from crystal to crustal: petrofabric-derived seismic

From crystal to crustal: petrofabric-derived seismic modelling

of regional tectonics

G. E. LLOYD1*, J. M. HALLIDAY1, R. W. H. BUTLER2, M. CASEY1†, J.-M. KENDALL3,

J. WOOKEY3 & D. MAINPRICE4

1Institute of Geophysics and Tectonics, School of Earth & Environment,

University of Leeds, Leeds LS2 9JT, UK2Geology and Petroleum Geology, School of Geosciences, University of Aberdeen,

Meston Building, King’s College, Aberdeen AB24 3UE, UK3Department of Earth Sciences, University of Bristol, Bristol BS8 1RJ, UK

4Geosciences Montpellier, CNRS & Universite Montpellier 2,

34095 Montpellier, France

*Corresponding author (e-mail: [email protected])

Abstract: The Nanga Parbat Massif (NPM), Pakistan Himalaya, is an exhumed tract of Indiancontinental crust and represents an area of active crustal thickening and exhumation. While themost effective way to study the NPM at depth is through seismic imaging, interpretationdepends upon knowledge of the seismic properties of the rocks. Gneissic, ‘mylonitic’ and cataclas-tic rocks emplaced at the surface were sampled as proxies for lithologies and fabrics currentlyaccommodating deformation at depth. Mineral crystallographic preferred orientations (CPO)were measured via scanning electron microscope (SEM)/electron backscatter diffraction(EBSD), from which three-dimensional (3D) elastic constants, seismic velocities and anisotropieswere predicted. Micas make the main contribution to sample anisotropy. Background gneisses havehighest anisotropy (up to 10.4% shear-wave splitting, AVs) compared with samples exhibitinglocalized deformations (e.g. ‘mylonite’, 4.7% AVs; cataclasite, 1% AVs). Thus, mylonitic shearzones may be characterized by regions of low anisotropy compared to their wall rocks. CPO-derived sample elastic constants were used to construct seismic models of NPM tectonics,through which P-, S- and converted waves were ray-traced. Foliation orientation has dramaticeffects on these waves. The seismic models suggest dominantly pure-shear tectonics for theNPM involving horizontal compression and vertical stretching, modified by localized ductileand brittle (‘simple’) shear deformations.

Knowledge of tectonic and geodynamic processesoccurring within the continental crust, particularlyat deeper levels, is traditionally obtained via detailedgeological analysis in both the field and subsequentlyin the laboratory of tracts of former middle and lowercrust now exposed at the surface. Although relevantto past processes, this approach provides onlyguides to what might be happening today in themodern middle and lower continental crust. In situtechniques involving remote sensing of the currentstate of the middle and lower continental crust havetherefore recently become increasingly popularand useful. The potentially most useful of these tech-niques is provided by seismic waves, as both naturaland controlled source seismology can be used toimage tectonic structures occurring at depth (e.g.Burlini et al. 1998; Khananehdari et al. 1998).However, such approaches demand that the control

geology exerts on the seismic properties is fullyunderstood. For example, as seismic resolutiondepends on the magnitude and spatial extent ofseismic anisotropy exhibited by the geology, thereare limits to thesizeof structures thatcanbeobserved.Shear zones, which are typically narrower than theseismic wavelength, are consequently not normallyvisible using standard seismic survey methodssuch as multi-azimuth wide-angle controlled-source seismic surveys. However, it has been sug-gested (e.g. Guest et al. 1993; Lloyd & Kendall2005) that vertical resolution can be improvedfurther using knowledge of reflection coefficientsand mode conversions at (lithological) interfaceboundaries, which demands detailed knowledge ofthe geological control on the seismology.

Large-scale continental structures are typicallyexhumed or emplaced on localized zones of

†Deceased 2008.

From: Prior, D. J., Rutter, E. H. & Tatham, D. J. (eds) Deformation Mechanisms, Rheology and Tectonics:Microstructures, Mechanics and Anisotropy. Geological Society, London, Special Publications, 360, 49–78.DOI: 10.1144/SP360.4# The Geological Society of London 2011. Publishing disclaimer: www.geolsoc.org.uk/pub_ethics

http://www.geolsoc.org.uk/pub_ethics

Fig. 1.

G. E. LLOYD ET AL.50

high-strain deformation that may persist to greatdepths, perhaps through the entire crust, and formboth simple and complex linked three-dimensionalarrays (e.g. Sibson 1977; Ramsay 1980; Coward1994). It has long been known that the deformationmechanisms and processes responsible for emplace-ment and/or exhumation are frequently preservedin the grain-scalemicrostructuresobserved insurfaceoutcrops (e.g. Nicolas & Poirier 1976; Passchier &Trouw 1996). More recently, it has been realizedthat these microstructures can also be used toestimate a variety of petrophysical properties and,in particular, their seismic characteristics (e.g.Babuska & Cara 1991; Kocks et al. 1998). Ductileshear zones in lower continental crust containstructures at the scale of the crystal lattice, grainsand lithological layering and larger. The seismic ani-sotropy of such zones is a function of all these struc-tures. Characterization of this seismic anisotropynecessitates information on the various scales anda method of integrating the effects to obtain the bulkproperties (e.g. Christensen 1984; Ben Ismail &Mainprice 1998). In contrast, concentrated shear inupper levels of the continental crust often occursalong cataclastic zones, the seismic properties ofwhich are determined by the fracture density orporosity of the fault rocks (e.g. Crampin 1981).

This contribution concentrates on the petro-fabric or crystal lattice preferred orientation (CPO)determination of seismic properties (e.g. Kern1982; Siegesmund et al. 1989; Barruol & Mainprice1993; Burlini & Kern 1994; Barruol & Kern 1996).It has long been recognized that CPO is a majorcause of seismic anisotropy (Hess 1964; Christensen1971). The question posed is therefore whether ornot it can be expected to be able to distinguishbetween different structural, tectonic and geody-namic configurations (e.g. shear and fault zones)at depth in crystalline rocks by measurement oftheir seismic properties. The Nanga Parbat Massif(NPM), Pakistan Himalaya (Fig. 1), is consideredan ideal example of the possibility of using seismicobservations to determine whether its emplacementwas/is accommodated solely on localized fault andshear zones via essentially simple-shear defor-mation (e.g. Burg 1999), or whether bulk pure-sheardeformation within the massif as a whole played/plays a significant role (e.g. Butler et al. 2002).CPO data from characteristic rock samples,measured via scanning electron microscope (SEM)electron backscattered diffraction (EBSD), areused to determine the overall elastic and seismicproperties of rocks from the NPM. These can then

be used to populate various seismic models to inves-tigate the regional geodynamic setting.

Regional geology and sample descriptions

Regional geology

The NPM (Fig. 1a) lies within the Indo–Asian con-tinental collision zone (Fig. 1b), which forms themajor current example of active continental col-lision and compressive intra-continental tectonics.It represents an exhumed tract of Indian continentalcrust and, as an area of active crustal thickening andexhumation, offers insights into how localizedthrust faults exposed at the Earth’s surface couplewith more distributed strain at depth. The NPMexhibits a remarkably high topography (.8 km)and forms an elongate, upright, north–south anti-form with c. 40 km wavelength that probablyinvolves the whole crust (Zeitler et al. 2001;Butler et al. 2002). It comprises exhumed young(,2 Ma) Indian continental granitic and meta-sedimentary gneisses and granitic orthogneissesbasement and calc-silicate and marble meta-sedimentary cover. However, there is a distinct geo-chemical, metamorphic and tectonic regime relativeto the surrounding terrain, which suggests an anom-alous structure beneath the massif (George et al.1993; Zeitler et al. 1993, 2001). The massif can bedefined by the shape of the contact, known as theMain Mantle Thrust (MMT), between Indian con-tinental gneisses with the overlying Kohistan arcrocks (Fig. 1c, d). The entire complex was sub-ducted and metamorphosed during collision toamphibolite grade typified by a quartz-muscovite-biotite-plagioclase assemblage (Zeitler et al.2001). Peak metamorphic conditions towards thewest of the massif, in the vicinity of the samplingregion (see below), are thought to have beenc. 7–11 kbars and 550–700 8C (i.e. 25–40 kmdepth), in contrast to c. 7–14 kbars and 675–800 8C (i.e. 40–50 km depth) on the eastern marginof the massif (Poage et al. 2000).

The NPM has been, and continues to be,exhumed on a series of ductile shear and brittlefault zones, resulting in at least 22 km of erosionduring emplacement (Butler et al. 2000; seeFig. 1c, d). It is assumed that the overall tectonicregime has not changed significantly over the lastfew million years (Zeitler et al. 2001), with geo-chronological data suggesting that exhumation isexposing rocks that are representative of the defor-mation kinematics currently operating at depth

Fig. 1. The Nanga Parbat Massif, Pakistan Himalaya. (a) General view of the Nanga Parbat Massif. (b) Generallocation, with the Nanga Parbat Massif region indicated by the white box. (c) Main geological features of the NangaParbat Massif region (after Butler et al. 2002), with the location of the study area indicated. (d) Schematic cross-sectionof the Nanga Parbat Massif (after Butler et al. 2002); note location of the study area.

FROM CRYSTAL TO CRUSTAL 51

Fig. 2.


(Zeitler et al. 1993). The central part of the massif isbound by two primary shear zones (the Liachar andRupal), although current exhumation is achieved ona cataclastic fault zone termed either the LiacharThrust (Butler & Prior 1988) or the Raikhot Fault(Zeitler et al. 2001) which emplaces gneisses ontoQuaternary gravels (e.g. Fig. 2a–c). The 2 kmwide Liachar Shear Zone occurs in the hangingwall of this structure (Fig. 1d), which led Butler &Prior (1988) to suggest that the surface cataclasticfaults pass downwards into a ductile simple-shearzone. However, an alternative model (Butler et al.2002) argues for more distributed deformation atdepth, partitioned between broadly simple shear inthe ductile shear and cataclastic fault zones andbroadly pure shear in the rest of the hanging wall.Various models currently exist to explain thekinematics of the exhumation and emplacementof the NPM. Nevertheless, as an area of activecrustal thickening and exhumation, the NPMoffers insight into how localized thrust faults atthe Earth’s surface couples with more distributedstrain at depth (Butler et al. 2002).

It has long been recognized that seismogenicfaulting often passes into aseismic creep withdepth (e.g. Sibson 1977). However, the kinematicsand distribution of strain at depth remains contro-versial (Butler et al. 2002). While simple-sheardeformation is localized into relatively narrowtracts of non-coaxial strain, anastomosing discretetracts could produce a thick zone of macroscopicallydistributed strain (e.g. Burg 1999). Alternatively,distributed strain at depth could be accommodatedby pure-shear subvertical stretching (e.g. Butleret al. 2002). The latter model is one of volumeconservation: layers parallel to the principle axesof strain do not rotate and experience no shearstrain, while layers perpendicular to compressionare shortened and thickened (e.g. Twiss & Moores1992). Butler et al. (2002) suggest that the bulkdeformation within the crust at Nanga Parbat maybe described as heterogeneous vertical stretching;it is only when strain gradients become pronouncedthat simple shear dominates. If this view is correct, acombination of pure and simple shear is responsiblefor the exhumation of the NPM.

Evidence for the mechanism(s) of exhumationof the NPM should be preserved within the rockrecord. For example, when a rock is deformed,one of the fundamental changes observed is are-orientation of the crystal components, usuallyinto a direction defined by the stress field, to form

CPO. It is well known that natural deformationmechanisms in rocks are responsible for inducinganisotropy of the inherent petrophysical properties(Babuska & Cara 1991; Wendt et al. 2003). As themajority of rock-forming minerals are elasticallyanisotropic, seismic anisotropy (i.e. variation ofseismic velocities with direction) is an intrinsicproperty of most minerals and rocks. Seismic aniso-tropy should therefore be present at all scales inthe Earth and hence can be used to infer informationon structures induced by large-scale geodynamicprocesses at depth (e.g. Davis et al. 1997; Shapiroet al. 2004). This contribution uses CPO measuredfrom rocks sampled from appropriate parts of theNPM to estimate their seismic properties and, inparticular, their anisotropy in order to predict themost likely mechanism(s) of exhumation.

