Weighted stacking of seismic AVO data using hybrid AB semblance and local similarity

AVO anomaly and the polarity reversal phenomenon

Gas sands cause seismic reflections to show a wide range of AVO characteristics (Rutherford and Williams, 1989). There are three main AVO classes for the seismic reflections from the gas sands. Class I gas sands have higher impedance than the encasing shale with large positive values for the normal-incidence reflection coefficient. Class II gas sands have nearly the same impedance as the encasing shale and have near-zero normal-incidence reflection coefficient. Class III gas sands have lower impedance than the encasing shale with large negative values for the normal-incidence reflection coefficient. Each class has a distinct AVO characteristic.

It is widely observed in the situation of class II gas sands, if the normal-incidence reflection coefficient is slightly positive, a phase change at near or moderate offsets will occur. This is often represented as a polarity reversal in the seismic reflections from class II gas sands, which is troublesome to NMO velocity analysis because the velocity spectrum using conventional equal weights appears to be zero at the polarity-reversal regions. This reversal will cause a failure in picking the true NMO velocity. A robust velocity analysis approach that can handle this type of AVO anomaly should be used to estimate the NMO velocity which will flatten the CMP gather for stacking. Because class II gas sands are challenging, a more robust weighted-stacking approach is also needed. In this study, we propose a hybrid framework for accurate velocity analysis and optimum-weighted stacking of AVO datasets with class II polarity-reversal anomalies.

In seismic numerical modeling including ray-based and wave-equation-based approaches, combinations of elastic parameters can produce polarity-reversal seismic data. We constructed a synthetic example for investigating the velocity analysis and CMP-stacking performance of the class II AVO anomaly, that show polarity reversals. Table 1 shows the values of the elastic parameters used for modeling: compressional-wave velocity, shear-wave velocity, and density in this four flat-layered model. Figure 1 is a shot gather produced by a 20 Hz Ricker wavelet convolved with the reflectivity calculated from the Zoeppritz equations. The output data clearly show that the second and third events modeled are polarity-reversed. In seismic exploration and interpretation, reflection data from gas reservoirs could often exhibit this phenomenon; therefore, this simple synthetic modeling example further motivates our study in accurate velocity analysis and weighted stacking of seismic data with this class II AVO polarity-reversal anomaly.

Table 1: Elastic parameters used for producing a synthetic data.

Depth (m)	Vp (m/s)	Vs (m/s)	Density (kg/m$^3$ )
0	2400	1100	1.9
0-200	2500	1200	1.95
200-500	2600	1300	2
500-700	2550	1200	1.98
700-1000	2700	1350	2.05

shot
Figure 1. A shot gather produced with the elastic parameters in Table 1.

Weighted stacking of seismic AVO data using hybrid AB semblance and local similarity