Radon变换频变自适应多次波压制方法

Frequency-dependent multiple attenuation method with Radon transform

Radon变换是常用多次波压制方法之一,然而有限、离散的采样空间导致一次波和多次波在Radon域混叠,并且混叠程度随频率降低愈加严重.高分辨率Radon变换可以改善一次波和多次波分离效果,但分辨率的提高会降低保幅性,导致多次波残留或一次波损伤.为此,本文在最小二乘反演基础上,提出一种随混叠程度自适应分离多次波的频变滤波方法.根据Radon变换褶积模型,分析一次波和多次波频变混叠机制;建立随频率调整的修正柯西函数混叠模型;依据一次波和多次波的混叠比例设计多次波滤波器,该滤波器能够根据一次波和多次波的混叠程度自适应调整滤波函数,提高多次波估计精度.合成数据和实际数据的多次波压制实验表明,这种频变滤波方法既可以改善多次波压制效果,又可避免高分辨率反演方法计算量大的问题.

The Radon transform is one of the commonly used methods for multiple attenuation.However,limited and discrete sampling causes the overlap of primaries and multiples in the Radon domain.The lower the frequency,the more severe the overlap.The high-resolution Radon transform can improve the separation of primaries and multiples,but the increased resolution will reduce the amplitude preservation performance,resulting in residual multiples or damage to primaries.Therefore,in this paper,a frequency-dependent filtering method for adaptive multiple separation based on least-squares inversion is proposed.Firstly,the frequency-dependent overlap mechanism of primaries and multiples is analyzed according to the Radon transform convolution model.Then,an overlap model adjusted with frequency is established using a modified Cauchy function.Finally,a filter is designed to extract multiples based on the overlap ratio of primaries and multiples.The filter can be adaptively adjusted according to the overlap degree of primaries and multiples,improving the accuracy of multiple estimation.Multiple attenuation experiments on synthetic and field data show that this frequency-dependent filtering method can improve the performance of multiple attenuation and avoid the high-computational complexity problem in high-resolution inversion methods.