| 基于f-x域流式预测滤波器的地震随机噪声衰减方法 |
| |
| 引用本文： | 国胧予,刘财,刘洋,郑植升,王青晗.基于f-x域流式预测滤波器的地震随机噪声衰减方法[J].地球物理学报,2020,63(1):329-338. |
| |
| 作者姓名： | 国胧予 刘财 刘洋 郑植升 王青晗 |
| |
| 作者单位： | 吉林大学地球探测科学与技术学院, 长春 130026 |
| |
| 基金项目： | 国家重点研发计划课题（2018YFC0603700），国家自然科学基金项目（41774127，41430322，41522404）以及吉林大学高层次科技创新团队建设项目（2017TD-14）资助. |
| |
| 摘 要： | 随机噪声的影响在地震勘探中是不可避免的,常规的随机噪声压制方法在处理中往往会破坏具有时空变化特征的非平稳有效地震信号,影响地震数据的准确成像.当前油气勘探的目标已经转变为“两宽一高”,随着数据量的增大,对去噪方法的处理效率也提出了更高的要求.因此,开发高效的非平稳地震数据随机噪声压制方法具有重要意义.预测滤波技术广泛用于地震随机噪声的衰减,本文基于流式处理框架提出一种新的f-x域流式预测滤波方法,通过在频率域建立预测自回归方程,运用直接复数矩阵逆运算代替迭代算法求解非平稳滤波器系数,实现时空变地震同相轴预测,提高自适应预测滤波的计算效率.通过与工业标准的FXDECON方法和f-x域正则化非平稳自回归(RNA)方法进行对比,理论模型和实际数据的测试结果表明,提出的f-x域流式预测滤波方法能更好地平衡时空变有效信号保护、随机噪声压制和高效计算三者之间的关系,获得合理的处理效果.
|
| 关 键 词： | 频率-空间域 流式预测滤波 随机噪声衰减 非平稳地震信号 |
| 收稿时间： | 2019-01-18 |
Seismic random noise attenuation based on streaming prediction filter in the f-x domain |
| |
| GUO LongYu,LIU Cai,LIU Yang,ZHENG ZhiSheng,WANG QingHan.Seismic random noise attenuation based on streaming prediction filter in the f-x domain[J].Chinese Journal of Geophysics,2020,63(1):329-338. |
| |
| Authors: | GUO LongYu LIU Cai LIU Yang ZHENG ZhiSheng WANG QingHan |
| |
| Institution: | College of Geo-exploration Science and Technology, Jilin University, Changchun 130026, China |
| |
| Abstract: | Random noise is unavoidable in seismic prospecting. Conventional random noise suppression methods will damage nonstationary seismic signals with characteristics of time-space variation and finally affect accurate imaging of seismic data. Meanwhile, the current target of oil and gas exploration becomes "two widths and one high", with the increase of data scale, higher requirements are proposed on the processing efficiency of denoising methods. Therefore, there is a great significance to develop an efficient random noise suppression method for nonstationary seismic data. Prediction filters are widely used for seismic random noise attenuation. In this paper, we propose a new f-x streaming prediction filter based on the streaming framework. By establishing a prediction autoregressive equation in the frequency domain, the direct complex matrix inverse operation is used instead of the iterative method to calculate the nonstationary filter coefficients, and the time-and space-varying seismic events are predicted, computational efficiency of adaptive prediction filtering is significantly improved. By comparing with standard FXDECON method in industry and f-x regularized nonstationary autoregression (RNA) method, the results of synthetic and field data show that the f-x streaming prediction filter is able to better balance the relationship between effective signal protection, random noise suppression, and high efficiency calculation, the proposed method can get reasonable results. |
| |
| Keywords: | f-x domain Streaming prediction filtering Random noise attenuation Nonstationary seismic signal |
| 本文献已被 CNKI 维普 等数据库收录！ |
| 点击此处可从《地球物理学报》浏览原始摘要信息 |
| 点击此处可从《地球物理学报》下载免费的PDF全文 |
|
