
浏览全部资源
扫码关注微信
1.中国科学院精密测量科学与技术创新研究院精密大地测量与定位全国重点实验室,湖北 武汉,430077
2.中国科学技术大学地球和空间科学学院,安徽 合肥,230026
Received:23 July 2026,
Published:05 November 2026
移动端阅览
陈冠芝,张成枫,张涵,等.DCMR-Net:一种面向复杂噪声地震记录的多尺度残差去噪网络[J].武汉大学学报(信息科学版),2026,51(11):2314-2328.
CHEN Guanzhi,ZHANG Chengfeng,ZHANG Han,et al.DCMR-Net: A Multi-scale Residual Denoising Network for Seismic Records with Complex Noise[J].Geomatics and Information Science of Wuhan University,2026,51(11):2314-2328.
陈冠芝,张成枫,张涵,等.DCMR-Net:一种面向复杂噪声地震记录的多尺度残差去噪网络[J].武汉大学学报(信息科学版),2026,51(11):2314-2328. DOI:10.13203/j.whugis20260325
CHEN Guanzhi,ZHANG Chengfeng,ZHANG Han,et al.DCMR-Net: A Multi-scale Residual Denoising Network for Seismic Records with Complex Noise[J].Geomatics and Information Science of Wuhan University,2026,51(11):2314-2328. DOI:10.13203/j.whugis20260325
随着深度学习在信号去噪领域的快速发展,基于卷积神经网络的数据驱动方法为复杂噪声背景下的地震信号恢复提供了新的技术途径。去噪卷积神经网络(denoising convolutional neural network,DnCNN)最初面向自然图像去噪任务设计,直接迁移至地震信号去噪时,存在感受野受限、特征尺度单一以及有效震相与噪声区分能力不足等问题。针对上述问题,提出扩张卷积多尺度残差网络(dilated convolution multi-scale residual network,DCMR-Net)。该模型利用不同扩张率的空洞卷积扩大网络感受野,通过多尺度特征融合提取局部震相细节和整体波形结构,并引入注意力机制,增强有效震相特征,抑制无关噪声响应。采用频率-波数域方法构建三分量FK合成地震波信号,分别叠加真实背景噪声、Gaussian噪声、Pink噪声和Combo混合噪声,并选取斯坦福地震数据集(Stanford earthquake dataset,STEAD)中的低信噪比实测地震记录进行模型评估。结果表明,DCMR-Net对4类噪声均具有较强的适应性,能够有效压制不同类型的噪声,同时较好保留原始合成有效信号的振幅和主要震相;在STEAD实测地震记录中,该模型能够有效提高信噪比,减弱P波到达前的背景噪声,并较好保留P波初至及主要震相特征。研究表明,DCMR-Net相较于DnCNN模型在合成数据和真实地震记录中都具有更好的去噪效果和应用潜力。
Objectives
2
With the rapid development of deep learning in signal denoising
data-driven methods based on convolutional neural networks have provided a new approach for seismic signal recovery under complex noise conditions. The denoising convolutional neural network (DnCNN)
originally designed for natural image denoising
has several limitations when directly applied to seismic signal denoising
including a limited receptive field
single-scale feature representation
and insufficient ability to distinguish effective seismic phases from noise. To address these limitations
this paper proposes a dilated convolution multi-scale residual network (DCMR-Net) for seismic signal denoising.
Methods
2
DCMR-Net employs dilated convolutions with different dilation rates to enlarge the receptive field
integrates multi-scale features to capture both local seismic-phase details and overall waveform structures
and introduces an attention mechanism to enhance effective seismic-phase features while suppressing irrelevant noise responses. Three-component synthetic seismic signals are generated using the frequency–wavenumber (FK) method and contaminated with four types of noise
including real background noise
Gaussian noise
Pink noise
and Combo mixed noise. In addition
low-signal-to-noise-ratio real seismic records from the Stanford earthquake dataset (STEAD) are selected to evaluate the denoising performance of the proposed model.
Results
2
The results show that DCMR-Net exhibits strong adaptability to all four types of noise and can effectively suppress different noise components while preserving the amplitudes of effective signals and the major seismic phases. For real seismic records from STEAD
DCMR-Net effectively improves the signal-to-noise ratio
reduces background noise before the P-wave arrival
and preserves the P-wave onset and major seismic-phase characteristics.
Conclusions
2
DCMR-Net demonstrates good denoising performance and stability for both synthetic seismic data and real seismic records. By combining dilated convolution
multi-scale feature fusion
and an attention mechanism
the proposed network improves the distinction between effective seismic phases and complex noise
showing promising potential for practical seismic signal denoising applications.
蔡振宇 , 盖增喜 . 人工智能在拾取地震P波初至中的应用: 以汶川地震余震序列为例 [J]. 北京大学学报(自然科学版) , 2019 , 55 ( 3 ): 451 - 460 .
