Journal of Volcanology and Geothermal Research 175 (2008) 231–240

Contents lists available at ScienceDirect

Journal of Volcanology and Geothermal Research

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j vo l g e o r e s

Resistivity structure of Unzen Volcano derived from time domain electromagnetic

(TDEM) survey
Wahyu Srigutomo a,⁎, Tsuneomi Kagiyama a, Wataru Kanda b, Hiroshi Munekane c, Takeshi Hashimoto d,
Yoshikazu Tanaka a, Hisashi Utada d,e, Mitsuru Utsugi a
a
Institute for Geothermal Sciences, Graduate School of Science, Kyoto University, Minamiaso, Kumamoto 869-1404, Japan
b
Disaster Prevention Research Institute, Kyoto University, Sakurajima, Kagoshima 891-1419, Japan
c
Geographycal Survey Institute, Kitago 1, Tsukuba, Ibaragi 305-0811, Japan
d
Graduate School of Science, Hokkaido University, Sapporo, Hokkaido 869-1404, Japan
e
Earthquake Research Institute, University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-0032, Japan

A R T I C L E I N F O A B S T R A C T

Article history: Time-domain electromagnetic (TDEM) surveys were carried out around Unzen Volcano, Shimabara
Accepted 24 March 2008 Peninsula, South-west Japan in 2001 and 2002 in the eastern part of the peninsula. The surveys were a
Available online 20 April 2008 completion of the previously interpreted TDEM data collected from surveys in 1995 in the western part of the
peninsula. Interpretation of the recorded transient magnetic fields resulted in a general feature of the
Keywords:
resistivity structure which are composed of three main layers: a resistive surface layer, a conductive layer
time-domain electromagnetic survey
Unzen Volcano
(below 10 Ωm) from about sea level to 2.5 km below the surface, and a third resistive layer. The conductive
resistivity structure layer is considered to be a complex of water-saturated and altered layers which spreads widely beneath
conductance distribution Shimabara Peninsula. The spatial conductance distribution shows a W–E trending pattern of high
conductance from Tachibana Bay to the summit of Unzen (Fugen–Dake). This evidence indicates that
volcanic gas was supplied to shallow layer in the western part of the peninsula, which is in accordance with
the path of magma migration derived from seismic and deformation studies. Another high conductance zone
was also found near Mayu–Yama in the eastern part of Shimabara Peninsula, indicating magma degassing
process due to the possibility of the presence of a magma system beneath it.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction including Heisei–Shinzan, a huge lava dome in the eastern flank of

Unzen is a composite volcano composed of a group of volcanic
cones, situated in the center of Shimabara Peninsula, Kyushu, SW
Japan (32.75 N, 130.30 E). It has developed within the E–W trending
regional tectonic graben system called Unzen Graben, of which
basement has subsided more than 1000 m below sea level due to N–S
extensional stress and is bounded by Chijiwa Fault in the north and
Futsu–Kanahama Faults in the south (Fig. 1). Unzen Volcano can be
divided into the Older Unzen (0.15–0.5 Ma) and the Younger Unzen
(0–0.15 Ma) volcanoes. The Younger Unzen volcano consists of No–
Dake, Myoken–Dake, Fugen–Dake and Mayu–Yama volcanoes, in
order from the oldest to the youngest (Hoshizumi et al., 1999). After
the 198 yr of dormancy, Fugen–Dake (Mt. Fugen), which is the main
peak of Unzen Volcano, began small phreatic eruption in November
1990, followed by successive different types of eruptions, and finally
ended in 1995 with a total volume of erupted lava of 2.1 × 108 km3
⁎ Corresponding author. Permanent address: Physics Dept., Bandung Institute of
Technology, Jl. Ganesha 10 Bandung 40132, Indonesia.
E-mail address: wahyu@fi.itb.ac.id (W. Srigutomo).
Fugen–Dake (Nakada et al., 1999).
Several models on magma supply system from deeper structure of
the region have been proposed which can be represented by two main
models: First, a W–E ascending magma supply system concentrated at
about 15 km beneath Tachibana bay and reached the shallow part of
Unzen beneath Fugen–Dake based on volcano-tectonic seismicity
(Umakoshi et al., 2001) supported by the results of leveling surveys
(Ishihara, 1993) and GPS surveys (Nishi et al., 1999) by which the main
pressure source was located to the west of the summit. The other one is a
magma supply system proposed by Ohmi and Lees (1995), inferred from
the study of three-dimensional P- and S-wave velocity structure below
the volcano. It was proposed that a NW–SE trending elongated system of
dykes starting at 30 km northwest of Unzen, striking to the southeast
and dipping to about 15 km below the Shimabara bay area.
Although the chemical and physical properties of dacite magma of
Unzen were similar to those of Mount Pinatubo and Mount St. Helens,
which erupted explosively in 1991 and 1980, respectively, the recent
eruption of Unzen was effusive (Nakada and Motomura, 1999; Nakada,
2000), and the difference was caused by the difference of the degassing
efficiency of the ascending magma. The volcanic gases may have escaped

0377-0273/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jvolgeores.2008.03.033
232 W. Srigutomo et al. / Journal of Volcanology and Geothermal Research 175 (2008) 231–240

Fig. 1. Location of Unzen Volcano in the center of Shimabara Peninsula, Kyushu Island, Japan. It is bounded by E–W trending Chijiwa Fault and Kanahama–Futsu Faults system and
located at Unzen Graben.

