You are on page 1of 5

GEOPHYSICAL RESEARCH LETTERS, VOL. 32, L06314, doi:10.

1029/2004GL021968, 2005

Interpretation of very long period tremors at Ambrym volcano,


Vanuatu, as quasi-static displacement field related to two distinct
magmatic sources
D. Legrand,1 D. Rouland,2 M. Frogneux,2 R. Carniel,3 D. Charley,4 G. Roult,5
and C. Robin6
Received 11 November 2004; revised 17 January 2005; accepted 7 February 2005; published 30 March 2005.

[1] A 3-component broadband seismometer (STS-2) has


been deployed on the SW slope of Ambrym, Vanuatu, a
large-caldera volcano with two persistent lava lakes from
July to November 2000. Well-polarized Very Long Period
Tremors (18 22s; VLPTs) occurred quite continuously,
whereas other seismic signal with periods of 9s, 6 7s, 5s,
2.3 2.7s, 1.95s and 1s were intermittent. VLPTs signals
show two clearly distinct directions pointing towards the
caldera. Polarization orientations suggest two sources: the
first one, and the most active, fits with the zone from which
silicic lava flows erupted in 1986, in the central-to-eastern
part of the caldera. The second one corresponds to the
active zone formed by the complex Marum volcanic center
and the Benbow cone, in the western part of the depression
structure. The depths are estimated between 2.7 2.9 km
with respect to sea level. VLPTs are interpreted as the quasi
static-displacement field of two nondestructive sources.
They indicate the persistent continuing magmatic activity
beneath the eastern floor of the caldera, although no
volcanic activity is observed in this area, and the more
obvious, but less active, corresponding to the lava lakes.
These data also corroborate the complex internal structure
of the Ambrym caldera, suggested by geological data
and bimodal magmatism (basaltic, 50% SiO2, and
silicic, 60% SiO2). Citation: Legrand, D., D. Rouland,
M. Frogneux, R. Carniel, D. Charley, G. Roult, and C. Robin
(2005), Interpretation of very long period tremors at Ambrym
volcano, Vanuatu, as quasi-static displacement field related to
two distinct magmatic sources, Geophys. Res. Lett., 32, L06314,
doi:10.1029/2004GL021968.

1. Introduction
[2 ] Since about ten years, broadband seismometers
(0.01 to 120 s) have been largely used to record volcanic
tremors, volcanic earthquakes, phreatic or magmatic
explosions [e.g., Kawakatsu et al., 1992; Chouet et al.,
1

Departamento de Geofsica, Universidad de Chile, Santiago, Chile.


Ecole et Observatoire des Sciences de la Terre, Strasbourg, France.
3
Dipartimento di Georisorse e Territorio, Universita` di Udine, Udine,
Italy.
4
Department of Geology, Mines and Water Resources, Port Vila,
Vanuatu.
5
Institut de Physique du Globe, Laboratoire de Sismologie, Paris,
France.
6
Laboratoire Magmas et Volcans, UR 163, IRD, Universidad de Chile,
Departamento de Geologa, Santiago, Chile.
2

Copyright 2005 by the American Geophysical Union.


0094-8276/05/2004GL021968$05.00

1994; Neuberg et al., 1994; Kaneshima et al., 1996; Rowe


et al., 1998]. Nevertheless, Very Long Period signals
(hereafter noted VLPTs) have been observed in only a
few cases, as on Aso, (Japan), where Kaneshima et al.
[1996] reported tremors with periods of 10 15s. VLPTs of
20s, 11 12s and 7s have also been recorded on Erebus,
Antarctica [Rowe et al., 1998; Aster et al., 2003] as well as
events with periods up to 20s on Stromboli, Italy [Chouet et
al., 1999].
[3] VLPTs may be explained by movements of a
nondestructive source and/or by propagation effects in a
two-phase medium (solid and fluids like water, gas,
bubbles, steam and/or magma) [e.g., Chouet et al., 1994;
Ripepe and Gordeev, 1999]. In order to illustrate the
probable occurrence of VLPTs in an active lava lake, we
installed a 3-component broadband seismometer (STS-2)
on Ambrym volcano (Vanuatu), from July to November
2000. This volcano is one of the few volcanoes in the
world to have at least one open continuously active lava
lake.

