Heat flow distribution and thermal structure of the Nankai subduction zone off the Kii Peninsula



Detailed heat flow surveys were carried out in the central part of the Nankai Trough southeast of the Kii Peninsula (off Kumano) for investigation of the thermal structure of the subducting plate interface. At stations in the Kumano Trough (forearc basin) and its vicinity, long-term monitoring of temperature profiles in surface sediments was conducted because bottom water temperature variations (BTV) significantly disturb subbottom sediment temperatures. Heat flow values were successfully determined at seven stations by removing the influence of BTV from temperature records for 300 to 400 days. The surface heat flow data were combined with estimates from depths of methane hydrate bottom simulating reflectors to construct a heat flow profile across the subduction zone. Heat flow decreases from 90–110 mW/m2 on the floor of the Nankai Trough to 50–60 mW/m2 at around 30 km from the deformation front, while it is rather uniform, 40–60 mW/m2, in the Kumano Trough. The values measured on the Nankai Trough floor are concordant with the value estimated from the age of the subducting Philippine Sea plate, about 20 m.y., taking into account the effect of sedimentation. The obtained heat flow profile was used to constrain thermal models of the subduction zone. The subsurface thermal structure was calculated using a two-dimensional, steady state model, in which the frictional heating along the plate interface and the radioactive heat production are treated as unknown parameters. Comparison of the calculated surface heat flow in the Kumano Trough with the observed data indicates that the effective coefficient of friction is small, about 0.1 or less, and thus the shear stress on the plate interface is very low in this subduction zone.

1. Introduction

The Nankai subduction zone is a convergent plate boundary where the Philippine Sea plate is subducting beneath the southwest Japan arc (Figure 1). Large thrust earthquakes have occurred along this boundary with a recurrence interval of 100 to 200 years. The last two large events were the 1944 Tonankai (Mw = 8.1) and 1946 Nankai (Mw = 8.3) earthquakes [e.g., Kanamori, 1972; Ando, 1975]. Knowledge of the thermal structure is critical for understanding the nature of the seismogenic zone because the physical and chemical properties of the crustal and mantle material are strongly temperature dependent. For example, the updip and downdip limits of the locked seismogenic thrust fault are thought to be controlled by various processes, including diagenetic/metamorphic changes, hydration/dehydration, consolidation, and transition in frictional property of rocks [e.g., Hyndman et al., 1997; Moore and Saffer, 2001], and most of them are temperature-controlled. The subsurface thermal structures of subduction zones can be estimated through numerical modeling. It should be noted that there are some parameters with large uncertainty in thermal models for subduction; e.g., frictional heating along the plate interface and the radioactive heat production in the overriding plate. The values of these parameters may be constrained by the surface heat flow distribution.

Figure 1.

(a) Bathymetry map of the middle part of the Nankai subduction zone. Broken line represents the deformation front. Shaded circle is the location of ODP Site 808. Depth contours are in meters. (b) Topographic profile and major subsurface structures along the line A-B (modified from Park et al. [2002]).

Heat flow measurements in the Nankai subduction zone have been conducted since the 1960s and relatively high values were reported on the floor of the Nankai Trough, which extends along the plate boundary (Figure 1) [Watanabe et al., 1970]. Detailed heat flow measurements were conducted in the central part of the Nankai Trough off eastern Shikoku [Kinoshita and Yamano, 1995; Yamano et al., 2003]. They revealed that heat flow on the trough floor is up to twice as high as the value expected from the age of the subducting plate, about 15 m.y.

In the area southeast of the Kii Peninsula (termed “off Kumano” below), where the rupture area of the 1944 Tonankai earthquake is located, heat flow data had been sparse. We conducted heat flow measurements in this area for estimation of the subsurface thermal structure. The estimated thermal structure would contribute to studies of the seismogenic zone of large thrust earthquakes along the Nankai margin. The area is one of the targets of seismogenic zone drilling by the Integrated Ocean Drilling Program (IODP) and various geophysical/geological surveys have been made extensively such as seismic reflection, microseismicity and electromagnetic surveys [e.g., Park et al., 2002; Kasaya et al., 2005; Obana et al., 2005; Moore et al., 2007]. Drilling itself also provides important information for thermal modeling, including thermal properties of rocks and temperature data in deep boreholes.

In this paper, we describe the result of heat flow measurements and the heat flow distribution in the area off Kumano, and estimate the amount of frictional heating along the plate interface and the thermal structure of the Nankai subduction zone off Kumano based on heat flow data.

2. Heat Flow Measurements

2.1. Tectonic Setting

The oceanic plate subducting along the Nankai Trough is the Shikoku Basin, which occupies the northern part of the Philippine Sea plate (Figure 1). The Shikoku Basin was formed as a back-arc basin of the Izu-Bonin subduction zone between 30 and 15 Ma and the spreading ceased around 15 Ma according to identification of lineated magnetic anomalies [e.g., Okino et al., 1994]. The fossil spreading ridge lying off eastern Shikoku is nearly perpendicular to the strike of the Nankai Trough. The convergence rate along the Nankai Trough is 4 to 6.5 cm/year according to models for the motion of the Philippine Sea plate estimated on the basis of earthquake slip vector, space geodetic data and geological constraints [e.g., Seno et al., 1993; Miyazaki and Heki, 2001; Loveless and Meade, 2010].

The Nankai Trough is characterized by a flat trough floor filled with thick turbidite deposits and a well developed accretionary prism. In the area off Kumano, the water depth of the trough floor is about 4500 m. The most seaward part of the prism forms a slope between the trough floor and a forearc basin (Kumano Trough) which spreads out flat with water depths of about 2100 m (Figure 1). Along the seaward edge of the Kumano Trough, an outer ridge extends in a direction parallel to the Nankai Trough, forming a bank to allow sediment to accumulate in the Kumano Trough.

Park et al. [2002] carried out multichannel seismic (MCS) reflection surveys in the area off Kumano and revealed detailed structure of the upper 10 km of the accretionary prism and the subducting oceanic crust. They reported the existence of “splay faults,” landward-dipping thrust faults branching upward from the plate interface at a depth of 10 km (Figure 1b). The splay faults reach the seafloor just seaward of the outer ridge, where cold seeps with chemosynthetic communities were discovered through submersible surveys [Ashi et al., 2002; Toki et al., 2004], suggesting that the splay faults may act as permeable conduits for upward pore fluid flow.

We conducted heat flow measurements mainly along a 100 km long NNW-SSE line across the Nankai Trough and the accretionary prism (line A-B in Figure 1), corresponding to a MCS survey line, line 5 of Park et al. [2002]. Most of the drill sites in the IODP NanTroSEIZE (Nankai Trough Seismogenic Zone Experiment) drilling program, which targets the seismogenic zone along the plate interface, are also located along this line [Tobin et al., 2009].

2.2. Measurement of Temperature Gradient in Deep Sea Areas

We used two methods for heat flow measurements according to the magnitude of bottom water temperature variation (BTV). In deep sea areas where the BTV is small, heat flow can be accurately measured in sediments just below the seafloor. If the BTV is large enough to disturb subbottom sediment temperatures, we need to remove the effect of BTV from measured temperature profiles before we can determine heat flow values. Results of previous surveys in the Kumano Trough [e.g., Yamano et al., 1984], where the water depth is about 2100 m, show that vertical temperature profiles within several meters of the seafloor are often nonlinear, indicating that they have been significantly disturbed by the BTV.

