The Study Of Three-dimensional Magnetotelluric Inversion Based On L-BFGS And Its Application In Volcano Basin Of Xiangshan

Magnetotelluric method(MT)plays an important role in many fields,such as structural research,petroleum exploration,mineral and hydrological exploration.With the practical application of the three-dimensional inversion of MT,the higher requirement of reducing the computer resources and increasing the convergence speed,a more optimized threedimensional inversion algorithm is needed.The limited memory Quasi-Newton method(LBFGS)only requires short computation time and can get high inversion resolution.It has certain advantages in the three-dimensional inversion of Magnetotellurics.Therefore,in this paper we study three-dimensional Magnetotelluric inversion based on the L-BFGS algorithm.After realizing the three-dimensional inversion algorithm based on the L-BFGS algorithm.We have designed four theoretical models with different degrees of complexity of electrical conductivity structures like: a single conductive model,a single resistive model,a combine conductive-resistive model,and a peak-valley symmetry combine model with synthetic terrain.By using the L-BFGS algorithm to do inversion of these theoretical models,the results show that the L-BFGS algorithm can effectively invert the isoelectric structural features of different theoretical models including morphology and location,plus it has a certain degree of noise resistance.Therefore,we do some research about the adaptability of three dimensional inversion parameters such as regularization factor ?,initial model,mesh subdivision,covariance parameters and it's a smoothing numbers by using the peak-valley symmetry combination models with synthetic terrain.In order to verify the practicability of the algorithm,the three-dimensional inversion algorithm based on L-BFGS was applied to the inversion of measured MT data in the volcanic basin of Xiangshan.Three-dimensional inversion of MT data from the study area was carried out,and the inversion result was better.The deep electrical structure of the study area is reflected,which provides support for deep geological structure and uranium metallogenic mechanism research in this area.The inversion results well reflect the deep electrical structure of the study area,provide support for the study of the deep geological structure research of the study area and the study of uranium mineralization mechanism.