The geometry of the Wadati Benioff zone and lithospheric kinematics in the Hellenic arc

Transcription

1 Tectonophysics 319 (2000) The geometry of the Wadati Benioff zone and lithospheric kinematics in the Hellenic arc B.C. Papazachos a,*, V.G. Karakostas a, C.B. Papazachos b, E.M. Scordilis a a Geophysical Laboratory, University of Thessaloniki, P.O. Box 352-1, GR Thessaloniki, Greece b Institute of Engineering Seismology and Earthquake Engineering (ITSAK), P.O. Box 53, GR Finikas, Thessaloniki, Greece Abstract Accurate locations of 961 shallow and intermediate depth earthquakes, which occurred between 1956 and 1995 in the Hellenic arc (34 39 N, E), are used to define the plate boundaries in this area. Reliable fault plane solutions for 77 shallow and intermediate depth earthquakes are also used in order to define the interaction between the different plates in the arc. An ocean continent type of interaction occurs on a curved surface, which is defined by the shallow branch ( km) of the Wadati Benioff zone. The intersection of this zone with the earth s surface is a curve which follows the convex (outer) side of the sedimentary arc (western Peloponnese west of Cythera south coast of Crete east coast of Rhodes) and dips at low angle (~30 ) to the Aegean sea. Coupling between the subducted oceanic crust and the overriding of the Aegean lithospheric plate takes place on this surface. The deep branch ( km) of the Wadati Benioff zone is dipping freely (without coupling) at a high angle (~45 ) beneath the south Aegean trough and the volcanic arc. The high shallow seismicity (h 20 km) which is observed in the southwestern convex (outer) side of the arc (Ionian sea) is attributed to the fast southwestward motion of the Aegean plate. Location of strong deep earthquakes (h>100 km) in the fore-arc area of the southwestern part of the Hellenic arc (west of Cythera) indicates that oceanic crust is destroyed in this part of the Hellenic trench due to roll-back of the descending lithospheric slab Elsevier Science B.V. All rights reserved. Keywords: subduction; Wadati Benioff zone; dip-parallel tension 1. Introduction vergence and subduction, like the Hellenic arc, is the accurate determination of the geometry of the Knowledge of the spatial distribution of earththe Wadati Benioff zones, because such zones define quake foci in an area of plate interaction is of boundaries of the subducting slabs. importance because the distribution of earthquake The identification of the Wadati Benioff seismic foci defines the boundaries of the lithospheric zone in the Hellenic arc was first accomplished by plates that interact. If reliable fault plane solutions accurate location of intermediate depth earth- are also available for this area, the patterns of quakes (60 km h 180 km) using the difference lithospheric motion can also be determined. An in arrival time at the seismological station of important problem for areas of lithospheric con- Athens of the P and PcP waves (Papazachos and Comninakis, 1970, 1971). This was the first important step for understanding the plate motion in * Corresponding author. Fax: this area because the existence of this seismic zone address: costas@itsak.gr (B.C. Papazachos) has been the strongest evidence for subduction of /00/$ - see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S ( 99 )

2 276 B.C. Papazachos et al. / Tectonophysics 319 (2000) Fig. 1. Topographic features of tectonic origin in the Aegean area (Papazachos and Papazachou, 1997). international network of permanent seismological stations which concern locations of earthquakes in the Hellenic arc, as well as data of dense portable networks of stations which have been in operation for several months in this area, are used to define as accurately as possible the spatial distribution of earthquakes in the Hellenic arc and particularly in its fore-arc area. In addition, reliable fault plane solutions of shallow and intermediate depth earthquakes are used to further investigate the lith- ospheric kinematics in this area. Fig. 1 shows a map of the Aegean and surround- ing area. The dominant tectonic feature of the area is the Hellenic arc. The Hellenic trench, the the eastern Mediterranean lithosphere (front part of the African plate) under the Aegean lithosphere (front part of the Eurasian plate). Although several attempts have been made to further investigate the Wadati Benioff zone in southern Aegean by the use of additional data (Comninakis and Papazachos, 1980; Papazachos, 1990), no detailed knowledge exists on the spatial distribution of the shallow and intermediate depth earthquakes beneath the sedimentary part of the Hellenic arc and in its convex (outer) part. This is due to the fact that no seismological stations of the regional network exist in the fore-arc area. For this reason the most reliable data of the

