Pith. sign in

REVIEW 3 major objections 6 minor 78 references

Neural machine translation of seismic waves for petrophysical inversion

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A Transformer language model treats seismic dispersion curves as a translation task, decoding them into textual descriptions of soil layers, compaction, and water-table depth, and produces petrophysical inversions that match conventional…

desk verdict A genuinely new text-based petrophysical inversion with a real field deployment, but the headline accuracy is self-consistency, not external validation. read the letter →

arxiv 2411.13491 v1 pith:QNR3DKLJ submitted 2024-11-20 physics.geo-ph

José Cunha Teixeira , Ludovic Bodet , Agnès Rivière , Santiago G. Solazzi , Amélie Hallier , Alexandrine Gesret , Sanae El Janyani , Marine Dangeard
show 2 more authors
Amine Dhemaied Joséphine Boisson Gaboriau
This is my paper · ORCID
classification physics.geo-ph
keywords petrophysicalinversionneuralmachinetranslationTransformersurfacewavedispersionpassiveseismicmonitoringwatertableestimationsinkholehazardsynthetictrainingdata
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper aims to show that petrophysical inversion—turning seismic velocity measurements into descriptions of soil nature, compaction, and water-table depth—can be treated as a machine-translation problem. The authors introduce Silex, a Transformer encoder-decoder model that takes a Rayleigh-wave dispersion curve (phase velocity versus frequency) and outputs a structured sentence describing the water-table level and up to four soil layers, each with a soil type, thickness, and a compaction number. Trained on more than a million synthetic curves generated by an adapted rock-physics model, Silex is validated on a railway embankment in France: its recomputed curves match the observed ones to 8 percent normalized RMSE, comparable to a conventional stochastic inversion, while inverting an entire daily geophone line in about 2 minutes instead of 65 hours. The practical stake is that daily, spatially dense monitoring of subsurface conditions relevant to sinkhole hazard becomes feasible.

What carries the argument

The load-bearing object is Silex, a Transformer encoder-decoder whose encoder accepts numerical dispersion curves through a convolutional embedding instead of a word embedding, and whose decoder generates tokens subject to a grammar that allows only valid next tokens (e.g., after the soil-type token only a thickness token is allowed). The physics enters through the training data generator: a multi-layer rock-physics model that combines an equilibrium water-retention profile, contact-elastic effective stress, fluid-substitution relations, and a layer-matrix wave propagator to compute the dispersion curve for each text-described soil column. This generator defines the vocabulary the model can ever output, so the forward model is doing the petrophysical inversion conceptually; the network learns its inverse as a translation.

What would settle it

Measure the dispersion curves and drill-core lithology at a second, independent railway site, run the trained Silex without retraining, and compare its predicted layer boundaries, soil types, and water-table depth to the cores and a piezometer: a systematic mismatch that exceeds the claimed 8% NRMSE or a water-table bias larger than one depth step (0.5 m) would falsify the claim that the synthetic training distribution generalizes to real embankments.

Watch

Extended reading notes

Core claim

Silex deterministically translates dispersion curves in the 15–50 Hz band into a fixed-vocabulary textual description: water table depth, then one to four layers, each with soil type, thickness, and average number of contacts per particle $N$. The mapping is learned entirely from synthetic training pairs produced by a multi-layer adaptation of a capillary-force rock-physics model, not from field measurements. On the study site, the inferred water-table levels track two piezometers through seasonal cycles (with a slight ~0.5 m overestimation at one of them), and the inferred soil types largely match borehole logs, though the gypsum substratum is not resolved. The paper's quantitative validation is the average 12 m/s RMSE (8% normalized) between input dispersion curves and curves recomputed from Silex's outputs, against 9 m/s for the conventional neighborhood-algorithm inversion used as benchmark.

Load-bearing premise

The load-bearing premise is that the synthetic training set, generated by the adapted rock-physics model with one to four homogeneous soil layers above a fixed rigid substratum, faithfully represents the real railway embankment's geology and saturation physics; if it does not, Silex can only express the vocabulary its synthetic training data contains.

