seismic imaging of magma chamber beneath an active volcano › people › yaoki › 2012epsl.pdf ·...
TRANSCRIPT
Earth and Planetary Science Letters 333-334 (2012) 1–8
Contents lists available at SciVerse ScienceDirect
Earth and Planetary Science Letters
0012-82
http://d
n Corr
E-m
knishid
takeo@e
journal homepage: www.elsevier.com/locate/epsl
Seismic imaging of magma chamber beneath an active volcano
Yutaka Nagaoka a,b, Kiwamu Nishida a, Yosuke Aoki a,n, Minoru Takeo a, Takao Ohminato a
a Earthquake Research Institute, University of Tokyo, 1-1 Yayoi 1, Bunkyo-ku, Tokyo 113-0032, Japanb Now at Japan Meteorological Agency, 3-4 Otemachi 1, Chiyoda-ku, Tokyo 110-8122, Japan
a r t i c l e i n f o
Article history:
Received 17 October 2011
Received in revised form
29 March 2012
Accepted 30 March 2012
Editor: P. ShearerCombining our result with independent observations suggests that the low velocity body is likely to
Keywords:
magma transport
seismic interferometry
magma chamber
seismic surface waves
1X/$ - see front matter & 2012 Elsevier B.V.
x.doi.org/10.1016/j.epsl.2012.03.034
esponding author. Tel.: þ81 3 5841 8283; fax
ail addresses: [email protected] (Y
[email protected] (K. Nishida), [email protected]
ri.u-tokyo.ac.jp (M. Takeo), [email protected]
a b s t r a c t
While the location and shape of magma chambers beneath active volcanoes play a key role in
understanding magma transport and forecasting volcanic activity, the nature of magma chambers,
particularly their shape, is not fully understood. Here we found a low velocity body too small to be
detected from conventional techniques by the aid of a modern technique called seismic interferometry.
represent a magma chamber. Our findings demonstrate the utility of seismic interferometry in imaging
a small scale feature with a size of less than 10 km.
& 2012 Elsevier B.V. All rights reserved.
1. Introduction
When magma rises up from depth, it is usually stored atcertain depths where neutral buoyancy acts to the magma (e.g.,Schmincke, 2006). Precise knowledge of the location and size ofsuch storage, generally called magma chamber, in active volca-noes are important to understand and forecast the behavior oferuptions in that the style of volcanic eruptions depends partiallyon the style of differentiation of magma, which depends on thepressure, temperature, thus depth, and the size of the magmachamber at its residence (e.g., Petford, 2003). However, they arepoorly understood mainly because the imaging capability withconventional local earthquake tomography is not good enough toimage magma chambers (e.g., Lees, 2007); it is rare to obtainoptimum earthquake distribution for local earthquake tomogra-phy in volcanic regions and it is difficult to obtain reliable firstarrival picks of S-waves, which is more sensitive to the existenceof magma chamber than P-waves. Here we find a robust lowvelocity region with a radius of about 5 km at several kilometersto the west of the summit. It is found from Rayleigh wave phasevelocity maps at a frequency of 0.1–0.2 Hz derived from seismicambient noise records, a technique emerging rapidly (Shapiroet al., 2005; Brenguier et al., 2007; Nishida et al., 2008, 2009). TheS-wave velocity structure derived from the phase velocity maps(Rawlinson and Sambridge, 2003) indicates that the low velocity
All rights reserved.
: þ81 3 3812 6979.
. Nagaoka),
okyo.ac.jp (Y. Aoki),
o.ac.jp (T. Ohminato).
zone is limited to depths at 5 km or deeper. Combining our resultwith seismic exploration results and geodetic data shows that theinferred low velocity region represents a magma chamber, whichis detected with such high resolution for the first time.
Mt. Asama is one of active volcanoes in central Japan (Fig. 1)with explosive eruptions with Volcano Explosivity Index (VEI;Newhall and Self, 1982) of 5 or more in 1108 and 1783. Recently,moderate-sized eruption with a VEI of 2 occurred in 2004 andminor ones in 2008 and 2009. With this background, Mt. Asamahas been well-instrumented with seismometers and Global Posi-tioning System stations. Shallow magma plumbing system is welldefined with this wealth of data; during the 2004 eruption,magma-filled cracks (dikes) intrudes to the several kilometersto the west of the summit preceding the summit eruption (Takeoet al., 2006). An active source seismic experiment imaged the areaof the dike intrusion by a high P-wave velocity anomaly (Aokiet al., 2009) which is interpreted to be the remnant of repeatingintrusions, indicating the persistence of the inferred magmapathway.
