TerraMosaic Daily Digest: April 28, 2026
Daily Summary
April 28 is unusually landslide-centered. The strongest papers move from first-principles mechanics to monitored failure: an evolutionary drag formulation challenges calibrated runout friction, while the Huangci reactivation, buried-ice debris-flow flumes, rock-avalanche impact theory, reservoir impulse-wave analysis, and Longxigou space-air-ground monitoring show how structure, water, ice, particles, and observation geometry control failure and consequence. The collection is not only a sequence of case studies; it defines a common technical direction in which landslide interpretation must preserve process history, not just map the final scar or probability surface.
The wider geohazard set extends that logic into multi-hazard data, coseismic landslide susceptibility, tsunami mobility, coastal flood levels, seismic-risk scenarios, volcanic unrest, freeze-thaw rockslides, tailings mechanics, and tunnel or mine safety. AI appears most useful where it is constrained by geomorphic priors, uncertainty calibration, physical equations, or operational monitoring: KAN-UNet, MPES-YOLO, tabular foundation models, SPH calibration, explainable ground-motion models, and nighttime-light damage mapping are strongest when they expose controls rather than replace them with opaque classification.
Key Trends
The selected papers converge on process-resolved geohazard assessment: not whether an area is hazardous in the abstract, but how failure initiates, moves, interacts, and becomes observable.
- Landslide mechanics are being recast as evolving systems: drag, buried ice, particle gradation, impulse overlap, mining-induced block instability, freeze-thaw cracking, and transitional tailings behavior all treat failure as a changing mechanical state rather than a fixed parameter choice.
- Monitoring is becoming multi-perspective and failure-specific: Huangci, Longxigou, Medellin, reservoir-margin landslides, loess hazards, and urban structural damage papers combine InSAR, UAV, TLS, geomorphology, and field evidence to match sensors to deformation geometry.
- Dynamic susceptibility is replacing static zoning: climate projections, land-use scenarios, bootstrap uncertainty, slope units, Newmark displacement, and time-dependent reliability shift landslide maps from present-day classification toward scenario-conditioned risk.
- Cascading hazards are treated as coupled mobility problems: reservoir landslides, moraine-dam overtopping, landslide and earthquake tsunamis, coastal total water levels, and tropical cyclone rain studies connect source mechanics to downstream exposure and evacuation.
- AI contributes most when paired with process constraints: tabular foundation models, KAN-UNet, MPES-YOLO, SPH calibration, explainable ground-motion models, UAV terrain segmentation, and nighttime lights illustrate a move from generic accuracy metrics toward interpretable hazard controls.
Selected Papers
This digest features 80 selected papers from 1,669 papers analyzed. The sequence opens with landslide drag theory and the 2025 Huangci loess-mudstone reactivation, then moves through buried-ice debris-flow initiation, rock-avalanche impact forces, reservoir cascading hazards, Longxigou space-air-ground monitoring, Medellin deep-seated urban landslides, dynamic susceptibility and reliability modelling, moraine-dam overtopping, tsunami and coastal-flood scenarios, volcanic unrest, freeze-thaw rockslides, tailings mechanics, and process-constrained AI for geohazard observation.
1. The landslide drag
Core Problem: Landslide runout models often treat drag as a calibrated constant even though deformation, acceleration, and flow state evolve during motion.
Key Innovation: The paper derives an evolutionary drag coefficient governed by a dimensionless acceleration number, reframing drag as a dynamic measure of energy inefficiency in landslides and debris flows.
2. The reactivated Huangci landslide at the Heifangtai terrace, Gansu Province, China, on December 10, 2025
Core Problem: The 2025 Huangci reactivation required separation of inherited structural controls, irrigation-driven weakening, freeze-thaw triggering, and short-term canal failure.
Key Innovation: Field surveys, UAV photogrammetry, multi-temporal InSAR, local monitoring, and historical hazard knowledge reconstruct the event and show how integrated observation enabled zero-casualty warning.
3. Initiation mechanisms of glacial debris flows considering buried ice: insights from flume experiments
Core Problem: Glacial debris-flow initiation is poorly constrained where source deposits contain buried ice that alters pore pressure, fissuring, and erosion.
Key Innovation: Flume experiments isolate buried-ice melt proportions and show that ice-driven water release and internal cavities more than double erosion volume relative to ice-free deposits.
