NANKA!

TROUGH:

A HOT TRENCH?

M. Y A M A N O , S. H O N D A , and S. U Y E D A

Earthquake Research Institute, Univ. of Tokyo, Tokyo, Japan

(Accepted 19 August, 1983) Abstract. Heat flow estimated from the gas hydrate layers on the landward slope of the Nankai Trough reveals that heat flow increases downslope toward the trench floor. This data plus six new heat flow values obtained by a conventional probe and two values available from DSDP drill holes give a fairly detailed heat flow distribution in the Nankai Trough area, when combined with the already existing data set. There appears to be a zonal pattern parallel to the trough axis, with a high heat flow zone on the floor of the trough that is quite anomalous for a subduction zone. It might be explained as a result of subduction of the hot portion of the Philippine Sea plate, i.e. the Shikoku Basin, and/or of more local effects such as heating due to intrusion of hot water from subducted sediments to shallow depth beneath the trough floor. Surface heat flow patterns landward of the trough were calculated for a simple thermal model of subduction. Perfect reproduction of the observed zonal pattern is difficult to achieve by the simple model, suggesting the necessity for further heat flow and other observations.

1. Introduction The Nankai Trough is located off the coast of the Southwest Japan and regarded as the place where the Philippine Sea plate is subducting beneath the Eurasian plate. The trough and the Southwest Japan arc have some anomalous characteristics compared with many other subduction zones (e.g. Yoshii and Kobayashi, 1981). For instance, in the Nankai Trough, the maximum water depth does not exceed 5000 m. The gravity anomaly is also smaller than in the other trenches. The seismic zone accompanied with subduction reaches only to a depth of about 70 km. Although there are a few Quaternary volcanoes, the volcanic front can only barely be defined. In contrast, in the middle Miocene time, unusual magmatism occurred much closer to the trough than the present volcanoes. There is a possibility that these features are related to the subduction of a young plate, the Shikoku Basin plate (Tatsumi, 1982; Sugi and Uyeda, 1984). Many heat flow measurements have been carried out in the vicinity of the Nankai Trough (see the compilations by Anderson et al. (1978) and Yoshii (1979)). The data shows that heat flow is very high, higher than 130 mW m -2, on the floor of the trough (Watanabe et al., 1970). On the landward ~lope of the trough, heat flow seems to be lower, at most 100 m W m -2. There have been, however, only a few data on the landward slope and the variation of heat flow from the trench to the arc has not been well-known. In this paper we add new heat flow data estimated from the depths defining the gas hydrate layers in the sediments and measured by a conventional geothermal probe and present a more detailed heat flow distribution in the Nankai Trough area. So far low heat flow values, about 40 mW m-2,'have been observed in many Marine Geophysical Researches 6 (1984) 187-203. 0025-3235/84/0062-0187 $02.55 9 1984 by D. Reidel Publishing Company.

~/

,#

70

72

132"

____.~++r ~

/

42-

c

,

9

9

1 3"

K54-I-2

51 9

134"

o9 )

50

-

+~

\

100405

9

o 32-46

135"

'

,3.

9 90

9

~.28

. v

136"

',,."(~

52

9

~Ss \~.4 137"

SHIKOKU~q~-72 55 ~ BAS'N~\~r ~J~147

/,2

~.1,:'r 75 ~ ~' ~COk"~

:7_._~......z____~

~

6~

83

138~

Fig. 1. Heat flow values estimated from gas hydrates and measured by conventional means (in mW m-2). Closed circles represent the former measurements (Anderson et aL, 1978; Yoshii, 19791 and unpublished data by Matsubara) and squares the measurements in this study (Table I) and stars on DSDP Leg 87 (Kagami el al., 1983). Depth contours are in meters.

30 31~

31"~

32~

33'

34"

35~ N

OO OO

A HOT TRENCH7

189

trench areas (e.g. Watanabe et al., 1977). Therefore the Nankai Trough, where heat flow is very high, appears to be quite anomalous. We attempt to explain this anomalous high heat flow and then discuss what the heat flow variation in the arc-trench gap may imply.

2. New Heat Flow Data

2.1. GAS HYDRATE METHOD On the landward slope of the Nankai Trough, prominent bottom simulating reflectors (BSRs) are found by multichannel seismic reflection surveys (Aoki et al., 1983; Nasu et al., 1982). These reflectors are believed to be the phase boundaries between gas hydrate and free gas. We have demonstrated in an earlier paper (Yamano et al., 1982) that heat flow values can be estimated from the depth of such BSRs. Temperature at the reflector is found by using the phase relation of the gas hydrate system and inferred density of the marine sediments. Then mean geothermal gradient between the sea floor and the BSR is calculated using the bottom water temperature. Finally heat flow can be obtained by multiplying the geothermal gradient by inferred thermal conductivity of the sediment. This new method (termed gas hydrate method below) enables us to obtain heat flow continuously along the seismic profiles which exhibit BSRs. Although the absolute accuracy of this method is not high, trends can be clearly established. Yamano et al. (1982) estimated the heat flow by this method along the three seismic profiles shown in Figure 1. An example showing the calculated heat flow values plotted on reflection records is reproduced in Figure 2. It can be clearly observed that the estimated heat flow increases downslope toward the floor of the Nankai Trough. This tendency is most pronounced on the easternmost line, N55-1. 2 . 2 . MEASUREMENTS BY GEOTHERMAL PROBE

