S receiver functions obtained from seismograms of teleseismic events recorded at 78 European permanent broad-band stations are used to estimate the thickness of the European lithosphere. Our results provide new, independent information about the lithospheric thickness beneath the Precambrian platform of Eastern Europe and the Phanerozoic platform of central Europe. Detailed high-resolution images of the lithosphere–asthenosphere boundary (LAB) reveal indications for a typical continental lithosphere of about 100 km thickness beneath a majority of stations within Central Europe, whereas in the vicinity of the Trans-European Suture Zone (TESZ), the lithosphere thickens to about 130 km. A relatively thin lithosphere of 80 km was found beneath the Upper Rhine Graben region suggesting that the Cenozoic extension affects the whole lithosphere. No clear signal from the LAB was detected beneath the Alps and Carpathians. The LAB Sp phase might be disturbed by complicated structure due to ongoing collision/subduction in these regions, or the data are not yet sufficiently dense. A relatively thicker lithosphere of about 120 km was found beneath the SW part of the Bohemian Massif that was formed during the Variscan orogeny. We found an LAB depth of about 190 km near a single station located in the Vrancea area/Eastern Carpathians, which is characterized by the occurrence of intermediate deep earthquakes. Beneath the stations located in the Precambrian platform of Eastern Europe, the LAB deepens to approximately >200 km, even though the converted phase from the LAB is not as sharp as found beneath other stations located in Central Europe or even is missing.
The thickness of mobile lithospheric plates is still an important question in debate. The geodynamics and kinematics of the plate tectonics strongly depend on the thickness and lithosphere–asthenosphere boundary (LAB) topography of the moving plates. Most probably, the lower boundary of the lithosphere represents the major shear zone within the Earth's mantle. However, so far it is problematic to detect and map this elusive boundary and up to now there is no general agreement between different geophysical methods in many regions.
The concept of ‘lithosphere’ goes back to the definition of an ‘elastic lithosphere’ studying the postglacial rebound. The lithosphere and asthenosphere were introduced to allow isostatic compensation (Barrell 1914; for a review see Jackson et al. 2008). Later on, heat flow and petrological studies lead to the definition of a ‘thermal lithosphere’, which base was defined as the 1300 °C isotherm. The calculated thickness of the thermal lithosphere strongly depends on heat production in the crust (lithosphere) and heat transport in sedimentary basins (Babuška & Plomerová 1993; Čermak 1993; Artemieva & Mooney 2001). The concept of thermal lithosphere seems to work especially well for oceanic plates, where an age dependent cooling leads to a thickening of the lithospheric mantle. The ‘electrical lithosphere’ was defined as the high-resistivity layer above a highly conducting zone in the upper mantle (Praus et al. 1990; Korja 2007; Smirnov & Pedersen 2009). Finally, the ‘seismic lithosphere’ represents a layer (shell) with high seismic velocities above a regional low velocity zone in the upper mantle. Gutenberg (1959b) found strong indications for the existence of this low-velocity zone in the upper mantle. The LAB might also be characterized by a change in seismic anisotropy. So far the main tool to detect the lower limits of the lithosphere seismically was the study of shear wave velocity distribution with depth from surface wave studies. The problem of this method is mainly the resolution of surface waves, which is in the order of 300–400 km laterally and 30–50 km vertically. Geodynamically, there is the question if the lithosphere is mechanically (rheologically) decoupled from the asthenosphere by a major shear zone.
Eaton et al. (2009) gave a review on available geophysical tools to study the LAB. Generally, if the base of the lithosphere—the LAB—would be a more or less sharp transition, then this seismic discontinuity should be observable studying converted phases. The problem of the Ps receiver function method (Vinnik 1977; Kind et al. 1995) is that multiple converted phases from the Moho discontinuity mask the time interval of the possible arrivals from the LAB. However, with the recently developed Sp receiver function method (e.g. Farra & Vinnik 2000; Yuan et al. 2006) separation of direct conversions and multiples is possible. Only in case of a more than 50-km-thick transition it might not be detectable by S receiver functions (Eaton et al. 2009).
