The Kuka'iau Cave, Mauna Kea, Hawaii, created by water erosion: A new Hawaiian cave type

Abstract

In 2000 and 2001, two large (each ca. 1000 m long) cave systems have been surveyed on the eastern, heavily eroded, flank of Mauna Kea: The Pa‘auhau Civil Defense Cave and the Kuka‘iau Cave (at first called ThatCave/ThisCave System). Both caves occur in the Hamakua Volcanics, 200-250 to 65-70 ka old. They are the first substantial caves documented for lavas of Mauna Kea and the first caves on Hawai‘i showing extensive morphological signs of water erosion.
The Pa‘auhau Civil Defense Cave is a lava tube, as attested by the presence of the typical morphological elements of lava tubes, including secondary ceilings, linings, base sheets, stalactites and lava falls. Subsequently, the cave was modified erosionally by a stream which entered upslope and traversed much, but not all, of the cave, leaving water-falls, water-fall ponds, scallops, gravel, rounded blocks and mud (see paper by Kempe et al., this volume). In contrast the Kuka‘iau Cave – a still active stream cave with a vadose and phreatic section - is essentially erosional in origin. This is concluded from the geology of the strata exposed in the cave and from its morphology: At the upper entrance the cave is situated in a thick series of ‘a‘ā and the lower section was created by removing ‘a‘ā and diamict layers, therefore excluding the possibility that the cave developed from a precursor lava tube. Also, in its phreatic section, the cave makes several right angle turns and moves upward through a series of pāhoehoe sheets, unlike any lava tube. Furthermore, a base layer can be followed along which the major section of the upper cave has developed. Allophane and halloysite – minerals produced by weathering - helped in sealing the primary porosity of this base layer causing a locally perched water table. Water moving along this base layer on a steep hydraulic gradient through the interstices of ‘a‘ā and through small pāhoehoe tubes exerted a high pressure on the porous diamict of the lower cave, causing its erosional removal. Our observations of water erosional caves in lavas of Hawai‘i offer a new perspective on deep-seated water courses in volcanic edifices.

Full text

Copyright © 2003 by The National Speleological Society Journal of Cave and Karst Studies, April 2003 • 53
Stephan Kempe and Marlin Spike Werner - The Kuka‘iau Cave, Mauna Kea, Hawaii, Created by water erosion, A new Hawaiian cave type. Journal of Cave and Karst Studies, v. 65, n.
1, p. 53-67.
On Hawaii, hundreds of small and large lava tubes have
been explored in the last 20 years, largely by members of the
Hawai‘i Speleological Survey and the Hawaiian NSS Grotto.
These tube systems are located on Kilauea, Mauna Loa and
Hualalai. Mauna Kea, however, remained somewhat of a blank
spot speleologically. Old reports of caves being intercepted by
water wells have not been documented sufficiently so far.
Along the old Mamalahoa Highway and along the road to the
Waipio Valley Outlook west of Honoka‘a there are a few
artificial caves, but they are former quarries where road
building material such as cinders and aa was mined.
Hawaiian microclimates vary widely from extremely
humid to extremely dry conditions on the same island. On the
windward side of the Big Island of Hawai‘i Mauna Kea’s NE
flank receives up to 90 inches of rain per annum (1961-1990
average annual precipitation Data, U.S. National Oceanic and
Atmospheric Administration). This has led, despite the highly
porous nature of volcanic rocks, to the formation of perennial
and episodic streams, which in turn have significantly
dissected the eastern flank of Mauna Kea. Its morphology is
characterized by several deep gulches and countless V-shaped
gullies and streamways, which funnel the water straight to the
ocean without forming large tributary river systems. The last
eruption on Mauna Kea occurred at the summit under the ice,
i.e., during the Last Glacial. Any lava tubes that have formed
on the lower flanks are therefore older and may have formed
several tens of thousands of years ago. Thus, such tubes, if any,
should be either invaded and plugged by younger lavas, filled
by ashes, or should have collapsed under the weight of
overlying strata. Therefore, no systematic search was made for
caves on Mauna Kea. When the first caves on Mauna Kea
came to the notice of the Hawaiian Speleological Survey in
1995, this came as quite a surprise. These were two caves in
Honoka‘a and the Civil Defense Cave at Pa‘auhau. First rough
maps were published by Werner, 1997. Further findings, three
small caves in Kalopa State Park and a larger system in the
Hamakua Forest Reserve were reported by Halliday (2000 a,
b). Pa‘auhau Civil Defense Cave was studied in detail in 2001
(Kempe et al. 2003) and turned out to be a lava tube heavily
impacted by water erosion.
In March 2000 Robert van Ells told M.S. Werner about
another cave on Mauna Kea (Werner et al., 2000; Kempe et al.,
2001). He reported having entered through a pit and then
having exited makai
1
from a resurgence by swimming through
a small pond. We relocated these two entrances on March 15th
2002. The cave was initially called ThisCave because we
suspected that the cave already had a local name. Because it
serves as the source of a stream at flood, we concluded that
THE KUKA‘IAU CAVE, MAUNA KEA, HAWAII, CREATED
BY WATER EROSION, A NEW HAWAIIAN CAVE TYPE
STEPHAN KEMPE
Institute for Applied Geosciences, Univ. of Techn. Darmstadt, Schnittspahnstr. 9, D-64287 Darmstadt, G
ERMANY; kempe@geo.tu-darmstadt.de
MARLIN
SPIKE WERNER
P.O. Box 11509 Hilo, Hawaii 96721-6509, USA; email mspike@ilhawaii.net
In 2000 and 2001, two large (each ca. 1000 m long) cave systems have been surveyed on the eastern,
heavily eroded, flank of Mauna Kea: The Pa’auhau Civil Defense Cave and the Kuka‘iau Cave (at first
called ThatCave/ThisCave System). Both caves occur in the Hamakua Volcanics, 200-250 to 65-70 ka
old. They are the first substantial caves documented for lavas of Mauna Kea and the first caves on
Hawai’i showing extensive morphological signs of water erosion.
The Pa’auhau Civil Defense Cave is a lava tube, as attested by the presence of the typical morphological
elements of lava tubes, including secondary ceilings, linings, base sheets, stalactites and lava falls.
Subsequently, the cave was modified erosionally by a stream which entered upslope and traversed much,
but not all, of the cave, leaving waterfalls, waterfall ponds, scallops, gravel, rounded blocks and mud
(Kempe et al. 2003). In contrast the Kuka‘iau Cave – a still active stream cave with a vadose and phreatic
section - is essentially erosional in origin. This is concluded from the geology of the strata exposed in the
cave and from its morphology: At the upper entrance the cave is situated in a thick series of aa and the
lower section was created by removing aa and diamict layers, therefore excluding the possibility that the
cave developed from a precursor lava tube. Also, in its phreatic section, the cave makes several right
angle turns and moves upward through a series of pahoehoe sheets, unlike any lava tube. Furthermore,
a base layer can be followed along which the major section of the upper cave has developed. Allophane
and halloysite – minerals produced by weathering - helped in sealing the primary porosity of this base
layer causing a locally perched water table. Water moving along this base layer on a steep hydraulic
gradient through the interstices of aa and through small pahoehoe tubes exerted a high pressure on the
porous diamict of the lower cave, causing its erosional removal. Our observations of water erosional
caves in lavas of Hawaii offer a new perspective on deep-seated water courses in volcanic edifices.
1
makai = Polynesian for “towards the sea” i.e., downslope

