in the southeastern part of the Park
along the Long Pine Key Nature Trail (Fig. 1c), and contains approx-
imately 20,000 hectares of pine rockland that is well-drained terrain
approximately 2 m above sea level. Mean watertable elevations fluc-
tuate from about 1.0–1.7 m above sea level during the wet season
(June through September), and approximately 0.0–1.4 m above sea
level during the dry season (October through May) (Bolster et al.,
2001). This study site was chosen following a previous investigation
along the Long Pine Key Nature Trail by Cunningham (2004), which
acquired a GPR profile, drilled two test coreholes and one rotary
roller bit test borehole, and collected geophysical logs in the test
boreholes to a maximum depth of approximately 13 m. In this study,
data obtained from the test corehole (G-3740) and test borehole (G-
3762) (Fig. 1c), are used to constrain the indirect GPR measures for
the upper 7 m of the Biscayne aquifer investigated here.
3. Methods
3.1. Ground penetrating radar (GPR)
GPR is a geophysical technique that typically uses high fre-
quency (10 MHz to 2 GHz) EM waves emitted from a transmitting
226 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236

antenna that penetrate the subsurface and return to the receiver as
a succession of reflections produced at certain interfaces. These
interfaces represent contrasts in physical properties in the rock
or sediment and typically correspond with stratigraphic or struc-
tural boundaries. The velocity of propagation of the EM waves is
dependent on the relative dielectric permittivity (er), a geophysical
property related to the ability of a material to store and release
electromagnetic energy in the form of an electric charge. This
capacity is strongly influenced by the presence of water; for exam-
ple, dielectric permittivity ranges from 1 in air to 88 in freshwater
(at 0 °C) (Buchner et al., 1999).
GPR data were acquired with a MALA ProEx™ system coupled
with 100 MHz unshielded antennas using two different surveying
methods: common midpoint and common offset (Fig. 2). The com-
mon midpoint method is widely used for vertical velocity estima-
tion (Greaves et al., 1996; Jacob and Hermance, 2004; Annan,
2009). In this study, the common midpoint approach used a GPR
transmitter (Tx) and receiver (Rx) that were increasingly separated
by a stepped distance of 0.1 m (Fig. 2a). The incremental increase
in separation of the Tx and Rx results in an increase in the two-
way travel time of the EM wave transmitting downward to a par-
ticular reflector (Annan and Davis, 1976) (Fig. 2b). Common offset
surveys use a transmitting and receiving antenna held at a fixed
distance from each other that move incrementally along a survey
line (Fig. 2c) (Dominic et al., 1995; Kruse et al., 2000; Grasmueck
et al., 2004; Neal, 2004). Common offset profiles are the most com-
mon display of GPR data because they resemble a geologic cross
section where the depth scale is expressed as a two-way travel
time (Fig. 2d). The unique presence of numerous hyperbolic
diffractions in the common offsets in this study was used to inves-
tigate the changes in velocity within the subsurface. Diffraction
hyperbolae originate from isolated point reflectors (i.e. areas repre-
senting a marked contrast in the dielectric properties as compared
to the surrounding host material). Examples of physical features
that could cause diffraction hyperbolas are anomalous cobbles
and boulders in an otherwise well sorted matrix (Busby and
Merritt, 1999), or buried man-made objects such as wires or
pipes (Neal, 2004). In karst environments, diffraction hyperbolas
can be caused by air-filled or water-filled voids in a solid matrix
(Bristow, 2009). Presumably, the presence of hyperbolic diffrac-
tions in the GPR data is due to the numerous karstic features in
the Miami Limestone that manifest as a series of air- and water-
filled voids, and these diffractions can be used to quickly character-
ize the spatial variability of porosity below the water table in the
Miami Limestone.
3.2. Velocity analysis
Processing of both GPR common offset and common midpoint
data was simplified to three components: (1) a dewow filter to
remove the low frequency noise and DC signal component; (2) a
time zero correction to remove the lag between the triggering of
the signal and the recording of the first arrival; and (3) an exponen-
tial gain filter to amplify the signal with depth (e.g., gain value of 2
and exponent of 0.2).
A common midpoint velocity analysis was conducted using
Reflex W™ software, and consisted of manually picking the arrival
time for the first negative break of the EM wave for each trace

Fig. 2. (a) Diagram of common midpoint (CMP) method of data acquisition by moving transmitting (Tx) and receiving (Rx) antennas at regular intervals or stations; (b)
example of CMP survey showing a series of hyperbolic-shaped reflectors resulting from the increased two-way travel time of EM wave from Tx to reflectors and then back to
Rx as antennas are separated; (c) diagram depicting the common offset (CO) method and showing the synchronous movement of Tx and Rx antennas at three different
acquisition locations or stations; and (d) example of CO data profile. Note prominent hummocky configurations (between 110 and 150 ns) and contorted reflections (between
130 and 135 ns).
 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236
along a particular reflected wavelet event in the profile. Reflection
events were chosen to match known sedimentary interfaces in the
Miami Limestone and Fort Thompson Formation. Reflection events
were used to calculate the root-mean-square velocity of the EM
waves from the ground surface to each reflector using a linear
regression routine. Error estimation in the common midpoint data
was calculated using the procedure outlined in Parsekian et al.
(2010) and based on methods in Jacob and Hermance (2004),
which calculate the standard deviation of the model parameters
for velocity and depth of each reflector. The 95% confidence limits
of those values were then determined using a Student’s t-test and
converted from confidence limits to interval confidence limits by
propagation of errors through the Dix equation (Dix, 1955). Error
values are provided on all 1-D model plots.
A velocity analysis for all common offset data also was com-
pleted using Reflex W™ to estimate diffraction velocities by fitting
a user-defined hyperbolic function to the diffraction geometry in
the radar profile. To further aid comparison, the diffraction analyses
were conducted at geographic locations where common midpoints
also were collected. Measurement error for the hyperbolic diffrac-
tions was calculated with a routine developed by ReflexW™ soft-
ware, (Karl Sandmeier, Sandmeier Scientific Software, personal
communication), and consisted of picking the lowest and highest
velocity possible to fit each particular hyperbola. An average veloc-
ity is assumed for each hyperbola, but the minimum and maximum
velocities are considered for estimating the error in the resulting
values. Estimated errors reached were less than 5.0% in all cases.
Velocities estimated from both GPR common offsets and com-
mon midpoints represent bulk values that account for the traveling
distance of the EM wave from the surface to each particular reflec-
tor and then to the receiver. The Dix equation (Dix, 1955) was
applied to estimate vertical changes in velocity within the column
and allows for the estimation of EM wave interval velocity
between two reflectors in a vertical sequence:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
v 2n tn  v 2n1 tn1
v i;n ¼ ð1Þ
t n  t n1
where v is the EM wave velocity, t is the two-way travel time, and
subscript n refers to the reflector number in the sequence of
reflectors.
3.3. Rock core data
GPR measurements were constrained by applying known
depths to geologic contacts. Profiles were collected adjacent to a
test corehole (G-3740, Figs. 1c and 3a) and a test borehole (G-
3762, Figs. 1c and 3a), both drilled as part of a previous study
(Cunningham et al., 2004a). Depths to the upper cyclostatigraphic
boundaries of HFC 4 and HFC 3b were used to constrain depth to
contacts and to estimate EM wave velocity to the depth of the
reflector in the GPR data (Fig. 3b and d).
Cunningham (2004) used whole-core helium-porosity methods
to measure porosity values for five whole-core samples acquired
from test corehole G-3740. Measured porosities ranged between
7.1% and 41.8% (Fig. 3b). Nomenclature for pore size classes is
based on Choquett and Pray (1970, p.247), e.g. ‘‘megapore’’ is used
for ‘‘equant to equant-elongate pores whose average diameter is
larger than 4 mm, and for tubular or platy pores whose average
cross-sectional diameter or thickness is larger than 4 mm.’’
4. Petrophysical modeling
Limestone is assumed to be a low-loss material, such that electri-
cal conductivity and magnetic properties are negligible, and the bulk
dielectric permittivity can be defined by the following equation:
227
erðbÞ ¼ ðc=v Þ2 ð2Þ
where c is the constant for the velocity of an EM wave in a vacuum
(0.3 m ns1), and v is the bulk EM velocity (m ns1) estimated from
the GPR data. The complex refractive index model (CRIM) (Birchak
et al., 1974) is used herein to convert the bulk relative dielectric
permittivity (dimensionless), into porosity. The CRIM is a three-
phase dielectric mixing model that expresses the bulk relative
dielectric permittivity (er(b)) as:
earðbÞ ¼ /Sw earðwÞ þ ð1  /ÞearðsÞ þ /ð1  Sw ÞearðaÞ ð3Þ
where er(s) is the relative dielectric permittivity of limestone or the
solid phase, er(a) and er(w) are the relative dielectric permittivity of
air (1) and water (79), respectively, based on an average water tem-
perature in the Biscayne aquifer of 23 °C (Buchner et al., 1999;
Renken et al., 2008), / is the porosity, and Sw is the water satura-
tion (values between 0 and 1, with 1 representing fully saturated
conditions). The exponent a represents the geometrical factor,
thereby accounting for the orientation of the electrical field with
respect to the geometry of the limestone, with values between
1 and 1.
There are two main sources of uncertainty in the application of
the CRIM to EM wave velocity data, the dielectric permittivity of
limestone and the exponent a. The dielectric permittivity of lime-
stone (i.e. solid phase), was not measured directly in the study and
therefore a range of values reported in the literature was assumed
(4–11) (Cassidy, 2009; Reynolds, 2011). Porosity is dependent on
EM wave velocity and porosity variability is typically minimized
at lower velocities throughout a range of solid dielectric permittiv-
ity of limestone ranging from 4 to 11 (Cassidy, 2009; Reynolds,
2011). A median value of 7.5 for the dielectric permittivity of lime-
stone is assumed for purposes of this investigation. The two
extreme values for the exponent a is another source of uncer-
tainty, whereas, 1 and 1, are opposite ends of the spectrum;
1 is related to the harmonic average or arrangement of dielectric
components in series, and 1 is the arithmetic average or arrange-
ment of dielectric components in parallel (Chan and Knight, 2001).
A value of 0.5 has been found applicable to most geologic materi-
als in the 100–1000 MHz range encountered by most GPR and
time domain reflectometry (TDR) studies (Birchak et al., 1974;
Roth et al., 1990; Knoll, 1996; West et al., 2003) and it is used
herein.
One additional source of uncertainty is related to the position of
the water table below the ground surface and the fact that the EM
wave travels through a section of unsaturated material within the
limestone column. This uncertainty was minimized by accounting
for the depth of the water table measured directly at wells G-3740
and G-3762 during data collection, and by assuming an average
velocity value for the unsaturated part of the column. The average
unsaturated EM wave velocity for the limestone column was calcu-
lated from the ground wave in the nine common midpoint surveys.
The ground wave is not a reflection, but a direct wave with a linear
relationship between distance between antennas and travel time.
The slope of this relationship represents an estimate of the EM
wave velocity for the topmost portion of the column and is
assumed here as representative of the velocity for the unsaturated
portion of the column, however a water saturation term (not avail-
able here), would be needed in order to properly quantify porosity
using the CRIM model along the unsaturated portion. Travel time
corresponding to the unsaturated part of the column using the
average ground wave velocity, was then isolated and subtracted
from the total travel time using the Dix equation (Eq. (1)), since
porosity estimates account only for fully saturated conditions.
After accounting for all corrections and using the interval velocity
returned from the Dix equation, porosity using the CRIM under
fully saturated conditions is measured as:
228 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236

