TerraMosaic Daily Digest: Jan 31, 2026
Daily Summary
This digest synthesizes 107 selected papers and focuses on seismic source-to-ground response pathways, landslide process mechanics and slope evolution, flood generation, routing, and hydroclimatic forcing. Top-ranked studies examine risk, fragility, and resilience assessment, satellite and LiDAR-based deformation monitoring, and earthquake-triggered slope response and liquefaction.
Across the full set, evidence converges on mechanism-constrained analysis with operational relevance, especially for infrastructure-focused hazard performance and high-resolution remote-sensing monitoring workflows. The strongest contributions pair interpretable process evidence with monitoring or forecasting workflows that support warning design and risk prioritization.
Key Trends
- Seismic hazard research links source behavior to ground response: Recurring topics connect rupture or loading conditions with geotechnical performance and consequence assessment.
- Landslide studies increasingly resolve process chains: Contributions connect triggering conditions, slope deformation, and mobility outcomes, improving the basis for warning thresholds and scenario testing.
- Flood analyses are becoming event-specific and process-based: Papers emphasize precipitation structure, antecedent wetness, and catchment controls rather than static hazard descriptors.
- Infrastructure-facing outputs are increasingly decision-ready: Asset performance is evaluated with uncertainty-aware frameworks to support mitigation and maintenance prioritization.
- Monitoring workflows rely on integrated remote-sensing products: Multi-source satellite and airborne observations are used for deformation retrieval, change detection, and rapid post-event mapping.
Selected Papers
This digest features 107 selected papers from 674 RSS items analyzed across multiple journals. Each paper has been evaluated for its relevance to landslide and broader geohazard research and includes links to the original publications.
1. Dynamics of population exposure to landslides under climate change scenarios
Core Problem: The insufficient study of the interaction between population dynamics and landslide processes, particularly in mountain regions experiencing increased landslide activity under climate change and rapid population growth.
Key Innovation: Proposal of a spatial, dynamic framework to estimate current and future populations exposed to landslides, integrating CNN with XAI for landslide susceptibility, projecting future landslide scenarios under climate change (SSP2-4.5, SSP5-8.5), and combining with demographic trajectories to produce fifty exposure scenarios.
2. STGCN-based inversion of landslide creep parameters using GNSS displacement time series
Core Problem: The significant challenges in accurately estimating landslide creep parameters due to the highly nonlinear and spatiotemporal nature of landslide deformation.
Key Innovation: Proposal of a novel GNSS-based method using a Spatiotemporal Graph Convolutional Network (STGCN) to invert landslide creep parameters, integrating spatial and temporal characteristics, which significantly improves accuracy (54.5% MAE reduction compared to BPNN) and effectively simulates spatiotemporal landslide evolution.
3. Forecasting tephra fall building impacts and losses at Awu volcano, Sangihe island (Indonesia)
Core Problem: The need to forecast tephra fall building impacts and losses at Awu volcano.
Key Innovation: Forecasting tephra fall building impacts and losses at Awu volcano, Sangihe island (Indonesia).
4. Influence of disturbance frequency on rockburst in circular openings: insights from cyclic loading experiments
Core Problem: Rockbursts induced by cyclic loading are a major hazard in deep underground excavations, and the role of cyclic loading frequency in their initiation and progression is not fully understood.
Key Innovation: Demonstrated that higher cyclic loading frequency significantly increases energy release rate and fragment ejection velocity, leading to increased pore density, accelerated strength deterioration, and earlier failure onset in sandstone specimens.
5. Sensitivity of hydrogeomechanical parameters in highly compressible aquitards via inverse modeling of one-dimensional subsidence
Core Problem: Land subsidence induced by groundwater extraction is a significant issue, especially in aquifer systems with thick, heterogeneous, and highly compressible aquitards, and identifying the key parameters controlling vertical deformation is crucial for model setup and reliability.
Key Innovation: Employed inverse modeling, coupling a one-dimensional nonlinear subsidence algorithm with stress-dependent parameters and the PEST tool, to evaluate the composite sensitivity of land subsidence to hydrogeological and geomechanical parameters in a stratified aquitard, revealing that total settlement and vertical deformation are more sensitive to hydraulic conductivity (K) than to compression index (Cc), with K controlling the rate of deformation.
6. Probabilistic active control of a seismically excited building using probabilistic fuzzy logic controller
Core Problem: Existing active vibration control systems for seismically excited buildings often struggle with linguistic and probabilistic uncertainties, leading to suboptimal performance in real-world applications.
Key Innovation: A novel Probabilistic Fuzzy Logic Controller (PFLC) that integrates fuzzy logic and probability theory to provide more accurate and robust active vibration control for seismically excited buildings with uncertain specifications, outperforming conventional controllers.
7. Rapid structural seismic response prediction using physics-informed inputs and scientific training strategies
Core Problem: Traditional methods for structural seismic response prediction are either computationally expensive (nonlinear time-history analysis) or lack accuracy and generalization (simplified models, conventional ML), hindering rapid and effective earthquake hazard mitigation.
Key Innovation: A novel deep learning method that integrates physics-informed input representations (response diagrams encoding time and frequency characteristics) with scientific training strategies and a hybrid transfer-learning framework, significantly enhancing computational efficiency, prediction accuracy, and generalization for rapid structural seismic response prediction.
8. Enhancing seismic resilience of precast segmental piers using external replaceable ring energy dissipator (ERRED)
Core Problem: Precast segmental piers (PSPs) have limited energy dissipation capacity and seismic resilience, restricting their use in high seismic regions and making post-earthquake repair challenging.
Key Innovation: A novel External Replaceable Ring Energy Dissipator (ERRED) integrated with precast segmental piers (PSP-ERRED) that significantly enhances lateral load capacity, energy dissipation, and stiffness, while reducing residual displacement and facilitating rapid post-earthquake repair for resilient bridge infrastructure.
9. Spatiotemporal evolution of deep fault slip and its coupled dynamic loading influence zone
Core Problem: Deep mining can induce fault slip and significant dynamic loading effects (rockbursts), but the mechanical mechanisms driving fault activation and the mutual feedback influence zone between mining and fault activation are not fully understood.
Key Innovation: A comprehensive analysis combining an elastic mechanical model, a fault slip evaluation model, FLAC3D numerical simulations, and in situ microseismic monitoring to elucidate the mechanical mechanisms of mining-induced deep fault activation, assess fault slip evolution, and delineate the coupled dynamic loading influence zone for enhanced support measures.
10. Analysis of dynamic response mechanism and control measures for tunnel crossing mountainous landslides: A shaking table test study
Core Problem: Understanding the dynamic response mechanisms and developing effective control measures for tunnels crossing mountainous landslides in seismically active regions, where coupled earthquake–tunnel–landslide interactions can lead to structural damage.
Key Innovation: A shaking table test study of a physical model (TULS) with a shock-absorbing layer (SAL) for tunnel lining, combined with VMD-HT analysis, revealing the four-stage damage evolution of the TULS and demonstrating SAL's effectiveness in mitigating seismic energy impact and slowing damage.
11. Research on Valley Deformation of High Arch Dam Based on Discontinuous Deformation Characteristics Analysis of Rock Mass Containing Structural Planes
Core Problem: The challenge of accurately analyzing and simulating the large, irreversible, and discontinuous valley (slope) shrinkage deformation in high arch dam reservoir areas, particularly due to the complex interaction of rock mass with numerous structural planes, which affects dam safety.
Key Innovation: Introduction and validation of the Nonlinear Spring Element method with a new structural plane constitutive model for analyzing discontinuous deformation of high slope rock mass, demonstrating its superior accuracy and convergence compared to FEM–SPH coupling, and its ability to truly reflect valley deformation for dam safety.
12. High predictability potential of highly synchronized widespread floods in monsoon regions
Core Problem: There is a noticeable absence of thorough investigations and analyses regarding the spatio-temporal characteristics, predictability of global widespread flood events, and the corresponding impact of climate indices.