54 • Journal of Cave and Karst Studies, April 2003
KUKA‘IAU CAV E , MAUNA KEA, HAWAI'I, CREATED BY WATER EROSION
there must be a stream sink mauka
2
as well. M.S. Werner then
scouted the area and found the sink in a creek bed leading to a
large vadose underground stream passage. It was provisionally
called ThatCave. None of the ranchers interviewed knew about
this entrance even though it is quite obvious and fences run
nearby. In 2000 ThisCave and ThatCave were explored and
surveyed by teams led by S. Kempe and M.S. Werner. In
addition geological investigations were carried out in
ThatCave by S. Kempe, I. Bauer and H.-V. Henschel in 2001
and 2002. In 2002, the sump, so far separating the two caves,
was explored physically by H.-V. Henschel, S. Kempe, and
M.S. Werner, linking the survey of the two caves. The story of
the discovery of these caves is related in Werner et al. (2002).
Initial reports were given in Kempe, 2000; and Kempe et al.,
2001. In June 2002 we also learned that Dr. Fred Stone of the
University of Hawai‘i, Hilo, had investigated ThatCave
biologically in 1992 (Stone 1992). He had named the cave
Kuka‘iau Piping Cave. Since older names take precedence, we
changed the name of the cave to Kuka‘iau Cave, dropping
“piping”, because the cave name should not give an
interpretation of the mechanism of its formation. We also
learned that Stearns and Macdonald (1946) mention the cave
(p. 244): “A large lava tube 2.3 miles S. 10°W (longitude
obviously in error, note of S.K.) of Kuka‘iau Post Office
constitutes an unusual source of water. Half a mile south of this
point a stream disappears underground, apparently flowing
into the tube. During heavy rains a large stream of water
emerges from the lower end of the tube, but it does not flow in
dry weather. A large body of water remains in the tube,
however, and is used to supply cattle tanks. About 50,000
gallons is siphoned from the tube every two months, but the
tube never has been emptied.” This citation explains the
presence of bent and rusted pipes in ThisCave and of a well
head above the sump in ThisCave.
L
OCATION AND
G
EOLOGICAL SETTING
The Kuka’iau Cave is located on the north-eastern flank of
Mauna Kea (Hamakua Coast), south of Kuka‘iau (Fig. 1). We
obtained valid GPS locations for the three entrances (Silva
Multinavigator) (Table 1).
The location indicates (pers. com. Dr. Frank Truesdale,
Hawaiian Volcano Observatory, USGS) that the cave is
situated in the Hamakua Volcanics. These are the oldest
volcanic rocks exposed on the Mauna Kea (Wolfe et al. 1997).
They are dated to between 200-250 to 65-70 ka BP and consist
mostly of alkalic and transitional basalts. In the upper part of
the volcano glacial deposits are found as well.
The cave offers an excellent opportunity to study the
structure of the upper Hamakua Series. It contains aa rubble
and aa core layers as well as thick series of tubiferous (tube-
bearing) pahoehoe, surface pahoehoe (with ropy surface
structures), diamict layers (a layer of mixed rock sizes in a
finer matrix of unclear genesis) and soil layers. The rocks are
generally very weathered and have lost much of there original
brittleness. All in all one can differentiate five units exposed in
the cave. These are (from mauka to makai):
1. The Entrance Pit Series (pahoehoe, soils, aa rubble and
core layer);
2. the Entrance Hall Series (a thick series of aa ruble and
core layers);
3. the Vadose Stream Series (several aa ruble and core
layers, paleosoil / red Marker Layer, tubiferous
pahoehoe and surface pahoehoe);
4. the Sump Series (pahoehoe sealed by halloysite); and
5. the ThisCave Series (overlaying the Sump Series,
consisting of a soil, an aa, and a diamict layer, and a
thick aa core, which serves as the roof of ThisCave).
Fig. 1: Location of Kuka‘iau Cave and Pa‘auhau Civil
Defense Cave and geological map of the Northern section
of Mauna Kea. The dark gray area represents the outcrops
of the Hamakua Volcanics and the white area represents
the distribution of the overlying Laupahoehoe Volcanics
(redrawn according to Wolfe et al. 1997).
Table 1. GPS locations of Kuka‘iau Cave entrances (map
datum: Hawaiian):
Date: 04.03.01 North West Altitude a.s.l.
ThisCave Entrance 20°00,033’N 155°21,316’W 740 m
ThisCave Pit 19°59,972’N 155°21,317’W 746 m
ThatCave St. 1 (post) 19°59,617’N 155°21,387’W 869 m
ThatCave St. 4 19°59,637’N 155°21,398’W 843 m
ThisCave to Pit Distance: 111 m Bearing: 180°N Vertical: 6 m
ThatCave 1 to ThisCave E. Distance: 780 m Bearing: 9°N Vertical: 103 m
ThatCave 4 to ThisCave E. Distance: 748 m Bearing: 10°N Vertical: 97 m
2
mauka = Polynesian for “towards the mountain” i.e., upslope

Journal of Cave and Karst Studies, April 2003 • 55
KEMPE & WERNER
MORPHOLOGY OF KUKA‘IAU CAV E
G
ENERAL DESCRIPTION AND SURVEY DATA
Kuka‘iau Cave consists in essence of one continuous
underground stream passage (Fig. 2), directed mostly S-N. It
has an upper entrance (a series of waterfalls in the gully of an
intermittent stream) and a lower exit (a portal at the head of
another gully, pirating the water from its neighbor). Only one
other entrance is present, a 5 m deep pit at the upper end of
ThisCave, possibly dug by the local ranchers to gain access to
the water supply of the head sump for cattle (Stearns &
Macdonald 1946).
This sump separates the two cave sections. We had to wait
for two years before the water fell low enough to wade through
the sump section and to tie the two surveys together
(07.06.02). Compared to the GPS locations the survey was in
error for 3.8 m in the N-S direction (mostly the tape and
inclination error) and for 15.5 m in the E-W direction (mostly
the compass error). Total main passage length of ThatCave is
719 m and of ThisCave is 187 m. The total cave length adds up
to 1073 m (inclusive of some unmapped side passages see
Table 1, Lines 3 and 4). Not many side passages exist (14 m in
ThisCave and 153 m in ThatCave). The ratio between the main
passage and the side passages is 5.2. The total vertical extent,
according to our survey, is 105 m (for details see Table 2). For
comparison, the vertical GPS difference between St. 4 (in the
creek of the upper entrance) and the exit of ThisCave is 97 m
(see Table 1), to which one has to add the drop from the roof
of the cave (where the GPS measurement was obtained) to the
floor of the stream bed which is ca. 6 m, yielding a very good
correspondence between the GPS altitude difference and our
underground survey. The slope of ThatCave is 6.9°. It is,
however, steeper in the upper part of the cave. The sinuosity of
the cave is 1.17.
G
ENERAL MORPHOLOGY
The importance of the cave system rests on the fact that the
morphology of the cave is almost entirely determined by
erosional features. The cave has morphologically four sections
(compare Fig. 2):
- the unroofed entrances waterfall series, comprised of
three up to 6 m tall water falls, water fall pools, steep and
partly overhanging walls and a prominent meander;
- the steep, vadose underground stream course, mostly of a
rectangular cross-section, dominated by water falls, water fall
ponds, polished and scalloped rock bed chutes, rounded
boulders and gravel chutes, undercut water fall lips, undercut
walls and polished and scalloped walls;
- the level and even upward sloping phreatic sump section,
mostly of an oval cross-section, showing several right angle
bends and dominated by polished and scalloped floors, walls
and ceilings, gravel banks and steep upward gravel and rock
chutes;
- the gently inclined vadose underground stream passage of
ThisCave, of a wide rectangular cross-section, with only a few
cascades, polished and scalloped rock chutes and a bed of
large, rounded boulders and gravel.
Regarding side passages, two types have to be discerned: a
very tight distributary side passage in ThisCave, which has
been carved out at the interface between the underlying
pahoehoe and the diamict, and those being most probably
small lava tubes in origin (all the side passages in ThatCave).
These latter side passages partly form oxbows and cut-
arounds, feeding back into the main cave and either serving as
high water bypasses or forming blind appendices blocked by
lava fills or collapses either mauka or makai. In these side
passages, some of the features of small lava tubes are
preserved, such as accretionary linings, glazing, and rarely
stumps of lava stalactites and stalagmites. For example, the
cave crosses a filled lava tube with recognizable linings
between St. 51 and 52. At the beginning of the high water
bypass at St. 32 there is glazing preserved at both sides of the
entrance. In the cupola at St. 28 there may by an open tube
crossing high up in the wall, out of which water seeps into the
Table 2. Survey statistics for the Kuka‘iau Cave, separate
for the ThisCave and ThatCave Sections, as of June 2002
(St. = survey station, see Fig. 2).
Measurement ThisCave ThatCave
1 Total added survey lines 272.1 m 1064.7 m
2 Total cave length (survey lines) 198.6 m 942.5 m
3 Total main passage survey 186.9 m 719.0 m
(St. 12-23A*)
4 Total side passages (see below) 14 m 153.0 m
5 Total cave length (3+4) 200.9 872.0m
6 Total main passage length horizontal 184.5 m 704.7 m
7 Total main passage length horizontal 160 m 679 m
(measured along center of passage)
(dripline to St. 23A*)
8 End-to-end distance 132.5 m 572 m
(dripline to St. 23A*)
8a End-to-end distance total 716 m
(not the sum of line 8) (dripline to dripline)
9 End-to-end direction (magnetic) 354°N 2.7° N
(St. 1-23A*)
9a Total 1.4° N
(dripline to dripline)
10 Sinuosity of total cave (horizontal) - 1.17
((160+679)/716)
11 Total altitude difference in caves +1.3 m -87.3m
(St.6-23) (St.12.-75)
11a Total altitude difference entrance to exit -85.4 m
12 Total altitude difference -104.8 m
(St. 5-75)
13 Entrance canyon vertical difference -17.42 m
(St. 5-12)
14 Maximal pressure above sump 65 m = 6.5 bar
15 Average slope of main passage (St.12-60) 9.4°
(tan
-1
(77/464))
16 Average slope of main passage (St.12-75) 6.9°
(tan
-1
(87/719)
*23A = Station 23A in ThisCave
The total of all surveyed and unsurveyed but explored side passages in ThatCave is:
A) At Station 23, Sand Tube: 31.8 m
B) At Station 24: 5 m
C) Stations 31-35: (- 4 m St. 31-32) 42 m
D) At Station 41: 5 m
E) Station 47 to 47C 26.8 m
F) Side passage between St. 53 and 56 40 m (not surveyed, horiz. estimate)
G) Station 77-78 (*0.5) 2.40 m
Total side passages 153.0 m (line 4 in Table)

