next up previous [pdf]

Next: Acknowledgments Up: Burnett & Fomel: Azimuthal Previous: Examples

Discussion and Conclusions

By extending time-domain velocity continuation to the azimuthally anisotropic 3D case, we have combined the concepts of azimuthal imaging and diffraction imaging. We assume a three-parameter migration slowness model that allows velocity to vary elliptically with azimuth. We have provided simple examples to illustrate the potential application of our method to fracture characterization through diffraction imaging. By treating diffractions as signal, our method performs azimuthal analysis on post-stack data, without the requirement for common-offset-vector or offset-vector-tile binning schemes. This is possible because, unlike reflections, diffractions can preserve azimuthal information post-stack. Post-stack data volumes have obvious advantages over pre-stack or vector-binned data for analysis, including smaller memory size, and improved signal-to-noise ratio.

Allowing azimuthal variation in the migration velocity will result in improved imaging, which is clearly a benefit of 3D velocity continuation. However, the potential for fracture characterization may be even more useful. Our method has many of the same ideas as the azimuthal imaging and fracture characterization flow proposed by Sicking et al. (2007) for reflection data. Under the velocity continuation framework, we can extend the azimuthal imaging idea to 3D diffraction imaging. Since diffraction-generating fractures and faults are often nearly vertical and preferentially aligned, they can be associated with azimuthal anisotropy. Fomel et al. (2007) demonstrate that it is possible to separate diffractions from specular reflections, and then image their associated discontinuities through the use of velocity continuation. Their method operates on post-stack data, as they show that diffractions are highly sensitive to migration velocity, even in the zero-offset case. Al-Dajani and Fomel (2010) have successfully demonstrated zero-offset diffraction image focusing as a fracture detection attribute on azimuth-sectored 3D field data. Our proposed method uses multi-azimuth image focusing primarily as a velocity analysis criterion, but kurtosis could also be used as an image attribute. In cases where subsurface fractures cause azimuthal anisotropy, kurtosis as an attribute may be indicative of fracture intensity (Al-Dajani and Fomel, 2010). By applying velocity continuation to 3D diffraction imaging, one may be able to estimate both the orientation and the intensity of fractures from the resulting anisotropic velocity model and maximum kurtosis volumes, respectively. This information can be useful in reservoir development, as it can provide insight to subsurface fluid flow behavior.

Although the spectral implementation of our method allows a range of possible images to be computed efficiently, it demands large amounts of memory to store a suite of images as well as the kurtosis volume. Modern computational hardware makes our approach feasible as-is, especially for target-oriented imaging and analysis strategies. Future studies may lead to better optimization-based approaches to finding local kurtosis maxima, in which case, our method could be practical for dense parameter estimation throughout full 3D volumes.

The underlying strategy of velocity continuation is to simultaneously estimate the velocity model as the data are imaged. This is beneficial in the case of azimuthal anisotropy discussed here, as the ambiguity between structural heterogeneity and anisotropy is handled without the need for iteration. Other multi-parameter seismic imaging problems, such as converted-wave imaging, which are also conventionally handled by iterative flows, could also benefit from pre-stack versions of the 3D velocity continuation strategy.


next up previous [pdf]

Next: Acknowledgments Up: Burnett & Fomel: Azimuthal Previous: Examples

2013-03-02