constraints on the seismic properties of the middle … › perso › mainprice › w_data ›...

Constraints on the seismic properties of the middle and

lower continental crust

G. E. LLOYD1*, R. W. H. BUTLER2, M. CASEY1†, D. J. TATHAM1,4 & D. MAINPRICE3

1Institute of Geophysics and Tectonics, School of Earth and 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, UK3Geosciences Montpellier, CNRS & Universite Montpellier 2, 34095 Montpellier, France

4School of Earth and Ocean Sciences, University of Liverpool, Liverpool

L69 3GP, UK

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

Abstract: For the past two decades geodetic measurements have quantified surface displacementfields for the continents, illustrating a general complexity. However, the linkage of geodeticallydefined displacements in the continents to mantle flow and plate tectonics demands understandingof ductile deformations in the middle and lower continental crust. Advances in seismic anisotropystudies are beginning to allow such work, especially in the Himalaya and Tibet, using passive seis-mological experiments (e.g. teleseismic receiver functions and records from local earthquakes).Although there is general agreement that measured seismic anisotropy in the middle and lowercrust reflects bulk mineral alignment (i.e. crystallographic preferred orientation, CPO), there is aneed to calibrate the seismic response to deformation structures and their kinematics. Here, wetake on this challenge by deducing the seismic properties of typical mid- and lower-crustalrocks that have experienced ductile deformation through quantitative measures of CPO insamples from appropriate outcrops. The effective database of CPO and hence seismic propertiescan be expanded by a modelling approach that utilizes ‘rock recipes’ derived from the as-measuredindividual mineral CPOs combined in varying modal proportions. In addition, different defor-mation fabrics may be diagnostic of specific deformation kinematics that can serve to constraininterpretations of seismic anisotropy data from the continental crust. Thus, the use of ‘fabricrecipes’ based on subsets of individual rock fabric CPO allows the effect of different fabrics(e.g. foliations) to be investigated and interpreted from their seismic response. A key issue is thepossible discrimination between continental crustal deformation models with strongly localizedsimple-shear (ductile fault) fabrics from more distributed (‘pure-shear’) crustal flow. The resultsof our combined rock and fabric-recipe modelling suggest that the seismic properties of themiddle and lower crust depend on deformation state and orientation as well as composition,while reliable interpretation of seismic survey data should incorporate as many seismic propertiesas possible.

Geodetic measurements quantifying displacementfields for the Earth’s surface illustrate a generalcomplexity that questions the existence of asimple relationship to mantle flow and plate tec-tonics (e.g. Fouch & Rondenay 2006; Caporaliet al. 2009; Thatcher 2009). Linking geodeticallydefined continental displacements to mantle flowand plate tectonics demands understanding ofductile deformation in the middle and lower con-tinental crust, for example, the discriminationbetween deformation models with strongly localized

‘simple-shear’ fabrics (e.g. Burg 1999) from thoseinvolving more distributed ‘pure-shear’ crustal flow(e.g. Butler et al. 2002). Studies of the in situ seis-mic characteristics of the middle and lower conti-nental crust should lead to enhanced understandingof ductile deformation in these regions. However,until recently the continental crust was consideredtoo heterogeneous mineralogically and tectonicallyto reflect coherent patterns of, for example, seismicanisotropy. Today, seismic anisotropy is now notonly recognized regularly in the continental crust

†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, 7–32.DOI: 10.1144/SP360.2# The Geological Society of London 2011. Publishing disclaimer: www.geolsoc.org.uk/pub_ethics

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

Vp km/s AVs % Vs1P

c m

a

-ar

zm

qu

artz

r r

r

m mm

z z

z

a a-a

c 10.00

.00

6.07.0

120.00

.00

10.0

40.0

crystalform

stereographicprojection

c m

a

am

v

vr

calc

ite

mm

m

v

v

vr

c

a

aa

10.00

.00

6.0

7.0

120.00

.00

10.0

50.0

pla

gio

clas

efe

ldsp

ar

ab

c

M m

p o a

bc

10.00

.00

6.08.0

120.00

.00

10.030.050.0

a

o oM M b

c

bio

tite

mic

a

a

bc

10.00

.00

6.07.0

120.00

.00

10.030.050.070.090.0110.0

aug

ite

pyr

oxe

ne

a

b

c

x

y

z

001111 010

100

110a

bc

10.00

.00

7.08.09.0

120.00

.00

10.0

fors

teri

teo

livin

e

a

b

c

x y

z001

101 021

110

010

a

bc

10.00

.00

8.09.0

120.00

.00

10.0

c

mm

yx

ba

ort

ho

clas

efe

ldsp

ar

a

bc

10.00

.00

6.07.0

120.00

.00

10.030.0

aM M b

c

e

e

mu

sco

vite

mic

a

a

bc

10.00

.00

7.08.0

120.00

.00

10.030.050.070.0

ho

rnb

len

de

amp

hib

ole

a

b

c

110010

011

xy

z

a

bc

10.00

.00

6.57.5

120.00

.00

10.030.0

n n

n

nn

nn

n

n

d

d

d da

a

a

gar

net

a

a

a

10.00

.00

120.00

.00

7.02

5.31

AVp = 27.7%

43.17

0.02

7.765.61

AVp = 32.1%

59.98

0.02

7.50

5.30

AVp = 34.5%

33.62

1.12

8.16

5.10AVp = 46.2%

53.24

0.47

8.20

6.13

AVp = 28.9%

72.09

0.36

7.81

5.60

AVp = 32.9%

113.82

0.15

7.89

6.01

AVp = 27.1%

30.69

0.80

9.01

6.92

AVp = 26.2%

19.31

0.24

9.77

7.65

AVp = 24.3%

17.95

0.52

8.88

8.85

AVp = 0.3%

0.72

0.00

Fig. 1.

G. E. LLOYD ET AL.8

but is also used increasingly to image zones ofdeformation (e.g. Meltzer et al. 2001; Shapiroet al. 2004; Sherrington et al. 2004; Moschettiet al. 2010). Seismic anisotropy studies (e.g. in theHimalaya and Tibet) are therefore beginning toinvestigate middle and lower continental crustaldeformation using passive seismological exper-iments, such as teleseismic receiver functions (e.g.Ozacar & Zandt 2004) and records from local earth-quakes (e.g. Schulte-Pelkum et al. 2005). Suchstudies claim that it is possible to distinguish differ-ent types and magnitudes of deformation via seismicanisotropy. If correct, there is much to be gainedfrom seismic (anisotropy) analysis of the (ductile)continental crust, provided the source of the aniso-tropy is rigorously constrained.

Although there is general agreement thatmeasured seismic anisotropy in the middle andlower continental crust reflects bulk mineral crystal-lographic preferred orientation (CPO), there is aneed to calibrate the seismic response to defor-mation structures and their kinematics (e.g. Okaya& Christensen 2002; Mahan 2006; Meissner et al.2006). Almost all of the common rock-formingminerals in the continental crust exhibit significantseismic velocity anisotropy as single crystals(Barruol & Mainprice 1993; Ji et al. 2002). Theseare illustrated in Figure 1 where the individualseismic properties for each mineral have been con-toured using the same scale ranges to emphasizerelative values and distributions. Plotting theseismic properties in this way indicates which min-erals are likely to control each seismic property(e.g. mafic minerals increase compressional wavevelocity Vp while most minerals have similarseismic anisotropy AVs except for the micas,which have very high maximum anisotropy).However, for many minerals (including quartz andfeldspars) crystal symmetry creates geometricallycomplex seismic responses that, with smalldegrees of misorientation within polycrystallineaggregates, interfere destructively to produce near-isotropic behaviour (Barruol & Mainprice 1993;Tatham et al. 2008; Lloyd et al. 2009). Key

exceptions are monoclinic mica and amphibole min-erals (e.g. biotite and hornblende) that commonlyalign both crystallographically and dimensionallyto create a single slow direction for seismic trans-mission. Thus, the seismic properties measured forthe (ductile) middle and lower continental crustare usually attributed to the CPO of mica (e.g.Kern & Wenk 1990; Barruol & Mainprice 1993;Nishizawa & Yoshino 2001; Shapiro et al. 2004;Mahan 2006; Meissner et al. 2006; Lloyd et al.2009) or amphibole (e.g. Siegesmund et al. 1989;Barruol & Mainprice 1993; Rudnick & Fountain1995; Kitamura 2006; Meissner et al. 2006;Barberini et al. 2007; Tatham et al. 2008). Itshould also be mentioned that other microstructuralparameters (e.g. grain shape fabric and variationsin spatial distribution of mineral phases, grain-boundary properties, porosity, etc.) may contributein particular to seismic anisotropy (e.g. Wendtet al. 2003). Furthermore, in the (brittle) uppercrust the effects of open fractures (e.g. Crampin1981; Kendall et al. 2007), sedimentary layering(e.g. Vernik & Liu 1997; Valke et al. 2006;Kendall et al. 2007) and/or grain-scale effects(e.g. Hall et al. 2008) must be considered in additionto CPO.

According to recent compilations (e.g. Rudnick& Fountain 1995; Rudnick & Gao 2003), increasingaverage P-wave seismic velocities with depth indi-cates increasing proportions of mafic lithologiesand increasing metamorphic grade (Fig. 2). Superpo-sition of single-crystal Vp values for common rock-forming minerals (e.g. Fig. 1) confirms this generaltrend. Thus, the lower continental crust (i.e. belowc. 20–25 km depth) is expected to be lithologicallydiverse but dominated by granulite facies maficlithologies with an average composition approach-ing that of primitive basalt, although amphibolitefacies may be significant where high water fluxesoccur and felsic-to-intermediate lithologies canalso be important locally (Rudnick & Fountain1995). In contrast, the average middle crust (i.e.between 10–15 and 20–25 km depth), where P-wave velocities are too low to be explained by

Fig. 1. Principal seismic properties of elastically anisotropic single crystals of major rock-forming mineralsrelative to mineral form and crystallography (Vp: compressional wave velocities; AVs: percentage shear-wavesplitting; Vs1P: polarization direction of fastest shear wave). The seismic property for each mineral has beencontoured using the same scale range to emphasize relative values and distributions (i.e. Vp, minimum 5.10 km/sfor plagioclase and maximum 9.77 km/s for olivine; AVs, minimum 0% and maximum 113.82% for biotite). Forexample, blues and reds indicate relatively high or low values respectively (NB contour lines are drawn and numberedonly for the precise range of that property for that mineral). The seismic properties were derived using the Mainprice(1990) suite of programs (i.e. Pfch5, Anis and VpGC) and appropriate mineral single-crystal elastic constants: quartz(McSkimin et al. 1965), calcite (Dandekar 1968), orthoclase (Aleksandrov et al. 1974), plagioclase (Aleksandrovet al. 1974), muscovite (Vaughn & Guggenheim 1986), biotite (Aleksandrov & Ryzhova 1961), hornblende(Aleksandrov & Ryzhova 1961), augite (Aleksandrov & Ryzhova 1961), garnet (Bonczar et al. 1977) and olivine(Abramson et al. 1997).

SEISMIC PROPERTIES OF CONTINENTAL CRUST 9

dominantly mafic lithologies, is considered toconsist of a mixture of mafic, intermediate andfelsic amphibolite facies gneisses (e.g. Rudnick &Fountain 1995; Rudnick & Gao 2003). Suggestionsthat micas control the seismic properties of thedeeper continental crust (e.g. Meltzer & Christensen2001; Meltzer et al. 2001; Takanashi et al. 2001;Chlupacova et al. 2003; Shapiro et al. 2004;Mahan 2006; Meissner et al. 2006) are difficult toreconcile with the view that there is little evidencefor metapelite-dominated layers in these regions(e.g. Rudnick & Fountain 1995). Nevertheless, mus-covite Vp values are clearly compatible with thoseexpected for the lower crust rather than the middlecrust (Fig. 2). In contrast, biotite values are typicallytoo low and are more appropriate to the middle crust.It is possible that high metamorphic grade formermetapelites that now comprise kyanite and/or silli-manite could also explain lower crustal Vp values;both of these minerals have high seismic velocitiesas well as large seismic anisotropy (e.g. Ji et al.2002), but they probably represent ,10% of thelower crust in general (Rudnick & Fountain 1995).

While single-crystal Vp characteristics of com-mon rock-forming minerals provide initial con-straints on the composition and nature of theductile continental crust (i.e. Figs 1 & 2), actuallithologies are polygranular and usually polymi-neralic in reality. Thus, the Vp properties need tobe derived from the whole-rock petrofabrics (i.e.the sum of the CPO for the individual constituentminerals) to provide more rigorous constraints(e.g. Barruol & Mainprice 1993). For example,

combining the appropriate single-crystal Vp charac-teristics shown in Figure 2 indicates that granitic(i.e. quartz + feldspars+micas), amphibolitic (i.e.hornblende + plagioclase + micas) and mafic (i.e.+amphibole + pyroxene + plagioclase + garnet)compositions are responsible for the Vp propertiesof the middle crust, lower–middle and lower crustand lower/lowermost ductile continental crust,respectively. However, this approach omits con-sideration of seismic anisotropy, which typicallyreflects the strength of individual mineral andwhole-rock CPO. In general, CPOs form in responseto deformation and their detailed distributions canreflect and hence distinguish the kinematic refer-ence frame (e.g. Schmid & Casey 1986; Passchier& Trouw 2005). Investigation of the completewhole-rock petrofabric-derived seismic propertiesshould therefore lead to a better understanding ofductile deformation in the middle and lower con-tinental crust, which ultimately should help inlinking geodetically defined continental displace-ments to mantle flow and plate tectonics. The presentcontribution takes up this challenge by providingconstraints on the seismic properties of the ductilemiddle and lower continental crust via quantitativemeasures of CPO in appropriate natural representa-tive samples.