While the active source seismic experiment is capable ofresolving shallow seismic structure, it does not resolve anystructures deeper than 3–4 km below sea level (Aoki et al.,2009). On the other hand, a regional seismic tomography studyimages broader-scale structures from the lower crust to uppermantle (Nakajima and Hasegawa, 2007) but is not capable ofimaging upper crust in details, where magma chambers likelyexist. Here we fill the gap to image the seismic structure of theupper crust beneath Mt. Asama using seismic ambient noise.
An idea of seismic imaging with random signals such asambient noise dates back several decades (Aki, 1957) but it is
Y. Nagaoka et al. / Earth and Planetary Science Letters 333-334 (2012) 1–82
only around the turn of the 21st century that it is started to bepractically applied for acoustics (Lobkis and Weaver, 2001),helioseismology (Duvall et al., 1993), atmospheric infrasound
138°00' 138°30' 139°00'
36°00'
36°30'
37°00'
0 10 20 30
km
region 3
region 1
region 2
Mt. Asama
Fig. 1. Spatial distribution of seismic sites used in this study. Three regions for
phase velocity measurements are circled. The location of Mt. Asama is also shown.
Geographical location of the area of this study is shown by a solid rectangle in
the inset.
0
20
40
60
80
100
-40 -20 0 20 40
Dis
tanc
e (k
m)
Time (sec)
0
20
40
60
80
100
-40 -20 0
Time (
Pha
se v
eloc
ity (k
m/s
)
Frequency (Hz)
0
1
2
3
4
5
0 0.2 0.4 0.6 0.8 1
Pha
se v
eloc
ity (k
m/s
)
Frequenc
0
1
2
3
4
5
0 0.2 0.4
Dis
tanc
e (k
m)
Fig. 2. An example of paste-ups of cross-correlations and variance reductions (VR) in
panels are an example of paste-ups with a band pass frequency between 0.1 and 0.4 Hz f
variance reductions for phase velocity measurements with various frequency and phas
(Haney, 2009), and seismology (Shapiro and Campillo, 2004).Ambient noise has extensively been employed to image theinterior of the Earth in the local (Brenguier et al., 2007), regional(Shapiro et al., 2005; Lin et al., 2008; Nishida et al., 2008), andglobal (Nishida et al., 2009) scales. The principle of imaging withambient noise is that the cross-correlation of ambient noiserecorded at two stations represents the wavefield as if a sourceis at one station and a receiver is at the other. The obtainedwavefield thus takes advantages of the sensitivity to the structureonly around the station pair.
2. Data analysis
We used vertical seismograms recorded between July 2005and July 2007 at 89 stations, 18 of which are broadband sensorsand 71 short-period with a natural frequency of 1 Hz. We takeadvantages of better station coverage by choosing this timeperiod when 6 broadband and 11 short-period sensors weretemporarily deployed. We first divided the whole region intothree subregions (Fig. 1) according to a priori knowledge from aprevious study (Nishida et al., 2008) which shows a clear velocitycontrast with slower northwest and faster southeast. We thentook cross-correlations of all possible pairs within each subregionwith appropriate preprocessing including instrument corrections,mean removal, detrend, and spectral whitening, as is done byprevious studies (Bensen et al., 2007).
Fig. 2A–C depict cross-correlations as a function of interstationdistance for each region. They show the propagation of theRayleigh wave trains in all regions but with different speed; the
20 40
sec)
0
20
40
60
80
100
-40 -20 0 20 40
Time (sec)
Pha
se v
eloc
ity (k
m/s
)
Frequency (Hz)
0
1
2
3
4
5
0 0.2 0.4 0.6 0.8 1 0
0.2
0.4
0.6
0.8
1
y (Hz) 0.6 0.8 1
Dis
tanc
e (k
m)
VR
phase velocity measurements as a function of frequency and phase velocity. Top
or (A) region 1, (B) region 2, and (C) region 3, respectively. The bottom panels show
e velocities for (D) region 1, (E) region 2, and (F) region 3, respectively.
Y. Nagaoka et al. / Earth and Planetary Science Letters 333-334 (2012) 1–8 3
main wave packet travels by about 3 km/s in region 2, muchfaster than that in other regions of about 2 km/s, consistent withprevious studies (Nishida et al., 2008). Note that wave packetstravel shorter distance in region 3 (Fig. 2C) than the region 2(Fig. 2B) because the region 3 is likely to exhibit more attenuationand scattering by thicker sediment as suggested by low velocitybodies (Nishida et al., 2008). Visual inspection of Fig. 2 also showsthat the causal and the acausal parts of the cross-correlations aresimilar, endorsing the quality of the obtained cross-correlations.
