Optimization Algorithm Of Reverse Time Migration In Frequency Domain

Petroleum and natural gas are the national strategic substance,the reduction of the storage capacity forces people to search for more difficult oil and gas reservoirs,the requirements for exploration technology are becoming higher and higher.Seismic exploration and seismic data processing are the main means to find petroleum and natural gas.Prestack depth migration is an important part in seismic data imaging technology.The main research methods are divided into two categories:one is the Kirchhoff integral method and the other is the wave equation method which is based on wave theory of recursive continuation method for wave field.The Kirchhoff method has more advantages and more flexibility in dealing with irregular data,migration velocity analysis,computing efficiency and local imaging aperture control,the disadvantage is that there is a high frequency approximation and it is difficult to deal with complicated wave phenomena.The wave equation method is more natural and easier than the Kirchhoff method in dealing with complex wave phenomena such as multiple travels,caustics and amplitude preserving and the imaging accuracy is higher in complex structures.The reverse time migration imaging based on wave equation has become a mainstream migration technique in recent years because it is not restricted by the inclination of subsurface structure,has higher precision and more accurate amplitude.The disadvantage of reverse time migration is that the efficiency is low and the finite difference method has dispersion.Therefore,the emphasis of this method is to improve the computational efficiency and denoise.In this paper,we choose reverse time migration in the frequency space domain,which can avoid the dispersion caused by the time difference.In theory,it has better imaging precision than the time domain migration.In the space domain by using five point and nine point difference schemes approximate the second-order derivative,and a sparse frequency domain wave equations is obtained.Comparing with the conjugate gradient method,the LU decomposition method is faster,although it consume more memory.Therefore,the LU decomposition method is used to solve this equations group to obtain the extended wave field.Finally,the imaging is done according to the maximum correlation imaging criterion.Numerical results show that the dispersion of the nine point difference scheme is better than the five point difference,in order to improve the computational efficiency,a synthetic source record technique is used in this paper.Numerical examples show that the method is correct and effective.