PHASE AND SPECTRAL RATIO DISCRIMINATION BETWEEN NTS EARTHQUAKES AND EXPLOSIONS .1. EMPIRICAL OBSERVATIONS

Seismic discrimination of small-magnitude (m(b) < 4) nuclear explosion s from earthquakes requires short-period (f > 0.5 Hz) regional measure ments. In this study we evaluate the most promising short-period discr iminants, P-wave to S-wave phase-amplitude ratios and low- to high-fre quency spectral-amplitude ratios, for a data set from the Nevada Test Site. The data consist of 130 underground nuclear explosions, one very large chemical explosion, and 50 earthquakes, ranging from magnitude 2 to 6, recorded at two broadband digital seismic stations operated by Lawrence Livermore National Laboratory. In contrast with much previou s discrimination work in the western United States, the common station s and paths for all the events in this study allow the discriminants t o be evaluated without applying propagation corrections, and differenc es between events can be ascribed to very near source effects. The sou rce effects found to have the most influence on the discriminants in t his study are depth for the earthquakes, and the source medium propert ies of gas porosity, density, and velocity for the explosions. The pha se-amplitude ratio discriminants tested are P-n/L(g) and P-g/L(g). Bot h show little dependence on magnitude and work better at higher freque ncies and when averaged over the two stations. At 6 to 8 Hz, P-n/L(g) discriminates all except the shallowest earthquakes, while P-g/L(g) di scriminates all but the high gas-porosity explosions. Low- to high-fre quency spectral-amplitude ratio discriminants were tested in each of t he phases P-n, P-g, L(g), and L(g) coda. The explosions show significa nt dependence on source material properties such as gas porosity. All the spectral ratios show dependence on magnitude and for (1-2 Hz)/(6-8 Hz), none performs well below magnitude 3.5. Of the spectral ratios, L(g)-coda performs the best, at least in part because of its extremely small interstation variability, discriminating all but the low gas-po rosity explosions. The best discriminant performance is given by combi ning phase and spectral ratios, A sum of the logarithmic values of P-n /L(g), P-g/L(g), and the L(g) coda discriminates all but two of the sh allowest earthquakes. The source medium dependence of the discriminant s suggests that discrimination performance differences between the wes tern United States and other regions of the world may be due more to s ource property differences than differences in the regional path. Simp ly put, western United States discriminant studies differ from those i n the rest of the world because of the lack of explosions in very high gas-porosity-low velocity materials outside of the Nevada Test Site.