https://arxiv.org/api//SOeyx8AfeKGlbmv9W4b6n6Ev+s 2026-04-09T08:27:27Z 8557 180 15 http://arxiv.org/abs/2512.10448v2 2026-02-15T17:34:29Z

Ambient noise tomography relies on the assumption that the seismic wavefield is equipartitioned. In practice, ambient noise sources are spatially and temporally heterogeneous, producing biased estimates of the Green's function between stations. We introduce a data-driven method, Coherent Source Subsampling (CSS), which selects and averages only cross-correlation time windows associated with excitation of sources in the stationary zone. By restricting the ensemble average to these windows, CSS mitigates the effects of nonuniform source distribution and restores causal-acausal symmetry in the retrieved interstation response. Applications to regional ambient-noise datasets show that CSS stabilizes surface-wave dispersion measurements even when source statistics violate the assumptions of standard seismic interferometry. For the central California dataset, CSS-derived group-velocity tomograms consistently image a high-velocity block between the Rinconada and San Andreas faults across multiple periods. In comparison, the full-ensemble (linear) average does not capture this block, which is well established. Our approach is particularly useful for short-duration passive surveys.

2025-12-11T09:13:28Z Sanket Narayan Bajad Pushkar Bharadwaj Pawan Bharadwaj http://arxiv.org/abs/2602.14217v1 2026-02-15T16:16:02Z

Lorentz-symmetry violation may be described via the CPT-odd, dimension-3, Carroll-Field-Jackiw term, which couples the electromagnetic fields to a constant 4-vector $k_{\rm AF}$ selecting a preferred direction in spacetime. We solve the field equations using the Green's method for a static point-like magnetic dipole and find the $k_{\rm AF}$-dependent corrections to the standard dipolar magnetic field that strongly dominates the near-Earth magnetic field. Given the very good agreement between current models and ground- and satellite-based geomagnetic data, our strongest constraints on the components of $k_{\rm AF}$ in the Sun-centered frame read $|(k_{\rm AF})_Z| \lesssim 4 \times 10^{-25} \, {\rm GeV}$ for $|(k_{\rm AF})_X|, |(k_{\rm AF})_Y| \lesssim 10^{-24} \, {\rm GeV}$ at the two-sigma level. This represents an improvement of about four orders of magnitude over earlier bounds based on other geophysical phenomena.

2026-02-15T16:16:02Z 58 pages, 16 figures. Constructive comments are welcome G. F. de Carvalho M. Fillion P. C. Malta C. A. D. Zarro http://arxiv.org/abs/2602.16857v1 2026-02-15T05:49:16Z

This paper introduces a distillation framework for an ensemble of entropy-optimal Sparse Probabilistic Approximation (eSPA) models, trained exclusively on satellite-era observational and reanalysis data to predict ENSO phase up to 24 months in advance. While eSPA ensembles yield state-of-the-art forecast skill, they are harder to interpret than individual eSPA models. We show how to compress the ensemble into a compact set of "distilled" models by aggregating the structure of only those ensemble members that make correct predictions. This process yields a single, diagnostically tractable model for each forecast lead time that preserves forecast performance while also enabling diagnostics that are impractical to implement on the full ensemble. An analysis of the regime persistence of the distilled model "superclusters", as well as cross-lead clustering consistency, shows that the discretised system accurately captures the spatiotemporal dynamics of ENSO. By considering the effective dimension of the feature importance vectors, the complexity of the input space required for correct ENSO phase prediction is shown to peak when forecasts must cross the boreal spring predictability barrier. Spatial importance maps derived from the feature importance vectors are introduced to identify where predictive information resides in each field and are shown to include known physical precursors at certain lead times. Case studies of key events are also presented, showing how fields reconstructed from distilled model centroids trace the evolution from extratropical and inter-basin precursors to the mature ENSO state. Overall, the distillation framework enables a rigorous investigation of long-range ENSO predictability that complements real-time data-driven operational forecasts.

