Imaging the Indian lithosphere beneath the Eastern Himalayan region

Geophys. J. Int. (2011) 187, 631–641 doi: 10.1111/j.1365-246X.2011.05185.x

GJI

Geo

dyna

mic

san

dte

cton

ics

Imaging the Indian lithosphere beneath the Eastern Himalayanregion

E. Uma Devi,1 P. Kumar2 and M. Ravi Kumar2

1Indian National Center for Ocean Information Services (INCOIS), Hyderabad 500090, India2National Geophysical Research Institute (Council of Scientific and Industrial Research), Hyderabad 500007, India. E-mail: [email protected]

Accepted 2011 August 3. Received 2011 August 3; in original form 2009 April 1

S U M M A R YLithospheric thickness is an important parameter to understand the nature of collision andsubduction between the Indian and Asian tectonic plates. In this study, we apply the S receiverfunction technique to data from a network of broad-band stations in the northeast India andEastern Himalayan regions and image the geometry of Indian Plate collision. This analysisreveals clear S-to-p conversions from the Moho and Lithosphere–Asthenosphere boundary(LAB) in the various tectonic units of the study region. The Indian lithosphere is found to beonly 90 km thick beneath the Shillong plateau deepening to 135 km on either side suggestiveof a lithospheric upwarp related to the plateau uplift. The lithosphere thickens northward,with values reaching ∼180 km beneath the Eastern Himalaya. The trend of the LAB northof the foredeep region indicates that the Indian Plate plunges beneath the Eastern Himalaya.The consistent northward-dipping character of the Indian Plate suggests that the Indian Plateis traceable until it gets subducted beneath Tibet just south of Bangong suture zone. Thedeepening of the LAB and its correlation with the topographic elevation is in agreementwith homogeneous thickening of the lithosphere in response to compressive forces due to thecontinental collision of India with Asia.

Key words: Seismicity and tectonics; Body waves; Computational seismology; Continentalmargins: convergent; Dynamics: seismology; Asia.

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

Imaging the lithospheric architecture of any region is regarded as animportant step forward to constrain the geodynamic processes thatshaped the overlying geology. Such processes are manifested in theform of tangibly diverse crustal rocks exposed on the surface withmarkedly contrasting physical and thermal characteristics. Charac-terizing the nature and depth to the Lithosphere–Asthenosphericboundary (LAB) is important to understand the plate kinematicsand mantle dynamics (Jordan 1981). The LAB was originally re-garded as a rheological boundary, with the lithosphere behaving asa mechanically strong elastic solid and the asthenosphere deform-ing plastically as a viscous fluid (Barrell 1914). Observations of lowseismic velocities in the upper mantle below a high-velocity lid havebeen first reported underneath oceanic regions by Gutenberg (1959)and hence the LAB is sometimes referred to as the G-discontinuity.Because this discontinuity is associated with variations in variousphysical properties, several observational means that include petro-logical, geochemical, thermal, electrical conductivity, elastic wavespeed and anisotropy are employed to characterize this layer (e.g.Eaton et al. 2009). Although global presence of LAB is widelyacknowledged, there is still an ongoing debate on the nature andsharpness of the velocity gradient across this boundary, particularlybelow the cratons (Fischer et al. 2010).

So far, most information about the thickness of the seismic litho-sphere comes from low-resolution surface wave observations (e.g.Knopoff 1983; Mitra et al. 2006). Although global tomographicmodels (e.g. Ritzwoller et al. 2002; Replumaz et al. 2004; Priestleyet al. 2006) provide glimpses of low-velocity zones in the subcrustallithosphere, their use in understanding the finer-scale regional tec-tonics is limited, because these images are often laterally smearedover a few hundreds of kilometres with a poor resolution in depth.Over the past few decades, receiver function analysis using P-to-sconverted phases has contributed immensely towards detailed imag-ing of the crust and the upper-mantle discontinuities in differenttectonic regions worldwide (e.g. Langston 1981; Ammon & Zandt1993; Saul et al. 2000; Bostock et al. 2002; Kind et al. 2002).However, detection of the LAB using the P receiver function tech-nique has remained mostly elusive, since arrival times of multiplesfrom the shallow interfaces, especially the Moho, arrive within thesame time window as converted phases from the deeper discontinu-ities, such as the LAB. Recently, the S receiver function technique(Farra & Vinnik 2000), which utilizes the S-to-p converted phasesfrom velocity discontinuities, has emerged as an effective tool toovercome this problem. This is because the converted and multi-ples phases are pre- and post-cursory, respectively, of the incidentdirect S wave. Application of the S receiver function technique tobroad-band data from regions like Hawaii (Li et al. 2004), Tien

C© 2011 Director, National Geophysical Research Institute 631Geophysical Journal International C© 2011 RAS

Geophysical Journal International

632 E.U. Devi, P. Kumar and M.R. Kumar

Shan-Karakoram, northwest Atlantic regions (Oreshin et al. 2002;Kumar et al. 2005a,b), Tibet (Kumar et al. 2006), Indian shield(Kumar et al. 2007), South West Ireland (Landes et al. 2007),Aegean and Dabie Shan (Sodoudi et al. 2006a,b), Eastern Turkey(Angus et al. 2006), Central Europe (Heuer et al. 2007), WesternAmerica (Li et al. 2007), North America (Rychert et al. 2007) andSouth America (Heit et al. 2007) has enabled distinct mapping ofthe LAB and provided new understanding of the mechanisms thatconstrain the dynamics of these regions.

