CAVES, SINKHOLES, AND FRACTURES IN THE EOGENETIC 
KARST OF FLORIDA, A GIS-BASED SPATIAL ANALYSIS
JAMA, VRTAČE IN RAZPOKE V EOGENEM KRASU FLORIDE, 
PROSTORSKA ANALIZA Z ORODJI GIS
Can DENIZMAN1*
Abstract  UDC  551.435.8:659.2:004:91(735.9)
Can Denizman: Caves, sinkholes, and fractures in the eoge-
netic karst of Florida, a GIS-based spatial analysis
The correlation between surface and subsurface karst develop-
ment was explored by comparing the directionality and spatial 
distribution of karstic depressions around twenty-two select 
caves in the eogenetic karst of Florida. Orientations of cave 
passages and major axes of depressions around cave centrelines 
imply varying degrees of correlation between them. Spatial 
distribution of karstic depressions was studied by standard de-
viational ellipses of sinkhole centroids and nearest neighbour 
orientations around caves using spatial statistics tools of Arc-
GIS. An overall analysis of the data shows close connections 
between some caves and the surrounding sinkholes in terms 
of their orientation and spatial distribution, suggesting the im-
portance of fracture systems in the development of karst.
Keywords: karst geomorphology, caves, sinkholes, GIS, spatial 
analysis, Florida.
Izvleček UDK 551.435.8:659.2:004:91(735.9)
Can Denizman: Jame, vrtače in razpoke na eogenem krasu na 
Floridi, prostorska analiza z orodji GIS
Z analizo smeri in prostorske porazdelitve vrtač v okolici 22 
izbranih jam na eogenem krasu na Floridi smo raziskovali 
povezavo med površinskim in podzemnim razvojem krasa. 
Primerjava smeri jamskih rovov in glavnih osi vrtač v okolici 
jam kaže na različno stopnjo povezave med njimi. Prostorsko 
porazdelitev vrtač smo raziskovali z orodji prostorske statistike 
v okolju ArcGIS, pri tem smo uporabili elipse standardnih od-
klonov centroidov vrtač in smeri osi vrtač, najbližjih sosedov 
izbranih jamskih rovov. Celotna analiza podatkov je poka-
zala tesno povezanost med nekaterimi jamami in okoliškimi 
vrtačami z vidika njihove orientacije in lege, kar kaže na pomen 
razpoklinskih sistemov za razvoj krasa.
Ključne besede: kraška geomorfologija, jame, vrtače, GIS, pro-
storska analiza, Florida.
ACTA CARSOLOGICA 51/1, 33-46, POSTOJNA 2022
1 Department of Physics, Astronomy, Geosciences, and Engineering Technologies, Valdosta State University, 1500 Patterson 
Street, GA-31698, Valdosta, Georgia, USA, e-mail: cdenizma@valdosta.edu
*  Corressponding author
Received/Prejeto: 18.10.2021 
COBISS: 1.01, DOI:10.3986/ac.v51i1.10454
CC BY-NC-ND
1. INTRODUCTION
Karst in Florida developed within the thick and porous 
Eocene and Oligocene limestones of a stable carbonate 
platform. The lack of significant tectonic activity gave 
rise to a well-developed karst plain with very little relief, 
which was later covered by siliciclastic material (White, 
1970; Scott, 1997). The development of a surficial drain-
age network on the impermeable siliciclastic cover not 
only increased hydraulic gradient by river downcut-
ting, contributing to karst groundwater circulation and 
conduit development but also eroded the impermeable 
cover material along the Suwannee River (Denizman 
& Randazzo, 2000). Allogenic recharge to the exposed 
carbonate platform and the subsequent hydrogeologic 
connection between surface and subsurface karst, a 
common characteristic of epigenetic karst areas, is 
readily observed in Florida by the sinkhole-cave-spring 
continuum (Kincaid, 1998) as well as groundwater 
flow within a maze of passages and rock matrix (Flo-
rea, 2006). Hypogenic karst development due to vari-
ous processes such as groundwater – seawater mixing 
during low sea-level stands was proposed to explain 
deeper cave horizons (Moore et al., 2010; Gulley et al., 
2013). With no significant burial diagenesis, carbonate 
rocks of Florida still retain their intergranular (matrix) 
porosity and form a perfect example of eogenetic karst 
as defined by Vacher and Mylroie (2012). Groundwa-
ter storage and flow take place within the dual porosity 
of the eogenetic karst aquifers which may not show a 
strong connection with the regional structural features 
and surficial karst development. Nevertheless, fracture 
patterns seem to be playing an important role in cave 
development with crude branchwork patterns within 
the undeformed Tertiary carbonates of the eogenetic 
Florida karst (Florea &Vacher, 2006; Palmer, 2009; Up-
church et al., 2019). The importance of structural fea-
tures such as faults, fractures, and bedding planes on 
karst development, as well as the connection between 
subsurface and surficial karst processes, have been re-
ported especially in telogenetic karst areas with little 
matrix porosity due to burial and diagenesis (e.g., Ford, 
CAN DENIZMAN
Figure 1: Major physiographic provinces and cave locations. Gray areas show minor physiographic provinces such as Coastal Swamps, 
Brooksville Ridge and Bell Ridge.