The first studies of S-to-P converted phases focused on conversions from the mantle transition zone discontinuities (Faber & Müller 1980, 1984; Bock & Kind 1991). Farra and Vinnik (2000); Bock (1994) and Wilson et al. (2006) warned that multiples or scattered waves might be erroneously taken for converted phases. However, this problem should be solved if many receiver functions recorded at one station from many different distances are summed after distance moveout correction was applied. More recently, the Sp receiver function method became a useful tool to map the thickness of the seismic lithosphere, for instance, in the western Bohemian Massif (Heuer et al. 2007), NW Atlantic (Kumar et al. 2005a), Tien Shan (Kumar et al. 2005b), Tibet (Kumar et al. 2006), the Aegean region (Sodoudi et al. 2006), Canada (Yuan et al. 2006) and in Hawaii (Li et al. 2004). Recently, first observations of P-to-S and S-to-P converted phases were reported from the centre of oceanic plates away from islands by Kawakatsu et al. (2009). All these studies found strong converted phases with negative polarity, which were interpreted as originating from the LAB.
Unfortunately, high-resolution seismic observations of the mantle lithosphere in Europe are rarely available. Among the first studies on shear wave velocity distribution beneath central and Eastern Europe is the work of Neunhöfer (1985). He found indications for the existence of a shallow low-velocity zone in the upper mantle beneath central Europe, whereas it seemed to be missing beneath the East European platform but also beneath the Bohemian Massif. In 1988, Babuška & Plomerová compiled a map of European lithosphere thickness, mainly from P residual studies. They mapped thick lithosphere (>200 km) beneath the southwestern margin of the East European Platform, beneath two zones in the eastern and western Alps, and beneath a narrow zone in the East Carpathians. The central parts of the Alps have a lithospheric thickness of approximately 140 km, comparable to the Moldanubian part of the Bohemian Massif. According to Babuška & Plomerová (1988, 2006), the central European Variscan fold belt has a lithosphere 100–140 km thick. Furthermore, they found thin lithosphere beneath regions of subsidence (Pannonian Basin, Po Plain, North German-Polish Basin and Belgo-Dutch Platform) as well as beneath the northern part of the Rhine Graben, the southern part of the Rhenish Massif, and beneath the Eger Graben in the NW part of the Bohemian Massif. These results more or less also concur with results from P-wave traveltime tomography of the European upper mantle (e.g. Spakman 1991; Koulakov et al. 2009), which shows relatively low seismic velocities in the upper 200 km beneath Central Europe and faster velocities beneath the craton and the active orogens. Geissler et al. (2008) showed that also differences in Ps delay times of conversions at the top and bottom of the mantle transition zone beneath central and Eastern Europe are caused by differing upper mantle structure/composition beneath both areas. Beneath the active collision zone of the Alpine–Carpathian belt the mantle transition zone itself is affected by subduction processes.
First regional P and S receiver function studies to map lithosphere thickness beneath the Bohemian Massif were carried out by Heuer et al. (2006, 2007). They found indications for a sharp transition from LAB at about 80–90 km depth in the Saxothuringian Zone towards 120–130 km depth in the Moldanubian Zone within the Bohemian Massif. In this work, we investigate the lithosphere beneath Central and Eastern Europe by means of S receiver functions and report on observations, which might image the base of the European lithosphere.
Tectonics of the Study Area
In surface geology, there exist huge differences between the Pre-Cambrian East European Craton (EEC), the Palaeozoic platform of central Europe (consolidated during the Caledonian and Variscan orogenies), and the recently active collision zone of the Alpine–Carpathian orogenic system (e.g. Banka et al. 2002; Fig. 1). However, it is still unresolved how deeply these different tectonic units continue into and related processes affect the upper mantle.
