Evolution of shallow, crustal thermal structure from subduction to collision:
An example from Taiwan
Wu-Cheng Chi†
Institute of Earth Sciences, Academia Sinica 128 Sec 2 Academia Road, Nankang, Taipei, Taiwan, ROC
Donald L. Reed*
Department of Geology, San Jose State University, One Washington Square, San Jose, California 95192-0102, USA
ABSTRACT
We study crustal thermal evolution by
examining heat flow patterns along a convergent boundary from a young subduction
zone to a more structurally mature collision
zone. More than 8000 km of seismic profiles
covering an offshore region of 45,000 km2 in
southern Taiwan show widespread bottomsimulating reflectors (BSRs). We derived
1107 BSR-based heat flows before combining
42 additional, published, offshore thermal
probe data and 86 on-land heat flow data to
document the shallow forearc thermal structures from the subduction zone to the collision
zone. In the subduction zone, the geothermal
gradient ranges mostly within 30–80 °C/km,
and decreases toward the arc due to slab cooling, intensive dewatering at the toe, sediment
blanketing, topographic effects, and other
processes. The geothermal gradient ranges
mostly from 30 to 90 °C/km in the collision
zone, and increases, instead of decreases,
toward the arc, possibly caused by exhumation, erosion, topographically induced groundwater circulation, and some upper mantle
processes related to collision. Heat flow in the
collision zone ranges from 80 to 250 mW/m2.
The high heat flow in the collision zone correlates with a shallower seismicity zone and
high seismic attenuation, while the lower heat
flow in the subduction zone might allow the
earthquakes to rupture to greater depth. The
heat flow increases along the topographic
high from subduction to collision zone due to
increasing geothermal gradients and higher
thermal conductivities of the exhumed basement rocks. This heat flow variation may generate an artificial exhumation rate pattern, if
a conventional 30 °C/km geothermal gradient
was used in fission-track studies.
†
E-mail: [email protected]
*E-mail: [email protected]
Keywords: bottom-simulating reflector, hydrates, geothermal gradient, heat flow, Taiwan,
subduction, collision, seismogenic zone, fission-track dating.
INTRODUCTION
Many mountain belts around the world are
the result of collisions preceded by subduction.
Arc-continent collisions have been interpreted
as important processes for the growth of continents. At the same time, many geological and
geophysical processes are affected by thermal
structures of the crust, e.g., metamorphism,
crustal rheology changes, the seismogenic process, seismic attenuation, and maturation of
hydrocarbons. Currently, there are few observational studies on crustal thermal evolution from
subduction to collision.
Taiwan is located along a convergent boundary with a subduction to the south and an arccontinent collision to the north. Thus, by studying transects from south to north, we can better
understand how thermal structures evolve from
subduction to collision. In the collision zone
on-land, there is an extensive heat flow data set
from exploration and thermal wells (Lee and
Cheng, 1986). However, regional-scale geothermal gradient and heat flow in the offshore region
in the initial collision zone and subduction zone
have not yet been studied systematically.
Here we used the gas-hydrate, bottom-simulating reflector (BSR) offshore southern Taiwan to derive 1107 geothermal gradients in
the accretionary prism in the subduction zone
and initial collision zone. We compared the
BSR-based thermal structures in the subduction zone with those in the collision zone. The
BSRs were found comparable with recently
published in situ heat flow measurements in the
Taiwan region. Different mechanisms affecting
the regional thermal patterns and the maximum
perturbations from these processes were discussed. We also speculated on the implications
of the different thermal structures on fissiontrack data interpretation, pressure-temperaturetime (P-T-t) paths of metamorphic rocks, frictional heat on the fault, seismogenic zone width,
seismic attenuation, and other geodynamic and
seismological processes. Overall, Taiwan is a
modern example that shows dramatic changes
in thermal structures from subduction to collision both in space and in time.
REGIONAL GEOLOGIC SETTING
The offshore Taiwan accretionary prism
(Fig. 1) is located along the boundary between
the Eurasian plate and Philippine Sea plate where
the oceanic lithosphere of the South China Sea
is subducting eastward beneath the Philippine
Sea plate (Bowin et al., 1978; Ho, 1986) with
a high rate of 7–9 cm/yr (e.g., Yu et al., 1997).
Oceanic crust in the subducting South China Sea
can be as old as 37 Ma, as inferred from marine
magnetic anomalies (Hsu and Sibuet, 2004).
The submarine accretionary prism (Fig. 2) contains three distinct structural domains (Reed et
al., 1992): (1) a lower slope domain composed
of mostly west-vergent ramp anticlines; (2) an
upper slope domain with highly discontinuous
reflections, suggesting intense deformation; and
(3) a backthrust domain located along the rear of
the prism. The prism is bounded on the west by
the Manila Trench and on the east by the forearc
basin of the North Luzon Trough.
Near Taiwan, subduction changes into an arccontinent collision where the Chinese continental margin enters the subduction zone (Suppe,
1981). The initial collision zone is probably near
latitude 21°N where the continent-ocean boundary (COB) enters the trench (Fig. 1). As the
Luzon arc approached the Taiwan and Chinese
passive margin, the volcanism started to cease.
The forearc basin is getting consumed until the
Luzon volcanic arc is juxtaposed next to the fold
and thrust belt of on-land Taiwan in the mature
collision zone. Although the timing of initial
GSA Bulletin; May/June 2008; v. 120; no. 5/6; p. 679–690; doi: 10.1130/B26210.1; 7 figures; 1 table.
For permission to copy, contact [email protected]
© 2008 Geological Society of America
679
Chi et al.
collision is still debated, Teng (1990) proposed
that it probably began in the late-middle Miocene and is currently propagating southward
(Suppe, 1984) toward the modern subduction
zone offshore Taiwan.
Further to the north in the mature collision
zone on Taiwan, the geologic provinces from
west to east include the Coastal Plains, Western
Foothills, Hsuehshan Range, Central Range,
and Coastal Range (Fig. 2). The Coastal Plains
are composed mostly of recent alluvium deposits overlying continental shelf materials of the
Chinese passive margin. The Western Foothills
are composed of Oligocene to Pleistocene shallow marine materials. The Hsuehshan Range
consists of quartz and carbonate sandstone,
argillites, and shales. The western side of Central Range consists of slatey materials, while
25°
CR
Cen R
Chinese
passive
margin
7-9
cm
/yr
22°
Luzon
Subduction
Arc
Slope
Backthrust
Upper
lope
er S
Low
SCS
h
renc
ila T
Man
21°
CO
B′
B
2005 Dec 10 10:13:03
119°
120°
121°
Initial
Collision
PSP
B
collision
A′
A
WF
CP
23°
Taiwan
PSP
SCS
24°
HR
EP
122°
Figure 1. Location map of Taiwan. To the south in the offshore region,
the South China Sea plate (SCS) is subducting underneath the Philippine
Sea plate (PSP), forming the Taiwan accretionary prism. The subduction
changes into an arc-continent collision when the Chinese passive margin
enters into the convergent boundary. The black arrow shows the direction
of the relative motion of the Philippine Sea plate with respect to a stable
Chinese continent. COB marks the continent-ocean boundary. The rapid
uplift forms the island of Taiwan in the collision zone. On Taiwan, from
west to east, are Coastal Plain (CP), Western Foothills (WF), Hsuehshan
Range (HR), Central Range (Cen R), and Coastal Range (CR). There is a
bathymetric and topographic high that extends from upper slope domain
offshore to the Central Range. Eurasian plate is marked as EP. A–A′ and
B–B′ show the locations of the two profiles in Figure 2.
680
the eastern part is mostly underlain by the preTertiary metamorphic rocks, including schist,
gneiss, carbonates, tuffs, and oceanic pelitic
and mafic schists (e.g., Ho, 1986). Overall the
metamorphic grade increases eastward in the
Central Range. Some of the metamorphic rocks
have been interpreted as the basement rocks of
the Eurasian continental crust (Lan et al., 1990).
The Central Range and the offshore upper slope
domain (Fig. 1) form a continuous bathymetric
and topographic high from collision to subduction. Further east is the Coastal Range, containing both volcanic and sedimentary rocks of
Luzon arc affinity. Slab earthquakes illuminate
a clear, east-dipping Benioff zone east of the
Coastal Range. However, the seismicity disappears abruptly north of 23.3°N. It is still a matter
of debate as to whether there is a slab between
the latitudes of 23.3°N and 24°N (e.g., Teng et
al., 2000; Lin, 2002).
In the northern part of Taiwan, the polarity of
the subduction flipped, and the Philippine Sea
plate subducts north underneath the Eurasian
plate along the Ryukyu Trench. The Benioff
zone is well imaged underneath the eastern part
of northern Taiwan, at least north of 24°N (Kao
et al., 1998). To reduce complications from
processes related to the Ryukyu Trench, in this
study we will focus mainly on thermal structures south of 24°N.
