Seismological evidence for shallow crustal melt beneath the Garhwal High Himalaya, India: implications for the Himalayan channel flow

Geophys. J. Int. (2009) 177, 1111–1120 doi: 10.1111/j.1365-246X.2009.04112.x

GJI

Sei

smol

ogy

Seismological evidence for shallow crustal melt beneath the GarhwalHigh Himalaya, India: implications for the Himalayan channel flow

Ashish,1 Amit Padhi,2 S. S. Rai1,3 and Sandeep Gupta1

1National Geophysical Research Institute, Hyderabad-500007, India. E-mail: [email protected] Institute of Technology, Kharagpur, West Bengal, India3Indian Institute of Science Education and Research, Kolkata, India

Accepted 2009 January 1. Received 2009 January 1; in original form 2008 August 25

S U M M A R YWe investigate the seismic attenuation along a 160 km profile from the Lower to the HighHimalaya in the Garhwal region, India, through the analysis of Lg waveforms from re-gional earthquakes recorded on 18 broadband seismographs with interstation spacing of 7–10 km. Lateral variability in attenuation is derived through the inversion of 36 two-stationQ0 (Lg Q at ∼1 Hz) measurements using a global optimization scheme. We observe a con-trasting attenuation property in the two Himalayan belts: the Lower Himalaya has Q0 of742 ± 235 similar to those observed in the Indian shield, whereas the High Himalaya ischaracterized by an unusually low Q0 value of 30–60. The seismic attenuation is also well cor-related with the P-wave teleseismic traveltime residual pattern: faster arrival (approximately0.2 s) at stations in the Lower Himalaya as compared to azimuthally independent time delayof ∼0.75 s for the High Himalayan stations. The high attenuation and low velocity in the HighHimalaya suggests that low viscosity and partial melt in the crust can be correlated to thepresence of Miocene leucogranite plutons in this Himalayan belt, a magmatic product of theIndo-Asian collision and presumably evidence of a partial melting event. This shallow lowviscosity channel possibly connects the mid-crustal channel beneath Tibet.

Key words: Inverse theory; Body waves; Seismic attenuation; Crustal structure; Asia.

1 I N T RO D U C T I O N

The Himalaya occupies a unique place among the world’s mountainbelts and has resulted from a long evolution, beginning about 50 Mawith the closure of the Tethys Ocean and collision of the Asian andthe Indian land masses (Basse et al. 1984; Patriat & Achache 1984).The geodynamic complexities of the Himalaya are manifested inseveral surface features such as the Southern Tibetan Detachment(STD), the Main Central Thrust (MCT), the Main Boundary Thrust(MBT) and the Main Frontal Thrust (MFT). The region between theMBT and the MCT is widely referred to as the Lower Himalaya,between the MCT and STD as the High Himalaya and further northas the Tethys Himalaya. The Main Central Thrust in the GarhwalHimalaya (Fig. 1) is a zone bounded by the Munsiari Thrust (MT)in the south and the Vaikrita Thrust (VT) in the north. These twothrust systems are also referred to as MCT1 and MCT2. The de-tailed geology of the Himalaya has been reviewed in Yin (2006).One of the most important magmatic products that post-dates theIndo-Asian collision is the High Himalayan Leucogranites (HHL)emplaced around 25–18 Ma at 400 ± 50 MPa pressure correspond-ing to depth of 8–15 km (Le Fort et al. 1987). Understanding thephysical property of these granites is consequential to models ofthe evolution of the continental crust as its composition may cor-respond to those of pure melts generated in the lower crust. Using

the petrologic constraints and the resistivity measurements on theHHL along with the geophysical measurements in southern Tibet,Gaillard et al. (2004) suggested that the Tibetan bright spot im-ages are the reflection of present day pluton assembly analogous tothe growth of leucogranites in the High Himalaya during Miocene.The Gangotri/Badrinath granite is one of the largest bodies of HHLbelt located in the Garhwal Himalaya (77◦–81◦E), 180 km west ofNepal border. Though numerous geological and geochemical in-vestigations (Scaillet et al. 1995) have been carried out on thesegranites, no geophysical measurements are available.

