Tomographic evidence for wholesale underthrusting of India ...easd.geosc.uh.edu/murphy/pdf/2005-zhou_murphy_jaes.pdfthick crust of Tibet from the surface down to 75 km depth. In the

Tomographic evidence for wholesale underthrusting

of India beneath the entire Tibetan plateau

Hua-wei Zhou*, Michael A. Murphy

Department of Geosciences, University of Houston, Houston, TX 77204-5007, USA

Received 15 April 2004; accepted 26 April 2004

Abstract

We analyzed a global tomographic model for the Tibet–Himalayan collision zone, which indicates that the Indian lithospheric slab has

been subducted subhorizontally beneath nearly the entire Tibetan plateau to depths of 165–260 km. Tibetan velocity structure is low in its

crust and high in its lithospheric mantle at depths between 75 and 120 km. We interpret an asthenospheric layer positioned above the

subducted Indian slab at depths between 120 and 165 km beneath the Tibetan plateau. Beneath the central portion of the plateau a low-

velocity anomaly exists from the crust down to 310 km depth, indicating mantle upwelling through a weakened part of the subducted slab.

We present a model, which explains that, the uplift history and low relief of the Tibetan plateau is a result of subhorizontal subduction and

heating of Indian lithosphere that is separated from Tibetan lithosphere by a thin channel of asthenosphere. Two predictions made by our

model are: (1) the amount of shortening in the Himalayas is equivalent to the amount of underthrusted Indian mantle lithosphere; and (2) a

young mantle geochemical signature should be present along the entire southern portion of the Tibetan plateau.

q 2004 Elsevier Ltd. All rights reserved.

Keywords: Seismic tomography; Tibetan plateau; Subduction; Underthrusting; India; Himalaya

1. Introduction

Continent–continent collision leading to orogenesis,

which may ultimately result in the development of super-

continents, is a fundamental tectonic process that has

significantly affected the character of Earth’s surface for at

least the past one billion years. In order to characterize the

dynamics of continent–continent collisions, it is essential to

define its present geometry and physical state. This is

especially true for the largest active continent–continent

collision zone on Earth, the Himalayan–Tibetan orogen.

The northward convergence of India into Asia over the past

50 Ma (Patriat and Achache, 1984; Dewey et al., 1988) is

intimately linked to the creation of the Tibetan plateau and

has impacted the tectonic framework throughout Asia

since Cenozoic time (Molnar and Tapponnier, 1975;

1367-9120/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.

doi:10.1016/j.jseaes.2004.04.007

* Corresponding author. Tel.: C1-713-743-3424; fax: C1-713-743-

9164.

E-mail address: [email protected] (H.-w. Zhou).

Yin and Harrison, 2000). The challenge of unraveling the

evolution of the Tibet–Himalayan orogen, and specifically

the uplift history of the Tibetan plateau, has been

significantly bolstered by the collection of a wide variety

of geologic and geophysical information. Nevertheless,

unresolved issues remain regarding the mechanisms by

which the convergence between India and Asia was

accommodated and the Tibetan plateau was uplifted.

Models that explain the Cenozoic uplift of the Tibetan

plateau can be differentiated from one another by the

predictions they make regarding the lithospheric structure

beneath the Tibetan plateau: (a) complete underthrusting of

the Indian plate (Argand, 1924; Powell and Conaghan,

1973; Powell, 1986) or portions of it beneath Tibet

(DeCelles et al., 2003); (b) lithospheric thickening by

distributed thrust faulting (Dewey and Burke, 1973; Chang

et al., 1986; England and Houseman, 1989) and attendant

convective removal of overthickened mantle lithosphere

(England and Houseman, 1989; Houseman and Molnar,

2001); and (d) intracontinental subduction along reactivated

Journal of Asian Earth Sciences 25 (2005) 445–457

www.elsevier.com/locate/jaes

http://www.elsevier.com/locate/jaes

Fig. 1. Map view of P1200 tomographic model (Zhou, 1996) for Tibet region at depth range of 165–260 km. Red lines are major faults and red dotted

lines are major suture zones in Tibet. IYSZ, Indus-Yalu suture zone; BNS, Bangong-Nujiang suture; JS, Jinsha suture. Purple crosses are earthquake

foci. Purple star is the location of the destructive earthquake near Jamnagar, India, on 26/01/01. Dark gray curve delineates the inferred northern

margin of the subducted Indian lithospheric slab in this depth range. Light gray curve outlines the high-velocity portion of the Tibetan lithospheric

mantle above 120 km depth.

H.-w. Zhou, M.A. Murphy / Journal of Asian Earth Sciences 25 (2005) 445–457446

sutures (Meyer et al., 1998). A necessary first step towards

assessing the merits of these models is a thorough

description of the geometry and physical state of the

lithospheric structure in this region.

This paper provides new insight into the lithospheric

structure beneath the Tibetan plateau based on analyzing a

high-resolution global tomographic model (Zhou, 1996).

Two surprising results are the presence of a slab-shaped

high seismic velocity anomaly beneath nearly the entire

Tibetan plateau that we interpret to be subducted Indian

lithospheric mantle, and an overlying wedge-shaped, slow

seismic velocity anomaly interpreted as an asthenospheric

layer (Fig. 1). Not only does this seismic structure set

boundary conditions for geodynamic modeling of the Tibet–

Himalayan collision zone, but also supports an inference for

the cause of the destructive earthquake that occurred in

western India on 26/01/01.

2. Observations from a global tomography

Seismic tomography provides a snapshot of the deep

Earth structure as expressed by its seismic properties, based

on traveltimes of seismic waves propagating from earth-

quake or man-made sources to seismological stations. In

deriving tomographic models at regional scales, data from

sources or receivers outside the model area are usually used

to enhance the data coverage. Since there are few

seismological stations in Tibet, for example, it is necessary

to use outside sources and stations that have rays traversing

beneath Tibet. A dilemma in using raypaths partially outside

of the model area is that the gain in data quantity is at the

expense of bringing potential contaminations from outside.

