论文部分内容阅读
目前在地震资料处理技术的相对保幅性评价方面,还缺少行之有效的方法。实际资料由于受到噪声和地表变化等因素的影响,不确定因素比较多,对处理技术的相对保幅性难以进行定量评价。针对这种情况,提出了基于正演模型的振幅曲线统计法、残差法、AVO属性分析法和振幅比法等4种保幅性评价方法,即利用正演数据,在影响因素比较单一的条件下,采用相应的保幅性评价方法,对比分析了处理前后正演数据的变化,对几何扩散补偿、预测反褶积和地表一致性振幅补偿处理技术的保幅性进行了评价。速度场精度对几何扩散补偿处理的相对保幅性有较大影响,采用准确的速度是提高处理保幅性的基础。预测反褶积虽然能提高分辨率,但相对保幅性较差,实际资料处理时应结合井资料进行细致的参数分析,提高处理的相对保幅性。对于地表一致性振幅补偿处理,参与运算的炮数越多,处理结果的相对保幅性越好,实际资料处理时,应该利用尽可能多的数据进行运算。
At present, there is still a lack of effective methods for the relative amplitude preserving evaluation of seismic data processing technologies. Due to the influence of noise and surface changes, the actual data have more uncertainties and it is difficult to quantitatively evaluate the relative amplitude-preserving effect of processing technologies. In view of this situation, four kinds of amplitude-preserving evaluation methods, such as amplitude curve statistics method, residual method, AVO attribute analysis method and amplitude ratio method, are proposed based on forward modeling. By using forward data, Under the conditions, the corresponding amplitude-preserving evaluation method was used to compare and analyze the changes of the forward data before and after the processing, and the amplitude-preserving performance of geometric diffusion compensation, predictive deconvolution and surface-consistent amplitude compensation processing was evaluated. The accuracy of the velocity field has a greater impact on the relative amplitude-preserving of the geometric diffusion compensation processing. The use of accurate velocity is the basis for improving the amplitude-preserving treatment. Although the prediction of deconvolution can improve the resolution, but relatively poor amplitude-preserving, the actual data processing should be combined with well data to conduct detailed parameter analysis to improve the processing of the relative amplitude-preserving. For the amplitude compensation of surface consistency, the more shots involved in the operation, the better the relative amplitude-preserving performance of the processing result. When the actual data is processed, the data should be calculated using as much data as possible.