To address irregular missing seismic data caused by environmental interference during field acquisition, this paper proposes a local learning-based reconstruction method that integrates particle swarm optimization and grey wolf optimizer (HPSOGWO) algorithm with the XGBoost model. The proposed method establishes a nonlinear mapping relationship between seismic trace spatial coordinates (trace number and sampling number) and amplitude values. By adaptively optimizing feature windows using the HPSOGWO algorithm, this method achieves intelligent selection of adjacent trace data and high-precision prediction of missing values. Compared with the traditional convex-set projection method based on the Curvelet transform (Curvelet-POCS), the proposed approach significantly improves reconstruction accuracy in complex structural areas. In contrast to deep learning methods such as U-Net, it reduces the reliance on large training datasets and lowers computational costs. Tests on a three-layer horizontal layered model with 20% random missing traces show that the proposed method achieves a peak signal-to-noise ratio (PSNR) improvement of 11 dB over Curvelet-POCS and 7 dB over U-Net. F-K spectrum analysis further confirms its effectiveness in preserving seismic wavefield characteristics in the frequency domain. Tests on real onshore 2D seismic data show that the reconstructed profile with 20% missing traces achieves a relative amplitude error of 5.72%, demonstrating high amplitude fidelity and phase consistency. The method thus provides an effective and practical solution for seismic data reconstruction under complex geological conditions.

To address the problem of limited generalization ability in traditional models for logging prediction of total organic carbon (TOC) content in the Longmaxi Formation shale reservoirs of the Sichuan Basin, this study systematically evaluates the performance of various deep learning models in both single-well and cross-well prediction tasks and identifies practical models suitable for different scenarios. Based on the acoustic time difference, density, and natural gamma-ray logging data from three wells (well1, well2, and well3) in the Longmaxi Formation, multiple models are constructed, including multiple linear regression (MLR), support vector regression (SVR), convolutional neural network (CNN), bidirectional long short-term memory network (BiLSTM), and hybrid CNN-BiLSTM and CNN-BiLSTM-Attention models. In the single-well prediction experiment, a random split strategy is applied to well1 for modeling and validation. The results show that the CNN model achieves the best performance, with the coefficient of determination (R²) reaching 0.9519 on the prediction set, demonstrating excellent local feature extraction capability and strong resistance to overfitting. To further evaluate model generalization ability, a leave-one-well-out (LOWO) cross-validation strategy is designed for cross-well prediction. The results indicate that the CNN-BiLSTM-Attention model exhibits the strongest generalization performance, achieving the highest R² of 0.9653 on the prediction set, with mean absolute error (MAE) and root mean square error (RMSE) as low as 0.131% and 0.170%, respectively, which significantly outperforms other models. The attention mechanism effectively integrates the local features extracted by CNN with the long-term sequential dependencies captured by BiLSTM, enhancing the model's ability to focus on key information and adapt to inter-well variations. This study verifies the effectiveness and robustness of deep learning models integrated with an attention mechanism for TOC prediction under complex geological conditions, emphasizes the importance of cross-well validation in practical applications, and provides a reliable methodological foundation for shale gas sweet-spot prediction.

