Analyzing the time-frequency features of seismic traces plays an important role in seismic stratigraphy analysis and hydrocarbon detection. The current popular time-spectrum analysis methods include short-time Fourier transform, continuous wavelet transform, S-transform, and matching pursuit (MP), among which MP is the most tolerant of the window/scalar effect. However, current MP algorithms do not consider the interfering effects of seismic events on the estimation of optimal wavelets in each decomposition iteration. The interfered reflection seismic events may result in inaccurate estimation of optimal wavelets during the whole decomposition procedure. We have developed a hybrid basis MP workflow to minimize the effect of event interference on the estimation of optimal wavelets. Our algorithm assumes that the wavelet features remain constant in a user-defined small time window. The algorithm begins with identifying the strongest reflection waveform. Next, we estimate the optimal wavelet and the corresponding reflectivity model for the selected waveform by using a basis pursuit algorithm. Then, we subtract the seismic traces from the waveform computed from the optimal wavelet and estimated reflectivity model. We repeat this procedure until the total energy of seismic traces falls below a user-defined value. We have determined the effectiveness of our algorithm by first applying it to a synthetic model and then to a real seismic data set.

We have developed and applied an inverse Q filter formulation using synchrosqueezed wavelet transforms for the compensation of attenuating and dispersive media. A damping criterion concerning the reconstruction of the effective components for controlling noise amplification and the separation of the noise and signal in the synchrosqueezed wavelet domain was generated. The proposed method provides a stable attenuation compensation without decreasing the seismic vertical and lateral resolution. The best property of the proposed method, unlike conventional inverse Q filtering methods, is that it carries out amplitude compensation for the effective components located at some time samples in the time-frequency domain. The spectral reconstruction contributes to the reconstruction of the trace in the time domain and suppresses the ambient noise located at high frequencies at later times, especially suppressing the ambient noise within the main frequency band. It is not a noise-level dependent method. We validated our approach with synthetic and real data. The comparison of the proposed method with the conventional stabilized inverse Q filtering method was also carried out to illustrate the particular features of the proposed method. The examples demonstrate that our proposed method can both effectively compensate for the amplitude attenuation by suppressing the ambient noise and further provide seismic images at high resolution while highlighting the effective details. Furthermore, it is a robust and easily tunable algorithm.

Lithology identification is crucial for characterizing complex unconventional reservoirs, where thin interlayers significantly influence hydrocarbon accumulation. Although deep learning-based methods utilizing well logs have become prevalent, most approaches treat well logs as generic 1D time series, frequently neglecting the multi-scale geological information inherent in the data. This oversight limits their accuracy and generalizability, especially in geologically complex environments. To overcome this limitation, we propose a novel geology-driven deep learning framework. Our key contribution is the transformation of 1D well logs into 2D multi-scale feature maps through multiresolution wavelet decomposition, a process designed to explicitly represent geological features resembling sedimentary cycles. These feature maps are subsequently processed by a novel Geology-Guided Hybrid Network with channel-spatial attention, which integrates a 2D CNN to capture geological patterns and a Bidirectional long short-term memory to model sequential dependencies. Evaluated on field data from complex reservoirs, our method achieves an outstanding F1-score of up to 0.966, outperforming four established deep learning benchmarks. Importantly, the approach demonstrates improved accuracy in identifying thin layers and enhanced generalization across wells with differing lithological distributions, attaining an F1-score of 0.885 on a challenging test well exhibiting significant data drift. This study validates the robustness of our geology-informed approach and offers an effective framework for high-precision lithology identification.

First-arrival traveltime tomography is an essential method for obtaining near-surface velocity models. The adjoint-state first-arrival traveltime tomography is appealing due to its straightforward implementation, low computational cost, and low memory consumption. Because solving the point-source isotropic eikonal equation by either ray tracers or eikonal solvers intrinsically corresponds to emanating discrete rays from the source point, the resulting traveltime gradient is singular at the source point, and we denote such a singular pattern the imprint of ray-illumination. Because the adjoint-state equation propagates traveltime residuals back to the source point according to the negative traveltime gradient, the resulting adjoint state will inherit such an imprint of ray-illumination, leading to singular gradient-descent directions when updating the velocity model in the adjoint-state traveltime tomography. To mitigate this imprint, we solve the adjoint-state equation twice but with different boundary conditions: one being taken to be regular data residuals and the other taken to be ones uniformly, so that we are able to use the latter adjoint state to normalize the regular adjoint state and we further use the normalized quantity to serve as the gradient direction to update the velocity model; we call this process ray-illumination compensation. To overcome the issue of limited aperture, we have developed a spatially varying regularization method to stabilize the new gradient direction. A synthetic example demonstrates that our method is able to mitigate the imprint of ray-illumination, remove the footprint effect near source points, and provide uniform velocity updates along raypaths. A complex example extracted from the Marmousi2 model and a migration example illustrate that the new method accurately recovers the velocity model and that an offset-dependent inversion strategy can further improve the quality of recovered velocity models.

Characterization of facies within a hydrocarbon reservoir is essential for potential prospect identification and evaluation. We have developed a practical workflow that integrates poststack seismic attributes and well-log facies analysis to understand the development and depositional setting of the Triassic-age Akekule Formation in Tahe field, Northwest China. The workflow begins with sequence and sedimentary cycle analysis on selected benchmark wells. We then identify sand bodies within each sedimentary cycle using well logs. The analysis from well logs and drilling cuttings together illustrate that there exist five sand bodies within the fourth member of Akekule Formation. We next predict the horizontal distribution of each sand body based on the facies analysis on well logs. We finally perform multiple poststack seismic attributes analyses to identify seismic facies. Multiattribute analysis is implemented either through corendering multiple seismic attributes or neural network clustering on multiple seismic attributes. The lithofacies obtained from seismic attribute analysis is calibrated with facies identified using well logs.

Seismic attenuation as represented by the seismic quality factor Q has a substantial impact on seismic reflection data. To effectively eliminate the interference of reflection coefficients for Q estimation, a new method is proposed based on the stationary convolutional model of a seismic trace using variational mode decomposition (VMD). VMD is conducted on the logarithmic spectra extracted from the time-frequency distribution of the seismic reflection data generated from the generalized S transform. For the intrinsic mode functions after VMD, mutual information and correlation analysis are used to reconstruct the signals, which effectively eliminates the influence of the reflection coefficients. The difference between the two reconstructed logarithmic spectra within the selected frequency band produces a better linear property, and it is more suitably approximated with the linear function compared to the conventional spectral-ratio method. Least-squares fitting is finally applied for Q estimation. Application of this method to synthetic and real data examples demonstrates the stabilization and accuracy for Q estimation.