Geophysical Journal International

SummaryReciprocal measurements are standard for data quality control in electrical resistivity surveys, but their potential as an interpretation tool is rarely exploited. Under the assumption of linear subsurface behaviour, normal and reciprocal readings should yield the same results, but when a systematic offset is present it can be indicative for additional processes. Here we present a time-domain induced polarization (TDIP) dataset from a volcanic hydrothermal system (Reykjanes, Iceland) where a strong, systematic offset between normal and reciprocal measurements is observed. The discrepancy is present as a positive shift of the reciprocal decay curves (>100 mV/V). While both normal and reciprocal inversions resolve a strong IP anomaly in the southern part of the profile, a substantial offset exists between the two ( >20 mS/m). The offset is spatially confined and persistent over 100 consecutive days, the anomaly itself coincides with a strong IP response attributed to the mutual occurrence of clay minerals and disseminated iron sulphides and oxides. Besides random noise, we evaluate four possible mechanisms for the offset: (1) polarization of the electrodes, (2) inherent sensitivity differences between normal and reciprocal configurations, (3) stray currents from the nearby powerplant and (4) nonlinear IP effects where the response depends on the current density. Given the geological context and the localized nature of anomaly, a non-linear effect is plausible but the ambiguity between systematic noise and a genuine subsurface signal cannot be fully resolved with field data alone. This study demonstrates that the systematic analysis of normal-reciprocal misfit in TDIP can serve as a widely accessible tool for identifying signals that, whether geologic, or anthropogenic in origin, are overlooked by standard data quality workflows.

AbstractTomographic reconstruction involves fundamental trade-offs between resolution, data coverage, and parameterisation. Conventional approaches commonly rely on regularised pixel-wise or cell-based parameterisations. These regularisations are valuable and can themselves be geology-informed when their assumptions match the target setting. However, they often encode structural information only indirectly through generic penalty terms and regularisation weights. As a result, they can be less suited to questions in which the objective is to test a specific structural hypothesis. We develop a Bayesian framework for tomographic reconstruction and uncertainty quantification that formulates such hypotheses explicitly and assesses them through posterior uncertainty. As an illustrative application of the framework, we consider ambient noise tomography. The phase velocity reconstruction is reformulated to address a targeted geological hypothesis by introducing a parameterised structural prior. The prior is constructed from a Whittle–Matérn latent Gaussian field, combined with spectral dimensional reduction via a truncated Karhunen–Loève expansion and a differentiable nonlinear pushforward mapping that promotes approximately piecewise-constant velocity structures. The prior parameters, including the correlation length, smoothness, pushforward slope, and velocity bounds, provide interpretable hyperparameters related to expected lateral scale, structural regularity, transition sharpness, and plausible phase-velocity contrasts. In this formulation, the prior is used as a parameterised structural assumption informed by geological context, rather than as hard geological knowledge. The resulting parameterisation substantially reduces dimensionality, making both maximum a posteriori estimation and Hamiltonian Monte Carlo sampling computationally feasible for the synthetic and observed datasets considered here. Synthetic experiments demonstrate accurate recovery of prescribed velocity contrasts and show that posterior uncertainty patterns reflect acquisition geometry and data coverage. Application to a ∼1000-sensor dense array dataset produces structures consistent with the imposed geology-informed prior assumptions, while providing spatially resolved uncertainty estimates that identify regions of robust inference and remaining ambiguity. The framework offers a proof-of-concept formulation for goal-oriented, uncertainty-aware seismic tomography when prior information supports an explicit structural hypothesis that can be encoded through the proposed Whittle–Matérn latent-field parameterisation.

