Statistical Study of Soviet Nuclear Explosions: Data, Results, and Software Tools

摘要

This final report summarizes our efforts in analyzing Soviet underground nucleartests, with emphases on our updated results and the statistical software developed. 28,875 carefully measured station magnitudes were fed into a maximum-likelihood inversion scheme which simultaneously determines the event size and the station correction, as well as the specific path correction for each source-station pair. The simultaneously-inferred path and station corrections are related to known geological/geophysical features. Applying these path and station corrections to the raw station magnitudes of any individual explosion yields a systematic reduction in the fluctuational variation of station magnitudes across the whole network with a reduction factor ranging from 1.2 to 3 for all Soviet events in our data set. The Mb(Pmax)-Mb(Lg)NORSAR bias between the southwest and northeast subregions of the Soviet's Balapan test site is assessed as 0.07 magnitude unit, which is significantly smaller than that of previous studies. First motion of the initial short-period P waves appears to be a very favorable source measure for explosions fired in hard rock sites underlain by a stable mantle such as Semipalatinsk. The mb(Pa)-based yield estimate for the JVE event of Sep 14, 1988, is 112kt. Between 100 and 150 kt, the mb bias between Eastern Kazakh and NTS using our Mb(Pmax) values is 0.35. Mb, Mb(Lg), Yield estimation, Mb Bias, General linear model, Iterative method, Maximum-likelihood inversion.