PREVIOUS STUDIES ON
CONVERGENT BOUNDARY ZONE
THERMAL STRUCTURES
Most subduction zones around the world
show high geothermal gradients and heat flows
near the trenches. The heat flows decrease
within the accretionary prism before increasing
toward the arcs. For example, across northern
Honshu in Japan, the heat flow is ~55 mW/m2
near the trench, decreasing to ~25 mW/m2 near
the bathymetric high region, then increasing to
more than 100 mW/m2 toward the volcanic arc
and backarc regions (Jessop, 1990). In Cascadia,
the heat flow is ~120 mW/m2 near the trench and
decreases to ~60 mW/m2 in the continental shelf
region (Hyndman and Wang, 1995; Hyndman
and Lewis, 1999; Lewis et al., 2003). The heat
flow then increases toward the Cascade volcanoes and reaches more than 100 mW/m2 (Jessop,
1990). The Chilean subduction zone near the
Chile triple junction also shows high geothermal
gradient near the trench and lower geothermal
gradient arcward (Brown and Bangs, 1995). In a
local scale near the toe of the accretionary prism,
heat flow studies in the Barbados Ridge accretionary prism show high heat flow associated
with thrust zones, implying focused fluid flow
at the décollement and some thrusts (Foucher et
Geological Society of America Bulletin, May/June 2008
Crustal thermal evolution from subduction to collision
Coastal
Plains
Western
Foothills
Central
Range
Coastal
Range
A′
(k m )
A
Hsuenshan
Range
Heat flow
Collision zone
(km)
Heat flow
Subduction zone
B
Upper slope
domain
Backthrust
domain
North Luzon
Trough
B′
Accretionary prism
Trench
Forearc b asin
(k m )
Manila Trench
Lower slope
domain
Figure 2. A schematic showing
structural domains, heat flow
patterns, and possible mechanisms causing different thermal
structures in a subduction zone
transect and a collision zone
transect in Taiwan. In the collision zone at a latitude of 23.8°N
(see A–A′ in Fig. 1 for location),
we have plotted previously
published heat flow (Lee and
Cheng, 1986) within 5 km of
the profile in small circles with
error bars, exaggerated topography, and seismicity within
35 km of the profile. Lin (2002)
proposed that the exhumation
of the crust will reduce the seismicity due to higher temperature at depth, thus the higher
heat flow. Other mechanisms,
including groundwater circulation, high erosion rate, higher
radiogenic heat production rate,
topographic effect, and other
processes can also increase the
heat flow. Lower panel shows
a profile in the subduction
zone at a latitude of 20.5°N.
See B–B′ in Figure 1 for location. We have plotted the heat
flows derived from this study
in large circles with error bars.
All these heat flows are within
5 km of the profile. Unlike the
heat flow profile in the collision
zone, here heat flow decreases
toward the arc in a pattern similar to other subduction zones
around the world, possibly
due to slab cooling, dewatering
near the toe of the accretionary
prism, increasing blanketing
effect toward the ridge, and
topographic effects. Triangles
are the theoretical heat flows
derived from our slab-cooling
modeling, which fit the overall
pattern of the BSR-based heat
flows. See text in Discussion
section for more details.
(km)
Geological Society of America Bulletin, May/June 2008
681
Chi et al.
al., 1990). Similar features have been reported at
Nankai accretionary prism (Yamano et al., 1992)
and Taiwan (Chi et al., 1998).
Observed data on thermal structures in the
collision zone are sparse, particularly for arccontinent collision zones; yet, Taiwan is one
of the best-studied regions in terms of crustal
thermal structures (cf. Barr and Dahlen, 1989;
Hwang and Wang, 1993; Lin, 2002). However,
the thermal structures in the initial collision
zone and subduction zone offshore southern
Taiwan have not been studied systematically
until this study (Fig. 2), in which we found
a dramatic difference in heat flow patterns
between the subduction zone and the collision
zone in the Taiwan region.
BOTTOM-SIMULATING REFLECTOR,
GAS HYDRATE, AND GEOTHERMAL
GRADIENT
A bottom-simulating reflector, or BSR
(Fig. 3), is a seismic reflector subparallel to
the topography of the seafloor, which, in some
places, cuts across reflectors generated by sediment layers (Tucholke et al., 1977; Shipley et
al., 1979; Bangs et al., 1993). BSRs are commonly observed in continental slope sediments,
particularly those associated with accretionary
prisms (Hyndman and Davis, 1992; Brooks et
al., 1991; Miller et al., 1991). These seismic
reflectors are generated by the acoustic impedance contrast at the base of a sediment layer
containing methane hydrate (Shipley et al.,
1979; Holbrook et al., 1996).
Methane hydrates form when water and methane combine to generate a crystalline substance
that looks like ice. Most hydrates have been
found in deep ocean sediments where there is a
sufficient supply of methane and where pressure
and temperature ranges are between 0.2 and
5 MPa and 0–25 °C, respectively, as inferred
from Kvenvolden and McMenamin (1980).
The increase in temperature with depth below
the seafloor causes methane hydrate to become
unstable and decompose, despite increasing
pressure. As a result, the base of the methane
hydrate defines a “phase boundary” that separates the stable gas hydrate above from a field
of instability below. The acoustic impedance
contrast across this phase boundary thus can be
imaged in seismic profiles as a BSR.
In particular, P-T conditions at the BSR have
been used to estimate heat flow in several accretionary prism settings (Yamano et al., 1982;
Yamano et al., 1992; Hyndman et al., 1992;
Ashi and Taira, 1993; Zwart and Moore, 1993;
Townend, 1997). The temperature at the BSR
can be inferred from two-component (methane and water), hydrate-phase equilibria and
estimated in situ pressures. By converting the
two-way traveltime of the BSR to depth and
using a temperature at the seafloor, the regional
geothermal gradient and heat flow can be estimated using the relation:
Q = -k dt/dz
where Q is heat flow in mW/m2, k is sediment
thermal conductivity in W/mK, dt/dz is geothermal gradient in °C/km, t is temperature in °C,
and z is subbottom depth in km.
DERIVED BOTTOM-SIMULATING
REFLECTOR–BASED HEAT FLOW
OFFSHORE TAIWAN
The bottom-simulating reflector (BSR) in the
Taiwan region and its relation with gas hydrate
was first documented in Reed et al. (1991). Chi
et al. (1998) systematically studied the BSR
offshore Taiwan and discussed the factors controlling its distribution. Schnurle et al. (2004)
used seismic reflection and ocean bottom seismic data to study the BSR in some of this area.
The gas component in the hydrates offshore
Taiwan is mostly methane. The gas seepage
from mud volcanoes near this region on land is
mainly composed of methane with minor ethane
and other gases; water column samples in this
Latitude: 22.17°N
Longitude: 120°E
Latitude: 21.17°N
Longitude: 120°E
B
A
Seafloor
9KIUTJYZ]U]G_ZXG\KRZOSK
Seconds (two-way traveltime)
Seafloor
Figure 3. (A) An example of a Q1 bottom-simulating reflector (BSR) with reversed polarity reflection located in an anticlinal ridge. The coordinate marks the location of the left edge of the profile. (B) Examples of Q2 and Q3 BSRs in this
region. The coordinate marks the location of the left edge of the profile.
682
Geological Society of America Bulletin, May/June 2008
Crustal thermal evolution from subduction to collision
offshore region also show high dissolved methane concentrations (Yang et al., 2006).
We use data acquired during two marine geophysical surveys (Fig. 4). The first data set was
collected in 1990 aboard the R/V Moana Wave
of the University of Hawaii. The second data
set is from the 160-channel seismic reflection
survey in 1995 aboard R/V Maurice Ewing of
Columbia University. The six-channel data were
processed using the SIOSEIS processing software. The 160-channel reflection data were processed using ProMAX (e.g., Chi et al., 2003).
The processing sequences of these two data sets
are listed in Table 1. More than 8000 km of seismic reflection profiles were examined covering
a region of 45,000 km2.
At least 15,000 km2, or 70% of the Taiwan
accretionary prism, is covered by a BSR that is
identified as a reflector subparallel to the seafloor with reversed polarity. Reversed polarity,
increasing subbottom depth with increasing
water depth, the distribution of the BSR, and
methane emitted on land in the nearby region
strongly suggest that this feature marks the base
of the methane hydrate stability field.
The BSR typically exhibits a single symmetrical pulse characteristic of a simple interface (Fig. 3). We have made a determined effort
throughout the study to separate reflectors associated with water-bottom reverberations from BSR
identifications. For example, shallow reflections
marked by constant subbottom depths beneath
a dipping seafloor reflection will be interpreted
as water-bottom reverberations, instead of a
BSR. The BSR was identified at more than 1000
locations along the survey lines, and each pick
was assigned to one of three categories according to the confidence level of the identification
(Q1, Q2, or Q3). The BSR was assigned into the
Q1 category if it showed the following reflector
characteristics: (1) subparallel to the seafloor in
shallow subbottom depths, (2) reverse polarity
with a single symmetrical pulse, and (3) a strong
coherent reflection that cuts across bedding. The
Q2 category, or probable BSR identifications,
showed similar characteristics to the Q1 picks
except that the reflections were either segmented
into pieces or did not cut bedding. The segment
lengths of the Q2 reflectors were usually longer
than the gaps between segments. The Q3 category, containing possible BSR picks, exhibited
very weak and segmented reflections that were
subparallel to the seafloor. BSRs in the Q3 group
usually could be traced from adjacent Q1 or Q2
categories without abrupt changes in the subbottom depth.