2. Ambrym Volcano
[4] Ambrym (168080E; 16150S), in the Vanuatu archipelago, is a 35  50 km wide basaltic volcano, rising about
1800 m from the adjacent ocean floor. Nearly 800 m above
sea level, the edifice is truncated by a 12 km-diameter
caldera. Pre-caldera subaerial volcanics essentially consist
of lava flows and subordinate pyroclastics, and form a
shield shaped edifice. A series of tuffs associated with the
caldera formation overlies this volcano and constitutes most
of the cone up to the caldera rim [Robin et al., 1993]. Then,
the caldera was largely filled by recent pyroclastics and lava
flows. Specially, volcanic activity from two large cones
located in its western part, -the Marum (1270 m)
and Benbow (1160 m) cones-, and products from several
phreatomagmatic vents and small cones in the eastern part,
has infilled the depression (Figure 1). A 2 to 3 km wide rift
crosses over the island in a W-E direction, passing by the
center of the caldera. Ambrym has had a large number of
historical eruptions: the largest recent eruptions occurred in
1888, 1894, 1913 14, 1929, 1937, 1942, 1952 53, 1957,
1986 and 1988 89. Since the latter eruption, which
produced extensive lava flows on the caldera floor, quite
constant lava lakes are present in Benbow and Marum
craters. In addition, it is noteworthy that the 1986 lavas
which erupted in the E-NE part of the caldera (Figure 1)
have an evolved, siliceous, composition (60% SiO2)
whereas the 1988 lavas, issued from the south flank of

L06314

1 of 5

L06314

LEGRAND ET AL.: VERY LONG PERIOD TREMORS AT AMBRYM VOLCANO

L06314

almost continuously seen over the whole survey period.


Four intense tremor activity periods occurred during the
survey (Figure 2 (bottom)). Unfortunately, the fourth one
and the most active started during the last part of the survey,
cancelling the possibility to analyze the sudden increase of
activity.

Figure 1. Sketch Map of Ambrym Island. The shortperiod (black square) and broad-band (black circle) stations
are shown.

22s VLPTs
4.2. Time Domain Analysis of the 18
[7] The polarization directions have been calculated
following the method described by Plesinger and Horalek
[1976] and Plesinger et al. [1986]. The instantaneous value
of the products Z(t)*N(t) and Z(t)*E(t) between the
3 components are performed continuously. If predominant
noise exists, these products will vary randomly, due to
incoherent signals between the vertical and horizontal
components. If linearly polarized signals (such as P waves)
exist, each of these products will be strictly positive (or

Marum cone, are basalts (50% SiO2) [Robin et al., 1991;


Picard et al., 1995].

3. The July November 2000 Seismic Experiment


[5] Two seismic sensors (a vertical short-period Mark
Product L4C and a 3-component Streckeisen STS-2 broadband seismometer) have been installed in July November
2000. The short period (square on Figure 1) was installed
near the eastern rim of Benbow crater (16.257S,
168.108E) and the broadband STS-2 seismometer (circle
on Figure 1) was installed near Lalinda village, on the
southwestern flank of the main cone at about 9 km from the
Benbow crater (loc. 16.326S and 168.058E). Both
seismometers were connected to a GEOSTAR 16-bit data
acquisition system. The short period seismometer supplied
measurements from 15 July to 11 August 2000 and results
have already been exposed in a previous article [Carniel et
al., 2003]. The broadband station has been deployed from
14 July to 26 November 2000. The continuous 3-component
signals were sampled at 75 samples per second, the GPS
time recorded on a separate channel, and the complete
data set stored on a 2Gb hard disk, allowing an independent, continuous acquisition for about 4 months. During
the 137-day data field acquisition, only 55 hours (end of
9 November to beginning of 11 November 2000) have
been lost, due to power problems.

4. Data Processing
4.1. Spectral Domain Analysis
[6] Figure 2 displays a time-frequency representation of
the ground velocity for the 135 days with the vertical axis in
logarithmic scale: Figure 2 (top) is between 0.6s and 10s,
and Figure 2 (bottom) between 15.8s and 31.6s. As we are
interested in studying local volcanic activity, we removed
part of the signal induced by teleseismic and regional
earthquakes reported by the NEIC agency. We observed
seven different dominant periods at 18 22s; 9s; 6 7s; 5s;
2.3 2.7s; 1.95s and 1s, some of them being intermittent
(Figure 2). The 18 22s VLPTs studied in this paper were

Figure 2. Time-frequency diagram (FFT) of the ground


velocity computed over one day, with a 24H averaged
amplitude without overlapping, for the 135 daily recorded
signals. (top) Time-frequency representation between
0.6 and 10s. (bottom) Same between 15.8 and 31.6s. Note
the three time laps of increasing activity before the day 120,
and a greater activity starting at the end of the survey. The
spikes that appear at regular positions are simply the log
effect on the ordinate axis (few points at the low frequency
part of the signal). We didnt apply interpolation to show
raw data. Both vertical axes are in log10. Vertical white
lines are drawn when no data are available. See color
version of this figure in the HTML.