However, this dilemma can be overcome by using a

global model that leaves no outside areas, if the model has

high enough resolution (Zhou, 1996; Bijwaard et al., 1998).

H.-w. Zhou, M.A. Murphy / Journal of Asian Earth Sciences 25 (2005) 445–457 447

In this study, we analyze a high-resolution global tomo-

graphic model called P1200 that describes the P-wave

velocity variations from the surface down to 1200 km depth

(Zhou, 1996), using cell size of 18 laterally and 35–50 km

vertically. Although lateral heterogeneities below 1200 km

depth could have small effects on steep-dipping features at

shallow depths, most velocity anomalies interpreted in this

paper are of subhorizontal geometry and therefore not

affected much by deep mantle anomalies through deep-

diving rays.

Fig. 2. Depth slices of velocity variations of the P1200 model (Zhou, 1996). The

75–120 km slice the high-velocity portion of the Tibetan lithospheric mantle is o

slices the inferred margin of the subducted Indian lithospheric mantle is indicate

Fig. 2 shows nine depth slices of the P1200 model

from the surface down to 410 km depth across the

Himalayan–Tibetan collision zone. The abundance of

earthquakes in the region and its surrounding areas

helps to constrain the velocity anomalies. The images

on the depth slices are dominated by coherent velocity

anomalies greater than 108!108 in lateral dimensions,

although there are also smaller anomalies. The resolution

of the data has been tested using checkerboard impulse

resolution tests (Zhou, 1988; Zhou and Clayton, 1990).

coastline is shown in green, and 4-km elevation is shown in black. On the

utlined by a shadowed light gray curve. On the 165–210 and 210–260 km

d by a shadowed dark gray curve. Other legends follow that of Fig. 1.

Fig. 3. Resultofcheckerboard resolution tests at six depth ranges, as indicated in the lower-left cornerof eachpanel. The lateraldimensionof the impulses is38!38.


As shown in Fig. 3, except some corner areas at shallow

depth ranges, excellent results are achieved throughout the

region at a resolution of 38 laterally and about 150 km

vertically.

The good resolution result as shown in Fig. 3 is largely

due to the use of global data set that contains many rays with

neither sources nor stations inside Tibet, such as earth-

quakes in central Asia that were recorded by Chinese


stations. The resolution tests suggest that the velocity

anomalies interpreted in this paper would be reliable if the

signal-to-noise ratio of the data is also sufficient. One

concern is from mantle anomalies below 1200 km depth.

For the subhorizontal, high-amplitude, and shallow struc-

tures interpreted in this paper, however, the contamination

from mantle anomalies below 1200 km depth is relatively

small. The processing of the ISC traveltime data for the

tomographic inversion included corrections for station

statics, hypocentral re-determination, and use of summary

ray culling (Zhou, 1996). The model was constrained by the

multi-scale inversion that improves the handling of uneven

ray coverage. Nevertheless, to reduce the bias of our

interpretation, the full array of depth slices and cross-

sections are presented here.

As shown in Figs. 1 and 2, low velocities characterize the

thick crust of Tibet from the surface down to 75 km depth.

In the 0–35 km depth slice two low-velocity zones extend

from the Tien Shan through western Tarim to western Tibet,

and from the eastern edge of Tibet southwards to the Gulf of

Thailand. These anomalies persist in the second depth slice,

which reveal two additional low velocity patches; one in

central Tibet between 82 and 938E, and the other beneath

western Mongolia and the Qilian Shan. The slow anomaly in

the central Tibet continues in deeper slices to at least

310 km depth, and is narrowing with depth to be within 84

and 918E around 300 km depth. In the 165–260 km depth

range, the region is actually faster than the layers’ average

values, but slower than the rest of the high-velocity slab.

We interpret the extremely coherent, slab-shaped high-

velocity anomaly in the four depth slices from 75 to 260 km

in Fig. 2 as Indian mantle lithosphere. On the 75–120 km

depth slice, for example, the appropriate shape of a

combined India and Tibet is discernable as the largest

high-velocity anomaly. In contrast, on the 120–165 km

depth slice, there is a northwest-trending low-velocity

anomaly from Tarim to central Tibet. This slow anomaly

appears to be connected with the low-velocity anomaly in

the Bay of Bengal. At the same depth range to the

southwest, a northwest-trending high-velocity feature

stretches from the Aral Sea to central India, and another

high-velocity anomaly exists from eastern Tibet to the

Burma arc. On the 165–210 and 210–265 km depth slices,

the Indian lithospheric mantle is the most coherent high-

velocity anomaly that extends northwards beneath nearly

the entire Tibetan plateau. Results from these two depth

slices have been stacked vertically to produce the depth slice

of the subducting Indian slab shown in Fig. 1. In the deepest

three depth slices of Fig. 2 the coherent high-velocity

anomaly in India and Tibet disappears, and the largest high-

velocity anomaly on these slices is north of 408N.

A comparison between the depth slices in Fig. 2 reveals

that the high-velocity Indian lithospheric mantle anomaly is

shifted north–northeasterly as it plunges deeper, manifest-

ing a subduction pattern. As seen in cross-sections A–A1,

B–B1, and D–D1 in Fig. 4, the high-velocity Indian slab

starts to dip towards the Tibetan plateau hundreds of

kilometers south of the Himalayas. Beneath the Tibetan

plateau a low-velocity layer, which we interpret as a thin

asthenospheric layer, exists between the high-velocity

Tibetan lithospheric mantle and the subducted high-velocity

Indian slab. Cross-section C–C1 traverses through the

anomalous low-velocity area in central Tibet and will be

compared later with another study. At lower-mantle depths

beneath India (cross-sections A–A1, B–B1, and D–D1)

there is a large, high-velocity anomaly that has been

interpreted as the subducted Tethyan oceanic slab (Van der

Voo et al., 1999). A complete set of cross-sections is shown

in Fig. 5, with 11 dip sections (E–E1 to O–O1) and three

strike sections (P–P1 to R–R1) with respect to the orogenic

front in the central Himalaya.

