科研之友
微信
新浪微博
Facebook
分享链接
摘要
Magnetotelluric method is a technique for imaging the electrical conductivity and structure of the Earth. It is widely used in mining exploration, the detection of deep geologic structure and the study of earth dynamics. As a key point of inversion and interpretation, forward is always focused on by many authors. Compared to other forwarding methods, for instance, finite difference method, finite element method and finite volume method, integral equation method has a better precision and it only need to discrete the anomalous bodies in the earth, which deserves to be implemented. Traditional integral equation approaches for magnetotelluric problems generally use regularly structured grids which are difficult to approximate geometrically complicated models. In addition, the traditional integral equation algorithms replace the cubic cell by a spherical cell of equal volume to approximately calculate singular integral in the dyadic Green function, which would reduce the accuracy of numerical solutions. In this paper, we propose a new approach for three dimensional magnetotelluric modeling. It uses the tetrahedral grids and the closed-form solutions for the singular dyadic Green volume integrals, to simulate the complicated anomalous bodies, as well as improve the accuracy of numerical solutions. Firstly, integral equation on anomalous body for magnetotelluric problem is derived from Maxwell's equations. The expression for linear equation is obtained by using tetrahedral grids to discrete the anomaly and assuming that the electrical field in each element is a constant. Then, the closed form solutions for the singular dyadic Green volume integrals on tetrahedron are applied to compute the singular integral in the dyadic Green function. PARDISO, a high-performance parallel solver, is chosen to achieve accurate results for three dimensional magnetotelluric problems. Finally, the 3D-1 model and the hexagonal prism model are tested to verify the accuracy of the new integral equation approach and validate its abilities for modeling high conductivity contrast problem up to 1000 : 1, and simulating complicated models.
- 出版日期2017-11
- 单位中南大学
