Journal of Volcanology and Geothermal Research 283 (2014) 73–81

Contents lists available at ScienceDirect

Journal of Volcanology and Geothermal Research

journal homepage: www.elsevier.com/locate/jvolgeores

Short communication

Possibility of effective magma degassing into groundwater flow systems

beneath Unzen volcanic area, SW Japan, inferred from the evaluation of
volcanic gas fluxes using electrical conductivity structures
Shogo Komori a,⁎,1, Tsuneomi Kagiyama b, Jerry P. Fairley c
a
Institute of Earth Sciences, Academia Sinica, 128 Academia Road, Section 2, Nankang, Taipei 115, Taiwan, ROC
b
Aso Volcanological Laboratory, Kyoto University, Minamiaso, Kumamoto 869-1404, Japan
c
Department of Geological Sciences, University of Idaho, 825 W 7th Street, Moscow, ID 83844-3022, USA

a r t i c l e i n f o a b s t r a c t

Article history: The mass and heat budget of volcanic gases released from magma is critical to understanding a number of
Received 25 January 2014 volcanic activities, including the ease with which magma can ascend. Due to its elevated temperature and salin-
Accepted 26 May 2014 ity, the crustal electrical conductivity in a groundwater flow system increases through the addition of hydrother-
Available online 21 June 2014
mal fluids which are produced by mixing of volcanic gases with meteoric-origin water. Therefore, the spatial
extent of high electrical conductivity regions within groundwater flow systems may be used to evaluate the
Keywords:
Unzen volcano
mass flux of volcanic gases to the systems. The present study attempts to estimate the mass flux of volcanic
Electrical conductivity gases beneath the Unzen volcanic area in Southwest Japan, by developing a simple flow model of hydrothermal
Pore water fluids and applying this model to the electrical conductivity structure of the area. The estimated mass
Surface conductivity flux of volcanic gases (104.8 ± 0.3 t/yr) yields results for CO2 flux (103.1 ± 0.3 t/yr) and magma input rate
Volcanic gas flux (100.1 ± 0.3 million m3/yr) that are consistent with those estimated by geochemical and geodetic observations.
Rainfall precipitation This suggests that volcanic gases are steadily released from magma into the overlying groundwater flow system
Effusive volcanism beneath the area, and that effective degassing may be one of the factors controlling the relatively effusive style of
recent volcanism at Unzen volcano.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction quantitative constraints on the dissipation of volcanic gases into a

Volcanic areas commonly develop hydrothermal systems, in which
circulating hydrothermal fluids are a mixture of meteoric water and vol-
canic gases released from magma (Henley and Ellis, 1983). The transport
of heat and mass within these systems is of great interest, because the
degassing of magma and the dissipation of volcanic gases control
the ease of magma ascent and the explosive potential of a volcano
(Eichelberger et al., 1986; Jaupart and Allégre, 1991; Shinohara and
Kazahaya, 1995; Kagiyama, 2008). In addition, hydrothermal activity
may be controlled by the influx of hydrothermal fluids (e.g., Ingebritsen
et al., 2010; Todesco et al., 2010; Matsushima, 2011). Hydrothermal fluids
have high salinity and temperature, causing significant rock alteration.
This leads to high electrical conductivity (E.C.) of pore waters and rock
matrices (e.g., Keller and Rapolla, 1974; Revil et al., 2002; Komori et al.,
2013). Therefore, the distribution of E.C. anomalies can provide
⁎ Corresponding author at: Central 7, Higashi 1-1-1, Tsukuba, Ibaraki 305-8567, Japan.
Tel.: +81 29 861 4556.
E-mail addresses: komori.shogo@aist.go.jp, komori@earth.sinica.edu.tw (S. Komori),
kagiyama@aso.vgs.kyoto-u.ac.jp (T. Kagiyama), jfairley@uidaho.edu (J.P. Fairley).
1
Now at Geological Survey of Japan (GSJ), AIST, Central 7, Higashi 1-1-1, Tsukuba,
Ibaraki 305-8567, Japan.
groundwater flow system (e.g., Aizawa et al., 2009; Rinaldi et al., 2011).
Unzen volcanic area is situated in a volcanotectonic depression in the
Shimabara Peninsula, SW Japan (Ohta, 1973; Chida, 1979, Fig. 1a). N–S
extension has formed Unzen Graben, defined by E–W trending
faults (Hoshizumi et al., 1999). The recent stage of Unzen volcano
(0–0.15 Ma) is characterized by relatively effusive volcanism, e.g. lava-
dome formation with minor dome-collapse pyroclastic flows, and hot
spring/fumarolic activities (Ohta, 1973; Nakada et al., 1999). Kagiyama
et al. (1999) proposed that effective magma degassing into the ground-
water flow system may induce the observed phenomena, including tem-
poral changes of the chemical composition of fumaroles, and volcanic
tremors after earthquake swarms. Srigutomo et al. (2008) conducted ex-
tensive TDEM surveys, and found an E–W trending high-conductance re-
gion (or a thick high E.C. region) at the western part of Shimabara
Peninsula, as shown in Fig. 1b and c, on the basis of one-dimensional
(1-D) inversion. This region is located above the epicenter of an earth-
quake swarm that occurred during the 1990–1995 eruption (Umakoshi
et al., 1994), and also above the path of magma traveling from a western
deeper pressure source to an eastern shallower source, inferred from
geodetic surveys (e.g., Ishihara, 1993; Kohno et al., 2008). Based on
these facts, the authors considered that the high conductance was due
to discharge of volcanic gases from magma beneath the area.

