Bend-ray Kirchhoff Prestack Time Migration And Parallel Computing

Migration and imaging in complex structure is one of main project in seismic exploration for long times, also there are many new methods such as CRS and people realize that prestack depth migration is the most efficient tool for complex structure imaging, but so far they are not used widely for many reasons .for depth migration ,the reason is it is more sensitive to velocity model.Prestack Kirchhoff time migration (PSTM) is the most popular migration technique in seismic data processing industry because of its simplicity, efficiency, feasibility and target orientated property, and geophysicists also pay more attention to improve the precision and other main problems of this method. In this paper , I will introduce some main problem in executing the pstm.Traveling time computation decide the precision of pstm in some way, and some seismic processing software such as Geodepth ,they calculate the traveling time based on straight ray, in this paper , I introduce the bend-ray traveling time computation that is on the basis of horizontal layer hypothesis, the bend-ray method is more close to the real situation of earth structure and can improve the precision of traveling time .besides, I also introduce the theory of non-symmetry traveling time by modern mathematic tool. It is derived by structure preserving algorithm of seismic wave propagation.Anti-aliasing is a main procedure in all migration methods, I analysis the kinds of alias that may appear in PSTM and pay more attention to the method of operator anti-aliasing, after compare many methods, I choose the Lumly's triangular smoothing to operator anti-aliasing.Besides the traveling time and anti-aliasing , I also show some important procedure in pstm such as amplitude , velocity interpolation and so on. these procedure is very important for the efficiency of this method. In this paper ,I also introduce the PC-Cluster and parallel computing and introduce two parallel methods that is parallel in"data domain"and"image domain".In last section, I present many read data results of this method to prove theefficiency.