Virtual receiver functions via conditional diffusion transformers for robust crustal imaging

2026-03-21T18:23:49Z

Receiver functions (RFs) are widely used to image crustal and upper-mantle structure, and their variation with backazimuth and epicentral distance contains key information about layering and azimuthal anisotropy. In practice, however, RFs are contaminated by nuisance effects from unknown earthquake source signatures and seismic noise, which obstruct reliable crustal imaging. Sparse RF coverage across backazimuths and epicentral distances also leads to biased anisotropy estimates. We address these challenges using conditional diffusion models, conditioned on backazimuth, epicentral distance, and station coordinates, to produce high-quality virtual radial and transverse RFs. RFs from earthquakes with similar backazimuths and epicentral distances share consistent crustal responses but differ in nuisance effects, allowing the model to suppress the latter. Our framework generates virtual RFs within gaps in backazimuth and epicentral distance coverage, enhancing the interpretation of crustal anisotropy and layering. On synthetic RFs with realistic non-Gaussian noise, virtual RFs correlate more strongly with the true RFs than traditional linear or phase-weighted stacking. Applied to the Cascadia Subduction Zone, virtual radial RFs sharply image scattered S-waves from the dipping slab, with enhanced phase clarity and backazimuthal coverage relative to previous work. In southern California, anisotropy parameters inferred from virtual RFs are spatially coherent and consistent with regional fault geometry. Our approach leverages all available RFs, regardless of quality, to increase spatial coverage and support robust, automated RF analysis.

Long-term outburst activity of comet 17P/Holmes and constraints on ejecta size distributions

2026-03-21T15:32:15Z

A quantitative understanding of cometary outbursts requires robust constraints on the size distribution of ejected particles, which governs outburst dynamics and underpins estimates of released gas and dust. In the absence of direct measurements of particle sizes, assumptions about the size distribution play a central role in modelling dust-trail formation, their dynamical evolution and observability, and the potential production of meteor showers following encounters with Earth. We analyse brightness amplitude variations associated with outbursts of comet 17P/Holmes from 1892 to 2021, with particular emphasis on the exceptional 2007 mega-outburst. During this event the comet underwent a rapid and substantial brightening: at its peak, the expanding coma reached a diameter exceeding that of the Sun and briefly became the largest object in the Solar System visible to the naked eye. We constrain the size distribution and total mass of porous agglomerates composed of ice, organics, and dust ejected during the outburst. The inferred particle size distribution is consistent with a power law of index q, yielding effective particle sizes between 10^-6 m for q = 4 and 5 x 10^-3 m for q = 2. Accounting for effective particle size, sublimation flux, and bulk density, we find that the total number of ejected particles increases with both q and sublimation flux. These results place constraints on the physical properties of outburst ejecta and provide physically motivated initial conditions for long-term dust-trail evolution modelling. They further indicate that cometary outburst brightness is determined primarily by the number of particles and their size distribution, rather than by the total ejected mass alone, with direct implications for the origin and evolution of meteoroid streams and the interplanetary dust population.

A quiet quantum revolution in Earth's deep interior

2026-03-21T13:15:49Z

The Earth's lower mantle hosts a subtle but pervasive quantum phenomenon: the pressure-induced spin crossover of iron in its dominant minerals, bridgmanite and ferropericlase. In this transition, iron ions gradually shift from high-spin to low-spin electronic states without structural change, altering their volume, compressibility, and elastic properties. Although long recognized experimentally and theoretically, its geophysical significance has only recently become clear through the integration of mineral physics and three-dimensional seismic imaging. The spin crossover reduces bulk modulus and P-wave velocities while leaving S-wave speeds largely unaffected, producing a distinctive decoupling between P- and S-wave anomalies. This signature is now observed in global tomography and reconciles seismic observations with realistic mantle temperatures and compositions. Rather than forming a sharp boundary, the crossover extends across most of the lower mantle, acting as a diffuse yet essential control on seismic structure. This work highlights how quantum-scale electronic transitions influence planetary-scale dynamics and interpretations of Earth's deep interior.

