Study On Seismic Data Reconstruction Methods Based On Seislet Transform

Seismic data reconstruction plays an important role in seismic data pre-stack processing.It is significant for subsequent processing steps such as surface-related multiple elimination,3D seismic migration and so on.Reconstruction methods based on sparsity-promoting transforms have been widely used.For these methods,different sparsity transforms,iterative algorithms,threshold models and noise-related problems are the main issues concerned.In this paper,considering the great advantage of Seislet transform in sparsity-promoting,we select Seislet transform as our sparsity-promoting transform and we compare it with other traditional transforms such as FFT and Curvelet transform,we analyze the advantages and disadvantages of Seislet transform.We mainly discussed four iterative methods including POCS,IHT,Bregman and JRSI.We also study on linear threshold model,exponential threshold model and data-driven threshold model.In noisy situation,regular reconstruction methods usually don't perform well.Thus we have to add a weighted factor in the iterative algorithm to reduce the effects of noise.In this paper,we discussed the calculation and selection of weighted factor and we analyzed constant model,linear model and data-driven model.In this paper,we studied the above key issues through numerical and real data examples and we analyzed the effects of different parameters on the results of seismic data reconstruction.