Mantle fabric of western Bohemian Massif (central Europe) constrained by 3D seismic P and S anisotropy

Tectonophysics 462 (2008) 149–163

Contents lists available at ScienceDirect

Tectonophysics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t e c t o

Mantle fabric of western Bohemian Massif (central Europe) constrained by 3D seismic P and S anisotropy Vladislav Babuška ⁎, Jaroslava Plomerová, Luděk Vecsey Geophysical Institute, Academy of Sciences of the Czech Republic, Boční II/1401, 141 31 Prague 4, Czech Republic

a r t i c l e

i n f o

Article history: Received 22 January 2007 Received in revised form 22 January 2008 Accepted 22 January 2008 Available online 22 August 2008 Keywords: Bohemian Massif P and S seismic anisotropy 3D fabric of mantle lithosphere Olivine preferred orientation

a b s t r a c t New teleseismic data from a dense network of temporary and permanent seismic stations were analyzed to investigate large-scale fabric of the lithosphere–asthenosphere system beneath the western part of the Variscan Bohemian Massif. We study three-dimensional orientation and strength of seismic anisotropy and model a fabric of the mantle lithosphere from shear-wave splitting and directional terms of relative P residuals. Fast shear-wave polarizations have mostly E–W orientation and split-delay times δt are around 1.2 s, on average. Small but systematic changes of the fast S split polarizations show a difference at about 20° to 30° between the Teplá-Barrandian (TBU)/Moldanubian (MD) tectonic units and the adjacent part of the Saxothuringian (ST), and about 40° between two parts of the ST, separated by a suture identified in the upper crust by different authors and by other methods. On the other hand, P-velocity anisotropy clearly indicates three different orientations of fossil olivine fabrics with inclined axes of symmetry in the mantle lithospheres of three tectonic units (ST, TBU, MD). The seeming contradiction between the results of the two independent data sets can be explained by different 3D anisotropic models: one with hexagonal ‘slow’ symmetry axis b and divergently dipping (a,c) foliations (ST, MD) and the other with hexagonal or orthorhombic symmetry with dipping high-velocity lineation a (TBU). These 3D self-consistent anisotropic models are compatible for both the P- and S-anisotropy observables. Our results suggest that a directionally varying constituent of the anisotropic signal is “frozen” in the mantle lithosphere. Regional changes in seismic anisotropy thus map tectonic boundaries, which cut the whole lithosphere, though with shifted crustal and mantle parts (TBU). Depending on symmetry and orientation of fabrics of the lithosphere domains, lateral changes are reflected in either P-velocity anisotropy and/or shear-wave polarizations. About half of the shear-wave split-time delays with predominantly E–W polarization azimuths represent a constant anisotropic signal which we associate with an olivine preferred orientation due to present-day flow in the asthenosphere. Our study emphasizes the importance of combining different methods of analysis and using complementary data sets in the three-dimensional analysis of mantle fabrics. © 2008 Elsevier B.V. All rights reserved.

1. Introduction Seismic anisotropy provides key information to our understanding of the tectonic fabric of the lithosphere–asthenosphere system. Thanks to systematic preferred orientation of olivine crystals and their elastic anisotropy, the European mantle lithosphere exhibits a consistent large-scale anisotropy of domains in dimensions of several hundred kilometres (Babuška and Plomerová, 2006) observed in spatial variations of both body- and surface-wave velocities. The presence of seismic anisotropy, as an almost ubiquitous property of the mantle, is now documented by extensive sets of data from temporary arrays of seismic stations covering a large variety of tectonic settings. One recently investigated region is the geodynamically active western part of the Bohemian Massif (Fig. 1). The

⁎ Corresponding author. E-mail address: [email protected] (V. Babuška). 0040-1951/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.tecto.2008.01.020

Bohemian Massif (BM), as a part of the Variscan orogenic belt representing a collage of magmatic arcs and microcontinents assembled by the collision of Laurasia (Laurentia–Baltica) and Africa (Gondwana), is the largest coherent surface exposure of basement rocks in central Europe. Geoscientists from 10 institutions in the Czech Republic, Germany and France cooperated in a multidisciplinary passive seismic experiment of the BOHEMA project (BOhemian Massif HEterogeneity and Anisotropy) to investigate the structure and dynamics of the lithosphere and asthenosphere in the western BM (Babuška et al., 2003). The region is the famous health and resort landscape of Bohemia, Saxonia and Bavaria, with more than 100 mineral water springs and several hundred gas vents in numerous mofette fields. The active tectonics is primarily manifested by frequently occurring weak to moderate earthquake swarms (Horálek et al., 2000), Cenozoic volcanism (Ulrych et al., 2000), emanations of CO2 and He of mantle origin (Weinlich et al. 1999; Bräuer et al., 2003), and by neotectonic crust movements (Bankwitz et al., 2003).

150

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

Fig. 1. Topographic map of the western Bohemian Massif (BM) and simplified boundaries of the crystalline basement of tectonic units (ST, MD, TBU) according to Kachlík (1997), Zulauf et al. (2002) and Mlčoch (2003). MLF — Mariánské Lázně Fault, WBSZ — West Bohemian Shear Zone, NBSZ — North-Bohemian Shear Zone, KHF — Krušné hory (Erzgebirge) Fault, CBP — Central Bohemian Pluton. Triangles labeled KH and ZH represent the Komorní and Železná Hůrka Quaternary volcanoes. K marks location of the Kozákov Neogene volcano, which provides lherzolite xenoliths for modelling mantle fabrics. Profiles C–D and C′–D′ indicate positions of the cross-sections shown in Fig. 7a,b. Insets show the BM position within the European Variscides and distribution of analyzed earthquakes, whose P-arrival times (epicenters marked by small circles) and shear-waveforms (SKS and SKKS phases — epicenters marked by small squares) were recorded by the BOHEMA stations (triangles and circles in the large map).

The concentration of various geodynamic phenomena in a small area calls for a systematic research of links between the near-surface structures and the deep tectonics. Central to the BOHEMA project was a passive seismic tomography experiment which operated from fall 2001 to spring 2003 (Babuška et al. 2003) and focused on construction of a tomographic model of the uppermost mantle, as well as the possible existence or non-existence of a “plume-like” low-velocity structure beneath the western BM (Plomerová et al., 2007). Besides the isotropic seismic tomography, we use the BOHEMA data to study seismic anisotropy. The dense network of seismic stations, which had not previously been operating in the BM and its surroundings, and the huge amount

of high-quality continuous recordings from the BOHEMA experiment, allowed us to examine the uppermost mantle fabric of the western BM in much greater detail than it had been done in previous studies (Plomerová et al., 1998; Babuška and Plomerová, 2001). In this paper, we use two types of data. We analyze in detail the splitting of core– mantle refracted shear waves and also use inferences on anisotropy derived from spatial variations of relative P residuals (Plomerová et al., 2007) to model independently the mantle lithosphere. A principal objective of this study is to present a 3D anisotropic model of the lithosphere–asthenosphere system in the western BM and to document the strength of combining P- and S-anisotropy investigations in studying mantle fabrics.

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

2. Tectonics of the region The western part of the BM is situated in the transition of three first-order tectonic units, the Saxothuringian (ST) in the north and the Teplá-Barrandian Unit (TBU) with the Moldanubian (MD) in the south. The lithosphere of the western Bohemian Massif was shaped by two major tectonic events: the Variscan orogeny and Cenozoic extensional tectonics and volcanic activity. The Tertiary Eger (Ohře) Rift (ER), an ENE-WSW striking structure about 190 km long characterized by high heat flow and Cenozoic volcanism, forms a boundary between the units (Fig. 1). The ER is a part of the European Cenozoic Rift System (ECRIS; Prodehl et al., 1995) and its formation is thought to be related to Alpine collision (Ziegler, 1992). The ECRIS may have a common source of “plume-like” volcanism in the mantle, manifested by the existence of “baby-plumes” beneath the French Massif Central (Granet et al., 1995) and the Eifel (Ritter et al., 2001). During the Variscan orogeny the lithosphere of the BM was affected by collisions of independent lithospheric blocks, which most probably retained their frozen olivine fabric of the mantle lithosphere formed prior to the assembly of microcontinents that created the

151

modern European landmass (Babuška and Plomerová, 2006). We can thus map approximate boundaries of individual mantle units by studying the 3D orientation of their intrinsic seismic anisotropy. The ST-TBU/MD mantle contacts most probably controlled the ER formation, local thinning of the crust (Geissler et al., 2005; Heuer et al., 2006), upwelling of the lithosphere–asthenosphere transition (Babuška and Plomerová, 2001), and localized the present-day geodynamic phenomena in the western BM. The most geodynamically active part of the region is situated on the crossing of the ER with the NNW running Mariánské Lázně fault (MLF), which follows the deep boundary between the MD in the west and the TBU in the east (Fig. 1). The existence of three major degassing fields in the western BM (Weinlich et al., 1999), as well as recently reported increasing 3He/4He ratio, which are interpreted as the first geochemical evidence for ascending, mantle-derived melt (Bräuer et al., 2005), raises a question, whether a “baby-plume” also exists beneath the BM. However, highresolution tomography down to about 250 km did not image any columnar low-velocity anomaly, which could be interpreted as a plume-like structure in the mantle (Plomerová et al., 2007). The mantle-derived fluids, as well as the Cenozoic volcanism, probably