SummarySubterranean rock masses are in true triaxial stress (TTS) fields, which can preferentially close internal microcracks, thereby inducing elastic anisotropy. However, the seismic response under TTS is poorly understood. Here, an analytical PP-wave reflection coefficient for triaxial stress-induced anisotropic media modelled by microcrack closure mechanism is proposed for seismic inversion. Considering a micromechanical model in which microcracks are represented by stress-dependent compliances, the stress magnitude and orientation are firstly incorporated to account for the elastic anisotropy resulting from triaxial stress-induced microcrack closure. A good agreement is obtained between the model predictions and the existing laboratory measurements. Based on weak anisotropy assumption, we then deduce the effective stiffness tensor of the triaxial stress-induced anisotropic media. Three sets of stress-related anisotropy indicators (SRAIs) are introduced to quantify the microcrack closure effect and anisotropy magnitude induced by triaxial stress. Furthermore, a linearized PP-wave reflection coefficient equation for triaxially stressed isotropic media is derived using the scattering theory. Numerical results validate the feasibility and accuracy of the proposed formula, and reveal the influence of TTS on the PP-wave amplitude variation with angle and azimuth (AVAz). Finally, due to the highly ill-conditioned AVAz inverse problem in triaxial stress-induced anisotropic media, a model and data driven inversion approach is proposed through building on a convolutional neural network. We stepwise estimate the isotropic elastic parameters and SRAIs based on azimuthal seismic amplitude difference inversion strategy. Tests on both synthetic and real seismic data indicate that the nonlinear AVAz inversion framework outperforms conventional approaches in terms of stability and accuracy. Our study allows the construction of elastic properties in triaxial stress-induced anisotropic media, and may provide new insights into determining in-situ stress from seismic data.

SummaryBy leveraging a significantly expanded GNSS dataset (997 vectors, 335 new), we construct a high-resolution crustal strain rate model for the Southeastern Tibetan Plateau (SETP) that approximately doubles the resolution of previous studies. Our model delineates a prominent J-shaped high-strain zone that follows the Xianshuihe-Xiaojiang fault system (XXFS) and extends ∼600 km southwestward. This feature, combined with unsupervised Euler pole clustering (EPC) machine learning results, provides further evidence for a mid-to-lower crustal linkage between the XXFS and faults southwest of the Red River fault (RRF), suggesting the active XXFS is propagating across this ancient boundary (RRF) at depth. Within the SETP interior, our model reveals widespread extensional deformation characterized by a systematic clockwise rotation of its principal axes from north to south, alongside discrete rotational domains partitioned by major faults. Shear strain patterns within the SETP indicate that major strike-slip faults, predominantly sinistral, drive regional strain localization. Moreover, the close association between modeled high-strain zones and historical M ≥ 6 earthquakes suggests that our high-resolution model can serve as a refined, physically-based perspective for regional seismic hazard assessment.

SummaryWhile Bayesian inference is widely applied to magnetotelluric, seismic, and magnetic datasets, its application to gravity anomalies remains relatively unexplored. To address this, we introduce InDIA (Inversion of Density Interface and its Application), a Python-based Bayesian inference framework utilizing a structural Gaussian process. InDIA is developed to estimate discretized subsurface layer depths and densities while explicitly quantifying uncertainty. Unlike previous techniques restricted to continuous depth profiles and localized issues, InDIA integrates diverse prior information to effectively resolve both local and regional gravity anomalies. This approach successfully mitigates the multi-parameter challenges and convergence issues common in local gradient-based optimization techniques, such as Adam. The superiority of this algorithm has been validated through various synthetic models (involving multi-prism configurations, two-layer models with lateral density variations, heterogenous subsurface model with lateral and vertical density variation along with faulted dipping models) featuring both constant and variable density distributions—whether lateral, vertical (prior-constrained), or both—incorporating Gaussian noise, Random-walk noise, Systematic noise and Salt and Pepper noise to replicate real conditions. Furthermore, the framework’s field applicability is demonstrated using two real gravity datasets. First, it tackles multi-layered subsurface profiling in the West Korea Basin. Second, it simultaneously resolves depth and prior-constrained vertical density variations in the Godavari Basin, India. In both field applications, the inverted parameters are geologically viable and closely align with previously established models, confirming InDIA as a reliable tool for gravity data inversion.

SummaryUnderstanding induced seismicity is a critical challenge for seismic risk mitigation, particularly in Europe where geo-resource exploitation increasingly occurs in densely populated areas. The Lacq site in southwestern France represents one of the most significant and long-lasting cases of induced seismicity in Western Europe. Despite decades of observations, the characterization of Lacq seismicity remains limited by insufficient instrumentation and monitoring, leading to poorly constrained earthquake locations and uncertainties regarding whether seismic events occur within or below the reservoir. In this study, we present a high-resolution seismicity catalog covering a five-year period (2021–2026), representing the most comprehensive dataset available for the Lacq area. To achieve this, a temporary seismic network was deployed through a collaboration between GFZ (Potsdam), OMP (Toulouse), and UPPA (Pau). Event detection and location were performed using two independent approaches: a detailed manual picking and relocation workflow, and a fully automated deep-learning-based method. Combined with the use of a pre-existing local 3D velocity model, this dual approach significantly reduces location uncertainties and improves the robustness and completeness of the catalog. The resulting dataset constitutes a valuable resource for future studies.

