TerraMosaic Daily Digest: April 25, 2026
Daily Summary
The April 25 literature is dominated by transient mountain and subsurface systems whose hazard state changes faster than conventional inventories can follow. The strongest contributions reconstruct a 2025 GLOF from dense satellite time series, model earthquake-driven sediment pulses as a time-varying landslide-supply problem, and trace thermokarst, avalanche weak-layer failure, rockburst damage, and outburst precursors back to measurable physical controls. Across these papers, hazard is not treated as a static class; it is a moving boundary among water storage, sediment availability, stress redistribution, cryospheric weakening, and infrastructural exposure.
A second line of work makes monitoring more diagnostic. InSAR phase-linking quality metrics, distributed fiber-optic sensing above normal faults, staged InSAR-laser-GPR settlement forensics, attenuation tomography, and digital twins for subway waterlogging all seek observables that can be trusted before failure. The AI papers are strongest when they encode a constraint--physics-guided thermal downscaling, bathymetry inversion with slope and radiative transfer, foundation-model adaptation, or uncertainty-aware geotechnical surrogates--rather than simply adding model capacity. The day's engineering studies converge on the same lesson: tunnels, dams, storage reservoirs, pipelines, railways, and urban lifelines become hazardous where boundary conditions change first.
Key Trends
The common methodological move is to replace static labels with constrained state variables: lake area and snow depletion, sediment supply, phase-linking reliability, fiber-optic strain, tunnel water pressure, fault slip potential, and exposure timing.
- Mountain hazards are becoming time-resolved systems: The GLOF, earthquake-sediment, thermokarst, avalanche, glacier, and proglacial papers all treat storage, release, and transport as evolving states rather than fixed susceptibility categories.
- Monitoring papers are asking whether signals are trustworthy: Phase-linking quality coefficients, DFOS settlement tests, InSAR-laser-GPR forensics, attenuation tomography, and seismic-noise site characterization turn detection into diagnostic evidence.
- Infrastructure risk is concentrated at changing interfaces: Fault-crossing tunnels and pipelines, karst tunnel hydraulics, fault-zone seepage, twin tunnelling, deep excavations, railway freeze-thaw, and dam material heterogeneity all locate failure at boundaries where stress, water, or stiffness shifts.
- AI is strongest when it carries physical structure: Physics-guided diffusion, PINN bathymetry, foundation-model adaptation, building-damage contrastive learning, HydroGRAF, AutoML tunnel reliability, and PGNN settlement analysis bind learning to process constraints or uncertainty.
- Exposure is being treated as dynamic data: Building-footprint timing, high-resolution inventory synthesis, water-supply functionality, avalanche work risk, wildfire mapping, and fire-risk review connect hazard processes to changing assets and response capacity.
Selected Papers
This digest features 65 selected papers from 1,381 papers analyzed. The sequence opens with GLOF diagnosis and earthquake-landslide sediment pulses, then moves through thermokarst, avalanche mechanics, rockburst and outburst precursors, megathrust and volcanic unrest, tunnel and fault-rupture response, InSAR and fiber-optic monitoring, constrained remote-sensing AI, and infrastructure risk chains that connect physical boundary changes to exposure.
1. Diagnosing the Controls of the 2025 Talidas GLOF Using Multi-Source Satellite Observations
Core Problem: Mountain GLOF diagnosis often depends on sparse post-event evidence, making it difficult to distinguish landslide impact, moraine erosion, and climate-driven lake growth.
Key Innovation: Dense PlanetScope, Landsat, MODIS, and ERA5-Land observations reconstruct the Talidas lake life cycle and point to anomalous meltwater expansion and moraine overtopping rather than a visible landslide trigger.
2. Suspended Sediment Pulses and Sediment‐Discharge Relationships in Earthquake‐Affected Rivers Along the Tibetan Plateau Margins
Core Problem: Post-earthquake sediment hazards are poorly represented by static rating curves because landslide-derived supply changes through time and interacts with flood transport capacity.
