What makes hydromagmatic eruptions violent? Some insights ...scott/Workshop_reading/...What makes...
Transcript of What makes hydromagmatic eruptions violent? Some insights ...scott/Workshop_reading/...What makes...
www.elsevier.com/locate/jvolgeores
Journal of Volcanology and Geothermal Research 137 (2004) 15–31
What makes hydromagmatic eruptions violent? Some insights
from the Keanakako’i Ash, Kılauea Volcano, Hawai’i$
Larry G. Mastina,*, Robert L. Christiansenb, Carl Thornbera,Jacob Lowensternb, Melvin Beeson1
aU.S. Geological Survey, Cascades Volcano Observatory, 1300 SE Cardinal Court, Building 10, Suite 100, Vancouver, WA 98683, USAbU.S. Geological Survey, MS 910, 345 Middlefield Road, Menlo Park, CA 94025, USA
Abstract
Volcanic eruptions at the summit of Kılauea volcano, Hawai’i, are of two dramatically contrasting types: (1) benign lava
flows and lava fountains; and (2) violent, mostly prehistoric eruptions that dispersed tephra over hundreds of square kilometers.
The violence of the latter eruptions has been attributed to mixing of water and magma within a wet summit caldera; however,
magma injection into water at other volcanoes does not consistently produce widespread tephras. To identify other factors that
may have contributed to the violence of these eruptions, we sampled tephra from the Keanakako’i Ash, the most recent large
hydromagmatic deposit, and measured vesicularity, bubble-number density and dissolved volatile content of juvenile matrix
glass to constrain magma ascent rate and degree of degassing at the time of quenching. Bubble-number densities (9� 104–
1�107 cm� 3) of tephra fragments exceed those of most historically erupted Kılauean tephras (3� 103–1.8� 105 cm� 3), and
suggest exceptionally high magma effusion rates. Dissolved sulfur (average = 330 ppm) and water (0.15–0.45 wt.%)
concentrations exceed equilibrium-saturation values at 1 atm pressure (100–150 ppm and f 0.09%, respectively), suggesting
that clasts quenched before equilibrating to atmospheric pressure. We interpret these results to suggest rapid magma injection into
a wet crater, perhaps similar to continuous-uprush jets at Surtsey. Estimates of Reynolds number suggest that the erupting magma
was turbulent and would have mixed with surrounding water in vortices ranging downward in size to centimeters. Such fine-scale
mixing would have ensured rapid heat exchange and extensive magma fragmentation, maximizing the violence of these eruptions.
Published by Elsevier B.V.
Keywords: Kılauea; Phreatomagmatism; explosive eruptions; vesicular texture; turbulence
1. Introduction hydromagmatic tephra deposits erupted from the sum-
The Keanakako’i Ash (Fig. 1) is the most recent,
best exposed and most thoroughly studied of several
0377-0273/$ - see front matter. Published by Elsevier B.V.
doi:10.1016/j.jvolgeores.2004.05.015
$ Supplementary data associated with this article can be found,
in the online version, at doi: 10.1016/j.jvolgeores.2004.05.015.
* Corresponding author. Tel.: +1-360-696-7518; fax: +1-360-
993-8980.
E-mail address: [email protected] (L.G. Mastin).1 formerly U.S. Geological Survey, MS 910, 345 Middlefield
Road, Menlo Park, CA 94025, USA.
mit of Kılauea volcano, Hawai’i, since the late Pleis-
tocene. Originally thought to have been produced
during a single eruption around A.D. 1790 (Dana,
1888), the number of eruptions represented by the
deposit has been repeatedly revised (Stone, 1926;
Powers, 1948; Christiansen, 1979; McPhie et al.,
1990). It is currently thought to have resulted from
at least a few eruptions over a period of perhaps a few
centuries that ended around 1790 AD (Swanson et al.,
1998; D. Swanson, written comm., 2001). The most
Fig. 1. (a) Index map of Kılauea showing the areal extent of the
Keanakako’i Ash (heavy dotted line; from R. Christiansen,
unpublished data), (b) close-up of summit showing sample
locations (numbered). Coordinates of numbered locations are given
in Table 2.
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–3116
extensive Keanakako’i units that likely resulted from
a single eruption (e.g. unit 2 of McPhie et al., 1990)
cover a few hundred square kilometers (R. Christian-
sen, unpublished data, 2002), similar in area to the
deposits of moderately large historical basaltic hydro-
magmatic events (e.g. Taal, 1965; Moore et al., 1966).
Late Pleistocene tephras at Kılauea, however, are
meters thick in exposures 10 km south of the caldera
(loc. 1, Fig. 1a; Easton, 1987; Clague et al., 1995a)
and may have originally covered thousands of square
kilometers.
The violence of eruptions that produced these
deposits is generally attributed to interaction of magma
with water, as inferred from their fine grain size, wide
range in vesicularity, cross-bedding, accretionary lap-
illi, and lithic components in the ejecta (Swanson and
Christiansen, 1973; Decker and Christiansen, 1984;
McPhie et al., 1990; Dzurisin et al., 1995; Clague et al.,
1995a). Mastin (1997) hypothesized that water influx
was caused by subsidence of the caldera floor below
the water table (currently atf 490 m depth; Hurwitz et
al., 2003). Since the early 1800s, the caldera floor has
fluctuated in elevation by hundreds of meters and has
come within less thanf 100 m of the water table on at
least three occasions (Mastin, 1997).
1.1. Explosivity
Subsidence of the eruptive vent below the water
table would allow water to accumulate within the
crater and would guarantee water-magma mixing dur-
ing summit eruptions; It would not, however, guaran-
tee that such eruptions be large and explosive. Along
Kılauea’s coast, lava commonly enters water without
extensive fragmentation or dispersal of tephra (Mattox
and Mangan, 1997). Non-violent effusion of lava into
water was also observed at Mauna Loa in 1950 (Finch
and Macdonald, 1953), at Soufriere of St. Vincent in
1972 (Shepherd and Sigurdsson, 1982) and Kick-’em
Jenny in the West Indies in 1974–1978 (Devine and
Sigurdsson, 1995), to name a few examples.
We surmise that violence (or non-violence) of these
and other hydromagmatic eruptions is affected not
only by the injection of magma into water but also the
circumstances under which they mixed. At least, two
mechanisms for explosive magma-water mixing have
been described during central-vent eruptions: (1)
drawdown of magma in a conduit below the water
table, followed by influx of water, and (2) jetting of
magma through surface water or through a wet crater
into which water is entering through the porous walls.
Mechanism (1) was noted at Ukinrek Maars, Alaska,
when a hydromagmatic explosion, observed from the
air on 3 April, 1977, was preceded a few minutes
earlier by the draining of a lava lake and collapse of
the crater walls (Kienle et al., 1980). The sequence
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–31 17
was similar to well-described non-juvenile steam
explosions at Halema’uma’u Crater in 1924, which
took place after a long-lived lava lake drained and hot
conduit walls collapsed (Stearns, 1925; Decker and
Christiansen, 1984).
The jetting of magma through water was described
during continuous-uprush phases of Surtsey Volcano,
Iceland (Thorarinsson, 1964; Moore, 1985). These
eruptions were characterized by cylindrical columns
100–250 m in diameter at their base (Moore, 1985),
that blasted fragmental debris upward at velocities of
f 100 m/s and persisted for hours. They occurred
when seawater access to the vent was ‘‘partially or
wholly blocked2’’ (Thorarinsson, 1964), and were
generally preceded by intermittent hydromagmatic
explosions that repeated with increasing frequency
until jetting became continuous. (The cause of inter-
mittent explosions has been the subject of disagree-
ment (Kokelaar, 1983) and may be a variant on
mechanism (1)). Water was apparently supplied to
the jet by seepage over and through the tephra pile
that lined the vent crater. Continuous-uprush phases
produced much more tephra and dispersed it more
widely than Surtsey’s intermittent explosions (Thor-
arinsson, 1964, p. 44).
