09:30 〜 09:45
[SSS09-03] 歪み蓄積の推定を目指したパークフィールドにおける速度構造変化の常時観測
キーワード：微動、速度変化、パークフィールド
Fault zones are subject to the long-term loading of tectonic strains, to the short-term release of fault fast and slow earthquakes, and to hydrological and thermoelastic modulation from environmental forces. Each of these stresses are thought to impact the fault-zone material properties, such as elastic wave speed. In this work, we perform the continuous monitoring using nearly two decades (2002-2020) of ambient seismic noise recorded by 3-components seismometers from the Berkeley Seismological Laboratory’s High Resolution Seismic Network (HRSN) to extract bulk properties, dv/v, near the rupture area of the 2004 Parkfield Earthquake on the San Andreas Fault.
Transient signals such as earthquakes and tremor are removed using a kurtosis and STA/LTA based algorithm, prior to computing the cross-correlation functions (CCFs). We compute the CCFs for all possible channel pairs and decompose them into narrow frequency bands using wavelet transform. During the stacking processes, we reject CCFs with low signal-to-noise ratio caused by the perturbation of noise sources and instrumental issues. The daily CCFs are stacked every 30 days with 15 days overlap. We explore the effects of various algorithms and waveform reference on our measurements.
This study aims to separate the potential tectonic and non-tectonic origins of the observed temporal variations in dv/v and to evaluate the effect of strain accumulation on its time history using geodetic measurements. The dv/v time history reveals the previously identified coseismic decrease in dv/v and the subsequent logarithmic recovery, as well as a newly observed long-term increase in dv/v. We show that the observed long-term increase exceeds the velocity healing model which is thought to explain the logarithmic postseimic behavior. We further show that this long-term increase cannot be explained by changes in ground water levels and does not follow the trend of decrease and recovery associated with the drought in California from 2011 to 2017. Instead, we discuss that the observed long-term increase in dv/v reflects strain change from quasi-static deformation at depth during the interseismic period.
Transient signals such as earthquakes and tremor are removed using a kurtosis and STA/LTA based algorithm, prior to computing the cross-correlation functions (CCFs). We compute the CCFs for all possible channel pairs and decompose them into narrow frequency bands using wavelet transform. During the stacking processes, we reject CCFs with low signal-to-noise ratio caused by the perturbation of noise sources and instrumental issues. The daily CCFs are stacked every 30 days with 15 days overlap. We explore the effects of various algorithms and waveform reference on our measurements.
This study aims to separate the potential tectonic and non-tectonic origins of the observed temporal variations in dv/v and to evaluate the effect of strain accumulation on its time history using geodetic measurements. The dv/v time history reveals the previously identified coseismic decrease in dv/v and the subsequent logarithmic recovery, as well as a newly observed long-term increase in dv/v. We show that the observed long-term increase exceeds the velocity healing model which is thought to explain the logarithmic postseimic behavior. We further show that this long-term increase cannot be explained by changes in ground water levels and does not follow the trend of decrease and recovery associated with the drought in California from 2011 to 2017. Instead, we discuss that the observed long-term increase in dv/v reflects strain change from quasi-static deformation at depth during the interseismic period.