Key Innovation: A dynamic supply-transport model couples seismic landslide inputs, rainfall-driven mass movement, channel storage, and discharge capacity, explaining 78%-90% of suspended-sediment variance in three Tibetan Plateau margin rivers.
3. From direct climatic forcing to permafrost-related restructuring: four decades of thermokarst lake dynamics in the northeastern Tibetan Plateau
Core Problem: Thermokarst lake evolution is increasingly controlled by both direct climate forcing and permafrost restructuring, but long observational records remain scarce for high mountain regions.
Key Innovation: Four decades of remote-sensing analysis in the northeastern Tibetan Plateau separate climatic controls from permafrost-related reorganization of lake dynamics.
4. Study on the microscopic failure mechanisms of snow under mixed loading conditions based on FDM-DEM
Core Problem: Avalanche initiation depends on weak-layer failure under mixed compressive and shear loading, a process difficult to measure directly in fragile snow.
Key Innovation: A coupled FDM-DEM snow model with spatial heterogeneity resolves the microscopic stages of weak-layer damage, crack growth, and failure under mixed loading.
5. Progressive failure mechanism and dynamic constitutive model of sandstone considering cyclic impact damage
Core Problem: Multiple-pillar rockburst failure under repeated dynamic disturbance remains hard to model because damage evolves before catastrophic release.
Key Innovation: Cyclic SHPB experiments quantify stiffness degradation, energy evolution, and progressive failure, then support a dynamic constitutive model for impact-damaged sandstone.
6. Coal and gas outburst prediction: A scale-invariant cascade framework
Core Problem: Outburst prediction needs indicators that remain transferable across scale-dependent mining signals rather than fitting isolated monitoring series.
Key Innovation: A scale-invariant cascade framework is proposed to characterize coal and gas outburst precursors and improve prediction across monitoring scales.
7. Shallow Creep in the Leaky Stress Shadow of Locked Zones of Subduction Megathrust
Core Problem: The shallowest portion of subduction megathrusts may creep even where deeper locked zones are accumulating earthquake and tsunami potential.
Key Innovation: Rate-and-state modelling explains shallow creep and slow earthquakes as the combined effect of prolonged afterslip and a leaky stress shadow around locked patches.
8. Seismic Attenuation Reveals Fault and Forearc Structure Across the Subduction‐Collision Transition in Southern Taiwan
Core Problem: Velocity models alone can miss fractured or fluid-rich structures that control earthquake hazard in subduction-collision settings.
Key Innovation: Dense-array attenuation tomography in southern Taiwan images sharp fault contrasts, a dipping offshore forearc block, and low-Q zones colocated with tectonic tremor.
9. Earthquake Swarm Activity in the Tokara Islands (2025): Statistical Analysis Indicates Low Probability of Major Seismic Event
Core Problem: Earthquake swarms require rapid statistical interpretation because public concern can rise long before a mainshock probability is clear.
Key Innovation: Statistical analysis of the 2025 Tokara Islands swarm evaluates whether the activity pattern implied a low probability of a major seismic event.
10. Seismo-Stratigraphic Data of the Gulf of Pozzuoli (Southern Tyrrhenian Sea, Italy): A Review and Their Relationships with the New Bradyseismic Crisis
Core Problem: The Gulf of Pozzuoli bradyseismic crisis demands integration of older seismo-stratigraphic constraints with present volcanic unrest interpretation.
Key Innovation: A review of seismo-stratigraphic data links subsurface structure in the Gulf of Pozzuoli to the ongoing bradyseismic crisis and volcanic hazard context.
11. Multiscale analysis of tunnel behavior in response to normal fault rupture: combining PFC3D–FLAC with centrifuge experiments
Core Problem: Tunnel response to active fault rupture involves micromechanical soil rearrangement that continuum models can underresolve.
Key Innovation: Centrifuge experiments are coupled with PFC3D-FLAC simulations to connect force-chain evolution and particle motion with tunnel lining deformation and settlement.
