Stuart Crampin and David C. Booth British Geoiogica/Survey, Murchison House, West Mains Road, Edinburgh EH93LA, Scotland

Transcription

1 Geophys. J. R. astr. SOC. (1986) 84, Shear-wave polarizations in the Peter the First Range indicating crack-induced anisotropy in a thrust-fault regime Stuart Crampin and David C. Booth British Geoiogica/Survey, Murchison House, West Mains Road, Edinburgh EH93LA, Scotland Maria A. Krasnova and Evgenie M. Chesnokov Instituteof Physics of the Earth, Bolshaya Grouzinskaya 10, Moscow , USSR Alexandr B. Maximov and Nikolai T. Tarasov Instituteof Physics of the Earth, Complex Seismological Expedition, Garrn, Tadzhikistan, USSR Accepted 1985 July 10. Received 1985 July 10; in original form 1985 April 24 Summary. Three-component seismograms of small local earthquakes recorded in the Peter the First Range of mountains near Garm, Tadzhikistan SSR, display shear-wave splitting similar to that previously observed near the North Anatolian Fault in Turkey. The Peter the First Range is in a region of compressional tectonics, whereas the North Anatolian Fault is a comparatively simple strike-slip fault. Detailed analysis of the Turkish records suggests that the splitting is diagnostic of crack-induced anisotropy caused by vertical microcracks aligned parallel to the direction of maximum compression. Preliminary examination of paper records from Garm shows that most shear waves arriving within the shear-wave window display shear-wave splitting, and that the polarizations of leading shear-waves are consistently aligned in a NE/SW direction. The area is complicated and the tectonics are not wellunderstood, but the NE/SW direction is approximately perpendicular to the compressional axis in many of the fault-plane mechanisms of the earthquakes. These earthquakes are usually at depths between 5 and 12 km, although there are some deeper events nearby. Parallel shear-wave polarizations, such as those observed, are expected to indicate the strike of nearly vertical parallel microcracks, which would be aligned parallel to the direction of maximum compression. Thus the shearwave polarizations in the Peter the First Range indicate that the directions of principal stress are reversed in the rock above the earthquake foci where thrust faulting is taking place. Key words: extensive dilitancy anisotropy, shear-wave splitting, thrust-fault regime, Peter the First Range 1 Introduction The Turkish Dilatancy Projects (TDP) recorded shear waves from small earthquakes near the North Anatolian Fault (NAF) in Turkey with closely-spaced networks of three-component

2 402 S. Crampin et al. seismometers. The shear waves displayed shear-wave splitting diagnostic of some form of effective anisotropy, and the horizontal polarizations (particle displacements) of the leading shear waves were parallel at all stations for all shear waves arriving within the shear-wave window (see the series of papers introduced by Crampin, Evans & fiser 1985). Six largely independent observations from these recordings (Crampin & Booth 1985) can be consistently interpreted in terms of crack-induced anisotropy resulting from vertical cracks (probably water-filled microcracks) aligned parallel to the direction of maximum compressive stress indicated by earthquake fault-plane mechanisms. The NAF is, however, a comparatively simple strike-slip feature, and similar phenomena may not occur in other types of tectonic activity. In order to examine these phenomena in different tectonic regimes, a Protocol has been signed by the Institute of Physics of the Earth, Moscow, and the British Geological Survey, Edinburgh, under the Scientific Exchange Agreement between the Academy of Sciences of the USSR and the Royal Society, London. The Protocol enables the Institute of Physics of the Earth and the British Geological Survey to collaborate on investigating shear-wave splitting phenomena within the USSR, particularly at the Complex Seismological Expedition (CSE) at Garm, Tadzhikistan SSR. This paper is the first product of this collaboration. The area surrounding the Peter the First Range of mountains has (since 1954) been exceptionally well instrumented for earthquake prediction research by CSE Garm. The instrumentation has included at various times a telemetered seismic network over the main centre of seismic activity beneath the range (Fig. 1). This paper reports a preliminary examination of paper records of earthquakes beneath this telemetered network. 2 Tectonic setting The Peter the First Range (PFR) is a mountain chain, about 30 km wide, forming the narrow eastern extension of the Tadzhik Depression, where Mesozoic and Cenozoic sediments have Figure 1. Topography of the Peter the First Range, Tadzhikistan, and the sites ot the GTSN seismic network radiolinked to CSE Garm. Contours are marked in metres.