In each region, reference dispersion curves were measuredwith assumptions (1) the seismic structure within the networkcan be approximated to vary only with depth, (2) the amplitude ofthe ambient noise is azimuthally isotropic, and (3) the phase ofambient noise is azimuthally uncorrelated. In that situation, thereal part of the cross-spectrum Fsyn (representation of the cross-correlation, the causal and acausal part stacked, in frequencydomain) at given frequency o can be given by a function of thepower spectrum Aðc,oÞ, interstation distance r, and phase velocityc as (Aki, 1957; Chavez-Garcia and Luzon, 2005; Sanchez-Sesmaand Campillo, 2006)
Fsynðc,oÞ ¼ Aðc,oÞJ0
oc
r� �
ð1Þ
where A and c are unknowns and J0 represents the zeroth orderBessel function of the first kind. The reference dispersion curve foreach region is defined as the variance reduction:
VRðc,oÞ ¼ 1�
Pij½F
obsij ðc,oÞ�Fsyn
ij ðc,oÞ�2Pij½F
obsij ðc,oÞ�2
ð2Þ
1.0 1.5 2.0 2.5 3.0 3.5Phase velocity (km/s)
0 10 20 30
km
138°00' 138°30' 139°00'
36°00'
36°30'
37°00'
1.0 1.5 2.0Phase ve
138°00' 1
36°00'
36°30'
37°00'
138°00' 138°30' 139°00'
36°00'
36°30'
37°00'
0 10 20 30
km
−20 −10 0 10 20Travel time anomaly (%)
138°00' 1
36°00'
36°30'
37°00'
−20 −10Travel time
Fig. 3. Perturbations in travel times with respect to the reference dispersions and the obta
expected from the reference dispersions and the obtained phase velocities at 0.1–0.2 Hz. (B
is maximized (Fig. 2D–F), where the subscript ij represents thepair of station i and j, the superscript obs and syn represent theobserved and calculated cross spectrums, respectively. Note thatgiven c and o, Aðc,oÞ is estimated by the least-squares method tomaximize VR. The obtained reference dispersion curves depictthat the phase velocity in region 2 is faster than the others, againconsistent with a previous study (Nishida et al., 2008) and visualinspection of cross-correlations. The dense deployment of seism-ometers in region 1 allows us to measure the phase velocity up to0.4 Hz, higher than the other regions. The less dense deploymentin regions 2 and 3 limits us to measure reliable phase velocitiesonly up to 0.3 and 0.25 Hz, respectively.
3. Phase velocity map
Once the reference phase velocity is obtained, the travel timeanomaly is measured as a perturbation from the reference phasevelocity for all available station pairs (Fig. 3A–C). Phase velocityanomalies were measured for three frequency bands in region 1(0.1–0.2, 0.15–0.3, and 0.2–0.4 Hz), three frequency bands inregion 2 (0.1–0.2, 0.15–0.3, and 0.2–0.4 Hz) and one frequencyband in region 3 (0.1–0.2 Hz), respectively.
We measured the phase shift between the observed waveformand the synthetic waveform from the reference dispersion curvein frequency domain. The optimum phase velocity perturbation gfor each given frequency band minimizes:
SðB,gÞ ¼XoðFobs
ðr,oÞ�BðgÞFsynðr,o,gÞÞ2 ð3Þ
2.5 3.0 3.5locity (km/s)
0 10 20 30
km
38°30' 139°00'
1.0 1.5 2.0 2.5 3.0 3.5Phase velocity (km/s)
0 10 20 30
km
138°00' 138°30' 139°00'
36°00'
36°30'
37°00'
38°30' 139°00'
0 10 20 30
km
0 10 20 anomaly (%)
138°00' 138°30' 139°00'
36°00'
36°30'
37°00'
0 10 20 30
km
−20 −10 0 10 20Travel time anomaly (%)
ined phase velocities. (A) and (D) represent the travel time perturbations from those
) and (E) are the same as (A) and (C) but for 0.15–0.3 Hz. (C) and (F) are for 0.2–0.4 Hz.