In the recent past, application of the S receiver function tech-nique to the data from INDEPTH experiments (Owens et al. 1993;Nelson et al. 1996; Kind et al. 2002) has revealed the collisionarchitecture of the Indian and Asian continental lithospheres (Ku-mar et al. 2006). The results of Kumar et al. (2006), suggest thatthe base of the Indian lithosphere dips northwards from a depth of150 km beneath the Himalaya to a depth of 210 km just south of theBangong suture zone. Their model of a well-defined, thick litho-sphere throughout the entire Tibetan plateau that resisted the con-vective processes, contradicts the existing models, which principallyinvoke a thin buoyant lithosphere. In this context, it is interestingto trace the Indian lithospheric plate further south and character-ize the variations in its thickness during the various stages of thecollision, starting from regions away from the collision front wherethe lithosphere is expected to be undeformed and shield-like, in theforedeep region just before collision, during the collision in the Hi-malaya and subsequently beneath southern Tibet. Although studiesusing P-to-s converted phases have imaged the northward-dippingnature of the Moho from the Shillong plateau to the Himalayanfoothills (Kumar et al. 2004b; Ramesh et al. 2005), mapping thelithospheric geometry itself, in terms of the geometry of the LAB,has remained elusive on the Indian side of the Himalaya.

In this study, we apply the recently developed S receiver functiontechnique to S and SKS waveforms recorded by a regional broad-band network in northeast India (Fig. 1) to obtain the first estimatesof the lithospheric thickness within this region. This regional net-work being operated by the National Geophysical Research Institutesince the end of 2001 consists of Gurlap CMG-3ESP seismometersequipped with REFTEK data loggers. With the exception of station

Figure 2. Distribution of earthquake events used in this study. S receiverfunctions are computed using events in the epicentral distance range of60◦–85◦ and SKS receiver functions are computed for those between 85 and120◦.

TEZ (Tezpur), which is a permanent station in operation since 1999,the other stations have been functioning intermittently. The data setis supplemented by data from five southernmost stations of the IN-DEPTH experiment (Owens et al. 1993; Nelson et al. 1996; Kindet al. 2002). Based on a 1-D P and S wave velocity model, an esti-mate of the conversion depth of a given S-to-p (or SKS-to-p) phaseand its offset from the recording station can be obtained by tracingthe ray along its backazimuth. However, the uneven distribution ofthe earthquake sources (Fig. 2) prohibits uniform sampling of thestudy region. In the absence of regional velocity models, it is acommon practice to adopt the IASP91 velocity model. The piercing

Figure 1. Topographic map showing the major tectonic units of the northeast India, Eastern Himalaya and Burmese arc regions. Inverted triangles indicatelocations of the broad-band stations on the Indian side and triangles show stations on southern Tibet. Inset map points to the location of the study region.

C© 2011 Director, National Geophysical Research Institute, GJI, 187, 631–641

Geophysical Journal International C© 2011 RAS

Imaging the Indian lithosphere 633

Figure 3. Geometry of the experiment showing the distribution of the piercing points (crosses) at 150 km beneath Himalaya and 100 km beneath India, basedon the approximate arrival of the negative converted phases. The region is divided into non-overlapping bins (shown with numbers) for stacking the S receiverfunctions. Lines AB and CD are the two profiles trending north–south and east–west, respectively along which the seismic sections are analysed.

points of the converted phases (Fig. 3), computed at a depth of 150km in the Himalayan region and 100 km in the Indian region, usingthe appropriate slowness values of the S and SKS phases, sample thediverse tectonic units of the study region and enable us to mapthe spatial variations of the Indian lithosphere as it underthrustsAsia in the north and Burma in the east.

2 S T U DY R E G I O N

The northeast India region is composed of disparate tectonic units,namely, the Eastern Himalaya, its foredeep, Shillong plateau, MikirHills and the Burmese arc region. While the northern portionis marked by a well-known collision environment involving theIndian and Asian plates resulting in the formation of the gigan-tic Himalayan mountains, the eastern margin is characterized bysubduction of the Indian Plate beneath the Burmese Plate, withseismicity extending down to intermediate depths (Ni et al. 1989).Away from the plate margins, the Shillong plateau and the MikirHills stand out as distinct topographic features south of the sedi-ment filled foredeep region that resulted from the shallow under-thrusting along the Himalayan front. These uplifted regions alongwith the ‘foreland spur’ represent the northeastern extension of theIndian shield elements (Nandy 2001) whose basement is comprisedof Archaean gneissic rocks. Various hypotheses have been proposedto explain the Shillong uplift, which include isostatic adjustment(Das Gupta & Biswas 2006), thermal disturbance during Jurassic,which resulted in uplift that is still continuing (Kailasam 1979)and pop-up due to tectonic forces (Rao & Kumar 1997; Bilham &England 2001). Also, two prominent faults to the south of the Shil-long plateau acted as vents to the eruption of Sylhet traps about 116Ma (Ray et al. 2005), as a consequence of the heating of the IndianPlate as it passed over the Kerguelen hotspot (Kumar et al. 2004a).