http://dx.doi.org/10.1016/j.jvolgeores.2014.05.024
0377-0273/© 2014 Elsevier B.V. All rights reserved.
74 S. Komori et al. / Journal of Volcanology and Geothermal Research 283 (2014) 73–81
(a)
Fig. 1. Unzen volcano, SW Japan. (a) Precisely-estimated hypocenters during the 1990–1995 eruption and the location of Unzen volcano (modified after Umakoshi et al., 1994). During the
eruptive event, magma moved from the deep source to the shallower one, as inferred from geodetic surveys (e.g., Ishihara, 1993; Kohno et al., 2008). (b) Conductance distribution,
obtained from TDEM surveys (Srigutomo et al., 2008). The high conductance region corresponds to the epicenter during the eruptive events. (c) 1-D electrical conductivity structure of
Srigutomo et al. (2008), along the lines P–P′ and Q–Q′.
The present study develops a simple model for the dissipation of vol-
canic gases into the systems, and attempts to evaluate their mass fluxes
using the E.C. structure of Unzen volcanic area. The estimated flux is
then compared to the other geophysical and geochemical studies to as-
sess the possibility of effective magma degassing and its influence on
volcanic activity.
2. Conceptual model
2.1. Fluid flow model at Unzen volcanic area
The western part of Shimabara Peninsula had about 10 sites for the
geothermal exploration drilled by NEDO (1988). A high temperature re-
gion (greater than 200 °C) was found at a depth of a few km (Fig. 1a);
this region corresponds to the high conductance zone from Srigutomo
et al. (2008). This suggests that fluids are maintained at high tempera-
ture within the high conductance zone. Ohsawa et al. (2002) and
Ohsawa (2006) conducted chemical composition and isotopic analyses
of hot springs and fumaroles, and showed that the liquid-dominated
hydrothermal fluids (thermal waters) are maintained at a temperature
of 300 °C beneath the western part of the peninsula. These are thought
to originate from NaCl-type deep fluids derived from deeply-seated
magma. The authors also showed that most of the hydrothermal fluids
are discharged laterally outwards within the shallow part of the penin-
sula, following the groundwater flow driven by meteoric water. Further,
Ohba et al. (2008) postulated that Unzen volcano has a three-stage
degassing system, indicated by temporal variations in the chemical
composition of the volcanic gases during the 1990–1995 eruption. The
authors inferred that the first degassing occurred through the depth in-
terval of 6–4 km; these depths correspond to two pressure sources B
and C, inferred from leveling and GPS data (Kohno et al., 2008, Fig. 1a).
On basis of all the abovementioned evidence, the following simple
dissipation model of volcanic gases may successfully describe the
(b)
(c)
Unzen volcanic area, as shown in Fig. 2a. Volcanic gas is released from
the magma at the first-degassing depths, and ascends along E–W
trending fragile sections; mixing with meteoric water may occur during
its ascent, to form the thermal water. This thermal water forms the high
temperature region at the shallow part, and is moving laterally accord-
ing to a groundwater flow driven by the injection of the thermal water
and rainfall precipitation.
Corresponding to this simple model, the E.C. structure also suggests
the lateral dissipation of the volcanic gas. Fig. 1c shows the 1-D E.C.
structures along the lines P–P′ and Q–Q′. The part enclosed by a rectan-
gle represents a high E.C. region greater than 0.3 S/m. The high E.C.
region extends toward the relatively shallower part near the high con-
ductance center; with increasing distance from the center, the region
extends downward. The region along the line Q–Q′ is not extending as
great a width as that along the line P–P′. The present study used the
above two distributions of the high E.C. region for estimation of the
volcanic gas flux.
Simplified model assumptions
In many instances, volcanic and geothermal areas may experience
large amounts of precipitation, and a significant proportion of meteoric
water can infiltrate the subsurface, contributing to the groundwater
flow system. Under such conditions, forced convection, resulting from
the mass of infiltrating water, is the predominant driver behind subsur-
face fluid flow (Holzbecher and Yusa, 1995). The present study there-
fore assumes that forced convection is a dominant factor controlling
the flow regime of the whole area (i.e., recharge-driven advection),
compared to buoyancy-driven convection. According to NEDO (1988),
temperatures of all the boreholes penetrated into the depths of 1000–
1400 m were significantly below vaporization curves. In addition, ther-
mal waters have neutral pH, and contain mainly Cl , HCO− +
3 , and Na ,
with minor quantities of SO2−4 and other ions (NEDO, 1988; Ohsawa
 S. Komori et al. / Journal of Volcanology and Geothermal Research 283 (2014) 73–81 75

(a) Table 1
Symbols used in 2-D flow model.

Symbol Name Unit

x Horizontal coordinate m
y Vertical coordinate (positive upwards) m
P Fluid pressure Pa
T Temperature °C
C Fluid salinity
ΔT Temperature difference (T1 − T0) °C
ΔC Salinity difference (C1 − C0)
F0 Volcanic fluid flux m2 s−1
u Horizontal Darcy flow m s−1
v Vertical Darcy flow m s−1
ρf Fluid density kg m−3
ρm Density of saturated matrix kg m−3
μ Fluid viscosity Pa s
g Gravity acceleration m s−2
L Horizontal extent of aquifer m
H Thickness of lateral flow region m
w Area of injection of thermal waters m
k Permeability of aquifer m2
ϕ Porosity of aquifer vol.%
cf Specific heat of fluid J kg−1 °C−1
cm Specific heat of saturated matrix J kg−1 °C−1
κ Thermal diffusivity m2 s−1
al Longitudinal dispersivity m
at Transverse dispersivity m
(b)

In the Cartesian model, the spatial distributions of fluid velocity, sa-

linity, and temperature are governed by the following four equations
(cf. Table 1 for symbols):

a) the equation of continuity, assuming an incompressible fluid:

∇
u¼0 ð1Þ

where: u is the Darcy velocity vector.

b) Darcy's law:

k
u ¼ − ð∇P−ρgÞ ð2Þ
μ
Fig. 2. Simple dissipation system of volcanic gas. Thermal waters, which are formed by
mixing between ascending volcanic gases and meteoric waters, are injected into the region c) the Advection–Diffusion Equation for solute transport:
beneath a volcanic center, and laterally transported by groundwater flow toward the right
boundary. Groundwater flow is driven by the mixing of the meteoric water from the upper ∇
ðuC Þ  ϕ∇
ðD∇C Þ ¼ 0 ð3Þ
boundary and the injected thermal waters from the leftmost lower boundary.
where: D is the dispersion tensor defined by:
 2  2  2  2

 aL ϕu þ aT ϕv aT ϕu þ aL ϕv
et al., 2002; Ohsawa, 2006). The present study therefore assumes that Dxx Dxy
D¼ ; Dxx ¼ r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2  2ffi ; Dyy ¼ r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2  2ffi ;
Dyx Dyy
the thermal waters can be modeled as a liquid-phase NaCl solution. u
þ ϕv u
þ ϕv
ϕ ϕ


 ð4Þ
u v
ðaL −aT Þ
2.2. Two-dimensional model development for heat and mass transfer ϕ ϕ
Dxy ¼ Dyx ¼ r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  ffi
2 2
u
ϕ þ v
ϕ
We define a simple system of the transfer of the thermal waters as
depicted in Fig. 2b. The groundwater flow system is symmetric about
its vertical axis, and has a horizontal length L and a thickness H of the Note that molecular diffusion is assumed to be negligible compared to
lateral-flow region. The thermal waters have a salinity of C1 = C0 + these dispersion terms. This is reasonable in the present case, because
ΔC, a temperature of T1 = T0 + ΔT, and a mass flux of F0; these fluids groundwater flow (advection) dominates over species diffusion.
are injected into the leftmost bottom of the model domain with width d) the Advection–Diffusion Equation for heat transport:
w. Meteoric water, with a salinity and temperature of C0 and T0,  
infiltrates at the upper boundary with a Darcy velocity of v0. Note that ∇
ρ f c f uT  ∇
ðρm cm κ∇T Þ ¼ 0 ð5Þ
ΔC and ΔT are the differences of salinity and temperature between the
meteoric water and thermal water. No fluid flow is allowed along the
left-hand side and bottom boundaries, and horizontal fluid flow is These equations apparently have various parameters which may
assumed at the right-hand side boundary. Adiabatic conditions are as- change significantly depending on models. However, to change all the
sumed along the left-hand, right-hand, and bottom boundaries, except parameters individually is not a realistic way to examine their influ-
at the injection center. On the basis of this conceptualization of the ences on a groundwater flow system. Because this study assumes
system, simple two-dimensional transport models were developed. the simplification described previously, non-dimensionalization can
76 S. Komori et al. / Journal of Volcanology and Geothermal Research 283 (2014) 73–81

organize the parameters into fewer dimensionless ones; which can Table 2
allow us to better understand the factors controlling a flow regime Set of parameters.