56 • Journal of Cave and Karst Studies, April 2003
KUKA‘IAU CAV E , MAUNA KEA, HAWAI'I, CREATED BY WATER EROSION
Figure 2. Map of ThatCave/ThisCave System. Note that the longitudinal Section has a slightly smaller scale than map and profiles. D

Journal of Cave and Karst Studies, April 2003 • 57
KEMPE & WERNER
Drawn by S. Kempe.

58 • Journal of Cave and Karst Studies, April 2003
KUKA‘IAU CAV E , MAUNA KEA, HAWAI'I, CREATED BY WATER EROSION
cave. Also Dr. W.R. Halliday noted during a visit in February
2002 a piece of upward facing glazing on the eastern wall near
St. 22. He collected a sample which he kindly forwarded to the
authors together with on-site photographs. The sample plus the
pictures and a site inspection in June 2002 shows that a filled
lava tube has been eroded into sideward. The piece of glazing
is part of a former lava tube representing the upper surface of
a bench or ledge, from the left-hand side of a tube (left-hand
looking makai). This is concluded from the fact that the
specimen submitted has a wedge-like rim, which could be a
flow line. This wedge-like rim is directed towards the present
day wall, showing - as explained above - that the present cave
has cut with its right-hand (eastern) wall through the left-hand
(western) wall of a former, completely filled lava tube. The
former floor of this ca. 1 m high lava tube is directed into the
main cave and appears to form a prominent ledge. Where this
small tube leaves the main cave could not be verified. Above
this site, another small lava tube, albeit obliterated by
breakdown, causes a widening of the main cave. This tube
(Sandy Tube) departs from the same wall a few meters makai
and was followed for ca. 30 m before becoming too tight. All
these examples show that part of the vadose section of the cave
was eroded into a stack of tubiferous pahoehoe, featuring
several smaller tubes on top of each other.
Other features, which could be mistaken for lava tube
formations such as structures similar to lava falls or flow lines
are present throughout the cave. Upon closer inspection, none
Figure 3.
Geological
situation at the
lip of the first
water fall at
the mauka
entrance to
Kuka‘iau cave.
The
underlying
pahoehoe is
superceded by
the quartz-
bearing soil
Layer 1 and
covered by an
aa flow. Its
rubble is
highly
weathered and
the core forms
the present
creek bed
(lettering
refers to XRD
analyses given
in Table 3).
Figure 4. Stratigraphic section exposed in Kuka‘iau Cave
at the plunge-pool series near Station 27 (sample numbers
refer to XRD analyses given in Table 3).

Journal of Cave and Karst Studies, April 2003 • 59
KEMPE & WERNER
of these features are, however, lava tube-related, rather they
represent exposed aa blocks or differentially eroded sheeted
lava flows.
The typical morphological elements of larger lava-tube
systems (reviewed, for example, in Kempe 2002; compare also
Allred 1997; Greeley et al. 1997; Kempe 1997) are however
missing. These include large tributary branches, primary tubes
near the ceiling of the tube, wide halls being eroded
underneath deep lava falls, secondary ceilings which separate
deeply eroded canyons into two or more passages, remains of
welded breakdown and others. Furthermore, the large change
in the gradient of the cave is atypical for lava tubes, and
upward gradients have never been documented.
On the mesoscale the absence of accretionary linings, lava
falls (which would have been eroded anyway) and lava-tube
floor formations such as tube-in-tube structures and others
should be noted. On a smaller scale, flow features along floor,
ceiling and walls (e.g., levees, accretionary ledges, lava balls,
glazed surfaces, stalactites and stalagmites, runners and drip
piles) or any other features caused by molten lava and
associated with lava tube formation are not seen.
Figure 5. The connection sump: (A) view makai at the
Rock Dam holding back Echo Lake. Note the notch cut by
water overflowing the Rock Dam back into the deeper part
of the sump (where person is standing). (B) View from the
crest of the Rock Dam, across Echo Lake into the lower
part of ThisCave (person at St. 23A). Even when the sump
is dry, Echo Lake is kept flooded because of the halloysite
filling all the interstices of the rock in this section of the
cave. (C) As the water rises out of Echo Lake it has to move
up this gravel chute into the Wellhead Room. Note size of
gravel (up to 10 cm in diameter) being moved uphill makai
by water when the cave floods. (D) View mauka, down the
last gravel chute. Person stands at the makai end of the
Wellhead Room. Here the water rises from the sump to
overflow into the vadose stream passage in ThisCave. Note
the size of the gravel and stones being moved up this steep
gravel chute. Pictures by S. Kempe.

60 • Journal of Cave and Karst Studies, April 2003
KUKA‘IAU CAV E , MAUNA KEA, HAWAI'I, CREATED BY WATER EROSION
The morphological elements of an erosional origin are, on
the other hand, ubiquitous. The cave floor is either composed
of solid rock sculptured by stream pots, chutes or scallops, or
it consists of beds of rounded boulders, cobbles, or gravel.
Garbage, plant residue, and parts of trees are found throughout
the cave, testifying to its function as a recent water course as
well. Farther down, where the passage narrows, one finds
Styrofoam debris wedged into cracks in the ceiling by high
water. The solid-rock sections include waterfalls with water
fall lips, undercut waterfall walls and large plunge pools. The
gravel sections include gravel chutes, leading upward and out
of plunge pools, boulder jams, behind which gravel is
impounded, and older gravel banks are consolidated by mud
(in the lower part of the cave).
The walls of the cave are either stream worn (in their lower
reaches) or show bare, joint-controlled faces making visible
the bedrock through which the cave has cut. The ceiling is
shaped either by breakdown, exposing horizontal and planar
Figure 7. Geological profile of the ThisCave Series at
eastern wall of the resurgence.
Figure 6. Geological profile of the ThisCave Series at the
head of the Large Passage in ThisCave.
Figure 8.
Schematic
longitudinal
section
through
Kuka‘iau
Cave with the
most
important
geological
units.

