Subsalt imaging in deepwater is still a challenging processes even with high quality data and sophisticated algorithms. The numerical modeling of wave propagation combined with the geological model build powerful tools to investigate proper imaging algorithms and survey acquisition geometries. In this paper we show how the modeling parameters can be determined to optimize the computational costs and the image resolution. The presented geological model covers about 1,150 square kilometers and reaches 15 km in depth. To image the deepest geological structures the required trace length is 18 s. Each of the 4,047 modeled shot gathers covers about 240 square kilometers.
Subsalt imaging in deepwater is still a challenging process even with high quality data and sophisticated algorithms. The reasons are usually poorly defined velocity models and illumination effects in general. Wide-azimuth data acquisition has proven to yield some of the required information (Regone, 2006; Herrmann et al., 2007). However, there is still more investigation required to find the proper survey acquisition geometry for different imaging algorithms. Geological models combined with synthetic data build a common dataset which can be used to benchmark imaging algorithms, to evaluate processing workflow, to design multiple elimination strategies, to design field data acquisition including wide azimuth and rich azimuth alternatives. The combined dataset can even be used to compare the performance of computer architectures. In this paper we will demonstrate the numerical modeling of shot gathers for a given geological model. We will show how we compromised the computational costs and the required resolution of the image. For the modeling we chose the wide-azimuth towed-streamer acquisition geometry (WATS).
Figure 1 shows the geological model used for this study. The geology behind the Repsol YPF model addresses both complex and common problems typically found in the subsalt exploration of the Gulf of Mexico. These problems include, imaging of salt feeders, steeply dipping subsalt reflectors, reflectivity changes in the subsalt section, faults and welds, rugose top of salt that originate multibranching and multipathing of seismic wavefronts and steep dips of the base of salt that cause illumination problems. In the near future, the model will be improved to benchmark anisotropy algorithms and reservoir characterization in full 3D. Velocity model Figure 2a shows the velocity model based on the geological model. Outside the pre-defined layers the velocity increases linearly with depth. Fine layers were added to the gridded velocity model, in seismic format, to simulate layered sediments (Figure 2b). The layering was added using a program that first interpolates, or extrapolates if necessary, the existing sediment layering through the salt. Once this is done the program then operates on one velocity trace at a time to insert a specified number of layers between existing layers. These layers vary randomly in thickness by 20 % of the average thickness with a velocity contrast of 3 % ± 0.6 %.
The two most common methods in modeling seismic wave propagation are the finite difference method (FD) and the pseudospectral method (e.g., Fornberg, 1987).