TerraMosaic Daily Digest: October 11, 2026
Daily Summary
The October 11 issue follows how slope failure develops into a changing hydraulic or fragmentation problem. Forty flume tests resolve the lateral widening of overtopping breaches in non-cohesive landslide dams, including toe scour, sidewall collapse and sediment-deposition feedbacks that can delay peak discharge. Moraine experiments locate sliding at a freeze–thaw interface where fines, water content and excess pressure accumulate; a water-blocking effect remains a proposed explanation. Rock-avalanche simulations connect prescribed launch velocities and source-area joints with flight, impact fragmentation and deposit sorting. Together, these studies distinguish the conditions that initiate movement from the mechanisms that govern its subsequent evolution.
Engineering assessment depends on both loading history and the representation of deterioration. A mining–rainfall loess test traces pre-damage, crack growth, toe slip and overall failure. Reservoir-slope surrogates accelerate repeated stability calculations, while ship-lock slope assessment links deformation indicators with laboratory aging and field observations. Coastal-cave analysis shows why empirical susceptibility, roof-failure mechanisms and three-dimensional hydraulic modelling are complementary rather than interchangeable. Supporting rock tests separate water–clay effects, seepage pressure, cavity geometry and support-interface shrinkage; their laboratory relationships require site-specific interpretation.
Seismic and coastal risk studies connect process understanding with assessment choices. Revisions to the 2023 U.S. national seismic hazard model adjust selected source, ground-motion and site-response components rather than introducing a new prediction paradigm. Costa Rica slow-slip modelling and Tibetan geodetic constraints address longer-term fault interaction and strain partitioning. In Kesennuma, simulated tsunami defences combine hydraulic protection with residents' preferences under sea-level-rise scenarios. A conservative flood-wave PINN improves joint state and roughness reconstruction in heterogeneous one-dimensional channels. Open rainfall-link data and a discussion-stage rock-property atlas broaden the inputs available for independent evaluation.
The AI papers are most consequential where they expose a gap between an attractive score and the quantity needed in practice. A cross-disaster audit finds that holding out whole events sharply reduces building-damage accuracy, while pre-event imagery and scene context remain predictive. Related preprints examine answer–evidence consistency, absent-target rejection and damage-evaluation protocols. Reservoir-policy analysis finds that release accuracy is only weakly related to learned storage quality. Sparse PDE surrogates, interpretable operators and depth representations offer supporting tools, but their benchmark gains are separate from demonstrated disaster forecasting. The common requirement is to evaluate the physical state, spatial transfer and decision-relevant error directly.
Key Trends
Failure evolution, changing hydraulic states and validation under a different event or observation regime connect the selected studies.
- Landslide consequences evolve after initiation: Breach widening, avalanche fragmentation and river-damming studies keep erosion, block interaction and hydraulic routing inside the hazard assessment.
- Prior disturbance changes later instability: Freeze–thaw interfaces, mining pre-damage, rock aging and water-sensitive clay-bearing rock create different starting states for subsequent loading.
- Assessment components need separate validation: Seismic-model revisions, complementary cliff-stability methods, slow-slip cycle models and tsunami-design scenarios constrain different parts of the risk problem.
- Conservation and observability matter in learned models: Flood-wave inversion, reservoir latent-state diagnosis and sparse PDE experiments distinguish physically useful information from a low aggregate prediction error.
- A new disaster can expose shortcuts hidden by familiar-event scores: Damage embeddings, image-grounding benchmarks and open-set rejection show why evaluation should test event transfer and the evidence supporting an output.
Selected Papers
Landslide-dam breaching, moraine failure, rock avalanches and disturbed loess slopes anchor this selection. Reservoir and coastal-cliff assessments connect failure mechanisms with infrastructure conditions, while seismic and tsunami studies extend the hazard view. Preprints on damage assessment and evidence-grounded remote sensing address evaluation reliability. Supporting methods and datasets cover rainfall inputs, rock properties, physical surrogates and geometric observations.
