Publication Abstract

not signed in
Log In
Title Determination of source parameters from noisy waveforms: Inversion of displacement seismograms versus accelerograms

Authors J. Sileny, V. Vavrycuk, C. Baker and R.P. Young

Publication Reference 1997, Rockbursts and Seismicity in Mines (ed. Gibowicz & Lasocki), Balkema, Rotterdam, pp. 95-100

Abstract Seismic data recorded in mining environments are always contaminated by noise. This noise may be either regular (white noise) or may exhibit a more complex pattern resulting from factors associated with natural phenomena or the recording instrumentation. At the frequencies of interest the signals are usually recorded as accelerograms and, therefore, a double integration must be performed in order to recover the source time function (STF). The noise in the waveforms results in a zero-line bias, so a high-pass filter must be used during the integration step. We highlight the importance of the choice of the cut-off frequency of the filter to the source infonnation which may be recovered from the waveforms. It is shown that a filter which suppresses the noise completely may result in a complete distortion of the recovered source parameters. We compare the effects of noise on two alternative approaches to the inversion of accelerograrn data. In the first the accelerograms are converted into displacement records before inverting them to retrieve the mechanism and STF. In the second the accelerograms are inverted directly and then the resulting time function is integrated to obtain the true STF .Using a series of synthetic data sets, contaminated with various levels of white and irregular noise, we show the second method is preferable. This method worlcs better for high noise levels and is less sensitive to any non-linear noise effects. The choice of cut off frequency for the filter is also less critical when the accelerograms are integrated a priori. However, in both approaches, the noise in the records is transformed into a spurious volumetric component in the mechanism. Fortuitously this appears to occur after the true STF has ended. The methodology is applied to real accelerogram data recorded at the Atomic Energy of Canada Ud's Underground Research Laboratory in Manitoba, Canada.

Key Figure Key Figure
Second derivative of the MTRFs obtained by inverting the accelerograms contaminated by the random noise (15% noise level) together with the irregular noise. The STF (right) was retrieved by using the filter with the cut-off frequency of 50 Hz.



Return to publications list

< Return to the home page   ::  Contact Us
Applied Seismology Consultants