Electromagnetic Forward Modeling And Analysis With Multigrid Method

Solving large linear equations is an important part of the magnetotelluric forward modeling work. For general iterative algorithms, the high-frequency error components can be smoothed effectively, but the smooth efficiency of the low-frequency error components are relatively poor, so, this seriously hampered the effects of forward simulation. In order to improve the two-dimensional magnetotelluric forward modeling efficiency, the multi-grid methods has been introduced to the two-dimensional magnetotelluric forward modeling. By analyzing the application of multi-grid method in one-dimensional problem, the basic principles of multi-grid method has been introduced, and it is proved that this method can efficiently smooth both high-frequency error component and low frequency error component.In this paper, the finite element method is used in the two-dimensional forward modeling. In meeting the simulation characteristics, the computing strategy of multi-grid method has been designed. Inorder to fully prove the superiority of multi-grid method, which is applied in the two-dimensional magnetotelluric forward modeling, three efficient iterative algorithms are selected to be compared with multi-grid method, they are the GMRES, the LSQR and the BICGSTAB.First, the horizontal layered model has been simulated, and the advantages of multi-grid method on calculation accuracy has been quantitative analyzed and proved; secondly, the model of underground heterogeneous has been simulated, and the advantages of multi-grid method on iteration times, calculate speed and accuracy have been analyzed and proved; finally, the model of rolling topography has been simulated, and proved that multi-grid method can effectively suppress the impact of topography.