On Detecting and Estimating Multiple Arrivals from Underground Nuclear Explosions
Abstract
Nine explosions were used to study the problem of specifying the number and characteristics of the arrival phases of nuclear explosion. Tentative identification of the amplitudes and time delays of the multiple arrivals are given. The method used is to scan the likelihood (L) for a single reflection model over amplitude (a) and time delay (tau). The likelihood is constructed in the frequency domain using von Seggern and Blandford's (1972) source signal spectrum and a noise spectrum estimated from a time window preceding the signal. The results demonstrated that pP was present along with additional multiple arrivals indicated by other peaks in the likelihood function, L. In the case of MAST and PILEDRIVER, the likelihood estimate for tau was found to be clearly superior to tau = 1/f sub n, where f sub n is the first null frequency. The result for PILEDRIVER, tau = .15, is the first accurate estimate of tau for this event, and the shortest estimated delay time in the literature. The delay time for CANNIKIN, .55 sec, is significantly shorter than twice the observed uphole time: 1.0 sec. This result is unexplained, but it may be related to a change of reflection coefficient with frequency which is not included in the model. Comparison of theoretical and observed WWSSN long-period explosion P waves provides evidence for such a variable reflection coefficient. Fair to good agreement was obtained between predicted and observed explosion periods measured on WWSSN long-period film and on LRSM short-period data which was deconvolved to appear as if recorded on the WWSSN long-period system.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 25, 1978
- Accession Number
- ADA079531
Entities
People
- R. H. Shumway
- R. R. Blandford