9:30 AM - 9:45 AM
[S02-03] Novel Method of Love Wave Detection and FK Analysis Using Distributed Acoustic Sensing (DAS) Data from the Sakurajima Network
Distributed Acoustic Sensing (DAS) has transformed seismic monitoring by providing high-resolution data over extensive distances, making it an invaluable tool for studying subsurface structures and seismic wavefields. However, applying frequency-wavenumber (FK) analysis to DAS data is challenging due to its directional sensitivity and the complex nature of seismic waveforms. Despite these challenges, DAS holds significant potential, particularly in fields such as oil and gas exploration, earthquake research, and environmental monitoring.
Our research introduces a novel method for the automatic detection of Love waves in DAS data. By integrating FK analysis, we accurately determine the slowness and azimuth of these waves. DAS captures horizontal strain data, which includes both Rayleigh and Love waves. A key observation in our study is that a 90-degree bend in the DAS cable causes a polarity flip in Love waves. This feature enables us to develop an automatic detection method that bypasses the need for manual inspection. Importantly, this method can be applied to other DAS data, provided there is a 90-degree bend along the cable.
We applied this method to ambient noise data collected between November and December 2022 from DAS observations at Sakurajima, one of Japan's most active volcanoes in Kyushu. The DAS cable, which is 4.4 km long with channel intervals of 4.79 meters, runs alongside the Nojiri River, and features a 90-degree bend at channel 250. Our analysis begins with cross-correlation between each channel and the average of all channels before the bend (50 channels), as well as between each channel and the average of all channels after the bend (50 channels). The average cross-correlation coefficient (CCC) serves as a critical indicator of data quality. If the quality is insufficient, we move to the next time window until we obtain good quality (coherent) data. We then examine CCC values between channel segments within the same time window, both before and after the bend, to identify polarity flips by detecting high inverse correlation with minimal time lag, which signals the presence of Love waves. Upon detecting these waves, we perform FK analysis to accurately determine slowness and azimuth. To address directional sensitivity issues in FK analysis, we adapt the polarity flip method used by Zhao et al. (2023) for MUSIC (Multiple Signal Classification) beamforming. Analyzing consecutive 10-second windows with an overlap of 5-seconds allows for precise localization of Love waves in the Sakurajima DAS data. We also analyzed Love waves across various frequency bands, enhancing our understanding of their characteristics.
To improve detection accuracy and reduce false positives, we developed an image processing technique that verifies detected polarity flips by comparing pixel color intensities of plotted waveforms before and after the cable bend. This ensures the reliability of our Love wave detection method and removes any false positives automatically.
Our analysis identified distinctive Love wave patterns within the 0.4 to 0.5 Hz frequency range. FK analysis revealed that these waves predominantly propagate from the northeast and southwest of Sakurajima, with apparent velocities ranging from 0.5 to 1 km/s. We hypothesize that this propagation pattern is influenced by the northeast monsoon winds and the complex subsurface structure around Kyushu, contributing to the generation of microseisms.
In summary, our research introduces a novel method for the automatic detection and analysis of Love waves in DAS data. This method enhances seismic monitoring capabilities and offers valuable insights for practical applications in various fields. By improving wave detection accuracy and providing a deeper understanding of seismic wavefields, our study significantly advances seismic research and monitoring.
Our research introduces a novel method for the automatic detection of Love waves in DAS data. By integrating FK analysis, we accurately determine the slowness and azimuth of these waves. DAS captures horizontal strain data, which includes both Rayleigh and Love waves. A key observation in our study is that a 90-degree bend in the DAS cable causes a polarity flip in Love waves. This feature enables us to develop an automatic detection method that bypasses the need for manual inspection. Importantly, this method can be applied to other DAS data, provided there is a 90-degree bend along the cable.
We applied this method to ambient noise data collected between November and December 2022 from DAS observations at Sakurajima, one of Japan's most active volcanoes in Kyushu. The DAS cable, which is 4.4 km long with channel intervals of 4.79 meters, runs alongside the Nojiri River, and features a 90-degree bend at channel 250. Our analysis begins with cross-correlation between each channel and the average of all channels before the bend (50 channels), as well as between each channel and the average of all channels after the bend (50 channels). The average cross-correlation coefficient (CCC) serves as a critical indicator of data quality. If the quality is insufficient, we move to the next time window until we obtain good quality (coherent) data. We then examine CCC values between channel segments within the same time window, both before and after the bend, to identify polarity flips by detecting high inverse correlation with minimal time lag, which signals the presence of Love waves. Upon detecting these waves, we perform FK analysis to accurately determine slowness and azimuth. To address directional sensitivity issues in FK analysis, we adapt the polarity flip method used by Zhao et al. (2023) for MUSIC (Multiple Signal Classification) beamforming. Analyzing consecutive 10-second windows with an overlap of 5-seconds allows for precise localization of Love waves in the Sakurajima DAS data. We also analyzed Love waves across various frequency bands, enhancing our understanding of their characteristics.
To improve detection accuracy and reduce false positives, we developed an image processing technique that verifies detected polarity flips by comparing pixel color intensities of plotted waveforms before and after the cable bend. This ensures the reliability of our Love wave detection method and removes any false positives automatically.
Our analysis identified distinctive Love wave patterns within the 0.4 to 0.5 Hz frequency range. FK analysis revealed that these waves predominantly propagate from the northeast and southwest of Sakurajima, with apparent velocities ranging from 0.5 to 1 km/s. We hypothesize that this propagation pattern is influenced by the northeast monsoon winds and the complex subsurface structure around Kyushu, contributing to the generation of microseisms.
In summary, our research introduces a novel method for the automatic detection and analysis of Love waves in DAS data. This method enhances seismic monitoring capabilities and offers valuable insights for practical applications in various fields. By improving wave detection accuracy and providing a deeper understanding of seismic wavefields, our study significantly advances seismic research and monitoring.