Research On The Method For Inversion Of 2.5D CSAMT Data

The current the inversion of controlled source audio magnetotelluric (CSAMT) data are mainly rely on one-dimensional. Two demension result is obtained from one dimensional inversion site by site. However, this result is clearly influenced subjective, and thus to improve the accuracy of inversion interpretation should continue to study the controlled-source electromagnetic sounding data of the two-dimensional inversionTo achieve 2.5D controlled source electromagnetic sounding data inversion, we must better understand the inversion theory, and related algorithms. Giving the results of previous research ,studying the theoretical framework of the inversion. The The realization of forward part based on the finite-element method developed by Martyn J.Unsworth, 2.5-D forward modeling algorithms and the inversion part on John Booker RRI inversion algorithm. Xinyou Lu (1999) made a detailed discussion of the algorithm. This paper studies the CSAMT-RRI forward and inversion which based on the combination of 2.5-dimensional numerical simulation technique and RRI inversion algorithm. The reliability of the algorithm through forward modeling and performance of inversion results were discussed. Inversion results, including REBOCC inversion,magnetotelluric RRI inversion and CSAMT-RRI inversion,of controlled source data from different model forward modeling.The inversion results show the significant source effect ,low resistivity are extended away from source and depth direction. The article discussed the source effect base on the designed model ,and the results show that source effect by different level under different algorithms. The low resistivity model modeling and inversion have been done under different source transmiter and the characteristics were discussed. Different transmiter has different inversion result, there is a need for a comprehensive analysis of different inversion results to get a more accurate judgments. In summary, the results show that the REBOCC algorithm is superior than the RRI algorithm, affected by the initial model is small, with a higher resolution, closer to the actual model of the inversion results. Model can better reflect the actual situation.