A general interpretation can be drawn from the depth

slices and cross-sections. The Indian lithospheric slab is

clearly subducted horizontally beneath the Tibetan plateau.

Above the subducted Indian slab, the Tibetan lithosphere

has an abnormally low-velocity crust and a high-velocity

lithospheric mantle above 120 km depth. In the depth range

of 120–165 km a low-velocity layer exists which we

interpret as an asthenospheric layer between the Tibetan

lithosphere and the subducted Indian slab. The subducted

Indian slab is very strong west of 858E and east of 938E

(Fig. 1). In central Tibet, between 85 and 938E and north of

the Indus-Yalu suture zone, there is a low-velocity area

within the subducted slab. This area is part of a low-velocity

anomaly that extends vertically from 35 km down to at least

310 km in depth. Interestingly, many previous studies in

Tibet were conducted in the area over this anomaly, and

have found features such as low seismic velocities and

indications of partial melting and a mantle-originated

geochemical anomaly (McNamara et al., 1994; Nelson et

al., 1996; Kind et al., 1996; Turner et al., 1996; Owens and

Zandt, 1997; Yuan et al., 1997; Makovsky and Klemperer,

1999; Kosarev et al., 1999; Hoke et al., 2000).

3. Comparison with previous studies

The unique nature of Tibetan plateau has attracted a large

number of previous seismologic studies. During the 1990’s,

for instance, there were findings of high S-wave velocities

under westernmost Tibet and the Karakorum, and low

velocities under central Tibet (Molnar, 1990; Woodward and

Molnar, 1995), low Pn velocity region in north central (Zhao

and Xie, 1993; McNamara et al., 1997), and high P-wave

velocities beneath southern Tibet (Pandey et al., 1991).

The deep seismic velocity structure analyzed in this paper is

consistent with many of the previous studies. For instance,

the lower mantle high-velocity anomaly beneath northern

India as shown in Fig. 4 is similar to the shallow portion of a

large high-velocity belt mapped by Van der Voo et al. (1999).

These investigators interpreted it as a subducted Tethyan

oceanic slab that extends laterally from the Mediterranean

Fig. 4. Four great-circle cross-sections of P-wave lateral velocity variations at locations shown in the insert map. Small purple crosses are earthquake foci. The

star in A–A1 denotes the destructive earthquake near Jamnagar, India. Shown above each section is topographic profile with a 20:1 vertical exaggeration.

Dashed lines in magenta are elevations of 0, 2, and 4 km.


Sea to the Andaman Sea and deepens to at least 2000 km in

depth. The low-velocity anomaly mapped and interpreted as

the signature of the Deccan plume by Kenneth and

Widiyantoro (1999) can also be seen in the 35–75 km

depth slice in Fig. 2, though the anomaly occurs slightly to

the south in the P1200 model. From the two depth slices by

these authors at 100 and 250 km depths, one can infer similar

patterns of the Indian slab as shown in the corresponding

depth slices in Fig. 2. In fact, the high-velocity Indian

slab in Fig. 1 matches well with the configuration of

Fig. 5. Fourteen great-circle cross-sections of P-wave lateral velocity variations at locations shown in the insert map. The lateral dimension is latitude for

sections E–E1 to O–O1, and longitude for sections P–P1 to R–R1. See caption of Fig. 4 for other legends.


the high-velocity anomaly in the 200 km panel of Plate 1 by

Bijwaard et al. (1998). The S-wave velocity map at 100 km

depth by Villasenor et al. (2001) also show that the high-

velocity Indian slab anomaly extends into the western part of

Tibet and that the central part of Tibet has a large low-

velocity region as seen in Fig. 1 in this paper.

Zhou et al. (1996) conducted a careful modeling study of

differential residual sphere data for four Tibetan earth-

quakes. Their best fitting model consists of a narrow, nearly

vertical, high velocity slab extending to at least 400 km

beneath the southern Tibetan plateau. As shown in Figs. 2

and 5, the P1200 model also contains a high velocity slab at

similar location, though the slab in the P1200 model

stays above 300 km and extends northward subhorizontally.

Fig. 6 shows the predicted differential residual spheres from

the P1200 model for the same four events (Zhou et al.,

1996). The level of fitness with the differential residual

sphere data is compatible between predictions by the P1200

model in Fig. 6 and that by the model of Zhou et al. (1996).

Hence a subhorizontal Indian slab offers a compatible

possibility for the differential residual sphere data.

Before the 1990’s most seismologic studies of the

Tibetan deep structure were aimed at constraining large-

scale features using stations mostly outside the plateau

(Roecker, 1982; Ni and Barazangi, 1983; Jobert et al., 1985;

Molnar, 1988). In the last decade, there have been several

deployments of seismologic arrays, such as the PASSCAL

experiment (McNamara et al., 1994; Owens and Zandt,

1997) and the INDEPTH II/GEODEPTH experiment

(Nelson et al., 1996; Kind et al., 1996; Yuan et al., 1997;

Fig. 6. Differential residual sphere predictions from the P1200 model for four events studied by Zhou et al. (1996). Each event pair is indicated below each

sphere. The differential traveltime residuals are plotted as a function of azimuth along the circumference and takeoff angle along the radius. The blue stars

indicate relatively fast arrivals in seconds, and the red crosses indicate relatively slow arrivals.


Makovsky and Klemperer, 1999) across the central and

southern Tibet. Interestingly, these studies were conducted

within the area over the low-velocity anomaly shown in

Figs. 1 and 2. Our study demonstrates that this low-velocity

feature extends down to at least 310 km depth. This feature

is consistent with the presence of volcanic activity (Turner

et al., 1996) and indications of partial melting in the

lower crust (Owens and Zandt, 1997; Nelson et al., 1996;

Kind et al., 1996).