Journal of Cave and Karst Studies, April 2003 • 61
KEMPE & WERNER
lava partings such as aa core – aa rubble interfaces, pahoehoe
surface roping or pahoehoe sheet separations. In a few
constrictions of the lower reaches of the cave and in the sump
section, the ceiling also shows marks of erosive action.
The cross-sections of the cave vary from wide canyons, to
narrow canyons, from rectangular habitus to a more oval
shape. In the main passage, they appear more to be determined
by breakdown enlargement rather than by down-cutting.
A
WALK THROUGH THE CAVE
The episodic surface stream, which feeds the cave, runs on
top of an aa core (in Hawaii called “blue rock”) that shows
vertical, wide-spaced, irregular columnar contraction joints. At
Station 5 (St. 5) of our survey this layer is breached and the
stream plunges down into a pit 6 m deep with two plunge pools
at the bottom. This and the two successive water falls, which
lead down to the cave entrance, constitute the impressive
entrance series. All water falls could be bypassed or free
climbed. On the walls various steeply sloping layers of loose
brownish or reddish paleosoils are exposed (Layers 1, 1a, 2
and 5 marked in the Longitudinal Section, Fig. 2) (Fig. 3).
Vegetation and moss make it difficult to follow these layers
entirely. Layer 5 can be traced from the bottom of the two
plunge pools of the first water fall to the foot of the second
water fall and along both sides of the widening below the last
of the three water falls, where it dips makai with 15°. Then one
stands in the gaping, 6.5 m high and 3 m wide, mouth of the
cave. Inside is another waterfall to be climbed down. It
exposes a thick layer of aa rubble, only partly consolidated by
fine-grained matrix and interspersed core layers making
climbing down the 6 m step easy and dangerous at the same
time since the abundant hand- and footholds can readily break
off. To the right, solid rock occurs and here the water normally
plunges down, having carved chutes. To the left, aa rubble and
thin, ledge-forming and steeply inclined (at first sloping at 18°
then at 25°) aa core layers compose the wall. The aa rubble
must belong to at least two different flows since the redbrown
paleosoil Layer 5 (termed Marker Layer) can be followed from
the entrance high up in the wall. The lower aa rubble could
represent a lateral levee of a large aa flow, the core of which
now forms the right-hand wall. The core layer is vertically
jointed (either tectonically or through contraction) and can
easily be mistaken as the remains of a thick lining.
The Entrance Hall is the largest room in the cave, 10 to 12
m high and 8 to 11 m wide. The plunge pit is filled with leaves
and tree trunks. Some large blocks from the aa bench litter the
floor. A chute with large, rounded blocks leads to a steeply
dipping solid platform (one of the blue rock layers) into which
a small canyon is cut. It is partly covered by large breakdown
blocks. The passage declines less steeply than the layering of
the rocks and the prominent red Marker Layer exposed on the
left wall meets the floor at about 70 m below the entrance
(Layer 5) for the first time.
This layer apparently played an important role for the
genesis of the cave, because it will be seen again and again up
to St. 65 either along the walls of the cave or at the base of
plunge pools. The solid lava sheet above it very often forms
the floor of the cave, sculptured by the running water. Above
this solid-lava layer is found another important layer, Layer 4.
It is composed of bedded but very weathered brown or gray
material. It can also be traced down to at least St. 59 and
appears to be the parting from which this section of the cave
originated. Consequently, from St. 21 onward, the cave
follows the local inclination of the strata which is actually
highly variable as indicated in the Longitudinal Section in Fig.
2.
Between St. 19 and 20, below the left wall, flotsam,
composed of plastic bottles, plastic parts, sandals, wood, and
the like is concentrated. Water apparently ponds up to this level
when the cave floods. This conclusion is substantiated by the
observation that the walls below this point appear clean-
washed while the walls of the upper Entrance Hall are
encrusted with dust.
Below St. 21, the cave widens as several small lava tubes
are intersected as described above. The upper one leads mauka
to a collapsed room, but can be penetrated for more than 30 m
makai. At first it is floored by muddy sand, which shows that
at high water the tube is also inundated. There are no signs of
erosion though. Towards the end, the sand disappears,
presumably washed into a crack in the floor and the tube
continues as a pristine small lava conduit. There is an air draft,
indicating that the tube does not end blindly, it is however too
small to be easily followed further.
At St. 26, the Plunge Pool Room opens up, offering the
most detailed view into the strata below the red Marker Layer
and an almost ten meter-high section of lava sheets and
intermittent aa layers can be studied (Fig. 4 and inset in Map
of Fig. 2). Further makai, the floor levels off as the red Marker
Layer descends down to the floor. Above, breakdown cupolas
open up to a height of 8 m. From one of them water seeps
down, possibly from an open lava tube intercepted at the
ceiling by breakdown.
Behind a prominent bend in the cave at St. 31, the first
high-water bypass opens up, a small lava tube developed at the
level of Layer 4. Its floor also forms the floor of the main tube
i.e. the lava sheet above the red Marker Layer. It is the base of
the stack of tubiferous pahoehoe. This sheet is relatively dense
and hard, showing that it was quickly cooled. Above it several
other sheets occur, on first inspection more vesicular and
possibly less quickly cooled. The next one up is the one in
which the small lava tube is situated. This layer could therefore
have been inserted between the base sheet and the next one up,
thereby forming a hot core in the inflating lava flow (for the
concept of how primary lava conduits form at the tip of tube-
fed pahoehoe flows compare Kauahikaua et al. 1998). The hot
core is the site where the proto-tubes form, often in parallel to
each other. The branching-off of the side passage suggests that
there may have been a small tube also within the main passage
mauka. There are remains of glazing on the right wall of the
main passage immediately below the branching, therefore a

