Research And Implementation Of Seismic Attribute Data Process System Based On Hadoop

Multiple seismic attribute data processing has important applications in geological exploring.Usually,multiple seismic attribute data processing owns characteristics as high computational complexity and large amount of calculation.Using signal computer to calculate needs too much computation time due to the limited calculation speed.In order to solve this time-consuming problem,distributed computing approach with Hadoop is proposed in this work.Single seismic attribute is saved as SEGY file format independently while multiple seismic attributes will be uploaded and distributed to different nodes in HDFS.Generally,a variety of seismic attributes will be used in calculation,causing the increasing cost of data transfer between nodes.By the way of altering save file format of seismic attribute data as well as extracting and combining the seismic attributes meanwhile,unnecessary data transfer between nodes in cluster can be avoided.To design the seismic attribute processing solution based on Hadoop,calculation process is divided into five stages: seismic attribute data format conversion,data upload,seismic attribute processing,data download and result data format conversion.Seismic attribute data format conversion and result data format conversion is in a single computer,data upload and data download is single computer and the Hadoop cluster for data interaction and seismic attribute processing is calculated by MapReduce program of Hadoop runing on Hadoop cluster graphs.Experimental results show that the seismic attribute processing solution is efficient and the time cost can be shortened by 70% by utilizing eight DataNodes MapReduce calculation to complete seismic attributes processing compared to the single stand-alone computing.