Weighted stacking of seismic AVO data using hybrid AB semblance and local similarity |
Since AVO II polarity-reversal phenomena widely exist in real seismic data (Castagna and Backus, 1993), accurate velocity analysis and weighted stacking for this type of seismic data is of the great significance for structural imaging and seismic interpretation. Our synthetic and field examples illustrate the successful applications of the proposed approach.
It has been pointed by Fomel (2009) that resolution loss is the price paid in obtaining the accurate velocity analysis for AVO II polarity-reversal data. In the comparison of Figures 3c and 3d, however, the effect from lower resolution of AB semblance on the velocity analysis for the first and second events without polarity reversal could be neglected. Additionally, through the comparisons of NMO-corrected CMP gathers in Figures 4a and 4b, and Figures 7a and 7b, the traces in the shallow and far-offset area (around 0.4 s for traces 11 and 12) get noise better attenuated after AB semblance. Here the NMO-corrected CMP gathers after conventional semblance and AB semblance went through the same frequency-stretch mute.
This study innovatively combines two recent ideas into a complete workflow for the weighted stacking of AVO II polarity-reversal data and achieves encouraging results in both synthetic and field testing examples. But it should be noted that our new approach at the current stage may not completely replace the conventional stacking, even if only for the AVO data. From the synthetic stacking examples in Figure 5 and Figure 8, the local-similarity-weighted stack does tell the locations of reflections and demonstrate high SNR, but may weaken the polarity-reversal information from prestack gathers. In the perspective of this point, the proper reference to conventional equal-weight stack and SNR-weighted stack may help us infer possible lithological changes from velocity and density changes. In the two examples using field CMP gathers, although the proposed method is suitable for both high- and low-SNR data, the stacking analysis largely depends on the SNR of prestack CMP gathers when the objective of stack is to detect deep reflectors.
It may be possible to apply the proposed framework to other types of seismic data, even in cases where the AVO phenomenon is not very obvious. Additionally, further research to verify this possibility could be accomplished with ease. It can be conveniently implemented based on the conventional semblance calculation and CMP stacking framework. The key elements in the framework are the AB semblance calculation and local-similarity-weighted stacking, which both can be found in an open-source software package, Madagascar ( ); thus, it is worth developing routine processing modules for implementation of the proposed framework.
Weighted stacking of seismic AVO data using hybrid AB semblance and local similarity |