Data-driven intelligent first-arrival picking methods primarily rely on supervised learning using labeled data. However, a significant disparity exists between the representational capacities of mainstream local-focus labeling methods and global-segmentation labeling methods. The former emphasizes the learning of local waveform details of first arrivals, while the latter focuses on capturing global structural features of the wavefield, resulting in constrained learning perspectives for traditional single-task models. Consequently, this paper proposes a Multi-task collaborative first-arrival picking framework based on a symmetric dual-decoders structure (MT-SDD picking). First, by leveraging the complementary representational strengths of these two labeling strategies, the MT-SDD framework employs a dual-path decoder with a symmetrical configuration and incorporates a feature fusion module that integrates Transformer blocks with a multi-scale convolutional pyramid.Second, through a multi-stage optimization strategy, the framework progressively directs the model in transitioning from learning single-perspective features to multi-perspective feature fusion, ultimately achieving a substantial advancement in both picking accuracy and stability. Finally, comprehensive ablation experiments substantiate the rationale and efficacy of the MT-SDD framework, demonstrating its superior performance in high-precision picking and cross-site generalization capabilities. Validation results based on active seismic data from the Halfmile and Brunswick mining areas in Canada indicate that, in comparison to traditional single-task picking models, the proposed method reduces the average first-arrival picking error by 20%. Accuracy rates across various error tolerances have been consistently enhanced: at error thresholds of 2 ms, 4 ms, 8 ms, and 10 ms, the accuracy achieved 95.6%, 97.8%, 98.9%, and 99.2%, respectively. These figures represent improvements of 0.7%, 0.3%, 0.1%, and 0.1% over the single-task baseline method.

As a crucial approach for constructing subsurface velocity models, traveltime tomography derives the velocity distribution of the subsurface medium by solving an inverse problem constrained by observed traveltime data. However, in physics-informed neural network (PINN) traveltime tomography, the weighting between the data misfit term and the physical constraint term often relies on manual experience, making adaptive balancing difficult. This can lead to slow convergence or even cause the network to become trapped in local optima. Therefore, a neural tangent kernel (NTK)based adaptive weight optimization method for PINN traveltime tomography is proposed. First, to address the challenge of optimizing multi-objective loss functions, a dynamic weight adjustment mechanism is constructed based on the NTK theory. Second, the trace of the NTK is used to characterize gradient flow and to adaptively balance the contributions from the data and physical constraint terms. Finally, this mechanism optimizes the training process by addressing gradient imbalance, accelerating convergence of the PINN, and enhancing inversion stability. Numerical experiments and real-data applications demonstrate that the proposed method improves training stability for complex velocity models and yields superior inversion results compared to traditional fixed-weight PINNs.

Rock fabric contains abundant geological information, and lithology identification from rock thin sections is of great importance for oil and gas exploration, mineral extraction, and related fields. To address issues such as low accuracy and high labor costs in rock image classification, this paper proposes a ConvNext V2-VMamba-based method for lithology identification in rock thin sections. First, based on the ConvNext V2 model, grouped convolution and channel shuffle strategies are adopted to combine the ConvNext module with the VSS module. Then, a Conv-VMamba module is proposed to replace the original ConvNext module, enabling the model to possess both a global receptive field and excellent local feature extraction capabilities. Finally, a spatial attention EMA module is integrated into the model to enhance cross-channel spatial information aggregation and improve the capture of texture information in different rock images. Experimental results show that the model achieves an accuracy of 81.5%, precision of 81.1%, recall of 81.4%, specificity of 96.3%, and F1-score of 81.2% on the test set. Compared with the original model, the accuracy is improved by 5%. This method demonstrates the highest algorithmic accuracy and best classification performance, providing a new approach and method for the field of lithology identification.

Accurate prediction of geological sweet spots is a core component for the efficient exploration and development of shale oil and gas. Intelligent prediction of sweet spots of shale oil and gas based on logging data holds significant importance for hydrocarbon development. Within the field of sweet spot prediction, traditional model-based prediction methods suffer from limitations such as weak generalization, insufficient feature representation, and cumulative error propagation. This study integrates novel deep learning frameworks with the aim of achieving accurate prediction of geological sweet spots using logging data. Porosity (POR) and total organic carbon (TOC) are key parameters defining geological sweet spots. This study establishes deep learning models between logging curves and these two critical parameters to achieve intelligent prediction of geological sweet spot parameters. Firstly, a convolutional neural network (CNN) model is constructed. By analyzing and comparing different combinations of input logging curves and varying CNN layer architectures, the model achieves high-precision predictions for both POR and TOC on the test set, with correlation coefficients exceeding 0.96. Secondly, innovatively fusing CNN with an attention mechanism, a composite Transformer architecture termed the CNN-Transformer compound (CTNet) model is proposed. This model possesses the dual capability of capturing local lithological features and global inter-layer dependencies. After validation and testing on the dataset, the CTNet model achieves a correlation coefficient above 0.91 on the test set. Experimental results indicate that the CNN model demonstrates higher overall prediction accuracy than the CTNet model. However, the CTNet model exhibits significantly superior prediction accuracy within limestone intervals compared to the CNN model (e.g., in the limestone section of well GY6-1, the correlation coefficient of the CTNet model and the CNN model is 0.848 and 0.300, respectively), showcasing its unique potential for addressing prediction challenges in complex heterogeneous reservoirs. The findings demonstrate that both the proposed CNN and CTNet deep learning mo-dels can effectively achieve high-precision prediction of key geological sweet spot parameters in shale oil and gas, offering new methodologies for intelligent exploration.