New heat flow measurements were carried out at seven stations (R-1 to 7 in Table I) on the cruise of Japan Meteorological Agency R/V Ryofu-Maru in December 1981. We used a 3 m Bullard type probe which has seven equally-spaced thermistors. Temperature-depth profiles were obtained at six stations and are shown in Figure 3. The estimated errors of temperature are shown by bars in Figure 3. These errors are estimated from the errors in calculation of equilibrium temperature in the sediments and those in thermistor calibration. The equilibrium temperatures are calculated by extrapolation using the theory of Bullard (1954) and the errors are generally 0.002 to 0.003 ~ C. Thermistor calibration is made using temperature records of water immediately above the sea floor. In some cases temperature records were unstable and the errors come up to 0.006~ C or more. Geothermal gradients were determined by the least-squares fit taking these errors into consideration (Table I). The temperature profiles seem non-linear at R-4 to 7, but

9---~.-

/72") , -

, t

B ~ C

?ig. 2. 24-fold migrated time sections on the landward slope of the Nankai Trough (line N55-1 in Figure 1; Nasu et al., 1982) and heat flow values estimated by the gas hydrate method (in mW m-2; from Yamano et al., 1982).

.........

.a

A HOT TRENCH?

R-I

R-2

R-/-.

191

R-6

R-5

R-7

0

~0.5 II.-

\

I..i.I

,,,I.0

\

i,,-,i

I-_,I

I.i.I r162

1.5

2.0 Fig. 3. T e m p e r a t u r e - d e p t h profiles m e a s u r e d on the cruise of R / V Ryofu-Maru. Their positions are listed in Table I. Estimated errors in temperature are shown by bars.

at the stations R--4, 5, and 7 they are linear if the uppermost thermistor is excluded. It appears that the temperature of the near surface sediment is disturbed by temporal change of bottom water temperature. Gradients at R-4 to 7 were calculated excluding the uppermost temperature data. The best fit lines are shown in Figure 3. Errors in these gradient values are estimated to be generally 5-10%. TABLE I Heat flow m e a s u r e m e n t s in the Nankai T r o u g h area

Station

North latitude

East longitude

Uncorrected water depth (m)

R-1 R-2 R-3 R-4 R-5 R-6 R-7

32056.6 ' 32024.6 , 32012.0 ' 32019.0 , 33015.6 ' 33024.5 ' 33o34.2 ,

134045.9 ' 134055.9 , 135033.5 ' 136033.8 ' 136"22.4' 136o22.4 ' 136o22.4 '

1500 4450 4380 4075 2025 2060 1930

aNumbers of thermistors in mud. b A s s u m e d values.

Gradient (inK m -1)

Conductivity (W m -1 K 1)

Heat flow (roW m -2)

32.8 100.6

1.01 0.92 b

33.1 92.6

100.8 72.2 114.9 104.7

0.92 b 1.09 0.84 b 0.84 b

92.7 78.7 96.5 87.9

Na 3 5 1 4 5 3 4

200

9

9

100

!

00

oO

9

+

0

9

AXIS

TROUGH

+.e e

+

oi ~ 00

9

I

oo 0

9

9

9

132"

9

9

9

-200

I

o 0

13z,"

DISTANCE (kin)

-100

o

28' 130"

30'

9

]36"

9

9

138"

i

-300

9

140"E

Fig. 4. Plot of heat flow values in the region shown in the inset versus the distance from the axis of the Nankai Trough. Crosses represent the values estimated from gas hydrates and closed circles those measured by conventional means (Anderson et al., 1978; Yoshii, 1979; unpublished data by matsubara; and this study).

0

7- 50

ILl

I-'-

1,100

b

O

E 50

~E200

250

32"

34"

35" N

0 01 ,4 >

>

to

A HOT TRENCH.'?

193

Core samples were taken at R-1 and 5 and thermal conductivity was measured by the needle probe method (von Herzen and Maxwell, 1959). From these data heat flow values were obtained as presented in Table I. At R-2, 4, 6, and 7, the conductivity values measured at nearby former stations were used. Errors in these assumed conductivity are probably about 10%. As a result the heat flow values are estimated to have errors of about 15%. The heat flow values estimated by the gas hydrate method and those newly measured by the conventional probe are plotted in Figure 1 (squares). It should be noted that the station R-2 is close to the profile N55-1 and the measured value, 93 mW m -2, agrees with the estimation from gas hydrate within the limit of errors.