Next, we compiled water depths and the BSR
subbottom depths. The two-way traveltime to
the seafloor was divided by two and converted
into water depth in meters (HWATER) by assuming
23°
Taiwan
22°
21°
119°
121°
120°
122°
Figure 4. Tracks of 1990 six-channel seismic data (small circles) (after Reed et
al., 1992) and 160-channel seismic data (large circles).
TABLE 1. REFLECTION DATA PROCESSING SEQUENCES
Ewing Dataset:
Moana Wave Dataset:
Resample to 4 msec
Resampling at 4 msec
Geometry
Geometry
Automatic gain control
True amplitude recovery
Spiking deconvolution
F-K demultiple in shot domain
Normal move out
F-K demultiple in receiver domain
Stack
Dip move out
Filter
Velocity analysis
Predictive deconvolution
Radon filter
Water-bottom mute
40-fold weighted stack at 12.5 m CDP spacing
Trace weighting
Time variant frequency filter
Filter
F-K or F-D migration
Automatic gain control
Automatic gain control
Fourfold stack section plot
Migrated section plot
F-K migration at 1500 m/s
Filter
Automatic gain control
Migrated section plot
Note: CDP—Common-Depth Point; F-D—Finite-Difference; F-K—frequency-wave number
1450 m/s as the seismic velocity of water. The
two-way traveltimes of BSR subbottom depths
were converted into BSR subbottom depths in
meters by using velocity-depth relations published by Hamilton (1980):
HBSR = VAVG × t
= [1511 + (1041 × t) − (372 × t2)] × t,
subbottom depth in sec. The regional two-way
traveltime of the BSR subbottom depth is, on
average, ~0.5 s, translating to 460 m of BSR
subbottom depth.
We fit the methane hydrate phase boundary
data published by Hyndman and Davis (1992)
using this equation:
TBSR = 2.03 × [Log(HWATER + HBSR) – 2] × 9.75 – 4,
where HBSR is BSR subbottom depth in m, VAVG
is the average velocity of sediments above BSR
in m/s, and t is the one-way traveltime of BSR
where TBSR is the temperature at BSRs in °C,
and Log is the logarithms to base 10.
Geological Society of America Bulletin, May/June 2008
683
Chi et al.
Another equation was derived to represent
the seafloor temperature by least-squares fitting
the water temperature and depth data provided
by the National Center for Ocean Research of
Taiwan:
shallow heat-flow boreholes
25°
geothermal exploration
wells and others
oil wells from CPC
TSF = 0.2597 × (LnH)3 – 3.802 × (LnH)2
+ 10.67 × (LnH) + 26.96,
thermo-electric
and BHT data
where TSF is temperature of the seawater right
above seafloor in °C. LnH is the natural log of
water depth in meter. The water depth in this
data set ranges from 600 to 3700 m; therefore,
the TSF ranges from 7.5 to 2.0 °C assuming that
the temperature on the seafloor sediment is similar to that of the seawater.
These two quantities (TSF and TBSR) were
used to determine the temperature difference
between the seafloor and the BSR, which then
was divided by the BSR subbottom depth to
estimate geothermal gradient. The BSR-based
regional geothermal gradient ranges from 17 to
150 °C/km. The mean and standard deviation of
the geothermal gradient are 44.7 ± 12.9 °C/km.
We have also included previously published
geothermal gradients in both subduction and
collision zones in Figure 5 (e.g., Shyu et al.,
2006; Lee and Cheng, 1986).
These previously published data sets usually
also include thermal conductivity measurements.
However, the conductivity data from offshore
BSR locations at several hundred meters subbottom depth are currently not available. We used a
published depth-porosity profile from a well in
the offshore region farther to the west (Lin et al.,
2003) to derive a 1D, depth-dependent, average
thermal conductivity model. The average seafloor conductivity of the 42 thermal probe measurements in this region is 0.97 W/mK (cf. Shyu
et al., 2006). With this value as the arithmetic
mean of the conductivities of the matrix (37%
volume at seafloor) and the pore fluid (63% volume), we used 0.6 W/mK as the pore fluid conductivity to derive the conductivity of the matrix
in this region, which is 2.2 W/mK. This matrix
conductivity value then was used to convert the
depth-porosity relation to a depth-conductivity
relation. Using an average regional heat flow
of 48 mW/m2, we then calculated the depthtemperature profile. For each subbottom depth,
we calculated its corresponding average conductivity by dividing the 48 mW/m2 heat flow
by the temperature increase from the seafloor
to that subbottom depth. In addition, we did a
sensitivity test by using 60 mW/m2, instead of
48 mW/m2, as the regional heat flow and found
that the average thermal conductivities differ by
less than 3%. For the whole BSR data set, the
derived thermal conductivities range from 1.11
to 1.29 W/mK with a mean of 1.25 W/mK. This
684
thermal probe
Q1 BSR
Q2 BSR
Q3 BSR
24°
23°
22°
21°
0
121°
120°
119°
20
40
60
80
100
122°
180
Geothermal gradient (°C/km)
Figure 5. Geothermal gradient pattern from subduction zone to collision zone in the Taiwan
region. Large stars indicate Q1 class, or highest confidence in BSR identification. Medium
stars indicate Q2 class, or probable BSR identification. Small stars indicate Q3 class, or
possible BSR identification. The large circles are from previously published thermal probe
measurements (Shyu et al., 1998; Shyu et al., 2006). Different sizes of the hexagons represent
different geothermal gradient data sets on land in the collision zone, which are discussed in
Lee and Cheng (1986). Note higher geothermal gradients along the trench in the subduction zone and along the topographic high in the collision zone. The three red boxes mark
the locations of the possible hydrothermal vents. BHT—bottom-hole temperature; BSR—
bottom-simulating reflectors; CPC—Chinese Petroleum Corporation.
Geological Society of America Bulletin, May/June 2008
Crustal thermal evolution from subduction to collision
translates to up to ±15% variations of thermal
conductivity. We then derived the BSR-based
heat flow using the geothermal gradient and
thermal conductivity. BSR-based heat flow values (Fig. 6) were estimated at 15–146 mW/m2,
while both the mean and standard deviation are
43.4 ± 12.5 mW/m2. The BSR-based heat flow
data set was combined with previous offshore
and on-land heat flow in situ measurements
(Fig. 7). Note that these data sets were derived
from different methods, but the first-order features for the crustal thermal structures from
the subduction zone to the arc-continent collision should be robust. All these data sets were
derived from shallow depths and range down
to several hundred meters. However, some of
the on-land measurements from Lee and Cheng
(1986) can reach kilometers in depth.
Taiwan
22°
21°
thermal probe
Q1 BSR
Q2 BSR
Q3 BSR
DISCUSSION
Uncertainty in Estimating Geothermal
Gradient
In this study, we discuss both geothermal gradient and heat flow. Overall, the geothermal gradients derived from this study are more robust
than the heat flow because of relatively larger
uncertainty on the thermal conductivity.
There is a large scattering of the BSR subbottom depths, ranging mostly from 250 to 600 m
(0.2–0.7 s in two-way traveltime). This scattering represents the wide range of thermal gradients derived from this study. The uncertainties
in measuring the two-way traveltime of the seafloor and the BSR are less than 0.05 s. Subbottom depths of BSRs in two-way traveltime can
be influenced by a number of factors, including
spatial and temporal changes in P-T conditions
and lateral variations in sediment velocity above
the BSR.
Variations in BSR subbottom depths, particularly along the surface of modern accretionary
prisms, are commonly attributed to changing
P-T conditions beneath the seafloor. The time
span for the hydrates to respond to P-T condition changes is still not clear. Torres et al. (2002)
showed that gas hydrate stability responds rapidly to tidal-sea–level changes, implying that
the BSR depths reflect the modern P-T regime.
An experiment conducted by a remotely operated vehicle in the deep waters also shows that
hydrate formation occurred within minutes
(Brewer et al., 1997).
We studied the uncertainty in the estimates
of BSR-based geothermal gradients. Assuming
the picked BSR correctly represents the hydrate
phase boundary, we estimated the uncertainty
based on error analysis of seafloor temperature,
BSR temperature, and BSR subbottom depth.
121°
120°
0
20
40
60
80
100
180
Heat flow (mW/m2)
Figure 6. Distribution of BSR identifications and the BSR-based heat flows. The legends are
the same as in Figure 5. Some very high heat flows from thermal probe are measured along
mud diapirs and mud volcanoes. Twenty-two out of 42 thermal probe data were measured
very near the BSRs. Except at shallow water depths, the discrepancy between BSR-based
and thermal probe heat flows are usually less than 30%. Note higher heat flows along the
trench in the subduction zone, especially near the continent-ocean boundary, where there is
a basin with thick sediments. Overall the heat flows decrease arcward. BHT—bottom-hole
temperature; BSR—bottom-simulating reflectors; CPC—Chinese Petroleum Corporation.
