模型驱动的高精度叠前地震反演方法及应用

Model driven high-precision prestack seismic inversion

线性化叠前地震反演方法由于具有高效率、低成本的特点,在叠前地震油气储层预测中得到广泛应用。为进一步提高线性化叠前地震反演方法的精度、稳定性及资料适应性,提出了模型驱动的高精度叠前地震反演方法。根据地震残差分布特点,采用最小二乘法构建目标函数,同时增加了L2范数正则化项确保反演解适定性,基于泰勒公式展开推导构建了反演迭代公式。反演计算中采用的雅克比矩阵是直接通过佐普里兹方程矩阵形式推导构建的,避免了小角度和弱反射界面假设,确保了雅克比矩阵的构建精度,从而提高了反演精度。在正则化参数选取方面,利用地震噪声和反演参数扰动量的高斯统计方差自动选取正则化参数。模型实验数据表明,模型驱动的高精度叠前地震反演方法在实现正则化参数自动化选取的同时确保了反演精度。胜利油田东部地区实际资料应用结果表明,该方法稳定性强,对复杂地震资料具有较强的适应能力,反演结果精度高,能够较好利用部分角度叠加数据体中的叠前地震振幅变化信息,提高了储层岩性和流体预测能力。

Linear prestack seismic inversion has been widely used in prestack seismic reservoir prediction because of its high efficiency and low cost.A model-driven,high-precision,prestack seismic inversion method was proposed to further improve the accuracy,stability,and adaptability for application to complicated seismic conditions.Based on the characteristics of seismic residuals,the least-square method was used to construct the inversion objective function,with the L2 norm regularization term ensuring the inversion results were well posed.The inversion iteration formula was finally deduced using Taylor's formula,based on the idea of linearization inversion.In this method,the Jacobian matrix was directly constructed by the matrix form of the Zeoppritz without the assumptions of a small incidence angle and weak reflection interface.This was done to ensure the accuracy of construction of the Jacobian matrix and improve the precision of inversion results.The regularization parameters were automatically selected using the Gaussian statistical variance of seismic noise and inversion parameter disturbances.Tests on model and field data from SL oilfield demonstrated that the method has good stability and high adaptability when applied to complicated seismic data,and it can improve the prediction of reservoir lithology and fluid by making full use of seismic partial angle stacked data.