The distribution of density plays a crucial role in interpreting subsurface geologic structures for resource exploration and deep-structure research. However, conventional gravity inversion often involves storing a dense, large sensitivity matrix, demanding significant computational resources. To enhance the memory efficiency of gravity inversion, we introduce an intelligent data-sensing-based inversion to minimize the data involved in the inversion process. Initially, we validate the effectiveness of our approach using two synthetic data examples. Our intelligent data-sensing approach uses a loss-controlled compression technique to sense the optimal subset from the original gravity anomaly. A user-specified threshold determines the degree of the information loss. Subsequently, we invert the optimal subset, but not all is used in each iteration. Instead, we implement an inverting-testing strategy during the iteration, when part of the optimal subset serves as the inverting data set and another part serves as the testing data set. The amounts of inverting and testing data sets are equal in each iteration but may vary across iterations based on the inversion process. We conduct statistical analyses to assess the impact of the intelligent data-sensing approach on memory requirement during gravity inversion. Our method is applied to two field data sets, demonstrating that it requires only a small percentage of the total gravity anomaly to reconstruct density distribution nearly identical to those obtained using the complete optimal subset.

Accurate evaluation of permeability in subsurface media is essential for solving problems in various fields, such as environmental and petroleum engineering. This study aims to numerically investigate the pore characteristics of spherical particle beds imitating compacted sandstones for the physical understanding of the permeability change with porosity. In this study, the structures of monodisperse and polydisperse particle beds are simulated and the pore regions are extracted using various numerical methods. The pore characteristics, such as effective porosity, tortuosity, and specific surface area in the bed, are quantified. The permeability is then calculated by substituting them into the Kozeny-Carman equation. The obtained permeability is compared with measurement results from actual sandstones. The respective pore characteristics are also evaluated against findings from previous studies. The role of pore characteristics in influencing permeability is investigated to examine changes in permeability in compacted sandstones. Consequently, the respective pore characteristics are modeled as functions of porosity. The effective porosity and tortuosity obtained from the present analysis matches with experimental results and the physical models developed in previous studies. The specific surface area also agrees with previous experimental results. However, the model developed in previous research does not match the experimental and present numerical results at low porosity. Therefore, we develop a new geometric model for specific surface area under low porosity conditions. Based on the numerical results, we assess the applicability of the original Kozeny-Carman equation to actual sandstones composed of deformed particles and analyze its physical interpretation. The results suggest that the rapid increase in tortuosity and the rapid decrease in specific surface area with decreasing porosity cause the Kozeny-Carman equation to maintain its original form even in compacted sandstones.

Audio-frequency magnetotelluric (AMT) inversion technology plays a crucial role in the exploration of deep resources, especially in the detection of geothermal resources. Although AMT inversion technology has significant advantages in vertical resolution, the inversion process still faces serious challenges of nonuniqueness. To improve the accuracy and reliability of inversion, this paper develops an AMT data inversion method based on an adaptive binary structure constraint. This method incorporates adaptive extraction techniques to conduct a cluster analysis of the resistivity model generated in each iteration, thereby identifying and distinguishing the target and background areas in the model. These target areas are then used as prior structural information for the next iteration, guiding the refined reconstruction of the resistivity model. During the iteration process, the AMT inversion algorithm focuses on calculating the target areas while the resistivity values of the background areas remain unchanged. This strategy effectively reduces the degrees of freedom in the inversion problem and lowers the dimension of the solution, thereby improving the computational efficiency and accuracy of the inversion. Model tests show that compared with the traditional smoothing-constrained AMT (S-AMT) inversion method, the adaptive binary structure-constrained AMT inversion method (ABS-AMT) can reconstruct high-resolution refined resistivity models, significantly reducing the uncertainty of inversion. Furthermore, the ABS-AMT method not only reduces the dependence on prior knowledge but also enhances the algorithm’s adaptability to different geologic conditions. Practical application in the geothermal area of Laosandui in Linjiang City, Jilin Province, China, fully demonstrates the method’s effectiveness in known geothermal areas.

