High-Order Time-Reassigned Synchrosqueezing Transform and Its Application in Tight Sandstone Gas Reservoir

Time-frequency analysis (TFA) can provide relevant information about subsurface reservoirs, which is essential in reservoir identification. Traditional TFA methods are limited by resolution, resulting in unreliable seismic interpretation. The time-reassigned synchrosqueezing transform (TSST) is a practical TFA approach for processing non-stationary signals. However, for a frequency-domain signal with a strong-varying group delay (GD), the TSST provides a blurred time-frequency representation (TFR). We propose a novel TFA method called high-order TSST (HTSST) to address this problem. First, to describe the signal with GD variations, we develop a frequency domain signal model. Next, the high-order partial derivative of the short-time Fourier transform (STFT) is obtained based on the high-order Taylor expansion of the signal, and a matrix determinant is constructed according to the results. The matrix determinant is then used to deduce an explicit GD estimation formula in the time-frequency domain. Finally, the high-resolution TFR of the HTSST is generated by squeezing the energy to the estimated GD. With a strong-varying GD, frequency-domain signals can thus benefit from a high-resolution TFR provided by HTSST. Numerical simulations are performed to investigate this method's efficiency. Through the analysis of seismic field data, the efficacy of this method for tight sandstone gas reservoir identification is verified.