The complex geological characteristics of shale reservoirs lead to stronger non-homogeneity in the corresponding regions, resulting in difficulties in capturing the internal sequence features implicit in logging parameters and porosity through deep learning. In response to the gradient problem inherent in long short-term memory (LSTM), gated recurrent unit (GRU), and other models, which leads to certain limitations in the prediction accuracy of the model in the task of predicting shale porosity in long sequences, the paper proposes a shale porosity prediction model based on Informer. The probabilistic sparse attention mechanism and the "distillation" operation are introduced into the model to efficiently extract the sequence features implicit in the long-sequence logging data and realize the accurate prediction of porosity. The sensitive logging parameters are obtained with the Pearson's correlation coefficient analysis method, and the sample set is constructed. The Informer model is set up with the predicted sequence lengths of 20, 40, 80, and 120 and compared with LSTM and GRU. The model is applied to the logging parameters of an actual working area in a shale reservoir in Songliao Basin for the test of the application effect. The experimental results show that with the increase of the predicted sequence length, the mean squared error (MSE) of the Informer model increases by 18.5%, 23.8%, and 10.7%, respectively, which has stronger model stability than LSTM and GRU. Through comparing the prediction results of each model in the actual working area when the predicted sequence lengths are 80 and 120, it is found that the R² and MSE of the Informer model reach the optima, which are 0.834, 0.179 and 0.789, 0.204, respectively, and the prediction accuracy of the Informer model is significantly improved compared with that of the other models.

In seismic exploration, it is generally believed that a higher fold density means a greater number of seismic wave samplings of underground media, and theoretically, the signal-to-noise ratio and resolution of seismic imaging will be higher. However, in actual production, the disconnect between the setting of acquisition parameters and imaging quality indicators has become an increasingly prominent issue. The final imaging results of many seismic acquisition data with high fold density are unsatisfactory. Excessive pursuit of high fold density not only substantially increases the cost of seismic acquisition but also leads to the gradual neglect of basic parameters such as excitation energy and receiving sensitivity. Therefore, an innovative method for estimating the efficiency of seismic acquisition parameters is proposed. First, by constructing geophysical models and conducting quantitative calculations, the actual contribution of the observation system to imaging points and the contribution of excitation and receiving links to effective signal energy are clarified in sequence. Second, a multi-dimensional evaluation system focusing on fold density, excitation parameters and receiving parameters is established, and a comprehensive evaluation scheme for parameter efficiency covering the entire acquisition process is formulated. Finally, a proper balance is struck among the selection of acquisition parameters, the control of acquisition costs and the quality of seismic imaging, so as to achieve the dual improvement of the efficiency and quality of seismic acquisition. The results of case comparative analysis show that the method proposed in this paper breaks through the previous one-sided approach of evaluating imaging effects solely by relying on fold density, and provides technical support for the high-quality and high-efficiency precise exploration of the "wide-azimuth, wide-band and high-fidelity" seismic acquisition technology.

