A Preliminary Study Of Low-frequency Data Enhancement For Land Moving-coil Geophone

With the development of global seismic exploration,exploration of oil and gas has gradually reached the stage of lithological explor4 ation and deep exploration.The improvement of oil and gas exploration has made the discovery of new types of oil and gas reservoirs more difficult.The problems such as complex fault blocks,thin interbeds,and small structure identification in lithologic exploration impose higher requirements on seismic data resolution.In the field data collection work,the natural frequency of the conventional type geophone is around 10 Hz,therefore,when the conventional moving-coil geophone is used to receive seismic data,the received low-frequency signal below 10 Hz is always distorted,which is because of the sensitivity of the geophone.It is difficult for the low-frequency information of seismic signals to be corrected and used.If reliable low-frequency components of seismic data are missing,the sidelobe of the wavelet will increase.The resolution of seismic data will be reduced and the ambiguity of data interpretation will increase.In addition,low-frequency components of seismic signals can effectively penetrate the strata and improve the quality and accuracy of deep seismic imaging.Therefore,the recovery of low-frequency seismic data in seismic exploration is becoming more and more important.Replacing conventional moving-coil geophones with broadband geophones can provide true and reliable broadband data,but it can also cause practical problems such as a long experimental cycle,the difficulty in field construction,and increases the high construction costs.In order to solve the problem that low-frequency signal is distorted when the conventional moving-coil geophones acquire the data.The thesis proposes a low-frequency data enhancement technique for conventional moving-coil geophones.Based on the difference of the natural frequencies between the moving-coil geophones and broadband geophones,the research matches the low-frequency information received by conventional geophones to the low-frequency information of broadband geophones that locates at the same locations.First,the lowest reliable frequency of the conventional moving-coil geophones needs to be determined,next,one applies the time-frequency transform on the seismic data received by the two types of geophones in the low-frequency band.Finally,a matching filter is used in the frequency domain or time-frequency domain,and one completes the low-frequency data enhancement of the conventional moving-coil geophone.The study of this technique is currently limited only on testing the one-to-one correspondence deployment of the conventional moving-coil geophone and broadband geophone.Theoretically,this method can also be applied to the situation that the broadband geophones are as the sparsely constrained condition,thus will improve the feasibility of practical applications.This technology can effectively enhance the low-frequency information of conventional moving-coil geophones,broaden the effective frequency range,and increase the time resolution of seismic data,which can provide broadband data to increase accuracy of velocity analysis and full waveform inversion.The technique helps us to reduce the interpretation ambiguity of seismic data.Because the wavelet matching and matching filters in the low-frequency enhancement are sensitive to random noise,the thesis also studies the problem of random noise suppression,which can fits with the low-frequency data enhancement.Random noise in prestack seismic data is unavoidable and always greatly influences the signal-to-noise ratio in seismic data.Conventional denoising methods will attenuate a part of low-frequency energy in the useful signal,which makes it hard to enhance and further use low-frequency components.This thesis proposes a hybrid wavelet-like transform for the attenuation of random noise.Due to the inability to characterize the plane wave,the traditional denoising method by using wavelet transform with thresholding will destroy seismic events,and a certain amount of residual noise will exist.The seislet transform is a wavelet-like transform method developed specifically for seismic data.It has better sparse representation for seismic data and can effectively compress seismic data in the spatial direction.After dealing with seislet transform,the seismic data meets the piece-wise smooth assumption,which is precondition for applying of the wavelet transform.Therefore,the thesis proposes to apply a discrete wavelet transform in the seislet domain to further compress the seismic coefficients in the time direction,which is called as a hybrid wavelet-like transform.In the hybrid wavelet-like transform domain,the energy of the useful signal is mainly distributed on the large-scale wavelet coefficients.By applying an appropriate thresholding function to suppress random noise in seismic data,one can avoid the loss of the useful low-frequency components and protect the useful low-frequency information to the most extent.Finally,the master's thesis develops a relatively complete set of processing framework for low-frequency data enhancement of conventional moving-coil geophone.