Key Innovation: Employed recurrence quantification analysis to evaluate the predictability potential of globally widespread flood events, finding that highly synchronized widespread flood events (HSEs) exhibit higher predictability potential in monsoon regions and are intricately influenced by climate indices, establishing a connection between predictive capacity and flood occurrence.
13. Configurable physics-informed operator network for real-time multi-scenario hydrodynamics in river networks
Core Problem: Real-time hydrodynamic prediction in plain river networks is essential for flood early warning, but traditional numerical models are computationally expensive, and purely data-driven approaches lack generalization and physical consistency.
Key Innovation: Proposed a physics-informed river operator network (PI-RONet) where a configurable operator network (DeepONet/MIONet) handles dynamic boundary conditions and a PINN embeds Saint-Venant equations for physical consistency, achieving high accuracy (R2 ≥ 0.8) and nearly 5000x acceleration compared to numerical models for multi-scenario hydrodynamic simulation, supporting flood early warning and water management.
14. Attenuation mechanisms of ultralow-frequency seismic metamaterials via complex band structure analysis
Core Problem: Limited understanding of complex band structures for surface waves and their application in analyzing attenuation mechanisms of ultralow-frequency seismic metamaterials (SMs).
Key Innovation: Application of complex band structure analysis to ultralow-frequency seismic metamaterials, revealing that zero-frequency-starting surface wave bandgaps result from multiple dissipative modes for stronger attenuation, and interpreting the effects of viscoelasticity and soil stratification within this framework.
15. Semi-active impact damper with genetic algorithm-optimized fuzzy control for structural vibration reduction under various excitations
Core Problem: Conventional control strategies for semi-active impact dampers (SAIDs) are limited by their reliance on precise prediction of structural return to static equilibrium, making them less effective under complex and diverse excitations like seismic events.
Key Innovation: A fuzzy control-enhanced semi-active impact damper (FSAID) with a relay genetic algorithm (GA) optimized fuzzy controller, which operates independently of physical models and demonstrates superior robustness and adaptability in mitigating structural vibrations under both seismic and wind excitations compared to conventional SAIDs.
16. Porous vegetated concrete for slope protection: A review
Core Problem: Traditional slope protection and ecological restoration methods for rock slopes often suffer from high resource consumption and low vegetation survival rates, necessitating new, resource-efficient, and environmentally friendly solutions for sustainable stability and restoration.
Key Innovation: A comprehensive review of porous vegetated concrete (PVC) as an innovative ecological material for slope protection, systematically investigating its design principles, material composition, performance indices, and construction methods, highlighting its advantages in enhancing slope stability and promoting ecosystem restoration.
17. Defining the Soil Freezing Characteristic Curve Directly from the Generalized Soil Water Characteristic Curve
Core Problem: Existing Soil Freezing Characteristic Curve (SFCC) models either neglect adsorptive pressure, are too complex for practical use, or require direct SFCC measurements.
Key Innovation: Developed a novel, closed-form SFCC model that unifies adsorbed and capillary water freezing mechanisms, fully considers different pressures and interfacial conditions, and uses parameters directly from the Generalized Soil Water Characteristic Curve (SWCC), thus bypassing the need for SFCC data measurement.
18. Ground Freezing for Sampling a Fluvial Sand in Norway
Core Problem: Obtaining high-quality undisturbed sand samples for accurate geotechnical characterization remains a challenge.
Key Innovation: Successfully designed and implemented a ground freezing campaign for sampling fluvial sand, demonstrating that this method can recover minimally disturbed, representative high-quality sand samples, as validated by detailed monitoring and microcomputed tomography (μCT) analysis.
19. Experimental study of the seabed responses induced by a floating column with a heave plate
Core Problem: Oscillatory flows from floating platforms can induce seabed pore pressure gradients and shear stresses, potentially triggering sediment transport and scour, which impacts marine infrastructure.
Key Innovation: Conducting laboratory experiments to investigate seabed response around a floating column, identifying a radial zoning mechanism for scour formation (vortex-induced shear stresses at periphery, vertical pressure gradients at center) and the role of high-frequency oscillations in restricting sediment transport.
20. Is daily extreme rainfall increasing in the Mediterranean basin? A critical review of the evidence
Core Problem: The need for a critical review of evidence regarding trends in daily extreme rainfall across the Mediterranean basin, given the spatial and temporal heterogeneity in reported findings and the lack of a unified methodological framework.
Key Innovation: A critical review of 175 peer-reviewed studies on extreme precipitation trends in the Mediterranean basin, highlighting the spatial and temporal heterogeneity of findings, concluding that evidence does not support a basin-wide intensification, and emphasizing the influence of local factors over global climate forcing.
21. DVGBench: Implicit-to-explicit visual grounding benchmark in UAV imagery with large vision–language models
Core Problem: Existing remote sensing visual grounding (VG) datasets primarily rely on explicit referring expressions, limiting the performance of large vision-language models (LVLMs) on implicit VG tasks that require scenario-specific domain knowledge, especially for drone imagery in diverse application scenarios including disaster.
Key Innovation: Introduction of DVGBench, a high-quality implicit VG benchmark for drones covering six major application scenarios including disaster, providing both explicit and implicit queries; and DroneVG-R1, an LVLM integrating Implicit-to-Explicit Chain-of-Thought (I2E-CoT) within a reinforcement learning paradigm to convert implicit references into explicit ones, enhancing reasoning capabilities for drone-based agents.
22. Freeze-thaw damage differences in saturated limestone: Macro-meso-micro response and mechanism driven by initial properties
Core Problem: The underlying mechanisms of freeze-thaw induced rock damage, particularly associated with fissures and mineral composition in limestone, remain incompletely understood.
Key Innovation: Characterized multiscale damage evolution in four types of limestone specimens after 100 freeze-thaw cycles using physical testing, CT, SEM, and XRD, elucidating distinct damage mechanisms (surface spalling, micro-crack propagation, transgranular cracking) driven by initial properties and providing a theoretical basis for stability assessment and hazard mitigation in cold regions.
23. High-precision numerical simulation framework for the integrated modeling of urban “Source-Plant-Network-River” water environment
Core Problem: Urban water environments face increasingly severe pollution challenges, and rigorous, computationally efficient numerical models are indispensable for mitigating urban water pollution and managing inundation.
Key Innovation: Developed an advanced coupled model integrating 2D surface-water hydrodynamics/water-quality transport, 2D non-point-source pollutant (NPSP) build-up/wash-off, and 1D pipe-network drainage/pollutant discharge, using high-resolution structured grids, DLL-based bidirectional coupling, and GPU acceleration to enable high-precision integrated simulation of urban 'Source-Plant-Network-River' systems for evaluation, forecasting, and early warning.
24. Time-dependent deformations in deep tunnels: Insights into uncertainty and variability of rheological behavior
Core Problem: The uncertainty in determining rock mass properties significantly impacts tunnel stability, and squeezing conditions worsen tunnel stability, causing gradual convergence over time, making deterministic methods unreliable for long-term tunnel designs.
Key Innovation: Introduced an analytical method to calculate tunnel convergence in a visco-elastic rock mass, considering the uncertainty of key parameters, and used both visco-elastic and visco-elasto-plastic models with risk-based analyses to determine the probability distributions of tunnel wall deformations over time, revealing that long-term tunnel convergence often follows a right-skewed Gamma distribution.
25. Tunnel excavation and swelling analysis of expansive bedrock with multiphysics elasto-plastic model capable of describing different swelling behavior due to exchangeable cation species
Core Problem: Swelling of smectite-bearing bedrock causes severe tunnel deformation, with behavior depending on exchangeable cation species, and existing models don't fully capture this distinction.
