摘要
Inversion of magnetotelluric(MT) responses has been used to explore the electrical conductivity distribution of the Earth’s interior. In three-dimensional(3-D) inversions, it is significant to use a good initial model, because final model obtained by most 3-D inversion methods is influenced by the initial model. Although uniform initial models are widely used in 3-D inversions, one-dimensional(1-D) initial models are alternatives, which could more appropriately represent the actual conductivity variations in the Earth’s interior. This study presents a two-step 3-D inversion method, especially for marine cases. This inversion method first concentrates on obtaining a 1-D initial model and then inverts for 3-D conductivity structures with it, in both of which the 3-D topography is carefully taken into consideration. This method was tested by synthetic models of different topography variations(depression-shaped, smoothly varying, channel-shaped and square-shaped plateau topography) and of heterogeneous layers with different checkerboard-type anomalies(sharp or smooth lateral conductivity variations) embedded in1-D model of depth-dependent conductivity. The comparisons were done about obtaining an initial model by the proposed method and that inverted from the corrected responses. The results of 3-D inversions by using the method of this study were also compared to that with different uniform initial models. Results of synthetic tests and comparisons were discussed by using directional information of newly introduced model-vector parameters. The performance and validity of this method was verified.It also revealed that some of the newly introduced model-vector parameters could be used to show the convergence of inversions and help to select inverse model.
基金
supported by the Scientific Instrument Developing Project of the Chinese Academy of Sciences(Grant No.ZDZBGCH2018006)
the National Natural Science Foundation of China(Grant No.41874088)
the Strategic Priority Research Program of the Chinese Academy of Sciences(Grant No.XDA14050100)。