[S-TT38] Seismic monitoring and processing system

[STT38-P05] Seismic data denoising via wavelet coefficient and pixel connectivity thresholding in sychrosqueezed domain

*Zhiyi Zeng¹, Peng Han¹, Da Zhang², Yaqian Shi², Ying Chang², Rui Dai² (1.Southern University of Science and Technology, Shenzhen, China, 2.Institute of Mining Engineering, Beijing General Research Institute of Mining &Metallurgy, Beijing, China)

Keywords:Seismic data denosing, Synchrosqueezed wavelet transform, Wavelet coefficient thresholding, Pixel connectivity thresholding

Random and coherent noises usually exist in seismic records, which make it difficult to utilize the information of signal waveform for imaging and inversion. We develop an effective denoising method based on wavelet coefficient and pixel connectivity threshold to enhance the quality of signal. The proposed method is different from the improved method about thresholding function based on the difference in energy between signal and noise. We take the time-frequency spectrum as an image, so that the image processing method can be introduced to remove the noise in time-frequency domain. The proposed method is mainly divided into two steps. First step is that the conventional wavelet hard-thresholding method is used to remove the dominant noise with low energy. In this step, we apply a fast and simple method, the amplitude variance ratio between two sliding time windows of waveform, to get the pure background noise range for a more accurate wavelet threshold. Compared with the effective signal energies, the remaining noise energies after hard-thresholding has smaller connectivity area. Thus, in the second step, pixel connected component area thresholding is used to erase the residual noise wavelet coefficient as much as possible. We test the performance of the proposed method on synthetic, field microseismic data and natural earthquake data. As the result show that the proposed method can efficiently improve the signal-to-noise ratio (SNR) of signal and clearly provide valuable information (the polarity, arrival time and amplitude) of signal after denoising.