3 Shear waves in the Peter the First Range 403 been compressed between the Pamir mountains and the more stable Tien Shan mountain block as the Pamirs advance northwards in front of the Indian subcontinent. Keith, Simpson & Soboleva (1 982) describe the general tectonic setting of the region. The area of the PFR is of high seismicity with many magnitude Ms > 5 earthquakes. The largely unconsolidated sediments of PFR display intense deformation and steep irregular topography rising 3000 m above the Surkhob and Obikhingou river valleys bounding the PFR. The highest points, where the telemetered seismic network is located, are usually associated with steep northward-facing escarpments sometimes nearly 1000 m in height. Investigations at CSE Garm indicate that PFR is rising relative to the Tien Shan block at between 5 and 15 mm yr-' (Lukk et al. 1980). The mountains are undergoing rapid erosion with huge erosional gullies, and rivers opaque with sediments cutting deeply into previous flood-plains. The base of the sediments of the PFR is not well defined, but from refracted arrivals appears to be a plane of decollement at about 8 km below sea-level, where the sediments of the Tadzhik Depression have been pushed in front of the advancing Pamirs. The earthquake foci are widely scattered, but there is a concentration of activity beneath the location of the telemetered network (Fig. l), where Ms = 5 earthquakes occur every 6 or 7 years (the largest was Ms = 5.5 on 1975 July 9). The foci of this centre of activity are between 6 and 12 km deep, presumably bounding the base of the PFR sediments, although there is some deeper activity down to 25 or 30 km nearby. 3 Instrumentation of the telemetered network The Garm Telemetered Seismic Network, GTSN, was originally set up in 1975 under the cooperative USSR/USA Program in Earthquake Prediction Research (Pelton & Fischer 1980). Several modifications have since been made to the network, and the configuration when the data used here were recorded is shown in Fig. 1. Three-component SM-2 seismometers of two seconds free-period were radio linked to Garm, sometimes using repeater stations. At Garm, triggered signals were written to a 1 inch 14 track analogue library tape using two NO-52 tape decks. The response of the system is flat to displacement between 0.3 and 20 Hz, and the instruments record at two gains, with a ratio of l:lo, to increase the effective dynamic range. The seismometers are located at high elevations to provide the essential line-of-sight connection for low-powered radio transmission. Most sites are accessible only by helicopter. It should be noted that these high sites have a harsh environment and are subject to extreme conditions including low temperatures, severe electrical storms, heavy icing on antennae, and damage by bears. The early recordings from the network were written by tape decks which are no longer easily accessible. Existing analogue tapes can usually be digitized, but when the research reported here took place the digital system was temporarily out of commission between changes of system. Consequently, the analysis was confined to existing jet-pen paper records, which had been played out for other purposes at paper speeds of 2.5 cm s-' for period I, from 1979 August to 1980 April, and for period 11, from 1981 September to 1982 March. The network did not have uniform coverage throughout these periods. It was possible to replay records from period I1 and some problematical recordings were replayed with more appropriate amplifications. 4 Analysis of paper records it is necessary to restrict the analysis of shear-wave polarizations to arrivals within the shearwave window, where waveforms observed at the surface are similar to the waveforms of the

4 404 S. Crampin et al. incident waves. (The shear-wave window is the area above the source where the angles of incidence on a plane surface are less than the critical angle sin-' Vs/Vp; this is about 35", for a Poissons ratio of 0.25, Booth & Crampin 1985.) Except for incident shear waves with purely SH polarization, the waveforms of shear waves arriving with incidence angles outside the shear-wave window are severely modified by interactions with the free surface (Evans 1984). Reliable locations of local earthquakes are essential for the analysis of the polarization of shear waves recorded at the surface as only arrivals within the shear-wave window are suitable for waveform analysis. The earthquake locations used in this study are shown in Fig. 2(a) for period I, and Fig. 2(b) for period 11. The source locations are those routinely determined by CSE Garm using arrivals from the conventional seismic network surrounding D.8 1 I I 0 I I A a" oag)o" ' 00 A I - 0 M=2.0 o Q 0 M= I 0 M u kp ib) Figure 2. Epiccntres ot' earthquakes beneath the GTSN seismic network during: (a) period I, 1979 August to 1980 April; and (b) period 11, 1981 September to 1982 March. The magnitudes are equivalent Ms values determined from the local magnitude scale by CSE Garm. Triangles arc the GTSN sites, Seven faultplane mechanisms are shown for period 11 as equal-area projections of the upper focal hemispheres. All earthquake parameters are from the routine earthquake catalogues of CSE Garm.

