Porosity is a critical petrophysical property for reservoir characterization. While conventional porosity inversion involves complex processes and factors, deep learning methods offer a more intelligent alternative. However, existing training dataset modeling strategies are inadequate for complex geological formations, whereas seismic facies-controlled modeling method enables fine characterization of underground structures. In order to achieve intelligent and fine porosity prediction, we propose a facies-controlled porosity prediction method constrained by seismic attributes. First, stochastic modeling is used to generate heterogeneous background models, enhancing the spatial variability of the dataset to reduce discomfort. Second, sensitive seismic attributes are selected as facies labels to construct seismogram and porosity training sets with facies-controlled significance. Finally, a designed neural network establishes the intrinsic relationship between seismogram and porosity, enabling petrophysical properties for other seismic sections in the same area. Validation using reservoir model data confirms the method’s feasibility and finer resolution compared to conventional method, offering enhanced accuracy in reservoir prediction. Field data from tight sandstone further demonstrates the superiority in reservoir characterization.

The detection of seismic small faults is vital in shale oil and gas exploration and development. Limited by the resolution of seismic exploration, it is difficult to effectively detect small faults. Recently, many fault characterization methods have been proposed. To overcome the obscurity of seismic resolution for small faults, time–frequency analysis algorithms and fault attributes have been employed to characterize small faults and stratigraphic inflection point. However, the traditional resolution of seismic time–frequency analysis algorithms greatly limits the accuracy of small fault identification. Therefore, there is a need to improve the resolution of seismic time–frequency analysis algorithms. Herein, we propose a new time–frequency analysis algorithm and workflow, high-order multichannel synchrosqueezing variational modal generalized S-transform (HMSVGST) based on variational mode decomposition and synchrosqueezing GST (SGST). The proposed algorithm differs from the original synchrosqueezing algorithm in that it decomposes and transforms the signal simultaneously, which preserves the original signal components and avoids interference between different signal components, thereby improving the time–frequency focusing ability. A high-order multichannel synchrosqueezing variational modal GST is employed to decompose the seismic data volume in the time–frequency domain, and the optimal surface voting technique is used to characterize small faults. We set the forward model with 5–30-m fault distance and the application of real seismic data; we show that the proposed method has a good ability to characterize small faults less than 10 m, which validated the proposed method.

To increase the accuracy and vertical resolution of seismic inversion for exploratory purposes, a new method was developed for P-wave velocity, S-wave velocity and density inversion using prestack seismic data based on the mayfly optimization algorithm (MA), exact Zoeppritz equations and Bayesian framework. A new form of an accurate Hessian matrix was successfully derived. We innovatively used the MA nonlinear AVA inversion based on the accurate Hessian matrix (MANAI-Hessian) method for prestack seismic inversion and highlighted two main challenges for the first time. The popular and recent whale optimization algorithm (WOA) and a conventional Levenberg–Marquardt (LM) method were introduced to demonstrate the existence of these two challenges. Comprehensive partial derivative tests were well designed to verify the existence of the second-order partial derivatives of the P-wave reflection coefficients. A 3D special wedge model was introduced to test the accuracy and vertical resolution of the new method. Next, we applied the proposed method to the field data of deep carbonate rock from a study area in China. Compared with the conventional LM method and the accurate Jacobian matrix-based nonlinear AVA inversion method, which provides foundational approaches to address the two main challenges, the proposed approach shows superior performance in terms of accuracy and vertical resolution.

Seismic resolution is a key indicator in seismic exploration technology and directly affects the reliability of seismic data interpretation. Seismic wavelet extraction plays a vital role in high-resolution seismic processing. Traditional seismic wavelet extraction methods, such as autocorrelation and homomorphic deconvolution, are limited to the minimum phase assumption. The wavelet extraction method that combines multitrace statistical wavelet extraction and logging data requires the reflection coefficients and wavelets of adjacent traces to be consistent. In view of these limitations, this study proposes a wavelet joint extraction method based on seismic velocity and acceleration signals. This method does not require assumptions regarding the phase of the wavelet, and an acceleration signal is introduced to avoid an excessive gap between adjacent traces. Independent component analysis (ICA) was used to reduce the influence of noise on the higher order cumulants. This method first performs ICA on the velocity and acceleration signals to obtain the preprocessed data and the corresponding high-order cumulants. The amplitude and phase spectra of the wavelet were then reconstructed using the amplitude and phase spectra of the high-order cumulant. Finally, a seismic wavelet was reconstructed based on the obtained amplitude and phase spectra. The theoretical model and actual seismic data were processed using the proposed method. The test results show that this method can extract a more accurate seismic wavelet and has feasibility and application potential for improving the resolution of seismic data.