In the present study, we conducted heat flow measurements with conventional heat flow probes in the area with water depths deeper than 2500 m, where the bottom water temperature is expected to be more stable than in the Kumano Trough. The probes are 4.5–6.0 m long and have seven temperature sensors. At some stations, temperature profiles were measured with sensors mounted on a piston core sampler. Temperature gradient data were obtained for 64 stations on 14 cruises (Table 1 and Figure 2). Multiple penetrations of the probe were made at most stations to examine local variability. The coordinates of the penetrations in Table 1 are the ship positions at the moment of penetration or the positions of an acoustic transponder attached about 50 m above the probe determined with the acoustic positioning system of the ship.

Figure 2.

Heat flow data in the area southeast of the Kii Peninsula (off Kumano). Circles are the values measured with conventional heat flow probes (larger diameters indicate data of higher quality, cf. Table 1), squares are those obtained through long-term temperature monitoring in surface sediment, and small dots are estimates from depths of methane hydrate BSRs [Ashi et al., 1999]. Colors of the symbols represent heat flow values. Asterisks (W1 and W2) are the stations where long-term bottom water temperature records in Figure 4 were obtained. Stars are IODP Sites C0011 and C0012. The data in the rectangle are projected on the line A-B in Figure 7.

Table 1. Results of Heat Flow Measurementsa
PenetrationLat. (N)Lon. (E)W.D. (m)Pos.NPen. (m)Qual.Grad. (mK/m)AreaH.F. (mW/m2)
  • a

    W.D.: water depth, Pos.: method of positioning (T.P.: acoustic transponder), N: number of temperature sensors used to determine temperature gradient, Pen.: minimum penetration depth of the lowermost sensor used to determine temperature gradient, Qual.: quality of temperature gradient data (see text for definition of classes A to C), Grad.: temperature gradient and uncertainty (±1 SD), Area: morphological division of the study area (T: Nankai Trough floor, S: accretionary prism slope, B: forearc basin; see text for the mean thermal conductivity for each area), H.F.: heat flow and uncertainty.

96BOTR HF03A32°36.43′136°42.05′4300Ship41.5C121.5±0.8T128.6±9.4
96BOTR HF03B32°36.31′136°42.01′4300Ship31.8C114.7±2.0T120.4±10.1
KT-98-09 HF10D32°36.57′135°46.03′4700Ship42.1B187.2±0.4T196.5±13.5
KT-98-09 HF11A32°36.36′135°46.50′4700Ship41.6C185.7±0.7T195.0±13.8
KT-98-09 HF11C32°36.35′135°46.49′4700Ship62.6B190.8±1.2T200.3±14.7
NT01-00 HF01A32°39.96′136°12.02′4624T.P.32.0C143.1±1.1T150.2±11.1
NT01-00 HF01B32°40.05′136°11.67′4619T.P.31.9C154.9±1.0T162.6±11.8
NT01-00 HF01C32°40.04′136°11.61′4620T.P.33.1C154.6±1.7T162.3±12.6
KY02-02 PH0132°50.00′136°53.09′4089T.P.52.3B111.8±1.0T117.4±8.9
KY02-12 HF-1B32°53.08′136°25.92′4477T.P.34.6C101.3±1.9T106.3±9.1
KY02-12 PC332°52.49′136°52.36′4332T.P.53.6A105.0±0.3T110.3±7.7
KY02-12 PC432°56.31′137°5.47′4202T.P.43.4A105.4±0.2T110.7±7.6
KT-02-1 HF02E32°54.46′136°32.37′4447Ship53.4A86.0±0.2T90.3±6.3
KT-02-1 HF02H32°54.17′136°32.33′4445Ship32.8C89.1±1.1T93.6±7.4
KT-02-13 HF01A32°43.80′137°10.67′4135Ship43.0A94.2±0.5T98.9±7.2
KT-02-13 HF01B32°44.01′137°10.57′4135Ship42.3B98.3±3.9T103.2±11.0
KT-02-13 HF01C32°44.22′137°10.49′4135Ship42.7B100.9±1.7T105.9±8.8
KT-02-13 HF02A32°48.76′137°8.90′4172Ship42.4B117.5±1.0T123.4±9.3
KT-02-13 HF02B32°49.04′137°8.94′4195Ship32.3C128.9±0.9T135.3±10.0
KT-02-13 HF02C32°49.59′137°8.19′4158Ship32.2C103.6±1.4T108.8±8.7
KT-02-13 HF02D32°49.46′137°8.31′4158Ship32.2C106.9±3.5T112.3±11.2
KT-02-13 HF02E32°51.17′137°7.88′4123Ship42.7B114.4±0.5T120.1±8.5
KT-02-13 HF02F32°51.18′137°7.81′4123Ship42.5B113.1±0.4T118.7±8.4
KT-02-13 HF02G32°51.19′137°7.74′4123Ship42.3B114.9±0.6T120.6±8.6
KT-02-13 HF04E33°3.34′136°31.73′3235Ship53.4A69.8±0.4S72.6±6.7
KT-02-13 HF04I33°4.27′136°30.73′3281Ship42.3B62.6±1.8S65.1±7.5
KR02-10 HF06A33°0.32′136°48.39′4347T.P.64.8A79.2±0.2T83.2±5.8
KR03-05 HF1A32°43.35′136°20.39′4569T.P.33.9C65.5±3.0T68.8±7.7
KR03-05 PC532°59.43′136°25.25′3943T.P.44.4A111.6±1.2S116.1±11.3
KR04-05 HF03A33°11.49′136°43.51′2934T.P.53.2A61.6±0.1S64.1±5.6
KR04-05 HF03B33°11.49′136°43.51′2934T.P.42.4B78.2±0.6S81.3±7.6
KR04-05 HF06A32°43.28′136°20.26′4580T.P.33.4C122.8±0.5T128.9±9.1
KY04-11 HF03A32°59.39′136°24.49′3850T.P.53.0A150.1±0.3S156.1±13.8
KY04-11 HF03B32°59.37′136°24.49′3850T.P.43.2A127.2±0.9S132.2±12.4
KY04-11 HF03C32°59.36′136°24.50′3850T.P.43.3A129.8±0.8S135.0±12.5
KY04-11 HF04A33°7.45′136°49.46′3840T.P.74.4A90.6±0.2S94.2±8.4
KY04-11 HF04B33°7.44′136°49.47′3840T.P.64.1A89.1±0.5S92.7±8.6
KY04-11 HF05A33°13.10′136°49.94′3280T.P.55.7A71.8±0.4S74.7±6.9
KY04-11 HF05B33°13.08′136°49.95′3280T.P.54.8A78.0±0.1S81.1±7.2
KT-05-2 HF02C33°12.05′136°43.07′2805Ship32.4C58.9±3.9S61.3±9.4
KT-05-2 HF02D33°12.00′136°43.08′2839Ship32.6C58.0±0.8S60.3±6.0
KT-05-2 HF02E33°11.51′136°42.87′2908Ship33.8C57.3±2.6S59.5±7.9
KT-05-2 HF04A32°34.95′136°4.02′4639Ship32.5C174.4±1.4T183.1±13.6
KT-05-2 HF04B32°34.94′136°4.02′4641Ship32.4C182.6±0.5T191.7±13.3
KT-05-2 HF05A32°38.54′135°44.97′4708Ship74.7A152.9±0.8T160.6±11.5
KT-05-2 HF05B32°38.54′135°44.99′4708Ship74.6A157.3±0.5T165.1±11.5
KR05-13 HFPC0232°35.59′135°53.60′4675T.P.53.6A148.2±0.7T155.6±11.1
KT-06-6 HF03A33°12.50′136°42.97′2775Ship32.1C100.7±0.7S104.8±9.8
KT-06-6 HF03B33°12.42′136°42.98′2750Ship32.4C80.6±0.4S83.8±7.7
KT-06-6 HF03C33°12.46′136°42.99′2750Ship32.5C78.4±1.6S81.6±8.7
KT-06-6 HF03D33°11.31′136°44.66′3050Ship35.4C81.4±0.5S84.7±7.9
KT-06-6 HF03F33°11.34′136°44.61′3035Ship52.5B81.5±0.9S84.7±8.2
KT-06-6 HF03G33°11.62′136°46.91′3100Ship32.5C70.3±0.4S73.2±6.7
KT-06-6 HF03H33°11.61′136°46.91′3110Ship32.8C69.1±0.3S71.9±6.6
KT-06-6 HF03I33°12.59′136°50.35′3340Ship32.6C78.9±0.2S82.0±7.3
KT-06-6 HF03J33°12.59′136°50.34′3345Ship32.5C76.2±2.1S79.3±9.0
KT-06-6 HF03K33°12.59′136°50.35′3350Ship32.6C79.5±0.4S82.7±7.5
KT-06-6 HF04A32°44.99′136°54.48′3345Ship61.8C127.0±0.3T133.3±9.2
KT-06-6 HF04B32°44.99′136°54.49′3540Ship51.7C125.7±0.4T132.0±9.2
KT-06-6 HF04L32°44.47′136°58.02′3500Ship41.4C82.8±1.3T86.9±7.2
KT-06-6 HF06E32°34.90′136°18.19′4540Ship74.7A129.1±0.3T135.5±9.4
KT-06-6 HF06F32°34.31′136°19.24′4540Ship74.5A127.5±0.6T133.9±9.5
KT-06-6 HF06G32°34.30′136°19.26′4545Ship74.6A128.9±0.2T135.3±9.2
KT-06-6 HFPC0332°39.82′137°5.97′4160Ship62.8B102.3±1.2T107.4±8.4