and distributions of temperature and salinity. Flow-regime parameters (dimensionless)

We define following non-dimensional variables: α 10−1–102
β 5, 20, 35
ω2 0.02, 0.05, 0.1
0 x 0 y 0 k ω1 1.1
x ¼ ; y ¼ ; P ¼ P;
L H μκ ð6Þ E.C.-related parameters
T−T 0 C−C 0 0 H F 16, 25, 44[1]
θ¼ ; ξ¼ ; u ¼ u σSmin 3 × 10−4, 9 × 10−4, 3 × 10−3[2]
ΔT ΔC κ S/m
σSmax 6 × 10−2, 10−1[2] S/m
σSH 10−2[2] S/m
Eqs. (1)–(5) are non-dimensionalized as follows: Tmax1 70, 85, 100[3] °C
Tmax2 120, 135, 150[3] °C

 0 TH 150, 175, 200[3] °C
H ∂u ∂v0 Parameters for thermal waters
þ ¼0 ð7Þ
L ∂x0 ∂y0 ΔT 300 °C
ΔC 0.011, 0.018, 0.026, 0.034
0
0  Other parameters for 2-D flow model
0 H ∂p 0 ∂p ρgHk ρmcm/ρfcf 0.65[4]
u ¼− ; v ¼− 0 − κμ ð8Þ
L ∂x0 ∂y al/H 0.03[5]
at/al 0.1[6]


  [1]
The values are calculated using m = 2 and ϕ = 15, 20, and 35%. [2]cf. Takakura (2000),
H 0 ∂ξ 0 ∂ξ ∂ a H 2 0 ∂ξ aL H 0 ∂ξ
u 0 þv 0− 0 L Dxx 0 þ Dxy 0 Revil et al. (2002), Komori et al. (2010), and Komori et al. (2013). [3]cf. Komori et al.
L ∂x
∂y ∂x H L  ∂x H L ∂y (2013). [4]The value uses ρfcf = 4.2 × 106 and ρmcm = 2.7 × 106, assuming 20 vol.% of
ð9Þ
∂ a H 0 ∂ξ a H 0 ∂ξ water content in the saturated matrix. [5]cf. Gelhar et al. (1992). [6]cf. Sato and Iwasa
− 0 L Dyx 0 þ L Dyy 0 ¼ 0
∂y H L ∂x H L ∂y (2002).


02 a 02 aT 02 02 a 0 0
u þ Tv u þv 1− T u v controlled conditions of ω1, ω2, α, and β, and the distribution of fluid ve-
0 aL 0 aL 0 0 aL
Dxx ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; Dyy ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; Dxy ¼ Dyx ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi locity is obtained from the pressure distribution via Darcy's Law. Finally,
02 02 02 02 02 02 non-dimensional temperature and salinity distributions are calculated.
u þv u þv u þv
ð10Þ In the simulations constructed for this study, the number of com-
putational nodes in the grid was 50 in the x-direction and 50 in the

2 2 y-direction. We used Successive Over Relaxation (S.O.R.) to obtain an
H ρ f c f 0 ∂θ ρ f c f 0 ∂θ 2
H ∂ θ ∂ θ
u 0þ v 0− 0 − 0 ¼0 ð11Þ iterative solution to the computational problem, and derived spatial
L ρm cm ∂x ρm cm ∂y L ∂x 2 ∂y 2 distributions of dimensionless temperature and salinity for various
combination of ω1, ω2, α, and β.
The non-dimensionalization process gives rise to the following five
dimensionless parameters in the governing equation: 2.3. Calculation of the E.C. distribution

H ρgHk ρm c m aL aT The calculated distributions of non-dimensional temperature and

¼ ω1 ; ¼ K; ; ; ð12Þ salinity are used for deriving those dimensional distributions, assuming
L μκ ρf cf H aL
the source temperature and salinity of injected thermal waters. The
spatial distribution of bulk E.C. is obtained from the above dimensional
and the boundary conditions give rise to three additional dimensionless distributions, based on the following steps.
parameters:
2.3.1. Calculation of pore water conductivity, σf
w F0 v0 H In general, pore water conductivity σf is positively correlated with
¼ ω2 ; ¼ α; ¼β ð13Þ
L κ κ temperature and salinity (Revil et al., 1998; Atkins and Paula, 2009).
Pore water conductivity is expressed as:
where: ω2, α, and β represent the normalized injection area, thermal h  i
water flux, and rainfall recharge, respectively. σ f ¼ γC 1 þ θ f T−T ref ð14Þ
Of the above eight dimensionless parameters, the permeability term
K does not actually affect fluid flow within this simple model, because where: C is the salinity (the mass fraction of solute), T is the dimensional
K defines dimensional fluid pressures, and disappears when Eq. (8) is temperature (°C), Tref is the reference temperature (Tref = 25 °C in this
substituted into Eq. (7). This reduction results from the dominant con- study), γ is the E.C. per unit salinity, and θf is the temperature depen-
tribution of forced convection to the flow regime (Holzbecher and dence [θf = 0.023 °C−1 in this study, according to Revil et al. (1998)].
Yusa, 1995). In this study, therefore, it is not necessary to consider In this study, the value of γ is calculated using the equivalent E.C. of
permeability within the fluid flow simulation. Further, variability in sodium and chloride ions as follows.
the parameters ρρm ccmf , aHL , and aaTL is small in comparison to the other non- E.C. of solutions is related to molar concentration:
f
dimensional parameters; therefore, these parameters are considered
fixed at the constant values noted in Table 2. For the purposes of this mNaCl
study, then, the dimensionless temperature and salinity distributions MNaCl
σ f 0 ¼ ðλNaþ þ λCl− Þ m ð15Þ
are controlled by changing the four dimensionless parameters ω1, ω2, aq
α, and β. ρaq
For the present study, the non-dimensionalized equations given
above are discretized using the finite difference method developed by where: σ f 0 is the E.C. of solutions (S/m), λNaþ and λCl− are the ionic con-
Patankar (1979), employing the power-law method. In the numerical ductivities of sodium and chloride ions (S m2 mol−1), mNaCl and maq are
simulations, the distribution of the fluid pressure is calculated, under the masses of NaCl and NaCl solution (kg), MNaCl is the molar mass of
 S. Komori et al. / Journal of Volcanology and Geothermal Research 283 (2014) 73–81 77