Fig. 3. (a) Common offset profile along Line 1 at the Long Pine Key Nature Trail study area (Fig. 1), showing the location of the USGS test corehole G-3740 and test borehole
G-3762 (vertical gray bars), and reflectors R1, R2 and R3 corresponding to reflection surfaces of the HFC 4 top, and inter-HFC reflectors of HFC 4, and HFC 3b, respectively. Note
the velocity sag area between 110.0 and 160.0 m distance along the transect characterized by a hummocky clinoform with a centrally located contorted reflector outlined by
black rectangles; (b) digital optical image log of borehole wall in the USGS test corehole G-3740 with arrows marking the R1, R2 and R3 reflections and locations (*) of five
whole-core laboratory helium-porosity values reported by Cunningham (2004); (c) subset of the Common Offset profile (shown in Fig. 3a) showing two-way travel time (ns)
and the numerous hyperbolic diffractions highlighted (with associated bulk electromagnetic velocities in m ns1) found within the profile; (d) digital optical image log of
borehole wall in the USGS test borehole G-3762 with arrows marking the R1, R2 and R3 reflection interfaces of the top of HFC 4, and inter-HFC 4 and 3b reflectors
respectively; and e) subset of Common Offset profile (3a). Profile subset is centered along test borehole G-3762 and shows reflectors R1, R2 and R3, and a close-up of the
velocity sag area. In addition, a sample of diffraction hyperbolas along with their corresponding bulk electromagnetic wave velocities are also highlighted.