2.0 2.5 3.0Phase velocity (km/s)
0 10 20 30
km
138°00' 138°30' 139°00'
36°00'
36°30'
37°00'
2.0 2.5 3.0Phase velocity (km/s)
0 10 20 30
km
138°00' 138°30' 139°00'
36°00'
36°30'
37°00'
2.0 2.5 3.0Phase velocity (km/s)
0 10 20 30
km
138°00' 138°30' 139°00'
36°00'
36°30'
37°00'
2.0 2.5 3.0Phase velocity (km/s)
0 10 20 30
km
138°00' 138°30' 139°00'
36°00'
36°30'
37°00'
tlusertupni
input result
Fig. 4. Results of a spike resolution test to demonstrate that the low phase velocity region we found is not numerical artifact but a real feature. (A) A spike resolution test
with velocity perturbations at where seismic sites are dense. The left panel represents the input velocity perturbations. The baselines in which the cross-correlations are
calculated are shown in Fig. 3. The right panel the inverted velocity anomalies. These panels indicate that the velocity perturbations in densely instrumented regions are
well recovered. (B) Same as (A), except that the input velocity perturbations are in sparsely instrumented regions, indicating that the velocity perturbations in sparsely
instrumented regions cannot be recovered.
Y. Nagaoka et al. / Earth and Planetary Science Letters 333-334 (2012) 1–84
where
Fsynðr,o,gÞ ¼ AðoÞJ0
oð1þgÞc0ðoÞ
r
� �ð4Þ
where we implicitly assume small perturbations, c0 representsthe reference phase velocity at frequency o, and B represents theamplitude correction factor to be estimated. Introduction of B
indicates that we discard the amplitude information.For each path, the velocity anomalies g were measured
from the lowest frequency band. At the lowest frequency band,0.1–0.2 Hz, we searched g assuming that the perturbation is lessthan 10%. In subsequent frequency bands, we assumed theperturbation is between g0�0:05 and g0 þ0:05 where g0 is theperturbation at the previous frequency band. When B is estimatedto be negative or no minimum is found within the search range ofthe phase velocity perturbation we judged that the measurementis failed and stopped measuring the phase velocity anomaly in thefollowing frequency bands.
Fig. 3D shows that the obtained phase velocity for a frequencyrange 0.1–0.2 Hz is slower to the west of the volcano by up to10%. Furthermore, the measured phase velocities between stationpairs were inverted for two-dimensional phase velocity maps.
We applied an iterative nonlinear inversion (Rawlinson andSambridge, 2003) to estimate phase velocities of a given frequencyrange at grid points with a spacing of 0.03 degrees or approxi-mately 3 km. Taking the phase velocity map from a previous study(Nishida et al., 2008) as a initial model, we iterated the inversionfor ten times to get the final model. The spatial variations of theRayleigh wave phase velocity at a frequency 0.1–0.2 Hz show anegative velocity anomaly by up to � 20% to the � 10 km west ofthe summit (Fig. 3D).
Now a question arises as to whether the localized low velocityseen in the 0.1–0.2 Hz phase velocity map can be resolved from theinversion procedure. To address this question, we conducted aspike resolution test (Aster et al., 2005; Nolet, 2008) by the methodof Rawlinson and Sambridge (2003). The result demonstrates that
Y. Nagaoka et al. / Earth and Planetary Science Letters 333-334 (2012) 1–8 5
our data have enough power to resolve the low velocity of this size(Fig. 4). While the mapped low phase velocity at 0.1–0.2 Hz has aradius of about 5 km (Fig. 3D), the actual size would be smaller
0
5
10
15
20
25
30
-0.5 0 0.5 1
Dep
th (k
m)
VsVp
ρ
0
5
10
15
20
25
30
-0.5 0
Dep
th (k
m)
Fig. 5. Sensitivity kernels for the density (green), S-wave velocity (red), and P-wave
(For interpretation of the references to color in this figure caption, the reader is referr
Fig. 6. Shear velocity around Mt. Asama. Upper panels represent the plan view of the
vertical view of the S-wave velocity along the A–B and C–D baselines, respectively, sh
because the analysis in this study does not take the effect of finite-frequency into account. Under an assumption of infinite frequency,which is employed here, geometrical rays have a resolution to the
0.5 1
VsVp
ρ
0
5
10
15
20
25
30
-0.5 0 0.5 1
Dep
th (k
m)
VsVp
ρ
velocity (blue). (A) Sensitivity kernels at 0.15 Hz, (B) 0.225 Hz, and (C) 0.3 Hz.
ed to the web version of this article.)
S-wave velocity at depths of 0 and 5 km, respectively. Lower panels represent the
own in the upper panels.