In ocean-bottom node (OBN) seismic exploration, natural factors such as ocean currents and tides, as well as human factors such as fishing vessel dragging, often interfere with node positioning results, causing the actual node positions to deviate from the initial positioning results and necessitating high-precision secondary positioning. Therefore, a secondary positioning method based on quadrant division and cross-correlation analysis is proposed. First, based on an in-depth analysis of quadrant positioning principles, the shot points around each node are divided into four quadrants, and linear normal moveout (LNMO) correction is applied to maximize coherent wavefield stacking within each quadrant, thus improving the signal-to-noise ratio (SNR) and reducing waveform distortion. Then, cross-correlation analysis is performed on the stacked results of opposite quadrants to obtain more accurate first-arrival time differences between quadrants, enabling high-precision node positioning. Case study results show that, compared with conventional secondary positioning methods, the proposed method improves the SNR through quadrant division and stacking of common receiver gathers within each quadrant, and applies cross-correlation of opposite quadrants to calculate first-arrival time differences, significantly enhancing both positioning efficiency and accuracy. The method is applicable to secondary positioning at different water depths and has broad application prospects.

Denoising is a critical step in seismic data processing. Traditional dictionary learning methods lack constraints on the physical mechanism of signals, easily misidentifying coherent noise with linear or curvilinear features as effective signals and incorporating them into dictionary atoms, which results in distortion of effective signals or residual noise during reconstruction. To address these problems, this paper proposes a hybrid noise suppression method based on sparse basis learning with physical constraints. Firstly, the plane-wave destruction (PWD) filter is used to estimate and update the local dip field, ensuring that the dip field matches the iteratively optimized signal. Then, the local dip field estimated by PWD is taken as a constraint to guide the sparse basis learning process, which combines the signal representation ability of sparse transform with the local dip characteristics of effective signals. This enables dictionary atoms to match only effective signals with corresponding dip characteristics, avoiding the mislearning of coherent noise atoms. Finally, effective signals can be accurately sparsely represented on the sparse basis matching their dip characteristics, while various noises are difficult to be matched by the sparse basis due to the absence of corresponding dip characteristics. Thereby, high-fidelity effective signal reconstruction and effective suppression of multi-type noises are achieved. Tests on both complex synthetic data and field data show that the proposed method has superior suppression performance and wide applicability in suppressing hybrid noises such as strong abnormal amplitude interference, surface waves, random noise, and coherent noise, compared with conventional methods and industrial standard denoising workflows.

As oil and gas exploration targets shift toward deep and ultra-deep formations, enhancing weak seismic reflection signals has emerged as a critical technical bottleneck restricting exploration breakthroughs. Due to the inherent limitations of traditional methods, weak reflection signals are highly susceptible to being misclassified as noise and thus suppressed during processing, which ultimately leads to their disappearance and thereby severely undermines the accuracy and reliability of reservoir prediction. To address this issue, this paper proposes a weak signal enhancement method based on adaptive threshold curvelet transform and compressed sensing. Adopting a joint processing strategy that combines curvelet transform with compressed sensing, this method improves upon existing static threshold-setting mechanisms by developing an adaptive threshold dynamic control approach rooted in the statistical characteristics of curvelet coefficient sensitivity and probabilistic modeling. This approach enables dynamic threshold adjustments across multiscale subbands, thereby enhancing the sensitivity to weak reflection signals and reducing misjudgment between signals and noise. The results from both synthetic data tests and real data applications indicate that this method can preserve signal details and enhance signal energy while effectively suppressing random noise. Compared with existing methods, it demonstrates a more pronounced advantage in terms of signal fidelity.

