Threshold magnitudes for a multichannel correlation detector in background seismicity

Colocated explosive sources often produce correlated seismic waveforms. Multichannel correlation detectors identify these signals by scanning template waveforms recorded from known reference events against target data to find similar waveforms. This screening problem is challenged at thresholds required to monitor smaller explosions, often because nontarget signals falsely trigger such detectors. Therefore, it is generally unclear what thresholds will reliably identify a target explosion while screening nontarget background seismicity. Here, we estimate threshold magnitudes for hypothetical explosions located at the North Korean nuclear test site over 6 months of 2010, by processing International Monitoring System (IMS) array data with a multichannel waveform correlation detector. Our method (1) accounts for low-amplitude background seismicity that falsely triggers correlation detectors but is unidentifiable with conventional power beams, (2) adapts to diurnally variable noise levels, and (3) uses source-receiver reciprocity concepts to estimate thresholds for explosions spatially separated from the template source. We find that underground explosions with body-wave magnitudes m (sub b) approximately 1.66 are detectable at the IMS array USRK with probability 0.99, when using template waveforms consisting only of Pwaves, without false alarms. We conservatively find that these thresholds also increase by up to a magnitude unit for sources located 4 km or more from the 12 February 2013 announced nuclear test.