ACTA CARSOLOGICA 51/1 – 202234
CAVES, SINKHOLES, AND FRACTURES IN THE EOGENETIC KARST OF FLORIDA, A GIS-BASED SPATIAL ANALYSIS
1964; La Velle, 1967; Palmer & Palmer, 1975; Kemmerly, 
1982; Barlow & Ogden, 1982; Shofner et al., 2001; Favre 
& Pahernic, 2007). This study aims to explore the rela-
tionship between the caves and karstic depressions in 
the eogenetic karst of Florida by analyzing the spatial 
distribution and directional patterns of selected caves 
and surrounding karstic depressions. Analysis of large 
sets of spatial data was made possible by utilizing GIS as 
in many other studies on the spatial analysis and mor-
phometric features of karst features (e.g., Denizman, 
2003; Angel et al., 2004; Lyew-Ayee et al., 2009; Komac 
& Urbanc, 2012; Öztürk et al., 2018).
2. PHYSIOGRAPHY AND STRUCTURAL GEOLOGY OF THE STUDY AREA
Karst in Florida has developed within the undeformed 
Eocene and Oligocene carbonates of high primary po-
rosity and rather uniform stratigraphy which can only be 
differentiated by biostratigraphy (Randazzo, 1997).
The Northern Highlands and Gulf Coastal Low-
lands constitute the major physiographic provinces in 
the area (Figure 1). As one of the most distinct physio-
graphic features in Florida, the Northern Highlands oc-
cupies most of the north and east of the study area. From 
the hydrogeologic standpoint, the Northern Highlands 
contains a thick confining unit of siliciclastic sediments 
and generates confining conditions for the Floridan aqui-
fer. The confining unit originally covered the entire study 
area and has been eroded by headward erosion through 
surface drainage as well as by karstic dissolution within 
the underlying carbonate units (Scott, 1997). 
The Gulf Coastal Lowlands consists of both ero-
sional and depositional features. Broad plains of a series 
of Pleistocene surfaces and shorelines are pitted with 
karstic depressions within the limestone at or near land 
surface. The Floridan aquifer is unconfined. It repre-
sents a typical mature karst terrane with a thin mantle 
of permeable marine terrace deposits. Because of the 
low topographic relief and rapid infiltration of rainfall 
by diffuse recharge to the karst aquifer, surficial runoff 
is limited to major rivers. Well-developed epikarst pro-
vides the initial stages of subsurface karst development 
(Upchurch et al., 2019). 
The Cody escarpment separates the Northern High-
lands from the Gulf Coastal Plain and plays a major role 
in karst development along the retreating marginal zone 
between the two physiographic provinces. Allogenic re-
charge from the noncarbonate cover sediments of the 
Northern Highlands accounts for extensive dissolution, 
resulting in disappearing streams and collapse features 
above major solutional conduits (Figure 2). Numerous 
cave diving expeditions and tracing experiments have 
revealed an intricate network of karst development rep-
resented by caves, disappearing streams, sinkholes, and 
springs (Figure 3; Kincaid, 1998).
Figure 2: A disappearing stream in 
Leon Sinks State Park, Florida. 
ACTA CARSOLOGICA 51/1 – 2022 35
CAN DENIZMAN
Structural control on the drainage patterns and 
sinkhole alignments have been reported in various stud-
ies on the geomorphology of the Florida platform. Ver-
non (1951) proposes stresses that form the folds of the 
Ocala Uplift (Platform) to be the cause of the regional 
fracturing and mentions NW and NE system of fractures 
paralleling stream patterns and sinkhole alignments. 
These fractures are thought to have been formed by the 
tensional stresses over the anticlinal flexure. 
Upchurch et al. (2019) suggest that the fractures 
due to tidal and tectonic stresses develop as soon as 
the carbonates are cemented and thus represent brittle 
characteristics. They propose tidal stresses as possible 
causes of fracture development rather than tectonic 
movement.
3. DATA 
Spatial correlation between subsurface and surface karst 
development was explored by comparing the alignment 
of cave passages with major axes of karstic depressions 
and spatial distribution of depression centroids around 
caves. 
3.1 CAVE DATABASE
Data on subsurface karst development comprise a Geo-
graphic Information System (GIS) database of twenty-
two phreatic caves located within the Gulf Coastal Low-
lands physiographic province where the Floridan aquifer 
is unconfined (Figure 1). GIS layers of caves, created by 
digitizing the centerlines of cave passages in ArcGIS con-
tain information on average dimensions and depths of 
cave segments. All the caves are phreatic and most are 
located along the Suwannee and Wakulla rivers. 
