Seismic anisotropy observed in upper oceanic crust

GEOPHYSICALRESEARCHLETTERS, VOL. 8, NO. 8, PAGES 865-868, SEISMIC ANISOTROPY OBSERVED IN UPPER OCEANIC AUGUST 1981 CRUST R. A. Stephen Woods Hole Oceanographic Abstract. ,, Seismic anisotropy 1500m. of oceanic Institution, in the upper basement has been observed Woods Hole, Ma. 02543 distinctly polarized phases with different velocities. In the experiment described below, azimuthally varying velocity could not be resolved from travel-time and synthetic seismogram analysis but both types of particle motion anomalies by polarization analysis of three-component borehole seismometer records. The most convincing evidence for the anisotropy is shear wave split- were ting for explosive sources at four azimuths. Compressional wave particle motion deviations observed. The borehole seismic experiment was carried out on Deep Sea Drilling Project Leg 52 at Site suggestive of anisotropy are also observed but they may be caused by lateral inhomogeneities. The anisotropy was not resolved by travel-time analysis. The observed velocities and particle motions in the horizontal plane can be modelled to within a standard deviation by assuming a perfectly elastic, homogeneous, anisotropic layer 2 with hexagonal symmetry and a horizontal symmetry axis. The most probable cause of the anisotropy is preferred crack orientation. 417 in the Western Atlantic (Figure 1) (Donnelly et al., 1979). The site is located at the southern end of the Bermuda Rise immediately north of the Verna Gap and lies on magnetic anomaly M0 with an age of 109 My. The nearest fracture zones north are the and the Blake Fracture Abaco Fracture Zone, Zone, 20km to 80km to the the south. The bathymetry in the area trends approximately north-south. At hole 417D where the experiment was performed the sediments were 343m thick. Drilling at hole 417A, located 450m east of 417D, revealed a buried basement hill about 150m high. In general, however, the basement topography is poorly resolved. The shooting pattern for the experiment was oriented parallel and perpendicular to the estimated spread- The structure of upper oceanic crust is generally considered to vary with direction . In addition to the rather obvious symmetry of the spreading process, this hypothesis has been supported by the observation of linear seafloor topography and well oriented faulting and fractur- ing at the surface of oceanic basement (Ballard ing direction and van Andel, 1977; Luyendyk and Macdonald, 1977). Hence, models for the formation of oceanic crust include aligned lava flows and dyke which was determined from magnetic lineations. A description of the experiment, the complete three-component data set and the results of the initial travel-time and amplitude analysis have structures in the shallowmost crust (Cann, 1974; elastic parameters (Stoneley, the elastic behaviour varies 228m) within 1949) and with analysis direc- shots considered anisotropic with the compressional wave velocity faster perpendicular to the than parallel to it different on the east-west ly the particle azimuth (N50+6øE) than lines (N19+15øE). Clear- motions were being consistently of the x-axis is N34øE, which is the average of The conventional method for detecting anisotropy is to measure the variation of velocity with direction by travel-time analysis. However, there are two other distinct features of seismic propagation in an anisotropic medium which can be detected by particle motion analysis. First, the particle motion of the first arrival or quasi-compressional arrival is deflected away from the propagation direction. Second, the quasi-shear wave consists of two the two estimmtes. motion deviation counter-clockwise For this value, the particle for north-south shots is and for east-west shots 16ø+6 ø is 15+15 ø clockwise. The strongest evidence for anisotropy is the decoupling of shear waves into quasi-horizontally polarized and quasi-vertically polarized phases which propagate at different velocities. Examples of this phenomenon in the data can be seen in the particle Copyright 1981 by the American Geophysical Union. The vertical 865 O0 The adverse deflected away from the energy propagation direction either by anisotropy or by some lateral inhomogeneity. The best estimate of the azimuth (Hess, 1964). Paper number 1L0996. 