The Physical Basis of Lg Generation by Explosion Sources

Stevens, J. L. ; Baker, G. E. ; Xu, H. ; Bennett, T. J. ; Rimer, N. ; Day, S. D.

Science Applications International Corporation

关键词: Research Programs;    Nuclear Explosions;    Mountains;    Amplitudes;    S Waves;   

DOI  :  10.2172/835251
RP-ID  :  NONE
RP-ID  :  FC03-02SF22676
RP-ID  :  835251

美国|英语

来源: UNT Digital Library

【 摘 要 】

The goal of this project has been to develop a quantitative predictive capability for explosion-generated Lg phases with a sound and unambiguous physical basis. The research program consisted of a theoretical investigation of explosion-generated Lg combined with an observational study. The specific question addressed by this research program is how the Lg phase is generated by underground nuclear explosions. This question is fundamental to how Lg phases are interpreted for use in explosion yield estimation and earthquake/explosion discrimination. To constrain modeling, we have extensively reviewed the existing literature and complemented that work with an examination of several explosion data sets, most notably: (1) Degelen Mountain explosions recorded between 7 and 57 km, with corresponding recordings at Borovoye, at approximately 650 km; (2) recordings from Russian deep seismic sounding experiments; (3) NTS explosion sources including the NPE and nuclear tests covering a range of source depths and media properties. A simple point explosion in an infinite medium generates no shear waves, so the Lg phase is generated entirely by non-spherical components of the source and conversions through reflections and scattering. We find that the most important contributors to the Lg phase are: (1) P to S conversion at the free surface and other near source interfaces, (2) S waves generated directly by a realistically distributed explosion source including nonlinear effects due to the free surface and gravity, and (3) Rg scattering to Lg. Additional effects that contribute significantly to Lg are scattering of converted S phases that traps more of the converted P-to-S in the crust, and randomization of the components of Lg. The pS phase from a spherically symmetric explosion source in media with P-wave velocity less than upper mantle S-wave velocity is trapped in the crust and can explain the observed radial and vertical Lg. The free surface pS converted phase from the same source in a high velocity medium, however, is not trapped in the crust. It, therefore, does not explain the Lg observations. A spherically symmetric explosion source also fails to explain near-source Sg and regional Sn observations. However, the observed shear waveforms and amplitudes can be explained by adding a CLVD source component, which is the lowest order non-spherical correction to the spherical source. The persistence of Rg to large distances in some regions argues against Rg scattering as the source of Lg in all regions. 2D nonlinear calculations of explosion sources quantify the amount of seismic radiation generated by the non-spherical parts of a realistic explosion source. A 2D nonlinear calculation modeled after the NPE source and structure produces Lg consistent with an explosion plus a CLVD source with about half the strength of the explosion. However, because of the very low velocities at the NPE source location, the explosion generates substantial Lg directly. This alone may be sufficient to explain the Lg observations in this case. The importance of the non-spherical component of the source to matching observed shear wave phases is demonstrated in 2D calculations of Degelen explosions, which are typically underburied in high velocity granite. Source calculations for an overburied event in high velocity medium do not produce the observed shear waves, and so we also investigate scattering mechanisms. 2D and 3D finite difference calculations indicate that the topography at Degelen traps much more of the surface P-to-S converted phase in the crust than does scattering from crustal heterogeneities. Topography also has the greatest impact on Rg. The effect of topography increases with frequency, and the primary effect is to disperse Rg. In the 3D calculations, there is significant scattering to the tangential component at 8 km and 4 Hz. We use energy conservation to determine an upper bound on Rg to Lg scattering. Rg to Lg scattering may contribute to Lg and more so to Lg coda at frequencies less than 1 Hz, but contributes less at higher frequencies than Lg generated directly by the explosion or by surface P-to-S conversion. This report is organized as follows. We first present a review of the extensive body of work on explosion generated shear waves, from which we have tried to extract the most robust observations that constrain the possible mechanisms. That is followed by a presentation of our own observations, from very near source, through local to regional distances for a range of source conditions, and a discussion of the implications of our observations. The observational work is complemented by three distinct types of simulations, one set focused on source physics, one on the effect of scattering due to topography and lateral heterogeneity on Lg, and a new type of modal calculation that simulates Rg to Lg scattering.