Permalink : http://hdl.handle.net/2115/45153

JaLCDOI : 10.14943/gbhu.74.67

Hypocenters are generally determined by a method which focuses on onsets of P and/or S waves. As the number, density and quality of seismic stations are recently enhanced, tremors with small amplitude and large duration without any clear onsets of P and/or S waves have been discovered. Epicenters of these events cannot be therefore determined precisely by conventional methods of epicenter determination. New approaches for epicenter determination are required without picking up onsets of P and/or S waves. In this study, we introduce a new method called Relative Maximum Algorithm（RMA ）．This is a hybrid method of the Source-Scanning Algorithm （SSA ）（Kao and Shan, 2004）and the Network Correlation Coefficient（NCC ）（Ohta and Ide, 2008）．This utilizes the maximum of amplitude in each seismogram and determines its focal point and origin time relative to those of a reference event. We determine focal points and theirdistribution of events without any clear phases by the three methods and compare their results each other. The events used in this study are fourteen events located in a 15-35 km depth range, that is, in the lower crust, beneath Tokachi-dake volcano located in the center of Hokkaido, Japan, which occurred in November 2002. We used seismograms recorded at up to twelve Hinet stations around this volcanic region. We determine their focal points with precision of about 2 km in all the directions and about 0.5 sec in origin time. Although the accuracy of Hi-net report appears better than that of RMA , we may enhance the accuracy of RMA by careful analyzes of waveforms such as bandpass and time window filtering, considering each quality of observed waveform data.