5 Shear waves in the Peter the First Range 405 PFR, but not using arrivals from the telemetered network. Thus, although the epicentres are sufficiently accurate to give reliable estimates of the azimuth of arrivals, the absence of direct upward-travelling arrivals to recording sites immediately above the earthquakes makes the determination of focal depths and hence the angles of incidence less accurate. In addition, the effective shear-wave window is difficult to estimate because of the very irregular relief. Consequently, in order to analyse as much data as possible we have included arrivals which had incident angles up to 50". This is believed to be justified when the polarizations at the edge of the projections show a similar pattern of orientations to those well within the window. This suggests that the effective angles of incidence at the surface are less than the estimated values, either because the ray paths of the shear waves are refracted upwards through low-velocity surface layers, or because the earthquake foci are deeper than those estimated by CSE Garm. Fig. 3 shows typical paper records of (low gain) seismic traces used in this study. Preliminary examination of corresponding peaks and troughs on the horizontal components of these records showed that the horizontal polarization of the leading shear waves within the shear-wave window is usually linear. Within two or three cycles, the linear motion is followed by 180" phase changes between the horizontal components indicating a 90" change in phase of the shear-wave polarization. This behaviour is very similar to that shown by recordings of shear waves arriving within the shear-wave window from small earthquakes near the North Anatolian Fault in Turkey (Booth et al. 1985; Crampin & Booth 1985), v n n n n n n n p Figure 3. Paper records of three-component (low gain) seismograms typical of those used in this study showing shear waves arriving within the shear-wave window at sites 36 and 33. Lines join corresponding cycles of the linear displacements of the leading split shear wave. 14

6 Figure 4. Equal-area projections, out to incidence angles of 50", of the lower hemisphere beneath the numbered GTSN sites showing the horizontal polarizations of the leading split shear waves for period I, , and period 11, Histograms of the polarization directions from N 180"W to N 180"E are shown beneath each equal-area projection. The average polarization (the preferred direction) of all arrivals is marked by arrowheads beneath the histograms, and the orthogonal direction by bars.

7 Shear waves in the Peter the First Range 407 which displayed polarization diagrams typical of shear-wave splitting in effectively anisotropic media. We do not have digital recordings of the GTSN data and cannot plot polarization diagrams. However, three factors made the examination of the correlation between horizontal components on these paper records particularly easy. The frequency of many of the shear waves (4-6 Hz) was sufficiently low, relative to the paper speed of 2.5 cm s-l to make it comparatively easy to correlate peaks and troughs of signals with wavelengths on the paper of 4-6 mm. The relatively low frequency of the shear waves is accentuated by the response of the instruments being flat to displacement. Secondly, the preferred alignment displayed by the polarizations was approximately NE/SW, so that peaks of the first cycles of the shear-wave motion had approximately equal amplitude on the NS and EW instruments. Such measurements of approximately equal amplitudes on two orthogonal components permit the greatest accuracy in estimating the polarization of the resulting particle motion. Thirdly, CSE Garm is confident that the response of the horizontal seismometers agrees to within 5 per cent. Amplitudes of corresponding peaks and troughs on the horizontal components were measured from an approximate midline of the background noise level and the polarization angles estimated for each arrival from the arctangent. About 90 per cent of the arrivals within the effective shear-wave window provided what appeared to be reliable polarization angles and are plotted in Fig. 4. The 10 per cent of arrivals whose polarizations were not plotted consist of those which were clearly identifiable as the local SP-phase (Evans 1984), and those where shear-wave arrivals appeared to be absent or difficult to identify, probably due either to a node in the radiation pattern from the source or to the signal-to-noise ratio being too small (the paper records only existed for comparatively well recorded events). The duplication of records at high- and low-gain settings kept the number of paper records, where we were unable to estimate reliable polarizations, down to a comparatively low level. However, it must be noted that this technique of correlating and measuring cycles on parallel time-series cannot be entirely objective. A more objective technique is to estimate polarization directions from polarization diagrams that have been plotted with radial and transverse axes, as in Booth er al. (1985), so that any subjective bias is eliminated when making the measurements. 5 Shear-wave polarizations The equal-area projections of the polarizations of the leading shear-wave arrivals in Fig. 4 show considerable scatter, but the histograms of all the arrivals display a marked concentration about the same parallel direction N 45"E/N 225"E. This preferred direction is marked by arrowheads beneath the individual histograms. The histograms are poorly sampled and the data display considerable scatter. Some histograms show a concentration about a direction perpendicular to the preferred direction (marked by bars between the arrowheads), which may be due to the shear-wave polarization radiated from the source exciting the slower rather than the faster shear wave, so that there is no energy propagating as the leading split shear wave. Some sites also appear to have arrivals polarized parallel to the instrument axes: north-south and east-west. Such polarizations are probably caused by misreading noisy records, particularly when one horizontal component is noiser than the other so that it is difficult to make objective estimates of the polarizations. The observed polarizations are summarized in rose diagrams in Fig. 5. Excluding the cardinal directions as misreadings, the rose diagrams at four sites, 36, 37, 38 and 39, clearly display azimuthal distributions about the preferred direction. Sites 33 and 34 show some polarizations in the preferred direction but also display a number of polarizations in an 14*