62 • Journal of Cave and Karst Studies, April 2003
KUKA‘IAU CAV E , MAUNA KEA, HAWAI'I, CREATED BY WATER EROSION
From here on, no remains of small tubes have been noticed,
the cave moves through a tightly sealed stack of pahoehoe
sheets. The “terminal sump” was reached first by M.S. Werner
and R. Elhard in May 2000. On June 7th 2002 we entered the
makai entrance when exploring the sump section and
investigated it geologically on a trip through the entire cave on
June 23rd. The passage leading toward it has the most unusual
form: First it rises up by 5 m and then becomes very low
because of gravel being transported up a chute; it then plunges
down by 7 m to a terminal pond filled with murky brown water
(in May 2000). Solid mud banks with cobbles line the passage,
and these have more recently been cut and partially removed.
Overall, the hall above the sump gives an impression of how
the water churns when at full flow.
In summer 2002 the water level fell about a meter below its
stand in May 2000, making the sump accessible. Still one has
to wade and in Echo Lake one is up to the belly in the water.
From the terminal sump the passage is level, much deeper on
the left than on the right side. It ends blindly at St. 78. A gravel
bank leads into the continuing passage which makes a sharp
right turn. Here the upward rising rock floor is encountered
again. A large stream pot, “The Tub”, interrupts the floor.
Beyond, a solid Rock Dam bars the passage, damming “Echo
Lake” from flowing back mauka into the sump. This is the
most unusual feature in the cave, difficult to envisage even
when described. There is even a small channel incised into the
dam, through which the water of Echo Lake trickles backward
into the lower-lying but dry sump. The crest of the Rock Dam
is horizontal, in level with Echo Lake. The lake is about 8 m
across and over 2 m deep at the far side and a sizeable cupola
helps to reflect the sound giving the lake its name. The Rock
Dam prevents the water level from falling lower, the level is
therefore no indicator if the sump mauka of the lake is passable
or not. We only understood that the sump opened up because
of a pulsating air flow.
The passage makes another sharp turn to the left and one
emerges from the lake into the lower part of ThisCave where
the passage turns right again (Fig. 5). A few logs rest on the
lake floor. At S. 23A the two surveys were tied together. The
floor of the passage rises out of the water and meets a steep
gravel chute leading upward into the Well Head Room. On the
far side there is a blind appendix again, carved out by the water
jet churning around. Here we found in June 2002 a truck tire
resting on our survey St. 21. It must have passed the cave
during a flood in the winter 2001/2002. The passage makes
one last left-hand turn and another gravel chute leads steeply
upward into the vadose passage of ThisCave, breaking through
the top of the pahoehoe Sump Unit. This unit is characterized
by the absence of any open space in the rock package: All
bedding planes, all contraction cracks and the vesicles of the
rock are filled with a white, waxy material, thereby effectively
sealing the rock. Due to this compact wall surface, scalloped
on ceiling, walls and floors, the passages reverberate with
sound, similar to the acoustic perception of limestone caves.
None of the wall sections inspected revealed any lava tube
small tube may also have continued down the main passage.
A little further makai, the base sheet of the tubiferous flow
has been cut through by the erosion of the stream, exposing the
red Marker Layer, lining the sides of the plunge pool below. A
small canyon, formed by the protruding ledges of the base
sheet of the pahoehoe flow, forms the lower part of the main
passage. Here molds of trees, encased by the base sheet are
exposed. They illustrate that the Marker Layer is in fact a
paleosoil, which carried a forest.
At the lower intersection with the bypass, a large pool is
situated (Horst’s Pool, named in honor of Horst-Volker
Henschel who took a plunge in it when one of the handholds
of the lateral traverse came loose).
Mauka of St. 40, a fault is crossed, down-faulted makai.
Then Spike’s Pool is encountered, below a 2 m high waterfall.
It is the best example of a deep, wall-to-wall pool. Seemingly
not climbable, Spike Werner found a way around it, needing to
follow one of the lower aa layers beneath the lip of the water
fall along the right wall. Below St. 47 (at a depth of 76.5 m
below the first waterfall in the entrance), several more lava
tube-bypasses occur on the right-hand side and the cave
becomes somewhat smaller and less steep. At St. 50, before a
small fault at which the mauka side is down-faulted, the red
layer reappears once more. Below St. 51, a filled lava tube is
cut through, as already mentioned. In this section the exposed
pahoehoe is rather thin-sheeted and bears rope marks typical of
pahoehoe solidified at the surface. Makai more small tubes
enter and leave the main passage at various levels. The cave
now is considerably lower than before and has more the
appearance of a phreatic tube than that of a canyon. The floor
consists of gravel.
The clean-washed floor reappears before St. 62, where the
cave narrows once more. This is the section which best
resembles the appearance of a lava tube. Since we have been
this low in the cave only twice (due to the high risk of being
caught by a sudden storm discharge) there is not enough
information to assure that this section of the cave is also
structurally a lava tube. Below this constriction (2*1.5 m) the
cave opens up once more into a larger plunge pool room.
Again the red Marker Layer is seen along the perimeter of the
pool below the waterfall lip. In spite of the otherwise wetter
conditions (compared to the spring of 2000), this pool was
empty in March 2001 (except for a thick deposit of decaying
leaves), suggesting that the water can escape into a deeper
groundwater body by means of the various aa layers
underlying the tubiferous pahoehoe. One has to climb upward
and out of the pool at the makai end, unlike the end of the
previous pools which terminate mostly in gravel chutes. This
raises the question where the material eroded from the pool
was transported. The passage beyond is horizontal and floored
with sandy mud. In April 2000, it was dry, but in March 2001
it was filled with water from a previous flood in the cave and
sumped at around St. 68, so that the terminal sump could not
be reached.

Journal of Cave and Karst Studies, April 2003 • 63
KEMPE & WERNER
Table 3. A) Semi-quantitative results of X-ray-diffraction analyses of rock samples (analyses courtesy of R. Apfelbach,
Darmstadt). Percent concentrations were estimated according to the reflex intensity as compared to known standards.
Percentages refer to X-ray active minerals only where there is a large amorphous fraction. In the samples with small
amorphous fractions the amorphous fraction was estimated as well. As an example the results of samples TC-21 to TC 23
are given in both notations, simple numbers denote overall composition, numbers in brackets denote composition of the
crystalline fraction only. Sample ThatCave 3 was analyzed for various grain size fractions.
Sample Color Plagioclase Augite Olivine Hematite Others Rest
of ground sample
Samples collected Aug. 2000 between St. 25 and 27 by M.S. Werner and W.R. Halliday
ThatCave 1, Gray, vesicular Light gray 25% 24% 42% 9%
(2-3 mm) lava, white pore fillings amorphous
ThatCave 2a. Brown, vesicular Dark brown 7% 58% - 4% 5% 26%
(ca. 1mm) lava meta-halloysite amorphous
ThatCave 2b brown mud on Brown - 43% 5 % 50%
surface of 2a amorphous
ThatCave 3a (3=red Marker Layer) Reddish brown - 48% 14% 28%
coarse component amorphous
ThatCave 3b Light greenish gray - 96%
Green shards of augite clasts
That Cave 3c fine-grained components Reddish brown 14% 52% - 12% 5% 17%
lighter than 3a) meta-halloysite amorphous
That Cave 3c fine-grained components (< 2µ) Reddish brown 13% 53% 6% 17% 11% Some
(forsterite) meta-halloysite amorphous
March 2001
TC 1 surface between This- and ThatCave, Light gray 64% 23% 10% - 3%
yellowish weathered rock Magnetite
TC3, Profile St. 26-27 Light brown 14 % 36% 47% 3%
Layer 4! (10cm crumbly rock)
TC4, ProfileSt.26-27 Light gray, brownish 21% 36% 40% 3%
above Layer 4 (13 cm harder layer)
TC5, Profile St. 26-27 Light gray 26% 45% 27% 2% Large X-ray
above Layer 4(1 m crumbly rock) amorphous
component
TC 6, Layer 4 Light gray-brownish 22% 53% 14% 5% Halloysite: Large X-ray
St. 31 begin of bypass 6% amorphous
component
TC 7, Red Layer, Layer 5, makai St. 31 Light gray brownish 16% 28% 19% 29% 8% Large X-ray
undeterminable amorphous
component
TC 9, Lava from wall, above St. 42, Brownish 23% 48% 13% 11% 5% Large X-ray
large white spec in pores undeterminable amorphous
component
TC 10, layer 1 at St. 7 Light brown red 12% 48% 13% 45% Large X-ray
amorphous
component
TC 11, layer 1a at St. 9 Brown red - - - 37 % Quartz! 19% Very large X-ray
Donathite* 44% amorphous
component
TC 12. Layer 2, at St. 9 Dark gray brown - 15% 41% 22% Illite 18% Very large X-ray
4% undterminable amorphous
component
TC 14, Layer 5 between St. 20 and 21 Brown-red 15% 45% 13% 22% 5% Very large X-ray
undeterminable amorphous
component
TC 15, Layer 5, red Marker Layer at St. 23 Intensive brown-red 0% 39% 16% 39% 6% Large X-ray
undeterminable amorphous
component
TC 16, solid lava above red Marker Layer at St. 22 Light gray 32% 39% 22% 0% 7% Medium X-ray
undeterminable amorphous
component
June 2002
TC 20 ThisCave, white waxy material Light grey - - - - 100% Large amorphous
sealing voids in Sump Pahoehoe Unit Halloysite component
TC 21, Layer 1a at entrance to ThatCave, St. 5 Reddish brown 5% 5%
(15%) - - (15%) Quartz 11 % (33%) 67% amorphous
Halloysite 5% (15%)
Donathite 7% (21%)
TC 22, weathered aa rubble above TC 21 Light brown 4% - - Halloysite 18% (72%) 75% amorphous
TC 23, weathered aa rubble above TC 22 Dark brown 7% - - - Halloysite18%(67%) 73% amorphous
Magnetite 2% (7%)
Note that samples from 2000 were dried at < 50°C transforming the halloysite into meta-halloysite. *Donathite (Fe,Mg)(Cr,Fe)
2
O
4
)
Table 3. B) Results of carbon, nitrogen, sulfur (C, N, S) total elemental analyses of rock samples from ThatCave, results are
given as weight percent.
Sample C(total) % C(inorg.) % C(org.) % Total N % C/N S %
org. matter %
ThatCave 2b brown mud on surface of 2a 3.67 0.33 3.34 8.35 0.43 7.8 1.19
That Cave 3c fine-grained components 0.04 0.00 0.04 0.1 0.11 0.36 0.76