Sample descriptions

The most recent phase of exhumation of the NPMhas occurred close to the surface on the LiacharThrust/Raikhot Fault (Butler & Prior 1988; Zeitleret al. 2001). In the hanging wall of this structurethere is a high-strain zone known as the LiacharShear Zone (Butler & Prior 1988), which emplacesaugen gneisses of the Indian continental crustonto the Kohistan Arc rocks (Fig. 2a–c). Althoughdeformation of these gneisses (as indicated byvarying degrees of lithological layering and grain-size reduction) effectively increases towards theLiachar Thrust, such deformation is not related tothe thrusting. Rather, it is related to the formationof the ductile Liachar Shear Zone that developedpreviously at deeper levels, which is now beingexhumed on the brittle Liachar Thrust. A suite ofrocks has been collected from the Liachar ShearZone in the Tato Ridge area above the LiacharThrust (Figs 1d & 2a–c). Two specific samples(B1 and B2) from this suite are considered in thiscontribution (Fig. 3). Both are augen orthogneissesand contain the same quartz + plagioclase + ortho-clase + biotite + muscovite mineralogy, althoughin varying proportions, while sample B2 also con-tains minor garnet.

Sample B1 (Fig. 3a) contains relict clasts of theoriginal igneous protolith, with a large range inclast and grain sizes. The clasts are mainly of pla-gioclase and orthoclase (although there are alsoregions of biotite + muscovite), while quartz hasbehaved in a ductile manner. The segregationof each of these phases, particularly the micas,

Fig. 2. Tato–Raikhot–Bulder study region, Nanga Parbat Massif. (a) Geological map; note the specific locationsof samples B1 and B2 along the sampling traverse. (b) General view of the Tato Road section from where the sampleswere collected; note steeply dipping foliation. (c) Schematic cross-section of the Tato Road region; note generallocations of samples B1 and B2.


imparts a crude lithological layering. Sample B2comprises two distinctive subsamples. Half of thesample (B2m) is considered to represent the ulti-mate development of lithological layering andgrain-size reduction due to ductile shearing(Fig. 3b). All minerals appear extended in the foli-ation except for garnet, which occurs as porphyro-blasts. In the other half of the sample (B2c), thelithological layering due to ductile shearing hasbeen completely destroyed by the effects of thebrittle deformation associated with the LiacharTrust (Fig. 3c). However, a weak ‘foliation’ due tovariations in fragment sizes is apparent within acement comprising very-fine-grained orthoclase.The kinematic indicators associated with the micro-structures of samples B1, B2m and B2c, whether

ductile or brittle, suggest a top-to-the-northwestsense of movement, consistent with the large-scalegeometry of the NPM (e.g. Figs 1d & 2b; seeButler et al. 2002 for details).

The varying degrees of lithological layering,grain-size reduction of quartz and feldspars byductile processes and other microstructures in theaugen gneiss samples (e.g. Fig. 3a, b) attest to differ-ent degrees of deformation under (at least) loweramphibolite facies temperatures. The samples aretherefore considered to indicate an increasingstrain gradient within the Liachar Shear Zone,partly responsible for the exhumation and emplace-ment of the NPM at depth (e.g. Figs 1d & 2c). Assample B1 was collected from the periphery of theLiachar Shear Zone, it is considered to represent

Fig. 3. SEM back-scattered electron (BSE) atomic number (Z) contrast images of the samples of Nanga Parbatorthogneisses used in this study, illustrating the textural variations between the samples. Minerals indicated: Q, quartz;K, orthoclase; P, plagioclase; B, biotite; M, white mica; G, garnet; and C, calcite. (a) Sample B1, representing the‘protolith’: note relatively coarser grain size and metamorphic layering or banding. (b) Sample B2m, developed fromthe ‘protolith’: note relatively finer scale shear-zone foliation defined by all minerals except porphyroblastic garnet.(c) Sample B2c, developed from the ‘protolith’ and/or the shear zone: cataclasite; note variation in size and compositionof individual fragments, set in an orthoclase cement seal cut by a late calcite vein.


the ‘background’ deformation state of the Indiancontinental gneisses protolith.

Sample B2m represents the maximum expres-sion of the (increasing) strain gradient associatedwith the shear zones involved in the exhumationof the NPM. In contrast, differences in orthoclasecomposition between relict gneissic fragments andcement within sample B2c suggest that faultingwas either associated with and/or succeeded byfluid flow that precipitated orthoclase in void spacescreated by fracturing, resulting in a cementedcataclasite. Such processes are typical of low temp-eratures and hence relatively near-surface defor-mation. Cataclasized sample B2c is thereforeconsidered to represent the brittle fault zonesresponsible for the exhumation/emplacement ofNanga Parbat at shallower crustal levels, a manifes-tation of shallow-level strain partitioning (e.g.Figs 1d & 2c). This sample is obviously nottypical of the fault zones, including those currentlyactive (e.g. the Liachar Thrust), during their move-ment episodes when fracturing dominates themicrostructure and hence petrophysical properties.Rather, the orthoclase cemented cataclasite istypical of the fossilized fault structures as theynow occur. However, the volume and distributionof orthoclase cement provides an estimate of faultzone porosity and hence fracture density duringsuch movement episodes.

Methodology

Factors that determine the seismic properties ofrocks include (e.g. Mainprice & Nicolas 1989;Babuska & Cara 1991): mineralogy and lithology,layering, grain shape, crack and fracture arraysand CPO. The latter is particularly significantbecause single-crystal seismic velocities of mineralsvary with crystal symmetry and direction due tovariations in the elastic properties of the crystals.Consequently, deformation processes such as dis-location creep that lead to CPO of anisotropicminerals in deformed rocks must also impact uponthe elastic stiffness matrix and hence seismic pro-perties of the rock aggregate (e.g. Lloyd et al.2011). Crack and fracture arrays (including poros-ity) are only likely to be significant, and thereforeimpact on seismic properties, at shallow depths(e.g. Crampin 1981). As samples B1 and B2m arefrom either the background gneisses or the shearzones developed at depth, it is therefore expectedthat their seismic properties (and hence those ofmost of the NPM) vary mainly with CPO. How-ever, the cataclastic fault zone (i.e. sample B2c)developed at shallow depths, where it was subse-quently cemented by orthoclase feldspar. The feld-spar has acted to seal the fault zone and thereby

removed the influence of fractures on the seismicproperties.

The methodology used to estimate seismic pro-perties from CPO follows standard procedure(e.g. Mainprice 1990; Barruol & Mainprice 1993;Mainprice & Humbert 1994; Lloyd & Kendall2005). The crystal orientation of each mineralgrain in a sample is measured automatically viaSEM/EBSD (e.g. Prior et al. 1999) to determinethe individual mineral CPO. The single-crystalelastic properties for each mineral CPO are rotatedinto the sample reference frame, such that theelastic parameters of the polycrystal are derivedby integration over all possible orientations in the3D orientation distribution function (e.g. Bunge1982; Ben Ismail & Mainprice 1998). Due tostress/strain compatibility assumptions, three dif-ferent averaging schemes are possible (Crosson &Lin 1971). The constant strain or Voigt (V) aver-age (Voigt 1928) and constant stress or Reuss (R)average (Reuss 1929) provide upper and lowerbounds. However, the arithmetic mean or Hill (H)average (Hill 1952) of V and R is often taken asthe best estimate of the elastic parameters as it isobserved to give results close to experimentalvalues (e.g. Bunge et al. 2000). The individualmineral CPO measurements are then combined intheir correct mineral proportions, from which thewhole-rock seismic properties are estimated viathe Christoffel equation (e.g. Babuska & Cara1991; Kendall 2000).

The seismic properties of interest include thecompressional wave (Vp) and shear waves (Vs1,Vs2) phase velocity distributions in three-dimensions, as well as the degree of shear-wavesplitting for a given direction. The latter is rep-resented as either the absolute difference in shear-wave velocities (i.e. dVs ¼ Vs1 – Vs2) or theshear-wave anisotropy, which is conventionallydefined (e.g. Mainprice & Silver 1993):

AVs% = 100(Vs1 − Vs2)/[(Vs1 + Vs2)0.5]

The absolute anisotropy of Vp, Vs1 and Vs2 canalso be calculated by substituting their appropriatemaximum and minimum values for Vs1 and Vs2

into this equation.Shear-wave splitting analyses of real data esti-

mate the degree of splitting and orientation of thefast shear wave for a given ray direction (e.g.Kendall 2000). Thus, the polarizations (or birefrin-gence) of the fast shear waves (Vs1P) are also calcu-lated from the whole-rock seismic properties viathe Christoffel equation (Mainprice 1990).

In practice, all constituent minerals contribute tothe overall seismic properties of a rock aggregatedepending on their single-crystal elastic parameters,volume fraction and CPO. Thus, the aggregate


Fig. 4.


seismic properties may be different from any of theindividual mineral properties in rocks where onemineral does not dominate in any particularmanner (i.e. modally, elastically and/or strength ofCPO). Indeed, there must be a significant degree ofcrystal alignment (most typically produced by dislo-cation creep due to tectonic stresses in deformedrocks) of at least one major mineral phase toproduce seismic anisotropy, as randomly orientedcrystals must generate an isotropic bulk rock.Where one mineral does dominate, the aggregateseismic properties are likely to reflect the character-istics of that mineral. For example, micas are one ofthe most anisotropic of the common rock-formingminerals, exhibiting a vertical transverse isotropy(VTI) parallel to the crystal c axis (i.e. normal tothe mineral cleavage plane). Depending on themodal content and/or CPO strength, micas aretherefore likely to contribute most to the seis-mic anisotropy of polymineralic rock aggregatesin which they occur (e.g. Lloyd et al. 2009, 2011).

CPO-derived seismic properties

CPO distributions

CPO data are conventionally represented in thetectonic or kinematic reference frame (XYZ, whereX ≥ Y ≥ Z ), usually from samples cut parallel tothe XZ plane. However, seismic data are viewed inthe natural or geographical reference frame(north–south/east–west). The SEM-EBSD CPOdata measured from XZ sections of samples B1and B2 have therefore been rotated into the geo-graphical reference frame by means of two (differ-ent) rotations per sample as follows: B1 is firstlyrotated by 222.58 about an axis plunging 908towards 0008 (i.e. vertical) and secondly by 258about 08/0678; and B2 is firstly rotated by 2408about 908/0008 and secondly by 508 about 08/0908.

The biotite and, to a lesser extent, muscoviteCPO for samples B1 and B2m clearly indicate theductile foliations present in these rocks (Fig. 4a, b).

In the former, the foliation is steeply NNE-dippingand strikes ENE–WSW while in the latter it is mod-erately south-dipping and east–west striking. Thenon-mica phases exhibit less distinct CPO patterns,possibly reflecting local constraints on CPO devel-opment imposed by neighbouring grains. TheCPOs for sample B2c are different and do not indi-cate a foliation (Fig. 4c). Rather, they show ten-dencies for small/great circle distributions centredon either maxima or minima that plunge consist-ently moderately towards the southwest.

Seismic properties

The seismic properties mimic the CPO distributions,particularly for biotite, in terms of the foliation insamples B1 and B2m (Fig. 4a, b). However, thesteep c. northeast-plunging stretching lineationcharacteristic of the ‘background’ gneisses is notreflected in the seismic properties of sample B1and the moderately c. southeast-plunging stretchinglineation characteristic of the ductile shear zones isonly indicated by the maximum in AVs for sampleB2m. These results support the suggestion (e.g.Mahan 2006; Lloyd et al. 2009, 2011) that, whenpresent, micas (in this case biotite) control theseismic properties of the rock, particularly aniso-tropy. Furthermore, because micas indicate mainlyfoliation and cannot recognize lineation, it is notpossible to differentiate between flattening andplane-strain (e.g. simple-shear) type deformations.Thus, the seismic properties for samples B1 andB2m could have arisen from similar or very differ-ent deformations. Sample B2c is therefore interest-ing because its seismic properties (Fig. 3c) areindicative of a non-foliation-forming deformation,such as expected for constriction.