3 S - R E C E I V E R F U N C T I O N A NA LY S I S

S waveforms from earthquakes in the epicentral distance rangeof 60◦–85◦ (Farber & Muller 1980; Yuan et al. 2006) and SKSwaveforms from earthquakes in the distance range of 85◦–120◦

are used for computation of receiver functions. The usefulness ofchoosing these epicentral distance ranges has been demonstrated byYuan et al. (2006) using synthetic and real data. The data has beenquality checked visually for the S and SKS phases and only thosewaveforms with a good signal-to-noise ratio have been retained forfurther analysis. While the focal depths of most selected events areshallower than 200 km, 18 events have depths between 200 and 250km. Of all the stations, TEZ has recorded the maximum number ofS and SKS waveforms, owing to its longest period of operation. Atotal of 133 SKS waveforms and 160 S waveforms have been usedin this analysis.

The methodology used for computing the S receiver functions issimilar to that adopted for P receiver functions, except that the roleof the respective components is interchanged. First the Z, N and Ecomponents of the waveforms are transformed into the Z (vertical),R (radial) and T (transverse) systems, respectively, using the the-oretical backazimuth. Subsequently, the P and SV components areisolated from the incident wavefield by further rotating the Z and Rcomponents automatically by an angle that minimizes the S-waveenergy (at its theoretical arrival time) on the P component of therotated trace (Kumar et al. 2005a,b; Kumar et al. 2006; Kumar &Kawakatsu 2011). To achieve this, we transform the ZRT compo-nents into LQT components by systematically varying the rotationangles and then measure the absolute amplitude of the S phase onthe Q component at the P onset time. A time window of ±1 s oneither side of the theoretical S/SKS wave (aligned at zero time) isused for estimating the maximum absolute amplitude. Consider-ing this optimal angle as the appropriate angle of incidence, theZRT components are again rotated and deconvolved to obtain the Sreceiver functions. Waveforms whose S phases have incidence an-gles >50◦ (Fig. 4) have been discarded because they are unrealisticand may be due to the presence of noise. Relying on the iterativerotation angle compared to the theoretical incidence angle derivedfrom a standard velocity model such as IASP91 is reasonable in atectonically complex region like the Eastern Himalaya, because thetheoretical incidence angles may deviate much from the actual ones.Fig. 5 showing the theoretical and estimated rotation angles revealsvariations among stations, in addition to backazimuth dependencyat individual stations. The possible causes for the differences in the




Figure 4. Histograms of the theoretical (white bars) and the iterative an-gles of rotation (grey bars) used for optimal isolation of the wavefield intoP-SV -SH components. The theoretical angles are derived based on theIAPS91 model.

theoretical and observed angles of incidence stem mostly fromthe velocity variations in the subsurface and noise content on theseismograms. Theoretical values are based on the velocity modelused (e.g. IASP91), which are constant for a given slowness. How-ever, the local velocity variations and the noise content may bequite different from station to station, and hence, the angle deter-mined from the waveforms may differ from the theoretical one.A potential source of noise in this context could be the interfer-ing teleseismic phases arriving within the same time window asthe S and SKS waves (Wilson et al. 2006). The iterative rotationscheme to minimize the S-wave energy is in effect a means ofreducing the impact of the interfering phases on the computedreceiver functions—suggesting that the rigorous epicentral (anddepth) constraints earlier imposed on the acceptable events for S-wave receiver function analysis (Wilson et al. 2006) can perhaps berelaxed.

Fig. 5 reveals that the iterative angle of incidence at most of thestations ranges from 0 to 35◦ for a majority of the events. This rangeis within the S-wave incidence angle corresponding to the epicentraldistances used in this study. Only two stations viz. Bhairabkunda(BKD) and Rupa (RUPA), show consistently lower iterative anglescompared to the theoretical. This can be reconciled by the presenceof low-velocity sedimentary layer(s) underneath these stations inthe lower Himalaya, which would lead to near vertical incidenceangles. Evidence for such sediments underneath most stations in

Figure 5. Diagrams showing the variation of rotation angles at individual stations with respect to the backazimuths. The blue and red circles denote anglesdetermined theoretically and using the iterative rotation method, respectively.




the sub-Himalayan regions comes from previous studies using Preceiver functions (Kumar et al. 2004b).

After rotation, the SV components are deconvolved from therespective P components to obtain the S receiver functions, whichessentially contain information about the S-to-p converted phases(Sp). To minimize the high-frequency noise, the traces have beentreated with a low-pass filter having a corner frequency of 0.3 Hz,before rotation. These converted phases being P waves, arrive asprecursors to the main S phase, whereas their corresponding freesurface multiples arrive after the S phase, because they essentiallytravel at the same speed. Deconvolution is performed to sourcenormalize the waveforms from different events. The deconvolutionis performed in time domain by finding an inverse time-series of theS phase in such a way that the SV component gets nearly transformedinto a spike. Thereby, it is implicitly assumed that the SV componentof the time-series essentially contains the source signature. Thisinverse series is then convolved with the P component, to obtain theS receiver function. The treatment of SKS waves is similar to thatof the S waves except that the SKS-to-p phases have a much lowerhorizontal slowness compared to the Sp phases. By convention, theamplitude of these Sp phases is negative for a positive velocity jumpdownwards. To make the Sp conversion comparable to the standardP-to-s conversions, we reverse the signs of the amplitudes as wellas the time axis of the receiver function traces with zero time beingthe arrival time of the S phase.