64 • Journal of Cave and Karst Studies, April 2003
KUKA‘IAU CAV E , MAUNA KEA, HAWAI'I, CREATED BY WATER EROSION
structure. The upward course of the passage, the fact that the
ceiling of the passage jumps upward from sheet parting to
sheet parting and the seemingly joint-controlled sharp turns of
the passage exclude any possibility that a precursor lava tube
has been enlarged by erosion in this section of the cave
Beyond the third and last chute the water boils upward into
the rectangular vadose passage of ThisCave, which is up to 14
m wide. Light filters down through the pit in the corner to the
left. From here on, the passage drops at about 6° to the stream
resurgence less than 200 m away. On its way it winds through
an S-shaped bend. The floor of the cave is formed by the sealed
pahoehoe of the Sump Unit (Fig. 6). The last sheet carries
many tree molds (best exposed at some of the cataracts),
suggesting that there was a longer hiatus between the
deposition of the bulk of the Sump Unit and the last flow
covering it. Below the layer with the tree trunk a thin bed of
soft material is found, possibly also a paleosoil. At places the
water has eroded several meters into the top of the Sump Unit,
forming cascades and showing scalloping and stream pots. The
wide and flat roof is formed by the lower interface of a thick
aa core layer. Only in a few places have blocks fallen out of the
ceiling. They are the source of the large rounded blocks in the
streambed. The walls are composed of a series of
unconsolidated rocks (Fig. 6). There are two red paleosoil
layers, one immediately on top of the pahoehoe Sump Series,
an aa rubble layer and a thicker diamict layer on top. The
diamict layer continues to the mouth of the cave. There it
increases in thickness causing the blue rock ceiling to rise on
the last few meters of the passage. On the left wall the diamict
shows an interesting internal structure with discontinuous red
soil layers incorporated in it (Fig. 7). It is this diamict layer
which has been removed to create the cave. Certainly no
precursor lava tube could have existed in this rock series. This
section of the cave is entirely erosional in origin.
Outside of the cave’s exit a cobble bank causes water to
pond inside the entrance, forming a temporary lake. This
temporary lake may have caused the ranchers to dig the
artificial pit in order to gain access to the water in the sump at
all times. From the mouth of the cave the stream flows first
over gravel but then the blue rock of the cave roof is
encountered again, possibly down-faulted by a small fault
crossing the creek. A few meters makai the creek plunges into
an impressive rock basin. There the diamict layer is exposed
once more, featuring also a red soil layer.
P
ETROGRAPHIC DATA
In spite of the morphological clues speaking of an erosive
origin for Kuka‘iau Cave, it nevertheless is challenging to
explain the origin of an erosive cave of such dimensions in
lava layers which at least in part lack the advantage of a
precursor lava tube. In order to advance our understanding of
the cave genesis we therefore took rock samples for
mineralogical, petrographic and geochemical analysis in
August 2000 (W.R. Halliday and M.S. Werner, designated
“ThatCave 1 to 3”), March 2001 (S. Kempe and I. Bauer,
labeled “TC 1 to 16”) and in June 2002 (labeled “TC 20-23”).
A subset of these samples was ground and X-ray-diffraction
(XRD) analyses were conducted (Philips PW 1949 powder
diffractometer, data evaluation according to the International
Centre for Powder Diffraction Data). Additionally some thin
sections were made and elemental carbon, nitrogen, and sulfur
concentrations (CNS Analyzer Vario El, Elementar) were
determined. Porosity was measured on one sample as well. All
results are listed in Table 3.
The analytical results show four general types of
composition: (i) samples with a general basaltic composition,
(ii) samples with a high hematite content, (iii) samples with a
composition unusual for Hawai‘i, and (iv) samples with a high
halloysite content.
The first group comprises the samples (ThatCave and TC1,
3, 4, 5, 6, 9, 16). They have high augite, olivine and plagioclase
contents; hematite may be present in small quantities. These
samples are gray or gray with a brownish hue in color and
represent the solid lava samples composed of tholeiitic and
alkalic basalt representing the Hamakua Volcanic Series of the
Mauna Kea volcanic edifice. In thin section, the olivine is
partly altered to iddingsite (MgFe
2Si3
O10*4H2O) or hematite
(Fe
2O3
) which form along the cleavages of the olivine crystals.
The samples (ThatCave 3 and TC 7, 10, 14, 15) from the
bright red brown, unconsolidated material of Layers 1 and 5
form the second group. They are also composed of augite,
olivine, and plagioclase but have significantly higher hematite
contents. These samples most probably represent paleosoils
developed on weathered volcanics.
Samples TC 11, TC 12, and TC 21 belong to a third group.
They represent unconsolidated, fine-grained material with a
composition atypical of Hawaiian lava, i.e., with significant
concentrations of quartz, donathite, and illite. Layer 1a, which
contains quartz, donathite and hematite, is a soil layer
containing continental (Asian) dust (the only known source of
quartz for Hawaii). Under the ESEM (environmental scanning
electron microscope) quartz particles were difficult to find
since they are coated with hematite and amorphous Al-
silicates. The few grains identified by EDX (energy dispersive
analysis of X-rays) had a short, rounded, columnar form, 1-2 µ
long. This habit would support the interpretation of continental
dust as a source for the quartz. Layer 2 contains illite in
addition to augite, olivine, and hematite. It again could be a
paleosoil in which illite could be either a dust-born addition or
a weathered equivalent of the plagioclase.
Table 3. C) Results of porosity analysis of rock sample from ThatCave.
ThatCave 2a. Brown, weight Envelope volume Sample volume Density porosity
vesicular (ca. 1 mm) lava 43.804 g 27.54 cm
3
13.78 cm
3
3.178 g/cm
3
49.9%

