HR: 1340h
AN: S33A-0311    [Abstracts]
TI: Signals embedded in the OBS records, in light of Gabor Spectral Analysis
AU: * Chang, T
EM: tychang@mail.ntou.edu.tw
AF: Institute of Applied Geosciences, National Taiwan Ocean University, No. 2 Peining Rd, Keelung, 202 Taiwan
AU: Wang, Y
EM: yetmen@ancad.com
AF: AnCAD, Inc., 5F, No. 67, Sec. 1, Yung-Ho Rd., Taipei, 305 Taiwan
AU: Chang, C
EM: cpchang@csrsr.ncu.edu.tw
AF: Center for Space and Remote Sensing Research, National Central University, No. 300, Chungda Rd., Chungli, 305 Taiwan
AU: Lee, C
EM: leecs@mail.ntou.edu.tw
AF: Institute of Applied Geosciences, National Taiwan Ocean University, No. 2 Peining Rd, Keelung, 202 Taiwan
AB: Since the last decades, seismological survey has been expanded to marine area, with goal of making up the deficiency of seismogenic study outside the land. Although teleseismic data can resolve plate boundaries­Ý location and certain seismic parameters for great earthquake, local seismogenic frame can be merely emerged by the seismic network in situ. The Ocean Bottom Seismometer (OBS), therefore, is developing for this kind of purpose and becoming an important facility for seismological study. This work introduces a synthesized spectral method to analyze the seismograms recorded by 15 OBS deployed at the Okinawa trough in 14 days (Nov. 19 ~Dec. 2, 2003). Geological background of Okinawa trough is well known to correspond with the back-arc spreading in the regime of the Philippine Sea plate subducting northward beneath the Eurasia plate. As the complex affections at sea bottom, for instance, strong current, slope slumping, turbidite flow, and even sea animal attack, the OBS seismogram show a rather noisy sequence in comparison with the record on land. However, hundreds of tectonic earthquake can be extracted from such noisy records (done by Drs. Lin and Sibuet). Our job is to sort out the signals with the distinguishable sources by means of a systematically spectral analysis. The continuous wavelet transform and short-term Fourier transform, both taking Gaussian function as kernel, are synthesized as the Gabor transform in data process. The use of a limited Gaussian window along time axis with negligible low frequency error can largely enhance the stability of discrete Fourier spectrum. With a proper window factor selection, the Gabor transform can improve the resolution of spectrogram in time domain. We have converted the OBS records into spectrograms to detect the variation of signal causes. Up-to-date, some tremors signals and strong current oscillations have been told apart from these continuous records with varied frequency composing. We anticipate the further application of spectral analysis can detect automatically onset of signal generating by the significant sources in the future.
DE: 0520 Data analysis: algorithms and implementation
DE: 0935 Seismic methods (3025, 7294)
DE: 3025 Marine seismics (0935, 7294)
DE: 3255 Spectral analysis (3205, 3280)
DE: 3280 Wavelet transform (3255, 4455)
SC: Seismology [S]
MN: Fall Meeting 2005