The actual magnitudes of the seismic propertiesfor the three samples are also interesting (Fig. 4). Interms of Vp their velocities are generally similar,reflecting their similar compositions. However, indetail, the ranges in Vp do vary between samples;B1 shows the absolute minimum (5.7 km/s) and

Fig. 4. SEM/EBSD-measured CPO and CPO-derived seismic properties plotted in the geographical reference frame(north–south/east–west) using the Mainprice (2003) suite of programs. All CPO lower hemisphere (except plagioclase,which is both upper and lower hemisphere), contoured in multiples of the uniform distribution (m.u.d.) as indicated(dotted line, 0.5 m.u.d.) with maximum and minimum values indicated by black squares and open circles, respectively.For quartz, the a(11–20), m(10–10), c(0001), r (1–101) and z(01–11) poles are plotted. For all other minerals therespective a(100), b(010) and c(001) poles are plotted. The seismic properties plotted are: the compressional (Vp) andshear (Vs1, Vs2) waves velocities; the azimuthal-dependent shear-wave anisotropy (either AVs or dVs); and thepolarization direction (birefringence) of the fast shear wave (Vs1P), indicated by the black/white lines for the wavepropagation direction as indicated by the plunge/azimuth position on the stereogram. (a) Sample B1: note the steepnorth-dipping ‘background’ foliation, principally defined by biotite which is mimicked by the seismic properties. (b)Sample B2m: note the moderate south-dipping shear-zone foliation, principally defined by biotite which is mimicked bythe seismic properties. (c) Sample B2c: note the generally weak fabrics, although a definite tendency for each mineral toshow small/great circle dispersions about a moderated southwest-plunging axis, which are mimicked by theseismic properties.


B2m the absolute maximum (6.3 km/s). Thesevariations in Vp impact on the AVp values persample, which is a maximum for B1 (8.8%) and aminimum for B2c (1.3%), with B2m intermediate(4.4%). Equivalent behaviour is shown by AVswith sample B1 exhibiting the shear-wave splittingof 10.43% (equivalent to a dVs of 0.35 km/s),while sample B2m has an AVs of 4.69% (dVs ¼0.16 km/s) and sample B2c has an AVs of 1.03%(dVs ¼ 0.04 km/s). Thus, the ductile Liacharsimple-shear zone exhibits significantly lowerseismic anisotropy than the ‘background’ gneissesof the NPM. Similar behaviour, whereby a shearzone is less anisotropic than its wall rocks, hasbeen observed by Michibayashi & Mainprice(2004) and Michibayashi et al. (2006), being attrib-uted in both cases to the presence and significance ofpre-existing mechanical anisotropy on shear-zonedevelopment. If correct, this suggests that the ‘back-ground’ gneisses of the NPM exhibit a significantdeformation fabric in terms of the seismic proper-ties. Although a brittle fault zone, sample B2c isalmost anisotropic. This is presumably due to thedispersion of the various CPO along small andgreat circles about the moderately southwest-plun-ging ‘rotation’ axis and the orthoclase cement seal,which has acted to remove any potential crack-induced anisotropy.

Finally in this section, in terms of polarizationthe fast shear waves (i.e. Vs1P) for samples B1and B2m are polarized parallel to the foliationwhen they propagate within the foliation (Fig. 4a, b);the behaviour of vertically propagating (e.g. tele-seismic) waves (Vs1Pv) is therefore different inboth samples. For sample B1, which is characterizedby a steeply dipping ENE–WSE foliation, Vs1Pv ispolarized ENE–WSW and has an AVs value closeto the maximum. In contrast, for sample B2m,which is characterized by a moderately dippingeast–west foliation, Vs1Pv is polarized northeast–southwest and has an AVs value of less than halfthe maximum for this sample. Sample B2c exhibitssimilar behaviour to sample B2m (Fig. 4c).

CPO-derived seismic modelling

The determination of seismic properties via CPOprovides elastic stiffness matrices for the rockaggregates involved. Such matrices can be used topopulate seismic models with realistic rock proper-ties. This approach is used to construct severalseismic models populated with the CPO-derivedelastic properties of the samples described pre-viously in order to investigate the impact of theseproperties (sic rocks) on the seismic characteristicsof the NPM. In these models, sample B1 is con-sidered to represent the background deformation

state of the Indian continental gneisses protolithaway from the localized high-strain zones, whetherductile or brittle. In other words, it is considered torepresent the rock fabric that comprises the bulk ofthe NPM. In contrast, samples B2m and B2c areconsidered to represent the maximum expressionsof the localized ductile shear and brittle faultzones, respectively, responsible potentially for the(near-surface) exhumation of the NPM. Theseismic models erected investigated the followingparameters: (1) effect of mineralogy, foliationorientation and deformation (partitioning); (2) wavepropagation; (3) seismic reflection coefficients;(4) controlled source surveys; and (5) deformationdetermination from reflection coefficients andmode conversions.

In the seismic models the aggregate elastic con-stants derived for each sample via CPO analysis areconverted into a density-normalized format. Onlytwo elastic constants are required for isotropicmedia but at least nine elastic constants must bespecified (e.g. Babuska & Cara 1991) for anyother form of anisotropy. A computer program(Atrak, Guest & Kendall 1993) is then used to con-struct the models based on the aggregate elastic con-stants, assuming the geometry of the NPM. Thisprogram is capable of: (1) deriving slowness sur-faces and reflection coefficients between adjacentlithologies; (2) tracking compressional and shear-wave ray-paths in the geological situations envi-saged where the effects of seismic anisotropy maybe important; (3) providing estimates of the effectof crustal anisotropy on either passive or teleseismicdata; and (4) generating three-component (3C) syn-thetic seismograms from the ray-traced models,using the reflection coefficients and mode conver-sions derived earlier to aid interpretations.

Effects of mineralogy, foliation

and deformation

Mineralogy. Although CPO relationships betweenindividual mineral phases in polymineralic rocksare often complex (e.g. Fig. 4), it is possible to deter-mine the impact of each mineral phase on the bulkrock seismic properties. This is achieved by recog-nizing that individual mineral CPOs represent a‘recipe book’ from which rocks of different modalproportions but with the same measured CPO canbe constructed (e.g. Tatham et al. 2008; Lloydet al. 2009). A single mineral phase can thereforebe chosen and its modal proportion varied between0 and 100%, with the other minerals left to comprisethe residual composition in their relative (i.e.measured) modal proportions.

Using this ‘recipe’ approach, the impact of modalproportion on the seismic anisotropy of samplesB1, B2m and B2c has been considered (Fig. 5). It


is clear that micas, and especially biotite, make themain contribution to the bulk seismic anisotropy(see also Takanashi et al. 2001; Chlupacova et al.2003). Due to their generally weak CPO in thesesamples, feldspars and quartz impose a dilutingeffect on the anisotropy introduced by micas. Theabsence of CPO in any of these minerals in sampleB2c means that the cataclastic fault zone exhibitsvery little anisotropy (Fig. 5).

Foliation orientation. The results of CPO andseismic property analysis of samples collectedfrom the NPM (Fig. 4) suggest that the orientationand strength of foliation play crucial roles in theextent of shear-wave splitting (see also Lloydet al. 2011). From the previous section, sucheffects are likely to be exacerbated as the phyllo-silicate content increases. To investigate the impactof (mica-controlled) foliation orientation, a simpleelastic model was therefore designed (Fig. 6a).Although model width can be arbitrary (in thepresent models, the X and Y dimensions were setat 10 and +5 km, respectively), depth was set atZ ¼ 50 km because the NPM is believed to

involve the whole crust which is estimated to bec. 48 km thick (Butler et al. 2002). The model waspopulated with the elastic constants of sample B1,considered as representative of the early fabricstate of the deformed NPM orthogneisses. Theeffect of foliation orientation was investigated byrotating the elastic constants of sample B1 suchthat the foliation changed progressively from verti-cal (i.e. 08) to horizontal (i.e. 908). Figure 6b illus-trates the end-member foliation configurations.Finally, a shear-wave source was positioned nearthe base of the model within an isotropic layer toavoid complications associated with sources inanisotropic media.

Figure 6c shows the travel times versus offsetplots of the shear waves for the end-member hori-zontal and vertical foliations situations (Fig. 6b).The original shear wave is clearly split into twoshear waves for both cases. However, there areobvious differences in the travel times and offsetsfor the two foliation orientations. This behaviouris shown more clearly by considering the magnitudeof the shear-wave splitting for each increment offoliation orientation (Fig. 6d). From 0–458 there is

B1

AV

p %

B1

AV

s %

35

0 0

45quartzbiotitemuscoviteorthoclaseplagioclase

B2m

AV

p %

B2m

AV

s %

0

20

0

40

B2c

AV

p %

B2c

AV

s %

0

3.5

0

4.5

Modal %0 100 Modal %0 100

Fig. 5. Rock-recipe modelling of the effect of composition on P- and S-waves anisotropy (AVp and AVs respectively)for samples B1, B2m and B2c. Note different anisotropy scales per plot. The line ornament per mineral is the samefor each plot (see key, top left plot).


little variation, but beyond 458 there is a trend ofsteadily decreasing shear-wave splitting to aminimum of ,0.15 s at 908. There is therefore adifference of c. 1.3 s between the maximum andminimum shear-wave splitting for vertical andhorizontal foliation orientations.

Although the impact of foliation orientation onshear-wave splitting is perhaps intuitive, themaximum splitting actually occurs at 108 to the ver-tical (Fig. 6d). This specific behaviour reflectsthe detail present in both the CPO distributions

(Fig. 4a) and the plunge of the lineation within thefoliation plane. The splitting observed thereforedepends on CPO, foliation orientation and ray geo-metry, which suggests that small-scale petrofabricobservations can be effectively used as proxiesto investigate regional-scale geodynamics in thecrust (see below). Indeed, as the potential forc. 1.3 s of shear-wave splitting due to CPO islarger than the average SKS splitting observed(e.g. Silver 1996), this suggests that the crust canmake a significant contribution to SKS splitting.

15.50

15.00

14.50

14.00

Trav

el ti

me

(sec

)

15.50

15.40

15.25

15.30

15.35

15.45

horizontalfoliation

Model x (km)0 10

verticalfoliation

(c)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

She

ar-w

ave

split

ting

(sec

)

Foliation angle(degrees)

0vertical

90horizontal

(d)

Mod

el z

(km

)0

Model x (km)0 10

50

ray-

trace

d m

odel

4 km

(a)

(b)

horizontalfoliation:simpleshear

verticalfoliation:

pureshear

Fig. 6. Seismic modelling of the effect of foliation orientation on shear-wave splitting characteristics. (a) Seismicmodel: the ‘crust’ is populated with the elastic properties of sample B1, while the seismic source region is isotropic.(b) End-member foliation configurations. Left: horizontal (908); right: vertical (08). (c) Fast (solid lines) and slow(broken lines) shear-wave travel-time plots for end-member configurations. Upper: horizontal (908); lower: vertical(08). (d) Amount of shear-wave splitting in terms of foliation angle (08, vertical; 908, horizontal) for verticallypropagating shear waves.


Furthermore, values of 1.3 s splitting would be rela-tively easy to distinguish in local events or teleseis-mic P–S conversions at the Moho, and highlightsthe importance in considering such factors in bothglobal and local seismic surveys.

