Reducing Numerical Dispersion with High-Order Finite Difference to Increase Seismic Wave Energy

Abstract

The numerical dispersion of 2D acoustic wave modeling has become an interesting subject in wave modeling in producing better subsurface images. Numerical dispersion is often caused by error accumulation with increased grid size in wave modeling. Wave modeling with high-order finite differences was carried out to reduce the numerical error. This study focused on variations in the numerical order to suppress the dispersion due to numerical errors. The wave equation used in modeling was discretized to higher orders for the spatial term, while the time term was discretized up to the second order, with every layer unabsorbed. The results showed that high-order FD was effective in reducing numerical dispersion. Thus, subsurface layers could be distinguished and observed clearly. However, from the modeling results, the wave energy decreased with increasing distance, so the layer interfaces were unclear. To increase the wave energy, we propose a new source in modeling. Furthermore, to reduce the computational time we propose a proportional grid after numerical dispersion has disappeared. This method can effectively increase the energy of reflected and transmitted waves at a certain depth. The results also showed that the computational time of high-order FD is relatively low, so this method can be used in solving dispersion problems.