Although the P1200 model is of lower resolution

compared to that of regional array studies, the former has

a much broader spatial coverage. Therefore, a comparison

between the two may assist in extrapolating the array data

laterally. Cross-section C–C1 in Fig. 4 features a compari-

son with the receiver function image of the eastern Tibetan

plateau by Kosarev et al. (1999), who found that on this

traverse the Moho depth reaches to its maximum depth of

80 km near 308N latitude, and becomes shallower toward

the north. Cross-section C–C1 shows a low-velocity

anomaly in the top 100 km depth range beneath most of

the Tibetan plateau. The lower boundary of this slow

anomaly follows closely to the interpreted Moho profile

from the receiver function study. The lower boundary also

reaches its deepest point near 308N and becomes shallower

toward the north.

An alternative explanation for the first-order seismic

structure beneath Tibet (high-velocity zones in the east,

west, and south and a slow-velocity zone underneath

north-central Tibet) is that it reflects temperature vari-

ations in the Tibetan upper mantle associated with small-

scale mantle convection (Molnar, 1990; Woodward and

Molnar, 1995; McNamara et al., 1997; Pandey et al.,

1991). As expressed by Molnar (1988) north-central Tibet

may reflect a region where previously thickened mantle

lithosphere is being convectively thinned and is flanked on

the east and west by downwelling mantle lithosphere.

Since subduction of the Indian plate is occurring, mantle

convection beneath parts of Tibet must certainly be


occurring also. Nevertheless, the geometry and position of

the Indian slab places limits on where convection may be

occurring. More conservative interpretations than ours

indicate that the Indian slab shallowly dips beneath Tibet

w350 km north of the Indus-Yalu suture zone (Owens and

Zandt, 1997; Kosarev et al., 1999). Although, as

mentioned earlier, these studies were conducted in the

central portions of the Tibetan plateau, extrapolating this

position to the west, places the Indian slab in a region

where downwelling of Tibetan mantle lithosphere is

inferred (Zhou et al., 1996). Another short-coming of the

mantle lithosphere convection model is that it does not

provide an explanation for a young mantle geochemical

signature observed in geothermal spring samples (Hoke et

al., 2000) nor early Miocene high-K volcanic rocks in

southwest Tibet (Miller et al., 1999).

4. Implications

Much of the Tibetan plateau is flat with an average

elevation of 5 km (Fielding et al., 1994), and its crust is

about twice as thick as the crust elsewhere (Jobert et al.,

1985; Molnar, 1988). The seismic images analyzed in this

study require that any theory seeking to explain the uplift

history of the Tibetan plateau include the existence of Indian

lithospheric slab beneath the western and eastern portions of

the Tibetan lithosphere. Less clear in the seismic images is

the presence of low-velocity channel, interpreted as

asthenosphere, sandwiched between the Indian and Tibetan

lithosphere.

Our proposed interpretation of the lithospheric structure

is similar in various aspects to other previously published

models involving underthrusting of Indian lithosphere

beneath Tibet (Argand, 1924; Powell and Conaghan,

1973; DeCelles et al., 2003; Chemenda et al., 2000). We

discuss below testable predictions made by our model

concerning the geologic evolution of the Himalaya and

Tibetan plateau.

4.1. Comparison with crustal shortening estimates

Assuming that a lithospheric-scale detachment separ-

ating Tibetan lithosphere from Indian lithosphere exists

along strike of the orogen (Zhao et al., 1993), our

interpretation that Indian mantle lithosphere has been

underthrust northward beneath Tibet to the Jinsha suture

(Fig. 7) predicts the magnitude of crustal shortening of rocks

within the Himalaya. On the 165–210 and 210–265 km

depth slices (Fig. 2), the Indian lithospheric mantle is the

most coherent high-velocity anomaly that extends north-

wards beneath nearly the entire Tibetan plateau to

approximately 368N in latitude. This latitude is, for the

western Tibet, around its boundary with the Tarim

Basin and, for the eastern Tibet, around its boundary

with the Qaidam basin. Zhu and Helmberger (1998) found

a 15–20 km step change in Moho depth between the thick

Tibetan crust and relative thinner Qaidam basin crust.

The above model places the Indian lithospheric slab

570 km north of the Indus-Yalu suture zone in southwest

Tibet (Fig. 7). DeCelles et al. (1998) calculated w228 km of

horizontal shortening of the Subhimalaya and Lesser

Himalaya, and later refined this estimate with detailed

mapping along the Seti River corridor to 460 km (DeCelles

et al., 2001). A minimum-shortening estimate for the

Dadeldhura thrust and Main Central Thrust is 117 km

(DeCelles et al., 2001). Along the Seti and Karnali river

corridors considerably more shortening is required if both

thrusts reached as far south as the Dadeldhura synform

(DeCelles et al., 2001; Upreti and Le Fort, 1999). Srivastava

and Mitra (1994) estimated between w193 and 260 km on

the Main Central Thrust and Almora thrust (equivalent to

the Dadeldhura thrust) in northern India. Combining the

shortening estimates for the Subhimalaya and Lesser

Himalaya along the Dadeldhura–Baitadadi road transect

(DeCelles et al., 1998) (Fig. 7b), the Main Central Thrust

and Almora thrust (Srivastava and Mitra, 1994), and the

Tethyan fold–thrust belt and Indus-Yalu suture zone

(Murphy and Yin, 2003), yields a total horizontal shortening

estimate across the central Himalaya in western Nepal and

southwest Tibet of 597–664 km. Alternatively, combining

the shortening estimates along the Seti River corridor

(DeCelles et al., 2001) with those in the Tethyan fold–thrust

belt and Indus-Yalu suture zone yields a total shortening

estimate of 763 km, which is clearly a minimum estimate

since internal deformation of the metamorphic rocks in the

High Himalaya is not accounted for. This amount of crustal

shortening is sufficient to explain Indian mantle lithosphere

to latitude of the Jinsha suture separating the Qiangtang

terrane from the Songpan Ganzi–Hoh Xil terrane (Yin and

Harrison, 2000) (Fig. 7). A similar conclusion was reached

by DeCelles et al. (2003) who used shortening estimates

from several parts of the Himalaya to suggest that portions

of Indian lithosphere have been underthrust beneath Tibet to

368N latitude. What is different about the two models is in

the interpretation of the lithospheric structure beneath north-

central Tibet. In the model by DeCelles et al. (2003) the

Indian mantle lithosphere has been delaminated leaving

behind lower crust that was previously beneath the Greater

Himalaya crystalline sequence. In our model, the Indian

mantle lithosphere has subducted at steeper angle in north-

central Tibet, and is separated from the shallowly subducted

slab to the west and east by lithospheric tears.

