HomeMy WebLinkAboutComm. No. 2019-12 John Olsen Testimony 8.6.19F Comm. No. 2019-12
•, Cycles of explbsive and effusive eruptions at Kilauea Volcano,, Hawaii
Donald A. Swanson', Timothy R. Rose2, Adonara E. Mucek3, Michael O. Garcia3, Richard S. Fiske2, and Larry G. Mastin4
'U.S. Geological Survey, Hawaiian Volcano Observatory, Hawaii National Park, Hawaii 96718, USA
'Department of Mineral Sciences, Museum of Natural History, Smithsonian Institution, Washington, D.C. 20013, USA
3Department of Geology and Geophysics, SOEST (School of Ocean and Earth Science and Technology), University of Hawai'i,
Honolulu, Hawai'i 96822, USA
'U.S. Geological Suivey, Cascades Volcano Observatory, Vancouver, Washington 98683, USA
ABSTRACT
The subaerial eruptive activity at Kilauea Volcano (Haivai`i) for
the past 2500 yr cAm be divided into 3 dominantly effusive and 2 domi-
nantly explosive periods, each lasting several centuries. The prevail-
ing; style of eruption for 60%, of this time was explosive, manifested
by repeated phreatic and phreatomagmatic activity in a deep summit
caldera. During dominantly explosive periods, the magma supply rate
to the shallow storage volume beneath the summit dropped to only a
few percent of that during mainly effusive periods. The frequency and
duration of cTplosive activity are contrary to the popular impression
that Kilauea is almost unceasingly effusive. Explosive activity appar-
ently correlates Kith the presence of a caldera intersecting the water
table. The'decrease in magma supply rate may result in caldera col-
lapse, because erupted or intruded magma is not replaced. Glasses
-with unusually high MgO, TiOs, and KBO compositions occur only in
explosive tephra (and one related lava flow) and are consistent with
disruption of the shallow reservoir complex during caldera formation.
Kilauea is a complex, modulated system in which melting rate, supply
rate, conduit stability (in both mande and crust), reservoir geometry,
water table, and many other factors interact with one another. The
hdaards associated with explosive activity at Klauea's summit would
have major impact on local society if a future dominantly explosive
period were to Iast several cenhn-ics. The association of lowered
magma supply, caldera formation, and explosive activity might char-
acterize other basaltic volcanoes, but has not been recognized.
INTRODUCTION
Kilauea (Hawai'i) is an iconic effusive volcano, known for its lava
flows and high fountains. Approximately 17% (250 km') of the volca-
no's subaerial narks has been resurfaced by lava flows in the past ?QO yr
(Fig. 1). and the ongoing Pu'u '0'6 eruption on the east rift zone, nearly
continuous since 1983, covered >125 km= with >4 km' of lava by 2014.
Our analysis of Klauea's past 2500 yr shows, however, that explo-
sive eruptions were dominant for periods lasting several centuries, not just
brief diversions at an otherwise effusive volcano. We find that Kilauea has
been in a dominantly explosive mode –60% of the past 2500 yr. The ef-
fusive style of the p ;tit 200 yr is, from that perspective, misleading.
For this paper we distinguish lava fountains, which at Kilauea in-
variably feed lava flows and contain only juvenile components, from ex-
plosive eruptions, which do not feed. lava flows and have at least some
lithic components. In this usage. most of Kilauea'S explosive eruptions are
phreatomagmatic or phreatic, though some may be products of overpres-
surized may=ntatic gats.
Kilauea had many explosive eruptions older than chose discussed
here (Easton, 1987), but details are lacking. We deal with only the past
2500 yr. for which many ages and stratigraphic. controls are available,
and examine only periods lasting centuries, not short-term events of sev-
eral years or less.
TWO DONIINANTLY EXPLOSIVE, PERIODS
Recent studies indicate two long periods of time during which ex-
plosive activity dominated ,the summit region and adjacent south slope
of Kilauea (Fig. 2A). More than 140'calendar-calibrated "C ages (Ta-
Sites of dated flows 155.25•W
1
0 1500-180Q CE
PACIFIC
• 1000-1500 CE
o P,
OCEAN
• 200 BCE -1000
CE V
• >200 BCE
P
J
P
9.5"N
of
oe
P 1P
tcilometers
KN
Ages of map units
19x5'
W Post -1800 CE
10 15th century
cE
PACIFIC OCEAN
= pOre-01000('%
Mad rnd from .r to arrcr
155.25° W
W". 2�J6 ,
Figure 1. Locations of all 14C -dated lava flow samples on Kuauea
Volcano (Hawai'f), color coded by time periods discussed in text.