Methodology

Seismic property determination

The velocity (Vp, Vs1, Vs2), anisotropy (AVp,AVs) and polarization (Vs1P) of seismic compres-sional (p) and shear (s) waves depend on the three-dimensional elastic ‘stiffness’ (Cij) of rocks, whichvaries with crystal direction in the constituentminerals (e.g. Babuska & Cara 1991). Bulk CPOand hence seismic properties are therefore deter-mined by: individual mineral crystallography(symmetry, etc.); whole-rock mineralogy (compo-sition, modal proportions); deformation mechan-isms (e.g. crystal slip); deformation kinematics(deformation type, magnitude and rate); and geo-logical environment (pressure, temperature, fluids,etc.). The influence of these variables is consideredhere via ‘end-member’ rock types that span typicalmiddle and lower continental crustal compositions(e.g. Rudnick & Fountain 1995; Rudnick & Gao2003), namely felsic (i.e. +quartz + feldspar +mica) and mafic (i.e. +amphibole + pyroxene +plagioclase + mica). Typical samples were ana-lysed via electron backscattered diffraction (EBSD)in the scanning electron microscope (SEM) tomeasure their CPOs, from which their seismicproperties were derived following conventionalprocedures (e.g. Mainprice 1990; Mainprice &Humbert 1994; Lloyd & Kendall 2005).

8

6

7

Sin

gle

crys

talV

pra

ng

e(k

m/s

)

5

MC

LMCLC

LsC

gar

net

9

10

Common rock-forming minerals

qu

artz

calc

ite

ort

ho

clas

e

pla

gio

clas

e

mu

sco

vite

bio

tite

ho

rnb

len

de

aug

ite

oliv

ine

Fig. 2. Summary of the average P-wave velocitiescompiled by Rudnick & Fountain (1995) for the middle(MC), lower middle (LMC), lower (LC) and lowestlower (LsC) continental crust. Also shown are the ranges(black columns) and approximate averages (whitecircles) of single-crystal Vp values for commonrock-forming minerals (see Fig. 1). The latter can becompared to the values suggested by Rudnick &Fountain (1995) to indicate the likely minerals presentand hence responsible for the seismic properties in eachcrustal region.

G. E. LLOYD ET AL.10

Samples

When modelling the seismic response of the conti-nents, we are aware that outcrop lithologies thatonce resided at deep crustal levels may not be repre-sentative of the lower crust in situ. Nevertheless, thesamples used in this study are considered to berepresentative of many of the lithologies that makeup the ductile middle and lower continental crust(e.g. Rudnick & Fountain 1995; Rudnick & Gao2003). Garnet-bearing lower crustal lithologieshave however been omitted from our analysis dueto the isotropic seismic properties and relativelyhigh density of garnet which predictably act todilute seismic anisotropy and increase seismic(P-wave) velocity (e.g. Figs 1 & 2), respectively,with increasing garnet content (e.g. Brown et al.2009) even if the garnet exhibits (rare) CPO. Inaddition, we consider orthopyroxene (specificallyhypersthene) to be representative of pyroxenes gen-erally due to the similarity in crystal structure andhence seismic properties of orthopyroxenes andclinopyroxenes (see Barruol & Mainprice 1993 orJi et al. 2002 for specific examples). The foursamples used in this study are described as follows.

(1) Felsic orthogneiss (Nanga Parbat, west Hima-laya; Butler et al. 2002). This rock compri-ses quartz–orthoclase–plagioclase–biotite–muscovite. Such orthogneisses are derivedlargely from a granodioritic protolith com-posed of feldspar, quartz and biotite, with sub-ordinate muscovite. Overall, this compositionis considered to represent a good analogue formuch of the Earth’s ductile hydrous middlecontinental crust. Furthermore, mica-bearingtectonites frequently exhibit S–C fabrics,which allow the impact of different texturaldevelopment on seismic properties to beinvestigated (Lloyd et al. 2009).

(2) ‘Banded’ amphibolite (south Harris, north-west Scotland; Lapworth et al. 2002). Thisrock comprises hornblende–plagioclase–quartz. The compositions of different ‘bands’(e.g. quartz–plagioclase, quartz–plagioclase–hornblende, plagioclase–quartz–hornblende,hornblende–plagioclase–quartz, hornblende–plagioclase and hornblende only) are con-sidered to be representative of a range ofhydrous ductile middle-to-lower continentalcrustal lithologies.

(3) Sheared mafic dyke (Badcall, northwestScotland; Tatham & Casey 2007; Tathamet al. 2008). This rock comprises hornblende–plagioclase–quartz and is therefore appar-ently compositionally similar to sample 2.However, it represents a suite of specimens(1–9; note specimen 1 contains also

appreciable clinopyroxene that presumablyreflects the original igneous composition)exhibiting progressively increasing deforma-tion. It is therefore used to study the impactof deformation on seismic properties. It is con-sidered to be representative of the hydrousductile upper–mid lower continental crust.

(4) Pyroxene granulite (Kerala, south India;Prasannakumar & Lloyd 2007, 2010). Thisrock comprises hypersthene–plagioclase–quartz–(biotite). It is considered to be repre-sentative of the anhydrous (hydrous wheremicaceous) ductile lower continental crust.

Results

Composition and texture

The effective database of whole-rock CPO andhence seismic properties can be expanded by a mod-elling approach that utilizes ‘rock recipes’ based onthe mineral modal proportions determined viaSEM-EBSD analysis (e.g. Tatham 2008; Tathamet al. 2008; Lloyd et al. 2009). In this approach,the modal content of one mineral is varied progress-ively from 0 to 100% while the other phases arevaried according to their original modal proportions.The seismic properties are then calculated per‘recipe’. The recipes considered here are based onnine specimens comprising hornblende, plagioclaseand quartz collected from sample 3 above, thesheared mafic dyke rock (Tatham & Casey 2007;Tatham 2008). The CPO (not shown here – seeTatham et al. 2008) of the individual minerals inthese specimens was determined via SEM-EBSD,from which the whole-rock seismic properties perspecimen were determined in the conventionalmanner (see above). In general, the specimensexhibit a trend in seismic properties from analmost isotropic aggregate (specimen 1) to astrong and ordered pattern of orthorhombic sym-metry (specimens 2–9) with increasing strain thataccords with the finite strain axes and indicates thedominant role of hornblende in controlling theseproperties (e.g. Tatham et al. 2008). However,the roles played by all three constituent mineralscan be investigated, both within and across therange of specimens from sample 3, via a rock-recipemodelling approach as follows.

Rock-recipe modelling. As the specimens of shearedmafic dyke rock typically comprise only threemineral phases (i.e. hornblende, plagioclase andquartz), modelled variations in their compositionscan be represented by simple ternary plots (seeFig. 3). Each plot is based on the interpolation ofmonomineralic aggregate elastic properties for thespecimen over all permutations of modal fractions,


including monomineralic, in each mineral phase.However, the elastic properties of each mineralphase are representative of their deformation

behaviour in a polyphase aggregate with the specificrelative modal volume fractions of the originalspecimen. In practice, both the strain distribution

Fig. 3. Ternary rock-recipe plots for mafic dyke samples 1–9 showing the sensitivity of various seismic propertiesto variations in modal composition for different shear strains (g) and ellipticity (R) of the finite strain ellipse in athree-phase aggregate system comprising hornblende, plagioclase and quartz. The shaded areas of the legend ternaryplots indicate the regions of confident extrapolation of data due to potential differences in the behaviour ofpolymineralic and monomineralic rocks (see text for discussion). Black dots indicate the actual modal composition ofeach sample. (a) Maximum in Vp; (b) AVp; (c) AVs.


in the original specimen and its partitioning intospecific mineral phases are likely to be such thatthe microstructure, petrofabric and hence elasticproperties of the constituent mineral phase fractionsare not representative of the behaviour of each phasefor a monomineralic rock of that phase at the strainobserved in the bulk specimen. For example, defor-mation of an aggregate comprising 60% hornblende,30% plagioclase and 10% quartz (similar to that ofspecimens 19; see Fig. 3) seems to preferentially par-tition strain into the hornblende phase as the plagio-clase and quartz phases exhibit poorly developedCPO (see Tatham 2008; Tatham et al. 2008). Defor-mation of approximately monomineralic aggregatesof either quartz or plagioclase to the same bulk strainwould almost certainly involve the development ofstrong CPO (e.g. Marshall & McLaren 1977a, b;Lister & Dornsiepen 1982; Olsen & Kohlstedt1984, 1985; Mainprice et al. 1986; Kruhl 1987).

The range of relative modal fractions (i.e. rockrecipes) for which the ternary plots (Fig. 3) can beapplied is therefore likely limited. From results oflow-law calculations in a selection of two-phaseaggregates, Handy (1994) postulated that the pres-ence of .10% of a weak phase is necessary forthat phase to govern the bulk strength of the aggre-gate. In detail, this prediction depends on a numberof factors including temperature, strain rate and thestrength contrast between adjacent phases. It istherefore suggested that the rock-recipe modellingresults are accurate only between 10 and 90% ineach phase for each possible two-phase system ofthe three-phase aggregate (i.e. the shaded areasindicated in the legend ternary plots in Fig. 3).

The ternary rock-recipe plots reveal the sensi-tivity of various seismic properties to variations inmodal composition for different deformation states(e.g. shear strain and ellipticity of the finite strainellipse). It is clear that the maxima in Vp, AVpand AVs are most strongly dependent upon hornble-nde modal fraction. For example, there are increasesof up to 1 km/s in Vp (Fig. 3a), 5% in AVp (Fig. 3b)and 8% in AVs between end-member plagioclaseand hornblende compositions.

Figure 3a, b also indicates that both AVp andAVs tend to increase with increasing strain, butonly up to a shear strain of c. 10. Beyond this valuethere tends to be little increase in either anisotropy.This behaviour suggests that seismic anisotropycannot increase indefinitely with deformation, butsaturates at a certain value. In general, Vp valuesshow little variation with increasing strain (Fig. 3a).

In contrast, the impact of changing quartz modalfraction on these seismic properties is minor.For example, from a starting aggregate of 50%hornblende and 50% plagioclase, progressivelyincreasing the quartz fraction up to 100% yieldsonly a 2% increase overall in AVp (Fig. 3b) and a

2% increase in AVs at 50% quartz, returning to noincrease at 100% quartz (Fig. 3c). The change inthe maximum in Vp is also minimal (Fig. 3a).Previous investigations into the relative effects ofcomponent phases on seismic properties have alsohighlighted the dominant role of mafic components(e.g. hornblende) compared to felsic components(plagioclase and quartz), with the latter oftenacting as ‘dilutants’ (e.g. Christensen & Fountain1975; Fountain & Christensen 1989; Tatham et al.2008; Lloyd et al. 2009).

Rock-recipe modelling approaches indicate thatthe seismic properties of (hydrous) mafic rocks aremost likely to be controlled by amphibole (e.g.Tatham et al. 2008). In contrast, for felsic rocks,micas (in particular biotite) are the controllingmineral phase (e.g. Lloyd et al. 2009, 2011). Otherminerals in both lithologies, such as plagioclaseand quartz but also pyroxene (see below), act asdilutants and tend to reduce seismic properties(particularly anisotropy) although some (e.g. plagio-clase and pyroxene) may increase velocities (seebelow).

Fabric-recipe modelling. Many rocks may comprisedifferent deformation fabrics (e.g. S–C foliations).Use of fabric recipes based on subsets of individualrock fabric CPOs therefore allows the impact ofdifferent fabrics on seismic properties to be investi-gated. In this recipe modelling approach, CPOs aremeasured (e.g. via SEM-EBSD) for each fabricelement and also for the whole rock, and are thenvaried in a similar manner to rock recipes. Theelastic properties of each fabric recipe are thenused to calculate the impact of different proportionsof each fabric on the whole-rock seismic properties.For example, Lloyd et al. (2009, 2011) have shownthat the typical transverse isotropy characteristicsassociated with individual mica fabric elementscan disappear in whole rocks comprising S–Ctype foliations.

Foliation and azimuthal effects

Seismic modelling. Rock and fabric-recipe model-ling suggest that foliation, and in particular itsorientation, exerts a significant impact on seismicproperties (see also Lloyd et al. 2011). To investi-gate this impact, a seismic model was constructedrepresenting a 10 km slice of deformed lower conti-nental crust (Fig. 4). This thickness of crust waschosen to provide practical values of shear-wavesplitting (dVs). The model was populated with theelastic stiffness properties of the nine specimens ofmafic dyke rock using a common rock-recipe com-position of 60% hornblende, 30% plagioclase and10% quartz, which represents the average compo-sition of these specimens.