3.2 KARSTIC DEPRESSIONS DATABASE
The surficial karst development is analyzed by the spa-
tial distribution of karstic depressions. Data on karstic 
depressions were compiled from four different sources:
1. Topographic maps: Most of the depressions used 
in this study were digitized as polyline layers from 
1/24,000 topographic maps. Each depression is repre-
sented by its GIS-determined centroid point.
2. Florida sinkhole database: Maintained by the Florida 
Geological Survey, this database is comprised of re-
ported subsidence incidents statewide.
3. Digital soil maps: SSURGO GIS soils data set includes 
some spatial point data that are potentially useful, 
especially since they are based on the direct field ob-
servations of the soil mappers.  The soils survey re-
ports use the terms "depression" and "sinkhole" inter-
changeably to an extent since they are both defined on 
the basis of closed depressions. 
4 2-ft topographic contours derived from the LIDAR 
data by the Florida Division of Emergency Manage-
ment. This GIS layer was available only for the north-
west corner of the study area around the following 
caves: Wakulla, Leon, Sally’s Ward, Indian, Shepherd, 
and McBride.
In the field, a total of 244 fracture orientations along the 
Suwannee River were measured and displayed in rose 
diagrams.
4. SPATIAL ANALYSIS OF THE DATA
Spatial analyses of the data include calculating length and 
orientation of cave passages and hundreds of depression 
major axes around the caves, determining points of de-
pression centroids, and their spatial distribution proper-
ties on a GIS platform. 
To explore the directional correlation between 
surface and subsurface karst development, depres-
sions within 2 and 3 km of cave centerlines were used. 
Major axis azimuths of depressions with major axis-
minor axis ratio >= 1.5, measured by ArcGIS using a 
Figure 3: Wakulla Springs cave entrance.
ACTA CARSOLOGICA 51/1 – 202236
CAVES, SINKHOLES, AND FRACTURES IN THE EOGENETIC KARST OF FLORIDA, A GIS-BASED SPATIAL ANALYSIS
fitted ellipse to the depression polygon, were calculat-
ed and compared with the azimuths of cave passages 
(Figure 2). 
Using the spatial statistics tools of ArcGIS 10.4, spa-
tial orientation of depression centroids located within 
distances of 1, 2, and 3 kilometers from the cave cen-
terline were analyzed. Spatial orientation, or trend of 
depression point distribution was determined by calcu-
lating the standard distance separately in X and Y direc-
tions. These two measures make up the axes of an ellipse 
that describe the distribution of features. The ellipse, re-
ferred to as standard deviational ellipse (SDE), represents 
the standard deviation of the features from the mean cen-
ter separately for the X and Y coordinates. It is expressed 
as (Equation 1):
   
(1)
where SDx and SDy are standard distances for the x and y 
axes, xi, yi, , and represent x and y coordinates of a feature 
and the mean x, y coordinates, and n is the number of 
features (Mitchel, 2005). 
Spatial distribution of depression centroids was also 
analyzed by calculating azimuth values between nearest 
neighbors.
Figure 4: Examples of depression 
major half axes and centroids. 
Numbers denote major half axis 
azimuth angles.
ACTA CARSOLOGICA 51/1 – 2022 37
CAN DENIZMAN
 5. RESULTS AND DISCUSSIONS
Table 1 shows an overall summary of the cave morphol-
ogy as related by Palmer (1991, 2009) to recharge type 
and the dominant porosity. The nearest neighbor index 
patterns for depressions within 3 km of each cave, the 
degree of alignment between each cave, and the spatial 
deviation ellipses (SDE) for depressions within 1, 2, and 
3 km of the cave centerline are also included in this table. 
All the porosity types cited in Palmer’s diagram –frac-
ture, bedding planes, and the intergranular porosity- are 
observed in caves examined in this study. 
Figure 5 allows a comparison of alignments among 
cave centerlines, nearest neighbor azimuths, SDE for 
depressions, and depression major axes within 2 and 3 
km of the cave centerlines. In general, moderate to good 
correlation of alignments are observed between the rose 
diagrams of cave passages and the nearest neighbor azi-
muths within 2 km of the following caves: Bonnet, Cow-
and the SDEs for the following caves: Bonnet, Cow, Leon, 
Litle River, Luraville, McBride, Morgan, Peacock, Shepa-
rd, Suwanacoochee, and Wakulla. 
Table 1: Cave patterns and spatial distribution of sinkhole centroids (# :Number, NN: Nearest Neighbor, SDE: Standard Deviational El-
lipse). Number of sinkholes within 2 and 3 km of caves are separated by semi-colon.