Data are available in Table DRi (see footnote 1). Two sample loca-
tions north of Kilauea Caldera (KC) with ages older than 200 BCE are
shown beyond the limit of Kilauea, because the dated flows are in
the subsurface overlain by tephra. Map colors indicate gene.. A ..ges
of lava flows compiled from map units of Wolfe and V ris (1996),
assuming that their unit poo is entirely younger than iJO0 CE. All
flows younger than 1800 CE were recorded during or shortly after
the eruption. The 151h century'Aili'au flow field is shown separately
to emphasize its large size; samples along Its margin date the flow
field (Clague et al., 1999). Other flow fields: PO—Pu'u 'O'6 flow field
(1983–presentL which has enlarged somewhat from depiction of
Wolfe and Morris (1996) used here; MU—Wuna Ulu flow field (1969-
74); KN—Kipuka Nene flow field (200-300 BCE).
bles DR2 and DR3 in the GSA Data Repository'; Stuiver and Reimer.
1993; Reimer et al., 2004) define the length -of each explosive period, as .
interpreted in Fiske et al. (2009) and Swanson et al. (2012a).
The Uwekahuna Ash contains deposits of explosive eruptions be-
tween ca. 200 BCE and 1000 CE (Fiske et al., 2009). Only three lava
flows have been, found interbedded with the Uwekahuna Ash. Two are
south of Kilauea Caldera; a third is interleaved with tephra low on the
caldera wall and may correlate paleomagneticadly with one of the other
flows (Fiske et al.; 2009). The Keanakako'i Tephra Member was produced
between ca. 1500 and 1800 CE (Swanson et al.. 201.2a). Only one lava
flow was erupted at Kilauea's summit during that time, from the outermost
ring. fault bounding the south caldera. Thus, for 2 periods of time lasting
1200 yr and 300 yr, 10auea's summit, normally the site of frequent lava
. flows (Holcomb, 1987; Neal and Lockwood. 2003), had little effusive ac -
'GSAIData Repository item 2014233, Tables DRI-DR3 and Figure DRi,
showing all "C ages of lava flows and tephra and their calendar -calibrated ages,
and Tables DR4 and DR5, presenting chemical data, is available online at www
.geosociety.org/pubs/ft'-)014.htm, or off request from editing@geosociety.org or
Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.
rGEOLOGY, July 2014: v. 42; no. 7: p. 631-634; Data Repository item 2014233 I doi:10.1130/G35701 1 1 Published online 22 May 2014
(V2014 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or editing@geosociety.org. 631
Figure 2. A: Histogram
Volume
(km')
showing number (No.)
N 21
00,L
of tephra ages, prepared
m a
from Tables DR2 and DR3
0o
(see footnote 1). B: Histo-
0.3
gram showing number of
Explosive
different dated lava flows
11
per century in each erup-
260
tive period. Numbers indf-
o
tate how many different
Z
flows were dated per pe-
5.5
riod. From 1800 CE, only
Effusive
flows outside the caldera
are counted, because in-
ZLC
tracaldera flows are not
E
recognizable for earlier
=i
periods. Analytical data
j
and original figure from
which histogram was pre-
pared are in Table DRi -sco 0 500 '000 1500 2000
and Figure DR11; entire Calendar year (BCE negative)
calendar ranges were
used except where constrained by stratigraphy. C: Histogram of esti-
mated volume (not corrected for pore space) erupted subaerially dur-
ing each eruptive period. Gray is dominantly effusive period. Black is
dominantly explosive period. UA—Uwekahuna Ash; KT—Keanakako'i
Tephra Member.
tivity. Instead, almost all summit eruptions were explosive: sporadic, vio-
lent, and brief. In contrast. most eruptions during the intervening periods
were frequent, effusive, and sometimes lasted for decades. This suggests
a cyclicity in summit activity. shifting from mostly effusive to mostly ex-
plosive and back again.
DISTRIBUTION OF LAVA FLOW AGES
Did the rest of the volcano, beyond the extent of marker ash beds for
the two dominantly explosive periods, behave similarly? To address this
question, we compiled and calendar -calibrated all 93 known r'C ages of
10auca lava flows younger tJhan ca. 500 BCE (only a few ages are older
than that -.Table DR 1.1. Figure 2B and Figure DR1 show the distribution of
the calendar -calibrated flow ages with time for the past 2500 yr, with the
two periods of major tephra production indicated. Lava flow ages cluster
in periods between explosive eruptions. This is best shown for the past
1000 yr. Numerous ages plot in the 100G-1500 CE time interval, when
the present summit shield was under construction (Holcomb, 1987; Neal
and Lockwood. 2003); few are within die 1500-1800 CE period, when
the Kea nakako'i tephra was deposited, and numerous lava flows have
been observed to form since 1800 (1823 CE is the actual age of the oldest
known post-Kcanakiko'i flow). The number of flow ages is small before
1000 CE, but there appears to be an increase in the period 200-500 BCE
(Fig. 2A), from 0.8 flows/100 yr to 3.3 flows/100 yr, bracketing the 200
BCE -1000 CE age of the Uwekahuna Ash.
