Joint interpretation of upper-mantle anisotropy based on teleseismic P-travel time delays and inversion of shear-wave splitting parameters

.• PHYSICS OFTHE EARTH

AN D PLANETARY INTERIORS

E L S E V I E R Physics of the Earth and Planetary Interiors 95 (1996) 293-309

Joint interpretation of upper-mantle anisotropy based on teleseismic P-travel time delays and inversion of shear-wave

splitting parameters v •

J. Plomerov~ a,*, j. Sfleny a, V. Babu~,ka b a Geophysical Institute, Czech Academy of Sciences, Bo~MH, 141 31 Praha 4, Czech Republic

b UNESCO, Division of Earth Sciences, 1, rue Miollis, 75732 Paris, France

Received 4 December 1994; accepted 31 August 1995

Abstract

We inverted shear-wave splitting parameters and simultaneously analysed delays of teleseismic longitudinal waves to obtain a self-consistent three-dimensional (3-D) image of the anisotropic upper mantle beneath the continents. Efficiency of the simultaneous 3-D analyses of P-residual spheres and shear-wave polarizations is demonstrated on data from two regions, southern Sweden and the western USA. The anisotropic inferences of the subcrustal lithosphere are, to a first approximation, represented by homogeneous hexagonal or orthorhombic media with plunging symmetry axes.

1. Introduction structure beneath stations in a broad fan of directions, from subvertically propagating PKP and SKS

Anisotropy of physical properties is inherent in phases to obliquely incident P or S. Whereas Pn rock-forming minerals, and their systematic preferred characterizes shallow subhorizontal propagations in orientation is reflected in the large-scale anisotropy the uppermost mantle, surface waves reach greater of physical parameters. From a seismological point depths depending on period. However, owing to their of view, seismic velocities depend locally on the long wavelength, they survey average crust and up- propagation direction and wave polarizations depend per-mantle structures with limited fine resolution. On not only on the type of waves, but also on the the other hand, information on the depth distribution orientation of the local symmetry of the elastic prop- of seismic anisotropy can be exploited from vectorial erties. Therefore, various types of seismic waves that tomography of dispersions of Rayleigh and Love differ in particle motion, direction and velocity of waves (Montagner and Nataf, 1988; Montagner, propagation, as well as wavelength, scan the upper- 1994) or from the modelling of converted phases mantle anisotropy differently. Teleseismic body (Farra et al., 1991; Girardin and Farra, 1992). It is waves sample large volumes of the upper-mantle evident that, for regional studies, the teleseismic

body waves can allow us to infer a more detailed picture of three-dimensional (3-D) anisotropic structures in the upper mantle and to map their lateral

* Corresponding author, variations.

0031-9201/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSDI 0031-920 1(95)031 22-7

294 J. PIomerov6 et a l . / Physics of the Earth and Planetary Interiors 95 (1996) 293 309

Although seismic waves differ in their basic char- ~, acteristics, realistic Earth models, including .,v e" anisotropic ones, should be compatible in that they _ o ~ , , ~ fit independent observations of comparable wave- ' ~ lengths. Concentrating on anisotropic structures of the subcrustal lithosphere, we have sought a model that would conform both to propagation of teleseis- ~ ~ % % mic longitudinal waves and to waveforms of shear ~ , ~ z ~ / ~ " ~ "~ waves.

Important information about P-velocity anisotropy t can be obtained from teleseismic P-residual spheres, ~ _ _ _ J _ __J which reveal that part of the relative travel time residuals that depends on azimuth and incidence angle (Babu~ka et al., 1984). In many regions we found the pattern of P-wave residuals plotted on the focal sphere, often of a bipolar character (Babu~ka et al., 1988a) throughout large tectonic units. By bipolar character we mean that negative residuals are concentrated prevailingly on one side of the lower- hemisphere projection whereas the positive ones are on the opposite side. Sudden changes of the P pattern are usually related to prominent tectonic boundaries (Babulka et al., 1984). However, the spatial dependence of P residuals commonly is not attributed to . . . . . . . . . . . . . . . . anisotropic propagation. Only azimuthal variations of the residuals with 7r periodicity have been associated with upper-mantle anisotropy (e.g. Dziewonski 1 / \ / ~ and Anderson, 1983). Instead, shear-wave splitting is generally accepted as a direct proof of anisotropy a, (Kind et al., 1985; Vinnik et al., 1989; Silver and Chart, 1991; Savage and Silver, 1993). ~ / Ii \\7~

There are various independent observations of SKS seismic anisotropy in the upper mantle. The majority Fig. 1. A scheme of teleseismic P- and SKS-wave propagation of them, however, approximate the anisotropy by through anisotropic continental lithosphere approximated by hexagonal symmetry with the symmetry axis coin- hexagonal symmetry with inclined symmetry axis (b') coinciding

with the low-velocity direction. The high-velocity direction deter- ciding with a high-velocity direction, and a priori mined from the bipolar pattern of the P-residual sphere (minuses assumed to be oriented horizontally. In contrast, stand for negative residuals, pluses for positive ones) is perpendic- anisotropy inferred from P-residual spheres implies ular to the fast SKS polarization azimuth, which follows the that 3-D anisotropic structures within the continental direction of the strike of the structures. lithosphere are often inclined (Babu~ka et al., 1984; Babu~ka and Plomerov~i, 1989, 1992), which results subvertically propagated SKS (Fig. 1). The mean in a different high P-velocity direction. This fact is polarizations are oriented along the strike of the sometimes interpreted as a discrepancy between the hexagonal structures within the lithosphere regard- directions of high P and S velocities (e.g. Makeyeva less of the inclination of the high P velocities et al., 1990; Vinnik et al., 1992). This seeming (Babu~ka et al., 1993). In this case, the fast shear- discrepancy occurs because in the shear-wave analy- wave polarization is computed with a use of elastic sis the anisotropic high-velocity direction is associ- constants and thus represents a high-frequency ap- ated directly with a mean polarization azimuth of proximation.

