Karst Geology of Aguijan and Tinian,
CNMI
Cave Inventory and
Structural Analysis of Development
Kevin W. Stafford, John E. Mylroie
& John W. Jenson
Department of Geosciences
Mississippi State University
Mississippi State, MS 39762
Water and Environmental Research Institute
of the Western Pacific
University of Guam
Mangilao, Guam 96923
Technical Report No. 106
Water & Environmental Research Institute
of the Western Pacific
University of Guam
September, 2004
Karst Geology of Aguijan and Tinian, CNMI
Cave Inventory and
Structural Analysis of Development
Kevin W. Stafford, John E. Mylroie
& John W. Jenson
Department of Geosciences
Mississippi State University
Mississippi State, MS 39762
Water and Environmental Research Institute
of the Western Pacific
University of Guam
Mangilao, Guam 96923
Technical Report No. 106
Water & Environmental Research Institute
of the Western Pacific
University of Guam
September, 2004
THE WATER RESOURCES RESEARCH INSTITUTE PROGRAM OF THE US GEOLOGICAL SURVEY, AWARD NO. 01HQPA0010,
SUPPORTED THE WORK REPORTED HERE. THE CONTENT OF THIS REPORT DOES NOT NECESSARILY REFLECT THE
VIEWS AND POLICIES OF THE DEPPARTMENT OF THE INTERIOR, NOR DOES THE MENTION OF TRADE NAMES OR
COMMERCIAL PRODUCTS CONSTITUTE THEIR ENDORSEMENT BY THE UNITED STATES GOVERNMENT.
2
ABSTRACT
Tinian and Aguijan, Commonwealth of the Northern Mariana Islands (CNMI), are
volcanic, back arc islands in the western Pacific formed by Pacific Plate subduction under
the Philippine Plate. The islands are composed of Eocene volcanic cores mantled by
Plio-Pleistocene carbonate facies and raised Holocene beach and reef deposits. The
entire sequence has been tectonically uplifted and contains high-angle normal faults,
while isostatic subsidence and scarp failures overprint tectonic brittle failure features.
A cave and karst inventory on Tinian and Aguijan surveyed 114 features and is
believed to adequately represent the megaporosity (cave) development. Two distinct
cave classes were identified: mixing zone caves (flank margin caves and banana holes)
and fissure caves. Most mixing zone caves were located in or near scarps and coastlines,
often at similar elevations to nearby caves. Fissure caves were located in regions of
brittle failure, forming linear features with narrow widths. Three previous sea-level
positions were identified based on horizons of mixing zone caves. Seventeen freshwater
discharge sites and four allogenic recharge sites were identified on Tinian.
Kolmogorov-Smirnov statistical analyses and rose diagram comparisons of
orientation trends found significant similarities between megaporosity and geologic
structure (brittle failure) on Tinian. Analyses of small regions showed distinct relations
between brittle deformation and megaporosity, while at larger scales similarities became
less obvious due to the complex geologic history and physiography of the island. Based
on similarities in populations of orientation trends, fissure cave development is primarily
controlled by brittle failure deformation with development along faults, fractures, and
joints, while mixing zone cave development is primarily controlled by fresh-water lens
position but significantly influenced by brittle failure deformation.
Tinian and Aguijan do not fit neatly into one classification of the Carbonate
Island Karst Model. Regions of Tinian best fit the Simple, Carbonate-Cover and
Composite Island Karst Models, but none easily fit the entire island. Aguijan must be
classified as a Simple Carbonate Island because no geologic data has proved the presence
of non-carbonate rocks interfering with the fresh-water lens, however it is probable that
Aguijan does contain basement rocks that extend above sea-level as on other carbonate
islands in the Marianas.
i
PREFACE
WERI (Water and Environmental Institute of the Western Pacific) Tech Report #96 was
published as a preliminary assessment of the karst development and water resources of Tinian and
Rota, CNMI (Commonwealth of the Northern Mariana Islands), based on a field reconnaissance
conducted on the islands in June 2002. Since that time, work in the CNMI has focused on the
Islands of Aguijan, Rota and Tinian with a primary emphasis on the cave and karst inventories of
these islands, which can be used to evaluate the hydrolgeologic evolution of the islands.
The work presented in this study provides a detailed cave and karst inventory of the
Islands of Aguijan and Tinian, CNMI with an analysis of structural controls on cave and karst
development on the island of Tinian. The authors conducted this work as thesis research at
Mississippi State University in conjunction with the University of Guam, while the US
Geological Survey, through the National Institutes for Water Resources Research program, award
no. 01HQGR0134, funded fieldwork. The data and findings in this report were originally
published as a master’s thesis by Kevin Stafford, titled: Structural Controls on Megaporosity in
Eogenetic Carbonate Rocks: Tinian, CNMI. Ultimately, the work presented herein was made
possible through the support of the Municipality of Tinian and Aguijan through: Tinian Mayors
Office, CNMI Department of Land and Natural Resources, CNMI Department of Historical
Preservation, Commonwealth Utilities Corporation, and Northern Marianas College.
Due to the large volume of data in this report, the appendices (Appendix A: Color
Figures; Appendix B: Cave and Karst Inventory; Appendix C: Orientation Data; and Appendix D:
Statistical Comparisons) are published as .pdf files in the attached CD or can be downloaded from
the WERI website (www.uog.edu/weri). These appendices include color images of significant
features, detailed maps of the inventoried karst features on Aguijan and Tinian, diagrams
representing the structural geology of Tinian, and data matrices of statistical analyses.
Additional research continues in the Mariana Islands, in order to further answer questions
about the karst geology and water resources of the islands. A similar study to the work reported
herein is currently underway by the authors and master’s thesis research by T. Montgomery Keel
at Mississippi State University. Ultimately this study and future studies will be used to evaluate
the fresh-water lens development on carbonate islands, which can be used to better manage the
water resources of these communities.
ii
CONTENTS
ABSTRACT ..........................................................................................................................
i
PREFACE .............................................................................................................................
ii
CONTENTS .......................................................................................................................... iii
LIST OF TABLES ................................................................................................................
v
LIST OF FIGURES............................................................................................................... vi
INTRODUCTION.................................................................................................................
1
LITERATURE REVIEW......................................................................................................
2
Geographical and Geological Setting .............................................................................
Historical Setting ...........................................................................................................
Carbonate Island Karst....................................................................................................
Karst Features .................................................................................................................
Epikarst .................................................................................................................
Closed Depression.................................................................................................
Caves.....................................................................................................................
Discharge Features................................................................................................
Water Resources .............................................................................................................
2
5
6
10
10
11
12
15
15
STUDY METHODOLOGY.................................................................................................. 17
Initial Site Investigation..................................................................................................
Data Collection ...............................................................................................................
Data Reduction ...............................................................................................................
Statistical Comparison of Data .......................................................................................
Small-Scale Test Site Evaluation....................................................................................
17
18
19
22
23
STUDY RESULTS ............................................................................................................... 23
Cave and Karst Inventory ...............................................................................................
Cave Orientations............................................................................................................
Brittle Deformation.........................................................................................................
Scarps and Coastlines .....................................................................................................
Rose Diagrams................................................................................................................
Statistical Comparison ....................................................................................................
Tinian Composite..................................................................................................
Central Plateau ......................................................................................................
Northern Lowland .................................................................................................
North-Central Highland ........................................................................................
Median Valley.......................................................................................................
Southeastern Ridge................................................................................................
Small-Scale Test Sites.....................................................................................................
Carolinas Limestone Forest...................................................................................
Puntan Diapblo......................................................................................................
Unai Dangkolo ......................................................................................................
iii
23
24
26
26
27
28
28
29
29
30
30
30
31
31
31
31
CONTENTS (cont.)
DISCUSSION AND CONCLUSIONS ................................................................................. 32
Tinian Cave and Karst Inventory.......................................................................
Aguijan Cave and Karst Inventory ....................................................................
Controls on Cave and Karst Development.........................................................
Island Scale Comparisons..................................................................................
Province Scale Comparisons..............................................................................
Small-Scale Test Site Comparisons ...................................................................
Structural Control of Caves ...............................................................................
Karst Development on Aguijan and Tinian .......................................................
32
35
35
38
38
39
40
40
SUMMARY .......................................................................................................................... 41
REFERENCES CITED ......................................................................................................... 44
APPENDIX (attached CD)
A.
B.
C.
D.
Color Figures......................................................................................................... 48
Cave and Karst Inventory: Maps and Descriptions.............................................. 63
Orientation Data.................................................................................................... 162
Statistical Comparison .......................................................................................... 248
iv
LIST OF TABLES
Table
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Page
Cave and karst features surveyed on Tinian, CNMI: UTM location,
physiography, cave type and geology ..................................................................... 65
Cave and karst features surveyed on Aguijan, CNMI: UTM location,
physiography, and cave type ................................................................................... 67
Fissure cave primary orientations ......................................................................... 162
Fissure cave segment orientations with segment length ....................................... 164
Mixing zone cave primary orientations................................................................. 168
Mixing zone cave segment orientations with segment length............................... 170
Mixing zone cave, entrance width segment orientations with segment length ..... 178
Mixing zone cave, penetration segment orientations with segment length........... 182
Mixing zone cave, maximum width segment orientations with segment
length..................................................................................................................... 186
Fault orientations................................................................................................... 190
Joint orientations ................................................................................................... 195
Fracture orientations measured during fieldwork ................................................. 197
Inland scarp segment orientations with segment length........................................ 200
Coastal scarp segment orientations with segment length...................................... 205
Coastline segment orientations with segment length ............................................ 213
Legend for column and row headings used in statistical comparison data
matrices ................................................................................................................. 324
Tinian Composite statistical comparison data matrix ........................................... 325
Central Plateau statistical comparison data matrix ............................................... 326
Median Valley statistical comparison data matrix ................................................ 327
Northern Lowland statistical comparison data matrix .......................................... 328
North-Central Highland statistical comparison data matrix.................................. 329
Southeastern Ridge statistical comparison data matrix......................................... 330
Carolina's Limestone Forest statistical comparison data matrix ........................... 331
Puntan Diapblo statistical comparison data matrix ............................................... 332
Unai Dangkolo statistical comparison data matrix ............................................... 333
v
LIST OF FIGURES
Figure
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Page
Location map of the Mariana Islands...................................................................... 2
Physiographic map of Tinian, CNMI...................................................................... 3
Physiographic map of Aguijan................................................................................ 4
Geology section....................................................................................................... 4
Carbonate Island Karst Model ................................................................................ 7
Schematic diagram illustrating the relationship between island size,
basement relationships to sea level and surface, and relative sea level within
the Carbonate Island Model .................................................................................... 9
Conceptual model of cave types that form in eogenetic rocks on carbonate
islands..................................................................................................................... 13
Model showing the fresh-water lens morphology showing the location of
basal and parabasal waters and the Ghyben-Herzberg principle............................ 16
Example of primary and segment orientation trends measured using the
apparent trend method for a typical fissure cave.................................................... 20
Example of primary and segment orientation trends measured using the
apparent trend method for a typical mixing zone cave .......................................... 20
Example of entrance, maximum width, and penetration trends measured
using the entrance width trend method for a typical mixing zone cave ................. 20
Flank margin caves develop more complicated morphologies as they grow
in size ..................................................................................................................... 32
Conceptual model for the growth of flank margin caves ....................................... 33
Horizons of flank margin cave development show previous fresh-water lens
positions ................................................................................................................. 34
Diagram showing the relationship between cave widths and lengths, which
represent two distinct populations for fissure caves and mixing zone caves ......... 36
Diagram showing the relationship cave entrance width and cave maximum
width....................................................................................................................... 36
Coastline orientations for Tinian show a wide range of trends because of the
elliptical shape of the shape of the island............................................................... 37
Map of Unai Dangkolo region showing the close proximity of flank margin
cave and fissure cave development ........................................................................ 43
Inland and coastal scarps on Tinian ....................................................................... 48
Areas of potential allogenic recharge..................................................................... 49
AMCS standard cave symbology........................................................................... 50
Location of test site areas....................................................................................... 51
Location of cave and karst features surveyed on Aguijan and Tinian ................... 52
Fresh-water discharge sites located on Tinian ....................................................... 53
Bamboo growing in the North Lemmai Recharge Feature with vines coating
the scarp that forms the non-carbonate / carbonate contact ................................... 54
Ponded water in the South Lemmai Recharge ....................................................... 54
Location of closed depression on Tinian................................................................ 55
Active quarry on Tinian ......................................................................................... 56
Fresh-water at the land surface at Hagoi in the Northern Lowland ....................... 56
Faults and non-carbonate rock outcrops................................................................. 57
South Unai Dangkolo represents a typical cove..................................................... 58
Hidden Beach Cave demonstrates well the transition from flank margin cave
to cove resulting from coastal erosion.................................................................... 58
vi
LIST OF FIGURES (cont.)
Figure
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
Page
Typical flank margin cave morphology .................................................................
Multiple levels of mixing zone dissolution exist on Tinian with at least three
identified in the North-Central Highland near Mount Lasu...................................
Fissure caves form narrow, linear passages that appear to be developed
along zones of brittle failure ..................................................................................
Flowstone deposits on ceilings and walls ..............................................................
Insect Bat Cave on Aguijan represents a paleo-discharge feature .........................
Scallops on the ceiling and walls of Liyang Atkiya, Aguijan................................
Diagram showing the location of modern carbonate beach deposits and
primary brittle failure types....................................................................................
Map of "600 Meter" Fracture System ....................................................................
Map of Almost Cave ..............................................................................................
Map of Andyland Cave ..........................................................................................
Map of Anvil Cave.................................................................................................
Map of Barcinas East Cave ....................................................................................
Map of Barcinas West Cave...................................................................................
Map of Barely Cave ...............................................................................................
Map of Bee Hooch Cave ........................................................................................
Map of Biting Mosquitoes Cave ............................................................................
Map of Body Repel Cave.......................................................................................
Map of Boonie Bee Sink........................................................................................
Map of Broken Stal Cave.......................................................................................
Map of Cabrito Cave..............................................................................................
Map of Cannon Cave .............................................................................................
Map of Carolinas Fracture Cave ............................................................................
Map of Cave Without a Cave.................................................................................
Map of Cave Without a Roof .................................................................................
Map of Cavelet Cave..............................................................................................
Map of Central Mendiola Cave Complex ..............................................................
Map of Cetacean Cave ...........................................................................................
Map of Chiget Fracture ..........................................................................................
Map of Cobble Cave ..............................................................................................
Map of Coconut Trap Cave....................................................................................
Map of Command Post Cave Complex..................................................................
Map of Cowrie Cave ..............................................................................................
Map of CUC Cave..................................................................................................
Map of Danko's Misery..........................................................................................
Map of Death Fracture Complex............................................................................
Map of Diamond Cave ...........................................................................................
Map of Dos Cenotes Cave......................................................................................
Map of Dos Sakis Cave Complex ..........................................................................
Map of Dove Cave .................................................................................................
Map of Dripping Tree Fracture Cave.....................................................................
Map of Dump Coke Cave ......................................................................................
Map of Dynasty Cave.............................................................................................
Map of East Suicide Cliff Cave .............................................................................
Map of Edwin's Ranch Cave ..................................................................................
vii
59
59
60
60
61
61
62
68
69
69
70
71
72
72
73
74
74
75
76
76
77
78
79
79
80
81
82
83
84
84
85
86
86
87
88
89
90
91
91
93
94
95
96
96
LIST OF FIGURES (cont.)
Figure
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
Page
Map of Elevator Cave ............................................................................................ 96
Map of False Floor Cave........................................................................................ 97
Map of Five Bee Cave Complex............................................................................ 98
Map of Flamingo Tail Caves.................................................................................. 99
Map of Fleming Point Cave .................................................................................. 100
Map of Full Bottle Cave........................................................................................ 101
Map of Gecko Cave .............................................................................................. 101
Map of Goat Cave ................................................................................................. 102
Map of Goat Fracture Cave................................................................................... 103
Map of Half-Dozen Cave ...................................................................................... 103
Map of Headless Tourist Pit.................................................................................. 104
Map of Hermit Crab Cave..................................................................................... 105
Map of Hidden Beach Cave .................................................................................. 106
Map of Hollow Column Cave ............................................................................... 106
Map of Insect Bat Cave......................................................................................... 107
Map of Isotope Cave ............................................................................................. 108
Map of John's Small Cave..................................................................................... 109
Map of Lasu Recharge Cave ................................................................................. 110
Map of Leprosy Caves .......................................................................................... 111
Map of Leprosy Discharge Feature....................................................................... 111
Map of "600 Meter" Fracture System ................................................................... 113
Map of Liyang Barangka ...................................................................................... 114
Map of Liyang Dangkolo ...................................................................................... 115
Map of Liyang Diapblo......................................................................................... 116
Map of Liyang Gntot............................................................................................. 117
Map of Liyang Lomuk .......................................................................................... 118
Map of Liyang Mohlang ....................................................................................... 119
Map of Liyang Popporput ..................................................................................... 120
Map of Liyang Sampapa ....................................................................................... 120
Map of Liyang Umumu......................................................................................... 121
Map of Lizard Cave .............................................................................................. 121
Map of Lower Suicide Cliff Cave Complex ......................................................... 123
Map of Masalok Fracture Cave Complex ............................................................. 124
Map of Mendiola Arch Cave Complex ................................................................. 125
Map of Metal Door Cave ...................................................................................... 126
Map of Metal Spike Cave Complex...................................................................... 127
Map of Metal Stretcher Cave ................................................................................ 128
Map of Modified Cave .......................................................................................... 129
Map of Monica Wants to be Like Kevin Cave...................................................... 129
Map of Natural Arch Cave.................................................................................... 130
Map of North Unai Dangkolo ............................................................................... 132
Map of Northern Playground Cave ....................................................................... 133
Map of Nuestra Señora de Santo Lourdes Cave Complex.................................... 134
Map of Orange Cave ............................................................................................. 135
Map of Orphan Kids Cave Complex..................................................................... 136
Map of Pebble Cave.............................................................................................. 137
Map of Pepper Cave.............................................................................................. 137
Map of Piña Cave Complex .................................................................................. 138
viii
LIST OF FIGURES (cont.)
Figure
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
Page
Map of Playground Cave ...................................................................................... 138
Map of Plunder Cave ............................................................................................ 139
Map of Radio Inactive Cave ................................................................................. 140
Map of Red Snapper Cave .................................................................................... 141
Map of Rock Hammer Cave ................................................................................. 141
Map of Rogue Cave .............................................................................................. 143
Map of Rootcicle Cave.......................................................................................... 143
Map of Scorpion Cave .......................................................................................... 144
Map of Screaming Bat Cave ................................................................................. 144
Map of Skip Jack Cave ......................................................................................... 145
Map of Skull Cave Complex................................................................................. 146
Map of Skylight Cave ........................................................................................... 147
Map of Solitary Cave ............................................................................................ 147
Map of South Mendiola Cave ............................................................................... 148
Map of South Unai Dangkolo ............................................................................... 149
Map of Spider Cave .............................................................................................. 149
Map of Swarming Termites Cave ......................................................................... 150
Map of Swiftlet Cave ............................................................................................ 151
Map of Swimming Hole Cave Complex............................................................... 152
Map of Toppled Column Cave.............................................................................. 153
Map of Tridactid Cave Complex........................................................................... 152
Map of Twin Ascent Caves................................................................................... 155
Map of Unai Chiget............................................................................................... 156
Map of Unai Lamlam ............................................................................................ 156
Map of Unai Masalok............................................................................................ 157
Map of Water Cave ............................................................................................... 158
Map of Waypoint Cave ......................................................................................... 159
Map of West Lasu Depression .............................................................................. 160
Map of West Suicide Cliff Cave Complex............................................................ 161
Rose diagrams of fissure cave primary orientations ............................................. 163
Rose diagrams of fissure cave, five-meter segment orientations .......................... 166
Rose diagrams of fissure cave, ten-meter segment orientations ........................... 167
Rose diagrams of mixing zone cave, primary orientations ................................... 169
Rose diagrams of mixing zone cave, five-meter segment orientations................. 173
Rose diagrams of mixing zone cave, ten-meter segment orientations .................. 174
Rose diagrams of composite cave, primary orientations....................................... 175
Rose diagrams of composite cave, five-meter segment orientations .................... 176
Rose diagrams of composite cave, ten-meter segment orientations...................... 177
Rose diagrams of mixing zone cave entrance width orientations ......................... 179
Rose diagrams of mixing zone cave entrance width, five-meter segment
orientations............................................................................................................ 180
Rose diagrams of mixing zone cave entrance width, ten-meter segment
orientations............................................................................................................ 181
Rose diagrams of mixing zone cave penetration orientations............................... 183
Rose diagrams of mixing zone cave penetration, five-meter segment
orientations............................................................................................................ 184
ix
LIST OF FIGURES (cont.)