The temperature profiles of all the penetrations are almost linear and show no obvious disturbance by the BTV. We should note, however, that even linear temperature profiles may have been influenced by BTV as discussed in 5.2. Temperature gradient and its error (standard deviation) in Table 1 were calculated through least squares fit to temperature profile with a straight line. The quality of the temperature gradient data depends on penetration depth of the probe and the number of penetrated temperature sensors used to calculate gradient as well as the linearity of the profile indicated by the error in the temperature gradient. We thus classified the data into three groups (A, B, and C) based on the following criteria (Table 1): class-A is the data with the probe penetration of over 3 m and more than three penetrated sensors, class-B is with the penetration of 2 to 3 m and more than three sensors, class-C is with the penetration of less than 2 m or three sensors.

2.3. Measurement of Temperature Gradient in Shallow Sea Areas

In relatively shallow seas, the BTV may significantly disturb the subsurface temperature distribution. We made temperature profile measurements with deep-sea probes at some stations in the Kumano Trough as well (not listed in Table 1) and obtained curved temperature profiles at most penetrations. Figure 3 shows an example of the curved profiles, which was measured at station W1 at a water depth of 2070 m (cf. Figure 2). Such curved profiles clearly reflect the effect of the BTV.

Figure 3.

Temperature versus depth profile obtained at station W1 (Figure 2).

We also conducted long-term monitoring of the water temperature on the seafloor for over one year at several stations for examination of the bottom water temperature stability in the area off Kumano. Bottom water temperature records for over two years obtained at stations W1 and W2 (2525 m in water depth) are shown in Figure 4. The amplitude of temperature variations reaches approximately 0.15 K at both stations. These results indicate that it is difficult to obtain reliable heat flow with conventional deep-sea probes in the area shallower than at least 2500 m.

Figure 4.

Long-term bottom water temperature records obtained at stations W1 and W2 (Figure 2).

In the Kumano Trough and its vicinity, we employed the method developed by Hamamoto et al. [2005] for determining heat flow in areas with large BTV. Pop-up type temperature monitoring instruments (hereafter termed “pop-up heat flow instruments”) were used to record temperature profiles in surface sediments for about one year. The instrument consists of a 2–2.5 m-long temperature probe with six to eight equally spaced sensors, weights, a recording unit and an acoustic release system for recovery. Long-term records of sediment temperature profiles for over 270 days were obtained at seven stations with water depths of 2000–2200 m (Table 2 and Figure 2).

Table 2. Results of Heat Flow Measurements Through Long-Term Temperature Monitoringa
StationLat. (N)Lon. (E)W.D. (m)Period (day)NGrad. (mK/m)AreaH.F. (mW/m2)
  • a

    W.D.: water depth, Period: measurement period, N: number of temperature sensors used to determine temperature gradient, Grad.: temperature gradient and uncertainty (±1 SD), Area: morphological division of the study area (see text for the mean thermal conductivity for each area), H.F.: heat flow.


We analyzed the temperature records to examine how temperature disturbance by BTV propagates downward through sediment and found that temperature variation is transferred only by thermal diffusion at all the stations. The influence of BTV at each sensor can then be calculated from the temperature variation at the uppermost sensor. A corrected temperature profile is obtained by subtracting the calculated temperature disturbance from the observed temperature variation at each depth and the temperature gradient is determined. An example of long-term temperature record and correction for the influence of BTV is shown in Figure 5 (station POP-UP3). Thermal diffusivity of sediments can be estimated through this analysis for each pair of temperature sensors and the obtained values mostly lie between 1.5 and 2.7 × 10−7 m2/s, consistent with the values reported for surface sediments [e.g., Goto and Matsubayashi, 2008].

Figure 5.

Example of analysis of long-term temperature data to determine the undisturbed temperature gradient. (a) Temperature variations recorded by the pop-up heat flow instrument at station POP-UP3 (Figure 2). CH1 is the shallowest sensor and CH8 is the deepest one. (b) Residual temperature variations after correction for the influence of BTV. (c) Plot of the undisturbed temperature versus depth. Solid line is least squares fit with a straight line.

The temperature gradient values obtained by the above method (Table 2) still include errors arising from longer period components in BTV, longer than the temperature profile records. We evaluated possible effect of long period components using the bottom water temperature record for three years at W1 (Figure 4). Disturbance in sediment temperature profile by BTV during the three years was calculated based on the bottom water temperature record. Supposing that the temperature profile in the upper two meters of sediment was recorded for 300 days with a pop-up heat flow instrument, 300-day long data sets were extracted from the calculated temperature profiles for the three years. Temperature gradient given by analysis of the data sets varies depending on selected period. The maximum variation reaches about 20% of the mean value. We therefore suppose that uncertainty in the temperature gradient values due to very long period BTV is less than 20% and the heat flow values may have an uncertainty of 20% considering the error in thermal conductivity (cf. 2.4).

2.4. Thermal Conductivity

Thermal conductivity of the sediment was measured in situ with the heat flow probes or on piston core samples in the laboratory (Table 3 and Figure 6). In situ thermal conductivity was obtained using the pulse heating method [Lister, 1979] by generating a calibrated heat pulse along selected temperature sensors. Core sample measurements were made using the transient line heat source method with two different types of probe: needle probe in a full-space configuration [Von Herzen and Maxwell, 1959] or box-shaped probe in a half-space configuration [e.g., Sass et al., 1984].