Fig. 2. Azimuths and split-delay times δt (length of the arrows is proportional to δt) of the fast shear-wave polarizations of weighted average of all SKS and SKKS phases at each station. Large open arrows show back-azimuths of arriving waves from the E/NE (blue/black) and W (red/gray). Splitting parameters for each direction area are also color coded. Standard deviations of the average polarization azimuths are shown as a fan between two short gray sticks. Thin gray arrows stand for less reliable measurements. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

152

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

ascend from the asthenosphere along boundaries of blocks of the mantle lithosphere, assembled in the Variscan orogeny. The boundaries were reactivated during the Cenozoic extension of the weakened and thinned lithosphere and predetermined a space for the volcanism and the present-day post-volcanic phenomena in the western BM (Babuška et al., 2007). 3. Data and methods The array of seismic stations forming the BOHEMA network (Fig. 1) covered a territory of about 270 × 150 km, with its long axis oriented perpendicular to the strike of major tectonic units and to the ER. The network consisted of 61 permanent and 92 temporary stations operating between October 2001 and the end of 2003, with a core of recordings in 2002 (Plomerová et al., 2003). About 2/3 of the network consisted of three component broad-band stations, the remaining were short-period stations. Spacing of stations was generally less than 30 km, while in the central part of the array the spacing was as dense as 10–15 km, hence allowing for lateral resolution of seismic velocity and anisotropy at a scale of about 20 km in the upper mantle. To analyze the shear-wave splitting we have extracted about 130 earthquakes with magnitude larger than 5.6 at epicentral distances larger than 85° with respect to the centre of the BOHEMA array and used records with a large signal/noise ratio and well separated shear phases. Only 14 events provided shear-waveforms of sufficient quality for reliable splitting analysis at stations of the array (Table S1, electronic supplement). The distribution of events used (Fig. 1) suffers from incomplete azimuthal coverage as they generate waves which arrive to the array mainly from the NE and from the W. We primarily analyzed core–mantle refracted SKS phases. Some SKKS phases were also included in the analysis to improve the incidence-angle and backazimuth coverage. In total, we evaluated splitting for 273 SKS phases and 126 SKKS phases (see Table S1) from records that were band-pass filtered. We use Butterworth filters of the 3rd order. To set a width of the filter (see Table S1), we need to know the frequency content of the analyzed waveform. For such purpose we have applied a time– frequency analysis, particularly the wavelet transformation. We use the Morlet wavelet (Daubechies, 1992), because it prefers frequency resolution to time resolution. The wavelet transformation visualises the whole frequency–time content and allows us to set different bandpasses. Then the filtering enables the splitting to be analyzed in separate frequency windows and thus to study potential frequency dependence of the splitting parameters. We evaluated the shear-wave splitting with the code SPLIT (Vecsey et al., 2008-this issue) based on a method by Šílený and Plomerová (1996), which is a 3D generalization of the method by Silver and Chan (1991). The code SPLIT takes into account the angle of incidence of the shear wave and determines the splitting parameters in the ray-parameter coordinate system LQT, where Q stands for the radial, T for the transversal and L for the longitudinal components (see also Vecsey et al., 2008-this issue). By rotating the coordinate system and by applying a time shift of the transverse and the Q components we determine the fast S polarization by an angle Ψ in the (Q,T) plane and time delay δt of the slow split shear-wave. Then the original elliptical motion in the (Q,T) plane, reflecting an interference of the two orthogonal polarized quasi-shear waves, transforms into the linear particle motion of the fast and slow split shear waves. The transformation into the new coordinate system is determined by the angle ψ (defined by two Euler angles — azimuth φ and inclination from vertical θ) and the time shift δt. The anisotropy evaluation can be considered as a process of correcting for anisotropy by searching for a new coordinate system. When analysing core-refracted mantle phases, we prefer to minimize energy on the transverse component, which provides more stable solutions compared to correlation methods (Savage and Silver, 1993). To verify the stability of splitting parameters with respect to

noise, the parameters Ψ and δt are tested by a bootstrapping procedure (Sandvol and Hearn, 1994). The SPLIT code has been successfully applied during recent years in several European regions (e.g., Plomerová et al., 2002, 2006; Babuška and Plomerová, 2004; Vecsey et al., 2007) and is described in detail in a companion paper by Vecsey et al. (2008-this issue), together with several examples of analyzed records. Standard interpretations of splitting parameters (fast S azimuth φ and δt) result in simple one or two layer anisotropic models with horizontal ‘fast’ symmetry axes. However, splitting parameters evaluated in 3D (Ψ, δt) vary with back-azimuths in a way which, along with P-velocity variations, calls for more complicated models to comply with independent observations of anisotropy. Independent information on mantle lithosphere anisotropy can be extracted from spatial variations of relative P-wave travel-time residuals. The relative residuals are mainly generated by seismic velocity perturbations underneath the station network. More than 13,500 P-arrival times were manually picked with a high precision (Plomerová et al., 2007) on recordings of different acquisition systems and types of seismometers (short period and broad-band, Plomerová et al., 2003), involved in the field measurements. Therefore, all recordings were normalized by deconvolving the seismometer responses and convolving the signal with the WWSSN-seismometer response, before starting the picking, in order to minimize travel-time effects related to differences in acquisition systems. The P-spheres, constructed for each station with a sufficient amount of data of a good azimuth coverage, show the azimuth-incidence-angle dependent

Fig. 3. Histograms of split-delay times divided according to quality of observations listed in Table S1.

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

153

Fig. 4. a) The best quality splitting measurements (marked g in Table S1) evaluated in 3D for stations of the BOHEMA network. Fast split polarizations are mapped at the 80 km deep piercing points of S-wave rays by dots with sticks oriented in polarization azimuths to highlight any back-azimuth dependence of the splitting parameters. Blue and green sticks stand for azimuths of SKS and SKKS waves, respectively, from the E/NE, red sticks for SKS waves from the W. Colored triangles mark stations in different parts of tectonic units. SGM — Saxonian Granulite Massif, UFCSL — Upper Franconian Central Saxon Lineament. b) Rose diagrams showing azimuths of the fast shear-wave polarizations in different parts of tectonic units indicated in (a) and corresponding histograms of split-delay times. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

154

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

Fig. 5. Azimuths and split times δt of the fast shear-wave polarizations of SKS phases of good (thick arrows) and fair (thin arrows) observations of an event from the W.

terms of relative residuals. The directional variable terms are obtained from the relative residuals by an additional normalization based on directional means. The directional mean, representing an azimuthincidence-angle filtered average velocity below each station, forms a reference level in the P-spheres (Babuška and Plomerová, 1992). Standard deviations (STD) of the means at most of the stations are below ±0.1 s and only at four stations of the network (B37, BDE, SCH, BG11) they do attain ±0.2 s. We attribute long-wavelength variations in the P-spheres to velocity anisotropy in the mantle below a station. Therefore, after minimizing effects of heterogeneities and concentrating on long-wavelength variations of residuals in the P-spheres, we evaluate P-wave anisotropy. To model fabrics of the three mantle lithosphere domains identified in Plomerová et al. (2007) we invert the P-spheres and search for orientation of symmetry axes of a priori peridotite aggregates (Šílený and Plomerová, 1996). We use two basic model aggregates, one with ‘slow’ hexagonal symmetry axis b and dipping high-velocity (a,c) foliation and another one, with dipping lineation a and ‘slow’ plane (b,c) as hexagonal approximation of orthorhombic symmetry. The average anisotropy of orthorhombic aggregates is about 9% (Ben Ismail and Mainprice, 1998). The average anisotropy of the hexagonal aggregates, created by rotation along the b and a axes are 5.4% and 8.2%, respectively. The inversion assumes a mantle lithosphere domain of constant thickness and anisotropy and searches

for orientation of the symmetry axes of the aggregates to meet variations of the P-residual terms in the spheres. We showed in several previous studies (e.g., Babuška et al., 1993; Šílený and Plomerová, 1996) that seemingly contradicting results from the two data sets (P-spheres and shear-wave polarizations) are caused by the fact that the 3D phenomenon (an anisotropy with inclined symmetry axes) is approximated by a 2D one (azimuthal anisotropy of shear-wave splitting). To avoid the seeming incompatibility Šílený and Plomerová (1996) introduced the joint inversion of anisotropic parameters, namely their variation with direction of propagation within the mantle lithosphere, providing thus self-consistent models with generally oriented symmetry axes in 3D. Such models are compatible with independent findings from both the longitudinal and shear-waves and narrow a family of plausible, yet ambiguous solutions obtained from anisotropy modelling based only on one type of waves. The authors also show that mutual orientations of the mean fast S polarization azimuth and the azimuth of the high velocity derived from the P-sphere govern the symmetry with which one can approximate the large-scale mantle lithosphere anisotropy. The two azimuths are close to each other if the medium symmetry is close to the orthorhombic with dipping lineation a, or its hexagonal approximation with the ‘fast’ a axis and ‘slow’ (b,c) plane. But they can be orthogonal, if the symmetry of a medium is hexagonal with dipping ‘slow’ axis b and the (a,c) foliation. In the present paper we

