Magnitude: Yield Relationship at Various Nuclear Test Sites--A Maximum- Likelihood Approach Using Heavily Censored Explosive Yields

Conventional methods for estimating underground explosion yields from seismic recordings are based on the use of some appropriate 'magnitude:yield' relationship. One of the most important parameters used to characterize the seismic signature of an underground explosion is the body-wave magnitude, mb. Thus obtaining an unbiased measurement of mb (auxiliarily Ms, pcoda, mb(Lg), Mo, and RMS Lg values) is obviously a key step in estimating the yield. During the past decade, the mb which is averaged over a well-distributed global network and which incorporates the maximum-likelihood technique into the inversion scheme has become widely accepted as a means to obtain mb estimates that avoid bias due to the detection threshold characteristics of individual network stations. Recently Soviet seismologists have published descriptions of 96 nuclear explosions conducted from 1961 through 1972 at the Semipalatinsk Test Site, in Eastern Kazakhstan. With the exception of releasing news about their 'peaceful nuclear explosions' PNE, the Soviets have never before published such a body of information. However, out of the 72 Degelen events with announced yields, only 9 events or 12.5% were of 'known' yields. The remaining were either left censored (66.7%) or bounded (20.8%). Similar heavy-censoring pattern can be found for other test sites. Thus the development of a procedure capable of making full use of such censored information would seem very timely and necessary.