As a common seismic facies visualization analysis method, self-organizing map (SOM) projects the waveform or seismic attribute vectors into a two-dimensional topological plane in a nonlinear way, which can effectively and efficiently discover the topological structure of a dataset. SOM does not need to set the number of classes in prior and has friendly visualization characteristics and excellent generalization, which are conducive to seismic facies interpretation using unlabeled data. However, due to the competitive learning used in SOM and the imbalance of data distribution in real world, the samples from majority classes are expanded on the topological plane and the minority classes are compressed. As a result, the plane cannot accurately describe the global structure of data distribution. To improve the visualization precision by modifying the topological relationship of the prototype vectors of SOM, we utilize uniform manifold approximation and projection (UMAP), a novel manifold learning technique for dimension reduction, to correct the prototype vectors generated by SOM. By combining the advantages of SOM and UMAP in the representation of data topological structure, the global relationship between seismic data samples can be properly established, and the internal relative spatial structure of majority class samples can be retained as much as possible, resulting in a more reliable classification. Meanwhile, the framework maintains the advantages of SOM in visualization. In the modeling tests and real data experiments, we have demonstrated the effectiveness and rationality of UMAP-SOM on the spatial structure representation of three-dimensional seismic data.

In recent years, significant advancements have been achieved in the exploration of oil and gas reserves within carbonate rock formations, particularly with respect to the considerable resources found in deep Ordovician fault-controlled karst fracture-cave reservoirs. Accurately identifying such reservoirs using effective geophysical methods is crucial, but it is often challenging due to low signal-to-noise ratio and strong background reflections shielding of raw seismic data. To fully extract the information of carbonate reservoirs contained in the seismic data and enhance interpretation accuracy, we innovatively employ variational Bayesian principal component analysis (VBPCA) technique to perform background modeling on the raw seismic data, aiming to effectively isolate the bead-like reflections of reservoirs from interfering signals. Subsequently, we conduct attribute analysis on the processed seismic data, and optimize the sweetness attribute to identify cave reservoirs. The identified reservoirs exhibit complete shapes with clear boundaries, providing an intuitive depiction of their locations. In comparison to traditional principal component analysis (PCA) and probabilistic principal component analysis (PPCA), VBPCA offers several advantages, including automatic determination of the number of principal components, eliminating the inconvenience of manual settings, more effective separation of reservoir reflections from interfering reflections, and greater robustness to noise. Testing on synthetic seismic records and actual data from an oilfield in northern China has validated the feasibility and effectiveness of the proposed approach for identifying carbonate karst cave reservoirs.

In multicomponent surveys, the travel path of seismic waves from the subface interfaces to geophones may not be vertical to the ground; consequently, vertical and horizontal geophones receive mixed wavefields comprising P- and S-waves. Vector separation of wavefields is crucial for subsequent imaging and shear wave splitting analyses. However, conventional methods cannot address the nonlinear polarization of pure wave and can also produce incorrect results due to the presence of dipping interface. To address the above issues, we propose a vector wavefield separation method based on the slope of events and polarizability. The proposed method calculates the slope of the events for determining the propagation direction of the wavefield. The polarizability is used to evaluate if the resulting wavefield is purely composed of one type of body wave while the wavefield type is determined based on the angle between the polarization and propagation directions. When the wavefield is mixed, P/S separation is realized based on vector decomposition in an affine coordinate system optimized by propagation direction. We use both numerical and field data to demonstrate the effectiveness and applicability of the proposed method. The numerical example confirms that the proposed method yields more accurate results than the conventional vector decomposition method. The field data application also indicates that the new method can effectively resolve the P- and S-waves from the noisy seismic wavefield.

With the dramatic growth and complexity of seismic data, manual seismic facies analysis has become a significant challenge. Machine learning and deep learning (DL) models have been widely adopted to assist geophysical interpretations in recent years. Although acceptable results can be obtained, the uninterpretable nature of DL (which also has a nickname “alchemy”) does not improve the geological or geophysical understandings on the relationships between the observations and background sciences. This article proposes a noble interpretable DL model based on 3-D (spatial-spectral) attention maps of seismic facies features. Besides regular data-augmentation techniques, the high-resolution spectral analysis technique is employed to generate multispectral seismic inputs. We propose a trainable soft attention mechanism-based deep dilated convolutional neural network (ADDCNN) to improve the automatic seismic facies analysis. Furthermore, the dilated convolution operation in the ADDCNN generates accurate and high-resolution results in an efficient way. With the attention mechanism, not only the facies-segmentation accuracy is improved but also the subtle relations between the geological depositions and the seismic spectral responses are revealed by the spatial-spectral attention maps. Experiments are conducted, where all major metrics, such as classification accuracy, computational efficiency, and optimization performance, are improved while the model complexity is reduced.

Sparse representation (SR)-based denoising method attenuates random seismic noise trace-by-trace, which may not be able to employ the coherency between neighboring traces. To address this, we propose a multiple measurement vector (MMV)-based algorithm for robust denoising of multichannel seismic gathers. Seismic reflectors in common offset gather (COG) is characterized by horizontal events, which satisfy the common sparsity assumption, making it ideal for the MMV approach. ℓ2,1-norm regularization, which enforces two constraints on source matrix, i.e., temporal sparsity and the horizontal continuity, is then adopted in the MMV model to stabilize the lateral variation between channels. Besides, ℓ2,1-norm regularization provides a reasonable intrinsic structure to reduce the multisolution of the algorithm. Celebrating the strengths of iteratively reweighted (IR), we present a novel MMV algorithm, IR ℓ2,1norm minimization (IR-ℓ2,1), that further improves the performance. The formulated IR-ℓ2,1can be minimized by the alternating direction method of multipliers. Both synthetic and field data applications confirm the effectiveness of the proposed denoising method.