The Research Of Random Noise Attenuation And Deconvolution In Time-frequency Domain

With the deepening development of oil and gas,seismic exploration techniques have become more sophisticated,such as: seismic profiles must be processed more accurately,so that we can characterize stratal structure,thin features,changes in lithology and fluid properties more finely,etc.This requires ensuring the signal to noise ratio of seismic data,and under the premise,as far as possible to improve the resolution of seismic data.This article will both study random noise and deconvolution in time frequency domain.The key to improve the signal to noise ratio of seismic data is noise suppression.Random noise in seismic record has a wide frequency without a certain speed,which is difficult to suppress using the differences in the spectrum or the direction of its propagation between random noise and significant wave.One focus of this paper therefore is to use time frequency thinking of curvelet transform and nonsubsampled contourlet transform to suppress random noise,improve signal to noise ratio of seismic data,At the same time,the advantages and disadvantages of the two methods are compared and analyzed to suppress the random noise of seismic data.Deconvolution is a necessary means to improve the resolution of seismic data,and conventional deconvolution is mostly based on stationary convolution model,which is assumed that seismic waves are time-invariant.But in fact due to the absorption and attenuation effect of subsurface media,seismic wavelets constantly change with time,which is a typical time-varying signal.Whether the accurate extraction of the time variant wavelet is the key to the success or failure of the time varying deconvolution.In this paper,we will focus on the time-frequency domain deconvolution method based on the S transform and modified S transform,in the meantime,giving analysis and comparison of the advantages and disadvantages of the two methods.Most of the traditional methods to improve signal to noise ratio and the resolution is in the frequency domain,without considering the frequency relationship with the transformation of the time.Both the suppression of random noise and the time varying deconvolution will be discussed in this paper,achieving a combination of the two methods for stacked seismic profiles,effectively improving the signal to noise ratio and resolution.