叠后MCMC法岩性反演算法研究  

Algorithm research of post-stack MCMC lithology inversion method

作　　者：王朋岩[1] 李耀华[1] 赵荣[1] 

机构地区：[1]东北石油大学地球科学学院,大庆163318

出　　处：《地球物理学进展》2015年第4期1918-1925,共8页Progress in Geophysics

基　　金：国家科技重大专项(2011ZX05028-002)资助

摘　　要：叠后MCMC(马尔科夫链蒙特卡罗)反演是一种地质统计学反演方法,该方法能够利用地震、测井等多尺度信息综合预测储层砂体,并使反演结果忠实于地震数据.算法的重点是在贝叶斯框架下,对样本点构建能够反映空间相关性的马尔科夫链,基于蒙特卡罗迭代算法实现对复杂后验分布空间进行有效搜索.文章以一维垂向视角阐述基于MCMC算法预测储层砂体的统计学原理,将反演过程所运用的统计学符号赋予地质含义,较为形象地将随机算法的数学意义与储层反演的地质意义联系起来.并在此基础上,以朝阳沟油田扶余油层未开发区朝65井所在剖面的砂体预测为例,验证叠后MCMC法岩性反演的可靠性.The post-stack MCMC（Markov Chain Monte Carlo）method is kind of a geological statistics inversion.It can be used to predict reservoir sand body with multi scale information which is more faithful to the seismic data. The MCMC algorithm emphasis on building the Markov chain within a Bayesian framework which can reflect the spatial correlation based on the Monte Carlo iterative algorithm.This paper concentrate on describing the sand body prediction＇s statistics principle based on the MCMC algorithm,which endows the statistical symbols with geological implication vividly.Basing on these,this paper take Chao65well＇s sand body prediction in Chaoyanggou Oilfield as an example to prove the reliability of the MCMC inversion method.

关 键 词：砂体预测 岩性参数 MCMC反演方法 贝叶斯框架 马尔科夫链 

分 类 号：P631[天文地球—地质矿产勘探]