自适应变分模态分解同态反褶积方法

Homomorphic deconvolution method based on adaptive variational mode decomposition

地震波在地层传播过程中会受到大地低通滤波的影响,分辨率显著降低。反褶积处理可以提高地震资料的分辨率,同态反褶积方法对子波相位没有要求,具有更强的适用性,但在实际应用中,同态域子波与反射系数不易彻底分离。针对这一问题,提出在同态域使用变分模态分解(Variational Mode Decomposition,VMD)方法分离子波与反射系数信息,提高了同态反褶积的抗噪能力。首先,对地震记录进行同态变换,获得同态域地震记录;然后,对同态变换后的结果做VMD,并对分解结果进行选择性重构,消除子波的影响;最后,对重构结果进行同态反变换,提取反射系数序列。针对VMD无法直接确定分解层数的问题,提出基于重构数据与原始数据能量差自动确定分解层数的方法,显著改善了VMD方法的使用效果。理论合成资料实验表明,在时变子波和信噪比较低的情况下,该方法仍可以获得理想的计算结果。对实际地震资料处理并进行地震属性提取,该算法能够有效提高实际资料的横向分辨率,取得较好的应用效果。

Seismic waves are affected by earth low-pass filtering during their propagation in strata,which greatly reduces the resolution.Deconvolution processing can improve the resolution of seismic data.The homomorphic deconvolution method has no requirement for wavelet phase and is more applicable.However,in actual proces-sing,wavelets and reflection coefficients are difficult to completely separate in a homomorphic domain.To solve this problem,this paper proposes using a variational mode decomposition(VMD)method to separate wavelets and re-flection coefficients in a homomorphic domain.The method improves the anti-noise ability of homomorphic deconvolution.Firstly,homomorphic transformation is performed on seismic records to obtain the data of a homomorphic domain.Then VMD is performed on the data,and specific decomposed data are selected for reconstruction to eliminate the influence of wavelets.Finally,the reflection coefficient sequence is recovered by the inverse homomorphic transformation of the reconstruction results.Considering that the VMD method cannot directly determine the decomposition levels,the paper uses a method based on the energy difference between the reconstructed data and the original data to automatically determine the decomposition level,which greatly improves the effect of the VMD method.Synthetic data experiments show that the method can still obtain ideal calculation results in the case of a time-varying wavelet and a low signal-to-noise ratio.Using this algorithm to process actual seismic data and extract seismic attributes can effectively improve the lateral resolution of actual data and achieve a good application effect.