Assessing Variations in Shear Wave Splitting Measurement Due to Scattered Waves Using Simulated Waveforms

ASSESSING VARIATIONS IN SHEAR WAVE SPLITTING MEASUREMENT DUE TO SCATTERED WAVES USING SIMULATED WAVEFORMS

1 Kenny Graham, 1 Martha K Savage & 2 Richard Arnold & 3Takuto Maeda 1 Institute of Geophysics, SGEES, VUW, NZ 2School of Mathematics and Statistics, VUW, NZ , 2Earthquake Research Institute, the University of Tokyo Email address of presenting author: [email protected]

MOTIVATION METHOD & RESULTS Empirical Dataset Measuring seismic anisotropy through shear wave splitting, SWS, is a potential tool to infer 1-D noise-free synthetic waveforms (Figure 2) were stress changes in the Earth's crust, which may assist in hazard prediction and resource computed using [11] anisotropic reflectivity code. A flat extraction. layer medium with arbitrarily orientated anisotropy was

assumed. This potentially powerful tool has yet to be realized in part because measurements on local earthquakes are sometimes challenging and give varying results. MFAST, Savage[3] was used for the estimation of SWS Yet many researchers have reported varying results for local earthquakes, suggesting possible parameters. It uses dynamic window ranges (based on causes (Table 1). cluster analysis) and multiple filters to find the inverse splitting operator which best removes splitting. It uses an Horizontal layers of anisotropy, dipping symmetry axes, lateral variations in anisotropy, automatic scheme for grading results. scattering due to heterogeneity and other sources of noise may affect the measurements. a b To explore the effects of these proposed physical mechanisms of variation in SWS measurements, with the ultimate aim of improving SWS measurements , we generate synthetic Figure 2: SWS: An incident shear wave (incoming wave) splits into two polarized waveforms using 1-D reflectivity and 3-D finite difference techniques. waves upon encountering the

anisotropic medium (box). The blue Amplitude PROPOSED MECHANISMS wave, S1 (fast-quasi shear) wave travels Figure 6: a. Local earthquake. b. Corresponding faster than the red wave , S2 (slow-quasi Time [sec] shear). Split shear waves are often RMS envelope, showing P and S- Coda

Table 1 : Possible factors affecting SWS measurements described in terms of two parameters, the delay time δt (the delay time Figure 3: a. Simple horizontal isotropic models based on the Alpine Fault velocity model Proposed Mechanism References between S1 and S2) and the fast axis (Boese, 2012) used for the 1-D studies. The model consists of anisotropic layer orientation (φ ), which is the polarization sandwitched between an isotropic layer above and isotropic half space below. b.Sample of simulated waveforms used for analysis Scattered seismic energy (which could interfere with the direction of the fast-quasi shear. Volti [1], Walsh [2], Savage direct waves and cause apparent anisotropy variations) due [3], Crampin[4 & 5] 3-D Studies to inhomogeneities in the wave’s propagation path For the 3-D study, we determine the effect of scattering from inhomogeneities on measurements of shear wave splitting. We started with synthetic seismograms calculated in isotropic media. The structure at Ontake volcano was used. We Walsh [2], Silver [9], used an open-source finite difference seismic wave propagation code “Open SWPC” (Maeda et al., 2017 [10]), which Multiple and non-horizontal layers Crampin[7] has been used extensively to determine synthetic seismograms in Japan. We used input moment tensors that had been

derived from real earthquake (figure 1). We used the Japan-wide model available on the ERI supercomputer eJIVSM. Varying source-time functions and locations Volti [1] , Crampin[5] eJIVSM is an extended version of the integrated velocity model for Japan (JIVSM). The grid is from -23.1 km in the x Figure 5: Seismograph deployment across the Whataroa Valley; Green circles and y directions, -10 km in z. The grids are 2800 grids long in the x and y direction, with 0.0165 km in each grid shows the 160, 3C seismometers surrounding the Deep Fault Drilling Project (DFDP-2) site with spacing of 10 m perpendicular and 20 m parallel to the Interaction of S-wave with surface and sub-surface Crampin[4 & 5], Savage [6], direction (i.e. 46 km long in the x and y directions, and z 16 km in the z direction. The time step was 0.00145 seconds, main strike of the Alpine Fault. The yellow and red circles shows the even and topography with a minimum velocity vcut = 1.5 km/s. We tried several different models, with maximum frequencies ranging from 1 odd patch number during the deployments. The red square in the inset map Figure 7: Plot showing coherency of wave-form from dense shows the study area. deployment across Whataroa valley Hz to 12 Hz. A comparison of SWS result of real and synthetic earthquake is shown in Figure 4 OBJECTIVES Synthetic Real FUTURE WORK 1. Use a more realistic 3-D simulated waveform algorithm like SpecFEM3D, which allows for an We aim to understand the effect of the following on SWS measurements: anisotropic Earth models and accounts for topography and heterogeneity.

1. Scattered waves 2. Test the proposed mechanisms on empirical dataset. A dense deployment of seismographs

around Whataroa valley (Figure 5) will be used. The waveform coherency of this dataset (Figure 2. Multiple anisotropic layers and non-horizontal layers 6-7) will offer us the opportunity to explore the effect of scattered waves on the SWS

measurements. 3. Varying source location REFERENCES [1] Volti, T. and Crampin, S. (2003a), ‘A four-year study of shear-wave splitting in iceland: 1. background and preliminary analysis’, Geological Society, London, Special Publications 212(1), 117–133. 4. Noise ( both empirical and synthetic noise [2] Walsh, E. (2012), Measuring shear wave splitting using the Silver and Chan method, Master’s thesis, Victoria University of Wellington.

[3] Savage, M., Wessell, A., Teanby, N. and Hurst, A. (2010), ‘Automatic measurement of shear wave splitting and applications to time varying anisotropy at mount Ruapehu volcano, New Zealand’, Journal of Geophysical Research 115

[4] Crampin, Stuart, and David C. Booth. "Shear-wave splitting showing hydraulic dilation of pre-existing joints in granite." Scientific Drilling 1.1 (1989): 21-26.

[5] Crampin, Stuart, and Yuan Gao. "A review of techniques for measuring shear-wave splitting above small earthquakes." Physics of the Earth and Planetary Figure 4: Example of Mfast main steps of measuring S-wave splitting parameters. Here we compare synthetic (left) and real earthquake (right). (a) Filtered 3C seismogram (e,n,and z) Interiors 159.1 (2006): 1-14. waveforms. The solid vertical line is the S arrival and the gray box shows the window used. (b) Measuring splitting parameters by rotating waveforms into the incoming polarization direction (p) and its perpendicular value (top plot) and bottom two are the waveforms corrected for the measured dt. (c) The (top) waveforms and (bottom) particle motion for the (left) [6] Savage, M., Ferrazzini, V., Peltier, A., Rivemale, E., Mayor, J., Schmid, A., Brenguier, F., Massin, F., Got, J.-L., Battaglia, J. et al. (2015), ‘Seismic anisotropy original and (right) corrected waveform for the best measuremnt. (d) Contours of the smallest eigenvalue of the covariance matrix for the final chosen splitting parameters. and its precursory change before eruptions at Piton de la Fournaise volcano, La Réunion’, J. Geophys. Res. Solid Earth 120.

Amplitude COMMENTS [7] Johnson, J. H., Prejean, S., Savage, M. K. and Townend, J. (2010), ‘Anisotropy, repeating earthquakes, and seismicity associated with the 2008 eruption of Okmok volcano, Alaska’, Journal of Geophysical Research 115. Multiple layers: Introduction of an extra isotropic horizontal layer to the model had a significant effect on the splitting parameters. This shows that multiple layers do affect SWS measurements, but a full 3-D study is [8] Levin, V. and Park, J. (1997), ‘P-SHconversions in a flat-layered medium with anisotropy of arbitrary orientation’, Geophys. J. Int. pp. 253–266.

planned to fully understand this observation. [9] Silver, Paul G., and Martha K. Savage. "The interpretation of shear-wave splitting parameters in the presence of two anisotropic layers." Geophysical Journal International 119.3 (1994): 949-963

Time [sec] The 1-D reflectivity and the 3-D finite difference simulated waveforms could not explain all the proposed [10] Maeda, T., T. Furumura, and K. Obara (2014), Scattering of teleseismic P-waves by the Japan Trench: A significant effect of reverberation in the seawater Figure 1: Comparison of synthetic and real waveform at station V.ONTV for event 150309175322 on March 9 column, Earth Planet. Sci. Lett., 397(1), 101-110, doi:10.1016/j.epsl.2014.04.037,. 2015 at 17:53.22. Solid line is the real waveform with the dashed line as the synthetic waveform physical mechanisms affecting SWS measurements ACKNOWLEDGMENTS: This project is funded by Victoria University of Wellington. We would like to thank the DFDP-2 VSP data acquisition team for providing this dataset for this studies. The author wishes to acknowledge Jeffrey Park and Takuto Maeda for providing the reflectivity and the 3-D finite Difference code.