Pith. sign in

REVIEW 3 major objections 5 minor 278 references

A self-supervised MAE pipeline can explore multi-year ocean sound recordings and turn them into usable signal classes with under an hour of human mapping.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-11 01:43 UTC pith:CQTT4JC6

load-bearing objection Solid practical pipeline for low-frequency hydroacoustic exploration; event-level extraction is the real engineering contribution, but its empirical thresholds are under-tested and the F1 gains are only partly attributed. the 3 major comments →

arxiv 2607.07733 v1 pith:CQTT4JC6 submitted 2026-07-07 cs.SD cs.LG

A Self-Supervised Approach for Minimal-Annotation Hydroacoustic Data Exploration

Pierre-Yves Raumer , Axel Marmoret , Dorian Cazau , Anatole Gros-Martial , Richard Dreo , Maelle Torterotot , Sara Bazin , Flore Samaran
show 1 more author
Jean-Yves Royer
This is my paper
classification cs.SD cs.LG
keywords self-supervised learningmasked autoencoderhydroacoustic monitoringevent-level clusteringmarine mammal vocalizationspassive acoustic monitoringspectrogram embeddingsminimal annotation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Long-term hydrophone arrays produce far more continuous low-frequency sound than anyone can label by hand, so most archives stay unexplored. This paper shows that a Masked AutoEncoder trained only to reconstruct masked spectrogram patches can learn useful local features of the soundscape without any labels. Adjacent informative patches are then grouped into event-level embeddings so that overlapping sources inside one window can be separated; those embeddings are clustered across the whole dataset into a few hundred groups. After roughly one hour of visual inspection that maps the groups onto a handful of semantic classes, the same clusters act as approximate detectors whose hourly scores match specialized and supervised baselines on known whale calls and seismic phases, recover the expected seasonal patterns, and surface several previously unstudied signals.

Core claim

On multi-year low-frequency hydroacoustic data, an MAE that learns patch embeddings, followed by within-spectrogram aggregation of neighboring patches into event embeddings and dataset-scale clustering, yields a few hundred clusters that a human can map to meaningful hydroacoustic classes in under an hour; the resulting lightweight classifiers perform comparably to existing detectors and recover both known seasonal activity and previously uncharacterized signals.

What carries the argument

Event-level disentanglement: after the MAE encoder produces a 14 imes14 grid of 256-dimensional patch embeddings, empty patches are discarded by energy and activation thresholds and remaining adjacent patches are merged by Chebyshev distance ≤3, so each spectrogram yields separate embeddings for co-occurring sources rather than a single window embedding.

Load-bearing premise

The empirical rules that throw away low-energy or low-activation patches and then merge any remaining patches within a Chebyshev distance of three actually isolate true single events rather than systematically merging or splitting overlapping sources.

What would settle it

Re-run the identical pipeline on a fully annotated multi-source subset where every overlapping event is timed and labeled; if the recovered event embeddings routinely fuse distinct concurrent sources or split single calls into multiple clusters, the central claim that event-level aggregation enables reliable exploration collapses.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper presents a self-supervised pipeline for large-scale exploration of low-frequency hydroacoustic recordings with minimal annotation. A Vision Transformer MAE is pre-trained (multi-site Indian Ocean data) then specialized on the target MAHY*2 hydrophone; patch embeddings are filtered by energy/activation rules and merged by Chebyshev distance ≤3 into event-level embeddings; these are clustered dataset-wide (MAE-K-Means or MAE-UMAP/HDBSCAN) and mapped by an annotator in ~1 h to 15 semantic classes. Quantitative evaluation converts cluster densities into hourly detectors and reports F1/ROC against a periodicity detector and a supervised YOLO baseline on seven classes with independent ground truth; qualitative evaluation recovers known seasonalities and isolates previously unstudied signals.