a
ððc=v Þ2 Þ  earðsÞ 5. Experimental design
/¼ ð4Þ
ea
rðwÞ  earðsÞ
Two GPR common offset transects (Line 1 and Line 2) were col-
(see Eqs. (2) and (3) for definition of terms). lected at the Long Pine Key Nature Trail study area (see Fig. 1c).
 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236
Line 1 coincides with a GPR profile collected by Cunningham
(2004), and was approximately 2 m from test corehole G-3740
and about 5 m from borehole G-3762. The Line 1 transect was col-
lected in 0.1 m increments in order to maximize lateral resolution,
over a total length of 200 m. Line 2 extended Line 1 westward
approximately 600 m and was collected in 0.2 m increments in
order to maximize time efficiency. All data traces were stacked 32
times to increase the signal to noise ratio, and the sampling time
window was set to 400 ns, resulting in a penetration depth of
approximately 6.5 m. A total of nine common midpoints (CMPs)
were collected at approximately equally distributed locations along
Line 1 and Line 2 (see Fig. 1c) to target potential areas of dissolution
versus non-dissolution as inferred from the common offset surveys.
CMP 1 and 2 were collected at 53 m and 132 m, respectively, along
the common offset and coincided with the locations of test corehole
G-3740 and test borehole G-3762. The remaining seven common
midpoints were collected at a distance along the survey transect
as follows: CMP 3 at 210 m, CMP 4 at 330 m, CMP 5 at 380 m,
CMP 6 at 420 m, CMP 7 at 510 m, CMP 8 at 570 m, and CMP 9 at
680 m. This spacing was designed to investigate vertical changes
in velocity and lateral changes in geology based on variations
within the reflection record seen in the common offset data set.
6. Results
6.1. Ground penetrating radar common offset data
At the Long Pine Key Nature Trail Site a common offset transect
was collected to assess the lateral continuity of GPR reflectors and
their correlation to known HFCs in the Miami Limestone (Fig. 3).
Fig. 3a corresponds to common offset Line 1, and shows three
prominent reflectors indicated with arrows R1 thru R3. Based on
the results from test corehole G-3740 (Fig. 3b) and test borehole
G-3762 (Fig. 3d), R1 corresponds to the top of the HFC 4, whereas
R2 and R3 correspond to inter-HFC reflectors for HFC 4 and HFC 3b
respectively. Two gray vertical bars, at 32 m and 132 m distance
along the transect mark the location of the test corehole G-3740
and test borehole G-3762, respectively (Fig. 3c and e). Note that
Fig. 2d provides a clean copy of Fig. 3e radargram. Notably, the
reflection surfaces are neither horizontal nor at the same elevation
across the entire survey. A slight undulation is indicated by differ-
ences in two-way travel time to the reflection surfaces, ranging
from 1 to 7 ns. In addition, there is a reflection sag between
115 m and 155 m distance along the transect, resulting in a down-
ward dip for reflectors 2 and 3 (R2 and R3) along the edges of the
sag and a convex area with a contorted reflection in the center of
the sag (Fig. 3a). To aid in clarity, two contrasting subsets of Line
1 are presented. Fig. 3c is an enlarged view of Line 1 between 28
and 64 m distance along the line showing an area away from the
sag area, whereas Fig. 3e is an enlarged view between 110 and
160 m distance along the line within the sag area. Both insets show
EM wave velocity estimates from numerous diffraction hyperbolas
and result in a contrasting set of values that range between 0.065
to 0.099 m ns1 (Fig. 3c) and 0.046 to 0.102 m ns1 (Fig. 3e). For
clarity, only a small sample of the diffractions present in the data
set is shown. The range of EM wave velocities estimated from
the hyperbolic diffractions is lower at test borehole G-3762
(Fig. 3b and c) than at test corehole G-3740 (Fig. 3d and e). This
change in EM wave velocity has implications for both the porosity
modeling and the depth of the R2 and R3 reflectors across the
reflection sag, as explained below.
6.2. Ground penetrating radar common midpoint data
A representative example of common midpoint GPR data from
the Long Pine Key Nature Trail site is presented in Fig. 4 to
229
illustrate the correspondence between GPR reflectors, velocity esti-
mates, and high frequency cycles at two contrasting locations
where lithology is well constrained with direct sampling. CMP 1
located at test corehole G-3740 (Fig. 4a) and CMP 2 located at test
borehole G-3762 (Fig. 4d) show an airwave, a linear ground wave,
and a series of reflectors where R1 corresponds to the top of HFC 4;
R2 and R3 correspond to inter-HFC reflectors for HFC 4 and HFC 3b,
respectively. Estimated bulk velocities for each reflector also are
shown for each reflector (Fig. 4a and d).
The 1-D velocity models (Fig. 4b and e) show interval velocities
(after Dix equation) and porosity estimates (after application of the
CRIM model). The 1-D models also show error bars corresponding
to the estimated error in both depth and velocity as noted in the
velocity analysis methods. Due to uncertainty in the saturation of
the limestone above the water table, the portion of the vertical col-
umn corresponding to unsaturated material in the 1-D models is
shaded in Fig. 4b and e, reflecting the inability of the CRIM model
to estimate porosity when saturation is unknown. Despite this lim-
itation, an average ground wave velocity of 0.082 m ns1 (repre-
sentative of the EM wave traveling directly through the ground)
was retrieved from CMP analysis and was used to estimate poros-
ity under fully saturated conditions. The 1-D model for CMP 1
(Fig. 4b) shows an interval between 1.8 and 2.6 m with a velocity
of 0.069 m ns1, followed by a slight decrease in velocity in the
interval between 2.8 and 4.1 m (0.067 m ns1) and a slight increase
between 4.1 and 5.2 m (0.073 m ns1). In comparison, CMP 2
shows a variation in both the travel time to the three reflectors,
increased by 6.0, 21.0 and 22.0 ns respectively, and interval veloc-
ities of 0.066, 0.068, and 0.059 m ns1 for the intervals of 1.8 to
3.25 m, 3.25 to 4.5 m and 4.5 to 6.0 m respectively. The area of
low velocity described in CMP 2 below 4.5 m depth (indicated with
an arrow in Fig. 4d) coincides with the sag area previously
described (Fig. 3a).
6.3. Comparison of porosity estimates
Interval velocities were calculated using travel times as mea-
sured from the land surface datum to the reflectors R1 and R3 near
test corehole G-3740 and test borehole G-3762 using the Dix equa-
tion, after which porosity was estimated for both the GPR common
midpoint and common offset data using the CRIM equation (Tables
1 and 2). Estimates for velocity and porosity from both common
midpoint and common offset acquisition techniques are consistent
for each of the two profiles. However, when the two sample loca-
tions are compared, marked differences between the test corehole
and test borehole locations are shown due to the placement of the
wells in areas of different porosity values. For example, estimated
porosity from the common midpoint along test corehole G-3740
increases slightly from 27% to 29% from the top to the middle of
HFC 4, and then decreases to 23% at HCF 3b (Table 1). Patterns of
estimated porosity for the common offset survey (Table 1) are sim-
ilar, with an overall decrease from 30% at HFC 4 to 25% at HFC 3b.
An initial decrease in porosity along HFC 4 was followed by a sharp
increase at HFC 3b (from 28% to 39% from the common midpoint;
and 32% to 41% from the common offset) along test borehole G-
3762 (Table 2). Direct porosity measurements (only available for
test corehole G-3740) are similar to GPR estimates and also indi-
cate a slight increase in porosity from 25.5% to 32.5% as depth
increases along HFC 4, and a much sharper decrease to 7.1% at
HFC 3b (Fig. 3b).
6.4. Porosity estimation at larger scales
The estimates between the common offset and common mid-
point techniques corresponded well; therefore, the common offset
survey was extended and additional CMPs were collected. These
230 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236

Fig. 4. Examples of common midpoint and common offset velocity data outside the velocity sag area: (a) the radargram for CMP1 at corehole G-3740, with the air wave,
ground wave, and reflectors R1–R3 shown with bulk velocity estimates (before applying the Dix equation); (b) 1-D model of interval velocity (and inferred porosity after use
of the complex refractive index model) of CMP1, showing the unsaturated zone and high frequency cycle (HFC) tops; (c) 1-D model of interval velocity based on hyperbolic
diffractions from common offset (CO) data. Within the velocity sag area: (d) radargram for CMP2 at borehole G-3762, with the air wave, ground wave, and reflectors R1–R3
shown with bulk velocity estimates; (e) 1-D model of interval velocity of CMP2, showing the unsaturated zone, high frequency cycle tops and a shift to lower bulk velocity
and thus increased porosity below 4.5 m; and (f) 1-D model of interval velocity based on hyperbolic diffractions from CO data, also showing the unsaturated zone, HFC tops
and the shift to lower bulk velocity below 4.5 m.

Table 1
Velocity measurements and porosity estimates at test corehole G-3740.

Reflector Depth to reflector (m) (b) Common midpoint reflections (c) Common offset diffractions
Interval velocity (m ns1) Porosity (%) Interval velocity (m ns1) Porosity (%)
R1 2.5 0.069 27 0.067 30
(HFC 4 top) (±0.002) (±2) (±0.003) (±3)
R2 4.1 0.067 29 0.066 30
(Inter-HFC 4) (±0.005) (±4) (±0.003) (±3)
R3 5.0 0.073 23 0.071 25
(Inter-HFC 3b) (±0.009) (±8) (±0.004) (±3)

Table 2
Velocity measurements and porosity estimates at test borehole G-3762.

Reflector Depth to reflector (m) (b) Common midpoint reflections (c) Common offset diffractions
1
Interval Velocity (m ns ) Porosity (%) Interval Velocity (m ns1) Porosity (%)
R1 3.0 0.066 30 0.059 38
(HFC 4 top) (±0.007) (±8) (±0.003) (±3)
R2 4.5 0.068 28 0.064 32
(Inter-HFC 4) (±0.002) (±1) (±0.003) (±3)
R3 6.0 0.059 39 0.057 41
(Inter-HFC 3b) (±0.005) (±4) (±0.003) (±3)

measurements were used to develop estimates of porosity at a composite profile from Lines 1 and 2 are recognized as the air wave
larger scale of measurement (greater than 100 m) and are pre- and direct waves, followed by ground coupling effects that obscure
sented in Fig. 5 with an 800 m long transect. In Fig. 5a, the upper- true reflections. Below these uppermost reflectors, the profile is
most set of 3–4 horizontal reflectors within the first meter in the mostly characterized by a sequence of laterally continuous parallel
 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236 231

to subparallel reflectors located between 0 and 110 m distance
along the profile (Figs. 5a and 3a) that contrast with a sharp change
in the reflection record characterized by a series of hummocky
reflection configurations between 110 and 150 m distance along
the profile and a depth of approximately 4–6 m. Similar variations
in the reflector signature also are found at approximately: (a)
250–350 m distance along the profile and 110–180 ns (approxi-
mately 3.5–6.0 m deep); (b) 490–510 m distance along the profile
and 150–180 ns (approximately 4.5–6.0 m deep); and (c) 590–
650 m distance along the profile and 150–180 ns (approximately
4.5–6.0 m deep) (Fig. 5a).
Fitting of diffraction hyperbolae (exemplified in Fig. 3c and e)
along the profile (Fig. 5a) allowed for estimation of velocity vari-
ability and the construction of a 2-D model of velocity based on
interpolation of velocity values (Fig. 5b). Variations in bulk
velocity in the 2-D model range from 0.055 to 0.085 m ns1, corre-
sponding to porosities between 45% and 15%, respectively. Velocity
decreases with depth (Fig. 5b); isolated areas of lower porosity
occur near the bottom of the profile and are indicated in blue. An
uppermost reflection unit that approximately corresponds to the
lowest part of HFC 5e lies beneath the unsaturated zone, and is
characterized by velocities that range between 0.070 and