J. Plomerov~ et aL / Physics of the Earth and Planetary Interiors" 95 (1996) 293-309 295

Here, we present alternative 3-D anisotropic mod- Having in mind the results of our P-residual els with plunging symmetry axes based on the inver- studies, we followed a different idea and developed sion of the shear-wave splitting parameters. We use an inversion method of polarization parameters in a the fast SKS and/or S polarization vectors, defined single layer (Sflen2~ and Plomerov~i, 1996), which by spherical angles ~b s (azimuth) and O s (measured allows us to find a generally oriented 3-D anisotropic upward from the positive z-axis oriented downward), structure, with either a hexagonal or orthorhombic and the time differences 8t between two split shear symmetry, and a degree of anisotropy assumed for waves (Sflen3~ and PlomerovL 1996). Moreover, the upper-mantle materials (Christensen, 1984; Main- main advantage of the new models is that they price and Silver, 1993). The corresponding inferred provide synthetic P-residual spheres in agreement anisotropic structures give synthetic P-residual with the observed ones, which is not the case if spheres that are compatible with the observed pattern anisotropic one- or two-layer models with horizontal of the P-residual spheres. symmetry axes are used (Savage and Silver, 1993; To evaluate shear-wave splitting, we modified the Ozalaybey and Savage, 1994). method by Silver and Chan (1991). By using all

three components of seismic digital signals, we performed the particle motion analysis to determine the splitting parameters: the polarization vector of the

2. Shear-wave particle motion analysis and inver- fast shear wave defined by spherical angles ths sion of splitting parameters (azimuth) and ~9 s (measured upward from the z-axis

oriented downward), i.e. its direction in three dimen- As mentioned above, the simplest hexagonal ap- sions, and time delay 8t between the two split shear

proximation of upper-mantle anisotropy assumes waves (Sflen# and Plomerov~i, 1996). The original horizontal orientations of the symmetry axis parallel signal elliptically polarized in the L-Q-T coordinate to the horizontal fast velocity direction in the upper system (L records the longitudinal component in a mantle (hereafter referred to as the symmetry axis ray direction, Q and T represent the transverse com- a'). Then the observed polarization azimuth of the ponents perpendicular to L in the vertical (SV) and fast split shear waves is directly associated with the horizontal (SH) planes, respectively) is corrected for high-velocity direction in the mantle. In such a anisotropy by simultaneous rotations of the coordi- medium the SKS polarization parameters should be nate system in a plane perpendicular to the ray of the independent of back azimuths, which, however, of- shear phase and by shifting the signal in time. The ten contradicts the observations. Accumulation of a resulting shear waveform exhibits the linear polariza- new material has allowed researchers to detect varia- tion. For S waves, this is done by minimizing the tions of the polarization azimuths with the back least eigenvalue of the cross-correlation matrix of the azimuth of arriving waves in many regions (Savage corrected particle motion. In the case of SmKS et al., 1990; Savage and Silver, 1993; McNamara et phases, the energy of the corrected transverse com- al., 1994; Vinnik et al., 1994; Ozalaybey and Sav- ponent is minimized. The problem is solved as the age, 1995). These findings are not compatible with fully 3-D one, as both the azimuth and inclination of models consisting of a single hexagonal anisotropic the fast shear-wave polarization vector are consid- layer with a horizontal symmetry axis, and have ered. given rise to an advanced development of the shear- In homogeneous anisotropic models with arbitrary wave polarization analysis by Savage and Silver plunging symmetry axes, the splitting parameters (1993, 1994) and Ozalaybey and Savage (1994). depend on back azimuth and incidence angle of the These workers considered two-layer models with arriving wave. To constrain the 3-D orientation of different magnitudes of anisotropy and orientations the anisotropy, we have to invert the splitting param- of the horizontal symmetry axis d to fit variations of eters. In all available directions defined by the back the polarization azimuth and time difference 8t with azimuth and angles of incidence of recorded shear the back azimuth, which is compressed to the inter- waves we minimize the differences between ob- val 0-7r /2 required by the two-layer model, served splitting parameters, i.e. polarization vectors

296 J. PIomerovA et a / . / Physic=s ~/'the Earth and Planetary Interiors 95 (1996) 293-309

and time delays, and synthetic parameters computed 3. 3-D anisotropie structures of the continental for peridotite aggregates (Babu~ka et al., 1993). In subcrustal lithosphere the inversions we use both the orthorhombic and hexagonal symmetries. In the hexagonal case, the 3.1 . F e n n o s c a n d i a

symmetry axis coincides with the low-velocity direction (b') and Vp anisotropy of 5%. It was created by To perform the joint interpretation of both P rotating elastic parameters along the b' axis of the residuals and S polarizations and to retrieve 3-D basic orthorhombic aggregate (Nicolas and Chris- orientation of anisotropic structures of the continen- tensen, 1987) with Vp anisotropy of 9%. This aggre- tal upper mantle from independent observations, we gate is considered as a representative of a 'typical' need both one-component short-period records of mantle peridotite with a preferred orientation of teleseismic P arrivals and three-component medium- olivine crystals (Fig. 2). or broad-band records of shear waveforms. We have

O R T H O R H O M B I C S Y M M E T R Y

anisotropy of olivine crystal k p- 25% (Chdstensen, 1984) peridotite aggregate k p =" 9%

B preferred odfdalion 51%

I L ~ olivine with otthorhombic B a ~ prefen'ed orientation 17%

mixture of randomly

[ I J.. = =% d

H E X A G O N A L S Y M M E T R Y

peridotite aggregates horizontal symmet~j axis a' symmetry axis b' low-velocity plane (b',c~ high-velocity plane (a',c')

d vertical axis inclined axis ,

b' b, i , ,," / . . . . . . . . . . . . . . . . . . b / N

a'

a' E

kp- 8% kp - 5% down I

Fig. 2. Approximation of the upper-mantle anisotropy by pcridotitc aggregates with orthorhombic or hexagonal symmetries caused by pncfcrrcd orientation of olivine crystals.

J. Plomerov6 et aL / Physics of the Earth and Planetary Interiors 95 (1996) 293-309 297

carried out a complete 3-D particle motion analysis east of the PZ. The mean polarization azimuth of of shear waveforms at the stations included in the steeply incident fast shear waves trends WSW-ENE 'Varmland '91' experiment (Plomerovh et al., and is almost perpendicular to the dip-azimuth of the 1993b,1996). The goal of that French-Swedish- high P velocities. According to our previous study of Czech experiment was to detect lithospheric the P-residual spheres in relation to the polarization anisotropy and to constrain its lateral variations as azimuth of the fast shear waves, such seeming incon- suggested by a preliminary P-residual study in sistency in the orientations of the fast P and S Fennoscandia(Babu~ka et al., 1988b). In the P-resid- velocities indicates that it is hexagonal symmetry ual studies we separated a static term of the relative with the axis b' that also matches the shear-velocity residuals at a station and azimuth-incidence angle data of appropriate wavelength. The orthorhombic dependent terms (e.g. Babu~ka et al., 1988a). From approximation of the anisotropy would give better spatial variations of the latter term, presented in results in cases of parallel orientation of both az- 'anisotropic' P-residual spheres, we can deduce cor- imuths (Babu~ka et al., 1993). responding velocity variations which can be related The inversion of all shear-wave splitting parame- to lithospheric anisotropy. Stations with almost no ters gives a 3-D solution (Fig. 3) which is almost difference in the absolute residuals can exhibit a identical with that inferred from anisotropic terms of different directional dependence of the relative resid- the P residuals. The dip azimuth of the high-velocity uals which attain values one order higher than mean plane of the resulting hexagonal model is 320 ° and errors of the relative residuals (Babu~ka et al., 1990). the dip itself is 49 °. This solution is stable with