2 of 5

L06314

LEGRAND ET AL.: VERY LONG PERIOD TREMORS AT AMBRYM VOLCANO

L06314

the Benbow and Marum craters (Figure 1 and picks B on


Figure 4). The tremor activity which corresponds to A was
predominant from July to October, whereas activity
corresponding to B becomes dominant in November
(Figure 4).

5. Discussion
5.1. Polarization and Quasi-Static Displacement Field
[8] A classical waveform in the near-field is the superposition of P, near-field, S, surface waves and the static
displacement field (SDF). The Near-Field waves and the
SDF have waveforms similar to the source time function. If
we suppose a source at a distance of 10 km, the difference in
arrival times between S and P waves is about 1 to 2s for
typical P and S wave velocities, corresponding to a factor of
order 10 compared to the 20s period. Hence, the 20s period
signals are mainly the filtered source time function. If we
suppose an isotropic source, the corresponding polarization
will be radial, exactly as for P-waves, even if they are not.
Only one radial (source-receiver) phase was observed on
Ambrym VLPTs, in particular no S-waves were recorded.

Figure 3. (top) Example of 1 hour three component Z, N


and E records (filtered between 18 22s period) and the
corresponding products Z*N and Z*E computed to calculate
the direction of polarization. (bottom) Raw particle motion
in the horizontal and vertical planes (respectively left and
right) for a 5-minute window selected in the 1 hour signal
between 1200 and 1500s. The horizontal components (and
the corresponding particle motion) are drawn before the
correction of the 12 degrees local magnetic declination.

negative), depending on the source azimuth, and the


maxima on both products will occur at the same time.
The time lag between these maxima gives a quantitative
quality criterion of the linear polarization of the signal: a
time lag of less than 1 sec has been applied in this study.
The polarization of the signals has been studied at different
frequencies, and it clearly appears that only the 18 22s
filtered signals (VLPTs) are well polarised (with the
criterion defined before). The 18 22s VLPTs consist of
wave trains with slightly varying amplitudes lasting up to
few minutes. They are very close in time to each other so
that they appear like almost continuous tremors. Figure 3
shows one hour record of the three components filtered
between 16 and 25s. The azimuth Az(t) and the dip angle
Dip(t) have been automatically calculated on a five minutes
sliding window. Figure 3 (bottom) shows an example of
particle motion in both horizontal and vertical planes. The
well-polarised 18 22s tremors indicate two preferential
directions: N54E and N32E, with 12 and 15 dip
upwards angles respectively. These directions are pointing
towards the central part of the volcano. Precisely, the first
N54E direction indicates the 1986 lava exit point (Figure 1
and picks A on Figure 4), while the N32 one points to

Figure 4. Monthly diagrams of the polarization angles


(azimuth and dip) of VLPTs observed during the survey and
performed on 5-min sliding windows. Each count (y-axis)
corresponds to the observation of simultaneous maxima (in
a 1 sec time window) on both products Z*N and Z*E. The
1 sec time window criterion guarantees phase coherence
and allows elimination of the signal disturbed by random
noise or distant earthquakes. The arrows, indicating the
polarization directions, are drawn with a length proportional
to the number of events.