As discussed previously, the elastic propertiesof rock sample 3 vary due to the increasing strainexhibited by specimens 1–9 (e.g. Fig. 3). One axisof the model can therefore be scaled according toincreasing strain, represented as either shear strain(g) or the length of the major axis of the strainellipse (S1, where S1 ¼ 1 defines an undeformedunit circle). Representing the variation in seismicproperties with respect to either g or S1 meansthat the plots can be used to consider either simpleor pure-shear-dominated deformations, respect-ively. The other axis of the model is scaled accord-ing to the orientation of sample foliation relative topropagating seismic waves. For the purposes of thismodel it is assumed that the waves propagate verti-cally through the 10 km of crust and can effectivelybe considered as teleseismic waves (Fig. 4). Samplefoliation is rotated from horizontal (08) to vertical(908) in both clockwise and anticlockwise directionsabout either the X or Y tectonic axes. In the formerconfiguration, the X or lineation axis remainsunchanged and horizontal while the foliationrotates from horizontal to vertical. In the latter con-figuration, both the lineation and foliation rotatebetween horizontal and vertical.

The variation in seismic properties with strainand foliation orientation in the model is determinedin three stages (e.g. Fig. 4). Firstly, the bulk elasticstiffness matrices (Cij) for each specimen ofdeformed mafic dyke rock are interpolatedbetween their pre-determined finite shear strains

(g) or the associated strain ellipse long axes (S1)to describe the variation of elastic stiffness withstrain. The interpolated grid is discrete but of highresolution, with interpolations at increments of 0.1for both g and S1. Secondly, the interpolated strainscale of Cij is incrementally rotated from 08 to 908either clockwise or anticlockwise with respect toeither the X or Y tectonic axes. Again, the inter-polated scale of petrofabric rotation is discretebut of high resolution, with increments of +58.Thirdly, the values of Vp and dVs are calculatedfor each node of the interpolated grids, fromwhich contoured plots are constructed (e.g. Fig. 4).Note that the calculated P-wave velocities referspecifically to vertically propagating waves andhence are not necessarily the maximum in P-wavevelocities. In addition, the fast shear-wave polariz-ation orientation (Vs1P) is also determined, butfor a more sparsely populated grid necessary fordisplay purposes.

Foliation orientation. Results of seismic modellingof the impact of foliation orientation on seismicproperties (Vp and dVs) for different strains (i.e.‘simple’ and ‘pure’ shears) are shown in Figure 5.In general, all relationships indicate a crude bilateralsymmetry about the line of zero rotation (horizontalfoliation). Deviations from perfect bilateral sym-metry are due to a combination of the slightly non-orthorhombic nature of the aggregate Cij, whichprobably reflects complex interactions between con-stituent phases with different seismic properties andsymmetries and other natural variations (e.g. micro-structure, CPO, etc.). Apparently anomalous stepsor jumps in the distribution of seismic propertiesare an amplification effect of specimen variations.Discontinuities in the smoothly varying distri-butions mark points of specimen control, whilesmooth variations denote regions of interpolationof physical properties between specimens. Simi-larly, complex patterns of ‘peaks’ and ‘cusps’,particularly within central bands adjacent to thesymmetry axis, are due to local effects around speci-men control points. The peaks and cusps are ofcomparatively low relief and are considered to beinsignificant with respect to the overall trendsindicated. The impact of each of these irregularitieswould be reduced by employing a greater speci-men density. If they are overlooked, a number of

Fig. 5. Seismic modelling of the effects of foliation orientation and deformation on P-wave velocity (Vp) andshear-wave splitting (dVs) for vertically propagating (‘teleseismic’) waves through 10 km of crust populated by thesamples of mafic dyke rock (see Fig. 4). For each pair of diagrams, deformation is represented in terms of shear strain(g) and the length of the major axis of the strain ellipse (S1, where S1 ¼ 1 defines an undeformed unit circle).(a) Variation in Vp for rotation about the X tectonic axis. (b) Variation in Vp for rotation about the Y tectonic axis.(c) Variation in dVs for rotation about the X tectonic axis. (d) Variation in dVs for rotation about the Y tectonic axis. Seetext for discussion.

Fig. 4. Schematic representation of the seismic modelused to investigate the impact of foliation orientation ondifferent seismic properties (see Figs 5 & 6).


(a) (c)

(b) (d)

80

60

40

20

0

20

40

60

80R

otat

ion

offo

liatio

nab

out X

0 5 10 15 20 25 30 35 40 45 50 55Shear strain ( )g

Vp km/s6.6

6.5

6.4

6.3

6.2

6.1

6.080

60

40

20

0

20

40

60

80

Rot

atio

nof

folia

tion

abou

tX

0 5 10 15 20 25 30 35 40 45 50 55Shear strain ( )g

dVs (secs)

0.18

0.16

0.14

0.12

0.10

0.08

0.06

0.04

0.020

2.00

80

60

40

20

0

20

40

60

80

Rot

atio

nof

folia

tion

abou

t X

Vp km/s6.6

6.5

6.4

6.3

6.2

6.11 2 3 4 5 6 7 8 9 10

Strain ellipse major axis (S )1

80

60

40

20

0

20

40

60

80

Rot

atio

nof

folia

tion

abou

t X

Vp km/s6.6

6.5

6.4

6.3

6.2

6.11 2 3 4 5 6 7 8 9 10


80

60

40

20

0

20

40

60

80

Rot

atio

nof

folia

tion

abou

t X

0 5 10 15 20 25 30 35 40 45 50 55Shear strain ( )g

Vp km/s6.6

6.0

6.5

6.4

6.3

6.2

6.1

80

60

40

20

0

20

40

60

80

Rot

atio

nof

folia

tion

abou

t X

dVs (secs)0.14

0.12

0.10

0.08

0.06

0.04

0.02

01 2 3 4 5 6 7 8 9 10


80

60

40

20

0

20

40

60

80

Rot

atio

nof

folia

tion

abou

t X

0 5 10 15 20 25 30 35 40 45 50 55Shear strain ( )g

dVs (secs)

0.18

0.16

0.14

0.12

0.10

0.08

0.06

0.04

0.02

0

2.00

80

60

40

20

0

20

40

60

80

Rot

atio

nof

folia

tion

abou

tX

dVs (secs)0.14

0.12

0.10

0.08

0.06

0.04

0.02

01 2 3 4 5 6 7 8 9 10


Fig. 5.


significant trends in the seismic properties can berecognized as follows (see Fig. 5).

(1) The bilateral symmetries recognized in allplots suggest that the direction of fabricrotation (i.e. clockwise or anticlockwise)about the kinematic X or Y axes does notaffect the pattern or magnitude of seismicvelocity or anisotropy to any significantdegree. Furthermore, there are notable simi-larities between the velocity and anisotropydistributions irrespective of whether the foli-ation is rotated about either the X or Y axes.

(2) There is a general decrease in the magnitude ofseismic velocity and anisotropy with increas-ing strain for subhorizontal foliation orien-tations. This contrasts with the tendency forvelocity and anisotropy values to increasewith strain for steep to vertical foliations orien-tations. The greatest rate of change occurs forvalues 0 ≤ g ≤ 10 or 1 ≤ S1 ≤3, beyondwhich increasing strain results in negligiblechanges in either velocity or anisotropy. Thisobservation supports the suggestion madeearlier that both petrofabric and hence petro-physical properties become saturated by gc. 10 (i.e. S1–3).

(3) Somewhat steeper gradients in the changein Vp and dVs with strain are observed forvalues 0 ≤ g ≤ 0.4 or 1 ≤ S1 ≤ 1.2. Thesebehaviours suggest that small initial strainsimposed upon initially isotropic protolithscan have dramatic effects upon the seismicproperties.

(4) The change from dominantly decreasing toincreasing trends of Vp and dVs values withstrain occurs for foliation rotations about Xof 20–408. Similar behaviour is shown byVp for rotations about Y. For dVs the changeoccurs closer to 608 due to details in theAVs distribution, whereby low values ofanisotropy occupy a greater proportion of theXZ plane compared to the YZ plane; thisleads to a comparatively wider band of lowdVs values about subhorizontal foliations.It therefore follows that, for a given strain,a move from subhorizontal to subvertical foli-ations is associated with an overall increase inVp and dVs values.

Based on these seismic modelling results, inter-polation of the petrophysical properties of a discretestrain-calibrated microstructurally characterizedand compositionally normalized suite of specimensallows the evaluation of seismic properties againstfinite strains (both simple and pure shears) andfoliation orientation in that material, across acontinuum. A similar conclusion, again based onpetrofabric-derived seismic modelling, is reported

elsewhere in this volume for mica-dominatedfelsic rocks (Lloyd et al. 2011). It thereforeappears that CPO observations and Vp and dVs(sic AVs) distributions are potentially useful proxiesfor each other in regional-scale crustal geodynamicinvestigations and models.

Azimuthal considerations. In the previous section itwas shown that accurate interpretation of seismicwave propagation through foliated rocks dependson knowledge of the source and propagation direc-tion of the seismic waves relative to the foliationorientation. The propagation direction was alwaysvertical in the seismic model described above,implying either teleseismic waves or possiblylocal seismic events. In nature, source locationsare likely to be variable and hence propagationdirections also vary in their orientations (i.e. azi-muths and plunges). For example, assuming a con-stant steeply dipping foliation, waves from ateleseismic source would propagate (sub-) parallelto foliation while waves from a remote (i.e. ‘wide-angle’) source would propagate (sub-) normal tofoliation. The same seismic array would thereforedetect very different seismic properties. Thissimple example emphasizes the need to also con-sider CPO-derived polarization/birefringence (i.e.Vs1P) effects in the seismic models.

Analysis of Vs1P behaviour for the seismicmodel (Fig. 4) reveals distributions similar tothose of Vp and dVs (Fig. 6). The vertical expres-sion of shear-wave polarization becomes moreapparent with increasing strain and as the foliationrotates towards the vertical. Furthermore, as indi-cated in the previous section, Vs1P is also associatedwith an increasing component of the total AVs as thefoliation rotates towards parallelism with the verti-cal wave propagation direction. However, the direc-tion of rotation does produce some differences.For rotations about X, the direction of polarizationis ‘parallel’ to the strain axis of the plots whicheffectively defines the orientation of X (Fig. 6a).In contrast, for rotations about Y, the direction ofpolarization is ‘normal’ to the strain axis of theplots which effectively defines the orientation ofY (Fig. 6b). These results are perhaps intuitiveand reflect the development of an increasingly well-defined foliation with strain, together with theincreasing parallelism of that foliation with theseismic ray-path. The Vs1P polarization plane isconsistently parallel to the plane of petrofabric foli-ation (Fig. 6 inserts), which in turn is parallel to thegirdle of greatest shear-wave anisotropy.

The seismic modelling described here illus-trates the significance of azimuth in the interpret-ation of seismic properties. A similar conclusion,again based on petrofabric-derived seismic model-ling, is reported elsewhere in this volume for


mica-dominated felsic rocks (Lloyd et al. 2011).However, in that contribution, it is shown thatrocks comprising multiple (mica-defined) foliationsexert complicated responses on Vs1P orientationsand magnitudes, varying with the relative pro-portions of the different foliations (see also Lloydet al. 2009). In particular, the VS1P orientationdoes not necessarily reflect variations in kinematicsindicated by changing foliation proportions. It there-fore appears that CPO observations and seismicproperty distributions are potentially usefulproxies for each other, but only in regional-scalecrustal geodynamic investigations and modelsinvolving a single and/or dominant foliation. Inregions comprising multiple foliations, which arelikely to generate variations in orientation and mag-nitude of both AVs and VS1P, ‘maps’ of variationsin these properties with depth need careful (geo-logical) consideration and interpretation.

CPO and deformation

CPO develops during ductile deformation via dislo-cation creep on crystal slip systems that are mainlytemperature dependent, but are also sensitive todeformation type (e.g. Nicolas & Poirier 1976;Passchier & Trouw 2005). Complex natural defor-mations are usually simplified (e.g. plane strain,pure and simple shear, flattening, constriction,etc.) and represented on diagrams such as Flinnplots (e.g. Flinn 1956; see below and Figs 7 & 8).However, these deformations may not always bedistinguished via the seismic properties of thedominant minerals, as the following examplesindicate.

