Detailed Analysis of the Far‐Regional Seismic Coda in Kazakhstan Using Array Processing
Résumé
An analysis of seismic coda phases recorded by five small aperture arrays in Kazakhstan and Russia is performed. The characteristics of the coda phases are obtained using array processing providing estimates of direction of propagation, frequency content, and apparent propagation velocity of coherent arrivals propagating through the arrays. A statistical analysis, using the coda characteristics of the whole dataset (far‐regional earthquakes) recorded by the arrays, is performed leading to a time‐azimuth distribution of the detected waves. Long Lg‐coda wavetrains clearly dominate the far‐regional seismic records. The later part of an Lg‐coda wavefield is generally expected to consist of omnidirectional and multiple scattered waves; still, in the present study, preferential propagation directions of scattered waves are identified in seismic signals, even for a lapse time larger than 1200 s. The comparison of the Lg‐coda characteristics for two different groups of earthquakes and for the different arrays shows that the geometrical configuration of the event array strongly influences the back‐azimuthal distribution of detected waves. More precisely, temporal variations of the Lg‐coda characteristics are identified: (1) as a transitory regime in the early part of the coda where arrival directions deviate from the epicenter back azimuth (eastward or westward); (2) as a final regime where the later part of the Lg coda observed by the two northern arrays exhibits incoherent phases with random propagation directions; and (3) as a final regime for the three southern arrays, which reveals coherent arrivals propagating from array‐specific back azimuths not related to the epicenter back azimuth.