Post-stack velocity analysis by separation and imaging of seismic diffractions

Separating diffractions

The underlying assumption that we employ for separating diffracted and reflected events is that, in a stacked data volume, background reflections correspond to strong coherent events with continuously variable slopes. Removing those events reveals other coherent information, often in the form of seismic diffractions. We propose to identify and remove reflection events with the method of plane-wave destruction (Claerbout, 1992; Fomel, 2002). Plane-wave destruction estimates continuously variable local slopes of dominant seismic events by forming a prediction of each data trace from its neighboring traces with optimally compact non-stationary filters that follow seismic energy along the estimated slopes. Minimizing the prediction residual while constraining the local slopes to vary smoothly provides an optimization objective function analogous to differential semblance (Symes and Carazzone, 1991). Iterative optimization of the objective function generates a field of local slopes. The prediction residual then contains all events, including seismic diffractions, that do not follow the dominant slope pattern. An analogous idea, but with implementation based on prediction-error filters, was previously discussed by Claerbout (1994) and Schwab et al. (1996). Although separation of reflection and diffraction energy can never be exact, our method serves the practical purpose of enhancing the wave response of small subsurface discontinuities.

Post-stack velocity analysis by separation and imaging of seismic diffractions