NaCl (kg mol− 1), and ρaq is the density of NaCl solution (kg m− 3). dependence (°C−1). The value of θs is about 0.04 for cations expected
Eq. (15) is rewritten as: to be contained within hydrothermal fluids in geothermal and volcanic
areas (Revil et al., 1998).
ðλNaþ þ λCl− Þ mNaCl
σ f0 ¼
M NaCl maq 2.3.3. Calculation of bulk conductivity, σ
ρaq The obtained pore water conductivity and surface conductivity are
¼ γC;
synthesized into bulk E.C. using Revil's model:
ðλNaþ þ λCl− Þ m 8 2 0 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s
13
>  
 2
γ¼ ; C ¼ NaCl ð16Þ >
> σf4 σs 1 σs @ 1 σs 1 σs 4F σ s A5
M NaCl maq >
> 1−t ðþÞ þ F þ t ðþÞ − 1− þ 1− þ
>
> σf σf t ðþ Þ σ f t ðþ Þ σ f t ðþÞ σ f
< F 2
ρaq
σ¼ for σ f ≧
σs
>
>
>
> t ðþ Þ
>
>
Using the values of λNaþ [5.01 × 10−3 S m2 mol−1, Atkins and Paula >
: σs þ
1−t ðþÞ
σf for σ f ≦
σs
F t ðþ Þ
(2009)], λCl− [7.64 × 10−3 S m2 mol−1, Atkins and Paula (2009)], MNaCl
(5.85 × 10− 2 kg mol−1), and ρaq (103 kg m−3, assuming dilute ð18Þ
solutions), the value of γ can be calculated as 216.
where: t(+) is the Hittorf transport number of a cation in the free
electrolyte [t(+) = 0.5 for KCl, and 0.38 for NaCl, for instance see Revil
2.3.2. Calculation of the surface conductivity of rock matrix, σs
et al. (1998)].
In general, the surface conductivity of rock matrix is positively corre-
F is the formation factor defined as:
lated with the amount of smectite produced by hydrothermal alteration
(e.g., Revil et al., 1998, 2002). Komori et al. (2010, 2013) examined the −m
F¼ϕ ð19Þ
relation between surface conductivity and temperatures to which a
rock matrix has been exposed, by E.C. measurements using drillcore where: ϕ is the porosity, and m is the cementation exponent. The ce-
samples from geothermal areas. The surface conductivity of rock matrix mentation exponent plays an important role in significantly changing
can be expressed as a function of the temperature to which rock matrix the contribution from σf and σs to σ.
has been exposed, because smectite is stable under certain tempera-
tures. In this study, it was thus assumed that the surface conductivity 2.4. Parameters and their uncertainties
of the matrix can be expressed as a function of the temperature of
pore water, following Komori et al. (2013). This study uses the following parameters to relate the E.C. structure
Fig. 3 shows the model describing surface conductivity. The amount with the flow of hydrothermal fluids. Four dimensionless parameters
of smectite is increased with increasing exposure temperature; the log- (ω1, ω2, α, β) and six dimensional parameters (Tmax1, Tmax2, Th, σSmin,
arithm of the surface conductivity linearly increases from σSmin (a). σSmax, σSH) are used for the flow pattern of groundwater and the spatial
When the exposure temperature is in the range between Tmax1 and distribution of E.C., respectively; which are referred to as “flow-regime
Tmax2, the rock matrix contains abundant smectite; the surface conduc- parameters” and “E.C.-related parameters” in this study. If the surface
tivity yields a high value σSmax (b). Under the condition of temperatures temperature (T0) and salinity (C0) are assumed to be 0 °C and 0 respec-
higher than Tmax2, smectite is changed into less conductive chlorite tively for simplification, two dimensional parameters, ΔT and ΔC, are
and/or illite (e.g., Pytte and Reynolds, 1989); the logarithm of the sur- defined as the temperature and salinity of original thermal waters.
face conductivity linearly decreases to σSH (c). After the breakdown of The values of the flow-regime parameters are assumed as follows
smectite, the surface conductivity has a constant value at temperatures (cf. Table 2). The horizontal length L was assumed to be 2 km, based
higher than Th (d). Note that the values of surface conductivity are those on the lines P–P′ and Q–Q′. The thickness of the lateral-flow region H
at the reference temperature Tref (25 °C in this study). was assumed to be 2.2 km, based on the average lowermost depth of
The surface conductivity is calculated on the basis of the above the high E.C. region (Srigutomo et al., 2008). Thus, the aspect ratio ω1
model, and modified to the value at the actual temperature, using the was assumed to be 1.1. The non-dimensional rainfall recharge β was
following formula: changed between 5 and 35, based on the several % to a few tens % of
h  i the annual rainfall at Unzen area. The non-dimensional flux was
σ SðT Þ ¼ σ SðT Þ 1 þ θs T−T ref ð17Þ changed between 10−1 and 102.
ref

The values of the E.C.-related parameters are assumed as follows.

where: T is the temperature during measurements (°C), σ SðT ref Þ is the
surface conductivity at reference temperature (S/m), σS(T) is the surface
conductivity at the temperature of T (S/m), and θs is the temperature
Fig. 3. Simple model for surface conductivity σs as a function of exposure temperature.
The formation factor F is assumed to be 16–44; which is calculated
using the porosity ϕ of 15–25% obtained from in-situ gamma-ray
logging at Unzen volcano (Ikeda et al., 2008), and the cementation
exponent m of 2 (e.g., Rinaldi et al., 2011). σSmin, σSmax, σSH,, Tmax1,
Tmax2, and Th are assumed their acceptable values with some uncer-
tainties as shown in Table 2, using the previous works as references
(e.g., Pytte and Reynolds, 1989; Takakura, 2000; Revil et al., 2002;
Komori et al., 2010, 2013).
In this study, it is assumed that the thermal waters have a possible
temperature of 300 °C (NEDO, 1988; Ohsawa et al., 2002; Ohsawa,
2006). According to phase diagrams, liquid-dominant thermal waters
with a temperature of 300 °C at deeper depths could have an enthalpy
of ~ 1300 kJ/kg. If volcanic gases and meteoric water are assumed to
have enthalpies of 4400 and 0 kJ/kg, respectively (e.g., Ohba et al.,
2010), thermal waters with an enthalpy of ~1300 kJ/kg could be formed
by mixing, with the ratio of volcanic gases:meteoric water = 1:2.4.
Ohba et al. (2008) showed the possible masses of main volcanic gas
components (H2O, CO2, S, Cl) released from the unit mass of magmas
at the first-degassing stage. This study assumed that these components,
78 S. Komori et al. / Journal of Volcanology and Geothermal Research 283 (2014) 73–81

Table 3
Volcanic gas components at the first degassing stage.

Components Lowest salinity case Highest salinity case

Before mixing with meteoric watera After mixing with meteoric water Before mixing with meteoric watera After mixing with meteoric water
(mol/l) (mol/l) (mol/l) (mol/l)

CO2/H2O 0.34 0.10 0.65 0.19

S/H2O 0.12 0.035 0.59 0.17
Cl/H2O 0.095 0.028 0.21 0.063
a
All the data before mixing with meteoric water are taken from Ohba et al. (2008).

