Application of Spectral Amplitude Decomposition to Discrimination of Anthropogenic and Natural Small-Magnitude Seismic Events (≤ M  3.5) in Utah, USA

Discrimination between small magnitude (M < 3.5) natural earthquakes and anthropogenic events is a key part of routine seismic network operations for accuracy of earthquake catalogs, seismic hazard analysis, and test-ban treaty verification. However, this remains challenging at local distances (< 125 km) likely because of increased sampling of heterogeneous Earth structure. Thus, correcting for propagation effects and site response is a key step in the discrimination process, and it has been applied at both regional and local distances. We apply a spectral amplitude decomposition technique that allows for a flexible data-driven fitting of site response and a combined geometrical spreading and anelastic attenuation term to discriminate small-magnitude earthquakes (EQs), explosions (EXs), and mining-induced seismicity (MIS) in Utah, United States. We measure root mean square amplitudes in the time domain for Pg and Sg phases in five frequency bands: 2–5, 5–10, 10–15, 15–20, and 20–30 Hz. We evaluate the performance of the site- and propagation-corrected amplitudes in each frequency band using low- to high-frequency Sg and Pg ratios and Pg/Sg cross-spectral ratios. EQs are best separated from anthropogenic events (EX and MIS) using cross-spectral ratios in the 2–5 and 5–10 Hz frequency range, owing to both EX and MIS being comparatively depleted in high-frequency Sg energy. EX and MIS are best separated using Pg/Sg ratios < 10 Hz, with EXs tending toward more Sg energy than MIS at these frequencies. We further analyze suspected single-fired (SFX) and delay- (ripple)-fired EXs and find SFXs separate more strongly from MIS using Pg/Sg ratios < 5 Hz. Finally, we find that the differential values between source excitation terms and reference magnitudes (Mw, ML, and Mc) provide additional discriminatory abilities warranting further study.