Analysis of ADCIGs and computation of anisotropic velocity spectra

We compute ADCIGs from the prestack migrated image by applying post-processing slant stacks Rickett and Sava (2002); Sava and Fomel (2003). The ADCIGs at three different locations are illustrated in Figure . It can be observed that up to 2 km depth, reflectors are imaged flat in ADCIGs, meaning that the migration velocity obtained by CGG is correct and that no anisotropy in those layers is needed. However, we can observe that below the Balder horizon (strong reflector at around 1.7 km depth), the reflectors start curving downward, indicating that the estimated isotropic migration velocity is larger than the true migration velocity.

 

ADCIGs
Figure 12 Line 750: ADCIGs at different locations obtained after isotropic migration.

That result is consistent with the sonic velocity recorded in well 3_9a-8, located at 32.83 km along the CMP axis on line 750. Figure  presents the estimates of the interval vertical velocity obtained from surface seismic data, measurement of sonic velocity and checkshot. It illustrates that the vertical velocity estimated from surface seismic data -- under the assumption of isotropy -- is accurately estimated down to a depth of around 2 km and is overestimated deeper between the Balder and BCU horizons. This overestimation is due to the presence of anisotropic rocks under the Balder layer. The velocity measured from surface seismic data (V_N) is equal to the vertical velocity (V_V) under the assumption of isotropy and flat layers. However, in an anisotropic medium, for flat layers and a VTI model, the expression of the NMO velocity becomes $V_N=V_V\sqrt{1+2\delta}$, where $\delta$ is the first Thomsen parameter. Since the velocity measured from seismic data, V_N, is larger than the velocity measured from sonic logs and checkshots, the anisotropic parameter $\delta$ probably can not be considered zero and takes significantly positive values.

 

Sonic_log Figure 13 Comparison of the vertical velocities at the location of well 3_9a-8. The different velocities are estimated from seismic surface data (assuming isotropy), measurement of sonic velocity and checkshots.

Finally, we computed anisotropic velocity spectra from the three different ADCIGs illustrated in Figure . The first series of velocity spectra are semblance panels of $\rho_V_N$(Figure ). The velocity spectra are computed by trying to fit the RMO curves in ADCIGs with only $\rho_V_N$. Figure  illustrates that the V_N was well estimated by CGG since high semblance values are centered around perturbation values close to zero.

 

VN_spectra
Figure 14 Velocity spectra computed at different inline locations by trying to fit the RMO curves in ADCIGs with only $\rho_V_N$.

Semblance panels of $\rho_V_H$ are presented in a second series of velocity spectra (Figure ). The second series of velocity spectra is are semblance panels of $\rho_V_H$. The velocity spectra are computed by trying to fit the RMO curves in ADCIGs with only $\rho_V_H$, assuming that V_N is correct. Figure  illustrates the presence of anisotropy at a depth greater than 2 km. More specifically, the fact that the energy is centered at negative values of the horizontal velocity perturbations indicates that the horizontal velocity we used for the migration is smaller than the true horizontal migration velocity. Since our migration was isotropic, we assumed that V_H=V_N=V_V. As a consequence, we can infer that below 2 km depth, the anisotropic parameter $\epsilon$ probably should not be set to zero and will take significant positive values ($V_H=V_V \sqrt{1+2\epsilon}$).

 

VH_spectra
Figure 15 Velocity spectra computed at different inline locations by trying to fit the RMO curves in ADCIGs with only perturbations in the horizontal velocities.