一种稳健的弹性阻抗反演方法  

A stable elastic impedance inversion approach

作　　者：张丰麒[1] 金之钧[1] 盛秀杰[1] 孔令武 

机构地区：[1]中国石化石油勘探开发研究院,北京100083 [2]中海油研究总院,北京100028

出　　处：《石油地球物理勘探》2017年第2期294-303,共10页Oil Geophysical Prospecting

基　　金：国家油气重大专项"陆相页岩油资源和选区评价技术与软件实现"(2016ZX05049001-003)资助

摘　　要：利用弹性阻抗分解,从弹性阻抗体中可以直接提取与储层性质更为密切的弹性参数体。常规弹性阻抗分解算法是基于最小二乘原理,通过求取弹性阻抗分解矩阵的广义逆提取弹性参数,由于该分解矩阵的条件数较大,导致在含噪声的情况下无法获取稳定的弹性参数。结合贝叶斯定理,引入弹性参数的先验分布,构建弹性阻抗分解的正则化项,从而有效提高了弹性阻抗分解的稳定性;考虑到弹性阻抗分解是个带限过程,并且同一时间采样点的弹性参数之间并非独立不相关,因此引入多变量高斯分布描述弹性参数的自然对数更为合理;最后结合基于低频软约束的叠后反演形成一套弹性阻抗反演的流程。模型试算和实际数据测试验证了该弹性阻抗反演流程具有较强的稳定性和较高的精度。Elastic parameters, which reflect reservoir features, can be extracted from inverted elastic impedance with elastic impedance decomposition. However conventional decomposition methods based on the least squares principle are implemented by inversing the elastic impedance decomposition matrix. It causes unstable results in the presence of noise due to large number of matrix conditions. We propose a stable elastic impedance inversion approach in this paper. According to the Bayesian theory, the regularization term for the elastic impedance decomposition is added by introducing the prior distribution of the elastic parameters, which can stabilize the process of the elastic impedance decomposition. Since elastic parameter extract process is band-limited and these parameters are not independent, a multi-variable Gauss distribution is used to describe statistical feature of the nature logarithm of the elastic parameters. At last, a flow of elastic impedance inversion is created by combining with a poststack inversion based on low frequency soft constraint. Model and real data tests reveal that this flow can own the great stability and the high accuracy. © 2017, Editorial Department OIL GEOPHYSICAL PROSPECTING. All right reserved.

关 键 词：弹性阻抗分解 多变量高斯分布 贝叶斯定理 

分 类 号：P631[天文地球—地质矿产勘探]