Figure 6.

Locations of thermal conductivity measurement sites. Circles are in situ measurements and squares are measurements on piston core samples. Colors represent thermal conductivity values. T, S, and B are the three areas for which the mean conductivity values were calculated (cf. Table 3).

Table 3. Results of Thermal Conductivity Measurementsa
PenetrationMethodLat. (N)Lon. (E)W.D. (m)AreaT.C. (W/m/K)
  • a

    Method: measurement method (in situ: in situ measurement by the pulse heating method, NP: measurement with a needle probe, BP: measurement with a box-shaped probe), W.D.: water depth, Area: morphological division of the study area (T: Nankai Trough floor, S: accretionary prism slope, B: forearc basin), T.C.: mean thermal conductivity for each penetration.

KT-98-09 HF07Bin situ33°40.48′136°34.95′2025B0.81
KT-98-09 HF08Bin situ33°40.02′136°33.81′2080B1.12
KT-98-09 HF09Cin situ33°39.88′136°34.21′2080B1.09
KT-98-09 HF11Cin situ32°36.35′135°46.49′4700T0.97
96BOTR HF03Ain situ32°36.43′136°42.05′4300T1.26
96BOTR HF03Bin situ32°36.31′136°42.01′4300T1.35
NT01-00 HF01Ain situ32°39.96′136°12.02′4624T1.10
NT01-00 HF01Cin situ32°40.04′136°11.61′4620T1.01
KT-02-1 HF01Cin situ33°37.91′136°40.27′1975B1.34
KT-02-1 HF01Fin situ33°38.84′136°39.03′2083B0.93
KT-02-1 HF01Gin situ33°38.87′136°38.89′2083B0.88
KT-02-1 HF01Kin situ33°38.49′136°36.72′2085B0.93
KT-02-1 HF01Lin situ33°37.93′136°35.75′2083B0.94
KT-02-1 HF01Min situ33°37.88′136°35.65′2083B0.91
KT-02-1 HF01Nin situ33°37.89′136°35.61′2083B0.93
KT-02-1 HF02Ein situ32°54.46′136°32.37′4447T1.17
KT-02-1 HF02Hin situ32°54.17′136°32.33′4445T1.13
KY02-02 PH01in situ32°50.00′136°53.09′4089T1.02
KY02-02 PH02in situ33°41.02′136°33.43′1943B1.24
KY02-02 PH03in situ33°43.98′136°33.97′1941B1.07
KY02-02 PH04in situ33°41.08′136°33.59′1929B1.04
KY02-02 HF02Cin situ33°41.11′136°33.75′1922B0.92
KR02-10 HF05Ain situ33°19.00′136°40.11′2086S1.02
KT-02-13 HF01Bin situ32°44.01′137°10.57′4135T0.96
KT-02-13 HF01Cin situ32°44.22′137°10.49′4135T0.98
KT-02-13 HF02Cin situ32°49.59′137°8.19′4158T1.06
KT-02-13 HF02Din situ32°49.46′137°8.31′4158T1.05
KT-02-13 HF02Fin situ32°51.18′137°7.81′4123T1.06
KT-02-13 HF04Kin situ33°5.28′136°32.86′3346S1.02
KY02-12 HF1Ain situ32°53.00′136°25.94′4462T1.21
KY02-12 HF-1Bin situ32°53.08′136°25.92′4477T1.16
KT-05-2 HF01Bin situ33°39.00′136°38.51′2080B0.99
KT-05-2 HF01Cin situ33°39.00′136°38.53′2082B0.97
KT-05-2 HF02Bin situ33°13.50′136°42.49′2396S1.12
KT-05-2 HF02Cin situ33°12.05′136°43.07′2805S1.05
KT-05-2 HF02Ein situ33°11.51′136°42.87′2908S1.02
KT-02-1 KK1PCNP33°18.18′136°29.36′1989S1.36
KT-02-1 KK2PCNP33°10.94′136°26.07′2190S1.15
KT-02-1 KK3PCNP33°3.28′136°31.75′3236S1.17
KT-02-1 MV4PCNP33°37.98′136°40.29′1986B1.44
KR03-05 PC1NP33°19.39′136°40.04′2030S0.94
KR03-05 PC3NP33°11.40′136°43.55′2930S1.00
KR03-05 PC5NP32°59.43′136°25.32′3950S0.89
KY02-12 PC-3NP32°52.49′136°52.36′4332T1.05
KY02-12 PC-4NP32°56.31′137°5.47′4202T1.04
KH-01-2 KMP1BP33°40.81′136°36.38′1954B1.07
KH-01-2 KMP2BP33°36.38′136°32.70′2065B1.05

The measured thermal conductivity shows no significant variation with depth at most stations. We calculated means for individual penetrations and listed them in Table 3. The values measured on core samples agree well with those measured in situ at nearby stations. The highest value, 1.44 W/m/K, was obtained on a mud volcano in the Kumano Trough (KT-02-1 MV4PC). Another high value, 1.34 W/m/K, was obtained on the same mud volcano (KT-02-1 HF01C). These outlying values may represent a local anomaly associated with mud volcanism. Excluding anomalous values, higher than 1.2 W/m/K or lower than 0.85 W/m/K, the mean and the standard deviation of thermal conductivity were calculated to be 1.03 and 0.08 W/m/K, respectively.

We divided the study area into three areas according to morphological features (Figure 6): the Kumano Trough (forearc basin), the slope on the frontal accretionary prism, and the Nankai Trough floor including the northernmost part of the Shikoku Basin. The three areas are covered by sediments of different origins, which may have different mineral compositions. Surface sediments in the Kumano Trough generally contain more terrigenous components than hemipelagites deposited on the prism slope. The Nankai Trough floor is filled with turbidites transported along the trough axis from the mountain ranges to the northeast [Taira and Niitsuma, 1986]. The means and the standard deviations of thermal conductivity are 0.99 ± 0.08 W/m/K, 1.04 ± 0.09 W/m/K, and 1.05 ± 0.07 W/m/K for the Kumano Trough, prism slope, and Nankai Trough floor, respectively. The mean thermal conductivity for each area was used to calculate heat flow from the temperature gradient measured at each penetration (Tables 1 and 2).

3. Heat Flow Distribution

Heat flow data obtained in the present study are plotted in Figure 2. In the area with water depths shallower than 2500 m (Kumano Trough and its vicinity), only the values determined through long-term temperature monitoring with pop-up heat flow instruments are shown, since those measured with deep-sea heat flow probes may have been disturbed by the BTV.

Figure 2 also shows heat flow estimated from depths of methane hydrate bottom simulating reflectors (BSRs) by Ashi et al. [1999]. Methane hydrate BSRs are widely distributed on the landward side of the Nankai Trough and have been detected along many seismic reflection survey lines crossing the Kumano Trough and the accretionary prism slope off Kumano. The BSRs are thought to correspond to the boundary between the gas phase and the hydrate phase in marine sediments [e.g., Shipley et al., 1979] and thus indicate the pressure and temperature condition at their depths. Heat flow can be estimated from the depths of the reflectors using the phase relation of the methane hydrate system and thermal conductivity between the seafloor and the reflectors inferred from P wave velocity [Yamano et al., 1982]. It should be noted that heat flow estimated from BSR depths (hereafter termed “BSR heat flow”) contains error of 20–25%, mainly due to uncertainty in velocity and thermal conductivity structures in sediments above the BSRs [e.g., Townend, 1997; Ganguly et al., 2000; Marcaillou et al., 2006]. In some areas where sediments have been actively deformed, BSRs may not be in thermal equilibrium and not reflect regional heat flow [e.g., Kinoshita et al., 2011].