1. Mechanisms and dimensionless prediction of lateral breach widening during overtopping failure of landslide dams
Core Problem: Breach widening depends on toe erosion, collapse, geometry and sediment grading.
Key Innovation: Forty flume tests resolve staged widening and deposition feedbacks, supporting dimensionless discharge and width relations checked on internal and external datasets; the framework concerns homogeneous non-cohesive dams.
2. The failure mechanism of moraine landslides induced by combined effect of freeze-thaw cycles and rainfall infiltration: Insight from the flume model test
Core Problem: Freeze–thaw interfaces can become weak zones during subsequent rainfall.
Key Innovation: Flume observations link interface fines, water content and excess pressure with sliding; hydraulic heterogeneity and a possible water-blocking effect are interpretations rather than directly proven universal mechanisms.
3. Influence of Source‐Area Rock Mass Structure on the Dynamic Evolution of Rock Avalanches With Ejection‐Dominated Initiation
Core Problem: Source structure and launch conditions jointly affect fragmentation and deposition.
Key Innovation: Prescribed ejection velocities and joint densities produce different flight, impact and sorting patterns; the framework resolves post-initiation transport rather than independently predicting the initial earthquake-driven launch.
4. Progressive failure mechanism of sandy loess slopes under mining disturbance-rainfall infiltration effects
Core Problem: Structural pre-damage can condition the response to later infiltration.
Key Innovation: A physical test follows mining pre-damage, crack growth, toe slip and overall sliding with pore-structure observations; scaled displacements and the tested loading sequence are distinct from measured prototype failure.
5. Analysis of stability influencing factors and failure mechanism of anti-dip rock slope under strong earthquake
Core Problem: Layer contrasts and seismic forcing can alter failure mode in anti-dip slopes.
Key Innovation: A 60-case synthesis and Hongshiyan analysis connect basal shear in softer rock with upper tensile cracking and toppling; factor rankings remain conditional on the examined slope and experimental design.
6. RBLA-Net: an efficient deep learning model for time-dependent probabilistic stability analysis of reservoir slope
Core Problem: Spatially variable strength and time-varying loads make repeated numerical assessment costly.
Key Innovation: RBLA-Net reproduces tested Baijiabao stability outputs and reduces trained-surrogate inference time; simulator agreement and a full-year calculation do not independently validate real-time warning or unseen future loads.
7. Predictive Stability Assessment of Ship Lock Rock Slopes: A Multi-level Early Warning Approach with Service-Life Integration
Core Problem: Short-term deformation indicators and long-term aging assumptions need consistency.
Key Innovation: A two-level warning scheme is linked to laboratory deterioration and field observations through aging-equivalence modelling; service-life estimates depend on that equivalence rather than directly observed decades of future performance.
8. Complementarity of Empirical, Analytical and Numerical Techniques for Assessing the Stability of a Natural Cave System in the Coastal Cliffs of Las Grutas (Río Negro, Argentina)
Core Problem: Heterogeneity, cavities, seepage and basal erosion produce different stability indicators.
Key Innovation: Field and UAV characterization combine empirical, analytical and hydro-mechanical analyses, distinguishing global stability under tested steady conditions from possible local degradation; assessment methods are complementary rather than interchangeable.
9. Revisions to the 2023 U.S. 50‐State National Seismic Hazard Model
Core Problem: Model weights, rupture implementation and site response require auditable revision.
Key Innovation: The revised 2023 U.S. model adjusts selected source and ground-motion components and supplies updated hazard products across site conditions; this is a model revision rather than a new earthquake-prediction paradigm.
10. Characteristic‐Aligned Control‐Volume Physics‐Informed Neural Network for Flood‐Wave Propagation and Roughness Estimation
Core Problem: Heterogeneous channels challenge local conservation and joint state–roughness estimation.
Key Innovation: Characteristic-aligned control volumes improve three one-dimensional benchmarks and sparse-observation inversion; integral conservation and station-placement tests are demonstrated within the examined channels and noise levels.