4. Cumulative impact forces of rock avalanches on bridge piers: role of particle-size distribution and impulse overlap
Core Problem: Bridge-pier impact design for rock avalanches usually relies on bulk-flow assumptions that miss particle-size distributions and overlapping impulse effects.
Key Innovation: An analytical model couples elastoplastic contact theory with statistical gradations to predict nonlinear cumulative impact forces and identify pier-collapse demand thresholds.
5. Potential cascading geohazard: El Arrecife landslide in Rules reservoir, Southeastern Spain
Core Problem: Active reservoir-margin landslides can cascade into impulse waves, road flooding, and dam overtopping, but threat assessment requires linked geological and hydraulic evidence.
Key Innovation: A multidisciplinary characterization of the El Arrecife landslide estimates an 8 million m3 unstable mass and simulates earthquake-triggered reservoir waves up to 46 m.
6. Decadal evolution and dynamics of the Longxigou landslide’s composite deforming surface revealed by space-air-ground investigation strategy
Core Problem: Composite landslides with differently oriented deforming surfaces are difficult to monitor with any single remote-sensing or ground technology.
Key Innovation: A space-air-ground strategy combines Sentinel-1, ALOS-2, UAV photogrammetry, and pile-based TLS to reconstruct ten years of Longxigou landslide evolution and rapid frontal displacement.
7. Neighbourhoods in motion: unveiling the drivers behind structural damage hotspots in Medellín (Colombia)
Core Problem: Urban structural damage hotspots can be misattributed to poor construction when slow hillslope deformation is the underlying driver.
Key Innovation: SBAS and PSI InSAR, active-deformation-area extraction, damage reports, and geomorphic mapping reveal deep-seated landslide sectors controlling damage in Medellin.
8. Characteristics and mechanism of an ancient landslide reactivation controlled by geostructure and extreme rainfall
Core Problem: Ancient landslide reactivation under extreme rainfall depends on buried geostructure, preferential flow, and inherited weak zones that are rarely observable directly.
Key Innovation: Integrated field, geophysical, and stability analysis links rainfall reactivation to structural controls and clarifies the mechanism of renewed motion in an ancient landslide.
9. Deformation and instability mechanism of catastrophic landslide triggered by multiple coal seams repeated mining in peak cluster landforms
Core Problem: Repeated coal-seam mining beneath peak-cluster landforms can initiate catastrophic slope collapse, but the mechanical transition from goaf subsidence to landslide is uncertain.
Key Innovation: A mechanical model of overlying strata explains collapse zones, ground fissures, key-block instability, and the shear-instability to fragmented-sliding collapse pathway.
10. Creation and analysis of a multi-hazard dataset: Tenerife (Canary Islands) as a case study
Core Problem: Multi-risk assessment is limited when volcanic, earthquake, flood, landslide, and extreme-weather records are held as separate historical narratives.
Key Innovation: A Tenerife dataset compiles and analyzes more than 500 years of multi-hazard evidence, providing a reusable basis for anticipatory risk assessment and mitigation planning.
11. Knowledge-Data Dually Driven Paradigm for Accurate Landslide Susceptibility Prediction under Data-Scarce Conditions Using Geomorphic Priors and Tabular Foundation Model
Core Problem: Landslide susceptibility prediction remains fragile in data-scarce regions where purely data-driven models lack geomorphic structure.
Key Innovation: A knowledge-data dual paradigm combines geomorphic priors with a tabular foundation model to improve susceptibility prediction under limited training data.
12. Energy and structural evolution process of high-altitude and long-runout landslides induced by a strong earthquake
Core Problem: High-altitude earthquake-induced long-runout landslides require explanation of how energy and structure evolve from initiation to deposition.
Key Innovation: The study analyzes energy conversion and structural evolution to clarify the movement process of strong-earthquake-triggered high-altitude long-runout failures.
13. Investigation on the overtopping failure of moraine dams under avalanche-induced surge waves
Core Problem: Avalanche-induced surge waves can breach moraine dams, but failure depends on repeated overtopping and erosion rather than only a single maximum wave.
Key Innovation: Flume tests identify effective surge-wave repetitions, net overtopping height, and duration as controls on incision, gully growth, accelerated erosion, and dam instability.