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

invert the P-spheres and compare the observed and the expected splitting parameters calculated from the resulting orientations derived from the P-sphere inversions. We also estimate independently the strength of anisotropy from relative-time delays along crossing rays. In these calculations we consider appropriate lengths of rays within the mantle lithosphere and assume an average velocity of 8.2 km/s. As the ray coverage is limited and the rays most probably do not sample the velocity extremes, the strength of anisotropy represents a lower limit of the Pvelocity anisotropy. 4. Results of shear-wave splitting In accord with the splitting results of previous papers focused on the Bohemian Massif (e.g., Bormann et al., 1996; Plomerová et al., 1998; Brechner et al., 1998; Plenefisch et al., 2001; Plomerová et al., 2005), the E–W polarization azimuths of the fast split shear-waves were evaluated mainly from recordings of both portable stations and permanent observatories during the BOHEMA experiment (Figs. 2, 3, 5, 6, Table S1, electronic supplement). The data provided 364 non-null splitting measurements, out of which 156 are of high quality (good — g) with the mean standard deviation of the time delay δt of ±0.15 s and ±6° for the fast S orientation (ψ). The remaining measurements comprise 154 fair (f) and 54 poor (p) evaluations. Only 35 evaluations (9% of all measurements) return null splits, out of which 9 are good and 26 poor. Values of δt below 0.3 s were considered as null measurements. We present average fast S polarization azimuth φ and split-delay δt (Fig. 2), calculated from all non-null measurements and weighted according to their quality (Table S1). The weighting scheme is based on synthetic standard deviations assigned as 0.2 s, 0.5 s and 1.0 s for good, fair and poor measurements, respectively, and summing the Gaussian distributions of the individual values. Because we consider anisotropy as a 3D phenomenon, we do not reduce our measurements into values with only π-periodicity as it is done routinely in case of azimuthal (2D) anisotropy. We calculate the weighted means in two

155

back-azimuth ranges −32°–85° (10 events referred to as from the NE) and 248°–293° (4 events referred as from the W; see distribution of earthquake epicentres in inset of Fig. 1). In general, the E–W polarizations prevail, but at most stations the mean azimuths, calculated for opposite back-azimuths, do not differ only by π. Similarly, the time delays differ for earthquakes from either side. This fact, along with the scatter of polarization azimuths, do not reflect only errors of the measurements, but also their dependence on direction of wave propagation through an anisotropic medium with generally inclined symmetry axes. Regardless of the quality of the measurements, the highest occurrence of δt was found in the interval 1.1–1.3 s, with an average value of about 1.2 s (Fig. 3). All these observations hold for SKS/SKKS waves with dominant periods of about 8 s. Though the E–W azimuths of the fast shear waves prevail in all three tectonic units, a closer look at individual splittings plotted at piercing points at depth of 80 km allows us to detect lateral changes in the polarizations (Fig. 4a). The E–W azimuths dominate in the TBU and MD (Fig. 4b). Polarizations of shear waves rotate to the WSW in the NW margin of the BM (region denoted as ST1), while in the part of the ST adjacent to the TBU (region denoted as ST2) azimuths rotate to the WNW. Rose diagrams constructed for the groups of stations in individual units show that average polarization azimuths in the regions ST1 and ST2 differ by about 40°. They both slightly differ from average azimuths of the fast polarizations in the remaining units. Similar 25° difference between the fast split polarizations at CLL, MOX (permanent observatories in ST1) and BRG (permanent observatory in ST2) was found by Brechner et al. (1998). Though distributions of polarization azimuths are statistically the same in the TBU and MD, there is a small difference between the split-delay times (Fig. 4b). To detect in more detail possible geographical changes of the mantle fabrics, we have examined lateral variations of the splitting parameters evaluated for shear waves of single events recorded at stations of the whole region. For this purpose we selected events

Fig. 6. a,b) Two events with almost identical back-azimuths show different orientations of SKS/SKKS polarizations. See the text for discussion of this observation.

156

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

with almost opposite back-azimuths and plot the polarizations for both the SKS and SKKS phases (Figs. 5, 6). Polarizations for the western back-azimuth (Fig. 5) follow the general trend as described above (see also Fig. 4), including the difference in the fastpolarization azimuths in regions ST1 and ST2. Moreover, the signals

in the central part of the TBU adjacent to the WBSZ (stations B50, B13, KLD) were weak and disturbed. Smaller δt were evaluated closer to the centre of the TBU. The most distinct feature in the polarizations of shear-waves with the NE back-azimuths is their sensitivity to incidence angles. Note in Fig. 6a,b and in Table S1 that a

Fig. 7. Schematic ray tracing along two profiles located in Fig. 1. The lithosphere–asthenosphere boundary is taken from Babuška and Plomerová (2001), the crust thickness from Geissler et al. (2005), Heuer et al. (2006) and Beránek and Zátopek (1981). The rays represented by blue dashed lines denote means of the negative directional terms of relative residuals b−0.1 s, red dot-dashed lines stand for means of the positive directional terms N 0.1 s, while the rays representing the directional terms close to zero are dotted. The zone of the mantle lithosphere limited by a change from negative to positive directional terms is shaded. We also give estimates of effective P-velocity anisotropy calculated from mean residuals of crossing rays. On the top are smoothed P-residual spheres, reflecting directional dependence of relative P velocities in the mantle lithosphere beneath all stations (triangles) shown for each of the three tectonic units. Blue triangles and red circles stand for negative and positive residuals, respectively, while crosses mark residuals close to zero. Most of the relative residuals are between −0.5 and 0.5 s, though exceptionally they reach values around ±1 s. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

difference in incidence angle of only 4° results in polarization changes as large as 10–15° in azimuths, considering only reliable splitting determinations. Deviations between azimuths of the fast SKS and SKKS increase systematically from the eastern TBU to its westernmost part, the region of disturbed signals for an event with the western back-azimuth (Fig. 5). 5. P-velocity anisotropy To estimate independently the strength and orientation of seismic anisotropy, we analyze spatial variations in relative travel times of P waves. In Fig. 7 we present two cross-sections through the lithosphere– asthenosphere running perpendicular to the ER over the three tectonic units with different P-sphere patterns. The cumulative P-spheres shown are summed from individual spheres of three groups of stations (see Table 1) situated along the profiles (Fig. 1) in belts of about 25 km wide. The ST residual pattern in the northern part of the investigated region exhibits the largest negative residuals for waves approaching stations from the north-northwest. The TBU pattern is characterized by negative residuals representing relatively high velocities from the east, or from directions forming a fan between the northeast and southeast. In the MD the P-residual spheres show the relatively high-velocity directions for south, or south-western propagations prevailingly. The three major types

Table 1 Inversion of anisotropy-related parts of relative P residuals a

157

Table 1 (continued) a bb Tectonic unit

‘Slow’ ‘Fast’ ab axis

‘Fast’ ‘Slow’(a,c) (b,c) foliation

Thickness

Stations

θ

φ

α

φ

(km)

60

355

60

175

40

Profile C–D b Saxothuringian TAU, MOX, ZEU, BG09, Tectonic unit BG10, OTR, GUNZ, WERN, ROHR, NKC, Stations BERN Profile C–D Teplá-Barrandian Saxothuringian BG22, LAC, B34 TAU, MOX, ZEU, BG09, BG10, OTR, GUNZ, WERN, ROHR, NKC, BERN Moldanubian Teplá-Barrandian B13, B47, KHC BG22, LAC, B34

Moldanubian Profile C′–D′ B13, B47, KHC Saxothuringian NIC, REU, BG07, B38, B41, B42

Profile C′–D′ Saxothuringian Profile C′–D′ NIC, REU, BG07, B38, Teplá-Barrandian B41, B42 B36, B37, B30, RBC, B45

‘Fast’ a axis

‘Slow’ (b,c) foliation

Thickness

θ

φ

α

φ

(km)

60 60

95 355

60 60

275 175

30 40

55 60

15 95

55 60

195 275

10 30

55

15

55

195

10

60

0

35

180

50

60 40

0 105

35 65

180 285

50 20

40 65

105 355

65 50

285 175

20 30

Fabric orientation

Fabric orientation

Tectonic unit

‘Slow’ b axis

‘Fast’ (a,c) foliation

Thickness

Stations

θ

φ

α

φ

(km)

40

175

40

355

70

Profile C′–D′ Teplá-Barrandian Moldanubian B36, B37, B48, B12 B30, RBC, B45

Teplá-Barrandian BG22, LAC, B34

50

310

50

130

40

Moldanubian 355 50 vertical 175 upward) 30 Azimuth φ and inclination65 θ (measured from are Euler angles B48, B12 the symmetry axis; α is inclination of the (b,c) foliation plane of olivine defining aggregate measured from horizontal downward; cumulative P-spheres constructed for (continued on next page) the named stations are inverted.

