Surface-Related Multiple Elimination With Consideration Of Sources And Receivers Properties

Occurrence of multiples is still an important problem in seismic data processing, especially in marine case. The influences of multiples include: (1) building velocity model, (2) seismic imaging, (3) seismic attribute analysis (including AVO analysis), and (4) structural interpretation and so on.Various methods have been proposed during the last decades to resolve the problem of multiples occurrence. The surface-related multiple elimination approach has became quite successful in the recent years. The conventional SRME method predicts and subtracts the multiples adaptively in the way of time-spatial convolution, without the consideration of the properties of the sources/receivers, but this conflicts with the practice to some extent.To resolve this problem, this thesis takes the sources/receivers properties into account and constructs the forward model for the seismic data and introduces two types of surface operator: (1) when parameterizing variations, it’s a diagonal matrix; (2) when parameterizing the directivity, it becomes a Toeplitz band matrix. Both types of the surface operator inversion were defined as a linear least-square problem and solved using the singular value decomposition method. Finally with the help of the surface operator, we can inverse and eliminate the multiples.In the surface operator inversion, this thesis introduces the time-windowing approach into the surface operators optimizing. With the new surface operators, the multiples can be inversed and eliminated.The method above has been tested on synthetic data, and with the consideration of variation and directivity of sources/receivers, we got the promising results of the multiple inversion and elimination in the iterative way. But the real data example doesn’t showing the same results as synthetic data. The main reasons are the length of the time window and the magnitude of condition number in SVD.