The individual S receiver functions from one of the grids (num-bered 14 in Fig. 3) obtained using the iterative and incidence an-gle approaches are shown in Fig. 6, for quality comparison. The

theoretical incidence angles are obtained using the IASP91 velocitymodel. As seen from the figure, the resulting S receiver functionsfrom both the approaches are consistent, revealing the positive con-version from the Moho at ∼5 s and the negative conversion fromthe LAB at ∼10 s. Moreover, the receiver functions obtained bythe iterative scheme (Fig. 6a) appear more coherent resulting in aless noisy summation trace; the Moho and LAB conversions beingabove the 2 sigma error level, compared to those obtained usingthe theoretical angles (Fig. 6b). Bootstrap stacking of the individualtraces obtained by both the methods (Fig. 7) adds statistical credi-bility to the Moho and LAB conversions obtained by the iterativerotation scheme. As can be seen, the iterative rotation scheme tendsto produce a better stack.

Because the main focus of this study is mapping of the LAB, werefrain from discussing the crustal features. In hindsight, to enablebinning of the individual traces to enhance the converted phasesfrom the LAB, we computed the piercing points for Sp phases ata depth of 100 km on the Indian side where the receiver functionsshow a negative phase near 10 s and 150 km in the Himalayan front,where indications of LAB are seen close to 15 s. Based on thesepiercing points, the entire study region is divided into a number ofnon-overlapping spatial bins for stacking (Fig. 3). We chose the binsin such a way that they sample the same geological unit and havesimilar Sp conversion times from the LAB. Because the distributionof the piercing points is not uniform, the number of traces usedfor stacking is variable. The choice of variable size of the bins hasbecome necessary to include more number of traces within eachbin, for enhancing the signal-to-noise ratio. Because the stacking

Figure 6. Individual S receiver functions for a subset of data from gird number 14 (see Fig. 3) with the summation trace shown in the top panel. S receiverfunctions computed using (a) the present methodology of minimizing SV energy at zero time on the P component and (b) the theoretical angle of rotation areshown. Both the receiver functions especially the stack traces show very good agreement. Traces on either side of the summation trace denote 2 sigma errorlimits.




Figure 7. Bootstrap stacks (grey wiggles) of the traces in grid 14 with themean (black line) and 2 sigma standard errors bounded by two grey linesrunning almost parallel to the zero line.

has been performed at a depth of 100 km, the phases around thatdepth will be coherently stacked.

4 R E S U LT S

To demonstrate the robustness of the results using the S receiverfunction technique, we first present the receiver functions at stationTEZ (Fig. 8) and then show them along a N–S profile spanning thestudy region from south of the Shillong plateau up to the Himalayancollision front (Fig. 10). The individual S receiver functions forstation TEZ moveout corrected for the Sp conversion phase andaveraged in narrow slowness bins of 0.2 s deg–1 with an overlap of0.3 s deg–1, are shown in Fig. 8(b). The receiver functions of the Sand SKS phases appear distinctly separated in view of the differencein their slowness values. The positive conversion from the Moho(red) is clearly evident close to 5 s followed by a negative conversionclose to 10 s, which we interpret as the Sp conversion from the LAB.For comparison, we show an image of the P receiver functionsat the same station moveout corrected for P-to-s conversions andsorted by slowness (Fig. 8a). In both the P and S receiver functionsections, the timings of conversions from the LAB (10 s), andthe Moho (5 s), are in excellent agreement. The slightly largeramplitudes of SKS-to-p receiver functions compared to the S-to-preceiver functions (Fig. 8b), could be possibly due to the dippingnature of the underlying structure, and its disposition (strike anddip) with respect to the backazimuth of the events. To demonstrate

Figure 8. Comparison of the (a) P receiver functions and (b) S receiver functions for station TEZ in the Himalayan foredeep. Both the receiver functions havebeen averaged in narrow slowness bins of 0.1 s/◦. Positive and negative amplitudes are colour coded with red and blue respectively. The phase labelled LABhas been observed using both the independent techniques showing the authentication of our observation.




Figure 9. (a) Amplitude variations of S-to-p and SKS-to-p conversions with increasing slowness for 0◦ backazimuth for a simple model shown in (b). Solidcurve represents a horizontal layer and dashed curves represent dipping layers with different strike and dip values indicated within the parenthesis.

this variation of amplitudes in S-to-p and SKS-to-p conversions,theoretical predictions of amplitude variations in the slowness rangeappropriate for SKS and S waves are generated for a simple model(Fig. 9b) with horizontal and dipping layers having 3◦, 5◦ and 10◦

dip. To simulate updip and downdip effects, the event backazimuthis considered as zero (north) while the strike of the layer is changedto have values of 90◦, 180◦ and 270◦. Instead of varying the eventdirections, the strike of the interface is changed to produce the sameeffect. Variation of the synthetic Sp/S amplitudes as a functionof slowness corresponding to the actual traces (in Fig. 8) showsthat events from the downdip direction produce larger conversionamplitudes compared to those from the updip direction (Fig. 9a).Hence, for a north-dipping layer for example, SKS waveforms fromthe southern azimuths albeit having a much lower slowness rangewill produce larger conversion amplitudes compared to S waveformsfrom northern azimuths. This scenario is similar to the presentdistribution of events shown in Fig. 2.

