TerraMosaic Daily Digest: September 13, 2026
Daily Summary
Mountain-slope studies resolve both precursory motion and the mechanics that control destructive runout. Before the 26 August Nepal disaster, enhanced optical pixel-offset tracking recovered more than 100 m of glacier-landslide deformation and identified a bifurcation beginning in January 2026 that coincided with anomalous warming; the study further proposes a flow-aligned landslide-tsunami-flood cascade. A new depth-averaged model makes basal friction state-dependent through μ(I) rheology and reproduces the runout and deposit geometry of the 2018 Baige landslide-clastic flow. Transparent-soil experiments on the Taping H1 reservoir-bank landslide show opposing deformation sequences during water-level rise and drawdown, while three-dimensional discrete-element simulations couple frictional heating, conduction-driven ice melting and particle fluidization to explain the high mobility of ice-rock debris flows.
Several studies couple process-based or spatially explicit hazard mapping with screening and adaptation decisions. In Taiwan's Dahan watershed, 335 typhoon simulations project a rise in maximum landslide probability from 14.0% to 22.4%, and translate the resulting impact maps into a stakeholder adaptation exercise. InSAR atmospheric correction coupled with constrained clustering improves unstable-slope screening in coal-mining terrain, while a geographically weighted random forest models spatially varying associations between ice avalanches and topographic, climatic and glacier-related factors on the Qinghai-Tibet Plateau. Complementary studies target response-adaptive rainfall thresholds from long-term alert records, rainfall-triggered debris-flow frequency-magnitude relations and geological controls on deep-seated-landslide mobility. Seasonal wetting-drying-freezing-thawing experiments relate fissure growth to strength loss in expansive-soil slopes, and CPT-based deterministic and probabilistic models extend liquefaction-induced lateral-spread estimates from soundings to spatial exceedance maps.
Earthquake and hydroclimate studies broaden the evidential chain from forcing to consequence. Joint InSAR and pixel offsets constrain the Wushi earthquake rupture and its Coulomb stress transfer, while fully dynamic finite elements simulate repeated rupture on a curved normal fault. A new study of the 2025 Tingri earthquake frames fatality outcomes around seismic hazard, structural damage and human behavior. For drought and flood hazards, the selected studies address distinct operational constraints: location-dependent forecast regimes across Iran, observed snow-drought impacts in 38 Italian catchments, ten-minute ensemble updates for heavy rainfall, long-term drought recurrence under ENSO and IOD modes, and spatial optimization or real-time operation of urban drainage and cascade reservoirs.
Several methods target the reliability limits of data-driven Earth observation. Earth-Agent-Pro joins acquisition, preprocessing, domain computation and evidence-backed repair in an executable agent workflow, whereas a Bayesian atmospheric transformer scales explicit aleatoric and epistemic uncertainty to global 0.25-degree forecasts. Physics-informed conformal prediction makes neural-operator intervals spatially adaptive to PDE residuals; physical-state-guided diffusion separates wave-equation and denoiser gradients in full-waveform inversion; and burned-area transfer, multimodal registration and band-level Shapley audits test performance or representation consistency across sensors and domains. Collectively, these methods introduce explicit checks on physical consistency, spatial or predictive uncertainty, and robustness to sensor or workflow shifts.
Key Trends
Across the strongest papers, evolving slope and flow mechanics are paired with satellite precursors, probabilistic forecasts and intervention-oriented maps.
- Evolving material state is entering runout and slope-response analysis: State-dependent μ(I) friction, frictional-heating and ice-melt coupling, and water-level-controlled transparent-soil experiments separately resolve basal resistance, fluidization and deformation pathways through time.
- Precursors are being interpreted within cascade geometry: The Nepal study combines multi-year deformation, its temporal coincidence with anomalous warming and flow-direction alignment to propose a tsunami-flood cascade, while regional InSAR screening and ice-avalanche inventories broaden analysis beyond individual failures.
- Climate-risk analysis is becoming ensemble-based and spatially differentiated: Hundreds of typhoon simulations estimate future landslide probability, observations from 38 catchments trace snow-drought impacts, and station-specific models reveal geographically varying drought-forecast regimes.
- Uncertainty is being attached to physical or spatial structure: PDE residuals adapt conformal intervals, Bayesian atmospheric ensembles separate aleatoric and epistemic uncertainty, random fields map lateral-spread exceedance, and inversion-ensemble variability concentrates near geological interfaces.
- Reliability is being tested across sensors, representations and workflows: Earth-Agent-Pro repairs execution plans from runtime evidence, burned-area models are evaluated across sensors and regions, band-level audits probe spectral consistency, and multimodal registration is tested under radiometric shift.