8 408 S. Crampin et al I I km Figure 5. Rose diagrams of the shear-wave polarizations at the GTSN sites. Solid triangles arc drawn every 10" with areas proportional to the number of polarizations in those directions. The cardinal directions are believed to be misreadings and are drawn with open triangles. orthogonal direction. These anomalies are, we suggest, caused by the shear waves radiated from the source being parallel to the second split shear wave so that the leading split shear wave with the preferred polarization is not excited. Site 32 has too few arrivals to provide an adequate sample. 6 Interpretation The shear waves display splitting with the leading shear wave having approximately parallel polarizations similar to the parallel polarizations of shear waves near the NAF in Turkey (Booth er al. 1985). Three principal phenomena determine the polarization of a shear wave recorded on the free surface: (I) the polarization radiated from the source; (2) the interaction with the free surface and internal interfaces along the wave path; and (3) the fixed polarizations imposed by possible anisotropy along the wave path. We shall show that the observed polarization alignments cannot be explained in terms of source or receiver effects alone, but can be explained by the presence of anisotropy along the wave path. (I) The polarizations of shear waves radiated by an earthquake source are aligned in directions which are fixed by the geometry and orientation of the focal mechanism and the direction of propagation. These alignments will be essentially preserved at a receiver, if the shear waves propagate through a homogeneous isotropic structure and are recorded within the shear-wave window at the free surface. CSE fault-plane catalogues give double-couple focal mechanisms determined by P-wave first motions for seven of the earthquakes in period 11. Thus, we can compute the shear-wave polarizations which would be observed on the surface of an isotropic half-space, and compare them with the recorded polarizations. The seven fault-plane solutions are displayed in Fig. 2(b), and the computed shear-wave polarizations radiated from these mechanisms within the shear-wave window at the surface

9 Shear waves in the Peter the First Range ?lllI' w ( (0) (b) * Figure 6. (a) Equal-area projections, out to incidence angles of 50", of shear-wave polarizations radiated from the seven fault-plane mechanisms determined for period 11. The observed polarizations for these earthquakes are shown by superimposed bars. The projections are of the lower focal hemisphere to give the observations the same relative positions as in Fig. 4. (b) Superposition of the nodal lines of the seven focal mechanisms in an equal-area projection, out to incidence angles of 90", of the upper focal hemisphere with the common compressional segment marked by P. are shown in Fig. 6. The polarization directions which were observed at the GTSN sites from these particular earthquakes are superimposed on the computed polarization projections. The polarizations radiated from the seven mechanisms examined here (Fig. 6) show a wide range of possible orientations and it is unlikely that they could induce nearly parallel polarizations over the wide range of azimuths and incidence angles corresponding to the observations in Fig. 5. (2) The topography around each CTSN station in the PFR varies widely from site to site (Fig. I), and in particular there is no uniform trend in a NE/SW direction. The remarkable consistency of the alignment of the leading shear-wave polarizations observed on the irregular topography near the NAF in Turkey (Booth et al. 1985) suggests that topography has only a secondary effect on the polarizations of the leading shear-wave arrivals. Cormier (1 984) has shown theoretically that realistic irregular internal interfaces will rarely alter the polarization of shear waves by more than 10". Thus the effects of topography and inter-