In principle, it should be possible to distinguish
mechanisms (1) and (2) using vesiculation textures
and dissolved-volatile contents of juvenile clasts.
Under mechanism (1), if magma existed within a
subaerial lava lake prior to the explosions, it should
contain dissolved volatiles in equilibrium with 1 atm
pressure. Magma quenched during withdrawal should
contain low vesicularity and low bubble-number den-
sities that typify low ascent rates or stagnation (Cash-
man and Mangan, 1994). Under mechanism (2),
juvenile clasts may contain dissolved volatiles equal
to or exceeding equilibrium values at 1 atmosphere
depending on the pressure and degree of equilibration
of the clasts at the time of quenching. Magma that
quenched during rapid ascent should exhibit abun-
dant, fine vesicles with high bubble-number densities
similar to lava-fountain tephras in subaerial environ-
ments (Cashman and Mangan, 1994; Mangan and
Cashman, 1996).
2 Water could not truly have been ‘‘wholly’’ blocked as
Thorarinsson describes, since he also characterizes these events as
hydromagmatic.
In this study, we examine vesiculation textures and
dissolved-volatile concentrations in the lower unit of
the Keanakako’i Ash to constrain magma-water con-
ditions at the time of quenching. The Keanakako’i
Ash is especially suited to this analysis because the
solubility and pre-eruptive volatile content of
Kılauean basalt are well constrained (e.g. Gerlach,
1986; Dixon et al., 1991, 1995; Wallace and Ander-
son, 1998) and vesiculation textures have been exten-
sively characterized (Mangan et al., 1993; Cashman et
al., 1994; Mangan and Cashman, 1996).
2. The Keanakako’i deposit
The Keanakako’i Ash consists of three major units
and several minor ones (Decker and Christiansen,
1984; McPhie et al., 1990). The major units include
(1) well-bedded ash and lapilli of inferred hydro-
magmatic origin containing primarily fresh, juvenile
sideromelane glass (the ‘‘lower, juvenile-rich beds’’ of
McPhie et al., 1990); (2) massive and cross-bedded
ash and lapilli, also of inferred hydromagmatic origin,
containing both juvenile glass and lithic fragments
(the ‘‘middle, mixed beds’’); and (3) block fall and
cross-bedded surge deposits composed almost exclu-
sively of lithic debris (the ‘‘upper lithic beds’’). These
units are named II, III and V, respectively, in the
nomenclature of Decker and Christiansen and 0–4,
6–10, and 11–16 in the nomenclature of McPhie et
al. (1990). The units are underlain by a massive
reticulite bed (unit I of Decker and Christiansen),
which in turn is locally underlain by a gray vitric-
lithic ash (D. Swanson, written comm. 2002). The
three units are overlain on the northwest and south-
west sides of the caldera by a pumice layer a few tens
of centimeters thick (unit VI of Decker and Christian-
sen; the ‘‘Golden Pumice’’ of Sharp et al., 1987).
Southwest of the caldera, along the contact between
units III and V of Decker and Christiansen are
scattered mats of Pele’s hair and tears (unit IV of
Decker and Christiansen), inferred to have erupted
from a fissure whose lavas are intercalated with the
Keanakako’i Ash (unit 1790f of Neal and Lockwood,
in press). In addition, at the base of the middle, mixed
beds (according to McPhie et al., 1990), or in the
upper part of unit II (according to Decker and Chris-
tiansen, 1984), is a scoria bed up to a few tens of
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–3118
centimeters thick (unit 6 of McPhie et al.) whose axis
of dispersal extends to the southeast, in contrast to the
southwestward dispersal of most units in the Keana-
kako’i deposit.
2.1. Evidence for depositional breaks
In one of the first written discussions of the
deposit, Dana (1888) assumed it resulted from a single
eruption in 1790 AD that was known from oral
accounts (Ellis, 1827; Dibble, 1843). Early 20th
century authors attributed only the upper lithic unit
to the eruption in 1790 and lower units to earlier
eruptions (Hitchcock, 1909; Powers, 1916; Stone,
1926; Stearns and Clark, 1930; Wentworth, 1938;
Finch, 1947; Powers, 1948). Later, Christiansen
(1979) and Decker and Christiansen (1984) proposed
that all members between the basal reticulite and the
Golden Pumice erupted during a single sequence in or
around 1790 AD. McPhie et al. (1990) concurred but
noted three horizons of erosion and reworking that
suggested pauses of unknown and possibly long
duration. Most recently, Swanson et al. (1998) cited
organic horizons, water-erosion features, archaeolog-
ical artifacts and carbon dates as evidence for at least
few large eruptive sequences, beginning shortly after
appearance of the present-day caldera in the late 15th
century and ending around 1790 AD (D. Swanson,
written comm., 2001). Horizons that show the stron-
gest evidence for non-depositional periods lie imme-
diately above the basal reticulite, slightly below and
above unit 6 of McPhie et al., and at the base of the
upper lithic beds (McPhie et al., 1990).
2.2. Field locations
Our study concentrates on unit II of Decker and
Christiansen, using samples collected in Sand Wash
(locations 61 and 84, Fig. 1b) and at a fissure a
kilometer north of Sand Wash (location 62, Fig. 1b).
At these locations, unit II contains three subunits: two
coarsening-upward sequences3 of well-bedded ash
and lapilli (units IIA and IIB) and a sequence of fine
3 Unpublished grain-size analyses give mdf =� 0.64 to 3.5,
rf = 1.22–1.86 for ash in the lower half of these sequences and
mdf =� 0.38–0.62, rf = 1.51–2.10 for ash and lapilli in the upper
half.
ash beds4 with sparse lapilli horizons (unit IIC; Figs. 2
and 3). Conformable bedding within subunits IIA and
IIB suggests that each was deposited during a single
eruptive pulse or rapid series of pulses with no
significant pauses. The contacts at the top of units
IIA and IIB are sharp and truncate underlying beds at
a low angle (Fig. 3b), suggesting some centimeters of
erosion, perhaps by surges or wind, between deposi-
tional pulses. The upper contact of unit IIA lacks
recognizable evidence of water erosion, suggesting
the pause was short and without rain. The upper
contact of unit IIB exhibits scarce water-erosion
rivulets and locally overlies cross-bedded and appar-
ently wind-reworked debris from underlying units
(McPhie et al., 1990), suggesting a longer pause of
unknown duration.
McPhie et al. (1990, p. 351) speculated that
juvenile-rich beds resulted from repeated explosions
that began with vesiculation-driven magma ascent
and ended with drainback of degassed magma.
Mastin (1997), used quantitative vesicularity and
bubble-number density measurements to argue that
some horizons in units IIA2 and IIB2 involved
sustained, high magma discharge. Whether other
beds involved sustained high discharge was not
investigated.
3. Sample collection and analysis
We collected samples by trowel from the outcrop
(Tables 1 and 2) and measured the volume percent of
fresh glass, altered glass, olivine and lithic fragments
by point-counting grains mounted in thin sections
(f 400 grains per section; Table 3). We analyzed
dissolved-volatile concentrations using the electron
microprobe for sulfur and Fourier transform infrared
(FTIR) spectrometry for water and CO2. Juvenile
glass in the deposit ranges in vesicularity from 0%
to >80%, requiring us to prepare samples for FTIR
analysis using two methods: (1) slightly-to-moderate-
ly vesicular clasts, 1–2 mm in diameter were embed-
ded in a 2.5-cm diameter wafer of epoxy that was then
ground and polished, flipped, remounted on a glass
slide, then ground to a thickness of 70–165 Am, and
4 mdf =� 0.67–3.71, rf = 1.31–2.53 based on unpublished
grain-size analyses.
