The Application of a Dynamic Correlation Processor for IMS detection screening

Abstract:

A recent study by Dodge and Harris (2020) found that the vast majority of seismic detections at the Pinedale, Wyoming International Monitoring System (IMS) station are associated with small events (predominantly mining explosions) that are not detected at other IMS stations. For every Pinedale station detection that is included in the IDC’s Reviewed Event Bulletin (REB) there are 30 to 40 times more detections that are associated with small events. Large numbers of similar small-event detections at each IMS station increases the probability that a false event can be built by the IDC automatic system. In this project we test the performance of detection associators if detections that are known to be associated with small, local events are screened from the association process. We use a dynamic correlation processor (DCP) (Harris and Dodge, 2011) to form groups of similar waveforms, and we manually associate groups to known and suspected man-made sources, such as mines, as well as small repeating events, that do not meet the IDC event-definition criteria. We tie the DCP detections to IMS detections that were input into the IDC automatic process. Such detections that are confidently determined to be from small repeating events that historically have not met the event definition criteria are screened (removed). The remaining IMS detections are input into the IDC automatic processing system – run on a research testbed – to determine whether detection screening reduces formation of false events, while maintaining or increasing the formation of events that meet the IDC event definition criteria. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract Number DE-AC52-07NA27344. LLNL-ABS-812628.