10:45 AM - 12:15 PM
[SSS12-P09] Detectability of tilt variations by tiltmeters in seafloor borehole at off the Kii Peninsula
Keywords:tiltmeter, Nankai trough, Seafloor Geodesy, Short-term Slow slip
We assessed detectability for tilt variation in seismic and geodetic frequency ranges by tilt meters installed into seafloor boreholes at the Kumano basin, off the Kii peninsula along the Nankai trough. In the seismic range, we compared power spectral density with broadband seismometers around the tiltmeters. In the geodetic range, we used standard deviation after filtering for the detection of short-term slow slip events.
In the Kumano basin, the Philippine Sea plate is subducting beneath the Amour plate. The subduction causes M8-class earthquakes with intervals of 100 – 200 years. VLFs and short-term SSEs are also detected in the region. To monitor seismic activities, an observation network to observe seismic waves and pressure “Dense Oceanfloor Network system for Earthquakes and Tsunamis” (DONET) is deployed in 2011 and is under operation.
We installed two tiltmeters to measure tectonic phenomena in a wide range of frequency bands relative to the existing network. First, an electrolytic tiltmeter was installed 6m below the seafloor in 2019. Based on the experience of the first one, we installed a more sensitive sensor in a deeper place in 2021. A pendulum tiltmeter was developed based on optical geodetic sensor (Zumberge et al., 2018) for more sensitive measurement and installed 19m below the seafloor (Kimura et al., 2021, JpGU Meeting).
To check the detectability in seismic frequency ranges, we compared noise power spectral density (PSD) of tiltmeters to that of four broadband seismometers (natural frequency of 360s) in DONET. We used continuous records in November 2022 and calculated ensemble PSDs. Continuous records are divided into each 1-hour trace and selected available traces so that traces do not include earthquake signals. Then we calculate PSD for each available trace and averaged them to obtain ensemble PSD.
As a result, the Pendulum type tiltmeter shows a lower noise level than DONET stations. The PSD of the Pendulum tiltmeter shows a similar shape in frequency range in microseism (0.8 – 10 Hz). Frequencies under 0.8 Hz the pendulum tiltmeter shows lower power than DONET stations. Especially, under 0.01 Hz, the pendulum tiltmeter shows approximately 100 times lower than that of the best DONET station. The Electrolytic tiltmeter shows almost same noise level as broadband seismometers under 0.1 Hz. This is caused by the limit of resolution of the tiltmeter. The lowest values of PSD in NS and EW components are less than the resolution on the spec sheet.
Next, we assessed detectability for short-term SSE by estimating standard deviation (STD) of tilt variation in a period after filtering. To pick up tilt change with SSE, we applied two filters. The first one is a low pass filter to cut off frequency components shorter than one day because clear tilt variations that are well correlated with tide are observed for both tiltmeters. The second one is moving average to enhance variations of longer than several days of duration and to reduce noise.
As a result, tilt changes over 0.5 micro radians are detectable for both the electrolytic and pendulum tiltmeter based on 95% confidence intervals in January 2023. For the electrolytic tiltmeter, STDs in NS and EW components are 0.23 and 0.059 micro radians, respectively. For the pendulum tiltmeter, those are 0.13 and 0.23 micro radians, respectively. We didn’t remove linear trends in this analysis. Thus, actual detectability may be higher than those values.
In the Kumano basin, the Philippine Sea plate is subducting beneath the Amour plate. The subduction causes M8-class earthquakes with intervals of 100 – 200 years. VLFs and short-term SSEs are also detected in the region. To monitor seismic activities, an observation network to observe seismic waves and pressure “Dense Oceanfloor Network system for Earthquakes and Tsunamis” (DONET) is deployed in 2011 and is under operation.
We installed two tiltmeters to measure tectonic phenomena in a wide range of frequency bands relative to the existing network. First, an electrolytic tiltmeter was installed 6m below the seafloor in 2019. Based on the experience of the first one, we installed a more sensitive sensor in a deeper place in 2021. A pendulum tiltmeter was developed based on optical geodetic sensor (Zumberge et al., 2018) for more sensitive measurement and installed 19m below the seafloor (Kimura et al., 2021, JpGU Meeting).
To check the detectability in seismic frequency ranges, we compared noise power spectral density (PSD) of tiltmeters to that of four broadband seismometers (natural frequency of 360s) in DONET. We used continuous records in November 2022 and calculated ensemble PSDs. Continuous records are divided into each 1-hour trace and selected available traces so that traces do not include earthquake signals. Then we calculate PSD for each available trace and averaged them to obtain ensemble PSD.
As a result, the Pendulum type tiltmeter shows a lower noise level than DONET stations. The PSD of the Pendulum tiltmeter shows a similar shape in frequency range in microseism (0.8 – 10 Hz). Frequencies under 0.8 Hz the pendulum tiltmeter shows lower power than DONET stations. Especially, under 0.01 Hz, the pendulum tiltmeter shows approximately 100 times lower than that of the best DONET station. The Electrolytic tiltmeter shows almost same noise level as broadband seismometers under 0.1 Hz. This is caused by the limit of resolution of the tiltmeter. The lowest values of PSD in NS and EW components are less than the resolution on the spec sheet.
Next, we assessed detectability for short-term SSE by estimating standard deviation (STD) of tilt variation in a period after filtering. To pick up tilt change with SSE, we applied two filters. The first one is a low pass filter to cut off frequency components shorter than one day because clear tilt variations that are well correlated with tide are observed for both tiltmeters. The second one is moving average to enhance variations of longer than several days of duration and to reduce noise.
As a result, tilt changes over 0.5 micro radians are detectable for both the electrolytic and pendulum tiltmeter based on 95% confidence intervals in January 2023. For the electrolytic tiltmeter, STDs in NS and EW components are 0.23 and 0.059 micro radians, respectively. For the pendulum tiltmeter, those are 0.13 and 0.23 micro radians, respectively. We didn’t remove linear trends in this analysis. Thus, actual detectability may be higher than those values.