Nuclear Explosion Monitoring

Abstract

Systematic monitoring of nuclear testing is a primary task that is critical for the success of any proposed comprehensive test ban treaty (CTBT). Current explosion monitoring methods use features of seismic and acoustic signatures to identify several thousand daily worldwide events as either naturally occurring or man‐made. The task is unique among risk analysis applications in that the political ramifications of event identification errors are profoundly serious and may persist for decades after the event in question. An approach to characterizing risk applies classical and Bayesian methods to signature features and summarizes the results in the event classification matrix (ECM). The approach stems from a desire to emulate a current expert system and has evolved as a collaboration between seismologists and statisticians.