Editorial extensions

If this is right

  • Silex can produce daily, per-geophone petrophysical sections across five 123-m lines at about two minutes per line on a standard CPU, a 2,000-fold speedup over the neighborhood-algorithm inversion used as benchmark.
  • The inferred water-table levels track seasonal rainfall and match piezometer measurements closely enough to support hydrogeological monitoring, with a residual overestimation of about 0.5 m at one piezometer.
  • The petrophysical outputs can be converted into drained shear modulus, making the method a structural-health indicator; no emerging weak zones were detected in the study period, consistent with the absence of observed sinkhole development.
  • The dispersion-curve recomputation error of 12 m/s (8% NRMSE) is close to the 9 m/s of the conventional method, indicating that the deterministic translation does not sacrifice accuracy for speed.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The translation design is not limited to surface-wave dispersion: the same encoder-decoder with a numerical-input encoder could ingest other geophysical observables (e.g., full waveforms or resistivity soundings) and emit the same textual soil vocabulary, giving a multi-physics inversion framework.
  • Because failure modes are determined by the training vocabulary, adding rock lithofacies and sharper impedance contrasts to the forward model—acknowledged in the paper as a limitation—would probably extend reliable inversion below 15 m more than any architectural change.
  • The 2,000x speedup turns inversion from a batch analysis into a near-real-time monitoring stream; one immediate use would be alarm generation when inferred water-table or compaction changes approach thresholds known to precede sinkhole collapse.
  • A straightforward way to address the paper's noted lack of uncertainty is to train an ensemble of Silex models with different seeds and treat the spread of output tokens as a proxy for prediction confidence, without the 65-hour stochastic sampler.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. Teixeira et al. present Silex, a Transformer-based encoder-decoder language model that maps Rayleigh-wave dispersion curves (numerical sequences of phase velocity versus frequency) to structured textual descriptions of near-surface soil layers (soil type, thickness, contact number N) and water-table (WT) depth. Training data are generated with a multi-layer adaptation of the Solazzi et al. (2021) rock-physics model, using one to four homogeneous soil layers from a vocabulary of four soil types over a fixed rigid substratum, with WT sampled from 0.5 to 10 m. The model is applied to daily passive-MASW data from five geophone lines at a railway site in France over 2020-2023, producing 2D/3D petrophysical sections and WT time series. The authors report an 81% token accuracy on a synthetic test set, an 8% NRMSE (12 m/s RMSE) computed by recomputing dispersion curves with the same rock-physics model used for training, a 2,000-fold speedup relative to a conventional neighbourhood-algorithm inversion, and qualitative agreement with two piezometers and two borehole logs. The paper emphasizes the method's potential for daily, high-resolution subsurface monitoring for sinkhole hazard assessment.

Significance. If the accuracy claims were independently validated, this would be a valuable contribution: it demonstrates a practical, deterministic, extremely fast alternative to stochastic petrophysical inversion, with the novel use of a transformer architecture to produce interpretable textual outputs. The paper is refreshingly explicit about limitations (no uncertainty quantification, rock-physics model cannot represent rock or gravel, gypsum substratum undetected) and makes code/data available on Zenodo. However, the central quantitative claim (8% NRMSE, 'closely rivaling conventional inversion') is not established by the reported metrics because the evaluation is circular, and the borehole validation shows systematic vocabulary mismatches. The speed advantage and the qualitative WT tracking are credible, but the paper's abstract overstates the accuracy and breadth of the inferred properties.

major comments (3)
  1. [Materials and Methods, Model evaluation] The headline 8% NRMSE and 12 m/s RMSE are computed by recomputing dispersion curves with the same rock-physics model (Solazzi et al., 2021) used to generate the training data, and the comparison with conventional inversion (12 m/s vs 9 m/s) uses the same recomputation procedure. This metric therefore measures internal consistency with the training generator, not accuracy against ground truth, and it does not support the claim that Silex 'closely rival[s]' conventional stochastic seismic inversion. Please report the misfit of the recomputed curves to the observed dispersion curves with proper uncertainty (e.g., using Eq. S24), and quantify agreement with independent measurements (piezometers, borehole lithology) rather than relying on the circular RMSE.
  2. [Results, Water table level; Table S2; Materials and Methods, Modeling parameters] There is an internal contradiction in the water-table vocabulary: the Results state WT levels range from 0.5 to 19.5 m, but Table S2 caps the allowed WT tokens at 10.0 m, and the Modeling parameters section says WT was sampled between 0.5 and 10 m. If the actual training vocabulary is 0.5-10 m, the model cannot make predictions below 10 m, so claims about deeper WT accuracy are unsupported; if 19.5 m is intended, the training data and table must be corrected. This must be resolved before the WT results can be assessed.
  3. [Results, Petrophysical and mechanical properties; Rock-physics model limitation] The borehole validation is partly negative: at DR1 and DR2, Silex predicts loam where drilling shows sandy clay and sand where drilling shows gravelly sand, and the gypsum substratum is not detected. The paper acknowledges these limitations, but the abstract claims that the method 'successfully delineates' soil nature. Because the training vocabulary is restricted to four soil types (sand, loam, silt, clay) and the rock-physics model cannot represent gravel or rock, the model cannot retrieve the actual lithologies at the site. Please either temper the claims to reflect the vocabulary-limited nature of the output or provide additional validation (e.g., classification against a broader borehole dataset) that quantifies the practical impact of these discrepancies.
minor comments (6)
  1. [Fig. 2D caption] The caption says 'September 31, 2023' which is not a valid date; it should be 'September 30, 2023'.
  2. [Eq. S22] The inline fraction in the RMSE definition is not rendered correctly; please fix the LaTeX so the sum and division are unambiguous.
  3. [Fig. S3] The label 'posority' in the figure should be 'porosity'.
  4. [Supplementary Text, Variability] The text states that spatial boundaries are 'variable over 10 cm (the depth increment)', which appears inconsistent with the 1 m thickness step in Table S1; please clarify the depth increment used in the sectional visualization.
  5. [References] References 5 and 9 are cited as 'in rev.'; if journal-accepted versions are available, please update the citations to their final publication details.
  6. [Results, Water table level] The Savitzky-Golay filter window sizes (10.75 m spatial, 137 days temporal) are introduced without justification; a brief explanation of how these values were chosen would help readers assess the smoothness of the WT products.

