Journal of the Korean Society of Mineral and Energy Resources Engineers ISSN:2288-0291(Print) 2288-2790(Online) 한국자원공학회지

Prestack migration is widely applied to imaging subsurface for complex stratigraphic area such as faults, folds and salt domes, but heavy computing time and data IO (Input & Output) are indispensable. Seismic problems consist of function of time, velocity, source, receiver, frequency, and wavenumber. These have intrinsically parallel characteristic. Phase-shift-plus-interpolation (PSPI) is widely used for post- and pre-stack migration with less computing time and efficiency. The PSPI has intrinsically parallel in frequency domain, too. In this study, we have developed a MPI_PSPI code to paralleize serial PSPI with Massage Passing Interface_Local Area Multicomputer (MPI_LAM), and then we have applied it to synthetic seismic data for a simple circular model. When we conducted the MPI_PSPI, extrapolated wavefields were calculated simultaneously as distributed to every computing node by domain decomposition method. The numerical model test shows that the computing time with the parallelized MPI_PSPI code for a single shot gather was about 2 times faster than serial PSPI code and for 59 shot gathers it was 50 timtes faster. This means that the more we have shot gathers, the MPI_PSPI is more efficiency. For variance of computing time of MPI_PSPI according to increasing processors, rate of decreasing computing time is high when processors were used 8. Though we used more processors than 8, the rate of decreasing computing time was insignificant. Through the common image gathers, subsurface images were better at the energy concentration area. This would be helpful for determining acuqusition parameters.