| With the development of mineral resources exploration,the geological condition of mining area has gradually shifted from simple tectonic areas to complex tectonic areas.The regional geological conditions of complex structures are complex,and the medium and small scale discontinuous geologic bodies are relatively developed,such as faults,cracks,etc.The safe mining of minerals is seriously affected by a variety of geological disasters,such as pit water inrush,coal and gas outburst,rock burst,etc.Constrained by Snell’s law,the traditional seismic exploration method,which is mainly based on reflections,is beneficial to characterize the interface of underground continuous strata,but it has limitations on the location and identification of medium and small scale tectonic geologic bodies,which seriously affects the safe mining and production of mineral resources.As seismic waves diffraction in geological discontinuity penalty on formation of secondary source form transmission wave field,abide by Huygens’ principle,with spherical wave spreading around,is an important information carrier of discontinuity underground geological body,has the advantages of uniform energy distribution and lighting area widely,and is helpful to identify underground discontinuous geologic abnormal body.Diffraction is weak in energy and is often shielded by reflection with strong energy.And conventional seismic exploration data processing methods are mainly aimed at reflection,which is not conducive to the recovery and enhancement of diffraction energy.Therefore,separating the weak diffraction from the strong reflection in the original seismic data and imaging the diffraction with high precision are the focus and difficulty of the diffraction research at present,which is of great significance for the accurate identification and detection of discontinuous structural geological bodies in underground space.Firstly,the characteristics of diffractions are analyzed,and the kinematic and dynamic characteristics of diffractions in different domains are summarized.In the poststack domain,the in-phase axis of reflections are linear with good continuity and strong energy,while the diffractions are hyperbolic with weak energy,which is often suppressed by the strong energy of reflections.Therefore,the diffractions can be separated by predicting the strong reflections in the post-stack domain.In the common shot domain,the diffractions and the reflections show similar curve shape.The reflections can be predicted by the property that the reflections energy can be focused to the virtual source point,and the diffraction wave can be extracted effectively.In the common imaging point domain,the diffractions are linear,while the reflections are curved with stagnation point,which is beneficial to the separation and imaging of diffractions.Just as the reflection coefficient determines the amplitude and polarity of the reflections,the variation of the diffraction coefficient also determines the energy strength and polarity of the diffractions,and the understanding of its variation is conducive to the accurate separation and imaging of the diffractions.Conventional rectangular coordinate system is not good for intuitionistic analysis of the spatial variation of diffractions with azimuth angle,incident angle,emergence angle and other angles in three-dimensional space.Based on the theory of geometrical physics diffraction,this thesis solved the three-dimensional diffractions in spherical coordinate system and deduced the expression of diffraction coefficient in threedimensional space.The variation law of diffraction coefficient with azimuth in common shot domain,common receiver domain and common depth domain are analyzed.The reflections in seismic records follow the assumption of plane wave incident.The traditional plane wave decomposition(PWD)method starts from the data and constructs a prediction operator by using adjacent seismic records to estimate the local dip field of reflections,so as to predict the reflections and extract diffractions.The traditional PWD method converts the local dip angle into the local dimensionless dip angle related to offset and sampling data,and calculates the local dip angle value of reflections by the time and space directional derivative operator.However,this process is driven by data.When the random noise in seismic data is serious,the calculation accuracy of local dip angle and the separation quality of diffraction wave will decrease.Therefore,an improved PWD diffractions separation method based on Hilbert transform is proposed in this thesis.Based on the traditional PWD method,the estimation method of local dip is improved,and the frequency response relationship between directional derivative operator and Hilbert transform is used to redefine the local dip angle of reflection.The gaussian smoothing operator is used as the regularization factor to smooth the local dip and enhance the stability of the local dip angle estimation,so as to effectively predict the reflections and separate the diffractions.In the prestack common shot gathers,the reflections and the diffractions are both hyperbolic,and it is difficult to separate the diffraction wave with the traditional PWD method,but relative to the post-stack domain prestack diffraction information more abundant,the reflection from the same shot or diffraction phase axis continuity is better,and diffractions are often damaged in CDP stack process.Therefore,in order to extract more abundant diffraction information,the PWD diffractions separation method based on travel time angle parameterization is proposed in the pre-stack shot domain.According to the different focusing properties of diffraction and reflection waves,the NMO velocity and near-surface ray parameters of the reflections are calculated to estimate the local dip field of the reflections in the prepile shot-domain,and then the hyperbolic reflections in the shot-domain are predicted accurately,and the reflections are suppressed in the PWD separation process to achieve the extraction of diffractions.The pre-stack PWD diffractions separation method based on travel time angle parameterization can obtain the two parameters of NMO velocity and near-surface ray parameters without complex calculation.The NMO velocity can be obtained by using the traditional velocity analysis method,and the ray parameters can be calculated on the stacked section.Using these two parameters,an accurate local dip angle prediction operator is obtained to filter the reflections and realize the diffractions separation.Numerical simulation and practical data application show that the PWD diffraction separation method proposed in this thesis can effectively separate the hyperbolic diffractions and hyperbolic reflectiosn in the pre-stack shot domain.In actual seismic exploration,the structure of underground media is often complicated,which makes the continuity of the reflection’s in-phase axis poor,makes the reflections prediction inaccurate.Because of the big difference in energy between the diffractions and the reflections,the influence of random noise on the weak diffractions are more prominent after removing the reflections.Therefore,in this thesis,after the separation of diffractions,the diffractions imaging method in angle domain based on correlation weighting is used to further suppress the noise in the process of diffractions migration imaging,so as to improve the diffractions imaging quality and the identification accuracy of discontinuous geological bodies.In this thesis,CIGs in angle domain is extracted from the separated diffraction wave field,and the dip angle range of CIGs from the same underground imaging point is analyzed by cross-correlation,and the correlation coefficients of diffractions in each angle group on CIGs are obtained.Group with different angle of correlation coefficient weighted imaging,grouping migration imaging effect by comparing the different angles to get gathers the best grouping dip angle,so as to achieve the purpose of removing noise,improve the quality of underground small geologic body imaging,realize the precise identification of discontinuous geologic bodies and positioning. |