except H2O, contribute to ionic conduction mechanism as follows:
CO2 → HCO− 2− −
3 , S → SO4 , and Cl → Cl ; which were assumed to be
solved into the magmatic H2O and mixing meteoric water. This study
used their molarity data provided by Ohba et al. (2008), as shown in
Table 3, and calculated the concentrations of NaCl-equivalent solutions,
which have electrical conductivities equal to those of volcanic compo-
nents. Note that Na+ was assumed as the counterion of the above
anions in the calculation to maintain electroneutrality, and that the cal-
culation is based on the principle of equivalent conductivity (e.g., Atkins
and Paula, 2009). Consequently, this study assumed that the thermal
waters have salinities of 1.1–3.4 wt.%, by using the results of the
calculation and taking into account mixing between volcanic gas and
meteoric water mentioned above; note that meteoric water is assumed
to have no salinity.
In addition, it was also assumed that the high conductivity region
can be located with a vertical uncertainty of ±100 m. This assumption
is deemed acceptable given the error of 1-D analysis performed by
Srigutomo et al. (2008).
3. Results
3.1. Spatial variations in the high E.C. region
Fig. 4a shows the spatial distributions of non-dimensional tempera-
ture and salinity together with the direction of groundwater flow,
which are simulated using a flow regime parameterized by α = 100.2,
β = 5, ω2 = 0.02, and ω1 = 1.1. Temperature is the highest at the
injection center, and decreases with distance from the center. High tem-
peratures extend horizontally, and there is relatively clear rapid
transition of temperatures near the mixing front between thermal
water and meteoric water. In contrast to the temperature distribution,
salinity has a relatively horizontal homogeneity, and the spatial distri-
bution of salinity has more significant rapid transition near their mixing
front. These features in the salinity distribution come from the smaller
effect of saline dispersion (diffusion) compared to the diffusion of heat.
To obtain the distribution of E.C., the non-dimensional distributions
of temperature and salinity are converted into their dimensional ones,
using the temperature and salinity of thermal waters assumed in the
simulation; furthermore, by applying the E.C.-related parameters, the
spatial distributions of pore water E.C., surface conductivity of rock ma-
trix, and bulk E.C. are obtained, as shown in Fig. 4b. Note that the results
in the figure are calculated using the following parameters: ΔT = 300,
ΔC = 0.034, Tmax1 = 100, Tmax2 = 150, Th = 175, σSmin = 3 × 10− 3,
and σSmax = 1 × 10−1. Pore water E.C. has a clear difference near the
mixing front between thermal water and meteoric water, and a high
pore water E.C. region extends horizontally; which is mainly due to
the salinity distribution. A high surface conductivity region (greater
than 0.3 S/m) develops, corresponding to the region with temperatures
of 100–170 °C assumed as the stable condition of smectite. As a conse-
quence of combined contributions of pore water and rock matrix, a
high bulk E.C. region (greater than 0.3 S/m) develops below the mixing
front.
Fig. 5 shows the distribution of the high bulk E.C. region (greater
than 0.3 S/m), under the controlled conditions of the non-dimensional
injection area ω2, thermal water flux α, and rainfall recharge β. Addi-
tionally, the figure also shows the case under the controlled condition
of aspect ratio of the lateral flow region ω1 to examine the influence
of its variation on the distribution. These distributions show the

(a)

(b)

Fig. 4. Examples of simulated results. (a) Spatial distributions of non-dimensional temperature and salinity, under certain conditions of flow-regime parameters. (b) Spatial distributions of
pore water E.C., surface conductivity of rock matrix, and bulk E.C., under certain conditions of E.C.-related parameters and parameters for thermal waters.
 S. Komori et al. / Journal of Volcanology and Geothermal Research 283 (2014) 73–81 79

(a) (b)

(c) (d)

Fig. 5. Simulated high E.C. region with an electrical conductivity greater than 0.3 S/m, under the controlled conditions of the thermal water flux α, rainfall precipitation β, injection area ω2,
and aspect ratio ω1. The spatial extent of the high electrical conductivity is controlled essentially by α and β.

following characteristics: (a) larger α extends the high E.C. region hori-
zontally and vertically; (b) in contrast, larger β shrinks the region to-
ward the injection area; (c) on the other hand, ω2 has little
contribution to the distribution of the high E.C. region; and (d) ω1 has
a smaller contribution to the distribution than α and β, under the possi-
ble variation in ω1 (0.75–1.25, equivalent to depths of 1.5–2.5 km). In
this case study at Unzen volcanic area, the simple dissipation model of
volcanic gases showed that the high bulk E.C. region is controlled essen-
tially by the thermal water flux α and the rainfall recharge β.
3.2. Evaluation of the mass flux of volcanic gases at Unzen volcanic area
3.2.1. Calculation of the structure similarity
The above non-dimensional parameters control the spatial extent of
the region with an E.C. greater than 0.3 S/m. The simulated high conduc-
tivity region was compared with the region inferred from TDEM, using
the “structure similarity” defined as:
intervals), 3 cases for σSmin, 2 cases for σSmax, 1 case for σSH, 3 cases for
Tmax1, 3 cases for Tmax2, 3 cases for Th, 3 cases for F, and 4 cases for ΔC;
which are summarized in Table 2. By changing the above parameters,
a total of 93,312 cases of E.C. distribution were generated. Three cases
of the distribution of high E.C. regions, which were generated by consid-
ering the uncertainty of the E.C. structure from TDEM, were compared
with the simulated results for each survey line. The simulated results
with the best 10% of high similarity were extracted from the total of
279,936 cases of comparisons. Fig. 6c shows the cumulative frequency
distributions as the function of non-dimensional thermal water
(a)

Govr
structure similarity ¼ ð20Þ (b)
Gcal þ Gobs

where: Gcal is the number of the computational grids of the high

conductivity region obtained from the numerical simulation, Gobs is
the number of its grids obtained from the observation data, and Govr is
the number of both grids of observation and simulation that are
overlapping each other. When both grids are overlapping completely,
the structure similarity has a maximum value of 1. Its value is decreased (c)
either when both regions of high conductivity are away from each
other, or when both sizes of the region are significantly different from
each other.