3. Pattern o| Heat Flow Distribution The heat flow values observed previously in this area are also plotted (closed circles) together with the data obtained in this study in Figure 1, which gives a fairly detailed heat flow distribution in the Nankai Trough area. Also included in this figure are two heat flow results of DSDP Leg 87 (stars; Kagami et al., 1983). These results taken along the multichannel seismic profile, N55-3-1, (63 to 66 mW m -2) appear to be consistent with the downslope increase in heat flow estimated by the gas hydrate method (Figure 1). As a whole a zonal pattern parallel to the axis of the trough is recognizable and contours of heat flow may be drawn as in Figure 1. There is a high heat flow zone (higher than 130 mW m -2) on the floor of the trough. On the landward slope heat flow decreases upslope to about 40 mW m -2. It, then, becomes higher, around 90 m W m -2, on the deep-sea terraces where the water depth is about 1500 to 2000 m. This moderately high heat flow zone seems to extend to Shikoku and Ki-i Peninsula. It decreases again further north around the Seto Inland Sea. Present data are insufficient to know whether the zonal pattern extends eastward further than about 136 ~ E. On the other hand, in the Shikoku Basin south of the Nankai Trough heat flow values show a large scatter and no distinct pattern can be found. But the heat flow seems to be lower than on the trough floor. In Figure 4 heat flow values in the region shown in the inset are plotted against the distance from the axis of the trough. The positive distance means landward of the trough and the negative seaward. The data in a rather wide area are plotted together, so that the pattern becomes more or less ambiguous, but still the zonal .distribution of heat flow can be seen. Especially the values estimated from gas hydrates (crosses) clearly demonstrate a landward decrease. It is also clear that heat flow in the Shikoku Basin is lower than on the floor of the trough and very scattered (see Figure 5 also).

194

M. Y A M A N O ET AL.

4. Possible Causes o! High Heat Flow on the Floor ot the Nankai Trough The high heat flow zone on the floor of the Nankai Trough is unusual since low heat flow is normally found in subduction zones. In the Nankai Trough, the Philippine Sea plate is bent downward at only a very shallow angle when subducting beneath the trough floor. In such a case, decreased heat flow may not be expected. However, heat flow is much higher than in the Shikoku Basin, Two interpretations may be suggested for this anomalously high heat flow: (1) the plate subducting in the Nankai Trough is essentially hot; (2) the cause of high heat flow is not in the subducting plate but in the sediments and/or the upper crust. First, the possibility that the subducting plate is hot enough to give a high heat flow of about 150 mW m -z is examined. It may be possible that only the part of the Philippine Sea plate beneath the trough floor has become hot locally (e.g. by intrusions of magma). However, some ad hoc thermal activity must then be assumed beneath the trough. This seems unjustified without some other evidence. It is more reasonable to assume that the whole Shikoku Basin is hot. At first thought, it may appear to contradict with the fact that heat flow on the trough floor is higher than in the Shikoku Basin. In young oceanic plates, however, observed heat flow values are almost always highly scattered and the mean values are lower than those predicted by thermal models of oceanic lithosphere (e.g. Parsons and Sclater, 1977; Lister, 1977). This fact is attributed to hydrothermal circulation in the crust. Since convection of sea water occurring in the crust transfers some heat advectively, the average conductive heat flux should be lowered. The circulation becomes sealed as the sediment thickness increases, so that heat is transfered only by conduction, and the observed heat flow approaches the theoretical value (see the review by Anderson and Skilbeck, 1981). Thus, if hydrothermal circulation lowers the observed heat flow values in the Shikoku Basin, and the thick sediments in the Nankai Trough seal off the convection, the theoretical high heat flow will be observed in the trough bottom. There are three problems with this explanation. The first is the maximum theoretical heat flow for the plate of the Shikoku Basin. Although it is still somewhat uncertain whether the heat flow versus age relationship holds in marginal basins such as the Shikoku Basin, in several marginal basins whose ages have been obtained, the mean reliable heat flow values seem to agree closely with the theoretical values for normal oceans (Sclater et al., 1980; Anderson, 1980). The age of the Shikoku Basin has been estimated from identification of magnetic anomalies (e.g. Kobayashi and Nakada, 1978; Shih, 1980). In Figure 5, the observed heat flow values in the Shikoku Basin north of 26 ~ N are plotted as black dots against the age given by Kobayashi and Nakada (1978) based on the magnetic reversal time scale by LaBrecque et al. (1977). The solid curve is the theoretical heat flow versus age calculated from the relation ' Q ( t ) = 4 7 3 t-~/2mWm -2' (t = age in m.y.) by Parsons and Sclater (1977). As expected, the heat flow values of the basin are scattered and generally lie below the theoretical curve. On the

A HOT TRENCH?

195

W

E

250 A

c~

E 200

E

150

AA

9

k 0 0 _2 Ii k-

<[ tu I

9

9

100

50

odl 9 o

9

~

9

9 0

I

I

24

22

9

IO

20

9

9 O0 O0 9

00

I

I

I

1

18

18

20

22

I

24

AGE (m. y.) Fig. 5. Heat flow values in the Shikoku Basin north of 26~ plotted versus the age estimated from interpretation of magnetic anomalies by Kobayashi and Nakada (1978) and the time scale by LaBrecque et al. (1977). Sources of data are the same as in Figure 4.