from the volcanic edifice to the surrounding fractured rock and aquifer
through the conduit wall beneath the summit ((Nakada and Motomura,
1999).
Extensive magneto–telluric (MT) observations had been conducted
around Unzen volcano since the first eruption on 17 November 1990
(Kagiyama et al., 1999). It was revealed that the conductivity structure
beneath the volcano comprises a very resistive layer of several
hundreds to several thousands Ωm from the surface to about 500 m
above sea level), a conductive second layer (about 10 Ωm), followed by
a third resistive layer which begins at about 1 ± 0.5 km below seal level
(about 100 Ωm), and another conductive layer (10–30 Ωm) starting
from depths of 10 ± 3 km below sea level. The conductive layer beneath
the surface layer was interpreted as water-saturated layer, which had
played important role in generating precursory phenomena and in
controlling the temporal change of the eruption types. Kagiyama et al.
(1999) explained that at a rate of about 20 m/day a magma column
reached the bottom of the saturated layer causing an intense volcanic
tremor and phreatic eruptions in January–February 1991, followed by
a phreatomagmatic explosion at the time the magma approached the
upper boundary of the layer on April 9, 1991. They also suggested that
volcanic gas was continuously released when the magma began to rise
at about 10 km depth in the western part of the Shimabara Peninsula.
The gas reached and heated the saturated layer instigating the first
occurrence of volcanic tremor in July 1990. However, there is still no
evidence regarding the supply of volcanic gas into the water saturated
layer in the western part of the peninsula.
The purpose of this study is to infer information on the general
feature of resistivity structure in the shallow part of Shimabara
peninsula from the interpretation of TDEM data set. We try to assess
signatures of magmatic volatiles supply from depth to the shallow

Fig. 2. Sketch of the TDEM field setup. Upper left is the illustration of the injected current wave form (in A) and upper right is the recorded magnetic response by a receiver (in pT).
 W. Srigutomo et al. / Journal of Volcanology and Geothermal Research 175 (2008) 231–240
surrounding in the peninsula based on the revealed resistivity
structure. The data set were obtained from TDEM measurements
conducted previously in 1995 in the western part of Shimabara
peninsula (Kanda, 1997), and from the measurements performed in
2001 and 2002 in the eastern part of the peninsula.
2. TDEM survey and data interpretation
2.1. TDEM method
Electrical conductivity σ, expressed in S/m (or its reciprocal ρ, the
resistivity, expressed in Ωm) significantly depends on temperature,
rock porosity, fluid saturation, and pore fluid conductivity (e.g.
Hyndman and Shearer, 1989; Llera et al., 1990; Hermance, 1995).
Commonly, a volcanic region is characterized by the presence of a high
conductivity contrast due to the combination of the mostly cold, dry
and resistive host rock and the high temperature, conductive melted
magma reservoir and the overlying hydrothermal system associated
with it (Wright et al., 1985; Newman et al., 1985). The contrast leads to
the high effectiveness of the application of electromagnetic methods
in such region. The time domain electromagnetic (TDEM) method that
uses an artificial source of electromagnetic excitation (Kaufman and
Keller, 1983; Keller et al., 1984; Ward and Hohmann, 1988; Strack,
1992; Kanda et al., 1996) is one of the most effective electromagnetic
exploration techniques, that is highly applicable for exploring 10 km
depths in a region where artificial electromagnetic noises are high and
electrical field measurement is difficult to carry out because of steep
topography and stiff lava on the surface.
As to compile the overall TDEM surveys around Unzen Volcano
which were conducted in the western part in 1995 (Kanda, 1997) and in
the eastern part in 2001 and 2002, we will first describe the method
briefly. At Unzen, a grounded horizontal electric dipole (HED) of about
1–2 km length was utilized as the transmitter where a bipolar square
wave current with 50% duty cycle was injected to the ground. One cycle
of current was completed in 20 s, consisted of 5 s of turn-on interval and
5 s of turn-off interval in an alternating polarity (Fig. 2). The transmitter
was fed by a three phased power supply (200 V AC) and the maximum of
output current was 20 A (peak-to-peak) at 800 V. The receivers were
located at about 2–8 km from the transmitter, where the three
components of the induced magnetic fields were measured as a function
time. The receivers were fluxgate magnetic sensors (Tierra Technica U43
System) powered by 12 V car battery, with resolution of 1 pT. Earlier
Fig. 3. Locations of TDEM receivers. UZ01–UZ13 (open circles) denote location of 2001 TDEM survey, whereas N101–N112 (rectangles) denote 2002 survey. AB is the current
transmitter used in both surveys. The solid circles in the western part of Unzen and CD denote TDEM sounding locations and the current transmitter of the 1995 surveys respectively.
The normal faults are drawn based on Hoshizumi et al. (2002).
233
fluxgate magnetometers version (U36 System) were used in the 1995
surveys. The transmitter and receivers were synchronized by a stable
clock system that has accuracy of 2 × 10− 8 s.
Although this method is less affected by the near-surface
inhomogeneities (Pellerin and Hohmann, 1990; Poddar 1999), the
transmitter–receiver offsets control the trade-off between the signal-
to-noise ratio (S/N) and the length of decaying signal. Small offset will
cause the transient signal to decay rapidly, containing only informa-
tion on the shallower part of the ground conductivity distribution and
it is strongly influenced by the characteristic of the measurement
system. On the other hand, large offset will cause the S/N to decrease.
The effects of measurement system can be removed by performing the
signal deconvolution procedure in the latest stage of data processing,
and the S/N can be improved by collecting sufficiently large data cycles
to stack. To obtain the filter for the deconvolution, calibration at the
vicinity of the transmitter (about 50–100 m) should be carried out
upon assuming that the induced magnetic fields can be ignored and
that the measured waveforms are regarded purely as the convolution
of the whole measurement system with the magnetic field generated
by the transmitted current.
2.2. Survey locations and data processing
The distribution of TDEM receiver stations at Unzen is shown in
Fig. 3. Solid circles denote the measurement sites in the western part
of Shimabara Peninsula in January and December 1995, where some of
them were close to the most active summit during the eruptions of
Unzen (KZA, AZM, NTG, and KNM) and some were near to the Unzen
Spa (SKM and SUS). A grounded-wire source of 1.8 km was installed in
the northwest of the volcano, indicated by the line CD. The current
intensity were about 17 A and 15 A with the ground resistance of 40 Ω
and 45 Ω in the first and second experiments respectively. About
1000–7000 cycles of transient wave forms of 8 Hz sampling rate were
collected in 2–4 days from the first survey, while about 2500–5000
cycles were obtained in the second. About 180–300 cycles of transient
signals measured with 128 Hz sampling rate were also collected to
resolve the shallower part of the conductivity structure. Only the early
time part of the 128 Hz is required for the analysis so the
measurements were carried out for only few hours.
The surveys in the eastern part were carried out in February–
March 2001 and 2002. The receivers were distributed as mainly to
cover the eastern part of Shimabara Peninsula, eastward from the
234 W. Srigutomo et al. / Journal of Volcanology and Geothermal Research 175 (2008) 231–240