Fig. 5. (a) composite radargram for Line 1 and Line 2 at the Long Pine Key Nature Trail study area showing parallel to subparallel reflectors with a velocity sag consisting of a
hummocky clinoform with centrally located contorted reflection between 110 and 150 m along the transect and at 4 to 6 m estimated depth outlined by a black rectangle.
Additional black rectangles outline similar velocity sag features with hummocky configurations or attenuated reflectors. High frequency cycle (HFC) tops are labeled on the
right side of the figure, with HFC 4 and HFC 3b outlined in red and blue, respectively; (b) 2-D velocity model based on hyperbolic diffractions from the common offset data,
with corresponding porosity color scale key. Fig. also shows the location of the unsaturated zone and the locations of CMP 1 through CMP 9, and the tops of HFC 4 and HFC 3b
outlined in red and blue respectively. The 2-D model shows areas in blue that correspond to areas of velocity sag (characterized by hummocky configurations, black polygons
in Fig. 5a), or areas where large changes in the porosity are encountered. In addition, a relationship between porosity within the HFC 4 and HFC 3b can be seen as porosity
increases in the HFC 3b with increasing depth, and porosity decreases in HFC 4 near the cycle top; (c) Radargrams for CMPs 1 through 9; and (d) corresponding 1-D models of
interval velocity for each of the CMPs with estimated porosity and HFC 4 and HFC 3b tops marked. Velocity sags in the HFC 3b are generally related to lower velocities and
thus higher porosity estimates. Note that the HFC tops of CMP1 and CMP2 are from direct measures, CMP3 through 9 HFC tops are based on data from the G-3740 core. (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
232 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236
0.060 m ns1 and porosity estimates ranging between 25% and
35%.
Common midpoint data sets (Fig. 5c) were used to create 1-D
models (Fig. 5d) of velocity and the CRIM-derived porosity to
investigate the correspondence with changes of velocity along
the common offset transect. These test locations are marked with
black lines (Fig. 5b) and were chosen in order to show areas of con-
trasting porosity. Because it is only possible to estimate porosity
for saturated conditions using the CRIM model, the unsaturated
part (upper 1.8 m) of the graphs are lightly shaded to indicate that
porosity estimates are uncertain within that area (Fig. 5d). There is
an increase in porosity below 4.5 m depth in the 1-D models,
which corresponded with isolated low velocity areas in the diffrac-
tion-based 2-D model (Fig. 5b and d). These areas also correspond
with sag zones typically above the HFC 3b, and areas of attenuated
amplitudes in the reflection record along the common offset
(Fig. 5a). This pattern is most prevalent in CMP 2, 4, 5 and 7.
CMP 3 and 9 show a slight increase in porosity in the 1-D model
coincident with a small low velocity area in the 2-D model. CMP
1, 6 and 8 show a slight increase in velocity and thus a decrease
in porosity in the 1-D model that coincides with areas where no
sags on the reflections were detected.
Velocity estimates from common offsets (Fig. 5b) and common
midpoints (Fig. 5d) are inherently different: estimates from com-
mon midpoints are based on interval velocities after applying the
Dix equation, whereas estimates from diffractions in common off-
sets are based on average bulk velocities from the surface to a par-
ticular diffraction. Application of the Dix equation to velocity
estimates from hyperbolic diffractions is not possible because dif-
fractions occur throughout the subsurface and are not stacked
along a particular point as they are for common midpoints. How-
ever, in some situations there is a distribution of diffractions that
appear vertically through the data set, much like a series of reflec-
tors from a CMP, so that an approximation to average velocities
with depth along an average positioning is possible, and the Dix
equation can be applied. For example, a 1-D model of velocity after
application of the Dix equation at test corehole G-3740 and test
borehole G-3762, respectively, produces average hyperbolas
within a range of 6 m at five particular depths coincident with
those in the common midpoint data (Fig. 4e and f). The models
are compared to those from common midpoints after application
of the Dix equation (Fig. 4e and f) and indicate a strong correspon-
dence. Unfortunately, use of the approach was not possible for the
other common midpoint locations (Fig. 5c) because there were too
few diffractions with which to compare the estimates of porosities
from common offsets.
7. Discussion
Common offset ground penetrating radar is a useful technique
in the estimation of porosity in near-surface carbonates when
numerous dissolution features create hyperbolic diffractions. Com-
parisons of porosity between the common offset and common
midpoint show similar and consistent results. Although common
midpoints are the standard GPR technique for velocity estimation,
common offsets provide the ability to assess EM wave velocity var-
iability and detect sag features more quickly.
7.1. Imaging of dissolution features in the Miami Limestone and Fort
Thompson Formation using GPR
GPR profiles collected throughout the Miami Limestone and the
upper part of the Fort Thompson Formation in the study area are
characterized by the unique occurrence of: (1) a large number of
hyperbolic diffractions, (2) relatively continuous parallel to
subparallel reflectors that in some areas become hummocky con-
figurations or contorted reflections, and (3) areas with attenuated
reflection amplitudes and sag zones. In all cases, diffractions, hum-
mocky configurations, sags, and zones of attenuated reflections are
interpreted as areas of low GPR velocity and increased porosity
when compared to areas with parallel to subparallel reflectors. Dif-
fractions represent point reflectors caused by karst dissolution fea-
tures. The data indicate that zones of increased porosity modeled
by both the common offset and common midpoint techniques cor-
respond to areas composed of hummocky or contorted reflection
configurations. Both of these conditions represent EM wave veloc-
ity pull-down areas due to dense touching-vug megaporosity, as
previously proposed by Cunningham (2004), and are not due to
structural sags or collapse features.
7.1.1. Hyperbolic diffractions and discontinuous reflections and the
relation to dissolution features
GPR velocities are mostly affected by the ratio of air to water in
the medium because the two materials are at opposite ends of the
dielectric permittivity spectrum (1 in air and 88 in freshwater). In
saturated conditions, the presence of a water-filled pore space will
cause the EM wave to slow down; for example, at 0.055 m ns1
porosity is estimated at 45%. The same pore space filled with air
under unsaturated conditions will cause GPR velocities to increase.
Hyperbolic diffractions are caused by sharp changes in the physical
properties of the subsurface such as contrasts in water content due
to changes in porosity, or due to the infill of voids by either water
or air that contrast with the surrounding matrix, and result in point
reflections originating from dissolution cavities (Neal, 2004). Given
the large number of diffractions recorded (Fig. 3c and e), the pres-
ence of point reflectors is most likely due to water-filled pore
spaces that result in decreases in the EM wave velocity. Large dis-
solution features may disrupt the continuity of reflections and
therefore could result in a variety of non-parallel reflectors in the
GPR reflection record or even create areas where the amplitude
may be attenuated (Mitchum et al., 1977; Neal, 2004). The hum-
mocky and contorted reflection configurations are consistent with
reflections from non-planar surfaces, such as those found in an
eroded or highly karstified surface, or from areas where higher lev-
els of amplitude attenuation are present (Mitchum et al., 1977;
Neal, 2004; Reynolds, 2011).
7.1.2. EM wave velocity pull down and the relation to areas of
enhanced porosity
Two simple velocity estimations for reflector 2 (R2 in Fig. 3c and
e) indicate that a velocity pull down is the most plausible explana-
tion for the sag feature in Fig. 3a and e, as previously proposed by
Cunningham (2004). At R2 along test corehole G-3740 (Fig. 3b), if
the average depth is 4.1 m and the two-way travel time is 115 ns
(as confirmed from GPR data), then the average interval velocity
is 0.071 m ns1 at that well (Fig. 3a and c). This velocity is similar
to interval velocities estimated between the ground surface and R2
from both common offsets (ranging between 0.069 and
0.073 m ns1) and common midpoints (0.069 m ns1) (Figs. 3c
and 4a, respectively). If the two-way travel time is approximately
160 ns between the ground surface and R2 at the deepest part of
the sag (Fig. 3e) and the true depth of R2 is 4.1 m, the calculated
interval velocity between the ground surface and R2 is
0.051 m ns1. This estimated velocity also is similar to interval
velocity estimates from common offsets within the sag area, which
ranged from 0.046 to 0.047 m ns1. These results indicate that: (1)
the sag structure in the reflector record can be explained by con-
sidering that reflectors are being pulled down because of a low
velocity (high porosity) area but are approximately horizontal in
reality; and (2) that similar low velocities along the sag area are
 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236 233