Key Innovation: An extended multiphysics elasto-plastic model incorporating a double-layer repulsive force based on Stern theory, capable of capturing distinct swelling behaviors induced by different exchangeable cation species in expansive bedrock during tunnel excavation.
26. Variable-angle shear-compression instability in coal: Mechanistic insights and precursor identification under hydrochemical-mechanical coupling
Core Problem: Understanding the instability mechanisms of coal under variable-angle shear-compression fracture in coupled hydrochemical-mechanical environments (e.g., mine water saturation), and identifying reliable precursors for dynamic disasters in deep coal mining.
Key Innovation: A systematic investigation using uniaxial staged loading tests, AE monitoring, SEM, and DEM to elucidate coal damage evolution under hydrochemical-mechanical coupling, revealing the influence of shear angles and water chemistry, and proposing a precursory warning method based on critical slowing down (CSD) theory.
27. Unraveling rate-strengthening and fracture-weakening effects in fractured rock masses: A hybrid bonded particle model- discrete fracture network and physics-inspired machine learning framework
Core Problem: The coupled impact of loading rate and stochastic fracture network characteristics on the dynamic behavior of fractured rock masses remains poorly understood, hindering accurate assessment of stability in mining, tunneling, and underground construction under dynamic loading.
Key Innovation: A novel hybrid framework combining the bonded particle model (BPM), discrete fracture network (DFN), and physics-inspired machine learning (PIML) to investigate rate-strengthening and fracture-weakening effects, achieving 97.43% accuracy in predicting dynamic uniaxial compressive strength and quantifying contributions of loading rate and fracture intensity.
28. Numerical investigation: Undrained capacity of a hybrid anchor under VHMT combined loading in clay
Core Problem: Meeting increasing bearing capacity requirements for anchors in floating systems, and understanding the failure mechanisms and holding capacity of novel hybrid anchors under complex combined loading in clay.
Key Innovation: Numerical simulation analyzing the failure mechanisms and holding capacity of hybrid anchors under VHMT combined loading in clay, quantifying the influence of gravity mats, establishing equations for uniaxial ultimate bearing capacity, and developing a practical framework for hybrid anchor design based on failure envelope analysis.
29. Future directions for data-driven approaches in pipeline integrity management: Risk assessment, in-line inspection, and machine learning
Core Problem: Limitations of current data-driven approaches in pipeline integrity management, specifically in third-party damage assessment, in-line inspection (ILI) data quality, and machine learning-based corrosion evaluation, leading to insufficient failure probability evaluation and unstable model performance.
Key Innovation: Proposes a new GIS-based probabilistic approach for third-party damage assessment, analyzes ILI data replicability issues, and highlights the need for more robust, context-sensitive models for pipeline safety using data-driven strategies.
30. Data-driven risk analysis and management framework for rail hazmat transportation in Canada: Machine learning approach
Core Problem: Significant safety and environmental risks associated with rail transportation of hazardous materials (hazmat) in Canada, requiring a data-driven predictive risk assessment framework for hazmat release following railway accidents.
Key Innovation: A data-driven predictive risk assessment framework for hazmat release following railway accidents, integrating multi-year incident records with operational, environmental, and geographic variables, using supervised machine learning models to classify release outcomes and inform targeted mitigation strategies.
31. Transferring soil moisture estimation skills to evapotranspiration and streamflow modeling through remote sensing data assimilation
Core Problem: Uncertainty in land surface model (LSM) parameterization limits the effectiveness of soil moisture (SM) data assimilation (DA) in accurately improving estimates of hydrological fluxes like evapotranspiration (ET) and streamflow.
Key Innovation: An improved SM DA framework that first calibrates SM-ET and SM-runoff coupling strengths in the VIC model using remote sensing data, and then assimilates RS SM retrievals, leading to enhanced DA efficiency and significantly improved hydrological flux simulations, particularly in (sub-)humid regions.
32. The great decline of suspended sediment load in the Po River (Italy) over the last 100 years
Core Problem: The Po River, Italy's largest fluvial system, has experienced profound alterations in suspended sediment yields due to a sequence of human impacts over the last 100 years, but a comprehensive reconstruction of the spatiotemporal trajectory of these dynamics and their geomorphological consequences is needed.
Key Innovation: Integration of long-term suspended sediment transport data with detailed analysis of anthropogenic drivers (land-use, mining, damming, river training) to reconstruct the spatiotemporal decline of suspended sediment yields in the Po River (exceeding -72% at catchment closure), highlighting asynchronous reductions and their contribution to substantial geomorphological transformations affecting the delta region.
33. A dynamic soft-constrained deep learning paradigm for spatial downscaling of satellite gravimetry terrestrial water storage
Core Problem: The limited spatiotemporal resolution of GRACE/GRACE-FO satellite gravimetry data (coarser than 300 km, monthly) hinders its ability to quantify hydrometeorological and climate episodes effectively.
Key Innovation: Introduced a novel dynamic soft-constrained deep learning paradigm for spatial downscaling of TWSA from 300 km to 50 km, effectively preserving basin-averaged temporal dynamics while incorporating high-resolution spatial variability from hydrological models, and demonstrating improved localization of groundwater depletion/recharging signals.
34. Implications of structural flexure on CO<sub>2</sub> plume distribution and long-term reservoir deformation in the Canadian Aquistore CCS project
Core Problem: The unpredicted migration of injected CO2 plumes following structural basement flexure patterns in CCS projects, and the lack of understanding of how these flexure characteristics influence plume evolution, pressure buildup, and long-term reservoir deformation and leakage risk.
Key Innovation: A study integrating eight years of field-monitored CO2 injection data with systematic simulations to quantify the influence of deep structural flexure characteristics (distance, permeability, dip) on CO2 plume shape, pressure dynamics, and leakage risk, providing practical guidance for optimizing CCS in structural-flexure reservoirs.
35. Geomechanical insights for enhanced wellbore stability in the Sarvak Formation: A comprehensive analysis of pore pressure and fracture gradient
Core Problem: Wellbore stability challenges in complex lithology formations (e.g., Sarvak Formation) where conventional pore pressure estimation methods are insufficient, leading to frequent stress-induced breakouts and drilling instability.
Key Innovation: An integrated approach combining empirical data analysis, geomechanical modeling, real-time drilling data, and Support Vector Machine (SVM) algorithms to enhance predictive capability for pore pressure and fracture gradient, achieving 95% accuracy and providing optimal mud weight recommendations for wellbore stability.
36. A review of microfluidic technology for CO<sub>2</sub> sequestration in saline aquifers
Core Problem: Understanding the complex CO2 trapping mechanisms, salt precipitation, and CO2–water–rock chemical interactions during CO2 sequestration in saline aquifers is crucial for ensuring safety and long-term storage, but real-time visualization is challenging.
Key Innovation: A review examining the application of microfluidic technology for real-time visualization of CO2 flow, mass transfer, and reaction processes in porous media for CO2 sequestration in saline aquifers, highlighting key findings related to trapping mechanisms, salt precipitation, and chemical interactions.
37. Numerical analysis of hydrogen fingering in underground hydrogen storage
Core Problem: The unique properties of hydrogen (diffusivity, density, viscosity) introduce distinctive hydrodynamic phenomena like fingering in underground hydrogen storage, which can dramatically decrease hydrogen saturation and recovery rates, and are not fully understood at the microscopic scale.
Key Innovation: Numerical simulations at the microscopic scale using micromodels with varying porosity and pore throat sizes to understand the evolution of hydrogen fingering patterns in underground hydrogen storage, identifying viscous, capillary, and crossover fingering, and proposing geometric descriptors to characterize finger shapes.
38. Melting dynamics of freely floating ice in calm waters
Core Problem: Lack of mechanistic models for the melting dynamics of small, freely floating ice bodies, limiting accurate predictions in Earth's climate models.
