5:15 PM - 7:15 PM
[SSS12-P04] Differences in Waveforms of off-Kesennuma Repeating Earthquakes in High-Frequency Bands
Keywords:repeating earthquakes, waveform cross-correlation, similar earthquakes
Repeating earthquakes occur in nearly the same location with similar mechanisms and exhibit almost identical seismic waveforms. They have been identified in various tectonic environments, including the plate boundary in the subduction zone off the Pacific coast of Tohoku. A catalog of repeating earthquakes across the Japanese archipelago has been compiled. Igarashi (2020) utilized the maximum of cross-correlation function (i.e., the correlation coefficient, CC) for pairs of different earthquakes in their catalog. For identifying repeating earthquakes of M3 or larger, the frequency band of 1-4 Hz was used, and CCs for frequencies above 4 Hz have not been revealed. Detailed analyses of the high-frequency band may enable us to investigate events within the same repeating earthquake sequence based on differences in source mechanisms and temporal changes. In this study, we performed waveform cross-correlation calculations using multiple frequency bands, including higher frequencies. We examined pairs within the same repeating earthquake sequence that exhibited distinctive results in both high- and low-frequency bands.
We analyzed 74 events occurring in a region offshore of Kesennuma, Miyagi Prefecture, which includes three repeating earthquake sequences. We used three-component waveforms from the Hi-net Onoda station, approximately 100 km from their hypocenters, ensuring that all 74 events are observed. The waveforms were bandpass-filtered using octave-wide frequency bands from 1-2 Hz to 16-32 Hz, and cross-correlation calculations were performed. Based on the distribution of CCs in the 2-4 Hz among all pairs, we classified all pairs into three groups: pairs with CC of 2.7 or higher, pairs with CC between 2.0 and 2.7, and pairs with CC below 2.0. Among these, pairs with CC of 2.7 or higher were exclusively from the same repeating earthquake sequence. However, in some pairs, the total CC across three components decreased by approximately 0.5 in the 1-2 Hz band compared to the 2-4 Hz band. In the 16-32 Hz band, CCs varied widely, ranging from nearly 3.0 to approximately 0.5. We examined the pairs with relatively low CCs in 1-2 Hz or 16-32 Hz regarding hypocenter locations, magnitude differences, and spectral characteristics. All pairs that exhibited the low CCs in the 1-2 Hz band included a specific event. It was found that another event, which occurred off Fukushima approximately 200 km from the Hi-net Onoda station just before this specific event, caused this reduction. The far-field event affected only the 1-2 Hz band. We also examined pairs with decreased CC in the 16–32 Hz band and found that these pairs were associated with four events. Comparing the four events with others, their spectral shapes were different, or the signal-to-noise ratio in the high-frequency band was lower. This result suggests differences in source processes among repeating earthquakes. Furthermore, in certain repeating earthquake sequences, the high-frequency characteristics of earlier events differed from those of later events. The effects of magnitude differences on CC in the 16-32 Hz band were not evident.
The catalog by Igarashi (2020) contains approximately 10,000 repeating earthquake sequences, leaving ample opportunity for similar analyses. By analyzing a larger number of repeating earthquake sequences, it is possible to investigate whether similar results are observed across different regions and sequences, thereby clarifying the characteristics of individual sequences.
We analyzed 74 events occurring in a region offshore of Kesennuma, Miyagi Prefecture, which includes three repeating earthquake sequences. We used three-component waveforms from the Hi-net Onoda station, approximately 100 km from their hypocenters, ensuring that all 74 events are observed. The waveforms were bandpass-filtered using octave-wide frequency bands from 1-2 Hz to 16-32 Hz, and cross-correlation calculations were performed. Based on the distribution of CCs in the 2-4 Hz among all pairs, we classified all pairs into three groups: pairs with CC of 2.7 or higher, pairs with CC between 2.0 and 2.7, and pairs with CC below 2.0. Among these, pairs with CC of 2.7 or higher were exclusively from the same repeating earthquake sequence. However, in some pairs, the total CC across three components decreased by approximately 0.5 in the 1-2 Hz band compared to the 2-4 Hz band. In the 16-32 Hz band, CCs varied widely, ranging from nearly 3.0 to approximately 0.5. We examined the pairs with relatively low CCs in 1-2 Hz or 16-32 Hz regarding hypocenter locations, magnitude differences, and spectral characteristics. All pairs that exhibited the low CCs in the 1-2 Hz band included a specific event. It was found that another event, which occurred off Fukushima approximately 200 km from the Hi-net Onoda station just before this specific event, caused this reduction. The far-field event affected only the 1-2 Hz band. We also examined pairs with decreased CC in the 16–32 Hz band and found that these pairs were associated with four events. Comparing the four events with others, their spectral shapes were different, or the signal-to-noise ratio in the high-frequency band was lower. This result suggests differences in source processes among repeating earthquakes. Furthermore, in certain repeating earthquake sequences, the high-frequency characteristics of earlier events differed from those of later events. The effects of magnitude differences on CC in the 16-32 Hz band were not evident.
The catalog by Igarashi (2020) contains approximately 10,000 repeating earthquake sequences, leaving ample opportunity for similar analyses. By analyzing a larger number of repeating earthquake sequences, it is possible to investigate whether similar results are observed across different regions and sequences, thereby clarifying the characteristics of individual sequences.