Spectral ratio discrimination studies carried out on events located in the western United States and Soviet Union (S.U.) illustrate that pronounced differences in radiated explosion-source spectra relative to nearby earthquakes exist between the two regions. Nevada Test Site (NTS) explosions are characterized by the existence of more low-frequency and/or less high-frequency energy (greater low- to high-frequency spectral ratios) than western U.S. earthquakes. The opposite pattern is observed in the S.U. with nuclear explosions appearing to have more high-frequency (and/or less low-frequency) energy than earthquakes. These observations may be caused by at least two principal effects that are probably acting in parallel: (1) variations in depth-dependent effects of attenuation acting between the shallow explosions and deeper earthquakes and (2) differences in the dynamic response of the near-source geology to the passing explosion shock wave. Anelastic synthetic seismogram calculations illustrate that depth-dependent attenuation effects may explain the spectral observations. However, a number of observations using near- and far-field data from NTS explosions suggest that near-source effects are the dominant factor. A quasi-empirical explosion source model is proposed that simultaneously fits the spectral ratio data from both the U.S. and S.U. relative to earthquakes in each of the respective regions.

Additionally, the model fits the trends of the spectral ratios observed as a function of magnitude. The key to the model is the shape of the pressure time history acting at the elastic radius. For explosions detonated in weak, porous rock, the radiated shock wave divides into a two-wave system consisting of an elastic precursor followed by a plastic wave. The generation of this two-wave system introduces a rise time into the pressure time history. In the frequency domain a second corner frequency is established in a third-order model (with an ω-3 high-frequency decay) whose value is inversely proportional to the time separation of the two waves. In higher-strength, saturated rocks (or for overburied explosions) the effective rise time is short, and a second-order model is appropriate (with an ω-2 high-frequency decay). The second-order model provides a good fit to the S.U. data. In contrast,a hybrid model is required to fit the NTS data with an ω-3 high-frequency decay for shallow explosions detonated in unsaturated tuff that evolves to an ω-2 decay as depth of burials reach higher-strength, saturated rocks below the water table. ¿1991 American Geophysical Union