Key Innovation: Development and validation of a theoretical model for ice melt rate, combining heat transfer and experimental results, identifying ice geometry and convective regime as key controls, and highlighting the ecological relevance of ice-driven convective volume flux.
39. Discussion of “Mechanical Properties of Fine-Grained Soils Treated with Fungal Mycelium of Trichoderma virens”
Core Problem: This is a discussion paper, likely related to the findings or methodology of a previous study on the mechanical properties of fine-grained soils treated with fungal mycelium.
Key Innovation: As a discussion, it likely provides critical analysis, alternative interpretations, or additional insights into the original research on soil treatment with fungal mycelium.
40. Localized anomaly detection and recovery of marine engine data to support digital twin development in shipping
Core Problem: Anomalies and missing values in marine engine data degrade the performance and credibility of Digital Twin (DT) models in shipping.
Key Innovation: Introduction of a data quality improvement framework for localized anomaly detection, isolation, and recovery of missing values in marine engine data, demonstrated with a real-world dataset to support digital twin development.
41. Validation and spatiotemporal analysis of multi-source potential evapotranspiration in Northwest Sichuan alpine grasslands
Core Problem: Accurate estimation of potential evapotranspiration (PET) is crucial for water resource evaluation and ecosystem modeling in alpine grasslands, but strong climatic heterogeneity and limited applicability of single-source PET data pose challenges in regions like Northwest Sichuan.
Key Innovation: Integrates multi-source PET data (MODIS, ERA5-Land, Hargreaves) using the Bayesian Triangle Cap (BTCH) method to construct a high-accuracy fusion product, which outperforms single-source datasets in stability and regional representativeness, and provides spatiotemporal analysis of PET patterns in alpine grassland ecosystems.
42. Coordinative optimization strategy for group track maintenance planning and train scheduling of railways
Core Problem: Current railway track maintenance planning overlooks structural dependencies among components and lacks coordination with train timetables, leading to inefficiencies and potential disruptions.
Key Innovation: A coordinative optimization strategy for group track preventive maintenance planning and train scheduling, considering structural dependencies, formulated as a mixed-integer linear programming model.
43. From text to network: A framework for identifying causal factors and risk propagation paths in maritime accidents
Core Problem: Systematically investigating complex causal mechanisms of maritime accidents from unstructured investigation reports to understand systemic risk and identify propagation paths.
Key Innovation: An automated analytical framework integrating NLP with complex network theory to transform unstructured accident reports into a quantifiable causal network, identifying causal factors, risk propagation paths, and key risk sources like 'Adverse Weather/Sea State'.
44. Translithospheric fault targeting for giant magmatic (-hydrothermal) ore deposit discoveries: recent advances and leading practices
Core Problem: Improving the discovery rates of giant magmatic ore deposits requires more accurate prediction of structural targets and mapping of translithospheric fault zones (LFZs).
Key Innovation: A refined, systems-based approach to mapping LFZs, based on a review of >120 global case studies, which identifies key structural predictors (e.g., proximity to secondary transverse LFZ intersections) and explains the dynamic fault-valve behavior of LFZs in channeling energy, fluid, and metal fluxes for ore formation.
45. An advanced decoupled polarimetric calibration method for the LuTan-1 hybrid- and quadrature-polarimetric modes
Core Problem: Low-frequency spaceborne PolSAR systems suffer from ionosphere-induced distortions coupled with system-induced polarimetric distortions, making high-precision decoupled polarimetric calibration essential for high-fidelity data, but existing methods lack generality and unbiased estimation under varying ionospheric conditions.
Key Innovation: A General Polarimetric Calibration Method (GPCM) for unbiased polarimetric distortion estimation across multiple modes and calibrator combinations, and an Enhanced Multi-Look Autofocus (EMLA) method for precise Slant Total Electron Content (STEC) inversion, enabling accurate Faraday rotation angle estimation and system distortion decoupling, confirmed effective for LuTan-1 data.
46. TUM2TWIN: Introducing the large-scale multimodal urban digital twin benchmark dataset
Core Problem: Creating Urban Digital Twins (UDTs) faces challenges in acquiring accurate 3D source data, reconstructing high-fidelity models, maintaining updates, and ensuring interoperability, with current datasets often limited to single parts of the processing chain.
Key Innovation: Introduction of TUM2TWIN, the first comprehensive multimodal Urban Digital Twin benchmark dataset, comprising georeferenced, semantically aligned 3D models and networks with diverse terrestrial, mobile, aerial, and satellite observations, supporting robust sensor analysis, advanced reconstruction methods, and downstream tasks like solar potential analysis and semantic segmentation.
47. <strong>City-Facade</strong>: A city-level large-scale point cloud building facade dataset for semantic & instance segmentation
Core Problem: Existing building facade datasets are predominantly image-based, lacking spatial information and robustness to lighting, while publicly available large-scale labeled 3D building point cloud datasets for urban scene understanding remain scarce and have limited coverage.
Key Innovation: Introduction of City-Facade, a city-level large-scale point cloud building facade dataset with approximately 200 million labeled 3D point clouds (over 60 km roads) for semantic and instance segmentation, facilitating the development and evaluation of algorithms for high-quality 3D urban scene understanding.
48. Roof-aware indoor BIM reconstruction from LiDAR via graph-attention for residential buildings
Core Problem: Reconstructing Building Information Models (BIMs) from terrestrial LiDAR scans remains challenging due to clutter, occlusions, and the geometric complexity of roof structures, particularly for residential buildings.
Key Innovation: A roof-aware scan-to-BIM pipeline tailored for residential buildings that processes indoor LiDAR data through four geometric abstractions using task-specific graphs, integrating LGNet for semantic segmentation, QTNet for floor plan reconstruction, and PPO for roof–floor fusion, producing watertight, Revit-compatible BIMs with high geometric accuracy on scenes with slanted roofs.
49. CityVLM: Towards sustainable urban development via multi-view coordinated vision–language model
Core Problem: Existing Vision-Language Models (VLMs) struggle with the complex needs of geoscience for comprehensive urban analysis across geographical, social, and economic dimensions, limiting their utility for sustainable urban development challenges.
Key Innovation: CityVLM, a multi-view coordinated vision–language model that integrates remote sensing and street-view imagery, along with CitySet (a multi-view vision–language dataset), to enable geospatial object reasoning, social object analysis, urban economic assessment, and sustainable development report generation, achieving superior performance in automated urban analysis and supporting sustainability efforts.
50. Structural barriers to complete homogenization and wormholing in dissolving porous and fractured rocks
Core Problem: Dissolution in porous and fractured media can lead to various patterns (uniform, channeling, wormholing), but the role of structural heterogeneity in limiting flow homogenization and its implications for upscaling dissolution kinetics are not fully understood.
Key Innovation: Quantified differences in dissolution patterns across regular pore, disordered pore, and discrete fracture networks using a unified flow focusing profile metric, demonstrating that structural heterogeneity sets a fundamental limit on flow homogenization.
51. Relating solution tests to pore-scale CaCO<sub>3</sub> crystal growth: Numerical simulation based on the phase field method
Core Problem: A lack of direct quantitative links between solution tests, pore-scale CaCO3 precipitation processes, and macro-scale performance in Microbially Induced Carbonate Precipitation (MICP).
Key Innovation: A coupled numerical framework integrating a saturation-dependent kinetic model for MICP chemical reactions with a phase-field model for pore-scale CaCO3 crystal growth, enabling direct translation of solution test parameters into pore-scale simulations.
52. Numerical insight into twin tunnelling-induced soil-structure interaction in battered pile-supported systems under lateral loading
Core Problem: The complex soil-structure interaction and ground deformation induced by twin parallel tunnelling, particularly its influence on the performance of existing battered pile-supported foundations in soft clay, are not fully understood.
Key Innovation: A series of 3D coupled consolidation finite element analyses, incorporating an advanced hypoplastic clay model, to numerically investigate twin tunnelling-induced soil-structure interaction, revealing how tunnel depth and sequence affect lateral displacements, differential settlements, and axial load distribution in battered pile foundations.
