TerraMosaic Daily Digest: September 16, 2026
Daily Summary
The featured landslide studies pair explicit probability estimation with integrated monitoring and prediction. A Bayesian analysis of 22 years of rainfall and shallow landslides in southern Italy yields a gradual posterior probability response rather than a sharp triggering threshold, with retrospective gains over prior probability but no prospective or external-region validation. At the Xiaolangdi Reservoir, a 2018-2024 Sentinel-1 Ministack-InSAR record is linked to RF/XGBoost-SHAP factor analysis and multi-source CNN-LSTM prediction; the best model reports an MAE of 3.0 mm, RMSE of 4.5 mm and R² of 0.885.
Volcanic and seismic hazards are being reconstructed from longer event histories and translated into probability maps. Two companion Stromboli studies synthesize 67 major explosions and paroxysms over roughly 150 years, then derive conditional and 10- and 50-year ballistic-fallout maps. At Mt. Mantap, 17 years of waveforms reveal delayed organization of post-test seismicity along shallow faults. A Croatia-specific PSHA combines 32 source zones, a ranked ground-motion logic tree and seven return periods, exposing period-dependent differences from Eurocode 8 spectra.
Climate-sensitive hazards are increasingly tested against independent events and future scenarios. A Landsat-XGBoost framework for 31,786 Mongolian Plateau lakes placed 987 of 1,188 observed 2021-2025 drainage events (83.1%) above the fixed 2020 top-decile risk threshold, while explicitly treating future outputs as relative risk rather than calibrated probability. Multi-temporal satellite and machine-learning models for Punjab transfer between the 2025 and 2023 floods but retain threshold and spatial-dependence limitations. Rainflow incorporates numerical wind fields and outperforms HRRR and NowcastNet, particularly for extreme rainfall, in the North Atlantic; shorter-lead western Pacific experiments indicate potential cross-basin generalizability.
Deployment constraints are entering hazard-system design. WISE combines weak supervision with a 0.12-million-parameter smoke model, reaches mean tile- and pixel-level F1 scores of 0.964 and 0.750, and executes on an ISS-mounted payload. A spatiotemporal autoencoder reduces turbulent atmospheric contamination in mining-area InSAR and transfers to an independent region. Piezoelectric bolts resolve distributed stress changes during field slope failure, while blasting pre-damage experiments quantify a transition from violent strainburst ejection to gravity-driven collapse.
Key Trends
The day's studies connect probabilistic forecasting, multi-sensor state estimation and deployment-aware hazard monitoring.
- Probability-based estimates sharpen hazard characterization: Several featured studies encode uncertainty or event rates probabilistically: Bayesian landslide forecasts, ballistic-hazard maps and national PSHA replace single-value summaries with probability-based estimates.
- Monitoring, factor interpretation and prediction are being coupled: The Xiaolangdi workflow links Ministack-InSAR, SHAP-based factor interpretation and short-term deformation prediction, while distributed piezoelectric anchors connect internal stress change to field-scale slope failure.
- Long records reveal delayed and time-dependent hazard behavior: Century-scale eruption histories, 17 years of seismic waveforms and multi-year InSAR series provide temporal context unavailable from single-event observations.
- Climate projections are separating risk ranking from event probability: Thermokarst-lake projections retain a fixed historical risk reference and cross-model agreement while avoiding claims of calibrated future drainage probability; flood-transfer studies similarly expose threshold sensitivity.
- Hazard AI is moving onto constrained platforms: Onboard smoke detection, cross-regional radar transfer and lightweight atmospheric correction treat latency, memory, sparse labels and geographic transfer as core scientific constraints.
Selected Papers
The 16 September selection is anchored by delayed fault reactivation after underground nuclear tests, Bayesian rainfall-landslide forecasting, integrated reservoir-landslide monitoring and field-scale slope sensing. Companion studies address volcanic ballistic probabilities, tunnel collapse, coastal subsidence, thermokarst-lake drainage, flood susceptibility, cyclone-rainfall nowcasting, wildfire-smoke detection, rockburst mitigation and national seismic-hazard mapping.
1. Nuclear tests at Mt. Mantap have reactivated intraplate faults
Core Problem: Explains persistent seismicity after underground nuclear testing.
Key Innovation: Seventeen-year waveform analysis showing delayed fault organization and reactivation.
2. Bayesian forecasting of triggered landslides
Core Problem: Forecasts landslide probability from rainfall while representing uncertainty.
Key Innovation: Bayesian posterior probabilities rise gradually with rainfall, with no sharp threshold emerging in this dataset.
3. Integrated Ministack-InSAR Monitoring and Multi-Source-Factor-Informed CNN-LSTM Prediction of Reservoir-Bank Landslide Deformation: A Case Study of the Xiaolangdi Reservoir, China
Core Problem: Connect long-term deformation mapping, factor interpretation and short-term reservoir-bank landslide prediction.