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–31 19
polished on the other side; (2) highly vesicular lapilli
were ground with a mortar and pestle and flakes f 1
mm in diameter were handpicked, mounted in epoxy
and doubly polished as described above.
3.1. Dissolved volatile analysis
Before removing the polished grain mounts from
their glass slides, each fragment was analyzed for
sulfur by electron microprobe using the JEOL JXA-
8900 Electron Probe Microanalyzer at the USGS
Menlo Park Laboratory. These samples were a subset
of more than 23 thin sections containing tephra frag-
ments analyzed for sulfur during 6 microprobe ses-
sions of 1–2 days each. For reference, we also
analyzed clasts from non-hydromagmatic units I, IV,
VI and the 1959 Kılauea Iki tephra (Table 1). Values
of S and K2O in Table 4 and the major-element data
(see supplementary data in the online version) repre-
sent averages of two to four analyses per clast (exact
numbers are listed online). During each session, we
used an acceleration voltage of 15 keV, a beam current
of 25 nA, a beam diameter of 10–15 Am, a barite
standard for sulfur, and a peak counting interval of 80
s. Results for elements other than sulfur, and standards
used, are provided in the online supplement. Of 343
analyses, 330 gave totals between 97% and 101%.
The precision of sulfur concentrations (ppm) is
estimated from counting statistics and reproducibility
of USNM basaltic glass standard VG-2 (Jarosewich et
al., 1979), which is similar in composition to the
analyzed glasses. The standard deviation in sulfur
values of VG-2 glass measured during individual
microprobe sessions is 57 ppm (the half-width of
the error bar in Fig. 2); however, mean VG-2 sulfur
analyses for a given session range from 1080 to 1350
ppm, suggesting some drift between sessions. Sulfur
calibration numbers were therefore adjusted so that
the average VG-2 analysis in each session equals
1340 ppm, the value obtained by Dixon et al. (1991).
After microprobe analysis, the grain mounts were
removed from their glass slides and bubble-and crystal-
free regions of glass were targeted for FTIR analysis
using apertures above and below the sample. Measure-
ments were performed using a Nicolet5 Magna 750
5 Use of trade names in this document does not imply
endorsement by the USGS.
spectrometer with an attached SpectraTechR Analyti-
cal-IR microscope that utilizes a liquid-N2-cooled
MCT-A detector. We collected 512 scans per analysis
on each sample and on backgrounds following each
sample analysis. The measured absorbance of infrared
radiation is directly proportional to the concentration of
H2O in the glass, adjusted for sample density and
thickness. Weight percent H2O was calculated accord-
ing to Eq. (15) of Ihinger et al. (1994) for the OH�
stretch at 3570 cm� 1, assuming a glass density of
2700F 100 g l� 1, an extinction coefficient (e) of
63F 3 l mol� 1 cm� 1 and sample thickness as mea-
sured by a MitutoyoR Digital Micrometer. Uncertainty
for e, sample thickness and density are all about 5%,
so that propagated errors are all between 10% and
15% relative.
We detected no molecular H2O (peak at 1630
cm� 1) or CO3�(1515 and 1430 cm� 1); we analyzed
two clasts in sample 517B twice, collecting 1024
scans during the second analysis, to verify the absence
of these peaks. The detection limit of dissolved
CO3�is f 50 ppm, suggesting that the melt degassed
to less than f 5 MPa (using solubility relations of
Dixon et al., 1995), or about 600 m depth (assuming a
pressure gradient of 25 MPa/km). The absence of a
molecular water peak is consistent with the experi-
mentally measured speciation of dissolved water in
basaltic glass at this concentration (Dixon et al.,
1995). Meteoric water adsorbed into the clasts at
magmatic temperature could also speciate of OH�,
but water absorbed at lower temperature would likely
remain in molecular form (P. Wallace and J. Dixon,
written comm., 2001). If water were absorbed into the
clasts during or after the eruption, we would expect it
to be absorbed over a range of temperatures and to
leave at least some water in molecular form. The
complete absence of molecular H2O in these clasts
therefore suggests to us that all dissolved water is
magmatic.
3.2. Vesicularity and bubble-number density
To obtain bubble-number densities, all of the
dense, hand-picked clasts analyzed by FTIR were
photographed in thin section under reflected light.
Printed photomicrographs were overlain with trans-
parencies and the outlines of the clasts, and of internal
bubbles, were traced in ink. The inked transparencies
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–3120
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–31 21
were then digitally scanned and the clast diameter,
area, and the number and size distribution of bubbles
were computed using the software Scion ImageR. Thenumber of bubbles per clast ranged from zero to
several dozen; too few to derive meaningful size-
distribution statistics but enough to estimate the or-
der-of-magnitude bubble-number density.
The highly vesicular clasts were destroyed when
ground by mortar and pestle and could not be ana-
lyzed for bubble-number density. Prior to grinding,
however, their vesicularity was measured by weighing
the samples in and out of water following the method
of Houghton and Wilson (1989). We calculated clast
density (q, kg/m3) using the formula q = qw(wa/
(wa�ww)), where qw is the density of water (1000
kg/m3), and wa and ww are the measured weights in air
and in water, respectively. The volume-fraction gas
(vg)—i.e. the fraction of the clast volume composed of
bubbles—was calculated using the formula vg = 1� q/qg, where qg is the density of basaltic glass, taken as
2700 kg/m3. We calculated the equivalent radius (r) of
each lapillus (assuming a spherical shape) from the
formula r = (3wa/(4pq))1/3. Densities are generally
repeatable to F 10%.
During the course of our analysis, we observed that
vesicularity varied systematically from one strati-
graphic unit to the next and from ash to lapilli-sized
clasts within a given unit. To quantify these variations,
we compare vg from 3000 clast density measurements,
published earlier (Mastin, 1997), with vg fromf 3000
ash-sized grains, estimated visually in thin section by
comparing the abundance of bubbles to diagrams of
modal percentage (Compton, 1985, Appendix 3).
Diagrams in Compton (1985) are given for percen-
Fig. 2. (left) Keanakako’i Ash, showing stratigraphic units exposed at San
center) total dissolved sulfur in glass from microprobe measurements; (right
right) components of these units from point counts (Table 3). To the left o
(1990) (left) and Decker and Christiansen (1984) (right). Tick marks at top
glass at saturation, calculated using a best-fit exponential curve through six
that contained no dissolved CO2 and were tested at pressures < 200 MPa
equilibrium concentration (0.09 wt.%) at p= 0.1013 MPa (e.g. Wallace an
best-fit curve has the form H2O= 0.295p.515 (r2 = 0.9998), where H2O
error = 0.003 wt.% H2O at p= 0.1013 MPa, 0.029 wt.% at 2.5 MPa). Th
Christiansen (1984): unit 6 of McPhie et al., which is thickest near Kean
correlate with unit IIC2 at Sand Wash. However, the isopach map of uni
Moreover, these units differ petrologically (IIC2 at Sand Wash contains
juvenile scoria). Scattered fragments of scoriaceous debris at the base of u
remnants of unit 6. Further refinements to this stratigraphy are underway. N
(1984) are shown in the stratigraphy of McPhie et al. because no units of
tages of 0.5, 1, 1.5, 3, 5, 7, 10, 15, 10, 25, 30, 35, 40,
45 and 50. We consider those intervals to indicate the
resolution of the estimation technique. Results of
these measurements, as well as microprobe data and
lapilli clast density, are posted online.