3 B.C. Papazachos et al. / Tectonophysics 319 (2000) sedimentary part of the Hellenic arc ( Hellenides instruments during four local seismicity monitormountain chain) and the volcanic part of the arc ing experiments in the southern Aegean area are also shown in this map. between 1986 and 1995 ( Hatzfeld et al., 1989; Hatzfeld et al., personal communication; Scordilis et al., 1995) were processed here using the 2. Data used HYPO71 computer program (Lee and Lahr, 1975). The configuration of the network of the Three different data sets were used in the present temporary stations, parts of which were occupied work to accurately define the Wadati Benioff zone during the previous experiments, is shown in Fig. 2. of the Hellenic arc (Appendix A). The first one Although not all stations were occupied simulta- consists of earthquakes recorded at local seismodifferent experiments. As a result, the aperture of neously, several stations were common between logical networks which were installed and operated in the southern Aegean area in different time each occupation was large enough to determine periods between 1986 and The second set the location of shallow and deep events with high includes earthquakes with focal depths h>100 km accuracy. In any case, events with large azimuthal as located by the International Seismological gaps and large errors in hypocentral parameters Center. The third set of data includes strong were not used, as will be described later in detail. earthquakes (M 5.5) for which reliable fault A velocity model of five layers above a half plane solutions have also been determined space (Table 1) was used to locate these earth- (Table 2). quakes. This model was based on the tomographic Data collected with both analog and digital results of a model proposed by Papazachos and Fig. 2. Configuration of the temporary stations installed in the southern Aegean area during several experiments (1986, 1988, 1993 and 1995).

4 278 B.C. Papazachos et al. / Tectonophysics 319 (2000) Table 1 that the focal coordinates of the earthquakes Velocity model used for the event relocation in the south located by the local network data are of high Aegean area accuracy and, therefore, can be used for the P velocity (km/s) Layer thickness (km) purposes of the present study. Despite the great number of earthquakes recorded by the local networks, the number of shocks with intermediate focal depths was quite small. This is probably due to the limited time span of recording, which is less than one year However, it has been observed (Comninakis and Papazachos, 1980) that the ISC data of the earthquakes of intermediate focal depth are accurate if Nolet (1997) for the Aegean area. A velocity ratio they have been estimated by the arrival of 26 or of V /V =1.78 was adopted from the same model more stations. In order to increase the number of p s for our calculations. earthquakes of intermediate focal depth, all these The location of each earthquake was derived shocks (73) which occurred in the studied area after several runs of the HYPO71 program, in between 1964 and 1995, with ISC focal depths order to reject the phases with the bigger errors. h>100 km, were included in the data set. The iterations in each run were not started from Due to the small number of available earththe coordinates of the closest station (as is standard quakes with h>100 km, only one earthquake was for HYPO71) but from the epicenter of the previous found to be located by both the local network location. Moreover, different initial depths data, as well as the ISC using the permanent were tried. The majority of time errors of the network data. This earthquake ( 1986 July 15, phases of the earthquakes located in this way are 15:15) is included twice in the catalog of Appendix less than 1.0 s, and only a small number of phases A, where the ISC location is denoted by an aster- have errors between 1.0 and 2.0 s. Since we did isk. The horizontal distance between the two locations not want our results to be contaminated by mislocations, is 12 km, while the depth gap is only 8 km. only data for events with at least 20 P and Although this is the only earthquake for which a S observations were used. comparison was possible, the small bias between The RMS error of the travel times obtained the determined hypocenters confirms that the previously from the location process was smaller than 1.2 s estimated location errors ( ERH and ERZ) in all cases, with a mean value of 0.43±0.17 s. have realistic values. Moreover, it suggests that no The corresponding epicentral error ( ERH ) varied systematic difference should be expected for deep between 0 and 10 km. Similar errors ( ERZ) were (h>100 km) events between the ISC and local determined for the hypocentral depth of located network determinations, although more data are events. In some cases both ERH and ERZ reached necessary in order to reach a final conclusion. values up to 20 km and in very few cases the Another data source is strong earthquakes, corresponding errors obtained even larger values. because their focal coordinates have been estimated In order to avoid the effect of mislocations, only on the basis of a great number of arrivals. events with both errors (horizontal and depth) less The accuracy of these data is also controlled by than 20 km were used. This resulted in a final data the available macroseismic observations and the set of 811 well-located earthquakes recorded by fault plane solutions. In the present work the data the local networks. The mean epicentral error for of 77 earthquakes with magnitude M 5.5, which this data set was ERH=2.5±2.5 km, while the occurred since 1956 in the studied area and for average depth error was estimated to be which reliable fault plane solutions have been ERZ=2.7±2.4 km. Although these quantities determined by NEIS, Harvard (CMT solutions) have a mathematical sense and do not necessarily or individual studies, were also selected. correspond to the true error estimates, they show Information on the fault plane solutions of these