4.2. Late Cenozoic mantle-derived magmatism

in southern Tibet

The presence of a thin channel of asthenosphere between

underthrusted Indian mantle lithosphere and Tibetan litho-

sphere provides a mechanism to generate mantle-derived

melts beneath the Tibetan plateau. Our interpreted seismic

structure predicts that a young mantle geochemical

Fig. 7. (a) Tectonic map of the Tibet–Himalaya collision zone modified from Yin and Harrison (2000). (b) Schematic cross-section (A–A 0) across Tibet–

Himalayan orogen showing tectonic interpretation based on surface geology and tomography results from this study. Indian mantle lithosphere beneath central

Tibet is interpreted to be underthrust at a steeper angle than in the western and eastern portions. AKMS, Ayimaqin-Kunlun Mutztagh suture; BNS, Bangong-

Nujiang suture; IYS, Indus-Yalu suture; JS, Jinsha suture; MFT, Main Frontal thrust; MBT, Main Boundary thrust; MCT, Main Central thrust; STD, South

Tibet detachment; GCT, Great Counter thrust; ATF, Altyn Tagh fault; LM, lithospheric mantle.


signature exists in along the entire southern portion of the

Tibetan plateau, which differs from other models that

restrict this signature to northcentral Tibet (Molnar et al.,

1993). Two studies have essentially tested this prediction.

Miller et al. (1999) conducted a major and trace element

study of volcanic rocks in southwest Tibet. Their results

indicate that mantle-derived high-K, calc-alkaline magma-

tism did not end until 17 Ma. A recent 3He study (Hoke et

al., 2000) of geothermal spring samples from the Tibetan

plateau, which included samples from southwest Tibet,

argues for degassing of volatiles from young mantle-derived

melts intruded into the crust. Their results indicate the

presence of mantle-derived helium exists along the entire

southern margin of the Tibetan plateau adjacent to the

Indus-Yalu suture zone. Combining the results from both

these studies indicate that conditions appropriate for melting

of mantle lithosphere persisted in southern Tibet since the

Miocene.

4.3. Thermal effect on buoyancy of Indian lithosphere

The coherent high-velocity Indian slab anomaly exists

beneath most of the Indian subcontinent and nearly the

entire Tibetan plateau. This indicates that the Indian

lithosphere is strong due to its great thickness, which is

more than 200 km at its core beneath northern India. In

contrast, the Tibetan lithosphere has a very thick and low

velocity crust and a thin lithospheric mantle. Although the

thickness and velocity of the Tibetan crust may have been

altered during the post-collision period, the Tibetan slab was


most likely warmer than the Indian slab prior to their

collision given that it has been a locus of deformation since

the Mesozoic (Yin and Harrison, 2000). The Indian

lithospheric slab is twice as thick as that of an oceanic

slab, but not as cold as the latter. The seismic velocity of

oceanic slabs is 3–5% higher than the ambient region at

shallow depths (Zhou and Clayton, 1990; Van der Hilst et

al., 1991), while the Indian slab is only 2–3% higher near

surface. In fact, as shown in cross-sections B–B1 and D–D1

in Fig. 4, the slab appears losing its amplitude of high-

velocity as it subducts to the north, suggesting the slab may

be heating up. This implies that, though the cold continental

Indian slab could subduct, its negative buoyancy is far less

than that of oceanic slabs. The Indian slab possibly reached

zero buoyancy at a depth around 210 km, as indicated by the

subhorizontal geometry of the subduction trajectory (Figs. 4

and 5). Further convergence of the two plates is interpreted

to have resulted in underthrusting-type horizontal subduc-

tion. As the subducted or underthrusted slab is heated up, it

is predicted to change to positive buoyancy that will

effectively lift the overlying Tibetan lithosphere and

asthenosphere. Heating of the Indian slab is certainly

enhanced by the slow convergence rate of w 20 mm/yr

between the two plates (Bilham et al., 1997). Hence, we

hypothesize that the support to the Tibetan plateau to

maintain its high elevation is most likely driven, at least

partially, by the buoyancy of the heated subducted Indian

slab. The much more yielding asthenospheric layer at 120–

165 km depth may provide a cushion to help maintain the

flatness of the plateau, similar to the mechanism suggested

by models involving hydraulic uplift by injection of Indian

continental crust into a fluid-like Tibetan lower crust (Zhao

and Morgan, 1987) or crustal thickening by lower crustal

flow (Royden et al., 1997).

We explore this hypothesis using a simple thermal model

by placing a slab of constant initial temperature into a

mantle of constant temperature. With time the temperature

Fig. 8. Time required for a lithospheric slab to lose half of its initial

temperature difference from that of ambient mantle, assuming thermal

diffusivity kZ1 mm2 sK1.

of the slab will approach to that of the mantle. Fig. 8 shows,

for different slab thickness, the time required for the slab to

lose one-half of its initial temperature difference from the

mantle. Since the Indian slab is a continental plate, its

buoyancy with respect to the mantle should turn from

negative to positive after losing a portion of its temperature

difference with respect to the mantle. Although we do not

know how much the portion is for the Indian slab, Fig. 8

predicts the trend. For instance, we estimate it took 36 Ma

for a 140-km-thick slab to lose half of its initial temperature

difference. This estimated time for the slab to switch from

negative to positive buoyancy, and therefore assist in

supporting the Tibetan lithosphere, is reasonable with

previous estimates on the timing of plateau uplift (Harrison

et al., 1995; Garzione et al., 2000).