4. Results
Sulfur values in matrix glass from unit II (Fig. 2)
average 330 ppm with a standard deviation of 80
ppm—significantly above the 100–150 ppm typically
measured for Kılauean glass that has equilibrated to
atmospheric pressure (Swanson and Fabbi, 1973;
Mangan and Cashman, 1996; Thornber, 2001). Glass
in non-hydromagmatic units I, IV, VI and the Kılauea
Iki tephra, all contain dissolved sulfur values consis-
tent with equilibrium degassing to 1 atm pressure.
Dissolved sulfur in glass inclusions in olivine ranges
from about 250 to 1530 ppm, with most around
1000–1200 ppm (Fig. 2). Considering the middle of
this range to represent undegassed inclusions and 100
ppm to be totally degassed, most hydromagmatic
glass fragments are about 70–80% degassed in sulfur.
The concentration of dissolved water in juvenile
glass also measurably exceeds equilibrium saturation
under atmospheric conditions (Fig. 2). Sulfur and
water values (Table 4) are moderately correlated
(r2 = 0.69) supporting the view that the water is
magmatic in origin. Measured H2O concentrations
correspond to quench pressures (under equilibrium
H2O saturation) of f 0.3–2.3 MPa (Fig. 2), equiva-
lent to 10–100 m depth at lithostatic pressure (as-
suming a pressure gradient of 25 MPa/km), or f 25–
d Wash plus units I and VI of Decker and Christiansen (1984); (left
center) total dissolved water in matrix glass, measured by FTIR; (far
f the stratigraphic section are unit names according to McPhie et al.
of the H2O plot indicate the equilibrium dissolved water in MORB
experimental measurements from Dixon et al. (1995), using samples
. One extra data point was added to this data set: the well-known
d Anderson, 1998; Mangan et al., 1993; Cashman et al., 1994). The
is in weight percent and p is pressure in megapascals (standard
is section contains one revision to the stratigraphy of Decker and
akako’i Crater, was thought by Decker and Christiansen (1984) to
t 6 in McPhie et al. suggests that it does not extend this far west.
pumice and lithics, whereas unit 6 at Keanakako’i Crater contains
nit IIC1 at Sand Wash (hollow diamonds in the sulfur plot) may be
o unit names corresponding to unit IIC of Decker and Christiansen
McPhie et al. appear to correlate with them.
Fig. 3. Photos of the Keanakako’i Ash at Sand Wash (location 61, Fig. 1b) showing stratigraphy and locations of some of the samples collected
there. (a) Upper part of section and (b) lower part of section. Three-digit numerals on photos indicate sample numbers. Stratigraphic names of
Decker and Christiansen are given in white lettering on the left side of photos; names from McPhie et al. are on the right side.
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–3122
250 m at hydrostatic pressure (100 MPa/km). The
amount of H2O lost to degassing can be assessed from
the ratio of H2O/K2O (Table 4, Fig. 4). Both of these
components are incompatible in a crystallizing
Kılauean melt and, in the absence of degassing,
their ratio should remain constant at about 1.3 (Dixon
et al., 1991; Wallace and Anderson, 1998). Average
K2O (0.44F 0.05 wt.%) in clasts from hydromag-
matic units imply pre-eruptive water content of
0.57F 0.06 wt.%. Measured H2O/K2O ratios (0.39–
1.02) suggest that the melt had lost 25–80% of its
H2O by the time it quenched.
Fig. 5 shows an inverse relationship between
dissolved water and vg, which is consistent with a
Table 2
Sample locations shown in Fig. 1
Number Easting Northing
1 259221 2146965
21 261979 2146795
61 259261 2145550
62 259240 2146443
67 259715 2148958
84 259237 2145737
Easting and Northing are given in UTM Zone 5 coordinates.
Table 1
Summary of samples collected and analyzed for this paper
Sample Unit Location Type of
numberD and C McPhie
number analysis
426 Iki Iki 21 mp
3HK215Q VI 1 mp
602 IV 84 mp
423 IIC4 62 mp, pc
525 IIC3 61 d, ves
422 IIC3 62 mp, pc
407b IIC3 61 pc
523 IIC3 61 d, ves
521 IIC3 61 d, ves
421 IIC2 62 mp
407a IIC2 61 mp, pc
520 IIC2 61 mp, FTIR, bnd, d
519 IIC1 61 d, ves
518 IIC1 61 ves
420 IIC1 62 pc
517 IIC1 61 mp, FTIR, bnd
516 IIC1 61 d, ves
419 IIC1 62 pc
514 6? 61 mp, d
513 IIB2 4 61 d, ves
418 IIB2 4 62 d
417 IIB2 4 62 mp, pc
512 IIB2 4 61 d, ves
406 IIB2 4 61 d
405 IIB2 4 61 mp, pc
511 IIB2 4 61 mp, FTIR, bnd, d, ves
416 IIB1 3 62 pc
509 IIB1 3 61 d
508 IIB1 3 61 mp, FTIR, bnd
403 IIB1 3 61 mp, pc
503 IIB1 3 61 mp, FTIR, bnd
413 IIA2 2 62 mp, pc
414 IIA2 2 62 d
502 IIA2 2 61 mp, FTIR, bnd, d
402 IIA2 2 61 d
412 IIA2 2 62 mp, d
401 IIA2 2 61 mp
501 IIA2 2 61 d
411 IIA1 1 62 mp, pc
526 IIA1 1 61 ves
409 IIA1 1 62 mp, FTIR, bnd
404 IIA1 1 61 pc
432 I 67 mp, d
Abbreviations in the right column include clast density measure-
ments (‘‘d’’), vesicularity measurements from thin section (‘‘ves’’),
microprobe glass analyses (‘‘mp’’), FTIR analyses (‘‘FTIR’’),
bubble-number density (‘‘bnd’’) and point counts (‘‘pc’’).
Table 3
Summary of point count results
Sample Unit Components
D and C McPhie Altered
glass
Fresh
glass
Lithic Crystal
n
423 IIC4 17% 63% 12% 8% 400a
422 IIC3 14% 80% 0% 6% 400
407b, 1f IIC3 18% 76% 4% 2% 400
407a, 1f IIC2 49% 43% 6% 2% 400
420 IIC1
(top)
11% 86% 0% 3% 401
419 IIC1
(base)
73% 24% 0% 3% 400
417 IIB2
(top)
4 6% 78% 15% 2% 400
405, 1f IIB2 4 11% 81% 5% 3% 390
416 IIB1
(top)
3 9% 91% 0% 1% 400
403, 1f IIB1 3 3% 96% 1% 0% 404
413 IIA2
(top)
2 8% 73% 9% 10% 400
411 IIA1
(top)
1 12% 87% 1% 1% 400
404, 1f IIA1 1 7% 86% 7% 1% 400
a The standard error for point counts of 400 grains is less than
4% (Van der Plas and Tobi, 1965).
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–31 23
scenario in which clasts were quenched during active
vesiculation. Dissolved H2O is lowest in the highly
vesicular clasts of unit IIB2. Nearly all data lie
below the line that represents closed-system gas
exsolution with total water (dissolved plus exsolved)
equal to 0.55 wt.%. Large, highly vesicular lapilli
appear to have suffered more gas loss than smaller,
less vesicular ash.
