摘要： 超临界CO2 注入到地下储集空间后会引起封存空间属性的强烈变化,此时会在叠后时移地震数据上表现出明显的时移响应信息.时移波阻抗反演是利用叠后时移差异数据计算差异波阻抗,识别储层变化的一种有效方法.储层参数变化通常是局部性的,因此差异波阻抗通常会表现出明显的块状特征,即差异波阻抗反射率具有稀疏性质.常规时移波阻抗反演方法中使用的L2 或L1 正则化约束项的稀疏性不足,导致对于时移差异的边界刻画不清晰的问题.本文提出将具有更加稀疏特性的L1-2 范数作为先验约束条件加入差异波阻抗反演中以增强反演结果在阻抗界面处的刻画清晰程度.通过模型测试分析,L1-2 范数约束相比L1 范数和L2 范数在垂向上具有最高的反演分辨率.同时为了克服采用单道反演方法可能造成的反演结果横向连续性差的问题,本文引用f-x滤波方法对结果进行改善.本文方法应用于模型数据和挪威Sleipner深部咸水层CO2 地质封存的结果表明,基于L1-2 范数约束的时移波阻抗反演方法能够有效且精确地表征时移地震差异响应,可以用作CO2 地质封存监测的一种手段.

关键词: 差异反演, L1-2范数, 贝叶斯框架, CO2地质封存, 稀疏

Abstract: Injecting supercritical CO2 into underground reservoir space significantly alters the properties of the sequestered area,resulting in obvious time-lapse responses on post-stack seismic data.Time-lapse wave impedance inversion is an effective method to identify reservoir changes by calculating differential wave impedance with post-stack time-lapse differential data.Since reservoir parameters lapse are usually localized,the differential wave impedance often displays clear,block-like features,meaning that the reflectivity of differential wave impedance tends to exhibit sparsity.And due to the insufficient sparsity of the L2 or L1 regularization constraint term used in the conventional time-lapse wave impedance inversion method,it results in the prob-lem of unclear boundary delineation in time-lapse difference imaging.In this paper,we propose to add L1-2 norms with stronger sparsity-promoting characteristics as a prior constraint into the differential wave impedance inversion to enhance the clarity of the inversion results at the boundary clarity.The L1-2 norm constraint improves the sharpness of the inversion results,especially at impedance interfaces.Through model testing and analysis,it is found that the L1-2 norm constraint provides the highest inversion resolution in the vertical direction when compared to L1 and L2 norms.Additionally,in order to overcome the problem of poor transverse continuity of the inversion results that may be caused by using the single-channel inversion method,this paper adopts the f-x filtering method for result enhancement.By applying the proposed method to the model data and the CO2 geological storage data in the deep saline layer of Sleipner,Norway,it is shown that the time-lapse wave impedance inversion method based on the constraints of L1-2 norms can characterize the time-lapse seismic differential response efficiently and accurately,and it can be used as a means to monitor the CO2 geological storage.

Key words: difference inversion, L1-2 norm, Bayesian framework, CO2 geological storage, sparse

中图分类号: 

• P631.4 P618.13