Research Of Noise Suppression Of Seismic Physical Simulation Data

Seismic physical modeling is in a small scale physical model,using ultrasonic simulation field seismic to research wave characteristics on geological,field acquisition,data processing and data interpretation theories.Seismic physical simulation system,will produce noise at working and produce multiple wave when seismic wave propagation in the physical model,in order to improve the signal-to-noise ratio and resolution,getting high quality prestack data,needing to suppress noise and try not to damage the signals effectively.Analysing system noise can be found that it has the characteristics of random white noise and different in the frequency domain.Applying of two-dimensional F-X prediction filtering technology,3d F-X prediction filtering and the singular value decomposition technique to suppress noise,contrast between the different methods of denoising effect,finding that f-x predictive filtering using modified covariance has the best effect of the system.Applying predictive deconvolution,selecting the appropriate prediction step can eliminate the full-process wave that has obvious periodicity.Using the linear Radon transform,parabolic Radon transform and the least squares high resolution Radon transform to eliminate internal multiple waves.The compared results show that the least squares high resolution Radon transform has the best effect on removing the internal multiple.Analysis the noise characteristics of data and considering the effect of different denoising method,choosing combination method which use predictive deconvolution to eliminate the full-process multiple firstly,then use least-square high-resolution Radon transform to suppress multiple wave between the layers.