Next-Generation MDAC Discrimination Procedure Using Multi-Dimensional Spectral Analyses

摘要

Theoretical understanding of regional P/S discriminants and their frequency dependence has been an important, long-standing research topic. Following Xie and Patton (1999), Fisk et al. (2005) and Fisk (2006) used modified Brune (1970) and Mueller and Murphy (1971) MM71 models to explain the frequency dependence of Pn/Lg and Pn/Sn discrimination performance for earthquakes and underground nuclear explosions (UNEs) near the Lop Nor, Semipalatinsk, and Novaya Zemlya test sites as mainly due to differences in explosion P and S corner frequencies. Fisk (2007) used source model fits to estimate Pn, Pg, and Lg corner frequencies for Nevada Test Site (NTS) explosions and found that Lg corner frequencies exhibit similar scaling with source size as for Pn and Pg, but shifted lower, analogous to observations by Fisk (2006) for other test sites. A key result is that increasing separation of regional P/S at higher frequencies between earthquakes and explosions at all of these test sites has a consistent model-based explanation mainly in terms of the difference between explosion P and S corner frequencies. Discrimination begins to improve at the explosion S-wave corner frequency and saturates at the explosion P-wave corner frequency, in agreement with empirical experience obtained over decades. We describe how grids of P/S spectral ratios for all combinations of frequencies for P and S wave spectra, which exhibit far greater differences in relative spectral amplitudes and shapes between explosions and earthquakes than considered before, and source models can be used to significantly enhance discrimination performance and confirm the physical basis of the results.