These data clearly show that hydromagmatic glass
did not fully degas in sulfur or water. In contrast, non-
hydromagmatic clasts are completely degassed in
sulfur. We did not measure dissolved water in non-
hydromagmatic tephra, but tephra from phase 1 of the
Kılauea Iki eruption, analyzed by Wallace and Ander-
son (1998, fig. 1), are in equilibrium with 1 atm
pressure (phase 1 tephra erupted before degassed lava
Table 4
Summary of FTIR and bubble-number density measurements
Sample Unit Thickness
(Am)
vg Diameter
(mm)
Clast area
(mm2)
Number of
bubbles
Number density
(cm� 3)
S
(wt.%)
H2O
(wt.%)
K2O H2O/K2O
520b IIC2 136 0.35 1.79 0.575 153 5.8E + 05 0.023 0.33 0.4585 0.7197
126 0.35 1.62 0.435 119 9.5E + 05 0.033 0.29 0.4427 0.6551
145 0.43 1.13 0.435 135 1.0E + 07 0.030 0.38 0.4383 0.8669
140 0.49 1.46 0.920 256 7.9E + 06 0.037 0.28 0.4387 0.6383
134 0.46 1.12 0.804 201 8.6E + 06 0.042 0.38 0.4400 0.8636
517b IIC1 152 0.39 1.63 1.536 161 2.1E + 06 0.042 0.37 0.4293 0.8618
132 0.42 1.65 1.755 136 1.0E + 06 0.042 0.33 0.4540 0.7269
125 0.44 1.21 1.027 53 1.1E + 06 0.033 0.36 0.4447 0.8096
136 0.43 1.4 1.157 69 1.0E + 06 0.042 0.45 0.4450 1.0112
128 0.42 1.35 1.027 105 1.9E + 06 0.035 0.29 0.4470 0.6488
140 0.42 1.31 0.934 55 8.3E + 05 0.034 0.34 0.4390 0.7745
147 0.44 1.23 0.048 0.41 0.4398 0.9323
511e IIB2 94 0.671 5.29 0.017 0.25 0.4002 0.6247
93 0.643 4.61 0.018 0.25 0.3958 0.6317
92 0.643 3.73 0.022 0.19 0.3903 0.4868
83 0.51 3.86 0.021 0.23 0.4052 0.5676
74 0.51 3.32 0.021 0.2 0.4053 0.4935
75 0.70 5.02 0.020 0.23 0.3977 0.5783
75 0.31 3.26 0.031 0.29 0.4103 0.7068
74 0.51 3.32 0.024 0.3 0.4021 0.7461
68 0.86 5.71 0.034 0.34 0.4165 0.8163
76 0.59 5.48 0.024 0.24 0.4168 0.5758
71 0.75 4.28 0.017 0.16 0.4101 0.3901
73 0.41 3.20 0.028 0.29 0.4009 0.7234
508f IIB1 93 0.67 0.49 0.1573 5 3.9E + 05 0.039 0.31 0.4273 0.7254
101 0.21 0.52 0.1591 3 9.5E + 04 0.034 0.33 0.4383 0.7529
90 0.66 0.56 0.1188 18 3.4E + 06 0.040 0.38 0.4283 0.8872
100 0.50 0.81 0.3902 13 3.2E + 05 0.041 0.32 0.4220 0.7583
90 0.32 0.66 0.2581 23 1.1E + 06 0.032 0.34 0.4227 0.8044
83 0.21 0.47 0.032 0.26 0.3950 0.6582
104 0.39 0.74 0.3237 18 7.2E + 05 0.027 0.36 0.4097 0.8788
100 0.52 0.41 0.048 0.33 0.4055 0.8138
502h IIA2 136 0.35 1.66 1.0158 200 5.5E + 06 0.041 0.3 0.4523 0.6632
143 0.44 1.26 1.0158 94 1.6E + 06 0.032 0.35 0.4270 0.8197
147 0.14 1.48 0.8732 79 1.1E + 06 0.027 0.42 0.4130 1.0169
148 0.24 1.04 0.6620 19 2.1E + 05 0.037 0.42 0.4373 0.9604
147 0.26 1.10 1.4885 36 1.4E + 05 0.047 0.4 0.4303 0.9295
409b IIA1 157 0.49 1.31 0.4768 35 9.2E + 05 0.041 0.34 0.4200 0.8095
167 0.31 1.71 1.2399 100 1.2E + 06 0.040 0.37 0.4303 0.8598
148 0.33 1.28 0.6712 139 4.5E + 06 0.041 0.37 0.4197 0.8817
152 0.46 1.16 0.5397 15 3.1E + 05 0.041 0.38 0.4050 0.9383
163 0.27 1.47 0.8240 30 2.8E + 05 0.035 0.4 0.4137 0.9670
Unit names are those of Decker and Christiansen (1984). Clast diameter (D) is calculated as D ¼ 2ffiffiffiffiffiffiffiffiA=p
p, where A is clast area. Clasts in sample
511e were prepared by grinding lapilli in a mortar and pestle and handpicking fragments for analysis. Other clasts were prepared by handpicking
small clasts from grab samples. Because of the difference in sample preparation, no number densities were measured for clasts in sample 511e.
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–3124
from lava ponds entered the vent). We therefore infer
that water, like sulfur, is elevated due to rapid quench-
ing by external water. Whether these clasts quenched
at elevated pressure or at 1 atm before they could
degas to equilibrium is the subject of later discussion.
4.1. Vesiculation textures
In general, pyroclasts of a given size are less
vesicular in units IIA1, IIB1, more vesicular in units
IIA2 and IIB2. Within individual units, average vesic-
Fig. 4. Dissolved H2O versus K2O in matrix glass from
Keanakako’i Ash. Melt that has not lost H2O to degassing plots
along the line H2O/K2O= 1.3, while completely degassed melt
would plot along a line having zero slope and an intercept at
H2O= 0.09 wt.%. Intermediate degrees of degassing are represented
by lines having intermediate slopes, intersecting the H2O/K2O= 1.3
line at H2O= 0.09 wt.%.
Fig. 5. Weight percent dissolved water versus volume fraction gas in
melt. Squares are measurements (symbol size is proportional to clast
diameter). Lines give volume-fraction gas versus dissolved water
for closed-system degassing for a total water content (dissolved plus
exsolved) of (top to bottom) 0.55, 0.44, 0.33 and 0.22 wt.%,
respectively. The lines represent loss of 0%, 20%, 40% and 60% of
the original gas, respectively.
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–31 25
ularity increases with average grain size (Fig. 6).
Although vg was estimated for the ash-and lapilli-sized
clasts using two different methods, we do not think
that sample bias in one or both methods is responsible
for the variation. Measurements made using both
methods in the same size range (for example, in unit
IIB2, 2–3 mm diameter clasts) yield the same range in
vg. Moreover, estimates by visual inspection are sim-
ilar to results from the more accurate image processing
technique (triangles, units IIA2, IIB1 and IIC1).
For clast diameters larger than a few tenths of a
millimeter, lower values of vg do not appear to result
simply from fragmentation at a scale smaller than the
average bubble diameter. Many clasts of this size (e.g.
Fig. 7) contain bubbles smaller than their own diam-
eter. The lower average dissolved water concentration
of the larger, more vesicular clasts (Fig. 5) and the
incomplete degassing of all hydromagmatic clasts
suggest that the smaller clasts were at an earlier stage
of vesiculation.
Bubble-number densities in unit II (Table 4, Fig. 8)
range from 9� 104 to 1�107 cm� 3 similar to those
measured in unit IIB2 by Mastin (1997). Though the
earlier data were obtained only from highly vesicular
lapilli, the new data come from ash-sized clasts with
vesicularity ranging down to about 20%. All these
values exceed number densities obtained from vent
samples of effusive eruptions (3000–5000 cm� 3;
Mangan et al., 1993) and most exceed values mea-
sured in lava-fountain tephra from Pu’uO’o (19,000–
180,000 cm� 3; Mangan and Cashman, 1996).
Unlike our Keanakako’i clasts, the lava-fountain
tephra rose many seconds in a hot plume before
cooling, raising the question of whether bubble coa-
lescence and escape could account for their lower
number densities. We do not think so. Such processes
recognizably alter texture by increasing vesicularity in
clasts interiors and reducing it on margins (Walker and
Croasdale, 1972). Mangan and Cashman (1996)
avoided this complication by selecting clasts that were
texturally homogeneous. From plots of log bubble
number versus bubble diameter, they inferred that
only about 13–40% of bubbles coalesced from two
or more smaller ones—not enough to account for the
< 100� lower number densities of lava-fountain
clasts relative to Keanakako’i clasts.
