Research On Denoising Method Based On Hankel Matrix

Seismic data denoising is an important content in seismic data processing, and the improvement of signal to noise ratio is the most important task for seismic data processing, and need to choose different method while processing different types of noises. The main methods to attenuate the coherent and random noise contain the prediction method in time domain and the eigenvalue method. Because of the rank of Hankel matrix is corresponded with the number of the events with different dips, and this advantage helps to solve the weakness of current methods using eigenvalue, so I chose this method into study.Decompose the Hankel matrix and get the singular value of it, the big singular values present data with good coherence and the small singular values present data with bad coherence. We can attenuate the noise and enhance the signal to noise ratio by choosing different rank reduction parameters according to the features of the data. In this article I draw the application workflow of denoising using Hankel matrix, and put forward four methods to get the rank reduction parameters, and choose one of them while processing different data. And after get the parameters, there also three filters according to the purpose of the processing similar with the filters in f-x domain. Choose the big singular values to reconstruct the matrix can attenuate the random noises, choose the singular values without big values in low frequency can attenuate coherent noises like ghost wave, and treat the missing traces as noise, the rank reduction step can recover some of the amplitudes expected on these missing traces.The model test and the result of real data processing all proved the denoising method based on Hankel matrix is effective and practical. In general, by constructing the Hankel matrix and reduce the rank to construct the new matrix can attenuate the random noise with high fidelity, attenuate most ghost wave without losing effective signal in low frequency, and can recover the missing traces successfully.