Significance. If the results hold, the work supplies a practical, open-source workflow (code and MAE weights released) that lowers the annotation barrier for multi-year passive acoustic monitoring archives containing biological, geophysical, and anthropogenic sources. The event-level aggregation step is a concrete response to source overlap, a recurring difficulty in low-frequency hydroacoustics. Independent ground-truth catalogs, ablations of pre-training and aggregation, and recovery of literature seasonal patterns strengthen the claim that the pipeline is useful both as an approximate detector and as an exploratory tool for unknown signals.

major comments (3)
  1. §V.B.1–2 defines “empty” patches by E_i < μ_E−σ_E or A_i < μ_A+σ_A and merges regions whose min Chebyshev distance ≤3; both rules are stated as chosen empirically. The abstract and contribution list attribute performance gains to this event-level disentanglement. Table II shows that event-level variants generally outperform whole-spectrogram aggregations, yet the max-deviation ablation already exceeds event extraction on EQ P, and §VIII acknowledges failure on time-frequency-overlapping sources. No sensitivity sweep on the three free thresholds, and no purity/completeness metric against the Dréo or Raumer catalogs, is provided. Without that check, attribution of the reported F1 gains specifically to “disentanglement” remains under-anchored and should be either quantified or more carefully qualified.
  2. §VI.B and Fig. 6: the hourly density score s(h) yields incomplete ROC curves by construction because the decision region is bounded by cluster support. The paper notes this, yet still claims performance “comparable to two existing detectors.” For classes where the operating point sits far from the full FPR range (e.g., AMW, OW), the comparison is only partial. The manuscript should state more explicitly what “comparable” means under this constraint (e.g., F1 at the optimal threshold only) and discuss whether the method can ever serve as a tunable detector.
  3. §V.D / Table I: cluster-to-class mapping of 317 clusters into 15 classes in one hour is central to both the quantitative F1 numbers and the qualitative seasonal plots. The procedure (eight nearest spectrograms per cluster) is reasonable but subjective and unreplicated. Different annotators, or the same annotator on a different day, could re-group clusters and change which events enter s(h). A short inter-annotator check or a leave-one-cluster-out stability note would strengthen that the reported metrics are not mapping-dependent.
minor comments (5)
  1. §V.C: HDBSCAN is incorrectly glossed as “Uniform Manifold Approximation and Projection” (copy-paste from the UMAP sentence).
  2. Discussion: typos “minimsl”, “reasults”, and “f (Hz)” column header in Table I should be cleaned.
  3. Fig. 4 and Fig. 5: spectrogram normalization and frequency axis limits are stated in captions but would benefit from explicit color-scale bars for reproducibility.
  4. §VII.A: pre-training used 12.20 % of available data with 50 % overlap; a one-sentence justification for the sampling fraction would help readers assess diversity.
  5. Related work could briefly situate the event-level step against recent whole-window SSL clustering in seismology (Rimpot et al. is cited but the distinction could be sharper).

Circularity Check

0 steps flagged

No load-bearing circularity: unsupervised MAE + event clustering + post-hoc mapping is evaluated against external catalogs; self-citations supply data/baselines only.

full rationale

The pipeline (MAE reconstruction pre-training, empirical patch discard + Chebyshev merge into event embeddings, UMAP/HDBSCAN or K-Means clustering, then ~1 h visual cluster-to-class mapping) is fully unsupervised with respect to the evaluation labels. Hourly density scores s(h) and the resulting ROC/F1 numbers (Table II, Fig. 6) are obtained by comparing the mapped clusters against two external sources (Dréo et al. 2025 weakly-annotated whale presence and Raumer et al. 2025 automatic P-phase catalog). Those catalogs pre-exist the present clustering and are not used to train the MAE, to set the discard/merge thresholds, or to form the clusters; the mapping step inspects spectrogram patches, not the ground-truth labels. Self-citations therefore supply the evaluation data and one baseline detector but do not force the reported F1 values or seasonal recoveries by construction. No equation equates a claimed prediction to a fitted input, no uniqueness theorem is imported to forbid alternatives, and no known empirical pattern is merely renamed. The empirical character of the patch rules is a methodological assumption (correctness risk), not a circular reduction. Score 1 reflects only the ordinary author-overlap on the evaluation corpora, which is non-load-bearing under the stated criteria.