3 of 5

L06314

LEGRAND ET AL.: VERY LONG PERIOD TREMORS AT AMBRYM VOLCANO

As the two sources are clearly distinct in space, the


assumption that they have the same focal mechanism
implies that the station is not in a nodal direction for
S-wave. Hence, the source is not a double-couple, but it
may more likely be isotropic, as for similar wavelength
tremors already reported on Aso [Kaneshima et al., 1996;
Legrand et al., 2000]. This has been our main hypothesis
in order to emphasize that the polarization direction points
towards the source (see Legrand et al. [2000] for a
detailed explanation). A double direction of polarization
in a vertical plan has already been observed at Erebus
volcano [Rowe et al., 1998; Aster et al., 2003]: the
VLPTs polarization directions start at shallow depth and
migrates downwards in time. At Ambrym, a double
polarization direction is also observed, but not in only
one vertical plane. We suggest that the source belonging
to the B (N32) direction corresponds to the Benbow and
Marum active zone and consequently, the other direction
of polarization (A, N54) is related to another source near
the 1986 lava vent. The directions of polarization seem
related in the sense that when one direction is important,
the other one is not.
[9] The VLPTs sources may be slow, nondestructive
movements like an increasing and then decreasing volume,
which returns to its original state, and without permanent
dislocation. The rapid amplitude-variations with distance (in
1/r2) of the NFWs and static or quasi static parts may be used
to perform a joint inversion for both the focal mechanism and
the location of very long period volcanic events. If this
method was possible on Aso [Legrand et al., 2000], it could
not be applied here with a single station.
5.2. Interpretation of the Signals in the Volcanologic
Context of Ambrym
[10] Tremor activity at Ambrym indicates the persistent
confined magmatic activity beneath the eastern floor of the
caldera, although no volcanic activity is presently observed
in this area. Whereas the occidental part of the edifice is the
centre of an open plumbing system shown by two active
lava lakes, the part of the volcano below the eastern part of
the caldera is the site of the most energetic tremors in a
close system. Thus, a first conclusion is that important
VLPTs tremors may also be generated within a volcanic
close system, as observed in only few cases [e.g.,
Kaneshima et al., 1996].
[11] The observed dips and directions suggest that source
A is at about 3.5 km (respect to the caldera), below the 1986
emission point, and source B is at about 3.7 km (respect to the
caldera) below the Marum and Benbow cones. Such
depths indicate two zones of important magmatic activity
at 2.7 2.9 km respect to sea level. This corroborates the
complex internal structure of the Ambrym caldera, emphasized by geological data [Robin et al., 1993] and bimodal
magmatism shown by the 1986 and 1988 eruptions. Probably,
the different rheologic properties (low and high viscosity) of
magmas in both systems may also be responsible for the
different seismological characteristics observed.

6. Conclusion
[12] At Ambrym, the recorded seismic signal shows a
wide range of periods at 18 22s, 9s, 6 7s, 5s, 2.3 2.7s,

L06314

1.95s and 1s. The 18 22s signals appear as almost


continuous tremors with quasi-constant amplitude, independently of signal duration, reflecting a steady activity of the
volcano. They are well polarized in two main directions,
N32E and N54E, with respective dip angles of 15 and
12, suggesting a regular and nondestructive process of
two VLPTs sources. The most active source fits with the
zone from which silicic lava flows erupted in 1986, in
the central-to-eastern part of the caldera, and indicates
persistent confined activity in this area, although no surface
activity is observed at this level. The second source, less
active, corresponds to the open-active lava-lake zone in the
west part of the caldera, formed by the complex Marum
volcanic centre and the Benbow cone. Such results corroborate the complex internal structure of the Ambrym caldera,
suggested by geological data and bimodal magmatism
(basaltic, 50% Sio2, and silicic, 60% SiO2) shown by
the 1986 and 1988 eruptions.
[13] Acknowledgments. We are very grateful to the people of
Lalinda village, Vanuatu, for their help in the field operations and to
J.Ph. Caminade, who successfully organized the whole logistics of the
survey, as well as IRD (Institut de Recherche et de Developpement). We
thank C. Ioan, Director of the Department of Geology, Mines and Water
Resources, Port Vila, Vanuatu. We thank M. Fulle and D. Nakedau, for their
participation during the survey and A. Cisternas for helpful comments. The
comments by Joshua Jones and an anonymous reviewer greatly enhanced
the quality of the manuscript.