Key Innovation: Ministack-InSAR plus SHAP-interpreted conditioning factors and multi-source CNN-LSTM forecasting.
4. Study on the field failure case and the monitoring effect of rock slopes based on piezoelectric bolts
Core Problem: Measures internal stress redistribution and abrupt changes during field-tested rock-slope failure.
Key Innovation: Distributed self-sensing piezoelectric anchors combining impedance and voltage signals.
5. Assessment of the stability of the Bonifacio cliff using 3D geomechanical modelling
Core Problem: Evaluates large-scale instability under fracture and basal-erosion uncertainty.
Key Innovation: 3D scenario modeling tied to sensor placement and monitoring optimization.
6. Active Control of Strain Rockburst via Blasting Pre-damage: Experimental Study on Energy Dissipation and Failure Mechanisms
Core Problem: Tests blast-like pre-damage as a means to dissipate elastic energy ahead of deep tunnel faces.
Key Innovation: Controlled blasting pre-damage with true-triaxial testing and a threshold-like suppression regime.
7. Ballistic projectile hazard of major explosions and paroxysms at Stromboli (Italy) with uncertainty quantification - Part 1: Mapping method and data analysis
Core Problem: Maps projectile reach and affected areas for major Stromboli explosions.
Key Innovation: Uncertainty-aware synthesis of 67 events over roughly 150 years.
8. Ballistic projectile hazard of major explosions and paroxysms at Stromboli (Italy) with uncertainty quantification - Part 2: Conditional and temporal probability maps
Core Problem: Estimates spatial and time-dependent ballistic-fallout probabilities.
Key Innovation: Conditional maps plus 10- and 50-year temporal probability maps.
9. Deep learning-based turbulent atmospheric delay correction and ground subsidence reconstruction in complex multi-source adjacent subsidence zones
Core Problem: Separates turbulent atmospheric delay from adjacent multiscale subsidence in InSAR.
Key Innovation: Spatiotemporal autoencoder with gradient-balanced image and semivariogram losses.
10. Rising Risk of Thermokarst Lake Drainage on the Mongolian Plateau Under Future Warming
Core Problem: Projects abrupt thermokarst-lake drainage risk under warming.
Key Innovation: Landsat-XGBoost framework with fixed baseline thresholds, independent event validation, climate ensembles, and sensitivity testing.
11. Multi-Temporal Satellite Observations and Machine Learning-Based Flood Susceptibility Assessment of the 2025 Punjab Flood
Core Problem: Maps flood susceptibility and tests event-to-event transferability.
Key Innovation: Multi-sensor Sentinel-1 inventory, refined runoff factors, multiple ML models, SHAP, and independent 2023 testing.
12. Probabilistic seismic hazard analysis for Croatia: hazard mapping and Eurocode 8 implications
Core Problem: Builds and validates a Croatia-specific probabilistic hazard model.
Key Innovation: Regional source model, ranked ground-motion logic tree, seven return periods, and code-spectrum comparison.
13. Tunnel collapses and adaptive support at rock-soil interfaces under low-to-medium overburden in tectonically disturbed rock masses
Core Problem: Identify collapse controls and support requirements where shallow tunnels cross faulted rock-soil interfaces.
Key Innovation: Integrates 26 collapse records, geological classification, construction evidence and convergence monitoring to derive an empirical interface-depth indicator and collapse-risk chart.
14. Integrated Sentinel-1 InSAR and IoT-GNSS sensors for monitoring ground movement in reclaimed coastal infrastructure: A case study at the Port of Brisbane
Core Problem: Separate vertical and horizontal ground movement reliably where single-track InSAR geometric assumptions can bias subsidence estimates.
Key Innovation: Cross-validates Sentinel-1 phase-linked InSAR with continuous three-dimensional IoT-GNSS to expose clay-controlled deformation and pseudo-vertical bias.
15. Thermal, Compositional, and Rheological Structures of the Lithosphere Beneath the Changbaishan Volcanic Field and Their Controls on Volcanic Activity
Core Problem: Characterize lithospheric thermal, compositional and rheological structures associated with contrasting volcanic activity.
Key Innovation: Probabilistic joint inversion combines heat flow, topography, geoid, and Rayleigh-wave dispersion to identify thin lithosphere, thermal anomaly, and rheological weakening.
16. Modular Deep Learning Mechanisms for Auditable Next-Day Wildfire Spread Prediction
Core Problem: Predict next-day active-fire spread while exposing assumptions and improving auditability.
Key Innovation: Tests wind- and slope-conditioned attention, retrieval-based correction, and fire-conditioned dual-stream gating across five backbones.
17. Interpretable Patch-Based Deep Learning for Wildfire Spread Prediction from Ensemble Simulations
Core Problem: Approximate costly physics-based fire-spread ensembles while retaining interpretability and transfer.
Key Innovation: Compares four architectures, tests physics constraints and interpretability, and evaluates zero-shot transfer to a second region.