5. Implications
These high bubble-number densities of unit II
suggest rapid magma ascent and eruptive activity
Fig. 7. Moderately vesicular ash from unit IIA2.
Fig. 6. Volume fraction gas (vg) versus clast diameter for ash and lapilli fragments collected from most juvenile-bearing units in the Keanakako’i
Ash. Dots represent vg estimated by visual comparison of clasts with modal area plots (e.g. Compton, 1985, pp. 366–367). Crosses are vg of
lapilli published earlier (Mastin, 1997). Triangles are obtained from digital image processing of photomicrographs.
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–3126
more akin to continuous-uprush events at Surtsey than
to drainback-induced explosions of Ukinrek or Hale-
ma’uma’u. Moreover, bubble-number data suggest
that all unit II subunits erupted rapidly, even those
containing only moderately or poorly vesicular tephra.
Whether all hydromagmatic phases during this period
were violent is less clear; less violent phases may have
failed to eject debris out of the caldera where it could
be preserved. The cause of repeated, high-flux rate
eruptive pulses is not known but could be related to
caldera subsidence.
High ascent rates could explain some but not all of
the fragmentation and dispersal that we associate with
Fig. 8. Bubble-number density of samples from the Keanakako’i
Ash (dots), from lava-fountain tephras (triangles; data from Mangan
and Cashman, 1996) and from effusive lavas (crosses; data from
Mangan et al., 1993). Inset is a photomicrograph of one clast (508f-
7), taken under reflected light, and the line-drawing interpretation of
the number of bubbles in this clast. The number of bubbles is 18,
and the clast area (excluding bubbles) is 0.00324 cm2, giving a
number density of ((18 bubbles)/0.00324 cm2))3/2 = 7� 105 cm� 3.
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–31 27
explosivity. Dry Kilauean high-flux eruptions (lava
fountains) certainly produce more fragmental debris
than low-flux eruptions, but the fragmentation process
in dry fountaining is primarily gas expansion. In the
Keanakako’i Ash, many juvenile clasts are poorly
vesicular and blocky in shape, implying fragmentation
by thermal fracture rather than bubble expansion. We
surmise that thermal fractures develop on clast surfa-
ces when they contact water or stream. Fracture
spacing, which determines the final grain size, is
influenced by the temperature contrast at the interface.
Rapidly formed surfaces are hotter and likely to host
more finely-spaced thermal cracks than surfaces
formed by slow, taffy-like stretching. The glassy rinds
between these fractures probably shed as the interface
strains and deforms (Fig. 9a); the more rapid the
deformation, the more extensive and fine the shedding
of debris. Fragmentation by this mechanism should be
finest and most extensive, and heat-transfer rates from
magma greatest, when ascent rates are high and when
melt is violently torn and deformed as it enters a lake
or the atmosphere. Fine particles are likely to originate
on the clast margins, which cool before clast interiors
and should, as our results suggest, contain the highest
dissolved volatile concentration.
5.1. The importance of turbulence
We hypothesize that the degree of fragmentation is
controlled by the turbulence of mixing. Photographs
of continuous-uprush phases at Surtsey (Thorarins-
son, 1964) show fully turbulent jets with exit veloc-
ities exceeding 100 m/s (Moore, 1985). Turbulence
occurs where perturbations along the jet margins are
not damped by viscous forces and develop into
eddies and vortices (White, 1991, p. 471). In circular
jets, the jet margin becomes unstable at Reynolds
numbers (Re) as low as 4 (Kaplan, 1964), though full
turbulence at the vent is not ensured unless flow is
turbulent in the upper conduit (i.e. Re>f 2300,
where Reu quD/l, q = fluid density, u = velocity,
D = conduit diameter and l is fluid viscosity. During
basaltic lava-fountain eruptions, flow in the upper
conduit may be either laminar or turbulent (Fig. 10),
with laminar or unstable flow during less vigorous
events and full turbulence during more vigorous
ones.
Engineering tests have long established that jet
turbulence increases rates of heat transfer and chem-
ical reaction between mixing fluids by orders of
magnitude (e.g. Burmeister, 1983, p. 394). Higher
rates of turbulent shear cause more efficient mixing
as eddies progress to ever smaller scales. The finest-
scale vortices dampen out at the so called Kolmogorov
length scale (ju l3d/q3U3)1/4, where d is boundary-
layer thickness and U is mean velocity, Kolmogorov,
1941; White, 1991, p. 471). For magma–steam–water
mixtures moving at tens of meters per second (based
on observed lava-fountains; Mangan and Cashman,
1996) within a boundary layer a meter or two wide, the
Kolmogorov scale would be decimeters or smaller
depending on whether the viscosity and density of
gas, water or magma are used in the calculation. The
scale of mixing is fully sufficient to incorporate
centimeter-to decimeter-sized volumes of water into
melt as required to generate molten fuel-coolant inter-
actions (Buettner and Zimanowski, 1998).
5.2. Water depth and volatile equilibration
Turbulent entrainment relations can be used to
constrain water depth from the deposit characteristics.
Immediately above the jet exit (‘‘A’’, Fig. 9b), jets
entrain ambient fluid within a turbulent boundary layer
Fig. 9. (a) Schematic illustration of the fragmentation sequence of a magma blob entrained in turbulent flow. Fragmentation is assumed to occur by extension and shear of the droplet,
supplemented by growth of cooling fractures and shedding of the glassy rind on clast margins. (b) Velocity profiles and boundary layers around an axisymmetric turbulent jet exiting
into ambient fluid.
L.G.Mastin
etal./JournalofVolca
nologyandGeotherm
alResea
rch137(2004)15–31
28
Fig. 10. Plot of magma viscosity versus mass flux (m), showing the
threshold for fully turbulent flow in a basaltic conduit. For a given
conduit diameter, diagonal lines separate regions of fully turbulent
flow (right) from laminar or unstable flow (left). The two dashed
vertical lines bound the range of m produced by lava-fountains at
Pu’u O’o (from volume-flow rates in Parfitt et al., 1995 assuming a
bubble-free magma density of 2500 kg/m3). Reynolds numbers are
calculated by substituting the equation for mass flux (m= pquD2/4)
into the Reynolds equation (Re= quD/l) to obtain Re= 4m/(pDl).Long tick marks give basalt viscosity at specified temperatures using
relations from Shaw (1972). The horizontal dotted line represents the
viscosity at temperatures of the Pu’u O’o melts (Thornber, 2001);
bubbles could increase viscosity up to a few times; Pal, 2003).
Holocene Kılauean basalt temperatures have ranged from about 1100
to 1300 jC (e.g. Clague et al., 1995b). Numerical calculations using
Conflow, a publicly available program (Mastin, 2002), indicate the
mass flow rates at Pu’uO’o could have been delivered through a
conduit a few to several meters in diameter.
6 Magma injection into water involves phase changes and
viscosity contrasts that may result different entrainment coefficient
than the a= 0.08, which is derived from jets of similar fluids.
Therefore, the implications of these relations are only presented
semi-quantitatively here. Experimental studies using magma and
water (or analogue fluids) are required for more quantitative
analysis.
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–31 29
whose thickness increases linearly until, at a height (h)
of several times the vent diameter (D), it has penetrated
into the potential core of the jet (‘‘B’’, Fig. 9b). Further
downstream (‘‘C’’, Fig. 9b), the jet diameter (d)
increases linearly and centerline velocity decreases
inversely with distance. Jet momentum remains con-
stant with distance but total mass flux increases as the
jet continues to entrain sorrounding fluid.