The Garhwal Himalaya offers an exceptional opportunity to re-solve the important scientific controversy about the channel flowmodel of the Himalaya–Tibet orogen (Bird 1991; Nelson et al.1996; Beaumont et al. 2004). The channel flow concept was in-troduced to explain the apparent partially molten middle Tibetancrust that is extruded southwards by buoyancy forces acting onthe elevated Tibetan crust. This process, which has been active atleast since the early Miocene, has a surface manifestation in theearly- to mid-Miocene rocks produced from mid-crustal melts andwhich are now found at the surface between the MCT and theSTD. Recent studies also suggest a close link between the extru-sion from channel flow, the rapid surface uplift between the MCTand the STD, and the Indian climate through erosion and isostaticuplift. Answering the aforementioned and other related questions

C© 2009 The Authors 1111Journal compilation C© 2009 RAS

1112 Ashish et al.

Figure 1. Regional tectonic map of the Garhwal Himalaya showing major thrust systems. MFT:, main frontal thrust; MBT, Main Boundary Thrust;MT, Munsiyari Thrust; VT, Vaikrita Thrust; MCT, Main Central Thrust, STD, Southern Tibetan detachment. Location of broadband seismographs op-erated during 2005–2008 is marked as filled triangle (�). Inset shows the location of earthquakes in Indian shield (E) recorded over the broad-bandseismographs (S).

requires a detailed and accurate image of the Himalayan crustalstructure.

During the last three decades, several geophysical imaging ex-periments have been made in the Himalaya, Tibet and Ladakh (re-viewed in Molnar 1988; Klemperer 2006). Most dense seismic net-works have been operated in central and eastern Tibet, where thepresence of low seismic velocity and low resistivity at mid-crustaldepth extending northwards from the Tethys Himalaya to South-ern Tibet has been inferred (Nelson et al. 1996; Owens & Zandt1997; Makovsky & Klemperer 1999; Unsworth et al. 2005). In thewestern Himalaya–Ladakh, a low velocity and low resistivity in themid-crust is observed only to the north of the Indus Zangbo Su-ture (Arora et al. 2007; Oreshin et al. 2008; Caldwell et al. 2009).In addition, Rai et al. (2009) observed a low Q0 (∼70) to thenorth of the Indus Suture and a high Q0 (∼700) beneath the Hi-malaya. In the central and eastern Tibet and the Himalaya, Fan & Lay(2002, 2003) and Xie et al. (2004) inferred high crustal attenuationwith Q0 varying from ∼500 beneath the Himalaya to ∼180 in theTethys zone, ∼90 beneath the Indus suture, and to ∼65 in southernTibet.

Effects of fluid or melt fraction are more dominant in attenua-tion (1/Q) than S-wave velocity. Whereas the wave velocity in thecrust varies by less than 20 per cent, Q varies by a factor of 3 onmajor continents (Mitchell 1995). The study of attenuation is, there-fore, important to investigate the presence of even small aqueousfluid/melt distributions in the crust. We follow here the interstationQ0 computation approach (Xie et al. 2004) to image the attenuationacross the major structural features of the Garhwal Himalaya fromthe MBT to the STD. In addition, we used teleseismic traveltimedata to investigate the relationship between measured attenuationand the velocity of the crust in this region.

2 L g AT T E N UAT I O N

The Lg phase is the most prominent seismic phase observed overcontinental paths at regional to teleseismic distances. It is com-monly thought to be generated by a superposition of higher modesurface waves (Oliver & Ewing 1957; Mitchell 1995) or multiplereflected shear energy in the crust (Gutenberg 1955). The Lg wavetravels with a group velocity of about 3.5 km s−1 and is prominentlyseen on all three components of ground motion recordings. Lateralheterogeneities are significant contributors to the characteristics ofthe Lg signal (Kennett et al. 1985) and consequently Lg carriesinformation about the average crustal shear wave attenuation along

Table 1. List of seismic stations, their location and operation period.

STN Lat. (◦N) Long. (◦E) Alt (m) Operational period

DKL 29.855 78.636 1234 2005/04/17–2008/06/28MRG 29.947 78.748 810 2005/04/18–2007/04/07BNK 30.091 78.938 1405 2005/04/13–2008/06/28SYT 30.172 79.045 1681 2005/04/12–2006/10/22KSL 30.285 79.207 1451 2005/04/17–2006/12/17SRP 30.349 79.267 1492 2005/04/13–2007/02/02NAL 30.412 79.353 1298 2005/04/14–2006/12/12PKH 30.461 79.439 1585 2005/07/07–2008/06/16LGS 30.504 79.486 1448 2006/02/28–2006/10/14HLG 30.519 79.509 1627 2005/04/11–2006/12/16ALI 30.531 79.572 2669 2005/07/06–2006/10/12LTA 30.495 79.717 2261 2005/07/07–2008/06/17TMN 30.594 79.783 2541 2006/05/28–2007/04/04JLM 30.639 79.825 2916 2005/10/04–2008/06/17KSP 30.707 79.876 3115 2006/05/27–2008/06/18NTI 30.777 79.84 3064 2006/05/27–2007/06/29

C© 2009 The Authors, GJI, 177, 1111–1120

Journal compilation C© 2009 RAS

Implications for the Himalayan channel flow 1113

its path. Lg Q is assumed to obey the relation

QLg( f ) = Q0 f η, (1)

where Q0 and η represent the Lg Q at 1 Hz and its power-lawfrequency dependence, respectively. The Q0 value is sensitive to thetectonic environment, and is found to be lower in the tectonicallyactive regions compared to the stable continental interiors. Highattenuation has been associated with regions of high temperatureand partial melt in the crust (Mitchell 1995; Fan & Lay 2002, 2003).