Marchenko imaging can effectively suppress migration artifacts associated with internal multiples while preserving amplitude fidelity. However, most existing Marchenko imaging studies are conducted under the assumption of isotropic media. Neglecting the anisotropic characteristics of real media can lead to inaccurate phase estimation of the initial focusing function, thereby reducing the accuracy of the reconstructed Green's functions and the final imaging results. To address this issue, this paper proposes a Marchenko imaging method tailored for tilted transversely isotropic (TTI) media. First, the initial focusing function from a subsurface imaging point to the surface is estimated using the pseudo-acoustic first-order velocity-stress equations in TTI media. The preprocessed shot records, together with the estimated initial focusing function, are then used as input to reconstruct Green's functions by solving Marchenko equation. Next, the reconstructed up- and down-going Green's functions are incorporated into an energy-compensated cross-correlation imaging condition to generate the imaging results. This procedure is repeated to achieve Green's function reconstruction and imaging for all imaging points within the subsurface target region. Synthetic model experiments demonstrate that the proposed method enables accurate reconstruction of Green's function in TTI media and produces clear imaging results with a high signal-to-noise ratio.

Currently, complex and concealed hydrocarbon reservoirs characterized by fragmentation, great depth, thinness, and subtlety serve as the primary targets for reserve growth in the Jiyang Depression, eastern Shengli Oilfield. Seismic exploration technologies, featuring small-bin, wide-azimuth, high-coverage, and high-density data acquisition as well as offset vector tile (OVT)-based wide-azimuth and wide-bandwidth data processing, have undergone rapid advancement. There is an urgent demand to develop OVT-domain data interpretation techniques tailored to high-density and wide-azimuth seismic data. Accordingly, taking the WJG area in the Dongying Sag of the Jiyang Depression as a case study, this research conducts well-seismic optimized processing on OVT seismic gathers. The work is grounded in seismic gather forward modeling using actual logging data, with multi-domain constraints implemented via amplitude-frequency-amplitude variation with offset (AVO) characteristics. Optimization of offsets and azimuth division for OVT seismic data are achieved through strategies including in-phase stacking of offsets and equal division of azimuth coverage counts. For wide-azimuth and large-offset seismic data, technical workflows are developed, encompassing prediction of azimuthal amplitude difference facies belts, description of offset-related amplitude difference thicknesses, and fluid prediction based on azimuthal AVO gradient differences. Ultimately, an integrated technical series for stepwise recursive reservoir prediction using difference attributes of Es₄ is established. The research findings indicate a departure from the previous understanding that extensive shore-shallow lake beach-bar reservoirs dominate this area. Instead, the study area is found to develop estuary sandbars and lake-wave modified deltaic beach-bar reservoirs. This technology effectively drives the iterative upgrading of interpretation techniques and provides reliable technical support for fine exploration in mature exploration areas and prediction of geological targets.

In the desert hinterland of the Junggar Basin, the seismic data fidelity and resolution are generally low, leading to unclear understanding of the genesis and distribution patterns of volcanic reservoirs, and difficulties in reservoir prediction and detailed characterization of hydrocarbon accumulations. To accurately delineate the complex internal characteristics of Carboniferous volcanic reservoirs and enhance the resolution of target horizons in seismic data, China's first onshore fiber-optic 3D-VSP seismic pilot test was conducted in the hinterland of the Junggar Basin. Techniques for acquisition, processing, and interpretation tailored for 3D-VSP data were proposed. First, a VSP data acquisition observation system was designed based on geological requirements and the characteristics of fiber-optic equipment, and high-quality VSP seismic data were obtained using efficient vibroseis sources and distributed fiber-optic receivers. Subsequently, by analyzing the characteristics of fiber-optic data and tackling key processing techniques, reliable borehole seismic data processing results were achieved. Finally, through high-resolution VSP imaging, a joint comparison and interpretation of surface and borehole seismic data were conducted to analyze the spatial characteristics of the target reservoir. Methods for describing and characterizing deep Carboniferous volcanic reservoirs were discussed, thereby improving reservoir prediction accuracy and reducing the multiplicity of interpretations in seismic data analysis. Application results demonstrate that this method provides a more reliable data foundation for the detailed characterization of Carboniferous hydrocarbon reservoirs in the Junggar Basin, offering significant support and implications for efficient oilfield exploration and development.