One of the most distinctive features of the relative respect to the orientation of the hexagonal symmetry P-residual spheres in Fennoscandia (Babu~ka et al., as well as to the data noise (Sflen2~ and Plomerovh, 1988b) is an almost reversed P pattern observed at 1996). The resulting thickness of 153km found for stations situated around the Protogine Zone (PZ) in the hexagonal symmetry is close to the lithospheric southern Sweden. The Protogine Zone is a deep thickness derived from static terms of the P residuals narrow suture separating the Trans-Scandinavian Ig- (Babu~ka et al., 1988b) or to that estimated from neous Belt and the southwest Scandinavian Domain surface waves (Calcagnile, 1982). The synthetic P- (Gaal and Gorbatchev, 1987). The reversed bipolar residual pattern (Fig. 3) computed for the anisotropic pattern is characterized by high-velocity eastern di- solution of the inversion of splitting parameters is rections marked by the negative residuals in a group close to the observed one which, of course, has a less of stations west of the PZ which change to positive complete back azimuth-incidence angle coverage. A residuals at stations deployed east of the suture. This solution of the inversion of the shear-wave splitting pattern was recorded by permanent observatories in data with the orthorhombic symmetry is unstable in the region as well as by the temporary digital sta- this region (Sflen2~ and Plomerovfi, 1996) and the tions operated on both sides of the PZ. Both the corresponding pattern of the synthetic P-residual P-residual study (Plomerovh et al., 1996) and particle spheres contradicts the observed one. motion analysis of shear waves confirm the presence of anisotropy in the subcrustal lithosphere and show 3.2. Western USA its lateral variations relative to the PZ across the array (Plomerovh et al., 1993b). The anisotropy seems There are distinct variations in the seismic struc- to be more developed in the lithosphere east of the ture of the continental upper mantle of North Amer- PZ. The P-residual spheres mark the northwesterly ica that result from the complex processes of conti- oriented azimuth of the high-velocity directions nental growth (Mooney and Braile, 1989). Four types plunging at angles between 30 and 60 ° (Babugka et of P-residual patterns affected by the upper-mantle al., 1993). Fig. 3 summarizes the results of the anisotropic structures were distinguished at stations P-residual study (Plomerov~i et al., 1996)and shows in the western USA (Babugka et al., 1993): (a) schematically also a smoothed P-residual sphere (a stations in the Cascade and Klamath Mts. region, two-parameter linear filter has been applied) at Sta- where the high P velocities plunge to the SE; (b) tion C as an example of the P pattern in the region stations around Berkeley (BKS), where the high P

298 J. Ph,merov6 et a l . / Physics q/'the Earth and Planetary Interiors 95 (1996) 293-309

(m)

/.2 13 ~4 15

,, ' " ,,,, A z i m u t h

Swedez

AA ~ ~ t P

V~nern

12 13 !.4 18

( i l ) CST - hexagonal model C - observed P ~:esiduals

- - 7 - ~'7- 6 -7 -7

/ -6 -5 4-5 - 6 - ~ - - 3 --'7 - 6 - 4 -2 -"~ ; 7 - 6 -4.~.-2 '-7-6-~:'D ",., 2 '4'I_ ~-6 -5 "-. -" 4 6

-4 2 6 " --2_~_ ~ -8

(a) (b) (c) (d)

Fig. 3. (1) A sketch map of a region in southern Sweden where the inversion of shear-wave splitting parameters was performed for three stations east of the Protogine Zone (Q). Bold arrows mark the azimuths in which the high P-velocity directions plunge as determined from the P-residual spheres (Plomerovft et al., 1996). The large arrow, axonometric display, marks the dip of the high-velocity plane of the hexagonal model obtained by the inversion of the shear-wave splitting data. (If) The orientation of the model is presented in projection of the lower hemisphere (a), where the triangle marks symmetry axis b' and the curve the high-velocity plane (a' ,c '); (b) the synthetic P residuals (in tenths of a second) for the hexagonal model in (a); (c) smoothed observed P-residual sphere with the anisotropic part of the residuals at Station C; (d) the same as (c), but for groups of events--minuses stand for negative residuals (early arrivals, relatively high velocity directions), crosses for positive residuals (late arrivals, relatively low velocity directions, the size proportional to the magnitude of the residual term).

J. Plomerov6 et al . / Physics of the Earth and Planetary Interiors 95 (1996) 293-309 299

velocities plunge to the SW; (c) stations of a transi- symmetry axis plunging to the W N W for stations in tional type (Coast Ranges, and also Stations MHC Groups (c) and (d) (Babu~ka et al., 1993). and ARN southeast of BKS, northern Great Central To explain the shear-wave splitting variations with Valley); (d) stations in a region of the Mojave back azimuth at Stations LAC (Mojave Desert; ac- Desert, with the highest Vp plunging towards N W cording to the P-residual pattern in Group (d)) and azimuths (Fig. 4). Synthetic tests with average SKS NSAN (three s t a t ions - -BKS, MHC and S A O - - a t polarizations and the P patterns found the hexagonal the Northern San Andreas; in Groups (b) and (c)), symmetry with the high-velocity plane dipping to the Savage and Silver (1993) proposed hexagonal mod- SE and SW as an optimum for stations in Groups (a) els consisting of two anisotropic layers with horizon- and (b), and the orthorhombic model with the a' tal symmetry axes ( d ) oriented along the San An-

-1 -2 -I

' J' 4 - 4 ' 0 , - t 0 1 2 o z - s a ~ ~ -4--s ARN 2 i L 0 4

m-a,-~ -~ ~ "£Jz t -s • -,e--a BRK A ~ - t - t

, C ~ASIN -I qGE / . . o - z - s

s ' P ' ; -2 I} ., 4 z so. ._~ _ ~ ~ - t - t NHC t .L .4

• -e-t BKS t - s -~ t o t

. o o

-t • -s ~ c - I -8 " " i l l Im | • z

- i t 6cc i t -m SAO mt ~ . 1 4 0 J

, t t t l • ~ , -4 -a

"~7-4 0 4 • ,-,es TPC

3 3 I ~ ~ -4 m ~'

- t 2 4 - t 2 2 - t 2 0 - t t 8 - i ~ ' 5

Fig. 4. Schematic representation of observations of seismic anisotropy in western North America. Open and filled triangles show stations with prevailingly early arrivals of P waves from SE and NW, respectively, half-black triangles represent a transition between the two types, and the triangles with dots are stations with early arrivals from SW. The patterns are also represented by filled and open arrows for directions of dipping relatively high and low velocities of P waves, respectively. Thin double arrows show mean azimuths of polarization of the fast split SKS waves, dashed bars azimuths of fast Pn velocities (Savage and Silver, 1993). Smoothed P-residual spheres (Babulka et al, 1993) are shown (in tenths of a second) for stations located in the northern San Andreas fault (left; BRS, BKS, PCC, GCC, SAO), northern Great Central Valley (MHC, ARN) and Mojave Desert (LAC, TPC).