14. A KAN-UNet identification method based on multi-source remote sensing data integration for potential landslide hazards in the Loess Plateau
Core Problem: Potential landslides in the Loess Plateau are difficult to identify because terrain, optical, and deformation signals are distributed across heterogeneous data streams.
Key Innovation: KAN-UNet integrates multi-source remote-sensing information to identify potential landslide hazards and improve spatial extraction of unstable loess terrain.
15. Complex dendrogeomorphic analysis of mass movements in the steep slopes of the Cretaceous canyon
Core Problem: Steep canyon slopes can host clustered debris flows, landslides, and rockfalls without the typical high-mountain setting used to infer hazard rates.
Key Innovation: Dendrogeomorphic analysis of 454 tree-ring series reconstructs process timing, magnitude-frequency relations, hydrometeorological triggers, and displacement rates.
16. A preliminary regional inventory of deep-seated gravitational slope deformations in the Central Western Carpathians: geological and geomorphological patterns
Core Problem: The Central Western Carpathians lacked a consistent inventory of deep-seated gravitational slope deformations despite their hazard relevance.
Key Innovation: High-resolution LiDAR and geomorphic-structural interpretation identify 298 DSGSDs and classify their lithostratigraphic, structural, kinematic, and evolutionary controls.
17. Reliable probabilistic landslide susceptibility mapping using a calibrated stacked ensemble with bootstrap-based uncertainty
Core Problem: Probabilistic landslide susceptibility maps often report unstable predictions without calibrated uncertainty.
Key Innovation: A calibrated stacked ensemble with bootstrap-based uncertainty improves reliability of susceptibility estimates and communicates spatial confidence.
18. Integration of machine learning model and CMIP6 analysis for climate change impact-led landslide susceptibility and population exposure assessments in the Nepal Himalaya
Core Problem: Future landslide exposure in the Nepal Himalaya depends on climate-driven susceptibility change and population distribution, not present-day hazard alone.
Key Innovation: Machine learning, CMIP6 forcing, and exposure analysis jointly project climate-change impacts on landslide susceptibility and population risk.
19. A machine learning-aided surrogate model for time-dependent reliability analysis of Baishuihe landslide under rainfall considering spatially variable soils
Core Problem: Time-dependent reliability of rainfall-affected landslides is expensive to evaluate when soil properties vary spatially.
Key Innovation: A machine-learning surrogate accelerates reliability analysis of the Baishuihe landslide while preserving rainfall forcing and spatial variability.
20. Integration of interpretable machine learning and PLUS model for dynamic landslide susceptibility mapping of urban settlements in the Three Gorges Reservoir area
Core Problem: Mountain urban settlements require landslide susceptibility forecasts that account for changing land use, triggers, and factor heterogeneity.
Key Innovation: Interpretable machine learning and PLUS land-use simulation generate scenario-specific dynamic susceptibility maps for the Three Gorges Reservoir area.
21. Slope unit-based susceptibility analysis in near-fault Zones: Integrating Newmark displacement, frequency ratio, and pulse-like seismic features
Core Problem: Coseismic landslide hazard mapping is sensitive to slope-unit delineation and often misses pulse-like near-fault ground-motion effects.
Key Innovation: A slope-unit framework integrates Newmark displacement, frequency ratio, and pulse-like seismic features to improve near-fault susceptibility analysis.
22. An SPH framework with dynamic parameter calibration for landslide hazard simulation and risk assessment
Core Problem: SPH landslide simulations are sensitive to uncertain parameters, limiting their use for hazard zonation and event reproduction.
Key Innovation: Dynamic parameter calibration links particle-system theory, footprint tracking, and iterative optimization to reproduce the 2017 Xinmo landslide hazard process.
23. MPES-YOLO: A Multi-Scale Lightweight Framework with Selective Edge Enhancement for Loess Landslide Detection
Core Problem: Loess landslides are spectrally and morphologically similar to surrounding terrain, especially when small, shallow, or boundary-poor.
Key Innovation: MPES-YOLO adds multi-scale partial convolution, selective edge enhancement, and SIoU geometry constraints for lightweight loess-landslide detection.