Heat flow data in the rectangle in Figure 2 are projected on the line A-B to construct a heat flow profile across the Nankai Trough, the prism slope, and the Kumano Trough (Figure 7). Heat flow generally decreases landward from about 110 mW/m2 around 15 km seaward of the deformation front to 40–60 mW/m2 in the Kumano Trough. There are few data on the Nankai Trough floor and around the deformation front (cf. Figure 2) because submarine telecommunication cables are densely laid in this area, prohibiting measurements in their vicinities. Highly scattered values are found about 15–25 km landward of the deformation front. The data set in the close vicinity of the line A-B still shows a high scatter, while local variability is low outside this zone. Heat flow data obtained with pop-up heat flow instruments are concordant with BSR heat flow except for one high value, about 90 mW/m2, at a station 70 km from the deformation front. General agreement of the pop-up instrument values with BSR heat flow indicates that errors in these data are relatively low, less than 20%.

Figure 7.

(top) Heat flow data with uncertainties within 30 km of the line A-B (Figure 2) plotted against the distance from the deformation front. Circles are the data obtained with conventional heat flow probe, which are classified into three groups (closed: class A, gray: class B, open: class C) according to the data quality. Squares are the values obtained through long-term temperature monitoring in surface sediment. Crosses are estimates from BSR depths [Ashi et al., 1999]. (bottom) Bathymetry profile along the line A-B.

The age of the Philippine Sea plate (Shikoku Basin) along the Nankai Trough off Kumano is estimated to be about 20 m.y. based on identification of lineated magnetic anomalies [Okino et al., 1994]. It is consistent with the minimum basement age at IODP Site C0012 (cf. Figure 2), 18.9 m.y., obtained from the biostratigraphic age of the overlying pelagic sediment [Underwood et al., 2010]. If we apply the relation between heat flow and seafloor age for normal oceanic lithosphere [e.g., Parsons and Sclater, 1977; Stein and Stein, 1992] to the Shikoku Basin lithosphere, heat flow is estimated to be 105–115 mW/m2 for the age of 20 m.y. On the floor of the Nankai Trough, where the sedimentation rate is high, correction for the effect of sedimentation on the surface heat flow needs to be made. We evaluated the sedimentation effect using the model of Hutchison [1985] and Wang and Davis [1992]. Sedimentation history deduced from the results of IODP drilling [Underwood et al., 2010] and the MCS profile [Park et al., 2002] gives decrease in heat flow at the surface due to sedimentation by about 20% and about 10% around the deformation front and around 20 km seaward of it, respectively. The surface heat flow on the trough floor off Kumano is, therefore, expected to be 85–105 mW/m2, consistent with the values measured along the line A-B, 90–110 mW/m2.

High scatter in the observed heat flow on the accretionary prism slope, 15–25 km landward of the deformation front (Figure 7), may be attributed to the influence of BTV. BSR heat flow, which is hardly affected by BTV, also shows some scatter in this area, though the amplitude is less than that of the surface probe measurements. It suggests that part of the scatter in the surface heat flow results from BTV. Fluid flow activity along faults cutting through the accretionary prism may also yield local heat flow anomalies. The scattered values were measured in the area where branches of the splay fault system reach the seafloor [Park et al., 2002; Moore et al., 2007]. Another possible cause is recent mass movement processes on the upper part of the prism slope in the vicinity of the splay faults [Kimura et al., 2011; Strasser et al., 2011]. Erosion and sedimentation should increase and decrease the surface heat flow. These possibilities are examined in detail in the discussion section. Whichever factor may be the cause, the high scattering cannot originate from a deeper part of the accretionary prism.

The high heat flow of 90 mW/m2 in the Kumano Trough (POP-UP2 in Table 2; 70 km landward of the deformation front) was measured at a station close to a mud volcano. In the Kumano Trough, seismic reflection and side scan sonar imaging surveys revealed the existence of many mud volcanoes. They have been investigated through various geophysical and geological approaches including submersible dives [e.g., Kuramoto et al., 2001; Morita et al., 2002; Sawada et al., 2002]. Bacterial mat and clam colonies, indicators of cold seeps, were found on some of them. Mud volcanoes are surface expressions of conduits that transport mobilized sediment from the deep and thus often accompanied by elevated surface heat flow [e.g., Henry et al., 1996; Kaul et al., 2006]. Goto et al. [2007a] conducted closely spaced temperature profile measurements with short probes on another mud volcano in the Kumano Trough. They corrected for the effect of BTV using a bottom water temperature record for about 300 days and showed that heat flow is highly variable across the mud volcano.

The mud volcano close to the high heat flow (POP-UP2), Kumano Knoll No. 8, is considered to be relatively young and active based on the results of submersible dive, side scan sonar, and subbottom profiler surveys [Morita et al., 2007]. High resolution seismic reflection record crossing this mud volcano shows conspicuous shallowing of BSR beneath it [Baba and Yamada, 2004], indicating a local high heat flow anomaly due to upward flow of fluid and/or mud. The station POP-UP2 is located only about 500 m away from Kumano Knoll No.8. It is, therefore, quite probable that the observed high heat flow originates from localized upward fluid/mud flow associated with activity of the mud volcano.

4. Thermal Structure of the Subduction Zone off Kumano

4.1. Thermal Model

Heat flow measurements at the surface provide important boundary conditions for estimating the subsurface thermal structure. Studies on thermal models of subduction have shown that the thermal structure of the frontal part of forearc, including the subducting plate interface, is controlled mainly by the convergence rate, subduction angle, and age (i.e., the temperature structure) and sedimentation history of the incoming oceanic plate. These parameters can be determined from plate motion models, seismic surveys, and magnetic anomaly lineation maps. Heat generation from the radioactive decay of K, U, Th in the overriding plate and frictional heat production along the plate interface also have significant influence on the thermal structure. However, heat productions cannot be directly measured at depths beyond the penetration of boreholes and hence have large uncertainties. They may be treated as unknown parameters in thermal models and estimated by comparing the calculated surface heat flow and the observation.

Hyndman et al. [1995] calculated temperature fields across the Nankai subduction zone and discussed thermal constraints on the downdip extent of the seismogenic zone along the plate boundary. They used a two-dimensional finite element thermal model of subduction developed by Wang et al. [1995a]. The model is time-dependent in order to take into account the increase in age of the subducting Shikoku Basin lithosphere with time in the past 15 m.y. In the marine area, their modeling results are not constrained by surface heat flow data since there were few reliable data on the landward side of the Nankai Trough. The frictional heating along the plate interface was assumed to be negligible based on the estimate in the Cascadia subduction zone, where the thermal structure is well constrained by surface heat flow data [Wang et al., 1995b]. Yoshioka and Murakami [2007] also neglected frictional heating at the plate interface in modeling temperature distribution of the upper surface of the subducted plate, on the ground that analysis of stress and strain rate data in southwest Japan indicate that the subduction fault is very weak [Wang, 2000].