Selected Papers
The September 13 collection is led by glacier-landslide precursor detection in Nepal, state-dependent runout modelling of the Baige landslide-clastic flow, transparent-soil experiments on reservoir-bank stabilization and three-dimensional simulation of ice-rock debris-flow fluidization. Direct applications span adaptive landslide-warning thresholds, debris-flow frequency-magnitude analysis, deep-seated-landslide mobility, climate-conditioned landslide probability, mining-slope InSAR, ice-avalanche zoning, expansive-soil degradation, liquefaction-induced lateral spread, post-fire erosion, earthquake rupture and fatality analysis, drought, flood and heavy-rainfall prediction. The wider methods set includes PDE-residual uncertainty intervals, executable Earth-observation agents, physics-guided geophysical inversion, cross-sensor burned-area transfer and explicit audits of remote-sensing representations.
1. Precursor of a powerful large glacier landslide causing tsunami floods in the 26 August 2026 Nepal disaster
Core Problem: Detecting pre-failure motion and resolving the water source and cascade mechanism of the 2026 Nepal glacier-landslide disaster.
Key Innovation: Enhanced optical pixel-offset tracking across hundreds of free satellite scenes reveals more than 100 m of deformation and a warming-synchronized bifurcation phase, supporting a directional landslide-to-tsunami-flood mechanism.
2. A depth-averaged numerical model incorporating variable friction for landslide-clastic flow dynamics
Core Problem: Fixed-friction depth-averaged models cannot represent the changing basal resistance of a landslide-clastic flow during transport, deceleration and deposition.
Key Innovation: Introduces state-dependent μ(I) friction into a depth-averaged Navier-Stokes model and validates runout and deposit geometry against experiments and the 2018 Baige event.
3. Transparent soil experiments on the deformation evolution of reservoir bank landslide stabilized by multi-row anti-slide pile groups
Core Problem: Dynamic interaction between large reservoir-bank landslides and multi-row pile groups under complex water-level forcing is poorly understood.
Key Innovation: A noninvasive transparent-soil model, informed by field investigation and long-term monitoring at Taping H1, resolves contrasting deformation propagation during reservoir-level rise and drawdown and tests pile-spacing effects.
4. Study on the solid–fluid transition mechanism of ice–rock debris flows using three-dimensional discrete element simulation
Core Problem: Quantitatively resolving how frictional heating, ice melting, and fluidization control ice-rock debris-flow dynamics.
Key Innovation: A PFC3D model couples heating, conduction-driven melting, and a second-order-work fluidization criterion, revealing ice-content thresholds and an encapsulation mechanism.
5. Assessing landslide probability changes under climate change scenarios as a basis for adaptation strategies
Core Problem: Communities need physically grounded estimates of how climate change will alter slope-hazard probability and impacts.
Key Innovation: TRIGRS simulations of 335 baseline and future typhoon events are coupled with settlement-scale impact mapping and agency interviews.
6. Ice Avalanche Hazard Assessment Using Multi-Source Geospatial Data and GWRF Model
Core Problem: Ice-avalanche research lacks regional inventories and models that capture spatially varying conditioning relationships.
Key Innovation: A multisource remote-sensing inventory and geographically weighted random forest integrate topographic, climatic, and glacier controls.
7. Improved Method for Unstable Slope Identification in Coal-Mining Mountainous Areas Combining InSAR and Clustering Techniques
Core Problem: Tropospheric contamination and weak geophysical constraints reduce the reliability of InSAR-based unstable-slope identification.
Key Innovation: Two-stage deformation-preserving atmospheric correction is coupled with spatiotemporally constrained density-peak clustering and an Unstable Slope Index using deformation and terrain-geology features.
8. Multi-scale investigation on shear behavior and degradation mechanisms of undisturbed highly expansive soils under seasonal climatic variations
Core Problem: How wetting-drying-freeze-thaw cycles degrade undisturbed expansive soil from microstructure to shear strength.
Key Innovation: Triaxial testing, microscopy, NMR, and CT jointly quantify fissure-network development and its linear relation to cohesion and friction loss.
9. CPT-based intelligent prediction of liquefaction-induced lateral spread: multi-scale analysis
Core Problem: Existing methods rely on SPT and laboratory soil properties and often lack uncertainty-aware spatial predictions.
Key Innovation: Interpretable ensemble CPT models provide deterministic equations, probabilistic exceedance curves, and random-field spatial maps validated with Christchurch earthquake data.
10. Coseismic Slip Distribution and Coulomb Stress Changes of the 2024 Mw 7.0 Wushi Earthquake, Xinjiang, China
Core Problem: Resolving the Wushi earthquake slip distribution and its implications for regional fault interaction and seismic hazard.