24. A μ(Im) model and SPH implementation for dense granular flows with coupled rate and bond effects
Core Problem: Dense granular-flow models need to represent rate effects and bonding without losing numerical robustness in landslide-scale deformation.
Key Innovation: A modified mu(Im) formulation with SPH implementation couples rate and bond effects and is demonstrated on column collapse and engineering-scale landslides.
25. Influence of debris flow gradation on movement characteristics and accumulation distribution: a case study of Huashan Mountain in Jinan, China
Core Problem: Debris-flow runout and deposition depend strongly on gradation, yet particle-size effects are often simplified in movement models.
Key Innovation: A Huashan Mountain case study analyzes how debris-flow gradation controls movement characteristics and accumulation distribution.
26. Area system failure probability and risk assessment of slope considering the spatial variability of soil properties
Core Problem: Slope risk assessments based on a single critical slip surface can underestimate failure probability in systems with multiple potential slip surfaces.
Key Innovation: Limit equilibrium, Monte Carlo simulation, and discretized meshes quantify area-level failure probability and map risk across multiple failure surfaces.
27. The impact of earthquake-induced debris on the near-field tsunami pedestrian mobility and safety: Tumaco Island case of study
Core Problem: Near-field tsunamis can move earthquake-induced debris through pedestrian evacuation routes, but mobility and safety impacts are rarely resolved.
Key Innovation: The Tumaco Island scenario links debris transport with pedestrian mobility to quantify how earthquake-tsunami debris changes near-field safety.
28. Integrated structural and nature-based defense systems to reduce tsunami inundation under a megathrust earthquake scenario
Core Problem: Megathrust tsunami risk reduction requires comparison of engineered barriers and nature-based defenses under the same inundation scenario.
Key Innovation: Integrated structural and nature-based defense systems are tested for their ability to reduce tsunami inundation under a megathrust earthquake scenario.
29. Estimating the source of a 3500–4500 year old tsunami in the South Pacific region based upon boulder transport modelling
Core Problem: The source of a 3500-4500 year old South Pacific tsunami remains uncertain where deposits are preserved mainly as transported boulders.
Key Innovation: Boulder-transport modelling is used to constrain plausible tsunami sources and improve long-term regional tsunami hazard interpretation.
30. Assessing extreme total water levels across Europe for large-scale coastal flood analysis
Core Problem: Large-scale coastal flood assessment needs consistent extreme total water levels across Europe rather than fragmented local boundary conditions.
Key Innovation: The study estimates extreme total water levels for continental-scale coastal flood analysis, supporting dynamic inundation and risk modelling.
31. Seismic hazard and risk assessment for Delhi-NCT considering future earthquakes along two major active faults: Strategic planning for disaster risk reduction
Core Problem: Delhi-NCT risk planning must account for future earthquakes on major active faults rather than relying only on historical shaking.
Key Innovation: Scenario-based seismic hazard and risk assessment evaluates two active-fault sources to support disaster-risk-reduction planning.
32. Accelerating unrest at Campi Flegrei signals a critical transition within the next decade
Core Problem: Campi Flegrei unrest requires detection of whether multiparameter acceleration signals indicate approach toward a critical transition.
Key Innovation: The study analyzes accelerating unrest signals and argues for a critical-transition interpretation within the next decade.
33. Daily Nighttime Lights for Rapid Post-Earthquake Damage Assessment: Multi-Scale and Azimuthal Differences from the Mw 7.7 Myanmar Earthquake
Core Problem: Rapid post-earthquake damage mapping must handle scale and azimuthal differences in nighttime-light observations after major earthquakes.
Key Innovation: Daily nighttime lights are analyzed for the Mw 7.7 Myanmar earthquake to assess multi-scale and azimuth-dependent damage signals.
34. Freeze-thaw cycles threaten high-altitude cold-region rock slopes stability: Advances in SPH simulation
Core Problem: Cold-region rock slopes are increasingly threatened by freeze-thaw cycles, but progressive rockslide simulation must couple thermal, mechanical, and damage processes.
Key Innovation: A thermal-mechanical-damage SPH framework captures thaw weakening, frost-heave cracking, ice-wedge propagation, and sliding-toppling failure in jointed rock slopes.
35. The mechanics of a mineral sand tailings with a transitional behaviour
Core Problem: Tailings-dam stability analyses often assume a unique critical-state line, yet mineral sand tailings can show transitional behavior.
