2:15 PM - 2:30 PM
[SVC25-03] Source location of volcanic earthquakes using ASL method in velocity structure having a constant vertical gradient
Keywords:Volcano Earthquake, Source location determination, ASL, ASL-VG
One of the most important aspects in monitoring volcanic activity is locating the source of volcanic earthquakes and tremor. A standard method uses P- and S-wave arrival times, but the method may not work well for seismic swarm consisting of continuous occurrence of VTs, low-frequency earthquakes, and/or sustained volcanic tremors which often do not show clear P- or S-wave arrivals. To overcome this problem, the amplitude source location (ASL) method using amplitude decay due to geometrical spreading and intrinsic attenuation is developed (Battaglia and Aki, 2003), and has been used in many active volcanoes. The method presumes a homogeneous structure, but it is realistic to consider the velocity changes with depths at volcanic areas. In the present study, we develop a new ASL method that works for the structure with seismic velocity increasing with depth with a constant velocity gradient (ASL-VG).
In this study, we compare the hypocenters determined from the ASL and ASL-VG methods. We use continuous seismic records of JMA stations at Sakurajima volcano when a seismic swarm occurred in the period from 08h00m to 22h00m JST on August 15th, 2015. We analyze the vertical component records at two frequency bands (4–8 Hz and 8–16 Hz) from 6 stations deployed on the Sakurajima volcano’s edifice.
We first determine the site amplification factor using the coda normalization method (Phillips and Aki, 1986). Correcting the observed amplitude with the estimated site amplification factors, we calculate the theoretical amplitudes for ASL and ASL-VG methods with intrinsic attenuation Q factors ranging from 25 to 100. In the ASL method, we assume the S-wave velocity is 2 km/s. In the ASL-VG method, we use the seismic velocity structure that is approximated from Miyamachi et al. (2013), in which S-wave velocity increases from 1.330 km/s at the ground surface with a vertical gradient of 0.4475 s-1.
We calculate the amplitude residual on 3D grids with a spacing of 0.001° in latitude and longitude, and 0.1 km in depth, respectively. We compare the 52 VT source locations estimated from ASL and ASL-VG methods with the hypocenters determined from P- and S-wave arrival times. The difference (misfit) shows about 1-2 km misfit in average both in lateral and vertical directions. Smaller misfit was obtained in the 8–16 Hz frequency band. The amplitude difference (observed-calculated) of ASL-VG shows smaller values than the ASL method. This result indicates better approximation is obtained by assuming the vertical velocity gradient in the ASL-VG method.
In this study, we compare the hypocenters determined from the ASL and ASL-VG methods. We use continuous seismic records of JMA stations at Sakurajima volcano when a seismic swarm occurred in the period from 08h00m to 22h00m JST on August 15th, 2015. We analyze the vertical component records at two frequency bands (4–8 Hz and 8–16 Hz) from 6 stations deployed on the Sakurajima volcano’s edifice.
We first determine the site amplification factor using the coda normalization method (Phillips and Aki, 1986). Correcting the observed amplitude with the estimated site amplification factors, we calculate the theoretical amplitudes for ASL and ASL-VG methods with intrinsic attenuation Q factors ranging from 25 to 100. In the ASL method, we assume the S-wave velocity is 2 km/s. In the ASL-VG method, we use the seismic velocity structure that is approximated from Miyamachi et al. (2013), in which S-wave velocity increases from 1.330 km/s at the ground surface with a vertical gradient of 0.4475 s-1.
We calculate the amplitude residual on 3D grids with a spacing of 0.001° in latitude and longitude, and 0.1 km in depth, respectively. We compare the 52 VT source locations estimated from ASL and ASL-VG methods with the hypocenters determined from P- and S-wave arrival times. The difference (misfit) shows about 1-2 km misfit in average both in lateral and vertical directions. Smaller misfit was obtained in the 8–16 Hz frequency band. The amplitude difference (observed-calculated) of ASL-VG shows smaller values than the ASL method. This result indicates better approximation is obtained by assuming the vertical velocity gradient in the ASL-VG method.