Research On Efficient Seismic Data Acquisition Methods Based On Compressed Sensing Theory

Nowadays,oil and gas exploration has been going deep from shallow to deep areas,and regional geological structure has become more and more complex,making exploration more difficult,and the seismic data obtained is often incomplete and irregular,which has a great influence on the subsequent processing and interpretation,also,the judgment of the result will also produce deviation,which will cause a lot of trouble and waste of cost and time.Because of the limitation of traditional Nyquist sampling theory,some traditional reconstruction methods can not get the ideal result for the serious missing data,meanwhile,the request of sampling rate is very high,so it is much difficult to operate in actual production.In recent years,the presentation of compressed sensing theory has provided a new way for this difficult problem and been developed rapidly,which combines the acquisition of seismic data in production with the compression of data in processing effectively,using the sparse characteristics of seismic signal in the transform domain to realize incomplete data reconstruction,breaking through the limitation of traditional sampling,and dramatically reducing the sampling cost and sampling points.It not only saves labor and material resources,but also improves efficiency.Based on the key element,sampling matrix in theory,this paper studies the different methods of seismic data acquisition,how to reconstruct the ideal data efficiently is the key of our research.According to the properties of compressed sensing sampling matrix,many kinds of sampling matrices have been proposed,which can mainly be divided into two types: random sampling matrix and deterministic sampling matrix.The kind of random matrix is simple to construct,but due to uncertainty elements,it occupies a large amount of memory space,which is not suitable for hardware implementation.The other kind of deterministic sampling matrix occupies less memory space,but under the same condition,the performance is worse than that of random sampling matrix.Both of two kinds of matrices are widely used in image reconstruction,communication engineering,medical imaging andother fields,but in seismic data acquisition,most of these matrices are not satisfied with the actual exploration situation,since the sampled points are regarded as 1,and missed points are regarded as 0,the elements of most matrices are not consilient.Therefore,a new random acquisition scheme is introduced in this paper,the LDPC sampling matrix,which is widely used in image processing,is here used into the seismic acquisition.It satisfies the requirements for being as compressed sampling matrices in seismic exploration.After testing by simulated and field data,the results show it is reasonable and feasible,at the same time,it is more suitable for the actual3 D exploration environment compared with other methods.Then,we extend efficient acquisition methods of single source data acquisition to multi-source seismic data acquisition system to improve acquisition efficiency.In modern seismic data acquisition,the source excitation methods are generally divided into two types,single source excitation and multi-source excitation.The main difference between these two ways is the number of excitation sources,generally speaking,both of them are based on the regular dense measurement network for data acquisition.Here,we introduce the jitter sampling method of single source data acquisition to multi-source data acquisition to arrange the positions of shot points and trace points in the traditional regular dense network,simulating the actual situation of bad track or missing points,finally via a reasonable algorithm to reconstruct the missing data,the test results can be seen that the reconstruction results and original complete records are not much different,in other words,there is no great error in the subsequent processing between them.Therefore,in actual production,we can decrease the acquisition cost by reducing the traces and shot points and minishing the work difficulty caused by the actual complex exploration environment.Multi-source has the advantage of time cost compared with single source,so collecting data by efficient methods,both exploration cost and acquisition efficiency will be improved greatly.