In a first experiment, we stacked the traces after correcting for NMO and normalizing by the CMP fold in each bin. As expected, NMO produced accurate amplitudes at the horizontal reflector and the flat top of the diffraction hyperbola. The NMO action, however, failed to preserve the steep flanks of the hyperbola. The CMP stacking process acted as a dip filter which destroyed the steep slopes of the diffraction. Moreover, the NMO stack action could not interpolate for the zero coverage area.
Partial stacking after AMO transformtion to a common offset of 850 meters better preserved the steep flanks of the hyperbola. As result of the narrow range of offset continuations and the efficient non-aliased implementation of the 2D AMO operator, the quality of the amo-stack as shown on Figure amo-stack is better than expected. Nevertheless, data aliasing effects contaminated the dipping arrivals and introduced phase distortions. Most noticeable are the amplitude distortions along the flat reflector (Figure amp-amo) and along the flanks of the diffraction and the dipping bed (Figure amo-stack-win).
The results of partial stacking after equalizing the input are shown in Figure equal-stack. The output is now smoother, it shows more continuity and better resolution than the unequalized result. The equalization step eliminated the phase distortions along the flanks of the dipping events (Figures equal-stack and equal-stack-win) and helped restore the amplitude scales for the horizontal reflector (Figure amp-norm) as well as the dipping arrivals (Figure equal-stack-win).
This simple inversion was inexpensive. Has anything been gained over the AMO stacking action? First, we reduced the phase distortions due to aliasing of the data. Second, we better preserved the true amplitude scales without ever bothering to think about the number of contributing traces.