3 E X P E R I M E N T A N D DATA

The seismic waveforms used for our attenuation study were recordedon 18 broadband seismographs during 2005 April –2008 June along

Figure 2. Record section of earthquakes from (a) Indian shield and (b) northern Tibet. Seismograms have been high passed with lower limit at 0.5 Hz. Pnvelocity for these two earthquakes is 8.9 and 7.8 km s−1 corresponding to up dip and down dip direction suggesting that Moho dips at ∼6◦. Well developedLg wave is recorded in seismograms for Indian shield event (a), whereas it is almost absent for the northern Tibet earthquake (b). Similar observation is madefor Sn also.

a NE–SW directed profile (Fig. 1) in the Garhwal Himalaya crossingimportant features such as the MFT, the MBT, the MCT zone andthe STD. This seismic profile is part of a 37 station network oper-ated in two phases by the National Geophysical Research Instituteduring 2005–2008 in the Garhwal Himalaya region (77◦–81◦E) tomap the seismicity, quantify earthquake hazard and study the earthstructure. Both, the seismological record and recent space geode-tic measurements (Bilham et al. 2001), suggest significant seismicpotential for this region, which has been the site of several re-cent, deadly, moderate sized earthquakes, the most prominent beingUttarkashi (1991 Oct 20, 30.75◦N 78.86◦E, M 6.6) and Chamoli(1999 March 29, 30.41◦N 79.42◦E, M 6.8) earthquakes apart froma possible great earthquake in 1803 that caused serious damageto lives in the Gangetic plain. Seismic waveforms were recorded

C© 2009 The Authors, GJI, 177, 1111–1120

Journal compilation C© 2009 RAS

1114 Ashish et al.

Figure 3. Amplitude spectra for an earthquake from (a) Northern Tibet on 2006 January 18 and (b) Indian shield on 2006 March 7 at selected stations. Decayof seismic energy at <1 Hz is much slower for Indian event compared to the northern Tibet event.

continuously at 50 samples s−1 by seismographs with an intersta-tion separation of ∼7–10 km. All the stations are equipped withCMG-3T (120 s period) sensor and REFTEK (RT 130-01) data log-gers with 4 GB swappable hard disk and GPS. Operational detailsof individual stations are presented in Table 1.

We first identified the Lg wave in the regional seismogram byhigh pass filtering the seismogram with corner at 0.5 Hz and check-ing in the time window appropriate for a wave travelling with agroup velocity between 3.0 and 3.6 km s−1 (Ruzaikin et al. 1977;McNamara et al. 1995). The part of the seismogram containing theLg wave was then cut out and a cosine taper of width 0.05 was addedto avoid spectral leakage. The instrument response was removed af-ter FFT. To stabilize the measurement of Q0, we applied a 15 pointsmoothing to the ground displacement spectra (Xie et al. 2004)and computed the spectral ratio in the frequency band 0.4–2 Hz.Fig. 2(a) shows the record section of an event located to the south-west at 23.78◦N, 70.90◦E where Lg, Pn and Sn are marked with theirrespective velocities. McNamara et al. (1995) reported that seismicwaveforms crossing northern Tibet do not show Lg. As an examplewe present the record section for an event from northern Tibet at34.53◦N, 87.77◦E (Fig. 2b). Unlike the event from the Indian shield,the northern Tibet event shows no energy content in the Lg timewindow at frequencies less than 1 Hz. This is further substantiatedby the contrasting decay pattern of amplitude spectra plot at se-lected stations for events located to the north-east and south-west(Fig. 3a and b) of the seismic profile. In view of the above, we areconstrained to use only earthquakes in the Indian shield.

4 M E T H O D O L O G Y

To study the lateral variability in the Lg-wave crustal attenuationin the Garhwal Himalaya, we first computed several two-stationQ0values and used them to determine values in individual cellsthrough a global optimization scheme. The interstation Q0 compu-tation assumes that the two stations must be aligned with the sourceand the Earth’s velocity structure is 1-D. Xie et al. (2004) arguedthat to minimize the imperfect source-station alignment the max-imum difference in azimuths of the source to the stations shouldbe less than 15◦. Accordingly, we choose an array of stations andearthquakes that roughly align along a great circle arc (Fig. 1).Earthquake parameters are given in Table 2.