SummaryHessian–vector products (HVPs) provide matrix-free access to second-order information in full-waveform inversion (FWI), enabling the use of truncated-Newton updates with improved convergence. This may result in enhanced imaging of small-scale scatterers in strongly heterogeneous, high-contrast media. In the time domain, standard adjoint-based HVP derivations are often presented either via perturbation theory or via a Lagrangian (Lagrange-multiplier) framework. Although these routes are formally equivalent, their correctness hinges on a precise treatment of transpose/adjoint placement, boundary/endpoint terms, and the discrete inner product used by the numerical solver. Revisiting these derivations, we show that a commonly used Lagrangian expression for the first HVP contribution can be incorrect when the wave equation operator L is not symmetric, and that rewriting the first term of the HVP, $\mathcal {H}_1$, using L† (or equivalently, swapping the contraction order) requires accounting for the associated boundary terms, where the L† indicates the adjoint of L. To resolve this ambiguity, we provide a complete Lagrangian formulation that explicitly incorporates initial and boundary conditions, together with a Green-identity-based treatment that distinguishes purely algebraic transpose identities from true operator adjoints obtained by integration by parts. We validate the resulting gradient and HVP expressions against reference HVPs computed by automatic differentiation (AD) using an independent JAX computational graph, which we term FDCompGraph. FDCompGraph reproduces the same discrete time stepping, source/receiver injection, and convolutional perfectly matched layer (CPML) treatment as the PDE solver, enabling apples-to-apples validation after matching wavefields under identical discretization and CPML settings. Using the first-order velocity–stress acoustic vertically transversely isotropic (VTI) system with CPML (a representative case where L is non-symmetric), we demonstrate that incorrect handling of $\mathcal {H}_1$ can yield errors up to approximately 400 per cent for HVP in ε and approximately 40 per cent for HVP in δ, whereas the HVPs in Vp and ρ are numerically unaffected. We further assess numerical errors unique to HVP evaluation, which requires perturbed wavefields and their time/space derivatives. In acoustic VTI tests using the damping sponge boundary condition, we observe HVP sensitivity losses of approximately 10 per cent amplitude compared with the results obtained using the CPML boundary condition. These errors may be difficult to diagnose because the commonly used finite-difference check, H δm ≈ g(m + δm) − g(m), is itself a linear approximation, particularly when the gradient g is computed using wavefields affected by the same imperfect boundary absorption. To make accurate HVPs practical at scale, we develop Devito–Hdm, a storage- and I/O-efficient Devito-based engine for co-simulating the required forward, adjoint, tangent, and second-adjoint system, with CPML-consistent checkpointing and partially on-the-fly correlations. As an application, we perform three-parameter VP–ε–δ acoustic VTI FWI using Newton–CG across models with increasing scatterer density and compare it against standard NLCG and L-BFGS. This application shows that incorporating HVP information overall yields a more reliable recovery of small-scale scatterer perturbations at approximately the same computational cost.

SummaryDip is often read from a regularized electrical resistivity tomography (ERT) image without a parameter-specific uncertainty. Image-resolution measures describe how an inversion blurs the resistivity field, but they do not give a noise-dependent limit for the dip of a geological interface. We derive that limit from the 2.5-D forward operator using Fisher information and a Cramér–Rao bound (CRB). The background resistivity is profiled out in log-data space. For a specified structural family, the dip standard error satisfies $\sigma _\theta \ge \sigma /\left\Vert \partial g_0/\partial \theta \right\Vert _{P_{C_0}}$. We also derive a marginal bound for unknown geometry and a first-order bias caused by structural misspecification. Tests with a 30-electrode dipole–dipole array show that the maximum-likelihood estimate attains the CRB when the fitted family is correct. At 3 % noise, the single-parameter floor is 0.14–0.50○ and the geometry-marginalised floor is 0.33–0.72○. Holding a coupled geometry parameter one cell away from its true value instead produces a 3–5○ dip bias. This is 6–25 times the formal floor. An ablation attributes most of the bias to thickness, width, or depth rather than resistivity contrast. The result is conditional on the chosen model family: the CRB measures noise-limited precision within that family, whereas the bias measures the cost of using the wrong geometry. For the cases tested here, improving an independent geometric constraint is therefore more useful than further reducing data noise.

