The Extraction And Optimization Of Pre-stack Seismic Attribute

Seismic data contains abundant information on lithology,physical property and hydrocarbon of subsurface. But it needs processors and interpreters extract them by all means of methods. Seismic attribute is subtracted from seismic data. It is specially used for the measurement of seismic data in its geometry, kinematics, dynamics or statistics feature. Seismic attribute represents the subset information contained in seismic data and so it is the most important means of obtaining stratum parameters. We can get relatively accurate information of stratum and hydrocarbon in the reservoir from seismic attribute analysis. Therefore, studying on seismic attribute has some theoretical and practical values.Firstly this paper discusses a pre-stack seismic attribute analysis with an optimal model-based linear AVO approximation method. By extending the validity range of linear approximation to larger angles, I verify if this approach helps when using information contained at near-critical offsets. The optimal basic functions are calculated by applying singular value decomposition to realistic modeled AVO reflect coefficients. I derive functions that use the coefficients represent new AVO attributes to obtain more accurate estimates of usual AVO attributes. I compare the difference when estimating seismic-parameter contrasts at reflecting interfaces by using Shuey approximation and optimal approximation through model data. I apply this method to practical data and compare the result estimated with two approximations. It proves the advantages of the new method. Secondly, this paper talks about new method of extracting prestack seismic attributes and then calculating their corresponding differences from partial angle stack seismic data. I analyze seismic attributes differences extracted from different angles and study cross equalization method on partial angle stack seismic data based on that of time lapse seismic data. I put out method of calculating seismic attribute difference. Moreover, I summary some kind of method used to calculate prestack seismic attributes from partial angle stack seismic data. I prove the feasibility of this method through the effectiveness on model and practical data. Finally, this paper discusses seismic attribute optimization methods, and put most of the stress on kernel principal component analysis method (KPCA). I give a detailed statement about it's theory and reveal it's differences from principal component analysis (PCA). I introduce three types of kernel functions that generally used and talk about the selection principal on kernel functions and the parameters of kernel functions. I apply this method to process practical seismic attributes data and prove it's superiority by comparing the result that calculated from PCA.