Circularity Check

1 steps flagged · score 4.0 of 10

The headline 8% NRMSE is a self-consistency check through the same rock-physics model that generated the training labels; independent field checks provide only partial support.

  1. fitted input called prediction [Results, 'Accuracy and error'; Materials and Methods, 'Model evaluation']
    "Silex was evaluated on 50,000 synthetic samples outside the training dataset, achieving an accuracy on the inferred tokens of 81%. Additionally, using Silex inference results, DCs were recomputed with the same rock-physics model used to generate the training data ... The average root mean square error (RMSE) on the DCs was estimated at 12 m/s (normalized RMSE of 8%). ... DCs were recomputed from Silex’s inferences using the same rock-physics model (41), allowing for a direct comparison with the input DCs."

    The supervised labels in training (petrophysical descriptions mapped to DCs) and the evaluation reference (DCs recomputed from Silex output) are both produced by the same adapted Solazzi et al. rock-physics model. The 81% token accuracy is measured on synthetic samples drawn from that same generator, and the 8% NRMSE is a cycle-consistency metric: it tests whether F(Silex(DC)) approximates the input DC, not whether the inferred soil type, N, thickness, or water-table level matches ground truth. Any model that learned a good inverse of F would score well even if F misrepresents the site. The paper itself states that the model cannot represent the gypsum substratum and that the vocabulary excludes gravel, so the 8% figure cannot validate real petrophysical accuracy.

full rationale

The paper's central inversion result is not entirely circular: Silex is applied to real field dispersion curves, and its water-table estimates are compared with piezometer data, its soil descriptions with boreholes, and its VS sections with a conventional neighborhood-algorithm inversion. These external checks give the central method independent content. However, the headline validation metric is substantially self-referential: training labels and the recomputed-DC RMSE both come from the same rock-physics model (Solazzi et al., 2021, ref. 41) used to generate the training data. The 8% NRMSE therefore measures consistency with the generative forward model, not agreement with independent subsurface truth. The paper's own 'Rock-physics model limitation' section concedes that the model is designed for soils rather than consolidated rock and cannot represent the gypsum substratum; the Results also note that at DR1 and DR2 the inferred loam/sand differ from the drilled sandy clay and gravelly sand. Additionally, there is an internal inconsistency in the water-table vocabulary: Results state WT levels range from 0.5 to 19.5 m, while Table S2 and Modeling parameters cap WT tokens at 10.0 m. This is a correctness risk rather than a circularity. No load-bearing uniqueness theorem or ansatz is smuggled in through self-citation; the rock-physics model is described in detail and externally published. Weighing these factors, the paper has one partially circular validation step, but the central claim retains independent support, so the circularity score is 4 rather than 0 or 6.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the fidelity of the synthetic forward model, the fixed vocabulary of possible outputs, and the undisclosed proprietary processing that produces the field input curves. The only fitted quantities introduced by the paper are the neural network weights and the smoothing windows, neither of which is calibrated against independent field measurements.

free parameters (3)
  • Silex transformer weights = 499,266 parameters trained on 1,166,446 synthetic samples with sparse categorical cross-entropy
    The inversion output depends entirely on these fitted weights. They are optimized on synthetic data generated by the forward model, and no independent calibration to field measurements is reported.
  • Fixed substratum sub-layer properties = Layer 1: thickness 10 m, VP 1500 m/s, VS 750 m/s, rho 2000 kg/m3; half-space: VP 4000 m/s, VS 2000 m/s, rho 2500 kg/m3
    These constants are imposed below 20 m in the forward model (Table S5). They shape all synthetic dispersion curves and therefore what Silex can infer at depth, but they are not constrained by field data.
  • Savitzky-Golay smoothing windows = Spatial window 10.75 m, temporal window 137 days
    Applied to water-table estimates before display. The 137-day temporal window smooths short-term variation and affects the visual validation against piezometers.
assumptions (5)
  • domain assumption The rock-physics forward model of Solazzi et al. (2021) accurately predicts real dispersion curves from soil properties.
    Training data are generated exclusively with this model, and the error metric recomputes dispersion curves with the same model, so its fidelity to the field site is assumed, not demonstrated. Introduced in Materials and Methods, Training data generation.
  • domain assumption Richards equilibrium and van Genuchten saturation profile describe the vadose zone at the site.
    Equations S1 to S6 assign saturation versus depth for the synthetic training data, assuming static water and no hysteresis in the vadose zone.
  • domain assumption The Thomson-Haskell propagator with the fundamental mode only, plus two fixed sublayers, approximates the measured dispersion.
    Used to compute synthetic dispersion curves from the rock-physics model. The authors acknowledge mode misidentification in field data and that high-order modes are ignored.
  • ad hoc to paper The subsurface is representable as one to four homogeneous soil layers, each of four soil types (sand, loam, silt, clay) with a constant contact number N.
    This vocabulary limits what Silex can output. Gypsum and rock cannot be represented, and the authors note the substratum remains undetected. Defined in Table S1 and the main text Petrophysical and mechanical properties.
  • domain assumption Passive-MASW processing yields reliable daily dispersion curves from train-induced noise.
    The input curves are produced by Sercel's proprietary processing; the authors state the specific segment-selection method used by Sercel is not known to them, so the quality and repeatability of the inputs rest on an external, partially undisclosed pipeline.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Neural machine translation of seismic waves for petrophysical inversion." pith.science (2026). https://pith.science/paper/QNR3DKLJ