Y. Nagaoka et al. / Earth and Planetary Science Letters 333-334 (2012) 1–86
structure only along the raypath. In reality, however, rays withfinite frequency has a resolving power to the structure of approxi-mately one third of the first Fresnel zone width (Yoshizawa andKennett, 2002; Nishida, 2011), which is approximately 7 km for the0.1–0.2 Hz case. Because the actual velocity anomaly off thegeometrical ray but within one third of the Fresnel zone will bemapped along the geometrical ray, the mapped velocity anomalywill be broader than the real one. We thus conclude that the radiusof the velocity anomaly would actually be smaller than 5 km. Therewill be a trade-off between the size and amplitude of the velocityanomaly where a smaller-sized anomaly gives an anomaly of largeramplitude. However, the size of the velocity anomaly should not bemuch smaller than 5 km because the amplitude of the anomaly weobtained is already large (Fig. 3).
Fig. 7. Uncertainties of dispersion measurements and S-wave velocity structure at
361240N 1381360E (blue), 361260N 1381260E (green), and 361000N 1381300E (red).
The top panel denotes the dispersion curve of Nishida et al. (2008) as our initial
model (thinner lines), measured phase velocities with 1s uncertainties (dots
with error bars), and inverted dispersion curves (thicker lines). The bottom
panel denotes the S-wave velocities of Nishida et al. (2008) as our initial
model (thinner lines) and inverted S-wave velocities (thicker lines) with their
uncertainties mapped from have velocity uncertainties. (For interpretation of the
references to color in this figure caption, the reader is referred to the web version
of this article.)
4. Three-dimensional S-wave velocity structure
The regionalized dispersion curves were inverted for local 1-Dvelocity models at each point (Tarantola and Valette, 1982;Montagner, 1986). Fig. 5 shows sensitivity kernels of the modelof Nishida et al. (2008) or our initial model at 361240N 1381270E,where a low phase velocity anomaly was found (Fig. 3) for 0.15,0.225 and 0.3 Hz calculated by the method of Saito (1988) with anassumption that the P-wave velocity and density are scaled as afunction of the S-wave velocity by Brocher (2005), indicating thatlower frequency has more sensitivity to structures at depth. Fig. 6shows the 3-D S wave tomographic model from a collection of thelocal 1D S-wave velocity structures. This figure indicates lowS-wave velocity at depths of 5–10 km. Such low velocity region tothe west of the summit disappears at higher frequencies (Fig. 3Eand F). Instead, Fig. 6 displays high S-wave velocity anomalies atshallower depths ðo3 kmÞ around the repeating dyking regioninferred from an active source seismic exploration (Aoki et al.,2009). The S-wave velocity structure is consistent with theP-wave velocity derived from an active source seismic experiment(Aoki et al., 2009, Fig. 4). In addition, the inferred S-wave velocityat depth of sea level shows the slower phase velocity to the east ofthe summit than the west, which is also consistent with the resultfrom an active source seismic experiment (Aoki et al., 2009).
Fig. 7 shows the difference among the measured phasevelocities, initial phase velocities and corresponding S-wavevelocities (Nishida et al., 2008), and the final phase velocitiesand corresponding S-wave velocities with their uncertainties atthree different locations. Uncertainties of the phase velocities aretaken to be 0.05 km/s. Choice of this value is rather subjectivebecause it is difficult to make a rigid assessment of the errors inphase velocities when the estimation of cross-spectra as a func-tion of interstation distance are not independent of each other.The absolute value of the uncertainties in S-wave velocities couldthus be biased but the relative magnitude of those with depth willbe preserved.
The mapped uncertainties of S-wave velocities are shown inthe bottom panel of Fig. 7 as error bars which represent a
posteriori covariance matrix of model parameters (Tarantola andValette, 1982); they are on the order of 0.1 km/s at 2 km belowsea level or deeper with higher uncertainties at shallower levels.This indicates that the low velocity body to the west of thesummit is not a computational artifact but a robust feature. Alsothe comparison between initial and final models shows that thelow velocity body is recovered from a initial model without suchanomalies.
What makes the low S-wave velocity from 5 to 10 km? Herewe suggest with some evidence that it is magma chamber. Weobserved tilt motions during the February 2009 eruption(Matsuzawa et al., 2009; Miyamura et al., 2009), much of which
are obtained from the horizontal record of broadband seism-ometers by taking advantage of the capability of broadbandseismometers in recording tilt motions (Rodgers, 1968; Graizer,
138°25' 138°30' 138°35'
36°20'
36°25'
36°30'
0 5
km
obscal
0.05 microrad
Magmachamber?