3 0 0 J. PIomerov6 et al. / Physics of the Earth anti Planetary Interiors 95 (I 996) 293-309

LAC NSAN tive to the centre of the diagrams (Fig. 5) are far

~ 2 ~ from the observed patterns (Fig. 4). Al though only ,~'23 44 4 33 ~'~_2, ~ /e__4 ~ 2 44 54 ~ 3", horizontal components of the shear waveforms were

~4 3 . 4 ~ ~ -~ ,~-2 4 _.~ ~ 2 ~' analysed, the published dependences of the fast S / -4 3 ~ ' - - ' ~ 2 -~% , ~ 2 ." ",

' ~ 4 . q ' e -% polarizat ion azimuth and splitting t ime 6 t can serve ~ ~ 2 " - q '~ -~, , z ~'~ -'~,-2~ ~ 4~ 4 ~ -~, , ~ " ~ z - J as an input of the inversion of the splitt ing parame-

/ 4. 4 - ~ ~ 3 4 4 3 ~ 4 4 -1 , -~- _ ~ ~. ~ ~ ~ ~ ters. * ' The invers ion of splitting data of the LAC station

Fig. 5. Synthetic P-residual spheres (residuals in tenths of a with the hexagonal symmetry ( b ' ) results in a solu- second) computed for two-layer anisotropic hexagonal models t ion (Fig. 6) with the high-veloci ty plane dipping (Savage and Silver, 1993) with horizontal 'high-velocity' symme- steeply (80 °) to the NNE, which gives the synthetic try axes (d) oriented along the fault in the upper layer and east-west in the lower layer at Station LAC (Mojave Desert) and P-residual pattern inconsistent with the observed one three stations at the northern San Andreas Fault (BKS, MHC, (Fig. 4). On the other hand, when invert ing the SAO; together NSAN). shear-wave splitting data with the or thorhombic

symmetry , we obtain a solut ion (Fig. 7) with the fast

dreas fault in the upper layer and e a s t -wes t in the d -ax i s p lunging at 23 ° from the horizontal in az- lower layer, The synthetic P-residual patterns com- imuth 283°; the thickness of the anisotropic layer is

puted for these models which are symmetr ical rela- 167km. The pattern of the synthetic P residuals for

(a) fast S polarization

azimuth 6t (s)

(b)

• .,6"4-3 . -2 _,-'2 _ / / - 6 " - 4 - ' a - ' 3 - ~ - 3 x

t- - 5 - . "" r ' l "~ " ~ - ~ ' 6 ; - i - 5 ~ - q ' r - - ~ ~ - -

~-3 - 3 " . . , - 4 - • , - 2 - 2 " - - 3 _-4 -~

\ - I - 3 / ~ . 2 _ . -~

Fig. 6. (a) Fast shear-wave polarization azimuth and the splitting time delay data ~t (open circles with error bars; Savage and Silver, 1993) and synthetic values (nets) according to the results of the inversion of the splitting data with the hexagonal model (b, left), together with the synthetic P-residual spheres (b, right) at Station LAC, Mojave Desert.

J. Plomerov6 et aL / Physics of the Earth and Planetary Interiors" 95 (1996) 293-309 301

(a) fast S polarization T azimuth 8t (s)

. . - ) .%<. %<" o

(b) 3 _ ~ 7 ~ 'f 5 7-

" , ' 6 " 2 1 3 6 6 7 . . /-7" 5 6 -/11.-2 3 4 5 .

' -H%2, , ' " . , 5 ? S~ - : -9"~-1 : p ',6 ; 6 ' -

',-7- , , 7 6 I - 2 4 " 7 7 /

35 2 6 8 ', 6 8 , = 8 s E; 8 .

I

Fig. 7. The same as Fig. 6, but with the orthorhombic model.

such a model (Fig. 7) agrees well with the observed pattem (Fig. 4). A stability test of the solutions also o ~ ~ favours the orthorhombic solution, which displays less scatter (Fig. 8) within the 10% and 30% levels 10% 30% above the minimum value of the misfit function (see also Sflen~ and Plomerov~i (1996)). The solution with hexagonal symmetry and a very steeply dipping LAC as v i typ suffer from t a i i y

Fig. 9 shows an integral solution for three stations, grouped as NSAN, for the hexagonal symme- 10% 30% try. The high-velocity plane dips steeply SW (dip- azimuth 210 °, dip 80 °, thickness of the layer 181 km). Synthetic P residuals exhibit the bipolar pattern, but

Fig. 8. Grid search minimization functions. The full triangles and the agreement with the observed: pattern is much squares, and solid curves correspond to the best solution Rmi n worse than in the case of the LAC station (ortho- (triangles pointing downward mark the orientation of the symme-

r h o m b i c symmetry). F o r t he i n v e r s i o n o f t he splitting try axis b' and the curve the plane (d,c') perpendicular to the

data with the orthorhombic symmetry (Fig. 10) we symmetry axis in the case of the hexagonal model, and the squares

obtain the following solution: the a',axis dips 59 ° and triangles pointing upward mark the axes d and c' in the orthorhombic model). Open symbols and dashed curves represent

from the horizontal in azimuth of 301°; the thickness all acceptable solutions less than 1.1 × Rmi n and 1.3X Rmi n (for

of the layer is 198 km. However, the bipolar pattern more details, see also .~flen~ and Plomerov~ (1996)).

302 J. Plomerov6 et al. / Physics of the Earth and Planetary Interiors 95 (1996) 293-309

of the P residuals differs from the observed one even Savage (1995) listed variations of the azimuth of the more than in the case of the hexagonal model. The fast S polarizations and the splitting time with the inversions of the splitting data result in orientations back azimuth for several stations, including BKS, of both hexagonal and orthorhombic symmetries that MHC and SAO, previously grouped together as are close to those found for synthetic tests with the NSAN. Their table contains 26 individual pairs of peridotite aggregate for the BRK (BKS) stations fast shear-wave azimuths and splitting times for Sta- (Babu~ka et al., 1993), but the orientations resulting tion BKS and 12 for Station MHC that we could from the inversion of splitting data exhibit high invert separately to test whether the approximation instability (Fig. 11). According to the P-residual of the upper-mantle anisotropy in terms of the sym- analysis mentioned above, the pattern at MHC (Group metry differs as predicted from the analysis of P-re- (c), 'orthogonal type') differs from that at BKS sidual spheres and both mean and synthetic fast S (Group (a), 'hexagonal type'), although the stations polarization azimuths (Babu~ka et al., 1993). Fig. 12 are close to each other. Thus, from this point of shows the resulting orientation of the fast plane of view, NSAN is an inconsistent group of stations, and the hexagonal model (b') of the upper-mantle this may be reflected in the low stability of the anisotropy beneath BKS (dip-azimuth 217 °, inclina- solutions, tion 72 °, thickness of the layer 185 km) which is in

In a recent study of the two-layer anisotropic agreement with the result for NSAN. However, as structure beneath the westem USA, Ozalaybey and we predicted for the MHC station, orthorhombic

T (a) fast S i polarization

I azimuth 6t (s)

,, - 7 - 4 . 2: i- 7_6 -6 ,~ -,-6 ,~-6~ p 4-~-3'- ~ . - ~ . . _~__~ 53-4 4-~_!