Figure 7a is a Flinn diagram illustrating the vari-ation in quartz c- and a-axes CPO with deformationtype and magnitude (Lister & Hobbs 1980). Note thetendency for dispersion rather than concentration

Fast shear wave orientation

80

60

40

20

0

20

40

60

80

Rot

atio

nof

f olia

tion

abou

t X

5 10 15 20 25 30 35 40 45 50 55Shear strain ( )g


80

60

40

20

0

20

40

60

80

Rot

atio

no f

folia

tion

abou

t X

1 2 3 4 5 6 7 8 9 10Strain ellipse major axis length (S )1

z

y

z

y

y

x

y

x


80

(a) (b)

60

40

20

0

20

40

60

80

Rot

atio

nof

folia

tion

abou

t X

5 10 15 20 25 30 35 40 45 50 55Shear strain ( )g


80

60

40

20

0

20

40

60

80

Rot

atio

nof

folia

tion

abou

t X

1 2 3 4 5 6 7 8 9 10Strain ellipse major axis length (S )1

y

x

y

x

z

x

z

x

Fig. 6. Relationship between the fast shear-wave polarization direction (Vs1P), deformation and CPO for a vertical ray(i.e. teleseismic wave) passing through a 10 km section of ‘crust’ populated with the elastic properties and estimatedshear strain (Tatham & Casey 2007) of the mafic dyke samples 1–9. Inserts to the right of each plot illustrate the CPOorientation in terms of the finite strain ellipsoid (i.e. X ≥ Y ≥ Z ). Polarization direction and intensity is indicated by theline lengths at each nodal point; points with no ticks exhibit no surface expression of shear-wave polarization.(a) Rotation of CPO about the X tectonic axis with respect to shear strain (g) and length of major axis of the strain ellipse(above and below, respectively). (b) Rotation of CPO about the Y tectonic axis with respect to shear strain (g) and lengthof major axis of the strain ellipse (above and below, respectively).


into unique directions of the crystal axes. Thesedispersions are responsible for the ‘diluting’ effectof quartz on whole-rock CPO. Simply stated, theelastic anisotropy due to minerals such as quartz(which do not form ‘quasi-single-crystal’ CPO) isdistributed rather than concentrated such that boththe whole-rock velocity and anisotropy of seismicwaves are reduced. This effect is indicated clearlyin Figure 7b, which is a Flinn plot representationof the Vp and AVs seismic properties for differentquartz CPOs due to different deformations.

The approach illustrated in Figure 7 for quartzcan be extrapolated to the seismically importantmica and amphibole minerals as follows. Ingeneral, natural CPO data for either mica or amphi-bole minerals are lacking when compared to thatavailable for quartz. However, it is possible tomodel the CPO expected for different deformationsdue to the simple relationship between mica/amphi-bole crystal shape and crystal axes. In effect, theshape of the mineral grains accurately reflects thedirection of the crystal axes. Both mineral groupstherefore tend to form quasi-single-crystal CPOduring deformation (e.g. Tatham et al. 2008; Lloydet al. 2009), involving the operation of relatively

few and simple crystal slip systems. These systemsare (001) ,110. and/or (001)[100] for micas and(100)[001], (010)[001] or (hk0)[001] for amphi-boles (e.g. Nicolas & Poirier 1976). Furthermore,due to the simple relationship between mineralgrain shape and crystal axes orientations, bothmineral groups may also develop similar CPO viarigid-body rotation rather than crystal slip (e.g.Meissner et al. 2006; Dıaz-Azpiroz et al. 2007;Tatham et al. 2008).

For flattening type deformations (i.e. where theFlinn K parameter approaches zero such that thetectonic axes have the relationship X � Y ≫ Z,leading to S-type tectonites), both micas and amphi-boles are expected to form strongly foliated rockfabrics. Consequently, their CPOs are characterizedby uniform distributions of a(001)/b(010) andb(010)/c(001), respectively, with the mica c[001]axis and the amphibole a(100) axis defining the foli-ation normal (Fig. 8). These CPOs lead to so-calledvertical transverse isotropy or VTI seismic distri-butions for Vp and AVs, with maximum values inboth symmetrically distributed parallel to foliation.The minimum in Vp is normal to foliation in bothmineral groups. However, the minimum in AVs is

c c

a

a

a

a

c

c

X/Y

Y/Z

1.50

0.68

1.96

0.38

6.20

5.79

7.83

0.12

2.35

3.23

0.15

6.20

5.79

7.83

0.12

6.17

5.81

6.00

.22

2.09

.46

2.35

0.07

0.22

3.80

0.02

6.29

5.77

8.36

0.12

Vp Vp

Vp

Vp

AVsAVs

AVs

AVs

AVp =6.0%

noda

taav

aila

ble

AVp =8.6%

2.27

AVp =7.0%

K = 0

AVp = 7.0%

K =

K = 1X

Y

Z

kinematiccoordinatesK =

cleftgirdle

c-axes

c-axes

c-axes

c-axes

c-axes c-axes

a-axes

a-axes

a-axes

X/Y constrictionalstrains

flatteningstrains

K = 1

Type I crossedgirdle

Y/ZK = 0

planestrains

0.37

Fig. 7. Left: idealized Flinn plot indicating the expected variation in quartz c- and a-axes CPO with different typesand magnitude of deformation (Lister & Dornsiepen 1982). Note the tendency for dispersion rather than concentrationof crystal axes. Right: Flinn plot representation of examples of natural quartz c- and a-axes CPO and the resultingVp and AVs seismic properties (data from G. E. Lloyd archive).

Fig. 8. Idealized ‘Flinn-type’ plots of (a) biotite and (b) hornblende CPO and their derived seismic properties for‘flattening’, ‘simple-shear’ and ‘constriction’ deformations. Note: 1. biotite-controlled Vp and AVs are sensitive only tofoliation due to mica VTI symmetry and therefore can distinguish only constriction deformation; 2. hornblende-controlled Vp and AVs are sensitive to foliation and lineation due to ‘orthorhombic’ symmetry and can thereforedistinguish different deformations.


Co

nst

rict

ion

K=

FlatteningK = 0

Plane st

rain

K=

1

X/Y(a)

(b)

Y/Z

seismic contours

CPO density hig

hlow

prolate ellipsoidX>>Y Z

constrictionL-tectonite

X

Z

Z

X Y

a(100) b(010) c[001]Vp AVs PS1

oblate ellipsoidX Y Z

flatteningL-tectonite

>>

Vp AVs PS1

a(100) b(010) c[001]X

X

YY Z

X>Y>Zsimple shearLS-tectonite

Vp

AVs

PS1

a(10

0)X

X Y

Y

Z

seismic contours

CPO density hig

hlow

Co

nst

rict

ion

K=

FlatteningK = 0

Plane st

rain

K=

1

X/Y

Y/Z

prolate ellipsoidX>>Y Z

constrictionL-tectonite

oblate ellipsoidX Y Z

flatteningL-tectonite

>>

X>Y>Zsimple shearLS-tectonite

XX

Y

Y

Z

a(100) b(010) c[001]

Vp

Y

AVs PS1

Vp

AVs

PS1

a(10

0)

b(010

)

c[00

1]

X

X Y

Y

Z

Vp AVs PS1

a(100) b(010) c[001]X

X

YY Z

b(010

)

c[00

1]

Fig. 8.


normal to foliation only for amphiboles; for micas, itdefines a small circle inclined to the foliationnormal. For both minerals, the orientation of Vs1Pis radial for waves propagating subvertical to foli-ation but becomes circumferential for waves propa-gating subparallel to foliation.

For plane strain deformations (i.e. K � 1 andX . Y . Z, leading to LS-type tectonites) micasand amphiboles are expected to form both foliatedand lineated rock fabrics. Consequently, theirCPOs are characterized by quasi-single-crystal dis-tributions in which (100) and [001] are parallel tolineation and foliation normal for micas, respect-ively, but this relationship is reversed for amphi-boles (Fig. 8). In both mineral groups, (010) alsolies within the foliation. Thus, as both (100) and(010) lie within the foliation and due to the VTIcharacteristics of the single crystal (Fig. 1), theseismic properties for micas exhibit exactlythe same distributions as for the previous case(Fig. 8). It is therefore not possible to distinguishbetween deformations defined by 0 , K � 1on the basis of mica seismic properties. However,amphibole single crystal do not exhibit VTI(Fig. 1); the CPO-based seismic properties are there-fore different from the previous case and can dis-tinguish both foliation (with the foliation normaldefined by the minimum in both Vp and AVs) andlineation (defined by the albeit slight maximum inboth Vp and AVs). The orientations of Vs1P foramphiboles are parallel to the YZ plane for wavespropagating subvertical to foliation and circumfer-ential for waves propagating subparallel to foliation(i.e. there is no indication of lineation).

For constrictional deformations (i.e. K � 1 andX ≫ Y � Z, leading to L-type tectonites) micas andamphiboles are expected to form strongly lineatedrock fabrics. Consequently, their CPOs are charac-terized by strong concentrations of the principalcrystal slip directions (i.e. (100) in micas and[001] in amphiboles) parallel to the tectonic exten-sion (X ) direction (Fig. 8). However, due to thefact that foliation development is impeded, neither[001] of micas nor (100) of amphiboles form orien-tation clusters but tend to form, with (010) in bothminerals, great circle CPO distributions normal toX. The resulting CPO-based seismic property distri-butions are therefore different to previous cases forboth mineral groups (Fig. 8). They both showmaximum in Vp parallel to X and a very diffusespread of low Vp velocities parallel-to-oblique tothe foliation. In contrast, it is the minimum in AVsthat aligns parallel with X for both mineral groups,while the foliation plane is defined by a narrowgreat circle of intermediate values for micas and amore diffuse great circle of alternating low-to-intermediate values for amphiboles. The orientationof Vs1P for micas is consistently parallel to XY for

all propagation directions subparallel to this plane;for all other propagation directions it is typicallycircumferential. For amphiboles, the orientationof Vs1P is parallel to either XZ or YZ for all prop-agation directions within these planes, leadingto a complex interference pattern for verticallypropagating (i.e. teleseismic) waves. For all otherpropagation directions, Vs1P is oriented circumfer-entially.

Figure 8 illustrates that it may be difficult (if notimpossible) to distinguish seismically betweenplane strain and flattening type deformations inmicaceous (i.e. felsic) rocks due to the similarityof their CPO-based seismic property distributions,which results from the VTI characteristics of micasingle crystals. However, constrictional type defor-mations, which are likely to be rarer in nature, canbe easily recognized via their distinctive CPO-basedseismic property distributions. In contrast, there is adifference between CPO-based seismic propertydistributions for plane strain and flattening defor-mations in amphibolitic (i.e. mafic) rocks, althoughthis may often be subtle, while constrictional defor-mations yield very different distributions. Theseobservations suggest that seismic property distri-butions can be used as proxies for deformationtypes and hence are potentially able to distinguishgeodynamic behaviour in mafic lithologies domi-nated by amphibole. Figure 8 also emphasizes thesignificant impact foliation development has oninfluencing the (observed) seismic properties, sup-porting observations presented above (e.g. Figs 5& 6). Indeed, it is the absence of foliation thatresults in the distinctive seismic responses of amphi-boles and particularly micas for constrictionaldeformations.

Modelling seismic properties of ductile

continental crust

This section considers the implications of the resultspresented above for the explanation and interpretationof the seismic properties of the ductile continentalcrust using the CPO-derived seismic properties(specifically Vp, AVp and AVs) of the four rocksamples described previously. It is emphasized thatthe controlling variables constraining the seismicproperties appear to be: (1) the single-crystal elasticconstants of the individual rock-forming mineralsas reflected in the modal composition; (2) CPO;(3) foliation development and orientation; and(4) deformation type.

Compressional wave velocities

Rudnick & Fountain (1995) suggested the followingVp ‘stratigraphy’ for the ductile continental crust


(e.g. Fig. 9): (1) middle crust, 6.20–6.50 km/s; (2)lower middle crust, 6.50–6.85 km/s; (3) lowercrust, 6.85–6.95 km/s; and (4) lowermost lowercrust, 7.15–7.25 km/s. Using the felsic orthogneiss,mafic dyke and pyroxene granulite as ‘rockstandards’ for the ductile continental crust, theconcept of rock recipes can be used to determinethe impact of varying modal composition on CPO-derived Vp values and, in particular, the rolesplayed by mica (biotite), amphibole (hornblende)and, to a lesser extent, pyroxene (augite).

It is clear from Figure 9a that biotite (or othermicas) cannot be responsible for the Vp valuesrecognized in the ductile continental crust. This isnot surprising given the single-crystal Vp valuesfor biotite (Fig. 1). Although muscovite has a some-what higher single-crystal Vp range than biotite,this is unlikely to be capable of increasing valuessufficiently to explain Vp values, particularly inthe middle crust. This situation becomes even

more problematical when the impact of foliationorientation is included. Based on observationsdescribed above (i.e. Figs 5, 6 & 8), regions of theductile continental crust dominated by (sub-) hori-zontal layering (i.e. foliation) should exhibit theminimum Vp values available through micas whenimaged via near-vertical (teleseismic) ray-paths(see e.g. Dıaz-Azpiroz et al. 2011). Such situationsdemand wide-angled seismic surveys.