3.2.2. Evaluation of volcanic gas flux

If it is assumed that the high E.C. region, shown in Fig. 1c, is present
continuously between the adjacent observation points, the regions
along the lines P–P′ and Q–Q′ can be represented as shown in Fig. 6a,
noting that a vertical distance is normalized by the distance between
ground surface and bottom of the high E.C. region of each 1-D profile
for simplification. Fig. 6b shows the simulated result with the similarity Fig. 6. Application of the simple dissipation model of volcanic gases to Unzen volcanic area.
of 92% for the line Q–Q′, given specific flow-regime and E.C.-related pa- (a) Simplified high E.C. regions along the lines P–P′ and Q–Q′ shown in Fig. 1. Horizontal
rameters as shown in the figure. Two flow-regime parameters, seven and vertical non-dimensional distances represent the normalized distance from the center
(P and Q) and the bottom of a lateral-flow region, respectively. (b) Example of simulated
E.C.-related parameters, and salinity of thermal waters are changed to results with a similarity of 92% for the case of the line Q–Q′, under the parameters de-
represent variations in the high E.C. region, as follows: 16 cases for α scribed in the figure. (c) Cumulative frequency distribution of the model with the best
(at 0.2 intervals by logarithmic expression), 3 cases for β (at 15 10% of similarity, as a function of non-dimensional thermal water flux.
80 S. Komori et al. / Journal of Volcanology and Geothermal Research 283 (2014) 73–81
fluxes α, using the simulated results with high similarities; from
which the average α was estimated to be 100.42 ± 0.35 for the line P–P′,
and 10−0.10 ± 0.35 for the line Q–Q′. Consequently, the average α was
estimated to be 100.16 ± 0.25.
The total mass flux of thermal waters is represented as:
total flux ¼ 2ακρ f d ð21Þ
where, 2 is the symmetry-based constant, and d is the length of the in-
jection area perpendicular to the coordinate system (m). In this study,
the value of d was assumed to be 2–3 × 103 m, referring to the extent
of the high conductance region (Srigutomo et al., 2008), and those of κ
and ρf were assumed to be 10−6 m2 s−1 and 103 kg m−3, respectively.
Using the equation, the total mass flux of thermal waters was estimated
as 105.3 ± 0.3 t/yr. As described previously, thermal waters are assumed
to be composed of volcanic gases and meteoric waters with a ratio of
1:2.4. Consequently, the total mass flux of volcanic gases is estimated
to be 104.8 ± 0.3 t/yr.
4. Discussion
4.1. Comparison with other data
At Unzen volcano, various observations have been made to investi-
gate volcanic activity, including geophysics and geochemistry. The va-
lidity of the fluxes of volcanic gases estimated above was examined in
light of observations and analyses from these previous studies.
According to Ohba et al. (2008), the first degassing process could re-
lease magmatic CO2 with a CO2/H2O mass ratio of 1.5–2.8 × 10−2 from
magma. Using the estimated mass flux of volcanic gases in this study,
the flux of magmatic CO2 is estimated to be 103.1 ± 0.3 t/yr. This value
is consistent with the diffusive flux of magmatic CO2 from the western
part of Unzen Graben (103.5 t/yr), estimated from soil gas and ground-
water analyses (Shinohara, 2005).
In addition, the first degassing process can decrease the H2O content
of magma from 6 (primary) to 4.8–3.7 wt.% (Ohba et al., 2008). Consid-
ering this H2O release, the volcanic gas flux of 104.8 ± 0.3 t/yr estimated
in this study requires a magma supply rate of 100.6 ± 0.3 million t/yr;
which is equivalent to 100.1 ± 0.3 million m3/yr of magmas, assuming
the density of magma to be 2.8 × 103 kg m−3. Leveling surveys,
started from 1983, showed that magma supply had continued at a rate
of 100.35 ± 0.05 million m3/yr from at least 1984 (Kohno et al., 2008);
this value is consistent with the rate estimated in this study. Further,
earthquake swarms occurring every few tens of years have suggested
an intermittent migration of magmas from a deeper source toward
the first degassing depths (JMA, 1970; Ohta, 1973, 1988). By incorporat-
ing the above, in the western part of the Shimabara Peninsula volcanic
gases could be steadily released from intermittently-supplied magma;
which might make a non-negligible contribution to decreasing the
H2O content of magmas before eruptions. Efficiency of magma
degassing is one of the important factors controlling volcanic activities;
for instance, if magmas easily release volcanic gases, they would not be
able to get enough buoyancy because of the lack of bubbled volcanic
gases, and their ascent would be slowed (effusive volcanism) or stopped
on the way to the surface (Kagiyama, 2008). Therefore, effective magma
degassing into overlying groundwater flow systems may be one of the
reasons that dome-forming eruptions have dominated at young
Unzen in the recent past, instead of explosive volcanism (Hoshizumi
et al., 2002).
Some thermal studies have revealed heat-discharging activity at
Unzen volcanic area. Yuhara et al. (1981) estimated the heat discharge
rate to be 101.85 ± 0.5 MW in the western part of Shimabara Peninsula
through helicopter-borne infrared measurements. Using the mass flux
of volcanic gases estimated in the present study and their possible en-
thalpies (4400 kJ/kg, assumed in the model), the released volcanic
gases are thought to have a heat flux of 100.9 ± 0.3 MW; which is smaller
than the estimate of Yuhara et al. (1981). On the other hand, a hydro-
thermal simulation by Fujimitsu et al. (2008) assumed deeply-seated
magma bodies as heat sources to explain the heat discharge rate in
the area, although the authors did not consider the contribution from
volcanic gases. Based on the above, from the viewpoint of heat transfer,
it would be necessary to consider combined contributions from both
volcanic gases and magma bodies to understand the nature of heat
discharge in volcanic areas.
4.2. Validity and uncertainty of the models
The models for fluid flow and E.C., developed in this study, represent
a smectite-rich region as a high surface conductivity (see Fig. 4b). Its
layerlike morphology is similar to those typically observed at volcanic
areas bearing hydrothermal fluids (e.g., Ogawa et al., 1998; Nurhasan
et al., 2006; Aizawa et al., 2009). In general, smectite significantly de-
creases formation permeability (e.g., Revil and Cathles, 1999; Komori
et al., 2013); which is not represented in this study, because isotopic
permeability is assumed in the models for simplification. However,
the E.C. structures simulated in this study yielded the volcanic gas
flux, which can consistently explain the magmatic CO2 flux and the
magma input rate at the western Unzen volcanic area. Therefore, the
models are likely sufficient to represent the liquid-dominated hydro-
thermal flow system in a first-order sense. It is known that low-
permeability of a cap rock plays an important role in developing a
vapor-filled region (e.g., Schubert and Straus, 1980). Their improved