18. Casualties analysis of Jishishan Ms 6.2 earthquake in Gansu Province, China
Core Problem: Characterizes structural and demographic patterns in earthquake fatalities.
Key Innovation: Disaggregates fatalities by immediate cause, building type, age, sex, fault distance and ground-motion context.
19. Experimental Investigation on the Role of Stress Difference in Rapid Unloading Response and Delayed Strainburst Failure Under True Triaxial Conditions
Core Problem: Determines how initial stress difference affects delayed strainburst after rapid unloading.
Key Innovation: True-triaxial tests linking energy release, AE, fragment velocity, and tensile cracking.
20. Stress Wave Response and In Situ Wave Absorption Performance of Rubber‒Polypropylene Fiber Anchoring Mortar for Deep-Buried Tunnels
Core Problem: Reduces blasting-wave transmission through tunnel support.
Key Innovation: Rubber-polypropylene anchoring mortar validated in laboratory and in situ.
21. Laboratory Injection-Induced Slip on Critically Stressed Granite Faults: Roles of Injection Rate and Temperature
Core Problem: Quantifies injection-rate and temperature controls on fault activation and seismic moment.
Key Innovation: Critically stressed hot-granite experiments linking microstructure, deformation, and moment release.
22. WISE: A Lightweight, Weakly-Supervised Model for Onboard Fire Smoke Detection and Localization
Core Problem: Deliver low-latency, spatially informative smoke detection under onboard compute constraints.
Key Innovation: Teacher-student weak supervision yields a 0.12M-parameter model producing tile labels and smoke maps in one pass.
23. Tropical Cyclone Precipitation Nowcasting Based on Flow Matching Model With Numerical Wind Field Constraints
Core Problem: Forecast complex cyclone rainfall motion and intensity at short lead times.
Key Innovation: Rainflow uses conditional flow matching with 500- and 850-hPa numerical wind constraints to predict 0.01-degree, 10-minute rainfall fields.
24. Dynamic processes and cascading mechanisms of the 8 February 2025 high-elevation long-runout landslide in Junlian, Sichuan, China, revealed by real-time seismic records
Core Problem: Title-level focus: resolve the event's dynamic sequence and cascading mechanisms.
Key Innovation: Title-signalled approach or contribution: Uses real-time seismic records to reconstruct a high-elevation landslide process. Methods, data and results could not be assessed because no reliable abstract was available.
25. Experimental study of the sediment source effect on debris-flow magnitude using a 3D miniaturized model
Core Problem: Title-level focus: examine how sediment-source conditions relate to debris-flow magnitude.
Key Innovation: Title-signalled approach or contribution: uses a three-dimensional miniaturized physical model to examine sediment-source effects. Methods, data and results could not be assessed because no reliable abstract was available.
26. Resilience-oriented restoration of distribution networks under typhoon-induced uncertainties with rolling-horizon co-optimization
Core Problem: Title-level focus: restore distribution networks resiliently as typhoon-related uncertainty evolves.
Key Innovation: Title-signalled approach or contribution: Uses rolling-horizon co-optimization for restoration decisions. Methods, data and results could not be assessed because no reliable abstract was available.
27. BlinkChange: Active perception bi-temporal reliable representation for disaster impact understanding from remote sensing imagery
Core Problem: Learn reliable bi-temporal representations for interpreting disaster impacts.
Key Innovation: Combines change-aware discrepancy construction, saliency-guided dynamic perception and progressive enhancement-suppression, achieving leading captioning results and strong building-damage performance with external-scene tests.
28. An explainable integrated early-warning method for coal and gas outburst risk based on multimodal transfer learning
Core Problem: Title-level focus: predict outburst risk early from heterogeneous mine data while retaining interpretability.
Key Innovation: Title-signalled approach or contribution: Integrates explainability, multimodal inputs, and transfer learning in an early-warning method. Methods, data and results could not be assessed because no reliable abstract was available.
29. Leveraging deep learning and kernel density representations for transformation-based regional flood frequency analysis
Core Problem: Title-level focus: estimate regional flood-frequency distributions under a transformation-based framework.
Key Innovation: Title-signalled approach or contribution: Combines deep learning with kernel-density representations. Methods, data and results could not be assessed because no reliable abstract was available.
30. Rapid Loss of the Sierra Nevada's Largest Trees Driven by Fire
Core Problem: Map large trees and characterize mortality, disturbance, and recovery across the Sierra Nevada.
Key Innovation: Combines a crown-delineation U-Net, sub-meter aerial canopy models, Sentinel-2 trajectories, BFAST breakpoints, and fire perimeters.
31. An online spectral nudging-based correction system: improving physical model forecasts by incorporating large-scale circulations derived from machine learning models
Core Problem: Combines ML circulation skill with physical-model fine structure.
Key Innovation: Online spectral nudging from ML into numerical forecasts.
32. Spatial variation of the present-day strain-rate and stress regime, from mid-Atlantic triple junction through north Africa to Aqaba transform zone
Core Problem: Maps crustal deformation and stress along segmented plate boundaries.