We modeled the thermal structure of the subduction zone off Kumano within 100 km of the deformation front (Figure 8). The line A-B in Figure 1, along which heat flow measurements were made, was chosen as a transect for modeling. The geometry of the plate interface was determined from a MCS profile along the same line (line 5 of Park et al., 2002). The convergence velocity at this transect was taken to be 4.8 cm/year in the direction of N55°W using the relative motion between the Philippine Sea plate and the southwest Japan forearc block estimated by Loveless and Meade [2010] through analysis of geodetic data obtained with GPS networks.

Figure 8.

Geometry and boundary conditions of the thermal model for the subduction zone off Kumano along the line A-B. Bold line represents frictional heating along the plate interface.

We used a thermal model similar to the one of Wang et al. [1995a] but assumed a steady state (no temporal change of the subducting plate age). The steady state thermal structure is considered to be an adequate approximation in the model region, because the overriding plate (accretionary prism) is less than 15 km thick (Figure 8) and consequently has a relatively short time scale for thermal diffusion, less than a few million years. It is also supported by the result of Wang et al. [1995a] that the difference between the temperature fields calculated with the time-dependent model and with the steady state model is not significant within 100 km of the deformation front.

We numerically solved the heat transfer equation

equation image

where T is temperature, k is thermal conductivity, ρc is volumetric heat capacity, and v is velocity. Q is heat generation consisting of radioactive heat production and frictional heating. The velocity (v) is nonzero only in the subducting-plate part of the model. The values of parameters used in modeling are summarized in Table 4. The conductivity value of 2 W/m/K for the accretionary prism is much higher than those reported in Table 3, because the latter values are determined for surface sediments with very high porosity.

Table 4. Parameter Values Used for the Thermal Model of Subduction
Heat capacity, ρc3.3 MJ/m3/K
Thermal conductivity (oceanic plate), ko3.0 W/m/K
Thermal conductivity (accretionary prism), kp2.0 W/m/K

The boundary conditions are shown in Figure 8. The subducting plate with a thickness of 70 km enters from the seaward boundary and exits from the landward boundary. The temperature structure of the incoming plate is calculated using a one-dimensional finite element model of cooling of half-space taking the effect of sedimentation and compaction into consideration [Wang and Davis, 1992]. The initial temperature of the half-space is 1400°C. The age of the incoming plate is 20 m.y. as inferred from magnetic anomaly lineations and the basement age at IODP Site C0012. The sedimentation history is based on the one obtained at Site C0011 [Underwood et al., 2010] (cf. Figure 2). The horizontal heat flow across the landward boundary is taken to be zero. The upper boundary (seafloor) is maintained at 2°C. The bottom boundary of the subducting slab is kept at a constant temperature, the temperature at the base of the incoming plate calculated with the one-dimensional model. Because the thickness of the slab is much larger than the thermal diffusion length for 2–3 m.y., in which the slab goes through the model region, the bottom boundary condition does not affect the model results.

The frictional heating along the plate interface is given by

equation image

where v is velocity of subduction and τ is shear stress on the interface. The shear stress τ is generally expressed as

equation image


equation image

where μ is the coefficient of friction, σ*n is effective normal stress, and μ′ is the effective coefficient of friction. Because of the very shallow dip of the megathrust, we assumed for simplicity that the normal stress, σn, equals to the lithostatic pressure P, and therefore

equation image

The effective coefficient of friction, μ′, represents influence of both material and pore pressure.

Radioactive heat production is small and negligible in the oceanic crust and mantle. Heat generation in the continental crust is generally concentrated in the upper part of the crust and thought to decrease exponentially with depth. In the Nankai subduction zone, however, the overriding plate is accretionary prisms of various ages mainly composed of terrigenous sediments derived from the Japanese Islands, suggesting that heat generation may not significantly change with depth. It was thus assumed that radioactive heat generation is constant, R, throughout the accretionary prism in the model. R and μ′ were treated as unknown model parameters.

The direction of the presumed plate convergence, N55°W, is oblique to the model section. It is therefore necessary to take account of the obliqueness in calculation of the thermal structure. In equation (1), we used the velocity component parallel to the model section (4.0 cm/year) for the magnitude of v, assuming that the component perpendicular to the model section has no significant effect on heat advection since spatial variation of the temperature structure along the trough axis is negligible. In equation (2), the magnitude of the convergence velocity (4.8 cm/year) was used to evaluate the frictional heating.

4.2. Estimation of the Effective Coefficient of Friction

We calculated the subsurface thermal structure off Kumano with the model described above for various values of the effective coefficient of friction along the plate interface, μ′, and the radioactive heat generation in the overriding plate, R. The surface heat flow distribution obtained for each set of μ′ and R was compared with the observed heat flow to estimate their actual values. The scattered heat flow at 15–25 km landward of the deformation front and the high heat flow at 70 km were neglected in the comparison since they are interpreted as local anomalies due to BTV, fluid/mud flow, or sedimentation/erosion.

Increases in μ′ and R have similar effects on surface heat flow. Consequently various combinations of these parameters give very similar heat flow distributions, all of which are consistent with observations from the deformation front to the Kumano Trough. Heat flow profiles calculated for examples of possible parameter combinations are shown in Figure 9 (μ′ = 0.00 and R = 3.0 μW/m3, 0.05 and 2.0, 0.10 and 1.0). These sets of parameters, however, give different subsurface temperature fields. Figure 10 show temperatures along the plate interface calculated for the three sets corresponding to the heat flow profiles in Figure 9. Temperature differences reach about 30 K at 80 km from the deformation front. One of the parameters needs to be constrained by some other observation for better determination of the thermal structure.

Figure 9.

Observed heat flow with uncertainties projected to the line A-B and heat flow profiles calculated for three combinations of the model parameters, effective coefficient of friction along the plate interface μ′ and radioactive heat generation in the accretionary prism R (μ′ = 0.00 and R = 3.0 μW/m3, 0.05 and 2.0, 0.10 and 1.0).

Figure 10.

Temperatures along the plate interface for the same combinations of the parameters μ′ and R as those used in calculations of the heat flow profiles shown in Figure 9.

Hyndman et al. [1995] estimated the radioactive heat generation of the Nankai accretionary prism to be 2.2 μW/m3 from K, Th and U contents of samples at ODP Site 808 located at toe of the prism off eastern Shikoku (Figure 1). The accretionary prism off Kumano may have a similar amount of heat generation since incoming sediments in both areas were supplied mainly from the central part of Honshu. Heat generation measured on sandstone and mudstone samples from older accretionary prisms exposed on land in southwest Japan is about 1.5 μW/m3 on average [Yamaguchi et al., 2001]. Granitic rocks in southwest Japan also have a similar average value, 1.8 μW/m3 [Miyake et al., 1975]. These data suggest that 2 μW/m3 is an adequate estimate of the heat generation within the accretionary prism off Kumano. Assuming R = 2.0 μW/m3, heat flow profiles calculated for various values of the effective friction coefficient are shown in Figure 11. The profiles corresponding to μ′ values between 0.00 and 0.10 appear to be most consistent with the observed heat flow.

Figure 11.

Heat flow profiles calculated for different values of the effective friction coefficient (μ′ = 0.00 to 0.20) when the radioactive heat production R is 2.0 μW/m3.