Key Innovation: Oedometer and triaxial tests reveal non-unique normal compression and critical-state lines and clarify how density affects transitional tailings mechanics.
36. Rock surface luminescence dating of the 2000 CE Yigong Flood Cobbles, southeastern Tibet: Implications for dating catastrophic outburst flood events
Core Problem: Catastrophic outburst-flood deposits are difficult to date because luminescence signals in rapidly transported coarse clasts are often incompletely reset.
Key Innovation: Rock-surface luminescence dating of 2000 CE Yigong flood cobbles tests bleaching mechanisms and defines depositional settings with higher dating potential.
37. A deep learning architecture and workflow for geomorphic feature extraction using digital terrain model‐derived land surface parameters
Core Problem: Geomorphic-feature extraction from DTMs usually requires separate preprocessing of terrain parameters and manual feature preparation.
Key Innovation: A deep-learning workflow embeds land-surface-parameter calculations within UNet-style architectures and releases R and Python implementations for terrain segmentation.
38. Evaluation of flood susceptibility through an artificial neural network-based differential evolution optimization algorithms and GIS techniques
Core Problem: Flood susceptibility mapping needs optimization strategies that can handle nonlinear interactions among terrain, hydrology, and land-use predictors.
Key Innovation: An artificial-neural-network model optimized by differential evolution is coupled with GIS to evaluate flood susceptibility.
39. Water vapor transportation features and key cause analysis of the “23.7” extreme rainstorm event in Hebei Province in 2023
Core Problem: The Hebei '23.7' extreme rainstorm requires attribution of water-vapor transport and key synoptic drivers for future flood-risk interpretation.
Key Innovation: The study diagnoses water-vapor pathways and causal features controlling the 2023 extreme rainfall event.
40. Empirical transfer function for enhanced seismic hazard modelling in the Northeast Himalaya
Core Problem: Synthetic ground-motion modelling in the Northeast Himalaya is limited by scarce recorded strong motions and uncertain site amplification.
Key Innovation: An empirical transfer-function correction from 447 accelerograms improves stochastic simulations and maps local amplification characteristics.
41. Interactive seismic hazard map of the Czech Republic
Core Problem: Regional seismic-hazard communication requires transparent, accessible tools rather than static maps alone.
Key Innovation: An interactive seismic hazard map of the Czech Republic provides an operational interface for exploring hazard outputs.
42. Explainable machine learning for ground motion modeling: predictive dominance of energy-based measures in the Southern Turkey seismic sequence
Core Problem: Ground-motion models need interpretable predictors that explain why some intensity measures dominate after major earthquake sequences.
Key Innovation: Explainable machine learning identifies energy-based measures as dominant predictors for the Southern Turkey seismic sequence.
43. Earthquake magnitudes depend on seismic history, as revealed by a neural network analysis
Core Problem: Earthquake magnitude estimation usually treats events independently, despite possible dependence on seismic history.
Key Innovation: A neural-network analysis shows that prior seismic history carries predictive information for earthquake magnitudes.
44. Seismicity, Repeating Earthquakes, and Tomographic Imaging of the Blanco Transform Fault System, Northeast Pacific
Core Problem: Transform-fault seismicity and repeating earthquakes require integrated catalog and tomographic constraints to resolve fault-system behavior.
Key Innovation: Seismicity, repeating-earthquake analysis, and tomographic imaging characterize the Blanco Transform Fault system.
45. The 1572 CE Santorini Eruption from Little-Known Historical Documents
Core Problem: The 1572 Santorini eruption remains poorly constrained because key historical documents have been underused.
Key Innovation: A 1588 Greek manuscript and historical maps revise the eruption timing, location, impacts, island formation, and likely VEI.
46. Research Progress on Intelligent Fault Recognition Technology in Seismic Exploration
Core Problem: Fault interpretation in seismic exploration remains labor-intensive and uncertain in deep or structurally complex targets.
Key Innovation: The review organizes intelligent fault-recognition methods and identifies needs for real benchmark datasets, geological priors, and interpretable models.
47. Fracture and fault characterization in the era of artificial intelligence
Core Problem: AI methods for fracture and fault characterization are advancing rapidly but remain fragmented across imaging, interpretation, and modelling workflows.