@misc{pith2026241113491,
  author       = {Pith},
  title        = {Pith review of: Neural machine translation of seismic waves for petrophysical inversion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QNR3DKLJ}},
  note         = {Machine review of arXiv:2411.13491}
}
read the original abstract

Effective structural assessment of urban infrastructure is essential for sustainable land use and resilience to climate change and natural hazards. Seismic wave methods are widely applied in these areas for subsurface characterization and monitoring, yet they often rely on time-consuming inversion techniques that fall short in delivering comprehensive geological, hydrogeological, and geomechanical descriptions. Here, we explore the effectiveness of a passive seismic approach coupled with artificial intelligence (AI) for monitoring geological structures and hydrogeological conditions in the context of sinkhole hazard assessment. We introduce a deterministic petrophysical inversion technique based on a language model that decodes seismic wave velocity measurements to infer soil petrophysical and mechanical parameters as textual descriptions. Results successfully delineate 3D subsurface structures with their respective soil nature and mechanical characteristics, while accurately predicting daily water table levels. Validation demonstrates high accuracy, with a normalized root mean square error of 8%, closely rivaling with conventional stochastic seismic inversion methods, while delivering broader insights into subsurface conditions 2,000 times faster. These findings underscore the potential of advanced AI techniques to significantly enhance subsurface characterization across diverse scales, supporting decision-making for natural hazard mitigation.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

78 extracted references · 48 canonical work pages

  1. [1]

    K. Aki, P. G. Richards, Quantitative Seismology, Theory and Methods. Geological Magazine 118 (2), 208 (1981), doi:10.1017/S0016756800034439

    work page doi:10.1017/s0016756800034439 1981

  2. [2]

    Voisin, S

    C. Voisin, S. Garambois, C. Massey, R. Brossier, Seismic noise monitoring of the water table in a deep-seated, slow-moving landslide. Interpretation 4 (3), SJ67–SJ76 (2016), doi: 10.1190/INT-2016-0010.1

    work page doi:10.1190/int-2016-0010.1 2016

  3. [3]

    S. Mao, A. Lecointre, R. van der Hilst, M. Campillo, Space-time monitoring of groundwater fluctuations with passive seismic interferometry. Nature Communications 13, 4643 (2022), doi:10.1038/s41467-022-32194-3

    work page doi:10.1038/s41467-022-32194-3 2022

  4. [4]

    Shen, et al., Fiber-optic seismic sensing of vadose zone soil moisture dynamics

    Z. Shen, et al., Fiber-optic seismic sensing of vadose zone soil moisture dynamics. Nature Communications 15 (1), 6432 (2024), doi:10.1038/s41467-024-50690-6

    work page doi:10.1038/s41467-024-50690-6 2024

  5. [5]

    Cunha Teixeira, et al., Physics-guided deep learning model for daily groundwater table maps estimation using passive surface-wave dispersion

    J. Cunha Teixeira, et al., Physics-guided deep learning model for daily groundwater table maps estimation using passive surface-wave dispersion. ESS Open Archive (in rev.), doi:10.22541/ essoar.171322609.99979575/v1

    work page arXiv

  6. [6]

    Bardainne, C

    T. Bardainne, C. Cai, T. Rebert, R. Tarnus, T. Allemand, Passive Seismic Monitoring Using Trains as Sources to Characterize Near-Surface and Prevent Sinkholes, in European Asso- ciation of Geoscientists and Engineers , vol. 2023 (2023), pp. 1–5, doi:10.3997/2214-4609. 2023101262

    doi:10.3997/2214-4609 2023

  7. [7]

    Rebert, C

    T. Rebert, C. Cai, A. Hallier, T. Bardainne, Rockfall alarm system for railway monitoring: Integrating seismic detection, localization, and characterization. Geophysics 89 (1), KS13– KS23 (2024), doi:10.1190/geo2023-0058.1

    work page doi:10.1190/geo2023-0058.1 2024

  8. [8]

    Bardainne, et al., Stimulated Noise and Surface Wave Interferometry Processing for Hybrid Seismic Imaging, in European Association of Geoscientists and Engineers, vol

    T. Bardainne, et al., Stimulated Noise and Surface Wave Interferometry Processing for Hybrid Seismic Imaging, in European Association of Geoscientists and Engineers, vol. 2023 (2023), pp. 1–5, doi:10.3997/2214-4609.202320206. 16

    arXiv 2023

Show all 78 references
  1. [9]

    Cunha Teixeira, et al., Nondestructive testing of railway embankments by measuring multi- modal dispersion of surface-waves induced by high-speed trains with linear geophone arrays

    J. Cunha Teixeira, et al., Nondestructive testing of railway embankments by measuring multi- modal dispersion of surface-waves induced by high-speed trains with linear geophone arrays. Seismica (in rev.)

