Crack Source Estimation for Long-Period Events Using Low-Frequency Seismic Amplitudes: Application at Kusatsu-Shirane Volcano, Japan

Long-period (LP) events observed at active volcanoes are interpreted as oscillations of fluid-filled resonators in magma and hydrothermal systems. The fluid-filled crack model has been used to interpret LP waveforms. However, crack source mechanisms have been estimated only for specific events by using full-waveform inversion in previous studies. The limited LP source estimates may originate in the difficulty to fit LP oscillatory waveforms by full-waveform inversion and the requirement of computational technical skills. To solve these problems, we developed the following simple method using only seismic amplitudes to obtain stable inversion results. We calculated envelope waveforms band-passed between 1 and 2 Hz at individual stations, in which observed amplitudes were corrected for site amplification factors estimated by the coda normalization method using regional earthquakes. In each 3-s sliding time window, we performed a spatial grid search to identify a grid position showing the maximum averaged observed amplitude, where the best-fit crack mechanism is estimated by fitting observed amplitudes to those of synthetic waveforms calculated with the discrete wavenumber method in a homogeneous half-space. The position and mechanism giving the minimum residual among different time windows were adopted as the LP source location and mechanism. We applied this inversion method to an LP event that occurred on 26 April 1992 at Kusatsu-Shirane volcano. Our result indicates that the estimated crack source is similar to that estimated from full-waveform inversion by Nakano and Kumagai (GRL, 2005). Our inversion results of other LP events in AprilNovember 1992, of which source mechanisms have not been estimated, point to similar locations and crack mechanisms, indicating that these events were excited by the same crack. Taguchi et al. (JGR, 2018) considered that LP events in this period were generated by a single crack, into which steam generated by degassing from magma supplied repeatedly. Therefore, our results support this LP source model. Our method is regarded as an extension of the amplitude source location (ASL) method to estimate the source mechanism using low-frequency amplitudes. This extended ASL method may be applicable to estimate crack sources of LP events observed at other active volcanoes.