Research On The Deghosting Method For Marine Seismic Data

In marine seismic exploration,the seismic data collected by geophones contains the primary reflection wave from the underground reflection interface,as well as the ghost generated by the influence of the sea-air interface.In seismic records,the ghost follows the reflected wave,and the genetated false event axis will change the waveform characteristics of the primary reflected wave,which has a great influence on the resolution of the seismic record.In spectrum analysis,the frequency notch effect caused by the ghost will reduce the effective bandwidth to a certain extent.The ghost in marine seismic data brings difficulties to subsequent seismic processing and interpretation.Deghosting,improving the signal-to-noise ratio and resolution of seismic data,and broadening the bandwidth of seismic data are urgent issues that we must solve.First of all,this article focuses on the deghosting methods for horizontal cables and variable depth cables.Starting from the principle of the ghost generation.Forward modeling simulates the seismic records with the ghosts,and discusses the frequency characteristics of the ghosts and the influence of the ghosts on seismic signals under different streamer conditions and different geophone depth conditions.Subsequently,the principles of conventional deghosting methods in frequency-space domain,frequency-wavenumber domain,and Radon domain is studied,and the expressions of the ghosts operators in different domains are deduced,and these conventional deghosting methods are very depending on the ghost delay time,when the ghost delay time is not accurate enough,interference ringing artifacts will be generated in the seismic record.Secondly,this paper proposes an adaptive deghosting method under different conditions.Under different constrained conditions,the quality of ghost wave suppression is evaluated by defining margin,skewness,pulse factor and crest factor.Perform an optimal search for the ghost delay time to optimize the ghost operator and achieve the purpose of deghosting adaptively.This method is applicable to both horizontal cable and slant cable containing ghost data.Among them,maximizing the crest factor has a better effect on the deghosting,while maximizing margin,maximizing pulse factor and maximizing skew have poor effects on the deghosting.They can not completely suppress the ghost on the seismic data,and there will be residual ghost on the near and far offset gathers.Based on the inversion type Radon domain deghosting method,this method is based on the least square linear inversion method to remove ghosts.The linear equations of the primary reflected wave field and the total wave field are established in the Radon domain,and the linear Radon equation is obtained by inversion wave field.This method has a better effect of deghosting and is suitable for horizontal cables and variable depth cables.Deghosting method based on deep learning,this method introduces the idea of deep learning to image denoising into deghosting,and constructs a training data set by forward modeling a large number of seismic records with the ghosts,and builds Dn CNN and U-net has two network structures.Data tests show that this method has achieved better effects on deghosting,and multiple waves have also been suppressed.