Journal of Cave and Karst Studies, April 2003 • 65
KEMPE & WERNER
Sample TC 6 (Layer 4), 20, 22 and 23 – the fourth group of
samples–have high contents of halloysite
(Al2
Si2O5
(OH)4*4H2O). In samples from March 2000
(Samples That Cave 2a, 3c) meta-halloysite (Al
2Si2O5(OH)4)
was detected. It most probably originated from halloysite
because the samples taken in 2000 were dried at higher
temperatures. Halloysite is the only X-ray detectable
component in the white, waxy material filling the voids in the
Sump Series pahoehoe (Sample TC 20). Halloysite was also
detected in several of the samples from Pa‘auhau Civil
Defense Cave (see Kempe et al. 2003). Halloysite is a clay
mineral which forms during weathering and is common in
Hawaiian soils and weathered basaltic rocks (so called
saprolite) (e.g., Patterson 1971; Vitousek et al. 1997).
To check if all of the white, waxy material in the vesicles
of the lava is halloysite, we scraped some of it from the pores
of rock sample TC 9 and X-rayed it. It proved to be
amorphous. We then looked at this waxy material under the
ESEM and analyzed its composition by EDX. It contains a
variable amount of Al and Si; thus it could either be the
amorphous forerunner of halloysite or represent the
amorphous Al/Si phase called allophane
((Al
2
O3)(SiO
2)1.3
*2.5H2O).
ESEM examination of a subsample from Layer 1a (the one
with the quartz, TC 11 and TC 21) and Layer 2 (with illite, TC
12) showed that layered, amorphous Al-Si phases are
ubiquitous in the paleosoils. They, together with the very fine-
grained hematite mantle all the surfaces of other minerals
almost completely. It is difficult to identify the other phases in
the sample by EDX, and only the larger grains of weathered
augite seem to be relatively clean. This, however, could be due
to the fact that they easily break in sample preparation,
producing fresh cleavage faces.
C-N-S data (Table 3b) show that the mud on the wall of the
cave (sample ThatCave 2b) is eroded recent soil with a high
organic C content (3.3 %) and a high C/N ratio (7.8 by weight).
In contrast, the red Marker Layer (sample TC 3c) has very low
C and N contents. This is either due to its older age or to the
fact that the soil was baked by the transgressing pahoehoe
sheets which oxidized all of the former soil matter.
C
ONCLUSIONS FROM THE MORPHOLOGICAL AND GEOLOGICAL
OBSERVATIONS
The XRD data illustrate that the weathering of feldspars,
olivine, and augite has created amorphous or poorly
crystallized clays, i.e., halloysite and allophane. These
accumulate in the voids of the rock, causing the rock layers to
become less permeable. This in turn gives rise to a local
perched water table. The more the sinking water is retarded,
the more weathering will occur facilitating the final closing of
vesicles and other interstices.
The geological observations in the cave allow us to identify
two major impermeable layers: The paleosoil Layer 5, i.e. the
red Marker Layer inclusive of the pahoehoe sheets
immediately above it and the Sump Series pahoehoe. The
paleosoil could have served as an initial water impediment to
retard further vertical drainage and cause lateral groundwater
flow in the small lava tubes above the base sheet and through
the partings between these lava sheets. This alone however is
not enough to form the cave.
A large hydrostatic pressure must be exerted as well. This
was possible because Layer 5 runs through a highly porous aa
rubble series near the entrance which was intercepted by a
gully. This water could rapidly infiltrate through the aa, fill the
voids in the stack of the tubiferous pahoehoe below and exert
a pressure of almost 10 bar on the distal part of the
groundwater body. This could have been enough pressure to
force the water along orthogonal joints upward through the
otherwise impermeable pahoehoe Sump Series into the
overlying diamict. The diamict has a high porosity (no exact
data yet available because the rock is very crumbly) and could
conduct the water along its layer to a new exit, the present
resurgence. In this way an erosive cave could have formed by
a set of fortuitous geological circumstances at this site (Fig. 8).
In this model it is not necessary to have a tubiferous pahoehoe
as an initial water conductor. aa rubble or any other rock types
of a high initial porosity would serve just as well. Therefore
the cave is not just an eroded lava tube where a stream invaded
at the mauka end and flowed out at the makai end (for an
example of this type see the Pa‘auhau Civil Defense Cave,
Kempe et al. 2003), but it is the product of a complex
interaction between a perched water table and the structural
components of the rock strata.
C
ONCLUSIONS DERIVED FROM CHANNEL MORPHOLOGY AND
S
OIL AGE FOR THE AGE OF THE SYSTEM
The morphology of bedrock channels is determined by
substrate and hydraulics, (e.g., Wohl & Merritt 2001).
According to Wohl (1998) the reach-scale morphology (i.e.,
the morphology of a section of the channel several times
longer than its width) of Kuka‘iau Cave can be classified as a
Figure 9. Comparison of the depth/length profile of
ThatCave (erosional) and Pa‘auhau Civil Defense (lava
tube).

66 • Journal of Cave and Karst Studies, April 2003
KUKA‘IAU CAV E , MAUNA KEA, HAWAI'I, CREATED BY WATER EROSION
single flow path, variable bed gradient channel, with a
dominant step-pool morphology (“downstream bed
undulations in form of vertical steps with pools between
them”). The step-pool morphology is aided by the
heterogeneity of the bedrock itself: lava layers are separated by
layers of loose material which can be ash layers, paleosoil
layers, aa rubble, or simply contact zones between pahoehoe
sheets. Also, vertical contraction cracks and possibly faults
play a role in bedrock strength in Kuka‘iau Cave. Overall the
bedrock is inhomogeneous, therefore aiding in the formation
of a step-pool morphology.
Overall, the depth-length relation of the cave is best
described by a polynomial fit (given in Fig. 9), consistent with
a water-related origin. Groundwater tends to sink quickly
toward the water table before flowing horizontally. In contrast,
lava tubes follow the slope of the mountain and tend (on the
hundred meter scale) to have a linear depth-length relation.
This is best shown by the comparison between the profiles of
the upper part of Kuka‘iau Cave (i.e. ThatCave) and Pa‘auhau
Civil Defense Cave (Fig. 9). Also, the general decrease of the
passage size with length is a feature more consistent with
water flow than of lava flow.
Total stream power (in watts) is calculated from maximal
flow, the specific density of water and bed gradient. The total-
reach gradient is ca. 100 m/1000 m, i.e., 0.1, or ca. 0.18 if
considering only the mauka vadose section of Kuka‘iau Cave.
This is comparable with other gradients of step-pool streams,
such as those given in Wohl and Merrit (2001, Table 2), who
list eight rivers with gradients between 0.02 and 0.2 and
maximal discharges of between ca. 4 and 700 m³/sec. The
overall gradient appears to be the most important factor
determining step-pool channels as illustrated by the statistical
analysis of Wohl and Merrit (2001) of over 40 river channels.
The analysis suggests also that step-pool channels develop
where there is a relatively low ratio between driving forces
(i.e., stream power) and rock resistance. “The presence of steps
and plunge pools, which may result from differential erosion
associated with substrate heterogeneity localizes and
maximizes erosional force in the plunge pools” (Wohl &
Merrit 2001, p. 1211). Since maximal flow of the creek feeding
Kuka‘iau Cave remains unknown at this stage of the
investigation, maximal stream power cannot be determined as
yet, but cobble size and inclination of pool-exit chutes should
provide clues to estimate this important stream characteristic
in further studies.
It is also interesting to consider the temporal aspect of the
evolution of the cave. Apparently no model exists as yet which
links down-cutting rates of bedrock channels with gradients,
water flow and substrate characteristics. Even if such models
would exist, it will be difficult to apply them to the ThatCave
case, simply because we do not know how active the creek is.
We do know - due to a rim of modern flotsam - that the creek
periodically floods the cave in present time up to a height of 65
m above the level of the sump. The cave flooded in November
of 2000 when the island of Hawai‘i experienced an
exceptionally heavy rainy season and in the winter of 2001-
2002, as indicated by the tire, which was washed through the
cave. However, the cave’s formation may well be linked to
past climatic conditions, for example, to the time when the ice
cap of Mauna Kea melted at the end of the Last Glacial
Maximum. Dethier (2001) published a recent overview of river
incision rates for the Western United States, using the Lava
Creek B Tephra, erupted from the Yellowstone caldera ca. 0.64
Ma ago, as a time Marker Layer. His results show that incision
rates ranged from 30 (for very steep terrain in the Rockies) to
< 5 cm/ka (along the plains west of the Mississippi/Missouri)
since the deposition of the tephra. If such values also apply for
ThatCave (average height assumed as 3 m) then the cave could
have formed in a period of less than 10 ka to more than 60 ka.
The finding of quartz and illite (a fine-grained mica) in the
paleosoil Layer 1 (TC 11 and TC 20) suggests that this soil was
exposed to higher dust-fluxes, such as occurring during glacial
times. Vitousek et al. (1997) describe sites on Mauna Kea with
20 ka old soils and on Kohala with a soil age of 150 ka. Both
sites contain dust-derived quartz. Kurtz et al. (1999) found a
total of 6 g dust per cm² at the 20 ka site and of 14-18 g/cm² in
the older sites. We found (sample TC 20) a quartz content of
ca. 11% of the total. If we assume a soil density of 1.5 g/cm³
and a layer thickness of ca. 20 cm, one can estimate the quartz
content to ca. 3 g/cm². Compared to the Mauna Kea site of
Vitousek et al. (1997), Layer 1 could have an exposure time of
around 10 ka. Total weathering time can be estimated by
looking at the feldspar contents. Vitousek et al. (1997) show
that feldspar is lost between the 20 ka and the 150 ka old sites.
Since feldspar is missing in the samples containing quartz (TC
11) and illite (TC 12) (but not in sample TC 21) and in one of
our samples from the red Marker Layer (sample TC 15) one
must assume that these soils have a weathering age older than
20 ka. Since exposure time and weathering time do not agree,
one can tentatively conclude that the weathering of the soils
continues even after they have been covered by the next lava
flow. This conclusion is also substantiated by the observation
that the aa layer above Layer 1 is well weathered (TC 22 and
TC 23, compare also Fig. 3), containing large fractions of
amorphous matter and only a few percent of feldspar.
Apparently the underlying soil served as a water retarder,
facilitating the continuation of the weathering reactions. At the
same time it is illustrated that water must have collected and
flowed above the soil through the interstices of the aa rubble.
The presence of magnetite in place of hematite and donathite
could suggest that this weathering occurred under reducing
conditions, again pointing at weathering within a perched
groundwater body.
All these observations allow concluding that the lavas into
which the cave system was eroded must be at least predating
the last glaciation. They could be as old as 100 ka, but not
much older than that, consistent with the published youngest
ages of the Hamakua Volcanic Series (see Chapter 2). This
limits the age of the cave to a few 10 ka fitting into the
considerations derived from the discussion of the erosion rate.