Considering that for Ai( f ) and Aj( f ) representing the amplitudespectra for the stations i and j, we can relate these spectra with the

Table 2. List of events used in the study.

Event Date Origin time Lat. Long. mb

1 2006066(07/03) 182046.1 23.78 70.90 5.52 2006096(06/04) 175916.4 23.32 70.48 5.53 2006174(23/06) 053406.4 23.46 70.12 4.8

attenuation through the following equation (Xie & Mitchell 1990):

(VLg/π�i, j ) ln(R( f )) = f 1−η/Q0, (2)

where R(f ) is the scaled spectral ratio calculated as

R( f ) =(�

1/2i

/�

1/2j

)[Ai ( f )/A j ( f )],

where �i and � j are the epicentral distances, �ij is the interstationdistance and VLg is the typical Lg group velocity. Further, due tolimited availability of earthquake–station pair measurements andconsidering the instability in computing η, we restrict our analysisto Q0 variation only. Accordingly, we used η = 0 in eq. (2) andcomputed the Q0 along with the associated error for individualtwo-station data. The station pairs are chosen to minimize the errorin the Q0 computation and proper spectral separation.

Consider that we have N interstation Q0 measurements and theregion is divided into M cells (model parameters). We can write thefollowing equation relating two-station measurements (Qn) to theQ0 value of the individual cell (Qm):

�n/Qn = ��mn/Qm + εn, n = 1, 2, . . . , N (3)

where �mn is the length of the nth ray in the mth cell, Qm is thequality factor value in the the mth cell and εn is the error termin each measurement. The N equations are now solved to get theindividual cell Q0 values using the least-squares criterion:∑

N

ε2n = min, with the constraints 1 < Qm < 10 000. (4)

We used a robust and fast converging evolutionary algorithmcalled the Differential Evolution (DE) for parameter estimation(Storn & Price 1997). DE is a method of mathematical optimizationof multi-dimensional functions with an evolution strategy optimizerthat is fairly fast and reasonably robust. The crucial idea behind DEis a scheme for generating trial parameter vectors, where a weighteddifference between two population vectors is added to a third vector,which undergoes mutation or random exchange of model parameterwith third vector. It is generally agreed that the number of initial

C© 2009 The Authors, GJI, 177, 1111–1120

Journal compilation C© 2009 RAS

Implications for the Himalayan channel flow 1115

Figure 4. Figures showing the best fit lines drawn to compute the interstation Q0 values for selected source and station pair. The corresponding Q0 and itsstandard deviation are also shown.

population should be 10 times the number of parameters. The trialvector is accepted for the next iteration, if and only if, it yields areduction in the value of the objective function. Iterations are con-tinued until we reach the targeted minima of the objective function.To quantify the number of iterations required for the solution toconverge, we examined the error at different iteration up to 200.We found that after 25 iterations there is very little reduction in theerror function. Mathematical details are available in Storn & Price(1997). We used the MATLAB code for DE available in Price et al.(2005) for data modelling.

Since the data consist of average two-station Q0 together withits upper and lower bounds, we used the Monte Carlo approachto find the error associated with model parameters. First, we findthe individual cell parameters with two-station mean Q0 values.Subsequently, we add random noise in observed Q0 values within±1σ to generate a large number of observations and find the modelparameters corresponding to each set. These modelled Q values foreach parameterized cell are used to compute the average value andthe corresponding associated error.

5 L g AT T E N UAT I O N R E S U LT S

We used waveforms from three regional earthquakes (Table 2)aligned with the seismic profile (Fig. 1) to compute interstationQ0 values. Fig. 4 shows an example of a linear regression fit forQ0 values based on eq. (2) with η = 0. Using the constraints men-tioned we computed 36 two-station Q0 values (Table 3). To assessthe data availability and lateral variability in Q0, we plotted all theinterstation values in Fig. 5. It is clear that the region between theMCT zone and the STD (High Himalaya) has characteristic lowQ0 compared to the region between the MBT and the MCT zone(Lower Himalaya). These two-station Q0 values are used to com-pute laterally varying Q0 with its associated error. To achieve this,we parameterized the 160 km length of the profile into four cellswith constant Q0. Parameterization is based on the study of spectralbehaviour of earthquakes at individual stations. We observed verysimilar behaviour for stations in the Lower Himalaya from DKL

Table 3. Two-station Q0 measurements.