CAI Zhenyu , GAI Zengxi . Using Artificial Intelligence to Pick P-Wave First-Arrival of the Microseisms: Taking the Aftershock Sequence of Wenchuan Earthquake as an Example [J]. Acta Scientiarum Naturalium Universitatis Pekinensis , 2019 , 55 ( 3 ): 451 - 460 .
STOCKWELL R G , MANSINHA L , LOWE R P . Localization of the Complex Spectrum: The S Transform [J]. IEEE Transactions on Signal Processing , 1996 , 44 ( 4 ): 998 - 1001 .
DONOHO D L , JOHNSTONE I M . Ideal Spatial Adaptation by Wavelet Shrinkage [J]. Biometrika , 1994 , 81 ( 3 ): 425 - 455 .
郑成龙 , 王宝善 . S变换在地震资料处理中的应用及展望 [J]. 地球物理学进展 , 2015 , 30 ( 4 ): 1580 - 1591 .
ZHENG Chenglong , WANG Baoshan . Applications of Stransform in Seismic Data Processing [J]. Progress in Geophysics , 2015 , 30 ( 4 ): 1580 - 1591 .
王海斌 , 周云耀 , 吕永清 , 等 . 小波阈值法在JCZ系列超宽频带地震仪记录的地震数据中的应用及探讨 [J]. 大地测量与地球动力学 , 2021 , 41 ( 3 ): 323 - 330 .
WANG Haibin , ZHOU Yunyao , LÜ Yongqing , et al . Application and Discussion on Wavelet Threshold Method inSeismic Data Recorded by JCZ Series Ultra-Wideband Seismographs [J]. Journal of Geodesy and Geodynamics , 2021 , 41 ( 3 ): 323 - 330 .
CANALES L L . Random Noise Reduction [C]// 1984 SEG Annual Meeting , Atlanta, USA , 1984 .
NEELAMANI R , BAUMSTEIN A I , GILLARD D G , et al . Coherent and Random Noise Attenuation Using the Curvelet Transform [J]. The Leading Edge , 2008 , 27 ( 2 ): 240 - 248 .
张恒磊 , 李名勇 , 刘天佑 . Contourlet域相关和阈值衰减联合去噪方法及其在地震资料处理中的应用 [J]. 武汉大学学报(信息科学版) , 2011 , 36 ( 9 ): 1047 - 1050 .
ZHANG Henglei , LI Mingyong , LIU Tianyou . Joint Denoising of Correlation and Thresholding in Contourlet Domain and Its Application to Seismic Data [J]. Geomatics and Information Science of Wuhan University , 2011 , 36 ( 9 ): 1047 - 1050 .
BEKARA M , VAN DER BAAN M . Random and Coherent Noise Attenuation by Empirical Mode Decomposition [J]. Geophysics , 2009 , 74 ( 5 ): V 89-V9 8.
OROPEZA V , SACCHI M . Simultaneous Seismic Data Denoising and Reconstruction via Multichannel Singular Spectrum Analysis [J]. Geophysics , 2011 , 76 ( 3 ): V 25-V32.
MOUSAVI S M , BEROZA G C . Deep-Learning Seismology [J]. Science , 2022 , 377 ( 6607 ): eabm4470 .
廖诗荣 , 张红才 , 范莉苹 , 等 . 实时智能地震处理系统研发及其在2021年云南漾濞MS6.4地震中的应用 [J]. 地球物理学报 , 2021 , 64 ( 10 ): 3632 - 3645 .
LIAO Shirong , ZHANG Hongcai , FAN Liping , et al . Development of a Real-Time Intelligent Seismic Processing System and Its Application in the 2021 Yunnan Yangbi MS6.4 Earthquake [J]. Chinese Journal of Geophysics , 2021 , 64 ( 10 ): 3632 - 3645 .
李帅 , 蒋策 , 田平 , 等 . 深度学习在地震数据处理中的应用 [J]. 物探化探计算技术 , 2025 , 47 ( 5 ): 623 - 640 .
LI Shuai , JIANG Ce , TIAN Ping , et al . Application of Deep Learning in Seismic Data Processing [J]. Computing Techniques for Geophysical and Geochemical Exploration , 2025 , 47 ( 5 ): 623 - 640 .
ZHANG K , ZUO W M , CHEN Y J , et al . Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising [J]. IEEE Transactions on Image Processing , 2017 , 26 ( 7 ): 3142 - 3155 .
陶立清 , 黄国满 , 杨书成 , 等 . 一种利用卷积神经网络的干涉图去噪方法 [J]. 武汉大学学报(信息科学版) , 2023 , 48 ( 4 ): 559 - 567 .
TAO Liqing , HUANG Guoman , YANG Shu-cheng , et al . A Interferogram Denoising Method Based on Convolutional Neural Network [J]. Geomatics and Information Science of Wuhan University , 2023 , 48 ( 4 ): 559 - 567 .
