Quasi Two-dimensional Joint Inversion Of CSAMT And Microtremor Survey Data

Joint inversion is to use a variety of geophysical observation data at the same time in the inversion,and solve the geological and geophysical models that meet the observation data of different methods through the inherent relationship of physical properties or structural coupling existing in different geophysical exploration methods.Compared with single inversion method,joint inversion can make full use of different physical parameter information,effectively reduce the multi solution of inversion,and realize the complementary advantages of different methods.Microtremor survey belongs to natural source surface wave exploration.It is a new geophysical exploration technology with strong anti-interference,high vertical resolution and large detection depth.In recent years,it has been widely used in urban areas with large interference.However,microtremor survey is a natural surface wave noise source,and the inversion accuracy is lower than that of conventional seismic exploration.Controlled source audio frequency magnetotelluric method belongs to high-power artificial source electromagnetic method.It has the advantages of strong anti-interference,little influence by terrain and high lateral resolution,but it is also affected by static effect and near-field effect.According to the characteristics of the two methods,the joint inversion code of CSAMT and microtremor survey is developed based on the similarity of the spatial distribution of resistivity and velocity model parameters.The emphasis of this paper is to add lateral constraint in joint inversion.Firstly,the theory of lateral constraint inversion is studied,the research idea of two-dimensional forward and one-dimensional inversion is established,and the construction formula of lateral constraint matrix is derived.Then,one-dimensional and two-dimensional forward modeling of CSAMT and microtremor survey are studied,focusing on the Fast Hankel transform algorithm,knopoff fast algorithm and the synthesis of twodimensional dispersion data.On this basis,the finite memory quasi Newton method is used to construct the objective function with lateral constraints,and the gradient expression is derived to complete the quasi-2D inversion of CSAMT and micro motion detection.Through the inversion of synthetic data of the designed theoretical model,the results show that,compared with the one-dimensional inversion,the quasi twodimensional inversion results can effectively reduce the sudden change of physical parameters between adjacent measuring points,and the lateral continuity is significantly enhanced;Compared with 2D and 3D inversion,the inversion speed is significantly improved.It is also found that the inversion speed is generally faster than1 D inversion by strengthening the lateral constraint weight.Based on the realization of single inversion and the spatial distribution similarity of resistivity and S-wave velocity,the objective function of joint inversion is constructed by using the finite memory quasi Newton method,and the gradient expression of the objective function is derived.The code of joint inversion of CSAMT and micro motion detection is completed,Through the inversion of the synthetic data of the theoretical model,compared with the quasi-2D inversion of the single method,the joint inversion improves the problem that the thickness inversion of the single method with two parameters is not accurate enough,and the inversion results are closer to the real model,which also verifies the correctness and effectiveness of the joint inversion algorithm of CSAMT and micro motion detection.