P and S wave Coda Calibration in Central Asia and South Korea

Empirically derived coda source spectra provide unbiased, absolute moment magnitude (Mw) estimates for events that are normally too small for accurate long-period waveform modeling. In this study, we obtain coda-derived source spectra using data from Central Asia (Kyrgyzstan networks - KN and KR, and Tajikistan - TJ) and South Korea (Korea Meteorological Administration, KMA). We used a recently developed coda calibration module of Seismic WaveForm Tool (SWFT). Seismic activities during this recording period include the recent Gyeongju earthquake of Mw=5.3 and its aftershocks, two nuclear explosions from 2009 and 2013 in North Korea, and a small number of construction and mining-related explosions. For calibration, we calculated synthetic coda envelopes for both P and S waves based on a simple analytic expression that fits the observed narrowband filtered envelopes using the method outlined in Mayeda et al. (2003). To provide an absolute scale of the resulting source spectra, path and site corrections are applied using independent spectral constraints (e.g., Mw and stress drop) from three Kyrgyzstan events and the largest events of the Gyeongju sequence in Central Asia and South Korea, respectively. In spite of major tectonic differences, stable source spectra were obtained in both regions. We validated the resulting spectra by comparing the ratio of raw envelopes and source spectra from calibrated envelopes. Spectral shapes of earthquakes and explosions show different patterns in both regions. We also find (1) the source spectra derived from S-coda is more robust than that from the P-coda at low frequencies; (2) unlike earthquake events, the source spectra of explosions have a large disagreement between P and S waves; and (3) similarity is observed between 2016 Gyeongju and 2011 Virginia earthquake sequence in the eastern U.S.