Moldanubian B13, B47, KHC

55

0

55

180

90

35

190

35

10

110 (5%)⁎ 80 (7%)⁎

Teplá-Barrandian B36, B37, B30, RBC, B45

65

315

65

135

40

Moldanubian B48, B12

50

10

50

190

50

Profile C–D Saxothuringian TAU, MOX, ZEU, BG09, BG10, OTR, GUNZ, WERN, ROHR, NKC, BERN

Profile C′–D′ Saxothuringian NIC, REU, BG07, B38, B41, B42

Fabric orientation

Azimuth φ and inclination θ (measured from vertical upward) are Euler angles defining the symmetry axis; α is inclination of the (a,c) foliation plane of olivine aggregate measured from horizontal downward; ⁎ strength of P anisotropy; cumulative P-spheres constructed for the named stations are inverted. (continued on next page)

of the residual pattern, reflecting the mantle lithosphere fabrics of the three tectonic units of the western BM, have been recognized not only along these profiles, but also throughout the whole BOHEMA array (Plomerová et al., 2007). Divergently dipping fabrics of the ST and MD mantle lithospheres were also modelled in the westernmost rim of the BM (Plomerová et al.,1998, 2005), where both units juxtapose each other. Three-dimensional orientation of the P anisotropy can be obtained by inverting the anisotropic residual perturbations in the P-spheres. We inverted the cumulative P-spheres (Fig. 7) using the hexagonal aggregates with ‘slow’ b or ‘fast’ a symmetry axes and P anisotropies of 5.4% and 8.2%. Three different fabrics in the three units resulted from the inversions (Table 1). We obtained minimum misfit functions for an isotropic velocity range of 8.2–8.3 km/s. When inverting the spheres, we do not consider a dip of the lithosphere–asthenosphere boundary, as we do in the estimates of P anisotropy from crossing rays along the profiles in Fig. 7. This can reduce the anisotropy estimate at some stations and thus reduce the thickness (MD) because of a tradeoff between strength of anisotropy and thickness of the modelled anisotropic block of mantle lithosphere. Nevertheless, the modelling with 8.2 km/s mean velocity provided thicknesses, which are compatible with more general models of the lithosphere thickness (Babuška and Plomerová, 1992) and orientations of dipping highvelocity (a,c) foliations (Plomerová et al., 2005). Only the thickness of the ST in Profile C′–D′ seems to be too large. Increasing the anisotropy reduces the lithosphere thickness. However, we have to keep in mind

158

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

that especially in cases when only P residuals are inverted, the resolution of the thickness is low and thickness estimates have to be considered suggestive only. A grid search with a step of 5° was applied to determine the symmetry axes (Table 1). However, a real uncertainty of axes orientations is probably larger due to limited azimuth and incidence-angle ranges of rays. We have also made estimates of the in-situ P-velocity anisotropy of the ST and MD mantle lithospheres from the observed differences in the anisotropic part of P residuals in volumes which are sampled by waves approaching stations from relatively high- and low-velocity directions. The TBU mantle domain is too narrow to allow such an estimate. We have calculated the residual perturbations in the spheres at single stations in two opposite azimuth fans in each unit. Strikes and azimuths of the three fabrics differ and cannot be either parallel or perpendicular to a profile which would be “optimal” for the anisotropy estimate. Therefore, we selected the azimuth intervals at 45° around the maximum and minimum residual perturbations in the P-spheres. To estimate anisotropy in lithosphere blocks along the profile, we considered appropriate length of the rays within the mantle lithosphere (Fig. 7) based on models of the crust thickness (Beránek and Zátopek, 1981; Geissler et al., 2005; Heuer et al., 2006) and the lithosphere–asthenosphere boundary (Babuška and Plomerová, 2001). Assuming the average P velocity in the mantle lithosphere at ∼ 8.2 km/s, we have obtained in-situ effective P-velocity anisotropy in the mantle lithosphere between 3% and 8% in the ST unit and between 2% and 6% in the MD unit. As the width of the TBU mantle lithosphere in the south-eastern part of the profiles (southward of the ER) is only 60–80 km, some of the ‘crossing’ rays sample both the TBU and MD lithospheres with different fabrics, which may lower the estimate of the MD anisotropy, namely around the contact of both units. When evaluating the effective anisotropies, we have to keep in mind that most of the considered seismic rays propagate within the mantle lithosphere at angles between 23° and 30° and most probably do not survey velocity extremes. We can thus consider the calculated strength as a lower limit of real in-situ anisotropy. Moreover, due to differences in strikes of the fabrics, we estimate the P anisotropy from synthetic rays off the profile, whose length estimates can be biased. All theses aspects result in a large variation of the P-anisotropy estimates in blocks of the lithosphere along the profiles. 6. Discussion 6.1. Depth location of anisotropy and anisotropy–heterogeneity trade-off It is well known that vertical resolution of anisotropy modelled from splitting of the core–mantle refracted waves (SKS) is limited due to their sub-vertical propagation in the mantle and the crust beneath seismic receivers. This leads to a trade-off between the magnitude of anisotropy and the thickness of anisotropic medium and affects their vertical resolution. The shear-wave splitting evaluated in the western BM attains 1.2 s on average (Figs. 3, 4; Table S1). A similar average delay time of 1.3 s was found by Plenefisch et al. (2001) in the German part of the western BM. The source of the splitting reflects wave propagation through anisotropic media that can reside at any depth in the mantle and also in the crust. The heterogeneous crust, with its small-scale anisotropic bodies (e.g., schistose metamorphic rocks) with different orientations can produce only small-scale heterogeneous anisotropy within a tectonic unit and usually contributes to the observed splitting delays by no more than 0.1–0.3 s (Barruol and Mainprice, 1993). Effective anisotropy of 6% caused by near-surface cracks sampled by the high-frequency shear waves was measured in the uppermost crust of west Bohemia (Vavryčuk, 1993). Similarly, Málek et al. (2005) found about 5% P-velocity anisotropy in the upper crust from controlled sources and local earthquakes. Růžek et al. (2003) modelled P

anisotropy of only about 1.5–2.5% from data of two refraction profiles crossing different tectonic units of the BM. However, these anisotropies are restricted to a few kilometres of the uppermost crust and can contribute only by a small fraction to the observed split times of long-wavelength teleseismic SKS phases analyzed on BB recordings, whose split-delay times exceed 1 s. We may ask, whether the major anisotropic signal of the observed S-wave anisotropy resides in the mantle lithosphere, or in the sub-lithosphere mantle. An intrinsic anisotropy of the upper mantle, caused mainly by the preferred orientation of olivine crystals, is at least partly constrained by anisotropy of mantle xenoliths. Nikitin et al. (2001) obtained 7.1% and 9.3% P anisotropy for two samples and Christensen et al. (2001) calculated 8% and 6% P- and S-wave anisotropies, respectively, for a larger set of spinel lherzolite xenoliths from the Neogene Kozákov volcano (see Fig. 1), the only locality in the BM providing samples suitable for calculations of anisotropy from olivine preferred orientations. These values, as well as average P and S anisotropies of 9.5% and 6.6%, calculated from 110 olivine petrofabrics from a variety of the upper mantle geodynamic environments worldwide by Ben Ismail and Mainprice (1998), represent maximum high-frequency anisotropy of hand specimens. However, large-scale mantle fabric strengths in the Earth differ. Christensen (2002) has demonstrated that the largescale anisotropy of peridotite massifs is smaller due to their heterogeneous fabrics. Therefore, 4% average S-velocity anisotropy estimated by Silver and Chan (1991) may be close to a typical largescale anisotropy of the mantle lithosphere. However, simple anisotropic models of the upper mantle assume horizontal symmetry axis a. Then, a split-delay time of 1 s is generated for sub-vertically propagated SKS by a 115 km thick anisotropic upper mantle layer. On the other hand, in some regions the mantle large-scale anisotropy can be larger (Vecsey et al., 2007) allowing for fabrics with dipping axes of symmetry (Babuška et al., 1993). From the perspective of 4% S anisotropy, the thickness of the mantle lithosphere in the western BM, modelled between about 50 km and 100 km (Babuška and Plomerová, 2001), need not be sufficient to explain completely the observed δt values averaging at 1.2 s, particularly in case of dipping fabrics. The dipping fabric reduces split-delay times as the angle between an incident wave and the symmetry axis decreases. However such lithosphere is thick enough to accommodate half of the average time delays, which we relate to the variable constituent of the shear-wave anisotropy. The second half we attribute to olivine preferred orientation due to present-day flow in the asthenosphere (Plomerová et al., 1998). An additive effect of both layers on the observed splitting is probably caused by both having approximately E–W orientations of the fast split azimuths due to the flow in the asthenosphere and the E–W strikes of the dipping mantle lithosphere fabrics. Our findings are in accord with results of other authors who show that stable continental regions generally contain seismic anisotropy that is located within both the lithosphere and the sub-lithospheric mantle (e.g., Levin et al., 1999; Fouch et al., 2000; Gung et al., 2003; Walker et al., 2004; Fouch and Rondenay, 2006). Flow in the asthenosphere as a mechanism of seismic anisotropy beneath and around the BM has been suggested also by Bormann et al. (1996), Brechner et al. (1998) and Plenefisch et al. (2001). Because the fast-polarization azimuths in central Europe do not correspond well to the direction of the absolute present-day plate motion of 130–140° (Minster and Jordan, 1978), Bormann et al. (1996) suggested that the flow in the asthenosphere may be deflected by the southward lithosphere thickening from the BM beneath the Alps (Babuška and Plomerová, 1992, 2001). In such a case, the E–W asthenosphere flow orients fast olivine a axes in the same direction and, according to Bormann et al. (1996), explains a major part of average delay times they observed (δt = 0.83 ± 0.31 s) and small changes in average fastpolarization azimuths. However, these authors assume only 2% mantle anisotropy of shear waves and thus increase thickness of the