Based on the data from the National Center
for Ocean Research of Taiwan, the maximum
seawater temperature variation is about ±3
°C. For an average BSR subbottom depth of
~500 m, it translates to ~6 °C/km of uncertainty
in geothermal gradient estimates.
For the BSR in situ temperature, previous
studies found that using hydrostatic pressure
yields a good result (e.g., Ashi and Taira, 1993).
But we also calculated the in situ lithostatic
pressure using 2.3 g/cm3 as the average density
of the sediment layer above the BSR. All the
geothermal gradients were recalculated using
lithostatic pressure and compared with our geothermal gradient model based on hydrostatic
pressure. From the histogram of the discrepancies between these two types of geothermal
gradients, we found that most of the geothermal
gradients will increase between 3–7 °C/km, if
lithostatic pressure is used.
It is difficult to estimate the errors on BSR
subbottom depth because currently there is no
detailed velocity analysis on a regional scale.
However, we found that, for an arbitrary 30%
error in estimating the BSR subbottom depth,
the geothermal gradients will change only
0.5–2.5 °C/km, which is relatively small. This
is because the BSR subbottom depth error will
also be mapped into error in hydrostatic pressure estimates, which in turn will change the
estimated BSR temperature, thus compensating for the effect that came from errors in BSR
subbottom depth in estimating the geothermal
gradient. The degree of lateral velocity variation
Geological Society of America Bulletin, May/June 2008
685
Chi et al.
shallow heat flow boreholes
geothermal exploration
wells and others
oil wells from CPC
25°
thermo-electric
and BHT data
thermal probe
Q1 BSR
Q2 BSR
Q3 BSR
24°
23°
22°
121°
120°
0
50
100
150
122°
200
600
Heat flow (mW/m2)
Figure 7. Surface heat-flow pattern in the collision zone. The legends are the same as in
Figure 5. For heat flow pattern in the offshore subduction zone, please refer to Figure 6.
Overall, the heat flow is low in the subduction zone and high in the collision zone. Along
the bathymetric and topographic high, the heat flow increases dramatically from subduction zone to collision zone. Note different color scales in heat flow in Figures 6 and 7. The
Benioff zone east of the Coastal Range disappears abruptly north of 23.3°N. Note very high
heat flows in the mountains between 23.3°N and 24°N. BHT—Bottom-hole temperature;
BSR—Bottom-simulating reflectors; CPC—Chinese Petroleum Corporation.
above the BSR in this region is still unclear.
The depth-velocity relation of Hamilton (1980)
might be better suited for the sediments in the
lower slope domain, while the sediments in the
upper slope domain might have higher P velocity at a given subbottom depth due to compaction from more intensive deformation. Saturation of the hydrates above the BSR could also
increase the velocity. However, this effect might
be relatively minor. In summary, the total error
should be less than 15 °C/km, whereas most of
the BSR-based geothermal gradients will have
error within 10 °C/km. The thermal probe geothermal gradients are consistently slightly higher
686
than those derived from the BSRs (Fig. 5), possibly due to the smaller thermal conductivity on
the seafloor compared to that in the sediments
above the BSR.
Regional Thermal Conductivity Model at
Shallow Depths
Local heterogeneity in thermal conductivity will also affect the local geothermal gradient pattern. Currently we do not have detailed
information on the thermal properties of the
crust in the offshore region. The 42 seafloor in
situ measurements that cover 70% of this region
(Shyu et al., 1998; Shyu et al., 2006) give thermal conductivities on the seafloor that range
from 0.79 to 1.19 W/m/°C. Most of them are
between 0.85 and 1.05 W/m/°C and show little
systematic spatial variation.
The thermal conductivities from this study
range from 1.11 to 1.29 W/mK with a mean of
1.25 W/mK. The higher conductivities occur
because we have considered the reduced porosity in sediments from seafloor to the BSR. The
±15% thermal conductivity variation derived
in this study is similar to that of thousands of
in situ thermal conductivity measurements
from drilling in Peruvian, Cascadia, and Blake
Ridge margins (cf. Figure 5 of Grevemeyer and
Villinger, 2001). However, larger variations of
±40% were found in the Nankai and Chile margins (Grevemeyer and Villinger, 2001). Without
drilling data from the offshore Taiwan region,
it is difficult to estimate the error ranges of our
conductivity estimates. However, because the
geothermal gradients in the offshore region are
mostly between 31 and 56 mW/m2, the errors that
propagate into our heat flow estimates should be
relatively small (mostly <20 mW/m2).
This thermal conductivity model does not take
into account the possible influence from different
lithology. However, the BSR is most likely found
in sedimentary layers, not in the basement rocks.
As a result, the thermal conductivity above the
BSR should be similar and does not vary a lot
such as that in the collision zone, where different
rock types were found in the outcrops.
Regional Heat Flow Model
We multiply the BSR-derived geothermal
gradient with depth-dependent thermal conductivity to get the heat flow. In total, 1107
BSR-based heat flows were derived from this
study (Fig. 6).
In situ heat flow measurements in and near this
region (Nissen et al., 1995; Xia et al., 1995; Shyu
et al., 1998; Shyu et al., 2006) are consistent with
the BSR-derived geothermal gradients and heat
flows from this study. Twenty-two out of the 42
thermal probe heat flows were measured near
the BSRs and, except at shallow-water depth
regions, the deviation ranges mostly from 0 to
16 mW/m2, which is within 30% of the regional
heat flow. Because most of the geothermal gradients fall between 31 and 56 °C/km, most of the
errors will be within 10–17 °C/km, which is similar to the results from our uncertainty analyses
from the previous section. The absolute values of
the resulting heat flow are less accurate, but the
spatial variations are well determined.
BSR-derived heat flow is an average heat
flow for the top 200–400 m sedimentary column, compared with thermal probe that samples
Geological Society of America Bulletin, May/June 2008
Crustal thermal evolution from subduction to collision
very shallow depths. As a result, BSR-derived
geothermal gradient is less influenced by periodic surface temperature variation. Actually, the
deep-sea stable temperature environment will
dramatically reduce the temperature perturbation
from periodic seasonal or even paleoclimatic
temperature changes that might have affected the
modern heat flow measurements as proposed by
Kohl (1998) and others. For the shallow-water
depth region where the seafloor temperature varies greatly, the average 300 m BSR subbottom
depth in the shallow depth region is still deeper
than the skin depth for temperature perturbation
with periods longer than 10,000 yr (cf. equation
4–90 of Turcotte and Schubert, 1982). In summary, the BSR-based heat flows derived from this
study are consistent with the thermal probe measurements, especially in the deep water regions.
By combining the BSR-based heat flow data sets
with the published heat flow measurements, we
will next discuss the thermal structures in both
subduction and the collision zones.
Crustal Thermal Structures
At a latitude of ~20.5°N in the subduction
zone, heat flow decreases from 75 to 40 mW/m2
from the trench to the upper slope domain of the
accretionary prism (Fig. 2). To the east in the
forearc basin, a BSR in a fold gives a heat flow
of ~25 mW/m2 (Fig. 6). The heat flow pattern
along this transect is consistent with the three in
situ heat flow measurements farther to the south
at ~19°N (Xia et al., 1995). Heat flows in the
satellite basins in the arc region are ~50 mW/m2
and are derived from some low-quality BSRs.
Farther to the north in the initial collision zone
(Fig. 5), the continent-ocean boundary (COB)
enters into the trench near 21.2°N. Nissen et al.
(1995) have collected 12 thermal probe measurements along a transect from continental shelf
(117°E, 22.8°N) to continental slope (118.1°E,
19.3°E) that is 220 km west of and parallel to
the trench. We can treat this data set as the initial condition before the Chinese passive margin
enters into this convergent boundary. Nissen et al.
(1995) found that heat flows are ~80 mW/m2 in
the continental slope and decrease to 70 mW/m2
in the abyssal plain. From BSR-based heat flow,
we found slightly increased heat flows once the
incoming sediments were scraped off and incorporated into the toe of the accretionary prism,
where intensive dewatering occurs.
High geothermal gradients (40–80 °C/km)
and heat flows (50–105 mW/m2) were found
in a thick basin near the toe in this region east
of the COB (Fig. 6). This might be a result of
intensive dewatering in this basin, which covers a circular region with a diameter of 60 km
centered at 119.8°E, 21.6°N. Seismic reflection
data show conjugate fault plane reflections
within the basin, even though the displacements
across the faults are small, suggesting possible
fluids within the fault zones. Cores recovered in
this region show manganese nodules, indicating
active hydrologic processes in this region (N.
