Magnetotelluric Regularized Inversion And Its Application Study

With the development of economy, the effectiveness of the inversion result is directly related to national production and economic benefits in the practical process of geophysical inversion. In order to improve the quality of inversion, geophysical workers have used a variety of geophysical exploration methods and inversion methods to reduce multiple solutions of inversion. In the process, it is often necessary for us to determine a more reasonable model, select the appropriate regularization factor, and use the advanced regularization method, which has been an important part of geophysical inversion workers'research.Under normal circumstances, if we can find the best overall solution of inversion objective function with various methods, and make the observation data fit the model data better, we can guarantee the accuracy of inversion solutions and further reduce the number of multiple solutions. Considered together, geophysical workers are confronted with more and more complex and practical problems, and the selection of inversion methods, inversion parameters, which determines the results of inversion to a large extent.Based on the previous results, the thesis discusses the geophysical inversion methods with different Gaussian noise and missing data, especially for the regularized inversion method on the basis of singular value decomposition. Then it analyzes the principles, describes the basic process of its inversion, and further discusses the effects of inversion for the field testing data. Firstly the thesis reviews the basic concepts about geophysical inversion and inversion methods under pathological conditions, including well-posed, over-determined, underdetermined together with the establishment of various data models about singular value decomposition and regularization. Then, the thesis introduces in detail the general process of the regularized inversion for the synthesized model of regularized inversion and the features of inversion structure, comprising the effects on overall function of inversion with different regularization factors, the estimated choice prior to regularization factor selection, principles of the flattest model, the smoothest model and the maximum model and their influence on regularized inversion. Finally, the thesis, taking sounding for an example, designs the algorithm, analyzes and compares field testing data for one-dimensional and two-dimensional models through regularized inversion.Researchers and production personnel have made great progress in geophysical inversion research, but they are still needed to choose different regularization inversion methods for specific geological problems, so that the regularization factor of the overall objective function is chosen more reasonably, and the prior information (including geology, drilling, geochemical exploration) is used adequately and is inversed effectively with constraint conditions. Only in this way can we make contributions to the actual production and deeper analysis of geologic origin.