Interactive and Automatic Modes of Long-Period Signal Extraction Using Cascaded Filters

摘要

The purpose of this work was to develop and evaluate an interactive long-period signal extraction filter suite. Five filters were included in the suite and were applied to test data in a cascaded mode, (i.e., the output of one filter served as the input to the next filter). The filters tested were a bandpass filter, a Wiener filter, a chirp matched filter, a polarization filter, and a Gaussian narrowband filter. It was found that, for the filters examined, the polarization filter dominates the cascaded filtering approach to signal extraction; specifically, processing gains as high as 14 to 15 dB were realized from the application of cascaded filters which included this filter. Further, tests performed on a Eurasian data set using one specific cascaded filter sequence yielded an improvement in surface wave detection capability equivalent to an improvement in bodywave detectability of approximately 0.6 to 0.8 m sub b units. Finally, cascaded filters were found to provide more detected signals from which surface wave magnitudes can be obtained; as such, the use of these filters can enhance an analyst's capability to discriminate between earthquakes and explosions. (Author)