Methodologies Of Multi-scale Seismic Full Waveform Inversion

Multi-scale full waveform inversion is an effective method for obtaining high-precision parameters of underground media stably and efficiently.With the difficulty of complex oil and gas reservoirs exploration and development grows continuously,reservoir geophysical methods and technologies are facing more severe challenges.Reservoir-oriented multi-scale full waveform inversion highlights the advantages of solving complex geological problems and has become one of the frontier topics in the field of exploration seismology.FWI covers the whole process of seismic forward modeling and inversion.There are many theory and application problems need to be solved.It is of theoretical significance and practical value to carry out the research of multi-scale seismic full waveform inversion method.This paper presents a thorough study of multi-scale full waveform inversion method.Based on the forward problem of multiscale medium model and combined with practical geological and geophysical problems,this paper studies the principle of multi-scale seismic full waveform inversion,establishes a prior constrained and high-order total variation regularized full waveform inversion method,and developes the strategy and method of joint inversion of full waveform and pre-stack seismic for reservoir.Multi-scale medium model seismic wave forward modeling is the basis of multi-scale full waveform inversion.Based on the theory of medium decomposition,this paper discusses the multi-scale characteristics and parameterization of the earth medium model,analyses the propagation characteristics of seismic waves in the stratum medium under resonance condition,and establishes a multi-scale response pattern of seismic waves based on energy flux density.Seismic wave forward modeling method is the basis of full waveform inversion.In order to improve the accuracy and efficiency of multi-scale medium seismic wave simulating,this paper derives attenuation matched layer wave equation,and establishes scalar wave equation high order finite difference forward modeling method.In the aspect of multi-scale full waveform inversion,this paper innovatively studies the multi-scale inversion method based on sparse representation by using the advantage of sparse representation in solving scale problems.Based on the theory of sparse representation,this paper realizes the solution process of sparse representation of seismic signal based on orthogonal matching pursuit,which improves the efficiency and accuracy of sparse decomposition of seismic signal.A multi-scale sparse decomposition method for seismic records is established,which overcomes the aliasing and interference effects of the seismic wave field and improves the time-frequency resolution of multi-scale decomposition of seismic records.Then this paper proposes a multi-scale full waveform inversion method based on sparse representation data decomposition.Combined with the multi-grid multi-scale strategy,the method recovers the small-scale disturbance components in the model through step-by-step multi-scale inversion on the basis of restoring the large-scale background of the model,and solves the problems of slow convergence speed,low computational efficiency and easy to fall into local minima in conventional full waveform inversion.Prior information constraints and regularization are effective methods to improve the accuracy of full waveform inversion.In this paper,abundant prior information of logging and geology in seismic exploration is introduced into full waveform inversion,and the initial model construction method of constrained correction of logging data is established,which provides a more reliable initial model for full waveform inversion.Well logging constraint term is added to the objective function of full waveform inversion to restrict the convergence process of full waveform inversion and improve the accuracy of full waveform inversion.Based on the characteristics of high-order total variation regularization constraining model sparsity,this paper proposes high-order total variation regularized full waveform inversion method.The split Bregman iteration algorithm is used to solve the objective function to improve the stability of the inversion and the accuracy of the inversion results.Reservoir-oriented seismic inversion is an important method and approach to obtain reservoir elastic parameters for reservoir prediction.In this paper,under the framework of Bayesian inversion theory,full waveform inversion and pre-stack AVO inversion are combined to improve the inversion accuracy of reservoir parameters by using full waveform inversion and pre-stack AVO inversion as prior information constraints and Cauchy prior constraints of model parameters.Joint inversion can solve geological problems that are difficult to solve by using only one method,and improve the ability of describing special lithologic bodies and predicting reservoirs.In summary,based on the multi-scale characteristics of the earth medium,this paper studies the multi-scale response of seismic wave based on energy flux density guidance,solves the seismic wave forward problem in multi-scale medium,proposes a multi-scale full waveform inversion method based on sparse decomposition,improves the calculation efficiency,solves the poor convergence effect and local extremum problem caused by strong nonlinearity.The accuracy of inversion is improved by synthesizing prior information constraints.This paper establishes a combined inversion method of full waveform and pre-stack AVO for reservoir,and achieves good application results.Full waveform inversion still faces many challenges,such as stability and computational efficiency of multi-parameter inversion,precise seismic wavelet acquisition and practical data preprocessing.However,it is believed that full waveform inversion will inevitably become one of the preferred technologies in oil and gas exploration and development.