It appears that neither mica content nor defor-mation state can account for the Vp values observedin the middle and lower continental crust, whichtherefore must be due to other minerals. FromFigure 1, the most likely candidates are quartz andparticularly feldspars, unless the rocks also consistof a mafic mineral. However, increasing the whole-rock Vp is not simply a matter of adding commonrock-forming minerals that possess relatively highVp as a rock-recipe model for the pyroxene granu-lite sample indicates (Fig. 9b). Although this rock

Micas - biotite (pyroxene granulite)

%M

od

alco

nte

nt

bio

tite

8

8

65

5

4

4

7

7

0

20

80

1006

40

60

10

10

12

12

8

8

6

6

4

4

2

20

0

20

80

100

14

14

0

%M

od

alcon

tent

bio

tite

16

16

40

60

1. Vp (max & min) km/s 2. AVp and AVs %10

10

12

12

8

8

6

6

4

4

2

20

0

20

80

100

14

14

0

%M

od

alcon

tent

ho

rnb

lend

e

16

16

40

60

8

8

65

5

4

4

7

7

0

20

80

100

%M

od

alco

nte

nt

ho

rnb

len

de

6

40

60

Amphiboles - hornblende

Orthopyroxene - Hypersthene

10

10

12

12

8

8

6

6

4

4

2

20

0

20

80

100

14

14

0

%M

od

alcon

tent o

rtho

pyro

xene

16

16

40

60

8

8

65

5

4

4

7

7

0

20

80

100

%M

od

alco

nte

nt

ort

ho

pyr

oxe

ne

6

40

60

Micas - biotite (felsic gneiss)

1. Vp (max & min) km/s 2. AVp and AVs %%

Mo

dal

con

ten

tb

ioti

te

8

8

65

5

4

4

7

7

0

(a) (b)

(c) (d)

20

80

1006

40

60

Lo

werm

ost

crust

Lo

wer

crust

Mid

dle

crust

Lo

wer

mid

dle

crust

Lo

werm

ost

crust

Lo

wer

crust

Mid

dl e

crust

Lo

wer

mid

dle

crust

Lo

werm

ost

crust

Lo

wer

crust

Mid

dl e

crust

Lo

wer

mid

dle

crust

Lo

werm

ost

crust

Lo

wer

crust

Mid

dle

crust

Lo

wer

mid

dle

crust

10

10

12

12

8

8

6

6

4

4

2

20

0

20

80

100

14

14

0

%M

od

alcon

tent

bio

tite

16

16

40

60

AVs

AVp

AVs

AVp

AVpAVs

AVpAVs

Fig. 9. Constraints on the seismic properties of the middle and lower ductile continental crust based on CPO androck-recipe modelling of: (a) felsic orthogneiss (biotite variable modal content); (b) biotite-bearing pyroxene granulite(biotite variable modal content); (c) mafic dyke (hornblende variable modal content); and (d) pyroxene granulite(hypersthene biotite variable modal content). All samples are deformed. Also shown are the typical Vp ranges (brokenlines) for different parts of the middle-to-lower crust as suggested by Rudnick & Fountain (1999). See textfor discussion.


comprises only minor biotite, its Vp values onlysatisfy those expected for the middle crust for theactual rock composition. Increasing the biotitecontent gradually decreases the maximum in Vpbut significantly decreases the minimum in Vp,such that most values of Vp fall below thoseexpected for the middle crust.

Amphibole is clearly capable of explaining Vpvalues throughout the whole ductile continentalcrust (Fig. 9c). Indeed, it does not matter whetherlayering (i.e. foliation) is flat lying or steep, Vpvalues are still well within limits. Rocks with veryhigh amphibole contents could account for theVp values observed in the deepest regions of thecontinental crust (i.e. in excess of 7 km/s). If thisinterpretation is correct, it implies that theseregions are also hydrous. However, mafic anhydrousminerals (and in particular pyroxene) can contributeeven more strongly to Vp values due to their single-crystal characteristics (Fig. 1) and are capableof achieving velocities up to 8 km/s in overtlypyroxene-dominated lithologies (Fig. 9d). It istherefore not possible to reconcile between amphi-bole and pyroxene as the specific cause of Vpvalues observed in the ductile continental crust,except where velocities exceed perhaps 7.5 km/s.Such high velocities might also be explained byincreasing garnet content, although with concomi-tant decrease in anisotropy (e.g. Brown et al. 2009).

P- and S-waves anisotropy

Unlike velocity, seismic anisotropy (both AVp andAVs) is not dependent directly upon depth as ittends to reflect CPO development and hence magni-tude of deformation rather than mineralogy per se.Nevertheless, it is clear that micas are the most seis-mically anisotropic of the common rock-formingminerals, with maximum AVp and AVs values(for biotite) of up to 64 and 114% respectively(see Fig. 1). This suggests that felsic rocks withwell-aligned micas should be the most anisotropic.The sample of felsic orthogneiss (Fig. 9a) supportsthis suggestion, showing rapid increase in AVpand particularly AVs with increasing biotite modalcontent. Similar behaviour is also shown by pro-gressively increasing the initial minor biotitecontent in the pyroxene granulite sample (Fig. 9b).Indeed, although mica (specifically biotite) contentsin excess of c. 30% are unusual (except perhapsin slates), even normal modal contents of say10–25% are more than capable of contributingsignificant anisotropy.

In contrast, amphibole modal contents of 10–25%appear to be capable of contributing only relativelylow values of AVp and AVs (Fig. 9c). However, formodal amphibole contents above c. 35%, both AVpand AVs increase rapidly and approach similar

values to those due to mica (biotite) for .60% amphi-bole. As amphibole modal contents of c. 60% may notbe unusual naturally, amphibole can therefore beregarded as a potential source of AVp and AVs inthe ductile continental crust.

The contribution of pyroxenes to anisotropy,as indicated by the pyroxene granulite sample,appears to be opposite to both micas and amphi-boles. For this particular sample, AVp remainsapproximately constant with increasing pyroxenecontent but AVs decreases progressively and even-tually converges on the AVp value (Fig. 9d). Asthis sample is deformed and the pyroxene exhibitsa relatively strong CPO, it therefore seems that pyr-oxene content may have little or no impact on AVpand a negative impact on AVs. This behaviour isperhaps explained by recalling that pyroxene exhi-bits significantly less single-crystal anisotropy, par-ticularly in AVs, compared to other commonrock-forming minerals (e.g. Fig. 1). For example,for augite AVp ¼ 24.3% and AVs ¼ 18.0% butfor hornblende AVp ¼ 27.1% and AVs ¼ 30.7%.This suggests that pyroxene-dominated lithologiesare not as anisotropic as their amphibolitic counter-parts, which is likely to have significant impli-cations for the seismic anisotropy of the lowercontinental crust.

Finally in this section, it should be recognizedthat AVs is critically azimuthally dependent,whereas AVp has no azimuthal dependence andmerely represents the absolute difference betweenthe maximum and minimum in Vp. For example,shear waves propagating normal to foliationexhibit significantly lower AVs than the maximumpossible (e.g. Fig. 5), which argues that crustalregions dominated by (sub-) horizontal layering(i.e. much of the lower continental crust accordingto Rudnick & Fountain 1995) should exhibit rela-tively low AVs values irrespective of mineralogy(e.g. Tatham 2008; Lloyd et al. 2011; Dıaz-Azpirozet al. 2011). It is therefore important to incorporatealso propagation direction (i.e. AVs combined withVs1P) when considering and/or interpreting AVsfrom continental regions. The samples describedin this section were considered simply in terms oftheir bulk AVp and AVs characteristics as functionsof composition alone (e.g. Fig. 9). In the nextsection, two of these samples – the ‘banded’ amphi-bolite and the pyroxene granulite – are consideredas microcosms of layered continental crustal sec-tions, so-called ‘seismic stratigraphy modelling’ inwhich case the azimuthal dependence of AVs canbe incorporated.

Seismic stratigraphy modelling

The sample of ‘banded’ amphibolite used to con-struct Figure 9c comprises hornblende, plagioclase


and quartz. However, these minerals have combinedtogether in different proportions to form alternatingbands of different compositions. In effect, thislayered structure mimics the compositionalchanges from felsic to mafic lithologies expectedwith increasing depth through the ductile continen-tal crust (e.g. Rudnick & Fountain 1999), althoughadmittedly not in actual sequence in the sample.By defining and analysing separately individuallayers/lithologies, their CPOs can be used toderive their particular seismic properties fromwhich a ‘seismic stratigraphy’ for Vp, AVp andAVs can be constructed and used to model theseismic response of the middle and lower continen-tal crust (Fig. 10).

Although not layered, a similar modellingapproach to that used for the banded amphibolitecan be applied to the pyroxene granulite based onrock-recipe variations in modal pyroxene, plagio-clase and biotite compositions. The seismic strati-graphy that can be constructed from this samplecan be used to model (Fig. 11) the seismic responseof not only mafic and felsic lower continental crustbut also potential anhydrous and hydrous variants

(i.e. pyroxene granulite or pyroxenite, as opposedto biotitic anorthosite granulite or biotitic pyroxenegranulite).

Middle-to-lower crust (‘banded’ amphibolite).Figure 10 is a potential CPO-derived Vp, AVp andAVs seismic stratigraphy for a middle-to-lower con-tinental crustal section. In terms of Vp (Fig. 10a),values increase progressively as compositionchanges from felsic (i.e. plagioclase–quartz domi-nated) to mafic (i.e. hornblende–plagioclasedominated) and eventually to ultramafic (i.e. horn-blende dominated). These values encompass therange of velocities suggested as being typical ofthe middle and lower crust (e.g. Rudnick & Fountain1999) and are also in close agreement with theresults obtained from rock-recipe modelling of thesheared mafic dyke sample (Fig. 9c).

In terms of anisotropy (Fig. 10b), both AVp andAVs values are relatively low for felsic (i.e. plagio-clase–quartz dominated) compositions, with AVpnoticeably smaller than AVs. With an increasingmafic (i.e. hornblende) component, AVp increasesinitially only slowly while AVs decreases slightly

10

10

12

12

8

8

6

6

4

4

2

20

P/S-wave Anisotropy (%)

H-P-Q %

92-5-3

54-35-11

31-38-31

8-49-43

40-32-28

14

14

0

S

S

S

S

P

P

P

P

P

P

16

16

N-S

‘banded’ amphibolite verticalAVs & orientation of Vs1PE-W

felsic

mafic

mid

dle

cru

stlo

wer

-mid

dle

cru

stlo

wer

cru

stlo

wes

tcr

ust

E-W

E-W

E-W

S

S

AVp

AVs

E-W

SPMaxima in AVp & AVs per

compositional band

8

8

65

5

4

4

7

7

(a) (b)P-wave Velocity (km/s)

H-P-Q %

92-5-3

54-35-11

31-38-31

8-49-43

40-32-28

‘co

mp

osi

on

ald

epth

’

6

Lo

werm

ost

crust

Lo

wer

crust

Mid

dle

crust

plagioclase-quartz

hornblende-plagioclase quartz-

plagioclase-quartz-hornblende

hornblende-plagioclase

‘ban

ded

’am

ph

ibo

lite

com

po

siti

on

(sca

led

to h

orn

ble

nd

e) ‘ban

ded

’amp

hib

olite

com

po

sition

(scaledto

ho

rnb

lend

e)

Lo

wer

mid

dle

crust

hornblende

Vp range percompositional

band

Fig. 10. Constraints on the seismic properties of the middle and lower ductile continental crust based on CPO androck-recipe modelling of individual compositional bands in a deformed ‘banded’ amphibolite, taken to simulate amiddle to lower continental crust compositional profile (i.e. felsic to mafic compositions). Also shown (grey) are theresults from Figure 9c of ‘rock-recipe’ modelling for amphibole (hornblende) content in the mafic dyke sample andtypical Vp ranges for different parts of the middle-lower crust as suggested by Rudnick & Fountain (1999). See textfor discussion. (a) P-wave velocity; (b) P- and S-waves anisotropy.


and becomes less than AVp for intermediate compo-sitions (i.e. c. 1/3 mafic content). At c. 50% maficcontent, AVs exceeds AVp and increases morerapidly towards ultramafic (i.e. so-called ‘hornblen-dite’) compositions. These results are in broadagreement with the behaviour exhibited by themafic dyke, including the tendency for AVs ,AVp for felsic compositions (Fig. 9c).

Due to the azimuthal dependence of AVs,simply plotting maximum values of shear-wavesplitting can be misleading (e.g. Fig. 6). This situ-ation is exacerbated by the impact of foliationand/or compositional layering orientation (e.g.Figs 5 & 8). To emphasize this effect, the layeringin the banded amphibolite sample has been con-sidered to be oriented horizontally in a geogra-phical sense. For a vertically propagating (i.e.teleseismic) shear wave, the value of AVs foreach composition is significantly smaller (i.e.from 2% to 3%) than the maximum recorded(Fig. 10b). Indeed, for most compositions it issmaller than the azimuthally independent absoluteP-wave anisotropy. Furthermore, the polarizationdirection of the fast (teleseismic) shear wave (i.e.Vs1P) is not constant and changes through 908

as the composition changes from felsic to inter-mediate, beyond which it remains constant. Thisbehaviour of Vs1P clearly reflects the increasinginfluence of the amphibole (hornblende). FromFigure 8b, this would indicate not only a horizon-tal foliation/layering (as initially defined) but alsoan ‘east–west’ (in the arbitrary coordinate systemassumed) tectonic Y-direction and hence a ‘north–south’ oriented lineation. Thus, this banded amphi-bolite sample can be interpreted from the CPO-derived seismic characteristics as being an LStectonite that developed under conditions close toplane strain (i.e. K � 1).