Key Innovation: Joint InSAR and pixel-offset inversion is combined with Coulomb stress, GNSS strain, and fault-locking evidence.
11. FEM simulations of seismic cycle along a complex normal fault
Core Problem: Robustly simulating full earthquake cycles on geometrically complex normal faults over long time spans.
Key Innovation: Adaptive Newmark FEM with rate-state friction and a healing-constrained slip-weakening law, tested over 1,100 simulated years.
12. Why Stratovolcanoes Are Mechanically Stronger than Shield Volcanoes
Core Problem: Why shield volcanoes experience more frequent dike-fed eruptions and large collapses than stratovolcanoes.
Key Innovation: A least-action mechanical argument links contrasting layered material properties to fracture arrest, fault confinement, and greater structural toughness.
13. Short-term drought forecasting in Iran using multi-source machine learning: an assessment of autoregressive, teleconnection-driven, and hybrid paradigms
Core Problem: Determining when local drought memory, global climate teleconnections, or their combination best predicts drought across Iran.
Key Innovation: Thirty years of rainfall and nine models reveal spatially varying optimal forecast paradigms rather than a single nationwide solution.
14. Impacts of Mediterranean snow droughts on mountain socio-ecohydrology
Core Problem: The coupled ecological and societal impacts of increasingly frequent snow droughts remain poorly quantified.
Key Innovation: Thirteen years of observations from 38 Italian catchments jointly resolve snow duration, melt, summer runoff, photosynthesis, and water-supply responses.
15. Impact of large-scale analysis constraints on C-band radar data assimilation during the simulation of two heavy rainfall events over southern peninsular India
Core Problem: High-resolution radar assimilation can introduce large- and convective-scale imbalances and unrealistic hydrometeor structures.
Key Innovation: Large-scale analysis constraints are integrated with C-band reflectivity and radial-wind assimilation and compared across two contrasting monsoon events.
16. Coal Fire Identification Based on Multisource Remote Sensing Time-Series Features and Hierarchical Feature Optimization: A Case Study in the Zhundong Coalfield
Core Problem: Coal-fire signals are obscured by heterogeneous backgrounds and redundant high-dimensional observations.
Key Innovation: Landsat-8, Sentinel-1 deformation, thermal, and DEM features are combined with spatial enhancement, SHAP-informed screening, metaheuristic optimization, and tuned XGBoost.
17. Hyperspectral Knowledge-Guided Cross-Sensor Burned-Area Mapping: Distilling AVIRIS Semantics to Sentinel-2 and Landsat-8
Core Problem: Multispectral imagery confuses fire effects with soil, shadows, agricultural disturbance, and other nonfire changes.
Key Innovation: Task-oriented semantic distillation transfers bitemporal AVIRIS supervision to compact Sentinel-2 and Landsat-8 students and is independently tested across regions.
18. Spatiotemporal Characteristics of Compound Hot-Dry Events and Lagged Impacts on Vegetation Productivity in Central Asia
Core Problem: Fine-scale compound hot-dry dynamics and delayed vegetation responses in Central Asia are poorly understood.
Key Innovation: A Vine-Copula daily compound index integrating satellite and reanalysis data with lag and driver attribution using satellite productivity observations.
19. The effectiveness of tree-log check dams to reduce post-fire sediment yields from terraced and contour-ploughed swales planted with maritime pine
Core Problem: Evidence for check-dam effectiveness after wildfire is extremely limited, especially across contrasting land treatments.
Key Innovation: Paired swales across terrace and contour-ploughed settings quantify two years of outlet sediment and retained deposits, showing treatment effectiveness varies with sediment supply.
20. Land Radiative Management Increases Heat Stress in Some Humid Regions
Core Problem: Albedo-based land cooling may suppress nearby rainfall and dry soils, but the resulting effect on human heat stress depends on humidity regime.
Key Innovation: Cloud-permitting simulations show that soil drying does not reliably lower relative humidity in humid climates, allowing wet-bulb temperature to rise despite reduced rainfall.
21. Dynamic response-adaptive rainfall thresholds for landslide early warning from long-term alert records using explainable artificial intelligence
Core Problem: Static rainfall thresholds cannot adapt to changing response patterns in long-term warning records.
Key Innovation: The verified title combines response-adaptive thresholds, long-term alert records and explainable AI; performance is not asserted without an abstract.
22. Frequency and magnitude of rainfall-triggered debris-flows in Mont Dore valley, Sancy Massif (French Central Massif)
Core Problem: Regional hazard assessment requires a defensible relationship between debris-flow occurrence frequency and event magnitude.