Event Station pair Qo values

1 DKL-PKH 157 ± 5DKL-LTA 88 ± 34DKL-JLM 123 ± 4SYT-LGS 50 ± 2SYT-LTA 56 ± 3SRP-LGS 20 ± 1SRP-LTA 71 ± 3

NAL-LGA 11 ± 1NAL-LTA 37 ± 1NAL-JLM 58 ± 3ALI-LTA 14 ± 1ALI-JLM 25 ± 1

2 DKL-LGS 128 ± 7DKL-HLG 124 ± 6DKL-LTA 121 ± 5MRG-LTA 159 ± 8BNK-HLG 100 ± 6SYT-PKH 70 ± 3SYT-LTA 77 ± 4KSL-PKH 37 ± 2KSL-LGS 39 ± 1KSL-LTA 50 ± 1SRP-LTA 61 ± 3NAL-PKH 12 ± 1NAL-LGS 17 ± 1NAL-LTA 32 ± 1NAL-JLM 45 ± 2ALI-LTA 14 ± 1ALI-JLM 23 ± 1

3 SRP-PKH 38 ± 3SRP-KSP 89 ± 7HLG-KSP 30 ± 1HLG-NTI 20 ± 1ALI-LTA 16 ± 1ALI-TMN 47 ± 6ALI-NTI 66 ± 7

C© 2009 The Authors, GJI, 177, 1111–1120

Journal compilation C© 2009 RAS

1116 Ashish et al.

Figure 5. Plot of all the computed interstation Q0 values used for modelling the attenuation in individual parameterized cell.

Figure 6. Amplitude spectra for a typical earthquake in Indian shield atstations in the Lower Himalaya. The spectra are almost same for thesestations.

to KSL (Fig. 6) and consequently the Lower Himalayan region isconsidered as a single cell (Fig. 7). Further north in the High Hi-malaya, the spectral character is well separated for even shorterpaths (Fig. 5) and hence we divided the region from SRP to NTIinto three cells (Fig. 7) considering the MCT zone as a single celland the region above it into two cells. The inversion results for Q0

are presented in Table 4 and Fig. 8 along with the associated error.The Lower Himalaya in the Garhwal region is characterized by a

Figure 7. Spatial parameterization used to compute Q0 value in individualcells (1, . . . , 4).

high Q0 value (∼742 ± 235) similar to those observed in the west-ern Himalaya (Rai et al. 2009) and the Indian shield. In contrast, theHigh Himalaya has strong attenuation of the Lg wave with Q0 ∼30–60. Interestingly, the MCT zone shows a similar higher attenuationcharacter as the High Himalaya. Low Q0 values (60–90) are well

C© 2009 The Authors, GJI, 177, 1111–1120

Journal compilation C© 2009 RAS

Implications for the Himalayan channel flow 1117

Table 4. Model parameters obtained from in-version.

Block Qm, m = 1,4

1 742 ± 2352 39 ± 23 30 ± 24 62 ± 14

documented in the Tibet and Ladakh region (Fan & Lay 2002, 2003;Xie et al. 2004; Rai et al. 2009). However, low values of 30–60 arepossibly being reported for the first time for any segment of the HighHimalaya. The high attenuation may be caused by several fac-tors such as the presence of fluid bearing material/partial melting(O’Connel & Budiansky 1977; Mitchell 1995) or due to small scalescattering by crustal heterogeneity (Bauomont et al. 1999).

To examine if the high attenuation is also accompanied by lowvelocity, we studied P-wave teleseismic traveltime residual alongthe path. We computed the relative residual with respect to theevent average. Fig. 9 shows the residuals from different azimuthsectors. It is interesting to note that independent of the azimuth ofthe incoming wave, all the stations north of the MCT show positiveresiduals (0.5–1.0 s) whereas those to its south have negative resid-ual (0–0.5 s). The stations in the High Himalaya are at an averageelevation of 2.5–3.0 km, whereas those in the Lower Himalaya are at1.7–2.0 km. This extra 1 km elevation could, at best, explain 0.25s of positive residual observed in the High Himalaya. In addi-tion, it may be noted that the observed positive residual pattern issimilar for earthquakes from two opposite direction SW and NE(Fig. 9). This, along with small interstation separation of7–10 km, suggests that the causative source for the observed timedelay over the High Himalaya is shallow. A few forward modellingtests constrain the depth of the low velocity to a depth of <20 kmfrom the surface. Assuming the average velocity of the crust in thisdepth to be 6 km s−1, the P-velocity in this layer is likely to bereduced by ∼6–10 per cent.

Figure 8. Q0 values for the individual cells obtained from inversion of several two-station pair data. The relative position of stations and the correspondinggeological domain have been shown.