Three factors could weaken this pattern. The spatial distribution of
lava flow ages is uneven (Fig. 1). The lower east rift zone is underrepre-
sented. For example, flows from Heiheiahulu (Holcomb, 1987) and the
so-called 1790 flow (Moore and Trusdell, 1991), both ascribed to the 18i1
century, are not dated.
Second, young flows cover old ones, so die temporal record becomes
obscured with age. Few ages, however, plot in the 1200 -yr -long period
of tephra production, in contrast to die many flow ages in the following,
much briefer, 500 -yr -long period.
Third, flows along die Puna Ridge, the 75 -km -long submarine ex-
tension of the east rift zone, etre not dated well enough for our analysis
(Smith ct al., 2002). Palagonite rind thicknesses suggest ages for dredged
samples of 700-24.000 yr, mostly 2000-7000 yr (Clague et al., 1995).
This suggests relatively little eruptive activity during the past 2500 yr. The
large flows at the base of the ridge are not young, based on sediment cover
(Clague et al., 1995). We discuss only the subaerial edifice in this paper,
but suspect that our conclusions apply to the Puna Ridge.
Acknowledging these caveats, we think that the clear pattern for the
summit area holds for the entire subaerial edifice. We interpret th-_ sub-
aerial volcano to have undergone alternating periods of mostly explosive
and mostly effusive eruptions for the past 2500 yr. Successive periods con-
stitute explosive -effusive cycles of varying duration.
MAGMA SUPPLY DROPS DURING PERIODS OF MAINLY
EXPLOSIVE ACTIVITY
How do eruptive volumes and rates of magma supply compare be-
tween the explosive and effusive periods? The volume of magma erupted
during periods of dominantly explosive activity is far less than that dur-
ing effusive periods (Fig. 2C), and the calculated magma supply rate is
correspondingly lower, only 1%-2% of the effusive rate (Table 1). Our
estimates of flow volumes (Table 1; Fig. 2C) are compromised by variable
flow thickness and coverage by later flows. A simple comparison of tephra
and flow thickness at the summit area, however, illustrates the disparity
between effusive and explosive volumes.
TABLE 1. ESTIMATED VOLUME OF LAVA ERUPTED ON SUBAERIAL KTLAUEA
DURING DOMINANTLY EFFUSIVE AND DOMINANTLY EXPLOSIVE PERIODS
Calendar age
range
Volume
(km')
Magma supply rate
(km°/yr•)
Dominant style
500r200 BCE
>0.6
not calculated
Effusive
200 BCE -1000 CE
0.3
2.5 x 10-`
Explosive
1000-1500 CE
11
2.2 x 10-2
Effusive
1500-1800 CE
0.15
5 x 10-'
Explosive
1600 -present
5.5
2.6 x 10-1
Effusive
Note: Effusive volumes estimated using mapped areas of flows and areas
projected beneath younger flows, as shown on geologic maps (Wolfe and Morns,
1996; Neal and Lockwood, 2003), assuming an age consistent with 14C data
and unit label; thickness was estimated from field observations and topographic
gradient. Explosive volumes were estimated from area and average thickness of
juvenile tephra. Volumes were not adjusted for vesicularity (lava flows) or pore
space (tephra).
'Average supply rate to ground surface for entire period. Not calculated for earli-
est period, which began before 500 BCE and so is incomplete.
The 140 -m -high wall of Kilauea Caldera is made almost entirely of
flows erupted between )000 and 1500 CE (Neal and Lockwood. 2003),
when the Observatory shield was built (Holcomb, 1987); its total thick-
ness is more, because the base of the shield is covered by caldera fill. In
contrast, the maximum exposed thickness of the Keanakako'i tephra is
only -11 m (McPhie et al., 1990; Swanson et al., 2012x). The flow thick-
ness is several meters thick 5 km southwest (downwind) of the summit,
and the tephra is only several centimeters.
A similar comparison can be made for the Uwekahuna Ash on
Kilauea's south flank. It is at most a few tens of centimeters thick, thin-
ning to only a few centimeters at the coastline (Fiske et al.. 2009), but the
overlying and underlying flows are each at least several meters thick.