Key Innovation: The review synthesizes how artificial intelligence is changing fracture and fault characterization and identifies methodological gaps.
48. Assessment of soil erosion processes by applying IntErO model within the Mkhdach catchment (Middle Atlas of Morocco)
Core Problem: Mediterranean mountain catchments require erosion-risk estimates that combine field gully monitoring with basin-scale model outputs.
Key Innovation: The IntErO model and GIS are validated with field monitoring to estimate severe sediment yield and rainfall-driven erosion in Morocco.
49. Water at risk in India: a review of national diagnostics and adaptation priorities for sea level rise and groundwater stress
Core Problem: Sea-level rise and groundwater stress interact in coastal and deltaic regions but are often assessed through separate diagnostic frameworks.
Key Innovation: The review links groundwater depletion, seawater intrusion, compound flooding, and Indian coastal-risk frameworks into adaptation priorities.
50. Satellite-Observed Acceleration in the Occurrence of Compound Marine Heatwave and Phytoplankton Bloom Events in the Global Coastal Ocean
Core Problem: Compound marine heatwave and phytoplankton-bloom events are increasing, but their drivers and ecological risk regimes vary by latitude and nutrients.
Key Innovation: Coastal satellite observations and interpretable machine learning quantify accelerating compound events and identify nutrient-light controls.
51. A global parametric rain model for landfalling tropical cyclones: a case study for the U.S.
Core Problem: Landfalling tropical-cyclone flood hazard needs parametric rain fields that can be applied consistently at broad scale.
Key Innovation: A global parametric rain model is developed and evaluated for U.S. landfalling tropical cyclones.
52. Tropical cyclone landfall intensity (Vmax) for western North Pacific nations: return period and trends
Core Problem: Western North Pacific nations require return-period estimates and trend analysis for tropical-cyclone landfall intensity.
Key Innovation: The study estimates Vmax return periods and trends for landfall intensity across exposed nations.
53. A Bidirectional Spatiotemporal Deep Learning Model with Integrated Vegetation–Thermal Features for Wildfire Detection
Core Problem: Wildfire detection from remote sensing must distinguish fire signals from vegetation and thermal background dynamics.
Key Innovation: A bidirectional spatiotemporal deep-learning model integrates vegetation and thermal features for wildfire detection.
54. High-wind weighted machine learning for multi-constellation and multi-polarization GNSS-R wind speed retrievals from the Tianmu-1 mission
Core Problem: Spaceborne GNSS-R wind retrievals often degrade under the high-wind conditions most relevant to marine storms.
Key Innovation: High-wind weighted machine learning improves multi-constellation and multi-polarization wind-speed retrieval from Tianmu-1 GNSS-R observations.
55. The formation of terraces and channel pattern changes in mountain rivers: Findings from the upper Sava valley (Slovenia)
Core Problem: Mountain-river terraces and channel patterns record floodplain evolution, glacial sediment supply, and human regulation but are difficult to date coherently.
Key Innovation: Geophysics, geology, terrain models, and historical maps reconstruct terrace formation and planform shifts in the upper Sava valley.
56. Experimental assessment of near‐Bank submerged rigid vegetation in low‐ and high‐velocity zones: Implications for nature‐based riverbank protection
Core Problem: Nature-based riverbank protection must account for whether vegetation patches perform differently in low- and high-velocity flow zones.
Key Innovation: Flume experiments show density-dependent vegetation effects on velocity, turbulence, Reynolds stress, and near-bed shear under contrasting hydraulic zones.
57. Critical dynamic stress and strain of coarse-grained soil fillers in high-speed railway subgrades under complex service environments
Core Problem: High-speed railway subgrades degrade under coupled moisture variation and high-cycle loading, but deformation-state thresholds are difficult to define.
Key Innovation: Dynamic triaxial tests and a modified shakedown criterion quantify critical stress and strain under wetting-drying and frequency effects.
58. Dynamic behaviour of shallow rectangular tunnel at different cover depths: shaking table testing and numerical modelling
Core Problem: Shallow rectangular tunnels show cover-depth, flexibility, frequency, and interface effects that are hard to capture from design equations alone.
Key Innovation: Shaking-table tests and calibrated numerical models quantify bending moments, earth pressures, settlements, and asymmetric corner demand.