4.4. The low-velocity anomaly beneath central Tibet

An intriguing feature in the tomographic model is a large

low-velocity anomaly north of the Indus-Yalu suture zone

and between 85 and 938E beneath the Tibetan plateau. This

anomaly can be traced from the crust down to at least 310 km

depth, though at depths of 75–120 and 165–260 km (Fig. 2) it

is a relatively slow area within the fast slab anomalies. This

implies that, though mantle upwelling might be occurring in

this area, the slab might just be weakened with fractured or

faulted zones. We suggest that the origin of the weakened

zone of the slab may be due to segmentation or fingering of

the slab during subduction (Fig. 7b). Alternatively, it may

have been weakened by mantle convection (Houseman and

Molnar, 2001) or slab break-off of Greater Indian lithosphere

(DeCelles et al., 2003). In response to the stresses caused by

subduction and/or uplifting of the heated Indian slab, the

fractured zone in the slab and over-pressure of the astheno-

spheric layer could produce seismic anisotropy observed

over this area (McNamara et al., 1994). It is fortunate that

previous geophysical and geochemical studies conducted

within this anomalous area have the opportunity to observe

the signature of the deep mantle. However, it is unfortunate

that, if this anomaly truly exists, we cannot generalize the

lithospheric structure beneath the Tibetan plateau based on

findings in this area alone.

4.5. Inference for the cause of a destructive intraplate

earthquake

The core of the Indian slab is a coherent high-velocity

anomaly as thick as 200 km, such as that seen beneath

northern India in cross-section B–B1 in Fig. 4. The coherent

high-velocity anomaly of the slab and its apparent bending

far south of the Himalayan thrust front indicates that the slab

is quite strong and rigid. An implication of this is that stress

generated by the resistance to the subduction due to

buoyancy of a heated slab can be carried along the slab.

The Mw7.5 destructive earthquake that occurred on

26/01/01 near Jamnagar, India, is located at the thinnest


point of the high-velocity slab layer (Fig. 4, A–A1). The

hypocenter of this intraplate earthquake is located at the

position where the slab just starts to thicken northwards

toward the subduction zone. The thinning of the slab at this

location is partially due to the low-velocity anomaly that

was interpreted as the signature of the Deccan plume

(Kenneth and Widiyantoro, 1999). According to the USGS

rapid moment tensor solution, the first focal plane of the

double-couple solutions (2928 strike, 368 dip, 1368 slip) of

this thrust event faces the direction of N228E, which is

parallel with cross-section A–A1. Thus, we propose that this

destructive earthquake is due to a failure of the Indian

lithosphere at its weakest point in response to the stress

originated from its subduction beneath Tibetan plateau.

5. Conclusions

Establishing the 3-D deep structure of the crust and

mantle is key to understanding the formation of the Tibetan

plateau. Our analysis of a high-resolution global tomo-

graphic model indicates the existence of a high-velocity

subducted Indian lithospheric slab beneath nearly the entire

Tibetan plateau. These results, coupled with existing field-

based crustal shortening estimates suggest that uplift of the

Tibetan plateau was assisted by subhorizontal subduction of

Indian lithosphere. In addition, a low-velocity astheno-

spheric layer might exist between the Tibetan lithosphere

and the subducted Indian slab, though this notion is highly

interpretive. The heating of the subducted continental slab

and the presence of the overlying asthenospheric layer likely

play a major role in the development and maintenance of the

high elevation and flatness of the Tibetan plateau. Our

interpretation is that, the uplift history and low relief of the

Tibetan plateau is a result of subhorizontal subduction and

heating of Indian lithosphere that is separated from Tibetan

lithosphere by a thin channel of asthenosphere. Two

predictions made by our model are: (1) the amount of

shortening in the Himalayas is equivalent to the amount of

underthrusted Indian mantle lithosphere; and (2) a young

mantle geochemical signature should be present along the

entire southern portion of the Tibetan plateau. The Tibetan

crust is not completed cut off from mantle flows beneath the

subducted slab. There is a large low-velocity anomaly north

of the Indus-Yalu suture zone between 85 and 938E that

extends from the crust all the way down to at least 310 km

depth beneath the plateau. This low-velocity anomaly is

indicative of mantle upwelling through a weakened zone of

the subducted slab. Interestingly, many previous studies,

including recent seismic array experiments, were conducted

within this anomalous region. At its strongest core west of

858E, the Indian slab begins its downward bending hundreds

of kilometers south of the Himalayas. Thus, the Indian slab

is probably quite rigid and the stress caused by the resistance

to the subduction could be carried along the Indian slab.

This scenario suggests that the destructive earthquake that

occurred on 26/01/01, a thrust event located at the weakest

point of the Indian slab, is likely a direct consequence of

subduction of the Indian slab beneath Tibetan plateau.

Acknowledgements

The authors thank S.A. Hall, K. Burke, K. Cooper, and J.

D’Andrea-Kapp for helpful discussions.

References

Argand, E., 1924. La tectonique de l’ Asie. Proc. 13th Int. Geol. Cong. 7,

171–372.

Bijwaard, H., Spakman, W., Engdahl, E.R., 1998. Closing the gap between

regional and global travel time tomography. J. Geophys. Res. 103,

30055–30078.

Bilham, R., Larson, K., Freymueller, J., Project Idylhim members, 1997.

GPS measurements of the present-day convergence across the Nepal

Himalaya. Nature 386, 61–64.

Chang, C., Chen, N., Coward, M.P., Deng, W., Dewey, J.F., Gansser, A.,

Harris, N.B.W., Jin, C., Kidd, W.S.F., Leeder, M.R., Li, H., Lin, J., Liu, C.,

Mei, H., Molnar, P., Pan, Y., Pan, Y., Pearce, J.A., Shackleton, R.M.,

Smith, A.B., Sun, Y., Ward, M., Watts, D.R., Xu, J., Xu, R., Yin, J.,

Zhang, Y., 1986. Preliminary conclusions of the Royal Society and

Academia Sinica 1985 Geotraverse of Tibet. Nature 323, 501–507.

Chemenda, A., Burg, J.-P., Mattauer, M., 2000. Evolutionary model of the

Himalaya–Tibet system; geopoem based on new modelling, geological

and geophysical data. Earth Planet. Sci. Lett. 174, 397–409.