Key Innovation: Integrates 784-event moment tensors, stress inversion, GNSS, and InSAR across 17 zones.
33. Cross-Regional Transfer Learning for Radar Echo Extrapolation in Northwestern Xinjiang, Western China
Core Problem: Improves radar-echo extrapolation where local severe-event samples are scarce.
Key Innovation: Cross-region pretraining plus architecture- and module-level transfer analysis.
34. HAF-Net: A Hierarchical Adaptive Fusion Network for Satellite-Based Radar Reflectivity Reconstruction
Core Problem: Reconstructs radar reflectivity from geostationary satellite data.
Key Innovation: Multi-receptive-field, saliency-guided, hierarchical echo fusion.
35. Thermo-hydro-mechanical response of the operation style in a faulted Malm reservoir: Storage triplet vs. hydrothermal doublet
Core Problem: Compares stress perturbation and fault stability for geothermal operation styles.
Key Innovation: Fully coupled reservoir-scale comparison of cyclic storage and continuous extraction.
36. Characterization and stability assessment of rare earth tailings under dry and paste stacking conditions
Core Problem: Determines confinement, breakage, and stacking effects on REE-tailings behavior.
Key Innovation: Integrated mineralogy, critical-state testing, breakage analysis, and PLAXIS modeling.
37. A Two-Stage Arrival-Time Picking Framework for Acoustic Emission Signals in Rock Failure and Validation for Source Spatial Localization
Core Problem: Picks AE arrivals robustly under low signal-to-noise conditions.
Key Innovation: Adaptive S-transform/AIC stage plus YOLO and 1D-CNN refinement.
38. Well-balanced lattice Boltzmann modelling of the depth-averaged age equation and its correlations with environmental indicators in a typical urban flood retention area
Core Problem: Title-level focus: models water age and environmental behavior in an urban flood-retention area.
Key Innovation: Title-signalled approach or contribution: Well-balanced lattice Boltzmann solution of the depth-averaged age equation. Methods, data and results could not be assessed because no reliable abstract was available.
39. Strain-dependent site amplification and its implications for the seismic bearing capacity of strip footings near slopes
Core Problem: Title-level focus: links nonlinear amplification to bearing capacity near slopes.
Key Innovation: Title-signalled approach or contribution: Strain-dependent seismic assessment. Methods, data and results could not be assessed because no reliable abstract was available.
40. Using Rapid Temperature Falls to Estimate Future Strong Cold Front Frequency in CMIP6 Climate Projections
Core Problem: Assess how strong cold fronts change under a high-emissions scenario.
Key Innovation: Uses rapid day-to-day maximum-temperature falls as a cold-front proxy and relates projected patterns to baroclinicity and storm tracks.
41. Internal Variability Dominates the Spatial Heterogeneity of Recent Trends in Eurasian Winter Cold Extremes
Core Problem: Explain the pronounced spatial heterogeneity of recent Eurasian cold-extreme trends.
Key Innovation: Quantitatively separates internal variability from external forcing and links regional trends to Atlantic and Pacific multidecadal variability.
42. Climate Models Accurately Predict Ocean-Going Meltwater Runoff From the Northwest Greenland Ice Sheet
Core Problem: Determine when surface-mass-balance runoff estimates agree with observed proglacial discharge.
Key Innovation: Compares multiple climate products with new multi-year discharge records in watersheds with contrasting subglacial and proglacial storage effects.
43. Coastal barrier response to energetic storms: A global process-based modelling overview
Core Problem: Title-level focus: synthesize process-based modeling of storm-driven coastal-barrier change globally.
Key Innovation: Title-signalled approach or contribution: Provides a global overview centered on process-based coastal response models. Methods, data and results could not be assessed because no reliable abstract was available.
44. From risk knowledge to action: A systematic review and evidence map of institutionalization deficits in non-structural urban flood risk reduction
Core Problem: Title-level focus: explain why non-structural flood-risk knowledge is not consistently institutionalized as action.
Key Innovation: Title-signalled approach or contribution: Combines systematic review with an evidence map focused on institutionalization deficits. Methods, data and results could not be assessed because no reliable abstract was available.
45. Time and tide: Mapping the changing 3D shape of Australia's dynamic intertidal zone using time series satellite data
Core Problem: Map intertidal elevation and morphological change through time at continental scale without in-situ calibration.
Key Innovation: Fuses Landsat and Sentinel-2 time series with global tide modeling and independent multi-temporal validation across Australia's coastline.
46. Gravitational energy released when Earth quakes: Heat supply for island-arc volcanism
Core Problem: Explain the energy source for intense island-arc volcanism at cool subduction zones.
Key Innovation: Proposes a first-principles gravitational-energy-to-heat conversion model tied to thrust-faulting earthquakes.