Axiom & Free-Parameter Ledger

5 free parameters · 3 axioms · 0 invented entities

The central claim rests on standard SSL and clustering machinery plus several empirical design choices (patch size, mask ratio, energy/activation thresholds, Chebyshev merge distance, target cluster count ~300) that are not derived from first principles. No new physical entities are postulated; the free parameters are algorithmic knobs fitted or chosen by hand for the Mayotte spectrograms.

free parameters (5)
  • patch energy discard threshold = μE − σE
    Patches with Ei < μE − σE are discarded; threshold chosen empirically (§V.B.1).
  • patch activation discard threshold = μA + σA
    Patches with Ai < μA + σA are discarded; threshold chosen empirically (§V.B.1).
  • Chebyshev merge distance = 3
    Events merged if min Chebyshev distance ≤ 3 (§V.B.2).
  • target number of clusters / HDBSCAN min cluster size = ~317 clusters
    Hyperparameters tuned to obtain ~300 clusters (min cluster size 50, min neighborhood 20) for feasible annotation (§VII.B).
  • MAE mask ratio and architecture size = 50 % mask, dim 256
    50 % masking, 16-layer encoder / 6-layer decoder, embedding dim 256 chosen for computational tractability (§V.A).
axioms (3)
  • domain assumption Reconstruction of masked spectrogram patches yields embeddings that capture hydroacoustic event structure useful for clustering.
    Core MAE pretext assumption imported from He et al. 2021 and applied to low-frequency spectrograms (§III, §V.A).
  • ad hoc to paper Adjacent informative patches after energy/activation filtering correspond to single hydroacoustic events that can be represented by one mean embedding.
    Event extraction procedure of §V.B; paper acknowledges failure on overlapping sources.
  • domain assumption Hourly density of cluster membership is a valid proxy for binary presence/absence of a semantic class.
    Evaluation protocol §VI.A–B that converts clusters into ROC/F1 scores.

pith-pipeline@v1.1.0-grok45 · 21049 in / 2743 out tokens · 26244 ms · 2026-07-11T01:43:32.250879+00:00 · methodology

0 comments
read the original abstract

Passive hydroacoustic monitoring often generates large volumes of continuous recordings that are only partially exploited due to the cost of manual annotation. Supervised detection methods perform well but require large labeled datasets, seldom available for rare signals or understudied environments. This work proposes a self-supervised exploration pipeline to address this limitation in low-frequency settings. A Masked AutoEncoder (MAE) is pre-trained on a reconstruction pretext task, then used to extract patch-level representations from spectrograms. Within each spectrogram, adjacent informative patches are aggregated into event-level embeddings, enabling the disentanglement of overlapping events. These embeddings are then clustered at the dataset scale using the dimension reduction algorithm UMAP and the clustering algorithm HDBSCAN to identify hydroacoustic patterns. The pipeline was applied to a multi-year hydroacoustic dataset collected near Mayotte Island, Indian Ocean, containing marine mammal vocalizations, seismo-volcanic signals, and anthropogenic noise. The 317 clusters were manually mapped to 15 hydroacoustic classes or noise in less than one hour. The method was evaluated in two ways. Quantitatively, when used as a classifier, it achieved performance comparable to two existing detectors. Qualitatively, it recovered known seasonal patterns of marine mammal acoustic activity. It also identified patterns of previously unstudied signals, thereby demonstrating its practical value.

Figures

Figures reproduced from arXiv: 2607.07733 by Anatole Gros-Martial, Axel Marmoret, Dorian Cazau, Flore Samaran, Jean-Yves Royer, Maelle Torterotot, Pierre-Yves Raumer, Richard Dreo, Sara Bazin.

Figure 1
Figure 1. Figure 1: FIG. 1. Schematic view of the architecture used in this work. The input spectrogram is partly [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Map of the southern Indian Ocean showing the locations of the hydrophones used in [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Overview of the proposed workflow. (1) A spectrogram containing both a 100-120 Hz [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Examples of MAHY*2 normalized spectrograms reconstructions by the MAE. Original [PITH_FULL_IMAGE:figures/full_fig_p021_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Time distribution of detected events with an example spectrogram, using the results of [PITH_FULL_IMAGE:figures/full_fig_p024_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Receiver Operating Characteristic curves of the presented MAE-based methods, the BioD [PITH_FULL_IMAGE:figures/full_fig_p026_6.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

278 extracted references · 278 canonical work pages · 15 internal anchors

  1. [1]

    UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction

    McInnes, Leland and Healy, John and Melville, James , year =. doi:10.48550/ARXIV.1802.03426 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1802.03426

  2. [2]

    Campello, Ricardo J. G. B. and Moulavi, Davoud and Sander, Joerg , editor =. Density-. Advances in. 2013 , note =. doi:10.1007/978-3-642-37456-2_14 , urldate =

    work page doi:10.1007/978-3-642-37456-2_14 2013

  3. [3]