The effective coefficient of friction of 0.00 to 0.10 is quite low as compared with the friction coefficients for rocks determined experimentally. If lower values of R are used in the model calculation, higher μ′ values are obtained. We, however, suppose R could not be lower than 1 μW/m3 considering that the accretionary prism is composed of terrigenous sediments. The effective coefficient of friction in the subduction zone off Kumano is therefore estimated to be 0.1 or lower and probably does not exceed 0.2, indicating that the shear stress on the plate interface is low. Estimation of the effective friction coefficient or shear stress based on surface heat flow data in forearc areas has been attempted in some other subduction zones as well, e.g., northeast Japan, Cascade, Mexico, and Costa Rica [Furukawa and Uyeda, 1989; Wang et al., 1995b; Currie et al., 2002; Harris and Wang, 2002]. In most of these subduction zones, similar low friction coefficients, generally less than 0.1, and low shear stress were obtained, consistent with our result in the Nankai subduction zone.

The above discussion was made with fixed values of other model parameters. As can be seen in Figures 9 and 11, the calculated heat flow in the Kumano Trough (beyond 30 km from the deformation front) is critical for the estimation of the effective frictional coefficient or shear stress along the plate interface. The thermal conductivity of the accretionary prism and the convergence velocity of the Philippine Sea plate may have significant influence on the surface heat flow in the Kumano Trough. We thus examined possible effects of uncertainties in these parameters on the above result.

Various models have been presented on the current motion of the Philippine Sea plate based on earthquake slip vectors and tectonic processes along the boundaries [e.g., Seno et al., 1993]. Recent studies presented global plate motion models or regional models in and around Japan using GPS station velocities as main constraint [e.g., Miyazaki and Heki, 2001; Sella et al., 2002; DeMets et al., 2010]. These models give convergence velocities in the area off Kumano in the direction of N50°W to N60°W with rates of 4.5 to 6.5 cm/year. Differences among the models partly result from variety in selection of the overriding plate or crustal block. Loveless and Meade [2010] treated the forearc of southwest Japan between the Median Tectonic Line and the Nankai Trough as a tectonic block. We used the relative motion of the Philippine Sea plate with respect to this block, because it is most appropriate to our model region. Inside of this forearc block, strain partitioning may be taking place, as Martin et al. [2010] demonstrated strike-slip faulting along the outer ridge of the Kumano Trough through analysis of 3-D seismic reflection data. It means the convergence velocity may vary within the block. We should also note that the motion derived from GPS velocity data for several years may be somewhat different from the average motion over hundreds or thousands of years.

We further need to consider the possibility that the motion of the Philippine Sea plate changed with time, since the temperature structure and surface heat flow in the model region should reflect subduction history for the last two to three million years. Studies on geology and structures on land indicate that subduction in a similar direction to the present one has continued for two million years or longer [e.g., Seno and Maruyama, 1984; Takahashi, 2006], while the convergence rate might have changed.

In order to evaluate the influence of variations in the convergence rate, we conducted calculation with μ′ of 0.05 and R of 2.0 μW/m3 for a convergence rate varying from 3.8 to 5.8 cm/year and found that the corresponding variation of the surface heat flow in the Kumano Trough is negligible, within 2 mW/m2. This means that enhancement of cooling by heat advection due to increase in the velocity of subducting slab nearly compensates the corresponding increase of frictional heating. We also examined the effect of varying the convergence direction for the same μ′ and R values. The surface heat flow in the Kumano Trough shows only a small change, about 4 mW/m2, for variation from N35°W to N75°W. Moderate change in the convergence rate and/or direction hence has no significant effect on the surface heat flow.

The thermal conductivity of the accretionary prism was assumed to be uniform, 2.0 W/m/K in our base model, whereas it actually varies depending on the porosity and mineral composition. Higher prism thermal conductivity should result in higher surface heat flow in the Kumano Trough. Calculation with various prism conductivity values showed that conductivity increase (or decrease) by 0.3 W/m/K yields increase (or decrease) in the calculated heat flow by about 4 mW/m2. We also made calculation with thermal conductivity increasing landward from 2.0 W/m/K at the prism toe to 2.5 W/m/K at 100 km from the deformation front. The heat flow at 40 to 70 km from the deformation front is practically the same as the one for the case with uniform conductivity of 2.3 W/m/K. Consequently, uncertainties in the convergence velocity and the thermal conductivity of the prism give an uncertainty of about ±5 mW/m2 in the heat flow in the Kumano Trough, which corresponds to varying μ′ by ±0.05 (cf. Figure 11). Taking these uncertainties into account, therefore, we can still conclude that the effective coefficient of friction in the study area is low, about 0.1 or lower.

Figure 12 is the subsurface temperature distribution calculated with our base model for one of the optimal combinations of μ′ and R values (0.05 and 2.0 μW/m3). It should provide a better estimate of the thermal structure of the seismogenic zone off Kumano than previous models, which were not constrained by heat flow data on the accretionary prism. Actually, the surface heat flow profile and temperature along the plate interface we obtained are very similar to the ones calculated by Hyndman et al. [1995] and Yoshioka and Murakami [2007], because they assumed frictional heating along the plate interface is negligible. We confirmed their assumption by comparing the calculated heat flow with the observation. NanTroSEIZE deep drilling may provide valuable thermal data such as in situ temperature, thermal conductivity, and radioactive heat generation, which can further constrain thermal models of this subduction zone.

Figure 12.

Thermal structure of the Nankai subduction zone off Kumano with 50°C isotherms (along the line A-B in Figure 2). The effective friction coefficient is 0.05 and the radioactive heat production is 2.0 μW/m3. The broken line represents the plate interface.

5. Discussion

5.1. Temperature Structure of the Shikoku Basin Lithosphere

The temperature structure of the incoming plate is a critically important factor in thermal models of subduction. In the model of the Nankai subduction zone off Kumano presented above, we assumed that the incoming Shikoku Basin lithosphere has normal temperature structure corresponding to its seafloor age (about 20 m.y.). The heat flow observed on the Nankai Trough floor along the line A-B agrees with that expected for the seafloor age considering the effect of sedimentation, though the number of data is limited. Kinoshita et al. [2008] conducted heat flow measurements in the area seaward of the trough floor (southeast of the rectangle in Figure 2). They found that heat flow in the northern margin of the Shikoku Basin, which must be less affected by sedimentation, ranges 120–147 mW/m2, appreciably higher than the value expected from the age. It indicates a possibility that the temperature structure of the Shikoku Basin lithosphere off Kumano is not normal for its age.

If we assume that the incoming plate has a temperature field corresponding to an age of 15 m.y., which is more consistent with the high heat flow observed in the northern margin of the Shikoku Basin, the heat flow profile calculated with our model is higher than that for the 20 m.y. case by 5–10 mW/m2 (Figure 13). In this case, lower values of μ′ and/or R are required to fit the data in the Kumano Trough and the corresponding subsurface temperature structure is different from that shown in Figure 12. The incoming plate with 15 m.y. temperature structure is a somewhat arbitrary assumption and there are other possibilities to account for the heat flow observed in the northern Shikoku Basin.

Figure 13.

Heat flow profiles calculated for the incoming plate age of 15 m.y. and 20 m.y. The effective friction coefficient is 0.05 and the radioactive heat production is 2.0 μW/m3.

Additional information on the temperature structure of the Shikoku Basin may be provided by heat flow distribution along the axis of the Nankai Trough. In our study area, heat flow on the trough floor in the western part, south of the Kii Peninsula, is higher than in the eastern part (Figure 2). The highest values in the western part, about 180 mW/m2, are comparable to anomalously high values observed on the trough floor off eastern Shikoku, 200 ± 20 mW/m2 [Yamano et al., 2003]. More detailed measurements should be made between the areas off eastern Shikoku and off Kumano in order to reveal how surface heat flow varies along the trough, which will be valuable information for investigation of the cause of the high heat flow anomaly and the temperature structure of the Shikoku Basin.