47. Every Fixed Metric Has a Blind Spot: A Learned Atmospheric Critic for Scoring Forecast Realism
Core Problem: Detect unphysical spatial artifacts that fixed forecast metrics miss.
Key Innovation: Trains an atmospheric discriminator whose logit acts as an adaptive divergence-like realism score.
48. Butterfly Effect and the Kinetic Energy Cascade in Probabilistic Machine Learning Weather Prediction Models
Core Problem: Determine whether leading probabilistic machine-learning weather models reproduce kinetic-energy cascades and the multiscale growth of forecast uncertainty.
Key Innovation: Four ML weather ensembles are compared with IFS-ENS using KE and difference-KE spectra, revealing model-specific failures in upscale transfer, mesoscale energy, and early spread growth.
49. Undercatch corrected gridded precipitation data to improve hydrological modeling in high-alpine orography
Core Problem: Corrects systematic precipitation undercatch in alpine terrain.
Key Innovation: Station- and terrain-informed gridded correction that improves runoff, snow, and glacier modeling.
50. Calibration of Event-Based Camera for High-Speed Observations of Lightning and Transient Luminous Events
Core Problem: Makes event-camera biases physically interpretable for microsecond lightning observations.
Key Innovation: Calibration of contrast, refractory period, and filter frequency with reconstruction code.
51. Displacement Reconstruction for Hybrid DOFSS FBG Inclinometer Monitoring with Angular Boundary Constraints
Core Problem: Controls angular-boundary error in distributed inclinometer reconstruction.
Key Innovation: Analytical uncertainty propagation and FBG angular referencing.
52. Contrasting effects of arsenic and cadmium on PFAS transport in soils: Mechanistic insights and mathematical models
Core Problem: Resolve metal-dependent PFAS transport mechanisms and their long-term implications for groundwater contamination.
Key Innovation: Combines batch and column experiments with a nonequilibrium Hydrus-1D model and 50-year transport scenarios.
53. Net Groundwater Recharge in California Occurs Primarily During Above-Average Snowpack Years
Core Problem: Determine when basin-scale net groundwater recharge occurs in California's Central Valley.
Key Innovation: Derives recharge from GRACE and GRACE-FO groundwater-storage changes and identifies an approximately 21-cubic-kilometre snow-water-equivalent threshold.
54. An uncertainty workflow for landsat TIRS split window surface temperature products incorporating atmospheric water vapor
Core Problem: Estimate atmospheric-water-vapor effects and propagate them into split-window surface-temperature uncertainty.
Key Innovation: Uses an XGBoost total-precipitable-water estimator from Landsat thermal bands and a water-vapor-aware uncertainty workflow.
55. Screening ANN models via qualitative laws for wave overtopping discharge at vertical walls: Implications for generalization deficits
Core Problem: Title-level focus: detect physically implausible generalization in ANN overtopping predictions.
Key Innovation: Title-signalled approach or contribution: Screens learned models against qualitative physical laws. Methods, data and results could not be assessed because no reliable abstract was available.
56. Achieve Medium-Range SST Forecast by Rolling 1-Day Deep Learning
Core Problem: Reduce error accumulation and discontinuity in medium-range sea-surface-temperature forecasting.
Key Innovation: Rolling one-day and multistep training in Earthformer deliver skill beyond 15 days and are checked against Argo observations.
57. Aligned Consensus Teaching for Label-Efficient Oriented Object Detection in Weakly-Aligned Visible-Infrared Imagery
Core Problem: Reduce dual-modality annotation needs for oriented object detection in weakly aligned visible-infrared imagery.
Key Innovation: Combines cycle-consistent alignment, cross-modal consensus pseudo-labeling, and text-guided paired augmentation.
58. Multi-View Mixture-of-Experts with Vision-Language Reranking for Cross-View Object Geo-Localization
Core Problem: Locate query objects in satellite imagery across multiple ground and aerial viewpoints.
Key Innovation: Unifies view-specific experts, vision-language reranking, contrastive alignment, and an elliptical spatial prior.
59. Understanding Dynamic Scenes at Gigapixel Scale: Wide-Area Spatio-Temporal Perception from UAVs
Core Problem: Support detection, tracking, and temporal scene understanding at gigapixel scale.
Key Innovation: Provides the HARD dataset with multilevel annotations and streaming-HOTA for latency-aware tracking.
60. A general lightweight global modeling framework for three-dimensional seismic exploration
Core Problem: Capture long-range dependencies in 3D seismic interpolation and denoising at manageable cost.
Key Innovation: Uses three-directional Mamba scanning, dual time-frequency features, and plug-and-play normalization.
61. PDA++: Field-Aligned Planning and Scene-Adaptive Insertion in Remote Sensing
Core Problem: Insert rare targets realistically into overhead imagery for few-shot and long-tailed recognition.
Key Innovation: Plans scene-compatible poses, conditions generation on target context, and aligns local texture distributions.
