Joint Inversion Of Magnetotelluric And Seismic Surface Wave Dispersion Datasets

The non-uniqueness of geophysical inversion problem is a phenomenon that multiple models can fit well in the same set of observation data due to the insufficiency of measured data or the equivalence of the forward modelling.The models,recovered by multi-geophysical data sets,can combine the complement advantages of different physical parameters,and have higher resolution than the inversion using only one type of geophysical data.It has been paid more attention because of its effective improvements in reducing the ambiguity of inversion problem,and in the comprehensive geophysical interpretation.Magnetotelluric(MT)has been widely used in the exploration of the deep electrical structure,oil gas,mineral resources,et al.,due to its advantages of wide frequency band,large exploitable depth,and no subject to high resistance shielding.However,its resolution is affected by electromagnetic noise interferences,volume effect,static displacement,et al.,in the exploring complicated geological environment.Surface wave is formed by the interference coupling of the body wave at the interface.Dispersion is one of its most important characteristics.The shear wave velocity structure with high transverse resolution can be obtained by the dispersion data sets.But its resolution decreases with increasing depth,and its transverse resolution is affected by the azimuth distribution of the surface wave travel path.Magnetotellurics(MTs)and seismic surface waves are two prominent geophysical methods for deep underground exploration.Joint inversion of these two datasets can help enhance the accuracy of inversion.Therefore,how to realize the complementary advantages of the two methods,reduce the non-uniqueness of inversion problem,and improve the inversion accuracy is the emphasis and difficulty of our study.In view of those,we make systematic studies on the joint inversion of magnetotelluric and surface dispersion data sets in 1D,2D,and 3D respectively.The research process adheres to the technical line of “methodology?programming?model experiences?application”,to test the stability,convergence,reliability,superiority and practicability of the algorithm.After careful investigation,some main findings of this study can be summarized as follow:(1)In the study of 1-D joint inversion of magnetotelluric and surface dispersion data sets,we describe a method for developing an improved multi-objective genetic algorithm(NSGA-SBX)and apply it to two numerical tests to verify the advantages of the algorithm.Our findings show that compared with the traditional binary encoding genetic process,the NSGA-SBX method yields better convergence results and avoids the lack of stability and precision caused by the selection of the discretization steps.Joint inversion with the NSGA-SBX method can improve the inversion results by strengthening structural coupling when the discontinuities of the electrical and velocity models are consistent,and in case of inconsistent discontinuities between these models,joint inversion can retain the advantages of individual inversions,in the 5-layers model experiments.By applying the algorithm to four detection points along the Longmenshan fault zone,we observe several features.The eastern margin of the Tibetan Plateau shows high velocity and high resistivity in the shallow crust,while two low velocity layers and a high conductivity layer are observed in the middle lower crust,probably indicating the mid-crustal channel flow.An extremely high resistivity layer beneath the middle segment might be interpreted as the effects of southeast-directed crustal flow beneath eastern Tibet blocked by the rigid Sichuan Basin.The Sichuan Basin demonstrates low S-wave velocity and high conductivity in the shallow crust probably due to thick sedimentary layers.Along the Longmenshan fault zone a high conductivity layer from ~8 km to ~20 km is observed beneath the northern segment and decreases with depth beneath the middle segment,which might be caused by the elevated fluid content of the fault zone.(2)In the study of 2-D joint inversion of magnetotelluric and surface dispersion data sets,an approach for direct inversion 2-D profile velocity structure with surface wave dispersion data based on OCCAM inversion method is presented,and then we jointly invert the magnetotelluric and surface wave dispersion data profiles with crossgradient constraints.Three synthetic models,including homogeneous or heterogeneous models with consistent or inconsistent discontinuities in velocity and resistivity,are presented to gauge the performance of the joint inversion scheme.Our findings show that the volume effect of magnetotelluric method results in the low resolution at the boundary of anomaly,especially the structure below the low resistivity anomaly.The seismic surface wave has higher resolution in the reorganization of the structure of the anomaly.But in the inhomogeneous model,it can still be found that the method has a low resolution to the upper boundary of anomaly.Based on the structure coupling constrains method,the resistivity models have greatly improved.At the same time,the models recovered from joint inversion method overcome the problems caused by signal method.Joint inversion technique can make full use of the advantages of the two methods,and then improve the inversion resolution.In the field test application,the joint inversion results have better resolution than the separate results,and the joint inverted velocity and electrical images show in good agreement with the geological units,which validate the practical worth of joint inversion technique.(3)In the study of 3-D joint inversion of magnetotelluric and surface dispersion data sets,based on L-BFGS inversion algorithm and 3-D cross gradient function theory,we realized 3-D velocity structure directly from surface wave dispersion data,and jointly inverted the magnetotelluric and surface wave dispersion data profiles with cross-gradient constraints.Two complex models,with full consistent or partial consistent structural coupling,are applied to test the performance of the joint inversion scheme.After numerical tests,we find that the velocity models recovered by joint method has been improved in detail,although it is hard to recognize in the velocity maps.Also,no false exceptions in velocity model are found at the structural dissimilarity after joint inversion process.The resistivity models obtained with structure coupling constrains from velocity models have higher resolution than signal method,which is similar to the two-dimensional models.In the field test of Sulu orogenic belt,the results of joint inversion show that the resistivity and velocity of the Luxi uplift,the high-pressure matemophic belt,the intersection area of YWF and TLF have been improved to some extent,which validate the practical worth of joint inversion technique.