The EEC, finally consolidated in the late Precambrian, is divided from Central Europe by the Trans-European Suture Zone (TESZ), which runs from the North Sea in the northwest to the Black Sea in the southeast. The TESZ is commonly split into the Teisseyre-Tornquist (TTZ) and Sorgenfrei-Tornquist zones (STZ). The Thor Suture is generally interpreted as a thrust of Caledonian nappes on Baltica (EEC). The basement of Central Europe was mainly consolidated after the Caledonian (early Palaeozoic) and Variscan (Late Palaeozoic) orogenies. During the Mesozoic, Central Europe was further consolidated and covered by platform sediments. Its southern periphery was influenced by the rifting and opening of the Tethys Ocean. During the late Mesozoic and Cainozoic, the study area southwest of the TESZ was influenced strongly by the Alpine orogeny and rifting in the North Atlantic. The Alpine–Carpathian belt is a result of the Tertiary African–European convergence and of the closure of the Tethys Ocean, associated with the collision of several microplates with the European plate (e.g. Nemcok et al. 1998; Sperner et al. 2001; Seghedi et al. 2005). Within the central Alps (Austrian, Swiss), subduction of the central European lithosphere is observed towards the south (e.g. TRANSALP Working Group 2002; Kummerow et al. 2004; Lüschen et al. 2006). In the Carpathians, subduction is (was) mostly directed towards the Pannonian basin (e.g. Sperner et al. 2001).
To study the LAB using Sp conversions we used data from 78 permanent seismic stations, mostly located in central and Eastern Europe (Fig. 1). Details about the stations and corresponding network can be found in Table 1. Data was mined mainly via GEOFON and IRIS data portals. Some stations are in operation for more than 20 yr as in the case of the stations of the Gräfenberg Array in SE Germany (stations GRA1, GRB1 and GRC1). All available events with a magnitude >5.7 (NEIC catalogue) were analysed. The amount of data from the different stations and networks vary significantly depending on the period of operation. For the three three-component stations of the Gräfenberg Array comprising data starting from 1980 about 280 from 1050 checked teleseismic events could be used for Sp analysis (epicentral distance range 60°–85°). The events were selected based on the signal-to-noise ratio of the SV wave (normally about a factor of 3 or 4), the overall signal quality and depending if the amplitudes of the SV were bigger than the SH waves. Since most of the analysed events occurred N–E from the stations, the results might be dominated by observations from these backazimuth range.
The S receiver function method (Farra & Vinnik 2000; Yuan et al. 2006) was applied in order to isolate S-to-P (Sp) converted phases from the incident S phases. The method comprises component rotation, deconvolution, moveout correction, and summation of many traces. The S-to-P converted waves arrive at the station earlier than the direct S waves, whereas the multiple reverberations arrive after the S onset. In Ps receiver functions potential primary conversions from the LAB are often masked by the arrival of strong multiple crustal reverberations in the same time window. This problem is avoided in the S receiver function technique.
The processing of the data, the moveout correction and stacking basically follow the description of, for example, Kumar et al. (2006). However, we did a slightly different rotation of the ZRT into the LQT coordinate system. After rotation from ZNE into ZRT coordinate system with theoretical backazimuth angles, we estimated the incidence angle through polarization analyses of the Z (P) and R (SV) components (equivalent to the P receiver function method, see Kind & Vinnik 1988). In contrast to the P receiver function method, both traces have to be rotated in that way that SV energy is maximal on the Q trace and minimal on the L component (see Fig. 2). This approach provides reasonable estimates of the incidence angle of the primary S waves.
In the isotropic case, the L component contains only Sp converted phases. The converted signals can be enhanced trough the deconvolution with the SV signal from the Q trace in time domain (window length used for analyses is about 100 s), which also achieves source equalization. The time axis and sign of amplitudes are reversed so that the S-to-P converted phases have positive arrival times and can be compared to the results of P receiver function analysis that analyses P-to-S converted phases. Further processing steps include moveout correction for slowness of 6.4 s deg−1 as it was previously applied also for Ps receiver function processing (e.g. Yuan et al. 1997; see also Yuan et al. 2006 for example, station CLZ –Fig. 3).