53. Enhancing fracture uniformity in hydraulic fracturing: A 3D simulation study on natural fracture networks and plugging optimization
Core Problem: Non-uniform propagation of multi-cluster fractures during hydraulic fracturing in deep shale gas reservoirs, driven by dynamic interaction between natural fracture networks and in-situ stress, reduces fracturing efficiency.
Key Innovation: A 3D discrete lattice numerical simulation study to systematically analyze dynamic regulatory mechanisms of multi-cluster fracture propagation under combined effects of natural fractures, differential in-situ stress, and temporary plugging, optimizing plugging parameters for balanced fracture extension.
54. Mechanisms of Efficient Three-Dimensional Fracture Network Construction in Deep Shale Reservoirs via Methane Multistage Explosive Fracturing
Core Problem: Efficiently constructing complex three-dimensional fracture networks in deep, low-permeability shale reservoirs to enhance hydrocarbon recovery, given the challenges of in situ stress and conventional fracturing techniques.
Key Innovation: High-fidelity 3D simulations integrating a characteristic methane–oxygen explosion load model with dynamic relaxation and full-restart methods to elucidate coupled interactions between explosive loading and in situ stress, proposing a novel dimensionless evaluation index (Fmef) for optimizing methane in situ multistage explosive fracturing (MISMEF).
55. Long-term irrigation water use datasets from multiple Earth Observation-based methods in major irrigated regions
Core Problem: The poor characterization of irrigation water use (IWU) at spatial and temporal scales needed for climate research, despite it being the largest human intervention in the terrestrial water cycle.
Key Innovation: Creation of a long-term archive of monthly IWU estimates (up to two decades, 0.25° resolution) for major irrigated regions, derived from multiple Earth Observation-based methods (SM-based Delta, SM-based Inversion, Model-observation integration), providing spatially explicit data for hydrological and climate studies.
56. Unraveling methane adsorption mechanisms in tectonically deformed coal: coupling roles of fissure width, deformation degree, and molecular structure
Core Problem: Clarifying the coupling mechanism of fissure width, tectonic deformation degree, and methane adsorption in tectonically deformed coal (TDC), which is crucial for efficient coalbed methane (CBM) development.
Key Innovation: Investigation using FTIR, XPS, 13C-NMR, and MD simulations to reveal how fissure width and pressure synergistically regulate adsorption energy, how deformation degree controls adsorption capacity via pore structure evolution, and proposing a unified 'multi-scale pore-fracture synergy adsorption mechanism'.
57. Effect of granular bentonite size and needle-punched fibers on virus transport through compacted and geosynthetic clay liners permeated with KCl
Core Problem: Limited understanding of pathogenic waste behavior in MSW landfill environments, particularly the impact of granular bentonite size and needle-punching on osmotic efficiency and virus permeation in geosynthetic clay liners.
Key Innovation: Evaluated sorption, hydraulic, and diffusion parameters of H1N1 in various bentonites and GCLs, showing that finer bentonite and powdered bentonite achieve optimal diffusion/retardation, while coarser bentonite and NP fibers increase permeation. Provides mechanistic insights for designing barrier systems for pandemic-related wastes.
58. Forecasting ballast performance under fast heavy-haul trains using an analytical-machine learning track (AMLT) model
Core Problem: Optimizing track maintenance for ballasted railway tracks under fast heavy-haul trains requires understanding the dynamic amplification of ballast permanent response with speed.
Key Innovation: Developed an Analytical-Machine Learning Track (AMLT) model combining a physics-based analytical model (elasto-dynamic response) with data-driven models (GA-ANN) to forecast permanent vertical strains and ballast breakage. Proposed a new performance-based limiting speed for heavy-haul trains.
59. SynPER: Synthesized prioritized experience replay for USVs formation control in island-reef waters via multi-agent reinforcement learning
Core Problem: Ensuring safe and reliable formation control for multi-unmanned surface vehicle (multi-USV) systems in complex marine environments, specifically addressing the challenge of low sample efficiency in multi-agent reinforcement learning.
Key Innovation: The SynPER (Synthesized Prioritized Experience Replay) algorithm, an advanced off-policy deep reinforcement learning method, which incorporates a hybrid KAN-based neural network, asynchronous multi-agent hindsight experience replay (AMAHER), and individual prioritized experience replay (IPER) to significantly improve convergence speed and performance for multi-USV formation control.
60. Practical considerations in directional decomposition of response spectra from time-domain measurements using relative RAO method
Core Problem: Practical application of the relative RAO-based directional decomposition method for reconstructing response spectra from time-domain motion measurements is challenged by phase distortion and statistical uncertainty from real-world data.
Key Innovation: Identification of key factors affecting the stability and accuracy of directional decomposition from time-domain measurements, and proposal of practical guidelines (phase-consistent preprocessing, normalization, adequate signal duration) for reliable implementation in digital-twin-based monitoring of marine structures.
61. Unified probabilistic model for axial strength of concrete-filled aluminum alloy tube columns with circular and square cross-sections
Core Problem: Traditional deterministic models for axial strength of concrete-filled aluminum alloy tube (CFAT) columns fail to account for inherent epistemic and aleatory uncertainties, challenging safety and durability assessment of offshore jacket platforms.
Key Innovation: A novel unified probabilistic model for predicting the axial strength of CFAT columns (circular and square cross-sections), developed by integrating a new unified deterministic model with a Markov Chain Monte Carlo Bayesian framework, which accounts for uncertainties and allows for calibration of existing deterministic models.
62. Load mitigation in floating wind turbines via active tuned mass damper using a physics-informed neural network based controller
Core Problem: Mitigating fatigue from coupled aerodynamic and hydrodynamic loads in floating offshore wind turbines (FOWTs) using active tuned mass dampers (ATMDs) is challenging due to nonlinear, high-dimensional dynamics.
Key Innovation: A fully data-driven control framework for ATMDs in FOWTs, utilizing a neural network-based surrogate model to capture dynamics and a physics-informed neural network (PINN) to solve the Hamilton-Jacobi-Bellman equation, resulting in substantial pitch fluctuation reduction and lower power consumption.
63. Analysis of collapse response of offshore pipes under external pressure and bending considering sizing effect
Core Problem: Reliable assessment of offshore steel pipelines' failure response to combined external pressure and bending is needed, considering the stress-strain history induced by the manufacturing process (ERW and sizing).
Key Innovation: Numerical simulation of ERW pipe manufacturing and collapse analysis, demonstrating improved collapse performance with increasing sizing ratios, and establishing a superior stacked ensemble learning (SEL) model for predicting failure pressure under combined loading.
64. Sparse Optoacoustic Sensing With Convolutional Dictionary Learning
Core Problem: Sparse optoacoustic sensing (SOS) requires advanced algorithms to compensate for under-sampled data to achieve high image reconstruction accuracy.
Key Innovation: Introducing a novel multi-layer convolutional dictionary-learning algorithm for SOS that eliminates pursuit algorithms and dictionary-wise parameters, enforcing slice-wise communication to achieve superior recovery accuracy and higher-fidelity reconstructions from sparse data.
65. Estimation of Ships’ Complex High-Resolution Range Profiles Based on Sparse Optimization Method in Non-Gaussian Sea Clutter
Core Problem: Recovering ships' high-resolution range profiles (HRRPs) from radar returns for classification and identification is challenging due to the complex sparse signal being interfered by non-Gaussian sea clutter.
Key Innovation: Proposes three sparse optimization methods (matching K-distribution, generalized Pareto distribution, and CGIG distribution) that specifically match the non-Gaussian characteristics of sea clutter to estimate complex HRRPs of ships, using the Anderson–Darling test for parameter searching and Kolmogorov–Smirnov distance for model selection, demonstrating superior performance over existing methods.