confirmed from both common offset and common midpoint saturation is 29–31%. When partially saturated conditions are con-
surveys. sidered, porosity estimates are reasonable for two reasons: (1) esti-
mated values of porosity are well within the ranges of reported
7.2. Porosity estimation in the Miami Limestone using GPR and porosity (50.6% and 41.8%) from whole core laboratory porosity
limitations of the method measurements from G-3740 at the top and bottom of the HFC 5e,
respectively (Cunningham, 2004); and (2) previous studies investi-
GPR common offset techniques were shown to be an effective gating the water content in the capillary fringe in soil-filled
and less time-consuming method of estimating lateral changes in trenches in the Miami Limestone have shown that moisture content
EM wave velocity and thus porosity, for the Miami Limestone and ranged between 24% and 43% (Barquin-Valle et al., 2011), and are
Fort Thompson Formation in the upper saturated part of the Bis- well within the range of conditions in this study. Finally, it is impor-
cayne aquifer, which is characterized by a large number of hyper- tant to consider the uncertainties involved in some of the constants
bolic diffractions. The common midpoint technique is widely used used in the model, and particularly the solid dielectric permittivity
to investigate subsurface changes in EM wave velocity, but by com- (Fig. 6). Uncertainty is minimized for lower velocities (and thus
parison it is substantially more time consuming. As an example, high porosities), but it increases substantially at higher velocity
nine common midpoint surveys across Lines 1 and 2 were collected (or low porosity). This may in part be the reason that low porosity
in 0.10 m increments over a distance of 10–18 m, and took about 6 values seem to be poorly modeled with this GPR approach.
hours to complete by a group of three people for a total of about 18
person-hours. The common offset technique, which typically is not
7.3. Effect of scale and support volume on porosity estimation
used for the characterization of subsurface EM wave velocity, was
conducted by one person and took approximately 2 hours to com-
Previous studies have discussed the potential efficacy of point
plete along Line 1 (collected by placing the antenna with direct con-
measurement methods at different scales of investigation, and
tact on the ground every 0.1 m for 200 m), and about 1 hour along
have shown that as scales increase, point data sets may not be suf-
Line 2 (collected by keeping antennas above the ground at 0.2 m
ficient for characterization; as a result, multiple scales of measure-
trace spacing for approximately 600 m) for a total of about 3 per-
ments may be required (Ford and Williams, 1989; Smart, 1999;
son-hours. Common offsets in areas with numerous point reflec-
Worthington, 1999; Bakalowicz, 2005). Robinson et al. (2008) sta-
tors, such as the Miami Limestone, show great potential for
ted that geophysical methods may prove most useful in groundwa-
quickly estimating lateral changes in porosity distribution.
ter research compared to other methods because multiple spatial
The GPR common offset approach presented herein, however,
scale measures can be collected to fill gaps in measurements at
does have several limitations. Estimates of porosity for the satu-
intermediate scales. In karstic environments, characterizations
rated part of the subsurface (HFC 4 and HFC 3b) are consistent
are even more difficult due to the prevalence of extreme variability
between the two independent GPR methods used (Tables 1 and
in the hydrologic properties of the subsurface beyond the pore
2), and similar to helium-porosity measurements of two whole-
scale. In order to find a statistically significant means of measuring
cores acquired from the limestone of the HFC 4 in test corehole
porosity and permeability in heterogeneous media, authors have
G-3740. For example, the porosity values of the two cores are
sought to determine the representative elementary volume (REV)
25.5% and 32.5% (Fig. 3b), whereas GPR porosity estimates range
for vuggy carbonates (Nordahl and Ringrose, 2008; Vik et al.,
from 29% to 30% (Table 1). The highest GPR-derived porosity values
2013). Research has shown that as sample sizes increase, variabil-
were expected in HFC 5e because two whole-core measurements
ity in the measurement decreases and average values between
reported by Cunningham (2004) (Fig. 3b) from test corehole G-
small samples are similar to values derived from larger values
3740 have values of 41.8% and 50.6%. The average value of porosity
(Vik et al., 2013). These results suggest that large scale GPR mea-
measured on four whole core samples from test coreholes in north-
surements may represent porosity values more accurately as they
central Miami-Dade County is higher than any helium-porosity
are more likely within the range of the REV of the Miami Limestone
values measured for 65 whole-core samples acquired from rocks
than point measures collected at specific locations along a column.
of the Biscayne aquifer below HFC 5e (Cunningham et al., 2004b).
The research presented herein demonstrates the ability of GPR to
However, estimates of porosity derived from GPR measurements
in the unsaturated part of the 5e HFC are lower when compared
to previously reported whole-core measures. The use of GPR in
the unsaturated zone, corresponding here to a major part of HFC
5e, is hindered by two factors: (1) the inability to measure water
saturation and the likely gradation of saturation due to capillary
action, which makes application of the CRIM model difficult; and
(2) the presence of EM wave ground coupling in GPR profiles,
which affects the reflection record for the topmost 0.5 m. For those
reasons, the CRIM model is limited to the saturated part of HFC 5e
and is unable to estimate porosity for the unsaturated part unless a
saturation component is assumed.
Despite this limitation, the ground wave velocity values can be
used with reflections and diffractions from common midpoints and
common offsets in the upper 1.8 m to estimate EM wave velocity
above the water-table elevation. By rearranging terms in Eq. (3)
and assuming a range of values of water content in the CRIM
model, estimates of porosity in the unsaturated part of HFC 5e
can be derived. Using this approach porosity values can be Fig. 6. Plot of the Complex Refractive Index Model (CRIM) estimated using
obtained from the CMP ground wave ranging between 43.8 and saturated porosity values from this study (ranging from 0.04 to 0.09 m ns1) for
solid dielectric permittivity ranging between 4 and 11 (Cassidy, 2009; Reynolds,
52.8%, when water saturation is at 50% and 45% respectively, and 2011). Curved lines were fitted to show upper and lower bounds of porosity
from common offset diffractions closer to the surface at estimates for velocities typically found in the saturated part of the Miami
43.8% and 54.5% porosity when velocity is 0.10 m ns1 and water Limestone.
234 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236

Fig. 7. Diagram of the changes in support volume related to measuring technique: (a) <1 m3 volume measures in geologic cores, with sample location (X) of porosity
measures (/); (b) support volume of 2–4 m3 in GPR common offset surveys; support volume is illustrated with the light gray spheroid labeled Fresnel volume 1, and the 2-D
foot print of that volume labeled Fresnel zone 1. Note that multiple sample locations (X) of porosity measures (/) are averaged within the Fresnel volume; and c) support
volume of 2–40 m3 in GPR common midpoint surveys (shaded volume illustrations are omitted for clarity). Note increased occurrence of sample locations (X) of porosity
measures (/) within support volume.

