Random noise of seismic exploration in desert modeling and its applying in noise attenuation

Random noise is one of key factors which influence valid signals of seismic exploration, and it weakens the signal to noise ratio (SNR) of seismic records seriously. It is the first request to study noise characteristics and generation while filtering methods are improved continuously. At this stage noise characteristics research emphasizes searching its regularity, but much less qualitative and quantitative analysis. In this paper, seismic random noise is classified into natural noise and cultural noise according to the reason that noise is generated. Considering convenience to the calculation, it is assumed that different kinds of noise sources are distributed around the geophones as point sources in their areas appointed. The noise source functions are decided according to the corresponding theories, experience and references. It is assumed that their excitation waves propagate by wave equation and seismic random noise is considered as a superposed wave-field excited by all of the source. The theoretical model of random noise is obtained by solving the non-homogeneous wave equations with different source functions and superposing all of the solutions. The characteristics of different kinds of noise in the seismic records can be shown through modeling random noise. A simulated noise record is compared with a section of field seismic data in the desert in Tarim, including spectral feature, kurtosis, skewness, root-mean-square, etc. From the comparative results, it can be seen the simulated noise is similar with the real noise, which shows the feasibility of the modeling method. The appropriate filtering method is chosen for noise attenuation according to analyzing all kinds of noise characteristics in seismic records. In the desert, near-field cultural noise is the key component of random noise, based on which, complex diffusion filtering is selected, and the filtered results including the artificial synthetic and actual seismic records are compared with the results of time frequency peak filtering which is a popular method in seismic random noise attenuation in recent years. The comparative results show that complex diffusion filtering is more suitable for the random noise in Tarim, which demonstrates that seismic random noise modeling can provide theoretical guidance to choosing appropriate filtering method.