" ,-2 - 4 I 7 - / /

Fig. 9. The same as Fig. 6, but with the NSAN stations.


(a) fast S I polarization azimuth 6t (s)

< <-LoO o°

(b)

/ - I 5 11~ - 1 4 J17~ : 8 1 8 . I -II . ~ II I - 14_ -6 "" 9 3 f -~I "L I I

-t-10 -3 P , 9 i:~- i.6-5 , iliJ

' ' 0 I I ,,~-I 3 7 10 1/1 ~ 2 6 9 l g l l 11/ "s. e il i4

i01.1J t

Fig. 10. The same as Fig. 7, but with the NSAN stations.

symmetry (Fig, 13) with the axis d dipping at 36 ° The least-squares misfit function of the orthorhombic from the horizontal in the azimuth of 304 ° (thickness model is 25% lower than that for the hexagonal of the layer is 198km) matches the data better than model, and its orientation is very close to that found the corresponding hexagonal model (dip-azimuth of in synthetic tests: symmetry axis d dipping at 43 ° the fast plane 35 °, inclination 74 °, thickness of the from the horizontal in the azimuth of 294 ° (Babugka layer 183 km; orientation is reversed relative to BKS). et al., 1993). The synthetic P-residual sphere com-

puted for the orthorhombic approximation of the ~ o m i ~ , ~ ~ ~ lithospheric anisotropy beneath MHC agrees with the

observed residual sphere. The best solutions accord- 10% 30% ing to their stability and consistency with the ob-

served P-residual spheres for the individual regions are presented in Table 1.

NSAN

a record of the continental growth by accretion of I0% ~ ~ 30% oceaniclithosphere

Recent intensive research of the upper mantle Fig, 11. The same as Fig. 8, but with NSAN stations, confirms the presence of seismic anisotropy as a

304 J. Plomerov6 et al. / Physics of the Earth and Planetary Interiors 95 (1996) 293-309

(a) fast S polarization ~t(s) .~ azimuth !

( b )

- " 7 " _ - 3 . 4 4 )~

~-6-~-4-, ~ ' ' - , 1 4 t -G; " ~ ~ 1 t 1 1 --5__-S' f f , - . -

4 -5 . ." -3 -

3 - - -S -~ ~ - 3 -4 ~-5 . - 5 / x x - 3 . 4 " ~ . 5 - ~ _ F

I

Fig. 12. (a) The fast shear-wave polarizations and the splitting time delay data Bt (open circles with error bars; Ozalaybey and Savage, 1995) and predicted values (nets) according to the results of the inversion of the splitting data with the hexagonal model (b, left), together with the synthetic P-residual spheres at Station BKS, northern San Andreas fault (right; compare with the sphere in Fig. 4). In the case of the polarization azimuths (a, left) the deviations are presented. The theoretical value of the difference is zero and is plotted as the planar net at the zero level.

ubiqui tous and impor tan t proper ty o f the cont inental al., 1992; Plomerovf i e t al., 1993b; Montagner , 1994; l i thosphere (e.g. Anderson , 1989; Babugka and G a o et al., 1994; M c N a m a r a e t ai., 1994; Al s ina and Plomerovfi , 1989; Savage and Si lver , 1993; Main- Snieder , 1995). The complex i ty o f the he te rogeneous pr ice et al., 1993). Studies o f lateral var ia t ions o f the and la tera l ly va ry ing anisot ropic structure of the upper -mant le an iso l ropy, its or ienta t ion and its distr i- cont inental upper mant le requires a comp lex seismic but ion with depth are e s sen t i a l for unders tanding the analys is based on independen t observat ions . Mode l s geodynamics of the l i thosphere and the Ear th ' s man- o f the upper mant le based on ly on the b road-band tie (Savage et al., 1990; Molnar , 1992; M a k e y e v a et shear -wave analys is m a y be mis lead ing as to the

Table 1 Orientations of the high-velocity directions in the 3-D anisoa-opic inferences of the upper mantle from the inversion of the shear-wave splitting parameters Station(s), region Type of P-velocity d-Axis (d,c')-Planc Required thickness

symmetry anisotropy Azimuth Dip Azimuth Dip (kin)

CST, S. Sweden Hexagonal (b') 5% - - 320 49 153 LAC. Mojave Desert Orthorhombic 9% 283 23 - - 167 BKS, San Francisco Bay Hexagonal (b') 5% - - 217 72 185 MHC, Great Central Valley Or~orhombic 9% 304 36 - - 198


(al fast S polarization St(s) ~- azimuth

0 ~

/ - 1 5 v - 6 .~ -17 !~_3 1 .| ~ .),, J19 -1

I - 14 - 1 6 -9' , " " , 4 71 ' ""~( P ' 6 6 a ' - --i-11 - 5 ~-7"7-2 "- ' 7 R j

3 " " 7 ~ 9 ~ 1 6 9 leg,, ? 1

..1;2~ ~.3J tzt~

Fig. 13. The same as Fig. 12, but with the orthorhombic model and Station MHC, northern Great Central Valley.

lithospheric anisotropy if only the azimuthal, i.e. good azimuthal coverage, direct shear waves are also horizontal, component of anisotropy is looked for. used in this study under the assumption mentioned

Shear waves traversing the Earth's core possess a above. convenient linearity in polarization at the bottom of From the analysis of P-residual spheres with an the mantle on the station side, but yield an integral anisotropic part of the relative residuals of teleseis- effect of anisotropy along the whole mantle path. On mic waves recorded by short-period sensors we de- the other hand, owing to their near-vertical inci- duced a dipping orientation of anisotropic structures dences they are sensitive to lateral changes beneath in the subcrustal lithosphere. On the basis of the P- the stations. Savage and Silver (1993) pioneered and S-wave velocities, and anisotropies of a typical consideration of splitting analysis of direct shear peridotite composed of olivine, orthopyroxene and waves also, although their polarizations may be con- clinopyroxene, both teleseismic P and S seismologi- taminated by the source mechanism and effects of cal observations in central Europe, western North anisotropic structures on the source side. Either cor- America, the Balkans and the southern part of central rections for source polarizations taken from centroid Sweden can be interpreted by a rotation of the moment tensor solutions (Ozalaybey and Savage, peridotite aggregate with hexagonal or orthorhombie 1995) or careful examinations of the 'inner consis- symmetry. Orientations of anisotropic structures with tency' with the SKS polarizations (Savage and Sil- horizontal or vertical symmetry axes do not fit the ver, 1993) minimize the source-side effects. As the spatial variations of the azimuth-incidence angle inversion of shear-wave splitting parameters to con- dependent terms of P residuals. Both the frequently strain the 3-D upper-mantle anisotropy with gener- observed bipolar pattern of the P-residual spheres ally oriented symmetry axes requires a broader fan and the observed variations of the shear-wave split- of incidences than for subvertical propagations, and a ting require an inclination of the symmetry axes. The

