Online Journal for E&P Geoscientists

Jiao, Jianwu ^*1; Saleh, Saleh M.¹; Verschuur, Eric ²
(1) Aramco, Dhahran, Saudi Arabia. (2) Delft University, Delft, Netherlands.

Forward modeling is an important step in seismic data processing and velocity model building. Modeling techniques can be divided into two major categories: ray-based methods and wave equation methods. Ray-based methods such as Kirchhoff modeling are computationally more efficient than wave equation methods. Ray-based methods usually produce inaccurate models when the subsurface geology is complex.

Wave equation modeling can either be based on the two-way wave equation (TWE) or one-way wave equation (OWE). Finite-difference, finite-element, boundary integral, and pseudo-spectral methods are examples of TWE modeling while split-step and phase-shift plus interpolation (PSPI) are examples of OWE modeling. For acoustic modeling, TWE is more accurate than OWE but generally requires a more accurate velocity model and more computational resources. Using an inaccurate velocity model in TWE modeling typically generates reflections and multiples from the wrong interfaces, which can lead to misleading results. OWE modeling can only generate primaries and is hence less sensitive than TWE to velocity errors, which is attractive for velocity model building.

We have developed an efficient forward modeling algorithm based on the PSPI algorithm. Our approach can handle both topographic and lateral velocity variations. In this presentation, we will discuss the PSPI theory followed by 2D and 3D comparisons with TWE synthetic data. Then, we will show how this approach can be used to simulate data recorded from topography. Finally, modeling results from real data velocity models will also be shown and discussed.