WANG H , ZHANG J . A Deep Learning Framework for Suppressing Prestack Seismic Random Noise Without Noise-Free Labels [J]. Energy Geoscience , 2024 , 5 ( 3 ): 100297 .
WANG F , CHEN S C . Residual Learning of Deep Convolutional Neural Network for Seismic Random Noise Attenuation [J]. IEEE Geoscience and Remote Sensing Letters , 2019 , 16 ( 8 ): 1314 - 1318 .
ZHU W Q , MOUSAVI S M , BEROZA G C . Seismic Signal Denoising and Decomposition Using Deep Neural Networks [J]. IEEE Transactions on Geoscience and Remote Sensing , 2019 , 57 ( 11 ): 9476 - 9488 .
YANG L Q , CHEN W , WANG H , et al . Deep Learning Seismic Random Noise Attenuation via Improved Residual Convolutional Neural Network [J]. IEEE Transactions on Geoscience and Remote Sensing , 2021 , 59 ( 9 ): 7968 - 7981 .
DUAN L , CAI J X , WANG L , et al . Research on Seismic Signal Denoising Model Based on DnCNN Network [J]. Applied Sciences , 2025 , 15 ( 4 ): 2083 .
WANG F Y , YANG Y , MA J W . Seismic Denoising by Deep Learning from Natural Repeating Earthquakes [J]. Journal of Geophysical Research: Machine Learning and Computation , 2025 , 2 ( 4 ): e2025JH000646 .
CHAI C P , ROSE D , STEWART S , et al . Denoising Seismograms in the Time Domain Using a Deep Learning Model [J]. Bulletin of the Seismological Society of America , 2026 , 116 ( 1 ): 192 - 204 .
DAHMEN N , CLINTON J , MEIER M A , et al . Toward Operational Earthquake Seismogram Denoising [J]. Bulletin of the Seismological Society of America , 2026 , 116 ( 3 ): 923 - 945 .
雷鸣 , 周云耀 , 向涯 , 等 . 结合短时傅里叶变换与注意力的震相识别模型 [J]. 地球科学 , 2026 , 51 ( 1 ): 104 - 115 .
LEI Ming , ZHOU Yunyao , XIANG Ya , et al . A Phase Picking Model Integrating Short-Time Fourier Transform and Multi-Scale Attention [J]. Earth Science , 2026 , 51 ( 1 ): 104 - 115 .
ZHONG T , CHENG M , DONG X T , et al . Seismic Random Noise Attenuation by Applying Multiscale Denoising Convolutional Neural Network [J]. IEEE Transactions on Geoscience and Remote Sensing , 2022 , 60 : 5905013 .
DONG X T , LIN J , LU S P , et al . Multiscale Spatial Attention Network for Seismic Data Denoising [J]. IEEE Transactions on Geoscience and Remote Sensing , 2022 , 60 : 5915817 .
HU H X , WEI Y H , CHEN H , et al . Seismic Data Denoising Based on Attention Dual Dilated CNN [J]. Scientific Reports , 2025 , 15 : 28123 .
YAO Z J , HAO L , QIN L , et al . Efficient Seismic Data Denoising via Multi-Scale Attention Network with Depthwise Separable and Residual Dilated Convolutions [J]. Scientific Reports , 2026 , 16 : 2818 .
李艳艳 , 殷海涛 , 韩林桥 . 利用MSMPCA去噪的CEEMD方法监测高频GNSS同震形变 [J]. 武汉大学学报(信息科学版) , 2022 , 47 ( 3 ): 352 - 360 .
LI Yanyan , YIN Haitao , HAN Linqiao . A Denoising Method of MSMPCA Based on CEEMD for Coseismic Deformation Monitoring with High-Frequency GNSS [J]. Geomatics and Information Science of Wuhan University , 2022 , 47 ( 3 ): 352 - 360 .
ZHU L P , RIVERA L A . A Note on the Dynamic and Static Displacements from a Point Source in Multilayered Media [J]. Geophysical Journal International , 2002 , 148 ( 3 ): 619 - 627 .
MOUSAVI S M , SHENG Y X , ZHU W Q , et al . STanford EArthquake Dataset (STEAD): A Global Data Set of Seismic Signals for AI [J]. IEEE Access , 2019 , 7 : 179464 - 179476 .
MENG F L , LI Y , LIU Y P , et al . Application of the Radon-FCL Approach to Seismic Random Noise Suppression and Signal Preservation [J]. Journal of Geophysics and Engineering , 2016 , 13 ( 4 ): 549 - 558 .
VOSS R F , CLARKE J . ‘1/ f noise’ in Music and Speech [J]. Nature , 1975 , 258 ( 5533 ): 317 - 318 .
0
Views
0
下载量
0
CSCD
Publicity Resources
Related Articles
Related Author
Related Institution
京公网安备11010802024621