11:00 AM - 1:00 PM
[SCG54-P01] Source process of initial to later phases and spatial characteristics of deep low-frequency earthquakes beneath Zao volcano
★Invited Papers
Keywords:Zao, deep low-frequency earthquake, source process
Deep low-frequency earthquakes (DLFs) beneath volcanoes are possible evidence for deep-seated magmatic activity in the mid-to-lower crust and uppermost mantle. After the 2011 Tohoku-oki earthquake, the number of DLFs beneath Zao volcano increased. The hypocenters of the DLFs form two clusters at shallower (20–28 km) and deeper (28–38 km) depths located near the side and lower parts of the high Vp/Vs zone, respectively (Okada et al., 2015). Ikegaya and Yamamoto (2021) classified the DLFs detected by the matched filter method into the seven groups (A: deeper cluster, B–G: shallower cluster) and found the focal mechanisms composed of dominant isotropic and small double-couple (DC) and CLVD components by using the S/P spectral ratio. In this presentation, based on the temporal evolution of the S/P ratio in the initial to later phases, we will discuss the source process of DLFs, and the spatial characteristics of different types of DLFs.
We used waveform data in the 1.5–3 Hz band at four stations, operated by NIED and Tohoku Univ. For the seven groups of the DLFs (Groups A–G), we manually picked the P and S wave arrival times of the events with the signal-to-noise ratios larger than two and used 3 s time windows from the onset of P and S waves in vertical and horizontal components, respectively. For the squared amplitude, by taking the time integral of the 0.5 s moving average, the temporal evolution of the S/P cumulative energy ratio (S/P ratio) was estimated for 157 DLFs. For comparison, we also calculated the S/P ratios for 142 ordinary crustal earthquakes in northeastern Japan.
We revealed that the value of the S/P ratio for the DLFs tended to reach a maximum value near the onset (0–1 s), then decrease and increase again or become constant. This behavior of the S/P ratio contrasts with the monotonous decrease for the early coda wave of ordinary earthquakes. For 86 DLFs (54.8%), the S/P ratio near the onset was higher than that from the entire waveform at multiple stations. The temporal change of the S/P ratio was robust for 51 DLFs (32.5%, hereafter referred to as Initial-High-S/P DLFs, IH-DLFs) even considering the uncertainty of the S/P ratio estimation caused by the phase picking error. The change rate of the S/P ratio at 2–3 s showed a statistically significant difference between IH-DLFs and ordinary earthquakes; IH-DLFs exhibited the larger increase rate of the S/P ratio at least at two stations.
The temporal change of the S/P ratio for IH-DLFs suggests a tensile-shear model. We thus reproduced the initial S/P ratio based on DC and tensile crack components in the focal mechanism of the entire waveform. We found the fraction of DC component for the initial phase of group A was 20–50% and that of B–G was 10–40%, which were larger than that obtained from the entire waveform. This may indicate that the source process of DLFs is a tensile-shear mechanism consisting of initial shear faulting and following volumetric change. Observed re-increase of the S/P ratio in the later phase of the DLFs may be attributed to the P-to-S conversion scattering from long-lasting non-DC components. To clarify the spatial distribution of IH-DLFs, we first examined the percentage of IH-DLFs in all DLFs detected by each template in the matched filter analysis. As a result, among sub-groups classified by hierarchical clustering of the templates, we identified sub-groups having high (56–72%) fractions of IH-DLFs (groups AH and SH) and low (5–10%) fractions (AL and SL). We further revealed that the hypocenters of AH and AL in the deeper cluster and SH and SL in the shallower cluster were robustly separated, even considering the errors in the hypocenter determination. This fact suggests that the source processes of DLFs are spatially segregated in a source cluster of <10 km scale.
Our further numerical simulations of the dynamic interaction between dikes and faults may contribute to understanding the source process of DLFs.
We used waveform data in the 1.5–3 Hz band at four stations, operated by NIED and Tohoku Univ. For the seven groups of the DLFs (Groups A–G), we manually picked the P and S wave arrival times of the events with the signal-to-noise ratios larger than two and used 3 s time windows from the onset of P and S waves in vertical and horizontal components, respectively. For the squared amplitude, by taking the time integral of the 0.5 s moving average, the temporal evolution of the S/P cumulative energy ratio (S/P ratio) was estimated for 157 DLFs. For comparison, we also calculated the S/P ratios for 142 ordinary crustal earthquakes in northeastern Japan.
We revealed that the value of the S/P ratio for the DLFs tended to reach a maximum value near the onset (0–1 s), then decrease and increase again or become constant. This behavior of the S/P ratio contrasts with the monotonous decrease for the early coda wave of ordinary earthquakes. For 86 DLFs (54.8%), the S/P ratio near the onset was higher than that from the entire waveform at multiple stations. The temporal change of the S/P ratio was robust for 51 DLFs (32.5%, hereafter referred to as Initial-High-S/P DLFs, IH-DLFs) even considering the uncertainty of the S/P ratio estimation caused by the phase picking error. The change rate of the S/P ratio at 2–3 s showed a statistically significant difference between IH-DLFs and ordinary earthquakes; IH-DLFs exhibited the larger increase rate of the S/P ratio at least at two stations.
The temporal change of the S/P ratio for IH-DLFs suggests a tensile-shear model. We thus reproduced the initial S/P ratio based on DC and tensile crack components in the focal mechanism of the entire waveform. We found the fraction of DC component for the initial phase of group A was 20–50% and that of B–G was 10–40%, which were larger than that obtained from the entire waveform. This may indicate that the source process of DLFs is a tensile-shear mechanism consisting of initial shear faulting and following volumetric change. Observed re-increase of the S/P ratio in the later phase of the DLFs may be attributed to the P-to-S conversion scattering from long-lasting non-DC components. To clarify the spatial distribution of IH-DLFs, we first examined the percentage of IH-DLFs in all DLFs detected by each template in the matched filter analysis. As a result, among sub-groups classified by hierarchical clustering of the templates, we identified sub-groups having high (56–72%) fractions of IH-DLFs (groups AH and SH) and low (5–10%) fractions (AL and SL). We further revealed that the hypocenters of AH and AL in the deeper cluster and SH and SL in the shallower cluster were robustly separated, even considering the errors in the hypocenter determination. This fact suggests that the source processes of DLFs are spatially segregated in a source cluster of <10 km scale.
Our further numerical simulations of the dynamic interaction between dikes and faults may contribute to understanding the source process of DLFs.