叠前正交分解去噪方法

Prestack noise suppression usir.g orthogonal decomposition,

本文介绍一种用于动校正之后多个道集上的叠前去噪方法。若道集内的各道按偏移距排列,则对任一时间T,取同一偏移距区间x_0上的连续N个CMP点的样点,组成N维空间上的一个向量V_(x_0)(N)。每一个偏移距区间x都对应一个向量V_x(N)。这些由不同偏移求得的向量就构成了N维空间上的向量集。应用特征分解方法可求出属于这个向量集的主分量u_1,然后将每个向量V_x(N)投影到主分量u_1上,即可得到较好的去噪效果。

Presented here is a prestack noise suppression method which is applied to multiple NMO-corrected trace gathers.If traces in a gather are arranged in offset order, a vector V_(x_0)(N) in N-dimensional space is formed by sampling at time T the common-offset(x_0) traces which are taken from successive N CMP gathers.Each common-offset (x) trace gather brings a vector V_x(N).The vectors obtained from different common-offset trace gathers form a vector set in N-dimensional space.The principal component u_1 of the vector set can be obtained by making eigenvalue decomposition.Then, projecting each vector V_x(N) on the principal component u_1 results in good noise suppression.