Cave Pattern Dominant Control
# of Sinkholes 
with L/W>=1.5 
in 2 and 3 km
NN 
Pattern in 
3 km
SDE 
Alignment 
Blue Hole Angular Fracture 16; 115 Random Poor
Bonnet Anastomoses  + angular 
Bedding plane partings  
+ fractures 51; 162 Random Poor
Cathedral-
Falmouth Angular Fracture 313; 927 Clustered Perfect 
Convict Angular  + rudimentary branchwork
Fracture  
+ intergranular 43; 152 Random Poor
Cow Angular  + rudimentary branchwork
Fracture  
+ intergranular 51; 116 Random Moderate 
Green Curvilinear passages Bedding plane partings 80; 245 Random Poor
Hart Anastomoses Bedding plane partings 50; 104 Clustered Perfect
Indian Angular  + Curvilinear passages
Fracture  
+ bedding plane partings 137; 430 Clustered Poor
Leon Angular  + rudimentary branchwork
Fracture  
+ intergranular 460; 745 Clustered Perfect 
Little River Angular  + rudimentary branchwork
Fracture  
+ intergranular 67; 195 Clustered Good 
Luraville Angular Fracture 172; 322 Clustered Good 
Madison Blue Angular  + rudimentary branchwork
Fracture  
+ intergranular 16; 39 Clustered Good
Manatee Angular Fracture 61; 140 Clustered Perfect
McBride Angular Fracture 192; 361 Clustered Good
Morgan Angular Fracture 43; 110 Clustered Good
Peacock Anastomoses  + angular 
Bedding plane partings  
+ fractures 64; 186 Clustered Poor
Rock Bluff Angular  + rudimentary branchwork
Fracture  
+ intergranular 35; 128 Clustered Moderate
Sally's Ward Angular Fracture 162;392 Clustered Good
Shepard Angular Fracture 211; 462 Clustered Moderate
Suwanacoochee Angular  + rudimentary branchwork
Fracture  
+ intergranular 99; 185 Clustered Perfect 
Telford Angular  + rudimentary branchwork
Fracture  
+ intergranular 79; 249 Random Good
Wakulla Angular  + Curvilinear passages
Fracture  
+ bedding plane partings 317; 607 Clustered Perfect 
ACTA CARSOLOGICA 51/1 – 202238
CAVES, SINKHOLES, AND FRACTURES IN THE EOGENETIC KARST OF FLORIDA, A GIS-BASED SPATIAL ANALYSIS
Fi
gu
re
 5
a:
 S
um
m
ar
y 
of
 D
ire
ct
io
na
l A
na
ly
se
s. 
I) 
Ca
ve
 n
am
e a
nd
 p
at
te
rn
, I
I) 
Ca
ve
 p
as
sa
ge
 ce
nt
er
lin
e, 
II
I) 
Ca
ve
 p
as
sa
ge
s (
%
 o
f t
ot
al
 le
ng
th
), 
IV
) N
ea
re
st 
ne
ig
hb
or
 a
zi
m
ut
h 
fo
r d
ep
re
ss
io
n 
ce
nt
ro
id
s w
ith
in
 2
 k
m
 of
 th
e c
av
e c
en
te
rli
ne
 (f
re
qu
en
cy
) V
) S
ta
nd
ar
d 
D
ev
ia
tio
na
l E
lli
ps
es
 fo
r d
ep
re
ss
io
n 
ce
nt
ro
id
s w
ith
in
 1,
 2
, a
nd
 3
 k
m
 of
 ca
ve
 ce
nt
er
lin
es
, V
I) 
M
aj
or
 a
xe
s (
%
 of
 to
ta
l l
en
gt
h)
 
fo
r d
ep
re
ss
io
ns
 w
ith
 m
aj
or
 a
xi
s/m
in
or
 a
xi
s >
= 
1.
5 
w
ith
in
 2
 k
m
 o
f t
he
 ca
ve
 ce
nt
er
lin
e, 
V
II)
 M
aj
or
 a
xe
s (
%
 o
f t
ot
al
 le
ng
th
) f
or
 d
ep
re
ss
io
ns
 w
ith
 m
aj
or
 a
xi
s/m
in
or
 a
xi
s >
= 
1.
5 
w
ith
in
 3
 k
m
 o
f 
th
e c
av
e c
en
te
rli
ne
.
ACTA CARSOLOGICA 51/1 – 2022 39
CAN DENIZMAN
Fi
gu
re
 5
.b
: S
um
m
ar
y 
of
 D
ire
ct
io
na
l A
na
ly
se
s. 
I) 
Ca
ve
 n
am
e a
nd
 p
at
te
rn
, I
I) 
Ca
ve
 p
as
sa
ge
 ce
nt
er
lin
e, 
II
I) 
Ca
ve
 p
as
sa
ge
s (
%
 o
f t
ot
al
 le
ng
th
), 
IV
) N
ea
re
st 
ne
ig
hb
or
 a
zi
m
ut
h 
fo
r d
ep
re
ss
io
n 
ce
nt
ro
id
s w
ith
in
 2
 k
m
 of
 th
e c
av
e c
en
te
rli
ne
 (f
re
qu
en
cy
) V
) S
ta
nd
ar
d 
D
ev
ia
tio
na
l E
lli
ps
es
 fo
r d
ep
re
ss
io
n 
ce
nt
ro
id
s w
ith
in
 1,
 2
, a
nd
 3
 k
m
 of
 ca
ve
 ce
nt
er
lin
es
, V
I) 
M
aj
or
 a
xe
s (
%
 of
 to
ta
l l
en
gt
h)
 
fo
r d
ep
re
ss
io
ns
 w
ith
 m
aj
or
 a
xi
s/m
in
or
 a
xi
s >
= 
1.
