基于密度的含噪声角道集波形聚类地震相分析  

Seismic facies analysis for angle gathers based on DBSCAN waveform clustering

作　　者：刘仕友 宋炜[2] 应明雄 孙万元 汪锐 牛悦 LIU Shiyou;SONG Wei;YING Mingxiong;SUN Wanyuan;WANG Rui;NIU Yue(Zhanjiang Branch,CNOOC Ltd.,Zhanjiang 524057,China;China University of Petroleum,Beijing 102249,China)

机构地区：[1]中海石油(中国)有限公司湛江分公司,广东湛江524057 [2]中国石油大学(北京),北京102249

出　　处：《石油物探》2019年第5期773-782,共10页Geophysical Prospecting For Petroleum

基　　金：国家科技重大专项(2016ZX05026-002)资助~~

摘　　要：基于地震沉积学理论的常规地震相划分主要利用均方根振幅属性,该划分方法未综合考虑与地震波形变化相关的频率和相位信息,因而导致地震相划分不精确。为此,将基于密度的含噪声应用波形空间聚类(DBSCAN)方法应用于地震相聚类分析,该方法的主要步骤为:①基于地震沉积学原理的基准参考面追踪和等时地层切片划分;②沿地层切片按固定时窗提取地震特征波形;③地震特征波形的标准化;④基于地震特征波形的DBSCAN参数优选及分析。由于DBSCAN方法无需提前确定聚类类别数,因此适用于无法提前确定地震相类别的地震相分析。基于密度的含噪角道集波形聚类地震相分析方法应用于三维地层物理模型和实际资料的结果表明,采用DBSCAN方法进行波形聚类分析,可有效地提高地震资料地震相分析的精度,在综合考虑AVO效应差异的前提下,该方法不仅可以提高地震相划分的可靠性,还可以区分储层含气性差异。Conventional division of sedimentology-based seismic facies uses mainly the root mean square amplitude attribute.Because the frequency and phase information related to the changes of seismic waveform are not comprehensively considered,the seismic phase division is not accurate.In this paper,a seismic facies analysis based on density-based spatial clustering of applications with noise(DBSCAN)is introduced.It includes four main steps:①tracking reference layers based on seismic sedimentology principles and division of isochronal stratigraphic slices;②extracting seismic characteristic waveforms along stratigraphic slices according to a specific time window;③standardizing seismic characteristic waveform;and④optimizing clustering parameter and clustering analysis based on DBSCAN.Because DBSCAN waveform clustering does not need the number of clustering categories to be specified ahead of time,it is especially suitable for seismic facies analysis in which the exact types of seismic facies cannot be known in advance.Application in a physical model and real data illustrated that DBSCAN waveform clustering,which uses seismic waveforms extracted along with stratum slices as input vectors,can effectively improve the accuracy of seismic facies analysis.When considering the difference of AVO effect,the proposed method can increase the reliability of seismic facies division and effectively distinguish the gas-bearing differences.

关 键 词：机器学习 DBSCAN 波形聚类 地震相 地震沉积学 地层切片 

分 类 号：P631[天文地球—地质矿产勘探]