    Geophysical Journal International , author =

    Self-supervised learning of seismological data reveals new eruptive sequences at the. Geophysical Journal International , author =. 2024 , pages =. doi:10.1093/gji/ggae361 , abstract =

    work page doi:10.1093/gji/ggae361 2024

  4. [4]

    The Journal of the Acoustical Society of America , author =

    Automated detection of. The Journal of the Acoustical Society of America , author =. 2015 , pages =. doi:10.1121/1.4934271 , abstract =

    work page doi:10.1121/1.4934271 2015

  5. [5]

    Geophysical Research Letters , author =

    Determining the orientations of ocean bottom seismometers using ambient noise correlation , volume =. Geophysical Research Letters , author =. 2013 , pages =. doi:10.1002/grl.50698 , abstract =

    work page doi:10.1002/grl.50698 2013

  6. [6]

    Rauch, Lukas and Heinrich, René and Moummad, Ilyass and Joly, Alexis and Sick, Bernhard and Scholz, Christoph , year =. Can. doi:10.48550/ARXIV.2504.12880 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2504.12880

  7. [7]

    Emerging

    Caron, Mathilde and Touvron, Hugo and Misra, Ishan and Jegou, Herve and Mairal, Julien and Bojanowski, Piotr and Joulin, Armand , month = oct, year =. Emerging. 2021. doi:10.1109/ICCV48922.2021.00951 , abstract =

    work page doi:10.1109/iccv48922.2021.00951 2021

  8. [8]

    Geophysical Journal International , author =

    Clock errors in land and ocean bottom seismograms: high-accuracy estimates from multiple-component noise cross-correlations , volume =. Geophysical Journal International , author =. 2018 , pages =. doi:10.1093/gji/ggy236 , abstract =

    work page doi:10.1093/gji/ggy236 2018

  9. [9]

    Journal of Geophysical Research: Solid Earth , author =

    Earthquake scaling relations for mid‐ocean ridge transform faults , volume =. Journal of Geophysical Research: Solid Earth , author =. 2004 , pages =. doi:10.1029/2004JB003110 , abstract =

    work page doi:10.1029/2004jb003110 2004

  10. [10]

    Journal of Geophysical Research: Solid Earth , author =

    780. Journal of Geophysical Research: Solid Earth , author =. 2021 , pages =. doi:10.1029/2021JB022152 , abstract =

    work page doi:10.1029/2021jb022152 2021

  11. [12]

    Earth and Planetary Science Letters , author =

    Controls on the magmatic fraction of extension at mid-ocean ridges , volume =. Earth and Planetary Science Letters , author =. 2020 , pages =. doi:10.1016/j.epsl.2020.116541 , abstract =

    work page doi:10.1016/j.epsl.2020.116541 2020

  12. [13]

    Journal of Geophysical Research: Solid Earth , author =

    Controls on. Journal of Geophysical Research: Solid Earth , author =. 2018 , pages =. doi:10.1029/2018JB015545 , abstract =

    work page doi:10.1029/2018jb015545 2018

  13. [14]

    The Journal of the Acoustical Society of America , author =

    Detecting, classifying, and counting blue whale calls with. The Journal of the Acoustical Society of America , author =. 2021 , pages =. doi:10.1121/10.0004828 , abstract =

    work page doi:10.1121/10.0004828 2021

  14. [15]

    MATPAC++: Enhanced Masked Latent Prediction for Self-Supervised Audio Representation Learning

    Quelennec, Aurian and Chouteau, Pierre and Peeters, Geoffroy and Essid, Slim , year =. doi:10.48550/ARXIV.2508.12709 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2508.12709

  15. [16]

    Quelennec, Aurian and Chouteau, Pierre and Peeters, Geoffroy and Essid, Slim , month = apr, year =. Masked. doi:10.1109/ICASSP49660.2025.10887666 , abstract =

    work page doi:10.1109/icassp49660.2025.10887666 2025

  16. [17]

    Comptes Rendus

    Initial results from a hydroacoustic network to monitor submarine lava flows near. Comptes Rendus. Géoscience , author =. 2023 , pages =. doi:10.5802/crgeos.119 , language =

    work page doi:10.5802/crgeos.119 2023

  17. [18]

    doi:10.1109/ICASSP49357.2023.10095642 , abstract =

    Hagiwara, Masato , month = jun, year =. doi:10.1109/ICASSP49357.2023.10095642 , abstract =

    work page doi:10.1109/icassp49357.2023.10095642 2023

  18. [19]

