Statistical features of the random noise in land seismic prospecting

In denoising of seismic prospecting, random noise is often assumed to be stationary and Gaussian. However, this is not always appropriate for real seismic data. Here, we use the modern statistical testing methods to investigate the properties of land-seismic-prospecting random noise in the stationarity, Gaussianity and linearity. The results show that the random noise is not strictly stationary as suggested traditionally. The time length and acquisition environment can pose effects on the stationarity of the random noise. Such stationarity becomes poorer with increasing length of the noise and complexity of the acquisition environment, whereas only the noise of short length can be approximately considered as a stationary process. The results also indicate that the complexity of the acquisition environment has influence on the Gaussianity and linearity of the random noise. The complex environmental conditions always result in well Gaussianity and poor linearity of the random noise. In general, the random noise should be attributed to a linear non-Gaussian stochastic process.