Fig. 4. (a) Example of the stacked data from the 1995 TDEM survey at AZM, merged from the 128 Hz and 8 Hz data and (b) The resampled AZM data and the inverted data. (c) The
stacked data from the 2002 survey at N106, obtained by merging the 128 Hz and the32 Hz data. (d) The resampled and inverted data of N106.

Fugen–Dake, forming an almost semi-circular shape of receiver
distribution, covering also Mayu–Yama (Mt. Mayu). The 2001
measurement sites are denoted by 13 open circles named as UZ01–
UZ13, while those in the 2002 survey are denoted by 12 rectangles
named as N101–N112. The closest station to the summit of Fugen–
Dake (Heisei Shinzan) is N112 in south of the summit (2.1 km from the
summit at elevation 756 m above sea level) and the farthest is N107 in
the east (5.5 km at elevation 81 m). AB in Fig. 3 is the position of the
1.4 km long transmitter's grounded edges. Approximately 7 A of
current at ground resistance of 85 Ω was injected during the surveys,

Fig. 5. Comparison between the observed data and the inversion result for several sites from the western part of Unzen (upper left) and the eastern part (upper right). Their rms
misfits for each iteration are shown in the lower diagram.
 W. Srigutomo et al. / Journal of Volcanology and Geothermal Research 175 (2008) 231–240
and transient magnetic data were collected at the total 24 sites. The
sampling rates of the receivers used were 128 Hz and 32 Hz and they
were operated for about 3–4 h and 3–4 days respectively. There were
total about 15,000 cycles of 32 Hz data and 1900 cycles of 128 Hz data
collected in the eastern part.
The data from each site were filtered using the pre-stack digital
recursive filter (Strack et al., 1989) that removes the periodic noises
without attenuating the amplitude of the transient signals and then
selectively stacked by rejecting the sporadic noises by using the
standard deviation of the recorded signals (Strack et al., 1989; Kanda
et al., 1996). For the 1995 data, after applying the deconvolution
process to remove the integrated system response effect, the early
time part of the 128 Hz data (up to 0.1875 s after the current
switching) were merged with the 8 Hz data. Similar way was applied
to the 2001 and 2002 data, where the early time part 128 Hz data (up
to 0.0234 s) were merged with the 32 Hz data. The last curve of
transient response for each station was obtained by resampling the
merged data in a log time scale. The examples of the merged and
resampled data for AZM (in the western part of the peninsula) and
N106 (in the eastern part) are shown in Fig. 4a–d. In most of the sites,
the amplitude of the 128 Hz data and those of the higher frequency
data have small differences and the data were connected smoothly.
2.3. 1-D inversion
The forward formulation for an earth model of horizontal layers
that relates the dipole moment of a grounded horizontal wire with the
generated vertical component of the magnetic field intensity at the
surface can be found in Kaufman and Keller (1983) and Ward and
Hohmann (1988). As mentioned, the TDEM interpretation in this
paper is intended mainly to seek for the general features or large
spatial scale of the conductivity structure, assuming that the
conductivity structure can roughly be approximated using individual
1-D inversion at each site. The inverse problem of transforming the
induced magnetic field as a function of transient time into
conductivity beneath the receiver as a function of depth is nonlinear.
Hence, we used an iterative procedure that solves a linearized
approximation of the full problem at each iteration using a weighted
Fig. 6. Comparison between resistivity value from two borehole sites, USDP-1 and USDP-2 with UZ01 and UZ09 respectively.
235
least-square scheme with a smoothing constraint (Constable et al.,
1987; Smith and Booker, 1988; Farquharson and Oldenburg, 1993). The
2001 and 2002 data were interpreted by adopting the Occam-type
algorithm (Constable et al., 1987) which uses maximum smoothness
constraints, yielding a smooth model of conductivities that fitted the
observed data within a certain level of misfit. The implementation of
the algorithm is straightforward, requires only the nonlinear forward
model and its Jacobian. We discretized the subsurface electrical
conductivity profile from the depth from 0 m down to 10 km into 40
layers with the logarithmically increasing thicknesses and it is
terminated by a semi-infinite layer below the 10 km depth. Thus we
have N = 41 model parameters to estimate.
The 1995 data previously were interpreted by Kanda (1997), based
on Bayesian statistic (Uchida, 1993; Mitsuhata, 1994; Ogawa and
Uchida, 1996) with an assumption that the spatial derivative of the
prior distribution for the model parameter is Gaussian. A Bayesian
likelihood is introduced to evaluate the validity of the model