Quantitative seismic prediction of hydrocarbon saturation remains a key technological bottleneck in exploration geophysics and has yet to achieve major breakthroughs. Previous studies continuously investigated seismic prediction methods for hydrocarbon saturation and developed various qualitative or quantitative approaches. However, practical applications indicate that the prediction accuracy at the current stage still fails to meet the urgent requirements for detailed reservoir characterization. To improve the reliability and accuracy of quantitative gas saturation prediction, this paper integrates well logging, petrophysics, inversion, and artificial intelligence and proposes a quantitative gas saturation prediction method based on three-dimensional phase-controlled pre-stack geostatistical inversion, artificial intelligence, and the Archie equation. First, high-precision pre-stack elastic parameters, including P-wave impedance, S-wave impedance, and the P-wave to S-wave velocity ratio, are obtained using three-dimensional phase-controlled pre-stack geostatistical inversion constrained by reconstructed lithological data, and sandstone porosity is predicted through petrophysical analysis. Next, artificial intelligence techniques are employed to establish a nonlinear mapping model from logging elastic parameters to undisturbed formation resistivity, enabling high-precision prediction of undisturbed formation resistivity using corresponding inverted elastic parameters. Finally, the predicted porosity and undisturbed formation resistivity are substituted into the Archie equation to calculate reservoir gas saturation. Application results demonstrate that this method enables accurate quantitative prediction of reservoir gas saturation. The reliability and accuracy of the prediction results are verified through practical applications, with a well-seismic correlation rate exceeding 90%. This method provides effective technical support for efficient exploration and detailed development of tight gas reservoirs.

The Yingmai 2 Area of the Tarim Basin, influenced by the superposition of multiple tectonic movements (e.g., Caledonian, Hercynian, and Indosinian), develops a complex fault network system. The interaction relationships and connectivity of these faults exert a key controlling effect on hydrocarbon migration and accumulation. Conventional geological analysis methods struggle to quantitatively characterize the interaction relationships between faults and multi-stage superposition effects, which limits in-depth understanding of fault connectivity. This study introduces topological theory, abstracting the fault network into a "node-branch" graph theory model. Based on seismic data, a fault topological model is constructed. A basic topological analysis is conducted by calculating geometric parameters such as strike, length, and line density, combined with node classification and branch types. Furthermore, advanced topological parameters including areal density, average branch connectivity, and normalized connectivity index C are used to quantitatively evaluate the connectivity of the fault network. The results reveal the combinatorial characteristics and developmental patterns of multi-stage fault systems in carbonate rocks within the study area. This research provides new technical methods and a theoretical basis for the quantitative characterization of fault systems in complex fault-block areas and the optimization of favorable hydrocarbon exploration targets.

When oil-based mud invades into a formation, the resistivity distribution and response characteristics of the invaded zone are complex and cannot be represented by a simple radial two-layer step model. Considering the actual structure and dimensions of the array induction logging tool, this study derives the expressions for the induced electromotive force and effective tool constant of the three-coil measurement structure with a built-in copper tube. The outer diameter of the copper tube is adjusted depending on the derived expression of the effective tool constant and the effective tool constant of the actual tool is simulated to approximate the radial detection characteristics of the actual tool. Subsequently, a forward model for the radial layered medium model of array induction logging is constructed. Based on the analysis of invasion response characteristics under oil-based mud conditions in the fractured tight sandstone formation of the Cretaceous Bashijiqike Formation in the Dabei and Keshen areas of the Kuqa Depression in the Tarim Basin, the radial step invasion model is selected for forward modeling of the complex invasion response characteristics of array induction logging in oil-based mud. The study shows that the radial three-layer step model can reasonably represent the complex response characteristics of formation invaded by oil-based mud, laying a foundation for the subsequent inversion application of array induction logging data.