11. LD-DETR: A Multi-Scale Landslide Detection and Segmentation Network for Complex Geological Environments
Core Problem: Small slides, irregular boundaries and imbalance challenge image inventories.
Key Innovation: LD-DETR combines ASPP, weighted Tversky loss and parameter-efficient adaptation, with Bijie–Luding tests and limited-sample Jiuzhai transfer; detection and segmentation gains differ and do not demonstrate triggering prediction.
12. Shortcut Learning in Geospatial Foundation Model Embeddings for Cross-Disaster Damage Assessment
Core Problem: Familiar-event accuracy can conceal scene and event shortcuts.
Key Innovation: Holding out whole disasters sharply lowers balanced accuracy across xView2 and BRIGHT; context, pre-event and pairing tests reveal non-damage cues, while training-only feature transformations improve tested transfer.
13. Modulation of Slow Slip Event Characteristics Throughout the Costa Rica Subduction Zone Earthquake Cycle
Core Problem: Slow slip and the locked earthquake zone exchange stress over the cycle.
Key Innovation: A sequence model reproduces observed Costa Rica slip patterns and explains cycle-dependent slow-slip changes through reciprocal stress transfer; it does not establish a next-earthquake warning rule.
14. Constraining Fault Kinematics and India‐Tarim Convergence Partitioning in Southwestern Tibet With Geodetic Block Modeling
Core Problem: Regional convergence is partitioned between faults and distributed deformation.
Key Innovation: Joint velocity fields constrain fault kinematics, coupling and slip-rate deficits in southwestern Tibet; the inferred seismic potential is a long-term tectonic constraint rather than an event-time prediction.
15. A Unified Quadratic Scaling Relation Between Potency and Local Earthquake Magnitudes for −1.8 ≤ M L ≤ 7.5
Core Problem: Local magnitude and physical source size are not interchangeable linear measures.
Key Innovation: Independent near-source, nodal and broadband datasets support a continuous quadratic potency–magnitude relation, while attenuation, passband and rupture scaling can produce similar nonlinearity.
16. Human-centred design approaches to tsunami mitigation under sea level rise in Kesennuma, Japan
Core Problem: Residents' infrastructure preferences and hydraulic protection need joint assessment.
Key Innovation: Three defence designs are compared in 150 source, sea-level and defence simulations; an open-feeling design approaches stronger containment in the tested scenarios, while eco-hybrid protection alone is limited.
17. One-dimensional settlement characteristics of loess high-fill foundations: insights from centrifuge model tests
Core Problem: Fill height, moisture and compaction affect settlement evolution and its spatial distribution.
Key Innovation: Centrifuge series find construction-stage consolidation dominates tested total settlement, with hysteresis and nonmonotonic depth patterns; the results are model-test evidence for the represented fills.
18. Coupled Effects of Lithology and Cavity Geometry on Brittle Failure and Rockburst-Like Instability in Hard Rocks
Core Problem: Material heterogeneity and cavity corners can localize failure differently.
Key Innovation: DIC, acoustic emission and simulations distinguish geometry-dependent tensile–shear failure from more distributed shear; laboratory brittle instability is not a calibrated field rockburst warning threshold.
19. Future Northeastward Shift of Extratropical Transition of Tropical Cyclones Over the North Atlantic in Downscaling Projections
Core Problem: A stable transition fraction can coexist with a changing preferred transition region.
Key Innovation: Synthetic tracks driven by twelve CMIP6 models project a northeastward extratropical-transition shift under three scenarios, with storm-count changes largely following cyclone frequency; the geographic result is scenario-conditioned.
20. Is the Number of Foreshocks Underestimated? The Role of Declustering Symmetry
Core Problem: Inferred foreshock counts depend on the clustering metric and assumed mechanism.
Key Innovation: A reversed nearest-neighbour metric is shown to correspond to time reversal of the standard metric and is compared in synthetic triggering scenarios; the official abstract truncates its findings, so no quantitative detectability gain is asserted.