An image of the individual S receiver functions along the N–Sprofile AB is shown in Fig. 10. This image is generated by binningthe waveforms whose piercing points lie within a narrow widthof 0.1◦ of longitude and sorting them from south to north. Boththe positive conversions from the Moho and the negative ones fromthe LAB can be traced continuously along the whole profile. Be-cause the piercing points are estimated based on the arrival times ofthe negative phase, the LAB would be coherently mapped although

the Moho may not correspond to its actual geographical location.Even the individual receiver functions bring out a north-dipping na-ture of the LAB that becomes more prominent beyond the foredeepregion (27.5◦ latitude) as it approaches the Himalayan front. Thisfigure demonstrates that subsequent stacking of the receiver func-tions to enhance the coherence of the otherwise weak conversionsis reasonable and robust. Also, it appears that the thickness of themantle lid of the Indian lithosphere in the south part of the sec-tion is reasonably consistent with that in the northern part (shownby vertical arrows). This suggests that the Indian Plate as a wholeshows a gentler dip from south to north, with an anomaly between26 and 27◦ latitude showing thinning of lithosphere underneath andin the vicinity of the Shillong plateau, which suffered volcanism inthe past around 116 Ma (Ray et al. 2005). Stronger amplitudes ofthe LAB conversions in the northern part of the profile can be rec-onciled by a dipping layer. To demonstrate this effect, we generatedsynthetics (Frederiksen & Bostock 2000) using a simple two-layercase with the LAB dipping northwards at an angle of 20◦ from26.5 to ∼29◦ latitude. As depicted in Fig. 11, we consider a simplecrust with a shear velocity of 3.8 km s–1, a mantle lid with velocity4.4 km s–1 and the drop in Vs across the LAB as ∼4.5 per cent(Artemieva 2009). Here our aim is not to model the sharpness ofany of the boundaries, but to qualitatively demonstrate the effect ofdip on the amplitudes of SRFs. The synthetic receiver functions re-veal a systematic increase in the amplitude of the LAB conversions

Figure 10. Image plot along the north–south profile AB (shown in Fig. 3). The image has been generated by binning S receiver functions within 0.1◦ latitude.The plot shows two prominent and coherent phases, the positive marked as Moho and the negative marked as LAB. Deepening of the LAB from South to Northindicates the dipping nature of the Indian plate as it approaches the Himalaya.




Figure 11. Synthetic S receiver functions (RFs) that demonstrate the vari-ation in conversion amplitudes due to a dipping structure. The traces arearranged by distance calculated using the event distribution shown in inset.

in the downdip direction. The negative phase observed in real datain Fig. 10, appears to be produced by a dipping interface becausethe amplitude of the synthetic S-to-p conversions show a graduallyincreasing trend from ∼10 (∼90 km) to ∼16 s (∼150 km) similarto the observation in Fig. 10.

Fig. 12 shows the stacked S receiver functions correspondingto the bins indicated in Fig. 3, arranged in increasing order of thearrival times of the negative phase. Because the Moho is usuallywell observed in P receiver functions due to their higher frequencycontent, we will focus on the signal from the LAB phase from the Sreceiver functions. It is evident that the data from all the bins revealclear positive Sp conversions from the Moho, followed by a negativeconversion from the LAB. It is pertinent to note that the SRFs stacksat individual stations also reveal clear Moho and LAB conversions(Fig. 13) and the timings of these Moho conversions agree wellwith the results from the P receiver functions (Kumar et al. 2004b)lending credence to our analysis. These conversion times can betranslated into approximate depths by multiplying them by a factorof 8 for the crust and 9 for the uppermost mantle based on the veloc-ity structure from the IASP91 standard earth model and a slownessvalue of 6.4 s deg–1. It is often difficult to constrain the depth cor-responding to the negative peaks, without a priori knowledge ofthe exact velocity model of the area. However, based on the aboveconversion times and the average velocity multipliers, we prepareda lithospheric thickness map for each bin (Fig. 14). The lithosphericthickness values indicated in Fig. 14 generally show a thick litho-sphere within the convergent margins namely Eastern Himalayaand the Burmese arc regions and a thinner lithosphere underneaththe Shillong plateau and foredeep regions. To systematically mapthe spatial variations of the Moho and the LAB from the IndianPlate interior to the convergent margins, we present the results ofour analysis along two profiles (shown in Fig. 3) in Fig. 15. For eachof the bins representing these profiles, the Moho and LAB phases areconsidered as reliable only if the amplitudes of the Sp conversionslie above the ±2 standard error (SE; as in Fig. 12). In case of bin 16for example, the negative phase at 10 s is below the error level ofthe stack, whereas the phase at 15 s is above it. Hence, we chose the

Figure 12. (a) Moveout corrected and stacked S receiver functions with their 2 sigma errors, corresponding to the spatial bins shown in Fig. 3. The number atthe top corresponds to the grid number. The bottom number shows the number of receiver functions used for stacking. The traces are arranged with increasingarrival times of the LAB. In each of these traces, the first positive peak corresponds to the Sp conversion from Moho followed by a negative phase interpretedas a conversion from the LAB.

Figure 13 S receiver function stacks at five stations along with their 2 sigma errors. The stations are sorted from south to north.




Figure 14. Depth to the Lithosphere–Asthenosphere boundary in the northeastern Himalayan region. The number in the box corresponds to the bin numbermarked in Fig. 3. The scale has been chosen converting the timings of LAB by multiplying with a factor of 9.

