一种基于曲波变换的自适应地震随机噪声消除方法

基于稀疏反演的随机噪声消除方法需要估计一个与噪声能量相匹配的阈值才能获得可靠的去噪结果。由于不同数据的噪声能量不同,因此通常采用人工调节的方法获得合理的阈值估计,这会耗费大量的计算资源和人力成本。为此提出一种自适应的随机噪声消除方法,以曲波变换为稀疏变换,通过迭代过程中解的稀疏性与拟合误差之间的内在关系确定合适的阈值,并且自动终止迭代,因而不依赖于对噪声能量的估计就能实现对噪声的消除。利用理论模型数据及两个地区实际地震数据验证了方法的有效性。

The conventional sparse inversion-based random noise elimination utilizes a thresholding operation to conduct denoising,on the basis that seismic signals are sparsely expressed in a transform domain.This would produce effective denoising when threshold values match noise energy.However,owing to variety in the noise energy of different data,the reasonable threshold is usually obtained by manual adjustment,which is time- and labor-consuming.This paper proposes an adaptive random noise elimination method that does not rely on noise energy estimation.The method uses a Curvelet transform as the sparse transform and chooses a proper threshold value through the inner relationship between solution sparsity and fitting error,thus terminating the iteration automatically.Testing on both synthetic data and field data demonstrate that the proposed method can eliminate random noise while preserving effective signal.

国家自然科学基金(41674114)、河北省自然科学基金(D2017403027)、河北省高等学校百名优秀创新人才支持计划(Ⅲ)(SLRC2017024)、河北省科技计划(15456011)共同资助。