Considering the difficulty in identifying faults in medium-deep strata with low signal-to-noise ratio(SNR),a multi-scale fault identification method based on Radon transform is developed.Specifically,the poststack seismic data are processed by noise suppression to improve the SNR of raw data.The noise suppression mainly deals with random interference and high-frequency noise,and removes weak boundary information caused by rock disturbance.The seismic responses at different frequency bands to different orders of faults are monitored through frequency division processing,and then sensitivity analysis is conducted to optimize the sensitive seismic attributes of faults of different scales.Finally,the Radon transform is used for edge detection enhancement to improve the fault edge effect.The method is essentially to introduce the Radon transform based on image reconstruction theory into the three-dimensional interpretation of faults,which is different from conventional edge detection algorithms.It is an integral opera-tion with the characteristics of linear enhancement and high noise resistance.This method was applied for the first time in the shal-low water area of the Zhu-I Sag,the Pearl River Mouth Basin.The results demonstrate that the proposed method significantly im-proves the accuracy and resolution of fracture prediction compared to conventional ant body slicing.It reveals that the LF_A buried-hill structure of the Lufeng 13 sag mainly develops two groups of fractures(NNW and NE),guiding the identification of high-quality granite reservoir distribution.The method helps confirm the fault development in the EP_B structure of Yangjiang East sag,which is believed to be a left echelon structure controlled by strike slip.There is a large NEE-trending central strike slip fault zone in the Lufeng C sag,which,together with the NW-trending regulating fault,controls the formation of central nose shaped belt traps.Under the new model,a number of structural traps have been discovered,improving the exploration potential of the sag.
维普
万方数据