Figure
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
Page
Rose diagrams of mixing zone cave penetration, ten-meter segment
orientations............................................................................................................ 185
Rose diagrams of mixing zone cave maximum width orientations....................... 187
Rose diagrams of mixing zone cave maximum width, five-meter segment
orientations............................................................................................................ 188
Rose diagrams of mixing zone cave maximum width, ten-meter segment
orientations............................................................................................................ 189
Rose diagrams of fault orientations....................................................................... 192
Rose diagrams of fifty-meter, fault segment orientations ..................................... 193
Rose diagrams of one hundred-meter, fault segment orientations ........................ 194
Rose diagrams of joint orientations....................................................................... 196
Rose diagrams of orientations of fractures measured during fieldwork ............... 199
Rose diagrams of inland scarp orientations........................................................... 202
Rose diagrams of inland scarp, fifty-meter segment orientations ......................... 203
Rose diagrams of inland scarp, one hundred-meter segment orientations ............ 204
Rose diagrams of coastal scarp orientations ......................................................... 207
Rose diagrams of coastal scarp, fifty-meter segment orientations........................ 208
Rose diagrams of coastal scarp, one hundred-meter segment orientations........... 209
Rose diagrams of composite scarp orientations .................................................... 210
Rose diagrams of composite scarp, fifty-meter segment orientations................... 211
Rose diagrams of composite scarp, one hundred-meter segment orientations...... 212
Rose diagrams of coastline orientations................................................................ 215
Rose diagrams of coastline, fifty-meter segment orientations .............................. 216
Rose diagrams of coastline, one hundred-meter segment orientations..................217
x
KARST GEOLOGY OF AGUIJAN AND TINIAN, CNMI
Cave Inventory and Structural Analysis of Development
INTRODUCTION
This study has three primary objectives:
1) conduct a cave and karst inventory on
Tinian and Aguijan, because no inventory
existed, 2) investigate the cave and karst
development on Tinian to determine if
influences on mega-porosity development
are associated with brittle deformation in
eogenetic rocks, and 3) evaluate of the
islands of Tinian and Aguijan in relation to
the Carbonate Island Karst Model (Jenson et
al., 2002; Mylroie and Jenson, 2002), in
order to further advance the understanding
of eogenetic karst development on carbonate
islands.
The Tinian municipality of the
Commonwealth of the Northern Mariana
Islands (CNMI) governs Tinian and
Aguijan. The Mariana Islands are a
volcanic, back arc island chain in the
western Pacific formed by Pacific Plate
subduction under the Philippine Plate along
the Mariana Trench. Aguijan covers 7.2
square kilometers and Tinian covers 102
square kilometers. A geologic study was
conducted on Tinian in 1960 (Doan et al.,
1960), which described the island as an
Eocene volcanic core mantled by PlioPleistocene coralliferous and algal carbonate
facies and raised Holocene beach and reef
deposits. The entire sequence has been
uplifted and contains high-angle normal
faults produced from arc tectonism. Tinian
has experienced greater than 100 meters of
uplift since the Pleistocene, with an
estimated 1.8 meters of uplift in the
Holocene (Dickenson, 1999). Aguijan has
only been studied briefly (Tayama, 1936)
and no geologic map has been produced for
the island; however, it is presumed to have a
depositional and tectonic history similar to
Tinian based on its proximity (~9 kilometers
southwest of Tinian).
The Carbonate Island Karst Model
predicts cave and karst development in
eogenetic limestone (Jenson et al., 2002;
Mylroie and Jenson, 2002; Mylroie and
Jenson, 2001). It provides a fundamental
and systematic framework for describing
hydrogeologic karst evolution in young
carbonate rocks, by incorporating the effects
of mixing zone dissolution, glacio-eustacy,
tectonism, and lithologic variations. In this
model distinctive cave morphologies
develop as eogenetic karst. These
distinctive morphologies include flank
margin caves, banana holes, pit caves, and
stream caves. Traditionally, the affects of
structural and lithologic controls on
eogenetic karst development have been
greatly overlooked because the original
model was developed to describe features
observed in the structurally and
lithologically simple islands of the Bahamas
(Mylroie and Carew, 1995a).
Structural control of karst development
is common in continental settings, where the
rock exhibits low porosity and is highly
fractured (Klimchouk and Ford, 2000).
Eogenetic karst is generally associated with
caves formed by mixing zone dissolution,
primarily independent of structural controls,
in rocks that have never been buried beyond
the range of meteoric diagenesis and retain
high primary and syngenetic porosities.
Recently, Jenson and coworkers (2002)
recognized the importance of structural and
lithologic controls on carbonate island karst
and modified the Carbonate Island Karst
Model to account for the effects of
deformation and lithology on dissolution
that occurs on carbonate islands that are or
were tectonically active and/or have intricate
interfingering of carbonate and noncarbonate rocks.
The cave and karst inventory of this
study was conducted in two intensive field
seasons based on information gathered
during a reconnaissance survey in June 2002
1
(Stafford et al., 2002). During fieldwork,
caves and karst features were documented,
surveyed and classified by cave type. Sites
of fresh-water discharge and allogenic
recharge were located. Orientations of
brittle failure structures were measured.
After fieldwork was complete, maps were
produced of the cave and karst features that
were surveyed. Maps of features on Tinian
were analyzed to determine cave orientation.
The cave orientation data was compared to
fault orientations, scarp orientations, and
coastline orientations for similarity. Data
comparisons were evaluated as independent
data pairs both visually (rose diagrams) and
statistically (Kolmogorov-Smirnov 2-sample
test). Using similar techniques, Nelson
(1988) and Barlow and Ogden (1992) have
shown that continental, telogenetic karst
development is related to the orientations of
regional faults, fractures, and lineaments,
implying that cave development is primarily
controlled by brittle deformation.
Evaluation of the cave and karst features
on Tinian and Aguijan allows them to be
evaluated with the Carbonate Island Karst
Model. However, the lack of previous
geologic studies on Aguijan and the
complex tectonic history of Tinian do not
allow for simple classification of these
islands.
LITERATURE REVIEW
maximum elevation of 157 meters (Figure
3). Both islands have wet-dry tropical
climates with a distinct rainy season (JulySeptember) and dry season (FebruaryMarch). Annual rainfall averages 200
centimeters and temperature ranges from 20
o
to 32o Celsius (Gingerich and Yeatts, 2000;
Butler, 1992; Tracey et al., 1964; Doan et
al., 1960; Cloud et al., 1956).
GEOGRAPHICAL AND
GEOLOGICAL SETTING
The Mariana Islands are located in the
western Pacific Ocean and comprise a total
of 17 islands (Figure 1). Guam is the largest
and southern-most island in the Marianas
and the only one that is not politically
affiliated with the Commonwealth of the
Northern Mariana Islands (CNMI). The
relative position of the islands north of
Guam in order of increasing distance is:
Rota, Aguijan, Tinian, Saipan and
Medinilla, which are carbonate islands. The
remainder of the island chain is volcanic
(Cloud et al., 1956). Tinian is located
approximately 3000 kilometers east of Asia,
approximately 180 kilometers north,
northwest of Guam, and approximately 10
kilometers south of Saipan, while Aguijan is
located approximately 9 kilometers
southwest of Tinian. Tinian (Latitude:
15.01oN, Longitude: 145.62oE) has a surface
area of 102 square kilometers with 51.2
kilometers of coastline and a maximum
elevation of 187 meters (Figure 2). Aguijan
(Latitude: 14.85oN, Longitude: 145.57oE)
has a surface area of 7.2 square kilometers
with 12.4 kilometers of coastline and a
Figure 1: Location map of carbonate
islands in the Marianas.
2
The Mariana Ridge, on which the
islands of Tinian and Aguijan are located, is
formed along a volcanic arc located
approximately 160 kilometers west of the
Mariana Trench, which is the world's
deepest trench with a maximum depth of
11,035 meters (Gross, 1982). This
subduction zone is created at the
convergence of the Pacific Plate to the east
and the Philippine Plate to the west. The
Mariana Islands are situated on an older
island arc that is separated by the Mariana
Trough from the younger Mariana West
Ridge, approximately 300 kilometers to the
west, which is developing in the Mariana
back-arc basin. This shift in ridge
development was initiated
by a change in Pacific Plate
subduction geometry
approximately 43 million
years ago (mya) (Dickinson,
1999; Mink and Vacher,
1997; Reagan and Miejer,
1984).
Doan and coworkers
(1960) used topography and
spatial relationships to
divide Tinian into five
physiographic regions:
Northern Lowland, NorthCentral Highland, Central
Plateau, Median Valley and
Southeastern Ridge (Figure
2). The Northern Lowland
comprises the broad, flat,
nearly horizontal surface
that slopes gently upward
from the west coast to
Sabanettan Chiget. Located
above the Central plateau
and midway between the
east and west coasts is the
North-Central Highland,
which contains the highest
point (162 meters) in
northern Tinian at Mount
Lasu. The Central Plateau
includes the central portion
of the island and is isolated
by steep slopes and
bounding scarps associated
with north-south faults. In
the south and east-central
regions, the Median Valley
expresses little relief, but
forms a broad depression
Figure 2: Physiographic map of Tinian, CNMI, with important
bounded by faults. The
features and locations identified (adapted from Doan et al., 1960).
Southeastern Ridge, which
3
includes Kastiyu, the highest point on Tinian
(187 meters), is developed on two principal
fault blocks (Doan et al., 1960).
Because little geologic work has been
performed on Aguijan (Butler, 1992;
Tayama, 1936), the following physiographic
provinces are proposed based on the
classification system used on Rota, which
exhibits a similar terraced topography
(Sugiwara, 1934): Upper Terrace, defined
by elevations greater than 100 meters, which
form a broad, relatively flat plateau and
reaches a maximum elevation of 157 meters;
Middle Terrace, defined by elevations
between 50 and 100 meters, which is nearly
absent on the northern side of the island and
best developed on the southeastern side; and
Lower Terrace, defined by elevations less
than 50 meters, which includes the steep
cliffs that form the coastline and the lowest
bench. This classification is proposed here,
in order to establish a distinction between
karst development located in different
regions on Aguijan (Figure 3).
Figure 4: Geology section (adapted
from Doan et al., 1960).
pyroclastics were probably ejected from a
submerged vent. The limestone units are
subdivided into two formations: Tagpochau
Limestone and Mariana Limestone. The
Tagpochau Limestone covers 16% of
surface area of Tinian and is early
Miocene in age. The formation is
composed of three contemporaneously
deposited facies: detrital (Tt), argillaceous
(Tta), and sandy (Tts). The detrital facies
comprises the majority of the formation
and is composed primarily of fragments of
biogenic calcium carbonate with calcite
cement. The other two facies comprise
only a small portion of the Tagpochau
Limestone. The Mariana Limestone
covers 83% of the surface area of Tinian
and was deposited in Pliocene and
Pleistocene time. The formation was
subdivided based upon the presence of
constructional or detrital compositions
into seven facies: constructional
coralliferous facies (Qtmcc), constructional
algal facies (QTmca), detrital coralliferous
facies (QTmc), detrital shelly facies
(QTms), detrital Halimeda facies (QTmh),
detrital argillaceous facies (QTmu), and
detrital undifferentiated facies (QTmu).
Overlying these deposits in coastal regions
are Holocene limestones, developing sands
and gravels, and reefs (Siegrest, 1988; Doan
Figure 3: Physiographic map of Aguijan.
The geology of Tinian was described by
Doan and coworkers (1960) and remains the
most detailed geologic study for the island.
Tinian is composed of volcanic tuffs and
breccias covered with coralline and algal
limestone (Figure 4). The igneous rocks
(Tinian pyroclastic) comprise less than three
square kilometers and retain only relict
structures and textures because of extensive
weathering. Based on the presence of
foraminifera within the sediments, the
4
et al., 1960; Burke, 1953). Recent fieldwork
indicates that late Pleistocene limestones
may also overlie the Mariana Limestone,
because some limestone present in coastal
areas appears to correspond with oxygen
isotope stage 5e, suggesting that Mariana
Limestone deposition does not extend to the
Pleistocene/Holocene boundary (Stafford et
al., 2002).
While no detailed study of the geology
of Aguijan has been conducted, the same
classification for rock units used on Tinian
will be applied, based on its close proximity,
until future work produces a more detailed
geologic map of Aguijan. In addition, it is
presumed that Aguijan is similar to other
islands in the southern Marianas, in that a
non-carbonate core of pyroclastic rocks
exists beneath the exposed carbonate rocks
that crop out, while the development of three
distinct terrace levels on Aguijan suggests a
similar tectonic history (Stafford et al., in
press; Doan et al., 1960; Sugiwara, 1934).
control lasted until World War One (Hunt
and Wheeler, 2000).
In 1914 Japan took control of the
Mariana Islands at the start of World War
One. After the war, Japan's control as
administrator was recognized by the League
of Nations and remained so until World War
Two. During the Japanese occupation, vast
tracts of land were cleared of coconut trees
and tropical forest, so that the extensive
sugar cane plantations could be developed
covering 58% (Bormann, 1992) to 90%
(McClure, 1977) of the land area. In 1944,
United States military forces secured the
islands in some of the bloodiest battles of
World War Two. Tinian became
instrumental in America’s Pacific war
campaign. The northern third of the island
was developed into a large airbase named
North Field, consisting of four major
runways. During full operation, this was the
largest and busiest airfield in the world with
two B-29’s taking off simultaneously every
45 seconds on mission days. From North
Field, the two atomic bombs, Fat Man and
Little Boy, were assembled and loaded into
the B-29 bombers Enola Gay and Bock’s
Car for drops on Hiroshima and Nagasaki,
respectively (McClure, 1977).
In 1947, the United Nations gave the
United States trusteeship over the Mariana
Islands, which continued until the 1970’s.
In 1970 discussion began between the
Marianas and the United States over the
termination of the trustee agreement, which
led to a plebiscite in 1975, which negotiated
a covenant with the Unites States. The
Covenant to Establish a Commonwealth of
the Northern Marianas Islands in Political
Union with the United States of America
negotiated an agreement where the citizens
of the Northern Mariana Islands became
self-governing as a Commonwealth, while
Guam remained a United States territory.
The covenant enabled the Commonwealth of
the Northern Mariana Islands (CNMI) to
retain the benefits of U.S. citizenship,
excluding the right to vote in U.S.
presidential elections, which ensured
continued economic support. In exchange,
HISTORICAL SETTING
The Mariana Islands were first settled
by the Chamorro people around 1500 B.C.E.
(before common error). In 1521, Ferdinand
Magellan, discovered the Marianas and
named them Islas de los Ladrones (Islands
of the Thieves), which were then renamed
Islas de las Marianas (Islands of the
Marianas) in the early 1600’s after Maria
Ana of Austria, the widow of the Spain’s
king, Phillip the IV. The islands remained
in Spanish control for almost four centuries,
during which time much migration by
indigenous people from the Caroline Islands
occurred. In 1899, after Spain’s loss of the
Spanish-American war, the Spanish
controlled islands of Micronesia (Mariana
Islands, Caroline Islands and Palau) where
sold to Germany. During German control,
the northern Marianas greatly developed
agricultural and fishing economies by
immigrating additional Carolinian people
from Chuuk and Yap, as well as Japanese
people from nearby islands like Okinawa,
who developed a lucrative agricultural trade
market of copra and sugar cane. German
5
the U.S. was given lease of 75 square
kilometers of land in the Northern Mariana
Islands, including over two thirds of Tinian
(Hunt and Wheeler, 2000).
Today, Tinian and Aguijan are governed
by a single municipality as part of the
Commonwealth of the Northern Mariana
Islands (CNMI). Aguijan is uninhabited and
Tinian hosts a population of approximately
2000 individuals. While the majority of the
Tinian remains under U.S. military control,
small farms and tourism, including scuba
diving and casinos, provide commerce
(Bormann, 1992).
1. The fresh water/salt water boundary
creates mixing dissolution, and
produces organic-trapping horizons at
both the upper and lower boundaries
of the fresh-water lens.
2. Glacio-eustacy has moved the freshwater lens up and down through a
vertical range of over 100 m in the
Quaternary.
3. Local tectonics can overprint the
glacio-eustatic sea level events,
adding complexity to the record.
4. Carbonate islands can be divided into
four categories based on basement/sea
level relationships:
i. Simple carbonate islands (no noncarbonate rocks).
ii. Carbonate cover islands (noncarbonate rocks beneath a carbonate
veneer).
iii. Composite islands (carbonate and
non-carbonate rocks exposed on the
surface).
iv. Complex islands (faulting and facies
interfingering create complex
carbonate/non-carbonate
relationships).
CARBONATE ISLAND KARST
This study enables the studies in karst
geology and hydrology performed on Guam
(Mylroie et al., 2001) and in the Caribbean
and Atlantic (Frank et al., 1998; Mylroie and
Carew, 1995a; Mylroie et al., 1995a) to be
applied and evaluated to different settings.
A primary objective of investigation on
Tinian is to advance the understanding of
the karst hydrology of carbonate islands,
while refining the general Carbonate Island
Karst Model (CIKM). The Carbonate Island
Karst Model has been designed as the
definitive model for the hydrologic
development of island karst. Modern
carbonate islands are unique due to
extensive interaction between fresh and
saline groundwater within young, porous
rock, which produces a unique geologic and
hydrologic history that is different than that
in continental settings (Vacher and Mylroie,
2002; Mylroie and Jenson, 2002; Mylroie et
al., 2001; Mylroie and Vacher, 1999, and
references therein).
Karst forming in marine conditions on
carbonate coasts and islands can be
explained by the Carbonate Island Karst
Model (Figure 5) (Stafford et al., 2003;
Jenson et al., 2002; Mylroie and Jenson,
2002; Mylroie and Jenson, 2001). Its main
aspects are:
5. The karst is eogenetic, i.e., it has
developed in carbonate rocks that are
young and have never been buried
below the range of meteoric
diagenesis.
Carbonate islands described by the
Carbonate Island Karst Model are composed
of young limestones and are heavily
influenced by meteoric waters and mixingzone dissolution, creating an environment
for eogenetic karst to develop because of
their close proximity to the site of
deposition. Vacher and Mylroie (2002, p.
183) define eogenetic karst as “the land
surface evolving on, and the pore system
developing in, rocks undergoing eogenetic,
meteoric diagenesis.” Eogenetic carbonate
rocks are diagenetically young and have not
undergone extensive cementation and
compaction, as opposed to telogenetic
carbonate rocks that are diagenetically
mature, have been buried below the range of
meteoric diagenesis, and are currently
6
Figure 5: Carbonate Island Karst Model (CIKM) (adapted from Mylroie et al., 2001).
