Simulation of Core Phases From Coda Interferometry

The extraction of core phases through seismic interferometry is important for exploring the seismic structure of the Earth's inner core because it can provide an unprecedented data coverage. However, low signal‐to‐noise ratios and uncertainties in phase arrivals hinder the application of seismic interferometry. With three arrays from the China Regional Seismic Network, South America, and the western United States, we simulate coda interferometry using the one‐dimensional synthetic coda of large earthquakes (magnitude ≥ 7.0) from 2005 to 2012. Compared with the cross correlations of real coda (10,000–40,000 s after earthquakes), the similarities among the simulated waveforms of the core phases (PKIKP, PKIIKP, PKPab, PKIKP2, and PKIIKP2) indicate that reverberations at first‐order discontinuities constitute the major source for coda interferometry. Relative to synthesized Green's functions, the core phases derived from coda interferometry provide reliable phase information but varying amplitudes. In particular, the slownesses are generally consistent with the theoretical calculations, thereby confirming the extracted phases. The simulations prefer the 15‐ to 50‐s period band and indicate that the earthquake distribution has an insignificant effect on the retrieval of core phases at small distances (0–8°). Coda interferometry is scarcely affected by white noise, but it is influenced by the attenuation of the inner core. Experiments on focal mechanisms further suggest that coda signals from dip‐slip earthquakes mainly contribute to the reconstruction of the core phases (except for the simulated PKPab). Key Points Core phases are extracted and confirmed through simulations of coda interferometry Reverberations at first‐order discontinuities from large earthquakes are the major source to reconstruct the core phases The retrieval of core phases is dependent on frequency band and focal mechanism but not significantly affected by source heterogeneities