0094-8276/81/001L-0996501. bedrock. described below (Kanasewich, 1973). significantly 1979) and the oceanic upper mmntle is spreading direction basaltic Since the geophone orientation was not known during the experiment, it was necessary to infer it from the data. Figure 2 shows the estimmted azimuth of the x-axis assuming that the compressional wave motion was in the direction of the shot. Shots on the north-south lines give a tion. For example, flat lying sediments can be considered anisotropic with the compressional wave velocity faster horizontally than vertical- ly (Levin, (N22øE) effects of inadequate coupling between the receiver and the ground motion, which complicate interpretation of ocean bottom seismometer records, are negligible in this experiment and it is meaningful to carry out the polarization in this study evidence for anisotropy has been observed by polarization analysis of threecomponent borehole seismometer data. Anisotropic media can be defined as media in which site been reported by Stephen et al., (1979) and Stephen (1979). High quality data was recorded because the geophone package was clamped (at Nisbet and Fowler, 1978; Sleep and Rosendahl, 1979). This letter reports the first observation of seismic anisotropy in the upper 1500 m of oceanic basement. The anisotropic structures appear to have too sinroll an effect on seismic velocities to be resolved using conventional travel-time and amplitude analysis. Particle motions, however, are more sensitive to aniso- tropic for crust at this vs. motion diagrams (Figure 3). transverse particle motion plot 866 Stephen: Anisotropy in Oceanic Crust The observed anisotropy has been modeled quantitatively assuming a flat, homogeneous, hexagonally anisotropic basement with all arrivals travelling horizontally just beneath the sediment-basement contact (Figure 4). 25 • 10' N This model is certainly no less valid than models assuming flat, isotropic structures with depthdependent velocity which are routinely used in seismic interpretation. Both sets of assumptions have limitations. In interpreting the particle motion behaviour however the former assumptions allow one to explain the observations. Hexagonally anisotropic media are a simple class of anisotropic media in which the elastic behaviour does not vary with direction in a plane normal to an axis of symmetry (Musgrave, 1970). These models are often a good approximation to anisotropic either Figure 1. Bathymetry in corrected metres at DSDP Site 417. The bathymetry is based on data collected by the R/V CHALLENGER on Legs 51, 52 and 53 and USNSLynch (Rabinowitz et al., 1979). Ships tracks are dashed. Hmavy solid lines represent the shooting pattern for the Oblique Seismic Experiment. for shear waves shows clearly polarized arrival seconds prior arrival faster media in which one axis or much slower than the is other two. Only five independent elastic constants are required to describe such media whereas 21 elastic constants are required to describe generally anisotropic media. Transversely isotropic media are hexagonally anisotropic with a N 16 • •4 a quasi-vertically 0 (qSV) occurring 0.09 to 0.13 to a quasi-horizontally (qSH). much / ß ! 0 •0. polarized 60 90 AZIMUTH (ø) The ranges in this experiment ; ; ; ; 5 km were not long enough to distinctly separate the quasi-vertically polarized shear waves from the quasi-horizontally polarized shear waves and qSV-qSH interference is present in the particle motion diagrams. The behaviour is observed on all four lines range. for shots from Of 51 identifiable 36 showed this 3.91kin to 12.11kin shear wave arrivals, distinctive particle motion pattern. No isotropic model can account for the different propagation velocities of the two shear wave polarizations. It is conceivable that the SH phases could be generated ESTIMATED AZIMUTH OF XAXIS by scat- tering of $V energy from inhomogeneities out of the sagittal plane, but the physics of this process is poorly understood. Anisotropy is the simplest and hence the most preferred explanation for the observed phenomena. In interpreting