3D velocity-independent elliptically-anisotropic moveout correction

Introduction

Common geological occurrences such as dipping interfaces, lateral velocity variations, or HTI media can lead to real or apparent azimuthal anisotropy, in which case the P-wave moveout velocity becomes elliptically dependent on azimuth (Grechka and Tsvankin, 1998). The symmetry axes of apparent azimuthal anisotropy often correspond to geologically meaningful parameters such as the strike and dip directions of the reflector (Levin, 1985),the directions of tectonic stress (Sicking et al., 2007), the preferred orientation of vertical fractures (Crampin, 1984), or any combination of these factors. Failure to account for azimuthal velocity variations often leads to stack degradation, improper time-to-depth conversion, inaccurate AVO/AVOA, and overall poorer image results (Williams and Jenner, 2002). Sicking and Nelan (2008) and Treadgold et al. (2008) demonstrate that migration algorithms which can handle azimuthal variations in the velocity model can visibly improve seismic imaging results.

Conventional manual velocity analysis procedures take up a significant part of the time needed to process seismic data. Even with semi-automated picking tools, this phase of a typical processing flow alone may take weeks or even months for modern 3D data sets. Accurate automated traveltime picking algorithms are the main tools for modern velocity analysis, and have greatly reduced the time and manual work required to hand-pick velocities (Siliqi et al., 2003). However, these tools still require significant manual inspection and editing for quality control.

The conventional production processing flow does not include picking azimuthally-dependent velocities, but two approaches are commonly used to handle and characterize azimuthal variations in velocity. The first, and historically more popular approach, is to sort CMP gathers into azimuth sectors, and then perform isotropic velocity analysis, processing, and migration on each sector. The individual moveout parameters from all sectors are plotted together, and then fit with a sinusoid to characterize the principal moveout directions and the percentage of anisotropy. Grechka et al. (1999) describe another approach, where NMO is first performed with a smooth global velocity model. If apparent anisotropy is detected, trace-to-trace traveltime shifts are estimated automatically, and the traveltime surface is fit with an ellipse characterized by the moveout slowness matrix $\tensor{W}$ . The second approach has become more popular in production because of its robustness, and in a case-study comparing the two, Lynn (2007) provides an example where the non-sectoring approach yielded a more reliable azimuthal velocity model.

The concept of velocity-independent imaging (Ottolini, 1983) is attractive because it can be very efficient when compared the time and manual work required to hand-pick velocities (Fomel, 2007). The underlying strategy of velocity-independent imaging relies on measuring traveltime slopes throughout the data set rather than hyperbolic traveltimes or velocities themselves (Wolf et al., 2004). Fomel (2002) demonstrates that plane-wave destruction filters provide an automated and effective way to measure local slopes in a seismic volume. Measured slopes can then be used to automate any common time-domain imaging step (Fomel, 2007). Previous work concerning automatic moveout corrections does not extend to the 3D case. In doing so here, we demonstrate that the azimuthal flexibility of automatic moveout correction in 3D is especially useful in the presence of real or apparent azimuthal anisotropy.

Rather than using a single picked velocity profile to apply the NMO correction, using the local slopes of a given 3D reflection event allows the event to be flattened regardless of azimuthal variations in NMO velocity. In practice, these slopes can be measured automatically throughout the volume, so no traveltime surfaces need to be picked. The velocity-independent approach can still be used to extract moveout or interval velocities throughout the data set as data attributes (Fomel, 2007). Our method also suggests that, by measuring local curvatures throughout the seismic data volume, the orientation of the symmetry axes can automatically be estimated with respect to the acquisition coordinates. We present theoretical expressions for azimuthal anisotropy moveout parameters as volumetric attributes, and demonstrate a practical inversion scheme for the same parameters using the velocity-independent approach. Synthetic and field examples are used to validate our proposed method and show the variety of potential applications.

3D velocity-independent elliptically-anisotropic moveout correction