Deformation partitioning. Deformation partitioning(e.g. between pure and simple shear) can occur onall scales (e.g. Twiss & Moores 1992; Passchier &Trouw 1996). To investigate the effect of defor-mation partitioning on seismic properties, assumingsimilar rock properties rotated into different relativeorientations (e.g. vertical or pure-shear and inclinedor simple-shear components), a 40 km deep elasticmodel was designed in which the lowest 4 km rep-resents a basal isotropic layer containing a shear-wave source located at 38 km depth (Fig. 7a).Initially, the remaining 36 km of the model waspopulated with the elastic properties of sampleB2m rotated into the vertical to represent 100%horizontal pure shear. Next, a shallowly dipping2 km thick simple-shear zone (i.e. comparable inorientation and thickness to the Liachar ShearZone in the NPM) was incorporated using theelastic properties of sample B2m rotated into theappropriate orientation (Fig. 7b). However, the rela-tive proportions of pure and simple shear in themodel can be varied progressively by increasingshear-zone width.

The travel-time-offset behaviours for the twoend-member configurations (i.e. 100% pure-shearand 100% simple-shear elastic properties) indicateconsiderable shear-wave splitting for both modelsand a slight increase in travel times for the simple-shear model (Fig. 7c, d). In detail, the degree ofshear-wave splitting measured for each incrementof increasing shear-zone width shows a negative(almost linear) relationship with the amount ofsimple shear present in the model (Fig. 7e). Moreshear-wave splitting is generated by the 100% pure-shear model compared to the 100% simple-shearmodel, which again emphasizes the influence ofsteep foliations on shear-wave splitting (i.e.Fig. 6d). Furthermore, the pure-shear model inc-reases velocity, as shown by the c. 0.4 s differencebetween the arrivals of the S1-waves between thetwo models (Fig. 7c, d). However, the differencein shear-wave splitting is only c. 0.6 s due to therocks having similar seismic properties. Greaterdifferences in seismic properties between the con-stituent rocks are therefore expected to inducelarger differences in shear-wave splitting.

The results of these pure- and simple-shearmodels (Fig. 7) suggest that it may be possibleseismically to not only recognize deformation parti-tioning but also to distinguish between differentstyles of deformation at depth. However, thereare certain provisos as follows: (1) the geology is

well-constrained; (2) the CPO characteristics arepervasive over sufficient distances depending onseismic resolution; (3) the measurements of shear-wave splitting are well constrained; and, crucially,(4) there is sufficient contrast between host andshear-zone rocks in terms of foliation, fabric andstructural relationships (e.g. Lloyd et al. 2011).

Seismic wave propagation

It is known that seismic wave propagation is morecomplex in anisotropic compared to isotropic media,while waveform effects due to anisotropy may bedramatic and unexpected (e.g. Guest & Kendall1993). Wave fronts are no longer spherical and thedirections of particle motion, rays and wave-frontnormal are generally not aligned. However, theprimary effect of wave propagation from isotropicto anisotropic media is shear-wave splitting orseismic anisotropy (e.g. Crampin 1981), resultingin the separation of shear waves into two orthogonalquasi-shear waves (qSH and qSV). The results ofthe previous sections indicate that shear-wave split-ting is to be expected in many of the samples fromthe NPM.

To investigate the potential effects of (shear)wave propagation through the NPM, the programSLWVEL (Guest & Kendall 1993) was used to cal-culate the slowness surfaces (including polarizationvectors and group velocity surfaces) for samplesB1, B2m and B2c based on their Voight-Reuss-Hill (VRH)-averaged elastic constants via numeri-cal decomposition of the Christoffel equation intoeigenvalues (phase velocities) and eigenvectors(displacements). Slowness is the reciprocal ofphase velocity; a slowness surface is an envelopethat encompasses all of the slowness vectors andtherefore provides a description of elastic aniso-tropy (e.g. Lloyd & Kendall 2005). Slownesscurves are directly analogous to CPO-derived vel-ocity pole figures (e.g. Fig. 4).

In sample B1, both foliation-parallel and -normalsections exhibit variable slowness and velocity,with off-axis shear-wave splitting illustrated by thepresence of two shear-wave surfaces resulting inanisotropic behaviour and almost orthorhombicsymmetry (Fig. 8a). In contrast, sample B2m isalmost isotropic with only slight splitting and almostconstant slowness and velocity (Fig. 8c). SampleB2c is clearly isotropic and exhibits no shear-wavesplitting and constant slowness and velocity seismi-cally (Fig. 8b).

Seismic reflection coefficients

Up to six new reflected and transmitted secondarywaves may be generated when a ray encountersan interface between two different media, including


both fast and slow shear waves. A unique feature ofsuch reflected-converted waves in anisotropic mediais that they have energy at normal incidence due to adifference in group and phase velocity producedby low-symmetry and anisotropy, which never

exists in isotropic cases (e.g. Lloyd & Kendall2005).

As an example of this phenomenon, the seismicreflection coefficients derived from the VRH elasticconstants were calculated for P-wave velocities at

0.45

0.30

0.35

0.40

She

ar-w

ave

split

ting

(sec

s)

Volume of shear zone (%)0 25 50 75 100

(e)

Trav

el ti

me

(sec

)Tr

avel

tim

e (s

ec)

Model x (km)

Model x (km)0 10

10.7

11.0

11.3

0 10

11.1

11.6

(d)

(c) S1

S1

S2

S2

Mod

el z

(km

)0

Model x (km)0 10

40

ray-

trac

ed m

od

el

(a)

(b)

4 km

Mod

el z

(km

)0

Model x (km)0 10

40

ray-

trac

ed m

od

el

4 km

2 km

Fig. 7. Seismic modelling of the effect of deformation partitioning on seismic anisotropy, assuming similar rockproperties rotated into different relative orientations (e.g. vertical or pure-shear and inclined or simple-shearcomponents). (a) Seismic model with vertical foliation defined by sample B2m properties rotated into the verticalorientation. (b) Seismic model with vertical foliation cut by a shallow-dipping 2 km thick simple-shear zone defined bysample B2m properties rotated into the appropriate orientation. (c) Travel-time plot for seismic model comprising 100%vertical foliation. (d) Travel-time plot for seismic model comprising 100% shear zone. (e) Variation in amount ofshear-wave splitting with variation in simple-shear zone content for vertically propagating shear waves.


(a) Horizontal slownesscross-section

Vertical slownesscross-section

Horizontal velocitycross-section

Vertical velocitycross-section

-0.6-0.6

x2 s

low

ness

(s/

km)

x2 s

low

ness

(s/

km)

x1 slowness (s/km) x1 slowness (s/km)

0.6 0.6

0.6 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

0 0

0 0

-0.2 -0.2

-0.4-0.4

x2 v

eloc

ity (

km/s

)x2

vel

ocity

(km

/s)

x2 v

eloc

ity (

km/s

)

x1 velocity (km/s) x1 velocity (km/s)

6

6

6

6 6

4

4

4

4 4

2

2

2

2

2

2

0

0

0

0 0

-2

-2

-2

-4

-4

-4

-6

-6

-6

Horizontal slownesscross-section

Horizontal slownesscross-section

Horizontal velocitycross-section

Horizontal velocitycross-sectionB2-cat B2-myl

0.3 0.3

0.2 0.2

0.1 0.1

0 0

-0.1 -0.1

-0.2 -0.2

-0.3 -0.3

x2 s

low

ness

(s/

km)

x2 s

low

ness

(s/

km)

x1 slowness (s/km) x1 slowness (s/km)0.3 0.30.2 0.20.1 0.10 0640

x1 velocity (km/s)

(b)

P

P P

P P

P

PP

S

S

S1

S1 S1

S1

S1S1

S2

S2 S2

S2

S2S2

x2 v

eloc

ity (

km/s

)

6

4

2

0

-2

-4

-6

x1 velocity (km/s)42 60

(c)

Fig. 8. Examples of seismic slowness and velocity surfaces and polarization directions (arrows) for P- and S-waves (occurrence of two S-waves indicates shear-wavesplitting). (a) Sample B1. Left: foliation (XY ) parallel or ‘horizontal’ sections; right: foliation (XZ ) normal or ‘vertical’ sections. (b) Sample B2c foliation (XY ) parallelsections. (c) Sample B2m foliation (XY ) parallel sections.

FR

OM

CR

YS

TA

LT

OC

RU

ST

AL

63

an imagined interface between samples B2m andB2c. Energy (as defined by the total displacementratio) clearly exists for normal incidence P-wavesand P-to-P conversions, but not for P-to-S wave con-versions (Fig. 9). However, from 15–538 incidenceangles, energy associated with P-to-S conversions isgreater than that associated with P-to-P conversions.This behaviour implies that converted phases arelikely to contain information about the propertiesof the interface boundary (see also Lloyd &Kendall 2005). Such clear azimuthal variation inreflections suggests also that multi-azimuth wide-angle reflection data can be used to study sense ofdeformation in deep-rooted deformation zones. Fur-thermore, the amplitude of reflected shear wavesis sensitive to anisotropy (Helbig 1993, 1994).

The ability to calculate reflection and trans-mission coefficients at boundaries between mediawith different material properties can be used toaid the interpretation of seismic datasets (e.g.Guest et al. 1993). The veracity of such interpret-ations depends critically upon knowledge of theseismic properties of the rocks involved. Theseproperties can be determined from seismic datasets,laboratory techniques or, as in this contribution,

from CPO. However, to properly gauge the effectsof anisotropy on seismic wave propagation, isotro-pic ‘control models’ are needed for each sampleconstructed from its VRH-averaged elastic con-stants. Models using the isotropic elastic constantsonly consider the impedance contrast in a layerand at a boundary, and generally yield consistentlylower P- and S-wave velocities than those observedin the actual (i.e. anisotropic) cases. This aspect isconsidered in the next section.

Seismic reflection coefficients and controlled

source surveys

The models described so far have considered onlyshear-wave splitting where a CPO is persistentover a large region. Typical shear-zone widths willnot be detected in such seismic surveys, as theyare too narrow to have a significant effect on shear-wave splitting. However, it has been suggestedthat reflection coefficients yield greater verticalresolution than shear-wave splitting at boundariesbetween anisotropic media (Guest et al. 1993;Lloyd & Kendall 2005). An elastic model was there-fore designed to investigate the effects of anisotropy

Ph

ase 180

0

–180

1.0

0.5

0.0

Tota

l dis

pla

cem

ent

rati

o

sagittalplane

Incident phase angle (degs)

0 2010 30 40 50 60 70 80 90

critical angle: phase & waveformchanges (analytical problems)

normal seismic reflection data angles

wideangle

reflectionsurveys

normalincidence

(nb energy)PP>PS

PP>PS

PS>PP

P-waves

P-to-Swaves

XTectonic reference frameZ

Fig. 9. Example of seismic modelling of reflection coefficients and mode conversions for P-waves at an interfacebetween samples B2m and B2c in terms of energy (total displacement) and incidence wave angle (08, vertical; 908,horizontal). For normal incidence (i.e. teleseismic waves), energy exists for P-waves and P-to-P conversions but not forP-to-S conversions. From 15–538 incidence angles, P-to-S conversions energy is greater P-to-P conversions energy.Note also the significant increase in energies for wide-angle (.708) reflections.


on the reflection, transmission and conversioncoefficients at various angles of incidence forboundaries between the background rock of theNPM, represented by the elastic properties of sampleB1, and a cross-cutting shear zone (Fig. 10a). For thepurposes of this model, the shear zone was popu-lated with the elastic properties of another sample(Gn7) that can be proven in the field to have beenderived from the protolith as represented bysample B1 (e.g. Butler et al. 2002). For control pur-poses, an isotropic version of the model was alsoconstructed.