other hand, the observed heat flow values in the Nankai Trough (triangles) appear to support our explanation, although they are somewhat higher than expected from the age. DSDP drilling, however, revealed that widespread off-ridge volcanism occurred in the Shikoku Basin from 13 to 15 m.y.b.p. (Klein et al., 1978; Klein and Kobayashi, 1980). The theoretical value corresponding to this age of off-ridge volcanism is about 1 3 0 m W m -2, which gives a better agreement with the observation. Moreover there may be a possibility that volcanism or thermal event without volcanism occurred in more recent times. Shiono et al. (1980) estimated the plate thickness in the Shikoku Basin to be about 30 km from surface wave dispersions. If the theoretical relationship between the age and the thickness of the lithosphere (e.g. Parker and Oldenburg, 1973; Kono and Yoshii, 1975) can be used, this value suggests the age of about 10 m.y., corresponding to the heat flow of around 150 mW m -2. Considering all these arguments, it seems possible that the total heat flow in the Shikoku Basin reaches 150 mW m -2. The second problem is whether hydrothermal circulation is actually active in the Shikoku Basin or not. The existence of convective heat transfer is suggested by the fact that the observed heat flow values are very scattered and the mean value is lower than that estimated from the magnetic-anomaly age as can be seen in Figure

196

r~. Y A M A N O ET AL.

5. On the other hand, the sediment thickness in the Shikoku Basin is around 300 m from seismic reflection records (e.g. Nasu et al., 1977; Klein et al., 1980). It may be too thick to permit convective heat transfer. In most oceanic plates hydrothermal circulations appear to become sealed when the sediment thickness reaches about 150 to 200 m. A sufficient condition for sealing, however, should depend not only on the sediment thickness but also on the ratio akb/ks (a = wave number of circulation, kb = oceanic crustal permeability, ks = sediment permeability) according to Anderson and Skilbeck (1981). Therefore 300 m thick sediments may still permit convective heat transfer, if akb/ks is smaller. Actually Herman et al. (1977) reported that the boundary between the regions of convective and conductive heat transfer roughly coincides with the 300 m sedimentary isopach in the eastern Atlantic Ocean. Critical thickness probably depends strongly on sediment type. In addition there is a possibility that the surface heat flow is lowered without active hydrothermal circulations through the sedimentary cover. Langseth and Herman (1981) found that, in the western flank of the Mid-Atlantic Ridge in the Brazil Basin, heat flow values observed above 200 m thick sediments are much lower than the theoretical values. According to their interpretation this phenomenon results from cold bottom water penetrating laterally into the upper igneous crust from basement outcrops and absorbing heat flux from below. Since the topography is rather rough and basement outcrops exist also in the Shikoku Basin, similar mechanism might be operating. The third problem is the effect of sedimentation. As the sedimentation rate in the trough floor is rapid, the influence on heat flow cannot be neglected. The seismic reflection records indicate that the sediment thickness increases by about 500 m while the Philippine Sea plate moves by 20 km in the Nankai Trough. Using a convergence rate of about 4 cm yr -j (Seno, 1977), the sedimentation rate is evaluated to be about 1000 m m.y. -~ (The averaged sedimentation rate at DSDP Site 582 on the trough floor is reported to be in the range of 300 to 900 m m.y.-t; Kagami et al., 1983). If it is assumed that sedimentation has continued at this rate for 0.5 m.y. and thermal diffusivity of the sediments is 2.5 • 1 0 _7 m 2 sec -~, the surface heat flow should be reduced by about 20% (Langseth et al., 1980). This estimation suggests that heat flow in the Nankai Trough should be still higher, around 170 mW m -2. If the above stated explanation on high heat flow in the Nankai Trough is correct, similar high heat flow may be observed in other trenches where subducting plates are hot. But there are few data in the trenches where young plates are subducting. A subduction zone which is possibly in the similar state to the Nankai Trough is that off the western North America where the Juan de Fuca plate is under thrusting beneath the American plate. Though there is no trench topography, considerable amount of evidence for active underthrusting exists and it is believed that the trench has been filled with sediments (e.g. Riddihough, 1978). Off northern Oregon many heat flow measurements have been made in an area

A HOT TI~ENCm

197

where the oceanic plate seems to begin to subduct and most of them show high values, 80 to 150 mW m -2 (Krogan et al., 1971). But they are lower than those expected from the age of 8m.y., about 1 7 0 m W m 2. It may be due to the sedimentation effect. Next we discuss the possibility that the origin of the observed high heat flow exists in sediments and/or upper crust. In this case, the original heat flow in the Nankai Trough is considered to be around 110 mW m-2 as expected from the magnetic-anomaly age of the Shikoku Basin, thus a heat flux of 20 to 40 mW m 2 must be explained by some process or processes occurring at shallow depth under the trough. Probable factors which may cause such heat flow anomalies are bottom water temperature variations, sedimentation, erosion, refraction of heat flux by a thermal conductivity contrast, advective heat transfer and so on. Among them sedimentation decreases surface heat flow and it appears unlikely that much erosion occurs on the trough floor. No structure which would produce a high conductivity contrast can be found in multichannel reflection records. Variation of bottom water temperature would be small in the deep trough region. Even if it occurred, it is required that similar pattern of changes of water temperature occurred before every measurement. This would be very unlikely. The remaining possibility would be the advective heat transfer. Some hot water from deeper levels may penetrate beneath the trough floor and heat up the sediments. The most probable origin of such hot water is the water contained in the subducted sediments which may be extruded during the subduction process. If this is the case, part of the flow may concentrate along the thrust faults seismically found beneath the structural steps on the landward slope of the trough (e.g. Nasu et al., 1982). Some of these faults were cut by drilling at DSDP Site 583 but no evidence for hot water flow was obtained (Kagami et al., 1982). Of course it does not completely disprove the possibility that advective heat transfer by pore water exists in this area. Temperature of the penetrating water can be evaluated from the amount of water being introduced into the subduction zone. We assume that the sediment thickness of the Shikoku Basin is 500 m just before subducting and the sediment contains 75% water by volume. Then the rate of water supply is about 4.8 • 1 0 -3 c m 2 sec -1 as the convergence rate is 4 cm yr -1. If this water heats up the sediments within a zone of 10 km width, the rate of water penetration would be 4.8 • 10 -9 c m s e c -1. At this rate, flux of water 1~ C warmer than the surroundings can increase heat flow by 0.20 mW m -2. Thus, in order to cause an increase in heat flux by 40 mW m -2, the temperature of the penetrating water should be at least 200 ~ C. This high temperature implies that the water should be extruded from a rather large depth, 4 km or more, otherwise much more water is flowing than expected from the amount of pore water. Therefore, though advective heat transfer by water could be a possible cause of the high heat flow zone, it is not a preferred explanation.