The salt lake carbonate rocks have the characteristics of strong heterogeneity and high formation water salinity, which easily leads to insignificant resistivity differences between oil and water layers, thereby causing difficulties in identifying reservoir fluids. In the traditional array induction logging data processing process, the electrical dispersion characteristics of the multi-frequency raw data are often eliminated during the skin effect correction, and thus are not effectively utilized. Therefore, starting from the microscopic mechanism of the interface polarization effect, this study reconstructs the real and imaginary parts of the conductivity spectrum based on the multi-frequency raw conductivity curves of the deep detection sub-array of array induction logging, achieving qualitative discrimination of the fluid properties of carbonate reservoirs in the study area. Meanwhile, the conductivity dispersion degree calculation formula is introduced, and the electrical dispersion characteristics of the insitu formation fluids are quantitatively evaluated based on the multi-frequency raw conductivity of the deep detection sub-array of array induction logging. A new set of fluid identification charts is constructed by combining the resistivity values after wellbore correction and porosity normalization. The results show that this method and model have good applicability in the identification of fluids in salt lake carbonate reservoirs, effectively compensating for the deficiency of traditional single reliance on the absolute value of resistivity for fluid discrimination, and have good application prospects.

Traditional wired hydrostatic level gauges suffer from issues such as complex probe cable structures, susceptibility to damage, high cost, and short service life. Conventional acoustic rangefinders, on the other hand, have limited measurement ranges (less than 20 m). Both are inadequate for dynamic liquid-level monitoring in deep wells. To address these issues, this paper proposes a rapid well liquid-level measurement method based on the waveguide effect of acoustic waves in confined wellbores. This method involves establishing a sound velocity correction model to correct acoustic velocity distortion, designing a dual-probe isolated receiving device to suppress noise interference and improve timing accuracy, employing an up-chirp linear frequency-modulated signal to enhance excitation energy, and developing a segmented time-delay accumulation positioning algorithm to overcome errors in variable-diameter wells. As a result, this method enables rapid and high-precision liquid-level measurement at the wellhead. Feasibility tests conducted at Honghe-1 Well demonstrate that this method is feasible and effective, providing a non-contact, low-cost, and universally applicable solution for safety monitoring in oil and water wells under various well conditions.

The Triassic reservoir in the Lunnan Oilfield is characterized by widespread low-resistivity oil reservoirs. However, the genesis of these reservoirs remains poorly understood, and the accuracy of reservoir fluid identification is still relatively low, posing a major challenge to efficient exploration and development in the area. Based on an integrated dataset comprising rock-physics experiments, formation-water salinity data, well logs, and well-testing results, this paper adopts an elimination-by-factors strategy to systematically investigate the dominant mechanisms controlling low-resistivity responses, with particular focus on reservoir properties and wettability, formation-water salinity, clay-related additional conductivity, and drilling-mud invasion effects. On this basis, the resistivity invasion factor Q is proposed as a preliminary indicator for distinguishing low-resistivity oil reservoirs from normal-resistivity oil reservoirs. Furthermore, an ADASYN-LightGBM optimization scheme is introduced, in which the resistivity invasion factor Q is jointly incorporated into a machine-learning framework to enhance the discrimination of dry layers.The results indicate that the low-resistivity phenomenon in the study area is primarily associated with formation pressure depletion caused by asynchronous development scheduling, as well as deep mud invasion resulting from untimely drilling-fluid adjustment. The proposed model achieves an average F1-score of 94.38% and an average accuracy of 94.41%, with anidentification accuracy of 96.77% for low-resistivity oil reservoirs, effectively improving the identification performance between low-resistivity oil reservoirs and dry layers. These findings suggest that Q can serve as a rapid screening parameter for low-resistivity oil reservoirs. However, its effectiveness is limited when used as a single parameter for dry-layer discrimination. By integrating ADASYN-LightGBM, the classification capability for complex reservoir types can be substantially improved, providing a reference for identifying Triassic low-resistivity oil reservoirs in the Lunnan Oilfield and for applications in analogous fields.