objectively, in which the optimum model is determined so as to
maximize the Bayesian likelihood by selecting a hyper-parameter
using the ABIC information criterion (Akaike, 1980), that combines the
data misfit and the prior information. In this inversion scheme, the
target region of 0.1 km to 15 km depths was equally devided into 30
layers.
In both inversion schemes, by assuming that the data errors are
unbiased, independent and Gaussian, the rms (root mean square)
misfit at each iteration, given by
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
rmsmisfit ¼ jj Wd  WGðmÞ jj 2 ð1Þ
M
is expected to be equal to 1. W is the diagonal weighting matrix whose
elements are the reciprocal of the uncertainty of each datum, d = (d1,
d2,…, dM)T is a set of the observed data whose total number is M, and G
(m) is a set of the calculated data which is a function of a model vector
m = (m1, m2,…, mN). Two examples of the comparison of the observed
and inverted data are shown in Figs. 4b, d, and 5. Generally the
calculated data fitted well to the observed ones except for the latter
time. This discrepancy is probably caused by the 3-D nature of the site
 236
W. Srigutomo et al. / Journal of Volcanology and Geothermal Research 175 (2008) 231–240
Fig. 7. (a) Cross section of resistivity structure at Unzen as shown in Fig. 3, Line L1 along Fugen–Dake, (b) L2 for the western part from ID2 to SUS. (c) L3 along Mayu–Yama, and (d) L4 for the southeastern part.
 W. Srigutomo et al. / Journal of Volcanology and Geothermal Research 175 (2008) 231–240
which cannot be represented by a layer. The rms misfits for each
iteration at several stations are shown in Fig. 5, for area in the western
and eastern part of Shimabara Peninsula, respectively, the value
generally tends to converge to the desired misfit tolerance.
A comparison between resistivity values from two borehole sites
namely USDP-1 and USDP-2 with the inverted results from sites UZ01
and UZ09, which were located very near from the boreholes is shown
in Fig. 6. Lithologies of the USDP-1 are mostly pyroclastic flow and
mud flow deposit, interbedded by lava blocks and pumice flow exist
between the strata, whereas the lithologies of USDP-2 are mostly mud
flow deposit and rock avalanches, with phreatomagmatic deposits
from about 600 to 800 m depth (Hoshizumi et al., 2002). The
comparison shows an acceptable agreement in values and pattern,
with decreasing tendency of resistivity with depth.
3. Results
Almost similar to the MT results, the most prominent results from
both TDEM measurements in the western and eastern part of Unzen is
the existence of the conductive layer beneath a resistive surface layer
in most of the sites. The resistivity structure can roughly be estimated
by a 3 layer model: i.e., a resistive or moderately resistive surface layer
(several tens to several hundreds Ωm) followed by a conductive layer
(b10 Ωm) and third resistive layer. The conductive layer is located at
about depths of 300 m to 3 km below the surface. They are equivalent
to depths of sea level to 2.5 km below sea level depending on the
altitude of each site. At several sites to the west of Fugen–Dake (KNM,
KT2, KT4, HKB, JJY, ID1, and ID2) and in the eastern part of Unzen
(UZ01, UZ03, UZ06, UZ13, N103, N104, N106, N107, N108, and N109)
minimum resistivity values attain 1 Ω m. The conductive fourth layer
detected at depths of ≥10 km by MT method was not found by the
present study probably because it is deeper than the depth of
penetration of TDEM method. To see the local features of the
resistivity structure in more detail, several 1-D sections denoted by
L1, L2, L3, and L4 are drawn as shown in Fig. 7. The location and
orientation of each section can be seen in Fig. 3. Along the L1 section
from the west to around the summit (Fugen–Dake), the depth of the
conductive layer from the sea level tends to decrease, whereas the
thickness of the layer increases. However, minimum value of
resistivity (1 Ω m) is distributed around KNM to KT2, indicating that
conductivity enhancement occurs at points aside from the summit.
Along L2 cross section in the western part, the conductive structure
was located at about 1to 2 km depth. The resistivity along the
northern side of this section is lower than the southern side and the
minimum value of resistivity is found at ID2 (1 Ωm). Beneath Mayu–
Yama the low resistivity layer extents to the deeper part as can be seen
at two sites, UZ02 and UZ03 in L3 section. The resistivity values show
decreasing pattern towards Mayu–Yama both from the UZ01 in the
north and UZ08 in the south. Increasing thickness of the low resistivity
layer is found from N104 near Mayu–Yama to UZ108, which is located
around Akamatsudani Fault. The L4 section in the southeastern part of
the peninsula also exhibits almost the same feature as that of the
