Calculation Of Common Imaging Point Set In Offset Domain Based On Reverse Time Migration

With the development of the oil and gas mineral industry,the amount of geological exploration data gradually becomes larger and the target body becomes more complex.In order to more accurately obtain more underground geological structure information and obtain better explanation effects,the time domain reverse time migration Imaging is a good choice,and it is an important part of the seismic data processing process.At this stage of processing,huge amounts of computation will be generated,which is the core factor that limits this method.Offset theory has been proposed very early,but has not been well applied due to computing bottlenecks.With the development of modern computer hardware,the reverse time migration method can be effectively applied to the actual data processing.With the development of seismic exploration techniques,migration methods have become more and more.Among these migration methods,the main selection in this paper is the time domain full wave field inverse time migration method.The paper not only analyzes the time domain reverse-time migration method,but also introduces the rapid development of GPU high-performance computing at this stage.Through the highly parallel computation of the GPU,the calculation speed of reverse time migration is improved.The combination of GPU computing technology and seismic reverse-time migration method reduces the computational time by orders of magnitude,so we can do more detailed processing.The core of the GPU optimization algorithm is to reduce the access to global variables and reduce the judgment statements.The algorithm has the potential of high parallel processing and makes full use of "shared memory." This paper proposes the computation of offset-domain common-image gathers based on GPU-accelerated reverse time migration.The implementation of this method depends on the maturity of the reverse-time migration computing capability.By using the equivalent offset imaging section of the reverse time migration,a new calculation of the gather is realized.The main research content has the following aspects:(1)The forward modeling of the second-order differential full-wave field equation in the 12-th order difference and time domain of the spatial domain,the analysis of the stability condition,the choice of wavelet and the choice of boundary conditions are introduced.(2)The GPU-based two-dimensional horizontal surface and undulating groundbased reverse-time migration programs were rewritten,and the implementation and flow of the reverse-time migration algorithm were studied.By improving the GPU occupancy rate and reasonable memory allocation,the calculation time of reverse time migration is improved.(3)A set of common-image point gathers based on GPU-optimized reverse-time migration offset domain was studied and proposed.Through the equivalent offset transformation and gun data segmentation according to the offset range,the regularization of the offset front gather is realized;the calculation time is saved by large step extension calculation;and the GPU-based implementation is achieved by changing the imaging conditions.The calculation and output of common offset gathers in the offset domain of accelerated reverse time migration.(4)When calculating the imaging profile of the shot collection data,the traditional method needs to output the data of each shot,and the storage space of the demand is large.This paper saves the storage space by storing the intermediate calculation results in the memory of the GPU.Through theoretical derivation,the common image point gathers in the offset domain are calculated and obtained.By comparing with other gathers,the advantages of this gather are obtained.Provides data for later seismic AVO and velocity analysis.