next up previous [pdf]

Next: Acknowledgments Up: Decker et al.: Oriented Previous: Field Data Example

Conclusions

We have developed and demonstrated a highly parallel and constructive procedure for time-domain velocity estimation. The method operates by decomposing data by slope and propagating slope components in velocity in the midpoint-time-slope domain. Semblance in slope gathers is used as a measure for selecting velocities that correspond to correctly migrated flat diffraction events, which applies even for single-offset data. This semblance measure is observed to achieve higher resolution when multiple offsets are considered in the oriented velocity continuation process. If data with multiple offsets are available, oriented velocity continuation could be used to better constrain the velocity model when used in conjunction with traditional reflection moveout analysis. Chosen velocities can be used to generate both diffraction and reflection images. The powerful ability of oriented velocity continuation to operate with only zero-offset data enables accurate migration velocity analysis in situations where only limited-offset data are available.

Flatness of slope-decomposed diffraction events is more responsive to velocity perturbation than diffraction focusing, because it does not require smoothing or windowing in space. Therefore, the proposed method has the potential for diffraction velocity estimation with superior resolution when compared to methods based on diffraction focusing.

Oriented Velocity Continuation is formulated as a type of time-migration, and is thus subject to constraints relating to image distortion from horizontal velocity changes in the subsurface. The presence of strong lateral velocity variations may alter diffraction moveout, making event slope change with azimuth. In such a case, OVC would be unable to completely flatten the diffraction signal in slope gathers and locally determine the correct migration velocity.

This method can be extended to three dimensions using data decomposition by azimuth and inclination for each image point. Operating on three dimensional data should improve velocity resolution by overcoming out of plane artifacts in the seismic image. Although extension to 3D adds the expense of additional spatial and slope dimensions, the Fourier-domain computation would also be parallel in these new dimensions, making the operation feasible in practice using computer clusters.


next up previous [pdf]

Next: Acknowledgments Up: Decker et al.: Oriented Previous: Field Data Example

2017-04-20