Key Innovation: The verified title establishes a joint rainfall-trigger and frequency-magnitude analysis for Mont Dore valley; record length and statistics are not inferred.
23. Geomorphological, lithological and age controls on the size and mobility of deep-seated landslides in the North Tanganyika–Kivu Rift region, Africa
Core Problem: The combined influence of geomorphology, lithology and landslide age on size and mobility is poorly constrained across rift terrain.
Key Innovation: The verified title integrates three geological controls across the North Tanganyika-Kivu Rift; inventory size, dating and quantitative effects are not inferred.
24. Mining-induced overburden failure and aquifer disturbance beneath karstified strata: Knowledge-guided neural-operator prediction
Core Problem: Rapid prediction is needed for interacting mining-induced rock-mass failure and aquifer disturbance.
Key Innovation: The verified title specifies a knowledge-guided neural-operator formulation; architecture, accuracy and validation are not inferred without an abstract.
25. A graph-based surrogate modeling framework for large-scale spatial optimization of low-impact development layouts in urban pluvial flood mitigation
Core Problem: Large-scale spatial optimization of flood-mitigation layouts is computationally expensive.
Key Innovation: The verified title specifies a graph-based surrogate for layout optimization; case scale, speed and mitigation gains are not inferred.
26. A spatiotemporal decomposition–coordination optimization method for multi-objective real-time flood-control operation of cascade reservoir systems
Core Problem: Cascade-reservoir decisions must coordinate multiple objectives across coupled spatial and temporal scales.
Key Innovation: The verified title specifies spatiotemporal decomposition-coordination optimization; basin scale and operational gains are not inferred.
27. Drought recurrence across China under diverse ENSO and IOD modes (1941–2023)
Core Problem: Long-term drought recurrence must be resolved across distinct coupled ocean-atmosphere modes.
Key Innovation: The verified title spans 1941-2023 and jointly stratifies ENSO and IOD modes; metrics and statistical effects are not inferred.
28. A Bayesian belief network model for seismic damage assessment of masonry buildings
Core Problem: Seismic loss analysis must represent uncertain and interacting drivers of masonry-building damage.
Key Innovation: The verified title specifies a Bayesian belief-network model; variables, data and predictive performance are not inferred without an abstract.
29. An integrated CFD-DEM framework for resolving the hydromechanical evolution of large-scale landslides induced by reservoir water level fluctuations
Core Problem: Resolving coupled fluid and granular dynamics during reservoir-induced slope failure.
Key Innovation: The title identifies an integrated CFD-DEM framework, but the supplied abstract contains only bibliographic metadata.
30. Fatality outcomes in the 2025 Ms 6.8 Tingri earthquake: Roles of seismic hazard, structural damage, and human behavioral response
Core Problem: Explaining fatality outcomes in the 2025 Tingri earthquake.
Key Innovation: The title indicates an integrated hazard-structure-behavior analysis, but the supplied abstract contains only bibliographic metadata.
31. A historical data curation (HDC) framework integrated with GIS-based multi-criteria decision analysis to inform flood risk prioritization in data-scarce regions
Core Problem: Curating historical evidence and prioritizing flood risk in data-scarce regions.
Key Innovation: The title specifies integration of historical data curation with GIS multicriteria decision analysis; the supplied abstract contains only metadata.
32. Earth-Agent-Pro: Towards Real-World Full-Chain Earth Observation with Agents
Core Problem: Executing open-world Earth-observation tasks from high-level scientific questions without pre-supplied imagery or prepared inputs.
Key Innovation: Expert-constrained planning, dependency-aware workflow memory with suffix repair, specialized planner/executor tuning, and a 744-question execution benchmark.
33. 4D Parallelism Unlocks Exascale Bayesian Neural Networks for High-Fidelity Atmospheric Modeling
Core Problem: High-resolution Bayesian atmospheric models and large ensembles are computationally prohibitive at global scale.
Key Innovation: BEAST combines a Bayesian Swin Transformer with orthogonal domain, tensor, and uncertainty parallelism, scaling to 3.96 EFLOP/s on 20,480 GH200 GPUs.
34. SeismoStats: A Python Package for Statistical Seismology
Core Problem: Making essential statistical-seismology analysis reproducible and accessible.
Key Innovation: A tested, documented Python package for catalog acquisition and manipulation, visualization, Gutenberg–Richter a- and b-values, and completeness magnitude estimation.
35. Airborne Hyperspectral Lidar Detection of Clear Air Turbulence Via Fabry–Perot Etalon Four-Edge Frequency Discrimination: Method and Simulation-Based Performance
Core Problem: Weak molecular backscatter and aerosol contamination hinder long-range clear-air-turbulence detection.