0
5
tsaEtseWSummit
Dep
th (k
m)
10
Dike
Fig. 8. Tilt motion associated with the 2 February 2009 eruption and the schematics of magma pathway. (A) Modeling of tilt motions associated with an eruption on
2 February 2009. The optimum location of intruded dike is shown in blue. Black and pink vectors represent observed and calculated tilting directions. Summit of Mt. Asama
is shown by a red triangle. (B) Schematics of the magma pathway integrating seismic and geodetic observations. Black dots represent seismicity. Magma starts to ascent
from the magma chamber as a dike. The intruded magma then follows a path represented by seismicity. (For interpretation of the references to color in this figure caption,
the reader is referred to the web version of this article.)
Y. Nagaoka et al. / Earth and Planetary Science Letters 333-334 (2012) 1–8 7
2006; Maeda et al., 2011). We excluded tilt records near site thesummit with tilting up to 7 mrad toward the summit because theyreflect the mass loss at the summit associated with the eruption,which is outside of our scope. Other sites generally tilt toward tothe west of the summit by up to 1:1 mrad (Fig. 8A), implying asource offset to the west. We modeled the observed deformationfield by a deflating dike embedded in a homogeneous, elastic, andisotropic halfspace (Okada, 1985) by applying Simulated Anneal-ing (Cervelli et al., 2001), a Monte-Carlo based optimization of anonlinear problem, to search for the optimum source geometryand amount of dike deflation. The optimum dike is at its top atabout 1 km below sea level with northwest-southeast strike(Fig. 8A), the location of which is right at above the low S-wavevelocity anomaly.
The existence of a dike participating in an eruption suggeststhat the low S-wave velocity anomaly represents the magmachamber beneath the dike deflating during the 2009 eruption.The crustal magma chamber has not been imaged at all (Aokiet al., 2009) or only as a broad low velocity zone (Lees, 2007) byconventional seismic imaging studies because of technical diffi-culties while our method is capable of delineating the size of acrustal magma chamber. Our results also demonstrate thatambient noise cross-correlations are capable of imaging ananomaly of a radius of � 5 km at a depth of 5 km or deeper withan optimal station coverage.
5. Discussion
Taken discussion above together and considering that thelocation of the deflating dike during the 2009 eruption almostcoincides with that intruded during previous eruptions (Takeoet al., 2006; Aoki et al., 2009), Fig. 8B depicts the magma pathwayin the upper crust beneath Mt. Asama. When magma rises up tothe upper crust, it is stored at the magma chamber. As discussedabove, the magma chamber is � 8 km offset to the west from thesummit and represented by a low velocity region. Then, themagma further moves up to a depth of � 1 km below sea levelas a dike. The shallow seismic and resistivity structures (Aizawaet al., 2008; Aoki et al., 2009) suggest that the magma pathway is
subject to structural controls to make a winding path to reach thesurface. Such interdisciplinary approach enables us to gain aunified understanding of the magma plumbing system of Mt.Asama from a crustal magma chamber to the surface.
Although many volcanoes have their magma chamber rightbeneath the summit (e.g., Masterlark et al., 2010), a magmachamber horizontally offset from the summit of a volcano couldbe a ubiquitous feature. In fact, tiltmeters in Kirishima volcano,southwest Japan, tilted toward a point � 8 km to the west of thesummit in response to an eruption on 26 January 2011. This pointcoincides with the location of a deflation source derived from GPSobservations. Although it is impossible to verify whether thisdeflation source is characterized by a low velocity body becauseof the lack in dense seismic observations, this tilt observationsdemonstrate the ubiquity of a horizontally offset magma reservoirand that the magma pathway of an active volcanoes is largelysubject to structural controls to make a winding path to make thesurface. Dense seismic deployment such as in Mt. Asama willenable the detection of such magma storages in other volcanoes,leading us to enhance the understanding of the mechanics ofmagma transport beneath active volcanoes.
Acknowledgments
Y.A., K.N., and M.T. are supported by the Grant-in-Aid forScientific Research (22540431, 21740319, and 22740289, respec-tively) from Japan Society for the Promotion of Science. Reviewsby Matthew Haney and an anonymous reviewer improved themanuscript.
References
Aizawa, K., Ogawa, Y., Hashimoto, T., Koyama, T., Kanda, W., Yamaya, Y., Mishina, M.,Kagiyama, T., 2008. Shallow resistivity structure of Asama volcano and itsimplications for magma ascent process in the 2004 eruption. J. Volcanol.Geotherm. Res. 173, 165–177, http://dx.doi.org/10.1016/j.jvolgeores.2008.01.016.