5 
w
ith
in
 2
 k
m
 o
f t
he
 ca
ve
 ce
nt
er
lin
e, 
V
II)
 M
aj
or
 a
xe
s (
%
 o
f t
ot
al
 le
ng
th
) f
or
 d
ep
re
ss
io
ns
 w
ith
 m
aj
or
 a
xi
s/m
in
or
 a
xi
s >
= 
1.
5 
w
ith
in
 3
 k
m
 o
f 
th
e c
av
e c
en
te
rli
ne
.
ACTA CARSOLOGICA 51/1 – 202240
CAVES, SINKHOLES, AND FRACTURES IN THE EOGENETIC KARST OF FLORIDA, A GIS-BASED SPATIAL ANALYSIS
Fi
gu
re
 5
c: 
Su
m
m
ar
y 
of
 D
ire
ct
io
na
l A
na
ly
se
s. 
I) 
Ca
ve
 n
am
e 
an
d 
pa
tte
rn
, I
I) 
Ca
ve
 p
as
sa
ge
 ce
nt
er
lin
e, 
II
I) 
Ca
ve
 p
as
sa
ge
s (
%
 o
f t
ot
al
 le
ng
th
), 
IV
) N
ea
re
st 
ne
ig
hb
or
 a
zi
m
ut
h 
fo
r d
ep
re
ss
io
n 
ce
nt
ro
id
s w
ith
in
 2
 k
m
 of
 th
e c
av
e c
en
te
rli
ne
 (f
re
qu
en
cy
) V
) S
ta
nd
ar
d 
D
ev
ia
tio
na
l E
lli
ps
es
 fo
r d
ep
re
ss
io
n 
ce
nt
ro
id
s w
ith
in
 1,
 2
, a
nd
 3
 k
m
 of
 ca
ve
 ce
nt
er
lin
es
, V
I) 
M
aj
or
 a
xe
s (
%
 of
 to
ta
l l
en
gt
h)
 
fo
r d
ep
re
ss
io
ns
 w
ith
 m
aj
or
 a
xi
s/m
in
or
 a
xi
s >
= 
1.
5 
w
ith
in
 2
 k
m
 o
f t
he
 ca
ve
 ce
nt
er
lin
e, 
V
II)
 M
aj
or
 a
xe
s (
%
 o
f t
ot
al
 le
ng
th
) f
or
 d
ep
re
ss
io
ns
 w
ith
 m
aj
or
 a
xi
s/m
in
or
 a
xi
s >
= 
1.
5 
w
ith
in
 3
 k
m
 o
f 
th
e c
av
e c
en
te
rli
ne
.
ACTA CARSOLOGICA 51/1 – 2022 41
CAN DENIZMAN
Fi
gu
re
 5
d:
 S
um
m
ar
y 
of
 D
ire
ct
io
na
l A
na
ly
se
s. 
I) 
Ca
ve
 n
am
e a
nd
 p
at
te
rn
, I
I) 
Ca
ve
 p
as
sa
ge
 ce
nt
er
lin
e, 
II
I) 
Ca
ve
 p
as
sa
ge
s (
%
 o
f t
ot
al
 le
ng
th
), 
IV
) N
ea
re
st 
ne
ig
hb
or
 a
zi
m
ut
h 
fo
r d
ep
re
ss
io
n 
ce
nt
ro
id
s w
ith
in
 2
 k
m
 of
 th
e c
av
e c
en
te
rli
ne
 (f
re
qu
en
cy
) V
) S
ta
nd
ar
d 
D
ev
ia
tio
na
l E
lli
ps
es
 fo
r d
ep
re
ss
io
n 
ce
nt
ro
id
s w
ith
in
 1,
 2
, a
nd
 3
 k
m
 of
 ca
ve
 ce
nt
er
lin
es
, V
I) 
M
aj
or
 a
xe
s (
%
 of
 to
ta
l l
en
gt
h)
 
fo
r d
ep
re
ss
io
ns
 w
ith
 m
aj
or
 a
xi
s/m
in
or
 a
xi
s >
= 
1.
5 
w
ith
in
 2
 k
m
 o
f t
he
 ca
ve
 ce
nt
er
lin
e, 
V
II)
 M
aj
or
 a
xe
s (
%
 o
f t
ot
al
 le
ng
th
) f
or
 d
ep
re
ss
io
ns
 w
ith
 m
aj
or
 a
xi
s/m
in
or
 a
xi
s >
= 
1.