7
exposed at the earth surface by tectonic
uplift and erosion of overlying strata
(Klimchouk and Ford, 2000). Typically,
small carbonate islands are the sites of
eogenetic karst undergoing meteoric
diagenesis, however, it is not possible to
state that carbonate islands develop solely
eogenetic karst and continental settings
develop solely telogenetic karst, because
fresh-water / sea-water interaction and a
fresh-water lens affected by glacio-eustacy
are also required for the formation of
carbonate island karst, as defined by the
Carbonate Island Karst Model (Jenson et al.,
2002; Mylroie and Jenson, 2002).
Eogenetic karst development occurs on
islands such as Bermuda, the Bahamas, and
the Marianas, as well as in continental
settings such as the Biscayne aquifer of
Florida, where the carbonate rocks are
young. Telogenetic karst development
occurs on islands such as Gotland (Sweden)
and Kephallenia (Greece), as well as in
continental settings such as Kentucky
(United States) and England (Great Britain),
where the rocks are of appreciable age and
have been buried beyond the range of
meteoric diagenesis (Vacher and Mylroie,
2002). Because eogenetic karst and island
karst are not necessarily synonymous, a
distinction should be made between island
karst and karst on islands. Island karst
occurs in specific environments that must
include three basic parameters: 1)
interaction of fresh-water and sea-water,
which produces mixing zone dissolution, 2)
a fresh-water lens that was affected by
glacioeustatic changes in the Quaternary,
and 3) karst that is eogenetic. Karst on
islands may exhibit some of the
characteristics of island karst, but these
islands are not true island karst unless all
three basic parameters are present. Islands
like Jamaica and Puerto Rico have karst
landforms (e.g. cockpits and mogotes) in
their interior that are developed in rocks of
appreciable age and that are not influenced
by glacio-eustacy or fresh-water / salt-water
interaction. This makes the karst landforms
karst on islands. True island karst occurs
on islands like the Bahamas, where the rock
is diagenetically young, fresh-water interacts
extensively with the salt-water, and the
fresh-water lens has been significantly
affected by glacio-eustacy. Therefore,
islands that have telogenetic karst, are
removed from the affects of glacio-eustacy,
and/or do not exhibit mixing zone
dissolution should have karst landforms and
features reported as karst on islands, while
islands that exhibit eogenetic karst, are
affected by glacio-eustacy, and exhibit
mixing zone dissolution should have karst
landforms and features reported as island
karst (Vacher and Mylroie, 2002).
In eogenetic rocks that have not
undergone compaction and cementation, the
rocks tend to initially exhibit high matrix
porosity and moderate permeability, but can
develop secondary vuggy porosity as a result
of meteoric and fresh-water diagenesis. The
matrix porosity will generally decrease with
age as secondary cementation infills the pore
space, while permeability increases as
preferential flow develops extensive
horizontal flow routes. Over time, the bulk
porosity remains the same or decreases, but
the vertical hydraulic conductivity of the
rock decreases while the horizontal
hydraulic conductivity increases (Vacher
and Mylroie, 2002). Due to the proximity of
eogenetic karst to marine waters in island
settings, it is highly susceptible to changes
in sea level, which results in the migration
of the fresh-water lens and flow routes in
response to glacioeustatic and tectonic
changes.
The four conceptual model
classifications of carbonate islands are based
on island composition (Figure 5). Simple
carbonate islands are composed of only
carbonate rocks at the surface and to a depth
below the base of the fresh-water lens
(Figure 5A). The Bahama Islands are a
good example. Carbonate cover islands are
composed of only carbonate rocks at the
surface, but have non-carbonate rocks that
interact with the fresh-water lens without
being exposed at the surface (Figure 5B).
Bermuda is a good example. Carbonate
8
cover and simple carbonate islands can
easily shift from one form to the other as a
result of changes in relative sea level (Figure
6). Composite islands are composed of both
non-carbonate and carbonate rocks on the
surface (Figure 5C). Barbados is a good
example. These three island types represent
a classification scheme where the two end
member environments are islands
completely composed of carbonate rocks at
one end and islands completely devoid of
carbonate rocks at the other end, with
intermediate compositions described by the
model (Figure 6). The fourth type are
complex islands, which are characterized as
having complex geologies as a result of
faulting, the interfingering of different
lithologies, and the presence of both
carbonate and non-carbonate rocks (Figure
5D). Guam and Saipan are excellent
examples (Jenson et al., 2002; Mylroie et al.,
2001).
In traditional continental settings,
structural and lithologic controls have been
recognized as having a significant influence
on karst development. Klimchouk and Ford
(2000, p. 57) state: “Bedding planes, joints,
and faults are planar breaks that serve as the
principal structural guides for groundwater
flow in almost all karstified rocks.” In
island karst settings the significance of
structural and lithologic controls have often
been overlooked because of the dominant
role of mixing-zone dissolution in the
hydrogeologic system and because the
original models for island karst were
developed on simple carbonate islands in the
Bahamas. Jenson and coworkers (2002) and
Mylroie and coworkers (2001) have recently
recognized the importance of structural and
lithologic controls on carbonate island karst
by modifying the CIKM with the addition of
the fourth type, the complex island. This
addition enables the effects of deformation
Figure 6: Schematic diagram illustrating the relationship between island size, basement relationships to
sea level and surface, and relative sea level for simple, carbonate-cover, and composite islands
within the Carbonate Island Karst Model (CIKM) (Mylroie and Jenson, 2001, Fig.2, p. 54).
9
and lithology to be accounted for on
carbonate islands that are or were
tectonically active and/or have intricate
interfingering of carbonate and noncarbonate rocks.
Lithologic controls include variations in
rock composition between rock formations
and within the same rock formation, where
beds subdivide the unit. Variations may
include grain size, grain origin, grain
sorting, and chemical composition. These
variations provide routes for preferential
dissolution within beds and formations of
favorable composition, while less favorable
beds and formations may restrict dissolution
and fluid movement (Klimchouk and Ford,
2000; Sasowsky and White, 1994; Palmer,
1991; White, 1988).
Structural controls include both brittle
and ductile deformation. Ductile
deformation results in the folding of rock
bodies, which adds to the complexity of
lithologic controls. Brittle deformation
includes joints, fractures, and faults, which
are often found in tectonically active,
carbonate islands where uplift and
subsidence create complex geology.
Fractures are surfaces where the rock has
been broken, occur over a range of
centimeters to meters and fall into several
categories, including: extension – relative
rock motion is perpendicular to the surface;
shear – relative motion is parallel to the
surface; and oblique – relative motion is a
both parallel and perpendicular to the
surface. Joints are specific types of
fractures, which show only minor extension
(Twiss and Moores, 1992). Fractures
reported on Tinian and Aguijan are
primarily associated with gravity slides
resulting from bank margin or cliff margin
failure along steep scarps. Joints reported
occur inland of the bank margin/cliff margin
fractures, possibly as a result of rock
expansion from a decrease in lateral pressure
from the scarp failure or as unloading
structures associated with isostatic
subsidence, and perpendicular to coastlines,
which may be associated with coastal
erosion or regional faulting (Stafford et al.,
2002). Faults are surfaces or narrow zones
that show relative displacement similar to
fractures, but over larger regions and
generally associated with regional tectonics
(Twiss and Moores, 1992). Faulting is
reported throughout Tinian, including the
boundaries between the Southeastern Ridge
and the Median Valley and between the
Central Plateau and the Northern Lowland,
where fault blocks moved independently
during island emergence (Doan et al., 1960).
Joints, fractures, and faults generally
enhance fluid movement, but may also
restrict it. Enhanced flow can occur along
planar features that provide a surface for
fluid movement both laterally and vertically
across lithologic boundaries, creating greater
connectivity within the subsurface.
Restricted flow can result if recrystallization
(slickensides) or secondary infilling (caliche
dikes) develop along the planar surface,
creating a barrier for fluid movement. Flow
restriction may also occur if faulting results
in the offsetting of different rock units,
which produces contacts between carbonate
and non-carbonate rocks that did not
previously exist (Klimchouk and Ford,
2000; Sasowsky and White, 1994; Palmer,
1991; White, 1988).
KARST FEATURES
Karst features on Tinian can be
classified into four broad categories;
epikarst, closed depressions, caves, and
discharge features. Identification,
classification, and spatial distribution of
different karst morphologies provide a basis
for understanding the hydrology of the
region.
Epikarst
Epikarst is the zone of dissolutional
sculpturing (karren) that is present on the
surface and upper few meters of bedrock in
carbonate regions. Karren has been
described as minor solutional forms, which
range from millimeters to meters in scale
(White, 1988). In general, epikarst is
independent of environmental setting;
however, in coastal regions, where salt spray
10
is in active contact with carbonate rocks, an
environment is created for the production of
biokarst (Viles, 1988) or phytokarst (Folk et
al., 1973); however, the biological affects
are often overstated (Mylroie and Carew,
1995b). The surface of the karren in
eogenetic karst is generally extremely rough
and may not support extensive soil profiles
because soil-forming material is in short
supply. Taborosi and coworkers (in press)
have recently reported additional karren
morphologies on eogenetic karst on Guam.
In the Mariana Limestone on Guam most
weathered surfaces are extremely jagged and
reminiscent of biokarst or phytokarst and is
attributed to a polygenetic origin that
includes mixing dissolution from rainwater
mixed with salt spray, salt weathering,
dissolution by meteoric waters, and
biological weathering (Taborosi et al., in
press). However, other karren morphologies
are also present, which include more
rounded karren in inland regions and
dissolution along brittle failure planes
producing enlarged vertical and horizontal
joints as wells as solution pans (kaminetzas)
(Taborosi et al., in press). Carbonate rocks
on most islands generally have little
insoluble material to produce soils,
suggesting that the insoluble material is
exclusively of eolian origin. The soils that
are present are generally piped downward
into voids and dissolutional cavities within
the karst system. Eogenetic karst regions
have specific vegetation types because of the
limited development of soils, which affects
infiltration rates of water entering the karst
system based on soil thickness, composition
and its presence, absence or modification
(human development). Beneath karren,
dissolutional bedrock debris, and soil, which
comprise the surficial epikarst, the
remainder of the epikarst zone is composed
of solutional fissures, holes, and shallow
small cavities in the bedrock (Mylroie et al.,
2001). Epikarst contributes water to the
deep vadose zone as diffuse flow and
through the integration of flow paths created
by cavities and fissures in the lower epikarst.
This integrated flow can effectively bypass
the deep vadose zone via fractures and pits
and supply water directly to the phreatic
zone (Mylroie et al., 1999). Epikarst
development and extent is a primary
controlling factor on the quantity of water
that enters phreatic storage via vadose paths
and it is possible that it may serve as a
location of significant water storage (Jocson
et al., 2002; Jocson et al., 1999; and Jenson
et al., 1997). Epikarst on Tinian appears
identical to that seen on Guam and Saipan
(Taborosi, 2000).
Individuals involved in the development
of land in karst regions dominated by
epikarst should be aware that any
modifications to the land surface and
epikarst would alter the drainage dynamics
of the area. Ponding basins may actually
exhibit lower infiltration rates because of
high sediment loads. Runoff events and soil
erosion may also increase as a result of the
modification of the natural landscape
(Mylroie and Carew, 1997).
Closed Depressions
In carbonate island environments, closed
depressions can be classified into three
general categories: dissolution, natural
construction, and human modification.
These different types can be extremely hard
to differentiate based on appearance and it is
possible that the features may have been any
or all of the three classification types at
some point in their development. In
carbonate islands, dissolutional depressions
are generally small to moderate in size as a
result of their young age and the nature of
autogenic recharge. However, in areas
where non-carbonate rocks are exposed at
the surface, streams can develop that
provide allogenic recharge to the karst
system, possibly forming large depressions
as are seen on Guam (Mylroie et al., 2001;
Taborosi, 2000). Natural construction
depressions are those that formed at the time
the rocks were deposited or are the result of
subsequent deformational processes.
Mylroie and coworkers (2001) report that
natural construction depressions are the
most common form on simple carbonate
11
islands, and anthropogenic depressions (e.g.
quarries, landfills, artificial drainage ponds,
and storage ponds) are often constructed in
these pre-existing features (Mylroie et al.,
1999).
Tinian exhibits areas where dissolutiontype closed depressions are formed.
Dissolution depressions have been seen at
the contacts between exposed volcanic
outcrops and carbonate outcrops (Stafford et
al., 2002). On Tinian, four exposures of
volcanic rocks are recorded near Sabanettan
Mangpang, Bañaderon Lemmai, and
Laderan Apaka. At the three northern
locations (two at Sabanettan Mangpang and
one at Bañaderon Lemmai), there are closed
depressions in the limestone outcrops, which
show allogenic streams descending into
them from the volcanic outcrops. An initial
field investigation of Bañaderon Lemmai
has shown that these features are similar to
those seen on Guam and are providing point
source recharge into the karst system
(Stafford et al., 2002). In the south-central
region near Laderan Apaka, volcanic
outcrops are exposed on a north-facing cliff.
To the north of the outcrops, a large closed
depression (Sisonyan Makpo) exists in
which the Municipal Wells are located
(United States Department of the Interior
Geological Survey, 1983), which may be a
natural construction feature formed by
complex faulting on the island, based on its
location between the two prominent ridges
(Kastiyu/Carolinas and Piña) of the
Southeastern Ridge and the low-lying
Median Valley. Weathering of volcanic
rocks and talus accumulation may be
partially armoring the slopes of the cliff and
the closed depression, increasing recharge to
the closed depression and possibly leading
to lateral corrosion of the closed depression.
Throughout Tinian, human-modified
depressions are also commonly reported
(United States Department of the Interior
Geological Survey, 1983) and may represent
modified constructional depressions similar
to those reported on other carbonate islands
(Mylroie et al., 1999). These include
modern and ancient quarries (latte stone),
borrow pits, refuse disposal sites, artificial
drainage ponds in residential areas, and
features associated with World War Two
(bomb pits, defensive positions, etc.). Initial
field investigations have revealed that
significant natural depressions are found in
association with volcanic outcrop exposures.
The large closed depressions seen elsewhere
are probably natural construction features or
features produced by human modification,
possibly to pre-existing natural depressions,
of the land surface (Stafford et al., 2002).
Caves
Caves are natural openings in the earth
that can be characterized based on their size,
shape, length and overall geometry (White,
1988). Solution caves, which have been
formed by the dissolution of bedrock by
circulating groundwater, are present
throughout the islands of Tinian and
Aguijan. Solution caves on carbonate
islands can be grouped into various
categories. In the Mariana Islands, five
distinct cave types have been documented:
banana hole, flank margin cave, fissure
cave, pit cave, and stream cave (Figure 7).
In some areas, it is difficult to discern the
exact origin and original extent of some
caves, because of the extensive modification
of some features for Japanese military
purposes during World War Two (Taborosi
and Jenson, 2002). In addition to human
modification, horizontal notches cut into
cliff faces create an additional classification
problem. Traditionally these have been
identified as bioerosion notches, but may
also form by lateral corrosion and cliff
retreat, or they may be the remnants of flank
margin caves (Mylroie et al., 1999).
Accurate classification of karst features is
integral to interpreting the hydrogeology of
island karst.
Banana holes (Figure 7) are shallow,
small chambers that formed at the top of the
fresh-water lens. They are isolated features
and exhibit a morphology with a width to
depth ratio greater than one (Harris et al.,
1995). On Tinian no feature has been
currently identified as a definite banana
12
Figure 7: Conceptual model of cave types that form in eogenetic rocks on carbonate islands.
hole. A feature located several hundred
meters to the northwest of the Lasso Shrine
appears to have a complex history that could
be explained as the stacking of two or more
banana hole features vertically on one
another as sea level fluctuated. Later
collapse by upward stoping could then form
the collapsed areas seen in the lower levels
of this feature (Stafford et al., 2002).
Flank margin caves (Figure 7) are
formed in the distal margin of the freshwater
lens where thinning of the lens and mixing
of fresh and saline waters creates
dissolutionally aggressive phreatic waters
(Mylroie et al., 1995a) and are the most
common morphology type that has been
reported on Tinian (Stafford et al., 2003,
2002). Flank margin caves exhibit globular
morphologies with a wide range of sizes that
may be connected, but often remain as
isolated chambers. Flank margin caves can
be used as indicators of previous sea-level
stillstands and have potential as tools for
evaluating differential rates of uplift in
tectonically active carbonate islands (Carew
and Mylroie, 1995; Mylroie et al., 1995a,b).
Flank margin caves may collapse in coastal
regions to form coves and caletas, which
retain only remnants of the original cave
morphology (Back et al., 1984). Further
complicating the morphology and
identification of flank margin caves are
bioerosion notches formed by wave erosion
and invertebrate borings at sea level during
stillstands (Mylroie and Carew, 1991).
Flank margin caves have been reported
at various locations. The most significant
flank margin cave development is in the
Suicide Cliffs on the southern end of Tinian,
were numerous entrances can be seen inland
from the road on the southern coastal
terrace. These caves are located
approximately 50 meters upslope from the
lower terrace and are developed along a
consistent horizon that is believed to reflect
a past sea-level stillstand. Along the
northern edge of the Southeastern Ridge,
near Laderan Apaka, a series of flank
margin caves are located, which are less
extensive due to more complete erosion and
slope retreat. Near Lasso Shrine on northern
Tinian, a series of modified flank margin
caves are located in the eastern cliffs. Along
the east coast, flank margin caves that are at
various stages of erosion due to cliff retreat
and coastal erosion are present near Unai
Masalok and Unai Dangkolo. At Unai
Dangkolo, the largest flank margin cave
documented on Tinian is located
approximately 200 meters inland. This cave
13
is breached on the surface by ceiling
collapse and can only be entered from above
via a 10-meter descent. The cave consists of
several large chambers that intersect,
creating a complex system that is laterally
extensive covering and area greater than
1300 square meters with cave passages
developed around a central chamber
approximately 15 meters tall and 35 meters
in diameter (Stafford et al., 2002).
Fissure caves (Figure 7) are developed
along joints, fractures, or faults, in which
preferential flow along the planar surface
has resulted in enhanced dissolution rates
(White, 1988). Three basic types of fissure
caves have been identified, based on their
morphology and spatial relationship to the
island. At the coastal edge, development
perpendicular to the coastline has been
reported, which in several cases has resulted
in caves that often discharge freshwater and
penetrate inland up to 30 meters as tubular
passages with distinct joints located in their
ceilings. The second type has formed as
fractures parallel to the coastline or scarps as
a result of cliff-margin or bank-margin
failure. These fractures may reach the water
table where they can affect the flow
dynamics of the lens by locally distorting
the lens or by providing vadose routes that
intercept the normal diffuse flow in the
phreatic zone (Mylroie et al., 1995c; Aby,
1994). Dissolutionally enhanced fractures
have produced caves over 40 meters deep on
Tinian. The third type occur where
dissolution along a fault plane, in
conjunction with collapse, has formed caves,
which tend to develop at moderate to steep
angles along the dip of the fault, and the
caves extend laterally along the strike of the
fault. All three types of fissure caves are
hydrologically important because they
provide vadose fast flow routes for water
through the subsurface, either as recharge or
discharge features and may distort the local
lens morphology and flow dynamics
(Stafford et al., in press; Stafford et al.,
2002).
