Parallel strategy and source encoding technology applied in the frequency domain full waveform inversion

With further development of oil industry and the deepening exploration, the exploration is getting more and more difficult. New technologies are urgently needed for meeting the requirement of exploration. FWI is a new technique which develops rapidly by following this requirement. It takes into account the kinematic and dynamic information of prestack seismic data to reconstruct underground physical parameters, therefore it has potential ability to rebuild and detail the structure and lithology characteristics under complex geological background and precisely outline the reservoir. In recent years FWI has been widely recognized as the most precise technique to build the subsurface model. However, the real data case studies faced many difficulties, among which, a significant factor is that the huge amount of computational time required in FWI has not been completed solved for large problems. Recently, thanks to the intelligent algorithm and high-performance computing techniques make 3D FWI application feasible. In this paper, we will emphasis on the computational efficiency of the frequency domain FWI by applying multiple parallel strategies and in combination with phase encoding of shot records to accelerate it and make it feasible to be realized for large 3D model. Finally we adopt the scheme to a complex geological model contained a logo ‘SINOPEC’, and the results show the acceleration scheme not only are able to improve the computational efficiency but can obtain excellent inversion results.