SummaryThe essence of earthquake early warning lies in the rapid and reliable prediction of the ground motion field after an earthquake occurs. While current numerical ground motion prediction methods based on data assimilation avoid the challenge of estimating source parameters, their interpolation processes do not adequately consider the source rupture directivity effects—a key physical factor controlling the spatially heterogeneous pattern of ground motions, resulting in limited prediction accuracy. To address this limitation, we break the traditional framework of numerical ground motion prediction methods and propose a novel real-time method that couples the source and ground motion field. This method incorporates a line-source model, dynamically extracts rupture directivity information, and estimates the line-source strike by comparing observed and predicted ground motion fields in real time, thereby continuously refining subsequent ground motion predictions. Tests using the 2016 Kumamoto earthquake (Mw 7.0) as an example demonstrate that the rupture direction inferred by this method within the early post-event period (within 28s) aligns closely with the results from detailed post-event inversion. After correction for the directivity effects, the accuracy of ground motion prediction is superior to that of traditional numerical prediction methods. The MAE of peak intensity predictions is reduced by 16.1%, 5.4%, and 19.6% for predictions of 5s, 10s, and 20s in advance, respectively.

SummaryGravity curvature is defined as the third-order derivatives of the gravitational potential. It has been proposed as a potential observable for future satellite gradiometry due to its enhanced sensitivity to short-wavelength components of Earth’s gravity field. Previous studies reported that gravity curvature exhibits apparently weaker spatial correlation with Moho geometry than gravity and gravity-gradient observations, raising concerns about its applicability for Moho inversion. In this study, we systematically assess the capability of gravity-curvature data for Moho recovery, with particular emphasis on the contrasting behaviors of spatial- and spectral-domain inversion schemes. The nonlinear spatial inversion is implemented using the iterative regularization, whereas the spectral inversion follows a spherical harmonic Parker-Oldenburg formulation. Synthetic and real-data tests reveal a clear methodological dependence. Spectral inversion demonstrates robust and stable performance, as it primarily exploits the inherent spectral content of the gravitational field and is insensitive to altered spatial patterns. In contrast, spatial inversion is strongly influenced by data-model correlation, with performance improving at higher observation altitudes but degrading under near-surface scenario. At satellite altitudes, with the effect of iterative stabilization, the spatial inversion yields roughly consistent performance to the spectral inversion. These findings highlight that the inversion capability of gravity curvature depends critically on the inversion framework. When spatial distribution and spectral content are properly considered, gravity curvature can provide a promising constraint for density interface inversion by enhancing sensitivity to specific spatial scales.

SummarySeismic imaging of the Earth’s interior has steadily improved in the past 40 years. Full-waveform inversion (FWI) has been the most recent driver of global and local tomography. Although mineral physics experiments link seismic properties to temperature, pressure and composition of the mantle, direct integration of seismic observations and thermodynamic calculations has not yet been achieved. This work illustrates a framework that integrates FWI techniques with mineral physics to directly invert seismic waveforms for mantle temperature and chemical composition. The forward model is based on interpolation of pre-calculated tables of mantle properties, namely density and P- and S-wave speeds. The inverse problem is formulated as an optimization problem. The gradient of the objective function with respect to temperature and composition is computed using the chain rule which allows us to split it into independent seismic and thermodynamic parts. For the seismic part, the gradient is computed with the adjoint method. For the thermodynamic part, gradients of the pre-calculated tables are obtained by analytically deriving the interpolation function, while accounting for the pressure–density coupling, under the assumption of hydrostatic loading. As a proof of concept, we applied this approach on 2-D synthetic models containing thermal and compositional anomalies where P–SV elastic wave propagation is simulated. The optimization problem is solved using the L-BFGS algorithm. We mitigate the trade-off between the effect of temperature and composition by applying a linear transformation to the model parameters before carrying out the minimization and then transforming back once a solution is found. The proposed framework enables the integration of diverse geophysical datasets as well as the incorporation of additional information on the potential origin of certain mantle anomalies based on petrological constraints, which are crucial to tackle the non-uniqueness of the inverse problem.

