SUMMARY

The conventional zero-lag crosscorrealtion imaging condition of reverse-time migration is subject to strong migration artifacts. This paper studies wavefield decomposition method under relatively complex subsurface. Although the method greatly suppresses the internal reflection noise, it is subject to residual noise. To suppress the residual noise, we apply a fan filtering on each wave field snapshot in space domain. The filtered wavefields are further decomposed into downgoing and upgoing components and into leftgoing and rightgoing components. The decomposition is carried on F-K domain.

INTRODUCTION

Reverse-time migration propagates wavefields in time through the use of two-way wave equation (Baysal et al., 1984; Whitmore, 1983). It correctly handles both multi-arrivals and phase changes. It’s main advantage over one-way wave equation techniques is that it has no dip limitation. Thus, reverse-time migration enables imaging of very complex subsurface. Twoway migration methods require significantly greater computational resources than one-way migration methods. However, as a consequence of improved computer hardware, there has been recent interest in reverse-time migration (Bednar et al., 2003; Yoon et al., 2003).

Prestack reverse-time migration is accomplished in two steps. First, a shot signature at surface is propagated forward in time using a two-way wave equation and saved on a disk file. Second, the recorded surface wavefield is time-reversed and backward propagated using the same wave equation. The time-reversed receiver wavefield is then time unreversed to produce full receiver wavefield. The migration image is computed by zero-lag crosscorreation of the source and receiver wavefields.

The imaging condition, however, is subject to image artifacts. Unwanted crosscorrelation of head waves, diving waves, and backscattered waves appear as image artifacts. Various methods are since proposed to suppress these noise. Mulder and Plessix (2003) used using a low cut filtering in space domain. Valenciano and Biondi (2002) proposed a deconvolution imaging condition which is based on inverse theory. Chang and McMechan(1986, 1990) suggested ray-traced imaging condition, in which source wavefield after the first arrival is limited to some fixed time duration. This automatically excludes all but postcritical reflections from the source wavefield. Fletcher et al. (2006) modified the wave equation to include a directional damping term in areas of the velocity model where unwanted reflections occur.

Yoon and Marfurt (2006) suggested using Poynting vectors which will determine the direction of wavefield propagation and to decompose into upgoing and downgoing waves. Liu et al. (2007) decomposed the full wavefields into their one-way components, and applied the imaging condition to the appropriate combinations of the wavefield components.

In this paper, we study wavefield decomposition method by migrating a synthetic seismic data over a complex subsurface. Although the wavefield decomposition method greatly suppresses the internal reflection noise, it is subject to significant residual reverse-time noise. The residual noise is analyzed. To suppress the residual noise, we apply a fan filtering on each wave field snapshot in space domain. The filtered wavefields are further decomposed into downgoing and upgoing components and into leftgoing and rightgoing components. The decomposition is carried on F-K domain. Different imaging condition is used depending on the dip of the target structure

This content is only available via PDF.
You can access this article if you purchase or spend a download.