The striking difference in erupted volume between periods domi-
nated by effusive and explosive activity must reflect a major disruption
to the supply system that lasts for centuries. The disruption could take
place anywhere between the point of melt accumulation in the mantle and
the shallow storage system beneath Krlauea's summit. Perhaps increased
magma supply to Mauna Loa volcano causes a drop in supply to Kilauea
(Gonneimann et al., 2012). Once magma enters the Kilauea plume, it
could be diverted away from the shallow reservoir, perhaps as intrusions
into the crust or lower shield (Lin et al., 2014). A subhorizontal mantle
pathway of magma transport at -30 km depth, interpreted by Wright and
Klein (2006; see also Wolfe et al., 2003), might be a zone within which
magma could stall or. -be divertedil Magma probably did not bypass the
summit reservoir system and immediately erupt on the Puna Ridge, be -
632 www.gsapubs.org I July 2014 1 GEOLOGY
.p
cause lava compositions on the ridge are highly fractionated, reflecting
shallow storage (Clagrrue et al., 1995). Whatever the cause, a major change
in the dynamics of Kilauea's magma supply system is needed to explain
the effusive -explosive cyclic behavior.
RELATION OF CALDERA TO CYCLES
Both explosive periods occurred when a deep caldera indented
Kilauea's summit. The Powers caldera existed during due time of depo-
sition of the Uwekahuna Ash (Powers, 1948: Holcomb, 1987) and only
began filling ca. 1000 CE, as estimated from the age of the oldest flow
overlying the Uwekahuna Ash south of the caldera (Fiske et al., 2009).
The modem caldera formed ca. 1500 CE, when the Kcanakako'i explo-
sive period started, and began to fill ca. 1500 CE (Swanson et al.. 2012x).
The phreatomagmatic and phreatic nature of most of the explosive erup-
tions suggest~ that the two calderas were often deep, at or below the water
table (Mastin, 1997; Mastin et al.. 2004). Today the water table is —615 m
below the highest point on the caldera rim, —490 m below the caldera
floor (Keller et al., 1979). We think it likely that parts of the floor were at
least that deep during the dominantly explosive periods. The water table is
unlikely to have been higher during the past few hundred years, to judge
from the presence of fresh basalt above, and altered basalt below, today's
water tabic (Hurwitz et al., 2002, 2003).
-The cause of caldera formation at Kilauea is uncertain. If the caldera
collapsed because a large volume of magma was rapidly erupted or in-
truded into the east rift zone, as conventionally thought (Holcomb et al.,
1988), then why did the average magma supph? rate drop by almost two
orders of magnitude and stay low for centuries? No relatively shallow,
top-down process in the edifice is likely to cause such a long-term change
in behavior.
Recognition of eruption cycles at Klauca raises a new possibility.
Perhaps caldera formation results from reduction in magma supply to the
shallow storage reservoir, a term we use for a configuration of multiple
magma storage volumes (Fiske and Kinoshita, 1969; Dawson et al., 2004;
Baker and Amelung. 2012: Poland et al., 2014). If magma supply dropped
significaritly, eruptions or intrusions might deplete the shallow reservoir,
eventually leading to collapse of the overlying edifice. This bottom-up
model seems best suited to explain the linked formation of the caldera and
the ensuing decline in magma supply.
Before each explosive period, an effusive eruption took place that
could have depleted the storage reservoir if magma supply were low. Ex-
trusion of the 4-6 km' -Ail.`au flow field (Fig. 1) during an —60 yr period
(Clague et al., 1999) immediately preceded collapse of the modem calde-
ra. The Kpuka NEnE flow field (Fig. 1; Holcomb, 1987) covers more than
130 km' (extrapolating back to its summit source) on Kilauea's south flank
and immediately underlies the Uwekahuna Ash. Much of its volume could
have flowed into the ocean, given its coastline width of —7 km (Fig. 1). We
estimate that at least 0.5 km -'remains on land, and the total volume erupted
could be several times larger.
An alternative bottom-up model is that an increase in mantle melt-
ing rate supplied the large flow fields preceding caldera formation. The
increase could have depleted a relatively large volume of its melt compo-
nents, so that little magma was available to enter the volcano for several
centuries.
Both altematives are at odds with the ongoing eruption of Pu'u '0'6,
during which the supply rate has generally been —0.12 km' yr'except for
a temporary increase in 2003-2007 (Poland et al.. 2012).