Journal of Cave and Karst Studies, April 2003 • 67
KEMPE & WERNER
We hope to get a further age constraint of the bedrock of the
cave from a charcoal sample recently recovered by Dr. Jack
Lockwood in ThisCave from the aa rubble below the diamict.
A
CKNOWLEDGEMENTS
We thank Ric Elhard, Horst-Volker Henschel, Phillip
Stankiewicz, Ingo Bauer, Christine and Herbert Jantschke,
Wolfgang Morlock, Andi Kücha, Marci Strait, and John Elhard
for continuous and profound help in the field. We are grateful
to R. Apfelbach, for XRD analyses, Dr. M. Ebert for ESEM
and EDX analyses, R. Brannolte for CNS analyses, G.
Schubert for porosity measurements, and E. Wettengl for help
in drawing the maps, all Inst. of Applied Geosciences,
Darmstadt. We thank Dr. Jack Lockwood and Dr. Oliver
Chadwick for their valuable discussion of the profiles in
ThisCave. We are indebted to Dr. R. Tilling and another
reviewer for critical remarks on the first version of this
manuscript.
R
EFERENCES
Allred, K. & Allred, C., 1997, Development and morphology of Kazumura
Cave, Hawaii: Journal of Cave and Karst Studies, v. 59, n. 2, p. 67-80.
Dethier, D.P., 2001, Pleistocene incision rates in the western United States
calibrated using Lava Creek B tephra: Geolog. v. 29, n. 9, p. 783-786.
Greeley, R., Fagents, S.A., Harris, R.S., Kadel, S.D., & Williams, D.A., 1998,
Erosion by flowing lava, field evidence: Journal of Geophysic Research
v. 103, n. B11, p. 27, 325-27, 345.
Halliday, W.R., 2000a, Kalopa State Park Stream Conduit Caves: Newsletter
Hawaii Speleological Survey of the National Speleological Society, v. 8,
p. 4.
Halliday, W.R., 2000b, A lava tube stream conduit system on Mauna Kea
Volcano, Hawaii: Newsletter Hawaii Speleological Survey of the National
Speleological Society 8: 5.
Kauahikaua, J., Cashmann, K.V., Mattox, T.N., Heliker, C.C., Hon, K.A.,
Mangan, M.T., & Thornber, C.R., 1998: Observations on basaltic lava
streams in tubes from Kilauea Volcano, island of Hawaii: Journal of
Geophysical Research, v. 103, n. B11, p. 27, 303-27, 323.
Kempe, S., 1997, Lava falls: a major factor for the enlargement of lava tubes
of the Ai-la‘au shield phase, Kilauea, Hawaii: Proceedings 12.
International Congress of Speleology. 10.-17. Aug. 1997, La Chaux-de-
Fonds, Switzerland, v. 1, p. 445-448.
Kempe, S., 2000, Expedition report, Hawaii March 2000: Newsletter Hawaii
Speleological Survey of the National Speleological Society, v. 2000, n. 7,
p. 25-29.
Kempe, S., 2002, Lavaröhren (Pyroducts) auf Hawaii und ihre Genese, in
Rosendahl, W. & Hoppe, A., eds.: Angewandte Geowissenschaften in
Darmstadt.- Schriftenreihe der deutschen Geologischen Gesellschaft, Heft
15, p. 109-127.
Kempe, S., Bauer, I. & Henschel, H.V., 2001, Expedition Report, Hawaii,
March 2001: Newsletter Hawaii Speleological Survey of the National
Speleological Society, v. 10, p. 3-10.
Kempe, S., Bauer, I, & Henschel, H.V., 2003, Pa‘auhau Civil Defense Cave on
Mauna Kea, Hawaii - A lava tube modified by water erosion: Journal of
Cave and Karst Studies, v. 65, p. 76-85.
Kurtz, A.C., Derry, L.A., & Chadwick, O.A., 1999, Trace element
redistribution and Asian dust accumulation in a Hawaiian soil
chronosequence: Abstracts 9th Annual V.M. Goldschmidt Conference
#7194 (http://www.lpi.usra.edu/meetings/gold99/pdf/7194.pdf).
Patterson, S.H., 1971, Investigations of ferruginous bauxite and other mineral
resources on Kauai and a Reconnaissance of ferruginous bauxite deposits
on Maui, Hawaii: U.S. Geological Survey Professional Paper, v. 656, 74
p.
Stearns, H.T. & Macdonald, G.A., 1946, Geology and Groundwater Resources
of the Island of Hawaii: Territory of Hawaii, Division of Hydrography,
Bulletin, v. 9, 363 p .
Stone, F., 1992, Kuka‘iau Piping Cave: Internal Report Prepared for Kuka‘iau
Ranch, 5 p., 1 fig., unpublished (pers. com. Stone, 17.06.02).
Vitousek, P.M., Chadwick, O. A., Crews, T. E., Fownes, J. H., Hendricks, D.
M. & Herbert, D., 1997, Soil and Ecosystem development across the
Hawaiian Islands: Geological Society of America Today, v. 7, n. 9, p. 1-8.
Werner, M.S., 1997: The Honoka‘a Pseudokarst: Hawaii Grotto News, v. 6, n.
1, p. 4, 14-16.
Werner, S., Kempe, S., Henschel, H.-V., Stankiewicz, P., Elhard, R., Strait, M.
& Elhard, J. 2000, ThisCave and ThatCave: Hawaii Grotto News, v. 9, n.
2, p. 13-17.
Werner, S., Kempe, S., Henschel, H.-V. & Elhard, R., 2002, Kuka‘iau Cave
(alias ThisCave and ThatCave): Exploration report of a lava cave eroded
by water, a new type of Hawaiian cave: National Speleological News,
December, p. 346-357.
Wohl, E.E. & Merrit, D.M., 2001, Bedrock channel morphology: Geological
Society of America Bulletin v. 113, n. 9, p. 1205-1212.
Wohl, E.E. 1998: Bedrock channel morphology in relation to erosional
processes, in Tinkler, K.J. & Wohl, E.E. eds., Rivers over Rock: Fluvial
Processes in Bedrock Channels: Washington, D.C., American.
Geophysics Union, p. 133-151.
Wolfe, E.W., Wise, W.S. & Dalrymple, G.B., 1997, The Geology and Petrology
of Mauna kea Volcano, Hawaii: A Study of postshield volcanism: U.S.
Geological Survey Professional Papers, v. 1557, 129 pp + 4 pl.