    Tang, Chengyu and Baskiyar, Sanjeev , year =. State. doi:10.48550/ARXIV.2512.03563 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2512.03563

  19. [20]

    The Thirty-Ninth Annual Conference on Neural Information Processing Systems workshop: AI for non-human animal communication , author =

    work page

  20. [21]

    PDMX: A large-scale public domain musicxml dataset for symbolic music processing

    Sarkar, Eklavya and Magimai.-Doss, Mathew , month = apr, year =. Comparing. doi:10.1109/ICASSP49660.2025.10889684 , abstract =

    work page doi:10.1109/icassp49660.2025.10889684 2025

  21. [22]

    PDMX: A large-scale public domain musicxml dataset for symbolic music processing

    Liang, Jingyong and Meyer, Bernd and Lee, Issac Ning and Do, Thanh-Toan , month = apr, year =. Self-supervised. doi:10.1109/ICASSP49660.2025.10890060 , abstract =

    work page doi:10.1109/icassp49660.2025.10890060 2025

  22. [23]

    and Brickson, Leandra and Titus, Alexander J

    Bermant, Peter C. and Brickson, Leandra and Titus, Alexander J. , month = oct, year =. Bioacoustic. doi:10.1101/2022.10.12.511740 , abstract =

    work page doi:10.1101/2022.10.12.511740 2022

  23. [24]

    A data-centric framework for crystal structure identification in atomistic simulations using machine learning

    Regional seismicity of the. Geophysical Journal International , author =. 2010 , pages =. doi:10.1111/j.1365-246X.2010.04815.x , language =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1111/j.1365-246x.2010.04815.x 2010

  24. [25]

    Geophysical Research Letters , author =

    Voluminous eruption from a zoned magma body after an increase in supply rate at. Geophysical Research Letters , author =. doi:10.1002/2016GL071327 , abstract =

    work page doi:10.1002/2016gl071327

  25. [26]

    Environmental Reviews , author =

    Underwater noise and. Environmental Reviews , author =. 2020 , pages =. doi:10.1139/er-2019-0033 , abstract =

    work page doi:10.1139/er-2019-0033 2020

  26. [27]

    Dosovitskiy, Alexey and Beyer, Lucas and Kolesnikov, Alexander and Weissenborn, Dirk and Zhai, Xiaohua and Unterthiner, Thomas and Dehghani, Mostafa and Minderer, Matthias and Heigold, Georg and Gelly, Sylvain and Uszkoreit, Jakob and Houlsby, Neil , year =. An. doi:10.48550/ARXIV.2010.11929 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2010.11929 2010

  27. [28]

    2024 , pages =

    Earth and Space Science , author =. 2024 , pages =. doi:10.1029/2023EA003332 , abstract =

    work page doi:10.1029/2023ea003332 2024

  28. [29]

    Niklasson, Siobhan Marion Yueh Ming and Rowe, Charlotte Anne , year =. Arctic

    work page

  29. [30]

    Acoustic

    Blondel, Philippe and Dell, Benedict and Suriyaprakasam, Cheran , year =. Acoustic. doi:10.1121/2.0001312 , urldate =

    work page doi:10.1121/2.0001312

  30. [31]

    Better quantifying inter-annotator variability:

    Dubus, Gabriel and Torterotot, Maëlle and Duc, Paul Nguyen Hong and Beesau, Julie and Cazau, Dorian and Adam, Olivier , month = jun, year =. Better quantifying inter-annotator variability:. doi:10.1109/OCEANSLimerick52467.2023.10244502 , urldate =

    work page doi:10.1109/oceanslimerick52467.2023.10244502 2023

  31. [32]

    and Kanamori, Hiroo , month = may, year =

    Hanks, Thomas C. and Kanamori, Hiroo , month = may, year =. A moment magnitude scale , volume =. Journal of Geophysical Research: Solid Earth , publisher =. doi:10.1029/jb084ib05p02348 , abstract =

    work page doi:10.1029/jb084ib05p02348

  32. [33]

    Science , author =

    Seismic constraints on caldera dynamics from the 2015. Science , author =. 2016 , pages =. doi:10.1126/science.aah5563 , abstract =

    work page doi:10.1126/science.aah5563 2015

  33. [34]