Figure 15. The stacked traces plotted together with elevation along thetwo profiles shown in Fig. 3 (AB: North–South and CD: East–West). Theblack dots are the local seismicity within a narrow width along the profiles.The first positive peak is Moho followed by the second negative phasefrom the lithosphere–asthenosphere boundary. In all the sections it is clearlyevident that the Indian lithospheric plate deepens below the high topographyof Eastern Himalaya. The right-side ordinates are the approximate depthscorresponding to the times shown in the left ordinates.

latter as a likely candidate for LAB. Although bin 2 lies in Tibet,where earlier publications by several authors (using different tech-niques) reveal a crustal thickness of ∼70–80 km, the Moho at ∼85km indicated by us seems unreliable because the Moho conversionfor this bin is barely above the noise level. The seismic sectionin Fig. 15(a) representing the region from the Shillong plateau tothe Eastern Himalaya shows a significant variation in LAB depthsfrom ∼90 km below the Shillong plateau to about ∼180 km be-neath the Eastern Himalaya. The second profile oriented in theeast–west direction (Fig. 15b), spans part of the Indian shield, theHimalayan foredeep and extends up to the onset of Indo–Burmaconvergence zone, where earthquakes hypocentres down to100 km demarcate a subduction zone. While the lithospheric thick-ness is close to 100 km beneath the foredeep, deepening of thelithosphere till a depth of ∼160 km in the Burmese arc region isin conformity with the presence of a cold subducted slab. Lack ofpiercing points further east of the Burmese arc refrains us fromtracing the Indian Plate in this region.

5 D I S C U S S I O N

Distinct signatures of the LAB and its varied depth underneath theShillong plateau, Himalayan foredeep, the Himalayan collision zoneand the Burmese arc regions suggests that S-to-p converted phasesare very effective in mapping the spatial variations in lithosphericthickness with a much more detail compared to the other prob-ing tools like seismic tomography. The Shillong plateau wedgedbetween the Himalayan collision in the North and the Burmese sub-duction in the East is an enigmatic geodynamic feature of the IndianPlate, in view of its elevation (average 1 km). Direct evidence for theplateau uplift comes from the results of repeated geodetic levellingconducted over the plateau by the Geodetic and Research Branch ofthe Survey of India, (Kailasam 1979). Also, based on the occurrenceof earthquakes at subcrustal depths (focal depths >40 km) beneaththe plateau, Chen & Molnar (1990) suggest a cold upper mantle. Allthese arguments point towards a high-velocity, cold, upper mantle,suggesting that a thermal source due to an earlier hotspot or plumeactivity, could not have caused the current uplift. Based on the anal-ysis of focal mechanisms and computation of strain rates, a newmodel for the uplift of the Shillong plateau has been proposed (Rao& Kumar 1997). It is suggested that the Shillong plateau was up-lifted under the influence of compressive stresses resulting from theIndia–Eurasia collision in the North, aided by a timely impetus from




the India–Burma thrust forces in the east, which is sustained evenat present. In this context, a thin lithosphere in the vicinity of theplateau could mean that the uplift represents a lithospheric up warprelated to flexure and not a crustal one, because the crustal thicknessis only 35 km. A ∼100 km thick Indian shield lithosphere (Kumaret al. 2007) gradually thickens as it reaches the Himalaya where theIndian Plate undergoes homogenous thickening. An abrupt increasein the plate thickness is seen at the northern end of the foredeepregion where the depth to LAB reaches about 135 km.

Estimates of lithospheric thickness values in Tibet derived fromapplication of the S receiver function technique to the INDEPTHdata (Kumar et al. 2006), reveal existence of strong distinct litho-spheres in the northern as well as southern Tibet with the Indianlithosphere subducting below the Asian lithosphere just south ofBangong suture zone. Overall, it emerges that both the crust andmantle lithosphere of the Indian Plate are homogeneously thick-ened in the Eastern Himalaya (Fig. 15a), with the compressivestress arising due to the continued collision in this region causingits upliftment. Bending of the lithosphere further north reachingdepths down to ∼180 km beneath the higher Himalaya, with a largetopography, is suggestive of thickening in an orogenic setting, due tocompression tectonics (Fig. 15a). The global (e.g. Ritzwoller et al.2002) and regional (Friederich 2003) tomographic models do bringout a thickening of the Indian lithosphere towards Himalaya alongour receiver function profiles, although with a poorer resolution.Although the tomography results of Friederich (2003) samplingthe region from 20 to 50◦ north latitude do indicate presence of ahigh velocity layer below our study region, this anomaly extends todepths deeper than 200 km. The results obtained from this study areconsistent with the results obtained earlier by Kumar et al. (2006)using essentially the same technique. This suggests that the IndianPlate is preserved till it gets subducted beneath Tibet just south ofBangong Suture zone. High resolution imaging of the crustal ar-chitecture using a dense network of seismic stations along a linearprofile from the Nepal Himalaya up to southern Tibet lend supportto such an interpretation (Nablek et al. 2009).

6 C O N C LU S I O N S

This study presents the first results of the collision architecture ofthe Indian and Asian lithospheric plates in the Eastern Himalaya andnortheast India obtained by application of the S receiver functiontechnique to waveforms from a network of eleven broad-band sta-tions. The bottom of the Indian Plate can be traced as a continuousentity from south of the Shillong plateau, through the Himalayanforedeep up to the centre of the Tibetan plateau, close to south ofBangong-Nujiang (BNS). A scenario of homogeneous northwardthickening of the lithosphere from ∼135 km in the foredeep to∼180 km beneath southern Tibet and its correlation with topogra-phy is consistent with continental collision between India and Asia.Thickening of the lithosphere towards east underneath the Burmesearc region is in conformity with the presence of a cold subductedIndian slab below the Burmese Plate. Interestingly, an upwelling inthe lithosphere in the vicinity of the Shillong plateau is observedsuggesting that the uplift is not confined to crustal depths.