DeCelles, P.G., Gehrels, G.E., Quade, J., Ojha, T.P., Kapp, P., Upreti, B.N.,

1998. Neogene foreland basin deposits, erosional unroofing, and

kinematic history of the Himalayan fold-thrust belt, western Nepal.

Geol. Soc. Am. Bull. 110, 2–21.

DeCelles, P.G., Robinson, D.M., Quade, J., Copeland, P., Upreti, B.N.,

2001. Stratigraphy, structure, and tectonic evolution of the Himalayan

fold–thrust belt in western Nepal. Tectonics 20, 487–509.

DeCelles, P.G., Robinson, D.M., Zandt, G., 2003. Implications of

shortening in the Himalayan fold–thrust belt for uplift of the Tibetan

plateau. Tectonics 21 doi:10.1029/2001TC001322.

Dewey, J.F., Burke, K., 1973. Tibetan, Variscan and Precambrian basement

reactivation: products of continental collision. J. Geol. 81, 683–692.

Dewey, J.F., Shackleton, R.M., Chang, C., Sun, Y., 1988. The tectonic

evolution of the Tibetan Plateau. Philos. Trans. R. Soc. London Ser., A

327, 379–413.

England, P., Houseman, G., 1989. Extension during continental conver-

gence, with application to the Tibetan Plateau. J. Geophys. Res. 94,

17561–17579.

Fielding, E., Isacks, B., Barazangi, M., Duncan, C., 1994. How flat is

Tibet?. Geology 22, 163–167.

Garzione, C.N., Quade, J., DeCelles, P.G., English, N.B., 2000. Predicting

paleoelevation of Tibet and the Himalaya from delta (super 18) O vs.

altitude gradients in meteoric water across the Nepal Himalaya. Earth

Planet. Sci. Lett. 183, 215–229.

Harrison, T.M., Copeland, P., Kidd, W.S.F., Lovera, O.M., 1995.

Activation of the Nyainqentanghla shear zone: implications for uplift

of the southern Tibetan Plateau. Tectonics 14, 658–676.

Hoke, L., Lamb, S., Hilton, D.R., Poreda, R.J., 2000. Southern limit of

mantle-derived geothermal helium emissions in Tibet: implications for

lithospheric structure. Earth Planet. Sci. Lett. 180, 297–308.

Houseman, G., Molnar, P., 2001. Mechanisms of lithospheric rejuvenation

associated with continental orogeny, in: Miller, J.A., Holdsworth, R.E.,

Buick, I.S., Hand, M. (Eds.), Continental Reactivation and Reworking

Geol. Soc. Lond. Spec. Publ., 184, pp. 13–38.


Jobert, N., Journet, B., Jobert, G., Hirn, A., Zhong, S.K., 1985. Deep

structure of southern Tibet inferred from the dispersion of Rayleigh

waves through a long-period seismic network. Nature 313, 386–388.

Kenneth, B.L.N., Widiyantoro, S., 1999. A low seismic wavespeed

anomaly beneath northerwestern India: a seismic signature of the

Deccan plume?. Earth Planet. Sci. Lett. 165, 145–155.

Kind, R., Ni, J.F., Zhao, W., Wu, J., Yuan, X., Zhao, L., Sandvol, E.A.,

Reese, C., Nabelek, J., Hearn, T., 1996. Evidence from earthquake data

for a partially molten crustal layer in southern Tibet. Science 274,

1692–1696.

Kosarev, G., Kind, R., Sobolev, S.V., Yuan, X., Hanka, W., Oreshin, S.,

1999. Seismic evidence for a detached Indian lithospheric mantle

beneath Tibet. Science 283, 1306–1309.

Makovsky, Y., Klemperer, S.L., 1999. Measuring the seismic properties of

Tibetan bright spot: evidence for free aqueous fluids in the Tibetan

middle crust. J. Geophys. Res. 104, 10795–10825.

McNamara, D.E., Owens, T.J., Silver, P.G., Wu, F.T., 1994. Shear wave

anisotropy beneath the Tibetan Plateau. J. Geophys. Res. 99, 13655–

13665.

McNamara, D.E., Walter, W.R., Owens, T.J., Ammon, C.J., 1997. Upper

mantle velocity structure beneath the Tibetan Plateau from Pn travel

time tomography. J. Geophys. Res. 102, 493–505.

Meyer, B., Tapponnier, P., Bourjot, L., Metivier, F., Gaudemer, Y.,

Peltzer, G., Guo, S., Chen, Z., 1998. Crustal thickening in Gansu-

Qunghai, lithospheric mantle subduction, and oblique, strike–slip

controlled growth of the Tibet plateau. Geophys. J. Int. 135, 1–47.

Miller, C., Schuster, R., Klotzli, U., Frank, W., Purtscheller, F., 1999. Post-

collisional potassic and ultrapotassic magmatism in SW Tibet:

geochemical and Sr–Nd–Pb–O isotopic constraints for mantle source

characteristics and petrogenesis. J. Petrol. 40, 1399–1424.

Molnar, P., 1988. A review of geophysical constraints on the deep structure

of the Tibetan Plateau, the Himalaya, and the Karakorum and their

tectonic implications. Phil. Trans. R. Soc. Lond. Ser. A 326, 33–88.

Molnar, P., 1990. S-wave residuals from earthquakes in the Tibetan region

and lateral variations in the upper mantle. Earth Planet. Sci. Lett. 101,

68–77.

Molnar, P., Tapponnier, P., 1975. Cenozoic tectonics of Asia: effects of a

continental collision. Science 189, 419–426.

Molnar, P., England, P., Martinod, J., 1993. Mantle dynamics, the uplift of

the Tibetan Plateau, and the Indian monsoon. Rev. Geophys. 31, 357–

396.

Murphy, M.A., Yin, A., 2003. Structural evolution and sequence of

thrusting in the Tethyan fold–thrust belt and Indus-Yalu suture zone,

southwest Tibet. Geol. Soc. Am. Bull. 115, 21–34.

