HR: 0800h
AN: S51B-1009    [Abstracts]
TI: Automated Monitoring of Non-Volcanic Tremors in Southwest Japan Subduction Zone
AU: * Suda, N
EM: suda@geol.sci.hiroshima-u.ac.jp
AF: Hiroshima University, 1-3-1 Kagamiyama, Higashi-Hiroshima, 739-8526 Japan
AU: Ishihara, Y
EM: y-ishihara@jamstec.go.jp
AF: JAMSTEC/IFREE, 3173-25 Showa-machi, Kanazawa-ku, Yokohama, 236-0001 Japan
AU: Nakata, R
EM: nakata@geol.sci.hiroshima-u.ac.jp
AF: Hiroshima University, 1-3-1 Kagamiyama, Higashi-Hiroshima, 739-8526 Japan
AU: Kusumi, T
EM: kusumi@geol.sci.hiroshima-u.ac.jp
AF: Hiroshima University, 1-3-1 Kagamiyama, Higashi-Hiroshima, 739-8526 Japan
AB: Activities of low-frequency tremors, which occur in the non-volcanic region of southwest Japan along the subduction zone of the Philippine Sea plate, often show slow migration of tremor sources with a speed of about 10 km/day and quasi-periodicity of occurrence with an interval of 2--6 months. Such characteristics seem to be related to fluid liberated from the subducting slab. Monitoring of tremor activities therefore may lead us to elucidation of fluid property and its role in the subduction zone. To monitor activities of low-frequency tremors in real-time, we have developed the automatic detection and hypocenter determination system using real-time data from the commutation-satellite seismic telemetry system. We divide the tremor region into 6 subregions, and use 8--12 vertical-component seismic data to monitor the tremor activity in each subregion. Before applying the detection method, we preprocess real-time data by applying the bandpass filter with a frequency band of 1--10 Hz, resampling them from 100 Hz to 20 Hz, calculating their envelopes, and finally applying the moving average with a time window of 3 seconds. The automatic detection method is based on a two-step numerical statistical test: the first test is for examining whether given two envelopes with a length of 2 minutes are correlated or not, and the second for examining whether an event occur or not in this time window using results of the first test. If an event is detected then the automatic hypocenter determination method is applied. In this method time lags for the maximum cross-correlations obtained in the detection method are utilized as traveltime difference data, and hypocenters are determined by minimizing the absolute residual using the simplex method. The results obtained by the automatic detection method are consistent to those determined by visual inspection of waveforms, and also the results obtained by the automatic hypocenter determination are consistent to those determined by the conventional traveltime analysis of envelope waveforms. Thus our system is useful for real-time monitoring of the tremors, and consequently will provide basic data for investigation of the subduction zone associated with great earthquakes.
DE: 7240 Subduction zones (1207, 1219, 1240)
SC: Seismology [S]
MN: Fall Meeting 2005