the results it is important to distinguish two types of events. The particle Figure 2. Estimated azimuth of the x-axis of the borehole geophone assuming the first compressional particle motion is in the direction motion elastic of surrounding timated events properties are a function of the crust the receiver on a scale of of approximately wavelength ( 300m).The travel hence velocities elastic properties travelled are however a time events and a function of the crust by the waves ( the immediately of the over ranges 4.0-12km). Since / the shot. At each / shot / location azimuth of the x-axis (+) is indicated. (.) the es- at the geophone The direction of particle mo- tion has been computed by polarization analysis (Kanasewich, 1973). All results are based on the first 0.1 sec of data which was bandpass filtered between 5 and 25Hz. Shots less than lkm in significant lateral inhomogeneity is present in the upper crust it is not necessary for these two types of events to be consistent with one another. For example, the shear wave polarizations in Figure 3 are affected by the elastic properties immediately around the receiver, but the travel time separation of the waves is affected by the elastic properties along the range have been omitted because they have near vertical ray paths which give inconsistent azimuths. Shots with power levels less than one whole north-south ray path. standard were also deviation above disregarded. the mean noise level The distribution of the azimuths is summarized in the histogram (inset). In an isotropic, homogeneous medium there be no difference between and east-west azimuths shots. estimated should from Stephen' Anisotropy vertical symmetry axis. here 001 the axis is dent values of the In the models described vertical is the symmetry axis. and In this tensor of the 010 elastic stiffnesses which must be satisfied 5.0 of the x-axis 4.5 4.0 for north-south shots and (19+_15)ø minus the azimuth of the x-axis for east-west shots, 2) the compressional wave velocity for both north-south and east-west lines must be 5.0+0.25 3.0 - km/sec (Stephen et al., 1979), 3) the velocity of the fastest shear wave phase (qSV)must be 2.6+0.1 km/sec and 4) the qSH arrivals must be 0.11+0.02 (I/$) 2.0 secs later than the qSV arrivals which implies a difference in velocity between 0.08 and 0.21km/sec. The variables in the modelling v ...... of the compressional wave particle motion direction from the phase propagation direction must be (50+6) ø minus the azimuth E-W •' (I/$) by the data for layer 2 are: 1) the deviation 867 5'5•N,, ' N-S axis case the indepen- devided by density are Fllll , F2222, Fl122 , F1212 and F1313. The observations in Oceanic Crust v • [ I I I I 120 150 180 20- V /1p o -2o T T T I P• 0 v v I 60 ENERGY , v I 90 PROPAGATION D/RECTION (ø) v Figure 4. Hexagonally anisotropic model ( which satisfactorily explain the observations. Compressional wave velocity (Vp), shear wave z •, 5.0,0 fv I :30 ' ' + v a.o 4oo , + v , + v velocity (Vs) and compressional wave particle motion deviation (Ap) are plotted against the 4oo energy propagation clockwise error bars south (N-S) direction from the 100 are the observed the particle H SH The values values and east-west (E-W) tropic model (eg qSV measured counter- axis. __) for lines. density (Fijkl) An iso- cannot explain motion observations. of the tensor of elastic with north- The values stiffness divided by used to model the anisotropy are: Fl111=30.25,F2222=16.00, F1212=8.50 ' V FR V FR V FR V Fl122=10.00,F1313=5.00(km/sec) 2 •, /5o, o , , T, 8.07 •m N Figure 3. ,7, I , + 6 0I •m $ Particle ,, I , + 6. 