Quantifying the Effectiveness Pg/Lg Ratios for Explosions and Earthquakes

Discriminating low magnitude earthquakes from low yield explosions at local distances is of increasing interest to the monitoring community. Modern methods compute P/S spectral amplitude ratios from seismic waveforms and can be susceptible to noise saturation and uncertainty in the group-velocity of both P and S phases. Here we investigate a catalog of 6475 analyst characterized seismic events recorded in Utah from (January 1-11, 2011) with magnitudes Mb(Lg) ranging from -1 to 4.1. We estimate spectral ratios as (z = log10(P/S)) as a network average of calibrated amplitude measurements. We then contaminate the waveform data with real additive noise and interference records collected from regions of monitoring interest and quantify source discrimination performance. Small events recorded by a sparse network of stations show that spectral ratios are strongly affected by complicated source and path relationships that drive substantial variability in locally recorded P/S ratios.