5 B.C. Papazachos et al. / Tectonophysics 319 (2000) earthquakes (with their corresponding references) (denoted by solid triangles) can be recognized at is given in Table 2. distances between 350 and 400 km in the corre- sponding cross-section (lower part of Fig. 4). Comparison with the upper part of Fig. 4 shows 3. Spatial distribution of the earthquake foci that the epicenters of these earthquakes are located south of Peloponnese and west/southwest of Figs. 3, 4 and 5 show the epicenters of the Cythera up to the Hellenic trench, which at this shallow and intermediate depth earthquakes point reaches a maximum depth of ~5 km. The (upper parts) and the corresponding projections of presence of these deep events at this area will be their foci on vertical planes normal to the arc (low discussed later. parts) for the western, central and eastern section Distribution of earthquake foci in the eastern of the Hellenic arc, respectively. Circles show part of the arc is shown in Fig. 5. Shallow seismi- shallow earthquakes (h 50 km), while shaded and city is also limited to the upper 20 km of the crust. black triangles show intermediate depth earth- The shallow part of the Wadati Benioff zone quakes with focal depths between 50 and 100 km (20 km h<100 km), however, does not show a and between 100 and 180 km, respectively. clear dipping from the fore-arc to the back-arc Symbols of different sizes are used to denote area (from point B to C in Fig. 5) in this part of earthquakes of different magnitudes. The horithe arc there is a gap in the seismicity between the the arc. On the other hand, in this eastern part of zontal line at a depth of 20 km shows the lower boundary of the shallow seismogenic layer, while deeper and the shallower segment of the Wadati Benioff zone, that is, at a depth of km, dashed lines show the dipping of the Wadati which is also partly observed in the central part Benioff zone. (Fig. 4) but is not expressed in the western part Fig. 3 indicates that in the western part of the (Fig. 3). It should be noted that this gap cannot arc the thickness of the shallow seismogenic layer be considered as an artifact of the location process covers the upper 20 km of the crust in the backsince the velocity model used ( Table 1) exhibits no arc area (Aegean sea) as well as in the fore-arc change in this depth range (strong velocity gradient part of the area ( Ionian sea). The Wadati Benioff or discontinuity) that would tend to force the zone starts to dip in western Peloponnese (point locations out of this depth range. Moreover, if B in Fig. 3) where the shallower seismogenic layer such a case occurred, it should be present in all becomes thicker. Up to a depth of about 100 km cross-sections (Figs. 3, 4 and 5) and not only in (point C in Fig. 3) this zone dips at a relatively the eastern part of the arc. low angle (~30 ) and after that the dip angle of Based on the accurate data used in the present the zone becomes larger (~45 ) up to its maximum study (Appendix A), isodepths of 20, 100 and focal depth (~180 km). In a large part of the fore- 170 km for the earthquakes in the Hellenic arc arc area (between western Peloponnese and Ionian have been drawn in Fig. 6. In the same figure the islands) no strong earthquakes with h>20 km (sec- epicenters of four complete samples of strong tion AB in Fig. 3) occur. intermediate depth earthquakes occurring during A similar distribution of the earthquake foci is the last two centuries (Papazachos and observed in the central part of the arc ( Fig. 4) Papazachou, 1997) are plotted. It is observed that where dipping of the Wadati Benioff zone starts the isodepths fit quite well with the distribution of ( point B in Fig. 4) in the convex side of the arc, the earthquakes, although these isodepths have under the Hellenic trench. An interesting difference been drawn by the use of a different sample of data. between the distribution of the earthquake foci in this part of the arc and the other parts is that some strong intermediate depth earthquakes with 4. Mechanisms of earthquakes in the Hellenic arc relatively large focal depths (h>100 km) occur in the convex (outer) side of the arc, beneath or very The first seismological evidence for the existence close to the Hellenic trench ( Fig. 1). These events of thrust faults in the Mediterranean lithosphere