59. Influence of P-waves and SV-waves on seismic response of submarine shield tunnels under seepage conditions
Core Problem: Submarine shield tunnels under seepage conditions may respond differently to P-waves and SV-waves during earthquakes.
Key Innovation: Numerical simulations compare seismic response under wave type and seepage conditions for submarine shield tunnels.
60. Dynamic Deformation and Stability Evaluation of Red-Bed Interbedded Rock Filler Under Cyclic Loading with Wetting–Drying Effects: A New Shakedown Criterion
Core Problem: Red-bed interbedded rock filler experiences wetting-drying and cyclic loading that can shift it from stable shakedown to progressive deformation.
Key Innovation: Dynamic testing supports a new shakedown criterion for deformation and stability evaluation under cyclic hydro-mechanical loading.
61. Spatio-temporal evolution of cross-layer pollution during groundwater recharge in abandoned mines: Experimental study and numerical modeling
Core Problem: Groundwater recharge in abandoned mines can mobilize pollutants across layers, but coupled transport pathways remain difficult to quantify.
Key Innovation: Experiments and numerical modelling resolve spatiotemporal evolution of cross-layer pollution during mine groundwater recharge.
62. A systematic review of unmanned aerial vehicles (UAVs) for coastal ecosystem monitoring
Core Problem: UAV coastal ecosystem monitoring is expanding rapidly but lacks a consolidated view of sensors, workflows, validation, and interpretability limits.
Key Innovation: A PRISMA-based review of UAV coastal monitoring synthesizes 406 studies and details platforms, sensors, algorithms, validation practices, and emerging fusion needs.
63. Progress and Trends in UAV-Based Soil Moisture Inversion: A Comparative Knowledge Mapping Analysis of CNKI and Web of Science
Core Problem: UAV soil-moisture inversion has grown quickly across hyperspectral, thermal, and machine-learning approaches but remains methodologically scattered.
Key Innovation: A CNKI and Web of Science knowledge-mapping review tracks progress and highlights future coupling of physical mechanisms with deep learning.
64. A Feature-Optimized Deep Learning Framework for Mapping and Spatial Characterization of Tea Plantations in Complex Mountain Landscapes
Core Problem: Tea expansion onto steep subtropical mountain slopes increases habitat fragmentation and soil-erosion pressure but is hard to map in cloudy rugged terrain.
Key Innovation: A feature-optimized Sentinel-1/2 and VGG16-UNet++ framework maps plantations and derives steep-slope exposure and tea-forest interface indicators.
65. Unraveling the Spectral–Spatial Mechanisms of Mineral Identification: A Case Study on CASI Data Using SpectralFormer and Traditional Classifiers
Core Problem: Hyperspectral mineral identification must balance diagnostic spectral continuity with spatial context without obscuring rare alteration zones.
Key Innovation: SpectralFormer and traditional classifiers are compared on CASI data to reveal spectral-spatial mechanisms for mineral mapping.
66. Global and boreal estimates of woody aboveground biomass for 2020: Filling GEDI'S northern data gap with ICESat-2 and harmonized Landsat Sentinel-2
Core Problem: Boreal biomass mapping has a data gap because GEDI's orbit undersamples high northern forests, weakening disturbance and carbon baselines.
Key Innovation: ICESat-2, harmonized Landsat-Sentinel-2, and topography fill GEDI's northern gap for global and boreal aboveground-biomass estimates.
67. Freezing, not thawing, indirectly controls soil organic carbon heterogeneity in the Source Areas of the Yangtze and Yellow Rivers
Core Problem: Soil organic carbon heterogeneity in the Yangtze and Yellow River source areas is controlled by freeze-thaw processes that are often simplified.
Key Innovation: The study shows freezing, rather than thawing, indirectly controls spatial SOC heterogeneity in cold high-altitude source regions.
68. Asymmetric Responses of Spring and Autumn Phenology to Permafrost Degradation in the Source Region of the Yangtze River
Core Problem: Permafrost degradation can alter spring and autumn phenology through thaw timing, active-layer thickness, and soil-water loss.
Key Innovation: NOAA AVHRR NDVI, freeze-thaw products, and Stefan-model active-layer estimates reveal asymmetric vegetation responses in the Yangtze source region.