In recent years, half-precision floating-point arithmetic has gained wide support in hardware and software stack thanks to the advance of artificial intelligence and machine-learning applications. Operating at half precision can significantly reduce the memory footprint compared to operating at single or double precision. For memory-bound applications such as time-domain wave simulations, this is an attractive feature. However, the narrower width of the half-precision data format can lead to degradation of the solution quality due to larger roundoff errors. We illustrate the negative impact caused by the accumulation of roundoff errors in wave simulations with carefully designed numerical experiments. Specifically, the energy-conserving property of the wave equations is used as a convenient diagnosis tool. The corresponding remedy in the form of compensated sum is then provided, with its efficacy demonstrated using numerical examples with acoustic and elastic wave equations on hardware that support half-precision arithmetic natively.

In both petroleum exploration and production, extracting petrophysical parameters of oil reservoirs from 3D seismic elastic data poses significant challenges. Applying machine learning (ML) algorithms to this issue promises much, but rigorous feasibility evaluation is crucial beforehand. More significantly, a comprehensive mathematical and logical justification is imperative to illuminate the ‘black box’ nature of ML, thereby bridging existing gaps in explanation for enhanced trust and insights. In this study, the input elastic data (compressional and shear wave velocities, along with density) for random forests originated from the Raymer-Dvorkin-Voigt rock physics model, with subsequent interventions introduced to emulate field application conditions. The modeled data, devoid of confounders, facilitates unambiguous causal interpretation. The algorithm demonstrated efficient and stable predictions for porosity, followed by clay volume, whereas water saturation predictions were less accurate. Subsequently, Shapley Additive exPlanations was employed to mathematically assess the relative importance of each input feature in determining the outcomes, thereby elucidating algorithm’s functionality. Furthermore, its causality was comprehended by referencing rock physics as the presumed truth. The results indicate irrational behavior in association-oriented ML, contradicting physics and logical principles. Overall, ML for petrophysical prediction is effective but constrained by several factors, including the problematic assumption of independent and identical distributed data, the unsatisfactory quality of data, and the complicated nature of the underlying physics. Our framework establishes a solid foundation for subsequent inquiries aimed at elucidating the explanation and causality behind ML-based petrophysical predictions.

The ocean-bottom cable (OBC) acquisition technique has become a significant geophysical method to simultaneously record geophone and hydrophone signals. The geophone signals have been used by many surface-wave dispersion inversion methods for seafloor imaging. However, geophones are not always perfectly coupled to the seabed, which can lead to low-quality or unavailable signals. In contrast, hydrophones are naturally coupled to seawater and record pressure disturbances induced by Scholte waves propagating on the ocean-bottom interface. We attempt to extract useful Scholte wave dispersion energy spectra from independent hydrophone or geophone signals of OBC systems by an extended frequency-Hankel transform method. Results show that this transform method can extract high-quality multimodal Scholte wave dispersion energy spectra from independent geophone or hydrophone data.

Diffractions generated by the wavefield response of small-scale discontinuous geologic bodies carry a wealth of information regarding complex geologic structures, which can be used for high-resolution imaging. However, the weak amplitudes of diffractions make them vulnerable to strong reflections. Thus, separating diffractions from strong reflections is crucial in the high-resolution imaging of underground structures. The low-rank (LR) characteristics of seismic wavefields enable LR methods to be used for diffraction extraction. However, traditional LR algorithms tend to disrupt the dynamic characteristics of diffraction during the separation process, which is detrimental to the subsequent high-precision imaging of diffractions. To address these shortcomings of traditional methods, we develop a weighted multichannel singular spectrum analysis (MSSA) algorithm designed to preserve the dynamic characteristics of diffractions. This algorithm fully considers the distribution characteristics of diffractions and reflections within the singular spectrum and redefines the singular values using a weighted approach, ultimately achieving amplitude-preserving separation of diffracted waves. Through a combined qualitative and quantitative analysis of synthetic and field examples, we demonstrate the effectiveness of this method in separating diffractions and suppressing reflections with robustness and high stability. Compared with the traditional MSSA method, our approach can provide better high-precision imaging results for small-scale geologic structures.

