Abstract
We developed a real-time seismic data processing system to detect and locate local earthquakes occurring in and around Hokkaido. The seismic network operated by the Research Center for Earthquake Prediction (RCEP), Hokkaido University consists of thirty four stations equipped with short period seismometers, including six stations of Tohoku University and three of Hirosaki University.
Our 32-bit computer-based processing system is composed of two parts, which are connected with a control data file and a seismic wave data file. In the first part, digital data of seismic waves telemetered to RCEP are compiled at every 1.387 sec and events are detected based on comparison of the Walsh spectrum of real-time input data with that of noise data at the system setup time. In the second part, arrival times of P waves are estimated by using the AIC (the Akaike's Information Criterion) method for AR (Autoregressive) model and the location and magnitude of the event are calculated. The hypocentral parameters are printed out within 2.5 to 6min after the first arrivals of P waves at the nearest three stations for the events. These results are automatically sent to the Earthquake Prediction Data Center, the University of Tokyo by telephone line.
4049 events were located in 1987 by using P and S phases on eye reading in the RCEP routine and 1993 events among the 4049 were located by the real-time system. The systematic differences between the epicenters located by the rouine and by the real-time system are less than 3km for the events occurring around the center of the network and more than several 10km for events occurring off the southern Kurile Islands. The large discrepancy for the second group of events may be explained in terms of the lateral heterogeneity of the upper mantle structure around Hokkaido.
