Abstract: Considering the ability of self-adaptive sparse representation for matching pursuit algorithm,we proposed a method for sedimentary boundary identification and fluid detection on the seismic data with strong background reflections shielding the effective information in weak reflections.Integrated of log data,uphole seismic trace and matching pursuit algorithm,the strong background reflections could be recognized and peeled away from the original seismic data;meanwhile,some interference of “false bright spots” and artificial continuous distribution of sand bodies could be weakened easily.Then,the elastic impedance inversion and fluid factor detection could be implemented to finely characterize the distribution of sand bodies and improve the accuracy of lithological trap detection and fluid identification.Applications on theoretical model and field seismic data demonstrate the reliability and effectiveness of our method.