Abstract:
The aim of this paper is to analyse how the scattered wavefield in a highly heterogeneous structure influences the image quality of structures below it and how the velocity control during processing of data from highly heterogeneous areas is important in the imaging problem. Statistical 2-D two-layer models are built based on deep borehole data. The layers are characterized by a fractal dimension (Hurst exponent) and by the standard deviation of the velocity fluctuations over a constant background velocity value. Shot gathers are generated in these models with a 2-D elastic and viscoelastic finite difference simulator. The synthetic data are first stacked and migrated using an average background velocity function, as would be done in a realistic situation during which only a smooth version of the velocity field can be resolved. Second, the full heterogeneous velocity models are applied to the migration scheme. The final sections are compared using a semblance-based technique to quantify the amount of contamination from an upper heterogeneous structure of a lower heterogeneous structure. In data that are not properly migrated, multiple scattering in the upper layer seems to dominate the lower part of the section. This contamination is controlled by the standard deviation of the velocity fluctuations and it is reduced when viscoelasticity is introduced in the simulations. Counterintuitively, a low Q value may improve deep imaging. When the seismograms are migrated with the real complex velocity field, this contamination is reduced to acceptable levels. Residual contamination cannot be corrected completely because of the single-scattering assumption inherent to all migration techniques.