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

The work has been performed under the project NGRI SIP0012–28(MRK). The Department of Science and Technology sup-ported the broad-band experiment in the northeast India region.

Most of the computations are performed using Seismic Handler [K.Stammler]. The paper benefited immensely from the comments bythe editor Randy Keller and two anonymous reviewers.

R E F E R E N C E S

Ammon, C.J. & Zandt, G., 1993. The receiver structure beneath the southernMojave Block, Bull. seism. Soc. Am., 83, 737–755.

Angus, D.A., Wilson, D.C., Sandvol, E. & Ni, J., 2006. Lithospheric structureof the Arabian and Eurasian collision zone in eastern Turkey from S-wavereceiver functions, Geophys. J. Int., 166(3), 1335–1346.

Artemieva, I.M., 2009. The continental lithosphere: reconciling thermal,seismic, and petrologic data, Lithos, 109, 23–46.

Barrell, J., 1914. The strength of the Earth’s crust, J. Geol., 22, 28–48.Bilham, R. & England, P., 2001. Plateau pop-up in the 1897 Assam earth-

quake, Nature, 410, 806–809.Bostock, M.G., Hyndman, R.D., Rondenay, S.J. & Peacock, S., 2002. An

inverted continental Moho and the serpentinization of the fore arc mantle,Nature, 417, 536–538.

Chen, W.P. & Molnar, P., 1990. Source parameters of earthquakes and in-traplate deformation beneath the Shillong Plateau and the Northern In-doburman ranges, J. geophys. Res., 95, 12527–12552.

Das Gupta, S. & Biswas, A., 2006. Geology of Assam, Geological Societyof India, Bangalore.

Eaton, D.W., Darbyshire, F, Evans, R.L., Grutter, H., Jones, A.G. & Yuan, X.,2009. The elusive lithosphere–asthenosphere boundary (LAB) beneathcratons, Lithos, 109, 1–22.

Farber, S. & Muller, G., 1980. Sp phases from the transition zone betweenthe upper and lower mantle, Bull. seism. Soc. Am., 70, 487–508.

Farra, V. & Vinnik, L., 2000. Upper mantle stratification by P and S receiverfunctions, Geophys. J. Int., 141, 699–712.

Fischer, K.M., Ford, H.A., Abt, D.L. & Rychert, C.A., 2010. The lithosphere-asthenosphere boundary, Annu. Rev. Earth planet Sci., 38, 551–575.

Frederiksen, W. & Bostock, M.G., 2000. Modelling teleseismic waves indipping anisotropic structures, Geophys. J. Int., 141(2), 410–412.

Friederich, W., 2003. The S-velocity structure of the East Asian mantlefrom inversion of shear and surface waveforms, Geophys. J. Int., 153, 88–102.

Gutenberg, B., 1959. Physics of the Earth’s Interior. Elsevier, New York.Heit, B., Sodoudi, F., Yuan, X., Bianchi, M. & Kind, R., 2007. An S-Receiver

function analysis of the lithospheric structure in South America, Geophy.Res. Lett., 34, L14307, doi:10.1029/2007GL030317.

Heuer, B., Kampf, H., Kind, R. & Geissler, W.H., 2007. Seismic evidencefor whole lithosphere separation between Saxothuringian and Moldanu-bian tectonic units in central Europe, Geophys. Res. Lett., 34, L09304,doi:10.1029/2006GL029188.

Jordan, T.H., 1981. Continents as a chemical boundary layer, Phil. Trans. R.Soc. Lond., A, 301, 359–373.

Kailasam, L.N., 1979. Plateau uplift in peninsular India, Tectonophysics, 61,243–269.

Kind, R. et al., 2002. Seismic images of crust and upper mantle beneathTibet: evidence for Eurasian plate subduction, Science, 298, 1219–1221.

Knopoff, L., 1983. The thickness of the lithosphere from the dispersionof surface waves, Geophys. J. Int., 74(1), 55–81, doi:10.1111/j.1365-246X.1983.tb01870.x.

Kumar, A., Dayal, A.M. & Padamakumari, V.M., 2004a. Kimberlitefrom Rajmahal magmatic province: Sr-Nd-Pb isotopic evidence forKerguelen plume derived magmas, Geophys. Res. Lett., 30(20), 2053doi:10.1029/2003GL018462.

Kumar, M.R., Raju, P.S., Devi, E.U., Saul, J. & Ramesh, D.S., 2004b. Crustalstructure variations in northeast India from converted phases, Geophys.Res. Lett., 31, L17605, doi:10.1029/2004GL020576.

Kumar, P., Yuan, X., Kind, R. & Kosarev, G., 2005a. The lithosphere–asthenosphere boundary in the Tien Shan-Karakoram region from S re-ceiver functions: evidence for continental subduction, Geophys. Res. Lett.,32, L07305, doi:10.1029/2004GL022291.




Kumar, P. et al., 2005b. The lithosphere–asthenosphere boundary in theNorth West Atlantic Region, Earth planet. Sci. Lett., 236, 249–257.

Kumar, P., Yuan, X., Kind, R. & Ni, J., 2006. Imaging the colliding Indian andAsian lithospheric plates beneath Tibet, J. geophys. Res., 111, B06308,doi:10.1029/2005JB003930.

Kumar, P., Yuan, X., Kumar, M.R., Kind, R., Li, X. & Chadha, R.K.,2007. The rapid drift of the Indian tectonic plate, Nature, 449, 894–897,doi:10.1038/nature06214.