66. An Anti-2D Deceptive Jamming Method for Bistatic SAR Based on Jammer Localization
Core Problem: Synthetic aperture radar (SAR) is vulnerable to 2D deceptive jamming, which introduces highly realistic false targets, severely impairing scene interpretation and remote sensing reliability, and existing anti-jamming methods require excessive channels or are limited to specific jamming types.
Key Innovation: Proposes a bistatic SAR-based method for anti-2D deceptive jamming that requires only two channels and a single-pass acquisition, effectively suppressing all types of deceptive jamming by extracting jamming discrepancy, establishing a unified mathematical relationship for accurate jammer 2D localization, and applying 2D minimum variance distortionless response digital beamforming.
67. Tectonic control on incised-valley geometry: Late Pleistocene examples along the longest active fault in Japan
Core Problem: The qualitative understanding of tectonic influences on incised-valley geometry, despite their formation being influenced by sea-level changes, fluvial discharge, and tectonic activity.
Key Innovation: Quantitative investigation of incised valleys along the Median Tectonic Line active fault system, revealing that valley depth and cross-valley gradient correlate with tectonic subsidence, while valley width correlates with fluvial discharge, providing quantitative insights into tectonic control on valley geometry.
68. Tectonic-climatic coupling in extensional landscapes: Quantifying divide migration and erosion-sedimentation feedbacks through bilateral sandbox experiments
Core Problem: The poorly quantified coupled feedbacks between tectonic subsidence, precipitation, and landscape evolution (divide migration, erosion-sedimentation) in extensional landscapes, leading to fundamental asymmetries.
Key Innovation: Use of bilateral sandbox experiments with varying subsidence rates and rainfall intensities, coupled with high-resolution morphometric analysis, to quantitatively demonstrate how drainage patterns diverge, divide migration is directed, and erosion scales more strongly with tectonic subsidence than precipitation, providing a predictive framework for landscape responses.
69. A synergistic approach: multi-purpose K-nearest neighbor and active learning Kriging for efficient failure probability function estimation
Core Problem: The high computational costs and limited efficiency of conventional and single-loop active learning Kriging (AK) methods for estimating failure probability functions (FPF) in reliability-based design, due to suboptimal sampling and inaccurate kernel density estimation (KDE).
Key Innovation: Introduction of the SL-AK-KNN method, a novel multi-purpose K-nearest neighbor (KNN) framework integrated with enhanced active learning Kriging (AK), which acts as a spatial-information-guided learning function and an adaptive nonparametric density estimator, significantly reducing computational costs and enhancing FPF estimation accuracy.
70. Advancing fatigue crack growth prognosis in metallic structures: A physics-informed sequential attention approach with uncertainty quantification
Core Problem: The challenge of accurate, physically consistent, and adaptable prediction of fatigue crack growth (FCG) in metallic structures under variable-amplitude loading, as conventional models lack adaptability and purely data-driven approaches lack physical coherence and generalization.
Key Innovation: Proposal of a Physics-Informed Sequential Attention Network (PI-STAN), a unified framework integrating LSTM-based temporal encoding with Transformer self-attention under a PINN formulation, which explicitly captures dynamic parameter evolution, enforces physical consistency, and quantifies predictive uncertainty for FCG prognosis.
71. A deep learning framework for aviation risk classification and high-order coupled risk modeling
Core Problem: Challenges in automated aviation risk analysis from narrative incident reports due to long-form format, class imbalance, and domain-specific semantics.
Key Innovation: A domain-adapted deep learning model (RoBERTa-based) for multi-label classification of contributing factors, integrating LLM data augmentation, text/metadata merging, composite loss, and domain adaptive pretraining, combined with N-K and Bayesian network models for high-order coupled risk propagation.
72. Component cascade utilization optimization based on an integrated framework with hybrid crack growth prediction
Core Problem: Managing failure risks and maintenance costs in complex engineering systems due to component degradation, requiring accurate remaining useful life (RUL) prediction and risk assessment for cascade utilization decisions.
Key Innovation: An integrated closed-loop framework for cascade utilization optimization, embedding Paris law into a Dynamic Gamma-Gamma process for hybrid crack growth prediction, using particle filtering for adaptive prognosis, and a two-stage decision-making model for optimal switching time and degradation thresholds.
73. Adaptive Wiener process modeling integrating physical causality and degradation states for high-reliability systems
Core Problem: Challenges in degradation modeling for high-reliability equipment due to coupled components and concealed early degradation data, making traditional data-driven methods difficult to apply.
Key Innovation: A novel degradation modeling framework based on physics-informed neural networks, integrating Bayesian causal attention for physical mechanisms and temporal convolutional networks for degradation states, fused into a nonlinear Wiener process with physics residual regularization and a two-stage online update strategy.
74. A multi-model fusion redundancy allocation method for subsea control systems with consideration of Bayesian stress-conditional importance
Core Problem: Achieving high reliability in subsea control systems under extreme conditions and multi-source uncertainties, where redundancy design can lead to a 'redundancy paradox' if not optimized.
Key Innovation: A multi-model fusion-based redundancy allocation method for subsea control systems, introducing a Bayesian stress-conditional importance metric and employing a multi-objective particle swarm optimization algorithm to maximize reliability while minimizing costs.
75. Dimension-mismatched adversarial network: a new feature distribution adaptation method for rolling bearing RUL prediction
Core Problem: Existing remaining useful life (RUL) transfer prediction methods assume equal sample dimensions between source and target domains, leading to distorted cross-domain data distribution measurements when operating conditions or fault types differ.
Key Innovation: A new feature distribution adaptation method called Dimension-Mismatched Adversarial Network (DMAN), which establishes a dimension selection rule, designs an adaptive empirical mutual information calculator, and uses an adversarial training mechanism to learn domain-invariant degradation features for RUL prediction.
76. An integrated framework for functional model-based safety assessment of process systems using Cloud-Bayesian network
Core Problem: Challenges in safety assessment of complex industrial systems due to difficulties in considering component interactions and dynamic behaviors, and limitations of traditional labor-intensive methods.
Key Innovation: An integrated framework for functional model-based safety assessment of process systems, enhancing Multilevel Flow Modelling (MFM)-based hazard analysis with probabilistic risk reasoning using a Cloud-Bayesian Network, demonstrated on an offshore oil & gas platform.
77. Uncertainty aware federated averaging approach for privacy secured collaborative remaining useful life prediction of rolling element bearing
Core Problem: Challenges in centralized RUL prediction for rolling element bearings in industrial scenarios due to difficulty in obtaining sufficient life-cycle data, data isolation concerns, and lack of uncertainty integration for reliable estimation.
Key Innovation: An uncertainty-aware federated averaging (UAFA) approach within a federated learning framework for RUL prediction, using Monte-Carlo Dropout based LSTM networks, a dynamic modulation factor for uncertainty-aware learning rate, and UAFA for model aggregation.
78. Approximate approaches for reliability evaluation and redundancy allocation for large-scale multi-state series-parallel systems
Core Problem: Computational intractability of reliability assessment and redundancy allocation problems for large-scale multi-state series-parallel systems (MSSPS-LN) due to high computational complexity.
Key Innovation: Two novel algorithms: an efficient reliability assessment algorithm integrating state-space reduction with dynamic programming, and an ϵ-approximate algorithm based on multi-objective dynamic programming to solve the redundancy allocation problem for MSSPS-LN.
79. Predictive risk analysis for leakage accidents with dynamic behaviour
Core Problem: Enhancing data foundation for risk analysis of leakage accidents with dynamic behavior, especially in situations of data and knowledge scarcities, by extending inductive risk analysis to predictive patterns.
Key Innovation: A predictive risk analysis approach for leakage accidents with dynamic behavior, based on time-series simulations, integrating FMEA, Bayesian and Event Tree Analysis, to calculate a risk index considering dynamic consequences, demonstrated on offshore hydrogen storage systems.
