New Forward And Inversion Algorithms And A Visual Integrated System For MT Data

Magnetotelluric (MT) sounding is an important geophysical means to study the Earth electrical structure using the natural EM field as its source. MT has obtained quite great progress in acquisition system, data processing and interpretation, and research on basic theory since its naissance in 1950's. However, with the in-depth theory research and wide practical application, many new problems are need to be solved, such as forward modeling technique, inversion algorithm, and data processing method.MT has been undergone "manual stage" and "digital stage" since 1950s. At the present, the new visual technique of scientific computation is springing up, which is pushing MT to the third stage: "visual stage". But, compared with the boom of visual technique in other scientific fields, the visual degree of MT data processing and interpretation is lower, especially in our country where the main means of MT data processing and interpretation is still the application running in the DOS system. Thus, processing and interpretation, save, maintainance of MT data are very difficult. And the work efficiency is low, and errors occur frequently.This thesis presents the development on the algorithm fundament, main function, theory and practical application of the Windows visual integrated system for MT data processing and interpretation: MTDATAMNG. In the course of MTDATAMNG development and application, this work has summed up the classical theory of MT, made some creative contributions to the MT forward and inversion method, and the data processing technique as well. The technique route of this dissertation is: classical theory synthesis, new algorithms research, realization and validation of new algorithms, development of visual integrated system by integrating algorithms, and applications of visual integrated system.1. Summary of MT classical theoryThis dissertation reviews some classical theories involved in the research, which includes forward and inversion calculations and data processing. As for the forward aspect, the boundary problems of ID, 2D and 3D model are emphasized. In the aspect of inversion, the optimal inversion theory applied in MT is presented. Some representative algorithms are discussed concretely, and the advantages and shortages of them are pointed out. Moreover, some other inversion algorithms not based on optimal theory are introduced. In the aspect of data processing, calculation methods of various processing parameters are presented, stressed on the rotation and calculation of transferfunctions and their variation. All above construct a solid algorithm foundation for the development of MTDATAMNG, and provide a wide knowledge background for MT theory research of this dissertation.2.Essential structure finite element (ESFE) algorithm and its application to ID MT forward modeling for continuous mediaFE commonly used has some drawbacks, such as large memory requirement and slow solution speed for large-scale problems, which limit its application to 3D problem seriously. These drawbacks are mainly attributed to the building, storing and solving of the stiffness matrix. To improve the conventional FE algorithm, Direct Iterative Finite Element algorithm (DIFE) is proposed. Its difference from conventional FE is to form directly an explicit iterative formulation similar to Finite Difference algorithm which solves the variation problem corresponding to the boundary problem of differential equations in the course of seeking the minimax of the functional equation. DIFE is effective in EM modeling.But DIFE has some drawbacks too. It only adapts to simple meshes and linear! shape function, and the speed of iterative methods is not stable because influence factors are too many. In this thesis, a new algorithm named Essential Structure Finite Element (ESFE) algorithm is presented. The shape function of neighbourhood of a point in the solution domain is defined thus the definition of essential structure is derived. Based on this, the essential structure equation in terms of the principle of classical Galer...