Denoise Investigation on Prestack Reverse time Migration Based on GPU/CPU Collaborative Parallel Accelerating Computation

摘要

Based on two-way wave equation, prestack reverse time migration (RTM) overcomes dip limitation, and can image complicated geological body with high precision. However, since the massy storage, large amount of computation, low frequency noise, the industrialization process of this approach develops slowly. High order finite difference algorithm is used to calculate reverse time migration in the paper. By GPU/CPU collaborative parallel accelerating computation, the computation efficiency of RTM has been greatly improved. The random boundary condition is adopted to obtain wavefield information. The paper discusses the theory and specific steps of Laplacian denoise approach. The tests on Marmousi and Sigsbee model illustrate that, by Laplacian operator denoise approach, prestack reverse time migration can image complicated geological body effectively and precisely. This approach has a wide application prospect.