A method of source-parameter estimation based principally on observations of teleseismic long-period surface waves is developed and applied to 71 underground nuclear explosions at the Balapan (Shagan River) test site in Kazakhstan. First, we quantify the observed radiation patterns of Love and Rayleigh waves, noting that such data place only three linear constraints on the five parameters of the standard point source model for each explosion. The three constraints are the strengths of the isotropic, sin 2 Phi, and cos 2 Phi components of the observed pattern (where Phi is azimuth). The five source parameters are the isotropic moment M(1); the double-couple moment M(o); and the strike, dip and rake angles, (Phi(s), delta, lambda). These last four parameters characterize the inferred faulting (tectonic release). Surface waves which propagate from different seismic events in a geographically tight cluster to a distant station are similar in shape, and for very shallow sources the propagation operator can be described by a single wave shape, a differential-propagation delay, and a single amplitude factor containing the effects of attenuation, scattering, focusing and geometrical spreading. The inverse problem is formulated which iteratively solves simultaneously for the path-amplitude factors and the three radiation-pattern parameters for all events using observed seismograms as the empirical wave shapes. We fit 1110 relative measurements of Rayleigh- and Love-wave amplitudes to determine path-amplitude factors to 29 globally distributed stations and radiation-pattern parameters for the 71 explosions.Second, we present a new method for interpreting such radiation patterns. In previous studies, two Of the five source parameters have typically been fixed by making the very restricted assumption of pure dip-slip faulting at 45 degrees dip. The assumption leads to estimates of strike that are mostly in the north-west-south-east direction at Balapan, and to estimates of M, that correlate very poorly with estimates based upon short-period signals (P, Lg). However, now that excellent yield estimates have become available, we can avoid the assumption of pure dip-slip faulting. Instead we can estimate M,from P and Lg-wave data, thus adding a fourth constraint to help in interpreting the teleseismic surface-wave amplitudes. The outcome is a method for estimating the strike Phi(s) of faulting associated with tectonic release for each explosion, and for estimating the two combinations M(o) sin delta cos lambda and 1/2M(o) sin 2 delta sin lambda, which, respectively, are the strike-slip and dip-slip components of the tectonic release.The strike directions of inferred faulting at Balapan are found to lie in the range 90 degrees-120 degrees, which conforms better with limited geological information on faulting in the region than does the trend that has long been inferred from assuming pure dip-slip faulting. We find that without additional information about individual explosions, we are not able to provide a high-precision isotropic moment or yield estimate from teleseismic surface waves.
Nd780Times Cited:16Cited References Count:28