62. GeoCond: A Conditioning-Aware Reliability Adapter for Feed-Forward 3D Reconstruction
Core Problem: Identify geometrically ill-conditioned camera and depth predictions and gate refinement.
Key Innovation: Adapts frozen 3D backbones using predicted geometry, orbit variance, pose errors, or cycle residuals.
63. Tabular Deep Learning vs Classical Machine Learning for Urban Land Cover Classification
Core Problem: Determine whether tabular deep-learning models improve nine-class urban land-cover classification under high dimensionality, heterogeneous features, and class imbalance.
Key Innovation: A unified reproducible benchmark compares classical ensembles and multiple tabular deep-learning models with imbalance-aware training and a broad metric suite.
64. Period-Inverted Blast-Radius Normalization for a Non-Ablating Hypersonic Reentry Source Using OSIRIS-REx Capsule Infrasound
Core Problem: Resolve the uncertain normalization linking hypersonic-source infrasound periods, energy deposition, and blast radius.
Key Innovation: Periods from 39 OSIRIS-REx reentry stations are inverted through a weak-shock model to calibrate blast radius without prespecifying the normalization, yielding uncertainty intervals and a rapid estimator.
65. Physics-Informed Neural Networks for Fast Multilayer Spectral Inversion of Hα 6562.8 Å and Ca II 8542.1 Å Spectra
Core Problem: Accelerate multilayer inversion of H-alpha and Ca II spectra while retaining interpretable radiative-transfer physics.
Key Innovation: A PINN predicts inversion parameters and passes them through a differentiable analytic forward model, reproducing conventional inversions while providing a reported 12–60× speedup.
66. Van der Waals interactions in supercritical water under Earth's mantle conditions
Core Problem: Quantify how exchange-correlation functionals and van der Waals corrections affect simulated water properties under mantle conditions.
Key Innovation: Ab initio molecular-dynamics comparisons across four functionals resolve structural, diffusion, vibrational, and proton-transfer sensitivities at 1–10 GPa and 1000 K.
67. Urban Wind Effects on UAV Operations in Building-Dense Low-Altitude Airspace
Core Problem: Identify where urban wind perturbations interact with terrain and buildings to create clearance-sensitive UAV trajectories.
Key Innovation: The method combines LBM wind fields with co-registered 20 m terrain-building grids and nominal versus wind-affected vehicle envelopes across Shanghai and Beijing.
68. LSR-Net: Learning the Forward Evolution Operator for Nonlinear Fluid Dynamics
Core Problem: Learn accurate and efficient forward evolution operators that capture both local and global interactions in nonlinear fluid systems.
Key Innovation: LSR-Net splits its integral kernel into short-range convolutions and long-range sum-of-exponentials Fourier multipliers, attaining O(n log n) computation and outperforming FNO and DeepONet baselines.
69. A Unified Hierarchical Framework for Fine-grained Cross-view Geo-localization over Large-scale Scenarios
Core Problem: Unify large-scale image retrieval and meter-level localization that are usually trained and executed as separate tasks.
Key Innovation: UnifyGeo jointly learns multi-granularity representations and re-ranks candidates within one hierarchical network, reporting meter-level recall on VIGOR.
70. Integral equations for flexural-gravity waves: analysis and numerical methods
Core Problem: Models flexural-gravity-wave scattering by heterogeneous sea ice or ice shelves.
Key Innovation: High-order FFT-accelerated integral-equation solver.
71. Full-scale investigation into top-down construction of fully prefabricated structures retaining deep excavation
Core Problem: Controls wall and ground deformation during deep excavation.
Key Innovation: Full-scale monitoring of prefabricated TAD-wall top-down construction.
72. Tunnelling-induced lateral pile response: a coupled analytical model accounting for tunnel-soil interaction
Core Problem: Predicts pile response near tunnel excavation.
Key Innovation: Coupled tunnel-soil-pile model including lining stiffness and screening zones.
73. Effect of critical state line curvature on cone penetration test-based state inference in sand
Core Problem: Quantifies bias from assuming a log-linear critical-state line in CPT interpretation.
Key Innovation: Full-geometry particle finite-element tests of curved critical-state behavior.
74. Micromechanical analysis of methane hydrate-bearing sediments during gas extraction considering hydrate dissociation and particle breakage
Core Problem: Explains deformation of hydrate-bearing sediment during depressurization.
Key Innovation: DEM coupling hydrate dissociation with irregular particle breakage.
75. Local scour evolution around monopile and pile-group foundations under unsteady flow
Core Problem: Title-level focus: examines local scour around offshore foundations.
Key Innovation: Title-signalled approach or contribution: Unclear from the metadata-only abstract. Methods, data and results could not be assessed because no reliable abstract was available.
76. Experimental study on wave attenuation and beach profile protected by rigid and flexible vegetation
Core Problem: Title-level focus: tests vegetation effects on waves and beach profiles.
Key Innovation: Title-signalled approach or contribution: Comparison of rigid and flexible vegetation protection. Methods, data and results could not be assessed because no reliable abstract was available.
