A Method For Processing Seismic Data Based On Hadoop

Seismic data is a kind of digital signal which is generated by the earth's crust.It is used as a medium to explore the structure of underground rock and soil,and to search for oil and gas resources.The value of the immeasurable seismic data is significance.With the development of science and technology,more and more artificial seismic experiment was carried out,a large number of seismic data are took back to the data processing center after each high cost of field test,the importance of data storage and processing is in doubt.This paper introduces the data processing flow from the data collection to the data processing firstly,as well as the problems existing in the three aspects of seismic data storage,extraction,processing,and then analyzes the problems that low efficiency in the traditional way of storage and processing of seismic data in the big data environment.With the seismic data constantly doubled,the traditional method cannot efficiently complete the work of data processing.Then this paper put forward a method for processing seismic data based on Hadoop on the basis of the existing hardware in the data processing center to improve the efficiency of data storage and the efficiency of data processing by the way of distributed parallel computing.In the face of this problem,this paper proposes a distributed framework using Hadoop to solve the problem of data calculation pressure caused by big data,connect multiple servers together to form a distributed system to implement parallel computing.Use its unique MapReduce parallel computing model to achieve data block processing to solve the performance bottleneck of a workstation.Then use the seismic data conversion to introduce the advantage of Hadoop,compare the experiment result with different treatment methods for analysis of data conversion efficiency,and then fully embodies the great advantage of Hadoop clusters in the big data environment.By comparing the data of the case,this paper proposes that the distributed parallel architecture based on Hadoop is also applicable to the seismic ambient noise processing,then design and implement the seismic data processing method based onprivate cloud computing from the aspects of data storage,query and extraction in the following chapters.Because the seismic data is stored in the SAN disk in the data center,and the security and redundancy of the data are hidden and can not meet the requirements of the client to extract the data efficiently,so that the paper puts forward a distributed storage scheme of seismic data.By comparing with RDBMS,the advantages of distributed storage of seismic data are fully demonstrated,the structure of data storage and related mechanisms are also designed and implemented.Based on the research of HBase,we know that the problem of the efficiency of client query is appeared.In order to carry out the subsequent processing of seismic data more efficiently,I have studied factors affect query efficiency carefully,and ultimately through the HBase coprocessor to achieve data secondary index query to improve client data query speed,as well as effectively compensate for the deficiencies of some aspects of seismic data query.Then,this paper designs and implements the rapid extraction of seismic data based on MapReduce in distributed data storage environment.This scheme greatly improves the efficiency of data extraction by the traditional serial I/O,which lays the foundation for the next step of seismic data processing.Under the condition of distributed storage of seismic data and efficient parallel extraction based on MapReduce,the last chapter designs and implements the background noise processing algorithm based on Private Cloud Computing,which use Hadoop cluster model to realize parallel processing,and then analyzes the factors that influence the efficiency of data processing by MapReduce operating nodes and draw a conclusion through experiment in the end.