2026-02-15T05:49:16Z Michael Groom Davide Bassetti Illia Horenko Terence J. O'Kane http://arxiv.org/abs/2601.02158v2 2026-02-14T22:11:57Z

This paper presents FormationEval, an open multiple-choice question benchmark for evaluating language models on petroleum geoscience and subsurface disciplines. The dataset contains 505 questions across seven domains including petrophysics, petroleum geology and reservoir engineering, derived from three authoritative sources using a reasoning model with detailed instructions and a concept-based approach that avoids verbatim copying of copyrighted text. Each question includes source metadata to support traceability and audit. The evaluation covers 72 models from major providers including OpenAI, Anthropic, Google, Meta and open-weight alternatives. The top performers achieve over 97% accuracy, with Gemini 3 Pro Preview reaching 99.8%, while tier and domain gaps persist. Among open-weight models, GLM-4.7 leads at 98.6%, with several DeepSeek, Llama, Qwen and Mistral models also exceeding 93%. The performance gap between open-weight and closed models is narrower than expected, with several lower-cost open-weight models exceeding 90% accuracy. Petrophysics emerges as the most challenging domain across all models, while smaller models show wider performance variance. Residual length bias in the dataset (correct answers tend to be longer) is documented along with bias mitigation strategies applied during construction. The benchmark, evaluation code and results are publicly available.

2026-01-05T14:36:02Z v2: expanded related work, added validation details, difficulty-domain table, community feedback website (at https://www.formationeval.no). 28 pages, 8 figures, 11 tables. Benchmark and code at https://github.com/AlmazErmilov/FormationEval-an-Open-Benchmark-for-Oil-Gas-Geoscience-MCQ-Evaluation Almaz Ermilov http://arxiv.org/abs/2602.13820v1 2026-02-14T15:33:07Z

The spinel phase (wadsleyite, ringwoodite) in the mantle transition zone (MTZ), can contain up to 1-2 weight percent of water. However, whether these water reservoirs in the MTZ are filled is debated. Here, we investigate water dynamics in the MTZ numerically by using a newly developed empirical model of deep hydrous mantle melting combined with 2D thermo-hydro-mechanical-chemical (THMC) upper mantle models. Numerical modeling results suggest that water-induced buoyancy triggers the development of hydrous solid-state mantle upwellings in the MTZ. On time scales of some tens of millions of years, they rise to and interact with the spinel-olivine phase transition. Depending on the water content and temperature of these thermal-chemical plumes, this crossing may trigger hydrous melting by water release from the wadsleyite upon its conversion to olivine. The melts are less dense than the solid matrix and continue rising upward in the form of either diapirs or porosity waives. Similar dehydration-induced melting process3 is also documented for the lower MTZ boundary, where hydrous downwellings (such as subducted slabs) generate buoyant melt diapirs rising through the MTZ. We therefore suggest that the MTZ operates as a transient water reservoir. Relatively small amounts of water (less than 0.1 weight percent, smaller than 0.2 ocean masses) and a geologically moderate duration (80-430 Myr) of the transient water storage should be characteristic for the MTZ, which may play a key role in stabilizing the surface ocean mass on Earth and Earth-like rocky exoplanets.

2026-02-14T15:33:07Z 26 Pages, 13 Figures, 2 Tables, submitted to Nature (under review) Taras V. Gerya Nickolas M. Bardi Shun-ichiro Karato Motohiko Murakami http://arxiv.org/abs/2509.04909v2 2026-02-13T12:21:04Z

Seismological and geodetic observations of fault zones reveal diverse slip dynamics, scaling, and statistical laws. Existing mechanisms explain some but not all of these behaviors. We show that incorporating an off-fault damage zone-characterized by distributed fractures surrounding a main fault-can reproduce many key features observed in seismic and geodetic data. We model a 2D shear fault zone in which off-fault cracks follow power-law size and density distributions, and are oriented either optimally or parallel to the main fault. All fractures follow rate-and-state friction with parameters enabling slip instabilities. We do not introduce spatial heterogeneities in frictional properties. Using quasi-dynamic boundary integral simulations accelerated by hierarchical matrices, we simulate slip dynamics and analyze events produced both on and off the main fault. Despite spatially uniform frictional properties, we observe a natural continuum from slow to fast ruptures, as seen in nature. Our simulations reproduce the Omori law, inverse Omori law, Gutenberg-Richter scaling, and moment-duration scaling. We observe seismicity localizing toward the main fault before nucleation of main-fault events. During slow slip events, off-fault seismicity migrates in patterns resembling fluid diffusion fronts, despite the absence of fluids. We show that tremors, Very Low Frequency Earthquakes, Low Frequency Earthquakes, Slow Slip Events, and earthquakes can all emerge naturally within this fault volume framework, making it an ideal digital twin for testing hypotheses, performing ground-truth inversions, and probing mechanical properties inaccessible with natural observations.