  2. [10]

    Tarantola, Inverse problem theory - and methods for model parameter estimation

    A. Tarantola, Inverse problem theory - and methods for model parameter estimation. (SIAM) (2005), doi:10.1137/1.9780898717921

    2005 doi

  3. [11]

    M. K. Sen, P. L. Stoffa, Global Optimization Methods in Geophysical Inversion (Cambridge University Press) (2013), doi:10.1017/CBO9780511997570

    2013 doi

  4. [12]

    Grana, L

    D. Grana, L. Azevedo, L. de Figueiredo, P. Connolly, T. Mukerji, Probabilistic inversion of seismic data for reservoir petrophysical characterization: Review and examples. Geophysics 87 (5), M199–M216 (2022), doi:10.1190/geo2021-0776.1

    2022 doi

  5. [13]

    LeCun, Y

    Y. LeCun, Y. Bengio, G. Hinton, Deep Learning. Nature 521, 436–44 (2015), doi:10.1038/ nature14539

    2015

  6. [14]

    Goodfellow, Y

    I. Goodfellow, Y. Bengio, A. Courville, Deep Learning (MIT Press) (2016)

    2016

  7. [15]

    Mousavi, G

    S. Mousavi, G. Beroza, Deep-learning seismology. Science 377 (2022), doi:10.1126/science. abm4470

    2022 doi

  8. [16]

    Araya-Polo, J

    M. Araya-Polo, J. Jennings, A. Adler, T. Dahlke, Deep-learning tomography.The Leading Edge 37 (1), 58–66 (2018), doi:10.1190/tle37010058.1

    2018 doi

  9. [17]

    Weinzierl, B

    W. Weinzierl, B. Wiese, Deep learning a poroelastic rock-physics model for pressure and satu- ration discrimination. Geophysics 86 (1), MR53–MR66 (2021), doi:10.1190/geo2020-0049.1

    2021 doi

  10. [18]

    Li, et al., Deep-Learning Inversion of Seismic Data

    S. Li, et al., Deep-Learning Inversion of Seismic Data. IEEE Transactions on Geoscience and Remote Sensing 58 (3), 2135–2149 (2020), doi:10.1109/TGRS.2019.2953473

    2020

  11. [19]

    A. P. O. Muller, et al., Deep pre-trained FWI: where supervised learning meets the physics- informed neural networks. Geophysical Journal International 235 (1), 119–134 (2023), doi: 10.1093/gji/ggad215

    2023 doi

  12. [20]

    V. Das, T. Mukerji, Petrophysical properties prediction from prestack seismic data using con- volutional neural networks. Geophysics 85, 1–85 (2020), doi:10.1190/geo2019-0650.1. 17

    2020 doi

  13. [21]

    Zhang, Z

    G. Zhang, Z. Wang, Y. Chen, Deep learning for seismic lithology prediction. Geophysical Journal International 215 (2), 1368–1387 (2018), doi:10.1093/gji/ggy344

    2018 doi

  14. [22]

    X. Chen, J. Xia, J. Pang, C. Zhou, B. Mi, Deep learning inversion of Rayleigh-wave disper- sion curves with geological constraints for near-surface investigations. Geophysical Journal International 231 (1), 1–14 (2022), doi:10.1093/gji/ggac171

    2022 doi

  15. [23]

    Talarico, W

    E. Talarico, W. Le ˜ao Neto, D. Grana, Comparison of Recursive Neural Network and Markov Chain Models in Facies Inversion. Mathematical Geosciences 53 (2021), doi: 10.1007/s11004-020-09914-w

    2021 doi

  16. [24]

    F. Liu, J. Li, L. Fu, L. Lu, Multimodal surface wave inversion with automatic differentiation. Geophysical Journal International 238 (1), 290–312 (2024), doi:10.1093/gji/ggae155

    2024 doi

  17. [25]

    Zhang, Y

    Z. Zhang, Y. Lin, Data-Driven Seismic Waveform Inversion: A Study on the Robustness and Generalization. IEEE Transactions on Geoscience and Remote Sensing PP, 1–14 (2020), doi: 10.1109/TGRS.2020.2977635

    2020

  18. [26]

    Vaswani, et al., Attention Is All You Need (2017), doi:10.48550/arXiv.1706.03762

    A. Vaswani, et al., Attention Is All You Need (2017), doi:10.48550/arXiv.1706.03762

    2017 doi arXiv

  19. [27]

    T. B. Brown, et al., Language Models are Few-Shot Learners. CoRR abs/2005.14165 (2020), doi:10.48550/arXiv.2005.14165

    2020 doi arXiv

  20. [28]