198

M. YAMANO

ET

AL.

5. Implication ot the Zonal Pattern of Heat Flow In many subduction zones, subnormal heat flow values have been obtained from the trench axis to the volcanic arc (Watanabe et al., 1977), but there are only a few trenches where heat flow variation in the forearc region have been investigated. Anderson (1980) discussed that, in the Japan Trench and Bonin arc, heat flow appears to have a minimum between the trench and the volcanic arc and to increase toward the volcanic front. As an example of a place where a young plate is subducting, Blackwell et al. (1982) presented a rather detailed heat flow profile in northern Oregon and demonstrated that heat flow is low and almost constant in the outer arc province. These data are important constraints on the thermal process of subduction. Many numerical models of thermal structure of subduction zones have been developed and some of them were used to explain the heat flow profiles in the forearc regions (e.g. Burch and Langseth, 1981; Honda, 1984). Here we consider the zonal pattern of heat flow distribution in the landward area of the Nankai Trough within one hundred kilometer from the axis of the trough. Surface heat flow patterns calculated from a numerical model are compared with the observations shown in Figure 4. A simple model (Figure 6) was used for calculation because the detailed structure of the Nankai subduction zone is unknown. The approximate dip angle of subduction is estimated from the depth of the upper plane of the underthrusting X

120 km

I

D

T=O

Z

8km

tO

V

.D

I:I

V

30kin F.--

0 L.

x r

q=q(x) 170 krn Fig. 6. Schematic representation of the model used for calculations,

A H O T TRENCH.'?

199

plate seen in the multichannel reflection records (e.g. Aoki et al., 1983). This plane can be traced as far as about 70 km from the trench axis. The subduction dip angle beyond there is uncertain. An alternating-direction implicit finite difference scheme was used to numerically solve the equation

pCp(OT/Ot+v. V T) = k V 2 T + H where p = density, Cp = specific heat, T = temperature, t = time, v = velocity, k = thermal conductivity, and H = heat generation per unit volume. The grid spacings in horizontal and vertical directions are 2 and 0.5 km respectively and the time step is 2.5 x 1 0 4 yr. Variation of these parameters had little effects on the results. Boundary conditions are also shown in Figure 6. At the top and the seaward side the temperature is fixed. At the landward side the temperature is extrapolated from internal grid points. Heat flux into the bottom, q, is constant in time but depends on x. The steady state solution when the plate does not move was used for the initial condition. For simplicity, the thermal conductivity and diffusivity are assumed to be uniform and 3 . 1 W m - I K -1 and 8 x l 0 - 7 m 2 s e c -1 respectively, for both sedimentary and hard rock parts in the model. The convergence rate is 4 cm yr -1. This model is unrealistic in some aspects. For instance, initial geothermal gradients were taken almost constant from the surface to the bottom. A calculation which starts from the error function type geotherm (e.g. Parker and Oldenburg, 1973) was also tried but its effects on the surface heat flow were inferred to be small. Both the thermal conductivity and diffusJvity of the sediments are lower than the assumed values of hard rocks. Accordingly, the time constant in which the effects of subduction reach the surface may be slightly longer than the following results. Calculations made first were for the case that the Shikoku Basin has a high heat flow, 145 mW m -2, more or less as observed in the Nankai Trough. Initial heat flow, namely the heat flow before the present episode of subduction, in the landward side of the trough was assumed to be 65 mW m -2. The distribution of calculated surface heat flow is shown in Figure 7a together with the observed values. The curve A is the initial state and B is at 2 m.y. after the initiation of subduction. Heat flow landward of the trough increases because of subduction of the hot oceanic plate. It takes about 6 m.y. for the region within 100 km from the trench axis to become steady state, the curve C. The calculated heat flow in this state is consistent with the observations in the area of deep-sea terraces, around 90 mW -2, but it seems to be too high within a few tens km from the trench axis corresponding to the landward slope of the trough. Pore water extruded from the sediments on the subducting plate might carry away heat and make the observed heat flow on the slope lower. According to the estimation in the preceding subsection, however, the amount of water which can be extruded is too small to lower heat flux by 20 to 40 mW m -2 within a zone of a few tens of km width. Moore and Karig (1976) deduced from the termination of turbidite deposition at DSDP Site 297 seaward of the trough that the present stage of subduction initiated about