Pit caves (Figure 7) are vertical shafts
that have developed by the dissolution of
descending meteoric waters and have a
depth to width ratio greater than one. Pit
caves often develop as a series of shafts with
connecting lateral sections, act as vadose
fast-flow routes for water entering the
subsurface, and effectively drain the epikarst
(Mylroie and Carew, 1995a). Although
common in certain settings (Harris et al.,
1995; Mylroie and Carew, 1995a), only one
pit cave has been reported on Tinian. It is
located 10 meters from the cliff edge on the
south coast, where it connects to a low-level
bench just above sea level. It is
approximately 25 meters deep with an
entrance width of 3 meters, but due to its
location and near-direct connection to the
ocean, it has little effect on aquifer recharge.
However, its presence indicates the
possibility of other similar features in the
area, which may be covered at the surface
by collapse and at shallow depths could be
open, acting as vadose fast flow routes
(Stafford et al., 2002).
Stream caves (recharge caves, Figure 7),
hydrologically active caves that are fed by
allogenic water, have been documented in
regions were contacts between carbonate
and non-carbonate rocks occur. Meteoric
water is unable to infiltrate efficiently into
the volcanic rocks, and forms surface
streams, which channel water to the outcrop
periphery where it descends into the
carbonate rocks as point source allogenic
recharge (Mylroie et al., 2001).
Tinian volcanic rocks outcrop in four
regions and closed depressions are
associated with the periphery of each of
these outcrops. Due to time and logistics,
only one of these features, located to the east
of the largest outcrop at Bañaderon Lemmai,
was investigated in June 2002 (Stafford et
al., 2002). This site contains a large
sinkhole that appears to be armored with
volcanic sediments that restrict infiltration,
but on the eastern edge, a small cave is
present that shows evidence of allogenic
recharge during rain events (Stafford et al.,
2002). It is expected that other closed
depressions associated with these volcanic
outcrops will exhibit similar features formed
14
by dissolution at the noncarbonate/carbonate contact, with variations
dependant on sediment infilling and the
volume of allogenic recharge that is
received.
WATER RESOURCES
Fresh-water is partitioned in a lens
above salt-water and is affected by the
presence of non-carbonate rocks interacting
with the fresh-water lens. Basal water
occurs where the fresh-water lens is directly
underlain by sea-water, while parabasal
water occurs where the fresh-water lens is
directly underlain by non-carbonate,
basement rocks (Figure 8; Mink and Vacher,
1997). Basal waters develop a lens
thickness defined by the Ghyben-Herzberg
Principle:
Discharge Features
Discharge volume and types vary
greatly on carbonate islands and spatially
within the same island. Three general
categories of discharge features can be
defined: seeps, springs, and submarine
freshwater vents. Of these types, seeps
represent diffuse discharge, while the other
types represent focused discharge points.
Seeps occur extensively in coastal areas
where calcareous sand covers the bedrock
and disperses emerging water over a large
area. Springs are locations in which water
emerges from the bedrock along preferential
flow paths generally defined by bedding
planes or fractures along coastlines. Springs
may be laterally extensive if controlled by
bedding planes or they may be restricted to a
single point discharge if controlled by
fractures. Submarine freshwater vents are
areas where freshwater discharge occurs
below tidal level along the island periphery
(Jocson et al., 2002; Mylroie et al., 1999).
Seeps and springs have been
documented in coastal regions on Tinian.
Minor seeps are present at Unai Dangkolo
and Unai Masalok on Tinian, while springs
have been documented in various coastal
areas. On Tinian, fracture caves were
reported along the west coast near Unai
Masalok where freshwater could be
observed mixing with salt water and
discharging from solutionally widened
bedding planes that extend approximately 30
meters inland (Stafford et al, 2002).
Submarine freshwater vents are not reported
for Tinian and Aguijan, but features are
expected to exist that are similar to those
that have been reported from Guam
(Mylroie et al., 2001; Jenson et al., 1997),
where freshwater discharges below sea
level.
Z = [(ρf)/(ρs - ρf)]h,
where Z is the depth of the fresh-water lens
below sea level, h is the height of the lens
above sea level, and ρf and ρs are the
densities of fresh-water and salt-water
respectively (Figure 8; White, 1988; Raeisi
and Mylroie, 1995). The fresh-water/saltwater lens thickens as the fresh-water head
depresses the interface below sea level,
relative to the density difference in the two
waters:
Z = αh,
where α = (ρf)/(ρs - ρf). With ρf = 1:00
g/cm3 and ρs = 1.025 g/cm3, α = 40;
therefore, for every one meter or foot of
fresh-water head, the fresh-water/saltwater interface is depressed 40 meters
or feet (1:40 ratio) (White, 1988; Raeisi
and Mylroie, 1995).
Island environments have limited water
resources and specific problems because of
the morphology of the fresh-water lens
(water with a chloride concentration <250
mg/L (Gingerich and Yeatts, 2000)).
Because of the limited extent of island
aquifers and the characteristics of eogenetic
karst, island aquifers are extremely
susceptible to contamination.
Contamination from the surface can include:
human and animal wastes, fertilizers,
detergents, pesticides, herbicides, petroleum
spills, and solvent spills. Subsurface
contamination can occur from salt-water
15
Figure 8: Model showing the fresh-water lens morphology showing the location of basal and parabasal
waters and the Ghyben-Herzberg principle (adapted from Mink and Vacher, 1997 and Raeisi
and Mylroie, 1995).
intrusion, where over pumping produces a
cone-of-depression in the lens, which then
up-cones salt-water into the fresh-water lens
at a one to forty ratio. Because of the ratio
of up-coned water to lens thickness above
sea level, a small decrease in the lens
thickness above sea level will create a large
cone of depression at depth in basal waters
(Gingerich and Yeatts, 2000; Mylroie and
Carew, 1997). When non-carbonate rocks
extend into or through the freshwater lens,
basement rocks restrict upconing of
parabasal water because salt-water does not
directly underlie fresh-water. Therefore, it
is important for island communities to
effectively manage their water resources in
order to continue to produce potable water,
by limiting water extraction at specific sites
and by utilizing zones of parabasal
groundwater (Mink and Vacher, 1997).
In the 1995 census, 2,631 people were
listed as living on Tinian primarily in the
Median Valley and parts of the adjacent
Central Plateau, which comprises 25% of
the total island surface area. This limited
region of population is partially the result of
the United States control of the northern
third of the island, the Northern Lowland,
for military purposes, while other regions
are less accessible due to terrain. Of the
inhabited region, 60% is an undeveloped,
public, rural area. The remaining 40% of
this region is composed of residential and
commercial lots that include a casino resort,
small businesses, farming, grazing and
housing. The residential and commercial
lots provide the largest concerns for
contamination from surface spills and
biological waste disposal. Tinian currently
has no sewer facility, instead human waste is
disposed of through septic and seepage
tanks, leaching fields or holding tanks,
which may not be adequate for preventing
groundwater contamination, because of the
thin soil profile and rapid infiltration rates
associated with eogenetic, carbonate rocks.
Other sources of contamination include the
airport, several quarries and a solid waste
disposal dump, which provide paths for
direct recharge into the aquifer by bypassing
the soil surface and epikarst, which have
been removed (Gingerich and Yeatts, 2000;
Bormann, 1992).
On Tinian, USGS investigations show
that the maximum lens thickness is
approximately 12 meters (40 feet) in the
16
center of the Median Valley, with slight
thinning near the Municipal Well and Marpi
Marsh (Gingerich and Yeatts, 2000). The
Municipal Well (a Maui-type, infiltration
well) supplies the majority of the islands
water needs at a rate of approximately 4.5 x
106 liters/day (~1.2 Mgal/day). Today water
produced from the Municipal well has a
chloride concentration of 180 mg/L, which
is 100 mg/L greater than when the well was
constructed in 1945. The USGS monitored
the aquifer thickness in relation to rain
events in the wet season of 1993 and
reported an aquifer thickening of 90-150 cm
(3-5 ft), while reports in the dry season of
1994 showed an aquifer thinning of 30-60
cm (1-2 ft). These results revealed that
annual fluctuations are minor and lens
thickness is dependent more upon long-term
rainfall patterns instead of annual
fluctuations, for maintaining overall
morphology (Gingerich and Yeatts, 2000).
This stable lens morphology indicates that
contamination problems have the potential
to produce long term effects due to long
residence time for water within the aquifer.
With the stable fresh-water lens
morphology on Tinian, the location and
identification of karst features becomes
important for preventing groundwater
contamination. Understanding the spatial
distribution and extent of fissure caves, pit
caves, and recharge features, which can
transport contaminants rapidly to the lens,
may enable government planners to regulate
activities near sensitive areas. Similarly,
identified discharge features can be used as
sampling points for monitoring possible
groundwater contamination. Identification
of allogenic recharge may be used to predict
regions where parabasal waters may be
thicker due to increased recharge, such that
salt-water intrusion risks from water
extraction may be reduced. Spatial
distributions of flank margin and banana
hole caves may provide insight into the
diffuse flow characteristics of previous
fresh-water lenses, making it possible to
better evaluate current lens morphology. If
positive correlations do exist between brittle
deformation and megaporosity, then
regional structure may provide insights into
groundwater behavior.
STUDY METHODOLOGY
This study had three objectives. It was
initially developed to inventory, survey, and
classify the cave and karst features on
Tinian and Aguijan, because no such
database existed for the islands. In addition
to the inventory, this study was developed to
evaluate whether or not a statistical
comparison could be made between
megaporosity and zones of brittle failure in
eogenetic rocks on carbonate islands, using
data collected from the island of Tinian
where basic geologic studies have been
conducted (Doan et al., 1960). The study
was conducted in five major phases, in order
to reach the two objectives: 1) initial site
investigation, 2) data collection, 3) data
reduction, 4) statistical comparison of data,
and 5) small-scale test site evaluation.
Results from the two primary objectives of
the study were used to evaluate the islands
of Tinian and Aguijan in relation to the
Carbonate Island Karst Model.
INITIAL SITE INVESTIGATION
The initial site investigation included a
reconnaissance of Tinian and analysis of the
physiography and geology of the island.
The reconnaissance was conducted in June
2002 (Stafford et al., 2002) and provided
basic information about the cultural,
physical and logistical aspects of Tinian,
while establishing relationships with local
government bodies that would be crucial for
in-depth studies on the island. During this
reconnaissance areas that were reported by
island residents as having significant cave
development were visited, including, but not
limited to: Suicide Cliffs, Unai Dangkolo,
17
and Mount Lasu. This initial site
investigation demonstrated that most known
caves, reported by local residents and
hunters, occur predominantly along scarps,
coastlines, and closed depressions, which
agreed with previous investigations on
carbonate islands (Stafford et al., 2002;
Mylroie et al., 1999).
After the initial reconnaissance, the
islands geography and geology were
analyzed with a geographical information
system (GIS) that was produced for the
island using ArcView 3.2 (ESRI, 2000). A
digital elevation model (DEM) was created
for the island using spatial data transfer
standard (SDTS) compliant raster data with
10-meter postings produced by the National
Mapping Program of the United States
Geological Survey (USGS, 2001a,b,c,d,e),
which provides greater resolution than the
1:25,000 topographic map produced by the
USGS (USGS, 1983). The DEM was then
overlain with a scanned geology map
produced by Doan and coworkers (1960). It
was geo-referenced using Image Analyst
(ESRI, 2000) in order to scale and align the
geology map with DEM, then all igneous
outcrops and faults were manually digitized
in order to create shape files of these
features. The DEM and digitized geologic
features were then used as the basic dataset
needed to delineate prominent scarps and
closed depressions on the island.
Scarps were defined as any change in
slope greater than twenty degrees, which
enabled the identification of all major scarps
in the island interior and coastline (Figure
19, Appendix A). Because of the DEM cell
size, smaller scarps that might contain cave
entrances were excluded; however, it did not
eliminate investigations of smaller coastal
scarps because coastlines were identified as
sites of cave development during the initial
reconnaissance. Closed depressions were
defined as regions that were lower than the
surrounding topography on all sides. Only
closed depressions greater than 10 meters in
diameter could be identified based on the
limitations of the DEM grid size. The
closed depressions where analyzed in
relation to outcrops of igneous rock. The
closed depressions that were found to be
proximal to the igneous outcrops were
identified as locations of possible allogenic
recharge developed by dissolution, while the
closed depressions that were distal to the
igneous outcrops probably had nondissolutional (i.e. constructional or human
modified) origin (Figure 20, Appendix A).
The reconnaissance and GIS
investigation of Tinian provided a
framework for fieldwork and data collection
on Tinian. Because there is little published
geology on Aguijan and a reconnaissance
visit was not possible, no detailed fieldwork
plan could be developed. However, this
project extended the cave and karst
inventory to Aguijan when the opportunity
for access to the island was available.
DATA COLLECTION
Two types of data were collected during
fieldwork: 1) cave and karst surveys, and 2)
structural orientations of zones of brittle
failure. These two datasets were collected
over the course of two intensive field
seasons (December 10, 2002 to January 07,
2003 and May 04, 2003 to June 1, 2003) and
provided the database for analysis and
correlation in this study. During the course
of fieldwork, coastal and scarp
investigations were limited to regions that
could be accessed within an acceptable risk
level that would not greatly endanger safety.
Cave and karst surveys were conducted
on Aguijan and Tinian based on the initial
site investigation and additional reports
provided by local residents during the course
of fieldwork. This fieldwork focused on
known caves, coastlines, scarps, and closed
depressions. Caves were surveyed in
accordance with current international
standards for cave cartography and mapping
established by the National Speleological
Society (NSS) (Dasher, 1994) and the
Association for Mexican Cave Studies
(AMCS) (Sprouse and Russell, 1980).
Individual surveys were conducted using a
Suunto compass, Suunto inclinometer, and
fiberglass tape, in association with a field
18
sketch recorded by experienced project
sketchers. Caves and other karst features,
including discharge and recharge features,
which did not warrant survey, were photodocumented and recorded. Discharge
volumes on discharge features were
estimated (minimal discharge and significant
discharge), because coastal conditions did
not allow for measurements of salinity and
temperature to be taken at most sites.
During fieldwork, features were
classified by cave type based on their
appearance in accordance with the
Carbonate Island Karst Model. The feature
types included: banana hole, discharge
cave/feature, fissure cave, flank margin
cave, recharge cave/feature, and pit cave.
When satellite coverage permitted,
Universal Transverse Mercator System
(UTM) coordinates and elevation were
recorded with the global positioning system
(GPS), in order to establish accurate location
information for the feature. When satellite
coverage was not possible, because of
vegetation or topography, cave locations
were identified on 1:25,000 USGS
topographic maps (USGS, 1983), which
were then used to determine UTM
coordinates and elevation for features.
Structural orientations of planes of
brittle failure were measured using a Suunto
compass during fieldwork on Tinian.
Aguijan was excluded from this phase of
data collection, because no published
geologic map for the island exists and
available time on the island was limited.
Orientations were taken in areas where a
joint or fracture could be observed in the
bedrock and was not obscured by karren or
phytokarst development. Orientations were
only taken where the joint or fracture
continued over a distance of several meters,
cut through more than one bedding plane or
several meters of bedrock, and in areas
where the bedrock appeared to be in situ.
on Aguijan and Tinian, 2) analysis of cave
maps and delineation of primary cave
orientations and cave segment orientations
for features on Tinian, 3) analysis of faults
and joints reported by Doan and coworkers
(1960), 4) analysis of scarp and coastline
orientations on Tinian, and 5) orientation
data reduction and production of rose
diagrams for structural and cave orientations
on Tinian.
Data from surveyed cave and karst
features was reduced using the software
package WALLS (McKenzie, 2002), which
enables compass, inclinometer and tape
measurements taken during surveys to be
plotted with corrections for minor loop
closure errors and regional magnetic
declination. The corrected line plots for
each cave were then used as a basis for
drafting spatially correct final maps of the
features using Corel Xara 2.0 (Xara, 1997).
Final cartographic products were created by
overlaying field sketch notes and corrected
line plots. The field sketch notes were
scaled and rubber-banded to match the
corrected line plots, then features recorded
on the field sketch notes were manually
digitized using standard cave symbology
established by the NSS and AMCS (Figure
21, Appendix A; Sprouse and Russell, 1980)
in order to create accurate maps of the cave
and karst features inventoried.
These maps were used to delineate
primary cave orientations and segment
orientations for features surveyed on Tinian,
using two methods: 1) apparent trends and
2) entrance oriented trends. Cave
orientations have been used in previous
studies in continental settings to correlate
cave and karst development with regional
structure (Nelson, 1988; Barlow and Ogden,
1982), where the orientations of individual
cave segments that have a consistent trend
are measured and compared against regional
brittle failure features to determine if a
correlation between the two populations
exist.
Using the apparent trend method, all
Tinian cave maps were analyzed
individually and a primary axis was defined
DATA REDUCTION
Data reduction included five phases: 1)
survey data reduction and production of
maps of the cave and karst features surveyed
19
through the cave based on the maximum
length of the cave and the location of the
breached entrance. Cave segments were
then defined by the orientation of individual
chambers, passages, and wall characteristics
such as large pockets and chamber alcoves
(Figures 9 and 10). This technique is highly
subjective, but was used because of its
similarity to studies in caves where passages
tend to be linear (Nelson, 1988; Barlow and
Ogden, 1982). The primary and segment
cave trends were then measured and
compiled. The segment cave trends were
length-weighted into 5 and 10-meter
increments, such that each 5 or 10-meter
section of cave segment was counted as an
individual orientation measurement in order
to give greater significance to longer
segment trends during data analysis (Nelson,
1988; Barlow and Ogden, 1982). Two
segment lengths were used to form two sets
of data that would reduce the subjectivity of
the parameters that were used to define
segment length, such that smaller cave
sections would be included in the 5-meter
segment data, while only larger cave
sections would be included in the 10-meter
segment data.
The entrance width trend method of
delineating cave orientations was developed
and performed in an attempt to reduce the
subjectivity in data reduction for flank
margin and banana hole type caves, which
tend to form globular or elliptical chambers
instead of the more linear passages seen in
fissure caves. Using the entrance orientation
method, up to three orientations were
measured for flank margin and banana hole
type cave maps on Tinian (Figure 11). In
this technique, the orientation of the
entrances to caves that were entered
horizontally was measured. Next a
maximum penetration measurement was
calculated near perpendicular to the entrance
(90o ± 15o from the entrance orientation). In
caves that had been breached by ceiling
collapse and were entered vertically, no
entrance orientation was measured, but the
penetration measurement was defined as the
longest dimension of the cave. Based on the
Figure 9: Example of primary and segment
orientation trends measured using the
apparent trend method for a typical
fissure cave.
Figure 10: Example of primary and segment
orientation trends measured using
the apparent trend method for a
typical mixing zone cave.
Figure 11: Example of entrance, maximum width,
and penetration trends measured using
the entrance width trend method for a
typical mixing zone cave.
20
penetration measurement, a maximum width
was measured near perpendicular to the
penetration orientation (90o ± 15o from the
penetration orientation). In cases where the
cave entrance was the maximum width, the
same measurement was reported for both
entrance width and the maximum width.
These measurements were then compiled as
primary entrance width, penetration length,
and maximum width orientations and
corresponding segment orientations that
were length-weighted in 5 and 10-meter
segments similar to the cave segments
defined in the apparent method.
Faults and joints reported by Doan and
coworkers (1960) were measured for length
and orientation. The data for faults was
length-weighted by 50 and 100-meter
segments in order to apply greater
significance to faults that extended over
greater distances. Two segment lengths
were used in order to reduce the subjectivity
in selecting segment lengths, such that
shorter lengths would be included in the 50meter segment data and only longer
segments would be included in the 100meter segment data. These two segment
lengths were defined because they are
proportional to the segment definition used
for cave passages, but measured at a scale
that is one order of magnitude greater,
because 5 and 10-meter lengths could not be
accurately measured from the geology
reported by Doan and coworkers (1960).