Two interfaces are present in the model, withrays shot from the surface passing from and inter-acting with B1-to-Gn7 and Gn7-to-B1. Reflectedwaves and mode conversions therefore originate atboth interfaces (Fig. 10a). The travel-time plotsfor the anisotropic cases are very similar for all azi-muths and indicate that fast and slow shear waves(i.e. shear-wave splitting) occur for both interfacesin the anisotropic models (Fig. 10b). In contrast,the isotropic control models for both wall-rock-to-shear-zone and shear-zone-to-wall-rock interfacesreveal relatively simple displacement-incidenceangle behaviours and no shear-wave splitting(Fig. 11a, b). Displacement (energy) increases pro-gressively with angle of incidence for P-waves atboth interfaces, although somewhat more rapidlyfor the wall-rock-to-shear-zone interface. Thedisplacement-incidence angle relationships for theshear waves are similar, with both showing zero

displacement at normal incidence and significantdisplacements at 30–408; there is considerablymore energy associated with the wall-rock-to-shear-zone interface, however. Displacements for bothinterfaces decrease to zero at c. 608 before increas-ing again significantly for wide-angle reflections.

Displacement-incidence angle behaviours forthe anisotropic model based on the petrofabric-derived elastic properties are significantly differentto the isotropic cases, with both interfaces exhibit-ing two shear waves and hence shear-wave splitting(Fig. 11c, d). In addition, the wall-rock-to-shear-zone and shear-zone-to-wall-rock interfaces alsoexhibit different behaviours, with the latter charac-terized by greater displacements for all three modeconversations. This suggests that it may be possibleto differentiate between interfaces via their modeconversion characteristics (e.g. Lloyd & Kendall2005). The P–S1 and P–S2 conversions for bothinterfaces show the greatest variation between 0–608 incidence angles, particularly for the formercase. In contrast, the P–P waves show little vari-ation with azimuth. Perhaps the best way to identifyanisotropic differences is therefore to observelarger azimuthal variations at different angles ofincidence (Guest et al. 1993; Lloyd & Kendall2005). Furthermore, from the reflection coefficientplots, high amplitudes are expected for theconverted waves.

To assist interpretation further by providinginformation on the amplitudes of seismic waves,

Azimuth = 0°

0 1 2 3 4 5Model x (km)

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

T rav

el t

ime

(sec

s)

P-P

P-P

wall rock to shear zone interface

shear zone to wall rock interface

P-S1

P-S1

P-S2

P-S2



0 1 2 3 4 5Model x (km)

Azimuth = 30°

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

Trav

el t

ime

(sec

s)

P-P

P-P

P-S1

P-S1

P-S2

P-S2



0 1 2 3 4 5Model x (km)

Azimuth = 60°

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

Trav

el t

ime

(sec

s)

P-P

P-P

P-S1

P-S1

P-S2

P-S2



0 1 2 3 4 5Model x (km)

Azimuth = 90°

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

Trav

el t

ime

(sec

s)

P-P

P-P

P-S1

P-S1

P-S2

P-S2

0 2 4Model x (km)

(a)

(b)

* 0

2

4

6

Dep

th k

m

B1

G7

B1

Fig. 10. Example of seismic modelling of reflection coefficients and mode conversions at the interfaces between a2 km wide shear zone (sample Gn7 elastic constants) and its wall rock (sample B1 elastic constants). (a) Elastic modeland ray paths. (b) Travel-time plots for anisotropic model interfaces for azimuths of 08, 308, 608 and 908.


and in particular their mode conversions, the elasticmodel shown in Figure 11a was used to producesynthetic three-component (3C) seismograms forthe two interfaces (Fig. 12). In both cases, the CPO-derived elastic constants were rotated into thecorrect geographical orientation for 0–908 azi-muths. The presence of the two interfaces is seenclearly in the 3C-seismograms, although P–P con-versions are only recognized on the vertical com-ponent. In particular, they show considerableS-wave converted phase energy on the transversecomponent. This behaviour is indicative of aniso-tropy as there is no energy on the transverse

component in an isotropic case. However, themain conclusion to be drawn is that boundariesbetween tectono-lithological units are likely toproduce significant azimuthal variations (as wellas polarity reversals) in 3C-seismograms.

Deformation determination from reflection

coefficients and mode conversions

Three elastic models were designed to test whetherit is possible to differentiate between different typesof CPO-dependent deformations via seismic aniso-tropy. All three models were populated with the

(a)

(b)

(c) (d)

dis

pla

cem

ent

0

0

0

0

0

20 40 60 80

0.01

0.02

0.03

angle of incidence

S-waves

0.50

1.00

0 20 40 60 80

0.12

0.06

dis

pla

cem

ent

dis

pla

cem

ent

angle of incidence

P-waves

S-waves

0.40

0.30

0.20

0.10

dis

pla

cem

ent P-waves

normalincidence

0

0.2

dis

pla

cem

ent P-P

0.2

dis

pla

cem

ent P-S1

dis

pla

cem

ent

0.2

00

0

20 40 60 80angle of incidence

P-S2

angle of incidence

0.50

1.00

dis

pla

cem

ent P-P

0.01

0.02

0.03

dis

pla

cem

ent P-S1

20 40 60 8000

0.01

0.02

0.03

dis

pla

cem

ent P-S2

0°30°

60°90°

azimuths

Fig. 11. Example of seismic modelling of reflection coefficients and mode conversions at the interfaces between a 2 kmwide shear zone (sample Gn7 elastic constants) and its wall rock (sample B1 elastic constants) for azimuths of 08, 308,608 and 908 (anisotropic cases only). (a) Isotropic reflection coefficients for P- and S-waves at the wall-rock-to-shear-zone interface. (b) Isotropic reflection coefficients for P- and S-waves at the shear-zone-to-wall-rock interface.(c) Anisotropic reflection coefficients for P–P, P–S1 and P–S2 waves for different angles of incidence at the wall-rock-to-shear-zone interface (note key below). (d) Anisotropic reflection coefficients for P–P, P–S1 and P–S2 waves fordifferent angles of incidence at the shear-zone-to-wall-rock interface (note key below).


elastic constants of sample B2c to represent defor-mation due to brittle faulting in the uppermostcrust. In the first model, a second lower layer waspopulated with the elastic constants of sample B1rotated such that the foliation was horizontal torepresent a simple-shear deformation (Fig. 13a). Incontrast, the lower layer of the second model waspopulated with the elastic constants of sampleB1 rotated so the foliation was vertical to representa pure-shear deformation (Fig. 14a). The thirdmodel was a hybrid of the other two models andrepresented a transition from coaxial pure-sheardeformation at depth into partitioned and localizedzones of ductile simple shear and brittle faultingdeformations at progressively shallower levels(Fig. 15a). The thickness of each layer was

arbitrarily chosen at 2 km, although this is not sig-nificant because it was shown previously that reflec-tions and mode conversions are most sensitive to theproperties of the interface. A seismic source wasplaced at the surface in the centre of each model,from which travel times (Figs 13b, 14b & 15b),geometrical spreading reflection coefficients andmode-converted reflections for P–P, P–S1 andP–S2 were calculated (Figs 13c, 14c & 15c)and used to construct 3C-synthetic seismograms(Figs 13d, 14d & 15d).

Although travel times do not differ betweenModels 1 and 2 and there is little or no shear-wavesplitting (Figs 13b & 14b), the 3C-seismogramsshow much greater variation. For Model 1 (Fig.13d), P-wave and mode-converted S-waves are

0

0

0

2.0

2.0

2.0

1.0

1.0

1.0

Trav

el t

ime

(sec

s)Radial

Transverse

Vertical

0 500 1000 1500 2000 2500

Offset (m)

(a)

0

0

0

2.0

2.0

2.0

1.0

1.0

1.0

Trav

el t

ime

(sec

s)

Radial

Transverse

Vertical

0 500 1000 1500 2000 2500

Offset (m)

(b)

0

0

2.0

2.0

1.0

1.0

1.0

Trav

el t

ime

(sec

s)

Radial

Transverse

Vertical

0 500 1000 1500 2000 2500

Offset (m)

0

2.0

(c)

0

2.0

2.0

1.0

1.0

1.0

Trav

el t

ime

(sec

s)

Radial

Transverse

Vertical

0 500 1000 1500 2000 2500

Offset (m)

0

2.0

0

(d)

Fig. 12. Example of seismic modelling of reflection coefficients and mode conversions at the interfaces between a2 km wide shear zone (sample Gn7 elastic constants) and its wall rock (sample B1 elastic constants) – travel-timegraphs and three-component (3C) synthetic seismograms for wall-rock-to-shear-zone-to-wall-rock (i.e. B1–Gn7–B1)interfaces using geographic elastic constants at azimuths of (see also Fig. 10b): (a) 08; (b) 308; (c) 608; and (d) 908. Note:(1) the presence of the two interfaces; (2) P–P conversions on the vertical component only; (3) considerable S-waveconverted phase energy on the transverse component; and (4) significant azimuthal variations and polarity reversals atthe boundaries between the tectono-lithological units.


present on the radial component, with the latterhaving the greatest amplitude. Only mode-converted waves are seen on the transverse com-ponent, while the P-wave is prominent on thevertical component with a small amount of mode-converted energy. There is significant variation inthe reflection coefficients, especially on the modeconversions but less so on the P reflection. In con-trast, Model 2 (Fig. 14d) has much less energy onthe vertical component of the 3C-seismogram. TheP-wave has the highest amplitude which decreaseswith offset, while the S-wave increases withoffset. There is a large amount of mode-convertedenergy on the transverse component. The radialcomponent has both P- and S-waves, although theformer is very small and the latter are far more pro-minent. Unsurprisingly, Model 3 shows the sameresults as Model 1 for the reflections from theupper interface between the cataclasite and horizon-tal foliation (compare Figs 13d & 15d). However,the lower interface between the horizontal and

vertical foliations has the most energy on the trans-verse component, although less than in Model 2(Fig. 14d) which corresponds to the mode-convertedwaves. Mode conversions are seen more clearly onthe radial component than the P-waves, which onlyhave a small amount of energy that disappears by600 m offset. The vertical component has onlyP-wave energy and this is stronger than that observedin Model 2. The transition from horizontal to verti-cal foliation therefore appears to enhance the P-waveon the vertical component and reduce the S-wave onthe transverse component of the seismogram.

It appears from the three models (Figs 13–15)that differences in amplitude of P-wave and mode-converted phases may help to determine defor-mation style. However, it is important to recognizethat the models represent simplified geometriesand consider only three types of interface. Forexample, as the inclination of an interface increases,there will be some energy on the transversecomponent. The seismic models described here

P-S1 reflectioncoefficients

0.1

0.2

0.3

Dis

pla

cem

ent


0

0.1

0.2

0.3

P-P reflectioncoefficients

0.1

0.2

0.3

0 20 40 60 80Incidence angle

0 degs30 degs60 degs90 degs

Dep

th k

m

0

2

4

B2-

cat

B1-

ho

riz

(a)

(d)

(b) (c)

T rav

el-t

ime

(s) 0.6

0.7

0.8

0.9

1.0

1.1

Model x (km)0 1 2 3 4 5

P-P

P-S1P-S2

00.51.01.52.0

2.5

00.51.01.52.0

2.5

0

0.51.01.52.02.5

Trav

el-t

ime

(s)

Radial

Transverse

Vertical

0 0.5 1.0 1.5 2.0 2.5 3.0Offset (km)

Fig. 13. Determination of CPO-dependent deformation styles using seismic reflection coefficients and modeconversions. Model 1: brittle cataclastic fault overlying horizontal ductile (simple-shear) foliation. (a) Model geometryand ray tracing. (b) Travel times for P–P, P–S1 and P–S2 wave conversions. (c) Three-component (3C) syntheticseismograms. (d) Reflection coefficients and mode-converted reflections for P–P, P–S1 and P–S2 waves (key is thesame for all plots).


therefore show only the potential of this approachrather than precise solutions. Nevertheless, theresults are encouraging and illustrate how the useof arrays of 3C-seismometers is crucial in enhancingunderstanding of crustal anisotropy. In the nextsection the experience gained from the CPO-derivedseismological modelling is used in an interpretationof the seismology and tectonics of the NPM.

