Magnitude Based Discrimination of Manmade Seismic Events From Naturally Occurring Earthquakes in Utah, USA

We investigate the feasibility of using the difference between local (ML) and coda duration (MC) magnitude as a means of discriminating manmade seismic events from naturally occurring tectonic earthquakes in and around Utah. Using a dataset of nearly 7,000 well-located earthquakes in the Utah region, we find that ML-MC is on average 0.44 magnitude units smaller for mining induced seismicity (MIS) than for tectonic seismicity (TS). MIS occurs within near-surface low-velocity layers that act as a waveguide and preferentially increase coda duration relative to peak amplitude, while the vast majority of TS occurs beneath the near-surface waveguide. A second dataset of more than 3,700 probable explosions in the Utah region also has significantly lower ML-MC values than TS, likely for the same reason as the MIS. These observations suggest that ML-MC, or related measures of peak amplitude versus signal duration, may be useful for discriminating small explosions from earthquakes at local-to-regional distances. ML and MC can be determined for small events with relatively few observations, hence an ML-MC discriminant can be effective in cases where moment tensor inversion is not possible because of low data quality or poorly known Green's functions. Furthermore, an ML-MC discriminant does not rely on the existence of the fast attenuating Rg phase at regional distances. ML-MC may provide a local-to-regional distance extension of the mb-MS discriminant that has traditionally been effective at identifying large nuclear explosions with teleseismic data. This topic is of growing interest in forensic seismology, in part because the Comprehensive Nuclear Test Ban Treaty (CTBT) is a zero tolerance treaty that prohibits all nuclear explosions, no matter how small. If the CTBT were to come into force, source discrimination at local distances would be required to verify compliance.