Temporal Variations of Seismic Wave Velocity Associated with 1998 M6.1 Shizukuishi Earthquake

We attempt to detect temporal variations of seismic wave velocity before and after 1998 M6.1 Shizukuishi, northeastern Japan, earthquake by using waveform data from explosions and earthquake doublets spanning the period immediately before and after the earthquake. Direct P waves of the second explosion are delayed by 20 ms at observation stations with epicentral distances less than 15 km. This tendency does not change if the analysis frequency band is changed. Our result suggests that the P-wave velocity decreased by at least 1% in the extremely shallow region of the hanging wall of the M6.1 thrust event after its occurrence. On the other hand, there was the frequency dependence of the coda wave delays for both artificial sources and for natural events. At 5–10 Hz, immediate sharp increases by more than 20 ms in time delays and lower coherency were observed at several stations. We estimated the region in which P-wave velocity might have decreased after the M6.1 earthquake. Maximum depth of the region is 13 km. The region includes the aftershock area of the M6.1 earthquake, but is eccentric to the earthquake in the west. Considering the frequency band analyzed (5–10 Hz), the scale of the spatial inhomogeneity which led to the coda wave delay is several hundreds meters. We investigated candidates for the cause of the direct P-wave and coda wave delay. Observed direct P-wave delay can be partly explained by the stress changes caused by coseismic fault slip. However, the coda wave delay cannot be explained by the stress changes that are limited to the superficial area. Crustal heterogeneity should have changed at coseismic time in the deeper area where aftershocks of the M6.1 earthquake occurred.