    Hadid, T

    A. Hadid, T. Chakraborty, D. Busby, When geoscience meets generative AI and large language models: Foundations, trends, and future challenges. Expert Systems 41 (10), e13654 (2024), doi:10.1111/exsy.13654

    2024 doi

  21. [29]

    C. Ning, B. Wu, Z. Zhu, Transformer and CNN Hybrid Neural Network for Seismic Impedance Inversion, in IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Sym- posium (2023), pp. 5543–5546, doi:10.1109/IGARSS52108.2023.10281611

    2023

  22. [30]

    Y. Dou, K. Li, 3D seismic mask auto encoder: Seismic inversion using transformer-based reconstruction representation learning. Computers and Geotechnics 169, 106194 (2024), doi: 10.1016/j.compgeo.2024.106194. 18

    2024

  23. [31]

    Wang, et al., Seismic velocity inversion transformer.Geophysics 88 (4), R513–R533 (2023), doi:10.1190/geo2022-0283.1

    H. Wang, et al., Seismic velocity inversion transformer.Geophysics 88 (4), R513–R533 (2023), doi:10.1190/geo2022-0283.1

    2023 doi

  24. [32]

    Bardaine, B

    T. Bardaine, B. Rondeleux, Method and device for monitoring the subsoil of the earth un- der a target zone (2018), https://patents.google.com/patent/WO2020021177A1/en? inventor=Bardainne&oq=Bardainne

    2018

  25. [33]

    T. Bardainne, et al., Permanent passive seismic monitoring of the near-surface ground beneath railways using trains as sources, in Fifth International Conference on Railway Technology: Research, Development and Maintenance, vol. 1 (2022), p. 14, doi:10.4203/ccc.1.10.14

    2022 doi

  26. [34]

    Tarnus, T

    R. Tarnus, T. Bardainne, L. Michel, N. Deladerri `ere, C. Cai, A case study for railway un- derground imaging using trains as seismic signal for sinkhole and subsidence phenomena prevention, in Fifth International Conference on Railway Technology: Research, Development and Mai...

    2022 doi

  27. [35]

    Tarnus, et al., A Case Study for Underground Imaging Using Trains as Seismic Signal to Investigate Subsidence Phenomena, in European Association of Geoscientists and Engineers, vol

    R. Tarnus, et al., A Case Study for Underground Imaging Using Trains as Seismic Signal to Investigate Subsidence Phenomena, in European Association of Geoscientists and Engineers, vol. 2022 (2022), pp. 1–5, doi:10.3997/2214-4609.202220033

    2022

  28. [36]

    Lavou ´e, et al

    F. Lavou ´e, et al. , Understanding Seismic Waves Generated by Train Traffic via Modeling: Implications for Seismic Imaging and Monitoring. Seismological Research Letters 92 (1), 287–300 (2020), doi:10.1785/0220200133

    2020 doi

  29. [37]

    Rebert, T

    T. Rebert, T. Bardainne, T. Allemand, C. Cai, H. Chauris, Characterization of train kinematics and source wavelets from near-field seismic data. Geophysical Journal International 237 (2), 697–715 (2024a), doi:10.1093/gji/ggae067

    2024 doi

  30. [38]

    Sambridge, Geophysical inversion with a neighbourhood algorithm—I

    M. Sambridge, Geophysical inversion with a neighbourhood algorithm—I. Searching a pa- rameter space. Geophysical Journal International 138 (2), 479–494 (1999), doi:10.1046/j. 1365-246X.1999.00876.x

    1999

  31. [39]

    Mavko, T

    G. Mavko, T. Mukerji, J. Dvorkin, The Rock Physics Handbook: Tools for Seismic Analysis of Porous Media (Cambridge University Press, Cambridge), 2nd ed. (2009). 19

    2009

  32. [40]

    Baevski, H

    A. Baevski, H. Zhou, A. Mohamed, M. Auli, wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations (2020), doi:10.48550/arXiv.2006.11477

    2020 doi arXiv

  33. [41]

    S. G. Solazzi, L. Bodet, K. Holliger, D. Jougnot, Surface-Wave Dispersion in Partially Saturated Soils: The Role of Capillary Forces. Journal of Geophysical Research: Solid Earth 126 (12), e2021JB022074 (2021), doi:10.1029/2021JB022074

    2021 doi

  34. [42]

    M. T. van Genuchten, A Closed-form Equation for Predicting the Hydraulic Conductivity of Unsaturated Soils. Soil Science Society of America Journal 44 (5), 892–898 (1980), doi: 10.2136/sssaj1980.03615995004400050002x

    1980

  35. [43]

    R. F. Carsel, R. S. Parrish, Developing joint probability distributions of soil water re- tention characteristics. Water Resources Research 24 (5), 755–769 (1988), doi:10.1029/ WR024i005p00755

    1988

  36. [44]

    Foti, et al

    S. Foti, et al. , Guidelines for the good practice of surface wave analysis: a product of the InterPACIFIC project. Bulletin of Earthquake Engineering 16 (2018), doi:10.1007/ s10518-017-0206-7

    2018

  37. [45]