Key Innovation: A compact three-etalon, four-edge frequency-discrimination system with correction for thermal and laser-frequency drift.
36. A global coastal tide model from multi-mission and wide-swath altimetry: GOAT-1C
Core Problem: Global tide priors retain substantial errors near complex coastlines where conventional altimetry coverage is limited.
Key Innovation: DeepONet learns continuous spatial admittance corrections to FES2022 from 30 years of nadir altimetry and wide-swath SWOT data at 4 km resolution.
37. From Wood–Foliar Semantics to 3D Fuel Characterization: Transferable Deep Learning for Terrestrial LiDAR Across Global Forest Ecosystems
Core Problem: Forest heterogeneity, occlusion, and platform differences limit transfer of wood-foliar segmentation across ecosystems.
Key Innovation: A platform-agnostic point-cloud framework trained on harmonized global benchmarks reduces forests to transferable wood-foliar semantics and then decomposes them into five fuel components.
38. A biopolymer-solid waste composite binder for stabilizing dredged sediments and its use in riverbank slope protection
Core Problem: Achieving mechanical stability, erosion resistance, environmental safety, and waste reuse with fluid dredged sediments.
Key Innovation: A xanthan-gum solid-waste binder, functionally graded field design, cyclic durability tests, sequential-rainfall erosion tests, and microstructural mechanism analysis.
39. Ensemble Forecast Updates without Model Re-integration Based on Ultra-rapid Data Assimilation: Idealized Experiments with a Heavy Rainfall Case
Core Problem: Assimilating high-frequency observations into ensembles without repeatedly integrating the forecast model.
Key Innovation: Ultra-rapid ensemble updates use forecast-error covariance to improve hourly rainfall and atmospheric state estimates in a realistic heavy-rain case.
40. Physical-State-Guided Diffusion Sampling for Full-Waveform Inversion
Core Problem: Full-waveform inversion is nonlinear and ill-posed, while direct denoiser-wave-solver coupling can provide unreliable physical guidance.
Key Innovation: A persistent physical velocity state is bridged to a diffusion geological prior, separating wave-equation and denoiser gradients while retaining FWI optimization history.
41. Physics-enriched neural solvers for transient ice-flow simulation
Core Problem: Repeated nonlinear higher-order ice-flow solves are expensive as glacier geometry evolves.
Key Innovation: Low-order physics-derived input fields enrich a neural online solver without changing its energy objective or requiring offline training data.
42. Hurdle-RMIL: Addressing Zero Inflation and Long-Tailed Imbalance in Infrared Rainfall Retrieval
Core Problem: Satellite infrared rain-rate retrieval systematically underestimates rare intense rainfall because labels are zero-inflated and long-tailed.
Key Innovation: A hurdle formulation separates zero rain from positive rainfall, while a Bayes-derived RMIL transformation learns a balanced positive-rain distribution from natural imbalanced samples.
43. Genesis: A Generative Engine for Hierarchical Satellite Image Synthesis
Core Problem: Existing satellite generators cannot jointly maintain cross-scale and spatial consistency over a tile pyramid.
Key Innovation: A quadtree engine composing vertical super-resolution and horizontal mask-based outpainting, accompanied by the dense500 multiscale benchmark.
44. PWVGAT: Graph Attention Network for Spatial Inference of GNSS Precipitable Water Vapor Over Irregular Station Networks
Core Problem: GNSS PWV is accurate but spatially sparse and poorly handled by isotropic or regular-grid interpolation.
Key Innovation: Masked self-supervised edge-enhanced graph attention learns anisotropic station relationships and predicts unseen network locations.
45. Deformation Prediction and Analysis of the Hong Kong–Zhuhai–Macao Bridge Based on Time-Series InSAR and TCN–Transformer–Cross Attention
Core Problem: Separating and forecasting long-term, seasonal, and meteorologically influenced deformation across a long cross-sea bridge system.
Key Innovation: Enhanced SBAS-InSAR from 172 Sentinel-1 scenes coupled with a TCN–Transformer–cross-attention forecast model incorporating meteorological variables.
46. TPHH: a long-term (1901–2023) high-resolution (1∕30°) near-surface humidity dataset for the Tibetan Plateau generated via spatial downscaling based on hybrid-structure deep learning
Core Problem: Long-term high-resolution near-surface humidity observations are lacking over complex plateau terrain.
Key Innovation: FourCastNet-based downscaling jointly maps total-field climate signals and terrain constraints for 1901–2023 temperature, humidity, and pressure.
