Curvelet域垂直地震剖面波场分离

针对垂直地震剖面（VSP）资料处理中上下行波重叠难以分离的问题，提出了一种基于奇异值分解（SVD）和Curvelet变换相结合的分离上下波方法．首先基于数据中随机噪声是统计平稳的假设，在Curvelet域通过分尺度选取阈值压制随机噪声，再利用奇异值分解方法滤除部分下行波能量，最后根据VSP资料中上下行波在Curvelet域分布区域不同，进行上下行波场分离．合成和实际资料处理结果表明，该方法与f-k滤波方法和SVD方法相比，不仅提高了信噪比，而且减弱了边界上的波形畸变，得到了更好的分离效果．

Wave Field Separation of Vertical Seismic Profiling in Curvelet Domain

To separate the up-going and down-going waves in data processing of vertical seismic profiling,a method of wave field separation is proposed based on a combination of singular value decomposition(SVD) and Curvelet transform.Firstly,based on the assumption that random noise in data is stationary in statistics,the threshold on scale of Curvelet domain is selected for attenuation of random noise.And then,SVD method is used to filter part of down-going wave energy.Finally,the wave field separation is done because up-going wave and down-going wave will be located in different region of Curvelet domain.The results of synthesized and real data show that the method can improve signal-to-noise ratio,and can weaken the malformation of waveform on boundary.The method can gain better wave field separation result than f-k filter method and SVD method.