Similarly, all scarps that were identified
during the initial site investigation and all
coastlines were divided into linear segments
in order to eliminate the effects of minor
variations in coastal erosion and mass
wasting. The linear segments were
measured for orientation and length; with
length weighted, 50 and 100-meter segments
applied to both scarp and coastline
orientations as was defined for fault
segments, because the DEM cell size does
not allow for accurate measurements less
than 10 meters.
In the final phase of data reduction, all
orientation measurements were reduced to
produce rose diagrams using the software
package GEOrient 9.2 (Holcombe, 2003),
using five-degree orientation sectors in order
to be able to visually compare the
orientation patterns of caves, zones of brittle
failure, scarps, and coastlines. Rose
diagrams were produced for each type of
orientation data in each of the five
physiographic provinces of Tinian (Central
Plateau, Median Valley, Northern Lowland,
North-Central Plateau, and Southeastern
Ridge) and for the entire island of Tinian
(Tinian Composite). These rose diagrams
include the following types:
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
21
fault orientations
fault, 5-meter segment orientations
fault, 10-meter segment orientations
joint orientations
orientations of fractures measured during
fieldwork
inland scarp orientations
inland scarp, 5-meter segment orientations
inland scarp, 10-meter segment
orientations
coastal scarp orientations
coastal scarp, 5-meter segment
orientations
coastal scarp,10-meter segment
orientations
all scarps orientations
all scarps, 5-meter segment orientations
all scarps, 10-meter segment orientations
coastline orientations
coastline, 5-meter segment orientations
coastline, 10-meter segment orientations
fissure cave orientations
fissure cave, 5-meter segment orientations
fissure cave, 10-meter segment
orientations
mixing zone cave orientations
mixing zone cave, 5-meter segment
orientations
mixing zone cave, 10-meter segment
orientations
all cave types orientations
all cave types, 5-meter segment
orientations
all cave types, 10-meter segment
orientations
mixing zone cave penetration orientations
mixing zone cave penetration, 5-meter
segment orientations
mixing zone cave penetration, 10-meter
segment orientations
30. mixing zone cave entrance width
orientations
31. mixing zone cave entrance width, 5-meter
segment orientations
32. mixing zone cave entrance width, 10meter segment orientations
33. mixing zone cave maximum width
orientations
34. mixing zone cave maximum width, 5meter segment orientations
35. mixing zone cave maximum width, 10meter segment orientations
7.
STATISTICAL COMPARSION
OF DATA
Statistical comparisons were calculated
in order to determine if different populations
of orientation data were similar within each
of the five physiographic provinces of
Tinian and the entire island of Tinian. The
comparisons were calculated in order to
determine if there is a significant
relationship between: 1) brittle deformation
of eogenetic rocks and the development of
eogenetic karst; 2) scarp orientations and the
development of eogenetic karst; and 3)
coastline orientations and the development
of eogenetic karst.
Statistical comparisons were conducted
using the Kolmogorov-Smirnov 2-sample
test (Burt and Barber, 1996; Till, 1974;
Miller and Kahn, 1962), which performs a
non-parametric test on two independent
sample populations to determine if they
represent similar populations or population
distributions. During the statistical
comparison phase, several populations of
orientations were evaluated, which
correspond to the 35 orientation types that
were used to produce rose diagrams during
the data reduction phase. Each of the 35
orientation types was evaluated for
correlation with:
15.
16.
1.
2.
3.
4.
5.
6.
8.
9.
10.
11.
12.
13.
14.
17.
inland scarp, 5-meter segment
orientations
inland scarp, 10-meter segment
orientations
coastal scarp orientations
coastal scarp, 5-meter segment
orientations
coastal scarp, 10-meter segment
orientations
all scarps orientations
all scarps, 5-meter segment orientations
all scarps, 10-meter segment
orientations
coastline orientations
coastline, 5-meter segment orientations
coastline, 10-meter segment
orientations
These correlations were performed in
order to test three different null hypotheses
(H0):
H0 (1): Regional brittle deformation and
karst development represent
significantly different populations
or population distributions and
develop independently.
H0 (2): Regional scarp positions and karst
development represent
significantly different populations
or population distributions and
develop independently.
H0 (3): Regional coastline positions and
karst development represent
significantly different populations
or population distributions and
develop independently.
The first null hypothesis (H0 (1)) was
formulated in order to determine if there is a
relationship between brittle deformation and
karst development on Tinian, while the
second and third null hypotheses (H0 (2 and
3)) were formulated to determine if there is a
relationship between karst development and
coastlines or scarps, where the edge of the
paleo fresh-water lens would be expected
based on the Carbonate Island Karst Model
(Mylroie and Vacher, 1999). Because of the
wide variation in orientations of the data in
fault orientations
fault, 5-meter segment orientations
fault, 10-meter segment orientations
joint orientations
orientations of fractures measured
during fieldwork
inland scarp orientations
22
this study, a high significance level (P ≤
0.01) was used to ensure that the populations
or population distributions that were
compared represented significantly similar
datasets. If the first null hypothesis is
rejected, then the regional populations of
joints, fractures, and faults are similar to the
regional karst development and a positive
relationship between the two populations
can be inferred. If the second null
hypothesis is rejected, then the regional
populations of scarp orientations are similar
to the regional karst development and a
positive relationship between the two
populations can be inferred. If the third null
hypothesis is rejected, then the regional
populations of coastline orientations are
similar to the regional karst development
and a positive relationship between the two
populations can be inferred.
province scale, and 3) site scale. Because of
the wide distribution of cave and karst
features, one square kilometer test sites were
chosen for additional analyses. In order to
compare the results of analyses from test
sites with the larger regions of Tinian, five
parameters where required to exist within
the one square kilometer boundary,
including: 1) three or more surveyed
features, 2) both fissure and mixing zone
caves, 3) coastline, 4) faults reported by
Doan and coworkers (1960), and 5) fracture
orientations measured during fieldwork.
The requirements for choosing smallscale test sites required the completion of
fieldwork and the cave and karst inventory
of this study. Once the inventory stage was
complete, three test sites were chosen for
analysis (Figure 22, Appendix A): Carolinas
Limestone Forest, Puntan Diapblo, and Unai
Dangkolo. At each of these sites, statistical
comparisons were evaluated using the
Kolmogorov-Smirnov 2-sample test, with
the same 35 parameters evaluated for each
of the physiographic provinces and the
entire island of Tinian.
SMALL-SCALE TEST
SITE EVALUATION
In addition to the analyses performed for
the five physiographic provinces and the
island of Tinian, smaller-scale test sites were
analyzed to determine if any relationships
existed at various scales: 1) island scale, 2)
STUDY RESULTS
zone type caves (banana hole and flank
margin type caves) were the most prominent
cave types found during the inventory (81%
on Aguijan; 77% on Tinian). While Tinian
exhibited a larger diversity and quantity of
caves, the ratio of fissure caves to mixing
zone caves on both islands was 0.24 (0.235
for Aguijan; 0.238 for Tinian), with fissure
caves accounting for almost 20% of the
caves recorded.
A total of seventeen locations were
found on Tinian, which exhibited freshwater discharge (Figure 24, Appendix A).
Most of these features were identified by
schlieren mixing at the coastline, but only
five of them were associated with cave
development and surveyed. Fresh-water
discharge observed at Unai Dangkolo, Unai
CAVE AND KARST INVENTORY
The mapping portion of cave and karst
inventory surveyed 114 caves or cave
complexes (green areas, Figure 23,
Appendix A; Appendix B): 26 on Aguijan
and 88 on Tinian. The features were
classified by morphology type and grouped
by physiographic province they were found
in (Table 1, Appendix B; Table 2, Appendix
B). On Aguijan, the cave survey produced
maps of 1 banana hole, 5 fissure caves, and
20 flank margin caves. On Tinian, the cave
survey produced maps of 3 banana holes, 5
discharge features, 12 fissure caves, 65 flank
margin caves, 1 pit cave, and 2 recharge
features; however, 4 of the discharge
features were also classified as fissure caves,
making a total of 16 fissure caves. Mixing
23
Masalok and Taga Beach appeared as
diffuse discharge through carbonate sand
beach deposits, while the discharge at other
sites appeared as focused discharge along
fractures and bedding planes. The two
discharge features that exhibited the greatest
discharge were located on the east and west
coasts of the island at Gecko Cave and
Barcinas Cove, respectively.
Investigation of closed depressions
identified four definite regions of allogenic
recharge in the North-Central Highland,
with small caves associated with two of
them (Lasu Recharge Cave and West Lasu
Depression Cave). The presence of
accumulated detritus and the lack of
sediment coatings on the feature walls is
evidence that recharge is rapid and that
water does not pond at these features during
recharge events. The other two recharge
features (North and South Lemmai Recharge
Features) covered larger areas, contained
contacts between carbonate and noncarbonate rocks, and showed vegetative
evidence of water ponding within the closed
depressions prior to entering the subsurface
(Figure 25, Appendix A). Ponded water was
present in portions of South Lemmai
Recharge Feature when investigated (Figure
26, Appendix A).
The other sixteen closed depressions
identified during the initial site investigation
that were not confirmed as allogenic
recharge features were also investigated
(Figure 27, Appendix A), either physically
or through communication with local
residents. Five closed depressions were
identified as quarries or borrow pits, eight
were identified as natural constructional
features, and three appear to be recharge
features. Four of the modified closed
depressions (quarries and borrow pits)
showed evidence of excavation in the past,
while the quarry located near Barcinas Cove
is being actively excavated today (Figure 28,
Appendix A). The eight natural
construction features showed no evidence of
excavation or allogenic recharge; however,
Hagoi in the Northern Lowland is less than 2
meters above mean sea-level and contained
fresh-water at the time of survey (Figure 29,
Appendix A), while the largest closed
depression on the island (Sisonyan Makpo)
is less than 3 meters above mean sea-level at
its lowest elevation and is the site of the
islands primary municipal well (Makpo
Wells). The three unconfirmed recharge
features are located north of Mount Lasu
near small outcrops of non-carbonate rocks,
but due to their small size and dense
vegetation in the area, a positive
confirmation could not be made.
CAVE ORIENTATIONS
The two methods of cave orientation
analysis (apparent trend and entrance width
trend) were applied to orientation data for
mapped caves on Tinian. Aguijan was
excluded from this analysis because no
geologic map has been published for the
area, which would compliment the cave
orientation data when comparing cave and
karst development to brittle failure features.
Each cave was analyzed separately,
including those that occurred on maps where
a complex or series of caves were surveyed
together as one map, making it possible to
have several datasets of orientation
measurements for some cave maps. Using
the apparent trend method all fissure caves
and mixing zone caves (flank margin and
banana hole type caves) were analyzed for
primary orientation and segment
orientations. The single pit cave and two
recharge caves were excluded from all
datasets except composite cave types (all
cave types) because of their small sample
size. Using the entrance trend method, the
mixing zone caves were analyzed for
entrance width, penetration length and
maximum width, while fissure caves, the
single pit cave and the two recharge caves
were excluded. The fissure caves were
excluded from the entrance width trend
method because they are linear features with
distinct segment orientations similar to
telogenetic caves (Nelson, 1988; Barlow and
Ogden, 1982), while the single pit cave and
two recharge caves were excluded again
because of their small sample size.
24
!
The apparent trend method was applied
because of its similarity to previous studies
(Nelson, 1988; Barlow and Ogden, 1982).
Apparent trend analysis of fissure caves
analysis yielded 18 primary cave
orientations (Table 3, Appendix C): 5 in the
Central Plateau; 5 in the Median Valley; 0 in
the North-Central Highland; 1 in the
Northern Lowland; and 7 in the
Southeastern Ridge. Fissure cave segment
analysis using the apparent trend method
yielded 147 orientations with 297 five-meter
segments and 135 ten-meter segments
(Table 4, Appendix C):
!
!
!
!
!
The entrance width trend method was
performed in order to reduce the subjectivity
of orientation measurements that exist in the
apparent trend method. Measurements used
in the entrance width trend method included
entrance width, penetration length, and
maximum width, with 10 caves excluded
from the entrance width analysis because
their entrances were entered vertically as a
result of ceiling collapse. Entrance width
analysis using the entrance trend method
yielded 107 entrance widths with 324 fivemeter segments and 163 ten-meter segments
(Table 7, Appendix C):
Central Plateau: 15 orientations, 24 fivemeter segments and 12 ten-meter
segments
Median Valley: 69 orientations, 100 fivemeter segments and 44 ten-meter
segments
North-Central Highland: 0 orientations
and segments
Northern Lowland: 2 orientations, 2 fivemeter segments and 2 ten-meter segments
Southeastern Ridge: 61 orientations, 171
five-meter segments and 77 ten-meter
segments
!
!
!
!
Apparent trend analysis of mixing zone
caves yielded 128 primary orientations
(Table 5, Appendix C): 46 in the Central
Plateau; 27 in the Median Valley; 8 in the
North-Central Highland; 1 in the Northern
Lowland; and 46 in the Southeastern Ridge.
Mixing zone cave segment analysis yielded
388 orientations with 980 five-meter
segments and 480 ten-meter segments
(Table 6, Appendix C):
!
!
!
!
Southeastern Ridge: 115 orientations, 213
five-meter segments and 128 ten-meter
!
Central Plateau: 40 orientations, 105 fivemeter segments and 59 ten-meter
segments
Median Valley: 20 orientations, 89 fivemeter segments and 43 ten-meter
segments
North-Central Highland: 5 orientations,
11 five-meter segments and 6 ten-meter
segments
Northern Lowland: 1 orientation, 22 fivemeter segments and 11 ten-meter
segments
Southeastern Ridge: 42 orientations, 97
five-meter segments and 44 ten-meter
segments
Penetration length analysis using the
entrance trend method produced 118
orientations with 369 five-meter segments
and 190 ten-meter segments (Table 8,
Appendix C):
!
Central Plateau: 123 orientations, 302
five-meter segments and 122 ten-meter
segments
Median Valley: 126 orientations, 424
five-meter segments and 212 ten-meter
segments
North-Central Highland: 20 orientations,
30 five-meter segments and 9 ten-meter
segments
Northern Lowland: 4 orientations, 11
five-meter segments and 9 ten-meter
segments
!
!
!
25
Central Plateau: 42 orientations, 116 fivemeter segments and 54 ten-meter
segments
Median Valley: 26 orientations, 116 fivemeter segments and 67 ten-meter
segments
North-Central Highland: 6 orientations,
15 five-meter segments and 6 ten-meter
segments
Northern Lowland: 1 orientation, 16 fivemeter segments and 8 ten-meter segments
!
!
Southeastern Ridge: 43 orientations, 106
five-meter segments and 55 ten-meter
segments
!
Maximum width analysis using the entrance
trend method yielded 118 orientations with
407 five-meter segments and 208 ten-meter
segments (Table 9, Appendix C):
!
!
!
!
!
!
!
Central Plateau: 42 orientations, 140 fivemeter segments and 72 ten-meter
segments
Median Valley: 26 orientations, 116 fivemeter segments and 60 ten-meter
segments
North-Central Highland: 6 orientations,
13 five-meter segments and 7 ten-meter
segments
Northern Lowland: 1 orientation, 22 fivemeter segments and 11 ten-meter
segments
Southeastern Ridge: 43 orientations, 116
five-meter segments and 58 ten-meter
segments
!
Central Plateau: 94 orientations, 944
fifty-meter segments and 472 one
hundred-meter segments
Median Valley: 50 orientations, 686 fiftymeter segments and 343 ten-meter
segments
Northern Lowland: 22 orientations, 324
fifty-meter orientations and 162 ten-meter
segments
North-Central Highland: 43 orientations,
394 fifty-meter segments and 197 one
hundred-meter segments
Southeastern Ridge: 65 orientations, 618
fifty-meter segments and 309 one
hundred-meter segments
Joint data reported by Doan and coworkers
(1960) produced 38 orientations in the
Central Plateau, 25 orientations in the
Median Valley, 0 orientations in the NorthCentral Highland, 16 orientations in the
Northern Lowland, and 33 orientations in
the Southeastern Ridge (Table 11, Appendix
C).
Measurements of planes of brittle failure
measured during fieldwork yielded 345
orientations for the island of Tinian (Table
12, Appendix C). Because of dense
vegetation, soil cover, and the criteria used
for orientation sampling, the orientations
measurements were primarily limited to
coastal areas where exposed bedrock could
be observed. However, measurements were
attained in all regions accept the NorthCentral Highland, which has no coastline.
Measurements included 103 orientations in
the Central Plateau, 106 orientations in the
Median Valley, 86 orientations in the
Northern Lowland, and 50 orientations in
the Southeastern Ridge.
BRITTLE DEFORMATION
Analysis of the data from the geologic
survey conducted by Doan and coworkers
(1960) identified 313 faults (Figure 30,
Appendix A) and 112 joints on Tinian. The
data reported by Doan and coworkers (1960)
was divided by physiographic province and
measured separately, with faults that
occurred at the boundary between two
physiographic provinces not being reported,
because Doan and coworkers (1960) defined
the five physiographic provinces based on
high-angle faults that divided the island into
distinct regions. The measurements for each
of the faults were length weighted to
produce a total of 3322 fifty-meter segments
and 1661 one hundred-meter segments with
segments rounded to the nearest 50 and 100meter increment. This process produced 39
orientations with 358 fifty-meter segments
and 179 one hundred-meter segments in the
boundary areas separating different
physiographic provinces (Table 10,
Appendix C), which included:
SCARPS AND COASTLINES
Analysis of scarps based on 20-degree
slopes derived from the ten-meter posting
DEM provided the basis for scarp
orientations on Tinian (Figure 19, Appendix
A). Scarps were divided into inland scarps
and coastal scarps and measured, resulting
in a total of 154 inland scarp orientations
with 725 fifty-meter segments and 379 one
hundred-meter segments and 147 coastal
26
!
scarp orientations with 539 fifty-meter
segments and 266 one hundred-meter
segments (Tables 13 and 14, Appendix C).
The measurements were grouped by the
physiographic province for both the inland
and coastal scarps for data analysis. Inland
scarp orientations included (Table 13,
Appendix C):
!
!
!
!
!
!
!
!
Central Plateau: 27 orientations, 126
fifty-meter segments and 68 one hundredmeter segments
Median Valley: 12 orientations, 46 fiftymeter segments and 24 one hundred-meter
segments
North-Central Highland: 30 orientations,
150 fifty-meter segments and 79 one
hundred-meter segments
Northern Lowland: 0 orientations
Southeastern Ridge: 85 orientations, 403
fifty-meter segments and 208 ten-meter
segments
The North-Central Plateau is surrounded on
all sides by other physiographic provinces
and has no coastline.
ROSE DIAGRAMS
Rose diagrams were plotted for each of
the 35 parameters investigated in this study,
with diagrams corresponding to each of the
physiographic provinces and the entire
island of Tinian (Appendix C). The data
was visually analyzed to determine if any
similar populations or population
distributions existed between independent
parameters within the same region. The
orientation data showed a wide range of
variability, while some datasets contained no
orientations or only one orientation trend,
which eliminated them from the comparison.
The rose diagrams for the Central Plateau
and the Northern Lowland showed no
distinct similarities, while the diagrams for
the entire island of Tinian, Median Valley,
North-Central Highland, and Southeastern
Ridge showed few distinct similarities.
The data for the entire island of Tinian
(Tinian Composite) showed similarity in
four parameter comparisons:
Coastal scarp orientations included (Table
14; Appendix C):
!
!
!
!
!
