Seismic Reflection Tomography Based On High Percision Velocity Analysis

As the important parament during the process of seismic exploration,the accuracy of velocity directly or indirectly determines the effect of tomography.In order to get a better inversion result through the tomography,the establishment of initial model is very important.We should build a velocity model which is more similar with the accrual layer so that it can reduce the number of iteration and also more convenient to calculate.So the velocity model that made by velocity analysis is a good choice.In order to get a high-accuracy initial model,we need considering the quality of the original seismic data and the accuracy of velocity analysis.In the process of seismic data collecting,there are many random noises in effective signal that caused by environmental factors.Denoising process is necessary to ensure the quality of seismic data.In this paper,a denoise method based on DDTF dictionary learning is used.We can get the sparse coefficient of data with noise.And get rid of smaller sparse coefficient by hard threshold.Then the denoised data is get by inverse transformation.On the other side,in order to avoid the energy clusters is non-convergence caused by a low resolution ratio of velocity spectrum and misidentification caused by adjoining energy clusters,this paper add least-square similarity in similarity-weighted velocity analysis.And we verified the effectiveness of this method by dealing with some simulate data.After improving the quality of seismic data and raise the accuracy of velocity,the velocity model is used in inversion of tomography.In this paper,we used the linear travel-time interpolation(LTI)method to calculate the ray travel-time matrix.Then use the first break inversion tomography to approach the real model.We apply this method in simple model,complex model and Marmousi model.Comparing with conventional method,the new method is more effective.