western part. A conductive layer was found at about 1 km to 2 km in
this section. It is also shown that the resistivity value of this layer
tends to decrease at points near from Mayu–Yama.
4. Discussion
There are several possible explanations on the existence of the
extensive conductor in the shallow part of a volcanic region. The
causes may be attributed to the presence of shallow magma chamber
(Newman et al., 1985), water-saturated layer (e.g. Kagiyama et al.,
1999), hydrothermal zones (Wright et al., 1985), hydrothermally
altered minerals (e.g. Jones and Dumas, 1993; Ogawa et al., 1998; Lenat
et al., 2000), or combination of the above features. The lava emergence
took place in the vicinity of the Fugen–Dake. The emergence is
237
associated with the presence of the smallest magma chamber located
just beneath the Fugen–dake at about 1.5 km depth as inferred from
geodetic study (Ishihara, 1993). A high temperature, ductile and low-Q
body exists only within 4 km west from the summit as suggested from
the seismicity study (Umakoshi et al., 2001). These features lead to an
unlikely imagery that magma would occupy a large horizontal extent
in shallow depth in the peninsula. Aside from the summit area, there
are three manifestations of geothermal system in Shimabara Penin-
sula: i.e. Obama hot spring, Unzen hot spring, and Shimabara hot
spring. The first two are located in the southwest direction from the
summit and the latter is located to the east from summit. These
geothermal expressions indicate the presence of fluids as media to
transfer heat to the surrounding or to the surface. NEDO (1988) also
suggested the presence of high temperature hydrothermal fluids
inferred from drilling data at several sites within about 5 km west
from the summit. Based on the above features and the previous MT
results, we interpret the conductive shallow layer derived from TDEM
data as a combination of a water-saturated layer and the thermally
altered rock resulted from long term interaction between the
hydrothermal fluid and the surrounding rock. The decrease of
resistivities to their minimum values at several sites might be
attributed to supply of hot ionic gases or fluids to the water-rich layer.
TDEM method has higher sensitivity to detect a conductive layer
(Kaufman and Keller, 1983) but it has lower reliability to determine
resistive zones above a conductor in volcanic environment as shown
in Lenat et al. (2000). Furthermore, electromagnetic methods are well
perceived to be more sensitive to conductance (conductivity–thick-
ness product) rather than to conductivity variations with depth (e.g.
Jones, 1987; Kusi et al., 1998; Korja et al., 2001). Hence, we use the
conductance distribution derived from the inversion results to allow
for locating the regional anomalous areas, where the cumulative
conductivity changes substantially caused by higher supply of
magmatic gas or volatiles into the water-saturated layer and/or the
presence of alteration products. The conductance value is obtained by
vertical integration of the conductivity with depth:
Z b X
k
rdz ¼ r i zi ð2Þ
a i¼1
where σi and zi are conductivity and thickness of the ith layer
respectively, and k is the number of layer in the interval (a,b). In this
study the value a is assigned as the height of the surface, whereas b
5000 m depth, which is assumed to cover the total thickness of the
water-saturated layer as well as the altered zone in the area.
The contour of spatial conductance is shown in Fig. 8a. Two zones
of high conductance are observed within the map: an elongated zone
trending W–E starting from around Tachibana Bay toward Fugen–
Dake and a zone in the northeastern part of the map. Fugen–Dake and
its surrounding are indicated by a dashed rectangle in the center.
Conductance in this zone ranges from about 300 to 450 S. The section
of resistivity along Fugen–Dake (Fig. 7a) shows that a low resistivity
layer is shallower and also thicker beneath the summit. This shallow
conductivity region below Fugen–Dake is probably related to the
volcanic activity including the circulating hydrothermal system in a
limited area (within 1 km from the summit) as inferred from the VLF-
MT (Kagiyama et al., 1999) and self-potential (SP) survey. Hashimoto
(1997) identified that a large SP anomalies of 1000 mV in the summit
vicinity (Fig. 8b) is attributed to the streaming potential caused by
subsurface hydrothermal upflows. In this conductance map, the
decreasing tendency of conductance in the radial direction from the
summit, which can be regarded as signature of the escape of volcanic
gas to the edifice through conduit walls, is not clearly seen. Maximum
value of conductance is not found around Fugen–Dake which is
located on the eastern tip of the W–E high conductance zone. This is
238 W. Srigutomo et al. / Journal of Volcanology and Geothermal Research 175 (2008) 231–240