SummaryInterfacial slip at fluid–solid boundaries can significantly modify hydraulic transport in tight porous media, yet its implications for representative wave-induced fluid-flow mechanisms have not been systematically investigated. In this study, we develop a physically explicit multiscale poroelastic framework to investigate how pore-scale interfacial slip propagates through representative poroelastic relaxation mechanisms. Starting from oscillatory viscous flow in a cylindrical pore with interfacial slip boundary conditions, we derive a slip-modified pore-scale transport model that provides a physically explicit realization of hydraulic transport under slip conditions. Rather than introducing a new wave-induced attenuation mechanism, the derived transport model is consistently incorporated into the Biot, Biot–squirt, and White–squirt formulations to systematically examine how the same pore-scale transport modification influences viscous flow, squirt-flow relaxation, and mesoscopic pressure diffusion. The results show that interfacial slip systematically reduces viscous resistance, enhances dynamic permeability, and primarily manifests as a shift of the characteristic frequencies governing wave-induced fluid flow. Although the same transport modification is introduced into each model, its macroscopic manifestation depends on the governing relaxation mechanism, leading to distinct responses in the Biot, Biot–squirt, and White–squirt frameworks. Comparisons with laboratory measurements on tight carbonate samples demonstrate that the proposed framework reproduces the principal trends of seismic dispersion and attenuation while maintaining physically reasonable model parameters. These results establish a physically explicit connection between pore-scale interfacial transport and multiscale seismic wave propagation, providing a physically consistent basis for incorporating interfacial effects into poroelastic wave-propagation models and for improving the interpretation of seismic responses in tight porous media.

SummaryThe acoustic approximation of seismic data is a widely used approach in marine seismic data processing. Continental shelf margins typically exhibit inclined interfaces. However, existing methods for calculating the P-wave wavefield beneath inclined seabed commonly rely on horizontal assumptions, which introduce errors into seismic data. In this study, an analytical recursive solution for P-wave propagation in inclined seabed formations is derived by establishing boundary condition equations at arbitrarily inclined interfaces using three-dimensional coordinate transformation. By applying three-dimensional coordinate transformation, boundary condition equations are established at arbitrarily inclined interfaces. Based on the acoustic seismic wave differential equations, a set of recursive equations is developed that enables the analytical computation of P-wave propagation in environments with inclined interfaces. This proposed analytical framework provides an effective approach for resolving wavefields in inclined multilayered formations.

SummaryThe Mount Isa Province in northern Australia represents one of the largest exposed Proterozoic crustal regions within the Australian continent and is recognized globally for its world-class mineral deposits. The underlying crustal architecture has been investigated using a series of geophysical profiles informed by geological mapping, gravity, magnetic, magnetotelluric, and seismic data. In this study, we reanalyze legacy passive seismic data recorded across the entire province to generate a high-resolution crustal shear wave velocity model using ambient noise tomography. These results are interpreted alongside reflection seismic profiles acquired along the same survey lines to provide complementary constraints on crustal composition, also with an association of surface geology. Our model reveals distinct velocity contrasts at some major crustal boundaries, including the Gidyea Suture Zone and several major fault zones. Mapped upper crustal intrusive bodies based on seismic reflections appear to cluster around high-velocity anomalies at 5 km depth in our model, and they seem to be linked to the broader higher velocities within the Mount Isa Inlier at 15–20 km depth. These fast, mid-to-lower-crust features probably represent mafic intrusions associated with a trans-lithospheric magmatic system that also involves felsic volcanism and associated metallogenic activity across the broader region. The research results of this area show how passive seismology can be integrated with other geophysical datasets to improve our understanding of fundamental structures for large mineral deposit systems worldwide.