77. A unified finite element upper-bound framework for the lateral bearing capacity of monopiles in undrained clays considering different failure mechanisms
Core Problem: Title-level focus: estimates lateral capacity of monopiles in clay.
Key Innovation: Title-signalled approach or contribution: Unified finite-element upper-bound framework for multiple mechanisms. Methods, data and results could not be assessed because no reliable abstract was available.
78. Experimental investigation of the monotonic and multistage cyclic lateral response of pile bucket composite foundations in calcareous sand
Core Problem: Title-level focus: characterizes composite-foundation response in calcareous sand.
Key Innovation: Title-signalled approach or contribution: Multistage cyclic testing of pile-bucket foundations. Methods, data and results could not be assessed because no reliable abstract was available.
79. Experimental study on the static and cyclic mechanical behavior and a generalized plasticity model for coral sand
Core Problem: Title-level focus: models static and cyclic coral-sand behavior.
Key Innovation: Title-signalled approach or contribution: Generalized plasticity formulation supported by experiments. Methods, data and results could not be assessed because no reliable abstract was available.
80. Beyond behavioural models: equifinality and overparameterisation undermine confidence in predictions by soil organic matter models
Core Problem: Diagnoses hidden uncertainty from equifinality and excess parameters.
Key Innovation: Shows identifiable-parameter limits even under data-rich conditions.
81. Multi-season evaluation of temperature and wind in the marine boundary layer along the United States northeast coast in the High-Resolution Rapid Refresh model
Core Problem: Quantifies systematic marine-boundary-layer forecast errors.
Key Innovation: Multi-year, multi-season evaluation using coastal observations.
82. Development of the CCPP-based GEFS-aerosols component in the Unified Forecast System for subseasonal prediction (UFS-Chem v1.0)
Core Problem: Improves coupled aerosol-weather subseasonal prediction.
Key Innovation: CCPP-based aerosol effects on radiation, clouds, and scavenging.
83. Recognition and characterization of blasting half-holes in hydraulic tunnels based on TLS point cloud
Core Problem: Measures blast quality and surrounding-rock effects in tunnels.
Key Innovation: TLS plus optimized RANSAC extraction of half-hole geometry.
84. A fractional-order viscoelastoplastic creep constitutive model for intensely altered rock accounting for moisture content
Core Problem: Predicts moisture- and confinement-dependent creep of altered rock.
Key Innovation: Fractional viscoelastoplastic model validated by large-scale triaxial creep tests.
85. SDFA-Net: Urban Manhole Cover Detection via Fisheye Image and LiDAR Point Cloud Fusion
Core Problem: Detects and diagnoses manhole-cover defects from mobile mapping.
Key Innovation: Depth-supervised fisheye-to-BEV transformation and fidelity-aware LiDAR-image fusion.
86. A Multi-Modal Deep Learning Framework for High-Resolution Alpine Land Use/Land Cover
Core Problem: Maps fine-grained alpine land cover under sparse labels.
Key Innovation: SAM-based multimodal fusion of RGB, multispectral, indices, and DSM with class-specific refinement.
87. Spatial Prediction and Performance Comparison of Soil Organic Carbon in High-Latitude Forest Soils Based on Multiple Feature Selection Methods and Machine Learning Models
Core Problem: Maps soil organic carbon by depth in discontinuous permafrost forest.
Key Innovation: Systematic comparison of 27 feature-model-depth combinations using SAR, optical, and terrain data.
88. TOTMSeg: A Texture-Aware Octree-Based Transformer-Mamba Framework for Large-Scale Urban Mesh Semantic Segmentation
Core Problem: Segment large textured urban meshes without losing fine structures or computational efficiency.
Key Innovation: Combines texture-aware sampling, octree-guided aggregation and Transformer-Mamba global modeling, with evaluation on two public urban datasets.
89. Uncertainty-Based Quality-Control Prioritization for High-Resolution Land-Cover Mapping Under Spatially Disjoint Evaluation
Core Problem: Determines whether model uncertainty can prioritize land-cover review.
Key Innovation: Spatially disjoint comparison of deterministic and Bayesian uncertainty ranking.
90. Reconstructed XCO2 Reveals Seasonal Moisture Limitation Sensitivity Across Typical Steppe, Forest, and Gobi Desert in Mongolia
Core Problem: Reconstructs sparse XCO2 and relates it to seasonal moisture limitation.
Key Innovation: Seamless LightGBM reconstruction combined with ecosystem-specific drought sensitivity analysis.
91. Effect of plasticity on fall cone testing of sand-clay mixtures
Core Problem: Relates mixture plasticity to fall-cone response and undrained strength.
Key Innovation: Two-zone interpretation plus random-forest strength prediction.
92. Acoustic Emission and 3D Fracture Morphology Evolution of Granite with Non-Parallel Prefabricated Fissures Under Uniaxial Cyclic Compression
Core Problem: Explains fissure-controlled damage under cyclic excavation and blasting loads.