5 
w
ith
in
 3
 k
m
 o
f 
th
e c
av
e c
en
te
rli
ne
.
ACTA CARSOLOGICA 51/1 – 202242
CAVES, SINKHOLES, AND FRACTURES IN THE EOGENETIC KARST OF FLORIDA, A GIS-BASED SPATIAL ANALYSIS
Fi
gu
re
 5
e: 
Su
m
m
ar
y 
of
 D
ire
ct
io
na
l A
na
ly
se
s. 
I) 
Ca
ve
 n
am
e 
an
d 
pa
tte
rn
, I
I) 
Ca
ve
 p
as
sa
ge
 ce
nt
er
lin
e, 
II
I) 
Ca
ve
 p
as
sa
ge
s (
%
 o
f t
ot
al
 le
ng
th
), 
IV
) N
ea
re
st 
ne
ig
hb
or
 a
zi
m
ut
h 
fo
r d
ep
re
ss
io
n 
ce
nt
ro
id
s w
ith
in
 2
 k
m
 of
 th
e c
av
e c
en
te
rli
ne
 (f
re
qu
en
cy
) V
) S
ta
nd
ar
d 
D
ev
ia
tio
na
l E
lli
ps
es
 fo
r d
ep
re
ss
io
n 
ce
nt
ro
id
s w
ith
in
 1,
 2
, a
nd
 3
 k
m
 of
 ca
ve
 ce
nt
er
lin
es
, V
I) 
M
aj
or
 a
xe
s (
%
 of
 to
ta
l l
en
gt
h)
 
fo
r d
ep
re
ss
io
ns
 w
ith
 m
aj
or
 a
xi
s/m
in
or
 a
xi
s >
= 
1.
5 
w
ith
in
 2
 k
m
 o
f t
he
 ca
ve
 ce
nt
er
lin
e, 
V
II)
 M
aj
or
 a
xe
s (
%
 o
f t
ot
al
 le
ng
th
) f
or
 d
ep
re
ss
io
ns
 w
ith
 m
aj
or
 a
xi
s/m
in
or
 a
xi
s >
= 
1.
5 
w
ith
in
 3
 k
m
 o
f 
th
e c
av
e c
en
te
rli
ne
.
Standard Deviational Ellipses of karstic de-
pression centroids show wide variations in cir-
cularity. For example, major axis/minor axis ra-
tios of SDEs for depressions within 3 km of each 
cave ranges from 1.04 for Bonnet cave to 2.99 
for Cathedral-Falmouth cave system, indicating 
various degrees of spatial trends for the surficial 
karst features. Spatial distribution of depression 
centroids, as revealed by SDEs around cave cen-
terlines, show moderate to good alignment with 
the centerlines of most of the caves except for 
Blue Hole, Bonnet, Convict, Green, Indian, and 
Peacock. 
Along with a visual comparison of the 
rose diagrams for the depression major axes 
and the cave passages given in Figure 5, a sta-
tistical comparison was also tested using Kol-
mogorov-Smirnov (K-S) two-sample technique 
as applied by Barlow & Ogden (1982) (Table 2). 
The calculations, performed using the statis-
tical package XLSTAT and verıfıed by DGOF 
in R, allow us to infer if the two samples fol-
low the same distribution. In this process, cave 
passage and depression major axis azimuths 
were compared as percentages of total length 
in 10-degree azimuth intervals. Except for the 
Sally’s Ward and Madison Blue two-km data 
sets, all the cave alignments and surrounding 
major axes are interpreted to be following the 
same distribution as the calculated p values are 
greater than the significance level of 5% (alfa 
= 0.05). P values confirm high levels of cor-
relation between the cave passage and depres-
sion major axis azimuths especially for Bonnet, 
Convict, Cow, Hart, Indian, Leon, Little River, 
Luraville, Manatee, Peacock, Suwanacoochee, 
and Wakulla.
Fracture control in cave development in 
Florida was mentioned in various studies. Flo-
rea (2006) describes cave passages along frac-
tures with preferred orientations in WNW-ES 
(100-120) and NNE-SSW (20-40) in Brooksville 
Ridge, Florida. Similar fracture trends were re-
ported by Vernon (1951) and Culton (1978). In 
this study, field measurements of joint systems 
along the Suwannee River show two distinct pat-
terns (see the rose diagrams in Figure 6). While 
a NE/SW orientation is dominant along the up-
per Suwannee, fractures along the north-south 
stretch of the river trend in NW/SE orientation. 
These patterns generally conform with the cave 
development as well as the orientation of the Su-
wannee River’s course.