    Geochemistry, Geophysics, Geosystems , author =

    Automatic. Geochemistry, Geophysics, Geosystems , author =. 2025 , pages =. doi:10.1029/2025GC012572 , abstract =

    work page doi:10.1029/2025gc012572 2025

  34. [35]

    Journal of Geophysical Research: Machine Learning and Computation , author =

    Self‐. Journal of Geophysical Research: Machine Learning and Computation , author =. 2025 , pages =. doi:10.1029/2025JH000958 , abstract =

    work page doi:10.1029/2025jh000958 2025

  35. [36]

    He, Kaiming and Chen, Xinlei and Xie, Saining and Li, Yanghao and Dollár, Piotr and Girshick, Ross , year =. Masked. doi:10.48550/ARXIV.2111.06377 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2111.06377

  36. [37]

    Frontiers in Marine Science , author =

    Global. Frontiers in Marine Science , author =. 2019 , pages =. doi:10.3389/fmars.2019.00067 , abstract =

    work page doi:10.3389/fmars.2019.00067 2019

  37. [38]

    Royal Society Open Science , author =

    Omura’s whales (. Royal Society Open Science , author =. 2015 , pages =. doi:10.1098/rsos.150301 , abstract =

    work page doi:10.1098/rsos.150301 2015

  38. [39]

    Marine Biology Research , author =

    Omura's whale off. Marine Biology Research , author =. 2016 , pages =. doi:10.1080/17451000.2015.1084424 , abstract =

    work page doi:10.1080/17451000.2015.1084424 2016

  39. [40]

    First specimen record of

    Jung, Jean-Luc , year =. First specimen record of

    work page

  40. [41]

    JASA Express Letters , author =

    Omura's whale (. JASA Express Letters , author =. 2025 , pages =. doi:10.1121/10.0037187 , abstract =

    work page doi:10.1121/10.0037187 2025

  41. [42]

    Animals , author =

    Distribution and. Animals , author =. 2024 , pages =. doi:10.3390/ani14202944 , abstract =

    work page doi:10.3390/ani14202944 2024

  42. [43]

    Acoustics Australia , author =

    Review of. Acoustics Australia , author =. 2017 , pages =. doi:10.1007/s40857-017-0101-z , abstract =

    work page doi:10.1007/s40857-017-0101-z 2017

  43. [44]

    Scientific Reports , author =

    Multiple pygmy blue whale acoustic populations in the. Scientific Reports , author =. 2021 , pages =. doi:10.1038/s41598-021-88062-5 , abstract =

    work page doi:10.1038/s41598-021-88062-5 2021

  44. [45]

    The Journal of the Acoustical Society of America , author =

    Description and seasonal detection of two potential whale calls recorded in the. The Journal of the Acoustical Society of America , author =. 2015 , pages =. doi:10.1121/1.4928719 , abstract =

    work page doi:10.1121/1.4928719 2015

  45. [46]

    Journal of Mammalogy , author =

    Occurrence of. Journal of Mammalogy , author =. 2020 , pages =. doi:10.1093/jmammal/gyaa130 , abstract =

    work page doi:10.1093/jmammal/gyaa130 2020

  46. [47]

    Marine Mammal Science , author =

    Rare or cryptic?. Marine Mammal Science , author =. 2017 , pages =. doi:10.1111/mms.12348 , abstract =

    work page doi:10.1111/mms.12348 2017

  47. [48]

    Omura’s whale (

    Ranjbar, Sharif and Dakhteh, Mohammad Sayed and Van Waerebeek, Koen , month = mar, year =. Omura’s whale (. doi:10.1101/042614 , abstract =

    work page doi:10.1101/042614

  48. [49]

    Water Biology and Security , author =

    Record of a juvenile. Water Biology and Security , author =. 2022 , pages =. doi:10.1016/j.watbs.2022.100056 , abstract =

    work page doi:10.1016/j.watbs.2022.100056 2022

  49. [50]

    Aquatic Mammals , author =

    First. Aquatic Mammals , author =. 2020 , pages =. doi:10.1578/AM.46.3.2020.301 , abstract =

    work page doi:10.1578/am.46.3.2020.301 2020

  50. [51]

    Aquatic Mammals , author =

    A. Aquatic Mammals , author =. 2016 , pages =. doi:10.1578/AM.42.2.2016.193 , abstract =

    work page doi:10.1578/am.42.2.2016.193 2016

  51. [52]