21. Is Earthquake Replenishment Sufficient for Unbiased b-Value Estimation?
Core Problem: Completing a catalog and obtaining an unbiased magnitude statistic are different objectives.
Key Innovation: Synthetic incompleteness tests and three earthquake sequences distinguish replenishment from positive-difference estimators; improved b-value estimation does not demonstrate an earthquake precursor.
22. Differences among preparatory processes preceding major earthquakes in Kamchatka, 2014-2025, including the MW 8.8 event
Core Problem: Metric changes before selected events can admit multiple physical interpretations.
Key Innovation: Seven Kamchatka episodes show different seismic-regularity and strain-dynamics histories; the proposed coupled-segment interpretation is retrospective and does not establish prospective accuracy or false-alarm rates.
23. Revisiting the Task of Building Damage Assessment by Satellite Imaging
Core Problem: Sensor choice, missing pre-event imagery and aggregate scores complicate comparison.
Key Innovation: A review proposes an instance-based formulation and multi-score evaluation that accommodates missing pre-event observations; these are task and protocol recommendations rather than a newly validated damage predictor.
24. Predictive modelling of time-dependent scour around inclined and vertical piles under steady and combined wave-current flows
Core Problem: The verified study models scour evolution around inclined and vertical piles under current and combined forcing.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
25. A Framework for designing Offshore Mega Artificial Shelters: optimisation of breakwater layout and hydraulic properties under directional extreme waves
Core Problem: The verified framework optimizes offshore shelter breakwater layout and hydraulic properties under directional extreme waves.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
26. A jointly optimized, multi-criteria peaks-over-threshold framework for regime-conditioned extreme wave analysis in the Persian Gulf
Core Problem: The verified study jointly optimizes a multi-criteria peaks-over-threshold approach for Persian Gulf wave extremes.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
27. Resilience assessment of pile-supported wharves under the effects of sequential earthquake and soil liquefaction incorporating vine copula captured dependence
Core Problem: The verified study examines pile-supported wharves under sequential earthquake and soil-liquefaction effects using dependence modelling.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
28. A time-dependent system reliability and lifecycle optimization framework for RC structures subjected to bidirectional corner-rebar corrosion: Seismic fragility, energy-based functionality loss, and optimal maintenance
Core Problem: The verified framework links corner-rebar corrosion, seismic fragility, functionality loss and lifecycle maintenance.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
29. Contrasting wave–surge dynamics under sea-level rise in the northern South China Sea
Core Problem: The verified study compares wave–surge dynamics under sea-level rise in the northern South China Sea.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
30. High-Resolution Burned Area Mapping Using Sentinel-2 Imagery and Robust Satellite Techniques
Core Problem: Burn products depend on resolution, phenology and the reliability of fire detections.
Key Innovation: A multitemporal workflow is evaluated in two contrasting region-wide experiments and multiple reference systems; larger mapped area and differing overlap scores require interpretation against each validation product.
31. LuTan-1 InSAR-Constrained Reconstruction of Concealed Coal-Mining Working-Face Geometry
Core Problem: Unknown working-face geometry is coupled with depth, thickness and subsidence assumptions.
Key Innovation: LuTan-1 deformation and constrained optimization recover a tested panel geometry consistently with records; agreement is conditional on empirical parameters and the represented active-mining period.
32. A systematic review of geospatial and meteorological approaches to flood risk assessment and monitoring
Core Problem: Diverse data, scales and metrics complicate comparisons of modelling approaches.
Key Innovation: A synthesis of 190 studies organizes geospatial, hydraulic, meteorological and AI methods, linking capability with validation and operational readiness; heterogeneous scores do not support a pooled algorithm ranking.
33. Potential early indication of tectonic ground movements by water-well monitoring datasets, with short-term forward modeling
Core Problem: Early anomalies require separation from common post-event or unrelated responses.
Key Innovation: A monitored-well study reports precursor-like signals for one Mw 5.5 event and short-term trajectory modelling; selected-event evidence does not establish prospective sensitivity, false-alarm performance or earthquake predictability.