ACTA CARSOLOGICA 51/1 – 2022 43
CAN DENIZMAN
Karst development in Florida is controlled not only by 
fracture systems and hydraulic gradients, but also by the 
type of porosity and the pattern of recharge to the car-
bonate bedrock. Allogenic recharge from non-karst areas 
accounts for extensive dissolution along the retreating 
escarpment, resulting in sinking streams and collapsed 
depressions above and around cave conduits with crude 
branchwork patterns. This pattern of karst development 
is further complicated by diffuse recharge through per-
meable soil cover on the bedrock, resulting in extensive 
epikarst-initiated cave development. In both types of re-
charge patterns, a connection between surface – subsur-
face karst development, and the effect of fracture systems 
are clear. In this study, this connection was established by 
an analysis of large data sets on caves and sinkholes on a 
GIS platform. 
Table 2: Results of the Kolmogorov-Smirnov analysis between the 
cave passage and sinkhole major axis orientations as percent of to-
tal length in 10-degree intervals. Data sets that do not follow the 
same distribution with a significance level of 0.05 are in bold.
Cave Distance from the Cave D
p-value 
(Two-tailed)
Blue Hole
2 km 0.39 0.13
3 km 0.39 0.13
Bonnet 
2 km 0.28 0.49
3 km 0.17 0.96
Cathedral-
Falmouth
2 km 0.28 0.49
3 km 0.33 0.27
Convict
2 km 0.28 0.49
3 km 0.17 0.96
Cow
2 km 0.28 0.49
3 km 0.17 0.96
Green
2 km 0.39 0.13
3 km 0.33 0.27
Hart
2 km 0.22 0.76
3 km 0.22 0.76
Indian
2 km 0.33 0.27
3 km 0.17 0.96
Leon
2 km 0.22 0.77
3 km 0.33 0.27
Little River
2 km 0.22 0.76
3 km 0.22 0.76
Luraville
2 km 0.28 0.49
3 km 0.22 0.76
Madison Blue
2 km 0.44 0.05
3 km 0.33 0.28
Manatee
2 km 0.17 0.96
3 km 0.17 0.96
McBride
2 km 0.33 0.27
3 km 0.31 0.39
Morgan
2 km 0.28 0.49
3 km 0.33 0.27
Peacock
2 km 0.22 0.76
3 km 0.33 0.27
Rock Bluff
2 km 0.28 0.49
3 km 0.33 0.27
Sally’s Ward
2 km 0.5 0.02
3 km 0.44 0.04
Shepard
2 km 0.28 0.49
3 km 0.33 0.27
Suwanacoochee
2 km 0.22 0.76
3 km 0.17 0.96
Telford
2 km 0.44 0.06
3 km 0.33 0.27
Wakulla
2 km 0.22 0.76
3 km 0.16 0.6
Figure 6: Fractures from Vernon (1951) and the rose diagrams for 
fracture azimuths along the Suwannee River. Ellipses show general 
areas of data collection.
ACTA CARSOLOGICA 51/1 – 202244
CAVES, SINKHOLES, AND FRACTURES IN THE EOGENETIC KARST OF FLORIDA, A GIS-BASED SPATIAL ANALYSIS
6. SUMMARY
In order to explore the connection between surficial and 
subsurface karst development in Florida, spatial corre-
lation of twenty-two cave passages and a large number 
of surrounding karstic depressions was carried out in a 
GIS-based spatial analysis (see Figure7 for an example 
of cave passage centerlines and the surrounding karstic 
depressions). Morphometric parameters evaluated in 
this study included cave passage orientations, depression 
major axis orientations, nearest neighbor depression di-
rections, as well as spatial distribution patterns of depres-
sions. Analysis of the data provides further support for 
the connection between the surficial and subsurface karst 
processes in the eogenetic karst of Florida. It also appears 
that the structural control plays an important role in 
karst development.
Figure 7: Leon Cave system and the surrounding karstic depres-
sions.
REFERENCES
Angel, J., Nelson, D., Panno, S., 2004. Comparison of a 
new GIS-based technique and a manual method for 
determining sinkhole density: an example from Illi-
noi’s sinkhole plain. Journal of Cave and Karst Stud-
ies, 66(1): 9-17.
Barlow, C., Ogden, A., 1982. A statistical comparison of 
joint, straight cave segment, and photo-lineament 
orientations. NSS Bulletin, 44: 107-110.
Culton, R., 1978. The subsurface geology of Hamilton 
County, Florida, with emphasis on the Oligocene 
age Suwannee Limestone. [MSc Thesis]. Florida 
State University, Tallahassee, 185 pp.
Denizman, C., Randazzo, A., 2000. Post-Miocene sub-
tropical karst evolution, lower Suwannee River ba-
sin, Florida. GSA Bulletin, 112(12): 1804-1813.
Denizman, C., 2003. Morphometric and spatial distribu-
tion of karstic depressions, Lower Suwannee River 
Basin, Florida. Journal of Cave and Karst Studies, 
65(1): 29-35.
Faivre, S., Pahernik, M., 2007. Structural influences on 
the spatial distribution of dolines, Island of Brac, 
Croatia. Zeitschrift für Geomorphologie, 51(4): 
487-503. 
Florea, L., 2006. Architecture of air-filled caves within the 
karst of the Brooksville Ridge, west-central Florida. 