2025-09-05T08:24:41Z M. Almakari N. Kheirdast C. Villafuerte M. Y. Thomas P. Dubernet J. Cheng A. Gupta P. Romanet S. Chaillat H. S. Bhat http://arxiv.org/abs/2602.12408v1 2026-02-12T20:57:02Z

Identifying systematic patterns in seismicity that precede large earthquakes remains a central challenge in statistical seismology. In this work, we present a methodological framework for detecting spatiotemporal anomalies in seismicity using the evolution of gridded b-values. Focusing on the Japanese subduction zone, we construct daily b-value fields on a fine spatial grid by aggregating local seismicity over moving time windows, yielding a continuous 2+1D representation of seismic-state evolution. We formulate the problem as a binary classification task in which spatiotemporal blocks extracted from these $b$-value fields are labeled according to the occurrence of a target earthquake with \Mw $\geq 5$ in the central region within the next day. To model this data, we introduce a hybrid deep-learning architecture that combines a spatial convolutional encoder with a temporal convolutional network, enabling joint learning of spatial structure and temporal dynamics. A progressive meta-epoch training scheme is employed, in which the model is iteratively updated using a time-forward strategy that mirrors operational deployment and mitigates issues related to nonstationarity. This paper is strictly methodological in scope. It describes the construction of b-value fields, the spatiotemporal sampling strategy, the network architecture, and the progressive training and internal validation framework used for model development and parameter selection.

2026-02-12T20:57:02Z Jonas Köhler Wei Li Johannes Faber Georg Rümpker Nishtha Srivastava http://arxiv.org/abs/2602.12075v1 2026-02-12T15:34:39Z

The Poisson equation governing a planet's gravitational field is posed on the unbounded domain, $\mathbb{R}^3$, whereas finite-element computations require bounded meshes. We implement and compare three strategies for handling the infinite exterior in the finite-element method: (i) naive domain truncation; (ii) Dirichlet-to-Neumann (DtN) map on a truncated boundary; (iii) multipole expansion on a truncated boundary. While all these methods are known within the geophysical literature, we discuss their parallel implementations within modern open-source finite-element codes, focusing specifically on the widely-used MFEM package. We consider both calculating the gravitational potential for a static density structure and computing the linearised perturbation to the potential caused by a displacement field - a necessary step for coupling self-gravitation into planetary dynamics. In contrast to some earlier studies, we find that the domain truncation method can provide accurate solutions at an acceptable cost, with suitable coarsening of the mesh within the exterior domain. Nevertheless, the DtN and multipole methods provide superior accuracy at a lower cost within large-scale parallel geophysical simulations despite their need for non-local communication associated with spherical harmonic expansions. The DtN method, in particular, admits an efficient parallel implementation based on an MPI-communicator limited to processors that contain part of the mesh's outer boundary. A series of further illustrative calculations are provided to show the potential of the DtN and multipole methods within realistic geophysical modelling.

2026-02-12T15:34:39Z Ziheng Yu Alex D. C. Myhill David Al-Attar http://arxiv.org/abs/2602.11777v1 2026-02-12T09:54:37Z

Reservoir simulations help the energy industry make better decisions by predicting how fluids like oil, gas, water, hydrogen, and carbon dioxide will flow underground. To keep these predictions accurate, engineers often need to update geological models quickly as new information becomes available. pycopm is a tool designed to make this process faster and easier. It allows users to adjust geological models in several ways, such as simplifying complex grids, focusing on specific parts of a reservoir, or changing the shape and position of the model. These capabilities help engineers test different scenarios efficiently. Although pycopm was first used on two well-known public datasets, it has since become useful in many other situations because of its easy-to-use features and recent extensions. Today, it supports studies involving model refinement, comparing coarse and detailed models, analyzing interactions between nearby sites, and speeding up troubleshooting in large simulations.