In experiments using jet and ambient fluids of
similar properties (e.g. water or saline solution
injected into water), the rate of entrainment follows
the relation (Eq. (6) of Wilson et al., 1995):
Mi ¼ pDhqaauj
ffiffiffiffiffiqj
qa
r
where Mi is the mass per unit time of ambient fluid
added to the jet between the jet exit and height h; D, qj
and uj are the average jet diameter, density and
velocity between the exit and h; qa is density of the
ambient fluid; and a is an empirical entrainment
coefficient, which is roughly 0.08 in the self-similar
region (‘‘C’’, Fig. 9b) and somewhat less in the initial
region (‘‘A’’).6 The mass influx calculated by this
method may be compared with the mass flux of the
jet, Mj = pD2qjuj/4.For a jet of magma and gas exiting into water,
realistic values for D (f 5 m, based on conduit
modeling of observed flow rates; Mastin, 2002), qa
(1000 kg/m3), qj (500–1500 kg/m3 assuming a mod-
erately to highly vesicular magma), uj (50–100 m/s
based on observed lava-fountain velocities; Mangan
and Cashman, 1996) and a (0.06–0.08), suggest that
the mass flux of water entrained will equal the mass
flux of the magma–gas mixture when water depth is a
few to several vent diameters. A magma:water ratio of
1:1 is much less than the optimal ratio (f 3–5.5:1)
for converting thermal to kinetic energy (Wohletz,
1986), and would leave most (60–80%) of the water
in liquid from (Mastin, 1995), very likely causing
deposits to remobilize into watery slurries. Optimal
magma:water ratios would likely require water depths
on the order of a single vent diameter, perhaps less.
Significantly, at such shallow depths, water would be
entrained only into the boundary layer and would mix
with magma in the core of the jet only after escaping
into the atmosphere. This inference is consistent with
observations from Surtsey, in which the margins of
continuous-uprush jets contained black ash and white
steam, but the jet core glowed a visible red color at
night (Thorarinsson, 1964; Moore, 1985).
Although accretionary lapilli are common in fine-
grained beds of unit II, coarse-grained beds of units
IIA2 and IIB2 contain none, and there is no evidence
for soft-sediment deformation or other remobilization.
We therefore suspect that that these coarse-grained
beds erupted through water depths on the order of a
vent diameter. Their elevated volatile concentrations
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–3130
therefore probably reflect disequilibrum rather than
elevated quenching at pressure in a deep lake. Could
the disequilibrium have resulted from high ascent
rates alone? We suspect not, as non-hydromagmatic
lava-fountain tephras (Fig. 2) appear to have equili-
brated before cooling. High volatile concentrations in
these glasses reflect both rapid ascent and rapid
quenching by water at the surface. In this sense, these
deposits provide a freeze-fame of degassing at the
time of eruption that is unattainable in non-hydro-
magmatic tephra.
Acknowledgements
This paper has significantly benefited from dis-
cussions with Don Swanson, Richard Fiske, Tim
Rose, David Clague, Paul Wallace and Jacqueline
Dixon. Pete Friedman provided a helpful review of
the discussion of jet entrainment. We also wish to
acknowledge the assistance of Robert Oscarson for
microprobe assistance; Margaret Mangan and Kathy
Cashman for advice on collecting and analyzing
vesicle data; and Marvin Couchman for advice on
collecting clast density measurements.
References
Buettner, R., Zimanowski, B., 1998. Physics of thermohydraulic
explosions. Physical Review Letters 57 (5), 5726–5729.
Burmeister, L.C., 1983. Convective Heat Transfer. John Wiley &
Sons, New York, 789 pp.
Cashman, K.V., Mangan, M.T., 1994. Physical aspects of magmatic
degassing II. In: Caroll, M.R., Holloway, J.R. (Eds.), Volatiles in
Magmas. Reviews in Mineralogy. Mineralogical Society of
America, Washington, D.C., pp. 447–478.
Cashman, K.V., Mangan, M.T., Newman, S., 1994. Surface degass-
ing and modifications to vesicle size distributions in active ba-
salt flows. Journal of Volcanology and Geothermal Research 61,
45–68.
Christiansen, R.L., 1979. Explosive eruption of Kılauea Volcano in
1790. In: Decker, R.W., Drake, C., Eaton, G., Helsley, C. (Eds.),
Hawai’i Symposium on Intraplate Volcanism and Submarine
Volcanism. U.S. Geological Survey, Hawaiian Volcano Obser-
vatory, Hawaii National Park, HI, p. 158 (abs.).
Clague, D.A., Beeson, M.H., Denlinger, R.P., Mastin, L.G., 1995a.
Ancient ash deposits and calderas at Kılauea. Eos 76 (46), F666
(abstr.).
Clague, D.A., Moore, J.G., Dixon, J.E., Friesen, W.B., 1995b. Pe-
trology of submarine lavas from Kılauea’s Puna Ridge, Hawaii.
Journal of Petrology 36 (2), 299–349.
Compton, R.L., 1985. Geology in the Field. Wiley, New York,
398 pp.
Dana, J.D., 1888. History of changes of the Mt. Loa craters. Amer-
ican Journal of Science 36 (212), 81–112 (ser. 3).
Decker, R.W., Christiansen, R.L., 1984. Explosive eruptions of
Kılauea Volcano, Hawaii. In: National Research Council, et al.,
(Eds.), Explosive Volcanism: Inception, Evolution, and Hazards.
National Academy Press, Washington, DC, pp. 122–132.
Devine, J., Sigurdsson, H., 1995. Petrology and eruption styles of
Kick ’em-Jenny submarine volcano, Lesser Antilles island arc.
Journal of Volcanology and Geothermal Research 69, 35–58.
Dibble, S., 1843. A history of the Sandwich Islands. T.G. Thrum,
Lahainaluna, Hawaii, 451 pp.
Dixon, J.E., Clague, D.A., Stolper, E.M., 1991. Degassing history
of water, sulfur, and carbon in submarine lavas from Kılauea
volcano, Hawaii, 1956–1983. Journal of Geology 99, 371–394.
Dixon, J.E., Stolper, E.M., Holloway, J.R., 1995. An experimental
study of H2O and carbon dioxide solubilities in mid-ocean ridge
basaltic liquids: Part I. Calibration and solubility results. Journal
of Petrology 36, 1607–1631.
Dzurisin, D., Lockwood, J.P., Casadevall, T.J., Rubin, M., 1995.
The Uwekahuna Ash member of the Puna Basalt: product of
violent phreatomagmatic eruptions at Kılauea Volcano, Hawaii,
between 2800 and 2100 years ago. Journal of Volcanology and
Geothermal Research 66, 163–184.
Easton,M., 1987. Stratigraphy of Kılauea Volcano. In: Decker, R.W.,
Wright, T.L., Stauffer, P.H. (Eds.), Volcanism in Hawaii. U.S.
Geological Survey Professional Paper, 1350, 243–260.
Ellis, W., 1827. Journal of William Ellis. Advertiser Publishing,
Honolulu, HI, 342 pp.
Finch, R.H., 1947. Kılauea in 1790 and 1823. The Volcano
Letter, (496), 1–2.
Finch, R.H., Macdonald, G.A., 1953. Hawaiian volcanoes during
1950. U.S. Geological Survey Bulletin 996-B, 88 pp.
Gerlach, T.M., 1986. Exsolution of H2O, CO2, and S during erup-
tive episodes at Kılauea Volcano, Hawaii. Journal of Geophys-
ical Research 91 (B12), 12177–12185.
Hitchcock, C.H., 1909. Hawaii and its Volcanoes. Hawaiian Ga-
zette, Honolulu, 314 pp.
Houghton, B.F., Wilson, C.J.N., 1989. A vesicularity index for
pyroclastic deposits. Bulletin of Volcanology 51 (6), 451–462.