80. A real-time reliability assessment framework for marine mechanical equipment integrating machine learning and physical knowledge: Toward applications in maritime autonomous surface ships
Core Problem: The increasing criticality of marine mechanical equipment reliability for safe and efficient operation of Maritime Autonomous Surface Ships (MASS), requiring a real-time reliability assessment framework.
Key Innovation: A real-time reliability assessment framework for marine mechanical equipment, integrating data-driven models (WGAN, PCA-LSTM) with physical knowledge to predict health indicators and quantify reliability based on Weibull distribution, enabling early fault detection and maintenance planning.
81. An interpretable multivariate remaining useful life prediction method of mechanical equipment based on adaptive threshold aggregation causal discovery
Core Problem: Challenges in interpretability and robustness of existing deep learning-based RUL prediction methods for mechanical equipment when dealing with complex multivariate time series data.
Key Innovation: A multivariate RUL prediction method based on adaptive threshold aggregation causal discovery, employing a Bayesian Information Criterion-based causal discovery method, an adaptive threshold mechanism for graph aggregation, causal effect estimation, and a Temporal Graph Convolutional Network for RUL prediction.
82. Digital twin evolution mechanism for individual aircraft life prognosis enhanced by on-line hybrid monitoring
Core Problem: Limited research on the evolution mechanism for digital twins (DTs) that can co-evolve multiple parameters using on-line structural health monitoring (SHM) methods for high-fidelity structural life prognosis.
Key Innovation: A novel life prognosis Digital Twin model evolution mechanism enhanced by on-line damage and load hybrid monitoring, enabling the co-evolution of multiple DT parameters (e.g., crack features, evolution rate, stress intensity factor) for improved life prognosis accuracy in structures like aircraft.
83. Enhancing route planning framework for indoor emergencies involving toxic gas release: Model and application
Core Problem: Existing route planning frameworks for indoor emergencies involving toxic gas release are strained by building complexity and dynamic incidents, leading to limitations in emergency response strategies.
Key Innovation: An enhanced route planning framework that optimizes indoor network graph structure (using m-CDT), cost model (Coupled Cost Function with Alternative Risk Set), and planning algorithm (multi-objective MERP with Bidirectional Optimization Mechanism) to reduce cumulative risk and eliminate congestion in emergency evacuations.
84. Innovations in underground hydrogen storage with multiphysics simulations, optimization, and monitoring: A review
Core Problem: Ensuring the safety and efficiency of Underground Hydrogen Storage (UHS) necessitates a comprehensive understanding of complex multiphysical interactions and geomechanical responses, including fault stability, under cyclic loading.
Key Innovation: This review synthesizes current knowledge, identifies critical knowledge gaps, and highlights future research directions for UHS, particularly in advancing geomechanical understanding under multiphysics-coupling, ML-driven theories, enhanced modeling, and robust optimization strategies.
85. Microvibration detection and compensation for SDGSAT-1 based on line-by-line bundle adjustment
Core Problem: Microvibrations degrade the geometric quality of optical Earth observation satellite imagery by introducing intra-scene spatial distortions, necessitating robust detection and compensation strategies.
Key Innovation: A novel microvibration detection and compensation framework for SDGSAT-1 based on line-by-line bundle adjustment, which minimizes directionally weighted residuals to estimate microvibrations at high temporal resolution and effectively suppresses geometric distortions, outperforming conventional methods.
86. Set-CVGL: A new perspective on cross-view geo-localization with unordered ground-view image sets
Core Problem: Existing cross-view geo-localization (CVGL) approaches are limited by using single images or fixed-view sequences, restricting perspective diversity and thus reliability, unlike human visual localization which integrates multiple perspectives.
Key Innovation: Introduction of the novel Cross-View Image Set Geo-Localization (Set-CVGL) task and the SetVL-480K benchmark dataset, along with FlexGeo, a flexible method that adaptively fuses image features and leverages geo-attributes for comprehensive scene perception, significantly outperforming existing methods in localization accuracy.
87. AnchorReF: A novel anchor-based visual re-localization framework aided by multi-sensor data fusion
Core Problem: Visual relocalization faces challenges in achieving robust pose estimations when query images exhibit significant changes compared to the reference scene, leading to inaccurate pose estimations that require verification and correction.
Key Innovation: AnchorReF, a novel anchor-based visual relocalization framework that achieves robust pose estimations through multi-view co-visibility verification and further refines poses using tightly-coupled multi-sensor data fusion, demonstrating state-of-the-art performance on challenging real-world urban driving datasets by significantly reducing translation and rotation errors.
88. BEDI: a comprehensive benchmark for evaluating embodied agents on UAVs
Core Problem: Evaluation methods for UAV-Embodied Agents (UAV-EAs) are constrained by a lack of standardized benchmarks, diverse testing scenarios, and open system interfaces, hindering the advancement of autonomous UAV tasks.
Key Innovation: BEDI (Benchmark for Embodied Drone Intelligence), a systematic and standardized benchmark for UAV-EAs, featuring a novel Dynamic Chain-of-Embodied-Task paradigm, a unified evaluation framework for six core sub-skills, and a hybrid virtual/real-world testing platform, facilitating objective model comparison and advancing embodied intelligence research.
89. Crowd detection using Very-Fine-Resolution satellite imagery
Core Problem: Accurate crowd detection (CD) is critical for public safety, but existing methods using ground/aerial imagery have limited spatio-temporal coverage, and very-fine-resolution (VFR) satellite imagery has not been explored for this task.
Key Innovation: CrowdSat-Net, a novel point-based convolutional neural network featuring a Dual-Context Progressive Attention Network (DCPAN) and a High-Frequency Guided Deformable Upsampler (HFGDU) for improved feature representation and high-frequency information recovery; and CrowdSat, the first VFR satellite imagery dataset for CD, demonstrating superior performance over state-of-the-art methods.
90. Intelligent and autonomous pipeline deposit tracking based on a multi-object tracking framework
Core Problem: Conventional pipeline maintenance relies on labor-intensive and time-consuming human inspection of CCTV records for deposit detection and tracking.
Key Innovation: Proposed an autonomous framework based on a multi-object tracking (MOT) algorithm (YOLOX + BYTE) for efficient and accurate deposit detection and tracking in pipelines, achieving high accuracy and significantly reducing manual intervention.
91. Climate modulation and intensity-stratified moisture sources in major global river basins
Core Problem: Understanding how atmospheric moisture pathways regulate precipitation variability is essential for basin-scale water security under climate change, but a global, intensity-stratified analysis of moisture sources is lacking.
Key Innovation: Presented a global, intensity-stratified analysis of moisture sources and transport across seven major river basins using ERA5 reanalysis and the WAM-2layers model, identifying dominant and higher-order moisture transport modes, classifying basins into ocean-dominated, mixed-source, and land-dominated regimes, and improving understanding of basin-specific hydroclimatic responses for hydrological predictions.
92. Analysis of runoff components and their responses to climate change in the Manas River Basin, Xinjiang, China
Core Problem: The response mechanisms of runoff components to climate warming and humidification in the Tianshan Manas River Basin remain unclear, and future runoff dynamics need to be projected to assess water scarcity risks.
Key Innovation: Employed the distributed hydrological model SPHY, integrated with remote sensing data and a multi-level calibration strategy, to quantitatively assess the composition and spatiotemporal variation of runoff components (rainfall, snowmelt, glacier meltwater, baseflow) and project future runoff dynamics under CMIP6 climate scenarios, revealing heightened risk of downstream water scarcity under high emission scenarios.
93. Towards a better understanding of river network dynamics in a glacierized catchment
Core Problem: River network dynamics (expansion and contraction) are poorly understood in high-mountain glacierized systems due to limited observations and the complexity of cryosphere-hydrology interactions, hindering understanding of ecosystem functioning.