Discussion

This section assesses whether it is possible to useCPO-derived seismic properties and models to dis-criminate between the different models to explainthe kinematics and geodynamics of the NPM. Itbegins with a consideration of the known seismol-ogy of the NPM before considering the knowntectonic configuration prior to a test of the pure-versus simple-shear models based on the lessonsand results gained earlier in this contribution.

NPM seismology

There have been several studies of the seismiccharacteristics of the NPM based on either directmeasurements of natural seismicity (e.g. Meltzeret al. 2001; Weeraratne et al. 2004) or ultrasoniclaboratory measurements of samples collectedfrom the region (e.g. Meltzer & Christensen2001). Natural seismicity indicates Vp and Vsvalues of 5.5–6.5 and 3.0–3.7 km/s respectively,similar to the ultrasonically measured values afterall (expansion) cracks have closed (Fig. 16a).These values are in excellent agreement with theCPO-derived measurements obtained in this study(see also Fig. 4). However, although there is goodagreement between ultrasonic- and CPO-derivedVp anisotropy estimates, the former indicate con-siderably greater shear-wave splitting (Fig. 16b). Ithas been shown previously that variation in micacontent is the main control on the amount of

0

2

4

Dep

th k

m

B2-

cat

B1-

vert

(a)

(d)

(b)(c)

)

00.51.01.52.02.5

00.51.01.52.02.5

00.51.01.52.02.5

Trav

el-t

ime

(s)

Radial

Transverse

Vertical

0 0.5 1.0 1.5 2.0 2.5 3.0Offset (km)

Trav

el-t

ime

(s) 0.6

0.7

0.8

0.9

1.0

1.1

Model x (km)0 1 2 3 4 5

P-P

P-S1P-S2



0.1

0.2

0.3


0.1

0.2

0.3

Dis

pla

cem

ent


0.1

0.2

0.3


Fig. 14. Determination of CPO-dependent deformation styles using seismic reflection coefficients and modeconversions. Model 2: brittle cataclastic fault overlying vertical ductile (pure-shear) foliation. (a) Model geometry andray tracing. (b) Travel times for P–P, P–S1 and P–S2 wave conversions. (c) Three-component (3C) syntheticseismograms. (d) Reflection coefficients and mode converted reflections for P–P, P–S1 and P–S2 waves (key is thesame for all plots).


anisotropy exhibited, with foliation orientation anddevelopment also making contributions (Figs 5–7;see also Lloyd et al. 2009, 2011). As the micacontent is up to 10% greater in the samplesmeasured via ultrasonics (Meltzer & Christensen2001), the discrepancy between the two estimatesis considered to be mainly due to differences inmineral composition between the two sample sets.

Meltzer & Christensen (2001) estimated c. 1.5 sof shear-wave splitting for a 40 km thick crust withvertical foliation, via ultrasonic measurements. Thisvalue is only slightly greater than the c. 1.3 s esti-mated from the CPO-derived model (Fig. 6).Again, the small discrepancy can be explained bydifferences in mica content and/or foliation devel-opment/orientation (Figs 5–7). As splitting delaytimes increase with distance travelled through ani-sotropic material, they can provide a means of

mapping rock fabric at depth (e.g. Kern & Wenk1990). However, the range of delay times can alsobe influenced by compositional heterogeneity,lateral variation in anisotropy, changes in regionalfoliation orientation and velocity variance due tonon-axial propagation through a wide range ofevent-station azimuths, as shown in this contri-bution. Meltzer et al. (2004) however argue thatbecause the composition of the NPM is basicallyhomogeneous and its structure is well constrained,while ray-paths are restricted to the crust andsource-receiver geometries sample a range of azi-muths with respect to structure, seismic data isideal for studying and quantifying the affect of non-axial propagation through the regional foliation.They argue further that as current tomographycodes do not generally account for anisotropiceffects, and may potentially under- or overestimate

* 0(a)

(b)

(c)

(d)

2

4

6

Dep

th k

m

B2-

cat

B1-

ho

riz

B1-

vert

T rav

el-t

ime

(s)

0.6

0.8

1.0

Model x (km)0 1 2 3 4 5

1.2

1.4

1.6

1.8

2.0

P-P

P-P

P-S1P-S2

P-S1, P-S2

upperinterface

lowerinterface

00.51.01.52.02.5

00.51.01.52.02.5

00.51.01.52.02.5

Trav

el-t

ime

(s)

Radial

Transverse

Vertical

0 0.5 1.0 1.5 2.0 2.5 3.0Offset (km)


0.1

0.2

0.3



0.1

0.2

0.3

Dis

pla

cem

ent


0

0.1

0.2

0.3


Fig. 15. Determination of CPO-dependent deformation styles using seismic reflection coefficients and modeconversions. Model 3: brittle cataclastic fault overlying horizontal ductile (simple-shear) foliation overlying vertical(pure-shear) foliation. (a) Model geometry and ray tracing. (b) Travel times for P–P, P–S1 and P–S2 waveconversions. (c) Three-component (3C) synthetic seismograms. (d) Reflection coefficients and mode convertedreflections for P–P, P–S1 and P–S2 waves (key is the same for all plots).


velocity structure in the crust, this type of analysishas important implications for understandingcrustal dynamics. In particular, Vp, Vs and Vp/Vsratios are typically used to infer both lithologyand rheology of subsurface materials providing con-straints for thermo-mechanical models of defor-mation (e.g. Christensen & Mooney 1995).

The potential for 1.3–1.5 s of shear-wave split-ting due to CPO-induced seismic anisotropyshould be easy to distinguish via natural seismicity,such as local events or teleseismic P–S conversionsat the Moho. Indeed, Weeraratne et al. (2004)observed such strong anisotropy in teleseismic andregional shear phases from a seismic array deploy-ment across the NPM. Stations outside the NPMrecorded 1.5–2.3 s delay times, with WNW–ESEfast directions in SKS and related core phases;regional S-phases from the Hindu Kush withsource depths of 200–300 km had similar c. east–west fast directions but delay times of ≤0.5 s. Asthe depth range sampled by the latter lies mainlyin the high-velocity lithosphere (which extends to.200 km beneath the NPM), it appears that thelithospheric contribution to the total shear-wavesplitting observed for the teleseismic phases isonly c. 0.5 s with c. 1.0–1.5 s originating in the sub-lithospheric mantle (Weeraratne et al. 2004). Fur-thermore, while SKS paths from a wide range ofback-azimuths produce null measurements withinthe interior of the NPM, laboratory studies ofgneiss samples suggest that as much as 21% shear-wave anisotropy with north–south fast axis mayexist in the crust (Meltzer & Christensen 2001). Inaddition, mantle lithosphere deformation consistentwith east–west compression of the NPM may alsocontribute to shallow north–south anisotropy(Weeraratne et al. 2004). The null observations inthe NPM interior may therefore be due to themutual cancellation of north–south and east–westshear-wave splitting effects (see below).

NPM tectonics

Seismic velocities and velocity ratios are typicallyused to infer both subsurface lithology and rheol-ogy, thereby providing constraints on thermo-mechanical models of deformation, tectonics andgeodynamics. A prominent low-velocity zone (com-pared to surrounding regions) has been recognizedbeneath the NPM and extends through the crustinto the upper mantle (e.g. Meltzer et al. 2001,2004). One explanation for this behaviour, espe-cially as composition is considered to be effectivelyhomogeneous, is the presence of hot rocks at depth.The variable seismic waveforms and slightly lowerVp/Vs ratios observed therefore suggest the exist-ence of super-critical fluids in small regions oflimited extent, which are more consistent with

partial melts and/or aqueous fluids rather than thepresence of large magma bodies. This is supportedby a shallow (c. 2–5 km bsl) brittle–ductile tran-sition that bows towards the surface under theNPM, which is consistent with rapid advectionfrom depth of hot crust into the massif along shal-low detachments. Furthermore, magnetotelluricsindicate that the lower crust is atypically resistive(Park & Mackie 1997, 2000). Each of these anom-alies exhibits a ‘bulls-eye’ pattern centred on theNPM. It has therefore been suggested that the mag-nitude and extent of the low-velocity zone withinthe NPM constrains crustal flow paths, thereby focus-ing exhumation and concentrating crustal strain andpotential zones of partial melting in the crust. Thisled Zeitler et al. (2001) to describe the geodynamicsof the NPM as a ‘tectonic aneurysm’.

In contrast to the ‘tectonic aneurysm’ modelinvolving hot rocks, high thermal gradients and/orpore pressures in typical plutonic and/or meta-morphic rocks, Meltzer & Christensen (2001) pro-posed an alternative explanation for the seismicstructure and characteristics of the NPM. Theysuggest that as the mid-lower crust is typicallylayered with well-defined foliations and fabrics onmultiple scales, in situ velocities from refracted orturning rays that spend substantial portions of theirtravel paths either parallel or normal to the foliationplane may systematically either over- or under-estimate seismic velocities. Whereas observedvelocities of 6.0–6.5 km/s that are interpretednormally as indicating rocks of intermediate compo-sition could also be indicative of waves propagating(sub-) parallel to foliation in felsic rocks, observedvelocities of 5.6–6.0 km/s could indicate propa-gation at high angles to foliation in similar litholo-gies (e.g. Figs 6 & 7; see also Lloyd et al. 2011).Such observed velocity variations only requirechanges in foliation properties (i.e. deformation)with depth rather than composition. Crucial to thevalidation of this alternative model are the seismicanisotropy characteristics of the NPM. AlthoughWeeraratne et al. (2004) observed up to 1.5 s ofshear-wave splitting within the NPM, they alsorecognized null observations within the interior ofthe massif, from which they derived a two-layeranisotropic model with north–south anisotropy inthe crust and lithosphere due to east–west com-pression of the Nanga Parbat orogen cancellingsplitting from c. east–west sublithospheric aniso-tropy (see above).

The results of the CPO-derived seismologicalmodels described in this contribution can be usedto test (although perhaps as yet only qualitatively)the various tectonic models for the NPM. The differ-ences in amplitude of P-wave and mode-convertedphases on the transverse component help to dis-tinguish deformation style, while mode conversions


(a)

(b)

7

6

5

4

3

B2m

B2m

Vel

oci

ty (

km/s

)

5

5

10

10

0

0

15

15

20

20

25

25

30

30

Depth (km)

Depth (km)

2

natu

ral

Vp

seis

mic

ity

natu

ral

Vs

seis

mic

ity

experimentalVp values

experimentalVs values

closureof

cracksetc.

B1

B2m

B2c

An

iso

tro

py

(%)

0

10

20

expe

rim

enta

l AV

p-m

ax

expe

rimen

tal A

Vs-

max

B1

B1

B2c

B2c

max

min

max

min

s1 s2

AVp

AVp

AVp

AVs

AVs

AVs

typical middle crust Vp

Vp/

Vs

= 1

.60

- 1.

95

Vp/

Vs

= 1

.69

- 1.

86

Vp/

Vs

= 1

.79

- 1.

81

Fig. 16.


and transverse component energy are diagnostic ofthe degree and orientation of foliation (which havedramatic effects on seismic waves). Steeply inclinedfoliation produces large mode-converted amplitudeson the transverse component of 3C-seismograms aswell as significant shear-wave splitting, comparedto shallowly inclined foliation. Such differencesshould be easy to distinguish in local events or tele-seismic P–S conversions at the Moho. Althoughhigh-amplitude P-to-S-wave converted reflectionshave been observed in wide-angle reflection datain the Himalayas, they have been interpreted as indi-cating the presence of partial melt accumulationbecause they are believed to be uncharacteristic ofcrustal reflections (e.g. Makovsky et al. 1996).However, the results of the elastic models presentedhere suggest that they can indeed be associatedwith interfaces between different lithologies and/or different deformation styles. These results notonly indicate the importance of considering seis-mic anisotropy in crustal seismic surveys, but alsothat it should be possible to discriminate betweendifferent deformation processes active at depthusing seismic measurements.

As a first approximation, the results of the CPO-derived seismological models described in this con-tribution appear to be more consistent with theMeltzer & Christensen (2001) model for NPMrather than other models such as the ‘tectonic aneur-ysm’ model of Zeitler et al. (2001).