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

anisotropic medium, which includes a large portion of the sublithospheric mantle. Though we have observed larger anisotropies reflected in average split-delay time δt = 1.2 ± 0.3 s, we argue that in agreement with our previous results (Plomerová et al., 1998; Babuška and Plomerová, 2001), the source of the P- and the variable constituent of S-velocity anisotropy is caused by fossil olivine fabric in the mantle lithosphere. Besides the large anisotropy of ultramafic xenoliths representing the BM mantle lithosphere (Christensen et al., 2002) and considering Fresnel zones of the waves, we have only indirect arguments to support this interpretation. A simple estimate of the depth of anisotropy from Fresnel zones (e.g., Fouch and Rondenay, 2006) assumes models with horizontal a axis and no variations of the splitting with back-azimuth. Then, from overlapping zones of waves recorded at an array of stations one can localize the anisotropy above or below the appropriate depth. But this depth estimate cannot be used in regions where we model inclined anisotropic structures. One of the most important indirect indications of locating a significant part of anisotropy into the mantle lithosphere is the large extent of an anisotropic pattern consistent within a tectonic unit and its often observed abrupt change near important tectonic boundaries (Babuška and Plomerová, 2006, for review). Though differences in orientations of the fast polarizations between the units of the BM are often small (Fig. 2), they correlate with the near-surface tectonics and indicate a continuation of geological features from the upper crust to the mantle, similarly to the P-anisotropy lateral variations. Both the azimuthal dependence of the fast split polarizations and the orientations of bipolar P-residual spheres, consistent for individual tectonic units or their parts, allow us to interpret the observed anisotropy by olivine aggregates with inclined symmetry axis. The

159

inclined fabrics are thus “frozen” and inherent to the domains of mantle lithosphere, separated by sutures, like the North-Bohemian Shear Zone (NBSZ) between the ST and TBU, or other tectonic boundaries like WBSZ between the MD and TBU (Fig. 1). Besides the small difference between the TBU and the ST adjacent to the ER (ST2 in Fig. 4), we also observe different orientations of the fast shear wave between two parts of the Saxothuringian (ST1 and ST2, Fig. 4). The transition between the parts has a general NE-SW trend of the Variscan belt in central Europe and coincides with the Upper Franconian Central Saxon Lineament (Tischendorf et al., 1995) and with a zone of discontinuous reflections, offset of upper-crustal seismic refraction velocity layers, and a different crust resistivities increasing towards the Krušné hory Mts. (Erzgebirge), found by Krawczyk et al. (2000). The latter authors claimed that all geophysical structures mark a structural change between the Saxonian Granulite Massif and the Erzgebirge down to 15 km, indicating a different origin and development of the two units. Our observations report different orientations of the fabric in the ST1 and ST2 mantle lithospheres, which means that the crustal boundary may have a mantle continuation. The two different fabrics within the mantle lithosphere of the ST unit reflect a long memory of preexisting olivine orientation and thus support plate–tectonic views of origin of the continental lithosphere and locate at least a part of the observed anisotropy into the mantle lithosphere (Babuška and Plomerová, 1989, 2006). Shear-wave splitting is generally accepted as a proof of anisotropy, but P travel-time residual variations tend to be often attributed solely to effects of heterogeneities, instead of searching for an adequate contribution to existing velocity anisotropy. Relative Presiduals, cleared from effects of the crust and of distant

Fig. 8. a,b) Conformity of the observed and synthetic splittings calculated for the aggregates retrieved in inversions of the P-spheres (Table 1). The observed fast shear-wave azimuths and split-delay times δt, colored according to the conformity, are shown for the broad-band stations along the two profiles (Fig. 1). The orientations of the hexagonal b axis model (part a) explain well directional variations of the splitting within the ST and MD, while this symmetry fails to explain the splitting at stations located above the TBU mantle lithosphere (see also Fig. 7). On the other hand, the hexagonal a axis model (part b) explains much better the polarizations at the three TBU stations, while it fails to model the observations at stations in the ST and MD. The dashed lines approximate the TBU mantle domain boundaries, which are north-westerly shifted (dashed double arrows) relative to the crustal boundaries of the unit and indicate and underthrusting of the MD mantle lithosphere beneath the south-eastern rim of the TBU.

160

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

Fig. 9. Schematic block-diagram shows the tectonic situation of the lithosphere–asthenosphere system in the western Bohemian Massif. Hatching indicates the divergently dipping (a,c) foliations of frozen olivine fabrics in the mantle lithosphere of the two major tectonic units (ST, MD) and an eastward dipping lineation a of the TBU fabric. While the short wavelength of P waves enables differently oriented fabric of the TBU mantle lithosphere to be mapped (see the diagrams in Fig. 7 and Table 1), the SKS wave alone does not recognize such relatively small structure of a different symmetry and fabric orientation. Below are different orientations of the olivine fabrics and corresponding schematization of the traveltime deviations related to the anisotropy (+ and − for relatively late and early arrivals) and average shear-wave splitting azimuths (left and right diagrams in the pairs, respectively). In anisotropic media approximated by hexagonal symmetry with ‘fast’ a axis (TBU and asthenosphere) the high-velocity directions derived from P-spheres (dashed arrows) and average polarization of the fast shear-wave (full double arrows) parallel. Note the seeming inconsistency between the high-velocity directions from the P-spheres and the average fast S polarizations in case of the hexagonal ‘slow’ b axis approximations (ST, MD). The ER has developed above the mantle transition between the ST in the north-west and the TBU/MD in the south-east.

heterogeneities, reflect both the isotropic and anisotropic structure of the upper mantle beneath stations. Isotropic velocity heterogeneities are imaged in isotropic teleseismic tomography (Plomerová et al., 2007) and in models of the lithosphere–asthenosphere boundary (LAB, Plomerová et al., 1998; Babuška and Plomerová, 2001). As neglecting anisotropy affects isotropic tomographies (Sobolev et al., 1999), also heterogeneities can bias anisotropy pattern in the Pspheres. The LAB represents the most distinct large-scale heterogeneity in the uppermost mantle. Therefore, we should ask to what extent can the lithosphere thinning beneath the western ER influence the pattern of relative P residuals, which show convergent relative low velocities pointing beneath the ER (Fig. 7). Babuška et al. (2002) showed that the updoming of the LAB above the plume beneath the French Massif Central (MC, Granet et al., 1995) affects Pvelocity difference between crossing rays by no more than 0.1 km/s, even if one assumes a velocity contrast between the lithosphere and a low-velocity asthenosphere as large as 0.6 km/s. As the lowvelocity anomaly in the asthenosphere beneath the ER is less pronounced and broader (Plomerová et al., 2007) than that beneath the MC (Granet et al., 1995), we can conclude that only a negligibly small portion of the anisotropic part of the relative residuals can be accounted for by the LAB relief beneath the western BM. Generally, most of methods dealing with upper mantle anisotropy (shear-wave splitting) and with mantle heterogeneity (velocity tomography) are used independently of one another (Fouch and Rondenay, 2006). Combining all available methods and data sets is thus also necessary to reduce the anisotropy–heterogeneity trade-off.