10 410 S. Crampin et al. action with internal interfaces are unlikely to produce the observed alignments in the NE/SW and NW/SE directions, although they may well contribute to the 515" scatter about these orthogonal directions (see the discussion in Booth et al. 1985). (3) Shear waves entering an anisotropic medium split into two phases with different velocities and different, approximately orthogonal, polarizations which are fixed by the anisotropic symmetry. An arbitrarily polarized incident shear wave will normally generate both split shear waves in the anisotropic medium, and the recorded polarization of the leading shear wave will be that of the faster shear-wave. However, if the incident shear wave polarization is orthogonal to the polarization of the faster split shear wave, only the slower shear wave will be excited, and the recorded polarization will then be nearly orthogonal to that of the faster split shear wave. In Fig. 6 the observed polarizations at site 34 from earthquake G and site 34 from earthquake E are in cardinal directions and are probably misreadings of noisy records. The polarizations radiated from the source in the directions of the other observations have, with one exception, large components of displacement in the preferred direction agreeing with the interpretation that the observations are the result of propagation along ray paths through effectively anisotropic rocks. The exception, the polarization radiated from the source in the direction of site 34 from earthquake t, is almost exactly perpendicular to the preferred direction and is parallel to the observed polarization. This supports the interpretation of the polarizations perpendicular to the preferred direction as being the result of preferential excitation of the slower shear wave at the source. Thus, we reject both the source radiation and the surface interactions as the cause of the nearly parallel alignments of the shear-wave polarizations, and conclude that the parallel shear-wave polarizations are the result of effective anisotropy along the ray path. Similar parallel shear-wave polarizations near the NAF in Turkey have been interpreted as the result of the effective anisotropy of parallel vertical water-filled cracks or microcracks aligned perpendicular to the direction of least compressive stress (Crampin & Booth 1985). Fault-plane mechanisms suggest that the least compressive stress is dominant in that area of the NAF (ul > u2 > u3). Individual mechanisms only restrict possible axes of stress to the appropriate compressional or tensional quadrants. However, it seems reasonable to assume that the main driving force of the earthquakes beneath the GTSN is the regional stress-field acting on a variety of fault planes, so that the direction of the dominating stress should be common to all mechanisms (Crampin & Booth 1985). Superimposing nodal lines of the mechanisms from Fig. 2(b) in an equal-area projection in Fig. 6(b), a small segment of the compressional quadrants is common to all solutions. This direction of compression is from an azimuth of N 150"E at an incidence angle of 40" downwards. This is in good agreement with the expected direction of stress where the northward advancing Pamir mountains are compressing sediments in a plane of decollement at about 8 km depth. However, these stress orientations are not common to the region above the earthquake foci where the sediments have been extruded upwards into the folds of the Peter the First Range. Joints and fractures in folded areas are usually parallel to the axis of the folds, which is in broad agreement with the polarizations of the shear waves, so we again interpret the shear-wave polarizations as indicating the strike of nearly vertical parallel water-filled cracks aligned perpendicular to the axis of tensional (least compressive) stress. In the short period of observation, we have been unable to detect any variation in the delay between the split shear-waves, or in the direction of polarization. which might indicate a change of stress preparatory to a larger earthquake. Maximov, Tarasov & Fischer (1979) displayed polarization diagrams for several arrivals at GTSN in Only one at-rival, at site 34, is within the shear-wave window, and this shows a polarization and delay similar to the preferred direction and delays in this study.