MgO-RICH AND FRACTIONATED VITRIC TEPHRA
Small volumes of MgO-rich juvenile ash occur in both the
Uwekahuna Ash and Keanakiko`i tephra. Microprobe glass analyses
show that vitric ash with MgO to 1 I wt% occurs at several levels in the
Kcanak5ko'i tephra (Martin et al.. 2004: Table DR4), and one thin vitric
ash in the Uwekahuna Ash (Fiske et al.. 2009) contains MgO values of
GEOLOGY I July 2014 1 Aww.gsapubs.org
12.5 wt% (Table•DR5A; Helz et al., 2014). Such high MgO contents in
erupted melt etre unprecedented among summit lava flows at Kilauea; the
highest published amount in glass is 1.0.2 wt9'o (Helz et al., 2014). The
normal summit glass compositions (6-9 wt% MgO; Garcia et i:, 2003)
indicate storage in the shallow reservoir and crystallization of mainly oliv-
ine before eruption (Powers, 1955; Wright, 1971).
We think it significant that the highest MgO melts were erupted dur-
ing explosive, not effusive, periods. We interpret these high Mg0 values to
record brief or no storage in a disrupted shallow reservoir not fully recov-
ered following caldera collapse. Helz et al. (2014) reached a similar inter-
pretation for the Kulanaokuaiki tephra and for the much olderPahalaAsh.
The most highly fractionated compositions known at Kilauea's sum-
mit also occur in vitric tephra. For example, in the Keanakako'i tephra,
thin vitric ash just above layer 6 (McPhie et al., 1990. Swanson et al.,
2012a) has only 4.0 wt% MgO and as much as 4.5 wt% TiO, and 1.0
wt% K 2 0 (Table DR4). We interpret such unusual compositions to reflect
isolated storage and advanced fractionation of small pockets of magma
undisturbed by fresh, mantle -derived magma during low magma supply.
A marker bed in the Kulanaokuaiki tephra (unit 2 of Fiske et al.,
20093 contains glass with unusually high values of TiO, (>3 wt%) and KO
(>0.7 wt%) for its moderately low Mg0 value (6.7 wtO1a; Table DR5B). A
temporally associated lava flow, one of two interbedded with the Kulana-
okuaiki tephra south of the caldera, has a similar composition (Wolfe and
Morris, 1996). This composition does not plot along the typical Kilauea
fractionation trend during effusive periods, although similar compositions
occur in lava flows earlier in Kilauea's history (Chen et al., 1996; Lipman
et al., 2006).
HAZARD INIPLICATIONS
Previous studies indicate that hazards of explosive eruptions at
Kilauea are substantial, including pyroclastic density currents as well as
tephra falls and ballistic showers (Decker and Christiansen, 1984; McPhie
et al., 1990; Dzurisin et al., 1995; Fiske et al., 2009; Swanson et al., 2012a,
2012b). Our work shows that the hazardous periods last much longer than
previously thought. When the next dominantly explosive period begins,
society may have to deal with centuries of repeated explosive activity at
Kilauea's summit.
CONCLUSIONS
Recognition of' the explosive -effusive cycle raises far-reaching ques-
tions about the d-vnamics of Kilauea volcano. Rather than erupting lava
flows almost continuously, Kilauea is instead a more complex, modulated
system in which melting rate, supply rate, conduit stability (in both mantle
and crust), reservoir geometry, water table, and many other factors inter-
act. The explosive -effusive cycle is the net result of this interaction. Future
studies approaching Kilauea from a broad perspective are necessary to
significantly advance our understanding of one of Earth's most studied
volcanoes. Whether unrecognized explosive -effusive cycles occur on oth-
er basaltic volcanoes is a topic for future research:
ACKNOWLEDGMENTS
We thank Dave Clague, Dan Dzurisin, Shaul Hurwitz, and Pete Lipman for
insightful manuscript reviews. Tom Wright's comments on a preliminary version
were provocative. Dave Sherrod supplied three unpublished 1°C ages. Discussions
with Frank Trusdell were valuable. Robin Holcomb proposed eruptive cycles that
differ from ours but stimulated our thinking. Garcia and Mucek were supported by
National Science Foundation grant EAR -11 18741. This paper is School of Ocean
and Earth Science and'rechnology (SOEST) contribution 9115.
REFERENCES CITED'
Bakcr, S., and Amehmg, F., 2012, Top-down inflation and deflation at the summit
of Kilauea Volcano, Hawai'i observed with inSAR: Journal of Geophysical
Research, v. 117, B 12406, doi:10.1029/201IJB0f)9123.
Chen, C. -Y., Frev, F.A., Rhodes, J.M., and Easton, R.M., 1996, Temporal geo-
chemical evolution of Kilauea Volcano: Comparison of Hilina and Puna
Basalt, in Basu, A., and Hart, S., eds., Earth processes: Reading the isotopic
633