Weakly supervised multimodal segmentation of acoustic borehole images with depth-aware cross-attention

2026-03-21T09:28:05Z

Acoustic borehole images provide high-resolution borehole-wall structure, but large-scale interpretation remains difficult because dense expert annotations are rarely available and subsurface information is intrinsically multimodal. The challenge is developing weakly supervised methods combining two-dimensional image texture with depth-aligned one-dimensional well-logs. Here, we introduce a weakly supervised multimodal segmentation framework that refines threshold-guided pseudo-labels through learned models. This preserves the annotation-free character of classical thresholding and clustering workflows while extending them with denoising, confidence-aware pseudo-supervision, and physically structured fusion. We establish that threshold-guided learned refinement provides the most robust improvement over raw thresholding, denoised thresholding, and latent clustering baselines. Multimodal performance depends strongly on fusion strategy: direct concatenation provides limited gains, whereas depth-aware cross-attention, gated fusion, and confidence-aware modulation substantially improve agreement with the weak supervisory reference. The strongest model, confidence-gated depth-aware cross-attention (CG-DCA), consistently outperforms threshold-based, image-only, and earlier multimodal baselines. Targeted ablations show its advantage depends specifically on confidence-aware fusion and structured local depth interaction rather than model complexity alone. Cross-well analyses confirm this performance is broadly stable. These results establish a practical, scalable framework for annotation-free segmentation, showing multimodal improvement is maximized when auxiliary logs are incorporated selectively and depth-aware.

A decade of airborne electromagnetic surveying Lake Menindee (Australia) under varying water levels

2026-03-20T13:21:09Z

Time domain airborne electromagnetic (AEM) surveying is a mature geophysical tool for imaging the Earth's shallow subsurface. It produces images of the electromagnetic conductivity structure of the earth, down to depths of a few hundred metres. The AEM method is fast, with aircraft acquiring data at speeds of 100-300 km/hr, making it an ideal near-surface reconnaissance tool. The physics of the AEM method are sensitive primarily to the subsurface conductivity, which is influenced by a range of geological factors such as mineral content, porosity, and water content and chemistry. In addition, the inferred subsurface conductivity depends on the accurate measurement and modelling of airborne transmitter and receiver geometries. In this work, we present inferences of the subsurface conductivity over Lake Menindee, New South Wales, Australia, using data from various AEM systems over the period 2014-2024. The lake storage has varied dramatically over this time and while this difference in storage volume undoubtedly influences the near surface conductivity, a remarkably consistent interpretation of the regional geology emerges. While the upper ten metres of the modelled depth sections exhibit the greatest time-variability in inferred electromagnetic conductivity, a correlation of lakebed near-surface conductivity with the lake water volume cannot robustly be established. We also provide information theoretic calculations for each inversion result to aid in their quantitative comparison. The implications of our study are that subtle, shallow, hydrogeological changes are difficult to image with repeat overflights. Conversely, we establish that different AEM systems robustly image the regional geo-electric structure of the near surface, validated by known stratigraphy and borehole conductivity logs.

On the role of memorization in learned priors for geophysical inverse problems

2026-03-20T04:22:26Z

Learned priors based on deep generative models offer data-driven regularization for seismic inversion, but training them requires a dataset of representative subsurface models -- a resource that is inherently scarce in geoscience applications. Since the training objective of most generative models can be cast as maximum likelihood on a finite dataset, any such model risks converging to the empirical distribution -- effectively memorizing the training examples rather than learning the underlying geological distribution. We show that the posterior under such a memorized prior reduces to a reweighted empirical distribution -- i.e., a likelihood-weighted lookup among the stored training examples. For diffusion models specifically, memorization yields a Gaussian mixture prior in closed form, and linearizing the forward operator around each training example gives a Gaussian mixture posterior whose components have widths and shifts governed by the local Jacobian. We validate these predictions on a stylized inverse problem and demonstrate the consequences of memorization through diffusion posterior sampling for full waveform inversion.