47. From 2D surface extent to 3D water levels: A SWOT-calibrated area–height framework for multidecadal reconstruction in hydraulically complex floodplain lakes
Core Problem: Two-dimensional surface extent must be converted into consistent long-term water levels where floodplain hydraulics are complex.
Key Innovation: The verified title specifies a SWOT-calibrated area-height framework; reconstruction accuracy and validation are not inferred.
48. Energy performance design and database of negative stiffness-equipped isolated structures with soil-structure interaction effects
Core Problem: Performance-based isolation design must include soil-structure interaction and energy response.
Key Innovation: The verified title reports a design framework and database for negative-stiffness-equipped isolated structures; scope and gains are not inferred.
49. Deep Learning Models for Mapping Surficial Geology in Selected Physiographic Regions of New York
Core Problem: Automating labor-intensive surficial geologic mapping across contrasting physiographic regions.
Key Innovation: Deep learning reproduces published deposit classes from only high-resolution lidar and existing geologic maps, with explicit tests of physiographic transfer.
50. Physics-Informed Conformal Prediction: Embedding PDE Consistency into Distribution-Free Uncertainty Quantification for Neural Operators
Core Problem: Providing rigorous uncertainty intervals for neural-operator PDE predictions.
Key Innovation: Adds PDE residuals to conformal nonconformity scores and identifies a boundary-condition limitation of FNOs resolved by coordinate channels.
51. Physically Consistent 2D to 3D Pore Space Synthesis: 1. Dynamic Grain Packing Under Strict Morphology Constraints
Core Problem: XCT resolution limits and artifacts compromise 3D reconstruction of heterogeneous pore geometry and flow properties.
Key Innovation: Newtonian dynamic grain packing, stereological size-distribution inversion, and virtual sub-particle insertion enforce contact, morphology, and flow constraints.
52. Morphological evolution and functional adaptation of a Submerged Sand Engine subjected to recurrent storms
Core Problem: How a submerged Sand Engine evolves and retains function under recurrent storm forcing.
Key Innovation: The abstract supplies only bibliographic metadata, so no method or result beyond the title can be verified.
53. RSGPNet: Geometric Prompting for Remote Sensing Open-Vocabulary Semantic Segmentation
Core Problem: Remote-sensing open-vocabulary masks are locally unstable, especially at complex boundaries.
Key Innovation: Text-guided coarse masks are converted to geometric prompts and refined under coarse-to-fine consistency verification.
54. Shipborne GNSS-IR for Sea Ice Freeboard Monitoring: A Case Study From the Chinese Polar Expedition
Core Problem: Obtaining accurate automated sea-ice freeboard from moving shipborne GNSS under harsh conditions.
Key Innovation: Two GNSS-IR retrieval approaches tested during an Antarctic expedition, including a draft-based method achieving sub-0.14 m RMSE.
55. A Geographically and Physics-Informed Convolutional Neural Network for Satellite-Derived Bathymetry Using ICESat-2 and Sentinel-2 Datasets
Core Problem: Sparse reference depths and spatially varying optical water and bottom properties limit satellite bathymetric inversion.
Key Innovation: A CNN fusing Sentinel-2, ICESat-2 samples, coordinates, and radiative-transfer-derived physics features.
56. Multimodal Remote Sensing Image Registration via Radiometric-Agnostic Geometric Discrepancy Estimation
Core Problem: Nonlinear radiometric differences obscure true geometric misalignment across remote-sensing modalities.
Key Innovation: A learned geometry-discrepancy evaluator trained from structure-preserving pseudo-pairs and reused for affine and deformable registration, plus the TRSR benchmark.
57. Are Geospatial Foundation Models Spectrally Consistent? A Band-Level Shapley Audit
Core Problem: Downstream accuracy does not reveal whether geospatial foundation-model representations are spectrally consistent with data or with one another.
Key Innovation: Band-level Shapley alignment with bootstrap uncertainty across four models and native Sentinel-2 bands.
58. Spectral–Texture Feature Fusion with Spatial Group Cross-Validation for Sea Ice Mapping Using HY-1C Satellite Imagery
Core Problem: Limited feature separability and spatial autocorrelation can inflate optical sea-ice classification accuracy.
Key Innovation: A spectral-texture fusion index combined with spatial group cross-validation, interpretable feature analysis, and cross-sensor and cross-region testing.
59. SemGeoNet: Semantic–Geometric Synergy Network for Optical Image-Guided Digital Surface Model Super-Resolution
Core Problem: Cross-modal misalignment, optical texture leakage, and inefficient attention degrade optical-guided DSM upscaling.
Key Innovation: Prototype-based semantic alignment, geometry-selective detail rectification, and gradient-prioritized sparse attention.
