CMT解と震源時間関数の迅速な自動決定とその結果の評価

抄録

Recent advances in broadband seismic observation enable us to analyze high quality waveform data just after the earthquake occurrence. An automated system for rapid determination of earthquake mechanisms for local events occurring beneath central Japan is developed with this point as background. When an earthquake occurs, its hypocentral parameters are determined automatically in several tens of seconds. In a case when an event occurring in a given area has a magnitude larger than a certain value, its source process is automatically analyzed by the use of a broadband waveform inversion technique. Considering the importance to shorten the computer process time to obtain the result, we calculate the Green's functions immediately after hypocentral parameters are calculated and when waveform data to be used in the inversion are being retrieved. The data of the Earth's response parameters, which are used in the calculation of the Green's functions, are stored on the hard disk so as to reduce the total lapse time. As a result, the CMT solution and source time function can be obtained in a few minutes after the earthquake occurrence. Inadequacy of theoretical waveform calculated by the 1-D Earth's model is corrected by introduction of station corrections to the observed arrival time and overparameterization to the source time function, where a number of double couple small sub-events are located in a certain time range. The CMT solution is represented by superimpose of moment tensors of small sub-events with very short duration. All the sub-events are assumed to occur at the same location (i. e., at the hypocenter) but to have different onset times. The source time function is also estimated from the temporal distribution of the sub-events. To evaluate the reliability of the results obtained from the present study, we compare them with the following two data sets: i) focal mechanism solutions determined by P first motion analysis, and ii) CMT solutions obtained from the other method for the same events. Our results are basically consistent with these two sets of mechanism solutions. However the agreement of the solutions is obviously better among the CMT solutions. Possible explanation for the difference between CMT and focal mechanism solutions is that the latter may be inaccurate because of insufficient push and pull first motion data, or that the total rupture may occur along the fault different from the initial rupture.