6 280 B.C. Papazachos et al. / Tectonophysics 319 (2000) Table 2 Information on the fault plane solutions of strong earthquakes (M 5.5) which occurred in the studied area since 1956 Date Time Latitude Longitude h M Plane A Plane B P axis T axis Reference Azimuth Dip Rake Azimuth Dip Rake Azimuth Dip Azimuth Dip a b c b c d c d e f d c f g c h c c c c b i j k b i f i l m l b g m n k l l o p p p q r,s m b,r k q t b t

7 B.C. Papazachos et al. / Tectonophysics 319 (2000) Table 2 (continued) Date Time Latitude Longitude h M Plane A Plane B P axis T axis Reference Azimuth Dip Rake Azimuth Dip Rake Azimuth Dip Azimuth Dip q q t u q q q q q q q q q q q v t q t q q q q t q q a Shirokova (1972). b Papadimitriou (1993). c McKenzie (1972). d Papazachos et al. (1992). e Lyon-Caen et al. (1988). f Baker et al. (1997). g Anderson and Jackson (1987). h Ritsema (1974). i McKenzie (1978). j Kiratzi and Langston (1989). k Karakostas (1988). l Papazachos (1975). m Taymaz et al. (1990). n Papazachos et al. (1991). o Ekstrom and England (1989). p Taymaz et al. (1991). q Harvard (CMT solutions). r Scordilis et al. (1985). s Kiratzi and Langston (1991). t NEIS determination. u Papazachos et al. (1988). v Karakaisis et al. (1993)

8 282 B.C. Papazachos et al. / Tectonophysics 319 (2000) Fig. 3. Epicenters of shallow and intermediate depth earthquakes in the western part of the Hellenic arc (upper part) and crosssection of the earthquake foci along the line ABCD (lower part).

9 B.C. Papazachos et al. / Tectonophysics 319 (2000) Fig. 4. Epicenters of shallow and intermediate depth earthquakes in the central part of the Hellenic arc (upper part) and cross-section of the earthquake foci along the line ABCD (lower part).

10 284 B.C. Papazachos et al. / Tectonophysics 319 (2000) Fig. 5. Epicenters of shallow and intermediate depth earthquakes in the eastern part of the Hellenic arc (upper part) and cross-section of the earthquake foci along the line BCD (lower part).