Improving moment tensor solutions under Earth structure uncertainty with simulation-based inference

2026-03-19T14:04:17Z

Bayesian inference represents a principled way to incorporate Earth structure uncertainty in full-waveform moment tensor inversions, but traditional approaches generally require significant approximations that risk biasing the resulting solutions. We introduce a robust method for handling theory errors using simulation-based inference (SBI), a machine learning approach that empirically models their impact on the observations. This framework retains the rigour of Bayesian inference while avoiding restrictive assumptions about the functional form of the uncertainties. We begin by demonstrating that the common Gaussian parametrisation of theory errors breaks down under minor ($1-3 \%$) 1-D Earth model uncertainty. To address this issue, we develop two formalisms for utilising SBI to improve the quality of the moment tensor solutions: one using physics-based insights into the theory errors, and another utilising an end-to-end deep learning algorithm. We then compare the results of moment tensor inversion with the standard Gaussian approach and SBI, and demonstrate that Gaussian assumptions induce bias and significantly under-report moment tensor uncertainties. We also show that these effects are particularly problematic when inverting short period data and for shallow, isotropic events. On the other hand, SBI produces more reliable, better calibrated posteriors of the earthquake source mechanism. Finally, we successfully apply our methodology to two well studied moderate magnitude earthquakes: one from the 1997 Long Valley Caldera volcanic earthquake sequence, and the 2020 Zagreb earthquake.

Towards Efficient and Stable Ocean State Forecasting: A Continuous-Time Koopman Approach

2026-03-19T10:03:30Z

We investigate the Continuous-Time Koopman Autoencoder (CT-KAE) as a lightweight surrogate model for long-horizon ocean state forecasting in a two-layer quasi-geostrophic (QG) system. By projecting nonlinear dynamics into a latent space governed by a linear ordinary differential equation, the model enforces structured and interpretable temporal evolution while enabling temporally resolution-invariant forecasting via a matrix exponential formulation. Across 2083-day rollouts, CT-KAE exhibits bounded error growth and stable large-scale statistics, in contrast to autoregressive Transformer baselines which exhibit gradual error amplification and energy drift over long rollouts. While fine-scale turbulent structures are partially dissipated, bulk energy spectra, enstrophy evolution, and autocorrelation structure remain consistent over long horizons. The model achieves orders-of-magnitude faster inference compared to the numerical solver, suggesting that continuous-time Koopman surrogates offer a promising backbone for efficient and stable physical-machine learning climate models.

Decadal sink-source shifts of forest aboveground carbon since 1988

2026-03-18T16:30:17Z

Forest ecosystems are vital to the global carbon cycle, yet their long-term aboveground carbon (AGC) dynamics remain uncertain. Here, we integrate multi-source satellite observations with probabilistic deep learning models to reconstruct a harmonized, uncertainty-aware global forest AGC record from 1988 to 2021 at 0.25-deg. We find that, although global forests sequestered 6.2 PgC, moist tropical and boreal forests have progressively transitioned toward carbon sources since the early 2000s. This shift coincides with a strengthening negative correlation between tropical AGC variability and atmospheric CO2 growth rates (r = -0.63 in 2011-2021), suggesting tropical forests increasingly modulate the global carbon cycle. Notably, in the Brazilian Amazon, the contribution of intact forests to the year-to-year variations in AGC losses increased from 33% in the 1990s to 76% in the 2010s, surpassing that of deforested areas (from 60% to 13%). Our findings highlight the vulnerability of carbon stocks in key biomes and provide a benchmark to track emerging sink-source shifts under anthropogenic climate change.