The frequency content of the signals used in the two methods is different: P receiver functions have dominating wave periods around 1 s, while for S receiver functions wave periods around 5 s prevail. Hence the resolution is also different. A thin layer of anomalous velocity might be detectable by P-receiver functions and invisible in S-receiver function data. A sharp gradient might be better visible in P-receiver function data, while a broad gradient might be better to detect in S-receiver function data.
We analysed the data without restitution of true ground displacement to keep the higher frequencies of the recordings. However, the data was filtered using a third-order Butterworth filter with corner periods of 3 and 50 s. Since we only choose high quality recordings, there is no influence from microseismic noise in the high-frequency band. Testing the influence of bandpass filtering we recognized, that already a low pass with corner periods of 7 s cause major loss of resolution (see Appendix A). For recognition of Sp converted phases the period band between 3 and 7 s is essential.
Fig. 4 shows the stack traces of Sp receiver functions for the European broad-band observatories. At all stations we can see clearly a S-to-P conversion in the time interval from 3 to 6 s lead time (grey, positive, meaning velocity increase downward) with relative amplitudes of 5–15 per cent (relative to the deconvolved primary S phase), which most probably stem from the Moho discontinuity. Furthermore, we observe strong multiple converted phases from the crust in the time interval after the S-wave arrival (negative lead times). Slightly before the arrival of the Moho Sp phase we observe (single/multiple) relatively strong phases with negative amplitude (relative amplitudes of about 3–6 per cent) indicating velocity decrease downwards. As it is indicated by studying this phase in specific backazimuth and epicentral distance windows (Appendices B and C), it is obvious, that this phase is not that coherently observed as the Sp converted phase from the Moho (crust–mantle transition). Making several tests, stacking different distance, focal depth, data from different time periods we estimate an uncertainty of the earlier Sp phases in the order of 0.5–1 s lead time.
We differentiate three quality classes of observation of this earlier Sp phase. Class A contains 32 stations with relatively clear (outstanding) single Sp phase with negative polarity in front of the Moho Sp phase. Class B contains 39 stations, which show multiple Sp phases with negative polarity in front of the Moho Sp phase. Class C contains five stations, which do not show any clear Sp phase with negative polarity in front of the Moho Sp phase. Sometimes, phases show up, which looks like side lobes of a strong Moho Sp arrival (e.g. at station STU); then normally there is a similar phase (side lobe) on the other side of SMohop phase. These stations are also classified as class B stations. Some stations marked by parentheses in figures or Table 1 might be more uncertain, even in case of class A, due to the small amount of stacked events (e.g. station HAM).
Fig. 5 shows the distribution of Sp piercing points for the interpreted LAB depths (listed in Table 1). It can be clearly seen, that most of the stations are dominated by observations from northern and northeastern backazimuths. The most western stations have also many events from the W. The lateral distance of the Sp piercing points from the station range from about 80 km for a 80 km deep LAB towards 150 km for a 140 km deep LAB. In case of a very deep LAB of about 200 km, the piercing points are more than 250 km away from the observing stations. This big lateral offset of observation points is a major problem for the detection of weak signals from not-pronounced lithosphere–asthenosphere transitions, especially in the case of complex lithospheric structure, as it is the case for active orogens (Alps, Carpathians) and the boundary between thin and thick lithospheres (TTZ).
Synthetic Receiver Functions
To better understand the observed signals we computed Sp receiver functions from synthetic seismograms using the reflectivity method (Kind 1985). We tested several models that might reproduce the differences in lithospheric structure beneath Europe (Fig. 6). As starting model we used the reference Earth model IASP91 (Kennett 1991) without any asthenosphere. We replaced the sharp Moho in the IASP91 model with a gradient zone between 25 and 34 km and introduced gradient zones ontop (G discontinuity; centred at 60, 80, 100, 120, 140 and 220 km depth) and at the bottom (L discontinuity; centred at 240 or 360 km depth) of a hypothetical asthenosphere. Generally, we modelled the LAB (G discontinuity) as a gradient zone with a thickness of 20 km (Table 2, Appendix E). The velocity contrast at the top is 3.3 per cent for P waves and 6.5 per cent for S waves. We tested three models representing the lithosphere of the EEC (all without an asthenosphere); in model EEC5D we introduced an intralithospheric discontinuity related to the change from depleted to undepleted upper mantle (see Geissler et al. 2008; Hacker & Abers 2004). Furthermore, we tested different velocity contrasts across the LAB assumed in 100 km depth (Fig. 7). We computed seismograms for epicentral distances from 60° to 84° and analysed them similar to the real data (see Appendix F).