estimate porosity reasonably well for certain units while failing in
others, particularly when changes in scale of measurement are
considered. For example, GPR porosity estimates for HFC 4 along
test corehole G-3740 were 29% from common midpoints and 30%
from common offsets (Table 1), and thus were almost the same
as the overall average porosity of 29% (n = 2) from whole-core
laboratory porosity measurements in unit HFC 4 (Fig. 3b). Alterna-
tively, the low (7.1%) whole-core laboratory porosity measurement
for HFC 3b from direct measurements (Fig. 3b) is overestimated by
bulk GPR estimates (23–25%) (Table 1), but corresponds well with
the arithmetic average based on two whole-core laboratory poros-
ity measurements (Fig. 3b) for the entire unit (i.e. 23%). The differ-
ences between estimates can be attributed to the fact that each
particular method implies a different scale of measurement and
therefore different methods sample a different support volume
(as defined below). Accounting for support volumes plays a critical
role when comparing porosity estimates from different methods in
the Miami Limestone, due to the highly heterogeneous nature of
this system.
GPR waves consider a 3-D volume as the waves propagate
through the subsurface materials (Spetzler and Snieder, 2004). This
support volume is known as the Fresnel volume, which is spheroid
in shape, and is reliant on the dielectric permittivity of the materi-
als, the wavelength of the antennas used, and the depth to the
reflection surface. Alternatively, the Fresnel zone is a 2-D area
where the reflection occurs, and is elliptical (Fig. 7b). Thus the Fres-
nel volume represents the 3-D equivalent of the 2-D Fresnel zone.
The individual techniques used in this study will have dissimilar
effects on the support volume being measured as the transmitted
EM wave travels through a 3-D volume (Fig. 7): (1) direct measures
(such as whole core laboratory measurements) differ from GPR
measurements; and (2) estimated support volumes vary among
GPR surveying techniques, thereby resulting in differences in EM
wave velocity estimates and porosity. Core-scale measurements
(Fig. 7a) are similar to those made from subsets of borehole images
or other direct observation techniques and represent smaller vol-
umes from a larger sample.
GPR techniques incorporate large support volumes and
therefore estimates represent averages over large sample sizes.
The common offset technique has support volumes typically
ranging from approximately 2–4 m3 based on depths to reflector
of 2 and 4 m respectively, 1 m antenna offset, and 1 m width
(Fig. 7b). The common midpoint technique has sample volumes
(omitted for clarity in Fig. 7c) ranging from 2 to 40 m3 given the
same depth to reflector and dimensions. In contrast to the common
offset measures, common midpoint techniques require the antenna
spacing to increase at a given interval during the measure. As the
antenna separation increases, the resulting velocity takes into con-
sideration larger support volumes. A core-scale volume measure-
ment may provide the ability to characterize a small volume, but
the inherently small size of the core may result in under- or overes-
timation of the core’s environmental properties. This inaccuracy is
particularly true in highly heterogeneous environments such as
the Miami Limestone. On the other hand, when support volume is
increased in the measurements, other problems arise. For instance,
in the case of porosity estimation of the Miami Limestone, if the
support volume is equal to the size of a dissolution feature, a poros-
ity of 100% may be estimated. If the size of the dissolution feature is
half of the support area, then average porosity may decrease to 50%.
Ideally, and particularly for highly heterogeneous systems, the sam-
pling method should account for the range of possible sizes and dis-
tribution patterns of features within the subsurface.
8. Implications and future work
This study has potential implications for: (1) the assessment
and modeling of flow dynamics; (2) potential routes of contami-
nant transport; and (3) constraint of hydrogeologic flow models
used in the Everglades Restoration Program. Moreover, this
research has implications for other near-surface karstic aquifers
and shallow karstic petroleum well field systems where hyperbolic
diffraction-laden carbonates are present. Although this research is
supported both by direct porosity measurements and comparable
results derived from multiple geophysical measurements, further
research concerning the suitability of the petrophysical model cho-
sen for this estimation and the validity of the assumptions made
for the values used in the model should be considered. For
instance, studies at the laboratory scale that include GPR measure-
ments of Miami Limestone and Fort Thompson Formation samples
under fully saturated and fully unsaturated conditions can be used
to estimate values of the solid dielectric permittivity for the rocks
of both formations. Results of such studies will allow for improved
 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236
constraint of the CRIM petrophysical model and will further refine
field-scale estimates of porosity for the Miami Limestone and Fort
Thompson Formation.
9. Conclusions
Common offset GPR is a hydrogeophysical technique that has
been used widely to characterize shallow, water-saturated carbon-
ate aquifers. The technique was used in the Miami Limestone and
Fort Thompson Formations of the Biscayne aquifer to: (1) rapidly
characterize 1-D and 2-D distributions of porosity; and (2) investi-
gate the presence and lateral extent of dissolution karst features.
The unique distribution of diffraction hyperbolae detected in com-
mon offset surveys was used to estimate changes in electromag-
netic (EM) wave velocity, and these changes were converted to
porosity using the complex refractive index model (CRIM), a petro-
physical model. Additional GPR measurements in the common
midpoint mode, which is a more time-consuming acquisition mode
when compared to common offset mode, and a set of digital bore-
hole and corehole images collected in a previous study were used
to constrain porosity estimates from GPR common offsets. Results
indicate a good correspondence between GPR and laboratory mea-
surements of porosity from whole-core samples, particularly when
support volumes for each technique are taken into consideration.
The results presented herein challenge the results of previous
studies investigating porosity distribution in highly heterogeneous
systems such as the Biscayne aquifer. Of particular importance is
the spatial scale of measurement and the support volume of the
measurement technique, because results presented herein indicate
that smaller support volumes can more accurately capture both
vertical and lateral changes in porosity. The results of this study
are in accord with previous studies in the Biscayne aquifer that
indicated that sag features in common offset profiles are not true
karst or structural sags but instead are manifestations of low EM
wave velocity and thus high porosity below the water table attrib-
utable to velocity pull-down areas in the reflection record. Mapped
in 3-D, these sag features could provide a means for visualizing the
spatial distribution of regional concentrated groundwater flow
through underground passageways in the Biscayne aquifer.
Acknowledgments
We would like to thank the Florida Atlantic University Environ-
mental Sciences Everglades Fellowship Initiative for financial sup-
port as well as the National Park Service for work conducted under
permit EVER-2010-SCI-0012. This work also has been partially sup-
ported by the U.S. Geological Survey under the Greater Everglades
Priority Ecosystems Science. In addition, this material is based
upon work supported by the National Science Foundation under
Grant No. DGE: 0638662. Tyler McNabb, of Florida Atlantic Univer-