Compressive Structures in the Foreshock of Collisionless Shocks

2026-03-18T16:05:27Z

Collisionless shocks are fundamental accelerators of energetic particles; yet, the observations of nonlinear foreshock structures, which are essential in acceleration processes, differ significantly between Interplanetary (IP) shocks and planetary bow shocks. We present a direct comparison of two high-Mach-number, quasi-parallel shocks: an IP shock observed by Solar Orbiter and the Earth's bow shock measured by the Magnetospheric Multiscale (MMS) mission during the 2024-2025 ``string-of-pearls'' campaign. We show that Foreshock Compressive Structures (FCSs) initiate upstream of both shocks at similar normalized distances ($\lesssim$50 ion inertial lengths, $d_i$) when the suprathermal ($>10$ keV) ion density exceeds $\sim$1\% of the background. However, the IP shock lacks the fully evolved, high-amplitude Short Large Amplitude Magnetic Structures (SLAMS) characteristic of the terrestrial foreshock. We demonstrate that the ``growth zone'' capable of sustaining these structures is spatially limited ($\sim$135 $d_i$), which, due to the high speed of the propagating IP shock, corresponds to a brief observational window of $<10$ s. Beyond this observational constraint, we suggest an additional physical mechanism that can inhibit foreshock maturity at IP shocks. The lack of global curvature prevents the lateral supply (``cross-talk'') of energetic ions from different shock regions. These findings suggest that while the fundamental physics of FCS initiation is unified across collisionless shocks, the achievement of full nonlinearity can be regulated by the unique shock geometry and upstream properties, while ultimately remaining observationally challenging to identify.

Deep Learning-Based 3D Seismic Velocity Inversion Under Dual-Domain Sparse Representation

2026-03-18T13:19:38Z

Three-dimensional seismic full-waveform inversion (FWI) provides high-fidelity subsurface velocity models but is restricted by high computational cost, strong nonlinearity, cycle-skipping, and heavy dependence on initial models. Although data-driven deep learning mitigates these issues, it still produces over-smoothed results with limited physical interpretability and low efficiency. To address these challenges, we propose a dual-domain sparse deep learning framework for 3D seismic velocity inversion using the discrete cosine transform (DCT). DCT compresses seismic data and velocity models into a sparse domain to remove redundancy while preserving key structural features. A geometry-adaptive network named SEDCN (Squeeze-and-Excitation Deformable Convolutional Network) is adopted to better capture irregular salt-dome geometries and sharp velocity boundaries. We train and validate the network on 676 samples from the 3D SEG/EAGE salt model, with two schemes for comparison: the proposed DCT-SEDCN and the baseline SEDCN without DCT. Numerical results show that DCT-SEDCN reduces training time by more than 90% and achieves higher PSNR and SSIM than conventional spatiotemporal-domain methods. It effectively suppresses over-smoothing, recovers salt body boundaries and stratigraphic details clearly, and generates geologically more reliable velocity models. This study confirms that DCT-based sparse representation combined with geometry-adaptive deep learning significantly improves the efficiency, accuracy, and robustness of 3D seismic velocity inversion. The framework offers a scalable solution for large-scale 3D FWI and can be extended to elastic/viscoelastic multi-parameter inversion and field data applications.

Joint Bayesian inference of Earth's magnetic field and core surface flow on millennial timescales

2026-03-18T11:04:22Z

Understanding Earth's core dynamics over millennial timescales requires models that jointly describe the evolution of the geomagnetic field and core surface flow, while accommodating the sparse, irregular, and uncertain nature of archaeomagnetic and palaeomagnetic data. We present a new Bayesian core field and core flow modelling framework that utilises archaeo/palaeomagnetic data directly, combining a reduced stochastic representation of core surface dynamics derived from numerical geodynamo statistics with a probabilistic treatment of observational and chronological uncertainties. A key innovation is an efficient discrete marginalisation of age uncertainties, which avoids the convergence difficulties associated with co-estimating ages in high-dimensional Hamiltonian Monte Carlo inversions. The framework aims to reconstruct the coupled evolution of the geomagnetic field and core surface flow over the past 9000 years while preserving dynamical correlations implied by the prior geodynamo time series. Tests using synthetic data generated from an Earth-like geodynamo demonstrate that the method reliably recovers large-scale geomagnetic field variations and key aspects of core dynamics, including long-term westward drift and the evolution of planetary-scale eccentric gyres. These results show that, when combined with physically informed priors, archaeo/palaeomagnetic data can constrain millennial-scale core flow, paving the way for reconstructions based on real data.