6 AT T E N UAT I O N I N T H E C E N T R A LA N D T H E W E S T E R N H I M A L AYA : AC O M PA R I S O N

Though the Himalayan–Tibetan orogenic belt shows remarkablecontinuity of its geological structure along its 2300 km stretch fromwest to east, there are several ways in which the western Himalayadiffers from the central and the eastern Himalaya: (1) less precipi-tation along the topographic front of the western Himalaya, leadingto less precipitation induced exhumation compared to tectonic ex-humation (Wobus et al. 2003); (2) evidence for distinct steep sub-duction geometry in the western Himalaya at Tso Morari as revealedby presence of high-pressure rocks, which are absent in central andeastern part (de Sigoyer et al. 2004); and (3) marked increase involume and decrease in age of granitoids that are possible signatureof channel flow, from west to east, along the Himalaya.

We compare the Lg attenuation results from the Garhwal Hi-malaya (Table 4) with those from the western Himalaya (Fig. 10)computed by Rai et al. (2009), following the same approach. TheHigh Himalaya segment of the Garhwal Himalaya is characterizedby unusual low Q0 value of 30–60 in contrast with a high Q0 insame segment of the western Himalaya. The low Q0 (∼90) in thewestern Himalaya is observed only to the north of the Indus Suturebeneath Ladakh. This suggests a relationship of tectonics and ge-ology with crustal attenuation. We speculate the presence of a lowviscosity channel in the Garhwal Himalaya continuing beneath theHigh Himalaya, whereas it is restricted to the Indus Suture beneaththe western Himalaya.

7 C O N C LU S I O N

We computed the lateral variation of the 1 Hz Lg Q (Q0) valuealong a profile in the Garhwal Himalaya from the MBT to theSTD covering both the Lower and the High Himalaya. Q is derivedfrom a set of two-station measurements using a global search al-gorithm. We observe contrasting signatures: efficient transmissionof seismic wave in the Lower Himalaya (high Q > 700) simi-lar to the Indian shield (Mitra et al. 2006) and strong attenuation

C© 2009 The Authors, GJI, 177, 1111–1120

Journal compilation C© 2009 RAS

1118 Ashish et al.

Figure 9. Variation of P-wave traveltime residual (with reference to event average) along the Himalayan seismic profile. Station elevation is plotted in (a). (b),(c) and (d) refer to data from all the azimuths, for NE and SW direction, respectively. Note the large time delay over stations in the High Himalaya for eventsfrom all the directions.

(low Q ∼ 30–60) in the High Himalaya. The low Q region is also azone of low P-wave velocity in the shallow crust. Using data fromTibet, we propose that the low velocity, low Q crust in the HighHimalaya is possibly a continuation of a similar structure observedin Tibet. These results are schematically presented in Fig. 11. Rocksbetween the MCT and the STD are widely referred to as the HigherHimalayan Sequence (HHS) and are composed of a mid-crustallayer of high-grade metamorphic rocks and migmatites with sheetsof crustal melts leucogranites. The upper contact of HHS is thesouthern Tibetan Detachment (STD), a low angle normal fault thatdips 5–20◦ beneath Tibet, whereas its lower boundary is the MainCentral Thrust (MCT). Using large-scale structures and pressure–temperature constraints, several authors (Gurjic et al. 2002;Beaumont et al. 2004; Searle et al. 2006) suggested horizontal

transport or flow of 100–200 km of the HHS rocks along the foot-wall of the STD through ductile channel flow during Miocene.Hodges (2006) suggested that at the deeper level of the crust (∼20–30 km) beneath Tibet, where temperatures and pressures are muchhigher, the rocks flow like toothpaste squeezed through a tube andthat the lower crust beneath Tibet is flowing towards the Himalayanmountain front between the MCT and the STD. This low viscos-ity, partially molten lower crust rock with characteristic low ve-locity and high attenuation beneath Tibet forms a continuous linkwith the HHS that also show very similar seismological features asobserved in the Garhwal region. In contrast with the Garhwal Hi-malaya, in the western Himalaya, the high attenuation, low viscositychannel is restricted to the north of the Indus Suture beneath theLadakh.

C© 2009 The Authors, GJI, 177, 1111–1120

Journal compilation C© 2009 RAS

Implications for the Himalayan channel flow 1119

Figure 10. Comparison of Q0 in the Garhwal Himalaya and the westernHimalaya (from Rai et al. 2009). Major fault systems, terranes and suturesinclude MBT, Main Boundary Thrust; MCT, Main Central Thrust; STD,Southern Tibetan detachment; IZS, Indus Zangbo Suture; KF, KarakoramFault; KB, Karakoram batholith.

A C K N OW L E D G M E N T S

This work was supported by a research grant from the erstwhileSeismology Division of the Indian Department of Science & Tech-nology, New Delhi. We greatly appreciate suggestions of ThorneLay and another anonymous reviewer that helped improving themanuscript.

