Interpolation and signal extraction of teleseismic wavefields with the linear radon transform

Charles K. Wilson and Antoine Guitton

wilsonck@sep.stanford, antoine@sep.stanford.edu

ABSTRACT

We present a new method for data interpolation and signal/noise separation of teleseismic wavefields recorded by regional seismic arrays. The method exploits the plane wave nature of direct arrivals and receiver-side arrivals from regional scale structure by decomposing the recorded wavefield into a plane wave basis using the linear radon transform. Casting the radon transform as an inversion problem allows the incorporation of time dependent weighting schemes and model variance tuning which are helpful in minimizing artifacts related to the transform process while enhancing lower amplitude arrivals. Following radon transformation, we mute portions of the radon panel that represent plane waves with significantly different moveout ($\sim\pm$.1 s/km) relative to the direct arrival. Transformation back to the data domain from the muted radon domain gives the original signal without (1) plane waves following undesired moveouts, (2) white ambient noise, and/or (3) arrivals not represented well by plane waves (diffractions). Interpolation follows from the inverse data spray operation computed upon return to the data domain and the implicit assumption that a plane wave basis provides the most compact representation of the teleseismic wavefield.