200

M. YAMANO ET AL.

250

250

200

~'200 E

E lSO

E ls0

~ \

o

~- ~oo

~ 50

9 , .r A

..-~ I ~

B...-+~'%~,: 9

o

9

,

9

-,

~oo E ....... "~+..*~+A~

;-

(a)

100 DISTANCE (km)

0 TROUGH AXIS

0 200

-100

0

(c)

200

-100

~200

_E1so

so

0 TROUGH AXIS

250

200

100

I00 DISTANCE (km)

250

~

9 9 9

i i

200

9

*'~-+~-rg ~ 9

5o

i 9 ..-

D I S T A N C E(kml

2

i ~ -

H:-;.~.; 100

E --150

o

0 TROUGH AXIS

,.

. 9

ql

~ |9

100

~ 5o 0

-100 ~)

200

I

""", q*++

9

§

100 D I S T A N C E(kml

TROUGH AXIS

-100

Fig. 7, Heat flow patternscalculated from the model shown in Figure 6 and the observed values (Figure 5). Curves A to J are explained in the text. 3 m.y.b.p. If this is the case, the steady state has not been attained yet and the present heat flow distribution should be compared with B rather than C. Then the calculated values are incompatible with those observed on the terraces. But this disagreement depends on the initial condition. If heat flow landward of the trough were 105 mW m -2 in the initial state (the curve D in Figure 7b), heat flow on the terrace would be about 75 mW m -~ at 2 m.y. (E). Next, in the case that the Shikoku Basin has a heat flow of 105 m W m -2 consistent with the magnetic-anomaly age, the results shown in Figure 7c can be obtained. Initial heat flow on the landward side is assumed to be 65 mW m -2 as the curve F demonstrates. The heat flow distribution is shown to be like G at 2 m.y. and like H in the steady state. This model can reproduce low heat flow on the slope, but cannot make the terrace values high enough. In Figure 7d frictional heat by a stress of 500 bars is imposed on the shearing boundary where the distance from the trench axis is larger than 70 km. It is not unrealistic that shear heating is negligible in shallower regions. The curve I represents heat flow at 2 m.y. and J in the steady state. T h o u g h heat flow on the terraces reaches about 85 mW m -~, the effect of

A HOT TRENCH?

201

heating is not sufficient at 2 m.y. In these cases, of course, some additional mechanism is necessary to explain the high heat flow (higher than 105 mW m -2) on the floor of the trough. The above results indicate that these simple models cannot satisfactorily reproduce the complicated heat flow pattern on the landward side of the trough. Many factors neglected in the models such as movement of water and sediments and time variation of the state of subduction (e.g. velocity and dip angle of subduction, and position of the trench) may be playing important roles. Conversely it is difficult to infer thc thermal structure of this subduction zone from the heat flow distribution obtained in this study alone.

6. Summary and Conclusions A fairly detailed heat flow distribution in the Nankai Trough area has been obtained from the results of previous measurements, the gas hydrate method and the values newly measured in this study. There appears to be a zonal pattern parallel to the axis of the trough. There is a high heat flow zone, exceeding 130 mW m -2, on the floor of the trough. On the landward slope heat flow decreases to about 40 mW m -2 from the trough toward the arc, and increases up to around 90 mW m -2 on the deep-sea terraces. This pattern is interesting as an example of heat flow variation in the arc-trench gaps. There is a possibility that subduction of the young hot Shikoku Basin plate causes the high heat flow zone on the trough floor. The high heat flow zone, together with some other features of the Nankai Trough, might be attributed to subduction of a young plate. An alternative explanation is the penetration of hot water extruded from the sediments on the Shikoku Basin during underthrusting, though it seems less probable. These mechanisms may be combined. The complicated zonal pattern on the landward side of the trough cannot be explained by simple subduction models. It appears necessary to consider the effects of advective heat transfer and time variation of the state of subduction. The present data are still insufficient to clarify the thermal structure of the Nankai Trough and the Southwest Japan arc. It is hoped that various additional investigations will be made in this area. For example, the existence and effect of water circulation should be tested by closely spaced measurements of surface heat flow and detailed surveys of sediment and basement structure. The determination of electrical conductivity by ocean bottom magnetometers will give estimates of the temperature of the lithosphere beneath the Shikoku Basin and the Nankai Trough. It is also important to investigate the detailed tectonic history of the Nankai subduction zone.