sity, provided valuable field support during collection of GPR data.
Andrew Parsekian provided the script used to process the common
midpoint data, as well as insight to the processing of the ground
penetrating radar data. We thank Tara Root from Florida Atlantic
University, Juan V Giraldez, Associate Editor and two anonymous
reviewers from the Journal of Hydrology for suggestions that
greatly enhanced earlier versions of this manuscript.
The use of trade, firm or product names in this paper is for iden-
tification purposes only and does not constitute endorsement by
the authors, Florida Atlantic University, or the U.S. Government.
References
Annan, A.P., 2009. Electromagnetic principles of ground penetrating radar. In: Jol,
H.M. (Ed.), Ground Penetrating Radar Theory and Applications. Elsevier,
Amsterdam, pp. 3–40.
235
Annan, A.P., Davis, J.L., 1976. Impulse radar soundings in permafrost. Radio Sci. 11,
383–394.
Bakalowicz, M., 2005. Karst groundwater: a challenge for new resources. Hydrogeol.
J. 13 (1), 148–160. http://dx.doi.org/10.1007/s10040-004-0402-9.
Ballard, R.F., 1984. Electromagnetic (Radar) techniques applied to cavity detection.
Geophysics 49 (5), 603–604.
Barquin-Valle, L.P., Migliaccio, K.W., Schaffer, B., Munoz-Carpen, R., Crane, J.H., Li,
Y.C., 2011. Predicting soil water content using the ‘‘drained to equilibrium’’
concept. Vadose Zone J. 10 (2), 675–682. http://dx.doi.org/10.2136/
vzj2010.0073.
Barr, G.L., 1993. Applications of Ground Penetrating Radar Methods in Determining
Hydrogeologic Conditions in a Karst Area, West-Central Florida, U.S. Geological
Survey Water-Resources Investigations Report 92-4141, pp. 26.
Birchak, J.R., Gardner, C.G., Hipp, J.E., Victor, J.M., 1974. High Dielectric Constant
Microwave Probes for Sensing Soil Moisture Proceedings of the Institute of
Electrical and Electronics Engineers, 62(1), 93–98. doi: 10.1109/proc.1974.9388.
Bolster, C.H., Genereux, D.P., Saiers, J.E., 2001. Determination of specific yield for the
Biscayne Aquifer with a canal-drawdown test. Ground Water 39 (5), 768–777.
Bradford, J.H., Clement, W.P., Barrash, W., 2009. Estimating porosity with ground-
penetrating radar reflection tomography: a controlled 3-D experiment at the
boise hydrogeophysical research site. Water Resour. Res. 45, W00d26 10.1029/
2008wr006960.
Bristow, C., 2009. Ground penetrating radar in aeolian dune sands. In: Jol, H.M.
(Ed.), Ground Penetrating Radar Theory and Applications. Elsevier, Amsterdam,
p. 271.
Buchner, R., Barthel, J., Stauber, J., 1999. The dielectric relaxation of water between 0
degrees C and 35 degrees C. Chem. Phys. Lett. 306 (1–2), 57–63. http://
dx.doi.org/10.1016/s0009-2614(99)00455-8.
Busby, J.P., Merritt, J.W., 1999. Quaternary deformation mapping with ground
penetrating radar. J. Appl. Geophys. 41 (1), 75–91. http://dx.doi.org/10.1016/
s0926-9851(98)00050-0.
Cassidy, N.J., 2009. Electrical and magnetic properties of rocks, soils and fluids. In:
Jol, H.M. (Ed.), Ground Penetrating Radar Theory and Applications. Elsevier,
Amsterdam, p. 41.
Chan, C.Y., Knight, R.J., 2001. Laboratory measurements of electromagnetic wave
velocity in layered sands. Water Resour. Res. 37 (4), 1099–1105. http://
dx.doi.org/10.1029/2000wr900356.
Choquett, P.W., Pray, L.C., 1970. Geologic nomenclature and classification of
porosity in sedimentary carbonates. Am. Assoc. Pet. Geol. Bull. 54 (2), 207–250.
Cressler, A., 1993. The Caves of Dade County, Florida, Georgia Underground:
Dogwood City Grotto. National Speleological Society, pp. 9–16.
Cunningham, K.J., 2004. Application of ground-penetrating radar, digital optical
borehole images, and cores for characterization of porosity hydraulic
conductivity and paleokarst in the Biscayne aquifer, southeastern Florida,
USA. J. Appl. Geophys. 55 (1–2), 61–76. http://dx.doi.org/10.1016/
j.jappgeo.2003.06.005.
Cunningham, K.J., Florea, L.J., 2009. The Biscayne Aquifer of Southeastern Florida,
Geography/Geology Faculty Publications. Western Kentucky University,
pp. 196–199.
Cunningham, K.J., Sukop, M.C., 2011. Multiple technologies applied to
characterization of the porosity and permeability of the Biscayne aquifer,
Florida., U.S. Geological Survey Open File Report 2011-1037, pp. 8.
Cunningham, K.J., Carlson, J., Hurley, N., 2004a. New method for quantification of
vuggy porosity from digital optical borehole images as applied to the karstic
Pleistocene limestone of the Biscayne aquifer, southeastern Florida. J. Appl.
Geophys. 55 (1–2), 77–90. http://dx.doi.org/10.1016/j.jappgeo.2003.06.006.
Cunningham, K.J., Carlson, J.I., Wingardm, G.L., Robinson, E., Wacker, M.A., 2004b.
Characterization of Aquifer Heterogeneity using Cyclostratigraphy and
Geophysical Methods in the Upper Part of the Karstic Biscayne Aquifer,
Southeastern Florida, U.S. Geological Survey Water-Resources Investigation
Report 03-4208, pp. 66.
Cunningham, K.J., Renken, R.A., Wacker, M.A., Zygnerski, M.R., Robinson, E., Shapiro,
A.M., Wingard, G.L., 2006a. Application of carbonate cyclostratigraphy and
borehole geophysics to delineate porosity and preferential flow in the karst
limestone of the Biscayne aquifer, SE Florida. In: Harmon, R.S., Wicks, C. (Eds.),
Perspectives on Karst Geomorphology, Hydrology, and Geochemistry – A
Tribute Volume to Derek C. Ford and William B. White. Geological Society of
America Special Paper 404, pp. 191–208.
Cunningham, K.J., Wacker, M.A., Robinson, E., Dixon, J., Wingard, G.L., 2006b. A
cyclostratigraphic and Borehole Geophysical Approach to Development of a
Three-Dimensional Conceptual Hydrogeologic Model of the Karstic Biscayne
Aquifer, Southeastern Florida., Scientific Investigations Report 2005-5235,
pp. 69.
Cunningham, K.J., Sukop, M.C., Huang, H.B., Alvarez, P.F., Curran, H.A., Renken, R.A.,
Dixon, J.F., 2009. Prominence of ichnologically influenced macroporosity in the
karst Biscayne aquifer: Stratiform ‘‘super-K’’ zones. Geol. Soc. Am. Bull. 121 (1–
2), 164–180. http://dx.doi.org/10.1130/b26392.1.
Dafflon, B., Barrash, W., 2012. Three-dimensional stochastic estimation of porosity
distribution: Benefits of using ground-penetrating radar velocity tomograms in
simulated-annealing-based or Bayesian sequential simulation approaches.
Water Resour. Res. 48, W05553 10.1029/2011wr010916.
Dicken, C.L., Nicholson, S.W., Horton, J.D., Foose, M.P., Mueller, J.A., 2005.
Preliminary Integrated Geologic map Databases for the United States, U.S.
Geological Survey Open-File Report 2005-1323, Reston Virginia, pp. Shapefile.
Dix, C.H., 1955. Seismic velocities from surface measurements. Geophysics 20 (18),
68–86.
236 G.J. Mount et al. / Journal of Hydrology 515 (2014) 223–236
Dominic, D.F., Egan, K., Carney, C., Wolfe, P.J., Boardman, M.R., 1995. Delineation of
shallow stratigraphy using ground penetrating radar. J. Appl. Geophys. 33 (1–3),
167–175. http://dx.doi.org/10.1016/0926-9851(95)90039-x.
Fish, J.E., Stewart, M., 1990. Hydrogeology of the Surficial Aquifer System, Dade
County, Florida, U.S. Geological Survey Water-Resources Investigations Report
90-4108, pp. 56.
Ford, D.C., Williams, P.W., 1989. Karst Geomorphology and Hydrology. Academic
Division of Unwin Hyman Ltd., London, England., 576pp.
Grasmueck, M., Weger, R., Horstmeyer, H., 2004. Three-dimensional ground-
penetrating radar imaging of sedimentary structures, fractures, and
archaeological features at submeter resolution. Geology 32 (11), 933–936.
Greaves, R.J., Lesmes, D.P., Lee, J.M., Toksoz, M.N., 1996. Velocity variations and
water content estimated from multi-offset, ground-penetrating radar.
Geophysics 61 (3), 683–695.
Halley, R.B., Evans, C.C., 1983. The Miami Limestone: A guide to Selected Outcrops
and Their Interpretation. Miami Geological Society, Miami, Florida, pp. 67.
Harbi, H., McMechan, G.A., 2011. Modeling 3D porosity and permeability from GPR
data in the Ellenburger Dolomite, central Texas. Geophysics 76 (6), J35–J46.
http://dx.doi.org/10.1190/geo2011-0051.1.
Harvey, R.W., Metge, D.W., Shapiro, A.M., Renken, R.A., Osborn, C.L., Ryan, J.N.,
