Removing abnormal environmental noise in nodal land seismic data using deep learning

• DOI: 10.1190/geo2023-0143.1
• Publication Date: 2023-08-31T13:19:30Z
• Authors: Xiaobin Li; Qiaomu Qi; Yuyong Yang; Pengfei Duan; Zhonglin Cao
• Keywords: Environmental Noise; Seismic Noise

In field seismic data, abnormal environmental noise from the acquisition environment often is unavoidable, and its characteristics are complex. Abnormal has a strong amplitude (tens of thousands times average amplitudes) masks effective signal. The traditional method for removing such relies on energy scanning to identify location. However, cannot always accurately due sudden change at locations first-breaks surface waves. A deep learning been developed. presence with large also results in an uneven distribution. existing denoising networks rarely consider this situation. Based received by nodal system, workflow automatically generating training data sets established. network built based Unet incorporation residual block. Batchnorm layer specifically omitted adapt Synthetic testing have confirmed effectiveness applicability proposed method. new indicates better performance greater flexibility than Comparison between different that Residual Block Connected much more contain noise.