Seismic Band Expansion Based On Compressed Sensing

In the era of increasingly difficult exploration and development of current oil and gas reservoirs,increasing the resolution in the entire seismic data processing process has become increasingly important.Conventional resolution-enhancing techniques generally use deconvolution methods or some modifications based on deconvolution.However,in actual data,the seismic wavelet does not fully satisfy the minimum phase hypothesis,and the real underground structure does not satisfy the assumption that the reflection coefficient is white noise.In this paper,firstly,the phase of the seismic wavelet is transformed into zero-phase by the time-varying phase solution method so that the uncertainty and inaccuracy brought about by the time-varying phase can be avoided in the subsequent steps.At the same time,the spectrum wavelet based on high-order statistics is used.The extraction method does not require any assumptions about the wavelet and satisfies the actual time-varying characteristics.The estimated wavelet is closer to the actual situation.Then the compressive sensing method is used to recover the seismic reflection coefficient.The method of this paper takes into account the existence of zero value in the spectrum of the wavelet.Adding a coefficient can ensure that the entire reflection coefficient can be solved by using the spectrum of the calculation part;The effects of random noise are also fully considered.Through numerical simulation and physical model tests,it was found that the reflection coefficient can be recovered and the error obtained is relatively small,and the continuity of the isotropic axis reflected in the two-dimensional model is very good.The processing results of the data also show the advantages of this method for expanding the frequency band of seismic signals,and the resolution of seismic data has been significantly improved,which is very critical for the thin layer recognition in subsequent processing.