6.2. Joint interpretation of P and S anisotropy evaluated in 3D We have been advocating for many years, in agreement with recent findings of Schulte-Pelkum and Blackman (2003) and Fouch and Rondenay (2006), that besides the traditional azimuthal shearwave anisotropy, it is necessary to study spatial variations of relative P velocities and to interpret them jointly (Babuška et al., 1993), to get a more thorough view of 3D anisotropy and a fabric of the lithosphere– asthenosphere system. Back-azimuthal and geographical variations of the splitting, deviations of the fast S polarizations between the SKS and SKKS using the same source–receiver pair in several regions (Plomerová et al., 1998; Fouch and Rondenay, 2006 and references therein; Vecsey et al., 2007), or the ‘bipolar’ pattern of the P-spheres (Babuška and Plomerová, 1992) suggest inclined fabrics in mantle domains. Plomerová et al. (1998) modelled directionally varying constituents of the splitting measured in the ST and MD by anisotropic mantle lithosphere with divergently dipping (a,c) foliations. This hexagonal model with ‘slow’ b axis explains the seeming contradiction between the fast S polarization azimuths and high-P-velocity directions. The 3D self-consistent models are compatible with results from the two independent sets of anisotropy observables, which would not be the case, if the mantle fabric is modelled by only azimuthal shear-wave anisotropy and approximated by models with horizontal symmetry axes a. Babuška et al. (1993) have demonstrated that shear waves, which propagate sub-vertically within anisotropic media with inclined

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

symmetry axes b split in such a way that the fast S is polarized in azimuths that form a narrow fan along strike of the structures. Usually, such variations are attributed only to errors of splitting evaluations caused by noise (Vecsey et al., 2008-this volume) and average polarizations are considered in modelling. Then, however, two lithosphere domains with either divergent or convergent dipping fabrics of similar strikes cannot be distinguished in shearwave azimuthal anisotropy (Babuška and Plomerová, 2001). The divergently dipping fabrics of the MD and ST can be detected only if anisotropy is evaluated as a 3D phenomenon and independent observables, such as shear-wave splitting and P-spheres are interpreted jointly. In the BM, both the frozen dipping olivine fabrics, modelled by azimuthal anisotropy, and the approximately E–W oriented flow in the asthenosphere probably work in the same direction to enhance the observed splitting. To model the mantle lithosphere constituent of the observed shear-wave anisotropy we calculated synthetic splittings in the aggregates retrieved from inversions of the P-spheres (Table 1). We use geometry of the real data (back-azimuths and incidence angles) evaluated at broad-band stations along the two profiles (Fig. 1) and show deviations between the observed fast S polarizations and the synthetics calculated for the hexagonal aggregates both with the ‘slow’ b axes and the ‘fast’ a axes of symmetry (Fig. 8). The orientations of the hexagonal b axis model (Table 1a) explain well the splitting variations within the ST and MD (Fig. 8a), while this symmetry fails to explain the splitting at stations located above the TBU mantle lithosphere. On the other hand, the hexagonal a axis model (Table 1b) explains much better the splitting at the three TBU stations, while, as expected, it fails to model observations at stations in the ST and MD (Fig. 8b). Almost no change of fabric orientations of the three mantle units would be detected if only the fast S polarization azimuths were mapped (Babuška and Plomerová, 2001). But distinct changes of the mantle lithosphere fabrics are clearly detected in the P-spheres (Plomerová et al., 2007). The joint interpretation of the P- and Sanisotropy observations results in 3D self-consistent models of different fabrics of the three units. The mantle domain boundaries inferred from the changes of the P anisotropy indicate a NW shift of the mantle lithosphere domain of the TBU relative to its crustal part as well as the MD underthrusting beneath the south-eastern margin of the TBU mantle lithosphere (Babuška et al., 2007; see also Fig. 7). Fitting the observed and synthetic splittings delimits the ST, TBU and MD domain boundaries (Fig. 8) similarly as they were derived from the P anisotropy. Dimensions of the ST and MD mantle lithospheres are much larger than that of the TBU. Therefore, the observables can be also more sensitive to the wavelength, particularly in the TBU. As we do not observe any particular difference in the observed shear-wave splitting between the TBU and MD, we can speculate that SKS wave does not “see” the difference. This can happen if the size of the sampled volume is comparable to or smaller than the wavelengths. Unlike the shear waves with average dominant period of about 8 s, the short-period P waves recognized different orientations of fabrics below the MD and TBU (Fig. 7, Table 1). Fig. 9 documents the importance of evaluating the anisotropy through “wavelength optics” (see also Marson-Pidgeon and Savage, 1997, for frequency-dependent anisotropy due to ‘wavelength/scale-length-of-heterogeneities’ relation). Only one of our events had enough energy at longer periods to allow us to analyze the SKS splitting for a wave with dominant period of 16 s. The analysis returned null splits at most of stations in the BM, which otherwise show split SKS waves arriving from similar azimuths and distances (Fig. 4 and Table S1), but with dominant periods about 8 s. Saltzer et al. (2000) and SchultePelkum and Blackman (2003) found that SKS apparent fast azimuths are weighted towards the upper portion of the model, however, when the upper layer thickness approaches a wavelength,

161

the average “fast” azimuth changes to that of a thicker bottom layer. Our observation of prevailing null splits at 16 s indicates that the thickness of the anisotropic mantle lithosphere might be less than the corresponding wavelength of this SKS wave (∼ 80 km). This is in accord with thickness of the mantle lithosphere in the western BM which does not exceed that value (Plomerová et al., 1998). On the other hand, resulting null splits do not necessarily mean an absence of anisotropy. Nulls also appear if waves propagate in special directions, which depend on the symmetry of the medium. Our data yield nulls for only 8% of the source/ receiver pairs and this is a surprisingly low score compared with studies in other tectonic environments showing, e.g., about 42% of null measurements in the Northern Apennines (Salimbeni et al., 2007), or a large number of unconstrained measurements in the Eifel region (Walker et al., 2005). In both settings, flow effects in the sub-lithospheric mantle related to subduction (Northern Apennines), or to an ascending and then deflected mantle plume (Eifel), are probably important. We found the short-period teleseismic P waves that survey the upper mantle at scales of their wavelengths (around 10 km, see Fig. 9), and under a broader cone of directions of propagations, compared with the SKS phases, to be very sensitive to dipping axes of anisotropy (see also a review by Fouch and Rondenay, 2006). The analysis of azimuth-incidence-angle (directional) terms of relative residuals (P-spheres) is thus very useful in modelling different orientations of fabric of the mantle lithosphere and especially in mapping boundaries of the lithosphere domains. Unlike the shearwave splitting, the P-velocity anisotropy clearly recognized three different patterns of the P-spheres (Plomerová et al., 2007 and Fig. 7), from which we modelled fabrics of the three domains – the ST, TBU and MD (Table 1) – divergently dipping (a,c) foliations in the ST and MD (Babuška and Plomerová, 1992; Plomerová et al., 1998; Babuška and Plomerová, 2001) and easterly dipping lineations a in the mantle lithosphere fabric of the TBU (Plomerová et al., 2005, 2007). The lateral extent of the TBU is small compared with the MD or ST and its vertical extent is a matter of debates. Because we have detected a clear signature of its mantle lithosphere, we do not consider this unit as only a crustal terrane (Babuška et al., 2007). We suggest that the TBU mantle lithosphere, which may represent a vestige of the TBU Cadomian block, is thinned after being partly resorbed by the Variscan intervention of the hot asthenosphere material (Zulauf et al., 1998). The domain is probably too small and thin to be “seen” by SKS waves, but large enough for surveying by the ∼ 10 km wavelength of teleseismic P waves. Another advantage of short-period P-wave propagations is that contrary to the anisotropic propagation of broad-band SKS phases, they are only weakly affected by a sub-lithosphere anisotropy. An asthenosphere flow orients olivine a axes sub-horizontally and thus the only effect could come from a velocity difference between the steepest propagations (∼20° from vertical) and less steep rays (∼ 50°). Resulting small travel-time deviations, however, are probably below the accuracy of the method. This is an additional reason, why the anisotropy derived from the P-spheres can be attributed to the mantle lithosphere. The apparent anisotropies, calculated from fans of crossing rays (Fig. 7), can be easily interpreted solely in terms of frozen fabric of the mantle lithosphere with a thickness between 50 and 100 km. Slightly higher apparent P anisotropies observed in the ST as compared to the TBU + MD (Fig. 7) may be explained by propagation of P waves through only one block of the mantle lithosphere in the ST, and through two blocks of the mantle lithosphere with different orientations of the TBU and MD fabrics (Figs. 7, 9 and Table 1). It is also possible that intrinsic anisotropy of the MD mantle lithosphere is smaller as its frozen fabric might have been partly destroyed by intensive heating processes during the Variscan orogeny (Zulauf et al., 1998). On the other hand, Vauchez et al. (2005) investigated the influence of a hot asthenospheric mantle on the base of the

