基于三维各向异性拉普拉斯滤波的随机噪声压制方法及应用

3D anisotropic Laplacian filtering based random seismic noise suppression

准噶尔盆地叠后地震资料随机噪声压制主要采用三维空间预测滤波(FXY)方法。由于假设同相轴在短距离内是线性的,因此该方法在提高强能量信号信噪比的同时,也会损伤相对弱小信号,模糊断层和裂缝等线性相关性较差的地质体的信号特征。为此,基于GeoEast系统研发了利用三维各向异性拉普拉斯滤波的随机噪声压制方法。模型试验和在准噶尔盆地不同地区的实际资料应用结果均表明,本文方法对随机噪声的压制效果优于FXY方法,能明显提高地震资料信噪比、较好地保护有效信号、清晰展现地质体边缘特征细节及提高断层成像精度,为后续储层反演和精细地震资料解释夯实了基础。

The 3 Drandom noise attenuation(3 D-RNA)is a spatial prediction filtering,which is widely applied to poststack seismic data in the Junggar Basin.With the linearity assumption of events in a short distance,this method improves strong energy signal-to-noise ratio(SNR),but it damages weak energy signals and blurs signals which are relative to faults and cracks.Therefore,a denoising method based on 3 Danisotropic Laplacian filtering is proposed to solve these problems.Tests on model and real seismic data show that the proposed method is better than 3 D-RNA in random noise suppression.Not only seismic data SNR is improved,but also weak signals are preserved.Meanwhile,geologicalbody edge characteristics become clearer and better fault imaging is achieved.The proposed method has been successfully applied in different areas in the Junggar Basin.This research provides a solid foundation for further reservoir inversion and seismic interpretation.