11 B.C. Papazachos et al. / Tectonophysics 319 (2000) Fig. 6. Isodepths of 20, 100 and 170 km for the earthquakes located in the Wadati Benioff zone of southern Aegean and epicenters of four complete samples of intermediate depth strong earthquakes occurring during the last two centuries. under the Hellenic trench was presented about the arc, respectively. Different symbols are used to three decades ago (Papazachos and Delibasis, denote different fault types. 1969). Since then fault plane solutions for several Normal faults and some strike slip ones are shallow and intermediate depth earthquakes in the shown in the inner part of the shallow crustal Hellenic arc have been published (McKenzie, 1970, seismogenic layer, which is due to the known 1972, 1978; Taymaz et al., 1990; Papazachos et al., expansion of the Aegean and the strike slip faults 1991 among others). in the Cephalonia area which is the contact A considerable number of reliable fault plane between the Aegean and the Adriatic microplates. solutions are now available for strong shallow and Thrust faults dominate in the outer part of this intermediate depth earthquakes in the Hellenic arc layer ( Hellenic trench) due to the convergence ( Table 2). Fig. 7 shows plots of the foci of these between the Aegean and the eastern Mediterranean earthquakes on three vertical planes normal to the lithospheres. In the descending lithospheric slab trend of the western, central and eastern part of strike slip faulting with thrust component occurs

12 286 B.C. Papazachos et al. / Tectonophysics 319 (2000) Fig. 7. Plot of the foci on the three cross-sections for earthquakes of the Hellenic arc for which reliable fault plane solutions are available. The type of faulting for each earthquake is also indicated.

13 B.C. Papazachos et al. / Tectonophysics 319 (2000) due to the fact that maximum tension trends crust, which is suggested by the data used in the parallel to the dip of the Wadati Benioff zone and present study, has also been observed during the maximum compression is almost horizontal and investigation of the spatial distribution of parallel to the arc ( Kiratzi and Papazachos, 1995; aftershocks of several strong earthquakes which Papazachos, 1996). were generated during the last two decades by It is of interest to observe that the earthquakes normal or strike slip faulting (Papazachos et al., with focal depths between 20 and 50 km which 1979, 1983, 1988; Scordilis et al., 1985). No such occur in the convex (outer) side of the sedimentary investigation of aftershocks of strong main shocks part of the arc are generated by thrust faulting, as generated by thrust faulting in the crust of the it occurs with the shallower earthquakes (0 eastern Mediterranean has been made. Therefore, 20 km), although they form part (the most shal- the result that the shallow seismogenic layer is also low) of the Wadati Benioff zone where the deeper 20 km in the fore-arc part of the Hellenic arc earthquakes (h 60 km) have a different focal indicates that it is a physical property of the crust mechanism. We can, therefore, separate each earth- to generate strong earthquakes only in its shallow quake according to its faulting type and define as part, independent of the type of deformation ( horishallow earthquakes those with h 50 km and as zontal expansion, horizontal shortening, etc.). This intermediate depth earthquakes those with result probably leads to the conclusion that accuh 60 km in the Hellenic arc. rately located foci of earthquakes at depths larger This can be more clearly seen in Fig. 8, where than 20 km indicate a dipping of the crust. the stress axes of the focal mechanisms of Table 2 The Wadati Benioff zone starts at a depth of are presented for the same cross-sections as in 20 km under the convex (outer) side of the sedi- Fig. 7. In this figure the projection of each stress mentary part of the arc (western Peloponnese axis is plotted for each cross-section, hence in west of Cythera south of Crete southeast of several cases the T (tension gray arrows) and Rhodes) and dips towards the back-arc area where P (compression solid arrows) axes are oblique it reaches a depth of 150 km under the volcanic and not normal. If the angle between the stress arc in the southern Aegean. Some earthquakes are axis and the cross-section exceeded 45, then the located even deeper, up to a focal depth of about stress axis is not shown. In all cross-sections we 180 km. This Wadati Benioff zone can be sepaclearly recognize a transition from thrust faulting (horizontal P vertical T ) in the fore-arc area to rated into two branches, one shallow (20 km h normal faulting (horizontal T vertical P) in the 100 km) with a dip angle of about 30 and the back arc area in the shallow part of the crust other deep ( 100 km h 180 km) with a dip angle (20 km). The orientation of the P axis remains of about 45. Of course, the existing scattering of characteristically constant (subhorizontal, dipping the intermediate depth seismicity does not allow towards the fore-arc area) even at larger depths the determination of these dip angles with high along the shallow part of the subduction, up to a accuracy ( less than 5 10 ). On the other hand, depth of 50 km or more in the eastern part of the the relatively small hypocentral errors previously slab. At larger depths ( km), the P axis is described, as well as the systematic increase of the not shown, as it is oriented horizontally and paral- dip angle in all three presented cross-sections lel to the strike of the arc. This section of the (Figs. 3, 4 and 5) suggest that this is a robust subduction is dominated by a strong extension, feature of the subducted slab. which more or less follows the dipping direction The observed slope change for the Hellenic of the slab in each cross-section. subduction is in excellent agreement with indepen- dent results in the area. Detailed tomographic results for the area (Papazachos and Nolet, 1997) 5. Conclusions and discussion along similar cross-sections confirm this change of dipping, although they indicate that it starts at a The confinement of the foci of the shallow shallower depth (~80 km). The same results also earthquakes to the upper 20 km of the Aegean suggest that this slope change for the central part