Cunningham, K.J., Landkamer, L., 2008. Pathogen and chemical transport in the
karst limestone of the Biscayne aquifer: 3. Use of microspheres to estimate the
transport potential of Cryptosporidium parvum oocysts. Water Resour. Res. 44
(8), W08431 10.1029/2007wr006060.
Jacob, R.W., Hermance, J.F., 2004. Assessing the precision of GPR velocity
and vertical two-way travel time estimates. J. Environ. Eng. Geophys. 9 (3),
143–153.
Klein, H., Hull, J.E., 1978. Biscayne Aquifer, Southeast Florida, U.S. Geological Survey
Water-Resources Investigations 78–107, pp. 52.
Klein, H., Sherwood, C.B., 1961. Hydrologic Conditions in the Vicinity of Levee 30,
Northern Dade County, Florida. Florida Bureau of Geology Report of
Investigations, pp. 24.
Knoll, M.D., 1996. A Petrophysical basis for Ground-Penetrating Radar and Very
Early Time Electromagnetics: Electrical Properties of Sand–Clay Mixtures.
NN14774 Thesis, The University of British Columbia, Canada, 337pp.
Kruse, S.E., Schneider, J.C., Campagna, D.J., Inman, J.A., Hickey, T.D., 2000. Ground
penetrating radar imaging of cap rock, caliche and carbonate strata. J. Appl.
Geophys. 43 (2–4), 239–249. http://dx.doi.org/10.1016/s0926-9851(99)00062-2.
Kruse, S., Grasmueck, M., Weiss, M., Viggiano, D., 2006. Sinkhole structure imaging
in covered Karst terrain. Geophys. Res. Lett. 33 (16), L16405, L16405 10.1029/
2006gl026975.
Lai, W.L., Tsang, W.F., Fang, H., Xiao, D., 2006. Experimental determination of bulk
dielectric properties and porosity of porous asphalt and soils using GPR and a
cyclic moisture variation technique. Geophysics 71 (4), K93–K102. http://
dx.doi.org/10.1190/1.2217730.
Light, S.S., Dineen, J.W., 1994. Water Control in the Everglades: a Historical
Perspective. In: Davis, S.M., Ogden, J.G. (Eds.), Everglades: The Ecosystem and its
Restoration. St. Lucie Press, Delray Beach, Florida, pp. 47–84.
Lucia, F.J., 2007. Carbonate Reservoir Characterization electronic resource: An
Integrated Approach. Springer-Verlag Berlin Heidelberg, Berlin, Heidelberg, pp.
226.
Malget, D., 2002. Everglades National Park Boundary. Colorado State University,
National Park Service, pp. Shapefile.
Manda, A., Gross, M., 2006. Identifying and characterizing solution conduits in karst
aquifers through geospatial (GIS) analysis of porosity from borehole imagery:
An example from the Biscayne aquifer, South Florida (USA). Adv. Water Resour.
29 (3), 383–396. http://dx.doi.org/10.1016/j.advwatres.2005.05.013.
McPherson, B.F., Miller, R.L., Haag, K.H., Bradner, Anne, 2000. Water Quality in
Southern Florida, 1996–1998, U.S. Geological Survey Circular 1207, pp. 32.
Mitchum, R.M., Vail, P.R., Sangree, J.B., 1977. Seismic straitgraphy and global
changes of sea levels, part 6: stratigraphic interpretation of seismic reflection
patters in depositional sequences. In: Payton, C.E. (Ed.), Seismic Stratigraphy –
Applications to Hydrocarbon Exploration, AAPG Memoir 26, pp. 117–133.
Murley, J.F., Lenore Alpert, Lindsay Walker, M.J. Matthews, MaryBeth Burton,
Christina Bryk, Kurry, M., 2006. Charting the Course: Where is South Florida
Heading? Center for Urban and Environmental Solutions at Florida Atlantic
University, pp. 84.
Neal, A., 2004. Ground-penetrating radar and its use in sedimentology: principles,
problems and progress. Earth Sci. Rev. 66 (3–4), 261–330.
Neal, A., Grasmueck, M., McNeill, D.F., Viggiano, D.A., Eberli, G.P., 2008. Full-
resolution 3d radar stratigraphy of complex oolitic sedimentary architecture:
Miami Limestone, Florida, USA. J. Sediment. Res. 78 (9–10), 638–653. http://
dx.doi.org/10.2110/jsr.2008.070.
Nordahl, K., Ringrose, P., 2008. Identifying the representative elementary volume
for permeability in heterolithic deposits using numerical rock models. Math.
Geosci. 40 (7), 753–771. http://dx.doi.org/10.1007/s11004-008-9182-4.
Parker, G.G., Ferguson, G.E., Love, S.K., 1955. Water Resources of Southeastern
Florida, with Special Reference to the Geology and Ground Water of the Miami
Area., U.S. Geological Survey Water Supply Paper 1255, pp. 965.
Parsekian, A.D., Slater, L., Comas, X., Glaser, P.H., 2010. Variations in free-phase
gases in peat landforms determined by ground-penetrating radar. J. Geophys.
Res. – Biogeosci. 115, G02002, G02002 10.1029/2009jg001086.
Perkins, R.D., 1977. Depositional framework of Pleistocene rocks in south Florida.
In: Enos, P., Perkins, R.D. (Eds.), Quartenary Sedimentation in South Floirda:
Geological Society of America Memoir 147, pp. 131–198.
Renken, R.A., Cunningham, K.J., Zygnerski, M.R., Wacker, M.A., Shapiro, A.M., Harvey,
R.W., Metge, D.W., Osborn, C.L., Ryan, J.N., 2005. Assessing the vulnerability of a
municipal well field to contamination in a karst aquifer. Environ. Eng. Geosci. 11
(4), 319–331.
Renken, R.A., Cunningham, K.J., Shapiro, A.M., Harvey, R.W., Zygnerski, M.R., Metge,
D.W., Wacker, M.A., 2008. Pathogen and chemical transport in the karst
limestone of the Biscayne aquifer: 1. Revised conceptualization of groundwater
flow. Water Resour. Res. 44 (8), W08429, W08429 10.1029/2007wr006058.
Reynolds, J.M., 2011. An Introduction to Applied and Environmental Geophysics.
Wiley-Blackwell, 696 pp.
Robinson, D.A. et al., 2008. Advancing process-based watershed hydrological
research using near-surface geophysics: a vision for, and review of, electrical
and magnetic geophysical methods. Hydrol. Process. 22 (18), 3604–3635.
http://dx.doi.org/10.1002/hyp.6963.
Roth, K., Schulin, R., Fluhler, H., Attinger, W., 1990. Calibration of time domain
reflectometry for water-content measurement using a composite dielectric
approach. Water Resour. Res. 26 (10), 2267–2273. http://dx.doi.org/10.1029/
90wr01238.
Schmoker, J.W., Hester, T.C., 1986. Porosity of the Miami Limestone (late
Pleistocene), lower Florida Keys. J. Sediment. Res. 56 (5), 629–634. http://
dx.doi.org/10.1306/062402730031.
Shapiro, A.M., Renken, R.A., Harvey, R.W., Zygnerski, M.R., Metge, D.W., 2008.
Pathogen and chemical transport in the karst limestone of the Biscayne aquifer:
2. Chemical retention from diffusion and slow advection. Water Resour. Res. 44
(8), W08430, W08430 10.1029/2007wr006059.
Smart, C.C., 1999. Subsidiary conduit systems: a hiatus in aquifer monitoring and
modelling. In: Palmer, A.N., Palmer, M.V., Sasowsky, I.D. (Eds.), Karst Modelling,
Symposium Proceedings. Karst Water Institute, Charlottesville, pp. 146–157.
Spetzler, J., Snieder, R., 2004. The Fresnel volume and transmitted waves.
Geophysics 69 (3), 653–663. http://dx.doi.org/10.1190/1.1759451.
Truss, S., Grasmueck, M., Vega, S., Viggiano, D.A., 2007. Imaging rainfall drainage
within the Miami oolitic limestone using high-resolution time-lapse ground-
penetrating radar. Water Resour. Res. 43 (3), W03405.
Turesson, A., 2006. Water content and porosity estimated from ground-penetrating
radar and resistivity. J. Appl. Geophys. 58 (2), 99–111. http://dx.doi.org/
10.1016/j.jappgeo.2005.04.004.
U.S. Geological Survey, 2003. Principal Aquifers of the 48 Conterminous United
States, Hawaii, Puerto Rico, and the U.S. Virgin Islands. U.S. Geological Survey,
Reston, Virginia.
Vacher, H., Mylroie, J., 2002. Eogenetic karst from the perspective of an equivalent
porous medium. Carbonates Evaporites 17 (2), 182–196.
Vik, B., Bastesen, E., Skauge, A., 2013. Evaluation of representative elementary
volume for a vuggy carbonate rock – Part: Porosity, permeability,
and dispersivity. J. Petrol. Sci. Eng. 112, 36–47, dx.doi.org/10.1016/j.petrol.
2013.03.029.
West, L.J., Handley, K., Huang, Y., Pokar, M., 2003. Radar frequency dielectric
dispersion in sandstone: implications for determination of moisture and
clay content. Water Resour. Res. 39 (2), 1026. http://dx.doi.org/10.1029/
2001wr000923.
Worthington, S.R.H., 1999. A comprehensive strategy for understanding flow in
carbonate aquifers. In: Palmer, A.N., Palmer, M.V., Sasowsky, I.D. (Eds.), Karst
Modelling, Symposium Proceedings. Karst Water Institute, Charlottesville,
pp. 30–37.