34. Process-resolving simulation of landslide-induced river damming under realistic open-channel hydraulics
Core Problem: The verified study resolves landslide damming under realistic open-channel hydraulic conditions.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
35. Characteristics and mechanisms of drilling-induced formation subsidence: Insights from quantitative identification of layered water–soil loss
Core Problem: The verified study identifies layered water–soil loss in drilling-induced formation subsidence.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
36. Dynamic mechanisms and 3D evolutionary simulation of earthquake-triggered composite Donghekou landslide under fault termination rupture effect
Core Problem: The verified study examines three-dimensional landslide evolution under a fault-termination rupture effect.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
37. Fault-zone architecture controls on CO₂ leakage and reactivation: A Field-informed hydro-mechanical analysis
Core Problem: The verified study examines how fault-zone architecture controls leakage and reactivation in a hydro-mechanical analysis.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
38. Analyzing and modeling land subsidence hazard in delta regions using time series InSAR and explainable artificial intelligence: A case study across the Nile Delta, Egypt
Core Problem: The verified study maps and models subsidence hazard across the Nile Delta using time-series InSAR and explainable AI.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
39. Multiphase CFD-DEM simulation of seepage failure in unsaturated earth-rock dams incorporating liquid-bridge capillary effects
Core Problem: The verified study includes liquid-bridge capillarity in multiphase CFD–DEM simulation of unsaturated dam failure.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
40. Basin Water Storage Degree for Risk-Informed Debris-Flow Warning Support: A Forward-Looking Proof of Concept in a Small Mountain Catchment
Core Problem: The verified proof of concept uses basin water storage degree to support risk-informed warning in a small catchment.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
41. Designing evidence-to-decision pathways for community-based landslide early warning: A scenario-based framework from Nepal
Core Problem: The verified Nepal framework connects evidence with community-based landslide warning decisions.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
42. Making Snow Avalanche Information Actionable: How Alaska Decision-Makers Perceive and Use Avalanche Science.
Core Problem: The verified study examines how Alaska decision-makers perceive and use avalanche information.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
43. Seismic social vulnerability assessment in cross-border regions: Application to Albania, Greece and North Macedonia
Core Problem: The verified study assesses seismic social vulnerability across Albania, Greece and North Macedonia.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
44. A Framework for Slope Failure Prediction Integrating Statistical-Kinematic Onset-of-Acceleration Identification with Sequential Probabilistic Time-to-Failure Updating
Core Problem: The verified framework combines statistical–kinematic acceleration-onset identification with probabilistic time-to-failure updating.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
45. A reference Landsat burned-area dataset over Mediterranean Algeria (1984–2025) with an assessment of historical and new-generation products
Core Problem: The verified study develops a Mediterranean Algerian Landsat burned-area reference from 1984–2025 and compares product generations.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
46. Satellite based multi-scale remote sensing system for infrastructure stability monitoring around an urban metro
Core Problem: The verified study develops satellite-based multiscale stability monitoring around an urban metro.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
47. Dynamic reliability analysis via stochastic unsteady seepage for high core-wall rockfill dams
Core Problem: The verified study assesses dynamic reliability using stochastic unsteady seepage.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
48. Morphological Characteristics of Tropical Precipitation Features Producing Extreme Instantaneous Rain Rates in GPM and DYAMOND Global Storm Resolving Models
Core Problem: The spatial organization of convection affects localized instantaneous rainfall.
Key Innovation: GPM feature analysis and four storm-resolving models link stronger peak rain rates to larger, more concentrated convective systems; instantaneous morphology does not directly specify catchment accumulation or landslide thresholds.
49. Diagnosing Latent Storage Representations in LSTM Reservoir Release Policies
Core Problem: Accurate release predictions need not imply physically meaningful storage representation.
Key Innovation: A large-sample LSTM analysis finds only weak association between predictive accuracy and learned storage quality, alongside limited pooled spatial transfer; latent interpretability must be evaluated separately from prediction scores.