Central Plateau: 84 orientations, 250
fifty-meter segments and 134 one
hundred-meter segments
Median Valley: 48 orientations, 169 fiftymeter segments and 94 one hundred-meter
segments
Northern Lowland: 40 orientations, 206
fifty-meter segments and 113 one
hundred-meter segments
Southeastern Ridge: 79 orientations, 320
fifty-meter segments and 160 one
hundred-meter segments
Central Plateau: 65 orientations, 539
fifty-meter segments and 101 ten-meter
segments
Median Valley: 3 orientations, 10 fiftymeter segments and 5 one hundred-meter
segments
North-Central Highland: 0 orientations
Northern Lowland: 0 orientations
Southeastern Ridge: 79 orientations, 320
fifty-meter segments and 160 one
hundred-meter segments
Coastline orientations were measured at
the 0 elevation contour on the DEM for
Tinian. Coastlines were measured similar to
island scarps, producing 251 orientations
with 945 fifty-meter segments and 501 one
hundred-meter segments for the island
(Table 15, Appendix C), which yielded 50.1
kilometers of coastline segments, closely
approximating the 51.2 kilometers of
coastline that exist (Doan et al., 1960). The
coastline measurements were grouped by the
appropriate physiographic province, with:
1.
2.
3.
27
inland scarp orientations and 10-meter
segment orientations of mixing zone cave
entrance widths
50-meter segment orientations of inland
scarps and 10-meter segment orientations
of mixing zone cave entrance widths
100-meter segment orientations of inland
scarps and 10-meter segment orientations
of mixing zone cave entrance widths
10. 50-meter segment orientations of inland
scarps and 10-meter segment orientations
of mixing zone cave entrance widths
11. 100-meter segment orientations of inland
scarps and 10-meter segment orientations
of mixing zone cave entrance widths
12. coastal scarp orientations and 5-meter
segment orientations of mixing zone cave
maximum widths
13. coastline orientations and 5-meter
segment orientations of mixing zone cave
maximum widths
4. 100-meter segment orientations of coastal
scarps and mixing zone entrance width
orientations
The data for the Median Valley showed
similarity in three parameter comparisons:
1.
orientations of fractures measured in the
field and 5-meter segment orientations of
mixing zone cave entrance widths
2. orientations of fractures measured in the
field and 10-meter segment orientations of
mixing zone cave entrance widths
3. composite scarp orientations and mixing
zone cave entrance width orientations
STATISTICAL COMPARISON
Orientation datasets were compared for
population similarities or population
distribution similarity by performing nonparametric tests on independent samples for
each of the physiographic regions and the
island of Tinian. Populations or population
distributions were considered similar if the
significance level between two independent
samples was less than 0.01 (P ≤ 0.01) when
compared using the Kolmogrov-Smirnov 2sample test. The data showed a high degree
of similarity with results varying for each
province and the island of Tinian.
The data for the North-Central Highland
showed similarity in three parameter
comparisons:
1.
inland scarp orientations and composite
cave primary orientations
2. inland scarp orientations and mixing zone
cave primary orientations
3. composite scarp orientations and
composite cave primary orientations
The data for the Southeastern Ridge showed
similarity in 13 parameter comparisons:
1.
2.
3.
4.
5.
6.
7.
8.
9.
Tinian Composite
Statistical comparisons for the entire
island of Tinian showed similarity for 266
(60.2 %) pairs of independent data. Joints
and fracture orientations measured in the
field were similar to the faults, while
segment datasets showed similarity to all
segment lengths in 37 dataset comparisons
(Table 16; Appendix D):
fault orientations and composite cave
primary orientations
fault orientations and 10-meter segment
orientations of mixing zone cave
penetrations
50-meter segment orientations of faults
and composite cave primary orientations
50-meter segment orientations of faults
and 5-meter segment orientations of
mixing zone cave penetrations
50-meter segment orientations of faults
and 10-meter segment orientations of
mixing zone cave penetrations
100-meter segment orientations of faults
and composite cave primary orientations
100-meter segment orientations of faults
and 5-meter segment orientations of
mixing zone cave penetrations
100-meter segment orientations of faults
and 10-meter segment orientations of
mixing zone cave penetrations
orientations of fractures measured in the
field and 10-meter segment orientations of
mixing zone caves
1.
2.
3.
4.
5.
6.
7.
8.
faults and inland scarps
faults and all scarps
faults and coastlines
faults and all cave types
faults and fissure caves
faults and mixing zone caves
faults and mixing zone cave penetrations
faults and mixing zone cave entrance
widths
9. faults and mixing zone cave maximum
widths
10. inland scarps and all cave types
11. inland scarps and fissure caves
12. inland scarps and mixing zone caves
28
8.
13. inland scarps and mixing zone cave
penetrations
14. inland scarps and maximum cave widths
15. coastal scarps and all cave types
16. coastal scarps and fissure caves
17. inland scarps and mixing zone caves
18. inland scarps and mixing zone cave
penetrations
19. inland scarps and mixing zone cave
entrance widths
20. inland scarps and mixing zone cave
maximum widths
21. coastal scarps and all cave types
22. coastal scarps and fissure caves
23. coastal scarps and mixing zone caves
24. coastal scarps and mixing zone cave
penetrations
25. coastal scarps and mixing zone cave
entrance widths
26. coastal scarps and mixing zone cave
maximum widths
27. all scarps and coastlines
28. all scarps and all cave types
29. all scarps and fissure caves
30. all scarps and mixing zone caves
31. all scarps and mixing zone cave
penetrations
32. all scarps and mixing zone cave entrance
widths
33. all scarps and mixing zone cave maximum
widths
34. coastlines and all cave types
35. coastlines and fissure caves
36. coastlines and mixing zone caves
37. coastlines and mixing zone cave
penetration
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
Northern Lowland
Statistical comparisons for the Northern
Lowland showed similarity for 72 (52.9 %)
pairs of independent data. Although only 72
datasets showed similarity, this is more than
half of the total comparisons for the
province, because data was limited for this
region. Analysis of the DEM produced no
scarps and the cave and karst inventory only
mapped two features (Rogue Cave and Unai
Lamlam) in the Northern Lowland. Joints
and fracture orientations measured in the
field were similar, while segment datasets
showed similarity for all segment lengths in
eight dataset comparisons (Table 18,
Appendix D):
1.
2.
3.
faults and all cave types
faults and mixing zone cave penetrations
faults and mixing zone cave entrance
widths
4. faults and mixing zone cave maximum
widths
5. coastlines and all cave types
6. coastlines and mixing zone cave
penetrations
7. coastlines and mixing zone cave entrance
widths
8. coastlines and mixing zone maximum
widths
Central Plateau
Statistical comparisons for the Central
Plateau showed similarity for 199 (45.0 %)
pairs of independent data. Joints and
fracture orientations measured in the field
are similar to the faults, while the segment
datasets showed similarity for all segment
lengths in 19 dataset comparisons (Table 17,
Appendix D):
1.
2.
3.
4.
5.
6.
7.
faults and mixing zone cave entrance
widths
faults and mixing zone cave maximum
widths
inland scarps and coastal scarps
inland scarps and all scarps
inland scarps and all cave types
coastal scarps and all cave types
coastal scarps and mixing zone caves
coastal scarps and mixing zone cave
entrance widths
all scarps and mixing zone cave entrance
widths
coastlines and all cave types
coastlines and mixing zone caves
coastlines and mixing zone cave entrance
widths
faults and inland scarps
faults and coastal scarps
faults and all scarps
faults and all cave types
faults and fissure caves
faults and mixing zone caves
faults and mixing zone cave penetrations
North-Central Highland
Statistical comparisons for the NorthCentral Highland showed similarity for 12
29
20. all scarps and mixing zone cave entrance
widths
21. all scarps and mixing zone cave maximum
widths
22. coastlines and all cave types
23. coastlines and fissure caves
24. coastlines and mixing zone cave types
25. coastlines and mixing zone cave
penetrations
26. coastlines and mixing zone cave entrance
widths
27. coastlines and mixing zone maximum
widths
(7.0 %) pairs of independent data. This
province showed the least similarity
between orientations datasets for all segment
lengths; however, data were limited for the
region. Because this province is completely
surrounded by other provinces, no coastal
scarps or coastlines exist for this province.
Additionally, structural data were limited to
the faults reported by Doan and coworkers
(1960) and no fissure caves were located in
this region during fieldwork. Segment
datasets showed similarity in only two
dataset comparisons (Table 19, Appendix
D):
1.
Southeastern Ridge
Statistical comparisons for the
Southeastern Ridge showed similarity for
236 (53.4 %) pairs of independent data.
Joints and fractures measured during
fieldwork were not similar to the faults for
this province, but segment datasets showed
similarity for all segment lengths in 23
dataset comparisons (Table 21, Appendix
D):
faults and inland scarps
2. faults and all cave types
Median Valley
Statistical comparisons for the Median
Valley showed similarity for 189 (42.8 %)
pairs of independent data. Joints and
fractures measured during fieldwork were
not similar to the faults for this province, but
segment datasets showed similarity for all
segment lengths in 27 dataset comparisons
(Table 20, Appendix D):
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
1.
2.
3.
4.
faults and inland scarps
faults and all scarps
faults and all coastlines
faults and all cave types
faults and mixing zone caves
faults and mixing zone cave penetrations
faults and mixing zone cave entrance
widths
faults and mixing zone cave maximum
widths
inland scarps and coastal scarps
inland scarps and coastlines
inland scarps and all cave types
inland scarps and mixing zone caves
inland scarps and mixing zone cave
penetrations
inland scarps and mixing zone cave
entrance widths
inland scarps and mixing zone cave
maximum widths
all scarps and all coastlines
all scarps and all cave types
all scarps and mixing zone caves
all scarps and mixing zone cave
penetrations
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
30
faults and all cave types
faults and fissure caves
faults and mixing zone cave penetrations
faults and mixing zone cave entrance
widths
faults and mixing zone cave maximum
widths
inland scarps and all cave types
inland scarps and fissure caves
inland scarps and mixing zone cave
penetrations
inland scarps and mixing zone cave
entrance widths
inland scarps and mixing zone cave
maximum widths
coastal scarps and all cave types
coastal scarps and fissure caves
coastal scarps and mixing zone cave
penetrations
coastal scarps and mixing zone cave
entrance widths
all scarps and all cave types
all scarps and fissure caves
all scarps and mixing zone cave
penetrations
all scarps and mixing zone cave entrance
widths
all scarps and mixing zone cave maximum
widths
20. coastlines and all cave types
21. coastlines and fissure caves
22. coastlines and mixing zone cave
penetrations
23. coastlines and mixing zone cave entrance
widths
Puntan Diapblo
Statistical comparisons for the Puntan
Diapblo test site showed similarity for 67
(16.4 %) pairs of independent data. The test
site included three fissure caves (Death
Fracture Complex) and nine mixing zone
caves (Cavelet Cave, Dos Sakis Cave
Complex, Flamingo Tail Cave Complex,
Monica Cave, Orange Cave, and Liyang
Diapblo). No joint data was reported for the
area, but fracture orientations measured in
the field showed similarity to faults, inland
scarps, all cave types, fissure caves, mixing
zone cave entrance widths, and mixing zone
cave maximum widths. Segment datasets
showed similarity to all segment lengths in
two dataset comparisons (Table 23,
Appendix D):
SMALL-SCALE TEST SITES
Three one square kilometer tests sites
(Figure 22, Appendix A; Carolinas
Limestone Forest, Puntan Diapblo, and Unai
Dangkolo) were analyzed separately to
determine if orientation data from
independent populations showed similarity
at a smaller scale. Statistical comparisons
were made for each of the three test sites
using the same criteria as was used for the
province and island scale comparisons. As
in the larger scale orientation comparisons,
datasets were considered to represent similar
populations or population distributions if
they showed a significance level less than
0.01 (P ≤ 0.01) when compared using the
Kolmogrov-Smirnov 2-sample test. The
results of the comparisons for the test sites
showed less similarity than the data
evaluated over larger regions, but did show
significant similarity.
1.
faults and inland scarps
2. faults and mixing zone cave penetrations
Unai Dangkolo
Statistical comparisons for the Unai
Dangkolo test site showed similarity for 63
(44.7 %) pairs of independent data. The test
site included one fissure cave (Dripping
Tree Fracture Cave) and five mixing zone
caves (Andyland Cave, John's Small Cave,
Liyang Dangkolo, North and South Unai
Dangkolo). No joint data was reported for
the area and no scarps could be resolved
from the DEM for the area. Fractures
measured in the field showed similarity to
the coastline, mixing zone cave penetrations,
mixing zone cave entrance widths, and
mixing zone cave maximum widths.
Segment datasets showed similarity to all
segment lengths in seven dataset
comparisons (Table 24, Appendix D):
Carolinas Limestone Forest
Statistical comparisons for the Carolinas
Limestone Forest test site showed similarity
for 64 (18.7 %) pairs of independent data.
The test site included three fissure caves
(Carolinas Fracture Cave, Plunder Cave, and
Water Cave) and one mixing zone cave
(Skip Jack Cave). No joint data was
reported for this area, but fracture
orientations measured in the field showed
similarity to faults. Segment datasets
showed similarity for all segment lengths in
three dataset comparisons (Table 22,
Appendix D):
1.
2.
1.
2.
faults and all cave types
faults and fissure cave
3. faults and mixing zone cave maximum
widths
3.
4.
5.
6.
7.
31
faults and coastlines
faults and mixing zone cave entrance
widths
faults and mixng zone cave maximum
widths
coastline and all cave types
coastline and fissure caves
coastline and mixing zone caves
coastline and mixing zone cave
penetrations
DISCUSSION AND CONCLUSIONS
margin caves that have been breached by
TINIAN CAVE AND
coastal processes (Figure 31, Appendix A),
KARST INVENTORY
The Tinian cave and karst inventory
with speleothems and remnant cave
surveyed eighty-eight caves (Figure 23.
chambers confirming their origin. One
Appendix A), located seventeen sites of
feature located north of Unai Dangkolo
freshwater discharge (Figure 24, Appendix
(Hidden Beach Cave, Figure 32, Appendix
A), and four allogenic recharge areas
A) confirmed that these were collapsed flank
(Figure 27, Appendix A). Cave
margin caves because it was breached at sea
development on Tinian is dominated by
level and contained a carbonate sand floor,
mixing zone type caves (flank margin and
but retained several regions of intact ceiling
banana hole type caves), but fissure caves
rock. Flank margin caves have a wide
account for twenty percent of the total.
range of sizes, from a few square meters to
Only one pit cave and two recharge caves
the archetypical Tinian flank margin cave
were identified (see Appendix B for maps
(Liyang Dangkolo, Figure 33, Appendix A)
and descriptions of individual cave and karst
that is over 1,300 square meters. Smaller
features).
caves have single chambers and larger caves
The lack of pit caves and recharge caves
have central chambers with smaller,
suggests that these features are uncommon
interconnected passages extending from
on Tinian; however, it is possible that more
them (Figure 12), demonstrating how
exist and that this is a sampling bias. Tinian
mixing zone caves become interconnected
experienced intense sugarcane farming
as they grow in size (Figure 13). Horizons
during the Japanese occupation and
of flank margin cave development (Figure
extensive military construction by the U. S.
14) occur at several locations, including
and Japan during and after World War Two.
Mendiola Cove, the southeastern portion of
If more of these features existed in
the past, it is probable that they
were infilled intentionally as part
of island development or infilled
by soil erosion from deforestation,
farming and construction.
Although soils may be limited in
eogenetic karst environments, it is
probable that the soil present will
erode more easily as a result of
human modification of the land
surface.
Mixing zone caves were
located in every physiographic
province and at elevations from
sea level to over 150 meters.
Three caves were classified as
banana holes because they are
small features located shallow in
the subsurface and are
significantly wider than tall;
however, these may represent
Figure 12: Flank margin caves develop more complicated
morphologies as they grow in size as observed on
small flank-margin caves. Most
Tinian (note these figures only include bedrock
coves represent collapsed flank
walls and columns in the cave maps).
32
Figure 13: Conceptual model for the growth of flank margin caves. At small scales (vug scale) mixing
zone dissolution forms simple ellipse because the rock is locally homogeneous. However,
at larger scales (cave scale) inhomogeneities in rock form more complex morphologies with
growth as a result of preferential dissolution of the bedrock.
Piña Ridge, Suicide Cliffs, Unai Dangkolo,
and Unai Masalok. These horizons indicate
at least three previous fresh-water lens
positions on Tinian (Figure 34, Appendix
A), but complex faulting prevents the direct
correlation of horizons of mixing zone
development across large regions.
Fissure caves show linear development
that appears to be associated with brittle
failure (Figure 35, Appendix A). They were
located in all physiographic provinces
except the North-Central Highland and at
elevations ranging from sea level to over
100 meters. These caves can extend to
significant depths when located in inland
regions and provide fast flow routes for
autogenic recharge. The deepest fissure
cave (Masalok Fracture Cave) is 42 meters
deep and has pools of fresh water at the
bottom. Coastal fissure caves generally
extend inland at angles near perpendicular to
the coastline and frequently have freshwater discharge associated with them. Other
fissure caves (Plunder Cave and Water
Cave) form broader, linear, dipping
chambers with much collapse and are
located near reported faults (Doan et al.,
1960). No evidence of offset resulting from
faulting could be seen in these caves, either
because it does not exist or because
extensive flowstone and speleothem
development covering the walls obscured it
(Figure 36, Appendix A).
Groundwater recharge on Tinian is
primarily autogenic. Closed depressions and
recharge caves in the North-Central
Highland indicate allogenic recharge occurs
where volcanic rocks crop out. The small
igneous outcrop near the municipal well
does not show direct evidence of supplying
waters for allogenic recharge, but clays
weathered from the volcanic exposure may
be armoring the carbonate rock slope below,
decreasing the effectiveness of autogenic
recharge in these rocks and providing some
allogenic recharge to the large closed
depression where the municipal well is
located. If this is occurring, this allogenic
recharge is likely to be minimal because the
igneous rocks crop out only on the steep
33
Figure 14: Multiple levels of mixing zone dissolution exist on Tinian with at least three identified in
the North-Central Highland near Mount Lasu. However, more probably exist, but island
faulting prevents correlation of horizons of development across large regions.
slope over a small area. In addition to the
closed depressions that showed evidence of
allogenic recharge, five closed depressions
were identified as quarries or borrow pits.
Although these features have autogenic
recharge, recharge rates in these closed
depressions is expected to be faster than in
other regions of autogenic recharge because
the soil cover and surface rock have been
removed.
Fresh-water discharge was observed
across much of the east and west coast.
Much of the coastline could not be
investigated due to strong surf; therefore, it
is expected that many more discharge sites
exist. The discharge sites that were located
were generally associated with focused
discharge along bedding planes or fractures,
indicating that much of the fresh-water
discharge is occurring along preferential
flow routes, which would be expected to
distort the lens morphology regionally.
Areas near fissure caves that were
discharging fresh-water did not have any
34
breached flank-margin caves nearby. This
may indicate that these features have
distorted the fresh-water lens to a great
enough degree that the mixing zone caves
are not forming in these areas or that they
are developing further inland and have not
been breached by coastal processes as have
other flank margin caves seen along
coastlines elsewhere.
The one problematic cave (Liyang
Atkiya) does not show typical island karst
development. It consists of a large entrance
chamber, which descends steeply and is
floored with breakdown. At the base of this
chamber, the walls and floor are coated with
a thick coating of dark sediment. The
chamber also contains several small pools of
water. Extending from the large chamber, a
linear passage continues for several hundred
meters, while continuing to descend
gradually. Old Scallops on the walls of this
linear passage indicate that in the past water
was flowing upwards from deeper in the
cave towards the entrance chamber (Figure
38, Appendix A). The linear passage
eventually splits and small mazelike tubes
are encountered, which were not completely
surveyed. The origin of this feature is
problematic because scallops are not
normally seen in island karst because of the
dominance of mixing zone dissolution.
