Aperture Optimized RTM Migration

Stacking is of paramount importance in seismic processing to improve the signal to-noise ratio (S/N) and the imaging quality of seismic data. The conventional stacking method that averages equally a collection of input traces cannot robustly suppress coherent noises. To attenuate this kind of noise and achieve an optimally stacked image remains an attractive and challenging topic in the seismic industry. The key point for ongoing research is to develop methods that can be used to reliably discriminate between “good” and “bad” data samples. To this end it is important to identify two key objectives for the process of optimal stacking. The first is to find a suitable domain where signal can be easily distinguished from noise and the second is to build a robust procedure that allows only the signal contribute to the final migrated stack. We describe a novel, iterative method to clean and enhance the stacked migration image. For bandlimited migration algorithms we define an original prestack image domain that is analogous to the aperture partitioned migration domain of Kirchhoff-type migration techniques. Through an automatic procedure that is based on a coherence analysis we show how, in this domain, signal can be separated from both coherent and incoherent noise in an effective way. With the aid of synthetic examples we show how this yields to a superior quality image compared to a conventional migration stack.