Lower crust (pyroxene granulite). Figure 11 is apotential CPO-derived Vp, AVp and AVs seismicstratigraphy for a lower continental crustal section.In terms of Vp (Fig. 11a), values increase progress-ively as composition changes from felsic (i.e.biotite–plagioclase dominated) to mafic (i.e. pyrox-ene dominated) and eventually to ultramafic (i.e.so-called ‘pyroxenites’). However, the most felsiccompositions (e.g. biotitic anorthosite granulite)exhibit Vp values less than those expected foreven the middle crust, while compositions between

felsic

mafic

low

erm

idd

lecr

ust

low

e rcr

ust

10

10

12

12

8

8

6

6

4

4

2

2

P/S-wave Anisotropy (%)0

20

100

14

14

S

S

P

16

16

pyroxene granulite verticalAVs & orientation of Vs1PNE-SW

40

NE-SW S

S

P

P

SP Maxima in AVp & AVs percompositional band

NE-SW

NE-SW

AVp AVs

E-W

8

8

65

5

4

4

7

7

(a) (b)P-wave Velocity (km/s)0

20

100

gra

nu

lite

com

po

siti

on

(sca

led

top

yro

xen

e)

‘co

mp

osi

on

ald

epth

’

6

40

mid

dle

cru

stlo

wes

tcr

ust

Lo

werm

ost

crust

Lo

wer

crust

Mid

dle

crust

biotitic pyroxenegranulite

pyroxenegranulite

biotitic anorthositegranulite

Lo

wer

mid

dle

crustpyroxenite

Vp range percompositional

band

‘hyd

rou

s’‘a

nh

ydro

us’

PPx Pl B Q %

15 51 20 14

50 30 15 5

60 26 7 7

80 13 3 4

gran

ulite

com

po

sition

(scaledto

pyro

xene)

Px Pl B Q %

15 51 20 14

50 30 15 5

60 26 7 7

80 13 3 4

Fig. 11. Constraints on the seismic properties of the middle and lower ductile continental crust based on CPO androck-recipe modelling of individual compositional bands in a deformed pyroxene granulite, taken to simulate inparticular a lowermost middle to lowest continental crust compositional profile. The presence of biotite in some ‘bands’permits not only felsic to mafic compositions but also hydrous to anhydrous conditions to be considered. Also shown(grey) are the results from Figure 9d of ‘rock-recipe’ modelling for pyroxene (hypersthene) content in this rock andtypical Vp ranges for different parts of the middle-lower crust as suggested by Rudnick & Fountain (1999). See textfor discussion. (a) P-wave velocity; (b) P- and S-waves anisotropy.


biotitic anorthosite granulite and biotitic pyroxenegranulite yield Vp values typical of the middlecrust (e.g. Rudnick & Fountain 1999). In general,essentially mafic compositions (i.e. biotitic pyrox-ene and pyroxene granulites) are required toproduce typical lower crustal Vp values. Ultramaficcompositions exhibit Vp values in excess of thoseexpected in typical lower crust.

In terms of anisotropy, AVp is relativelysmall for felsic (i.e. biotite–plagioclase dominated)compositions but AVs is much larger (Fig. 11b). Asthe mafic component steadily increases towardspyroxene-dominated compositions, AVp remainsessentially constant but AVs reduces significantlyand therefore trends towards AVp. These resultsare obviously in broad agreement with thoseshown in Figure 9c due to the fact that the samesample is involved in both. However, there aresome differences because the seismic-stratigraphycompositions have been defined via rock recipes.

To investigate the impact of azimuthal depen-dence on AVs, the sample of pyroxene granulite(which is an XZ tectonic section) is considered tohave been oriented with X horizontal and Z verticalgeographically (i.e. horizontal foliation and/orlayering). For a vertically propagating (i.e. teleseis-mic) shear wave, the value of AVs for each compo-sition considered is not only significantly smallerthan the maximum recorded but it is also smallerthan the equivalent and azimuthally independentabsolute AVp value (Fig. 11b). The magnitude ofthe vertical AVs also decreases with increasingmafic (pyroxene) content, although a slight increasemay be significant at the change from effectivelyfelsic to effectively mafic granulite compositions.The polarization direction of the fast (teleseismic)shear wave (i.e. Vs1P) is essentially constant ‘north-east–southwest’ for all compositions except for theslight increase in AVs values, where it is ‘east–west’. The behaviours of the vertical (teleseismic)AVs and Vs1P clearly reflect the increasing influ-ence of pyroxene and suggest that mafic and ultra-mafic lithologies with (sub-) horizontal layeringmay exhibit little recognizable and/or interpretableshear-wave splitting anisotropy (e.g. see Dıaz-Azpiroz et al. 2011).

Discussion

While recent studies suggest that crustal defor-mation is distinguishable via seismic profilingusing velocity and/or anisotropy (e.g. Meltzeret al. 2001; Meltzer & Christensen 2001; Chlupa-cova et al. 2003; Brown et al. 2009; Moschettiet al. 2010), the source(s) of these seismic propertiesmust be constrained. In the ductile continental crust,the usual source is the CPO characteristics of the

rocks through which the seismic waves travel. Itappears to be tacitly assumed that CPO not onlyindicates foliation but is also mimicked by theseismic symmetry (e.g. Siegesmund et al. 1989;Okaya & Christensen 2002; Kitamura 2006; Meiss-ner et al. 2006; Barberini et al. 2007). Thus, micasare generally regarded as controlling the seismicproperties of much of the ductile continental crust(e.g. Vernik & Liu 1997; Meltzer & Christensen2001; Nishizawa & Yoshino 2001; Mahan 2006;Valcke et al. 2006; Lloyd et al. 2009). However,there is little evidence for meta-pelitic layers indeep continental crust as increasing Vp valueswith depth are interpreted as indicating increasingmafic component (e.g. Rudnick & Fountain 1999;Rudnick & Gao 2003). Furthermore, the presentcontribution has shown (i.e. Figs 1–8) that whilerelationships exist between mineralogy, CPO,deformation and seismic symmetry, the precisenatures need to be established rigorously, usuallyfor each geodynamic situation.

Figures 9–11 summarized these results andinitially considered the potential origins of theseismic properties of the middle and lower ductilecontinental crust, which form the basis for manytectonic and/or geodynamic interpretations. Thesimple questions to be posed/answered whenconsidering the seismic properties of the ductilecontinental crust are therefore: (1) what seismic vel-ocity and/or anisotropy is/are observed/needed;(2) what mineralogy/composition and/or CPO/deformation state is/are observed/required; and/or (3) what tectonic/geodynamic setting(s)explain(s) the seismic properties and/or geologyobserved? The rest of this discussion considersthese questions based on the results presented inthis study and provides some caveats for the accu-racy and reliability of some current geodynamicmodels based on seismic profiling of the ductile(middle and lower) continental crust.

Seismic velocity (Vp) profiling

The typical mica modal contents of most rocks is,30% and alone cannot account for observed mid-crustal Vp values of 6.2–6.8 km/s, particularly iffoliation/layering is (sub-) horizontal (e.g. Fig. 9a,b). Consequently, other faster common rock-forming minerals (such as feldspars) must makethe most significant contributions (e.g. Figs 1, 2,10a & 11a). Micas therefore cannot be regardedas being responsible for the Vp values observedin the ductile continental crust. Furthermore, whilemicas are amongst the most significant and sensitivefoliation-forming minerals, their VTI characteristicsmake them insensitive to most deformation states(e.g. Fig. 8a). Indeed, it appears that micas arecapable only of distinguishing constrictional


deformations, ironically due to the lack of foliationdevelopment. Exceptional care must therefore betaken when attempting to infer tectonic states frommicaceous rocks, particularly where the orientationand/or number of any foliation(s) is/are unknown.

In contrast to micas, the wide range of amphibolemodal contents possible naturally (i.e. up to andexceeding 65%) could explain observed Vp valuesin not only the middle but also much of the lowercontinental crust (e.g. Figs 9c & 10a). Indeed, Vpvalues in excess of 7 km/s and regarded as indica-tive of the lowermost continental crust can beexplained by hydrous rocks with very high amphi-bole contents (i.e. so-called ‘hornblendites’) andsubvertical foliation(s). However, as with micas,amphiboles form foliation/layering relativelyeasily and (sub-) horizontally oriented foliation/layering acts to reduce Vp values significantly(e.g. Figs 9c & 10a). Nevertheless, amphiboles(unlike micas) are sensitive to most deformationstates due to their lack of VTI characteristics (e.g.Fig. 8b). The Vp characteristics of most of theductile continental crust can therefore be explainedby amphiboles alone, but conditions must beregarded as being hydrous.

The highest Vp values recognized are providedby pyroxenes (e.g. Figs 9d & 11a). Indeed, so-called‘pyroxenites’ are capable of exhibiting velocities ofc. 8 km/s, perhaps associated more usually withupper mantle lithologies (e.g. Ben Ismail & Main-price 1998). Even relatively low pyroxene modalcontents can explain Vp values associated with themiddle continental crust, while pyroxene granulitecompositions readily account for most lower conti-nental crustal values (e.g. Figs 9a & 11b). Although(sub-) horizontally oriented foliation/layering willagain act to reduce Vp, values are likely to remainsufficiently high to explain most middle and espe-cially lower crustal observations. The Vp character-istics of most of the ductile continental crust cantherefore be explained by pyroxenes alone, but con-ditions must be regarded as being anhydrous.

In summary, it is unlikely that seismic profilingobservations based on Vp characteristics aloneare capable of providing accurate and/or reliableinterpretations of the middle or lower ductile conti-nental crust. Both amphiboles and pyroxenes yieldacceptable Vp values but micas do not and requiresignificant contribution from other common rock-forming mineral phases (e.g. feldspars).

Seismic anisotropy (AVp, AVs) profiling

Both mica-dominated felsic and amphibole-dominated mafic lithologies are capable of gen-erating the levels of AVp and AVs recognizednaturally (e.g. Figs 9a–c & 10). In detail, micasand amphiboles contribute differently to whole-rock

anisotropy. In felsic rocks, relatively low (i.e.,30%) modal mica contents are offset by relativelyhigh mica single-crystal elastic anisotropy (Fig. 1).In mafic rocks, relatively high (.50%) modalamphibole contents offset relatively low amphibolesingle-crystal elastic anisotropy. However, theimpact of micas on both AVp and AVs appears tobe more significant overall (compare Figs 9a & 9c).

In contrast to micas and amphiboles, pyroxenesdo not appear to contribute significantly to whole-rock anisotropy (e.g. Figs 9d & 11). This behaviouris clearly related to the relatively low single-crystalelastic anisotropy of pyroxenes (Fig. 1).

In summary, significant AVp and AVs valuesobserved in the ductile middle to lower continentalcrust are due to either mica (felsic) and/or amphi-bole (mafic hydrous) lithologies. Furthermore, rela-tively low or intermediate values of anisotropy couldalso be due to pyroxene (mafic anhydrous) litholo-gies. It will therefore be difficult and/or impossibleto distinguish between contrasting felsic and maficcompositions and/or conditions using seismic aniso-tropy profiling observations alone.

Seismic shear-wave splitting polarization

(AVs, Vs1P) profiling

AVs values measured normal to foliation and/orcompositional layering are typically significantlylower than the maximum AVs values possible(e.g. Figs 5 & 6) due to the composition and intrinsicproperties of the single-crystal elastic anisotropyand whole-rock CPO (e.g. Figs 1 & 8). In addition,the polarization direction of the fast shear wave (i.e.Vs1P) may also vary with shear-wave propagationdirection, which similarly depends on composition,single-crystal elastic anisotropy and CPO. Thus,variations in AVs–Vs1P (e.g. Figs 10 & 11) maynot necessarily reflect different kinematics, as isoften assumed in many geodynamic interpretationsof seismic profiling data (e.g. Ozacar & Zandt2004; Shapiro et al. 2004; Sherrington et al. 2004;Schulte-Pelkum et al. 2005). For example, if thelower continental crust comprises layered maficlithologies (whether amphibole or pyroxene domi-nated) and the layering is oriented (sub-) horizon-tally as is often assumed (e.g. Rudnick & Fountain1999), it is likely to exhibit only low shear-wavesplitting anisotropy and may even appear essentiallyisotropic irrespective of its actual composition anddeformation state (e.g. see Dıaz-Azpiroz et al.2011).

Seismic properties of ductile continental

crust: an example

Recently, Moschetti et al. (2010) used ambientnoise tomography to argue for strong, deep


(middle-to-lower) crustal radial seismic anisotropybeneath the western USA, confined mainly to geo-logical provinces that have undergone significantextension in the last 65 Ma. They consider that thecoincidence of crustal radial anisotropy with exten-sional provinces suggests that the anisotropy resultsfrom the CPO of anisotropic crustal minerals causedby extensional deformation. Consequently, theirobservations provide support for the hypothesisthat the deep crust within these regions has under-gone widespread and relatively uniform strain inresponse to crustal thinning and extension. Wenow consider the results of Moschetti et al. (2010)in terms of the observations presented in thiscontribution.