SummaryWe present a comprehensive numerical investigation of the sensitivity of horizontaltovertical spectral ratio (H/V) measurements of ambient seismic noise to the spatial distribution of noise sources and to strong lateral structural contrasts. Our approach relies on the computation of realistic 3D viscoelastic Green’s functions between a dense set of potential surface sources and a few surface receivers, allowing the synthesis of ambient noise wavefields corresponding to arbitrary spatiotemporal source distributions. The study focuses on a simplified 3D representation of the Mygdonian Basin (northern Greece), where significant temporal variations of H/V amplitude, polarisation, and peak frequency have been observed at a site located near a blind fault. Comparisons between the synthetic H/V and the predictions of the elastic Diffuse Field Theory (DFT) show good agreement under isotropic source conditions, confirming the usefulness of DFT as a reference framework in diffuse or quasidiffuse regimes. However, attenuation, lateral heterogeneities, and uneven source distributions lead to substantial deviations from DFT predictions. We find that sourcerelated effects can dominate both amplitude and polarisation patterns, sometimes masking the structural signature of the fault. Nevertheless, the simulations reveal robust spatial trends of H/V polarisation relative to the fault geometry, suggesting that polarisation mapping may help identify lateral heterogeneities when combined with dense, longduration measurements. Our framework provides a physically consistent way to test the limits of the diffuse field assumption and to explore the applicability of ambient noise–based observables in complex geological environments.

SummaryNear-surface geological faults play a crucial role in subsurface systems because they influence fluid flow and transport and can cause geotechnical failure. Topography can sometimes indicate the presence of fault zones, particularly in active tectonic settings where geomorphic features such as triangular facets are preserved. However, particularly for older inactive faults, characteristic surface geomorphic signatures may have been completely removed by erosion, or the fault may be buried beneath younger sediments. Although direct observations from outcrops or borehole data can confirm the presence of a fault very reliably, such data are commonly too sparse and/or distant to precisely constrain fault location in the target area. In this context, near-surface geophysical methods, such as electrical resistivity tomography (ERT) and seismic refraction tomography (SRT), may be useful to locate faults. However, these geophysical data are typically inverted using smoothness-constrained approaches, which can obscure sharp structural features unless the latter are explicitly accounted for in the models. In this study, we present a workflow that enhances the identification of fault zones by jointly inverting collocated ERT and SRT data, considering a parameterization approach based on a layered subsurface model explicitly including a fault. Unlike conventional approaches, we run multiple inversions in which the starting models are initialized with randomly generated fault geometries, guiding the search toward different structural scenarios that yield comparable data misfits. In the absence of additional prior information, these solutions should be interpreted as equally plausible representations of the subsurface. The resulting ensemble of models provides a practical indication of where the structure is well constrained (i.e. features that appear consistently across solutions) and where ambiguity remains (i.e. features that vary among solutions). We showcase the applicability of our workflow with synthetic and field data, where collocated ERT and SRT measurements were collected across expected fault zones. Our results show that the joint inversion of ERT and SRT data consistently narrows the range of plausible fault locations, particularly in the shallow subsurface. Iterative incorporation of prior information (e.g., one vs. two boreholes) leads to significant improvements in model geometry and physical property estimates. However, consistent with the intrinsic ambiguity of geophysical inversion, the results remain non-unique; even models depicting opposite fault types (normal vs. reverse) may fit the data comparably well. This outcome highlights the strength of the workflow in identifying competing structural scenarios that are equally consistent with the observations. Thus, the resulting ensemble provides essential guidance for targeted ground-truthing and follow-up investigations aimed at resolving the remaining ambiguity.

SummaryTo address the challenge of assessing the reliability of three-dimensional (3D) magnetotelluric (MT) inversion results, we have developed a variational inference (VI) inversion framework (VI-MT) based on the Stein Variational Gradient Descent (SVGD) method. Through parallel particle optimization, we efficiently approximate the model posterior distribution, overcoming the computational limitation of traditional Markov Chain Monte Carlo (MCMC) methods. Synthetic tests show that the VI-MT inversion can effectively recover the synthetic model while reducing the tailing effect, and it can approximate the multimodal posterior distribution, providing quantitative uncertainty estimates. Furthermore, the VI-MT inversion is applied to the field MT data collected at the Weishan volcano in northeast China, with the posterior mean model consistent with the deterministic inversion model. A clear low resistivity body of ∼5 Ω·m with small uncertainties is imaged at depths of ∼2-6 km beneath the volcanic crater, suggesting the existence of a shallow magma chamber. Our study shows that the VI-MT inversion based on the SVGD method can efficiently solve the 3D MT Bayesian inversion, providing reliable model uncertainties.

