The Study On Passive Source Full Waveform Inversion In Non-uniform Distribution And Weak Body Wave Signals

With the improvement of underground exploration accuracy and depth in the field of exploration geophysics,seismic exploration is gradually developing towards multi source joint exploration.In recent years,ambient noise passive source seismic data has attracted extensive attention because of its rich low-frequency information and flexible reconstruction.It shows a good application prospect in migration imaging and velocity modeling.Full waveform inversion is an advanced technology for velocity modeling in the field of seismic exploration.However,the application of passive source seismic data still faces some limitations,such as low signal-to-noise ratio and lack of source spatio-temporal information.The full waveform inversion of passive source based on cross-correlation seismic interferometry uses the virtual seismic records reconstructed by cross-correlation seismic interference as the observation data,which avoids the problem of accurate positioning of underground passive source.However,the inversion quality and effect are largely limited by the accuracy and signal-to-noise ratio of virtual records.But the signal-to-noise ratio of ambient noise passive source seismic data is very low,resulting in only weak body wave signals in the reconstructed records.In addition,the accuracy of cross-correlation reconstruction records needs the uniform distribution of passive sources underground,but in practice,the passive sources are basically in the state of non-uniform distribution,and their transverse distribution range usually does not wrap the surface survey line,which may only exist in the medium on one side below the survey line.At this time,the effective body wave signals in the cross-correlation reconstruction record will also be weakened,and the inversion effect will also be significantly negatively affected.Therefore,it is necessary to design a full waveform inversion method more suitable for passive source seismic data.Based on the background noise passive source seismic data,from the perspective of practical exploration,this paper studies from three angles to realize the robust high-precision passive source full waveform inversion.The first is to suppress the influence of background noise in passive source reconstruction records on inversion.The second is to alleviate the influence of relative amplitude distortion in cross-correlation reconstruction records on inversion.The third is to overcome the influence of the lateral non-uniform distribution of passive sources on inversion.Using the proposed passive source full waveform inversion technology,this paper studies the joint inversion of active and passive sources of different geological models,and further explores the characteristics and advantages of passive source full waveform inversion.The methods,technologies and research results proposed in this paper are summarized as follows:(1)Due to the low signal-to-noise ratio of ambient noise passive source seismic data,the full waveform inversion can only rely on the weak body wave signals to restore the underground velocity structure.According to the characteristics of interferometry noise in passive source reconstruction records,a passive source full waveform inversion method based on random interferometry reconstruction is proposed.The contribution of weak body wave signals to inversion is enhanced by dispersing the influence of interference noise on inversion,and the signal-to-noise ratio and velocity structure continuity of the final inversion results are improved.Aiming at the distortion of body wave amplitude information reconstructed by cross-correlation seismic interferometry,the global cross-correlation objective function is introduced into the passive source full waveform inversion,which effectively improves the accuracy of the passive source full waveform inversion results.(2)Aiming at the problem that the passive source full waveform inversion based on cross-correlation seismic interferometry is limited by the distribution conditions of passive sources,a robust passive source full waveform inversion method based on multi-dimensional deconvolution is developed,which overcomes the dependence on the transverse uniform distribution of passive sources,and the final imaging accuracy of passive source inversion is significantly improved.According to the characteristics of Green's function obtained from multi-dimensional deconvolution reconstruction,a passive source full waveform inversion objective function with independent source wavelet ability in L2 norm framework is constructed.To improve the main frequency of convoluted wavelet to carry out multi-scale passive source full wave inversion can enrich the effective velocity structure information of the final inversion results.(3)The full waveform inversion of active source seismic data and background noise passive source seismic data is compared from the aspects of observation data,frequency distribution and inversion effect.It is found that the passive source has obvious low-frequency advantages and can stably use the low-frequency information to restore the underground large-scale structure.The joint or series full waveform inversion of active and passive sources has high requirements for the inversion results of passive sources.In this paper,the joint inversion test of active and passive sources is carried out on Marmousi model.It is found that the passive source full waveform inversion method proposed in this paper can provide a reliable initial model for active source full waveform inversion without more low-frequency data under the condition of horizontal non-uniform distribution and small number of passive sources,and effectively improve the effect of active source full waveform inversion.In view of the problem that the conventional active source full waveform inversion can not invert the strong contrast velocity structure,this paper extracts the ultra-low frequency information from the passive source reconstruction record to construct the underground large-scale velocity structure,and successfully constructs the strong contrast velocity model under the assumption of weak scattering through the joint inversion of active and passive sources.