5:15 PM - 6:45 PM
[SSS07-P08] Spatio-temporal analysis of amplitude variation of microseisms over a long period of time
Keywords:seismic noise, microseisms
The solid earth constantly vibrates even without fault motion, which has been recorded continuously by high-sensitivity seismographs as seismic noise. The seismic noise is mainly generated by interaction between solid earth and meteorological phenomena, as well as by human activities. In particular, seismic waves excited by ocean waves dominant at 0.02-1.0 Hz are called microseisms.
Since it has been more than 20 years after development of the dense high-sensitivity seismic observation network Hi-net in Japan, continuous records of long-term seismic waveforms have been accumulated. In this study, we analyzed the characteristics of seismic noise throughout Japan over about 13 years using seismic waveform records of Hi-net.
This study analyzed the data processed by Maeda (2022). The raw continuous three-component velocity traces of a Hi-net station were first corrected for instrumental response, band-pass filtered, and decimated to one-minute sampling by taking median of the RMS amplitude. Then, median of night-time data between 22:00 to 8:00 JST was adopted as the characteristic amplitude of the day at the station. We analyzed three frequency bands of 0.04-0.1 Hz, 0.10-0.25 Hz, 0.25-1.0 Hz among seven frequency bands in the original dataset.
First, we focused on the long-term periodicity of the observed amplitude. Assuming annual, six-month, and four-month periodicities are dominant in the observed amplitude record at a station, we estimated their amplitudes and phases as an inverse problem. Since periodic fluctuations that dominant at each station would be different, we selected the optimal model among the combination of the periodic oscillations including constant by using the AIC (Akaike Information Criterion).
The results showed that the annual periodic variation was dominant in the low frequency band below 1.0 Hz at throughout Japan. There was a clear regional variation in the amplitude values; larger on the Sea of Japan side and weaker on the Pacific side. The phase changes continuously in space, and its direction of change varied with frequency. The amplitudes of variation with six and four-month periods were also considerably large, particularly in the coastal area. As a result, most of stations selected the model with annual, six- and four-months periodic change as their best-fit model. The behavior of the phase of six- and four-month periods changed significantly across the inland stations where the AIC estimated that the model with only annual variation was optimal. It was found that the phase changed continuously from the Sea of Japan side and the Pacific side to inland, respectively.
At 0.1-0.25 Hz, regions with large amplitude values of annual change matched well with locations of sources of strong microseisms. On the other hand, at 0.25-1.0 Hz, the amplitude values did not correlate with the sources. Seismic noise is considered to originate from microseisms sources, but it also would be affected by wave height and wind. Additionally, seismic noise with period of six and four months, which propagated from the Pacific- and the Japan Sea side, may interfere around inland. This interference would be a reason why the stations for which a model with only annual variation were selected via the AIC were found inland.
Since it has been more than 20 years after development of the dense high-sensitivity seismic observation network Hi-net in Japan, continuous records of long-term seismic waveforms have been accumulated. In this study, we analyzed the characteristics of seismic noise throughout Japan over about 13 years using seismic waveform records of Hi-net.
This study analyzed the data processed by Maeda (2022). The raw continuous three-component velocity traces of a Hi-net station were first corrected for instrumental response, band-pass filtered, and decimated to one-minute sampling by taking median of the RMS amplitude. Then, median of night-time data between 22:00 to 8:00 JST was adopted as the characteristic amplitude of the day at the station. We analyzed three frequency bands of 0.04-0.1 Hz, 0.10-0.25 Hz, 0.25-1.0 Hz among seven frequency bands in the original dataset.
First, we focused on the long-term periodicity of the observed amplitude. Assuming annual, six-month, and four-month periodicities are dominant in the observed amplitude record at a station, we estimated their amplitudes and phases as an inverse problem. Since periodic fluctuations that dominant at each station would be different, we selected the optimal model among the combination of the periodic oscillations including constant by using the AIC (Akaike Information Criterion).
The results showed that the annual periodic variation was dominant in the low frequency band below 1.0 Hz at throughout Japan. There was a clear regional variation in the amplitude values; larger on the Sea of Japan side and weaker on the Pacific side. The phase changes continuously in space, and its direction of change varied with frequency. The amplitudes of variation with six and four-month periods were also considerably large, particularly in the coastal area. As a result, most of stations selected the model with annual, six- and four-months periodic change as their best-fit model. The behavior of the phase of six- and four-month periods changed significantly across the inland stations where the AIC estimated that the model with only annual variation was optimal. It was found that the phase changed continuously from the Sea of Japan side and the Pacific side to inland, respectively.
At 0.1-0.25 Hz, regions with large amplitude values of annual change matched well with locations of sources of strong microseisms. On the other hand, at 0.25-1.0 Hz, the amplitude values did not correlate with the sources. Seismic noise is considered to originate from microseisms sources, but it also would be affected by wave height and wind. Additionally, seismic noise with period of six and four months, which propagated from the Pacific- and the Japan Sea side, may interfere around inland. This interference would be a reason why the stations for which a model with only annual variation were selected via the AIC were found inland.