10:00 AM - 10:15 AM
[SSS09-05] Seismic velocity response to ground water level revealed by ACROSS and seismic array
Keywords:ACROSS, array, surface wave, groundwater level
Temporal variation in propagation property of seismic wave was monitored at the landslide area using Accurately Controlled Routinely Operated Signal System (ACROSS) and a seismic array. The ACROSS, which is a precisely-controlled seismic source, is deployed in Mori-machi, Shizuoka Prefecture, Japan, and used in this study.
Before this experiment, a preliminary observation was conducted for three weeks in each of 2017 and 2018. We set up two seismometers at a site which is 3 km away from the ACROSS source to investigate the response of seismic velocity to rainfalls by analyzing the transfer function obtained from the ACROSS signal. The temporal variation of the arrival time of the later phases showed a significant response to rainfalls. The rainfall is converted into the groundwater level to investigate the relation to the propagation property of seismic waves. We assumed a tank model in the conversion by the exponential decay function with a time constant of 20 days.
We found that there is a strong correlation between the arrival time variation of these waves and the groundwater level, and also found the sensitivity function of the two wave groups is different.
A small seismic array is deployed to identify the origin of the phases and to monitor the temporal variation. 14 seismometers were deployed following the typical array configuration of SPAC methods, which consists of a center and three triangles on concentric circles with radii of 4, 8, and 16 meters. Two additional seismometers on each of the south and north sides of the array to increase the wavenumber resolution in the source direction. The total size of the array is 70 m in NS and 30m in EW. This array arrangement is intended to estimate the shallow shear-velosity profiles by inversing the surface wave dispersion curve by SPAC method using ambient noises and analysis the ACROSS signal. The seismometers are Mark Products L-28-3D (sensitivity 120 V/m/s) of three-component velocity-type ones with a natural frequency of 4.5 Hz. The sampling frequency is 1 kHz, and the inter-channel skew of sampling is maintained within 15 µs.
The results of the array analysis of ACROSS signal showed that the phases arriving around 2.8 and 3.8 seconds, whose response to the rainfall was significant, had the same apparent velocity as the body waves (~2.5 km/s). This observation contradicts the result of the analysis of the 2017 and 2018 observations, in which the later phases are identified as Love and Raiyleigh waves based on their particle motions. The wave at 2.8 seconds had a dominant transverse component, which was regarded as a Love wave, and the wave at 3.8 seconds had retrograde motion in the vertical plane, which was regarded as Rayleigh wave.
The variation of the arrival time of the phase around 2.8 and 3.8 seconds were estimated by applying the cross-spectrum method to the transfer function during the observation period, and a strong correlation with rainfall was found in 2017 and 2018 experiments. For all observations from 2017 to 2020, there was a positive linear relationship between the change in the arrival time of the phase around 2.8 seconds and the groundwater level that is converted from the rainfalls. The sensitivity of the change in velocity [%] to the increase in groundwater level [m] was 1.04 [%/m]. In contrast, there was a third-order relationship between the change in the arrival time of the phase arriving around 3.8 seconds and the groundwater level. The sensitivity of the velocity change became stronger when the groundwater level increased above a certain level.
The amount of precipitation during the observation period of 2020 was much less than the previous two years, and the relation cannot be extended outside the range of the previous two years' observation. We will continue the observation during the high precipitation season and pursue the mechanism of these variations.
Before this experiment, a preliminary observation was conducted for three weeks in each of 2017 and 2018. We set up two seismometers at a site which is 3 km away from the ACROSS source to investigate the response of seismic velocity to rainfalls by analyzing the transfer function obtained from the ACROSS signal. The temporal variation of the arrival time of the later phases showed a significant response to rainfalls. The rainfall is converted into the groundwater level to investigate the relation to the propagation property of seismic waves. We assumed a tank model in the conversion by the exponential decay function with a time constant of 20 days.
We found that there is a strong correlation between the arrival time variation of these waves and the groundwater level, and also found the sensitivity function of the two wave groups is different.
A small seismic array is deployed to identify the origin of the phases and to monitor the temporal variation. 14 seismometers were deployed following the typical array configuration of SPAC methods, which consists of a center and three triangles on concentric circles with radii of 4, 8, and 16 meters. Two additional seismometers on each of the south and north sides of the array to increase the wavenumber resolution in the source direction. The total size of the array is 70 m in NS and 30m in EW. This array arrangement is intended to estimate the shallow shear-velosity profiles by inversing the surface wave dispersion curve by SPAC method using ambient noises and analysis the ACROSS signal. The seismometers are Mark Products L-28-3D (sensitivity 120 V/m/s) of three-component velocity-type ones with a natural frequency of 4.5 Hz. The sampling frequency is 1 kHz, and the inter-channel skew of sampling is maintained within 15 µs.
The results of the array analysis of ACROSS signal showed that the phases arriving around 2.8 and 3.8 seconds, whose response to the rainfall was significant, had the same apparent velocity as the body waves (~2.5 km/s). This observation contradicts the result of the analysis of the 2017 and 2018 observations, in which the later phases are identified as Love and Raiyleigh waves based on their particle motions. The wave at 2.8 seconds had a dominant transverse component, which was regarded as a Love wave, and the wave at 3.8 seconds had retrograde motion in the vertical plane, which was regarded as Rayleigh wave.
The variation of the arrival time of the phase around 2.8 and 3.8 seconds were estimated by applying the cross-spectrum method to the transfer function during the observation period, and a strong correlation with rainfall was found in 2017 and 2018 experiments. For all observations from 2017 to 2020, there was a positive linear relationship between the change in the arrival time of the phase around 2.8 seconds and the groundwater level that is converted from the rainfalls. The sensitivity of the change in velocity [%] to the increase in groundwater level [m] was 1.04 [%/m]. In contrast, there was a third-order relationship between the change in the arrival time of the phase arriving around 3.8 seconds and the groundwater level. The sensitivity of the velocity change became stronger when the groundwater level increased above a certain level.
The amount of precipitation during the observation period of 2020 was much less than the previous two years, and the relation cannot be extended outside the range of the previous two years' observation. We will continue the observation during the high precipitation season and pursue the mechanism of these variations.