Well-log And Seismic Constained Joint Inversion Of Magnetotelluric And Gravity Data

The study of the joint nonlinear inversion between 2D/3D magnetotelluric(MT)and gravity data constrained on the known logging,seismic and geological data has important theoretical significance and practical value,which can reduce the ambiguity of the geophysical data interpretation and improve the resolution to deep target.In this paper,the Well-log and seismic constrained nonlinear joint inversion of MT and gravity datais studied.The research is carried out from the following aspects.Firstly,a real-time and human-computer interaction 2D and 3D modeling software with multi-attribution and multi-parameter for this study is developed based on the Qt4 platform and C++ language compiler.It can be used to build 2D and 3D MT-gravity model quickly and accurately,which provides a basic support for the 2D and 3D MT-gravity constrained joint inversion.Twice interpolations method is proposed and realized,that is,different interpolation methods are chosen for the geometric and attribution values of the mesh to ensure the internal attributions of the strata is not changed when the geological boundary changes.A progressive modeling method is proposed.The 3D model is yielded by building several main 2D sections.This modeling method is simple and reliable,easy to be implemented without considering the complex techniques of 3D curved surface cutting and topological polyhedron constructing,which can satisfy the need of 3D MT-gravity data modeling.Secondly,the parallel forward modeling of 2D and 3D MT,gravity is developed.The parallel forward algorithms of the 2D and 3D MT,gravity data are designed respectively based on MPI(Message Passing Interface)platform.Through parallel design by frequency,the calculation speed of 2D MT finite difference forward modeling is tens to hundreds of times faster than the serial finite element method.The relative error between 3D MT finite difference modeling results and the 3D integral equation method is less than 2.2%.The calculation speed of 2D and 3D MT forward modeling is linearly related to the number of CPU involved in the parallel computing.The speed of 2D and 3D gravity forward modeling is improved by the parallel design according to the measured station.Thirdly,the parallel joint inversion methods of 2D and 3D MT-gravity artificial fish swarm algorithm based on well-log and seismic constraints are proposed and realized.The joint inversion is based on the relationship between resistivity and density.The artificial fish swarm intelligence algorithm is developed for the inversion of 2D MT,2D gravity,3D MT,3D gravity,2D joint MT-gravity and 3D joint MT-gravity.The correctness and feasibility of the inversion methods are verified by several 2D and 3D models.For the 2D and 3D joint MT-gravity parallel inversion,the resistivity and density values from joint inversion are more concentrated than those obtained by single MT data or single gravity data inversion.This shows that the joint inversion of 2D and 3D MT-gravity data is better than single method inversion,and the joint inversion can effectively improve the inversion accuracy of single method inversion.Lastly,the parallel joint inversion technique for 2D MT-gravity data is successfully used to find the deep rift and the favorable oil and gas target in central Sichuan basin.Based on the analysis of logging,core and field measured data,the relationship between resistivity and density for each formation is proposed and established,which provides the bridge for the joint inversion of MT and gravity data.Two lines are processed by 2D MT-gravity joint inversion.The results show that based on well-log and seismic constrained,the precision and resolution of deep target are improved.The thickness of Nanhua period strata in the study area is interpreted,and the favorable oil and gas accumulation zone is inferred.The predictions are confirmed by two wells.