Ultra-deepwater and ultra-shallow unconsolidated gas reservoirs, which have good exploration and development prospects, are widely distributed in the Qiongdongnan Basin. However, weak cementation and gas-bearing properties make it highly challenging to accurately determine the formation compressional-wave slowness. Therefore, a method for extracting the compressional-wave slowness of ultra-shallow unconsolidated gas reservoirs using low-frequency monopole acoustic array logging data is proposed. First, the validity of the reservoir velocity is verified by comparing core experimental data with numerical simulation results from an effective medium model. Second, based on borehole acoustic field theory in porous media, array acoustic waveforms excited by sources with different frequencies are simulated. The analysis shows that low-frequency leaky P-waves are more suitable for measuring formation compressional-wave velocity. Based on this understanding, acoustic logging data are processed to obtain formation compressional-wave velocity. Finally, the accuracy of the low-frequency compressional-wave velocity is verified by comparing VSP and acoustic logging velocities, as well as calibrating synthetic seismograms with actual seismic profiles. The results show that low-frequency compressional waves can accurately determine the compressional-wave velocity of ultra-deepwater ultra-shallow unconsolidated sandstone gas reservoirs, providing reliable support for subsequent reservoir horizon calibration and characterization.

The demand for refined and new exploration of hydrocarbon reservoirs necessitates the development of novel geophysical technologies to monitor remaining oil distribution and dynamic production processes. Surface-to-Borehole Electromagnetic (SBEM) method employs a three-dimensional observation by deploying power sources on the surface and sensitivity probes near the target layer within the well, which possesses the advantages of strong anti-interference capabilities and high resolution, making it potentially effective for remaining oil exploration and monitoring extraction. However, SBEM observations are constrained by the limited measurement data available from the receiving well, the inversion imaging represents a high-dimensional, nonlinear, and ill-posed inverse scattering problem, posing computational challenges and theoretical completeness. Therefore, efficient 3D SBEM inversion imaging with multi-source and multi-frequency surface excitations alongside observations in favorable borehole areas—has become a current research hotspot. We utilize the Incomplete Fast Fourier Transform (IFFT) algorithm to accelerate the updated calculation of the dyadic Green's function. By employing the Distorted Born Iterative Method (DBIM), both the Green's function and the contrast function are updated iteratively, ensuring stable convergence while preventing computational costs. To validate the proposed method, we constructed homogeneous half-space and layered models based on typical volcanic reservoir characteristics to conduct synthetic data inversions. The spatial distribution patterns of the anomalies accurately match the true models, aligning well with SBEM inherent sensitivity to vertical interfaces and the immediate borehole vicinity. The results verify the algorithm's feasibility and effectiveness, providing a theoretical basis for the practical application of SBEM in hydrocarbon reservoir exploration.

The low resolution of seismic signals is one of the main problems currently facing seismic exploration, mainly manifesting as insufficient bandwidth and signal distortion. This paper focuses on micro-distortion in wideband signals captured by geophones. By analyzing the operating principles and equivalent circuits of existing geophones, this paper reveals the fundamental reason why they cannot achieve wideband signal reception with minimal distortion. On this basis, this paper proposes a circuit balance theory. This theory no longer treats the formation of a traditional circuit loop as the criterion for design. Instead, it evaluates the rationality of a circuit design based on whether the energy or other parameters in the circuit reach a state of balance. Guided by this theory, this paper utilizes the piezoelectric effect of piezoceramics to redesign and develop a new type of piezoelectric geophone. Field experiments indicate that the effective seismic frequency in well shooting can reach 400 Hz, and that the side-lobe suppression effect of the Ricker wavelet for this geophone is superior to that of high-sensitivity geophones. To address the relatively low sensitivity revealed during experimental testing, technical improvements are implemented to optimize the structural design parameters, ultimately increasing the geophone sensitivity to 31.7 V·m^-1·s and meeting the requirements of practical production. This paper provides high-quality equipment support for wideband micro-distortion seismic exploration and offers new research directions for researchers in related fields.