Study On Acoustic Full Waveform Inversion Method Based On First-arrival Wave Time Modeling By Fast Sweeping Method

Full waveform inversion(FWI)is a high-resolution seismic inversion method.The core idea is to continuously update the initial model by transmitting the residual wave field back,so as to achieve the optimal match between the observed record and the simulated record.Therefore,it is necessary to provide a high quality initial model.While the first-arrival wave traveltime tomography has long been used in seismic inversion,although its fine structure imaging accuracy is lower,but has high computing efficiency and convergence speed,so the imaging results for full waveform inversion,is a good way to solve the dependence on initial model to the full waveform inversion and first-arrival wave imaging problem of low precision.Therefore,the research content of this paper is mainly divided into two parts.The first part is traveltime tomography: starting from forward modeling,this paper calculated the traveltime of first-arrival and reflected wave in 2D and 3D space based on fast sweeping method(FSM).For the problem of poor fitting of rectangular grid near the seismic source and wave front,the method of source thinning was used to compare and optimize.After calculating the departure time field,ray tracing is carried out by using the fastest descending direction of the travel time gradient,and the accuracy of the computed travel time field is also proved by the relationship between the ray path and the position in front of the wave.In the inversion part,the steepest descent method is used,and the adjoint state method based on FSM algorithm is used to obtain the gradient of the objective function.The simulation tests are carried out with different complexity models.The results show that the FSM algorithm has high efficiency and stability in traveltime tomography.The second part is full waveform inversion: firstly,the forward part deduces the high-order difference scheme of the acoustic wave equation,and analyzes the stability condition and dispersion phenomenon of the discrete process,and gives the difference formula within the PML absorption boundary.In the inversion part,L-BFGS method is used to calculate the Hessian matrix,the search step is obtained by backtracking method,and multi-scale inversion strategy is introduced to improve the calculation efficiency and stability of the inversion.Finally,the traveltime inversion results are used as the initial model for trial calculation,which can better reflect the overall structure and velocity variation of the model.The main research achievements of this paper are as follows :(1)FSM algorithm is implemented to obtain the travel time fields of first arrival and reflection waves in 2D and 3D space,and ray paths of these two waves are traced in 2D space;(2)The adjoint state method travel time inversion based on FSM algorithm is realized.(3)The multi-scale full-waveform inversion of acoustic wave in time domain is realized,and the trial calculation using traveltime imaging results shows that the results are good.