Seismic resonance: Wavelet-free reflectivity retrieval via modified cepstral decomposition

Abstract

Spectral decomposition is a proven tool in seismic interpretation. It helps interpreters highlight channels and map temporal bed thickness and other geologic discontinuities. In spectrally decomposed seismic data, amplitude spectra contain notch patterns. The period of these notch patterns inversely correlates to the reservoir layer’s thickness and/or its interval velocity. Additional cepstral decomposition will directly extract the bed time thickness or arrival time under simple reflectivity setups. Based on these observations, we have adopted a new workflow for reflectivity retrieval in stably phased seismic. There is no need to understand the details of the wavelet in this workflow. We find that the reflector time and its “apparent strength” can be identified in a transformed seismic resonance domain resulting from a modified cepstrum analysis. We find that linear hot spots in this domain lead to direct quantification of the reflectivity series. The timing and strength of the linear hot spots reveals reflector times and scaled reflectivity coefficients. We applied this method for thickness prediction for a target deepwater reservoir with a complex geologic setting. Because of the large reservoir thickness variations and weak impedance contrast with underlying lithology, identification of the base reservoir and therefore the reservoir thickness has been historically challenging in this field. In a blind test, the reservoir thicknesses evaluated from this method are close to the true vertical thickness found in wells.