models, for instance, including the permeability-reduction associated
with water–rock interaction (e.g., Xu et al., 2003), will be applicable to
more complicated hydrothermal systems at volcanic areas.
As mentioned in Section 2.3, there are various parameters control-
ling bulk E.C. This section examines the effect of their uncertainties on
the precision of the estimated volcanic gas fluxes. In many cases, the
upper part of the high E.C. region is defined by high surface conductivity
of rock matrices (σs), like Fig. 4b. Basically, the upper boundary of the
high σs is controlled by Tmax1, σSmin, and σSmax; on the other hand,
Tmax2 and TH control the lower boundary. The present study found that
the uncertainties of Tmax1, σSmin, and σSmax can fluctuate the upper
boundary within a few percent of vertical length. This suggests that fluc-
tuations can affect the estimation of volcanic gas fluxes within less than
±1/10 order of magnitude.
Formation factor F controls the contribution from pore water to bulk
E.C., as shown in Eqs. (18) and (19). In the present study, the contribu-
tion can be changed up to ~2.8 times by the uncertainty of F. This means
that the estimate of bulk E.C. of a fluid-bearing region beneath a
smectite-rich (high σs) layer is significantly changed. For instance, if
low-salinity fluids are contained in rock matrices with a high F, the
bulk E.C. maintains a small value; which does not extensively develop
the high E.C. region as seen at the Unzen volcanic area.
In addition, the uncertainty of the vertical extent of high E.C. regions
is about ±5% of vertical length. This can also affect the same degree of
estimation errors of the fluxes as the uncertainties of σs.
5. Conclusions
In this study, we developed a simple model of fluid flow and E.C. of
pore water and rock matrix, and attempted to estimate the mass flux
of volcanic gases from magma using the E.C. structure. It appears that
the spatial extent of the high E.C. region is controlled essentially by
the thermal water flux and rainfall precipitation. The application of
this simple model to Unzen volcanic area successfully estimated the vol-
canic gas flux, consistent with other geophysical and geochemical data.
Therefore, our method to estimate the flux of volcanic gases is consid-
ered to be suitable in this simple case study. The possibility of effective
magma degassing into overlying groundwater flow systems provides
some insights into the diversity of volcanic activities, associated with
the ease of magma ascent.
 S. Komori et al. / Journal of Volcanology and Geothermal Research 283 (2014) 73–81
Acknowledgements
We thank K. Takemura, S. Ohsawa, M. Utsugi, T. Ohkura, T.
Kawamoto, T. Shibata, T. Goto, N. Oshiman, R. Yoshimura (Kyoto
Univ.), A. Terada (Tokyo Institute of Technology), and W. Kuang
(NASA) for their valuable discussions. We also thank K. Aizawa (Kyushu
Univ.), H. Hase (Tokyo Institute of Technology), Y. Yamaya (GSJ),
T. Hashimoto (Hokkaido Univ.), H. Utada (Univ. of Tokyo), and B.F.
Chao (Academia Sinica) for their valuable comments on our research.
We would like to appreciate one anonymous reviewer and G.F.
Zellmer (Massey Univ.) for critical review and constructive comments,
and A. Aiuppa for editorial support. This work was supported by the
Grant-in-Aid for Scientific Research (no. 23310120, T. Kagiyama) from
the Ministry of Education, Culture, Sports, Science, and Technology,
Japan. Some figures were made using the GMT program (Wessel and
Smith, 1998).
Appendix A. Supplementary data
Supplementary data associated with this article can be found in the
online version, at http://dx.doi.org/10.1016/j.jvolgeores.2014.05.024.
These data include Google maps of the most important areas described
in this article.
References
Aizawa, K., Ogawa, Y., Ishido, T., 2009. Groundwater flow and hydrothermal systems
within volcanic edifices: delineation by electric self-potential and magnetotellurics.
J. Geophys. Res. 114 (B1), B01208. http://dx.doi.org/10.1029/2008JB005910.
Atkins, P.W., Paula, J.D., 2009. Elements of Physical Chemistry, 5th ed. Oxford U.P., Oxford
(627 pp.).
Chida, N., 1979. Late Quaternary tectonic movements in central Kyushu district, South-
west Japan, with special reference to the active faults distributed in the Quaternary
volcanic area. Annu. Rep. Fac. Educ. Iwate Univ. 39, 97–135 (in Japanese).
Eichelberger, J.C., Carrigan, C.R., Westrich, H.R., Price, R.H., 1986. Non-explosive silicic vol-
canism. Nature 323, 598–602. http://dx.doi.org/10.1038/323598a0.
Fujimitsu, Y., Ehara, S., Oki, R., Kanou, R., 2008. Numerical model of the hydrothermal sys-
tem beneath Unzen Volcano, Japan. J. Volcanol. Geotherm. Res. 175, 35–44.
Gelhar, L.W., Welty, C., Rehfeldt, K.R., 1992. A critical review of data on field-scale disper-
sion in aquifers. Water Resour. Res. 28 (7), 1955–1974.
Henley, R.W., Ellis, A.J., 1983. Geothermal systems ancient and modern: a geochemical re-
view. Earth-Sci. Rev. 19, 1–50.
Holzbecher, E., Yusa, Y., 1995. Numerical experiments on free and forced convection in
porous media. Int. J. Heat Mass Transfer 38 (11), 2109–2115.
Hoshizumi, H., Uto, K., Watanabe, K., 1999. Geology and eruptive history of Unzen volca-
no, Shimabara Peninsula, Kyushu, SW Japan. J. Volcanol. Geotherm. Res. 89 (1–4),
81–94. http://dx.doi.org/10.1016/S0377-0273(98)00125-5.
Hoshizumi, H., Uto, K., Matsumoto, A., Xu, S., Oguri, K., 2002. Geology of Unzen volcano
and core stratigraphy of the flank drillings. Extended Abstract of International
Unzen Workshop, pp. 4–7 (in Japanese).
Ikeda, R., Kajiwara, T., Omura, K., Hickman, S., 2008. Physical rock properties in and around a
conduit zone by well-logging in the Unzen Scientific Drilling Project, Japan. J. Volcanol.
Geotherm. Res. 175, 13–19. http://dx.doi.org/10.1016/j.jvolgeores.2008.03.036.
Ingebritsen, S.E., Geiger, S., Hurwitz, S., Driesner, T., 2010. Numerical simulation of mag-
matic hydrothermal systems. Rev. Geophys. 48, RG1002. http://dx.doi.org/10.1029/
2009RG000287.
Ishihara, K., 1993. Continuous magma supply inferred from discharge rate of magma and
ground-deformation rate at Mt. Unzen, Japan. Ann. Disaster Prev. Res. Inst., Kyoto
Univ. 36 (B1), 219–230 (in Japanese with English abstract).
Japan Meteorological Agency (JMA), 1970. Implementation Report of Immediate Volcano
Observation — Unzen Volcano. Japan Meteorological Agency, Tokyo, pp. 13–17
(in Japanese).
Jaupart, C., Allégre, C.J., 1991. Gas content, eruption rate and instabilities of eruption re-
gime in silicic volcanoes. Earth Planet. Sci. Lett. 102, 413–429.
Kagiyama, T., 2008. Eruption dominant volcanism vs. geothermal activity dominant
volcanism — new aspect in volcanism. J. Geotherm. Res. Soc. Jpn. 30 (3), 193–204
(in Japanese with English abstract).
Kagiyama, T., Utada, H., Yamamoto, T., 1999. Magma ascent beneath Unzen Volcano, SW
Japan, deduced from the electrical resistivity structure. J. Volcanol. Geotherm. Res. 89,
35–42. http://dx.doi.org/10.1016/S0377-0273(98)00120-6.
