Deconvolutive Improved S Transform and Its Application in Hydrocarbon Detection

Seismic signals are usually nonlinear and nonstationary. The Fourier transform (FT) based on stationary signal processing theory cannot depict the frequency components at any moment. However, the time–frequency analysis (TFA) methods have the capability of describing the partial features of signal both in time and frequency domains. S transform (ST), as a common TFA method, has great time–frequency (TF) combination characteristics, but the changing trend of the window function is fixed and the TF resolution cannot be adjusted. In addition, for seismic signals, the peaks of the frequency distribution in the TF spectrum bias the actual Fourier spectrum, which will affect the accuracy of data analysis. Therefore, we propose a new TFA method called the deconvolutive improved S transform (DIST). The DIST introduces one parameter to the window function other than multiple parameters to improve the flexibility in the application process. The normalization factor is also removed from the window function to avoid the frequency bias. Moreover, the deconvolution in DIST can further improve the accuracy of TF representation. The comparison of the TFA results of synthetic seismic signals shows that the DIST has better TF resolution and energy aggregation than other TFA methods in this article. By adding different degrees of noise to synthetic seismic signals, we conclude that DIST has better noise robustness. Finally, we apply DIST to different field data for hydrocarbon detection, and the results are basically consistent with the drilling data.