Generalized S-Transform Based on Sparse Representation and Its Application on Attenuation Gradient Estimation

Absorption attenuation caused by the fluid is an essential indicator for reservoir delineation. The attenuation gradient estimation method related to time-frequency (TF) decomposition (TFD) is an effective way for reservoir prediction. However, the attenuation gradient estimation method heavily relies on spectral decomposition, which directly impacts the accuracy of reservoir localization. To improve the resolution, a new TFD method called sparse generalized S-transform (SGST) is proposed in this letter. The optimal scale parameters of the window function in SGST can be flexibly determined based on the specific characteristics of the given signal. The analysis of the synthetic seismic signal indicates that the proposed method has higher TF resolution and energy concentration. On this basis, a high-precision attenuation gradient calculation method based on SGST (AGSGST) is proposed and it is used for reservoir prediction. The application of field seismic data has shown that the AGSGST provides higher resolution results when compared with other methods, offering more accurate results in characterizing reservoir edges, which are more consistent with actual well-logging curve interpretation results. The proposed method can describe the spatial distribution of the hydrocarbon reservoirs more clearly, and help to promote reservoir development.