Fig. 8. (a) Distribution of conductance in Shimabara Peninsula from the TDEM data. Dashed line indicates the distribution of hypocenters from 1989–1991 (Umakoshi et al., 1994).
(b) Area of contour of SP anomaly in the vicinity of the summit (from Hashimoto, 1997) as denoted by a dashed rectangle around Fugen–Dake in a. The anomaly reached as high as
1000 mV near the dome. (c) Spatial distribution of CO2 flux in soil in Shimabara Peninsula (after Takahashi et al., 2004). (d) Spatial distribution δ13C of dissolved inorganic carbon
values in the groundwater in the peninsula (after Ohsawa et al., 2002).

probably due to higher contribution of ionic substances in some
locations west from the summit area.
The elongated high conductance zone trending W–E from around
Tachibana Bay to the summit can be seen in Fig. 8a, with values from
450 to 700 S. This elongated zone is in accordance with the W–E fault
systems near from Take Fault, around ID1, ID2, KT2, KT3, KNM, and
several other sites near by. This high conductance region corresponds
to high seismic region around the shore of Tachibana Bay to Take Fault
as indicated by a dashed line in Fig. 8a. The dashed line indicates the
distribution of earthquake hypocenters from 1989–1991 (Umakoshi
et al., 1994). There are no TDEM data to the north from this elongated
zone for comparing the conductance value there. However, results of
MT survey by Kagiyama et al., (1992) along the Chijiwa Fault revealed
that thick accumutation of resistive layer exists, which might confirm
that the elongated zone is bounded between two resistive zones
within the graben. The seismic activity in Unzen has long been
recorded before the 1990 eruption. Two large events of Shimabara
Earthquake (M 6.9 and M 6.5) took place during the sequence of
earthquakes in 1922. Since then earthquake occurrences took place in
all the following decades, including the earthquake swarm during
1968–1974 whose hypocenters were distributed beneath the Tachi-
bana Bay and in the Unzen Graben. Another earthquake swarm
occurred in 1984–1985, of which magnitudes varied from M 4.3 to M
5.7 and widely distributed in the western shore of the peninsula and an
earthquake at a depth of 6.6 km with magnitude of 4.2 occurred in
1985. An earthquake swarm with maximum magnitude of 3.5 also took
place in the central part of Tachibana Bay in 1987. On November 1989,
one year before the eruption, an earthquake swarm with a maximum
magnitude of 3.3 took place beneath the SW part of Tachibana Bay and
the depth of hypocenters ranged between 12 and 15 km (Umakoshi et
al., 2001; Japan Meteorological Agency, 2006). Most of earthquakes
occurred around Unzen Graben were identified as volcano-tectonic
earthquakes (Nakada et al., 1999) and all of the above frequent
earthquake occurrences were not accompanied by eruptions, which
indicated that the occurrences were related to the intrusive events of
the magma system in the western part of Shimabara Peninsula. These
earthquake occurrences before the eruption probably were seated in
many fracture zones in the western part of the peninsula, and these
fracture zones became the pathways of volcanic gases to escape and
dissolve into the water-saturated layer. The elongated high conduc-
tance zone which is related to a high ionic concentration might be
attributed to the continuous supply of magmatic volatiles that initiated
long before the eruption. Whereas, the conductance zone around
Fugen–Dake whose value lower than that of the elongated zone was
mainly due to the latest eruption activity. This also implies that during
the magma ascent process the escape route of magma degassing was
established mainly in region to the west of the summit and thus
contributed to the non-explosiveness of the recent eruption type.
Within the investigated area, a very high conductance zone occupies
the northeastern part of the investigated area, near from Mayu–Yama
 W. Srigutomo et al. / Journal of Volcanology and Geothermal Research 175 (2008) 231–240
with conductance value higher than 1000 S. In relation to the 1990–1995
eruptive activity, the presence of magmatic chamber in shallow part of
Shimabara Peninsula is related to the migration of magma from beneath
Tachibana Bay to the summit of Fugen–Dake as previously mentioned.
Therefore, high conductance in the eastern part of the investigated area
can be attributed to other causes that might not be directly correlated to
the ascending magma in the western part.
Beside salts, dissolved CO2 gas in crust fluid can have significant
impact in raising the conductivity (Nesbitt,1993). The study of magmatic
CO2 concentration, flux, and dissolved inorganic carbon from soil and
groundwater around Unzen Volcano indicated higher contribution of
magmatic gas to the eastern region of Unzen Volcano, especially around
Chijiwa Fault and Akamatsudani Faults, which implies that the volcanic
gas may migrate from magmatic source along the fracture zones
(Ohsawa et al., 2002; Takahashi et al., 2004). The high concentration of
CO2 fits with the presence of high conductance in the area as can be seen
from Fig. 8a, b, and c. Notsu et al. (2001) noted that the helium isotopic
ratio (3He/4He), which is the tracer to identify source materials of
volcanic products, shows an increasing pattern from west to east inside
the Unzen Graben with Obama hot spring at the west side, Unzen hot
spring around the center, and Shimabara hot spring at the east side. They
analyzed the isotopic ratio from the bubbling gas of the hot springs and
suggested that since the ratio should be highest in the shortest distance