306 J. Plomerov6 eta I. / Physics qf the Earth and Planetary Interiors 95 (1996) 293-309

research conducted in several continental regions variable orientations; (2) the more ductile lower crust, (e.g. Babugka et al., 1984, 1988b,1993; Plomerov~ et characterized by sub-horizontal structural 'layering' al., 1993a,b,1994) results in an attempt to confront in many places; (3) the high-velocity subcrustal the inferences with other data and methods and to lithosphere, with predominantly frozen-in olivine ori- find a self-consistent 3-D image of the anisotropy of entations as the main cause of large-scale anisotropic the upper mantle that would fit independent observa- structures and frequently with plunging axes of sym- tions. This can be achieved by the inversion of metry (Babu~ka et al., 1984, 1993); (4) convecting splitting parameters (see also Sflen27 and PlomerovL low-velocity asthenosphere and the upper mantle, 1996). with high velocities oriented in the direction of the

The inversion itself, dealing with a real spatial present-day flow. coverage of the input data, does not allow us to McNamara and Owens (1993) showed that only decide reliably about the type of symmetry with 0.1-0.2 s of the magnitude of the observed splitting, which the anisotropy in the subcrustal lithosphere is which is usually one order larger, can be accounted approximated. However, in combination with the for by the crustal anisotropic structures. Owing to analysis of the P-residual spheres, the type of sym- differences in dimensions, all seismic observations metry can be distinguished. Generally, the anisotropy related to the structure depend on frequencies. in the lower lithosphere can be approximated by Anisotropic models of the subcrustal lithosphere with dipping structures, where oriented a' and c' axes, i.e. plunging symmetry axes are proposed here as a the fast-velocity directions and intermediate-velocity result of joint interpretation of body-wave character- directions, are within the dipping planes and the b' istics related to the short-period range (P or PKP at axis (the low-velocity direction) is perpendicular to approximately 1-2 s) and to the intermediate period the plane. As the preferred orientation of olivine range (S or SKS at approximately 4-5 s). There are crystals is likely to be the source of anisotropy of the several reasons why the major part of the large-scale upper-mantle materials, which are predominantly of anisotropy is bound to originate in the subcrustal peridotite type (Christensen, 1984), the past or pre- lithosphere. Most of the tomographic studies agree sent existence of large stress fields is bound to that down to a 300-400km depth, the structure is produce the anisotropic effect. We interpret the dip- closely related to plate tectonics and the distribution ping anisotropic structures as remnants of various of continents (Gudmundsson et al., 1990; Grand, palaeosubductions of the oceanic lithosphere which 1994; Montagner, 1994). Also, observed lateral vari- retained original olivine orientations (Babu~ka and ations of anisotropy are related to prominent tectonic Plomerovfi, 1989). Many geological, volcanological units of the continents (Babu~ka et al., 1984, 1993; and geochemical data suggest that various parts of Savage and Silver, 1993; Plomerovfi et al., 1993b). continents originated in oceanic settings through plate Although surface waves have a low resolving power, tectonic processes such as subduction and accretion they show that the amplitude of anisotropy sharply at convergent boundaries or island arcs. decreases with increasing depth. At a depth of 370 km

Characteristics of the P residuals and the shear- an average radial anisotropy amounts to about 25% wave splitting (e.g. lateral variations related to main of its average value at a depth of 100kin (Montagner, tectonic units or the degree of anisotropy) allow us 1994). to assume that a substantial part of the observed Mantle materials are strongly anisotropic, and anisotropic effects can be accounted for by structures realistic petrological models consider coefficients of within the subcrustal lithosphere. On the other hand, P-velocity anisotropy to be higher than 5% (Babu~ka there definitely are observable contributions to the and Plomerovfi, 1992). Estimates of the magnitude of anisotropy which originate in other parts of the lithospheric anisotropy based on P residuals and SKS crust-upper-mantle system of continental provinces, particle motion analyses result in values of about 8% In terms of the different origins, scales and orienta- and 4%, respectively (Babu~ka and Plomerovfi, tions of anisotropic structures, four dominant depth 1995). These values are in agreement with observa- regions can be distinguished: (1) the upper brittle tions of the upper-mantle rocks (Nicolas and Chris- crust, with small-scale, often consistent fabrics of tensen, 1987; Mainprice and Silver, 1993). In the

J. Plomerovd~ et aL / Physics of the Earth and Planetary Interiors" 95 (1996) 293-309 307

inversion of shear-wave splitting parameters, the in- crystal alignment origin of the upper-mantle ferences of the required thicknesses are about 20- anisotropy. However, if we leave the assumption that 30% higher than are the assumed thicknesses of the the high velocities are a priori oriented in the hori- subcrustal lithosphere, deduced from the static term zontal, then neither the fast polarization azimuth nor of the residuals (Babulka et al., 1988a). Under the the 7r term of the P residuals itself describes directly assumption that this estimate of the lithosphere the high-velocity directions. Then the absence of the thickness is not far from reality according to a correlation is understandable. Anisotropic structures comparison with other fields--geomagnetic, heat- with dipping symmetry axes produce 27r periodicity flow data or gravity modelling (Praus et al., 1990; of the splitting parameters with the back azimuth, Babu~ka and Plomerov~, 1993; Lillie et al., which is the same periodicity as in the case of lateral 1994)--either there is a higher anisotropy of the real heterogeneities. subcrustal lithosphere than that in the hexagonal models we considered ( k p = 5%) or a corresponding part of the anisotropic signal comes from the as- 5. Conclusions thenosphere.

