Gas Reservoir Detection Using Mixed Components Short Time Fourier Transform (MC-STFT) as a new attribute

Identification of gas reservoirs as a main natural resource due to its economic importance has always been one of the most important issues in research studies in the oil and gas field. Accurate localization of a gas reservoir through seismic data has been broadly studied. The final destination of all seismic attributes is to distinguish a specific feature. Accordingly, many seismic attributes have been developed among which short time Fourier transform (STFT)-based methods play an important role. The location of gas reservoirs can be detected taking advantage of its particular criteria in seismic data. Generally seismic signals are nonstationary as their frequency responses vary with time. So we propose an attribute which utilizes mixed components of STFT (MC-STFT). The novelty about this method is that without altering STFT method or adding any complexity, MC-STFT is able to detect gas reservoirs at high resolution. Simplicity and time efficiency can make a method superior. In fact, this method takes advantage of extracting three frequency components obtained by STFT. In the next step, we can do the second iteration of the procedure, this will represent the degree of sharpness of reduction in amplitude and again do the same jobs as before and it leads to this, make it more specific. We apply this method on three data sets, first, Marmousi model and then two other real seismic zero-offset sections. To evaluate the proposed method compared with the Synchrosqueezing STFT (SSTFT). The results confirm good performance of MC-STFT in high resolution gas reservoir detection.
The location of gas reservoirs can be detected taking advantage of its particular criteria in seismic data. Generally seismic signals are nonstationary as their frequency responses vary with time. There are some techniques called Time-Frequency Decomposition (TFD) which map a 1D signal into a 2D plane of time and frequency. In this way the frequency content of the signal with respect to time can be revealed. So TFD methods used as spectral decomposition in both seismic processing and interpretation (Reine et al., 2009; Chen et al., 2014). For example, Partyka et al., (1999) adopted the windowed discrete Fourier transform (DFT) for reservoir characterization. To detect low frequency shadows beneath hydrocarbon reservoirs, Castagna et al., (2003) applied the matching-pursuit decomposition. Sinha et al., (2005) proposed a novel method of taking a Fourier transform of the inverse continuous wavelet transform (CWT) as a time-frequency map to identify subtle stratigraphic features (Zhang et al., 2019). Wu and Zhou (2018) developed synchrosqueezing wavelet transform (SWT) to reallocate the wavelet transform values to different points and produce a sharp spectral decomposition for the input signal (Mateo et al., 2020). Li and Zheng (2008) took advantage of the smoothed pseudo-Wigner-Ville distribution (SPWVD) for carbonate reservoir characterization.

Spectral decomposition has been applied in exploration feilds such as hydrocarbon detection, seismic attenuation analysis, channel identification, and thin-layer thickness estimation (Quan & Harris 1997, Gao et al. 1999, Liu & Marfurt 2007, Zhou et al. 2019, Odegard et al. 1997). Conventional spectral decompositions have some restrictions such as Heisenberg uncertainty principle and cross-terms which limit their applications in signal analysis. In an effort to overcome some of the limitations, use has been made of the STFT (Siddique et al., 2023; Yang et al., 2019). Recently, valuable efforts are done to cover these limitations, Barabadi et al., (2024) used synchroextracting transform for AVO analysis in time frequency and Shirazi et al., (2023) employed Multi-synchrosqueezing transform to detect shallows gas.

In this article we propose a novel seismic attribute to detect gas reservoirs which is based on STFT (Cohen, 1989). The superiority of this method relies not only on its simplicity (which doesn’t add any mathematical burden to STFT method) but also on the high resolution characterization it provides. This method takes advantage of seismic low frequency shadows as a gas reservoir indicator. The novelty behind this algorithm is in seismic signal transformation from time domain to time-frequency domain using STFT and then extraction of three frequency sections of each signal. This approach converts seismic zero-offset section into a 2D image of gas reservoir.

We assess the performance of the proposed algorithm against three models including Marmousi model and other two real data. The results show that the first iteration of this algorithm can locate gas reservoirs at high resolution which can also be much more accurate by applying the second iteration in comparison to the method SSTFT.