In general, the seismic results of Moschettiet al. (2010) recognize seismic anisotropy in themiddle-to-lower crust and hence support our basicconclusion that the ductile continental crust can con-tribute significantly to the overall seismic anisotropydetected in the Earth. However, can their seismicresults be interpreted geologically by the obser-vations and relationships established in the presentstudy? To attempt to answer this question we haveincorporated the best-fit results of Moschetti et al.(2010, supplementary fig. 6a, b), which predictAVs of c. 5%, into our observations on the impactof different modal contents of mica (biotite), amphi-bole (hornblende) and pyroxene (hypersthene) onseismic properties (i.e. Figs 9–11). For the purposesof this comparison, these figures have been amendedto include the maximum and minimum in the CPO-derived Vs1 and Vs2. To comply with the analysispresented by Moschetti et al. (2010), these aretermed VSV and VSH here to represent a horizontallypropagating shear wave split into vertical and hori-zontal components (Fig. 12).

According to the results presented previously(Fig. 9), felsic lithologies with mica contents of,18% are required to explain the middle crust Vsvalues indicated by Moschetti et al. (2010) whiletheir AVs of c. 5% require ,5% modal mica(Fig. 12a). However, only the latter compositioncan result in typical middle crustal Vp values (e.g.Rudnick & Fountain 1995). Felsic lithologies withmica contents of up to 18% require the presenceof one or more seismically ‘fast’ minerals, that donot impact on anisotropy, to achieve middlecrustal Vp values. The lower crustal Vs and AVsvalues indicated by Moschetti et al. (2010) cannotbe achieved by variation in modal mica content inthe felsic lithology considered here (Fig. 12a). Inaddition, micas cannot explain the overall resultsindicated by Moschetti et al. (2010) due to theirVTI characteristics, which force isotropy in theradial plane for flattening and plane strain (simple-shear) deformations (e.g. Fig. 8a). Consequently,any anisotropy observed in the radial plane must

be due to other minerals, which may be possiblefor low biotite contents.

While hydrous mafic lithologies with amphibolecontents of 27–72% (Fig. 10) readily explain thelower crustal Vs values indicated by Moschettiet al. (2010), even the lowest amphibole contentsyield shear-wave velocities that are too high fortheir middle crust (Fig. 12b). Notwithstandingthese comparisons, increasing amphibole contentcan account for the Vp values expected for themiddle and lower crust (e.g. Rudnick & Fountain1995). The AVs value of c. 5% predicted byMoschetti et al. (2010) can be satisfied by amphi-bole contents of 35–40%, which are compatiblewith both middle and lower crustal Vp valuesand the Vs values predicted by Moschetti et al.(2010) for their lower crust (Fig. 12b). In addition,amphiboles exhibit radial VTI characteristics forflattening type deformations but, for deforma-tions where K � 1 (e.g. plane strain, simple shear,etc.), they exhibit slight radial anisotropy. Any ani-sotropy observed in the radial plane must thereforebe due to either other minerals for flattening typedeformations, which implies relatively low amphi-bole contents, or implies some type of sheardeformation.

Anhydrous mafic lithologies with pyroxene con-tents of ,18% and 17–37% explain the Vs valuesof Moschetti et al. (2010) for the middle andlower crust, respectively (Fig. 12c). However, Vpvalues for the former are less than expected for themiddle crust while for the latter Vp is only thatexpected for the middle crust (e.g. Rudnick &Fountain 1995). Furthermore, AVs values are con-siderably higher than the 5% predicted by Moschettiet al. (2010), probably due to the lack of impactof pyroxene.

The ratio Vp/Vs is often used seismologically tocharacterize rock types and, in particular, to recog-nize hot rocks and/or the presence of melt.Moschetti et al. (2010) also calculated the variationin this parameter with depth in their seismic ana-lyses but observed little change (Fig. 12 insets).In terms of the CPO-derived seismic properties,the ratio Vp/Vs can be defined either by Vp-max/Vs2-min to give a maximum, or by Vp-min/Vs1-max to give a minimum value. A range ofVp/Vs values can therefore be plotted with vari-ation in, for example, modal content of a specificmineral (Fig. 12 insets). The ranges of Vp/Vsexpand significantly with increasing modal contentof, particularly, mica in felsic lithologies and amphi-bole in hydrous mafic lithologies (Fig. 12a, b insets).In contrast, the increasing modal content of pyrox-ene in anhydrous mafic lithologies results in acontraction of this range (Fig. 12c inset). Neverthe-less, the ranges predicted from the CPO-derivedseismic properties generally bracket the values


Pyroxenes - hypersthene

2.5

2.0

1.5

1.0

Vp/

Vs

VV

p-max

s2-min

VV

p-min

s1-max

0 20 40 60 80 100

M-S6b

Pyroxene%

10

10

12

12

8

8

6

6

4

4

2

20

0

20

80

100

14

14

0

%M

od

alcon

tent o

rtho

pyro

xene

16

16

40

60

orthopyroxeneAVsAVp

8

8

65

5

43

43

7

7

0

20

80

100

%M

od

alco

nte

nt

ort

ho

pyr

oxe

ne

6

40

60

Middle

crustV

p-max

orthopyroxene

Vp-m

in

Vs-m

ax

orthopyroxene

Vs-m

in

Lower

middle

crust

1. Vp (max & min) km/s 2. AVp and AVs %

VSV

VSVVSH

VSH

lowercrust

middlecrust

Amphiboles - hornblende

2.50

2.00

1.50

1.00

Vp/

Vs

VV

p-max

s2-min

VV

p-min

s1-max

0 20 40 60 80 100

M-S6b

Hornblende%

10

10

12

12

8

8

6

6

4

4

2

20

0

20

80

100

14

14

0

%M

od

alcon

tent

ho

rnb

lend

e

16

16

40

60

maficdyke

AVp AVs

8

8

65

5

43

3 4

7

7

0

20

80

100

%M

od

alco

nte

nt

ho

rnb

len

de

6

40

60

Vp-max

Vs-max

Vp-min

Vs-min

1. Vp (max & min) km/s 2. AVp and AVs %

VSV

VSV

VSH

VSH

lowercrust

middlecrust

Micas - biotite

AVs

AVs

Middle

crust

Lower

middle

crust

Lowerm

ost crustLow

ercrust

2.50

2.00

1.50

1.00

Vp/

Vs

VV

p-max

s2-min

VV

p-min

s1-max

0 20 40 60 80 100Biotite%

%M

od

alco

nte

nt

bio

tite

1. Vp (max & min) km/s 2. AVp and AVs %8

8

65

5

4

3

3

2

2

4

7

7

0

20

80

1006

40

60

Lowerm

ost crustLow

ercrust

Middle

crustV

p-max

biotitefelsic

gneissV

p-minV

s1-max

biotitefelsic

gneissV

s2-min

Lower

middle

crust

10

10

12

12

8

8

6

6

4

4

2

20

0

20

80

100

14

14

0

%M

od

alcon

tent

bio

tite

16

16

40

60

biotite AVs

(felsic gneiss)

biotite AVp

(felsic gneiss)

VSV

VSVVSH

VSHAvs

lowercrust

middlecrust

M-S6b

Lower

crust

Lowerm

ostcrust

(a)

(b)

(c)

Fig. 12.


suggested by Moschetti et al. (2010), except forpyroxene contents in excess of 80%.

Based on the observations presented in this con-tribution, we can therefore explain geologically theseismic results described by Moschetti et al. (2010)in terms of a dominantly amphibolitic and hencehydrous middle/lower continental crust beneaththe current western USA. A composition of c. 40%amphibole and c. 60% felsic minerals (e.g. plagio-clase and quartz) is compatible with our results. Fur-thermore, the extensional deformation indicated isdue to horizontal (simple?) shear rather than verticalflattening, accommodated by the development ofCPO in the amphibole to produce the observedradial crustal anisotropy of c. 5%.

Seismic profiling strategy for the ductile

continental crust

The observations made and experiences gained inthis contribution have led us to suggest the follow-ing strategy for in situ seismic profiling of theductile continental crust. Essentially, there is onebasic question that needs to be considered: whatseismic velocities and/or anisotropies are observedand/or needed to explain a particular geodynamicsetting? To answer this question it is important tomake use of as much seismic information as poss-ible as individual parameters (e.g. Vp, Vs, AVp,AVs, etc.) are sensitive to different geological con-straints and hence provide information on differentaspects of the geodynamical setting, as the follow-ing considerations illustrate.

Relatively low mica modal contents are capableof generating strong mica CPO in felsic rocks,resulting in relatively large seismic anisotropy butrelatively low P-wave velocity. Thus, anisotropicfelsic rocks with typical middle crustal P-wavevelocities must consist of mica(s) and seismicallyfaster minerals. However, if any associated foliationis oriented normal to the seismic wave propagationdirection, both anisotropy and velocity will be mini-mized. Nevertheless, due to the VTI characteristicsof mica(s) (which tend to dominate seismic propertydistributions in three dimensions), it will be difficultto distinguish specific types of deformation (except

constriction) unless the other minerals are able toimpart significant modifications.

To generate similar seismic anisotropy in maficrocks as micaceous felsic rocks requires large butnot unrealistic amphibole modal contents, whichcan also account for most middle and lowercrustal P-wave velocities. Such lithologies implyhydrous conditions. Again, foliation orientation issignificant in terms of the velocity and particularlyanisotropy detected. The non-VTI characteristicsof amphibole mean that different types of defor-mation (e.g. flattening, simple shear, constriction)can be distinguished.

The fastest P-wave velocities are associated withpyroxene-dominated lithologies which can accountfor velocities associated with both the lower crustand upper mantle, particularly when garnet is alsopresent (see Brown et al. 2009). Such situationsimply anhydrous conditions. However, it appearsthat such lithologies tend to exhibit relatively lowanisotropy (whether normal or parallel to foliation),especially when garnetiferous. Although in princi-pal the non-VTI characteristics of pyroxene meanthat different types of deformation can be distin-guished, in practice this may prove difficult dueto the lack of detail in the seismic propertydistributions.

Conclusions

Results of rock- and fabric-recipe modelling usingrepresentative rock samples suggest that theseismic properties of the middle and lower continen-tal crust depend on CPO and deformation as well ascomposition, while reliable interpretation of seismicsurvey data should incorporate as many seismicproperties (i.e. Vp, AVp, AVs, Vs1, Vs2, Vs1P) aspossible. Crustal AVs is controlled by the dominantrock fabric-forming mineral: micas provide themost significant values, even at low concentrations,and amphiboles are significantly weaker except athigher concentrations. As lithology and anisotropyvary with depth, AVs can have a significantcrustal component (e.g. steep, pervasive coaxialmica fabrics). Micas do not contribute significantlyto Vp, however; observed Vp and AVs values imply

Fig. 12. Explanation of the seismic results of Moschetti et al. (2010) in terms of the potential geology of themiddle-to-lower crust beneath the western USA using observations presented in this contribution (see Figs 9–11)amended to include the CPO-derived shear-wave velocities. The black boxes in the velocity graphs represent thesplitting of horizontally propagating shear waves into vertical and horizontal components, while in the anisotropygraphs they represent the best estimate AVs as reported by Moschetti et al. (2010, fig. 6a, b) for their best-fitting results,plotted in order to intercept with the respective CPO-derived values. The inset in each part compares their Vp/Vsratio with the ranges predicted via CPO (NB minor inflection points indicate change from upper, middle and lower crustand mantle). (a) Felsic lithologies (varying modal biotite content); (b) hydrous mafic lithologies (varying modalamphibole content); and (c) anhydrous mafic lithologies (varying modal pyroxene content).


the middle/lower crust is dominated by mafic lithol-ogies, including ‘amphibolites’. Seismic symmetryis crucial in distinguishing strain type. Micasexhibit VTI and are unlikely to distinguish eitherradial deformation typical of flattening strains (i.e.X � Y ≫ Z; K � 0) or ductile deformation typicalof simple-shear zones (i.e. X . Y . Z; K � 1), wheremultiple fabrics may also interfere. Amphibolesexhibit ‘orthorhombic’ symmetry and distinguishpotentially all common strain types, providingmore sensitive tests of anisotropy-derived tectonicmodels (e.g. anisotropic lower crust with high Vpvalues can be explained by foliated amphibolites,although pyroxene granulites with only minorbiotite content can achieve similar impact). Theseresults pose significant constraints for many cur-rently extant seismically based geodynamic models.

The samples of Nanga Parbat gneisses were collectedduring fieldwork funded by a Royal Society researchgrant (RWHB). We are grateful to Tamsin Lapworth andV. Prasannakumar for the samples of ‘banded’ amphiboliteand pyroxene granulite, respectively, used in this study.We thank Craig Storey and Stanislav Ulrich for their con-structive reviews and Dave Prior for his editorial com-ments that have helped to improve the original versionof this contribution. The UK NERC supported part of theSEM/EBSD facilities via Small Grant GR9/3223 (GEL,MC) and a PhD studentship to DJT.