In the plastic part of the upper mantle, the stress In this study we demonstrated, using as examples field tends to align the fast axis of the olivine grains two regions for which simultaneous fully 3-D analy- in the direction of the mantle flow. However, this ses of both the P residuals and shear-wave splitting process, which is related to the large mantle convec- parameters were performed, that there is a solution tion cells, is of a much larger scale in comparison which allows us to find a coherent orientation of the with the size of the structures within the continental fast velocity directions in the anisotropic upper man- lithosphere. Thus the asthenospheric contributions to tle from independent observations of teleseismic body the anisotropic effects should be, in general, observ- waves. To a first approximation, this solution is able with longer periods (Vinnik et al., 1989; Alsina represented by a homogeneous hexagonal or or- and Snieder, 1995) and should be reflected in a thorhombic medium with symmetry axes plunging significant correlation between the high-velocity di- from the horizontal. rections and the absolute plate motions. The depth resolution of the individual methods

It has been pointed out that the structure of the based on teleseismic shear-wave splitting is very upper mantle beneath the continents differs from that low. There is a well-known trade-off between the beneath the oceans. Montagner (1994) compared the thickness of the anisotropic medium and the magni- directions of maximum shear velocities (SV) at a tude of its anisotropy. Inferences from the inversion depth of 200km with absolute plate directions by of shear-wave splitting parameters are compatible Minster and Jordan (1978). Whereas a good correla- with those obtained from P-residual spheres and are tion was observed below the oceanic plates, Montag- realistic from the petrological view. Both P-residual net did not find such a good correlation for continen- sphere analyses and the inversion of the shear split- tal plates. This discordance is accounted for by the ting attribute most of the anisotropic effects observed complex history of continents, resulting from the in an adequate period range to the subcrustal litho- collage of different-aged blocks, which favours mod- sphere. In combination with P tomographic studies, els of the complex continental subcrustal lithosphere modelling of converted phases, or surface-wave stud- with plunging symmetry axes (Babugka and Plome- ies, and taking into consideration the knowledge of rovfi, 1989). velocity anisotropies of a realistic petrological mate-

Owing to the absence of correlation in a his- rial under corresponding P - T conditions, and a togram of the angular difference between the fast possible genesis of the large-scale anisotropy, the SKS polarization azimuth, directly assigned to the 3-D inferences about the anisotropy of the continen- fast direction of anisotropy, and the fast directions of tal subcrustal lithosphere with plunging symmetry the teleseismic P waves derived from a residual term axes represent a convenient solution. These orienta- with ,r periodicity (Dziewonski and Anderson, tions match both the P-residual spheres and shear- 1983), Vinnik et al. (1992) argued against the olivine wave splitting variations.

308 J. Plomerov6 eta l . /Physics o f tire Earth and Planetary Interiors 95 (1996) 293-309

Specifically, the anisotropic structures of the sub- Anderson, D.L., 1989. Theory of the Earth. Blackwell, Oxford,

crustal lithosphere east of the Protogine Zone, south- 366 pp. e m Sweden, can be approximated by the hexagonal Babu~ka, V. and Plomerov~i, J., 1989. Seismic anisotropy of the

subcrustal lithosphere in Europe: another clue to recognition symmetry (b') with the plane containing the high- of accreted terranes? Geophys. Monogr. Am. Geophys. Union, velocity directions dipping to the NW. In the western 50: 209-217. USA, the lithospheric structures beneath the Mojave Babulka, V. and PlomerovL J., 1992. The lithosphere in central

Desert exhibit orthorhombic symmetry with the Europe--seismological and petrological aspects. Tectono-

high-velocity directions (a ') plunging to the WNW, physics, 207: 141-163. Babulka, V. and Plomerovfi, J., 1993. Lithospheric thickness and

whereas beneath northern San Andreas Fault both velocity a n i s o t r o p y - seismological and geothermal aspects. types of anisotropic symmetry were found. Beneath Tectonophysics, 225: 79-89. the BKS station, hexagonal symmetry with the high- Babu~ka, V. and Plomerov~, J., 1995. Regional fabric of the deep velocity plane dipping to the WSW satisfies the lithosphere in central Europe from seismic anisotropy. Stud.

body-wave data, whereas at the nearby station MHC Geophys. Geod., 39: 219-226. Babu~ka, V., Plomerovfi, J. and Sflen2~, J., 1984. Large-scale

the symmetry is orthorhombic with the d axis plung- oriented structures in the subcrustal lithosphere of central ing to the WNW. Europe. Ann. Geophys., 2: 649-662.

.Three-dimensional self-consistent anisotropic im- Babu~ka, V., Plomerov~i, J. and Pajdu~ik, P., 1988a. ages of the upper mantle with properly oriented Lithosphere-asthenosphere in central Europe: models derived

from P residuals. In: G. Nolet and B. Dost (Editors), Proc. 4th symmetry axes in the subcrustal lithosphere beneath Workshop on the European Geotraverse (EGT) Project: the the continents should be preferred in comparison Upper Mantle. European Science Foundation, Strasbourg, pp. with those comprising alternating high- and low- 37-48. velocity layers or widespread velocity hetero- Babu~ka, V., Plomerov~, J. and Pajdu~fik, P., 1988b. Seismologi- geneities in the upper mantle, interpreted in long- cally determined deep lithosphere structure in Fennoscandia.

Geol. Foren. Stockholm Forhandl., 110: 380-382. range refraction profiling or body-wave tomography, Babu~ka, V., Plomerov~, J. and Granet, M., 1990. The deep that are difficult to explain by reasonable petrologi- lithosphere in the Alps: a model inferred from P residuals.

cal models. Tectonophysics, 76: 137-165. Babu~ka, V., Plomerov~, J. and Silent, J., 1993. Models of

seismic anisotropy in deep continental lithosphere. Phys. Earth Planet. Inter., 78: 167-191.

Acknowledgements Calcagnile, G., 1982. The lithosphere-asthenosphere system in Fennoscandia. Tectonophysics, 90: 19-35.

We wish to gratefully acknowledge P. Molnar, G. Christensen, N.I., 1984. The magnitude, symmetry and origin of Poupinet and an anonymous reviewer for their c o m - upper mantle anisotropy based on fabric analyses of ultramafic merits and suggestions to improve the manuscript, tectonites. Geophys. J. R. Astron. Soc., 76: 89-11 I. Special thanks are due to K. Kli'rna and K. Spa~ek Dziewonski, A.M. and Anderson, D.L., 1983. Travel times and

• station corrections for P waves at teleseismic distances. J. for~ their technical~ assistance with some figures. This Geophys. Res., 88: 3295-3314.

work was Supported by the Czech Academy of Sci- Farra, V., Vinnik, L.P., Romanowicz, B., Kosarev, G.L. and Kind, e n c e s grant No. 312115. The international experi- R., 1991. Inversion of teleseismic S particle motion for az- ment, data from which were partly used in this imuthal anisotropy in the upper mantle: a feasibility study.

paper, was partially supported by grants of the Geophys. J. Int., 106: 421-431. Gaal, G. and Gorbatchev, R., 1987. An outline of the Precambrian

Swedish Natural Science Research Council, French evolution of the Baltic Shield. Precambrian Res., 35: 15-52. Institut National des Sciences de l'Universe and Gao, S., Davis, P.M., Lin, H., Slack, P.D., Zorin, Yu.A., Mordvi- Czech Academy of Sciences grant No. 31225. nova, V.V. and Meyer, R.P., 1994. Seismic anisotropy and

mantle flow beneath the Baikal rift zone. Nature, 371: 149- 151.