Fig. 6 shows the stack traces from the stacked synthetic receiver functions. The Moho and its multiples are the most impressive phases as also observed in the real data. Clear negative signals are observed from the top and base of the asthenospheric layer in the models (see also coherent negative phases in Appendix F). As it is obvious from Fig. 6, the models without any asthenosphere show no significant negative phase in the stack traces, however, looking at single distance traces (Appendix F) it cannot be ruled out, that spurious phases might be misinterpreted as LAB in case of real data only from a narrow distance range. Model EEC5D show a broad negative phase stemming from the velocity contrast related to the change from depleted to undepleted lithosphere. Model LAB80B shows the effect of a slab-like structure within the asthenospheric upper mantle, as it might be the case in subduction or collision zones.
Fig. 7 shows the effect of different velocity contrast across the LAB. The velocity contrast as well as the Vp/Vs ratio contrast has a major effect on the amplitudes of the related Sp converted phases (see also Table 2). As it is obvious from this figure the velocity contrast modelled with model LAB100-4 (no contrast in Vp/Vs ratio) is already that small that the primary conversions has only the same amplitude as the side-lobe from the strong Moho conversion. This implies that it might be difficult to detect weak LABs.
The conversions at 3–6 s lead time can be attributed to the Moho discontinuity (or better crust–mantle transition) or in few cases to the base of sediments, as it was observed also by P -to-S receiver functions at the same set of stations (Geissler et al. 2008). Within the uncertainty range of the methods the Ps delay and Sp lead times from the seismic boundaries fit nicely. The multiple Sp phases from these crustal discontinuities arrive at negative lead times (Fig. 4), allowing the study of the sub-Moho uppermost mantle.
The interpretation of the phases with negative amplitudes arriving just in front of the Moho Sp phases is more difficult. If we exclude the origin as side lobes and scattered or multiple S phases, they can be interpreted as conversions at lithospheric discontinuities, most probably at the base of the lithosphere, following previous works of Yuan et al. (2006) or Sodoudi et al. (2006). The amplitudes of the assumed LAB converted phases (about 20–40 per cent of the Moho Sp phase) agree with synthetic receiver functions (Yuan et al. 2006), where the LAB was assumed as an isotropic 5 per cent velocity drop.
No corresponding phases could be observed in Ps receiver functions for most of the stations (see Geissler et al. 2008; see also Appendix A). This is not surprisingly, since Ps receiver functions are dominated by mid-crustal and Moho reverberations in this time range following the primary crustal conversions (see also Geissler et al. 2005).
Depth of Lab Beneath Europe from Single Station Analysis
Interpreting the negative phases between about 8 and 22 s lead time observed in single station analysis (Fig. 4) as primary converted phases from the base of the lithosphere we can map the lithospheric thickness in Europe with Sp receiver functions with higher lateral resolution then previously possible (Fig. 8). The lead time can be translated into depth using the IASP91 reference earth model (Kennett 1991). We multiplied the observed lead times by the empirical value 8.94 km s−1.