Kalabera Cave on Saipan contains scallops
(Jenson et al., 2002), which indicate that
water was rising as a lift tube along a
lithologic barrier of non-carbonate rocks.
However, the lack of evidence of noncarbonate rocks in Liyang Atkiya prevents
applying the Kalabera Cave model to this
cave. Another investigator sampled black
sediments from Liyang Atkiya. Analysis of
this material may provide further insight into
the origin of this cave.
AGUIJAN CAVE AND KARST
INVENTORY
The cave and karst inventory surveyed
26 features on Aguijan (see Appendix B for
maps and descriptions of individual cave
and karst features); however, no site of
fresh-water discharge was identified because
the time was limited on the island and the
coastline consists of large scarps that are
subject to constant heavy surf. The
inventory identified two banana holes,
nineteen flank margin caves, five fissure
caves, and one problematic cave.
According to local guides, these 26 features
represent the majority of the caves on
Aguijan. The general morphology of
Aguijan appears similar to the Carolinas
Ridge on Tinian and is expected to have a
similar geologic history, although no
detailed geologic studies have been
conducted on Aguijan to confirm this.
Fissure and mixing zone caves are
located at all three terrace levels and on all
sides of the island. Boonie Bee Sink may be
a small flank margin cave, but because of its
general morphology it was classified as a
banana hole. The flank margin caves range
in size from a few square meters to hundreds
of square meters. The fissure caves show
three distinct morphologies, which appear to
correspond with the morphologies seen on
Tinian. Two of the fissure caves (Insect Bat
Cave and Toppled Column Cave), located at
the edge of the middle terrace, are similar to
freshwater discharge caves seen on Tinian
and Guam (Taborosi, 2002), but do not
discharge freshwater. These two features
have been identified as paleo-discharge
features (Figure 37, Appendix A).
CONTROLS ON CAVE AND
KARST DEVELOPMENT
The second objective of this study is to
determine if comparisons of cave
development orientation trends to brittle
failure trends show similarity that would
indicate structural influence on eogenetic
karst development or if karst development is
related to scarp and coastline positions
where the margin of the paleo or current
fresh-water lens is expected to be. Cave and
karst development in continental settings is
significantly influenced by geologic
structure and lithology (Klimchouk and
Ford, 2000; White, 1988), while island karst
is dominated by mixing zone dissolution
35
with the fresh-water lens position
significantly influencing porosity
development. Jenson and coworkers
(Jenson et al., 2002; Mylroie et al., 2001)
have recently recognized the importance of
structural and geologic controls on
eogenetic, island karst development, but the
extent of the influence that geologic
structure and lithology have on carbonate
island karst development has not been well
studied. The
morphological
difference in fissure
caves and mixing
zone caves suggests
that each class of
cave development is
dominated by
different controls on
dissolution. A
simple comparison
of the length to
width ratio for
fissure caves and
mixing zone caves
shows that these
two general cave
types represent
Figure 15: Diagram showing the relationship between cave widths and
different
lengths, which represent two distinct populations for fissure caves
populations (Figure
and mixing zone caves. Note that as flank margin caves grow in
15).
size they have a length to width ratio close to one.
This portion of
the study was only
conducted on data
from Tinian, where
the geology and
structural
deformation had
been mapped (Doan
et al., 1960),
because of the focus
on the relationship
between zones of
brittle failure and
karst development.
Several biases may
have been
introduced into the
data that would
affect the results of
the comparison of
orientation
Figure 16: Diagram showing the relationship cave entrance width and cave
populations. It is
maximum width. Note that a large portion of the caves plot with a
inevitable that in
ratio of one, indicating that the entrance width and the maximum
the cave and karst
width are the same.
36
inventory some features were not
inventoried because of time constraints in
the field, dense vegetation concealing
entrances, safety issues, or because the caves
must be breached in order to enter and
survey them. The methodology used for
determining cave orientations was subjective
and depended on the accuracy of the cave
survey and the parameters used to determine
orientations, especially in mixing zone caves
that are globular or elliptical. The segment
lengths that were used in comparisons had
the potential of introducing error using the
Kolmogorov-Smirnov 2-sample test
(Nelson, 1988; Barlow and Ogden, 1992),
therefore two sets of segment data were used
for all orientation categories. Data types
were only considered to be similar if they
showed similarity amongst all segment
lengths for those data pairs being compared.
The degree of breaching may have biased
the data, where collapse, cliff retreat or
coastal erosion removed portions of the
original cave, leaving an incomplete
remnant. If a flank margin cave were
modeled as a simple ellipse or if flank
margin caves are linear features developed
parallel to the cliff, then removal of one half
or more of the feature by erosion would
make the entrance width and the maximum
width of the feature the same. In many
cases, it appears that at least 50% of the
original flank-margin caves had been
removed by cliff retreat (Figure 16). The
cell size of the DEM limited the resolution
of scarps and coastlines. The measurement
of fracture orientations in the field was
subject to human error, as were the faults
and joints reported by Doan and coworkers
(1960), which were based on interpretation
of data gathered during fieldwork conducted
in the 1950's. All of these biases may have
affected the comparison of independent data
populations, therefore only data pairs that
showed a high degree of similarity (P ≤
0.01) were considered to be related.
Comparisons between orientations of
brittle failure features, scarp and coastline
orientations, and cave primary and segment
orientations show a wide range of
orientation trends at the island and province
scale, but more distinct trends at smallerscale test sites. This wide range of
variability at the larger scales may be the
result of the physiographic nature of islands.
Coastline and scarp orientation trend
datasets show a wide range of orientation
trends at large region or entire island scales
because of the roughly elliptical shape of
islands. Island coastlines have orientation
trends between 0o and 360o because islands
are surrounded on all sides by water,
therefore a wide range of orientations exist
for coastline orientations at the island scale.
Similarly, any ridge that may produce scarps
will be roughly elliptical in shape and also
show a wide range of orientations.
However, trend orientations associated with
the long axis of these features will show
greater dominance as the length to width
ratio increases (L/W > 1) for the island and
scarps (Figure 17).
The orientations of brittle deformation
structures reported by Doan and coworkers
Figure 17: Coastline orientations for Tinian show a
wide range of trends because of the
elliptical shape of the island. Note that
the rose diagram pattern for coastlines
resembles the coast outline of Tinian.
37
(1960) and those measured during fieldwork
show a wide range of trends. This is
probably the result of Tinian's complex
tectonic setting. Tinian is a Paleogene
volcanic edifice mantled by younger
carbonate rocks with complex, high-angle
normal faulting throughout the island. The
geomorphology of Tinian is primarily
controlled by three factors: 1) the original
volcanic depositional regime, 2) the original
carbonate depositional regime, and 3)
structural deformation primarily in the form
of brittle failure. The coastlines of Tinian
are primarily erosional, with modern
carbonate beach deposits representing less
than 5% of the coastline (light blue regions
in Figure 39, Appendix A), and strongly
influenced by geologic structure. Because
of the erosional nature of the coastline and
the geomorphology of Tinian, three types of
brittle failure are expected to exist (Figure
38): 1) regional faulting associated with
island arc tectonism, 2) brittle failure nearparallel to coastlines associated with margin
failures, and 3) brittle failure nearperpendicular to coastlines associated with
tension release structures that form
perpendicular to margin failures (Doan et
al., 1960). In addition to the three primary
types of brittle failure expected, rock units
may also be fractured by passive, isostatic
subsidence (>0.05 mm/yr, Dickinson, 1999).
The combination of the faulting and
fracturing from subsidence and margin
failure creates great variability in the
orientations of brittle failure features
observed on Tinian.
development, because scarps, coastlines,
brittle failure features and cave development
all show similarities indicating that at large
scales the interaction of these variables is
too complex to define cause and affect.
However, rose diagrams show that, at the
island scale, scarps are the only features that
show similarity to cave development,
suggesting that cave development is
primarily controlled by the position of the
fresh-water lens if the scarps do represent
paleo-coastlines, as predicted in the
Carbonate Island Karst Model (Stafford et
al., 2003; Jenson et al., 2002; Mylroie and
Jenson, 2002; Mylroie and Jenson, 2001).
PROVINCE SCALE COMPARISONS
At the physiographic province scale,
fewer independent pairs showed similarity;
however, rose diagrams showed more
similar pairs. Data for the Northern
Lowland and North-Central Highland
showed the least degree of similarity in
comparisons, which is probably due to a
sampling bias of cave and karst features.
Only two features were located in the
Northern Lowland (one fissure cave and one
mixing zone cave) and eight mixing zone
caves were located in the North-Central
Highland. Therefore, these small sample
sizes suggest that these regions do not have
enough data to be analyzed with any
confidence. The other physiographic
provinces produced larger datasets and
appeared to have less sampling bias.
The cave inventory of the Central
Plateau surveyed 5 fissure caves and 46
mixing zone caves. Rose diagrams show no
distinct similarities amongst independent
orientation populations. However, statistical
analysis of the data showed significant
similarity in 45% of the independent sample
pairs. Analysis of the data showed
similarity between faults, scarps, coastlines,
fissure caves and mixing zone caves. This
presents the same problem as the islandscale analyses. The Central Plateau, which
extends from the east coast to the west coast,
exhibited too much variability in the data to
determine if brittle failure features,
ISLAND SCALE COMPARISONS
At the island scale, comparisons showed
60% similarity in paired, independent
samples. This high degree of similarity
appears to indicate that the complex nature
of brittle deformation and the great
variability of coastline and scarp
orientations make it impossible to
differentiate between relationships at the
island scale. The similarity also does not
enable the elimination of variables that may
not be influencing cave and karst
38
coastlines or scarps were the dominant
control on cave development. The data for
the Central Plateau suggests that brittle
failure does significantly influence cave
development in the region. In the Central
Plateau similarity with caves to scarps and
coastlines (paleo and modern fresh-water
lens) appeared to dominate in all cases
except fissure caves, which only showed
significant similarity to faults.
The cave inventory of the Median
Valley surveyed 5 fissure caves and 27
mixing zone caves. Rose diagrams showed
similarities between fractures, scarps, and
mixing zone cave entrance widths,
suggesting that the breaching of mixing zone
caves is associated with scarp failure as
expected. Statistical analysis of data
showed significant similarity for 43% of the
independent data pairs, with similarity
between faults, scarps, coastlines and cave
development. The only statistical similarity
seen for fissure caves was with coastlines.
The Median Valley, which extends from the
east coast to the west coast, again showed
too much variability in the data to determine
if brittle failure features, coastlines or scarps
were the dominant control on cave
development. The similarity between
fissure caves segments and coastlines,
suggests that the fissure caves in this region
are associated with bank-margin failure and
not regional faulting; however, the segment
data is being highly biased towards one cave
(Dripping Tree Fracture Cave) that is
significantly longer than the other fissure
caves combined.
Seven fissure caves and forty-six mixing
zone caves were inventoried on the
Southeastern Ridge. Rose diagrams showed
more distinct similarities in this province
than in others. Fault orientations appeared
similar to mixing zone primary orientations
and penetrations, while scarps and coastlines
showed similarity to mixing zone cave
widths. Statistical analysis showed
similarities for 53% of the independent data
pairs, with similarity between faults, scarps,
coastlines and cave development. As in the
Central Plateau and Median Valley, the data
showed too much variability to determine
the dominant control on cave development.
However, the data for the Southeastern
Ridge suggests that the fissure caves in this
region are associated with scarp failures,
while mixing zone caves are controlled by a
combination of fault and fresh-water lens
position. Mixing zone cave penetration
shows significant similarity to faults, while
mixing zone cave widths are significantly
similar to scarps and coastlines. This
suggests that mixing zone dissolution at the
edge of the fresh-water lens expanded
laterally in relation to the edge of the lens
and inland in relation to regional fault
patterns.
SMALL-SCALE TEST SITE
COMPARISONS
The one square kilometer test sites
showed a lesser degree of orientation
variability than was observed in the largerscale comparisons. Because these were only
test sites used to evaluate whether or not
orientation similarities exist at different
scales on the island, no rose diagrams were
used. Instead, analyses were limited to
statistical comparisons. The three test sites
chosen for small-scale analyses all contained
fissure caves, mixing zone caves, faults, and
coastline, and were locations where fractures
had been measured in the field.
The Carolinas Limestone Forest test site
contained three fissure caves and one mixing
zone cave. Data analyses showed
similarities for 19% of the independent data
pairs. Fractures measured in the field
showed significant similarity to regional
faults. Segment analyses only showed
significant similarity for all segment lengths
in three data pairs. The significantly similar
segment orientations were all related to
regional faults and caves, with faults being
similar to fissure caves, mixing zone cave
maximum widths and all cave type
orientations. This data suggests regional
faulting controls fissure caves in this area.
The similarities between all cave types and
mixing zone caves are not considered
reliable because of the small sampling size
39
of mixing zone caves for this regions, but
the do suggest that regional brittle
deformation may be affecting mixing zone
development.
The Puntan Diapblo test site contained
three fissure caves and nine mixing zone
caves. Data analyses showed similarities
for 16% of the independent data pairs.
Fractures measured in the field showed
similarity to faults, scarps, fissure caves and
mixing zone cave widths. As in the
Carolinas Limestone Forest test site,
similarity between segment orientations
were only seen in relation to faults. Faults
were similar to inland scarps and mixing
zone cave penetrations, suggesting regional
jointing and faulting influence mixing zone
cave development. The lack of similarity
between faults and the fissure cave
development does not mean that these
fissure caves are not controlled by brittle
deformation. Similarity seen between
fractures measured in the field and the
fissure caves suggests that these fissure
caves are controlled by fractures that are
near perpendicular to the coastline and may
be related to unloading structures associated
with regional isostatic subsidence or tension
release features.
The Unai Dangkolo test site contains
one fissure cave and five mixing zone caves.
Data analyses showed similarities for 45%
of the independent data pairs, which appears
high. Data for this test site was limited
because analyses of the DEM showed that
no major scarps exist in this area. Fractures
measured during fieldwork and faults both
showed similarity to the coastline and
mixing zone cave development. Coastline
orientations showed similarity to fissure
caves and mixing zone caves. This data
suggests that mixing zone caves in this area
are influenced by both the coastline (edge of
the fresh-water lens) and regional faulting.
Fissure cave development appears to be
completely associated with coastlines,
suggesting that Dripping Tree Fracture cave
is related to bank-margin failure parallel to
the coastline.
STRUCTURAL CONTROL OF CAVES
Eogenetic karst development on Tinian
is dominated by mixing zone dissolution;
however, brittle deformation appears to have
a significant influence. Seventy-seven
percent of the caves surveyed on Tinian are
mixing zone caves, while only twenty
percent are fissure caves. These two groups
of caves are significantly different and their
developmental controls are different.
Orientations of fissure caves showed
significant similarity to zones of brittle
failure; however, the type of brittle failure is
varied. Independently, orientations of
fissure caves show similarities to regional
faulting, jointing, coastline and scarp
position. The statistical comparisons imply
that fissure cave development is controlled
by brittle failure that results from island
tectonism (high-angle faulting), passive
isostatic subsidence (joints), and scarp
failure (fractures). Additionally, most freshwater discharge observed on Tinian was
associated with fissure caves or smallerscale dissolutionally widened bedrock
fractures.
Mixing zone caves appear to be
primarily controlled by the fresh-water lens
position; however, orientation similarities
between mixing zone caves and brittle
deformation features exist in analyses
ranging from the small-scale test areas to the
entire island of Tinian. Although the
relationships are not consistent from region
to region, analyses of the orientation data
suggest that brittle deformation does
significantly affect mixing zone
development.
KARST DEVELOPMENT ON
AGUIJAN AND TINIAN
This study indicates that karst
development on Tinian is dominated by
mixing zone dissolution, but that geologic
structure, in the form of brittle deformation,
plays a significant role. Significant
similarities between independent
populations or population distributions
confirm that the structural controls on
fissure cave development inferred from cave
40
morphology are correct and that mixing
zone caves are being significantly
influenced by brittle failure. Based on the
presence of at least three zones of mixing
zone cave development, it is believed that
Tinian and Aguijan have experienced
several stable sea-level stillstands that lasted
for significant periods. These horizons
probably represent previous sea-level
stillstands related to glacio-eustacy with
constant uplift; however, they may indicate
episodic island uplift. Some flank-margin
caves are vertically exaggerated; suggesting
slow uplift, slow sea level change or a
combination of the two at times.
The data indicate that Tinian does not
conveniently fit in to any single model for
carbonate island karst. The data do not
suggest that Tinian is a Complex Carbonate
Island because no intricate interfingering of
carbonate and non-carbonate facies were
identified, although the island does show a
high degree of faulting. The high-angle
faulting does appear to separate the island
into distinct regions and significantly distort
the fresh-water lens as indicated by freshwater discharge sites identified near the
boundaries between the Median Valley and
Southeastern Ridge and between the
Northern Lowland and Central Plateau. The
Northern Lowland best fits the Simple
Carbonate Island Karst Model, with fresh-
water exposed at the surface at Hagoi and no
non-carbonate rocks cropping out. The
Southeastern Ridge best fits the CarbonateCover Island Karst Model, with several
areas of mixing zone cave development and
non-carbonate rocks that outcrop, but do not
show any direct evidence of allogenic
recharge. The Central Plateau, NorthCentral Highland and Median Valley cannot
be easily fit into separate models, but
together best fit the Composite Island Karst
Model with allogenic recharge occurring at
the non-carbonate / carbonate rock contacts
in the North-Central Highland. Because the
central portion of the Tinian best fits the
Composite Island Karst Model, the entire
island must be classified as this although the
northern and southern portions of the island
do not represent this model well.
Aguijan can only be classified as a
Simple Carbonate Island because only
carbonate rocks are known. However, the
geomorphology of Aguijan is similar to that
of the Southeastern Ridge of Tinian,
indicating that it is most likely a CarbonateCover Island. Further complicating the
interpretation of Aguijan is Liyang Atkiya,
which may indicate some complex
interaction between carbonate and noncarbonate rocks.
SUMMARY
This study inventoried and surveyed 26
cave and karst features on Aguijan and 88
cave and karst features on Tinian and is
believed to have adequately sampled the
cave and karst development on the islands,
although it is probable that more features
exist. Two distinct classes of cave and karst
features were identified: mixing zone caves
(flank margin and banana hole type caves)
and fissure caves (linear caves associated
with planes of brittle failure). Most mixing
zone caves are located in or near scarps and
coastlines, at elevations from sea level to
160 meters, have areas from a few square
meters to more than 1300 square meters, and
are often at consistent levels with nearby
caves. At least three distinct horizons of
breached mixing zone caves occur on
Tinian, representing previous fresh-water
lens positions that have been tectonically
uplifted and breached by erosional
processes. Fissure caves were located in
regions that showed evidence of brittle
failure produced from active tectonic uplift,
passive isostatic subsidence, or bank margin
and smaller scale scarp failure. The fissure
caves range in length from a few tens of
meters to hundreds of meters and are
41
relatively narrow. Some of these features
reached significant depth. The fissure caves
show direct evidence of vadose fast flow
routes, which can rapidly transmit fluids
more than 40 meters into the subsurface and
can distort the fresh-water lens morphology.
Although some fissure caves extended over
longer distances and to greater depths than
the mixing zone caves, mixing zone caves
were found to be significantly more
abundant (mixing zone cave / fissure caves
= 4).