12. A Digital Twin-Enabled 3D Real-Time Monitoring System of Subway Tunnel Waterlogging for Disaster-Resilient Infrastructure
Core Problem: Urban underground flood response needs real-time spatial awareness inside complex tunnel networks rather than static emergency maps.
Key Innovation: A digital-twin monitoring system integrates 3D tunnel representation and real-time waterlogging information to support disaster-resilient subway infrastructure.
13. Heuristic quality coefficients for interferometric phase linking
Core Problem: Distributed-scatterer InSAR workflows need reliability indicators for phase-linking estimates, especially when non-maximum-likelihood methods are used for deformation mapping.
Key Innovation: Three heuristic quality coefficients are formalized within a common phase-linking framework, giving practical uncertainty screens for multitemporal InSAR deformation products.
14. PGDM: Physically guided diffusion model for land surface temperature downscaling
Core Problem: Land-surface temperature downscaling often gains spatial detail at the cost of physical consistency, limiting use in mountain and urban hazard contexts.
Key Innovation: A physically guided diffusion model embeds thermal constraints into super-resolution downscaling, providing a transferable architecture for high-resolution environmental risk mapping.
15. Use of Distributed Fiber Optic Sensing to Measure Differential Settlements Induced by a Normal Fault
Core Problem: Critical infrastructure needs spatially continuous monitoring of differential settlement induced by fault displacement.
Key Innovation: Laboratory fault-deformation tests compare buried fiber-optic cables and a strip sensor, showing how distributed strain and curvature can reconstruct subsurface settlement profiles.
16. Satellite based forensic MSRS monitoring system for detecting settlement of urban underground
Core Problem: Satellite-detected subway settlement anomalies need field-scale confirmation before they can support forensic diagnosis.
Key Innovation: A staged MSRS workflow combines InSAR, laser scanning, and GPR to move from corridor-scale settlement screening to shaft-level deformation and near-surface condition assessment.
17. Insights into the rockburst control mechanism by slit-filling-based advanced pressure relief technology under biaxial stresses
Core Problem: Pressure-relief measures for rockburst control must be tied to the actual stress redistribution mechanism around excavations.
Key Innovation: Biaxial tests clarify how slit-filling pressure relief modifies damage development and supports advanced borehole design for rockburst prevention.
18. Prediction of progressive sandstone fracturing under freeze-thaw cycles and seepage-stress coupling via cepstrum features and deep learning network optimization
Core Problem: Progressive rock fracturing under freeze-thaw and seepage-stress coupling is difficult to predict from raw acoustic or mechanical signals.
Key Innovation: Cepstrum features and optimized deep-learning networks are used to forecast sandstone damage evolution under coupled cold-region deterioration processes.
19. Water pressure distribution around tunnels in pipeline-type karst and lining response characteristics induced by rainfall
Core Problem: Rainfall-driven water pressure around tunnels in pipeline-type karst can overload linings in ways not captured by steady seepage assumptions.
Key Innovation: Experiments and modelling quantify pressure distribution and lining response under rainfall-controlled tunnel-karst hydraulic connectivity.
20. Seepage of a deep-buried tunnel in large-scale fault zones: Effects of fault heterogeneity and excavation process
Core Problem: Deep-buried tunnel seepage risk depends on fault heterogeneity and excavation sequence, not simply on average permeability.
Key Innovation: A numerical framework tests how fault-zone spatial variability and excavation process control groundwater inflow and seepage redistribution.
21. A quantitative risk assessment framework for fault reactivation in underground hydrogen storage: Coupled simulation and deep learning approach
Core Problem: Underground hydrogen storage can perturb stresses around faults, creating coupled leakage and induced-slip risks that are hard to evaluate rapidly.
Key Innovation: Coupled simulations are paired with deep learning to quantify fault reactivation risk under storage scenarios.
22. Numerical evaluation of fault reactivation and leakage risk in CO2 geological storage
Core Problem: CO2 injection can raise pore pressures in fault zones, but many leakage-risk assessments remain two-dimensional or over-simplified.