11 7 Conclusions Shear waves in the Peter the First Range 41 1 The polarizations of shear waves from small earthquakes beneath the Garm Telemetered Seismic Network in the Peter the First Range display preferred orientations for arrivals within the shear-wave window. Most polarizations are aligned N 45'E/N 225"E, but there are a few polarizations at some sites which are orthogonal to this preferred direction. The polarizations appear to be the result of propagation through the effective anisotropy of vertical parallel water-filled cracks or microcracks aligned parallel to the direction of least compressive stress. This causes the shear waves to split into two components with nearly perpendicular polarizations. The preferred direction is the polarization of the leading shear wave, and the orthogonal direction is the result of the radiation from the source exciting only the second, slower, arrival. The principal stress causing the earthquakes beneath GTSN is compressional, whereas, in the volume of rock vertically extruded above the earthquakes, the principal horizontal stress is tensional resulting in cracks parallel to the axis of the folds perpendicular to the direction of compression of the deeper earthquakes. The shear waves propagating from the earthquakes through this extruded material have polarizations parallel to the strike of the cracks which are perpendicular to the direction of compression at the depth of the earthquakes in this Compressional regime. In contrast, similar observations of parallel polarizations of shear waves near the strike-slip regime of the North Anatolian Fault in Turkey are aligned parallel to the compressional axis of the fault-plane mechanisms. Not surprisingly, this application of shear-wave polarizations to monitor cracks and stress alignments in a compressive regime shows a completely different configuration of cracks and stress from those in a region where strike-slip movements are dominant. The observations reported here are further evidence supporting the hypothesis of extensivedilatancy anisotropy (EDA) that there are stress-aligned watei--filled microcracks throughout the upper km of the cool brittle crust (Crampin. Evans & Atkinson 1984; Crampin & Atkinson 1985). The new results from the PFR suggest that the aligned cracks and pores exist in the 8-10 km of poorly consolidated sediments above the plane of dicollement and they appear to have the same vertical parallel orientations above thrust-fault earthquakes as they have in the TDP strike-slip regime. Acknowledgments This work was partially supported by a Protocol between the Institute of Physics of the Earth and the British Geological Survey under the Scientific Exchange Agreement of the Academy of Sciences of the USSR and the Royal Society, London. We thank Academician V. A. Magnitsky and Professor I. L. Nersesov for their help and encouragement in setting up the Protocol and making this project possible. The work of SC and DCB was also supported by the Natural Environment Research Council and is published with the approval of the Director of the British Geological Survey (NERC). References Booth, D. C. & Crampin, S., Shear-wave polarizations on a curved wavefront at an isotropic free surface, Ceophys. J. R. astr. Soc., 83,3145. Booth, D. C., Crampin, S., Evans, R. & Roberts, G., Shear-wave polarizations near the North Anatolian Fault - I. Evidence for anisotropy-induced shear-wave splitting, Ceophys. J. R. asp. SOC., 83, Cormier, V. F., The polarization of S waves in a heterogeneous isotropic earth model, J. Ceophys., 56,20-23.

12 412 S. Crampin et al. Crampin, S., Evidence for aligned cracks in the Earth's crust, First Break. 3, 12~- 15. Crampin, S. & Atkinson, B. K., Microcracks in the Earth's crust, First Break, 3, Crampin, S. & Booth, D. C., Shear-wave polarizations near the North Anatolian Fault Interpretations in terms of crackinduced anisotropy, Geophys. J. R. ustr. Soc., 83, Crampin, S., Evans, R. & Atkinson, B. K., Earthquake prediction: a new physical basis, in Roc. First int. Workshop Seismic Anisofropy Suzdal, 1982, eds Crampin, S., tlipkin. R. G. & Chesnokov, E. M., Geophys. J. R. astr. SOC Crampin, S., Evans, R. & User, S. B., Analysis of records of local earthquakes: the Turkish Dilatancy Projects (TDPI and TDP2), Geophys. J. R. ustr. Soc., 83, Evans, R., Effects ot the free surface on shear wavetrains. Geophys. J. R. asrr. Soc., 76, Keith, C. M., Simpson, D. W. & Soboleva, 0. V., Induced seismicity and style of deformation at Nurek reservoir, Tadjik SSR, J. geophys. Res., 87, Lukk, A. A., Nersesov, I. L., Pevnev, A. K. & Yunp, S. L., Present-day motions of the Westcrn Part of the Peter I Range from geodetic and seismological data, Izu. Earth Phys., 16, Maximov, A. B., Tarasov, N. T. & Fischer, F. G., The structure of first arrivals of body waves of local earthquakes (in Russian), in Colln. Sovier-Am. Res. Eurthq. Predn., 2, Pellon, J. & Fischer, F., An earthquake catalog and velocity model for the USGS Peter the Iht Range seismic array, Tadjikistan, USSR, Open File Rep. U.S. geol. Surv.,