Lundberg, 2003, personal commun.). Similar
high heat flows also were found across continent-ocean transform boundaries around the
world (e.g., Vogt and Sundvor, 1996). Vogt and
Sundvor (1996) proposed that the great sedimentary thickness of the continental slope will
enhance dewatering, thus generating high heat
flow on the Norwegian-Barents-Svalbard continental slope. Moreover, along the Australia margin, the high heat flow was modeled as a result
of thermal uplift, local isostatic rebound, and
other processes (Lorenzo and Vera, 1992).
Away from the toe in the initial collision
zone, the heat flows decrease toward the arc
as the sediments stack thicker, especially from
lower slope domain to the upper slope domain.
Some clear BSRs were identified in the backthrust domain with heat flows ranging from 30
to 60 mW/m2 and increasing toward the arc.
On Taiwan in the mature collision zone, the
heat flows are high in the mountains, mostly
between 80 and 250 mW/m2. Whether the heat
flow data from some of the “geothermal wells”
are representative of the regional heat flow pattern is still under debate. But several studies are
able to fit the high heat flow pattern by thermal
modeling using different crustal kinematic models (Barr and Dahlen, 1989; Hwang and Wang,
1993; Lin, 2002). The high heat flow pattern
is also consistent with observed shallowing of
seismicity (Lin, 2002). Overall, the high heat
flow in the mountainous regions is interpreted
as a result of rapid uplift and exhumation of the
warmer material from depth. Other processes,
including groundwater circulation, topographic
effects, and higher radiogenic heat production
rates in the continental crust will also contribute to the high heat flow. Complex deformations
in lower crust and upper mantle following the
collision might have also affected the thermal
structures in this region.
FACTORS AFFECTING SHALLOW
THERMAL STRUCTURES
Cooling Effect of the Subducting Plate
We compared the theoretical heat flow values derived from the age of the subducting plate
with the BSR-derived heat flow. The predicted
heat flow ranges from 90 to 84 mW/m2, based
on the ages of the subduction plate (32–37 Ma;
Hsu and Sibuet, 2004) and the updated seafloor
age–heat flow relation (Stein and Stein, 1992).
The average 44.5 mW/m2 derived from BSR is
about half the average predicted heat flow of
87 mW/m2.
Here we studied the cooling effect due to the
rapid subduction of the South China Sea plate
based on Molnar and England (1990) that was
slightly modified by Von Herzen et al. (2001):
Q=
QO +V τ
,
S
where Qo is the reference heat flow in the
region, ν is the rate of orthogonal plate convergence, and τ is the shear stress associated
with slip on the fault. The term ντ is for frictional heating on the fault. S = 1 + b(zfνsinδ/α),
where zf is depth of the fault below the heat
flow measurement, δis the dip of the fault separating the upper and lower plates, and α is the
average thermal diffusivity of the plate.
We used the following parameters to calculate
the heat flow profile at the latitude of 20.5°N
in the subduction zone: Qo = 87 mW/m2, ν =
85 mm/yr *cos(50°), which is the full convergent rate times cosine of the angle between relative plate motion and the strike of the trench axis.
Based on a crustal model derived from a seismic
profile and gravity modeling (Chi et al., 2003),
we used an average dip of 8° to calculate the
depth of the fault at different distances from the
trench axis. We used a diffusivity of 1.1*10–6 m2/s
as used by Von Herzen et al. (2001). We used
one (1) for the dimensionless factor b, because
Molnar and England (1990) found this a reasonable number for most of the cases. The results
are plotted in Figure 2, which shows a first-order
match to the BSR-based derived heat flows in
this region. Similar observations and full numerical simulations have been applied to subduction
zones in other regions (cf. Hyndman and Wang,
1995; Wang et al., 1995; Currie et al., 2004).
Because the BSR-based and in situ heat flows
are low in this region, we excluded the frictional
heating term, i.e., we set the shear stress τ as
zero. If we include the frictional heating term,
a uniform stress of 25 MPa will increase the
heat flow by 18 mW/m2. We have also tested
the cases of increasing stress with depth, but the
higher frictional heating from deeper fault segments will increase the heat flow toward the arc,
opposite to the decreasing trend derived from
the BSR-based heat flows.
Near the collision zone to the north, we
observed very high heat flows between the latitudes of 23.3°N and 24°N (Fig. 7), where there
is no Benioff seismicity east of Taiwan island.
Whether the slab is detached in this region is still
under debate (e.g., Teng et al., 2000), and its influence on the heat flow pattern is still unknown.
Other tectonic events might have influenced
the thermal structures in this region, including
Geological Society of America Bulletin, May/June 2008
687
Chi et al.
subduction initiation or initiation and cessation
of arc volcanism. Wakabayashi (2004) has discussed the effects of these processes on other
orogenic belts. Rapid accretion near the trench
axis might have caused shortening and thickening of the accreted sediments, causing isotherm
spacing to stretch vertically faster than thermal
equilibrium can be reestablished, as proposed
by Hyndman et al. (1993). But the effect probably is not very important for the Taiwan region
because seismic profiles do not show strong
evidence of simple shear-style thickening of the
sediments arcward of the trench.
Sedimentation and Erosion
We used the relation derived by Von Herzen
and Uyeda (1963) to study the range of heat flow
reduction due to sedimentation in this region:
Q*
2X − X2
= 1 − erf ( X ) −
e + 2 X 2erfc ( X ) ,
Q
π
where Q* is the modified heat flow due to the
sedimentation, Q is the initial surface heat flow,
α is the diffusivity,
X = 0.5 × U ×
t
α ,
and erf() and erfc() are the error function and its
complement. The time (t) is in seconds. We use
a thermal diffusivity of 2 × 10−7 m2/s. U is the
sedimentation rate.
Although no sedimentation rate data are
available in this study region, a wide range of
sedimentation rates (0.17–7.3 cm/ka) were
derived from the nearby regions (cf. Table 8 of
Chang, 2002). This translates to a 0.5%–18%
reduction in heat flow based on our calculations. It is likely that the sedimentation effect is
less important than the cooling effects from the
subducting slab. The frictional heating and sedimentation calculations are assumed to increase
and decrease the heat flow of the whole study
region, respectively. Thus they can partially
cancel out each other. In general, they will generate a “direct current” (DC)–like shift in heat
flow profiles that could not explain the observed
decreasing heat flow toward the arc in the subduction zone. However, if the sedimentation
effect is small, the frictional heating must be
small. Large frictional heating will not be consistent with the low heat flows observed in the
offshore Taiwan region.
Opposite to the sedimentation processes
reported in the offshore region, one of the highest erosion rates ever documented in the literature was found in the mountain region, on land,
in the Taiwan collision zone (e.g., Dadson et
688
al., 2003). These high erosion and exhumation
rates will likely contribute to the high heat flow
observed in the collision zone.
Topographic Effects
Heat flow that escapes from the crust will take
the shortest path to the low-temperature regions,
in this case, the seafloor. The irregular bathymetry will cause nonvertical heat transfer, and the
geothermal gradient increases under the valleys
while it decreases under the bathymetric highs
in the upper slope domain. The dense heat flows
derived from this study provide one of the first
opportunities for us to directly study the topographic effects over a wide range of scales.
To quantitatively study the effects of the
topography of the accretionary prism as a whole
on the geothermal gradient, we have set up a
finite-element numerical experiment using a
transect in the subduction zone at the latitude of
20.5°N (i.e., the lower panel of Fig. 2). Preliminary results give more than 20 °C/km of perturbation from topography alone, which is consistent with the results from the analytical solution
using a sinusoidal topography (cf. equation
4–66 of Turcotte and Schubert, 1982).
On a smaller scale, some of the local variation in the general trend of BSR subbottom
depths is due to its occurrence in regions of
folded strata. The curvature of the BSR is usually slightly less than that of the seafloor topography. A similar pattern is shown in seismic profiles from the western North Atlantic (Tucholke
et al., 1977). Ganguly et al. (2000) have also
quantitatively studied the topographic effects
on 1–2 km scale based on BSR data from the
Cascadia margin, and they found ~50% variation in heat flow due to topographic effect in
that region. This local observation suggests a
lower geothermal gradient on the crest than
on the limbs of a fold, similar to the patterns
from topographic effects. But this could also
be influenced by the more intense upward fluid
migration on the limbs, due to higher strain. In
addition, note that many processes generating
the topography, including thrusting and erosion, will also affect the heat flow pattern.
We found strong correlations between high
geothermal gradient and valleys over scales
ranging from hundreds of kilometers to kilometers. Topographic effects, along with some
processes generating the topography, can be
important factors contributing to the observed
heat flow pattern.
Fluid Migration
Figure 5 shows higher geothermal gradient at
the toe of the accretionary prism. Several studies
have shown that sediment dewatering and presumably fluid flow are most extensive along
the toe of accretionary prisms (Bray and Karig,
1985; Suess et al., 1998). Fluids are also released
by dehydration reactions within sediments
(Moore and Vrolijk, 1992). These fluids are
forced to migrate upward to the seafloor, advecting heat. Other evidence of overpressuring and
fluid migration within the lower slope domain,
especially in the collision zone, includes the presence of mud diapirs and volcanoes, on-land gas
seepages, bright spots beneath BSRs and fault
zones, and the small taper angle of the prism
(Chi et al., 1998). This implies that the fluid is
trapped in some locations and can only migrate
in several specific locations like fault zones or
diapirs, causing spatial variations in heat flow.
