A Study On Denoising Method Of F-X Domain

Seismic exploration is a very effective geophysical method in the search for oil and gas and coal fields.After years of exploration,geological structures with shallow depth,simple structure have been basically identified;the remaining undiscovered areas are basically geological structures with deep burial,complex structures.This will increase the demand for seismic data exploration methods and data processing and interpretation.At the same time as the acquisition method is improved,the requirements for improving the signal-to-noise ratio of seismic data processing are also increasing.Therefore,high-fidelity,high-resolution,high-signal-noise and other requirements are correspondingly proposed in seismic data processing.The high fidelity and high resolution are based on the high signal to noise ratio.This paper is based on the study of denoising in the FX domain with improved signal-to-noise ratio of seismic data,in order to reach a certain effect in the denoising of the FX domain through the study of this thesis.In the process of seismic data processing,seismic data is usually transformed into different domains according to the distribution law of noise in different domains in seismic data,and statistical denoising is performed according to the distribution law of noise in different domains.The FX domain predictive filtering and denoising technique is a relatively effective method to remove random noise.Its theoretical basis is that the straight line or approximate linear signal in the FX domain has a regular pattern in the X direction,that is,these signals can be calculated.Random noise is not regular in the X direction and cannot be extrapolated.Based on this,the predictive filter operator at each frequency point can be found,and the filter operator can be applied to the original data at this frequency point to predict the data containing only the valid signal,thereby suppressing the random noise.In this study,various coherent signals and random noise were first transformed from the TX domain to the FX domain,and their differences in the FX domain from the effective signal or the regular signal were analyzed,and the characteristics of the random noise in the FX domain were obtained.According to its characteristics,the FX domain predictive filtering is realized by programming;and the programmed program is tested by the theoretical model,and the distribution law of random noise in the FX domain and the selection rule of the relevant parameters in the FX domain denoising are obtained.The effective reflection signal in actual seismic data is a hyperbola.Although the FX domain prediction filter has strict requirements on the data,that is,the effective signal is required to be linear or nearly linear,the phase of the effective signal in the common offset gather is relatively flat.Therefore,you can first extract the seismic data into a common offset gather,and then perform FX prediction filtering.This study first theoretically deduces the seismic records containing multiple linear events and obtains a predictable pattern of linear events.This model is in good agreement with the AR model.Therefore,in this paper,the data can be predicted using the autoregressive AR model,and the optimal solution of the AR model can be obtained by the least squares method.The autoregressive AR model can be applied to the prediction of seismic records with multiple linear events.Obtain the required predictors,further convolute with the original data,and get the predicted data to achieve the purpose of improving the signal-to-noise ratio of seismic data.In the verification of the theoretical model data,this study conducted a large number of experimental verifications from the aspects of operator length,signal-to-noise ratio,distribution of random noise,and unbalanced energy of events,and obtained the law of selection of relevant parameters.The FX domain prediction filter was applied to the theoretical and practical total offset data.Finally,the FX domain prediction filtering was performed on the stacking profile,and an ideal result was obtained.By comparing the F-X filtering module in VISTA seismic data processing software with the filter results of the F-X domain prediction filtering program in this study,the F-X domain prediction filtering program in this study has better processing effect than the F-X filtering module in VISTA seismic processing software.