SummaryCharacterizing electronically conductive minerals with spectral induced polarization (SIP) has been a longstanding goal of mineral exploration. A more comprehensive understanding of how mineral texture (e.g. disseminated grain versus veinlet forms) and mineral type influence SIP signals is required, as the polarization mechanisms at the electron conductor-fluid interface and their effect on SIP signatures remain incompletely understood. We performed laboratory measurements on synthetic copper (Cu) and graphite (Gr) samples prepared in disseminated grain and veinlet forms by mixing fixed concentrations of the electron conductor with a non-polarizable sand. Current density dependent SIP (complex impedance) responses were assessed by measuring at multiple current densities, and non-equilibrium was investigated by making measurements as a function of time. SIP responses that remained independent of current density were interpreted as evidence of non-Faradaic processes, with possible minor Faradaic contributions, being dominant. Conversely, current density dependent complex impedance responses were interpreted as an increased contribution from Faradaic polarization processes involving redox reactions at the electrolyte-electron conductor interface. Veinlet Cu samples exhibited strong current density dependent SIP responses and prolonged equilibration, in contrast to the negligible current density dependence of veinlet Gr and disseminated grain Cu and Gr samples. All samples that exhibited negligible current density dependence could be fit to a Pelton-Cole model, whereas samples exhibiting strong current density dependence could only be fit with a more flexible Debye decomposition model. Gr samples displayed a fluid-resistivity dependent peak in the phase response at a specific frequency consistent with mechanistic models, while Cu samples deviated from predicted behaviour. Measurements on rock cores were consistent with observations on synthetic samples, with the SIP response of a Cu bearing rock core exhibiting a clear dependence on current density, whereas the Gr bearing rock core showed negligible current dependence. The observed dependence of the phase and complex resistivity on current density, mineral type, and texture is consistent with differences in the relative importance of Faradaic and non-Faradaic polarization processes. These results suggest that distinguishing current-density-dependent Faradaic responses from current-independent non-Faradaic responses may improve the discrimination of mineralogical and textural characteristics using SIP measurements, although field-scale applications will be limited due to the lower current densities relative to those used in this laboratory study.

SummaryFull Waveform Inversion (FWI) leverages the amplitude and phase of full wavefield seismic data to obtain high-resolution subsurface structures by minimizing an objective function. However, FWI encounters severe cycle-skipping issues when the seismic data lack low-frequency components or the initial velocity model is significantly different from the true velocity model. Furthermore, an incorrect source wavelet can also directly affect the accuracy of inversion results. To overcome these challenges, we introduce the waveform-preserving source-independent wave-equation-based local-scale traveltime inversion (WPS-LTI) method. This approach aims to recover the low-wavenumber velocity structures as an initial velocity model, while reducing the influence of an incorrect source wavelet. In the WPS-LTI method, seismic data are processed via both convolution and deconvolution with reference traces, resulting in modified observed and synthetic seismic data that share the same seismic wavelet. This approach helps avoid waveform distortions typically introduced by conventional source-independent methods that rely solely on convolution. By preserving the waveform characteristics, we can use cross-correlation to compute more accurate local-scale traveltime differences, thus recovering accurate low-wavenumber velocity structures. Tests conducted on the Marmousi model and the Chevron 2014 blind test dataset demonstrate that the WPS-LTI method can significantly alleviate cycle-skipping issues and reduce the dependence on the source wavelet in FWI.

SummaryWith the shale gas exploration and development into deep strata (> 3500 m in depth), weak signal detection becomes more challenging and vital in surface-based microseismic monitoring. Most existing deep learning approaches are trained on synthetic or single-station data, which limits their performance in detecting weak events in field seismic recordings. Here, we propose a deep learning approach based on the computer vision object detection model YOLOv5 via transfer learning and train it on a dataset from field observation to better detect microseismic events in multi-channel seismic data. First, we convert continuous seismic data into waveform images in grayscale with a record length of 20 s, and manually annotate the microseismic events containing the first arrival signals using rectangular boxes. To train the model, the transfer learning strategy is used by using parameters of YOLOs in computer vision to initiate the model. After training and validation, we conducted comprehensive tests to examine the performance of the model. Compared with the single-station short-term average/long-term average (STA/LTA) detector, our method reduces false positives markedly, yet it also delivers higher recall rate than the available multi-station detection scheme. When applied to continuous dataset from a different area, the model achieved good performance with a precision of 0.9832 and a recall of 0.9702, indicating it learns representative features of microseismic signals. All these tests illustrate that the proposed method achieves high detection accuracy, low false positive rates, and robustness. Meanwhile, the method possesses the superiority for real-time monitoring and can be easily updated by adding a few new samples to the training dataset.