消除全局层析反演累计误差的反向递推方法

常规层析反演方法基于地震偏移成像道集剩余曲率计算深度误差, 忽略了深度误差自上而下的积累, 随着深度的增加, 累计误差也随之增加, 且速度更新后对应叠前深度偏移剖面的层位深度也会改变。从道集曲率深度误差积累的原理出发, 提出一种从下至上逐层消除累计误差的递推方程, 并根据深度残差来更新之前拾取的层位。根据角度域共成像点道集的剩余曲率来计算剩余时差, 利用斯奈尔定律推导出相邻层位入射角与层速度的关系, 计算每一层与上一层入射角的对应关系, 消除该层的累计误差并更新层位深度。模型测试与实际资料应用结果表明, 该方法可行有效, 与常规层析反演方法相比较, 有效改善了速度模型整体的精度并提高了反演效率, 为后续叠前深度偏移成像提供了可靠的速度模型。

The conventional tomographic inversion calculates the depth errors based on residual curvatures of gathers and ignores the accumulation of depth errors from top to bottom.Cumulative error increases with depth, which affects depth positions of horizons on prestack depth migration profiles obtained using the updated velocity field.Based on the principle of curvature depth error accumulation, a recursive equation is formulated to eliminate cumulative errors layer by layer from bottom to top, and horizon picking is updated according to depth residuals.Residual time differences are calculated by using the residual curvatures of common imaging-point gathers in the angle domain, and the relation between incident angle and interval velocity derived from the Snell's law is used to calculate incident angles and eliminate cumulative errors for horizon updating.Model tests and field data application prove the feasibility and effectiveness of the proposed method.Compared with conventional tomographic inversion methods, the new method effectively improves the overall accuracy and efficiency to invert the velocity model for subsequent prestack depth migration.