5.2. Scattered Heat Flow on the Accretionary Prism Slope

Possible causes of the scattered heat flow observed around 15–25 km landward of the deformation front (Figure 7) are pore fluid flow along thrust faults, disturbance by BTV, and/or deformation processes near the surface. Thrust faults cutting through the accretionary prism may act as permeable paths for pore fluid expulsion from the prism and subducted sediments and oceanic crust. Heat flow and subsurface temperature anomalies associated with thrust faults were found in other accretionary prisms such as Barbados and Cascade margins and interpreted as results of upward fluid flow along the faults [e.g., Fisher and Hounslow, 1990; Zwart et al., 1996]. In the Nankai prism off eastern Shikoku, a local high heat flow anomaly was detected just on a thrust fault at the prism toe [Fujino and Kinoshita, 2011].

In the present study area, submersible surveys revealed the existence of chemosynthetic communities and cold seeps around the seafloor scarps associated with the splay faults [Ashi et al., 2002; Toki et al., 2004]. Goto et al. [2007b] obtained high heat flow over 100 mW/m2 through long-term monitoring experiments with short temperature probes in a biological community at station W2 (Figure 2). One of the temperature profile records is better explained by a model taking account of advective heat transfer by upward fluid flow. These results indicate that upward fluid flow along the splay fault system might result in the observed high and variable heat flow.

BTV with large amplitude significantly affects surface temperature gradient and may cause scattering in heat flow measured with conventional probes. Influence of BTV can be evaluated based on long-term water temperature data. We, however, have no long-term record in a water depth range of 2750–3350 m, where the scattered values were obtained. At a station with a shallower depth, 2525 m (W2 in Figure 2), large BTV with amplitude of about 0.15 K was observed (Figure 4), while the bottom water temperature was very stable at a station on the Nankai Trough floor south of the Kii Peninsula (at 4685 m).

We analyzed the temperature record at W2 for about 900 days to evaluate possible variation in measured heat flow. Subbottom temperatures down to 4 m below the seafloor were calculated every 1 day based on the bottom water temperature record supposing that a heat flow probe with eight temperature sensors at intervals of 0.5 m penetrates into sediments by 4 m (Figure 14). The undisturbed temperature gradient and the thermal diffusivity of sediment were assumed to be 50 mK/m and 3 × 10−7 m2/s respectively. We then determined the temperature gradient at each time by fitting a straight line to the temperatures of the eight sensors. Eliminating obviously nonlinear profiles, as the temperature profiles actually observed were apparently linear, the gradient ranges from 43 to 57 mK/m. It means that heat flow variation up to about 15 mW/m2 can arise from BTV at this water depth (2525 m). Heat flow variation is probably smaller in the area with deeper water depths (2750–3350 m) but may still account for part of the observed scatter, 20–30 mW/m2. For further examination, monitoring of the bottom water temperature should be conducted on the prism slope. Repeated measurements with conventional probes at the same site will also be effective in evaluation of influence of BTV.

Figure 14.

Temporal variation of subbottom temperature profile at station W2 (Figure 2) calculated from the long-term bottom water temperature record (Figure 4), assuming that the undisturbed temperature gradient is 50 mK/m and the sediment thermal diffusivity is 3 × 10−7 m2/s.

Some recent deformation near the seafloor may have disturbed the temperature structure of surface sediment, which has not reached thermal equilibrium with deeper part of the accretionary prism. Analysis of seismic reflection data, surface topography, and geological data obtained through IODP drilling suggests that some branches of the splay fault are currently active and slumping and mass transport repeatedly occurred over the upper part of the prism slope [e.g., Moore et al., 2007; Sakaguchi et al., 2011; Kimura et al., 2011; Strasser et al., 2011]; the most recent event was possibly caused by the 1944 Tonankai earthquake. On the accretionary prism in the eastern part of the Nankai subduction zone, Martin et al. [2004] found anomalies in BSR heat flow correlated with topographic features and showed that they can be attributed to active erosion and sedimentation processes. Kinoshita et al. [2011] revealed that BSRs are anomalously shallow near the axis of anticlines in the middle part of the prism slope off Kumano, just adjacent to the area where we obtained the scattered heat flow data. They interpreted the anomalies as a result of recent uplift and erosion due to thrust faulting. It is probable that similar tectonic activities cause local heat flow variations with the observed amplitude.

We suppose that the scatter of the measured heat flow is most probably a complex product of the factors discussed above. More detailed heat flow measurements in the prism slope area taking account of relative positions to topographic and tectonic features will allow better understanding of this problem.

6. Conclusions

We conducted heat flow surveys in the central part of the Nankai subduction zone off Kumano with ordinary probes in deep sea areas and with pop-up type monitoring instruments in shallower sea areas. The obtained data are generally consistent with the heat flow estimated from BSR depths around the stations. The surface and BSR heat flow was projected onto a MCS profile crossing the Nankai Trough and the accretionary prism to construct a heat flow profile across the subduction zone. Heat flow decreases landward from about 100 mW/m2 on the floor of the Nankai Trough to about 50 mW/m2 in the Kumano Trough (forearc basin), beyond 30 km from the deformation front. The values obtained on the Nankai Trough floor are concordant with that estimated from the seafloor age, considering the effect of recent rapid sedimentation. The measured heat flow shows a high scatter on the prism slope, 15–25 km from the deformation front, which probably results from processes near the seafloor such as fluid flow along thrust faults, disturbance by BTV, and sedimentation/erosion.

We calculated the temperature field along the MCS and heat flow profiles off Kumano using a two-dimensional, steady state finite element model. Two unknown parameters, the amounts of frictional heating along the plate interface and radioactive heat production in the accretionary prism, can be constrained by the surface heat flow observed in the Kumano Trough. Combinations of the parameter values which fit the measured heat flow profile, however, still yield a variety of subsurface thermal structures. We may assume that the radioactive heat generation is about 2 μW/m3 based on the values obtained at the prism toe off Shikoku and old prisms on land in southwest Japan. The amount of frictional heating is then better constrained and the effective coefficient of friction along the plate interface is estimated to be around 0.1 or lower. Similar low friction coefficient values were reported in other subduction zones, indicating that the shear stress on the subduction plate interface is generally low.


We are grateful to the crew and scientific parties of the cruises KT-98-9, KT-02-1, KT-02-13, KT-05-2, KT-06-6 (R/V Tansei-maru), KY02–02, KY02–12, KY04–11 (R/V Kaiyo), KR02–10, KR03–05, KR04–05, KR05–13 (R/V Kairei), 96BOTR (R/V Bosei-maru), NT01–00 (R/V Natsushima) and YK03–03 (R/V Yokosuka) for their assistance in heat flow measurement and core sampling operations. Discussion with O. Matsubayashi on heat flow measurement in shallow sea areas was very helpful. We thank S. Saito for providing information on sedimentation history in the Shikoku Basin. We also thank R. Harris, B. Marcaillou, anonymous reviewers and the Associate Editor for valuable comments for improvement of the manuscript. This work was supported by a Grant-in-Aid for Creative Basic Research from Ministry of Education, Culture, Sports, Science and Technology, Japan (09NP1101) and a Grant-in-Aid for Scientific Research from Japan Society for the Promotion of Science (16340126).