NPM: pure- versus simple-shear tectonics

This contribution has referred throughout to twosimple alternatives (perhaps ‘end-member’ models)for the tectonics of the NPM, namely the penetrativepure-shear model (e.g. Butler et al. 2002) and thelocalized simple-shear model (e.g. Burg 1999). Ifpure shear is persistent at depth, it would require asteep to subvertical fabric to persist throughout theNPM due to horizontal shortening and verticalextension. In contrast, the simple-shear modelwould induce a crystal alignment in the directionof shearing at much shallower angles, perhapswith much narrower length scales measured in the(sub-) vertical sense. The question therefore arisesas to whether it is possible to discriminatebetween the two models and hence to make infer-ences about the kinematics using real seismic data

based on the experienced gained via the CPO-derived seismic modelling described above. Tobegin to answer this question it is necessary to con-sider the effect of foliation orientation on the magni-tude of shear-wave splitting, as steep foliationsresult in considerably greater shear-wave splittingthan shallow foliations (i.e. Figs 6 & 7).

Butler et al. (2002) envisaged an increasingstrain gradient from the core of the NPM towardsthe northwest, within which subvertical pure-shearstretching flanks and passes into an inclined simple-shear zone (e.g. Fig. 17a). The impact of the changesin foliation orientation due to this strain gradient onshear-wave splitting can be assessed from the resultsof the seismic modelling described previously.From Figure 6d, most of the NPM (including itscore) is dominated by steep vertical foliations andtherefore should exhibit .1.0 s shear-wave splitting(e.g. Fig. 17b). However, a zone of lower shear-wave splitting values (i.e. ,1.0 s) should also beobserved, particularly near to the surface towardsthe northeast (e.g. Figs 1 & 2).

In practice, the regional variation betweenpure-shear and simple-shear tectonics shown inFigure 17a can be recognized on all scales. Forexample, anastomosing simple-shear zones occuron many scales and separate domains of relativelylow deformation that may be described as pure shear.Furthermore, many tectonites exhibit S–C foliationsin which planar C-surfaces of concentrated defor-mation (i.e. ‘shear zones’) separate broader regionsor ‘lithons’ containing S-surfaces inclined to theC-planes (e.g. Lister & Snoke 1984). Such scale-invariant pure-shear–simple-shear deformation geo-metries are illustrated in Figure 17a, superimposedupon the regional tectonic traverse across the NPM.Their impact on the shear-wave splitting character-istics can be inferred from the seismic model des-cribed in Figure 7. This model considered the effectof varying the proportion of sample B2m elasticproperties Elastic properties, which can be con-sidered as a proxy for a simple-shear zone, to sampleB1 elastic properties, which can be considered as aproxy for a pure-shear fabric. Together, theseproxies could also represent various combinationsof S–C foliation intensities (see also Lloyd et al.2009). As the proportion of shear zones/S–C foli-ation increases, the amount of shear-wave splittingdecreases (Fig. 7e). Thus, variations in shear-wave

Fig. 16. Comparison of seismic properties for the NPM. (a) Compressional (Vp) and shear (s1, s2) wave velocities:rainbow bars, CPO-derived values for samples B1, B2m and B2c (see Fig. 4); rectangular boxes, natural velocitiesmeasured in situ at NPM (Meltzer et al. 2001); solid and broken curves, ranges of experimental ultrasonic values forNPM samples (Meltzer & Christensen 2001). Also shown are the Vp/Vs ratio ranges per sample and the typical valuesof Vp expected in the middle crust according to Rudnick & Fountain (1995). (b) Compressional (AVp) and shear (AVs)wave anisotropies: open circles, absolute AVp values; black bars, range of AVs values; solid and broken double arrows,experimentally measured AVp and AVs values (Meltzer & Christensen 2001) (higher AVs values reflect higher micacontents and/or stronger mica CPO).


Fig. 17.


splitting characteristics must also reflect the defor-mation state in terms of the development of shearzones and/or S–C foliations on the scale of theseismic wavelength and/or the volume of rock pene-trated (Fig. 17b).

On the basis of these data only, it appears that thetectonics and kinematics of the NPM can beexplained by the model of Butler et al. (2002).Thus, pervasive pure-shear vertical stretching dom-inates the tectonics and is responsible for most ofthe exhumation and topography of the NPM (e.g.Fig. 17a). Superimposed onto this fabric is a loca-lized (simple) shear and brittle faulting deformation,particularly near to the surface. However, confir-mation of the viability of this model requires morefield data, both geological and seismological. In par-ticular, 3C-seismic array studies of reflected wavesand mode conversions could be used to recognizeand distinguish the two styles of deformation, andhence strain gradients, due to deformation changesand associated variations in petrofabric-derivedelastic constants.

Conclusions

(1) Polymineralic high-strain zones responsiblefor large-scale orogenic displacement havedifferent CPO and seismic properties dep-ending on whether they are mylonitic orcataclastic.

(2) ‘Mylonitic’ shear zones exhibit specific non-random CPO and seismic characteristics.Girdle distributions develop typically parallelto mylonitic foliation, while maxima and/orminima in CPO develop parallel to myloniticlineation. Such CPO characteristics result inanisotropic elastic properties and hence rela-tively large seismic anisotropy, with AVsgirdles and AVs-maxima developing parallelto mylonitic foliation and lineation, respect-ively. The Vs1-min is typically normal to themylonitic foliation.

(3) Where ‘mylonitic’ shear zones have evolvedfrom a previously deformed (gneissic) proto-lith, they may exhibit uncharacteristicallylow seismic anisotropy compared to theirwall rocks. It is therefore not sufficient touse high anisotropy as indicative of localizedzones of concentrated ductile deformation.

(4) Cataclastic fault zones exhibit random CPOfabrics and hence very weak seismic aniso-tropy with isotropic symmetry, although afracture-related and/or fault cement CPOcan develop. The maximum seismic velocitiesprobably lie in the fault plane and Vs2-maxmay develop parallel to the fault movementdirection.

(5) Seismic modelling and ray-tracing has shownthat it is possible to recognize differences indeformation style. Foliation orientation andintensity have dramatic effects on seismicwave propagation. A rock with 10% AVsand a vertically aligned foliation persistentthroughout a 40 km thick crust inducesc. 1.2 s of shear-wave splitting compared toa horizontal alignment that induces only0.2 s splitting. This degree and/or differencein shear-wave splitting caused by CPO-induced anisotropy should be easy to dis-tinguish in local events or teleseismic P–Sconversions at the Moho.

(6) The model results also highlight the impor-tance of considering anisotropy in crustalseismic surveys. Inclining foliation to matchgeological observations at the surface indi-cates that relative differences in shear-wavesplitting may permit a strain gradient tobe measured and mapped using seismicmeasurements.

(7) The results of the petrofabric-derived seismo-logical models are consistent with a pervasivepure-shear vertical stretching model for thetectonics of the NPM (e.g. Butler et al.2002), rather than the conventional shear andfault zones localization model (e.g. Burg1999). In addition, they challenge the inter-pretation of low velocities beneath the NPMin terms of partial melts and hence castdoubt on the concept of ‘tectonic aneurysm’.

(8) Petrofabric-derived seismological modellingrepresents a combination of micro–meso–macro scale observations that can provide aquantification of the petrophysical propertiesinvolved in large-scale geodynamic pro-cesses. Small-scale petrofabric observationspotentially can be effectively used as aproxy to investigate regional-scale geody-namics in the crust. However, this contri-bution has concentrated only on the input

Fig. 17. Relationship between deformation-induced shear-wave splitting and tectonics of the NPM according tothe pure-shear vertical stretching model of Butler et al. (2002). (a) Symmetrical regional subvertical stretchingpassing into a restricted zone dominated by simple shear. Note also variation in pure-shear (P) and simple-shear (S)S–C foliations (Pc, Ps, Sc and Ss respectively) and/or shear zones that can occur on all scales. (b) Effect offoliation orientation on magnitude of shear-wave splitting in terms of relative proportions of pure- (P) and simple- (S)shears and S–C foliations. Shaded regions indicate range and relative probabilities of splitting for each type ofdeformation.


from CPO. Other (micro-) structural elementsare also known to contribute to the seismicproperties of rocks (e.g. grain shape, fractures,lithological layering, grain boundaries, etc.)and must be included for a complete analysis.

Fieldwork to Nanga Parbat was funded by a Royal Societyresearch grant (RWHB). JH thanks the UK NERC forMRes funding. Part of the SEM/EBSD facilities was sup-ported by the UK NERC Small Grant GR9/3223 (GEL,MC). We are grateful to K. Michibayashi, an anonymousreviewer and the Special Editor, D. Prior, for theirreviews and comments that have helped to improve theoriginal version of this manuscript.

References

Babuska, V. & Cara, M. 1991. Seismic Anisotropy in theEarth. Kluwer Academic, Dordrecht.

Barruol, G. & Mainprice, D. 1993. 3D seismic vel-ocities calculated from CPOs and reflectivity of alower crustal section – example of the Val Sesia(Ivrea Zone, Northern Italy). Geophysical JournalInternational, 115, 1169–1188.

Barruol, G. & Kern, H. 1996. Seismic anisotropy andshear-wave splitting in lower-crustal and upper-mantlerocks from the Ivrea Zone – Experimental and calcu-lated data. Physics of Earth and Planetary Interiors,95, 175–194.

Ben Ismail, W. & Mainprice, D. 1998. An olivine fabricdatabase: an overview of upper mantle fabrics andseismic anisotropy. Tectonophysics, 296, 145–158.

Bunge, H. J. 1982. Texture Analysis in Materials Science.Butterworths, London.

Bunge, H. J., Kiewel, R., Reinert, T. & Fritsche, L.2000. Elastic properties of polycrystals – influenceof texture and stereology. Journal of the Mechanicsand Physics of Solids, 48, 29–66.

Burg, J.-P. 1999. Ductile structures and instabilities: theirimplications for Variscan tectonics in the Ardennes.Tectonophysics, 315, 251–267.

Burlini, L. & Kern, H. 1994. Special issue: seismic prop-erties of crustal and mantle rocks – Laboratorymeasurements and theoretical calculations. Surveys inGeophysics, 15, 439–672.

Burlini, L., Marquer, D., Challandes, N., Mazzola,S. & Zangarini, N. 1998. Seismic properties of highlystrained marbles from the Splugenpass, central Alps.Journal of Structural Geology, 20, 277–292.

Butler, R. W. H. & Prior, D. J. 1988. Tectonic controlson the uplift of Nanga Parbat, Pakistan Himalaya.Nature, 333, 247–250.

Butler, R. W. H., Wheeler, J., Treloar, P. J. & Jones,C. 2000. Geological structure of the southern part ofthe Nanga Parbat massif, Pakistan Himalaya, and itstectonic implications. In:Khan, M. A., Treloar,P. J., Searle, M. P. & Jan, M. Q. (eds) Tectonics ofthe Nanga Parbat Syntaxis and the Western Himalaya.Geological Society, London, Special Publications,170, 123–136.

Butler, R. W. H., Casey, M., Lloyd, G. E., Bond, C. E.,McDade, P., Shipton, Z. & Jones, R. 2002. Vertical

stretching and crustal thickening at Nanga Parbat,Pakistan Himalaya: a model for distributed continentaldeformation during mountain building. Tectonics, 21,doi: 10.1029.2001TC901022.

Chlupacova, M., Skacelova, Z. & Nehybka, V. 2003.P-wave anisotropy of rocks from the seismic area inWestern Bohemia. Journal of Geodynamics, 35,45–57.

Christensen, N. I. 1971. Fabric, seismic anisotropy andtectonic history of the Twin Sisters dunite, Washing-ton. Geological Society of America Bulletin, 82,1681–1694.

Christensen, N. I. 1984. The magnitude, symmetry andorigin of upper mantle anisotropy based on fabric ana-lyses of ultramafic tectonites. Geophysical JournalRoyal Astronomical Society, 76, 89–111.