Three possible hydrothermally active sites with
BSR-based geothermal gradients exceeding 100
°C/km are identified in this region (Fig. 5).
In summary, the arcward-decreasing heat
flow pattern in the Taiwan subduction zone
can be best explained by the slab cooling
effect. We also found evidence of topographic
effects and focused fluid migration. It is less
clear how much sedimentation and shear heating from the fault affect the heat flow pattern.
On the other hand, the arcward-increasing heat
flow in the collision zone has been previously
modeled as a result of rapid exhumation. In
addition, topography, groundwater migration,
higher radiogenic heat, and some lower crust
and upper mantle processes might also affect
the heat flow pattern there.
Implication for Other Geological and
Geodynamical Processes
Zircon fission-track studies (Liu et al.,
2001; Willet et al., 2003) in the collision zone
show very fast cooling rates, some as large as
220 °C/m.y. for the past 2 m.y. Actually, Liu et
al. (2001) found the width of the zircon reset
zone, marked as a 260 °C isotherm, increases
as the collision becomes more mature to the
north. Although the increased cooling rate
could be caused by increased exhumation rate,
it could also be partially related to the northward increased geothermal gradient. In northern Taiwan where the fission-track studies were
mainly conducted, the average geothermal gradient at shallow depth is slightly higher than
the assumed 30 °C/km. The presumably higher
thermal conductivity at greater depths might
reduce the geothermal gradients at depth where
the fission-track data sampled. However, the
much higher geothermal gradients in southern
Taiwan might cause overestimates of exhumation rates, while the lower geothermal gradients in the initial collision zone and subduction
Geological Society of America Bulletin, May/June 2008
Crustal thermal evolution from subduction to collision
zone might cause underestimates. The large
variation of the geothermal gradient, from the
upper slope domain in the subduction zone, to
the Central Range in the collision zone, implies
that the 30 °C/km assumption in calculating the
uplifting rate might not be correct at all times
even though it has been shown to be able to capture the overall evolution of thermal structure
in this region. The geothermal gradients derived
from this study might provide a closer approximation. However, even these spatially varying,
geothermal gradients might not be representative at greater depths. New techniques and data
sets are needed to better constrain the temperature field at depths.
Heated materials will have higher seismic
attenuation. We found that the high geothermal
gradient region derived from this study correlates with the high seismic attenuation regions
proposed by Liu and Tsai (2005). The correlation is strongest in the southern part of the Central Range where geothermal gradient measurements in the mountainous regions are available.
This suggests that this high geothermal gradient
region can extend to at least mid-crust depth
where the seismic waves have sampled.
The high heat flow in the collision zone will
make the overall seismogenic zone on land
shallower (Lin, 2002). The lower limits of the
seismogenic zone in the convergent boundary is
proposed to be defined by the 350 °C isotherm,
which marks the change from velocity weakening to the velocity strengthening as the temperature increases (Tse and Rice, 1986). The
slip distribution of the 1999 Chi-Chi, Taiwan
earthquake on an east-dipping fault has a width
of ~40 km (e.g., Chi et al., 2001). The deeper
part of the mainshock slip terminated to the
east at a region where the heat flow increases
dramatically. No Chi-Chi aftershocks were
found that ruptured into this region (Chi and
Dreger, 2002; Chi and Dreger, 2004). Actually,
very few earthquakes were found at greater
depths in that region (Fig. 2). These results
suggest that the high heat flow in some of the
collision zone constrains the rupture extent of
large earthquakes. On the other hand, the lower
heat flow in the subduction zone implies that
the seismicity in the offshore region probably
can go deeper, thus increasing the width of the
seismogenic zone.
were derived offshore southern Taiwan. We then
combined this data set with previously published in situ heat flow measurements to study
the possible thermal evolution from subduction
to arc-continent collision.
In the subduction to the south, most of the
BSR-derived heat flows range from 35 to
85 mW/m2. The heat flow is high along the
trench and decreases toward the arc. On the
other hand, the in situ heat flow measurements
in the collision zone mostly range from 70 to
250 mW/m2. The heat flows are low near the
frontal thrust, and increase, instead of decrease,
toward the mountains.
The increased heat flow from subduction
zone to collision zone results from increasing
rock exhumation, reduced sedimentation and
increased erosion, and topographically induced
groundwater circulation after the upper slope
domain emerged from the sea surface.
Other geodynamic processes at depth might
have also affected the thermal structures. The
observed arcward-decreasing heat flow pattern
in the subduction zone can be modeled with
slab cooling, which shows ~50% reduction in
heat flow. We found highest heat flows located
above the regions without Benioff zones or
near the edges of the slabs, which might relate
to the lack of slab cooling. Other lower crust
and upper mantle processes might also affect
the thermal structures.
From this study, we found the thermal structures vary both in space, and very likely in time,
when the crust evolves from subduction to collision. This dense data set depicts low heat flows
in the subduction zone, suggesting low frictional
heat on the décollement in this region. The
increased geothermal gradient from subduction
to collision along the bathymetric and topographic high will give faster exhumation rates
for fission-track studies, if a constant geothermal gradient is used. The decreasing heat flow
toward the subduction zone also implies that
the seismogenic zone in the subduction zone in
Taiwan can be wider and reach to greater depth.
This dense and large heat flow data set might
provide a unique opportunity to study some
upper mantle, crust, and ocean thermal processes, in addition to P-T-t paths of metamorphic rocks in convergent boundaries.
ACKNOWLEDGMENTS
CONCLUSIONS
Most of the mountainous regions around the
world usually only have sparse heat flow data
due to their rough terrain. This study is one of
the first to document dense heat flow data from a
region that covers both subduction and collision.
In total, more than 1107 BSR-based heat flows
We thank George Spence, Brendan Murphy, and
reviewers Kelin Wang and Ingo Grevemeyer for
their unselfish help, which dramatically improved
this manuscript. This research is partially funded by
National Science Foundation grant OCE-9416583
to Donald Reed and National Science Council of
Taiwan grant 96-2119-M-001-015 to WCC. This is
contribution number IESAS1208 of the Institute of
Earth Sciences, Academia Sinica.
REFERENCES CITED
Ashi, J., and Taira, A., 1993, Thermal structure of the Nankai accretionary prism as inferred from the distribution
of gas hydrate BSRs, in Underwood, M.B., ed., Thermal evolution of the Tertiary Shimanto Belt, southwest
Japan: An example of ridge-trench interaction: Geological Society of America Special Paper 273, p. 137–149.
Bangs, N.L.B., Sawyer, D.S., and Golovchenko, X., 1993,
Free gas at the base of the gas hydrate zone in the
vicinity of the Chile Triple Junction: Geology, v. 21,
p. 905–908, doi: 10.1130/0091-7613(1993)021<0905:
FGATBO>2.3.CO;2.
Barr, T.D., and Dahlen, F.A., 1989, Brittle frictional mountain building: 2. Thermal structure and heat budget:
Journal of Geophysical Research, v. 94, p. 3923–3947.
Bowin, C., Lu, R.S., Lee, C.S., and Schouten, H., 1978,
Plate convergence and accretion in the Taiwan-Luzon
region: The American Association of Petroleum Geologists Bulletin, v. 62, p. 1645–1672.
Bray, C.J., and Karig, D.E., 1985, Porosity of sediments in
accretionary prisms and some implications for dewatering processes: Journal of Geophysical Research,
v. 90, p. 768–778.
Brewer, P.G., Orr, F.M., Friederich, G., Kvenvolden, K.A.,
Orange, D.L., McFarlane, J., and Kirkwood, W., 1997,
Deep-ocean field test of methane hydrate formation
from a remotely operated vehicle: Geology, v. 25,
p. 407–410, doi: 10.1130/0091-7613(1997)025<0407:
DOFTOM>2.3.CO;2.
Brooks, J.M., Field, M.E., and Kennicutt, M.C., II, 1991,
Observations of gas hydrates in marine sediments,
offshore northern California: Marine Geology, v. 96,
p. 103–109, doi: 10.1016/0025-3227(91)90204-H.
Brown, K.M., and Bangs, N.L., 1995, Thermal regime of the
Chile Triple Junction: Constraints provided by downhole temperature measurements and distribution of
gas hydrate: Proceedings, the Ocean Drilling Program,
Scientific Results, v. 141, p. 259–275.
Chang, H.J., 2002, Settling particulates and sediments in the
Northern South China Sea: Distributions of mass flux
and Pb-210 [M.S. thesis]: Kaohsiung, Taiwan, National
Sun Yat-Sen University, 64 p.