Key Innovation: Applied the SWAT-GL model to simulate river network dynamics in the Valsorey catchment (Swiss Alps) over 19 years, revealing strong seasonality (up to 50% contraction in winter, full connectivity in summer) and significant interannual variability, with flow intermittency most prevalent in low-order headwater streams, underscoring the sensitivity of alpine river networks to climatic variability.
94. TKLE-BPINN: A Bayesian physics-informed inversion framework for high-dimensional parameter identification in geotechnical subsurface systems
Core Problem: Characterizing spatially variable hydraulic properties in geotechnical subsurface systems is a high-dimensional and ill-posed inverse problem, especially in unsaturated flow conditions.
Key Innovation: TKLE-BPINN, a novel Bayesian framework integrating Bayesian physics-informed neural networks (B-PINNs) with truncated Karhunen–Loève expansion (KLE), for accurate parameter estimation and robust uncertainty quantification in subsurface flow and transport, outperforming standard B-PINNs.
95. Investigation of pipe–soil interface effects on the stress–deformation behavior of buried high-density polyethylene pipes: full-scale test and analytical solution
Core Problem: Existing analytical models for buried HDPE pipes often assume fully bonded or perfectly smooth pipe-soil interfaces, which inaccurately represent the actual stress-deformation behavior and interface slip under load.
Key Innovation: A full-scale experimental investigation using a self-developed three-direction loading platform and interface slip monitoring apparatus, combined with a novel analytical solution incorporating a pipe-soil slip interface, to accurately describe the stress-deformation response of buried HDPE pipes.
96. Image-traced flow behavior and frequency-dependent response of ballasted trackbed under dynamic loads
Core Problem: The concealed nature of ballast particle movement and its frequency-dependent response under dynamic train loads makes it challenging to observe and understand the mechanisms leading to trackbed settlement.
Key Innovation: A full-scale model test incorporating image-assisted measurement techniques (dyed tracer particles, high-speed cameras) to trace ballast particle flow behavior and analyze the frequency-dependent response of ballasted trackbed under varying train speeds, axle loads, and long-term loading conditions.
97. Nonlinear unsaturated shear strength behaviour of compacted crushed rock class IV material: implications for corrugation in unsealed roads
Core Problem: Corrugation in unsealed roads, exacerbated by moisture loss, is linked to progressive shear failure in the surface layer, but the nonlinear unsaturated shear strength behavior of compacted crushed rock class IV material (a common road material) has not been properly characterized.
Key Innovation: A systematic experimental investigation into the nonlinear unsaturated shear strength behavior of compacted crushed rock class IV material under varying saturation levels and fines content, proposing nonlinear failure envelopes and an empirical Gaussian-based cohesion model to capture saturation-dependent behavior, with implications for improving unsealed road durability.
98. Consolidation of unsaturated composite foundation with permeable short piles and impermeable long piles
Core Problem: The consolidation characteristics of unsaturated composite foundations, particularly those integrating permeable short piles and impermeable long piles, remain insufficiently understood, hindering optimal design for reinforcing layered unsaturated ground.
Key Innovation: A consolidation model and semi-analytical solutions for unsaturated composite foundations with centrally located permeable short piles surrounded by impermeable long piles, validated against existing methods, to analyze and optimize consolidation performance by varying pile parameters.
99. Cross-model feature-importance analysis of soil properties for predicting optimum moisture content and maximum dry unit weight of fine-grained soils
Core Problem: Accurately predicting optimum moisture content and maximum dry unit weight of fine-grained soils, which are critical compaction parameters, based on routine soil index properties, and understanding the relative importance of these properties using interpretable machine learning models.
Key Innovation: A cross-model feature-importance analysis using GAM, RF, and XGBoost models on a curated database of fine-grained soils, consistently identifying LL and PL as the most influential predictors for wopt and γdmax, with GAM offering superior interpretability and comparable accuracy.
100. Study on the vertical bearing characteristics and influencing factors of threaded piles
Core Problem: The mechanism by which thread parameters (height, pitch, shape, thickness) influence the load-bearing performance and vertical bearing capacity of threaded piles is not fully understood, hindering optimal design for foundation reinforcement.
Key Innovation: A combined laboratory half-pile model test with Digital Image Correlation (DIC) and numerical simulation study to systematically examine the influence of thread parameters on threaded pile bearing capacity, identifying optimal thread height (8-10 mm), pitch ratio (1.0), and trapezoidal shape for superior performance.
101. Pile installation effects in natural soft clays: A semi-analytical solution using strain path method
Core Problem: Accurately modeling pile installation effects in natural soft clays, particularly the anisotropic evolution and destructuring nature of these soils, is challenging for predicting stress, pore pressure, and fabric changes around piles.
Key Innovation: A semi-analytical solution for pile penetration in natural soft clays using the strain path method (SPM) integrated with the S-CLAY1S model, capturing anisotropic evolution and destructuring, and providing insights into the distribution of stresses, excess pore water pressure, and fabric anisotropy.
102. Human Interference Quantification and Representation in Hydrological Models—Introduction to the Special Collection on Quantifying Human Interferences in Hydrology
Core Problem: The need for more reliable quantification and representation of human interferences in hydrological models to improve their realism and predictive capability.
Key Innovation: Highlighting advances in data availability and collection technologies that enable better quantification of human interferences, and the promise of data-driven, generalizable modeling approaches for realistic representation.
103. Hydrology in the Age of Artificial Intelligence: From Fragmentation to Coherent Terrestrial Hydrosphere Science
Core Problem: The fragmentation of hydrological science, leading to inconsistent and incomplete understanding of the terrestrial hydrosphere, despite advances in machine learning for specific tasks like streamflow prediction.
Key Innovation: A call for a more coherent terrestrial hydrosphere science, leveraging the success of machine learning in streamflow prediction to overcome the limitations of fragmented hydrological research.
104. Peat core research in Western Siberia: methods applied, regions studied, and future prospects
Core Problem: The fragmented and evolving nature of palaeoecological research based on peat cores in the Western Siberian Lowland, leading to spatial, temporal, and proxy-specific gaps in understanding peatland dynamics and past environmental conditions.
Key Innovation: Creation of the Western Siberian Peat Core Database (WSPC), the most extensive compilation of peat-core-based palaeoecological data for the region, synthesizing information from 654 cores and 156 publications, identifying research trends, spatial coverage gaps (especially in permafrost regions), and future research challenges.
105. GlobalRice20: A 20 m resolution global paddy rice dataset for 2015 and 2024 derived from multi-source remote sensing
Core Problem: Lack of a consistent global, medium-to-high resolution paddy rice map due to challenges in processing multi-source satellite archives (cloud contamination, temporal irregularity).
Key Innovation: Development of GlobalRice20, the first global 20m resolution paddy rice dataset for 2015 and 2024, using a "Time-Series-to-Vision" framework (T2VRCM) that integrates multi-source optical and SAR time-series.
106. A Cambrian soft-bodied biota after the first Phanerozoic mass extinction
Core Problem: Limited understanding of the evolutionary and ecological dynamics of the Cambrian explosion due to the rarity of high-diversity Burgess Shale-type biotas.
Key Innovation: Discovery of the Huayuan biota, a lower Cambrian (Stage 4) Burgess Shale-type Lagerstätte, which illuminates the impact of the Sinsk event (first Phanerozoic mass extinction) and offers insights into early Cambrian global ecosystem transformation.
107. Grain size, geochemical characteristics, and transport patterns of surface sediments in the Dongsha sea area, South China sea
Core Problem: Lack of comprehensive understanding of the characteristics, material sources, and transport patterns of surface sediments in the Dongsha area of the South China Sea.
Key Innovation: Comprehensive analysis of grain size, major elements, and REEs in surface sediments, identifying three main end-member components, predominant transport trends, and establishing Taiwan as the dominant source with a minor contribution from the Pearl River.