Christensen, N. I. & Mooney, W. D. 1995. Seismicvelocity structure and composition of the continentalcrust: a global review. Journal of GeophysicalResearch, 100, 9761–9788.

Coward, M. P. 1994. Continental collision. In: Hancock,P. L. (ed.) Continental Deformation. Pergamon,New York, 264–288.

Crampin, S. 1981. A review of wave motion in anisotropicand cracked elastic media. Wave Motion, 3, 242–391.

Crosson, R. S. & Lin, J. W. 1971. Voigt and Reuss pre-diction of anisotropic elasticity of dunite. Journal ofGeophysical Research, 76, 570–578.

Davis, P., England, P. & Houseman, G. 1997.Comparison of shear wave splitting and finite strainfrom the India–Asia collision zone. Journal of Geo-physical Research-Solid Earth, 102, 27 511–27 522.

George, M. T., Harris, N. B. W. & Butler, R. W. H.1993. The tectonic implications of contrasting granitemagmatism between the Kohistan arc and the NangaParbat-Haramosh Massif, Pakistan Himalaya. In:Treloar, P. J. & Searle, M. P. (eds) HimalayanTectonics. Geological Society, London, SpecialPublications, 74, 173–191.

Guest, W. S. & Kendall, J.-M. 1993. Modelling seismicwaveforms in anisotropic inhomogeneous media usingray and Maslov asymptotic theory: applications toexploration seismology. Canadian Journal of Explo-ration Geophysics, 29, 78–92.

Guest, W. S., Thompson, C. J. & Spencer, C. P. 1993.Anisotropic reflection and transmission calculationswith application to a crustal seismic survey from theEast Greenland Shelf. Journal of GeophysicalResearch, 98, 14 161–14 184.

Helbig, K. 1993. Simultaneous observation of seismicwaves of different polarization indicates subsurfaceanisotropy and might help to unravel its cause.Journal Applied Geophysics, 30, 1–24.

Helbig, K. 1994. Next moves in anisotropy. OilfieldReview, 6, 1–1.

Hess, H. H. 1964. Seismic anisotropy of theuppermost mantle under oceans. Nature, 203, 629.

Hill, R. 1952. The elastic behaviour of a crystalline aggre-gate. Proceedings of the Physical Society, London, A,65, 351–354.

Kendall, J.-M. 2000. Earth’s Deep Interior: MineralPhysics and Tomograpy from the Atomic to theGlobal Scale. Geophysical Monograph Series, 117,American Geophysical Union, Washington.


http://www.geolsoc.org.uk/pub_ethics

Kern, H. 1982. P- and S-wave velocities in crustal andmantle rocks under the simultaneous action of highconfining pressure and high temperature and theeffect of microstructure. In: Schreyer, W. (ed.)High Pressure Researches in Geoscience. Verlags-buchhandlung, Stuttgart, 15–45.

Kern, H. & Wenk, H. R. 1990. Fabric-related velocityanisotropy and shear wave splitting in rocks from theSanta Rosa mylonite zone, California. Journal ofGeophysical Research (Solid Earth), 95, 11 213–11 223.

Khananehdari, J., Rutter, E. H., Casey, M. &Burlini, L. 1998. The role of crystallographic fabricin the generation of seismic anisotropy and reflectivityof high strain zones in calcite rocks. Journal of Struc-tural Geology, 20, 293–300.

Kocks, U. F., Tome, C. N. & Wenk, H.-R. 1998. Textureand Anisotropy, Preferred Orientation in Polycrystalsand the Effect on Material Properties. CambridgeUniversity Press, Cambridge.

Lister, G. S. & Snoke, A. W. 1984. S-C mylonites.Journal of Structural Geology, 6, 617–638.

Lloyd, G. E. & Kendall, J.-M. 2005. Petrofabric derivedseismic properties of a mylonitic quartz simple shearzone: implications for seismic reflection profiling. In:Harvey, P. K., Brewer, T., Pezard, P. A. &Petrov, V. A. (eds) Petrophysical Properties ofCrystalline Rocks. Geological Society, London,Special Publications, 240, 75–94.

Lloyd, G. E., Butler, R. W. H., Casey, M. & Main-

price, D. 2009. Deformation fabrics, mica andseismic properties of the continental crust. Earth andPlanetary Sciences Letters, 288, 320–328.

Lloyd, G. E., Butler, R. W. H., Casey, M., Tatham, D.J. & Mainprice, D. 2011. Constraints on the seismicproperties of the middle and lower continental crust.In: Prior, D. J., Rutter, E. H. & Tatham, D. J.(eds) Deformation Mechanisms, Rheology andTectonics: Microstructures, Mechanics and Anistropy.Geological Society, London Special Publications, 360,7–32.

Mahan, K. 2006. Retrograde mica in deep crustal granu-lites: implications for crustal seismic anisotropy. Geo-physics Research Letters, 33: Art. No. L24301 Dec 16.

Mainprice, D. 1990. An efficient Fortran program to cal-culate seismic anisotropy from the lattice preferredorientation of minerals. Computers and Geosciences,16, 385–393.

Mainprice, D. 2003. http://www.gm.univ-montp2.fr/PERSO/mainprice

Mainprice, D. & Nicolas, A. 1989. Development ofshape and lattice preferred orientations: applicationto the seismic anisotropy of the lower crust. Journalof Structural Geology, 11, 175–189.

Mainprice, D. & Silver, P. G. 1993. Interpretation ofSKS waves using samples from the subcontinentallithosphere. Physics of Earth and Planetary Interiors,78, 257–280.

Mainprice, D. & Humbert, M. 1994. Methods ofcalculating petrophysical properties from lattice pre-ferred orientation data. Survey Geophysics, 15,575–592.

Makovsky, Y., Klemperer, S. L., Ratschbacher, L.,Brown, L. D., Li, M., Zhao, W. J. & Meng, F. L.

1996. INDEPTH wide-angle reflection observation ofP-wave-to-S-wave conversion from crustal brightspots in Tibet. Science, 274, 1690–1691.

Meltzer, A. & Christensen, N. 2001. Nanga Parbatcrustal anisotropy: implications for interpretation ofcrustal velocity structure and shear-wave splitting.Geophysical Research Letters, 28, 2129–2132.

Meltzer, A., Sarker, G., Beaudoin, B., Seeber, L. &Armbruster, J. 2001. Seismic characterization of anactive metamorphic massif, Nanga Parbat, PakistanHimalaya. Geology, 29, 651–654.

Meltzer, A., Christensen, N. & Okaya, D. 2004.Crustal seismic anisotropy: implications for under-standing crustal dynamics. In: Proceedings of Ameri-can Geophysical Union, Fall Meeting 2003, abstract#S11C-0302. http://adsabs.harvard.edu//abs//2003AGUFM.SIICO302W.

Michibayashi, K. & Mainprice, D. 2004. The role of pre-existing mechanical anisotropy on shear zone develop-ment within oceanic mantle lithosphere: an examplefrom the Oman ophiolite. Journal of Petrology, 45,405–414.

Michibayashi, K., Ina, T. & Kanagawa, K. 2006. Theeffect of dynamic recrystallization on olivine fabricand seismic anisotropy: insight from a ductile shearzone, Oman ophiolite. Earth and Planetary SciencesLetters, 244, 695–708.

Nicolas, A. & Poirier, J.-P. 1976. Crystalline Plasticityand Solid-State Flow in Metamorphic Rocks. Wiley,New York.

Park, S. K. & Mackie, R. J. 1997. Crustal structure atNanga Parbat, northern Pakistan, from magnetotelluricsoundings. Geophysics Research Letters, 24,2415–2418.

Park, S. K. & Mackie, R. J. 2000. Resistive (dry?) lowercrust in an active orogen, Nanga Parbat, northernPakistan. Tectonophysics, 316, 359–380.

Passchier, C. W. & Trouw, R. A. J. 1996. Microtec-tonics. Springer-Verlag, Berlin.

Poage, M. A., Chamberlaine, C. P. & Craw, D. 2000.Massif-wide metamorphism and fluid evolution atNanga Parbat, northwestern Pakistan. AmericanJournal of Science, 300, 463–482.

Prior, D. J., Boyle, A. P. et al. 1999. The application ofelectron backscatter diffraction and orientation con-trast imaging in the SEM to textural problems inrocks. American Mineralogist, 84, 1741–1759.

Ramsay, J. G. 1980. Shear zone geometry. A review.Journal of Structural Geology, 2, 83–89.

Reuss, A. 1929. Berechnung der Fließgreze von Mis-chkristallen auf Grund der Plastizitatsbedinggungenfur Einkristalle. Zeitschrift fur Angewandte Mathema-tik unf Mechanik, 9, 49–58.

Rudnick, R. L. & Fountain, D. M. 1995. Nature andcomposition of the continental crust – a lower crustalperspective. Reviews in Geophysics, 33, 267–309.

Shapiro, N. M., Ritzwoller, M. H., Molnar, P. & Levin,V. 2004. Thinning and flow of Tibetan crust constrainedby seismic anisotropy. Science, 305, 233–236.

Sibson, R. H. 1977. Fault rocks and fault mecha-nisms. Journal Geological Society of London, 133,191–213.

Siegesmund, S., Takeshita, T. & Kern, H. 1989. Aniso-tropy of Vp and Vs in an amphibolite of the deeper


http://www.gm.univ-montp2.fr/PERSO/mainprice

http://www.gm.univ-montp2.fr/PERSO/mainprice

http://adsabs.harvard.edu//abs/

http://adsabs.harvard.edu//abs/

crust and its relationship to the mineralogical. micro-structural and textural characteristics of the rock.Tectonophysics, 157, 25–38.

Silver, P. G. 1996. Seismic anisotropy beneath the conti-nents: probing the depths of geology. Annual Review ofEarth and Planetary Science, 24, 385–432.

Takanashi, M., Nishizawa, O., Kanagawa, K. &Yasunga, K. 2001. Laboratory measurements ofelastic anisotropy parameters for the exposed crustalrocks from the Hidaka Metamorphic Belt, CentralHokkaido, Japan. Geophysical Journal International,145, 33–47.

Tatham, D. J., Lloyd, G. E., Butler, R. W. H. & Casey,M. 2008. Amphibole and lower crustal seismic proper-ties. Earth and Planetary Sciences Letters, 267,118–128.

Twiss, R. J. & Moores, E. M. 1992. Structural Geology.WH Freeman & Company, New York.

Voigt, W. 1928. Lehrbuch der Kristallphysik: mit Auss-chluss der Kristalloptik. B.G. Teuber, Leipzig.

Weeraratne, D. S., Fischer, K. M., Manners, U. &Meltzer, A. 2004. Constraints on the vertical vari-ation of seismic anisotropy beneath the Nanga ParbatHaramosh Massif From S and SKS Splitting. In: Pro-ceedings of American Geophysical Union, SpringMeeting 2004, abstract #S33B-05. http://adsabs.harvard.edu/abs/2004 AGUSM.S33B.05W.

Wendt, A. S., Bayuk, I., Covey-Crump, J., Wirth, R. &Lloyd, G. E. 2003. An experimental and numericalstudy of the microstructural parameters contributingto the seismic anisotropy of rocks. Journal of Geophy-sical Research, 108, 2365.

Zeitler, P. K., Chamberlain, C. P. & Smith, H. 1993.Synchronous anatexis, metamorphism, and rapid denu-dation at Nanga Parbat (Pakistan Himalaya). Geology,21, 347–350.

Zeitler, P. K., Koons, P. O. et al. 2001. Crustal rework-ing at Nanga Parbat, Pakistan: metamorphic conse-quences of thermal–mechanical coupling facilitatedby erosion. Tectonics, 20, 712–728.


http://adsabs.harvard.edu/abs/

http://adsabs.harvard.edu/abs/

from crystal to crustal: petrofabric-derived seismic

Documents