Shear Wave Splitting in Foliated Rock

Sergei Stanchits (USGS), David Lockner (USGS), Jens Zinke (Stanford University)


Orientation of sample foliation

Figure 1. Sample assembly for measuring seismic wave velocities.

During an earthquake, compressional waves (P waves) and shear waves (S waves) radiate outward in all directions through the interior of the earth. Under certain conditions, S waves can be split, that is, polarized into different directions as they propagate. In rocks that have a strong foliation or grain fabric, the velocities of these waves often depend on their orientation. "Shear wave splitting", as it is termed, has been studied in many areas of the earth, and variations in the delay times of arriving wave signals have been suggested as possible earthquakes precursors. However, it has not yet been proven that the measured effects can be associated with some physical parameter such as a change in the pressure of fluids in the cracks and pore of a rock (pore pressure) or of changes in stress before an earthquake, because of complications with direct field measurements. While theoretical advances have been made related to this topic, there are no direct reliable laboratory measurements relating changes in shear wave splitting to changes in the physical characteristics of rocks. Accordingly, we have measured P- and S-wave velocities of wet and dry rock samples during varying confining pressures to 240 MPa and pore pressures to 80 MPa using a newly-developed wavespeed measurement system (Fig.1). Fast and slow shear wave velocities were measured separately using two pairs of orthogonally mounted polarized shear wave piezoelectric transducers (760 kHz resonance).


For a test material, we used a foliated quartz phyllite from Alta, Norway with an intrinsic shear wave velocity anisotropy of about 10%. Analyses of thin section images showed that naturally-existing cracks were mostly related to micas, and that mean crack size was about 100 microns in the direction parallel to the foliation plane and about 30 microns in the orthogonal direction. Crack orientation distribution showed alignment of naturally-existing cracks with the foliation plane.

We prepared three sets of samples (Fig. 2):

Orientation of sample foliation

Figure 2. Orientation of sample foliation relative to P- and S- wave directions.

Normalized P- and S-wave velocities

Figure 5. Normalized P- and S-wave velocities, showing constant ratios (i.e., crack closure) for non-thermally-treated samples at 50 MPa pressure., and at 120 MPa for the thermally-treated samples.

P and S wave velocities as a function of pressure for all samples

Figure 4. P- and S- wave velocities as a function of pressure for all samples (dry).

P- and S- waveforms as a function of pressure for sample B

Figure 3. P- and S- waveforms as a function of pressure for sample B.


P- and S-waveforms determined at different confining pressures are presented in Figure 3 for sample B. P-wave arrivals were a function of confining pressure (Figure 3a). A comparison of Figures 3b and 3c demonstrates that the S-wave polarized parallel to the foliation plane arrived at the receiver earlier than the S-wave polarized orthogonally to the foliation plane, indicative of shear wave splitting in the sample.

We observed that both P- and S-waves showed increasing velocities with increased confining pressure, as shown in Figure 4, and also that thermally treated samples (C) have lower velocities at confining pressures below 100 MPa. This can be explained by the influence of cracks that are opened at lower confining pressures. At higher pressures almost all of these cracks are closed, accounting for the fact that there is almost no difference in the P- and S- wave velocities of samples (B) and (C). Comparison of P-wave velocities of orthoganally-cored samples (A) and (B) in Figure 4a demonstrate an intrinsic P-wave anisotropy of about 10% relative to matrix properties. We found no S-wave splitting in sample (A), and the value of the S-wave velocity was within 1% of the slow velocity of cores measured parallel to foliation. Figure 5 shows P- and S- velocities normalized to the values of the fast velocities of (B). This figure demonstrates that for the non-thermally treated samples (A) and (B), the ratio of velocities became constant at confining pressures higher than 50 MPa. This can be explained by the closure of naturally existing cracks with pressure. For the thermally treated Samples (C), a constant velocity ratio was reached only at pressure level of 120 MPa, since the additional cracks induced by thermal treatment closed at higher confining pressures. In case of water saturated samples, we observed that the shear wave splitting changed as a function of effective confining pressure (confining pressure minus pore pressure) as shown in Figure 6. Other observed features of the wet samples were very similar to the dry samples.

Difference in slowness as a function of pressure

Figure 7. Difference in slowness as a function of pressure between (a) S- and P- waves and (b) two shear waves, showing that most shear wave splitting is below 100 MPa, a result of microcrack closure.

P- and S-wave velocities as a function of 
                        confining pressure

Figure 6. P- and S-wave velocities as a function of confining pressure and effective confining pressure for saturated samples. Shear wave splitting changes as a function of effective confining pressure.

We have also analyzed the data in terms of the difference in slowness between two shear waves and between S- and P-waves (Fig. 7). It is clear that most variations in shear wave splitting occurred below 100 MPa confining pressure, presumably the result of microcrack closure. At higher pressures the shear wave splitting remained constant, influenced only by the anisotropy of the sample. Thermally treated samples with higher crack densities showed larger changes in shear wave splitting.


We have carried out laboratory investigations that show the existence of shear wave splitting in foliated quartz phyllite rock samples. The dual nature of the observed splitting is related to: 1) anisotropic properties of the matrix, and 2) the system of opened crack aligned with the foliation. Shear wave splitting is sensitive to crack closure from either increasing confining pressure or decreasing pore pressure. The threshold level of confining pressure when most cracks become closed was higher when additional cracks were induced by thermal treatment. Based on the results of our laboratory observations, we would expect to be able to monitor changes in the delay time of fast and slow shear waves from earthquake sources only if the effective pressure (confining pressure minus pore pressure) was low.