基于叠前地震反射模式分析的复杂储层识别方法研究

Pre-stack seismic data based seismic reflection pattern analysis methods can make full use of reflection information (such as amplitude changes with offset or azimuth), which facilitates the study of complex reservoirs, especially unconventional reservoirs, and has great significance to promote the fine and quantitative development of complex reservoir prediction. In this project, a variety of prestack reflection feature extraction methods are proposed based on the prestack superset optimization processing method learning from the methods of image processing and multidimensional data analysis; Multi-parameter fusion, discriminant analysis and deep learning algorithm are used to study the optimization method of reflection feature; Focusing on the problems of prestack data noise, horizon interpretation error, stratum thickness variation and interpretation information, lack of interpretation information, the study of high adaptation prestack seismic reflection pattern analysis method is carried out to realize fine characterization and recognition of complex reservoirs. The main innovations include: multi-parameter and multi-attribute complementary fusion method, robust prestack seismic reflection pattern analysis method for horizon interpretation noise, prestack seismic reflection pattern analysis method for adaptive formation thickness, semi-supervised prestack seismic reflection pattern analysis method.

基于叠前地震数据进行地震反射模式分析可以充分利用叠前数据中包含的振幅随着偏移距和方位角变化等信息，有利于复杂储层特别是非常规储层的研究，对促进复杂储层预测向精细化、定量化发展具有重要的意义。本研究在叠前超道集优化处理的基础上，借鉴图像处理和多维数据分析的若干手段，提出多种叠前反射特征提取方法；采用多参数融合、判别分析和深度学习方法对反射特征优化方法进行研究；围绕叠前数据噪声、层位解释误差、地层厚度变化和解释信息难以融入等问题，开展高适应叠前地震反射模式分析方法的研究，实现复杂储层精细刻画和识别。主要创新包括：多参数多属性互补融合方法；层位噪声鲁棒的叠前地震反射模式分析方法；自适应地层厚度的叠前地震反射模式分析方法；半监督的叠前地震反射模式分析方法。

针对复杂油气藏开发难度较大，储层分析多解性强的问题，本项目研究基于叠前反射模式的复杂储层识别方法。围绕叠前数据噪声、层位解释误差、地层厚度变化和解释信息难以融入等问题，开展高适应叠前地震反射模式分析方法的研究，实现复杂储层精细刻画和识别。本项研究取得的主要成果包括：基于动态时间规整的道集叠加方法；地层空间位置约束的层位追踪方法；不敏感于层位噪声的地震相分析方法；适用于储层厚度变化的地震相分析方法等。