162

V. Babuška et al. / Tectonophysics 462 (2008) 149–163

lithosphere from mantle rocks of the Ronda Massif and xenoliths from Tanzania and Polynesia volcanoes and concluded that the tectonic fabric inherited from previous deformation and the resulting seismic anisotropy was only slightly modified by advection of heat from the asthenosphere. 7. Conclusions We present 3D self-consistent models of three mantle lithosphere domains inferred from both the shear- and longitudinal-wave anisotropy evaluated at a dense network of seismic stations in the western Bohemian Massif. • Shear-wave splitting parameters show only small but systematic changes of the prevailingly E–W fast split polarizations over the whole region with average delay times of 1.2 s. On the other hand, the P-velocity anisotropy has recognized three distinctly different orientations of fabrics in the mantle lithosphere of the three tectonic units — the Saxothuringian (ST), Moldanubian (MD) and TepláBarrandian (TBU). • Average fast shear-wave polarization azimuths differ only at about 20° to 30° between the MD/TBU tectonic units and the adjacent part of the ST, and about 40° between two parts of the ST. The two ST subregions are separated by a suture identified in the upper crust by different authors and by other methods. Seismic anisotropy thus maps tectonic boundaries, which cut the whole lithosphere, though with shifted crustal and mantle parts (TBU). • The seemingly contradicting results from the two data sets (only weak variations in the shear-wave splitting, but distinct differences in anisotropy from the P-spheres) occur only if the 3D phenomenon (fossil anisotropy of mantle lithosphere domains with inclined symmetry axes) would be interpreted as a 2D one (azimuthal anisotropy of shear-wave splitting) and without a change of medium symmetry. • The splitting delays contain a contribution from preferred olivine orientation due to a present-day generally E–W flow in the asthenosphere, while the anisotropic part of the P travel-time delays is affected only negligibly by such sub-horizontal flow. • We model fabrics of the mantle lithosphere of the three units by peridotite aggregates of hexagonal symmetry with easterly dipping lineations a (TBU) and divergently dipping foliations (a,c) (to the NNW in the ST, to the S-SW in the MD). Such self-consistent anisotropic models result from joint interpretation of inversion of the P-residual spheres and shear-wave splitting synthetics. The resolved fabrics are compatible for the two independent data sets of anisotropic parameters. We emphasize the importance of combining different methods of analysis and using complementary data sets, as well as of considering sensitivity of different wavelengths to mantle fabrics in 3D imaging of anisotropy of individual domains of the continental mantle lithosphere. Acknowledgements The authors thank to U. Achauer for cooperation during the field work and data processing. Thoughtful reviews of Martha Savage, Donna Blackman and an anonymous reviewer, as well as careful editorial work of Andrew Frederiksen, substantially improved the original manuscript. Funding from the Czech Grant Agency (grants Nos. 205/04/0748 and 205/07/1088) and the Grant Agency of the Czech Academy of Sciences (No. KJB300120605) provided financial support to this work. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.tecto.2008.01.020.

References Babuška, V., Plomerová, J., 1989. Seismic anisotropy of the subcrustal lithosphere in Europe: another clue to recognition of accreted terranes? In: Hillhouse, J.W. (Ed.), Deep Structure and Past Kinematics of Accreted Terranes. AGU Geophys. Monograph 50, vol. 5. American Geophysical Union, Washington DC, pp. 209–217. Babuška, V., Plomerová, J., 1992. The lithosphere in central Europe — seismological and petrological aspects. Tectonophysics 207, 141–163. Babuška, V., Plomerová, J., 2001. Subcrustal lithosphere around the Saxothuringian– Moldanubian Suture Zone — a model derived from anisotropy of seismic wave velocities. Tectonophysics 332, 185–199. Babuška, V., Plomerová, J., 2004. Sorgenfrei-Tornquist Zone as the mantle edge of Baltica lithosphere: new evidence from three-dimensional seismic anisotropy. Terra Nova 16, 243–249. Babuška, V., Plomerová, J., Šílený, J., 1993. Models of seismic anisotropy in deep continental lithosphere. Phys. Earth Planet. Inter. 78, 167–191. Babuška, V., Plomerová, J., 2006. European mantle lithosphere assembled from rigid microplates with inherited seismic anisotropy. Phys. Earth Planet Inter. 158, 264–280. doi:10.1016/J.pepi.2006.01.010. Babuška, V., Plomerová, J., Vecsey, L., Granet, M., Achauer, U., 2002. Seismic anisotropy of the French Massif Central and predisposition of Cenozoic rifting and volcanism by Variscan suture hidden in the mantle lithosphere. Tectonics 21 11-1-20. doi:10.1029/2001TC901035,2002. Babuška, V., Plomerová, J., BOHEMA, W.G., 2003. BOHEMA seismic experiment: search for an active magmatic source in the deep lithosphere in central Europe. EOS, Trans. AGU 84, 409–417. Babuška, V., Plomerová, J., Fischer, T., 2007. Intraplate seismicity in the western Bohemian Massif (central Europe): a possible correlation with a paleoplate junction. J. Geodyn. 44, 149–159. doi:10.1016/j.jog.2007.02.004. Bankwitz, P., Schneider, G., Kämpf, H., Bankwitz, E., 2003. Structural characteristics of epicentral areas in central Europe: study case Cheb Basin (Czech Republic). J. Geodyn. 35, 5–32. Barruol, G., Mainprice, D., 1993. A quantitative evaluation of the contribution of crustal rocks to shear-wave splitting of teleseismic SKS waves. Phys. Earth Planet. Inter. 78, 281–300. Beránek, B., Zátopek, A., 1981. Earth's crust structure in Czechoslovakia and in central Europe by methods of explosion seismology. In: Zátopek, A. (Ed.), Geophysical Syntheses in Czechoslovakia. Veda, Bratislava, pp. 243–264. Ben Ismail, W., Mainprice, D., 1998. An olivine fabric database: an overview of upper mantle fabrics and seismic anisotropy. Tectonophysics 296, 145–157. Bormann, P., Grünthal, G., Kind, R., Montag, H., 1996. Upper mantle anisotropy beneath central Europe from SKS wave splitting: effects of absolute plate motion and lithosphere–asthenosphere boundary topography? J. Geodyn. 22, 11–32. Bräuer, K., Kämpf, H., Strauch, G., Weise, S.M., 2003. Isotopic evidence (3He/4He, 13CCO2) of fluid triggered intraplate seismicicty. J. Geophys. Res. 108 (B2), 2070. doi:10.1029/ 2002JB002077. Bräuer, K., Kämpf, H., Niedermann, S., Strauch, G., 2005. Evidence for ascending upper mantle-derived melt beneath the Cheb basin, central Europe. Geophys. Res. Lett. 32, 1–4. doi:10.1029/2004GL022205 L08303. Brechner, S., Klinge, K., Krüger, F., Plenefisch, T., 1998. Backazimuthal variations of splitting parameters of teleseismic SKS phases observed at the broadband stations in Germany. Pure Appl. Geophys. 151, 305–331. Christensen, N.I., 2002. Continental mantle seismic anisotropy: a new look at the Twin Sisters massif. Tectonophysics 355, 163–170. Christensen, N.I., Medaris, L.G., Wang, H.F., Jelínek, E., 2001. Depth variation of seismic anisotropy and petrology in central European lithosphere: a tectonothermal synthesis from spinel lherzolite. J. Geophys. Res. 106, 645–664. Daubechies, I., 1992. Ten Lectures on Wavelets. SIAM, Philadelphia, PA., USA, p. 357. Fouch, M.J., Rondenay, S., 2006. Seismic anisotropy beneath stable continental interiors. Phys. Earth. Planet Inter. 158, 292–320. Fouch, M.J., Fischer, K.M., Parmentier, E.M., Wysession, M.E., Clarke, T.J., 2000. Shear wave splitting, continental keels, and patterns of mantle flow. J. Geophys. Res. 105 (B3), 6255–6275. Gung, Y.C., Panning, M., Romanowicz, B., 2003. Global anisotropy and the thickness of continents. Nature 422 (6933), 707–711. Geissler, W.H., Kampf, H., Kind, R., Bräuer, K., Klinge, K., Plenefisch, T., Horálek, J., Zedník, J., Nehybka, V., 2005. Seismic structure and location of a CO2 source in the upper mantle of the western Eger (Ohře) Rift, central Europe. Tectonics 24, TC5001. doi:10.1029/2004TC001672. Granet, M., Wilson, M., Achauer, U., 1995. Imaging a mantle plume beneath the Massif Central (France). Earth Planet. Sci. Lett. 17, 1109–1112. Heuer, B., Geissler, W.H., Kind, R., Kämpf, H., 2006. Seismic evidence for asthenospheric updoming beneath the western Bohemian Massif, central Europe. Geophys. Res. Lett. 33, L05311. doi:10.1029/2005GL025158. Horálek, J., Fischer, T., Boušková, A., Jedlička, P., 2000. The western Bohemia/Vogtland region in the light of the WEBNET network. Stud. geophys. geod. 44, 107–125. Kachlík, V., 1997. The Kladská unit. In: Vrána, S., Štědrá, V. (Eds.), Geological Model of Western Bohemia Related to the KTB Borehole in Germany. Sborn. Geol. Věd, Geologie, 47, pp. 70–80. Krawczyk, C.M., Stein, E., Choi, S., Oettinger, G., Schuster, K., Götze, H.J., Haak, V., Prodehl, C., Schulze, A., 2000. Geopysical constraints on exhumation mechanisms of highpressure rocks: the Saxo-Thuringian case between the Franconian Line and Elbe Zone. In: Franke, et al. (Ed.), Orogenic Processes: Quantification and Modelling in the Variscan Belt. . Geol. Soc., vol. 179. Spec. Publ., London, pp. 303–322. Levin, V., Menke, W., Park, J., 1999. Shear wave splitting in Appalachians and Urals: a case for multilayer anisotropy. J. Geophys. Res. 104, 17975–17994.