Acknowledgements We would like to thank M. Yasui, T. Akiyama, and K. Shuto, Japan Meteorolo-

202

M. Y A M A N O E T AL.

gical Agency, for giving us an opportunity of measuring heat flow on the cruise of R/V Ryofu-Maru. The officers, crew and scientic staff of Ryofu-Maru assisted us in obtaining the data. We are grateful to H. Kinoshita, K. Nakamura, R. Hyndman, and B. Loncarevic for helpful comments. This study was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Culture, Japan. References Anderson, R. N.: 1980, '1980 Update of Heat Flow in the East and Southeast Asian Seas', in Hayes, D. E. (ed.), The Tectonic and Geologic Evolution of Southeast Asian Seas and Islands, Amer. Geophys. Union, Geophys. Mono. Ser. 23, pp. 319-326. Anderson, R. N. and Skilbeck, J. N.: 1981, 'Oceanic Heat Flow', in Emiliani, C. (ed.), The Sea Vol. 7, Wiley-Interscience, New York, pp. 489-523. Anderson, R. N., Langseth, M. G., Hayes, D. E., Watanabe, T., and Yasui, M.: 1978, 'Heat Flow, Thermal Conductivity, Thermal Gradient', in Hayes, D. E. (ed.), Geophysical Atlas of the East and Southeast Asian Seas, Map and Chart Ser. Vol. MC-25, Geol. Soc. Amer. Aoki, Y., Tamano, T., and Kato, S.: 1983, 'Detail Structure of the Nankai Trough from Migrated Seismic Sections', in Watkins, J. S. and Drake, C. L. (eds.), Studies in Continental Margin Geology, Amer. Assoc. Petr. Geol. Memoir 34, pp. 309-322. BiackweU, D. D., Bowen, R. G., Hall, D. A., Riccio, J., and Steele, J. L.: 1982, 'Heat Flow, Arc Volcanism, and Subduction in Northern Oregon', J. Geophys. Res. 87, 8735-8754. Bullard, E. C.: 1954, 'The Flow of Heat through Floor of the Atlantic Ocean', Proc. Roy. Soc. London A 222, 408--429. Burch, T. K. and Langseth, M. G.: 1981, 'Heat Flow Determination in Three DSDP Holes near the Japan Trench', J. Geophys. Res. 86, 9411-9419. Herman, B. M., Langseth, M. G., and Hobart, M. A.: 1977, 'Heat Flow in the Oceanic Crust Bounding West Africa', Tectonophysics 41, 61-78. Honda, S.: 1984, 'Thermal Structure beneath Tohoku, Northeastern Japan - A Case Study for Understanding the Detailed Thermal Structure of the Subduction Zones -', submitted to Tectonophysics, Kagami, H., Karig, D. E., Akiba, F., Bray, C. J., Cadet, J. P., Charvet, J., Coulbourn, W., Fujioka, K., Kinoshita, H., Lagoe, M., Lang, T. H., Leggett, J. K., Lombari, G. A., Lundberg, N., Machihara, T., Matsumoto, R., Mukhopadhyay, P., Niitsuma, N., Smith, A. J., Stein, C. L., and Taira, A.: 1982, 'Results of Deep Sea Drilling in the Nankai Trough and the Japan Trench', Programme and Abstracts, Seismol. Soc. Japan 1982, No. 2, 66 (in Japanese). Kagami, H., Karig, D. E., Akiba, F., Bray, C. J., Cadet, J. P., Charvet, J., Coulbourn, W., Fujioka, K., Kinoshita, H., Lagoe, M., Lang, T. H., Leggett, J. K., Lombari, G. A., Lundberg, N., Machihara, T., Matsumoto, R., Mukhopadhyay, P., Niitsuma, N., Smith, A. J., Stein, C. L., and Taira, A.: 1983, 'In the Orient Leg 87 Drills off Honshu & SW Japan', Geotimes 28 (1), 15-18. Klein, G. DeV. and Kobayashi, K.: 1980, 'Geological Summary of the North Philippine Sea, Based on Deep Sea Drilling Project Leg 58 Results', in Initial Reports of Deep Sea Drilling Project 58, U.S. Government Printing Otfice, Washington D.C., pp. 951-961. Klein, G. deV., Kobayashi, K., Chamley, H., Curtis, D. M., Dick, H. J. B., Echols, D. J., Fountain, D. M., Kinoshita, H., Marsh, N. G., Mizuno, A., Nisterenko, G. V., Okada, H., Sloan, J. R., Waples, D. M., and White, S. M.: 1978, 'Off-ridge Volcanism and Seafloor Spreading in the Shikoku Basin', Nature 273, 746--748. Klein, G. deV., Kobayashi, K., Chamley, H., Curtis, D. M., Dick, H. J. B., Echols, D. J., Fountain, D. M., Kinoshita, H., Marsh, N. G., Mizuno, A., Nisterenko, G. V., Okada, H., Sloan, J. R., Waples, D. M., and White, S. M.: 1980, Initial Reports of Deep Sea Drilling Project 58, U.S. Government Printing Office, Washington, D.C. Kobayashi, K. and Nakada, M.: 1978, 'Magnetic Anomalies and Tectonic Evolution of the Shikoku Inter-arc Basin', J. Phys. Earth 26 Suppl., $391-$402.

A HOT TRENCH?

