Attenuation (1/Q) estimation in reflection seismic records

Despite its numerous potential applications, the lack of a reliable method for determining attenuation (1/Q) in seismic data is an issue when utilizing attenuation for hydrocarbon exploration. In this paper, a new method for measuring attenuation in reflection seismic data is presented. The inversion process involves two key stages: computation of the centroid frequency for the individual signal using a variable window length and fast Fourier transform; and estimation of the difference in the centroid frequency and travel time for paired incident and transmitted signals. The new method introduces a shape factor and a constant which allows several spectral shapes to be used to represent a real seismic signal without altering the mathematical model. Application of the new method to synthetic data shows that it can provide reliable estimates of Q using any of the spectral shapes commonly assumed for real seismic signals. Tested against two published methods of Q measurement, the new method shows less sensitivity to interference from noise and change of frequency bandwidth. The method is also applied to a 3D data set from the Gullfaks field, North Sea, Norway. The trace length is divided into four intervals: AB, BC, CD, and DE. Results show that interval AB has the lowest 1/Q value, and that interval BC has the highest 1/Q value. The values of 1/Q measured in the CDP stack using the new method are consistent with those measured using the classical spectral ratio method.