60. Integrating electrical resistivity tomography–derived metrics with ensemble machine learning to predict the depth-to-water in earthen levees
Core Problem: Predicting internal phreatic conditions from electrical-resistivity observations.
Key Innovation: The title indicates integration of ERT-derived metrics with ensemble machine learning; the supplied abstract contains metadata only, so details cannot be verified.
61. A review of phase-transition representation in permafrost land surface models: Mechanisms, algorithms, and impacts
Core Problem: Synthesizing how permafrost land models represent freezing and thawing and how those choices affect simulations.
Key Innovation: The title promises an integrated review of mechanisms, numerical algorithms, and impacts; the log provides no substantive abstract.
62. Ensembling or not? On the value of boosting and bagging for physics-aware machine learning in hydrology
Core Problem: Whether bagging and boosting materially improve physics-aware streamflow models over a well-configured single model.
Key Innovation: Controlled experiments across CAMELS-US catchments isolate effects of capacity, epochs, ensemble size, and training length.
63. Burial state controls the impact force and travelling distance of a movable cubic object in granular flows
Core Problem: Hazard models need to resolve how an object's initial burial state changes both its impact loading and mobility within granular flow.
Key Innovation: The verified publication identity establishes a coupled analysis of burial state, impact force and travel distance; quantitative mechanisms are not asserted until the correct abstract is available.
64. A Cover-based Enhanced Direct Search Approach for Contact Detection of Convex Polyhedral Blocks and Its Application with 3D DDA
Core Problem: Robust and efficient contact search among convex polyhedral blocks in 3D DDA.
Key Innovation: A cover-based enhanced direct-search approach is identified by the title; the supplied abstract contains only metadata.
65. Ground deformation dynamics and environmental controls in discontinuous permafrost regions: Insights from InSAR in the Da Xing'anling Mountains
Core Problem: The spatial and temporal controls on deformation in discontinuous-permafrost terrain remain difficult to resolve.
Key Innovation: The verified title identifies an InSAR analysis in the Da Xing'anling Mountains; time span, precision and causal attribution are not inferred.
66. Risk-based collaborative planning for cyber-physical-social distribution systems against extreme weather
Core Problem: Physical, cyber and social actions must be coordinated when extreme weather disrupts distribution systems.
Key Innovation: The verified title specifies a risk-based collaborative planning framework; hazard scenarios, network scale and gains are not inferred.
67. Assessing the global metal mining pollution risk to the marine environment
Core Problem: Global assessment is needed for how metal-mining pollution threatens the marine environment.
Key Innovation: The verified title establishes a global risk assessment; data coverage, synthesis design and results are not inferred.
68. First vegetation optical depth product retrieved from FengYun-3D X-band observations
Core Problem: FengYun-3D X-band observations lack a dedicated vegetation-optical-depth product.
Key Innovation: The verified title reports the first such FengYun-3D X-band product; coverage, resolution and validation are not inferred.
69. Emergence of macroscopic hydraulic behaviors from data: a hybrid physics-AI universal differential equation framework for spatio-temporal soil moisture dynamics
Core Problem: Macroscopic hydraulic behavior must be learned from data without discarding governing physical structure.
Key Innovation: The verified title specifies a hybrid physics-AI universal differential equation; constraints, scale and generalization are not inferred.
70. Feature Recovery for Object Understanding After Irreversible Fire Damage
Core Problem: Identifying objects whose geometry, material, and appearance have been irreversibly altered by fire.
Key Innovation: Introduces the TRACE benchmark and a frozen-host feature-recovery module aligning degraded objects with pristine representations.
71. Preemptive Ensemble Forecast Sensitivity to Observations
Core Problem: Estimating observation impact and selective-assimilation forecasts without model reintegration.
Key Innovation: PEFSO and two ADD-SEL-PRE approximations use tangent-linear updates to reproduce short-lead ensemble effects at lower cost.
72. An End-to-End Automated Pipeline for Controllable Crack Data Synthesis
Core Problem: Generating realistic cracks with controllable morphology, boundaries, and context.
Key Innovation: Procedural Bezier skeletons drive GAN masks and dual-ControlNet appearance/geometry synthesis.
73. Aligned Radiometric RGB-Thermal Fusion for UAV Facade Anomaly Screening
Core Problem: Detecting surface and subsurface facade anomalies despite sensor misalignment and weak thermal contrast.
Key Innovation: A sensor-level registration and signed radiometric contrast pipeline with a project-grouped UAV dataset.
74. AquaCubeAI-Powered Monitoring Turbidity on-board Φsat-2
Core Problem: Reducing latency and bandwidth demands in coastal turbidity monitoring.