Key Innovation: A 3D Ordos Basin saline-aquifer model evaluates pressure response, leakage rates, and controlling factors for fault reactivation during long-term storage.
23. A simplified physics model for estimating subsurface CO2 storage resources constrained by fault slip potential
Core Problem: Regional CO2 storage screening needs fast estimates of subsurface capacity that also respect fault-slip constraints.
Key Innovation: A simplified physics model constrains storage resources with fault slip potential, linking capacity assessment to geomechanical safety.
24. Modelling of shear creep behaviour in unsaturated rough rock joints using the grain-based stress corrosion model
Core Problem: Rock slopes and underground openings can fail through delayed shear creep along rough joints under unsaturated conditions.
Key Innovation: A grain-based stress-corrosion model simulates shear creep behavior in unsaturated rough rock joints and clarifies moisture-sensitive time-dependent deformation.
25. Shear Behavior and Constitutive Models of Rock-Like Samples with Different Joint Roughnesses: A Novel Methodology Based on 3D Printing
Core Problem: Shear behaviour of rough rock joints remains hard to standardize because natural joint geometry is difficult to reproduce.
Key Innovation: 3D-printed rock-like samples create controlled joint-roughness series for shear testing and constitutive model calibration.
26. Mesoscopic characterization of visualized damage-induced crack propagation in calcareous-cemented medium-fine grained conglomerate
Core Problem: Weakly cemented conglomerate roofs in coal-bearing strata can transition from local cementation damage to wider seepage-connected fracture networks.
Key Innovation: Mineralogy, acoustic emission, micro-CT, and seepage tests resolve a cementation-crushing to crack-coalescence damage chain relevant to roof stability and rockburst prevention.
27. Cluster analysis of rock block shapes: A novel quantitative method for characterising rock mass structures
Core Problem: Rockfall and rock-mass analyses need quantitative block-shape descriptions rather than subjective geometric classes.
Key Innovation: Cluster analysis provides a reproducible method for classifying rock block shapes and characterising rock mass structure.
28. Application prospect of machine learning bridging rock/lithology identification and engineering rock mass characterization: A review
Core Problem: Machine-learning studies on lithology identification and rock-mass characterization remain fragmented across sensors, scales, and engineering contexts.
Key Innovation: A question-driven review synthesizes 248 studies and identifies multimodal data fusion, transformers, interpretability, and physically grounded lightweight models as the main path forward.
29. Object Detection in Optical Remote Sensing Images: A Systematic Review of Methods, Benchmarks, and Operational Applications
Core Problem: Operational remote-sensing object detection faces annotation bottlenecks, scale variation, and cross-region domain shifts that directly affect disaster mapping.
Key Innovation: A systematic review of methods, benchmarks, and applications identifies foundation models, efficient architectures, and size-specific metrics as priorities for operational deployment.
30. Efficient Adaptation of Vision Foundation Model for High-Resolution Remote Sensing Image Segmentation via Spatial-Frequency Modeling and Sparse Refinement
Core Problem: Vision foundation models are powerful but difficult to adapt efficiently to high-resolution remote-sensing scenes with sensor and texture variability.
Key Innovation: Spatial-frequency modelling and sparse refinement are used to adapt a foundation model for remote-sensing segmentation without full retraining.
31. Physics-Informed Neural Network for Bathymetry Inversion Coupling Seafloor Slope Effects and Radiative Transfer Constraints Using ICESat-2 and Sentinel-2 Data
Core Problem: Satellite bathymetry inversion must account for seafloor slope and radiative-transfer constraints to avoid terrain-induced depth bias.
Key Innovation: A physics-informed neural network couples slope effects, ICESat-2 elevation, Sentinel-2 reflectance, and radiative-transfer constraints for bathymetry inversion.
32. FD-SC: focus-diffusion self-contrastive framework empowers efficient remote sensing building damage detection
Core Problem: Remote earthquake damage mapping often struggles with domain shift and limited labeled examples.
