Time-lapse Seismic Data Are Mutually Parallel Processing Algorithm

With the oil and gas exploration targets increasingly complex, the people realizeseismic data processing more and more importance, even more urgent need for highprecision prospecting technology. Parallel processing technologies obtains the extensiveattention of petroleum geophysics day by day, there is one of the urgently needed solutionmajor issues that how to parallel processing of massive earthquake data fast and efficiently.In this article, we take advantage of the CUDA platform to achieve operation of largeamounts of data at high complexity; based on Hadoop technology to handle the largeamount of seismic data; and apply a new technology of HadoopDB to processing of thetime-lapse seismic data, to deal with the seismic data with more complex relativelyrelationship.In recent years, high-speed and parallel pipeline of GPU draw and programmablefunction make GPU have great potential for application in the field of general-purposecomputing. In the article, we take use of GPU parallel computing technology to parallelprocesses the arduous task of computing seismic data of the phase correction andamplitude correction. The experimental results show that the GPU parallel computing canimprove the operational efficiency of the algorithm, as same as the method can be appliedto a variety of time-lapse seismic data inter-homogenized correction algorithm.Cloud computing is an efficient processing computing model for data-intensiveapplications; the Hadoop is a cloud computing software platform which is based onMapRecuce programming model. In the article, we base on Hadoop to achieve parallelalgorithm of time delay adjustment and matching correction. Large data set is divided intosmall data set and assigned to heterogeneous computer cluster, in result of large-scaleparallel processing of massive data. The experimental results show that the parallelalgorithm based on the Hadoop has obvious advantages in dealing with large data sets.HadoopDB is a technology with high-throughput, large concurrent, storage andanalysis calculated of mass data. It is a hybrid of mass data storage computing technologywhich has made the improvement on the basis of the Hadoop and used the RDBMS to save the data. In this paper, we presents a solution to apply the HadoopDB technology tocross equalization processing of time-lapse seismic data, and combine fundamentalconstruction and characteristic of HadoopDB with cross equalization processing oftime-lapse seismic data.Finally, we summary the paper work and discuss the outlook for further work.