Girardin, N. and Farra, V., 1992. Inversion of P-to-SH convened References phase for azimuthal anisotropy in the upper mantle: applica-

tion to the Australian GEOSCOPE station of Canberra (ab- Alsina, D. and Snieder, R., 1995. Small-scale sublithospheric stract). In: 10 Years of GEOSCOPE-Broad Band Seismology.

continental mantle deformations: constraints from SKS split- CNRS, Paris. ting observations. Geophys. L Int., 123: 431-448. Grand, S.P., 1994. Mantle shear structure beneath the Americas

J. Plomerovfz et al. / Physics of the Earth and Planetary Interiors 95 (1996) 293-309 309

and surrounding oceans. J. Geophys. Res., 99(B6): 11591- Ozalaybey, S. and Savage, M.K., 1995. Shear-wave splitting 11621. beneath western United States in relation to plate tectonics. J.

Gudmundsson, O., Davies, J.H. and Clayton, R.W., 1990. Geophys. Res., 100(B9): 18135-18149. Stochastic analysis of global traveltime data: mantle hetero- PlomerovL J., Payo, G. and Babulka, V., 1993a. Teleseismic geneity and random errors in the ISC data. Geophys. J. Int., P-residual study in the Iberian Peninsula. Tectonophysics, 221: 102: 25-43. 1-12.

Kind, R., Kosarev, G.L., Makeeva, L.I. and Vinnik, L.P., 1985. Plomerovfi, J., Sflen~, J., Babu~ka, V., Hor~lek, J., Pajdu[~ik, P., Observations of laterally inhomogeneous anisotropy in the Poupinet, G., Granet, M., Kulh~nek, O. and Arvidsson, R., continental lithosphere. Nature, 318: 358-361. 1993b. A small array study of S-wave anisotropy of the deep

Lillie, R.J., Bielik, M., Babulka, V. and Plomerovfi, J., 1994. lithosphere across the Protogene Zone in southern Sweden. Gravity modeling of the lithosphere in the eastern Alpine- Proc. XXIII General Assembly of the ESC, Prague 1992. western Carpathian-Pannonian Basin region. Tectonophysics, Geophysical Institute, Czech Acad. Sci., Prague, pp. 307-312. 231: 215-235. PlomerovL J., Babulka, V., Dorbath, C., Dorbath, L. and Lillie,

Mainprice, D. and Silver, P.G., 1993. Interpretation of SKS-waves R.J., 1994. Deep lithosphere structure across the Central using samples from the subcontinental lithosphere. Phys. Earth African shear zone in Cameroon. Geophys. J. Int., 231: 215- Planet. Inter., 78: 257-280. 235.

Mainprice, D., Vauchez, A. and Montagner, J.-P. (Editors), 1993. Plomerovfi, J., Silent, J., Arvidsson, R., Babu~ka, V., Granet, M. Dynamics of the subcontinental mantle: from seismic and Poupinet, G., 1996. A small scale study of lithospheric anisotropy to mountain building. Phys. Earth Planet. Inter., structure across the Protogine Zone, southern Scandinavia-- 78(Special Issue): 57-354. signs of a paleocontinental collision. In preparation.

Makeyeva, L., Ple~inger, A. and Hor,'ilek, J., 1990. Azimuthal Praus, O., P/~/5ov~i, J., Petr, V., Babulka, V. and Plomerov~i, J., anisotropy beneath the Bohemian Massif from broad-band 1990. Magnetotelluric and seismological determination of the seismograms of SKS waves. Phys. Earth Planet. Inter., 62: lithosphere-asthenosphere transition in central Europe. Phys. 298-306. Earth Planet. Inter., 60: 212-228.

Makeyeva, L.I., Vinnik, L.P. and Roecker, S.W., 1992. Shear-wave Savage, M.K. and Silver, P.G., 1993. Mantle deformation and splitting and small-scale convection in the continental upper tectonics: constraints from seismic anisotropy in western mantle. Nature, 358: 144-147. United States. Phys. Earth Planet. Inter., 78: 207-227.

McNamara, D.E. and Owens, T.J., 1993. Azimuthal shear wave Savage, M.K., Silver, P.G. and Meyer, R.P., 1990. Observation of velocity anisotropy in the Basin and Range Province using teleseismic shear-wave splitting in the Basin and Range from Moho Ps converted phases. J. Geophys. Res., 98(B7): 12003- portable and permanent stations. Geophys. Res. Lett., 17: 12017. 21-24.

McNamara, D.E., Owens, T.J., Silver, P.G. and Wu, F.T., 1994. Silent, J. and Plomerovfi, J., 1996. Inversion of shear-wave Shear wave anisotropy beneath the Tibetan Plateau. J. Geo- splitting parameters to retrieve three-dimensional orientation phys. Res., 99(B7): 13655-13665. of anisotropy in continental lithosphere. Phys. Earth Planet.

Minster, J.B. and Jordan, T.H., 1978. Present-day plate motions. J. Inter., 95: 277-292. Geophys. Res., 83: 5331-5354. Silver, P.G. and Chan, W.W., 1991. Shear wave splitting and

Molnar, P., 1992. Crust in mantle overdrive. Nature, 358: 105-106. subcontinental mantle deformation. J. Geophys. Res., 96: Montagner, J.-P., 1994. Can seismology tell us anything about 16429-16454.

convection in the mantle? Rev. Geophys., 32:115-137. Silver, P.G. and Savage, M.K., 1994. The interpretation of shear- Montagner, J.-P. and Nataf, H.-C., 1988. Vectorial tomography--l, wave splitting parameters in the presence of two anisotropic

Theory. Geophys. J., 94: 295-307. layers. Geophys. J. Int., 119: 949-963. Mooney, W.D. and Braile, L.W., 1989. The seismic structure of Virmik, L.P., Farra, V. and Romanowicz, B., 1989. Azimuthal

the continental crust and upper mantle of North America. In: anisotropy in the Earth from observations of SKS at Geoscope A.E. Bally and A.R. Palmer (Editors), The Geology of North and NARS broadband stations. Bull. Seismoi. Soc. Am., 79: America--an Overview. Geological Society of America, 1542-1558. Boulder, CO, pp. 39-52. Virmik, L.P., Makeyeva, L.I., Miler, A. and Usenko, A. Yu.,

Nicolas, A. and Christensen, N.I., 1987. Formation of anisotropy 1992. Global patterns of azimuthal anisotropy and deforma- in upper mantle peridotites--a review. Am. Geophys. Union tions in the continental mantle. Geophys. J. Int., 111: 433-447. Geophys. Ser., 16. Vinnik, L.P., Krishna, V.G., Kind, R., Bormarm, P. and Stammler,

Ozalaybey, S. and Savage, M.K., 1994. Double-layer anisotropy K., 1994. Shear wave splitting in the records of the German resolved from S phases. Geophys. J. Int., 117: 653-664. regional seismic network. Geophys. Res. Lett., 21: 457-460.

Joint interpretation of upper-mantle anisotropy based on teleseismic P-travel time delays and inversion of shear-wave splitting parameters

Documents