References

Abramson, E. H., Brown, J. M., Slutsky, L. J. & Zaug,J. M. 1997. The elastic constants of San Carlos olivineto 17 GPa. Journal of Geophysical Research, 102,12 253–12 263.

Aleksandrov, K. S. & Ryzhova, T. V. 1961. The elasticproperties of rock-forming minerals, I: pyroxenes andamphiboles. Izvestiya Academy of Sciences, USSR,Geophysics, Series 9, 1339–1344 (in Russian).

Aleksandrov, K. S., Alchikov, V. V., Belikov, B. P.,Zaslavskii, B. I. & Krupnyi, A. I. 1974. Velocitiesof elastic waves in minerals at atmospheric pressureand increasing of precision of elastic constants bymeans of EVM. Izvestiya Academy of Sciences,USSR, Geophysics, Series 10, 15–24 (in Russian).

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

Barberini, V., Burlini, L. & Zappone, A. 2007. Elasticproperties, fabric and seismic anisotropy of amphibo-lites and their contribution to the lower crust reflectiv-ity. Tectonophysics, 445, 227–244.

Barruol, G. & Mainprice, D. 1993. 3-D seismic vel-ocities calculated from lattice-preferred orientationand reflectivity of a lower crustal section: examplesof the Val Sesia section (Ivrea zone, northernItaly). Geophysical Journal International, 115,1169–1188.

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

Bonczar, L. J., Graham, E. K. & Wang, H. 1977.Pressure and temperature dependence of elastic-constants of pyrope garnet. Journal of GeophysicalResearch, 82, 2529–2534.

Brown, D., Llana-Funez, S., Carbonell, R., Alvarez-

Marron, J., Marti, D. & Salisbury, M. 2009.Laboratory measurements of P-wave and S-wave vel-ocities across a surface analog of the continental crust-mantle boundary: Cabo Ortegal, Spain. Earth andPlanetary Science Letters, 285, 27–38.

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

Butler, R. W. H., Casey, M., Lloyd, G. E., Bond, C. E.,McDade, P., Shipton, Z. & Jones, R. 2002. Verticalstretching and crustal thickening at Nanga Parbat,Pakistan Himalaya: a model for distributed continentaldeformation during mountain building. Tectonics, 21,doi: 10.1029/2001TC901022.

Caporali, A., Aichhorn, C. et al. 2009. Surfacekinematics in the Alpine–Carpathian–Dinaric andBalkan region inferred from a new multi-networkGPS combination solution. Tectonophysics, 474,295–321.

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. & Fountain, D. 1975. Constitution ofthe lower continental crust based on experimentalstudies of seismic velocities in granulite. GeologicalSociety of America Bulletin, 86, 227–236.

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

Dandekar, D. P. 1968. Elastic constants of calcite.Journal of Applied Physics, 39, 3694–3699.

Dıaz-Azpiroz, M., Lloyd, G. E. & Fernandez, C. 2007.Development of lattice preferred orientation inclinoamphiboles deformed under low-pressure meta-morphic conditions – a SEM/EBSD study of metaba-sites from the Aracena metamorphic belt, SW Spain.Journal of Structural Geology, 29, 629–645.

Dıaz-Azpiroz, M., Lloyd, G. E. & Fernandez, C. 2011.Deformation mechanisms of plagioclase and seismicanisotropy of the Acebuches metabasites (SW Iberianmassif). 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, 79–95.

Flinn, D. 1956. On the deformation of the Funzie Con-glomerate, Fetlar, Shetland. Journal of Geology, 64,480–505.

Fouch, M. J. & Rondenay, S. 2006. Seismic anisotropybeneath stable continental interiors. Physics of theEarth and Planetary Interiors, 158, 292–320.

Fountain, D. & Christensen, N. 1989. Compositionof the continental crust and upper mantle; a review.In: Pakiser, L. & Mooney, W. (eds) GeophysicalFramework of the Continental United States. Geolog-ical Society of America, Boulder, Memoir, 172,711–742.

Hall, S. A., Kendall, J.-M., Maddock, J. & Fisher, Q.2008. Crack density tensor inversion for analysis of


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

changes in rock frame architecture. GeophysicalJournal International, 173, 577–592.

Handy, M. 1994. Flow laws for rocks containing two non-linear viscous phases: a phenomenological approach.Journal of Structural Geology, 16, 287–301.

Ji, S., Wang, Q. & Xia, B. 2002. Handbook of SeismicProperties of Minerals, Rocks and Ores. PolytechnicInternational Press, Canada.

Kendall, J.-M., Fisher, Q. et al. 2007. Seismic aniso-tropy as an indicator of reservoir quality in siliclasticrocks. In: Jolley, S., Barr, D., Walsh, J. J. &Knipe, R. J. (eds) Structurally Complex Reservoirs.Geological Society, London, Special Publications,292, 123–136.

Kern, H. & Wenk, H.-R. 1990. Fabric-related velocityanisotropy and shear wave splitting in rocks fromSanta Rosa Mylonite Zone, California. Journal ofGeophysical Research, 95, 11 213–11 223.

Kitamura, K. 2006. Constraint of lattice-preferred orien-tation (LPO) on Vp anisotropy of amphibole-richrocks. Geophysical Journal International, 165,1058–1065.

Kruhl, J. 1987. Preferred lattice orientations of plagio-clase from amphibolite and greenschist facies rocksnear the Insubric Line (Western Alps). Tectonophysics,135, 233–242.

Lapworth, T., Wheeler, J. & Prior, D. J. 2002.The deformation of plagioclase investigated usingelectron backscatter diffraction crystallographic pre-ferred orientation data. Journal of StructuralGeology, 24, 387–399.

Lister, G. & Dornsiepen, U. 1982. Fabric transitions inthe Saxony granulite terrane. Journal of StructuralGeology, 4, 81–92.

Lister, G. S. & Hobbs, B. E. 1980. The simulationof fabric development during plastic deformation andits application to quartzite – the influence of defor-mation history. Journal of Structural Geology, 2,355–370.

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 &Planetary Sciences Letters, 288, 320–328.

Lloyd, G. E., Halliday, J. M., Butler, R. W. H., Casey,M., Kendall, J.-M., Wookey, J. & Mainprice, D.2011. From crystal to crustal: petrofabric derivedseismic modelling of regional tectonics. In: Prior,D. J., Rutter, E. H. & Tatham, D. J. (eds) DeformationMechanisms, Rheology and Tectonics: Microstructures,Mechanics and Anistropy. Geological Society, London,Special Publications, 360, 49–78.

McSkimin, H. J., Andreatch, P. Jr. & Thurston, R. N.1965. Elastic moduli of quartz versus hydrostaticpressure at 258C and 2195.88C. Journal of AppliedPhysics, 36, 1624–1632.

Mahan, K. 2006. Retrograde mica in deep crustalgranulites: implications for crustal seismic anisotropy.

Geophysics Research Letters, 33, Art. No. L24301Dec 16.

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

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

Mainprice, D., Bouchez, J., Blumenfeld, P. & Tubia,J. 1986. Dominant c slip in naturally deformedquartz: implications for dramatic plastic softening athigh temperature. Geology, 14, 819–822.

Marshall, D. & McLaren, A. 1977a. The direct obser-vation and analysis of dislocations in experimentallydeformed plagioclase feldspars. Journal of MaterialsScience, 12, 893–903.

Marshall, D. & McLaren, A. 1977b. Deformationmechanisms in experimentally deformed plagioclasefeldspars. Physics and Chemistry of Minerals, 1,351–370.

Meissner, R., Rabbel, W. & Kern, H. 2006. Seismiclamination and anisotropy of the lower continentalcrust. Tectonophysics, 416, 81–99.

Meltzer, A. & Christensen, N. 2001. Nanga Parbatcrustal anisotropy: implications for interpretationof crustal velocity structure and shear-wavesplitting. 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.

Moschetti, M. P., Ritzwoller, M. H., Lin, F. & Yang,Y. 2010. Seismic evidence for widespread western-USdeep-crustal deformation caused by extension. Nature,464, 885–889.

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

Nishizawa, O. & Yoshino, T. 2001. Seismic velocityanisotropy in mica rich rocks: an inclusion model.Geophysical Journal International, 145, 19–32.

Okaya, D. A. & Christensen, N. I. 2002. Anisotropiceffects of non-axial seismic wave propagation in foli-ated crustal rocks. Geophysical Research Letters, 29,doi: 10.1029/2001GL014285.

Olsen, T. & Kohlstedt, D. 1984. Analysis of dis-locations in some naturally deformed plagioclasefeldspars. Physics and Chemistry of Minerals, 11,153–160.

Olsen, T. & Kohlstedt, D. 1985. Natural deformationand recrystallisation of some intermediate plagioclasefeldspars. Tectonophysics, 111, 107–131.

Ozacar, A. A. & Zandt, G. 2004. Crustal seismic aniso-tropy in central Tibet: implications for deformationstyle and flow in the crust. Geophysical ResearchLetters, 31, L23601, doi: 10.1029/2004GL021096.

Passchier, C. W. & Trouw, R. A. J. 2005. Microtec-tonics. Springer-Verlag, Berlin.

Prasannakumar, V. & Lloyd, G. E. 2007. Develop-ment of crystal lattice preferred orientation andseismic properties in Bhavani shear zone, Southern


http://dx.doi.org/10.1029.2001GL014285

http://dx.doi.org/10.1029.2004GL021096

India. Journal of Geological Society India, 70,282–296.

Prasannakumar, V. & Lloyd, G. E. 2010. Application ofSEM-EBSD to regional scale shear zone analysis: acase study of the Bhavani shear zone, South India.In: Mamtani, M. (ed.) Structural Geology – FromClassical to Modern Concepts. Special Issue Journalof Geological Society India, 75, 183–201.

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

Rudnick, R. L. & Gao, S. 2003. The Composition of theContinental Crust. In: Rudnick, R. L. (ed.) TheCrust, Treatise on Geochemistry. Elsevier–Pergamon,Oxford, 3, 1–64.

Schmid, S. M. & Casey, M. 1986. Complete fabric analy-sis of some commonly observed quartz c-axis patterns.In: Hobbs, B. E. & Heard, H. C. (eds) Mineral andRock Deformation: Laboratory Studies – The PatersonVolume. American Geophysical Union, Washington,D.C., Monograph, 36, 263–286.

Schulte-Pelkum, V., Monsalve, G., Sheehan, A.,Pandey, M. R., Sapkota, S., Bilham, R. & Wu, F.2005. Imaging the Indian subcontinent beneath theHimalaya. Nature, 435, 1222.

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

Sherrington, H. F., Zandt, G. & Frederiksen, A. 2004.Crustal fabric in the Tibetan Plateau based on wave-form inversions for seismic anisotropy parameters.Journal of Geophysical Research Solid Earth, 109,B02312.

Siegesmund, S., Takeshita, T. & Kern, H. 1989. Aniso-tropy of Vp and Vs in an amphibolite of the deepercrust and its relationship to the mineralogical, micro-structural and textural characteristics of the rock.Tectonophysics, 157, 25–38.

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. 2008. Towards using seismic anisotropy tointerpret ductile deformation in mafic lowercrust. Unpublished PhD thesis, University of Leeds,UK.

Tatham, D. J. & Casey, M. 2007. Inferences from shearzone geometry: an example from the Laxfordianshear zone at Upper Badcall, Lewisian Complex, NWScotland. In: Ries, A. C., Butler, R. W. H. &Graham, R. H. (eds) Deformation of the ContinentalCrust: The Legacy of Mike Coward. GeologicalSociety, London, Special Publications 272, 47–57.

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

Thatcher, W. 2009. How the continents deform: the evi-dence from tectonic geodesy. Annual Review of Earthand Planetary Sciences, 37, 237–262.

Valcke, S. L. A., Casey, M., Lloyd, G. E., Kendall,J.-M. & Fisher, Q. 2006. Lattice preferred orienta-tion and seismic anisotropy in sedimentaryrocks Geophysical Journal International, 166,652–666.

Vaughan, M. T. & Guggenheim, S. 1986. Elasticity ofmuscovite and its relationship to crystal-structure.Journal of Geophysical Research, 91, 4657–4664

Vernik, L. & Liu, X. 1997. Velocity anisotropy in shales:a petrophysical study. Geophysics, 62, 521–532.

Wendt, A. S., Bayuk, I. O., Covey-Crump, S. J., Wirth,R. & Lloyd, G. E. 2003. An experimental and numeri-cal study of the microstructural parameters contribut-ing to the seismic anisotropy of rocks. Journal ofGeophysical Research, 108, 2365. doi: 10.1029/2002JB001915.


http://dx.doi.org/10.1029.2002JB001915

http://dx.doi.org/10.1029.2002JB001915

constraints on the seismic properties of the middle … › perso › mainprice › w_data ›...

Documents