    Marine Biodiversity Records , author =

    First record of. Marine Biodiversity Records , author =. 2017 , pages =. doi:10.1186/s41200-017-0121-2 , abstract =

    work page doi:10.1186/s41200-017-0121-2 2017

  52. [53]

    Animal Systematics, Evolution and Diversity , author =

    First. Animal Systematics, Evolution and Diversity , author =. 2018 , pages =. doi:10.5635/ASED.2018.34.3.162 , abstract =

    work page doi:10.5635/ased.2018.34.3.162 2018

  53. [54]

    The Journal of the Acoustical Society of America , author =

    Source levels of. The Journal of the Acoustical Society of America , author =. 2025 , pages =. doi:10.1121/10.0036847 , abstract =

    work page doi:10.1121/10.0036847 2025

  54. [55]

    The Journal of the Acoustical Society of America , author =

    Automatic detectors for low-frequency vocalizations of. The Journal of the Acoustical Society of America , author =. 2020 , pages =. doi:10.1121/10.0001108 , abstract =

    work page doi:10.1121/10.0001108 2020

  55. [56]

    Nature , author =

    A newly discovered species of living baleen whale , volume =. Nature , author =. 2003 , pages =. doi:10.1038/nature02103 , abstract =

    work page doi:10.1038/nature02103 2003

  56. [57]

    Nature , author =

    Proxy evidence for an. Nature , author =. 2003 , pages =. doi:10.1038/nature02101 , language =

    work page doi:10.1038/nature02101 2003

  57. [58]

    Journal of Geophysical Research: Solid Earth , author =

    Global observation of vertical‐. Journal of Geophysical Research: Solid Earth , author =. 2013 , pages =. doi:10.1029/2012JB009721 , abstract =

    work page doi:10.1029/2012jb009721 2013

  58. [59]

    Marine Pollution Bulletin , author =

    The variable influence of anthropogenic noise on summer season coastal underwater soundscapes near a port and marine reserve , volume =. Marine Pollution Bulletin , author =. 2023 , pages =. doi:10.1016/j.marpolbul.2023.115406 , abstract =

    work page doi:10.1016/j.marpolbul.2023.115406 2023

  59. [60]

    , month = dec, year =

    Wenz, Gordon M. , month = dec, year =. Acoustic. The Journal of the Acoustical Society of America , publisher =. doi:10.1121/1.1909155 , abstract =

    work page doi:10.1121/1.1909155

  60. [61]

    Journal of Geophysical Research: Biogeosciences , author =

    Seasonal. Journal of Geophysical Research: Biogeosciences , author =. 2023 , pages =. doi:10.1029/2023JG007409 , abstract =

    work page doi:10.1029/2023jg007409 2023

  61. [62]

    Earth and Planetary Science Letters , author =

    Ring faulting and piston collapse in the mantle sustained the largest submarine eruption ever documented , volume =. Earth and Planetary Science Letters , author =. 2024 , pages =. doi:10.1016/j.epsl.2024.119026 , language =

    work page doi:10.1016/j.epsl.2024.119026 2024

  62. [63]

    Earth System Science Data , author =

    A 30-month data set of glider physico-chemical data off. Earth System Science Data , author =. 2025 , pages =. doi:10.5194/essd-17-4535-2025 , abstract =

    work page doi:10.5194/essd-17-4535-2025 2025

  63. [64]

    Simultaneous evolution of the virial parameter and star formation rate in molecular clumps undergoing global hierarchical collapse

    Seismic moment distribution revisited:. Geophysical Journal International , author =. 2002 , pages =. doi:10.1046/j.1365-246x.2002.01594.x , language =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1046/j.1365-246x.2002.01594.x 2002

  64. [65]

    The Seismic Record , author =

    Estimating the. The Seismic Record , author =. 2023 , pages =. doi:10.1785/0320230017 , abstract =

    work page doi:10.1785/0320230017 2023

  65. [66]

    Science Advances , author =

    Earthquake frequency-magnitude distribution at. Science Advances , author =. 2025 , pages =. doi:10.1126/sciadv.adx9873 , abstract =

    work page doi:10.1126/sciadv.adx9873 2025

  66. [68]