Coherent noise separation stands as a crucial step in seismic data processing. The morphological component analysis (MCA)-based separation method, which treats coherent noise and signal as distinct components and represents them sparsely with dictionaries, has been widely adopted for noise suppression. Typically, constructing effective fixed dictionaries for MCA-based methods involves necessary expert knowledge to meticulously select appropriate transform basis functions from an extensive dictionary library and fine-tune their parameters. To reduce time consumption and ensure optimal dictionary construction, we introduce an adaptive framework for identifying optimal dictionaries used in MCA-based coherent noise separation. Initially, we define a fixed dictionary library comprising dictionaries constructed using various transform basis functions with their corresponding parameters. Subsequently, we formulate a relative sparsity minimization problem (RSMP) to identify the optimal fixed dictionaries that minimize relative sparsity within this predefined library. Finally, we design a genetic algorithm to solve RSMP. The identified dictionaries are then applied to MCA-based coherent noise separation. Synthetic and field data examples demonstrate the effectiveness of our method.

In general, the traditional spatial autocorrelation (SPAC) method uses the fundamental-mode surface-wave dispersion curve for inversion. Ignoring the higher-mode surface-wave dispersion curves affects the accuracy of the SPAC method in detecting the S-wave velocity structures, especially for the inversion of deeper structures. To fully and accurately use the multimode nature of surface waves, we investigate a direct inversion method of subsurface S-wave velocity structures from the SPAC coefficients of multimode surface waves extracted from the seismic ambient noise. In this new method, the SPAC coefficient is directly inverted to obtain the S-wave velocity. The key step is to obtain the theoretical SPAC coefficients of a layered model by taking into account the energy ratio of each mode of surface waves in the studied frequency band and the influence of the finite station pairs and by weighting the Bessel functions corresponding to the phase velocities of each mode with the energy ratio as the weighting factor. The direct inversion method for the SPAC coefficient significantly improves the efficiency of data processing by omitting the extraction of the dispersion curves. Meanwhile, joint inversion of the fundamental and higher-mode SPAC coefficients of surface waves ensures the reliability of inversion results, increases the exploration depth, strengthens the stability of inversion, and significantly reduces the inversion misfit.

Microseismic events play a crucial role in mapping fault and fracture distributions in natural and induced earthquakes. Detecting and localizing microseismic events is challenging due to low signal-to-noise ratios. Waveform stacking imaging location is a practical approach for automatically detecting and localizing microseismic events, assuming that the traveltime-corrected seismic waveforms will stack and enhance coherently. However, coherent stack enhancement is susceptible to polarity reversal caused by the nonexplosive components of the source mechanism, which can lead to an unfocused source image, making it difficult to retrieve the optimal location accurately. In this study, we develop a new polarity correction method to address this issue based on the fact that the instantaneous phase difference between two seismic signals with opposite polarity is [Formula: see text]. First, the Hilbert transform is applied to the original seismic record to obtain the instantaneous amplitude [Formula: see text] and phase [Formula: see text]. Then, a new signal is constructed by multiplying [Formula: see text] with [Formula: see text]. The constructed signals for different receivers have the same polarity; therefore, they can be used to refocus the source image. Furthermore, they preserve positive and negative amplitudes, which contributes to noise suppression. Synthetic tests show that our method can effectively achieve polarity correction and noise suppression, enabling a high-resolution source image. Application to real hydraulic fracturing data demonstrates that our method can detect and locate more microseismic events at the fracturing depths, suggesting its effectiveness and potential advantage in microseismic data processing. Because the polarity correction is performed in the data domain without relying on specific receiver layouts, the method is computationally efficient and could be applied to real-time microseismic monitoring across various sites, such as hydraulic fracturing and volcano monitoring.