from the magma, the results might indicate either the presence of
magma below Shimabara Bay or the migration of magmatic gas through
the faults system from the volcanic edifice to the Shimabara hot spring.
The spatial correlation between the high concentration of CO2, high
helium isotopic ratio, and high conductance distribution suggest that
magmatic gas is supplied to the water-saturated layer reflecting the
possibility of the presence of deep magmatic source beneath Mayu–
Yama, suppoting the magmatic model of Ohmi and Lees (1995) in the
eastern part of Shimabara Peninsula. Geological studies also suggested
the presence of an old magma in the eastern part of the peninsula that
created Mayu–Yama about 4000 years ago. This magma has not been
cooled down yet and caused the low seismicity in the eastern part of the
peninsula (Nakada, 1996).
5. Conclusions
TDEM surveys in Unzen region have revealed the large scale
resistivity structure of the Shimabara Peninsula. In the shallow part of
the of the peninsula, a conductive layer is found and distributed
widely around Shimabara Peninsula at depth ranging for about 100 m
to 2.5 km below sea level indicating the existence of a water-saturated
and altered layers in the area.
High conductivity due to the supply of volcanic gases into the
water saturated layer, represented as high conductance zones is found
between inside the Unzen Graben. A high conductance zone trending
W–E direction in the shallow part is also found at around the western
part of the peninsula toward the summit. This zone is correlated
strongly with the pattern of the frequent swarm earthquakes that
started relatively long before the 1991–1995 eruption indicating the
repeating emission of volcanic gases into the shallow water layer
through the fault system in the western part of Unzen, which affects
the degassing efficiency during the magma migration of the latest
eruption. Another zone, the highest conductance zone, is located
around the northern part of Mayu–Yama. This zone is correlated with
the high spatial distribution of volcanic gases as inferred from
geochemical studies that traced the escape of volcanic gas from a
deeper magmatic source, and might indicate the presence of magma
beneath Mayu–Yama.
Acknowledgements
The authors wish to thank Prof. Setsuya Nakada, Dr. Takao
Ohminato, Dr. Hisayoshi Shimizu, Dr. Yuji Takahashi, Earthquake
239
Research Institute,Tokyo University, for the support and encourage-
ment during the research. We are grateful to two anonymous
reviewers whose comments and revisions greatly improved the
manuscript. We also greatly appreciate Prof. Hiroshi Shimizu,
Shimabara Earthquake and Volcano Observatory, Kyushu University,
Dr. Shinya Sakanaka, Akita University and all members of Volcano
Research Center and Ocean Hemisphere Research Center, ERI, Tokyo
University for their kind assistant during the field experiments.
References
Akaike, H., 1980. Likelihood and Bayes procedures. In: Bernard, J.M., deGroot, H.M., Lindley,
D.V., Smith, A.F. (Eds.), Bayesian Statistic. University Press, Valencia, pp. 143–166.
Constable, S.C., Parker, R.L., Constable, C.G., 1987. Occam's inversion: A practical
algorithm for generating smooth models from electromagnetic sounding data.
Geophysics 52, 289–300.
Farquharson, C.G., Oldenburg, D.W., 1993. Inversion of time-domain electromagnetic
data for a horizontally layered earth. Geophys. J. Int. 114, 433–442.
Hashimoto, T., 1997. Self-potential changes and subsurface hydrothermal activity
accompanying the 1990–1995 eruption of Unzen volcano. J. Geomagn. Geoelectr.
49, 977–993.
Hermance, J.F., 1995. Electrical conductivity models of the crust and mantle. AGU Ref.
Shelf 1, Global Earth Phys., Handbook of Phys. Constants, pp. 190–205.
Hoshizumi, H., Uto, K., Matsumoto, A., Xu, S., Oguri, K., 2002. Geology of Unzen volcano
and core stratigraphy of the flanks drillings. Extended abstract volume of
International workshop on Unzen Scientific Drilling Project (USDP), pp. 4–8.
Hoshizumi, H., Uto, K., Watanabe, K., 1999. Geology and eruptive history of Unzen
volcano, Shimabara Peninsula, Kyushu, SW Japan. J. Volcanol. Geotherm. Res. 89,
81–94.
Hyndman, R.D., Shearer, P.M., 1989. Water in the lower continental crust: modelling
magnetotelluric and seismic reflection results. Geophys. J. Int. 98, 343–365.
Ishihara, K., 1993. Continuous magma supply inferred from discharge rate magma and
ground-deformation rate at Mt. Unzen, Japan. Annuals, Disas. Prev. Res., Inst. Kyoto
Univ., 36 B-1, pp. 219–230 (in Japanese).
Japan Meteorological Agency, 2006. Unzendake, National Catalogue of Active Volcanoes
in Japan, 3rd edition. , pp. 504–518 (in Japanese).
Jones, A.G., 1987. MT and reflection: an essential combination. Geophys. J. R. Astron. Soc.
89, 7–18.
Jones, A.G., Dumas, I., 1993. Electromagnetic images of a volcanic zone. Phys. Earth
Planet. Inter. 81, 289–314.
Kagiyama, T., Utada, H., Masutani, F., Yamamoto, T., Murakami, H., Tanaka, Y., Masuda, H.,
Hashimoto, T., Honkura, Y., Mishina, M., Shimizu, H., 1992. MT observations around
Unzen volcano and the estimated process of magma ascent. Report of Grant-in-Aid
Scientific Research (No. 03306009; Ohta, K.), pp. 73–86 (in Japanese).
Kagiyama, T., Utada, H., Y, amamoto, T., 1999. Magma ascent beneath Unzen Volcano,
SW Japan, deduced from the electrical resistivity structure. J. Volcanol. Geotherm.
Res. 89, 35–42.
Kanda, W., 1997. On a deep transient electromagnetic sounding-measurement and
modeling. Ph.D Thesis, The University of Tokyo, Japan.