46 •m œ motion diagrams for •oo , I;-qs , ,v procedure are: 1) the azimuth of the x-axis of first the geophone, 2) the azimuth of the 100 axis of the anisotropic medium and 3) the five elastic compressional (P) and first shear (S) wave arrivals for four shots. The vertical (V), radial (R) and transverse (T) components are shown. A constantsFllll, trial and error dot indicates the beginning of the arrival and 0.2 seconds have been plotted. The observation fit of quasi-vertically servations. arriving prior polarized shear waves (qSV) to the quasi-horizontally ized shear waves (qSH) indicates elastic behaviour. The azimuth polar- anisotropic of the x-axis for all diagrams was assumed to be N34øE. The number in the lower left corner of the V-T plot is the amplification factor used in each case. The P-wave arrival is below the noise for for at 10.62km on the west line level. The S-wave arrivals ranges less than 7.0km are from layer 2 and ranges greater than 7.0km are from layer 3. F2222 , Fl122, F1212 and F1313. The variables were adjusted by the until the modeled parameters observations. The model shown in Figure The azimuth sumed to be the average within a standard clockwise 4 satisfies the the ob- x-axis was as- of the closest deviation mates for north-south E). The azimuth of anisotropy of of the and east-west the 100 axis of values mean lines the esti- (N39 was chosen to be N67 E which is 45 from the north-south line. In this model the selected elastic parameters which satisfy the data give compressional wave anisotropy between 4.0 and 5.5km/sec and shear wave anisotropy between 2.2 and 3.0km/sec. 868 Stephen' Anisotropy in Oceanic Crust Hess, H. Nature• Lond. 203, 629-631, 1964. In order to relate the velocity anisotropy observed on the scale of 300 m. to spreading direction and crustal formation one requires a geological explanation for the anisotropic behaviour. Drilling in the upper 366m showed that the upper crust consisted of basaltic pillow lavas and massive flows (Donnelly et al., 1979). A probable cause of anisotropy in these structures may be preferred orientation of cracks. No definitive orientation observations for preferred crack have been made on cored material. H•wever, comparison of laboratory velocities of hand samples, sonic log velocities and seismic experiment velocities indicates that about 5% porosity is caused by cracks size (Salisbury et al., scale porosity anisotropy. larger 1979). than core Aligned large may cause the observed Seismic anisotropy is present in upper oceanic crust. A thorough quantitative measurement of the anisotropy and an accurate determination of the direction of the principal axes will require more extensive and detailed experiments multi-component borehole seismometers. using Kanasewich, E.R. Time sequence analysis in geophysics (University of Alberta Press, Edmonton, 1973). Levin, F.K. Geophysics 44, 918-936, 1979. Luyendyk, B.P. & Macdonald, K.C. Bull. geol. Soc. Am. 88, 648-663 1977. Musgrave, M.J.P. Crystal Acoustics (Holden-Day, San Francisco, 1970). Nisbet, E.G. & Fowler, C.M.R. Geophys. J. R. ß astr. Rabinowitz, in Initial and the United States National Science & van Andel, T.H. Bull. LIII Francheteau, N.I., Part 1. (eds. J., Bryan, W., Francheteau, J., Hamreno,Y., & Johnson,D. in Deep Drilling Atlantic Ocean: Ocean Crust Hobart, M. Results in the (ed. Talwani, Geol. Soc. (American Geophysical Union, Washington, N.H. & Rosendahl, B.R.J. geophys. Res. 8_•.4,6832-6839, 1979. Stephen, R.A., Louden, K.E. & Matthews, D.H., Geophys.J.R. astr. Soc60, 289-300(1980) Reports of the Deep Sea Drilling T., Francheteau, LIII Part 1. (eds. J., Bryan, W., Robinson, P., Flower, M., Salisbury, M.) 675704 (U.S. GovernmentPrinting Office, 1979). Stephen, R.A. Marine Geophysical Researches 4, 213-226, Soc. 39, 169-187, 1974. Stoneley, 1979. R. Mon. Not. R. astr. Soc. geophys. Suppl. _5, 343-353, 1949. Donnelly, T.W. et al. in Initial Reports of the Deep Sea Drilling Project, Volume LI, LII, LIII Part 1. (eds. Donnelly, T., Francheteau, J., Bryan, W., Robinson, P., Salisbury, M.) 23-350 (U.S. Printing Office, 1979). Flower, M., Harrison, C.G. & Hayes, D.E.) 113-134 Project, VolumeLI, LII, References R.D. T., Robinson, P. Flower, M. Salisury, M.) 629-669 (U.S. GovernmentPrinting Office, 1979). Salisbury, M.H., Stephen, R.A., Christensen, Donnelly, Am. 8__•8, 507-530, 1977. Cann, J.R. Geophys. J.R. astr. 1978. P.D., Hoskins, H. and Asquith, S.M. Rmports of the Deep Sea Drilling - in Initial Foundation (Contract # OCE79-09351). (W.H.O.I. Contribution #4672). Ballard, 631-660, Project, VolumeLI, LII, Donnelly, Sleep, London 54, 1979). Acknowledgements. This research was supported by the Natural Environment Research Council - Soc. M., Government {Received March 51, 1981; accepted June 9, 1981.}