14 288 B.C. Papazachos et al. / Tectonophysics 319 (2000) Fig. 8. Stress field along the same cross-sections presented in Fig. 7. Notice that the extension is parallel to the dipping of the Wadati Benioff zone in all cross-sections (depth >50 60 km).

15 B.C. Papazachos et al. / Tectonophysics 319 (2000) of the subduction is quite sharp, as also seen in to the back-arc part of the arc and overriding of Fig. 4. Similar changes in the slope of the subduct- the Aegean lithosphere from northeast to southing slab at depths of km have been west contribute to this coupling. Subduction preobserved in other subduction areas, such as the vails at the low part of this branch ( km) Japan island arc ( Wesnousky et al., 1982) or the since the maximum tension trends parallel to the Aleutian subduction zone ( Ekstrom and Engdahl, dip of the zone at these depths ( Kiratzi and 1989). Moreover, theoretical modeling (e.g. Tao Papazachos, 1995). This dip-parallel extension is and O Connell, 1992) also favors the shape of also present at larger depths ( km), as seen kinked slab profiles, especially in oceanic conti- in Fig. 8. nental subduction systems, such the Mediterranean This stress pattern is very consistent with the subduction beneath the Aegean. stress pattern seen in other subduction areas. The shallow branch of the Wadati Benioff zone Subduction parallel tension has been well docuis not well developed, particularly in the eastern mented for similar depth ranges ( km) in part where its dipping from the fore-arc to the other subduction areas, e.g. Philippines (Cardwell back-arc part of the arc is doubtful. This gap in et al., 1980) and used in theoretical modeling (e.g. the slab observed in its eastern section is further Tao and O Connell, 1992). For the Central supported by recent regional tomographic results Aleutians, Ekstrom and Engdahl (1989) find that ( Papazachos and Nolet, 1997), which clearly show the thrust faults with subhorizontal P axes (also a discontinuity in the slab between 75 and 95 km, dipping towards the fore-arc side) remain practialmost exactly at the same place where no intermecally constant along the main thrust zone at depths diate depth earthquakes are detected. This between 15 and 50 km, similar to what is found in agreement between the seismicity distribution and the Hellenic arc. According to the same authors, tomography suggests that this slab tear-up is a down-dip tension dominates at large depths robust feature of the eastern section of the Hellenic (>150 km), although they confirmed the presence arc. Small earthquakes with h>20 km occur even of down-dip compression at depths between 70 in the area between the shallow seismogenic layer and 120 km (Frohlich et al., 1982). ( 0 20 km) of the southern Aegean and this branch The orientation of the compression parallel to of the Wadati Benioff zone. Strong earthquakes the strike of the arc (and the subduction) seen in in the shallow part (20 55 km) of this branch of the Wadati Benioff zone are also caused by horiin the previously mentioned or other subduction the Hellenic arc at depths >55 km is not identified zontal compression (thrust faulting) which trends in a northeast southwest direction while deeper areas. For example, detailed study of the seismicity strong earthquakes ( km) are caused by of the subducted Nazca plate (Cahill and Isacks, tension trending parallel to the dip of the Wadati 1992) confirms that the T axis for intermediate Benioff zone (strike slip faulting with a thrust and deep earthquakes (up to 350 km) has also a component). more or less down-dip orientation along All these observations indicate that this shallow ~3000 km of subduction (10 35 S). However, the branch (20 km h 100 km) of the Wadati P axis was found to be oriented almost vertically Benioff zone defines the area of coupling between or normal to the subducted plate, similarly to the the Aegean and the Mediterranean lithospheres. deeper (>150 km) Aleutian events. The origin of This is strongly supported by the fact that all big the in-plane horizontal compression at depths intermediate depth earthquakes with magnitudes >55 km in the Hellenic arc is not clear, and further up to 8.2 (see Fig. 6) occur in this branch. It is modeling is necessary in order to explain such also supported by the existence of the aseismic special behavior. layer ( km), at least in the eastern arc, The shallow earthquakes ( 0 55 km) in the outer which is probably an area of low rigidity on which (convex) side of the sedimentary arc are mainly lithospheric plates slip easily. It seems that both due to the overriding of Mediterranean by the subduction of oceanic lithosphere (front part of Aegean lithospheric plate which moves fast south- the Mediterranean lithosphere) from the fore-arc westwards, as shown by geodetic data (Smith et al.,