50. TerraAgent: Demonstration-Grounded Multi-Agent Reasoning for Long-Context Remote Sensing Visual Question Answering
Core Problem: Correct answers may rely on priors rather than cited imagery.
Key Innovation: Demonstration filtering, tool-routed evidence and counterfactual audits improve tested visual dependence and reduce hallucinations at higher inference cost; the preprint does not demonstrate disaster-specific decision performance.
51. MemReg: Streaming Outdoor LiDAR Point Cloud Registration with Hybrid Memory Buffers
Core Problem: Pairwise registration discards temporal context in sparse scans.
Key Innovation: Feature and pose buffers improve registration on three outdoor benchmarks and indoor transfer tests; vehicle-scanning results do not independently validate deformation measurements on unstable slopes.
52. Spatial Context or Memory? Sizing the Inputs of Neural Surrogates for Partially Observed Fields
Core Problem: Forecast error can arise from missing information as well as model limitations.
Key Innovation: Analytic and PDE experiments show that useful spatial window and memory allocation depend on lead time and field dynamics; conclusions are benchmark-conditioned, with chaos and linear instability not fully separated.
53. Structure-Aware Matching for Robust UAV Thermal-to-Satellite Geo-Localization
Core Problem: Cross-modal appearance differences hinder georeferenced image matching.
Key Innovation: Structure enhancement and affine remapping improve two thermal–satellite benchmarks, including a released paired dataset; successful localization does not establish hazard detection or operational navigation safety.
54. Diffusion Precipitation Generation with Medium-range Weather Forecasting
Core Problem: Reanalysis-trained weather models can retain precipitation biases.
Key Innovation: A conditional diffusion generator maps frozen-model atmospheric forecasts to satellite–gauge-aligned daily rainfall and is evaluated globally and against Chinese stations; improved precipitation scores do not independently validate flood or landslide warnings.
55. Low-Rank Kolmogorov-Arnold Generator Flows for Interpretable Operator Learning
Core Problem: Dense learned operators can obscure coefficient structure and scale poorly.
Key Innovation: Low-rank generator symbols and exponential integration improve synthetic and quantum benchmarks and recover coefficients; demonstrations do not establish general discovery of field-scale geomorphic laws.
56. SBNO : Schrödinger Bridge Neural Operator for the Forward and Inverse Problems under Missing Information
Core Problem: Sparse measurements and incomplete equations complicate forward and inverse modelling.
Key Innovation: A diffusion-bridge operator uses masked paired data for both task directions and compares favorably on five sparse-observation PDE benchmarks; accuracy and resource gains are restricted to the reported settings.
57. PDE-PFN: Prior-Data Fitted Neural PDE Solver
Core Problem: Noisy approximate solution priors may support inference without gradient updates.
Key Innovation: A pretrained attention-based solver demonstrates in-context prediction across tested two-dimensional equations; empirical equation transfer does not establish universal PDE solving or geohazard validation.
58. PlacePulse: Learning Transferable Place Representations from Structural Geographic Data
Core Problem: Reusable place embeddings can encode structural attributes beyond imagery or mobility.
Key Innovation: PlacePulse evaluates U.S. cell embeddings against held-out external targets and feature-selection baselines; social-vulnerability utility is reported, but disaster-loss prediction is not validated.
59. Knowing Whether, Not Just Where: Anchor-Probing Distillation for Open-Set Remote Sensing Visual Grounding
Core Problem: A plausible box can be returned even when the queried object is absent.
Key Innovation: Anchor-probing distillation combines localization with unsupported-query rejection on tested imagery; its visual-support objective is distinct from calibrated disaster-risk confidence.
60. Benchmarking Answer-Grounding Consistency in Remote Sensing Visual Question Answering
Core Problem: Answer accuracy and localization accuracy can hide unsupported answers.
Key Innovation: A rule-based evidence protocol and 15027-sample benchmark expose correct answers lacking supporting grounding; the finding concerns evaluated VQA models rather than every remote-sensing workflow.
