1999 年 52 巻 2 号 p. 241-254
Many earthquakes occur after a destructive shallow earthquake or before a volcanic eruption. It is very important to determine accurate hypocenters as early as possible at a time of such a huge seismic activity, since seismicity data are essential for the understanding of the crustal activity. We developed an automatic data processing system of seismic waves which has a swarm mode processing and can locate accurate hypocenters even for a huge seismicity. The main difference between the swarm and ordinary modes is that the former assumes hypocenters for all events to be in a small area. Event detection is made by estimating approximate origin times calculated from each picked arrival time of P and S waves for several low noise stations close to the swarm area and each theoretical travel time from the swarm area to these stations. We assume events having similar origin times to be seismic events. The system picks P and S wave arrival times not only for these stations but also for all stations by setting time windows at their arrivals which are calculated from the approximate origin time. The application of swarm mode to waveform data at a huge seismicity after the 1996 Onikobe earthquake sequence showed that the new system is very effective and can determine more precise hypocenters rather than manual pickings.