Key Innovation: A compact multispectral regressor trained with spatial-block validation, reformulated for dense onboard mapping, and deployed on a low-power VPU.
75. Surrogate Modeling of 3D Rayleigh-Benard Convection with Equivariant Autoencoders
Core Problem: Large three-dimensional PDE systems require sample-efficient, symmetry-aware surrogate models across multiscale dynamics.
Key Innovation: An end-to-end equivariant autoencoder and ConvLSTM using horizontally D4-steerable kernels and partial vertical kernel sharing.
76. An intelligent method for predicting the escape characteristics of blowout leaks from offshore CO2 storage wells based on improved transformer and Bayesian-optimized SVR
Core Problem: Efficient prediction of multiphase CO2 leakage mass flow, bottom-hole pressure, and surface-reaching fraction across offshore well conditions.
Key Innovation: A 10,270-case simulation dataset and hybrid feature-wise Transformer with Bayesian-optimized SVR, supplemented by Sobol sensitivity analysis.
77. Assessing annual change in forest canopy mortality in Southern Finland, using very high-resolution satellite imagery and deep learning
Core Problem: Sparse observations limit large-area tracking of standing tree mortality.
Key Innovation: A U-Net trained on more than 21,000 segments and validated with satellite, drone, inventory, and canopy-height data across seven sites.
78. LTS-ProtoNet: Local–Temporal–Semantic Prototype Network for Remote Sensing Change Detection
Core Problem: Boundary ambiguity, temporal appearance interference, and object-scale variation impair bitemporal change maps.
Key Innovation: Learnable change-prototype bases drive local refinement, prototype-mediated temporal interaction, and multiscale semantic enhancement.
79. Initial sedimentary anisotropy-induced creep anomalies in calcareous sand and mechanism exploration
Core Problem: How depositional anisotropy and particle shape jointly control creep response and failure.
Key Innovation: Multistage triaxial creep and X-ray CT link macroscopic deformation to breakage, fabric, pores, and load-bearing frameworks.
80. An integrated snowfall-drifting snow model for roof snow distribution: Multi-case validation and parametric analysis
Core Problem: Predicting spatial roof snow loads under coupled snowfall and drifting.
Key Innovation: The title indicates an integrated model with multi-case validation and parameter analysis, but the supplied abstract contains only metadata.
81. Bioconstruction of Buffer Barrier: a new paradigm for enhancing Earth surface soil resilience to extreme climates
Core Problem: Earth-surface soils require protective systems that retain function under extreme climatic forcing.
Key Innovation: The verified title proposes bioconstruction of a buffer barrier; composition, scale and performance are not inferred without an abstract.
82. Assessment of multi-driver human interventions on streamflow and its components based on information and complexity metrics
Core Problem: Multiple human drivers must be separated in streamflow and component changes.
Key Innovation: The verified title combines information and complexity metrics; direct extreme-event evaluation is not inferred.
83. Optimizing evapotranspiration products by merging multi-source datasets with three no-prior-knowledge methods over China
Core Problem: Multiple evapotranspiration datasets require fusion without assuming prior product quality.
Key Innovation: The verified title compares three no-prior-knowledge fusion methods over China; resolution and gains are not inferred.
84. An automated adaptative framework for weather data interpolation
Core Problem: Weather interpolation must adapt automatically across variables and observation configurations.
Key Innovation: The verified title specifies an automated adaptive framework; variables, scale and comparative performance are not inferred.
85. Improvement and testing of the SWAT model for simulating hydrological processes and non-point source pollution in the paddy dominated polder
Core Problem: SWAT requires adaptation for coupled hydrology and non-point pollution in polder landscapes.
Key Innovation: The verified title specifies model improvement and testing; modifications and hazard performance are not inferred.
86. Explainable method for advanced early-warning of hypoxic events in estuaries: coupling Kolmogorov-Arnold networks with self-attention mechanism
Core Problem: Hypoxic-event warning needs both lead time and interpretable drivers.
Key Innovation: The verified title couples a Kolmogorov-Arnold network with self-attention; lead time and event validation are not inferred.
87. Operator learning for predicting bulk wave parameters of spectral wave models
Core Problem: Bulk parameters from spectral wave models are costly to predict repeatedly.
Key Innovation: The verified title applies operator learning to spectral-wave outputs; extreme-sea coverage and error are not inferred.
88. Reflection analysis for nonlinear shoaling waves with an opposing current
Core Problem: Opposing currents alter the reflection of nonlinear shoaling waves.
Key Innovation: The verified title couples nonlinear shoaling, reflection and opposing current; analytical or experimental validation is not inferred.