10:15 AM - 10:30 AM
[SVC30-06] Source-based seismic amplitude correction for determining source locations of Volcano-Tectonic events at Izu-Oshima volcano
Keywords:Seismic amplitude correction, Volcano Earthquake, Source location determination, ASL
Locating the source of volcanic earthquakes and tremors is one of the most important aspects in monitoring volcanic activity. One of the methods to locate the source is the Amplitude Source Location (ASL) method, which assumes the seismic amplitude decays due to geometrical spreading and intrinsic attenuation (Battaglia and Aki, 2003) in a homogeneous structure. This method has been widely used to locate volcano-tectonic (VT) events, tremors, rockfalls, and lahars. Because the observed waveform amplitude is much affected by the site amplification factor, the coda normalization method is often used. However, our preliminary analysis of earthquakes at Izu-Oshima using ASL did not work well, probably because the internal structure of the volcano is heterogeneous and/or other local effects. In the present study, we present a new site amplification correction method to better determine the earthquakes using ASL by examining VT events that occurred in Izu-Oshima island.
We use 74 events of VT events at Izu-Oshima volcano that were located within the seismic network coverage of JMA (4 stations) and NIED (4 stations) in the period from 2014 to 2021. We analyze the vertical component records at a frequency band of 8-16 Hz. The hypocenters of those VTs were previously determined from the P- and S-wave arrival times. We then divide those events into several clusters, each of which consists of at least 2 VTs within a grid area of 0.01x0.01 degree. We estimate the attenuation parameter (Q) by applying the curve fitting to the observed amplitude, and the amplification correction factor (AF) for each station is then calculated based on the comparison of the theoretical and the normalized observed amplitudes. That is, Q and the site amplification factors are determined for the grids. Applying an exponential variogram model to the obtained values at the grid points, we generate a map of the amplification correction factor (AF) and Q distribution map. As a result, the site amplification factor at each station and Q change with source location.
We select 32 events as reference events to determine the AF and Q. Reference events are determined by ASL, in which the maximum amplitude of averaged envelopes and a constant S-wave velocity (2.8 km/s) are used. We grid search the source location that shows the minimum residual on the 3D grid points with a spacing of 0.001° in latitude and longitude, and 0.1 km in depth, respectively. The difference (misfit) for the reference events shows about 1 km misfit in average both in lateral and vertical directions. We further determine the other 42 events, called as the test data. The results show that the misfits are 1.3 and 2 km in lateral and vertical directions, respectively. Horizontal distribution shifts significantly and is better correlated to the reference compared to the source locations determined by ASL using site amplification factors determined from the coda normalization. Thus, it is important to utilize local events within the research area for waveform amplitude correction, which better represents the local medium condition.
We use 74 events of VT events at Izu-Oshima volcano that were located within the seismic network coverage of JMA (4 stations) and NIED (4 stations) in the period from 2014 to 2021. We analyze the vertical component records at a frequency band of 8-16 Hz. The hypocenters of those VTs were previously determined from the P- and S-wave arrival times. We then divide those events into several clusters, each of which consists of at least 2 VTs within a grid area of 0.01x0.01 degree. We estimate the attenuation parameter (Q) by applying the curve fitting to the observed amplitude, and the amplification correction factor (AF) for each station is then calculated based on the comparison of the theoretical and the normalized observed amplitudes. That is, Q and the site amplification factors are determined for the grids. Applying an exponential variogram model to the obtained values at the grid points, we generate a map of the amplification correction factor (AF) and Q distribution map. As a result, the site amplification factor at each station and Q change with source location.
We select 32 events as reference events to determine the AF and Q. Reference events are determined by ASL, in which the maximum amplitude of averaged envelopes and a constant S-wave velocity (2.8 km/s) are used. We grid search the source location that shows the minimum residual on the 3D grid points with a spacing of 0.001° in latitude and longitude, and 0.1 km in depth, respectively. The difference (misfit) for the reference events shows about 1 km misfit in average both in lateral and vertical directions. We further determine the other 42 events, called as the test data. The results show that the misfits are 1.3 and 2 km in lateral and vertical directions, respectively. Horizontal distribution shifts significantly and is better correlated to the reference compared to the source locations determined by ASL using site amplification factors determined from the coda normalization. Thus, it is important to utilize local events within the research area for waveform amplitude correction, which better represents the local medium condition.