Key Innovation: A focus-diffusion self-contrastive framework is proposed for efficient remote-sensing building damage detection.
33. MineFormer: A lightweight edge-aware transformer for high-resolution semantic segmentation of open-pit mining area
Core Problem: High-resolution open-pit mining segmentation is computationally expensive and sensitive to edge errors around haul roads, benches, and disturbed slopes.
Key Innovation: MineFormer uses a lightweight edge-aware transformer to improve semantic segmentation of open-pit mining areas from remote-sensing imagery.
34. HydroGRAF: Hybrid discharge reconstruction and basin-aware streamflow forecasting in the Himalayas
Core Problem: Snow-monsoon Himalayan basins often have incomplete discharge records yet require probabilistic short-term flood forecasting.
Key Innovation: HydroGRAF reconstructs missing discharge with hybrid seasonal predictors and uses a basin-aware graph attention network for streamflow forecasting.
35. Climatology and Trends in Spatial Scales of Extreme Precipitation Over Land in the Contiguous US
Core Problem: Extreme precipitation risk is often summarized by intensity or frequency, while spatial extent controls the scale of flood impacts.
Key Innovation: Connected-component analysis of 1980-2024 U.S. precipitation shows that extremes are concentrating into fewer days with larger event footprints, especially in the eastern United States.
36. Multi-index assessment of flash drought hazards in North Khorasan province, northeastern Iran
Core Problem: Flash drought detection in semi-arid regions requires indices that capture rapid onset, vegetation stress, and extreme tails under sparse observations.
Key Innovation: SPEI, ESI, FDII, ARIMA-GEV modelling, and wavelet analysis are integrated to map drought severity and tail risk in North Khorasan, Iran.
37. Cosmogenic 36Cl Dating of Fault Activity in East Messinia, Greece
Core Problem: Long-term slip rates are needed for seismic hazard assessment where instrumental seismicity is too short to characterize fault loading.
Key Innovation: Cosmogenic 36Cl exposure dating constrains late Holocene slip rates along the Eastern Messinia Fault Zone in Greece.
38. Morphotectonic characteristics of the Xiongpo anticline in the southwestern Sichuan Basin, China: Implications for active folding mechanisms
Core Problem: Active folds can redistribute deformation across basins, complicating seismic hazard interpretation from surface morphology alone.
Key Innovation: Morphotectonic analysis of the Xiongpo anticline clarifies active folding mechanisms in the southwestern Sichuan Basin.
39. Relative Uplift Rates Along the Central Mindoro Fault, Philippines
Core Problem: Fault segments with limited instrumental records still require uplift-rate constraints for long-term seismic hazard context.
Key Innovation: Geomorphic analysis estimates relative uplift rates along the Central Mindoro Fault in the Philippines.
40. Characterizing soil dynamic parameters through ambient seismic noise in the Sakarya region of the north anatolian fault system
Core Problem: Local soil dynamic parameters are often poorly constrained in active fault regions where they control ground-motion amplification.
Key Innovation: Ambient seismic noise is used to characterize dynamic soil parameters in the Sakarya region of the North Anatolian fault system.
41. Improved methods for site classification using statistical analysis of ground motion records
Core Problem: Seismic site classification can be unstable when ground-motion records are sparse or statistically inconsistent.
Key Innovation: Improved statistical methods are proposed for site classification from ground-motion records.
42. Experimental evaluation of axial restraint and pretension on the response of fusible PVC pipes crossing a normal fault
Core Problem: Buried lifelines crossing normal faults must accommodate permanent ground deformation without rupture or joint opening.
Key Innovation: Large-scale tests evaluate how axial restraint and pretension affect fusible PVC pipe bending response during normal-fault offset.
43. Investigating the impact of strike-slip faults on adjacent buildings
Core Problem: Adjacent urban buildings near strike-slip faults may experience foundation interaction and pounding under permanent ground displacement.
Key Innovation: Numerical experiments analyze how strike-slip fault displacement affects closely spaced rigid mat foundations and building response.
