2:15 PM - 2:30 PM
[STT42-03] Comparison of teleseismic strain fields obtained by array analysis of seismometers and direct measurement of strainmeters at seafloor and deep seafloor observatory
Keywords:Fiber optic strainmeter, DONET, Dynamic strain records, Strain tensor
Recent progress in fiber optic sensing, such as Distributed Acoustic Sensing (DAS), has made strain observation popular. Many studies use DAS records as records of ground motion observed by a dense seismic array. To fully utilize these strain records, comparisons with existing observatories are necessary. This requires comparing strain records obtained from fiber sensing with velocity or acceleration records from seismometers.
One method for comparison is the estimation of the strain tensor through array analysis, using more than three stations with three-component seismometers (Spudich et al., 1995). Array processing can resolve the strain tensor from the displacement at each station, obtained by integrating the velocity records. Therefore, if the strainmeters are located within or near the array, strain records can be compared with synthesized strains produced by the array processing.
We applied this method to three linear fiber optic strainmeters on the seafloor and one borehole fiber optic strainmeter located 450 meters below the seafloor. These strainmeters were installed around Node-2F of DONET2, off the Kii Channel. Each seafloor strainmeter uses a 200 m fiber optic cable extended on the seafloor as a strain sensor. Normal strain in the extended direction is observed. Thus, the strain records were compared to the component in the synthesized strain tensor. Two of the three were extended in the same direction and the other was extended 60 degrees from the direction of the others. The borehole strainmeter uses fiber optic cables wrapped around a stainless steel mandrel like a coil to install a long-baseline sensor into the limited space of the borehole. Thus, strain used for the comparison was calculated based on the strain tensor and geometrical shape of the sensor fiber after Tsuji et al. (2024, JpGU). In addition, the coefficient of sensitivity of the sensor 0.58 (Tsuji et al. 2024, SSJ) was applied. This coefficient accounts for the effects of surrounding structures during the installation process on the strain measurements.
To perform the comparison, we synthesized the strain field of teleseismic signals using four broadband seismometers of the DONET2 around the strainmeters. Two earthquakes were selected based on signal-to-noise ratios: the M7.4 earthquake in Taiwan on April 3rd, 2024, and the M7.3 earthquake in Vanuatu on December 17th, 2024. Since the maximum distance between the stations was approximately 20 km, a bandpass filter with a range of 20 to 100 seconds was applied.
As a result, most phases in the strain records were obtained in the synthesized strain waveforms with almost equal amplitude. S phase and surface waves were major phases due to the sensor characteristics and frequency ranges of the band pass filter. For the seafloor strainmeters, synthesized strain and observations showed almost the same waveforms. Correlation coefficients were over 0.8 and the residuals in energy were less than 30%. For the borehole strainmeter, most of the phases matched well except the surface wave of the earthquake in Taiwan. In the phase, several packets did not seen in the synthesized strain. Thus, the residual in energy was 60%. Possible causes are effects of the sensor structure and geological properties around the borehole observatory. Since asymmetric structures exist in the sensor, azimuthal dependency in sensitivity might be created. The sedimentary layer around the borehole observatory is laid in the EW direction in an accretionary prism. This can also affect propagation of surface wave.
These results demonstrate the potential for quantitative comparison between strain fields estimated by a seismometer array and strain records obtained through fiber sensing. This approach could allow for amplitude comparisons of DAS records, although effects of the surrounding media were suggested by the results from the borehole observatory.
One method for comparison is the estimation of the strain tensor through array analysis, using more than three stations with three-component seismometers (Spudich et al., 1995). Array processing can resolve the strain tensor from the displacement at each station, obtained by integrating the velocity records. Therefore, if the strainmeters are located within or near the array, strain records can be compared with synthesized strains produced by the array processing.
We applied this method to three linear fiber optic strainmeters on the seafloor and one borehole fiber optic strainmeter located 450 meters below the seafloor. These strainmeters were installed around Node-2F of DONET2, off the Kii Channel. Each seafloor strainmeter uses a 200 m fiber optic cable extended on the seafloor as a strain sensor. Normal strain in the extended direction is observed. Thus, the strain records were compared to the component in the synthesized strain tensor. Two of the three were extended in the same direction and the other was extended 60 degrees from the direction of the others. The borehole strainmeter uses fiber optic cables wrapped around a stainless steel mandrel like a coil to install a long-baseline sensor into the limited space of the borehole. Thus, strain used for the comparison was calculated based on the strain tensor and geometrical shape of the sensor fiber after Tsuji et al. (2024, JpGU). In addition, the coefficient of sensitivity of the sensor 0.58 (Tsuji et al. 2024, SSJ) was applied. This coefficient accounts for the effects of surrounding structures during the installation process on the strain measurements.
To perform the comparison, we synthesized the strain field of teleseismic signals using four broadband seismometers of the DONET2 around the strainmeters. Two earthquakes were selected based on signal-to-noise ratios: the M7.4 earthquake in Taiwan on April 3rd, 2024, and the M7.3 earthquake in Vanuatu on December 17th, 2024. Since the maximum distance between the stations was approximately 20 km, a bandpass filter with a range of 20 to 100 seconds was applied.
As a result, most phases in the strain records were obtained in the synthesized strain waveforms with almost equal amplitude. S phase and surface waves were major phases due to the sensor characteristics and frequency ranges of the band pass filter. For the seafloor strainmeters, synthesized strain and observations showed almost the same waveforms. Correlation coefficients were over 0.8 and the residuals in energy were less than 30%. For the borehole strainmeter, most of the phases matched well except the surface wave of the earthquake in Taiwan. In the phase, several packets did not seen in the synthesized strain. Thus, the residual in energy was 60%. Possible causes are effects of the sensor structure and geological properties around the borehole observatory. Since asymmetric structures exist in the sensor, azimuthal dependency in sensitivity might be created. The sedimentary layer around the borehole observatory is laid in the EW direction in an accretionary prism. This can also affect propagation of surface wave.
These results demonstrate the potential for quantitative comparison between strain fields estimated by a seismometer array and strain records obtained through fiber sensing. This approach could allow for amplitude comparisons of DAS records, although effects of the surrounding media were suggested by the results from the borehole observatory.