Kanda, W., Utada, H., Mishina, M., Sumitomo, N., 1996. A deep transient EM experiment
in the northern part of Miyagi Perfecture, northeastern Japan. J. Geomagn.
Geoelectr. 48, 1265–1280.
Kaufman, A.A., Keller, G.V., 1983. Frequency and transient soundings. Elsevier,
Amsterdam.
Keller, G.V., Pritchard, J.I., Jacobson, J.J., Harthill, N., 1984. Megasource time domain
electromagnetic sounding method. Geophysics 49, 993–1009.
Korja, T., Engels, M., Zhamaletdinov, A.A., Kovtun, A.A., Palshin, N.A., Smirnov, M.Y.,
Tokarev, A.D., Asming, V.E., Vanyan, L.L., Vardaniants, I.L., BEAR working group,
2001. Crustal conductivity in Fennoscandia — a compilation of a database on crustal
conductance in the Fennoscandian Shield. Earth Planets Space 54, 535–558.
Kusi, R., White, A., Heinson, G., Milligan, P., 1998. Electromagnetic induction studies in
the Eyre Peninsula, South Australia. Geophys. J. Int. 132, 687–700.
Lenat, J.F., Fitterman, D., Jackson, D.B., Labazuy, P., 2000. Geoelectrical structure of the
central zone of Piton de la Fournaise volcano (Réunion). Bull. Volcanol. 62, 75–89.
Llera, F.J., Sato, M., Nakatsuka, K., Yokoyama, H., 1990. Temperature dependence of the
electrical resistivity of water-saturated rocks. Geophysics 55, 576–582.
Mitsuhata, Y., 1994. Application of prior information of smooth resistivity structure to 1-D
inversion of magnetotelluric data by using the ABIC minimization method. Butsuri
Tansa 47, 358–374.
Nakada, S., 1996. On the magma reservoir of Unzen Volcano. Report of Grant-in-Aid
Scientific Research (A) (No. 06306011; K. Ohta) on “Research on the preliminary survey
for Joint Exploration Experiment of Unzen Volcano”, pp. 93–102 (in Japanese).
Nakada, S., 2000. Eruption mechanisms of Unzen Volcano. In: Yamashita, T. (Ed.),
Observations of Earth's lively motions. Iwanami, Tokyo, pp. 107–128.
Nakada, S., Motomura, Y., 1999. Petrology of the 1991–1995 eruption at Unzen: effusion
pulsation and groundmass crystallization. J. Volcanol. Geotherm. Res. 89, 173–196.
Nakada, S., Shimizu, H., Ohta, K., 1999. Overview of the 1990–1995 eruption at Unzen
Volcano. J. Volcanol. Geotherm. Res. 89, 1–22.
Nesbitt, B.E., 1993. Electrical resistivities of crustal fluids. J. Geophys. Res. 98, 4301–4310.
New Agency Development Organization (NEDO), 1988. Western Area of Unzen. Report
of promotional exploration for geothermal research, p. 15 (in Japanese).
Newman, G.A., Wannamaker, P.E., Hohmann, G.W., 1985. On the detectability of crustal
magma chambers using the magnetotelluric method. Geophysics 50, 1136–1143.
240 W. Srigutomo et al. / Journal of Volcanology and Geothermal Research 175 (2008) 231–240
Nishi, H., Ono, H., Mori, H., 1999. Global positioning system measurements of ground
deformation caused by magma intrusion and lava discharge: the 1990–1995
eruption at Unzendake volcano, Kyushu, Japan. J. Volcanol. Geotherm. Res. 89,
23–34.
Notsu, K., Nakai, S., Igarashi, G., Ishibashi, J., Mori, T., Suzuki, M., Wakita, H., 2001. Spatial
distribution and temporal variation of 3He/4He in hot spring gas released from
Unzen volcanic area, Japan. J. Volcanol. Geotherm. Res. 111, 89–98.
Ogawa, Y., Uchida, T., 1996. A two-dimensional magnetotelluric inversion assuming
Gaussian static shift. Geophys. J. Int. 126, 69–76.
Ogawa, Y., Matsushima, N., Oshima, H., Takakura, S., Utsugi, U., Hirano, K., Igarashi, M.,
Doi, T., 1998. A resistivity cross-section of Usu volcano, Hokkaido Japan, By
audiomagnetotelluric soundings. Earth Planets Space 50, 339–346.
Ohmi, S., Lees, J.M., 1995. Three-dimensional P- and S-wave velocity structure below
Unzen Volcano. J. Volcanol. Geotherm. Res. 65, 1–26.
Ohsawa, S., Kazahaya, K., Yasuhara, M., Kono, T., Kitaoka, K., Yusa, Y., Yamaguchi, K.,
2002. Escape of volcanic gas into shallow groundwater system at Unzen volcano
(Japan): Evidence from chemical and stable carbon isotope composition of
dissolved inorganic carbon. Limnology 3, 169–173.
Pellerin, L., Hohmann, G.W., 1990. Transient electromagnetic inversion: a remedy for
magnetotelluric static shifts. Geophysics 55, 1242–1250.
Poddar, M., 1999. Grounded-source TEM modeling of some deep-seated 3D resistivity
structures. Geophys. Prospect. 47, 945–958.
Smith, J.T., Booker, J.R., 1988. Magnetotelluric inversion for minimum structure.
Geophysics 53, 1565–1576.
Strack, K.M., 1992. Exploration with deep transient electromagnetics. Elsevier,
Amsterdam.
Strack, K.M., Hanstein, T.H., Eilenz, H.N., 1989. LOTEM data processing for areas with
high cultural noise levels. Phys. Earth Planet. Inter. 53, 261–269.
Takahashi, H.A., Kazahaya, K., Sinohara, H., Nakamura, T., 2004. Pathways for escape of
magmatic carbon dioxide to soil air at Unzen Volcano, SW Japan. Radiocarbon 46,
491–496.
Uchida, T., 1993. Smooth 2-D inversion for magnetotelluric data based on statistical
criterion ABIC. J. Geomagn. Geoelectr. 45, 841–858.
Umakoshi, K., Shimizu, H., Matsuwo, N., 1994. Magma ascent path in the 1990–1994
eruption of Fugendake, Unzen volcano, as inferred from precisely determined
hypocentral distribution. Bull. Volcanol. Soc. Japan. 39, 223–235 (in Japanese with
English abstract).
Umakoshi, K., Shimizu, H., Matsuwo, N., 2001. Volcano-tectonic seismicity at Unzen
Volcano, Japan, 1985–1999. J. Volcanol. Geotherm. Res. 112, 117–131.
Ward, S.H., Hohmann, G.W., 1988. Electromagnetic theory for geophysical applications.
In: Nabighian, M.N. (Ed.), Electromagnetic Methods in Applied Geopjysics. Soc.
Expl. Geophys., Tulsa, vol. 1.
Wright, P.M., Ward, S.H., Ross, H.P., West, R.C., 1985. State-of-the-art geophysical
exploration for geothermal resources. Geophysics 50, 2666–2699.