44. Evaluation of post-earthquake functionality of water supply network based on dual indices
Core Problem: Water-supply functionality after earthquakes is misestimated when pressure and node-supply metrics are treated separately.
Key Innovation: A probabilistic Monte Carlo framework with PSO-calibrated dual indices evaluates post-earthquake network functionality.
45. A Voronoi-cell stochastic finite element method for seismic analysis of gravity dams with material spatial variability
Core Problem: Material spatial variability can alter gravity-dam seismic response but is expensive to propagate through full finite-element simulations.
Key Innovation: A Voronoi-cell stochastic finite-element method is developed for seismic analysis of gravity dams with heterogeneous material fields.
46. Comparative analysis of two evapotranspiration landfill covers incorporating long-term environmental effects on soil permeability
Core Problem: Long-term permeability change can undermine both percolation control and slope stability in landfill covers.
Key Innovation: Numerical comparison of monolithic and capillary-barrier covers shows how permeability increase shifts critical slip surfaces and long-term instability risk.
47. The geomorphology of the mature, fluviokarst-dominated Mesijune salt extrusion (Zagros Mountains, Iran). Insights into the evolution of salt fountains
Core Problem: Salt extrusions can form complex fluviokarst landscapes whose cavities and collapse features are poorly represented in standard karst models.
Key Innovation: Geomorphic mapping of the Mesijune salt extrusion reconstructs salt-fountain evolution and fluviokarst landform development in the Zagros Mountains.
48. Does bed relief index inform about sediment supply in proglacial multi-channel systems? The Rutor case study (Italian Alps)
Core Problem: Proglacial channels need diagnostics that separate sediment-starved from transport-limited reaches under changing glacier inputs.
Key Innovation: A normalized Bed Relief Index is tested with high-resolution photogrammetric DSMs to compare sediment-transport regimes across Rutor Glacier reaches.
49. Timing of deglaciation and slope evolution in the Furkotská Valley, the High Tatra Mountains, Central Europe
Core Problem: Postglacial slope evolution controls sediment delivery and debris-flow inheritance in high mountain valleys.
Key Innovation: Chronology and geomorphic reconstruction in the High Tatra Mountains link deglaciation timing to slope and valley evolution.
50. Detecting Glacier Dynamics During 2016–2024 Using Planet Imagery in the Upper Zarafshon River Basin, Tajikistan
Core Problem: High mountain glacier dynamics require dense optical time series where field access is limited.
Key Innovation: Planet imagery from 2016-2024 is used to detect glacier dynamics in the upper Zarafshon River Basin.
51. Enhancing geomorphological mapping with multi-source data in a changing climate: Insights from a high-elevation experimental basin in the Italian Alps
Core Problem: Climate-sensitive alpine basins need geomorphic maps that integrate field, remote-sensing, and terrain evidence.
Key Innovation: Multi-source data are fused to improve geomorphological mapping in a high-elevation experimental basin in the Italian Alps.
52. Morphodynamic adjustments of the Lower Yangtze River during 1988–2023: Integrated effects of sediment reduction and rocky outcrops
Core Problem: Large dams can trigger long downstream adjustments in bank erosion, island morphology, and vegetation that affect floodplain stability.
Key Innovation: Landsat-based analysis of the Lower Yangtze River shows a shift from pre-dam accretion to post-dam scouring under sediment reduction and rocky-outcrop controls.
53. Soil erosion and erosion-induced lateral soil organic carbon loss associated with photovoltaic development across China
Core Problem: Photovoltaic development may create erosion-induced lateral soil carbon losses that remain undercounted in renewable-energy land assessments.
Key Innovation: China-wide modelling quantifies soil erosion and associated organic-carbon export linked to photovoltaic development.
54. Highway construction accelerates topsoil organic carbon loss in permafrost on the Qinghai-Tibet plateau
Core Problem: Highway construction in permafrost can accelerate soil carbon loss and alter thermal-hydrological stability along corridors.
Key Innovation: Qinghai-Tibet Plateau field and soil analyses quantify topsoil organic carbon loss associated with highway construction in permafrost terrain.
