2D CG-OCCAM Inversion Of MT Data

The computational efficiency and computer requirements(such as memory)are key issues to consider in the inversion method.In this paper,based on the advantages of CG(Conjugate Gradient)and OCCAM inversion methods,a new CG-OCCAM inversion method is proposed.The CG-OCCAM inversion method uses the OCCAM inversion process as the basic framework.,with using "pseudo-forward" idea of CG inversion solves the computational problem of the Jacobian matrix.It does not need to calculate and store each element of the Jacobian matrix,directly obtains the update quantity of the inversion model,greatly reduces the memory usage and improves the Computational efficiency.At the same time,this paper improves the searching process of OCCAM optimal regularization parameter and obtains a new set of inversion process.2D forward modeling uses the finite element method and 2D inversion uses the CG-OCCAM method.This paper implements the 2D magnetotelluric CG-OCCAM inversion method and uses the Fortran programming language to developed the 2D magnetotelluric CG-OCCAM inverse.In this paper,two theoretical models with low-resistance anomaly and high-resistance anomaly were designed.The results of the theoretical model with synthesis data verified the correctness and reliability of the 2D magnetotelluric CG-OCCAM inversion.