R E F E R E N C E S

Arora, B.R., Unsworth, M.J. & Rawat, G., 2007. Deep resistivity structureof the northwest Indian Himalaya and its tectonic implications, Geophys.Res. Lett., 34, L04307, doi:10.1029/2006GL029165.

Figure 11. Schematic representation of the Himalayan channel flow model and the crustal properties. HHS, High Himalayan Sequence; MHT, Main HimalayanThrust; MFT, Himalayan Frontal Thrust; MBT, Main Boundary Thrust; MCT, Main Central Thrust; STD, Southern Tibetan detachment; IZS, Indus ZangboSuture.

Basse, J., Courtillot, V., Pozzi, J.P., Westphal, M. & Zhou, Y.X., 1984. Pa-leomagnetic estimates of crustal shortening in the Himalayas thrusts andZangbo suture, Nature, 311, 621–626.

Bauomnt, D., Paul, A., Beck, S. & Zandt, G., 1999. Strong crustal hetero-geneity in the Bolivian Altiplano as suggested by attenuation of Lg waves,J. geophys. Res., 104, 20 287–20 305.

Beaumont, C., Jamieson, R.A., Nguyen, M.H. & Medvedev, S., 2004.Crustal channel flows: numerical models with application to the tec-tonics of Himalayan–Tibetan orogen, J. geophys. Res., 109, BB 06406,doi:10.1029/203JB002809.

Bilham, R., Gaur, V.K. & Molnar, P.,2001. Himalayan seismic hazard, Sci-ence, 293, 1442–1444.

Bird, P., 1991. Lateral extrusion of lower crust from under high topography,in isostatic limit, J. geophys. Res., 96, 10 275–10 286.

Caldwell, W.B., Klemperer, S., Rai, S.S. & Lawrence, J.F., 2009. Melts/fluidsin the mid-crust of the northwest Indian Himalaya revealed by Rayleighwave dispersion, Tectonophysics, doi:10.1016/j.tecto.2009.01.013.

de Sigoyer, J., Guillot, S. & Dick, P., 2004. Exhumation of the ultrahighpressure Tso Morari unit in the eastern Ladakh (NW Himalaya): a casestudy, Tectonics, 23, TC3003, doi:10.1029/2002TC001492.

Fan, G.W. & Lay, T., 2002. Characteristics of Lg attenuation in the TibetanPlateau. J. geophys. Res. 107(B10), 2256, doi:10.1029/2001JB000804.

Fan, G.W. & Lay, T., 2003. Strong Lg attenuation in the Northern andEastern Tibetan Plateau measured by a two-station/two-event stackingmethod, Geophys. Res. Lett., 30, 1530, doi:10.1029/2002GL016211.

Gaillard, F., Scaillet, B. & Pichavant, M., 2004. Evidence for present-day leucogranite pluton growth in Tibet, Geology, 32, 801–804,doi:10.1130/G20577.1

Gurjic, D., Hollister, L. & Parrish, R.R., 2002. Himalayan metamorphicsequence as an orogenic channel: insight from Bhutan, Earth planet. Sci.Lett., 198, 177–191.

Gutenberg, B., 1955. Channel waves in the Earth’s crust, Geophysics, 20,283–294.

Hodges, K., 2006. Climate and the evolution of mountains, Sci. Am., 54–61.Kennet, B.L.N., Gregerson, S. Mykkeltveit, S. & Newmark, R., 1985. Map-

ping of crustal heterogeneity in the North Sea via the propagation ofLg-phases. Geophys J. R. astr. Soc., 83, 299–306.

Klemperer, S.L., 2006. Crustal flow in Tibet: geophysical evidence for thephysical state of Tibetan lithosphere, and inferred patterns of active flow,in Channel Flow, Ductile Extrusion and Exhumation in Continental Col-lision Zones, Vol. 268, pp 39–70, eds Law, R.D., Searle, M.P. & Godin,L., Geological Society London, Special Publication.

Le Fort, P., Cuney, M., Deniel, C., France-Lanord, C., Sheppard, S.M.F.,Upreti, B.N. & Vidal, P., 1987. Crustal generation of the Himalayanleucogranites, Tectonophysics, 134, 39–57.

Makovsky, Y. & Klemperer, S.L., 1999. Measuring the seismic properties ofTibetan bright spots: evidence for free aqueous fluids in Tibetan Middlecrust, J. geophys. Res., 104(B5), 10 795–10 825

C© 2009 The Authors, GJI, 177, 1111–1120

Journal compilation C© 2009 RAS

1120 Ashish et al.

McNamara, D.E., Owens, T.J. & Walter, R., 1995. Observation of regionalphase propagation across the Tibetan plateau. J. geophys. Res., 102,22 215–22 229.

