9:45 AM - 10:00 AM
[SSS07-04] Estimation of shallow subsurface structure of elastic wave velocity before and after rainfall by active seismic monitoring on a slope
Keywords:Slope, Active seismic monitoring, Velocity structure, Rainfall
In recent years, studies have begun to be conducted to investigate changes in water content and groundwater table in a slope to detect precursors of slope surface failure triggered by rainfall or snowmelt through active seismic monitoring. For example, Nakayama et al. (2022, SSJ) embedded a vibration speaker and accelerometers in a slope (Figure 1) and repeatedly applied linear sweep (chirp) signals to the speaker to observe elastic waves for periods including rainfall. They extracted fundamental tone components from stacked complex spectrograms and investigated temporal variations in phase spectra. The phase spectra were sufficiently stable before and after the rainfall. If there is no velocity dispersion, that is, waves at any frequency propagate through a uniform medium on the same paths, the phase change is proportional to the product of frequency and the change in travel time between two receivers. Note that the phase change obtained at receivers included the ambiguity of cycles, that is, integer (n) multiple of 2π. Using a linear line fitting between the unwrapped phase and the frequency, they estimated the velocity at which the residual at any n becomes minimum by grid search. In the results, the velocity decreased after the rainfall.
However, residuals were almost the same even if n changed, which made the reliability of the absolute value of the estimated velocity questionable although the phase data were relatively consistent with the approximate linear line. In this study, we reanalyzed the data obtained by Nakayama et al. (2022, SSJ) and tried to estimate the velocity structure before and after the rainfall.
We re-extracted the fundamental tone components (91-300 Hz) every 1 Hz from the spectrograms of the stacked waveforms calculated by Nakayama et al (2022, SSJ). Assuming an average apparent velocity between a reference receiver and another receiver, the corresponding phase change can be calculated for each receiver. Calculating cross-spectrum between the spectrum for each receiver pair, its phase components are expected to be zero in the entire frequency band of the input waveform when the average apparent velocity is appropriate. We calculated the cross-spectra between AC8 (reference receiver) and each accelerometer (AC3-7) for the band (210-230 Hz) before the rainfall (about 10 hours) where the phase was stable.
The estimated average apparent velocity for each accelerometer from AC8 tended to decrease with distance (Figure 2). When the transmitted waves are surface waves with a single frequency or body waves through a uniform media, the average apparent velocity should be independent of distance. Changing the average apparent velocity with distance suggested that the apparent velocity was for a body wave propagating a depth-dependent structure. We will estimate the velocity structure before and after the rainfall, and discuss how the velocity structure changed.
Acknowledgments: We thank JSPS Grants-in-Aid for Scientific Research JP15H02996 and 26750315.
However, residuals were almost the same even if n changed, which made the reliability of the absolute value of the estimated velocity questionable although the phase data were relatively consistent with the approximate linear line. In this study, we reanalyzed the data obtained by Nakayama et al. (2022, SSJ) and tried to estimate the velocity structure before and after the rainfall.
We re-extracted the fundamental tone components (91-300 Hz) every 1 Hz from the spectrograms of the stacked waveforms calculated by Nakayama et al (2022, SSJ). Assuming an average apparent velocity between a reference receiver and another receiver, the corresponding phase change can be calculated for each receiver. Calculating cross-spectrum between the spectrum for each receiver pair, its phase components are expected to be zero in the entire frequency band of the input waveform when the average apparent velocity is appropriate. We calculated the cross-spectra between AC8 (reference receiver) and each accelerometer (AC3-7) for the band (210-230 Hz) before the rainfall (about 10 hours) where the phase was stable.
The estimated average apparent velocity for each accelerometer from AC8 tended to decrease with distance (Figure 2). When the transmitted waves are surface waves with a single frequency or body waves through a uniform media, the average apparent velocity should be independent of distance. Changing the average apparent velocity with distance suggested that the apparent velocity was for a body wave propagating a depth-dependent structure. We will estimate the velocity structure before and after the rainfall, and discuss how the velocity structure changed.
Acknowledgments: We thank JSPS Grants-in-Aid for Scientific Research JP15H02996 and 26750315.