3D Marine Controlled-source Electromagnetic Modeling And Inversion Theory And Research On The Mechansim Of Anisotropic Effect

As one of the main predrill reservoir appraisal methods, Marine Controlled-source Electromagnetic(MCSEM) has been widely applied in marine oil/gas exploration because of its high efficiency, low cost, and high resolution for targets. With the rapid development and high requirement of MCSEM in practical exploration, the forward and inversion theory of three-dimensional(3D) MCSEM has been a hotspot in electromagnetic exploration research. The current 3D forward researches in MCSEM are mostly based on the assumption that the seabed is isotropic or transversal isotropic(TI). The inversion techniques both in China and abroad are still inmature. The efficiency and accuracy of inversion are to be improved. Aiming at these problems, 3D forward modeling with arbitrary anisotropy has been studied in this dissertation. The impact mechanism of anisotropy on MCSEM is also analyzed. Furthermore, 3D inversion for MCSEM is realized through the L-BFGS method. The impact of different factors on 3D MCSEM inversion is also discussed, with the hope to provide support for improving the interpretation of MCSEM data.In this dissertation, the 3D forward modeling for MCSEM with arbitrary anisotropy is realized through the staggered finite difference method combined with two kinds of volume weighted methods. In the forward modeling, the electric field is first split into background field and secondary field. The background field is calculated based on the Green’s function of dipole source in layer media, while the secondary field is calculated through the finite difference method. Starting from the Maxwell’s equations in frequency domain, the Helmholtz equations is obtained and is discretized by staggered finite difference method, resulting in the linear equations system with the coefficients being a large complex sparse matrix. In the discretization procedure of Helmholtz equations, the portion of the current density related to the elements of the anisotropic conductivity tensor is discretized with the direct spatial weighted volume method, while the portion related to the off-diagonal elements is discretized with the weighted average of spatial current density method. To reduce the condition number of the equations system, the pre-conditioning method with D-ILU decomposition is applied to the large complex sparse matrix. The divergence correction technique is simultaneously introduced to improve the accuracy and converge speed of the solution. The results obtained by the algorithm proposed in this dissertation agree well with the 1D semi-analytical results, with the maximal relative error no more than 1.7%, which validates the high accuracy of this algorithm.The forward modeling of some typical 3D marine anisotropic geo-electric models is undertaken in this dissertation. Through the analysis of the Magnitude-versus-Offset(MVO) and Phase-versus-Offset(PVO) curve of Ex response, the surface distribution of the three components of electric field, and the distribution features of the electric field and current density in horizontal and vertical cross-sections, the following conclusions can be drawn. The anisotropy of seabed has obvious impacts on MCSEM. With the change of the reference conductivity in the xdirection of the seabed anisotropy, the distribution of both electric field and current can reveal the anisotropic features of the seabed, the flow pattern of electric current tremendously changes in the anomalies with high resistivity, and the magnitude of electric field is higher in the side including the highly resistive anomalies, which benefits the recognition of the anomalies with high resistivity. When highly resistive anomalies are included in the anisotropic seabed, the magnitude and phase dramatically vibrate, even without regular pattern. This shows the severe impacts of anisotropy on the electromagnetic responses of the anomaly. In shallow water case, with the change of the reference conductivity in anisotropic media, the intensity and direction of the guide wave in highly resistive anomalies show regular patterns and has stack and counteracting effects with the air wave. When the seabed is TI media, the guide wave in the anomalies has the largest intensity, hence the largest relative anomaly. Through the analysis of the shallow water case, the response feature of MCSEM differs tremendously in shallow and deep water case. Hence, the experience of data processing, inversion and interpretation in deep water cannot directly be applied to shallow water.3D MCSEM inversion is also realized in this dissertation through the BFGS optimization method with limited storage. Compared with the traditional inversion method, L-BFGS calculates the sensitivity matrix and its multiplication with vector through adjoint method, replacing the explicit approach in the sensitivity matrix calculation, which merely requires one forward and one adjoint forward in each iteration, thus requires relatively less storage. Based on the theory of regularization inversion, the calculation of the gradient of the objective function is briefly discussed in the dissertation. The basic principles and iteration method of L-BFGS optimization method is investigated systematically. The control condition of linear step searching and the step selection method based on Wolfe condition are also discussed. To improve the inversion accuracy, the Hessian matrix is pre-processed in this dissertation, that is, replacing the numerical field with the field calculated through the 1D layer media, so that the calculation error can be reduced. The inversion test of different inversion models, different measure configurations and different meshes shows that the inversion results is close to the real model in resistivity, spatial distribution and anomaly location, which validates the effectiveness of the L-BFGS algorithm in the MCSEM inversion.The research progress in this dissertation is a hotpot in current research. The achievement of this dissertation has significance in improving the interpretation of MCSEM data. As the development of MCSEM in China has begun later, the achievement will provide technical supports for the development of MCSEM technique in China.