Chi, W.C., and Dreger, D., 2002, Finite fault inversion of
the September 25, 1999, (Mw=6.4) Taiwan earthquake:
Implications for GPS displacements of Chi-Chi, Taiwan,
earthquake sequence: Geophysical Research Letters,
v. 29, no. 14, p. 1694, doi: 10.1029/2002GL015237.
Chi, W.C., and Dreger, D., 2004, Crustal deformation in
Taiwan: Results from finite source inversions of six
Mw>5.8 Chi-Chi aftershocks: Journal of Geophysical
Research–Solid Earth, v. 109(B7), Article no. B07305.
Chi, W.C., Reed, D.L., Lundberg, N., and Liu, C.S., 1998,
Distribution of the bottom-simulating reflector in the
offshore Taiwan collision zone: Terrestrial Atmospheric and Oceanic Sciences, v. 9, no. 4, p. 779–794.
Chi, W.C., Dreger, D., and Kaverina, A., 2001, Finite source
modeling of the 1999 Taiwan (Chi-Chi) Earthquake
derived from a dense strong motion network: Bulletin
of the Seismological Society of America, v. 91, no. 5,
p. 1144–1157, doi: 10.1785/0120000732.
Chi, W.C., Reed, D.L., Moore, G., Lundberg, N., and Liu,
C.S., 2003, Tectonic wedging along the rear of the
offshore Taiwan accretionary prism: Tectonophysics,
v. 374, p. 199–217, doi: 10.1016/j.tecto.2003.08.004.
Currie, C.A., Wang, K., Hyndman, R.D., and He, J., 2004,
The thermal effects of steady-state slab-driven mantle
flow above a subducting plate: The Cascadia subduction
zone and backarc: Earth and Planetary Science Letters,
v. 223, p. 35–48, doi: 10.1016/j.epsl.2004.04.020.
Dadson, S.J., Hovius, N., Chen, H., Dade, W.B., Hsieh, M.L.,
Willett, S.D., Hu, J.C., Horng, M.J., Chen, M.C., Stark,
C.P., Lague, D., and Lin, J.C., 2003, Links between
erosion, runoff variability, and seismicity in the Taiwan orogen: Nature, v. 426, p. 648–651, doi: 10.1038/
nature02150.
Foucher, J.P., Le Pichon, X., Lallemant, S., Hobart, M.A.,
Henry, P., Benedetti, M., Westbrook, G.K., and Langseth, M.G., 1990, Heat flow, tectonics, and fluid circulation at the toe of the Barbados Ridge accretionary
prism: Journal of Geophysical Research–Solid Earth,
v. 95, B6, p. 8859–8867.
Geological Society of America Bulletin, May/June 2008
689
Chi et al.
Ganguly, N., Spence, G.D., and Chapman, N.R., 2000, Heat
flow variations from bottom-simulating reflectors on
the Cascadia margin: Marine Geology, v. 164, p. 53–
68, doi: 10.1016/S0025-3227(99)00126-7.
Grevemeyer, I., and Villinger, H., 2001, Gas hydrate stability and the assessment of heat flow through continental
margins: Geophysical Journal International, v. 145,
p. 647–660, doi: 10.1046/j.0956-540x.2001.01404.x.
Hamilton, E.L., 1980, Geoacoustic modeling of the seafloor:
The Journal of the Acoustical Society of America,
v. 68, p. 1313–1339, doi: 10.1121/1.385100.
Ho, C.S., 1986, A synthesis of the geologic evolution of Taiwan: Tectonophysics, v. 125, p. 1–16, doi: 10.1016/00401951(86)90004-1.
Holbrook, W.S., Hoskins, H., Wood, W.T., Stephen, R.A.,
and Lizarralde, D., 1996, Methane hydrate and free
gas on the Blake Ridge from vertical seismic profiling:
Science, v. 273, no. 5283, p. 1840–1843, doi: 10.1126/
science.273.5283.1840.
Hsu, S.K., and Sibuet, J.C., 2004, Continent-ocean transition
of the northern South China Sea and off southwestern Taiwan: Marine Geophysical Researches, v. 25,
no. 1–2, p. 1–4, doi: 10.1007/s11001-005-0729-1.
Hwang, W.T., and Wang, C.Y., 1993, Sequential thrusting
model for mountain building: Constraints from geology and heat flow of Taiwan: Journal of Geophysical
Research, v. 98, p. 9963–9973.
Hyndman, R.D., and Davis, E.E., 1992, A mechanism for
the formation of methane hydrate and seafloor bottomsimulating reflectors by vertical fluid expulsion: Journal of Geophysical Research, v. 97, p. 910–924.
Hyndman, R.D., and Lewis, T.J., 1999, Geophysical consequences of the Cordillera-Craton thermal transition in
southwestern Canada: Tectonophysics, v. 306, p. 397–
422, doi: 10.1016/S0040-1951(99)00068-2.
Hyndman, R.D., and Wang, K., 1995, The rupture zone of
Cascadia great earthquakes from current deformation and the thermal regime: Journal of Geophysical
Research, v. 100, no. B11, p. 22,133–22,154.
Hyndman, R.D., Foucher, J.P., Yamano, M., Fisher, A., and
Scientific Team of Ocean Drilling Program Leg 131,
1992, Deep-sea bottom-simulating reflectors: Calibration of the base of the hydrate stability field as used for
heat flow estimates: Earth and Planetary Science Letters,
v. 109, p. 289–301, doi: 10.1016/0012-821X(92)90093-B.
Hyndman, R.D., Wang, K., Yuan, T., and Spence, G.D., 1993,
Tectonic sediment thickening, fluid expulsion, and the
thermal regime of subduction zone accretionary prisms:
The Cascadia margin off Vancouver Island: Journal of
Geophysical Research, v. 98, p. 21,865–21,876.
Jessop, A.M., 1990, Thermal geophysics, developments in
solid earth geophysics: Amsterdam, Elsevier Science
Publishers, v. 17, p. 306.
Kao, H., Shen, S.J., and Ma, K.F., 1998, Transition from
oblique subduction to collision—Earthquakes in the
southernmost Ryukyu arc–Taiwan region: Journal of
Geophysical Research, v. 103, no. B4, p. 7211–7229,
doi: 10.1029/97JB03510.
Kohl, T., 1998, Palaeoclimatic temperature signals—Can
they be washed out?: Tectonophysics, v. 291, p. 225–
234, doi: 10.1016/S0040-1951(98)00042-0.
Kvenvolden, K.A., and McMenamin, M.A., 1980, Hydrates
of natural gas: A review of their geologic occurrence:
U.S. Geological Survey Circular 825, p. 1–11.
Lan, C.Y., Lee, T., and Lee, C.W., 1990, The Rb-Sr isotope
record in Taiwan gneiss and its tectonic implications:
Tectonophysics, v. 183, p. 129–143, doi: 10.1016/00401951(90)90412-2.
Lee, C.R., and Cheng, W.T., 1986, Preliminary heat flow
measurements in Taiwan: Proceedings, Singapore,
Fourth Circum-Pacific Energy and Mineral Resources
Conference, p. 1–9.
Lewis, T.J., Hyndman, R.D., and Fluck, P., 2003, Heat flow,
heat generation, and crustal temperatures in the northern Canadian Cordillera: Thermal control of tectonics: Journal of Geophysical Research, v. 108, no. B6,
p. 2316, doi:10.1029/2002JB002090.
Lin, A.T., Watts, A.B., and Hesselbo, S.P., 2003, Cenozoic
stratigraphy and subsidence history of the South China
Sea margin in the Taiwan region: Basin Research,
v. 15, no. 4, p. 453–478.
Lin, C.H., 2002, Active continental subduction and crustal
exhumation: The Taiwan orogeny: Terra Nova, v. 14,
p. 281–287, doi: 10.1046/j.1365-3121.2002.00421.x.
690
Liu, K.S., and Tsai, Y.B., 2005, Attenuation relationships
of peak ground acceleration and velocity for crustal
earthquakes in Taiwan: Bulletin of the Seismological
Society of America, v. 95, no. 3, p. 1045–1058, doi:
10.1785/0120040162.
Liu, T.K., Hsieh, S., Chen, Y.G., and Chen, W.S., 2001,
Thermo-kinematic evolution of the Taiwan oblique-collision mountain belt as revealed by zircon fission-track
dating: Earth and Planetary Science Letters, v. 186,
p. 45–56, doi: 10.1016/S0012-821X(01)00232-1.
Lorenzo, J.M., and Vera, E.E., 1992, Thermal uplift and erosion across the continent-ocean transform boundary
of the southern Exmouth Plateau: Earth and Planetary
Science Letters, v. 108, p. 79–92, doi: 10.1016/0012821X(92)90061-Y.
Miller, J.J., Lee, M.W., and von Huene, R., 1991, An analysis of a seismic reflection from the base of a gas hydrate
zone, offshore Peru: The American Association of
Petroleum Geologists Bulletin, v. 75, p. 2204–2213.