    R. D. Mindlin, Compliance of Elastic Bodies in Contact. Journal of Applied Mechanics16 (3), 259–268 (1949), doi:10.1115/1.4009973

    1949 doi

  38. [46]

    Savitzky, M

    A. Savitzky, M. J. E. Golay, Smoothing and Differentiation of Data by Simplified Least Squares Procedures. Analytical Chemistry 36 (8), 1627–1639 (1964), doi:10.1021/ac60214a047

    1964 doi

  39. [47]

    Wathelet, An improved neighborhood algorithm: Parameter conditions and dynamic scaling

    M. Wathelet, An improved neighborhood algorithm: Parameter conditions and dynamic scaling. Geophysical Research Letters35 (9) (2008), doi:10.1029/2008GL033256

    2008 doi

  40. [48]

    Pasquet, L

    S. Pasquet, L. Bodet, SWIP: An integrated workflow for surface-wave dispersion inversion and profiling. Geophysics 82 (6), WB47–WB61 (2017), doi:10.1190/geo2016-0625.1

    2017 doi

  41. [49]

    Moufad, et al., Variational Diffusion Posterior Sampling with Midpoint Guidance (2024), doi:10.48550/arXiv.2410.09945

    B. Moufad, et al., Variational Diffusion Posterior Sampling with Midpoint Guidance (2024), doi:10.48550/arXiv.2410.09945

    2024 doi arXiv

  42. [50]

    Cunha Teixeira,et al., Neural machine translation of seismic waves for petrophysical inversion (2024), doi:10.5281/zenodo.14192468

    J. Cunha Teixeira,et al., Neural machine translation of seismic waves for petrophysical inversion (2024), doi:10.5281/zenodo.14192468. 20

    2024 doi

  43. [51]

    C. B. Park, R. D. Miller, Roadside Passive Multichannel Analysis of Surface Waves (MASW). Journal of Environmental and Engineering Geophysics 13 (1), 1–11 (2008), doi:10.2113/ JEEG13.1.1

    2008

  44. [52]

    D. A. Quiros, L. D. Brown, D. Kim, Seismic interferometry of railroad induced ground motions: body and surface wave imaging. Geophysical Journal International 205 (1), 301–313 (2016), doi:10.1093/gji/ggw033

    2016 doi

  45. [53]

    Cheng, J

    F. Cheng, J. Xia, Y. Xu, Z. Xu, Y. Pan, A new passive seismic method based on seismic interferometry and multichannel analysis of surface waves.Journal of Applied Geophysics117 (2015), doi:10.1016/j.jappgeo.2015.04.005

    2015 doi

  46. [54]

    Cheng, et al., Multichannel analysis of passive surface waves based on crosscorrelations

    F. Cheng, et al., Multichannel analysis of passive surface waves based on crosscorrelations. Geophysics 81 (5), EN57–EN66 (2016), doi:10.1190/geo2015-0505.1

    2016 doi

  47. [55]

    Mi, et al., Near-surface imaging from traffic-induced surface waves with dense linear arrays: An application in the urban area of Hangzhou, China

    B. Mi, et al., Near-surface imaging from traffic-induced surface waves with dense linear arrays: An application in the urban area of Hangzhou, China. Geophysics 87 (2), B145–B158 (2022), doi:10.1190/geo2021-0184.1

    2022 doi

  48. [56]

    Czarny, T

    R. Czarny, T. Zhu, J. Shen, Spatiotemporal evaluation of Rayleigh surface wave estimated from roadside dark fiber DAS array and traffic noise.Seismica 2 (22) (2023), doi:10.26443/seismica. v2i2.247

    2023 doi

  49. [57]

    Rezaeifar, et al., Imaging shallow structures using interferometry of seismic body waves generated by train traffic

    M. Rezaeifar, et al., Imaging shallow structures using interferometry of seismic body waves generated by train traffic. Geophysical Journal International 233 (2), 964–977 (2023), doi: 10.1093/gji/ggac507

    2023 doi

  50. [58]

    B. You, B. Mi, B. Guan, H. Zhang, Y. Liu, High-quality surface wave retrieval from vibrations generated by high-speed trains moving on viaducts.Journal of Applied Geophysics212, 105005 (2023), doi:10.1016/j.jappgeo.2023.105005

    2023

  51. [59]

    G. D. Bensen, et al., Processing seismic ambient noise data to obtain reliable broad-band surface wave dispersion measurements.Geophysical Journal International169 (3), 1239–1260 (2007), doi:10.1111/j.1365-246X.2007.03374.x. 21

    2007 arXiv

  52. [60]

    Bohlen, S

    T. Bohlen, S. Kugler, G. Klein, F. Theilen, 1.5D inversion of lateral variation of Scholte-wave dispersion. Geophysics 69 (22), 330–344 (2004), doi:10.1190/1.1707052

    2004 doi

  53. [61]

    Socco, C

    L. Socco, C. Strobbia, Surface-wave method for near-surface characterization: a tutorial. Near Surface Geophysics 2 (4), 165–185 (2004), doi:10.3997/1873-0604.2004015