The typical continental lithosphere beneath central Europe has a thickness of about 90–110 km (10–12 s lead time) that can be observed at a great number of stations using Sp converted phases. A thinner lithosphere of only about 75 km (8.3 s lead time) was found beneath parts of the Pannonian Basin (station BUD). Another region of lithospheric thinning was observed beneath the northern part of the Upper Rhine Graben (station ECH, BFO). This area is characterized by a 75–85 km deep LAB (8.5–9.3 s lead time) suggesting that the Cenozoic extension may influence the whole lithosphere. No consistent signal from the LAB was detected at stations in the Alps and Carpathians. The LAB phase might be disturbed (scattered) by a complicated structure due to the subduction occurring in these regions. Complicated Ps receiver functions were reported for instance by Kummerow et al. (2004) in the Eastern Alps. There is a weak indication for a deep LAB beneath station MLR (190 km, 21 s lead time). Since the piercing points are located mostly NE of the stations, this observation might be affected already by the SW edge of the EEC. A relatively thicker lithosphere was found beneath the SW part of the Variscan Bohemian Massif (stations WET, GEC2). The LAB is estimated at about 115 km depth (13 s lead time) beneath this area. In the vicinity of the TESZ, the lithosphere probably thickens to about 115–130 km (stations RGN, KWP; 13–14.5 s lead time).
More difficult is the situation at stations in the East European Craton. Stacked S receiver functions clearly show converted phases from the Moho and confirm the presence of a thick crust beneath the old craton (∼5–6 s, about 40–50 km thick; Figs 2 and 4b). However, we did not observe a strong phase from a deep LAB, as it can be assumed from diverse studies (e.g. Shapiro & Ritzwoller 2002; Babuška & Plomerová 2006; Artemieva 2007; Koulakov et al. 2009). Only at two stations on the craton we observe phases, which might be converted phases from a deep LAB (KIEV and SUW; 20–22 s lead time; 180–200 km depth; Fig. 2b) but the amplitudes of these phases are rather low. At many stations situated on the craton we observe converted phases in the time range from 10 to 14 s lead time (just in front of the Moho Sp phase). Considering previous results on lithospheric thickness beneath cratons, we cannot interpret these phases as Sp conversions from the LAB. If we accept that they are real Sp conversion from the upper mantle they might be related to intralithospheric layering (or structure).
Depth of Lab Beneath Central Europe from Common Conversion Point (CCP) Stacking
Whereas station density in Eastern Europe is low, there exist a large number of stations in central Europe (mainly Germany and Czech Republic). In this area piercing points at the assumed LAB depth partly overlap allowing a common conversion point (CCP) stacking of single traces. For CCP (box) stacking we calculated the piercing points for Sp rays in 100 km depth and stacked them in boxes as shown in Figs 9 and 10 (only if there is minimum of 10 traces per box; see also Table 3). However, it has to be stated that the bias caused by the majority of events in the northeast of the study area is also existent for CCP stacking. The resulting pattern in Fig. 9 does not differ too much from the single station results shown in Fig. 8.
There are negative phases between 10 and 14 s observed beneath northern Germany and Denmark (boxes Y, Z, A) close to the contact between the Palaeozoic platform and the Scandinavian shield. Generally the density of rays is not yet sufficient, so only first indications for the depth of LAB can be achieved.
Throughout the main part of Germany and the northern part of Czech Republic (boxes B–G) Sp phases between 8 and 11 s lead time indicate a shallow LAB between 70 and 100 km depth. Even if we observe some scattering there is a quite homogenous area in SW Germany with observations of less than 10 s lead time (LAB in less than 90 km depth).
In the Alpine–Carpathian foreland and in the area of the southern Bohemian Massif (boxes G11–16, H–J) Sp phases, which might stem from the LAB, are observed at 13–17 s, indicating a deepening of the LAB in that area up to 150 km depth. However especially in the area of the Alps and Carpathians observed Sp receiver functions are complex and show multiple negative phases. So our interpretation is not unique.
New Observations in Context of Existing Data
Previous studies mapped the European LAB at depths between 50 km beneath the Pannonian basin and the Eifel Plume (Rhenish Massif) up to more than 200 km beneath the Alps and the East European craton (Babuška & Plomerová 1988, 2006). Their results partly concur with P-wave tomography studies (Spakman 1991; Koulakov et al. 2009) and surface wave studies (e.g. Shapiro & Ritzwoller 2002). However, the problem of traveltime tomography studies is how to define the LAB. Values of about 80 km for the depth of the LAB beneath central Germany were also found by a recent surface wave study (Bischoff et al. 2004). Farra & Vinnik (2000) already studied Sp conversions beneath station GRA1, but found no clear indications for an isotropic LAB. We also do not observe a clear isotropic signal at this station, which might be caused by a complex crustal and lithospheric structure as it is also indicated by Ps receiver functions (e.g. Geissler et al. 2008).
