Research On High-density Seismic Weak Signal Extraction And3D Processing Methods

With the increasing requirement of seismic prospecting accuracy and rapid developmentof computer technology, the high density seismic exploration has received more and moreattention. Because it costs too much for good high density seismic data, how to make full useof it is very important. This paper makes an attempt on methods study of high density seismicdata processing from both sides. On the one hand, how to extract weak signals with massnoise in low signal-to-noise ratio seismic data is studied. On the other hand, in order to protectthe3D information of high density seismic data, the paper studied some3D processingmethods.Firstly, the paper introduces the features and the layout designing method of high densityseismic exploration. As the theory evidence of following high density forward modeling andcharateritics analysis, several notable issues on layout designing are analyzed in theory. Inaddition, we also introduced the advantages of high density seismic exploration insignal-to-noise ratio, resolution and fidelity.In order to extract weak signals, suppressing the effect of noise as much as possible isnecessary, at the same time; we must see that effective signals are still needed to be wellprotected. Curvelet transform is good amplitude preserved denoising technology. Nevertheless,it has some drawbacks when applied to weak signals extraction. Therefore, the paperimproved Curvelet transform from three aspects:1) threshold algorithm was applied to noisedominant IMFs after EMD instead of the whole data, which can avoid the damage foreffective waves caused by using global threshold in Curvelet transform;2) When denoisingusing curvelet transform, a smaller decision threshold than that estimated was used. By doingthis, smaller curvelet coefficients related to effective weak signals were maintained while some undesired wild values appeared;3) median filtering method was presented to eliminatewild values in curvelet domain.For study on3D processing methods, three parts including3D high density forwardmodeling,3DFK and HOSVD were selected. Forward modeling is a very efficient methodused in seismic wave filed analysis. Based on3D high order finite difference forwardmodeling in according with high density layout design, the paper studied the characteristicsand advantages of high density seismic exploration. Next,3DFK conical filtering wasdeveloped to suppress surface wave based on cross-spread gather sorted from high densityseismic data after research on surface wave propagation based on surface wave forwardsimulation and linear disturbance suppression using3D FK method. Hyperbolic surface wavefrom those lines far away from shot location got well suppression. Finally, SVD methodwidely used in post-stack denoising was extended from2D to3D. HOSVD was proposed todenoising for3D seismic data which can be viewed as a three order tensor and improved S/N.