Abstract:Forward modeling of the frequency-domain wave equation represents an essential foundation for full waveform inversion in the frequency domain, the accuracy and efficiency of which rely heavily on the forward modeling method employed. To reduce the numerical dispersion, anisotropy, and number of grids per the shortest wavelength in forward modeling methods, rotating coordinate systems have been successfully applied to establish finite-difference (FD) schemes for the forward modeling of the frequency-domain wave equation. However, rotated optimal FD schemes are incapable of handling rectangular sampling grids, which are ubiquitous in practice. Fortunately, optimal FD schemes based on the average-derivative method (ADM) overcome this restriction on different directional sampling intervals. However, the ADM itself is merely an algebraic approach and therefore does not inherit the geometrical properties of the rotating coordinate system. Based on the principle of a rotating coordinate system, a novel optimal directional derivative method (DDM)-based 4th-order, 17-point FD scheme is developed in this paper for the forward modeling of the frequency-domain, two-dimension scalar wave equation to approximate the spatial derivatives. The conventional 4th-order, 9-point scheme and rotated optimal 17-point FD scheme can be derived as special cases of the proposed scheme. Compared with the rotated optimal 17-point FD scheme, the proposed scheme is capable of addressing arbitrary rectangular sampling grids, including equal and unequal directional sampling intervals; moreover, the optimized weighted coefficients can reduce the number of grids per the shortest wavelength from 2.56 to less than 2.4 with maximum phase velocity errors of 1%. Furthermore, the proposed scheme is superior to the ADM-based optimal 17-point FD scheme in suppressing numerical dispersion due to the inherited geometrical properties of the rotating coordinate system. A perfectly matched layer boundary condition is applied to the final FD equation to attenuate boundary reflections. Numerical examples demonstrate the validity and adaptability of our 17-point FD scheme.

Average-Derivative Optimized 21-Point Forward Modeling and FWI in Frequency Domain

Average‐derivative optimized 21‐point and improved 25‐point forward modelling and full waveform inversion in frequency domain

Full Waveform Inversion Based on Time-Integral-damping Wavefield

Quasi-P wave forward modeling in viscoelastic VTI media in frequency-space domain

Frequency Domain Full Waveform Inversion Based on Generalized FD Method

Scalar wave forward modeling method in Laplace-Fourier domain using finite-difference method

A Developed Nearly Analytic Discrete Method for Forward Modeling in the Frequency Domain

Comparisons and analysis of several optimization finite-differencing schemes in 2D acoustic frequency-domain numerical modeling

Time Domain Full Waveform Inversion with Low Frequency Wavefield Decompression

Strategy and Application of Waveform Inversion Based on Seismic Data Subset

Full waveform inversion method based on automatic differentiation and graph space optimal transport

Adaptive Multi-Step Full Waveform Inversion Based on Waveform Mode Decomposition

Three-dimensional Frequency-Domain Full Waveform Inversion Based on the Nearly-Analytic Discrete Method

Frequency-domain Full-Waveform Inversion with an Efficient Iterative Solver Approach

Full Waveform Inversion Without Low Frequency Using Wavefield Phase Correlation Shifting Method

Time-Domain Elastic Full Waveform Inversion With Frequency Normalization

A GENERALIZED 17-POINT SCHEME BASED ON THE DIRECTIONAL DERIVATIVE METHOD FOR HIGHLY ACCURATE FINITE-DIFFERENCE SIMULATION OF THE FREQUENCY-DOMAIN 2D SCALAR WAVE EQUATION

Full Waveform Inversion Method Using Envelope Objective Function Without Low Frequency Data

Forward modeling by optimized equivalent staggered-grid finite-difference method for time-space domain

Waveform inversion with structural regularizing constraint based on gradient decomposition

Time Domain Full Waveform Inversion Based on Gradient Preconditioning with an Angle-Dependent Weighting Factor