Mitchell, B.J., 1995. Anelastic structure and evolution of the continentalcrust and upper mantle from seismic surface wave attenuation, Rev. Geo-phys., 33, 441–462.

Mitra, S., Pristley, K., Gaur, V.K. & Rai, S.S., 2006. Frequency-dependentLg attenuation in the Indian Shield, Bull. seism. Soc. Am., 96, 2449–2456.

Molnar, P., 1988. A review of geophysical constraints on the deep structureof the Tibetan Plateau, the Himalaya and the Karakoram, and their tectonicimplications, Phil. Trans. R. Soc. Lond., 326, 33–38.

Nelson, K.D., Zhao, W. & Brown, L.D., 1996. Partially molten middle crustbeneath southern Tibet: synthesis of project INDEPTH results, Science,274, 1684–1688.

Oliver, J. & Ewing, M., 1957. Higher mode surface waves and their bearingon the structure of the Earth’s mantle, Bull. seism. Soc. Am., 47, 187–204.

Oreshin, S., Kiselev, S., Vinnik, L., Prakasam, K.S., Rai, S.S., Makeyeva, L.& Savvin, Y., 2008. Crust and mantle beneath western Himalaya, Ladakhand western Tibet from integrated seismic data, Earth planet. Sci. Lett.,271, 65–87, doi:10.1016/j.epsl. 2008.03.048

O’ Connel, R.J. & Budiansky, B., 1977. Viscoelastic properties of fluidsaturated cracked solids, J. geophys. Res., 82, 5719–5735.

Owens, T.J. & Zandt, G., 1997. Implications for crustal property variationsfor models of Tibetan plateau evolution, Nature, 387, 37–43.

Patriat, P. & Achache, J., 1984. India–Asia collision chronology has impli-cations for crustal shortening and driving mechanism of plates, Nature,311, 615–621.

Price, K., Storn, R. & Lampinen, J., 2005. Differential Evolution—A Prac-tical Approach to Global Optimization. Springer, Berlin.

Rai, S.S., Ashish, Padhi, A. & Sharma, P.R., 2009. High crustal seismic

attenuation in Ladakh-Karakoram, Bull. seism. Soc. Am., 99(1), 407–415,doi:10.1785/0120070261.

Ruzaikin, A.I., Nersesov, I.L., Khalturin, V.I. & Molnar, P., 1977. Propaga-tion of Lg and lateral variations of crustal structure in Asia, J. geophys.Res., 82, 307–316.

Scaillet, B., Pecher, A., Rochette, P. & Champenois, M., 1995. The Gan-gotri granite (Garhwal Himalaya) Laccolith emplacement in an extendingcollisional belt, J. geophys. Res., 100, 585–607.

Searle, M.P., Law, R.D. & Jessup, M.J., 2006. Crustal structure, restorationand evolution of Greater Himalaya in Nepal–South Tibet: implication forchannel flow and ductile extrusion of middle crust, 2006, in ChannelFlow, Ductile Extrusion and Exhumation in Continental Collision Zones,Vol. 268, pp 39–70, eds Law, R.D., Searle, M.P. & Godin, L., GeologicalSociety London, Special Publication.

Storn, R. & Price, K., 1997. Differential Evolution—a simple and efficientheuristic for global optimization over continuous spaces, J. Global Optim.,11, 341–359.

Unsworth, M.J., Jones, A.G., Wei, W., Marquis, G., Gokaran, S.G. & Spratt,J.E., 2005. Crustal rheology of the Himalaya and southern Tibet inferredfrom magnetotelluric data, Nature, 438, 78–81.

Wobus, C.W., Hodges, K.V. & Whipple, K.X., 2003. Has focused denudationsustained active thrusting at the Himalayan topographic front?, Geology,31, 861–864.

Xie, J. & Mitchell, B.J., 1990. Attenuation of multiphase surface waves inthe Basin and Range province, part I: Lg and Lg coda, Geophys. J. Int.,102, 121–138.

Xie, J., Gok, R., Ni, J. & Aoki, Y., 2004. Lateral variations of crustal seismicattenuation along the INDEPTH profiles in Tibet from Lg Q inversion, J.geophys. Res., 109, B10308, doi:10.1029/2004JB002988.

Yin, A., 2006. Cenozoic tectonic evolution of the Himalayan orogen asconstrained by along strike variation of structural geometry, exhumationhistory, and foreland sedimentation, Earth-Science Review, 76, 1–131.

C© 2009 The Authors, GJI, 177, 1111–1120

Journal compilation C© 2009 RAS