基于贝叶斯理论的多尺度地震反演方法  

Multi-scale seismic inversion method based on Bayesian theory

作　　者：杨千里[1] 吴国忱[1] 

机构地区：[1]中国石油大学(华东),青岛266000

出　　处：《地球物理学进展》2016年第3期1246-1256,共11页Progress in Geophysics

基　　金：国家科学基金石油化工联合基金重点项目(U1562215)资助

摘　　要：典型沉积旋回的模型分析表明,薄层调谐作用的影响抑制了地震相对高频的反射能量,降低了地震资料的分辨率.基于该资料的常规地震反演难以识别较薄的储层.为了提高地震反演的分辨率,本文利用小波变换分频技术,将地震资料分解为大、中、小三个尺度的数据体,有效减弱了调谐作用,增强了地震资料的相对高频和相对低频,提高了分辨率.在贝叶斯理论框架下,依据地震资料与沉积层序体的多尺度对应关系,逐级进行地震反演,把大尺度地震反演结果作为中尺度地震反演的模型约束,中尺度反演结果又作为小尺度地震反演的模型约束,最终完成整个反演.理论模型试算展示了该方法的逐级寻优特性,实际地震资料的多尺度反演表明,同常规稀疏脉冲反演方法相比,基于贝叶斯理论的多尺度反演对较薄储层的识别能力有实质性提升.The typical sedimentary cycle models analysis shows that thin-layer tuning effects inhibited seismic relatively high frequency reflection energy and reduced the seismic data resolution. The conventional inversion based on such data is difficult to identify the thinner reservoirs. In order to improve seismic inversion resolution,the wavelet-transform frequency division technology was used to decompose the seismic data into large,medium and small scale data volume in the study,effectively weakening the tuning function,enhancing seismic data relative high frequency and relative low frequency,improving the seismic resolution. In the Bayesian theory framework,according to the multi-scale correspondence relationship between seismic data and sedimentary sequence bodies,seismic inversion was carried out step by step, the large-scale seismic inversion results were taken as model constraint of medium-scale seismic inversion,while medium-scale inversion results were taken as model constraint of small-scale seismic inversion to complete the inversion in the end. Theoretical models trial verified step by step optimization feature of this method. Actual multi-scale inversion suggested that compared with conventional sparse pulse inversion method, multi-scale inversion based on Bayesian theory has substantial improvement in the identification ability of thinner reservoirs.

关 键 词：贝叶斯理论 小波变换分频 多尺度反演 薄储层 

分 类 号：P631[天文地球—地质矿产勘探]