Kumar, P. & Kawakatsu, H., 2011. Imaging the seismic lithosphere–asthenosphere boundary of the oceanic plate, Geochem. Geophys.Geosyst., 12(G3), Q01006, doi:10.1029/2010GC003358.

Landes, M., Ritter, J.R.R. & Readman, P.W., 2007. Proto-Iceland plumecaused thinning of Irish lithosphere, Earth planet. Sci. Lett., 225(1–2),32–40.

Langston, C.A., 1981. Evidence for the subducting lithosphere under south-ern Vancouver Island and western Oregon from teleseismic P wave con-versions, J. geophys. Res., 86, 3857–3866.

Li, X., Kind, R., Yuan, X., Wolbern, I. & Hanka, W., 2004. Rejuvenation ofthe lithosphere by the Hawaiian plume, Nature, 427, 827–829.

Li, X., Yuan, X. & Kind, R., 2007. The lithosphere–asthenosphere bound-ary beneath western United States, Geophys. J. Int., 170(2), 700–710.doi:10.1111/j.1365-246X.2007.03428.x.

Mitra, S., Priestley, K., Gaur, V.K. & Rai, S.S., 2006. Shear-wave structure ofthe South Indian lithosphere from Rayleigh wave phase-velocity measure-ments, Bull. seism. Soc. Am., 96, 15551–15559, doi:10.1785/0120050116.

Nablek, J. et al., 2009. Underplating in the Himalaya–Tibet collision zonerevealed by the Hi-CLIMB experiment, Science, 325, 1371–1374.

Nandy, D.R., 2001. Geodynamics of Northeastern India and the AdjoiningRegion, 1–209, ACB Publications, Kolkata, India.

Nelson, K.D. et al., 1996. Partially molten middle crust beneath south-ern Tibet: a synthesis of project INDEPTH initial results, Science, 274,1684–1688.

Ni, J.F., Speziale, M.G., Bevis, M., Holt, W.E., Wallace, T.C. & Seager,W.R., 1989. Accretionary tectonics of Burma and the three dimensionalgeometry of the Burma subduction, Geology, 17, 68–71.

Oreshin, S., Vinnik, L. & Peregoudov, D., 2002. Lithosphere and astheno-sphere of the Tien Shan imaged by S receiver functions, Geophys. Res.Lett., 29, 81191, doi:10.1029/2001GL014441.

Owens, T.J., Randall, G.E., Wu, F.T. & Zeng, R., 1993. PASSCAL instrumentperformance during the Tibetan Plateau passive seismic experiment, Bull.seism. Soc. Am., 83, 1959–1970.

Priestley, K., Debayle, E., McKenzie, D. & Pilidou, S., 2006. Upper mantlestructure of eastern Asia from multimode surface waveform tomography,J. geophys. Res., 111, B10304, doi:10.1029/2005JB004082.

Ramesh, D.S., Kumar, M.R., Uma Devi, E., Solomon Raju, P. & Yuan,X., 2005. Moho geometry and upper mantle images of northeast India,Geophys. Res. Lett., 32, L14301, doi:10.1029/2005GL022789.

Rao, N.P. & Kumar, M.R., 1997. Uplift and tectonics of the Shillong plateau,J. Phys. Earth, 45, 167–176.

Ray, J.S., Pattanayak, S.K. & Pande, K., 2005. Rapid emplacement ofthe Kerguelen plume–related Sylhet Traps, eastern India: evidencefrom 40Ar-39Ar geochronology, Geophys. Res. Lett., 32, L10303,doi:10.1029/2005GL022586.

Replumaz A., Karason K., Van Der Hilst R., Besse J. & Tapponnier P., 2004.4-D evolution of SE Asia mantle structure from geological reconstructionand seismic tomography, Earth planet. Sci. Lett., 221(1–4), 103–115.

Ritzwoller, M.H., Shapiro, N.M., Barmin, M.P. & Levshin, A.L., 2002.Global surface wave diffraction tomography, J. geophys. Res., 107(B12),2335, doi:10.1029/2002JB001777.

Rychert, C.A., Rondenay, S. & Fischer, K.M., 2007. P-to-S and S-to-Pimaging of a sharp lithosphere–asthenosphere boundary beneath easternNorth America, J. geophys. Res., 112, doi:10.1029/2006JB004619.

Saul, J., Kumar, M.R. & Sarkar, D., 2000. Lithospheric and upper man-tle structure of the Indian Shield, from teleseismic receiver functions,Geophys. Res. Lett., 27(16), 2357–2360.

Sodoudi, F. et al., 2006a. Lithospheric structure of the Aegean obtainedfrom P and S receiver functions, J. geophys. Res., 111, doi:10.1029/2005JB003932.

Sodoudi, F., Yuan, X., Liu, Q., Kind, R. & Chen, J., 2006b. Lithosphericthickness beneath the Dabie Shan, central eastern China from S receiverfunctions, Geophy. J. Int., 166, 1363–1367.

Wilson, D.C., Angus, D.A., Ni, J.F. & Grand, S.P., 2006. Constraints on theinterpretation of S-to-P receiver functions, Geophys. J. Int., 165, 969–980.

Yuan, X., Kind, R., Li, X. & Wang, R., 2006. The S receiver functions:synthetics and data example, Geophys. J. Int., 165, 555–564.



Imaging the Indian lithosphere beneath the Eastern Himalayan region

Documents