The Application Of Improved EMD In Seismic Random Noise Suppression

Seismic prospecting is an important geophysical method for mineral resources, oil,and gas. Lots of noises exist in seismic data as there are so many influences caused by the behaviors of people and environment. These noises and useful signal mixed together,which makes that the data cannot be used to be further processed directly. So an effective noise reduction method is necessary to improve the quality of seismic signal.This paper mainly researches on seismic random noise reduction. Many effective methods have been used to reduce seismic random noise so far. EMD has significant advantages in processing non-stationary signal. However, the EMD method also has its own fatal shortage. In this paper, aiming at the mode aliasing problems caused by EMD,two improved algorithms: Ensemble EMD and Complementary Ensemble EMD were researched. By comparing the results of numerical simulation, we see that these two improved algorithms can remit the mode aliasing problems effectively. In addition,CEEMD can keep the completeness of EMD better and has higher computational efficiency.Firstly, we research one dimensional noise reduction method based on CEEMD.The decomposition property of CEEMD is used to decompose the original signal into finite IMFs from high frequency to low frequency. Then we calculate the cross correlation coefficient of each IMF and the original signal to estimate which IMFs should be filtered. Lastly, we add the remaining components and the components after filtering by improved threshold function method to achieve the purpose of noise reduction. Compared to traditional wavelet soft or hard threshold denoising method, our improved algorithm can protect the amplitude of effective signal better while suppressing random noise effectively by adjusting parameters. Besides, our improved algorithm can retain edges and details of signal better. The experiment results of simulated seismic signal and real seismic signal verify the superiority of the proposed method.The traditional NLM algorithm need to presuppose parameters according to experience. So it ignores the randomness of the noise intensity in real seismic signal,which cannot adjust filtering parameter adaptively. We propose an local noise of standard variance-assisted adaptive NLM filtering algorithm. In this algorithm, the firsttwo dimensional noise component after CEEMD is used to estimate the standard deviation of the noise in search window approximately. According to the estimated standard deviation, we adjust filtering parameter adaptively to achieve the target that signals in different regions can be filtered in different degree and improve the denoising performance. Simulation results show that the improved algorithm can suppress random noise of seismic data and recover reflection events more effectively. Besides, the proposed method is more flexible than the traditional NLM algorithm and more suitable to process seismic signal with nonuniform noise intensity.