    Seismological Research Letters , author =

    A. Seismological Research Letters , author =. 2022 , pages =. doi:10.1785/0220210279 , abstract =

    work page doi:10.1785/0220210279 2022

  67. [69]

    Bulletin of Volcanology , author =

    No depth limit to hydrovolcanic limu o. Bulletin of Volcanology , author =. 2010 , pages =. doi:10.1007/s00445-009-0315-5 , abstract =

    work page doi:10.1007/s00445-009-0315-5 2010

  68. [70]

    Nature , author =

    Dynamics of a seafloor-spreading episode at the. Nature , author =. 2016 , pages =. doi:10.1038/nature20116 , abstract =

    work page doi:10.1038/nature20116 2016

  69. [71]

    Geophysical Research Letters , author =

    Empirical relationship of. Geophysical Research Letters , author =. 2001 , pages =. doi:10.1029/2001GL012939 , abstract =

    work page doi:10.1029/2001gl012939 2001

  70. [72]

    and Lee, W

    Dziak, Robert P. and Lee, W. S. and Haxel, J. H. and Matsumoto, H. and Tepp, G. and Lau, T.-K. and Roche, L. and Yun, S. and Lee, C.-K. and Lee, J. and Yoon, S.-T. , month = jul, year =. Hydroacoustic,. Frontiers in Earth Science , publisher =. doi:10.3389/feart.2019.00183 , urldate =

    work page doi:10.3389/feart.2019.00183 2019

  71. [73]

    and Fowler, Matthew and Matsumoto, Haruyoshi and Bohnenstiehl, DelWayne and Park, Minkyu and Warren, Kyle and Lee, Won Sang , year =

    Dziak, Robert P. and Fowler, Matthew and Matsumoto, Haruyoshi and Bohnenstiehl, DelWayne and Park, Minkyu and Warren, Kyle and Lee, Won Sang , year =. Life and. Oceanography , publisher =. doi:10.5670/oceanog.2013.20 , number =

    work page doi:10.5670/oceanog.2013.20 2013

  72. [74]

    Bulletin of the Seismological Society of America , author =

    P- and. Bulletin of the Seismological Society of America , author =. 2004 , pages =. doi:10.1785/0120030156 , language =

    work page doi:10.1785/0120030156 2004

  73. [75]

    Geology , author =

    Rapid dike emplacement leads to eruptions and hydrothermal plume release during seafloor spreading events , volume =. Geology , author =. 2007 , pages =. doi:10.1130/G23476A.1 , abstract =

    work page doi:10.1130/g23476a.1 2007

  74. [76]

    Nature Geoscience , author =

    Seismic precursors and magma ascent before the. Nature Geoscience , author =. 2012 , pages =. doi:10.1038/ngeo1490 , abstract =

    work page doi:10.1038/ngeo1490 2012

  75. [77]

    Bulletin of the Seismological Society of America , author =

    Practical. Bulletin of the Seismological Society of America , author =. 2007 , pages =. doi:10.1785/0120060193 , language =

    work page doi:10.1785/0120060193 2007

  76. [78]

    pure and applied geophysics , author =

    Universality of the. pure and applied geophysics , author =. 1999 , pages =. doi:10.1007/s000240050277 , language =

    work page doi:10.1007/s000240050277 1999

  77. [79]

    Nature , author =

    The seismic moment budget of slowly spreading ridges , volume =. Nature , author =. 1988 , pages =. doi:10.1038/334058a0 , language =

    work page doi:10.1038/334058a0 1988

  78. [80]

    Journal of Geophysical Research , author =

    Seismology and the new global tectonics , volume =. Journal of Geophysical Research , author =. 1968 , pages =. doi:10.1029/JB073i018p05855 , language =

    work page doi:10.1029/jb073i018p05855 1968

  79. [81]

    Physics of the Earth and Planetary Interiors , author =

    Seismicity of oceanic and continental rifts—a geodynamic approach , volume =. Physics of the Earth and Planetary Interiors , author =. 1999 , pages =. doi:10.1016/S0031-9201(98)00165-4 , language =

    work page doi:10.1016/s0031-9201(98)00165-4 1999

  80. [82]

    Nature Geoscience , author =

    Continuous exhumation of mantle-derived rocks at the. Nature Geoscience , author =. 2013 , pages =. doi:10.1038/ngeo1771 , language =

    work page doi:10.1038/ngeo1771 2013

Showing first 80 references.