V. Babuška et al. / Tectonophysics 462 (2008) 149–163 Málek, J., Horálek, J., Janský, J., 2005. One-dimensional qP-wave velocity model of the upper crust for the West Bohemia/Vogtland earthquake swarm region. Stud. Geophys. Geod. 49, 501–524. Marson-Pidgeon, K., Savage, M.K., 1997. Frequency-dependent anisotropy in Wellington, New Zealand. Geophys. Res. Lett. 24, 3297–3300. Minster, J.B., Jordan, T.H., 1978. Present-day plate motions. J. Geophys. Res. 83, 5331–5354. Mlčoch, B., 2003. Character of the contact between the Saxothuringian and TepláBarrandian unit. Geolines 16, 75. Nikitin, A.N., Ivankina, T.I., Burilichev, D.E., Klíma, K., Lokajíček, T., Pros, Z., 2001. Anisotropy and texture of olivine-bearing mantle rocks at high pressures. Izv. Phys. Solid Earth 37, 59–72. Plenefisch, T., Klinge, K., Kind, R., 2001. Upper mantle anisotropy at the transition zone of the Saxothuringicum and Moldanubicum in southeast Germany revealed by shear wave splitting. Geophys. J. Int. 144, 309–319. Plomerová, J., Babuška, V., Šílený, J., Horálek, J., 1998. Seismic anisotropy and velocity variations in the mantle beneath the Saxothuringicum–Moldanubicum contact in central Europe. Pure Appl. Geophys. 151, 365–394. Plomerová, J., Babuška, V., Vecsey, L., Kouba, D., 2002. TOR Working Group, 2002. Seismic anisotropy of the lithosphere around the Trans-European Suture Zone (TESZ) based on teleseismic body-wave data of the TOR experiment. Tectonophysics 360, 89–114. Plomerová, J., Achauer, U., Babuška, V., Granet, M., BOHEMA, W.G., 2003. Passive seismic experiment to study lithosphere–asthenosphere system in the western part of the Bohemian Massif. Stud. Geophys. Geod. 47, 691–701. Plomerová, J., Vecsey, L., Babuška, V., Granet, M., Achauer, U., 2005. Passive seismic experiment MOSAIC — a pilot study of mantle lithosphere of the Bohemian Massif. Stud. Geophys. Geod. 49, 541–560. Plomerová, J., Babuška, V., Vecsey, L., Kozlovskaya, E., Raita, T., SSTWG, 2006. Proterozoic-Archean boundary in the upper mantle of eastern Fennoscandia as seen by seismic anisotropy. J. Geodyn. 41, 400–410. doi:10.1016/j.jog.2005.10.008. Plomerová, J., Achauer, U., Babuška, V., Vecsey, L., BOHEMA, W.G., 2007. Upper mantle beneath the Eger Rift (Central Europe): plume or asthenosphere upwelling? Geoph. J. Int. 169, 675–682. doi:10.1111/j.1365-246X.2007.03361.x. Prodehl, C., Mueller, S., Haak, V., 1995. The European Cenozoic rift system. In: Olsen, K.H. (Ed.), Continental Rifts: Evolution, Structure, Tectonics. Elsevier, pp. 133–212. Ritter, J.R.R., Jordan, M., Christensen, U.R., Achauer, U., 2001. A mantle plume below the Eifel volcanic fields, Germany. Earth Planet. Sci. Lett. 186, 7–14. Růžek, B., Vavryčuk, V., Hrubcová, P., Zedník, J., CELEBRATION Working Group, 2003. Crustal anisotropy in the Bohemian Massif, Czech Republic: observations based on Central European Lithospheric Experiment Based on Refraction (CELEBRATION) 2000. Journ Geophys. Res. 108 (B8), 2392. doi:10.1029/2002JB002242. Salimbeni, S., Pondrelli, S., Margheriti, L., Levin, V., Park, J., Plomerová, J., Babuška, V., 2007. Abrupt change in mantle fabric across northern Apennines detected using seismic anisotropy. Geophys. Res. Lett. 34, L07308. doi:10.1029/2007GL029302. Saltzer, R.L., Gaherty, J.B., Jordan, T.H., 2000. How are vertical shear wave splitting measurements affected by variations in the orientation of azimuthal anisotropy with depth? Geophys. J. Int. 141, 374–390.

163

Sandvol, E., Hearn, T., 1994. Bootstrapping shear-wave splitting errors. Bull. Seismol. Soc. Am. 84, 1971–1977. Savage, M.K., Silver, P.G., 1993. Mantle deformation and tectonics: constraints from seismic anisotropy in western United States. Phys. Earth Planet. Inter. 78, 207–227. Schulte-Pelkum, V., Blackman, D.K., 2003. A synthesis of seismic P and S anisotropy. Geophys. J. Int. 154, 166–178. Šílený, J., Plomerová, J., 1996. Inversion of shear-wave splitting parameters to retrieve three-dimensional orientation of anisotropy of continental lithosphere. Phys. Earth Planet. Int. 95, 277–292. Silver, P., Chan, W.W., 1991. Shear wave splitting and subcontinental mantle deformation. J. Geophys. Res. 96, 16429–16454. Sobolev, S., Gresillaud, A., Cara, M., 1999. How robust is isotropic delay time tomography for anisotropic mantle? Geophys. Res. Lett. 26, 509–512. Tischendorf, G., Förster, H.J., Frischbutter, A., Kramer, W., Schmidt, W., Werner, C.D., 1995. V.B.3 igneous activity. In: Dalmeyer, et al. (Ed.), Pre-Permian Geology of Central and Eastern Europe. Springer-Verlag, Berlin Heidelberg, pp. 249–259. Ulrych, J., Cajz, V., Pivec, E., Novák, J.K., Nekovařík, Č., 2000. Cenozoic intraplate alkaline volcanism of western Bohemia. Stud. geophys. geod. 44, 346–351. Vauchez, A., Dineur, F., Rudnick, R., 2005. Microstructure, texture and seismic anisotropy of the lithosphereic mantle above a mantle plume: insights from the Labait volcano xenoliths (Tanzania). Earth Planet. Sci. Lett. 232, 295–314. Vavryčuk, V., 1993. Crustal anisotropy from shear-wave splitting observations in West Bohemia, Czech Republic. Bull. Seismol. Soc. Am. 83, 1420–1441. Vecsey, L., Plomerová, J., Babuška, V., 2008-this issue. Shear-wave splitting measurements — problems and solutions. Tectonophysics. doi:10.1016/j.tecto.2008.01.021. Vecsey, L., Plomerová, J., Kozlovskaya, E., Babuška, V., 2007. Shear-wave splitting as a diagnostic of variable anisotropic structure of the upper mantle beneath central Fennoscandia. Tectonophysics 438, 57–77. http://dx.doi.org/10.1016/j.tecto.2007.02.017. Walker, K.T., Nyblade, A.A., Klemperer, S.L., Bökelmann, G.H.R., Owens, T.J., 2004. On the relation between extension and anisotropy: constraints from shear-wave splitting across the East African Plateau. J. Geophys. Res. 109 (B8), B08302. Walker, K.T., Bokelmann, G.H.R., Klemperer, S.L., Bock, G., 2005. Shear-wave splitting around the Eifel hotspot: evidence for a mantle upwelling. Geophys. J. Int. 163, 962–980. Weinlich, F.H., Brauer, K., Kampf, H., Strauch, G., Tesař, J., Weise, S.M., 1999. An active subcontinental mantle volatile system in the western Eger rift, central Europe: gas flux, isotopic (He, C, and N) and compositional fingerprints. Geochim. Cosmochim. Acta 63, 3653–3671. Ziegler, P.A., 1992. European Cenozoic rift system. Tectonophysics 208, 91–111. Zulauf, G., Fiala, J., Kotková, J., Maluski, H., 1998. Rheologischer Kollaps eines unterkarbonischen Hochplateaus in der Böhmischen Masse: Die Rolle der Nordböhmischen Scherzone. Freib. Forsch.hefte, C 471, 261–263. Zulauf, G., Bues, C., Dörr, W., Vejnar, Z., 2002. 10 km minimum throw along the West Bohemian shear zone: Evidence from dramatic crustal thickening and high topography in the Bohemian Massif (European Variscides). Int. J. Earth Sci. 91, 850–864.