Hurwitz, S., Goff, F., Janki, C.J., Evans, W.C., Counce, D.A.,
Sorey, M.L., Ingebritsen, S.E., 2003. Mixing of magmatic vol-
atiles with groundwater and interaction with basalt on the
summit of Kılauea Volcano, Hawaii. Journal of Geophysical
Research 108 (B1), 2028 (doi: 10.1029/2001JB001594).
Ihinger, P.D., Hervig, R.L., McMillan, P.F., 1994. Analytical meth-
ods for volatiles in glasses. In: Carroll, M.R., Holloway, J.R.
(Eds.), Volatiles in Magmas. Reviews in Mineralogy. Mineral-
ogical Society of America, Washington, D.C., pp. 67–122.
Jarosewich, E., Nelen, J.A., Norberg, J.A., 1979. Electron micro-
probe reference samples for mineral analyses. Smithsonian Con-
tributions to the Earth Sciences 22, 68–72.
Kaplan, R.E., 1964. The stability of laminar incompressible bound-
ary layers in the presence of compliant boundaries. Aeroelastic
and Structures Research Lab Report. ASRL-TR 166-1, MIT,
Cambridge, MA.
L.G. Mastin et al. / Journal of Volcanology and Geothermal Research 137 (2004) 15–31 31
Kienle, J., Kyle, P.R., Self, S., Motyka, R.J., Lorenz, V., 1980.
Ukinrek Maars, Alaska: I. April 1977 eruption sequence, petrol-
ogy and tectonic setting. Journal of Volcanology and Geother-
mal Research 7, 11–37.
Kokelaar, B.P., 1983. The mechanism of Surtseyan volcanism.
Journal of the Geological Society (London) 140, 939–944.
Kolmogorov, A.N., 1941. The local structure of turbulence in in-
compressible viscous fluid for very large Reynolds numbers.
Doklady Akademii Nauk SSSR 30, 299–303.
Mangan, M.T., Cashman, K.V., 1996. The structure of basaltic
scoria and reticulite and inferences for vesiculation, foam for-
mation, and fragmentation in lava fountains. Journal of Volca-
nology and Geothermal Research 73 (1–2), 1–18.
Mangan, M.T., Cashman, K.V., Newman, S., 1993. Vesiculation of
basaltic magma during eruption. Geology 21, 157–160.
Mastin, L.G., 1995. Thermodynamics of gas and steam-blast erup-
tions. Bulletin of Volcanology 57 (2), 85–98.
Mastin, L.G., 1997. Evidence for water influx from a caldera lake
during the explosive hydromagmatic eruption of 1790, Kılauea
Volcano, Hawaii. Journal of Geophysical Research 102 (B9),
20093–20109.
Mastin, L.G., 2002. Insights into volcanic conduit floor from an
open-source numerical model. Geochemistry, Geophysics, Geo-
systems 3 (7) (doi:10.1029/2001GC000192).
Mattox, T.N., Mangan, M.T., 1997. Littoral hydrovolcanic explo-
sions: a case study of lava-sea water interaction at Kılauea
Volcano. Journal of Volcanology and Geothermal Research
75, 1–17.
McPhie, J., Walker, G.P.L., Christiansen, R.L., 1990. Phreatomag-
matic and phreatic fall and surge deposits from explosions at
Kılauea Volcano, Hawaii, 1790 A.D.: Keanakako’i Ash mem-
ber. Bulletin of Volcanology 52, 334–354.
Moore, J.G., 1985. Structure and eruptive mechanisms at Surtsey
Volcano, Iceland. Geological Magazine 6, 649–661.
Moore, J.G., Nakamura, K., Alcaraz, A., 1966. The 1965 eruption
of Taal Volcano. Science 151, 955–960.
Neal, C.A., Lockwood, J.P., in press. Geologic Map of the summit
region of Kılauea Volcano, Hawai’i. USGS Map I-2542.
Pal, R., 2003. Rheological behavior of bubble-bearing magmas.
Earth and Planetary Science Letters 207, 165–179.
Parfitt, E.A., Wilson, L., Neal, C.A., 1995. Factors influencing the
height for Hawaiian lava fountains: implications for the use of
fountain height as an indicator of magma gas content. Bulletin
of Volcanology 57, 440–450.
Powers, S., 1916. Explosive ejectamenta of Kılauea. American
Journal of Science 41, 227–244.
Powers, H.A., 1948. Chronology of the explosive eruptions of
Kılauea. Pacific Science 2, 278–292.
Sharp, R.P., Dzurisin, D., Malin, M.C., 1987. An early 19th century
reticulite prime from Kılauea Volcano. In: Decker, R.W.,
Wright, T.L., Stauffer, P.H. (Eds.), Volcanism in Hawaii. US
Geological Survey Professional Paper, 1350, 395–404.
Shaw, H.R., 1972. Viscosities of magmatic silicate liquids: an Em-
pirical method of prediction. American Journal of Science 272,
870–893.
Shepherd, J.B., Sigurdsson, H., 1982. Mechanism for the 1979
explosive eruption of Soufriere Volcano, St. Vincent. Journal
of Volcanology and Geothermal Research 13, 119–130.
Stearns, H.T., 1925. The 1924 eruption of the Hawaiian volcano.
Scientific American 132, 242–243 (Apr.).
Stearns, H.T., Clark, W.O., 1930. Geology and water resources of
the Kau District, Hawaii. U.S. Geological Survey Water-Supply
Paper 616, 194.
Stone, J.B., 1926. The products and structure of Kılauea. Bernice P.
Bishop Museum Bulletin 33, 59 pp.
Swanson, D.A., Christiansen, R.L., 1973. Tragic base surge in 1790
at Kılauea Volcano. Geology 1, 83–86.
Swanson, D.A., Fabbi, B.P., 1973. Loss of volatiles during foun-
taining and flowage of basaltic lava at Kılauea Volcano,
Hawaii. Journal of Research of the U.S. Geological Survey 1
(6), 649–658.
Swanson, D.A., Fiske, R.S., Rose, T.R., Kenedi, C.L., 1998. Pro-
longed deposition of the Keanakako’i Ash member, Kılauea.
Eos 79 (45), F937 (abstr).
Thorarinsson, S., 1964. Surtsey, the New Island in the North At-
lantic. Viking Press, New York, 47 pp.
Thornber, C., 2001. Olivine– liquid relations of lava erupted by
Kılauea Volcano from 1994 to 1998: implications for shallow
magmatic processes associated with the ongoing East-Rift-Zone
eruption. The Canadian Mineralogist 39, 239–266.
Van der Plas, L., Tobi, A.C., 1965. A chart for judging the reliabil-
ity of point counting results. American Journal of Science 263,
87–90.
Walker, G.P.L., Croasdale, R., 1972. Characteristics of some basal-
tic pyoclasts. Bulletin Volcanologique 35 (2), 303–317.
Wallace, P.J., Anderson, A.T.J., 1998. Effects of eruption and lava
drainback on the H2O contents of basaltic magmas in Kılauea
Volcano. Bulletin of Volcanology 59 (5), 327–344.
Wentworth, C.K., 1938. Ash Formations of the Island of Hawaii.
Hawaiian Volcano Research Association, Honolulu, 183 pp.
White, F.M., 1991. Viscous Fluid Flow. McGraw Hill, New York,
614 pp.
Wilson, L., Partfitt, E.A., Head, J.W., 1995. Explosive volcanic
eruptions: VIII. The role of magma recycling in controlling
the behavior of Hawaiian-style lava fountains. Geophysical
Journal International 121, 215–225.
Wohletz, K.H., 1986. Explosive magma–water interactions: ther-
modynamics, explosion mechanisms, and field studies. Bulletin
of Volcanology 48, 245–264.