61. Constrained Probabilistic Numerics: Gaussian Process Solvers for Linear PDEs under Linear Inequality Constraints
Core Problem: Equality constraints alone do not enforce nonnegativity or other solution bounds.
Key Innovation: Gaussian-process conditioning incorporates finite-dimensional inequalities and improves selected reconstruction and constraint diagnostics; it does not guarantee continuum-wide admissibility or uniform accuracy gains.
62. Intra- and Cross-Scale Feature Interaction Network for Referring Remote Sensing Image Segmentation
Core Problem: Limited scene diversity and absent-target coverage constrain text-guided segmentation.
Key Innovation: NWPU-Refer and a multiscale network extend multilingual, multi-country annotation with no-target cases; the preprint benchmarks general remote-sensing targets rather than a disaster inventory.
63. Green-Routed Neural Operators: Physics Determines Where the Network Reads
Core Problem: Treating a transport field only as a value may ignore its role in selecting read locations.
Key Innovation: Equation-derived departure coordinates improve long-horizon comparisons on five PDE systems, with intervention and ablation tests; the structural prior is demonstrated on benchmarks rather than operational hazard models.
64. DeFM: Learning Foundation Representations from Depth for Robotics
Core Problem: Depth representations need metric awareness and transfer across sensors and scales.
Key Innovation: A depth model pretrained on 60 million images releases reusable representations and tests robotic tasks and sim-to-real transfer; geohazard mapping remains a potential application rather than a demonstrated breakthrough.
65. A Topography-Structure-Aware and Similarity-Enhanced Network for DEM Super-Resolution
Core Problem: Upsampling can confuse global terrain structure with fine-scale detail.
Key Innovation: Frequency-separated branches and local similarity improve two 30-to-10 m benchmarks; learned terrain refinement is not equivalent to a new independent elevation survey.
66. The OpenSat4Weather dataset: Ku-band satellite link data for precipitation monitoring
Core Problem: Restricted sensor ownership can obstruct independent rainfall-method evaluation.
Key Innovation: OpenSat4Weather provides 215 satellite links with concurrent gauges, radar and rain-height inputs over five months in southern France; geographic and temporal coverage limit direct global transfer.
67. Atlas of geochemical, petrophysical and mechanical rock properties of the crystalline basement in Germany – first release
Core Problem: Fragmented rock-property measurements complicate consistent subsurface models.
Key Innovation: The AMPEDEK discussion preprint compiles about 25700 data points from 8200 samples with linked provenance; the national database supports parameter exploration but is not a global landslide dataset.
68. Improving the CLASSIC (v1.8) snow model to better simulate Arctic snowpacks
Core Problem: Snow thermal exchange and compaction influence snowpack and underlying soil conditions.
Key Innovation: CLASSIC revisions improve tested snow and temperature errors across alpine and Arctic sites; blowing-snow effects vary with exposure and forcing uncertainty remains important.
69. An abrupt shift towards a hotter and drier climate regime in the Western Eurasian Steppe
Core Problem: The verified study examines an abrupt hotter and drier climate regime in the Western Eurasian Steppe.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
70. Multi-constraint collaborative seismic imaging and interpretation method for fine characterization of small faults in complex coal mining areas
Core Problem: Small faults are difficult to trace beneath complex near-surface conditions.
Key Innovation: Dense acquisition, azimuth-preserving migration and attribute fusion improve fault interpretation at Daqiang Mine, checked against mining-revealed data; structural imaging is not itself a hazard forecast.
71. Automatic Extraction of Discontinuity Traces from Rock Mass Point Cloud Data Using PhaseVote
Core Problem: Contrast and illumination can obscure fracture traces in mapped rock surfaces.
Key Innovation: Phase-based edges and tensor voting improve two case studies, using RGB structure alongside geometric segmentation; trace extraction does not independently measure joint strength or slope stability.
72. Continuous Radar QPE Bias Field Reconstruction Through the Implicit Neural Representation Framework: QPE-BiasINR
Core Problem: Radar biases can vary sharply and depend on the local precipitation context.