2026-02-12T09:54:37Z David Landa-Marbán http://arxiv.org/abs/2602.11631v1 2026-02-12T06:28:14Z

Accurate and efficient wavefield modeling underpins seismic structure and source studies. Traditional methods comply with physical laws but are computationally intensive. Data-driven methods, while opening new avenues for advancement, have yet to incorporate strict physical consistency. The principle of reciprocity is one of the most fundamental physical laws in wave propagation. We introduce the Reciprocity-Enforced Neural Operator (RENO), a transformer-based architecture for modeling seismic wave propagation that hard-codes the reciprocity principle. The model leverages the cross-attention mechanism and commutative operations to guarantee invariance under swapping source and receiver positions. Beyond improved physical consistency, the proposed architecture supports simultaneous realizations for multiple sources without crosstalk issues. This yields an order-of-magnitude inference speedup at a similar memory footprint over an reciprocity-unenforced neural operator on a realistic configuration. We demonstrate the functionality using the reciprocity relation for particle velocity fields under single forces. This architecture is also applicable to pressure fields under dilatational sources and travel-time fields governed by the eikonal equation, paving the way for encoding more complex reciprocity relations.

2026-02-12T06:28:14Z Caifeng Zou Yaozhong Shi Zachary E. Ross Robert W. Clayton Kamyar Azizzadenesheli http://arxiv.org/abs/2602.11621v1 2026-02-12T06:15:29Z

This study introduces Differentiable Graph Neural Network Simulators (Diff-GNS) as a physics-informed and automated framework for estimating post-liquefaction residual strengths ($S_r$). Traditional approaches to estimate $S_r$ rely on simplified physics, manual iterations, and assumptions about runout development. Diff-GNS overcomes these limitations by integrating a Graph Neural Network Simulator (GNS) that simulates granular flows, with gradient-based optimization through automatic differentiation. GNS accelerates forward runout simulations that are otherwise computationally intensive with conventional numerical methods, while gradient-based optimization automates the inversion to back-calculate $S_r$. The GNS is trained on simulations with the material point method on geometries informed by case-history runout failures, enabling focused learning of realistic runout mechanisms and the ability to simulate slopes across small and large scales. The Diff-GNS framework is validated using two well-documented liquefaction-induced flow failure case histories: the Lower San Fernando dam and La Marquesa dam. In the two cases, the inferred $S_r$ agrees closely with published estimates and reproduces physically consistent runout behaviors. The framework also has the ability to jointly infer multiple interacting parameters, extending beyond single-parameter back-analyses. By embedding the physics of runout processes, minimizing manual intervention, and accelerating the inversion process to estimate $S_r$, Diff-GNS provides an efficient, reproducible, and physically grounded approach for geotechnical analysis of liquefaction-induced flow failures.

2026-02-12T06:15:29Z Yongjin Choi Jorge Macedo http://arxiv.org/abs/2602.11313v1 2026-02-11T19:34:50Z

Machine learning (ML)-based models have demonstrated high skill and computational efficiency, often outperforming conventional physics-based models in weather and subseasonal predictions. While prior studies have assessed their fidelity in capturing synoptic-scale atmospheric dynamics, their performance across timescales and under out-of-distribution forcing, such as +3K or +4K uniform-warming forcings, and the sources of biases remain elusive, to establish the model reliability for Earth science. Here, we design three sets of experiments targeting synoptic-scale phenomena, interannual variability, and out-of-distribution uniform-warming forcings. We evaluate the Neural General Circulation Model (NeuralGCM), a hybrid model integrating a dynamical core with ML-based component, against observations and physics-based Earth system models (ESMs). At the synoptic scale, NeuralGCM captures the evolution and propagation of extratropical cyclones with performance comparable to ESMs. At the interannual scale, when forced by El Niño-Southern Oscillation sea surface temperature (SST) anomalies, NeuralGCM successfully reproduces associated teleconnection patterns but exhibits deficiencies in capturing nonlinear response. Under out-of-distribution uniform-warming forcings, NeuralGCM simulates similar responses in global-average temperature and precipitation and reproduces large-scale tropospheric circulation features similar to those in ESMs. Notable weaknesses include overestimating the tracks and spatial extent of extratropical cyclones, biases in the teleconnected wave train triggered by tropical SST anomalies, and differences in upper-level warming and stratospheric circulation responses to SST warming compared to physics-based ESMs. The causes of these weaknesses were explored.