Key Innovation: Combines mechanics, energy, acoustic emission, and 3D fracture morphology.
93. Experimental and Numerical Investigation on the Mechanical Behavior of Extremely Soft Coal-Like Material
Core Problem: Characterizes soft-coal damage and post-peak behavior.
Key Innovation: Improved sampling, calibrated analog material, PFC3D, and piecewise constitutive model.
94. Toward Practical Rock Mass Classification and Conversion: An Interpretable Data-Driven Framework for Mapping Q-System Parameters to RMR
Core Problem: Converts Q-system observations to RMR reliably.
Key Innovation: Interpretable OGSA-XGBoost benchmarked against classical and spline relations.
95. Quantitative Characterization of Hydraulic Parameters from Two-Dimensional Rough Fracture Geometry: Analytical Solution, Validation, and Application
Core Problem: Derives nonlinear hydraulic parameters from realistic rough-fracture geometry.
Key Innovation: Analytical discretized flow model extended to shear-seepage and fracture networks.
96. Physics-Consistent Monotonic Machine Learning for RMR Prediction from Q-System Parameters Under Duplicate-Aware Validation
Core Problem: Prevents leakage and engineering-inconsistent RMR predictions.
Key Innovation: Duplicate-aware grouped validation, monotonic boosting, boundary diagnostics, and external testing.
97. A Micromechanics-Driven Model for Normal Deformation of Rough Rock Joints
Core Problem: Predicts nonlinear closure of natural rough joints.
Key Innovation: Mechanically derived two-parameter exponential model validated on more than 500 datasets.
98. Self-similarity criticality of the world’s large river systems
Core Problem: Title-level focus: examines critical scaling in large river networks.
Key Innovation: Title-signalled approach or contribution: Self-similarity or criticality framing. Methods, data and results could not be assessed because no reliable abstract was available.
99. Transfer evolution of 3D principal stress magnitude and direction and the mechanical mechanism of asymmetric failure during coal-rock drift excavation
Core Problem: Title-level focus: explains asymmetric failure during coal-rock drift excavation.
Key Innovation: Title-signalled approach or contribution: Tracks three-dimensional principal-stress magnitude and direction. Methods, data and results could not be assessed because no reliable abstract was available.
100. A probabilistic framework for the estimation of specific stream power and its implications for sediment transport processes
Core Problem: Title-level focus: estimate specific stream power while representing uncertainty.
Key Innovation: Title-signalled approach or contribution: Frames stream-power estimation probabilistically and examines implications for sediment transport. Methods, data and results could not be assessed because no reliable abstract was available.
101. Urban building height mapping in Beijing-Tianjin-Hebei megalopolis by synergy of spaceborne LiDAR and multisource geospatial datasets
Core Problem: Title-level focus: map building heights consistently across a large urban agglomeration.
Key Innovation: Title-signalled approach or contribution: Fuses spaceborne LiDAR with multiple geospatial datasets. Methods, data and results could not be assessed because no reliable abstract was available.
102. Long-term reconstruction of daily global OMI NO2 product between 2005-2023 with spatiotemporally constrained compressive sensing
Core Problem: Title-level focus: fill and reconstruct daily global OMI nitrogen-dioxide observations over 2005-2023.
Key Innovation: Title-signalled approach or contribution: Applies spatiotemporally constrained compressive sensing to the satellite product. Methods, data and results could not be assessed because no reliable abstract was available.
103. An enhanced model for the prediction of freeze-thaw-induced settlements and compression properties of fine-grained soils during the freeze-thaw cycle
Core Problem: Title-level focus: model settlement and compression changes through freeze-thaw cycles.
Key Innovation: Title-signalled approach or contribution: An enhanced predictive model jointly addresses settlement and compression properties. Methods, data and results could not be assessed because no reliable abstract was available.
104. Ice accretion growth of high-speed railway contact wires under periodic traction load
Core Problem: Title-level focus: characterize ice growth on contact wires under realistic cyclic loading.
Key Innovation: Title-signalled approach or contribution: Examines coupling between periodic traction load and ice accretion. Methods, data and results could not be assessed because no reliable abstract was available.
105. Spatial heterogeneity and drivers of PFASs distribution in an arid-semiarid city waters system
Core Problem: Title-level focus: characterize spatial heterogeneity and drivers of PFAS contamination across an arid-semiarid urban water system.
Key Innovation: Title-signalled approach or contribution: City-scale spatial driver analysis across connected urban waters. Methods, data and results could not be assessed because no reliable abstract was available.
106. Finite-source filtering and Doppler modulation of infragravity waves generated by breakpoint forcing
Core Problem: Explain how the finite extent and migration of realistic surf-zone breakpoint forcing shape radiated infragravity waves.
Key Innovation: Decomposes breakpoint and bound-wave forcing, then links source width and motion to interference filtering and directional Doppler shifts in controlled numerical experiments.