References
Aster, R., S. Mah, P. Kyle, W. McIntosh, N. Dunbar, J. Johnson, M. Ruiz,
and S. McNamara (2003), Very long period oscillations of Mount
Erebus Volcano, J. Geophys. Res., 108(B11), 2522, doi:10.1029/
2002JB002101.
Carniel, R., M. Di Cecca, and D. Rouland (2003), Ambrym, Vanuatu
(July August 2000): Spectral and dynamical transitions on the hoursto-days timescale, J. Volcanol. Geotherm. Res., 128, 1 13.
Chouet, B., R. Page, C. Stephens, J. Lahr, and J. Power (1994), Precursory
swarms of long-period events at Redoubt volcano (1989 1990), Alaska:
Their origin and use as a forecasting tool, J. Volcanol. Geotherm. Res.,
62, 95 135.
Chouet, B., G. Saccorotti, P. Dawson, M. Martini, R. Scarpa, G. De Luca,
G. Milana, and M. Cattaneo (1999), Broadband measurements of the
sources of explosions at Stromboli volcano, Italy, Geophys. Res. Lett.,
26, 1937 1940.
Kaneshima, S., et al. (1996), Mechanism of phreatic eruptions at Aso
volcano inferred from near-field broadband seismic observations,
Science, 273, 642 645.
Kawakatsu, H., T. Ohminato, H. Ito, Y. Kuwahara, T. Kato, K. Tsuruga,
S. Honda, and K. Yomogida (1992), Broadband seismic observation at
the Sakurajima volcano, Japan, Geophys. Res. Lett., 19, 1959 1962.
Legrand, D., S. Kaneshima, and H. Kawakatsu (2000), Moment tensor
analysis of near field broadband waveforms observed at Aso volcano,
Japan, J. Volcanol. Geotherm. Res., 101, 155 169.
Neuberg, J., R. Luckett, M. Ripepe, and T. Braun (1994), Highlights from a
seismic broadband array on Stromboli volcano, Geophys. Res. Lett., 21,
749 752.
Picard, C., M. Monzier, J.-P. Eissen, and C. Robin (1995), Concomitant
evolution of tectonic environment and magma-geochemistry, Ambrym
volcano (Vanuatu - New Hebrides arc): Volcanism associated with extension at Consuming Plate Margins, J. Geol. Soc. London, 81, 135 154.
Plesinger, A., and J. Horalek (1976), The seismic broadband recording
and data processing system FBV/DPS and its seismological applications,
J. Geophys., 42, 201 217.
Plesinger, A., M. Hellweg, and D. Seidl (1986), Interactive high-resolution
polarization analysis of broad-band seismograms, J. Geophys., 59, 129
139.
Ripepe, M., and E. Gordeev (1999), Gas bubble dynamics model for
shallow volcanic tremor at Stromboli, J. Geophys. Res., 104, 10,639
10,199.
Robin, C., M. Monzier, J.-P. Eissen, C. Picard, and G. Camus (1991),
Coexistence de lignees HK et MK dans les pyroclastites associees a` la
caldera dAmbrym (Vanuatu, Arc des Nouvelles Hebrides), C. R. Acad.
Sci., Ser. II, 313, 1425 1432.

4 of 5

L06314

LEGRAND ET AL.: VERY LONG PERIOD TREMORS AT AMBRYM VOLCANO

Robin, C., J.-P. Eissen, and M. Monzier (1993), Giant tuff-cone and
12-km-wide associated caldera at Ambrym volcano (Vanuatu, New
Hebrides Arc), J. Volcanol. Geotherm. Res., 55, 225 238.
Rowe, C., R. Aster, P. Kyle, and J. Schlue (1998), Broadband recording of
Strombolian explosions and associated very-long-period seismic signals
on Mount Erebus volcano, Ross Island, Antarctica, Geophys. Res. Lett.,
25, 2297 2300.


R. Carniel, Universita` di Udine, Dipartimento di Georisorse e Territorio,


Via Cotonificio 114, I-33100, Udine, Italy. (roberto.carniel@uniud.it)

L06314

D. Charley, Department of Geology, Mines and Water Resources, PO


Box 001, Port Vila, Vanuatu.
M. Frogneux and D. Rouland, Ecole et Observatoire des Sciences de la
Terre, 5 Rue Rene Descartes, 67000 Strasbourg, France.
D. Legrand, Universidad de Chile, Departamento de Geofsica, Blanco
Encalada 2002, Santiago, Chile. (denis@dgf.uchile.cl)
C. Robin, Universidad de Chile, Departamento de Geologa, Plaza Ercilla
803, Santiago, Chile, IRD, UR 163 (Laboratoire Magmas et Volcans).
G. Roult, Institut de Physique du Globe, Laboratoire de Sismologie,
4 Place Jussieu, 75000 Paris, France.

5 of 5

You might also like