We cannot observe any clear correlation of LAB depth and the position of Palaeozoic tectonometamorphic units at the surface. This is in agreement with missing deep-seismic observations of any subduction/collision related structures in the lower lithosphere beneath central Europe. This indicates a re-organization of the upper mantle after the Variscan convergence possibly related to the Permo-Carboniferous extensional processes accompanied by wide-spread volcanic activities. The observed differences in LAB depth might be more related to the more recent geodynamic processes as the Cenozoic rifting (e.g. Upper Rhine Graben – shallow LAB) and Alpine collision (Alps, Carpathians – deeper LAB). We might speculate that the slightly thicker lithosphere beneath the SW part of the Bohemian Massif represent the normal central European lithosphere before the onset of the Upper Cretaceous–Cenozoic extension instead of being an old lithospheric root from the Variscan orogeny.
The biggest question remains for station on the East European craton. Normally, the thickness of the old craton is assumed as >200 km. We observe strong and coherent phases at 10–14 s lead time (∼90–125 km depth), which are too late to be converted from the deep LAB. Assuming that these phases are primary converted phases they might stem from intralithospheric boundaries. Indications for such discontinuities were previously found in the NW part of the East European craton (see Artemieva et al. 2006 and references therein). It is assumed that the cratonic lithospheric upper mantle consists of 2 or even three layers of depleted and undepleted composition having different seismic velocities (but still higher than in IASP91 reference Earth model; Kennett 1991). The difference in seismic velocities between both types of lithospheric mantle might be sufficient to cause conversions of teleseismic phases. The conversions might be also related to changes in seismic anisotropy; however, these interpretations would need further investigations.
Using S receiver function method a clear converted positive phases could be observed, which stem from the Moho. A negative signal in front of the Moho phase can most probably be interpreted as the conversion at the LAB. Notable variations of the arrival times of LAB phase suggest important differences in the lithospheric thickness beneath the whole study area. The lithosphere is thinnest beneath the Upper Rhine Graben region, a Cenozoic extensional structure. In most parts of the central Germany, the NE Bohemian Massif, the W Carpathians and the Pannonian Basin we observe a lithospheric thickness of about 100 km. In northern Central Europe the LAB might be as deep as 120 km, similar to the SW Bohemian Massif. Beneath the EEC no clear Sp conversion from the base of lithosphere could be observed. Weak phases indicate a depth of LAB at approximately 200 km. The results of lithospheric thickness obtained from this study mostly support previous assumptions on lithospheric thickness derived from surface wave studies, magneto-telluric investigations and the analyses of P residuals.
The LAB seems to be not as sharp as it was found in other tectonic settings worldwide. The mostly and sometimes missing weak Sp phases in Central and Eastern Europe might be caused by complex lithospheric structure, a broader lithosphere–asthenosphere transition zone thickness, or topography at the LAB. Maybe even multiple transitions (velocity gradient zones) might exist. Only in case of uniform lithospheric structure and flat-lying LAB there might be observations of strong Sp phases. Multiple Sp phases from the deep lithosphere might indicate petrological or tectonical layering.
We thank our colleagues from the SZGRF Erlangen, the GRSN, GEOFON, IG CAS Prague, ReNaSS, ZAMG Vienna, PAN Warsaw, Bratislava, Hungary, Zurich and MedNet for providing data. Most data were achieved at SZGRF, GEOFON and IRIS data centres. This research was supported by the Deutsche Forschungsgemeinschaft and the EU project NERIES. We used the software packages SeismicHandler (Stammler 1993) and GMT (Wessels & Smith 1998) for most of the data processing and plotting.