    2004

  54. [62]

    L. V. Socco, S. Foti, D. Boiero, Surface-wave analysis for building near-surface velocity models — Established approaches and new perspectives. Geophysics 75 (5), 75A83–75A102 (2010), doi:10.1190/1.3479491

    2010 doi

  55. [63]

    Bergamo, D

    P. Bergamo, D. Boiero, L. V. Socco, Retrieving 2D structures from surface-wave data by means of space-varying spatial windowing. Geophysics 77 (4), EN39–EN51 (2012), doi: 10.1190/geo2012-0031.1

    2012 doi

  56. [64]

    Rebert, T

    T. Rebert, T. Allemand, T. Bardainne, C. Cai, H. Chauris, Seismic emissions from a passing train: turning ambient noise into a controlled source, in EGU General Assembly Conference Abstracts, EGU General Assembly Conference Abstracts (2023), pp. EGU–7289, doi:10.5194/ eguspher...

    2023

  57. [65]

    Cheng, J

    F. Cheng, J. Xia, Z. Xu, Y. Hu, B. Mi, Frequency–Wavenumber (FK)-Based Data Selection in High-Frequency Passive Surface Wave Survey.Surveys in Geophysics39 (4), 661–682 (2018), doi:10.1007/s10712-018-9473-3

    2018 doi

  58. [66]

    Cheng, J

    F. Cheng, J. Xia, M. Behm, Y. Hu, J. Pang, Automated Data Selection in the Tau–p Domain: Application to Passive Surface Wave Imaging.Surveys in Geophysics 40 (2019), doi:10.1007/ s10712-019-09530-2

    2019

  59. [67]

    Ning, et al., High-Frequency Surface-Wave Imaging from Traffic-Induced Noise by Se- lecting In-line Sources

    L. Ning, et al., High-Frequency Surface-Wave Imaging from Traffic-Induced Noise by Se- lecting In-line Sources. Surveys in Geophysics 43 (6), 1873–1899 (2022), doi:10.1007/ s10712-022-09723-2

    2022

  60. [68]

    C. B. Park, R. D. Miller, J. Xia, Multichannel analysis of surface waves. Geophysics 64 (3), 800–808 (1999), doi:10.1190/1.1444590. 22

    1999 doi

  61. [69]

    Nazarian, K

    S. Nazarian, K. Stokoe, W. Hudson, Use of Spectral Analysis of Surface Waves Method for Determination of Moduli and Thicknesses of Pavement Systems. Transportation Research Record (1983)

    1983

  62. [70]

    A. W. Bishop, G. E. Blight, Some Aspects of Effective Stress in Saturated and Partly Saturated Soils. G´eotechnique 13 (3), 177–197 (1963), doi:10.1680/geot.1963.13.3.177

    1963 doi

  63. [71]

    L. A. Richards, Capillary Conduction of Liquids through Porous Mediums. Physics 1 (5), 318–333 (1931), doi:10.1063/1.1745010

    1931 doi

  64. [72]

    Hill, The Elastic Behaviour of a Crystalline Aggregate.Proceedings of the Physical Society

    R. Hill, The Elastic Behaviour of a Crystalline Aggregate.Proceedings of the Physical Society. Section A 65 (5), 349 (1952), doi:10.1088/0370-1298/65/5/307

    1952 doi

  65. [73]

    M. A. Biot, Mechanics of Deformation and Acoustic Propagation in Porous Media. Journal of Applied Physics 33, 1482–1498 (1962), doi:10.1063/1.1728759

    1962 doi

  66. [74]

    Gassmann, Elastic Waves Through a Packing of Spheres

    F. Gassmann, Elastic Waves Through a Packing of Spheres. Geophysics 16, 673–685 (1951), doi:10.1190/1.1437718

    1951 doi

  67. [75]

    W. T. Thomson, Transmission of Elastic Waves through a Stratified Solid Medium.Journal of Applied Physics 21 (2), 89–93 (1950), doi:10.1063/1.1699629

    1950 doi

  68. [76]

    N. A. Haskell, The dispersion of surface waves on multilayered media. Bulletin of the Seismo- logical Society of America 43 (1), 17–34 (1953), doi:10.1785/BSSA0430010017

    1953 doi

  69. [77]

    O’Neill, Full-waveform reflectivity for modeling, inversion and appraisal of seismic surface wave dispersion in shallow site investigations , Ph.D

    A. O’Neill, Full-waveform reflectivity for modeling, inversion and appraisal of seismic surface wave dispersion in shallow site investigations , Ph.D. thesis, University of Western Australia (2003). 23 Acknowledgments We thank Thomas Bardainne, Renaud Tarnus, Thibaut Allemand,...

    2003

  70. [2022]

    The shown examples correspond to the red annotations in fig. S17. S30 July, 2023 N L4-X22-Z13 L4-X22-Z13 Soil type July, 2022 NSoil type Average WT level and standard deviation L4-X37-Z6 L4-X37-Z6 Normalized frequency [-] Figure S17: Occurrence normalized frequency sections. N...

    2023

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.