Nelson, K.D., Zhao, W., Brown, L.D., Kuo, J., Che, J., Liu, X.,

Klemperer, S.L., Makovsky, Y., Meissner, R., Mechie, J., Kind, R.,

Wenzel, F., Ni, J., Nabelek, J., Chen, L., Tan, H., Wei, W., Jones, A.G.,

Booker, J., Unsworth, M., Kidd, W.S.F., Hauck, M., Alsdorf, D.,

Ross, A., Cogan, M., Wu, C., Sandvol, E.A., Edwards, M.A., 1996.

Partially molten middle crust beneath southern Tibet: synthesis of

project INDEPTH results. Science 274, 1684–1688.

Ni, J., Barazangi, M., 1983. High-frequency seismic wave propagation

beneath the Indian Shield, Himalayan arc, Tibetan Plateau, and

surrounding regions: high uppermost mantle velocities and efficient

Sn propagation beneath Tibet. Geophys. J. R. Astron. Soc. 72, 665–689.

Owens, T.J., Zandt, G., 1997. Implications of crustal property variations for

models of Tibetan plateau evolution. Nature 387, 37–43.

Pandey, M.R., Roecker, S.W., Molnar, P., 1991. P-wave residuals at

stations in Nepal: evidence for a high velocity region beneath the

Karakorum. Geophys. Res. Lett. 18, 1909–1912.

Patriat, P., Achache, J., 1984. India–Eurasia collision chronology has

implications for crustal shortening and driving mechanism of plates.

Nature 311, 615–621.

Powell, C.M., 1986. Curvature of the Himalayan arc related to Miocene

normal faults in southern Tibet. Geology 14, 358–359.

Powell, C.M., Conaghan, P.G., 1973. Plate tectonics and the Himalayas.

Earth Planet. Sci. Lett. 20, 1–12.

Roecker, S.W., 1982. Velocity structure of the Pamir–Hindu Kush region:

possible evidence of subducted crust. J. Geophys. Res. 87, 945–959.

Royden, L.H., Burchfiel, B.C., King, R.W., Wang, E., Chen, Z., 1997.

Surface deformation and lower crustal flow in eastern Tibet. Science

276, 788–790.

Srivastava, P., Mitra, G., 1994. Thrust geometries and deep structure of the

outer and lesser Himalaya, Kumaon and Garhwal (India): implications

for evolution of the Himalayan fold-and-thrust belt. Tectonics 13, 89–

109.

Turner, S., Arnaud, N., Liu, J., Rogers, N., Hawkesworth, C., Harris, N.,

Kelley, S., VanCalstren, P., Deng, W., 1996. Post-collision, shoshonitic

volcanism on the Tibetan plateau: implications for convective thinning

of the lithosphere and the source of oceanic island basalts. J. Petrol. 37,

45–71.

Upreti, B.N., Le Fort, P., 1999. Lesser Himalayan crystalline nappes of

Nepal: problems of their origin, in: MacFarlane, A., Sorkhabi, R.B.,

Quade, J. (Eds.), Himalaya and Tibet: Mountain Roots to Mountain

Tops Geol. Soc. Am. Spec. Pap., 328, pp. 225–238.

Van der Hilst, R., Engdahl, R., Spakman, W., Nolet, G., 1991. Tomographic

imaging of subducted lithosphere below northwest Pacific island arcs.

Nature 353, 733–739.

Van der Voo, R., Spakman, W., Bijwaard, H., 1999. Tethyan subducted

slabs under India. Earth Planet. Sci. Lett. 171, 7–20.

Villasenor, A., Ritzwoller, M.H., Levshin, A.L., Barmin, M.P.,

Engdahl, E.R., Spakman, W., Trampert, J., 2001. Shear velocity

structure of central Eurasia from inversion of surface wave velocities.

Phys. Earth Planet. Interiors 123, 169–184.

Woodward, R.L., Molnar, P., 1995. Lateral heterogeneity in the upper

mantle and SS–S traveltime intervals for SS rays reflected from the

Tibetan Plateau and its surroundings. Earth Planet. Sci. Lett. 135, 139–

148.

Yin, A., Harrison, T.M., 2000. Geologic evolution of the Himalayan–

Tibetan orogen. Annu. Rev. Earth Planet. Sci. 28, 211–280.

Yuan, X., Ni, J., Kind, R., Mechie, J., Sandvol, E., 1997. Lithospheric and

upper mantle structure of southern Tibet from a seismological passive

source experiment. J. Geophys. Res. 102, 27491–27500.

Zhao, W.L., Morgan, W.J., 1987. Injection of Indian crust into Tibetan

lower crust: a two-dimensional finite element model study. Tectonics 6,

489–504.

Zhao, L.-S., Xie, X., 1993. Lateral variations in compressional velocities

beneath the Tibetan Plateau from Pn traveltime tomography. Geophys.

J. Int. 115, 1070–1084.

Zhao, W., Nelson, K.D., Che, J., Borwn, L.D., Xu, Z., Kuo, J.T., 1993.

Deep seismic reflection evidence for continental underthrusting beneath

southern Tibet. Nature 366, 557–559.

Zhou, H., 1988. How well can we resolve the deep seismic slab with

seismic tomography?. Geophys. Res. Lett. 15, 1425–1428.

Zhou, H., 1996. A high-resolution P-wave model for top 1200 km of the

mantle. J. Geophys. Res. 101, 27791–27810.

Zhou, H., Clayton, R.W., 1990. P and S wave travel-time inversions for

subducted slab under the island arcs of the northwest Pacific.

J. Geophys. Res. 95, 6829–6852.

Zhou, R., Grand, S.P., Tajima, F., Ding, X.-Y., 1996. High velocity zone

beneath the southern Tibetan plateau from P-wave differential travel-

time data. Geophys. Res. Lett. 23, 25–28.

Zhu, L., Helmberger, D.V., 1998. Moho offset across the northern margin of

the Tibetan Plateau. Science 281, 1170–1172.

Documents

Tomographic evidence for wholesale underthrusting of India ...easd.geosc.uh.edu/murphy/pdf/2005-zhou_murphy_jaes.pdfthick crust of Tibet from the surface down to 75 km depth. In the