Keller, G.V., Rapolla, A., 1974. Electrical prospecting methods in volcanic and geothermal
environments. In: Civetta, L., Gasparini, P., Luongo, G., Rapolla, A. (Eds.), Physical
Volcanology. Elsevier, Amsterdam, pp. 133–166.
Kohno, Y., Matsushima, T., Shimizu, H., 2008. Pressure sources beneath Unzen Volcano in-
ferred from leveling and GPS data. J. Volcanol. Geotherm. Res. 175, 100–109. http://
dx.doi.org/10.1016/j.jvolgeores.2008.03.022.
Komori, S., Kagiyama, T., Hoshizumi, H., Takakura, S., Mimura, M., 2010. Vertical mapping
of hydrothermal fluids and alteration from bulk conductivity: simple interpretation
81
on the USDP-1 site, Unzen Volcano, SW Japan. J. Volcanol. Geotherm. Res. 198,
339347. http://dx.doi.org/10.1016/j.jvolgeores.2010.09.019.
Komori, S., Kagiyama, T., Takakura, S., Ohsawa, S., Mimura, M., Mogi, T., 2013. Effect of the
hydrothermal alteration on the surface conductivity of rock matrix: comparative study
between relatively-high and low temperature hydrothermal systems. J. Volcanol.
Geotherm. Res. 264, 164–171. http://dx.doi.org/10.1016/j.jvolgeores.2013.08.009.
Matsushima, N., 2011. Estimation of permeability and degassing depth of Iwodake volcano at
Satsuma–Iwojima, Japan, by thermal observation and numerical simulation. J. Volcanol.
Geotherm. Res. 202, 167–177. http://dx.doi.org/10.1016/j.jvolgeores.2011.02.005.
Nakada, S., Shimizu, H., Ohta, K., 1999. Overview of the 1990–1995 eruption at Unzen Vol-
cano. J. Volcanol. Geotherm. Res. 89, 1–22.
New Energy Development Organization (NEDO), 1988. Western area of Unzen. Report of
Promotional Exploration for Geothermal Research, 15 (1060 pp., in Japanese).
Nurhasan, Ogawa, Y., Ujihara, N., Tank, S.B., Honkura, Y., Onizawa, S., Mori, T., Makino, M.,
2006. Two electrical conductors beneath Kusatsu–Shirane volcano, Japan, imaged by
audiomagnetotellurics, and their implications for the hydrothermal system. Earth
Planets Space 58, 1053–1059.
Ogawa, Y., Matsushima, N., Oshima, H., Takakura, S., Utsugi, M., Hirano, K., Igarashi, M.,
Doi, T., 1998. A resistivity cross-section of Usu volcano, Hokkaido, Japan, by
audiomagnetotellurics soundings. Earth Planets Space 50, 339–346.
Ohba, T., Hirabayashi, J., Nogami, K., Kusakabe, M., Yoshida, M., 2008. Magma degassing
process during the eruption of Mt. Unzen, Japan in 1991 to 1995: modeling with
the chemical composition of volcanic gas. J. Volcanol. Geotherm. Res. 175, 120–132.
Ohba, T., Sawa, T., Taira, N., Yang, T.F., Lee, H.F., Lan, T.F., Ohwada, M., Morikawa, N.,
Kazahaya, K., 2010. Magmatic fluids of Tatun volcanic group, Taiwan. Appl. Geochem.
25, 513–523. http://dx.doi.org/10.1016/j.apgeochem.2010.01.009.
Ohsawa, S., 2006. A hydro-geochemical study on hydrothermal system and formation
processes of hot springs in Unzen Graben of the Shimabara Peninsula, Kyushu,
Japan. J. Geotherm. Res. Soc. Jpn. 28 (4), 361–371 (in Japanese with English abstract).
Ohsawa, S., Kazahaya, K., Yasuhara, M., 2002. Geochemistry of fluids from hot springs,
mineral springs, wells and fumarolic area in the Shimabara Peninsula, Middle
Western Kyushu, Japan. J. Hot Spring Sci. (OnsenKagaku) 52, 51–68 (in Japanese
with English abstract).
Ohta, K., 1973. A study of hot springs on the Shimabara Peninsula. The Science Reports
of the Shimabara Volcano Observatory, the Faculty of Science, Kyushu University, 8,
pp. 1–33 (in Japanese with English abstract).
Ohta, K., 1988. Damages due to the 6 August 1984 earthquake occurred in the Shimabara
Peninsula. The Science Reports of the Shimabara Volcano Observatory, the Faculty of
Science, Kyushu University, 14, pp. 53–65 (in Japanese with English abstract).
Patankar, S.V., 1979. A calculation procedure for two-dimensional elliptic situations. Num.
Heat. Transfer 2.
Pytte, A.M., Reynolds, R.C., 1989. The thermal transformation of smectite to illite. Thermal
History of Sedimentary Basins: Methods and Case Histories. Springer-Verlag,
New York (319 pp.).
Revil, A., Cathles III, L.M., 1999. Permeability of shaly sands. Water Resour. Res. 35 (3),
651–662. http://dx.doi.org/10.1029/98WR02700.
Revil, A., Cathles III, M., Losh, S., 1998. Electrical conductivity in shaly sands with geophys-
ical applications. J. Geophys. Res. 103 (B10), 23925–23936. http://dx.doi.org/10.1029/
98JB02125.
Revil, A., Hermitte, D., Spangenberg, E., Cochemé, J.J., 2002. Electrical properties of
zeolitized volcaniclastic materials. J. Geophys. Res. 107 (B8). http://dx.doi.org/10.
1029/2001JB000599.
Rinaldi, A.P., Todesco, M., Vandemeulebrouck, J., Revil, A., Bonafede, M., 2011. Electrical
conductivity, ground displacement, gravity changes, and gas flow at Solfatara crater
(Campi Flegrei caldera, Italy): results from numerical modeling. J. Volcanol.
Geotherm. Res. 207, 93–105.
Sato, K., Iwasa, Y., 2002. Hydrology. Maruzen, Japan (321 pp.; in Japanese).
Schubert, G., Straus, J.M., 1980. Gravitational stability of water over steam in vapor-
dominated geothermal systems. J. Geophys. Res. 85 (B11), 6505–6512. http://dx.
doi.org/10.1029/JB085iB11p06505.
Shinohara, H., 2005. Understanding the distribution and transportation of deep-origin
fluids. Unzen Volcano: International Collaborative Research for Understanding the
Eruptive Mechanisms and Activities by Scientific Drilling. 11 pp. (in Japanese).
Shinohara, H., Kazahaya, K., 1995. Role of volatile components in eruption processes. Bull.
Volcanol. Soc. Jpn. 40, S59–S77 (in Japanese with English abstract).
Srigutomo, W., Kagiyama, T., Kanda, W., Munekane, H., Hashimoto, T., Tanaka, Y., Utada,
H., Utsugi, M., 2008. Resistivity structure of Unzen Volcano derived from time domain
electromagnetic (TDEM) survey. J. Volcanol. Geotherm. Res. 175, 231–240.
Takakura, S., 2000. Relationships of resistivity to physical, chemical and mechanical prop-
erties of clay-bearing samples. Butsuri-tansa 53 (5), 415–426 (in Japanese with En-
glish abstract).
Todesco, M., Rinaldi, A.P., Bonafede, M., 2010. Modeling of unrest signals in heterogeneous
hydrothermal systems. J. Geophys. Res. 115 (B09213). http://dx.doi.org/10.1029/
2010JB007474.
Umakoshi, K., Shimizu, H., Matsuwo, N., 1994. Magma ascent path in the 199094 eruption
of Fugendake, Unzen Volcano, as inferred from precisely determined hypocentral dis-
tribution. Bull. Volcanol. Soc. Jpn. 39, 223–235 (in Japanese with English abstract).
Wessel, P., Smith, W.H.F., 1998. New, improved version of Generic Mapping Tools re-
leased. EOS Trans. Am. Geophys. Union 79 (47), 579.
Xu, T., Sonnenthal, E., Spycher, N., Pruess, K., 2003. TOUGHREACT: a new code of the
TOUGH family of nonisothermal multiphase reactive geochemical transport in vari-
ably saturated geologic media. Proceedings of the TOUGH Symposium 2003. Law-
rence Berkeley National Laboratory, Berkeley, California (12–14 May 2003).
Yuhara, K., Ehara, S., Tagomori, K., 1981. Estimation of heat discharge rates using infrared
measurements by a helicopter-borne thermocamera over the geothermal areas of
Unzen Volcano, Japan. J. Volcanol. Geotherm. Res. 9, 99–109.