多尺度地震资料稀疏贝叶斯联合反演方法

针对地面和井中地震资料在不同尺度上置信度不同的问题，提出了多尺度地震资料稀疏贝叶斯联合反演方法。利用尺度分解方法分析多尺度地震资料与井旁合成地震记录的匹配度，以此为标准确定各种资料对应的尺度约束算子。从贝叶斯理论出发，进一步假设模型参数服从柯西先验分布，保证了反演解的稀疏性。引入尺度约束算子，建立多尺度地震资料稀疏联合反演目标泛函，采用PRP（Polak-Ribiere-Polyak）共轭梯度进行优化求解。模型测试和实际资料处理表明，多尺度地震资料稀疏贝叶斯联合反演方法有效利用了地面地震资料和井中地震资料中高置信区间内的尺度信息，为面向储层的综合研究提供了高精度的反演结果。

A sparse Bayesian joint inversion of multi-scale seismic data

Joint inversion of multi-scale geophysical datasets is an effective way to improve accuracy and resolution of seismic inversion.Considering different confidence of data in different scale,we propose a sparse Bayesian joint inversion method of multi-scale seismic data.Firstly we obtain scale-constrained operator for different scale data through matching analysis for well-side synthetic seismogram with surface seismic data.Based on the Bayesian inversion framework,assuming that model parameters obey Cauchy prior distribution to retreive the sparse results,we derive a cost function for sparse Bayesian joint inversion for multi-scale seismic datasets with scale-constrained operator,and employ Polak-Ribiere-Polyak (PRP) conjugate gradient algorithm to solve the optimization problem.Model and real data tests show that the proposed inversion can highlight reservoir information from surface seismic data and borehole seismic data in high-confidence scale,and provide the high-accuracy inversion results for reservoir-oriented integrated investigation.