Domain-informed explainable boosting machines for trustworthy lateral spread predictions

2026-03-17T22:09:28Z

Explainable Boosting Machines (EBMs) provide transparent predictions through additive shape functions, enabling direct inspection of feature contributions. However, EBMs can learn non-physical relationships that reduce their reliability in natural hazard applications. This study presents a domain-informed framework to improve the physical consistency of EBMs for lateral spreading prediction. Our approach modifies learned shape functions based on domain knowledge. These modifications correct non-physical behavior while maintaining data-driven patterns. We apply the method to the 2011 Christchurch earthquake dataset and correct non-physical trends observed in the original EBM. The resulting model produces more physically consistent global and local explanations, with an acceptable tradeoff in accuracy (4--5\%).

Magnetoconvection in a spherical shell: Equatorial symmetry during the transition from the weak- to the strong-field regime

2026-03-17T16:47:17Z

At small but supercritical Rayleigh numbers, simulations of dynamos in spherical shells often separate into two broad regimes characterised either by their relative magnetic field strength (weak/strong) or by their dominant force balance (VAC/MAC). These regimes can tend smoothly from one to the other but can also be bistable, a phenomenon which occurs particularly at large $\Pm$. We show that in either case the transition correlates with a breaking of equatorial symmetry. Nonlinear simulations of the geodynamo cannot be performed at accurate parameters and hence it is important to ensure that the correct (strong-field) branch is tracked as a distinguished limit is tracked towards a correct parameterisation from the simulations that we can perform. In order to understand the transition to strong-field dynamos, and better understand the mechanisms that occur in both branches, we report on a series of magnetoconvection simulations (that is, with the magnetic field fixed at the outer boundary) with which we bridge the gap between the strong- and weak-field regimes, and show that symmetry-breaking is triggered by the sudden growth of the magnetic field and in turn supports the dynamo in the strong-field regime.

Polarity transitions induced by symmetry-breaking outer boundary heat flux in rapidly rotating dynamos

2026-03-17T10:21:55Z

This study investigates, analytically and numerically, the role of an equatorially anti-symmetric lateral variation in heat flux at the outer boundary in polarity transitions in rapidly rotating dynamos. In an unstably stratified fluid, the frequencies of vertical and horizontal (lateral) buoyancy complement each other such that a polarity transition is induced by the boundary anti-symmetry through the suppression of the slow magnetic-Archimedean-Coriolis (MAC) waves at relatively small vertical buoyant forcing. A dipole-dominated dynamo in the low-inertia limit transitions to polarity reversing and multipolar states in succession as the relative intensity of horizontal buoyancy is progressively increased for a fixed vertical buoyancy. In the same parameter space, an equatorially symmetric heat flux variation does not induce a polarity transition. A composite boundary heterogeneity consisting of comparable magnitudes of symmetric and anti-symmetric variations induces the polarity transition at a horizontal buoyancy of the same order as that for the transition induced by a purely anti-symmetric variation. This makes the analysis relevant to Earth's core, which convects in response to a composite heat flux variation in the lowermost mantle. A heterogeneity with a dominant equatorially symmetric component does not favour polarity transitions, likely producing long periods in Earth's history without reversals. The fact that compositional buoyancy is much stronger than thermal buoyancy, together with the known order of magnitude of the peak field intensity in the inertia-free limit, indicates a large lower-mantle heat flux heterogeneity of O(10) times the mean superadiabatic heat flux at the core--mantle boundary.