High-frequency earthquake swarm associated with the May 1991 dome extrusion at Unzen Volcano, Japan

Abstract

The 1990–1995 eruption of Unzen Volcano in southwestern Japan was characterized by the extrusion of a dacite lava dome and frequent pyroclastic flows during the dome growth. Associated with the dome emergence on May 20, 1991, an intense swarm of high-frequency (HF) microearthquakes occurred just beneath the crater at very shallow depths. We used data from FG3, a seismic station located 500 m SSW of the crater, to identify 29,401 HF earthquakes between May 11 and 31, 1991. The rate of HF earthquakes increased starting on May 12 and peaked on May 17. The high seismicity continued until May 26, then dropped sharply, coinciding with a marked decline in the swelling on the southern upper flank of the volcano. The seismicity increased and decreased repeatedly within a 1- to 2-h period, which was correlated with tilt cycles observed 680 m west of the crater in such a way that the seismicity increased during uplifting on the craterward side. Defining an earthquake group as a series of earthquakes with waveforms that are similar or vary only slightly over time, we identified 10 such groups, each containing more than 300 events. The largest group comprised 3,214 events over 18 days. Seismicity rates of eight groups, including the largest, increased and decreased repeatedly, correlated with the tilt cycles. As the waveform data from station FG3 were considerably clipped for relatively large events, we analyzed data from two additional stations, KRA and CJA, located 3.5 and 8.7 km from the crater, respectively, and identified a total of five new groups. The activity of two groups recorded at KRA was correlated with the tilt cycles, but those of three groups recorded at CJA did not always show such a correlation. Hypocenters for the groups recorded at KRA were distributed to the east side of the conduit and a focal mechanism suggested that the events in this area occurred due to compressional stress produced by the inflation of the conduit. The groups recorded at CJA occurred to the north side of the dike trending westward from the conduit. The focal mechanisms have P-axes roughly trending to the dike, which can be explained by the compressional stress generated by the increasing thickness of the dike.