Application of the finite-difference contrast source inversion method to multiparameter reconstruction using seismic full-waveform data

• We extend FD-CSI method (Abubakar et al., 2009) to reconstruct density and velocity.
• FD-CSI needs to solve full forward problem only once, making it highly efficient.
• Multiplicative constraint is incorporated to improve the inversion results.
• FD-CSI method is more efficient than the conventional nonlinear CG methods.
• Two-phase inversion strategy can alleviate cross-talk effects between parameters.

In this paper, we present an efficient inversion method to reconstruct the velocity and density model based on the acoustic wave equation. The inversion is performed in the frequency domain using the finite-difference contrast source inversion (FD-CSI) method. The full forward problem is required to be solved only once at the beginning of the inversion process, which makes the method very computationally efficient. Furthermore, the flexibility of the finite-difference operator ensures FD-CSI the capability of handling complex geophysical applications with inhomogeneous background media. Moreover, different parameters are automatically normalized, avoiding the numerical difficulty arising from the different magnitudes of the parameters. A variant of total variation regularization called multiplicative constraint is incorporated to resolve the sharp discontinuities of the parameters. We employ a two-phase inversion strategy to carry out the FD-CSI method. After simultaneously reconstructing the bulk modulus and density, we obtain a relatively reliable bulk modulus, which is used as the background in the next phase to retrieve more accurate bulk modulus and density. A simple experiment is carried out to present the capability of the FD-CSI method in dealing with the cross talk effect between different parameters. The application on the Marmousi model further emphasizes the performance of the method for more complex geophysical problems.