Three-dimensional target-oriented pre-stack migration

The main limitations of current 3D migration techniques do not lie in the zero-offset migration algorithm itself but in the preceding stacking process which generates the zero-offset input data. In conventional common-midpoint (CMP) stacking four problems may occur: 1. The members of a CMP gather may have different individual midpoints and, therefore, the concept of 'common midpoint' may not actually exist. 2. In situations with complex geology, reflection times in a CMP gather may not be described by a hyperbolic relationship and so the concept of 'stacking velocity' may not exist. 3. Conflicting structural dips require different stacking velocities. 4. One stacked event may represent information from different reflection points. In the seventies and early eighties refined stacking procedures were developed, based on the dip moveout (DMO) concept (Bolondi et al. 1982; Hale 1984). These DMO-oriented stacking techniques (followed by post-stack migration) can be successfully applied if the aforementioned second problem does not occur. However, in 3D data sets generally all four problems occur and, as a consequence, the quality of post-stack migrated data is often far from optimum. Therefore, particularly for complicated subsurface structures, alternative techniques must be developed. The ideal procedure would be full pre-stack migration by singleshot record inversion (SSRI), followed by zero-offset (ZO) stacking (Berkhout 1984). However, for the 3D case this promising technique is still too laborious. It is obvious that, given the limitations of today's computational power, a more practical approach to 3D prestacking migration is required. In many practical situations seismic interpreters are mainly interested in an accurately positioned and well-resolved image of a prespecified target zone. Hence, much work can be saved by the following two-stage procedure: 1. Apply conventional CMP processing and two-pass migration for an initial 3D evaluation of the total area under investigation. 2. Apply full 3D pre-stack migration to specific areas of interest ('target-oriented' stage). In this paper we propose a full 3D target-oriented prestack migration scheme and discuss various theoretical and practical aspects. Also we demonstrate the validity of the target-oriented approach with the aid of a physical scale model (water tank) data example.