2026-02-11T19:34:50Z 48 pages, 9 figures Ziming Chen L. Ruby Leung Wenyu Zhou Jian Lu Sandro W. Lubis Ye Liu Chuan-Chieh Chang Bryce E. Harrop Ya Wang Mingshi Yang Gan Zhang Yun Qian http://arxiv.org/abs/2601.08356v2 2026-02-11T17:46:48Z

It was conjectured for a long time that the tectonic plates are in a self-organized state of criticality and that the Gutenberg-Richter law is a manifestation of that. It was recently shown that for a system near criticality, the inequality of their responses due to external driving would sharply rise and show universal behavior that could indicate proximity of the system to a critical point. As a result, measures such as the Gini and Kolkata indices that quantify inequality, can also serve as indicators of imminent criticallity and that of diverging (system spanning) responses. In the context of earthquakes, such a large response would correspond to events of high magnitudes. In this work, we show with numerical simulations and seismic data analysis that large earthquake events have a tendency to follow events that are highly unequal, similar to the case of a system near a critical point. Therefore, a continuous monitoring of the inequality indices of the earthquake time series could be useful for hazard estimates. We have applied this framework to models of earthquakes as well as to the earthquake time series from various tectonically active regions, such as North America, Southern Japan, parts of South-East Asia and Indonesia. The findings also indicate a quantitative estimate of the distance from criticality, when the tectonic plates are viewed as a self-organized critical system.

2026-01-13T09:13:08Z 18 pages, 18 figures Sudip Sarkar Soumyajyoti Biswas http://arxiv.org/abs/2602.10557v1 2026-02-11T06:11:03Z

Absolute gravimeters deliver drift-free, high-precision measurements but are typically bulky and difficult to deploy, whereas relative gravimeters are lightweight and mobile but intrinsically limited by time-dependent drift. We demonstrate a hybrid quantum-enabled gravimetry approach in which an on-site atomic gravimeter provides routine, $μ$Gal-level calibration of two mobile spring gravimeters during a field survey spanning 24 km$^2$ of dense tropical terrain. The atomic reference enables high-precision, asynchronous cross-comparison of relative measurements acquired over seven days, effectively suppressing instrumental drift to a level required for demanding geophysical applications. This deployment captures regional gravity gradients with high fidelity under challenging environmental conditions, illustrating how field-operable quantum sensors can extend quantum-grade gravimetry beyond laboratory settings and serve as scalable calibration backbones for large-area, high-precision geophysical surveys in remote or logistically constrained environments.

2026-02-11T06:11:03Z Nathan Shettell Kai Sheng Lee Fong En Oon Elizaveta Maksimova Hong Hui Chen Rainer Dumke http://arxiv.org/abs/2509.03629v3 2026-02-11T03:52:26Z

We introduce SeisDiff-denoNIA, a diffusion-based seismic denoising framework that trains directly on noise extracted from field data, eliminating the dependence on synthetic datasets. Unlike conventional denoising methods that require clean signal labels, our approach uses field noise recorded as training targets, enabling the diffusion model to explicitly learn the true noise distribution. We further demonstrate the framework on a field DAS-VSP survey, where the model effectively suppresses multiple types of noise, including production, instrument, and environment related noise, while preserving key seismic events. By denoising shot gathers with SeisDiff-denoNIA prior to migration, the resulting images exhibit improved event continuity and enhanced reflection visibility without artificial artifacts. In addition, synthetic experiments show that the method substantially outperforms traditional signal-based diffusion models under low-SNR conditions. These results suggest that explicitly modeling noise is not only viable but advantageous for a broad class of seismic denoising tasks, particularly in challenging field environments.

2025-09-03T18:30:48Z This manuscript has been submitted to Geophysics and is currently under review Donglin Zhu Peiyao Li Ge Jin