High-resolution velocity spectra using eigenstructure methods
, by Biondo Biondi and Clement Kostov
Stacking velocity analysis provides optimal estimates for the parameters of a few, well
separated reflections in additive white noise. To resolve close, interfering reflections,
we adapt a theory developed originally for radar and sonar applications to the particular
reflection seismics experiment. In seismic data the wavefronts are wide-band and the
curvature of the wavefronts is significant across the receivers array. We introduce a family
of estimators for the wavefront parameters based on the eigenstructure of the data covariance
matrix. The traditional stacking method can be related to the new high-resolution estimators
and the superiority of the latter is explained with a simple geometric construction.
A field-data example shows that the high resolution estimators are particularly attractive
for local spectra estimation where short arrays are considered. Several realistic synthetic
examples illustrate the new method and show its superiority in estimating stacking velocity
spectra.