Optimization of staggered grid finite- difference coefficients based on conjugate gradient method

KEJIE JIN,JIANPING HUANG,QIANG ZOU,ZIYING WANG,SIYOU TONG,BIN LIU,HU ZIDUO

2021-01-01

Journal of seismic exploration

Abstract:Jin, K., Huang, J.P., Zou, Q., Wang, Z., Tong, S., Liu, B. and Hu, Z., 2021. Optimization of staggered grid finite-difference coefficients based on conjugate gradient method. Journal of Seismic Exploration, 31: 33-52. The implementation of difference coefficients optimization strategy can effectively suppress numerical dispersion and improve the modeling accuracy. The conventional difference coefficients calculation method based on Taylor-series Expansion exists serious numerical dispersion. In this paper, we derive a new dispersion error function from the dispersion relation, and the optimal difference coefficients are obtained iteratively by using the conjugate gradient method, thus a staggered-grid difference coefficients optimization method based on the conjugate gradient is developed. We compare dispersion curves, snapshots and single shot records using low-velocity model, high-velocity model and Marmousi model, the results show that the new method can effectively reduce the numerical dispersion compared with the difference coefficients of the conventional Taylor-series Expansion method. The 8th-order optimized difference operators can achieve the modeling precision of 12th-order Taylor-series Expansion difference operators, which can effectively save calculation time and internal storage. The optimization method performs well for both simple model and complex model forward modeling.