Journal of Cave and Karst Studies, 68(2): 64-75.
Florea, L., Vacher, H., 2006. Cave levels, marine terraces, 
paleoshorelines, and the water table in Peninsular 
Florida. Archives of Climate Change in Karst, Karst 
Waters Institute Special Publication 10, 188-192 pp.
Ford, D., 1964. Origin of closed depressions in the cen-
tral Mendip Hills, Somerset, England. In: Hamilton, 
F.E.I. (Eds.), Abstract of Papers, 20th International 
Geographical Congress, 6th July – 20th August 1964, 
London, pp. 105-106.
Gulley, J., Martin, J., Moore, P.J., Murphy, J., 2013. Forma-
tion of phreatic caves in and eogenetic karst aqui-
fer by CO2 enrichment at lower water tables and 
ACTA CARSOLOGICA 51/1 – 2022 45
CAN DENIZMAN
subsequent flooding by sea-level rise. Earth Surface 
Processes and Landforms, 38(1): 1210-1224. https://
doi.org/10.1002/esp.3358
Kemmerly, P., 1982. Spatial analysis of a karst depression 
population: Clues to genesis. GSA Bulletin, 93(11): 
1078–1086.
Kincaid, T., 1998. River water intrusion to the uncon-
fined Floridan aquifer. Environmental and Engi-
neering Geosciences, 4: 361-374.
Komac, M., Urbanc, J., 2012. Assessment of spatial 
properties of karst areas on a regional scale using 
GIS and statistics – the case of Slovenia. Journal of 
Cave and Karst Studies, 74(3): 261-261. https://doi.
org/10.4311/2010ES0188R 
La Valle, P., 1967. Some aspects of linear karst devel-
opment in South Central Kentucky. Annals of 
the Association of American Geographers, 57(1): 
49-71. https://doi.org/10.1111/j.1467-8306.1967.
tb00590.x
Lyew-Ayee, P., Viles, H.A., Tucker, G.E., 2006. The use of 
GIS-based digital morphometric techniques in the 
study of cockpit karst.- Earth Surface Processes and 
Landforms, 32(2): 165-179. https://doi.org/10.1002/
esp.1399
Mitchell, A., 2005. The ESRI guide to GIS analysis. Vol-
ume: 2 spatial measurements and Statistics. ESRI 
press, Redlands, California, 238 pp.
Moore, P., Martin, J., Screaton, E., Neuhoff, P., 2010. Con-
duit enlargement in eogenetic karst aquifer. Jour-
nal of Hydrology, 376(3-4): 143 – 155. https://doi.
org/10.1016/j.jhydrol.2010.08.008
Öztürk, M., Şener, M., Şimşek, M., 2018. Structural 
controls on distribution of dolines on Mount Ana-
mas (Taurus Mountains, Turkey). Geomorpho-
logy, 317: 107-116. https://doi.org/10.1016/j.geo-
morph.2018.05.023
Palmer, A., 1991. Origin and morphology of limestone 
caves. Geological Society of America Bulletin, 103: 
1-21.
Palmer, A., 2009. Cave Geology. Cave Books, Dayton, 
454 pp.
Palmer, M., Palmer, A., 1975. Landform development in 
the Mitchel Plain of southern Indiana: Origin of a 
partially karstified plain. Zeitschrift für Geomor-
phologie, 19: 1-39.
Randazzo, A, 1997. The sedimentary platform of Florida: 
Mesozoic to Cenozoic. In: Randazzo, A., Jones, D. 
(Eds.), Geology of Florida. University Press of Flor-
ida, Gainesville, Florida, pp. 39-56.
Shofner, G., Mills, H., Jason, E., 2001. A simple map in-
dex of karstification and its relationship to sinkhole 
and cave distribution in Tennessee. Journal of Cave 
and Karst Studies, 63(2): 67-75.
Scott, T.M., 1997. Miocene to Holocene history of Flor-
ida. In: Randazzo, A., Jones, D. (Eds.), Geology of 
Florida. University Press of Florida, Gainesville, 
Florida, pp. 57-68.
Upchurch, S., S. Thomas., M. Alfieri, B., Fratesi, Dobecki, 
T., 2019. The Karst Systems of Florida, Understand-
ing Karst in a Geologically Young Terrain. Springer, 
New York, 450 pp.
Vacher, H., Mylroie, J., 2002. Eogenetic karst from the 
perspective of an equivalent porous medium. Car-
bonates and Evaporites, 17: 182-196.
Vernon, R., 1951. Geology of Citrus and Levy Counties, 
Florida. Florida Geological Survey Bulletin, 3, Flor-
ida Geological Survey, Tallahassee, Florida, 256 pp.
White, W., 1970. The Geomorphology of the Florida Pen-
insula. Florida Geological Survey Bulletin, 51, Flor-
ida Geological Survey, Tallahassee, Florida, 164 pp.
ACTA CARSOLOGICA 51/1 – 202246