基于时间和空间方向差分稀疏约束的叠前AVO反演

Prestack AVO inversion with space and time directional difference sparse constraints

在叠前AVO反演中,常规的吉洪诺夫正则化方法不能突出异常体在纵向上的边界,并且忽略了地层的横向联系,导致反演结果产生边缘模糊现象。为了解决这一问题,这里基于Zeoppritz方程的Fatti近似公式和贝叶斯理论框架,以模型参数柯西分布作为先验约束,通过引入时间和空间方向的差分稀疏约束保护地层边界,并最终采用迭代重加权最小二乘算法实现纵、横波阻抗的同时多道反演。模型反演结果具有明显的地层边界和分层特征,表明该方法能够解决边缘模糊问题、保护不连续体的横向边界,大大提高了油气预测的准确度。

In seismic prestack AVO inversion,the conventional Tikhonov regularization method can not highlight the vertical edge of geologic anomalies and usually ignores the lateral connection of strata,which will cause fuzzy edges in inversion results.To address this issue,based on the Fatti approximate formula of Zeoppritz equation and Bayesian framework,this paper introduces the space and time directional difference sparse constraints expressed by the Cauchy norm to protect edges,and realizes the simultaneous multitrace inversion of P-impedance and S-impedance with iteratively reweighted least squares algorithm.The inversion results for model have obvious stratigraphic edge and layered characteristics,and it indicates that the new method can solve the edge fuzzy problem,protect the lateral edge of discontinuities,and largely improve the accuracy of predicting reservoirs.