69. Three-dimensional modeling reveals lateral recharge dominates seasonal dynamics of dry soil layers under exotic vegetation on the Loess Plateau
Core Problem: Exotic vegetation on the Loess Plateau can form dry soil layers, but recharge pathways are often inferred from one-dimensional assumptions.
Key Innovation: Three-dimensional modelling shows lateral recharge dominates seasonal dry-soil-layer dynamics beneath exotic vegetation.
70. Hydraulic Effects of Channel Realignment and Floodplain Reconnection in a Headwater Stream
Core Problem: Channel realignment and floodplain reconnection change hydraulic conditions, but their effects are difficult to isolate in headwater restoration.
Key Innovation: Hydraulic analysis evaluates how channel realignment and reconnection alter flow in a headwater stream.
71. Experimental investigation on the mechanism of gripper TBM cutterhead clogging by mudstone in coal mines
Core Problem: Gripper TBM cutterhead clogging by mudstone in coal mines is poorly explained compared with clay clogging in EPB shields.
Key Innovation: Controlled tests and SHAP analysis identify moisture as the dominant clogging factor and define operational thresholds for mitigation.
72. Underground mine tunnel fire dynamics: Experimental investigation and AI-driven characterization of flame behavior
Core Problem: Mine tunnel fire behavior is difficult to characterize under ventilation, smoke, and visibility constraints.
Key Innovation: Full-scale experiments and multi-view computer-vision features classify fire size and ventilation conditions with high accuracy.
73. Machine Learning-Based Real-Time prediction of mass loss rate in tunnel pool fires under longitudinal ventilation
Core Problem: Real-time tunnel-fire assessment needs rapid estimates of mass loss rate under longitudinal ventilation.
Key Innovation: Machine learning predicts tunnel pool-fire mass loss rate from experimental and ventilation features.
74. Drained deformation behavior of in situ soda residue soil via long-term cyclic simple shear tests
Core Problem: Soda-residue fills in reclaimed wharf yards may undergo progressive settlement and abrupt deformation under drained cyclic traffic loading.
Key Innovation: Long-term cyclic simple-shear tests identify a cyclic-stress threshold and empirical deformation models for in situ soda residue soil.
75. Research on Strength Degradation and Crack Development in Defective Concrete
Core Problem: Tunnel linings with pre-existing cracks can lose strength through damage paths controlled by crack length and orientation.
Key Innovation: Acoustic emission, fractal crack analysis, and DEM simulation quantify strength degradation and crack development in defective concrete.
76. Dynamic evolution mechanism of mine accident risks: An interpretable machine learning framework for causal analysis
Core Problem: Mine accident risk evolves dynamically through interacting causal factors that are difficult to separate with descriptive statistics alone.
Key Innovation: An interpretable machine-learning framework analyzes causal evolution mechanisms for mine accident risk.
77. How warnings diffuse among the public: factors that influence self-organized evacuation decisions
Core Problem: Public warning diffusion can generate delayed, incomplete, or shadow evacuations when emergency communication networks are weak.
Key Innovation: A self-organized evacuation model evaluates how warning channels, loudspeaker range, and online groups shape evacuation decisions.
78. Rapid visual screening for earthquake risk reduction: application to institutional and residential buildings in MIT Muzaffarpur campus Bihar, India
Core Problem: Data-limited campuses and cities need rapid ways to prioritize buildings for detailed seismic assessment.
Key Innovation: Rapid visual screening integrates structural, non-structural, and preparedness indicators to classify earthquake risk for buildings in Bihar.
79. Numerical simulation of oblique incident S1 wave propagation at layered saturated frozen soil interface
Core Problem: Layered saturated frozen soils have frequency-, temperature-, and angle-dependent wave transmission that affects cold-region dynamic response.
Key Innovation: An analytical model derives transmission, reflection, mode conversion, and energy partition for obliquely incident S1 waves.
80. TAPping into homeowners’ decision making and emotions: A think-aloud experimental study of homeowner responses to flood buyouts
Core Problem: Flood buyout programs depend on homeowner risk perception, emotion, and decision processes that are not captured by economic exposure metrics alone.
Key Innovation: A think-aloud experimental study examines homeowner responses to flood buyouts and clarifies behavioral constraints on risk-reduction policy.