科研之友
微信
新浪微博
Facebook
分享链接
摘要
Three dimensional surface-related multiple elimination (SRME) is one of the important topics in the processing of seismic data from marine exploration, theoretically, the data-driven SRME based on wave-equation, can suppress all surface-related multiples from complex structure, both in 2D and 3D sense. But actually, because of the high requirement for seismic data acquisition, it is usually difficult to apply 3D SRME for field data demultiple processing. 3D multiple suppression approach based on sparse inversion is analyzed.
We classify multiple suppression method into two categories, filter and SRME method respectively. For the seismic data from complex geological structure, filter approach doesn't work well. However, the data-driven SRME approach based on wave equation, can suppress multiple better, which has no requirement for velocity information. For SRME, the full wavefield data requirement is an important disadvantage, which cannot be meet for almost all marine field data. Therefore the data reconstruction is necessary for traditional multiple suppression using SRME. For current marine acquisition geometries, the data is densely sampled in the inline direction, but very sparsely in the crossline direction, we introduce contribution gather concept, and calculate sparsely sparsely sampled crossline multiple contribution by means of sparse inversion algorithm. Therefore, the data reconstruction is unnecessary before demultiple, comparing with traditional 3D SRME algorithm, which decrease storage cost greatly. Based on the assumption that the crossline time-distance curves are hyperbolic or parabolic, i.e., after integrated along inline direction after finished the first step processing of contribution gathers, we applied the phase correction algorithm which based on the principle of stationary phase approximation, predicted multiples by the proposed method and result by the full 3D SRME method are basically the same on kinematics and dynamics characteristics.
Multiple prediction and adaptive subtraction are two crucial steps for 3D multiple suppression using SRME. We simulate complex model data to test the proposed 3D multiple prediction algorithm, and 10 thousands shot records are modelled, the shot interval and trace interval are both 25 m. The result comparison show the 3D multiple prediction approach can predict the multiple's amplitude and phase correctly, and also, the subtraction result is superior than 2D algorithm. The horizontal four-layered media is also designed to test the sparse inversion 3D multiple suppression, algorithm, there are 3136 shot records in total, the trace and shot intervals are both 25 m, the line interval is 75 m. The single trace and common-offset result show that the proposed approach can predict the multiple's amplitude and traveltime correctly, and they are very close to the full data circumstance. The test on field data from some area in China show that the proposed sparse inversion method is applicable and effective, where the trace interval is 12.5 m, shot interval is 50 m, and the line interval is 100 m. After summation along inline direction, the partial integration data is transformed to Radon domain using apex-shifted Radon approach based on the assumption of hyperbolic or. parabolic events, stacked the Radon imaging, and also applied the phase correction to the sparse inversion solution, the predicted multiple is acquired.
After theoretical investigation and data tests, we have the following conclusions, (1) the proposed method is suitable for simple and complex 3D model data, (2) for real seismic data, the inline direction reconstruction is needed, (3) because only forward Radon transform, not inverse Radon transform is used, phase correction is demanded, and (4) the algorithm do not rely on the assumption of full data using sparse inversion approach, and also, widen the application extent of 3D multiple suppression. The result show that the proposed 3D multiple suppression algorithm can improve the S/N ratio during the course of preprocessing, and provide the high quality data for the subsequent high resolution imaging.
- 出版日期2015-7
- 单位东北石油大学; 中国石油大学（北京）