Key Innovation: An implicit bias field outperforms tested mean-field and Kriging corrections in one basin, while consistency regularization reduces local overcorrection; downstream warning performance is not independently established.
73. Depth-dependent thermo-hydraulic responses of unsaturated silty soil: laboratory thermal cycling and field observations
Core Problem: Atmospheric thermal forcing attenuates and lags with soil depth.
Key Innovation: Columns and landfill-cover observations show depth-dependent temperature and moisture responses under specified water content and density; cover-system evidence does not directly validate slope-trigger thresholds.
74. Steady-state hydro-mechanical recovery of microbially induced self-healing synchronous grout: experimental characterization and optimization
Core Problem: Grout cracking can weaken contact and hydraulic performance around linings.
Key Innovation: Orthogonal mixture tests and constrained optimization balance fresh properties, permeability and strength recovery; the recommended mixture window is specific to the tested materials and healing conditions.
75. Biot-Cosserat PFEM for progressive failure and large deformation analysis of saturated soil and its application
Core Problem: Mesh distortion and localization dependency hinder coupled soil-failure simulation.
Key Innovation: A Biot–Cosserat particle finite-element implementation regularizes progressive failure and tests marine-anchor behaviour; benchmark and anchor comparisons do not validate a field landslide forecast.
76. Coupling Effects of Surface Roughness and Volumetric Stability on the Bonding Performance of the Shotcrete–Rock Interface
Core Problem: Rougher contact geometry can interact unfavorably with shrinkage.
Key Innovation: Pullout tests and CT observations show high-shrinkage matrices developing damage at roughness peaks; a coupled material–geometry factor improves tested bond prediction without establishing universal interface strength.
77. Experimental Study on the Dynamic Response of Deep Rocks Under Seepage Pressure
Core Problem: Seepage pressure and loading rate can influence different response measures.
Key Innovation: Coupled sandstone tests associate higher seepage pressure with lower dynamic strength and altered energy dissipation, within the examined confinement and loading conditions.
78. Extrinsic Cohesive Zone Model (ECZM): Implementation, Verification, and Comparison
Core Problem: Dynamic insertion and contact transitions can introduce artificial stress noise.
Key Innovation: Force-consistent insertion and transition strategies improve multiple fracture benchmarks and comparisons; numerical artifact reduction is distinct from validation of site-specific hazard parameters.
79. Experimental Investigation on Damage Evolution of Rock Joints Combining Electrical Resistivity and 3D Morphology Analysis
Core Problem: Natural-rock conductivity can limit resistivity-based joint monitoring.
Key Innovation: Conductive analog specimens link electrical response with shear stages and measured morphology; applying the inferred relations to natural field rocks requires independent conductivity and mechanical validation.
80. The Macro-Meso-mechanism of Splitting Failure of Diorite Under Water Action: The Dominant Role of Clay Minerals
Core Problem: Clay content changes how water affects tensile damage and monitoring signals.
Key Innovation: Clay-bearing and clay-free samples show different nonlinear strength, acoustic and fracture responses with saturation; tested material contrasts do not define a universal water–strength relationship.
81. Iterative Analysis of Diaphragm Wall Displacement Under Cyclic Earth Pressure in Servo Strut Excavations
Core Problem: Reloading support may not reverse plastic deformation accumulated after force loss.
Key Innovation: An iterative non-limit pressure model is compared with three field excavations and FE calculations, identifying residual displacement after compensation; beam and soil assumptions bound the proposed reloading relation.
82. Multi-metric optimization of Mohr–Coulomb parameters in SPH granular flow simulation
Core Problem: The verified study jointly optimizes multiple metrics for Mohr–Coulomb parameters in granular-flow simulation.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.
83. Soil erodibility assessment and model improvement based on soil microstructure in a typical watershed of the loess hilly-gully region, China
Core Problem: The verified study assesses soil erodibility in a loess watershed using microstructural information.
Key Innovation: Not assessed: a reliable abstract matching the verified record was unavailable. Quantitative findings and validation cannot be established from the title alone.