Molnar, P., and England, P., 1990, Temperatures, heat flux,
and frictional stress near major thrust faults: Journal
of Geophysical Research–Solid Earth, v. 95, no. B4,
p. 4833–4856.
Moore, J.C., and Vrolijk, P., 1992, Fluids in accretionary
prisms: Reviews of Geophysics, v. 30, p. 113–135.
Nissen, S.S., Hayes, D.E., Yao, B.C., Zeng, W.J., Yongqin, C.,
and Xiaupin, N., 1995, Gravity, heat flow, and seismic
constraints on the processes of crustal extension: Northern margin of the South China Sea: Journal of Geophysical Research, v. 100, no. B11, p. 22,447–22,484.
Reed, D.L., Lundberg, N., Liu, C.S., McIntosh, K.D., Lieske,
J.H., and Kuo, B.Y., 1991, Strain domains, fluid/gas
migration, and protothrusting in the offshore Taiwan
accretionary wedge: Geological Society of America
Abstracts with Programs, v. 23, p. 365.
Reed, D.L., Lundberg, N., Liu, C.S., and Kuo, B.T., 1992,
Structural relations along the margin of the offshore
Taiwan accretionary wedge: Implications for accretion and crustal kinematics: Acta Geologica Taiwanica,
v. 30, p. 105–122.
Shipley, T.H., Houston, M.H., Buffler, R.T., Shaub, F.J.,
McMillen, K.J., Ladd, J.W., and Worzel, J.L., 1979,
Seismic evidence for widespread possible gas hydrate
horizons on continental slopes and rises: The American
Association of Petroleum Geologists Bulletin, v. 63,
p. 2204–2213.
Schnurle, P., Liu, C.S., Hsiuan, T.H., and Wang, T.K., 2004,
Characteristics of gas hydrate and free gas offshore
southwestern Taiwan from a combined MCS/OBS
data analysis: Marine Geophysical Researches, v. 25,
p. 157–180, doi: 10.1007/s11001-005-0740-6.
Shyu, C.T., Hsu, S.K., and Liu, C.S., 1998, Heat flows off
southwestern Taiwan: Measurements over mud diapirs
and estimated from bottom-simulating reflectors: Terrestrial Atmospheric and Oceanic Sciences, v. 9, no. 4,
p. 795–812.
Shyu, C.T., Chen, Y.J., Chiang, S.T., and Liu, C.S., 2006,
Heat flow measurements over bottom-simulating
reflectors, offshore southwestern Taiwan: Terrestrial Atmospheric and Oceanic Sciences, v. 17, no. 4,
p. 845–869.
Stein, C.A., and Stein, S., 1992, A model for the global variation in oceanic depth and heat flow with lithospheric age:
Nature, v. 359, p. 123–129, doi: 10.1038/359123a0.
Suess, E., Bohrmann, G., von Huene, R., Linke, P., Wallmann, K., Lammers, S., Sahling, H., Winckler, G.,
Lutz, R.A., and Orange, D., 1998, Fluid venting in
the eastern Aleutian subduction zone: Journal of Geophysical Research, v. 103, p. 2597–2614, doi: 10.1029/
97JB02131.
Suppe, J., 1981, Mechanics of mountain building and metamorphism in Taiwan: Geological Society of China
Memoir, v. 4, p. 67–89.
Suppe, J., 1984, Kinematics of arc-continent collision, flipping of subduction, and backarc spreading near Taiwan:
Geological Society of China Memoir, v. 6, p. 21–33.
Teng, L.S., 1990, Geotectonic evolution of late Cenozoic arccontinent collision in Taiwan: Tectonophysics, v. 183,
p. 57–76, doi: 10.1016/0040-1951(90)90188-E.
Teng, L.S., Lee, C.T., Tsai, Y.B., and Hsiao, L-Y., 2000,
Slab breakoff as a mechanism for flipping of subduction polarity in Taiwan: Geology, v. 28, no. 2,
p. 155–158, doi: 10.1130/0091-7613(2000)28<155:
SBAAMF>2.0.CO;2.
Torres, M.E., McManus, J., Hammond, D.E., de Angelis,
M.A., Heeschen, K.U., Colbert, S.L., Tryon, M.D.,
Brown, K.M., and Suess, E., 2002, Fluid and chemical
fluxes in and out of sediments hosting methane hydrate
deposits on Hydrate Ridge, OR, I: Hydrological provinces: Earth and Planetary Science Letters, v. 201,
p. 525–540, doi: 10.1016/S0012-821X(02)00733-1.
Townend, J., 1997, Estimates of conductive heat flow
through bottom-simulating reflectors on the Hikurangi
and southwest Fiordland continental margins: New
Zealand: Marine Geology, v. 141, p. 209–220, doi:
10.1016/S0025-3227(97)00073-X.
Tse, S.T., and Rice, J.R., 1986, Crustal earthquake instability in relation to the depth variation of frictional slip
properties: Journal of Geophysical Research, v. 89,
p. 5801–5816.
Tucholke, B.E., Byran, G.M., and Ewing, J.I., 1977, Gashydrate horizons detected in seismic-profiles data
from the western Northern Atlantic: The American
Association of Petroleum Geologists Bulletin, v. 61,
p. 698–707.
Turcotte, D.L., and Schubert, G., 1982, Geodynamics:
Applications of continuum physics to geological problems: New York, John Wiley and Sons, 450 p.
Vogt, P.R., and Sundvor, E., 1996, Heat flow highs on the
Norwegian-Barents Svalbard continental slope: Deep
crustal fractures, dewatering, or “memory in the
mud”?: Geophysical Research Letters, v. 23, p. 3571–
3574, doi: 10.1029/96GL03259.
Von Herzen, R.P., and Uyeda, S., 1963, Heat flow through
the Eastern Pacific Ocean floor: Journal of Geophysical
Research, v. 68, p. 4219–4250.
Von Herzen, R.P., Ruppel, C., Molnar, P., Nettles, M., Nagihara, S., and Ekstrom, G., 2001, A constraint on the shear
stress at the Pacific-Australian plate boundary from heat
flow and seismicity at the Kermadec forearc: Journal of
Geophysical Research, v. 106, no. 4, p. 6817–6833.
Wakabayashi, J., 2004, Tectonic mechanisms associated
with P-T paths of regional metamorphism: Alternatives
to single-cycle thrusting and heating: Tectonophysics,
v. 392, p. 193–218, doi: 10.1016/j.tecto.2004.04.012.
Wang, K., Mulder, T., Rogers, G.C., and Hyndman, R.D.,
1995, Case for very low coupling stress on the Cascadia subduction fault: Journal of Geophysical
Research, v. 100, no. b7, p. 12,907–12,918, doi:
10.1029/95JB00516
Willett, S.D., Fisher, D., Fuller, C., En-Chao, Y., and Lu,
C.Y., 2003, Erosion rates and orogenic-wedge kinematics in Taiwan inferred from fission-track thermochronometry: Geology, v. 31, no. 11, p. 945–948, doi:
10.1130/G19702.1.
Xia, K.Y., Xia, S.G., Chen, Z.R., and Huang, C., 1995,
Geothermal characteristics of the South China Sea, in
Gupta, L., and Yamano, M., eds., Terrestrial heat flow
and geothermal energy in Asia: Rotterdam, Netherlands, A.A. Balkema Publishers, p. 113–128.
Yamano, M., Uyeda, S., Aoki, Y., and Shipley, T.H., 1982,
Estimates of heat flow derived from gas hydrates:
Geology, v. 10, p. 339–343, doi: 10.1130/00917613(1982)10<339:EOHFDF>2.0.CO;2.
Yamano, M., Foucher, J.P., Kinoshita, M., Fisher, A.,
Hyndman, R.D., and ODP Leg 131 Shipboard Scientific Party, 1992, Heat flow and fluid flow regime
in the western Nankai accretionary prism: Earth and
Planetary Science Letters, v. 109, p. 451–462, doi:
10.1016/0012-821X(92)90105-5.
Yang, T.F., Chuang, P.C., Lin, S., Chen, J.C., Wang, Y., and
Chung, S.H., 2006, Methane venting in gas hydrate
potential area offshore of SW Taiwan: Evidence of gas
analysis of water column samples: Terrestrial Atmospheric and Oceanic Sciences, v. 17, no. 4, p. 933–950.
Yu, S.B., Chen, H.Y., and Kuo, L.C., 1997, Velocity field of
GPS stations in the Taiwan area: Tectonophysics, v. 274,
p. 41–59, doi: 10.1016/S0040-1951(96)00297-1.
Zwart, G., and Moore, J.C., 1993, Variations in temperature
gradient of the Oregon accretionary prism: Eos (Transactions, American Geophysical Union) Spring Meeting
Supplement, v. 74, p. 222.
MANUSCRIPT RECEIVED 27 FEBRUARY 2007
REVISED MANUSCRIPT RECEIVED 2 NOVEMBER 2007
MANUSCRIPT ACCEPTED 2 NOVEMBER 2007
Printed in the USA
Geological Society of America Bulletin, May/June 2008