In addition to the karst inventory, freshwater discharge and allogenic recharge sites
were investigated. Along the coastline of
Tinian, 17 fresh-water discharge sites were
identified, many associated with sea level
caves. No active discharge sites were
identified on Aguijan because strong surf
conditions prevented exploration, but two
paleo-discharge sites were identified on the
Middle Terrace. Four closed depressions
were identified as sites of allogenic recharge
on Tinian at contacts between non-carbonate
igneous outcrops and carbonate rocks,
including two with caves that receive direct
recharge. No allogenic recharge sites were
identified on Aguijan because only
carbonate rocks outcrop on the island.
Orientation data showed that brittle
failure features significantly influence karst
development on Tinian. At the island scale
and regional scale, it is difficult to determine
if megaporosity (cave) development is
associated with geologic structure or with
fresh-water lens position (scarps and
coastlines), because of the wide range of
data orientations on Tinian, resulting from
the island geomorphology and complex
tectonic setting. However, similarities
between data populations or population
distributions suggest that geologic structure
does affect karst development. At smaller
scales, more direct evidence of structural
controls on megaporosity development are
seen, probably resulting from the narrower
range of orientation data present at these
scales because of smaller sample sizes.
Analyses of cave maps in relation to
brittle failure and island geomorphology
indicates that the interpretations for origin of
fissure caves that have been inferred from
cave morphology are correct. Fissure cave
orientation trends in the vicinity of faulting
show significant similarity to faults,
implying that the caves formed along
preferential flow paths created by faulting.
Fissure cave orientation trends near
coastlines and scarps show significant
similarities to nearby fractures, implying
that the caves formed along preferential flow
paths created by bank-margin and smallerscale scarp failure or in relation to tension
release structures. Orientations of mixing
zone caves show significant similarities to
scarps and coastlines, implying that the
caves formed in relation to the edge of the
fresh-water lens. However, mixing zone
caves often show similarities with brittle
failure features, suggesting that mixing zone
caves are significantly influenced by
geologic structure, although mixing zone
dissolution along the edge of the fresh-water
lens is the dominant controlling factor.
Clearly, the interaction between
megaporosity development and brittle
deformation on carbonate islands is
complex, often resulting in features that are
primarily controlled by different factors
(fresh-water lens position or brittle
deformation) developing close to each other
(Figure 18).
Tinian and Aguijan show that the
Carbonate Island Karst Model cannot be
simply applied to entire carbonate islands
that occur in complex tectonic settings. In
order to apply the Carbonate Island Karst
Model to Tinian it must be divided into
different regions. The Northern Lowland
best fits the Simple Carbonate Island Karst
Model. The Southeastern Ridge best fits the
Carbonate-Cover Island Karst Model. The
North-Central Highland, Central Plateau,
and Median Valley cannot be easily
separated, but when grouped together best
fit the Composite Island Karst Model.
Aguijan must be classified as a Simple
Carbonate Island because no proof has been
found to confirm that the island has a core of
non-carbonate rocks that extend above sea
42
Figure 18: Map of Unai Dangkolo region showing the close proximity of lank margin cave
and fissure cave development.
level and partition the fresh-water lens.
However, based on Aguijan's similarity with
the Southeastern Ridge of Tinian it probably
best fits the Carbonate-Cover Island Karst
Model.
Geologic investigations on the islands of
Tinian and Aguijan have been limited in the
past and much remains to be done in future
research, which is beyond the scope of this
study. The cave and karst inventory
surveyed the majority of known cave and
karst features on Tinian and Aguijan;
however, more features exist which need to
be identified and surveyed. Basic geologic
mapping is still needed for the island of
Aguijan. During this study, the author and
other investigators sampled speleothems,
paleosols, and cave sediments, which may
provide greater insight into the geology and
hydrogeology of the Mariana Islands.
Future isotope analysis of speleothems and
petrographic analysis of paleosols may
provide important information about the
paleoclimatic history of the region, while
future analysis of cave sediments may
elucidate the origin of Liyang Atkiya.
43
REFERENCES CITED
Aby, S.B., 1994. Relation of bank-margin
fractures to sea level change, Exuma
Islands, Bahamas. Geology, v. 22, pp.
1063-1066
Back, W., Hanshaw, B.B., and Van Driel,
J.N., 1984. Chapter 12, Role of
groundwater in shaping the Eastern
Coastline of the Yucatan Peninsula,
Mexico. in LaFleur, R.G. (Editor),
Groundwater as a Geomorphic Agent,
Allen & Unwin, Inc., Boston, pp. 281293.
Barlow, C.A., and Ogden, A.E., 1992. A
statistical comparison of joint, straight
cave segment, and photo-lineament
orientations. NSS Bulletin, v. 44, pp.
107-110.
Bormann, N.E., 1992. Ground-water
resources management on Tinian,
CNMI. American Water Resources
Association technical publication series,
v. 92-1, pp. 35-44.
Burke, H.W., 1953. The Petrography of the
Mariana Limestone, Tinian, Mariana
Islands. Graduate dissertation, Stanford
University, 111 p.
Burt, J.E., and Barber, G.M., 1996.
Elementary Statistics for Geographers:
Second Edition. The Gulford Press,
New York, NY, 640 p.
Butler, B.M, 1992. An Archaeological
Survey of Aguiguan (Aguijan), Northern
Mariana Islands: Micronesian
Archaeological Survey Report Number
29. The Micronesian Archaeological
Survey. Division of Historic
Preservation, Dept. of Community and
Cultural Affairs, Saipan, MP, 260 p.
Carew, J.L., and Mylroie, J.E., 1995.
Quaternary tectonic stability of the
Bahamian Archipelago: evidence from
fossil coral reefs and flank margin
caves. Quaternary Science Reviews, v.
14, pp. 145-153.
Cloud, P.E., Jr., Schmidt, R.G., and Burke,
H.W., 1956. Geology of Saipan,
Mariana Islands, Part 1. General
Geology. 280-A, U.S. Geological
Survey Professional Paper, U.S.
Government Printing Office,
Washington, D.C., pp. 126.
Dasher, G.R., 1994. On Station. National
Speleological Society, Inc., Huntsville,
Alabama, 242 p.
Dickinson, W.R., 1999. Hydro-isostatic and
tectonic influences on emergent
Holocene paleoshorelines in the
Mariana Islands, Western Pacific Ocean.
Journal of Coastal Research, v. 16, no.
3, pp. 735-746.
Doan, D.B., Burke, H.W., May, H.G.,
Stensland, C.H., and Blumenstock, D.I.,
1960. Military Geology of Tinian,
Mariana Islands. Chief of Engineers,
U.S. Army, 149 p.
ESRI, 2000. ArcView GIS 3.2.
Environmental Systems Research
Institute, Inc., software.
Folk, R.L., Roberts, H.H., and Moore, C.C.,
1973. Black phytokarst from Hell,
Cayman Islands, British West Indies.
Geological Society of America Bulletin,
v. 84, pp. 2351-2360.
Frank, E.F., Mylroie, J.E., Troester, J.,
Alexander, E.C., and Carew, J.L., 1998.
Karst development and speleologensis,
Isla de Mona, Puerto Rico. Journal of
Cave and Karst Studies, v. 60, no. 2, pp.
73-83.
Gingerich, S.B., and Yeatts, D.S., 2000.
Ground-Water Resources of Tinian,
Commonwealth of the Northern
Mariana Islands, Water-resources
Investigations Report 00-4068. U.S.
Department of the Interior, 2 sheets.
Gross, M.G., 1982. Oceanography: A View
of the Earth. Prentice-Hall, Inc.,
Eaglewood Cliffs, NJ, 498 p.
Harris, J.F., Mylroie, J.E. and Carew, J.L.,
1995. Banana holes: Unique karst
features of the Bahamas. Carbonates and
Evaporites, v. 10, no. 2, pp. 215-224.
Holcombe, R., 2003. GEOrient 9.2.
Department of Earth Sciences,
University of Queensland, Australia.
44
http://www.earth.uq.edu.au/~rodh/softw
are/
Hunt, E., and Wheeler, T., 2000. South
Pacific. Lonely Planet Publications,
Oakland, CA, 928 p.
Jenson, J.W., Jocson, J.M.U., and Siegrist,
H.G., 1997. Groundwater discharge
styles from an uplifted Pleistocene
island karst aquifer, Guam, Mariana
Islands. in Beck, B.F. and Stephenson,
J.B. (Editors), The Engineering Geology
of Karst Terraines, Balkema,
Springfield, Missouri. pp. 27-32.
Jenson, J.W., Mylroie, J.E., Mylroie, J.R.,
and Wexel, C., 2002. Revisiting the
carbonate island karst model. Geological
Society of America Abstracts with
Program, v. 34, no. 6, p. 226.
Jocson, J.M.U., Jenson, J.W., and
Contractor, D.N., 2002. Recharge and
aquifer responses: Northern Guam Lens
Aquifer, Guam, Mariana Islands.
Journal of Hydrology, v. 260, pp 231254.
Jocson, J.M.U., Jenson, J.W., and
Contractor, D.N., 1999. An Integrated
Numerical Modeling Study and Field
Investigation of the Northern Guam
Lens Aquifer. Technical Report #88,
Water and Environmental Research
Institute of the Western Pacific,
University of Guam, Mangilao, 34 p.
Statistic. in. Miller, R.L., and Kahn,
J.S., Statistical Analysis in the
Geological Sciences. John Wiley and
Sons, Inc., New York, NY, pp. 464-468.
McClure, G.E., 1977. Tinian, Then and
now..: 5th edition. Glenn E. McClure,
Universal City, TX, 28 p.
McKenzie, D., 2002. WALLS Project
Editor – Beta Version 2.0 B6.
http://davidmck.home.texas.net/walls/,
shareware software.
Mylroie, J.E., and Carew, J.L., 1997. Land
use and carbonate island karst. in Beck
and Stephenson (Editors), The
Engineering and Hydrology of Karst
Terranes. Balkema, Rotterdam, pp. 3-12.
Mylroie, J.E., and Carew, J.L., 1995a. Karst
development on carbonate islands. in
Budd, D.A., Harris, P.M., and Staller, A.
(Editors), Unconformities and Porosity
in Carbonate Strata. American
Association of Petroleum Geologists,
pp. 55-76.
Mylroie, J.E., and Carew, J.L., 1995b.
Geology and karst geomorphology on
San Salvador Island, Bahamas.
Carbonates and Evaporites, v. 10, no. 2,
pp. 193-206.
Mylroie, J.E., and Carew, J.L., 1991.
Erosional notches in Bahamian
carbonates: bioerosion or groundwater
dissolution?. in Bain, R.J. (Editor),
Proceedings of the Fifth Symposium on
the Geology of the Bahamas. Bahamian
Field Station, San Salvador, Bahamas,
pp. 185-191.
Mylroie, J.E., Carew, J.L., and Vacher, H.L.,
1995a. Karst development in the
Bahamas and Bermuda. In: Curran,
H.A., and White, B. (Editors),
Terrestrial and Shallow Marine Geology
of the Bahamas and Bermuda, pp. 251267.
Mylroie, J.E., Carew, J.L., and Moore, A.I.,
1995c. Blue hole: definition and
genesis. Carbonates and Evaporites, v.
10, no. 2, pp. 225-233.
Mylroie, J.E., and Jenson, J.W., 2002. Karst
flow systems in young carbonate
islands. in Martin, J.B., Wicks, C.M.,
Klimchouk, A., and Ford, D., 2000. Chapter
3.2, Lithologic and structural controls of
dissolutional cave development. in
Klimchouk, A., Ford, D., Palmer, A.,
and Dreybrodt, W. (Editors),
Speleogenesis: Evolution of Karst
Aquifers. National Speleological
Society, Huntsville, AL, pp. 54-64.
Mink, J.F., and Vacher, H.L., 1997.
Hydrogeology of northern Guam. in
Vacher, H.L., and Quinn, T. (Editors),
Geology and Hydrogeology of
Carbonate Islands. Developments in
Sedimentology 54, Elsevier Science, pp.
743-761.
Miller, R.L., and Kahn, J.S., 1962.
Appendix G: The Kolmogorov-Smirnov
45
and Sasowsky, I.D. (editors),
Hydrogeology and biology of postPaleozoic carbonate aquifers. Karst
Waters Institute Special Publication
Number 7, Karst Waters Institute, Inc.,
Charles Town, WV, pp. 107-110.
Mylroie, J.E., and Jenson, J.W., 2001. The
carbonate island karst model applied to
Guam. Theoretical and Applied
Karstology, v. 13-14, pp. 51-56.
Mylroie, J.E., Jenson, J.W., Jocson, J.M.U.,
and Lander, M., 1999. Karst Geology
and Hydrology of Guam: A Preliminary
Report, Technical Report #89. Water
and Environmental Institute of the
Western Pacific, University of Guam,
Mangilao, 32 p.
Mylroie, J.E., Jenson, J.W., Taborosi, D.,
Jocson, J.M.U., Vann, D., and Wexel,
C., 2001. Karst features of Guam in
terms of a general model of carbonate
island karst. Journal of Cave and Karst
Studies, v. 63, no. 1, pp. 9-22.
Mylroie, J.E., Panuska, B.C., Carew, J.L.,
Frank, E., Taggart, B.E., Troester, J.W.,
and Carrasquillo, R., 1995b.
Development of flank margin caves on
San Salvador Island, Bahamas and Isla
de Mona, Puerto Rico. in Boardman, M.
(Editor), Proceedings of the Seventh
Symposium on the Geology of the
Bahamas. Bahamian Field Station, San
Salvador, Bahamas, pp. 49-81.
Mylroie, J. E., and Vacher, H. L., 1999. A
conceptual view of carbonate island
karst. in Palmer, A. N., Palmer, M. V.,
and Sasowsky, I. D. (Editors), Karst
Modeling, Karst Waters Institute Special
Publication 5, Karst Waters Institute,
Inc., Charles Town, WV, pp. 48-57.
Nelson, J.W., 1988. Structural and
Geomorphic Controls of the Karst
Hydrogeology of Franklin County,
Alabama [MS Thesis]. Mississippi
State University, 165 p.
Palmer, A.N., 1991. Origin and morphology
of limestone caves. Geological Society
of America Bulletin, v. 103, pp. 1-21.
Raeisi, E., and Mylroie, J.E., 1995.
Hydrodynamic behavior of caves
formed in the fresh-water lens of
carbonate islands. Carbonates and
Evaporites, v. 10, no. 2, pp. 207-214
Reagan, M.K., and Meijer, A., 1984.
Geology and geochemistry of early arcvolcanic rocks from Guam. Geological
Society of America Bulletin, v. 95, pp.
701-713.
Sasowsky, I.D., and White, W.B., 1994.
The role of stress release fracturing in
the development of cavernous porosity
in carbonate aquifers. Water Resources
Research, v. 30, no. 12, pp. 3523-3530.
Siegrest, H.G., 1988. Miocene reef
carbonates of the Mariana Islands.
American Association of Petroleum
Geologists Bulletin, v. 72, no. 2, p. 248.
Sprouse, P., 1991. Proyecto Espeleologico
Purificacion: Standard cave map
symbols. The Death Coral Caver, no. 2,
p. 37.
Sprouse, P., and Russell, W., 1980. AMCS
cave map symbols: AMCS Activities
Newsletter, no. 11, pp. 64-67.
Stafford, K.W., Mylroie, J.E., Mylroie, J.R.,
and Jenson, J.W., 2003. Tinian, CNMI:
A carbonate island karst model
evaluation. Geological Society of
America abstracts with programs, v. 35,
no. 1, p. 54.
Stafford, K.W., Mylroie, J.E., and Jenson,
J.W., 2002. Karst Geology and
Hydrology of Tinian and Rota (Luta),
CNMI: Technical Report No. 96. Water
and Environmental Institute of the
Western Pacific, University of Guam,
Mangilao, 31 p.
Stafford, K.W., Mylroie, J.E., Mylroie, J.R.,
and Jenson, J.W., 2003. Tinian, CNMI:
A carbonate island karst model
evaluation. Geological Society of
America Abstracts and Programs, v. 35,
no. 1, p. 3.
Sugiwara, S., 1934. Topography, Geology,
and Coral Reefs of Rota Island. Pacific
Geological Surveys, Military Branch,
United States Geological Survey, 59 p.
Taborosi, D.S., 2000. Karst Features of
Guam [MS thesis]. University of Guam,
Mangilao, 19 p.
46
Taborosi, D.S., and Jenson, J.W., 2002.
World War II artefacts and wartime use
of caves in Guam, Marian Islands.
Capra 4,
http://www.shef.ac.uk/~capra/4/.
Taborosi, D., Jenson, J.W., and Mylroie,
J.E., in press. Karren features in island
karst: Guam, Mariana Islands.
Zeitschrift fur Geomorphologie.Tayama,
R., 1936. Geomorphology, Geology,
and Coral Reefs of Tinian Island
Together with Aguijan and Naftan
Islands. Institute of Geology and
Paleontology, Tohoku University,
Japan. 72 p.
Till, R., 1974. Chapter 7: Non-Parametric
Statistics. in Till, R., Statistical
Methods for the Earth Scientist. John
Wiley and Sons, Inc., New York, NY,
pp. 117-138.
Twiss, R.J., and Moores, E.M., 1992.
Structural Geology. W.H. Freeman and
Company, U.S., 532 p.
Tracey, J.I., Jr., Schlanger, S.O., Stark, J.T.,
Doan, D.B., and May, H.G., 1964.
General Geology of Guam. 403-A, U.S.
Geological Survey Professional Paper,
U.S. Government Printing Office,
Washington, D.C., 104 p.
United States Department of the Interior
Geological Survey, 2001a.
1697184.DEM.SDTS.TAR.GZ.
National Mapping program of the U.S.
Geological Survey,
ftp://sdts.er.usgs.gov/pub/sdts/datasets/r
aster/dem/
United States Department of the Interior
Geological Survey, 2001b.
1697335.DEM.SDTS.TAR.GZ.
National Mapping program of the U.S.
Geological Survey,
ftp://sdts.er.usgs.gov/pub/sdts/datasets/r
aster/dem/
United States Department of the Interior
Geological Survey, 2001c.
1697336.DEM.SDTS.TAR.GZ.
National Mapping program of the U.S.
Geological Survey,
ftp://sdts.er.usgs.gov/pub/sdts/datasets/r
aster/dem/
United States Department of the Interior
Geological Survey, 2001d.
1697337.DEM.SDTS.TAR.GZ.
National Mapping program of the U.S.
Geological Survey,
ftp://sdts.er.usgs.gov/pub/sdts/datasets/r
aster/dem/
United States Department of the Interior
Geological Survey, 2001e.
1697338.DEM.SDTS.TAR.GZ.
National Mapping program of the U.S.
Geological Survey,
ftp://sdts.er.usgs.gov/pub/sdts/datasets/r
aster/dem/
United States Department of the Interior
Geological Survey, 1983. Topographic
Map of the Island of Tinian,
Commonwealth of the Northern
Mariana Islands: U.S. Geological
Survey, scale 1:25,000.
Vacher, H.L., and Mylroie, J.E., 2002.
Eogenetic karst from the perspective of
an equivalent porous medium.
Carbonates and Evaporites, v. 17, no. 2,
pp. 182-196.
Viles, H. A., 1988. Biogeomorphology.
Basil Blackwell, Oxford, United
Kingdom, 325 p.
White, W.B., 1988. Geomorphology and
Hydrology of Karst Terrains. Oxford
University Press, New York, 464 p.
Xara, 1997. CorelXara 2.0. Xara Ltd.,
Hertfordshire, United Kingdom,
http://www.xara.com, software.
47