16 290 B.C. Papazachos et al. / Tectonophysics 319 (2000) ; Kastens et al., 1996). Such southwestward overriding of the Aegean plate explains the fact Appendix A that no clear dipping of the shallow branch of the Wadati Benioff zone is observed in the concave Table 3 (inner) part of the eastern section of the arc. It Focal parameters of the earthquakes used in the present study. also explains the observation that no earthquakes The earthquake of the 15th July 1986, 15:15 is included twice with h>20 km occur in a large area of the southand ISC (ISC solution denoted by 1) for comparison of the solutions obtained by the local network westernmost part of the arc (sections AB in Figs. 3 and 4) where shallow seismicity is very high, while this is not valid for the eastern part of the arc (no Date Origin time Latitude Longitude h M AB seismic section exists in Fig. 5). The Aegean :11: :08: plate, due to its fast southwestward motion, has :45: surpassed the subducted lithospheric slab and :59: reached the Ionian sea where the boundary :34: between this plate and the African plate is located :12: today. Therefore, the interaction between the :57: Aegean and the African plate in this western :09: :18: boundary is of continent continent type now :26: The deep branch (100 km h 180 km) of the :42: Wadati Benioff zone in the Hellenic arc is sepa :39: rated from its shallow part by an aseismic layer :40: ( km), at least in the eastern part where the :47: presence of a slab discontinuity is suggested at :52: :12: these depths. This branch of the zone ( :21: km) is probably due to the free sinking of an :13: oceanic lithospheric slab, as also indicated by the :09: dip-parallel extension axis presented in Fig. 8. Such :50: a sinking, without the presence of strong coupling, :39: :13: explains the fact that no earthquakes with magni :07: tudes larger than about 7.0 occur in this deep :49: branch of the Wadati Benioff zone. This sinking :57: occurs, mainly, under the southern Aegean sea :16: and particularly under the volcanic arc :44: There are, however, several well located strong :45: :59: intermediate depth earthquakes in the southwest :59: ern part of the sedimentary arc (in the Ionian sea, :27: west of Cythera) with focal depths larger than :19: km (see Fig. 4). This indicates that the oceanic :59: lithosphere of the Eastern Mediterranean is sink :59: ing in this part of the Ionian sea. This sinking :34: :41: of the oceanic lithosphere in this part of the :53: subduction is also indicated by recent tomographic :35: information (Papazachos and Nolet, 1997). These :58: results support the idea that roll-back of the :39: descending lithospheric plate towards the remain :19: ing scarp of oceanic crust beneath the Ionian sea :20: :39: may cause gravitational spreading or gravitational :18: collapse of the expanding Aegean lithosphere :37: ( LePichon and Angelier, 1981; Dewey, 1988) :18:

17 B.C. Papazachos et al. / Tectonophysics 319 (2000) Date Origin time Latitude Longitude h M Date Origin time Latitude Longitude h M :48: :14: :17: :34: :22: :35: :41: :55: :12: :48: :41: :15: :28: :25: :41: :26: :15: :20: :58: :49: :59: :07: :59: :55: :14: :51: :04: :21: :02: :25: :48: :47: :57: :09: :29: :47: :43: :54: :19: :14: :11: :51: :41: :56: :08: :38: :40: :26: :40: :47: :45: :32: :26: :31: :49: :45: :46: :54: :09: :00: :32: :44: :20: :49: :33: :04: :39: :48: :12: :33: :24: :21: :42: :23: :50: :11: :54: :24: :12: :25: :58: :50: :25: :34: :40: :21: :49: :43: :42: :47: :21: :20: :07: :27: :07: :35: :42: :38: :43: :12: :44: :30: :57: :50: :18: :52: :52: :05:

18 292 B.C. Papazachos et al. / Tectonophysics 319 (2000) Date Origin time Latitude Longitude h M Date Origin time Latitude Longitude h M :18: :48: :45: :18: :16: :48: :52: :31: :52: :47: :45: :40: :48: :59: :19: :34: :41: :00: :58: :46: :44: :55: :51: :31: :20: :39: :39: :04: :07: :38: :53: :28: :18: :32: :21: :14: :46: :40: :06: :33: :46: :03: :55: :28: :14: :57: :19: :59: :52: :37: :04: :33: :42: :03: :37: :57: :39: :32: :45: :05: :51: :20: :24: :40: :40: :32: :55: :49: :25: :06: :06: :19: :54: :34: :30: :55: :14: :06: :07: :52: :45: :09: :45: :24: :52: :29: :52: :24: :54: :29: :36: :41: :44: :08: :14: :15: :21: :28: :20: :54: :32: :23: :10: :23: :20: :33: :28: :57:

19 B.C. Papazachos et al. / Tectonophysics 319 (2000) Date Origin time Latitude Longitude h M Date Origin time Latitude Longitude h M :30: :02: :57: :59: :16: :29: :34: :31: :47: :17: :24: :32: :14: :18: :43: :32: :54: :39: :13: :54: :14: :44: :48: :55: :17: :54: :47: :58: :57: :52: :20: :29: :41: :25: :35: :44: :16: :00: :22: :51: :42: :52: :15: :34: :05: :34: :30: :09: :56: :54: :31: :33: :07: :00: :12: :33: :15: :03: :15: :50: :04: :25: :42: :19: :10: :39: :09: :02: :50: :28: :28: :19: :40: :26: :25: :20: :05: :26: :24: :22: :10: :11: :34: :36: :56: :45: :40: :09: :28: :47: :50: :52: :28: :27: :45: :29: :55: :25: :20: :56: :17: :40: :34: :21: :03: :00: :17: :55:

20 294 B.C. Papazachos et al. / Tectonophysics 319 (2000) Date Origin time Latitude Longitude h M Date Origin time Latitude Longitude h M :58: :28: :51: :40: :30: :39: :46: :29: :39: :30: :48: :22: :08: :27: :48: :28: :46: :28: :30: :37: :16: :22: :33: :08: :30: :49: :09: :02: :12: :10: :09: :39: :35: :10: :18: :58: :26: :07: :00: :07: :07: :01: :50: :13: :06: :00: :43: :10: :40: :25: :46: :07: :02: :17: :21: :25: :07: :52: :53: :04: :04: :00: :25: :16: :37: :29: :22: :30: :01: :32: :35: :30: :00: :23: :58: :25: :49: :55: :39: :47: :23: :05: :07: :14: :27: :40: :28: :50: :09: :07: :48: :51: :10: :34: :06: :21: :04: :30: :08: :06: :36: :32: :39: :13: :19: :14: :48: :59: