10:15 AM - 10:30 AM
[SSS10-16] Ubiquity of ground motion pulses near a seismic source fault
Keywords:ground motion pulse, source fault, ubiquity, directivity effect
Since it was discovered in the Kobe earthquake etc., ground motion with a pulse-like waveform ("ground motion pulse") has been considered a special phenomenon, and its origins have been debated. However, here we show that ground motion pulses are ubiquitous, appearing at any locations and components near the source fault. In the Kumamoto earthquake, ground motions were recorded at several locations near the source fault. The displacement records have waveforms like smoothed ramp functions and their differentials, the velocity waveforms, formed ground motion pulses (e.g., Nozu et al., 2019). In addition, many nearby records were observed for the 2023 earthquake in southern Turkey, with similar results (e.g., Nagasaka, 2023). In the latter case, ground motion pulses also appear in the displacement records. They will be interpreted at the end of this abstract.
Ground motion simulations were performed to investigate the ubiquity of ground motion pulses. Following Dreger et al. (2011), we did not use analytical or semi-analytical methods, but used OpenSWPC (Maeda et al. 2017) based on the finite difference method. In a Cartesian coordinate system, the x-, y-, and z-axes point northward, eastward, and downward. An Mw 5.0 point source of right-lateral strike-slip faulting was placed at (0, 0, 3km), referring to the Kumamoto earthquake. Since it is well established that the function of Nakamura and Miyatake (2000) is appropriate as a source time function, we selected the texp function, which is similar to it, from the OpenSWPC options. The top waveforms in the figure are the results for a case where the seismic basement is semi-infinitely spreading, and the smoothed ramp shape is reproduced in both components. It is similarly reproduced in the vertical component. In the rising part of the ramp shape, a displacement in the opposite direction appears, and this can also be seen in observed records. Since this coincides with the P wave arrival, it can be due to different polarity between the P and S waves.
Since a velocity structure can be a primary factor that disturbs this reproduction, we extracted a 1D structure in Mashiki from the JIVSM V1R. This was introduced into the ground motion simulation, and the results are the second waveforms in the figure. Due to the amplification by the velocity structure, the maximum amplitude increases about three times, and the ramp-like waveform is maintained. However, since the amplification in the crustal deformation part is weaker, which is about twice, a swing-back peak appears after the ramp shape part. Since the finiteness of the source fault can be another factor that disturbs the reproduction, we located a 2.5km long line source (Mw 5.5) centered on the point source and extended horizontally in the halfspace and 1D velocity structure. It consists of 5 subfaults 0.5km long. The fault rupture was assumed to propagate eastward from the western end at 3.0km/s. Looking at the results in the third and fourth rows of the figure, the amplitude has increased by about 5 times in proportion to M0, but the ramp-like waveform is maintained. However, the time length of the ramp shape part is longer and the pulse width of the velocity waveform is longer. The influence of the finiteness of the source fault is called the directivity effect, but this does not produce ramp-like displacement waveforms and ground motion pulses. This affects the ubiquitous ground motion pulses, expanding the pulse width and lengthening the period. Furthermore, since the amplification of the ground motion pulse is commensurate with M0, the amplification due to the effect is not significant near the center of the source fault.
Finally, we consider the ground motion pulse in ground displacements. In the fault-normal displacements due to the point source, the entire waveforms have the same shape on the left and right sides of the nodal plane parallel to the x-axis, but the polarity is reversed. Therefore, in a displacement on the central nodal plane due to the line source, the crustal deformation part becomes zero as the left and right contributions cancel out. However, because time differences of rupture propagation are not symmetrical, the sum does not become zero, remaining as a pulse waveform.
Acknowledgments: Drs. S. Takemura, T. Furumura, and H. Miyake helped us about OpenSWPC and for motivation of this prensentation.
Ground motion simulations were performed to investigate the ubiquity of ground motion pulses. Following Dreger et al. (2011), we did not use analytical or semi-analytical methods, but used OpenSWPC (Maeda et al. 2017) based on the finite difference method. In a Cartesian coordinate system, the x-, y-, and z-axes point northward, eastward, and downward. An Mw 5.0 point source of right-lateral strike-slip faulting was placed at (0, 0, 3km), referring to the Kumamoto earthquake. Since it is well established that the function of Nakamura and Miyatake (2000) is appropriate as a source time function, we selected the texp function, which is similar to it, from the OpenSWPC options. The top waveforms in the figure are the results for a case where the seismic basement is semi-infinitely spreading, and the smoothed ramp shape is reproduced in both components. It is similarly reproduced in the vertical component. In the rising part of the ramp shape, a displacement in the opposite direction appears, and this can also be seen in observed records. Since this coincides with the P wave arrival, it can be due to different polarity between the P and S waves.
Since a velocity structure can be a primary factor that disturbs this reproduction, we extracted a 1D structure in Mashiki from the JIVSM V1R. This was introduced into the ground motion simulation, and the results are the second waveforms in the figure. Due to the amplification by the velocity structure, the maximum amplitude increases about three times, and the ramp-like waveform is maintained. However, since the amplification in the crustal deformation part is weaker, which is about twice, a swing-back peak appears after the ramp shape part. Since the finiteness of the source fault can be another factor that disturbs the reproduction, we located a 2.5km long line source (Mw 5.5) centered on the point source and extended horizontally in the halfspace and 1D velocity structure. It consists of 5 subfaults 0.5km long. The fault rupture was assumed to propagate eastward from the western end at 3.0km/s. Looking at the results in the third and fourth rows of the figure, the amplitude has increased by about 5 times in proportion to M0, but the ramp-like waveform is maintained. However, the time length of the ramp shape part is longer and the pulse width of the velocity waveform is longer. The influence of the finiteness of the source fault is called the directivity effect, but this does not produce ramp-like displacement waveforms and ground motion pulses. This affects the ubiquitous ground motion pulses, expanding the pulse width and lengthening the period. Furthermore, since the amplification of the ground motion pulse is commensurate with M0, the amplification due to the effect is not significant near the center of the source fault.
Finally, we consider the ground motion pulse in ground displacements. In the fault-normal displacements due to the point source, the entire waveforms have the same shape on the left and right sides of the nodal plane parallel to the x-axis, but the polarity is reversed. Therefore, in a displacement on the central nodal plane due to the line source, the crustal deformation part becomes zero as the left and right contributions cancel out. However, because time differences of rupture propagation are not symmetrical, the sum does not become zero, remaining as a pulse waveform.
Acknowledgments: Drs. S. Takemura, T. Furumura, and H. Miyake helped us about OpenSWPC and for motivation of this prensentation.