Aki, K., 1957. Space and time spectra of stationary stochastic waves, withspecial reference to microtremors. Bull. Earthquake Res. Inst. Univ. Tokyo 35,415–456.
Aoki, Y., Takeo, M., Aoyama, H., Fujimatsu, J., Matsumoto, S., Miyamachi, H.,Nakamichi, H., Ohkura, T., Ohminato, T., Oikawa, J., Tanada, R., Tsutsui, T.,
Y. Nagaoka et al. / Earth and Planetary Science Letters 333-334 (2012) 1–88
Yamamoto, B.L.N., Yamamoto, M., Yamasato, H., Yamawaki, T., 2009. P-wavevelocity structure beneath Asama Volcano, Japan, inferred from active sourceseismic experiment. J. Volcanol. Geotherm. Res. 187, 272–277, http://dx.doi.org/10.1016/j.jvolgeores.2009.09.004.
Aster, R.C., Borchers, B., Thurber, C.H., 2005. Parameter Estimation and InverseProblems. Elsevier.
Bensen, G.D., Ritzwoller, M.H., Barmin, M.P., Levshin, A.L., Lin, F., Moschetti, M.P.,Shapiro, N.M., Yang, Y., 2007. Processing seismic ambient noise data to obtain
reliable broad-band surface wave dispersion measurements. Geophys. J. Int.169, 1239–1260, http://dx.doi.org/10.1111/j.1365-246X.2007.03374.x.
Brenguier, F., Shapiro, N.M., Campillo, M., Nercessian, A., Ferrazzini, V., 2007. 3-Dsurface wave tomography of the Piton de la Fournaise volcano using seismicnoise correlations. Geophys. Res. Lett. 34, L02305, http://dx.doi.org/10.1029/
2006GL028586.Brocher, T., 2005. Empirical relations between elastic wave speeds and density in
the Earth’s crust. Bull. Seismol. Soc. Am. 95, 2081–2092, http://dx.doi.org/10.1785/0120050077.
Cervelli, P., Murray, M.H., Segall, P., Aoki, Y., Kato, T., 2001. Estimating sourceparameters from deformation data, with an application to the March 1997earthquake swarm off the Izu Peninsula, Japan. J. Geophys. Res. 106,
11,217–11,237, http://dx.doi.org/10.1029/2000JB900399.Chavez-Garcia, F.J., Luzon, F., 2005. On the correlation of seismic microtremors.
J. Geophys. Res. 110, B11313, http://dx.doi.org/10.1029/2005JB003671.Duvall, T.L., Jefferies, S.M., Harvey, J.W., Pomerantz, M.A., 1993. Time-distance
helioseismology. Nature 362, 430–432, http://dx.doi.org/10.1038/362430a0.Graizer, V., 2006. Tilts in strong ground motion. Bull. Seismol. Soc. Am. 96,
2090–2102, http://dx.doi.org/10.1785/0120060065.Haney, M.M., 2009. Infrasonic ambient noise interferometry from correlations
of microbaroms. Geophys. Res. Lett. 36, L19808, http://dx.doi.org/10.1029/
2009GL040169.Lees, J.M., 2007. Seismic tomography of magmatic systems. J. Volcanol. Geotherm.
Res. 167, 37–56, http://dx.doi.org/10.1016/j.volgeores.2007.06.008.Lin, F.-C., Moschetti, M.P., Ritzwoller, M.H., 2008. Surface wave tomography of the
western United States from ambient seismic noise: Rayleigh and Love wavephase velocity maps. Geophys. J. Int. 173, 281–298, http://dx.doi.org/10.1111/j.1365-246X.2008.03720.x.
Lobkis, O.I., Weaver, R.L., 2001. On the emergence of the Green’s function in thecorrelations of a diffuse field. J. Acoust. Soc. Am. 110, 3011–3017.
Maeda, Y., Takeo, M., Ohminato, T., 2011. A waveform inversion including tilt:method and simple tests. Geophys. J. Int. 184, 907–918, http://dx.doi.org/
10.1111/j.1365-246X.2010.04892.x.Matsuzawa, T., Obara, K., Hirose, H., Kumagai, H., Ukawa, M., 2009. Tilt Change
Associated with the Eruption of the Asama Volcano on February 2, 2009Observed by a Hi-net High-Sensitivity Accelerometer. Japan Geoscience UnionMeeting, Abstract V159-P033.