55. A numerical study on the influence of peat layer thickness on ground thermal regime along the Hudson Bay Railway
Core Problem: Peat layers can buffer or amplify ground thermal change along northern railways, affecting track stability under warming.
Key Innovation: Numerical simulations test how peat thickness controls ground thermal regime along the Hudson Bay Railway.
56. The influence of frozen climatic conditions on railway track settlement rate
Core Problem: Cold-climate trackbeds experience settlement through freezing persistence and thaw cycles, but maintenance models rarely include this forcing explicitly.
Key Innovation: A train-track-ground interaction framework links frozen climatic conditions to future railway track settlement rates.
57. Enhancing ground settlement prediction in deep excavations in soft soils through Bayesian updating of time-dependent model biases
Core Problem: Ground-settlement prediction in soft-soil excavations is biased by time-dependent soil behavior and model discrepancy.
Key Innovation: Bayesian updating with MCMC and a Bi-LSTM surrogate reduces settlement-prediction errors in a 30.2 m deep excavation case.
58. Prediction of Railway Concrete Track Settlement Using Deep Learning Algorithms
Core Problem: High-speed rail settlement under subgrade-dominated deformation requires uncertainty-aware long-term prediction from monitoring records.
Key Innovation: LSTM and GRU models are trained on 321 field monitoring locations to predict railway concrete track settlement.
59. Reliability assessment of front tunnel stability in twin tunnelling: A framework using an AutoML surrogate model accounting for soil spatial variability
Core Problem: Twin tunnelling can destabilize the front tunnel through disturbance superposition and spatially variable soil resistance.
Key Innovation: An AutoML surrogate built from parametric numerical models estimates serviceability and ultimate limit states while accounting for soil spatial variability.
60. Prediction of compressive index of clayey soils: A novel physics-guided neural network with consistency-gated fusion and ensemble uncertainty
Core Problem: Clay compressibility prediction needs generalization and uncertainty estimates before it can support risk-aware embankment design.
Key Innovation: A physics-guided neural network with consistency-gated fusion and ensemble uncertainty propagates compressive-index predictions into consolidation settlement analysis.
61. The Second Shift in Mountain Guiding and Avalanche Work: Intersectional risk, human-caused hazards, and risk-management effectiveness among non-dominant groups
Core Problem: Avalanche risk management can miss human-caused exposure and vulnerability that shape safety in professional mountain work.
Key Innovation: Interview-based analysis defines intersectional risk as a combination of mountain hazards, human-caused hazards, exposure, and vulnerability.
62. Making footprints move: Temporal disaggregation of building footprint data using Sentinel-2 imagery and Bayesian deep learning
Core Problem: Building-footprint inventories are often temporally static, weakening exposure models in rapidly changing settlements.
Key Innovation: Sentinel-2 imagery and Bayesian deep learning are used to temporally disaggregate building footprint data.
63. Exposure matters: A synthesis framework for high-resolution building inventory development
Core Problem: Regional hazard-impact models depend on building inventories, but data sources and synthesis methods remain inconsistent across studies.
Key Innovation: A synthesis framework is proposed for high-resolution building inventory development to support footprint-level disaster risk modelling.
64. Early Post-Fire Assessments of Wildfires in a Natural Mixed Forest in Northeastern Japan Using Sentinel-2 dNBR and UAV RGB Imagery
Core Problem: Early wildfire assessment needs to reconcile landscape-scale burn severity with fine-scale canopy damage before erosion and runoff impacts emerge.
Key Innovation: Sentinel-2 dNBR, UAV RGB imagery, and field plots characterize burn severity and tree health after the 2024 Nanyo wildfire.
65. A survival guide for assessing global fire risks to natural and human systems
Core Problem: Fire-risk studies across natural and human systems remain difficult to compare because exposure, vulnerability, and response are defined inconsistently.
Key Innovation: A survival-guide review synthesizes concepts and assessment choices for global fire-risk analysis.