203

Kono, Y. and Yoshii, T.: 1975, 'Numerical Experiments on the Thickening Plate Model', J. Phys. Earth 23, 63-75. Krogan, B. J., Bodvarsson, G., and Mesecar, R. S.: 1971, 'Heat Flow through the Floor of the Cascadia Basin', J. Geophys. Res. 76, 4758--4774. LaBrecque, J. L., Kent, D. V., and Cande, S.C.: 1977, 'Revised Magnetic Polarity Time Scale for Late Cretaceous and Cenozoic Time', Geology 5, 330-335. Langseth, M. G. and Herman, B. M.: 1981, 'Heat Transfer in the Oceanic Crust of the Brazil Basin', J. Geophys. Res. 86, 10805-10819. Langseth, M. G., Hobart, M. A., and Horai, K.: 1980, 'Heat Flow in the Bering Sea', J. Geophys. Res. 1t5, 3740-3750. Lister, C. R. B.: 1977, 'Estimates for Heat Flow and Deep Rock Properties Based on Boundary Layer Model', Tectonophysics 41, 157-171. Moore, J. C. and Karig, D. E.: 1976, 'Sedimentology, Structural Geology, and Tectonics of the Shikoku Subduction Zone, Southwestern Japan', Geol. Soc. Amer. Bull. 87, 1259-1268. Nasu, N., Tomoda, Y., Kobayashi, K., Kagami, H., Uyeda, S., Nagumo, S., Kushiro, I., Ozima, M., Nakazawa, K., Takayanagi, Y., Okada, H., Murauchi, S., Chujo, J., and Ishii, Y.: 1977, 'Multichannel Seismic Reflection Data across the Shikoku Basin and the Daito Ridges, 1976, part 1', IPOD-Japan Basic Data Series No. 1, 17 pp., Ocean Research Institute, Univ. of Tokyo, Tokyo, Japan. Nasu, N., Tomoda, Y,, Kobayashi, K., Kagami, H., Uyeda, S., Nagumo, S., Nakamura, K., Kushiro, I., Ozima, M., Nakazawa, K., Takayanagi, Y., Okada, H., Murauchi, S., Kinoshita, H., Ishiwada, Y., Tamano, T., Toba, T., and Aoki, Y.: 1982, 'Multi-channel Seismic Reflection Data across the Nankai Trough', IPOD-Japan Basic Data Series No. :4, 34 pp., Ocean Research Institute, Univ. of Tokyo, Tokyo, Japan. Parker, R. L. and Oldenburg, D. W.: 1973, 'Thermal Model of Ocean Ridges', Nature Phys. Sci. 65, 137-139. Parsons, B. and Sclater, J. G.: 1977, 'An Analysis of the Variation of Ocean Floor Bathymetry and Heat Flow with Age', J. Geophy. Res. 1t2, 803-827. Riddihough, R. P.: 1978, 'The Juan de Fuca Plate', EOS Trans. Amer. Geophys. Union 59, 836-842. Sclater, J. G., Jaupart, C., and Galson, D.: 1980, 'The Heat Flow through Oceanic and Continental Crust and the Heat Loss of the Earth', Rev. Geophys. Space Phys. 18, 269-311. Seno, T.: 1977, 'The Instantaneous Rotation Vector of the Philippine Sea Plate Relative to the Eurasian Plate', Tectonophysics 42, 209-226. Shih, T. C.: 1980, 'Magnetic Lineations in the Shikoku Basin', in Initial Reports o[ Deep Sea Drilling Project 58, U.S. Government Printing Office, Washington, D.C., pp. 783-788. Shiono, K., Sacks, I. S., and Linde, A. T.: 1980, 'Preliminary Velocity Structure of Japanese Islands and Philippine Sea from Surface Wave Dispersion', Carnegie Inst. Washington Year Book 79, 498-505. Sugi, N. and Uyeda, S.: 1984, 'Subduction of Young Oceanic Plates without Deep Focus Earthquake', Bull. Soc. Gdol. France (in press). Tatsumi, Y.: 1982, 'Origin of High-magnesian Andesites in the Setouchi Volcanic Belt, Southwest Japan, II. Melting Phase Relations at High Pressures', Earth and Planet. Sci. Letters 60, 305-317. yon Herzen, R. and Maxwell, A. E.: 1959, 'The Measurement of Thermal Conductivity of Deep-sea Sediments by a Needle-probe Method', J. Geophys. Res. 64, 1557-1563. Watanabe, T., Epp, D., Uyeda, S., Langseth, M., and Yasui, M.: 1970, 'Heat Flow in the Philippine Sea', Tectonophysics 10, 205-224. Watanabe, T., Langseth, M. G., and Anderson, R. N.: 1977, 'Heat Flow in Back-arc Basins of the Western Pacific', in Talwani, M. and Pitman III, W.C. (eds.), Island Arcs, Deep Sea Trenches, and Back-arc Basins, Maurice Ewing Ser., Vol. 1, Amer. Geophys. Union, Washington, D.C., pp. 137-167. Yamano, M., Uyeda, S., Aoki, Y., and Shipley, T. H.: 1982, 'Estimates of Heat Flow Derived from Gas Hydrates', Geology 10, 339-343. Yoshii, T.: 1979, 'Compilation of Geophysical Data around the Japanese Islands (I)', Bull. Earthquake Res. Inst. Univ. Tokyo 54, 75-117, (in Japanese with English abstract). Yoshii, T. and Kobayashi, K.: 1981, 'Geophysics of Subduction Zones', Kagaku 51, 472-478 (in Japanese).