Masterlark, T., Haney, M., Dickinson, H., Fournier, T., Searcy, C., 2010. Rheologic andstructural controls on the deformation of Okmok volcano, Alaska: FEM, InSAR,
and ambient noise tomography. J. Geophys. Res. 115, B02409, http://dx.doi.org/10.1029/2009JB006324.
Miyamura, J., et al., 2009. Eruptive Activity at Asamayama in February 2009. JapanGeoscience Union Meeting, Abstract V159-019.
Montagner, J.-P., 1986. Three-dimensional structure of the Indian Ocean inferredfrom long period surface waves. Geophys. Res. Lett. 13, 315–318, http://dx.doi.org/10.1029/GL013i004p00315.
Nakajima, J., Hasegawa, A., 2007. Deep crustal structure along the Niigata-KobeTectonic Zone, Japan: its origin and segmentation. Earth Planet. Space 59,e5–e8.
Newhall, C.G., Self, S., 1982. The Volcanic Explosivity Index (VEI): an estimate ofexplosive magnitude for historical volcanism. J. Geophys. Res. 87, 1231–1238,http://dx.doi.org/10.1029/JC087iC02p01231.
Nishida, K., 2011. Two-dimensional sensitivity kernels for cross-correlation func-tions of background surface waves. Comp. Rend. Geosci. 343, 584–590, http://dx.doi.org/10.1016/j.crte.2011.02.004.
Nishida, K., Kawakatsu, H., Obara, K., 2008. Three-dimensional crustalS-wave velocity structure in Japan using microseismic data recorded by Hi-net tiltmeters. J. Geophys. Res. 113, B10302, http://dx.doi.org/10.1029/2007JB005395.
Nishida, K., Montagner, J.P., Kawakatsu, H., 2009. Global surface wave tomographyusing seismic hum. Science 326, 112, http://dx.doi.org/10.1126/science.1176389.
Nolet, G., 2008. A Breviary of Seismic Tomography: Imaging the Interior of theEarth and the Sun. Cambridge University Press, Cambridge, UK.
Okada, Y., 1985. Surface deformation due to shear and tensile faults in a half-space. Bull. Seismol. Soc. Am. 75, 1135–1154.
Petford, N., 2003. Rheology of granitic magmas during ascent and emplacement.Ann. Rev. Earth Planet. Sci. 31, 399–427, http://dx.doi.org/10.1146/annurev.earth.31.100901.141352.
Rawlinson, N., Sambridge, M., 2003. Irregular interface parametrization in 3-Dwide-angle seismic traveltime tomography. Geophys. J. Int. 155, 79–92, http://dx.doi.org/10.1046/j.1365-246X.2003.01983.x.
Rodgers, P.W., 1968. The response of the horizontal pendulum seismometer toRayleigh and Love waves, tilt, and free oscillations of the Earth. Bull. Seismol.Soc. Am. 58, 1384–1406.
Saito, M., 1988. DISPER80: a subroutine package for the calculation of seismicnormal mode solutions. In: Doombos, D.J. (Ed.), Seismological Algorithms,pp. 293–319.
Sanchez-Sesma, F.J., Campillo, M., 2006. Retrieval of the Green’s function fromcross correlation: the canonical elastic problem. Bull. Seismol. Soc. Am. 96,1182–1191, http://dx.doi.org/10.1785/0120050181.
Schmincke, H.-U., 2006. Volcanism. Springer.Shapiro, N.M., Campillo, M., 2004. Emergence of broadband Rayleigh waves
from correlations of the ambient seismic noise. Geophys. Res. Lett. 31,L07614, http://dx.doi.org/10.1029/2004GL019491.
Shapiro, N.M., Campillo, M., Stehly, L., Ritzwoller, M.H., 2005. High-resolutionsurface-wave tomography from ambient seismic noise. Science 307,1615–1618, http://dx.doi.org/10.1126/science.1108339.
Takeo, M., Aoki, Y., Ohminato, T., Yamamoto, M., 2006. Magma supply path beneathMt. Asama volcano, Japan. Geophys. Res. Lett. 33, L15310, http://dx.doi.org/10.1029/2006GL026247.
Tarantola, A., Valette, B., 1982. Generalized nonlinear inverse problems solvedusing the Least Squares criterion. Rev. Geophys. Space Phys. 20, 219–232,http://dx.doi.org/10.1029/RG020i002p00219.
Yoshizawa, K., Kennett, B.L.N., 2002. Determination of the influence zone forsurface wave paths. Geophys. J. Int. 149, 440–453, http://dx.doi.org/10.1046/j.1365-246X.2002.01659.x.